https://t.

me/MBS_MedicalBooksStore
Molecular Hematology
Molecular
Hematology
FOURTH EDITION

Edited by
Drew Provan MD FRCP FRCPath
Emeritus Reader in Autoimmune Haematology
Department of Haematology
Barts and The London School of Medicine and Dentistry
Queen Mary University of London, UK

John G. Gribben MD DSc FRCP FRCPath FMedSci

Professor of Medical Oncology
Barts Cancer Institute
Barts and The London School of Medicine and Dentistry
Queen Mary University of London, UK
This edition first published 2020
© 2020 by John Wiley & Sons
Edition History [3rd edition, 2010]
All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or
transmitted, in any form or by any means, electronic, mechanical, photocopying, recording or
otherwise, except as permitted by law. Advice on how to obtain permission to reuse material from this
title is available at http://www.wiley.com/go/permissions.
The right of Drew Provan to be identified as the author of editorial in this work has been asserted in
accordance with law.
Registered Office(s)
John Wiley & Sons, Inc., 111 River Street, Hoboken, NJ 07030, USA
John Wiley & Sons Ltd, The Atrium, Southern Gate, Chichester, West Sussex, PO19 8SQ, UK
Editorial Office
9600 Garsington Road, Oxford, OX4 2DQ, UK
For details of our global editorial offices, customer services, and more information about Wiley
products visit us at www.wiley.com.
Wiley also publishes its books in a variety of electronic formats and by print-on-demand. Some content
that appears in standard print versions of this book may not be available in other formats.
Limit of Liability/Disclaimer of Warranty
The contents of this work are intended to further general scientific research, understanding, and
discussion only and are not intended and should not be relied upon as recommending or promoting
scientific method, diagnosis, or treatment by physicians for any particular patient. In view of ongoing
research, equipment modifications, changes in governmental regulations, and the constant flow of
information relating to the use of medicines, equipment, and devices, the reader is urged to review and
evaluate the information provided in the package insert or instructions for each medicine, equipment,
or device for, among other things, any changes in the instructions or indication of usage and for added
warnings and precautions. While the publisher and authors have used their best efforts in preparing this
work, they make no representations or warranties with respect to the accuracy or completeness of the
contents of this work and specifically disclaim all warranties, including without limitation any implied
warranties of merchantability or fitness for a particular purpose. No warranty may be created or
extended by sales representatives, written sales materials or promotional statements for this work. The
fact that an organization, website, or product is referred to in this work as a citation and/or potential
source of further information does not mean that the publisher and authors endorse the information or
services the organization, website, or product may provide or recommendations it may make. This work
is sold with the understanding that the publisher is not engaged in rendering professional services. The
advice and strategies contained herein may not be suitable for your situation. You should consult with a
specialist where appropriate. Further, readers should be aware that websites listed in this work may have
changed or disappeared between when this work was written and when it is read. Neither the publisher
nor authors shall be liable for any loss of profit or any other commercial damages, including but not
limited to special, incidental, consequential, or other damages.
Library of Congress Cataloging-in-Publication Data
Names: Provan, Drew, editor. | Gribben, John editor.
Title: Molecular hematology / edited by Drew Provan, John G. Gribben.
Description: Fourth edition. | Hoboken, NJ : Wiley-Blackwell, 2020. | Includes bibliographical
references and index.
Identifiers: LCCN 2019034376 (print) | LCCN 2019034377 (ebook) | ISBN 9781119252870 (hardback) |
ISBN 9781119252955 (adobe pdf) | ISBN 9781119252931 (epub)
Subjects: MESH: Hematologic Diseases | Molecular Biology–methods
Classification: LCC RC636 (print) | LCC RC636 (ebook) | NLM WH 120 | DDC 616.1/5–dc23
LC record available at https://lccn.loc.gov/2019034376
LC ebook record available at https://lccn.loc.gov/2019034377
Cover Design: Wiley
Cover Image: © JUAN GAERTNER/SCIENCE PHOTO LIBRARY/Getty Images
Set in 10/12pt MinionPro by Aptara Inc., New Delhi, India

10 9 8 7 6 5 4 3 2 1
Dedication

We would like to dedicate this book to two people: our dear friend and colleague, Professor Sir David Weatherall, who sadly
passed away on 8 December 2018. He was truly a pioneer of molecular biology and was the first physician scientist to use
molecular techniques to study hematological disease. We will all miss him very much.
In addition, we would like to dedicate the book to Val Provan. Always in our thoughts and much missed.
Contents

Contributors, ix 13 The molecular basis of iron metabolism, 161

Preface to the fourth edition, xiii Nancy C. Andrews & Tomas Ganz

Further reading, xv 14 Hemoglobinopathies due to structural mutations, 173

D. Mark Layton & Steven Okoli
Acknowledgments, xvi
15 Molecular pathogenesis of malaria, 193
David J. Roberts, Arnab Pain & Chetan E. Chitnis
1 Beginnings: the molecular pathology of hemoglobin, 1
16 Molecular coagulation and thrombophilia, 207
David Weatherall
Björn Dahlbäck & Andreas Hillarp
2 Stem cells, 21
17 The molecular basis of hemophilia, 221
David T. Scadden
Daniel P. Hart & Paul L.F. Giangrande
3 The genetics of acute myeloid leukemias, 37
18 The molecular basis of von Willebrand disease, 235
Amy M. Trottier & Carolyn J. Owen
Luciano Baronciani
4 Molecular diagnostics and risk assessment in
19 Platelet disorders, 251
myeloid malignancies, 49
Kenneth J. Clemetson
Christian Scharenberg & Torsten Haferlach
20 The molecular basis of blood cell alloantigens, 267
5 Molecular basis of acute lymphoblastic leukemia, 59
Cristina Navarrete, Louise Tilley, Winnie Chong
Bela Patel & Fiona Fernando
& Colin J. Brown
6 Chronic myeloid leukemia, 71
21 Functions of blood group antigens, 285
Hagop Kantarjian, Jorge Cortes, Elias Jabbour &
Jonathan S. Stamler & Marilyn J. Telen
Susan O’Brien
22 Autoimmune hematological disorders, 297
7 Myeloproliferative neoplasms, 87
Drew Provan & John W. Semple
Jyoti Nangalia, Anthony J. Bench, Anthony R.
Green & Anna L. Godfrey 23 Molecular therapeutics in hematology: gene
therapy, 319
8 Lymphoma genetics, 101
William M. McKillop & Jeffrey A. Medin
Jennifer L. Crombie, Anthony Letai & John G. Gribben
24 Pharmacogenomics, 339
9 The molecular biology of chronic lymphocytic
Leo Kager & William E. Evans
leukemia, 111
John G. Gribben 25 History and development of molecular biology, 353
Paul Moss
10 The molecular biology of multiple myeloma, 121
Wee Joo Chng & P. Leif Bergsagel 26 Cancer stem cells, 363
Sara Ali & Dominique Bonnet
11 The molecular basis of bone marrow failure
syndromes and red cell enzymopathies, 131 27 Molecular basis of transplantation, 373
Deena Iskander, Lucio Luzzatto & Anastasios Francesco Dazzi & Antonio Galleu
Karadimitris
12 Anemia of chronic disease, 155 Index, 389
Tomas Ganz

vii
Contributors
Sara Ali MD
Haematopoietic Stem Cell Laboratory, The Francis Crick
Institute, London, UK
Nancy C. Andrews MD, PhD
Duke University School of Medicine, Durham, NC, USA
Luciano Baronciani PhD
Fondazione IRCCS Ca’ Granda Ospedale Maggiore Policlin-
ico, Angelo Bianchi Bonomi Hemophilia and Thrombosis
Center, Milan, Italy
Anthony J. Bench MA, PhD
Laboratory Medicine, NHS Lothian, Edinburgh, UK
P. Leif Bergsagel MD
Division of Hematology-Oncology, Comprehensive Cancer
Center, Mayo Clinic Arizona, Scottsdale, AZ, USA
Dominique Bonnet PhD
Haematopoietic Stem Cell Laboratory, The Francis Crick
Institute, London, UK
Colin J. Brown PhD, FRCPath
Histocompatibility and Immunogenetics Laboratory, NHS
Blood and Transplant; Faculty of Life Sciences & Medicine,
King’s College London, London, UK
Chetan E. Chitnis MSc, MA, PhD
Malaria Group, Pasteur Institute, Paris, France
Wee Joo Chng MB ChB, MRCP, FRCPath
National University Cancer Institute, National Univer-
sity Health System of Singapore; University of Singapore,
National University Hospital, Singapore
Winnie Chong PhD
Histocompatibility and Immunogenetics Service Develop-
ment Laboratory, NHS Blood and Transplant, London, UK
Kenneth J. Clemetson PhD, ScD, CChem, FRSC
Department of Haematology, Inselspital, University of Berne,
Berne, Switzerland
Jorge Cortes MD
Department of Leukemia, University of Texas MD Anderson
Cancer Center, Houston, TX, USA
Jennifer L. Crombie MD
Department of Medical Oncology, Dana-Farber Cancer
Institute, Harvard Medical School, Boston, MA, USA
Björn Dahlbäck MD, PhD
Department of Translational Medicine, Section of Clinical
Chemistry, Lund University, University Hospital, Malmö,
Sweden
Francesco Dazzi MD, PhD
School of Cancer & Pharmaceutical Sciences, King’s College
London; King’s Health Partners Cancer Research UK Centre,
London, UK
William E. Evans PharmD
St Jude Children’s Research Hospital, Memphis, TN, USA
Fiona Fernando MD
Centre of Haemato-Oncology, Barts Cancer Institute, Queen
Mary University of London, London, UK
Antonio Galleu MD, PhD
School of Cancer & Pharmaceutical Sciences, King’s College
London; King’s Health Partners Cancer Research UK Centre,
London, UK
Tomas Ganz PhD, MD
Department of Medicine, David Geffen School of Medicine
at UCLA, Los Angeles, CA, USA
ix
x Contributors
Paul L.F. Giangrande MD, FRCP, FRCPath, FRCPCH
Formerly of Oxford Haemophilia and Thrombosis Centre,
Churchill Hospital, Oxford, UK
Anna L. Godfrey PhD, MRCP, FRCPath
Department of Haematology, Cambridge University Hospi-
tals NHS Foundation Trust, Cambridge, UK
Anthony R. Green PhD, FRCP, FRCPath, FMedSci
Department of Haematology, Cambridge Institute for Medi-
cal Research; Wellcome Medical Research Council Stem Cell
Institute, Cambridge, UK
John G. Gribben MD, DSc, FRCP, FRCPath, FMedSci
Barts Cancer Institute, Barts and The London School of
Medicine and Dentistry, Queen Mary University of London,
London, UK
Torsten Haferlach MD
MLL Munich Leukemia Laboratory, Munich, Germany
Daniel P. Hart FRCP, FRCPath, PhD
The Royal London Hospital Haemophilia Centre, Barts and
The London School of Medicine and Dentistry, Queen Mary
University of London, London, UK
Andreas Hillarp PhD
Department of Clinical Chemistry and Transfusion
Medicine, Halland County Hospital, Halmstad, Sweden
Deena Iskander MD, PhD, MRCP
Centre for Haematology, Imperial College London, Ham-
mersmith Hospital, London, UK
Elias Jabbour MD
Department of Leukemia, University of Texas MD Anderson
Cancer Center, Houston, TX, USA
Leo Kager MD
Department of Pediatrics, St. Anna Children’s Hospital,
Medical University Vienna, Austria
Hagop Kantarjian MD
Department of Leukemia, University of Texas MD Anderson
Cancer Center, Houston, TX, USA
Anastasios Karadimitris PhD, MRCP, FRCPath
Department of Haematology and Blood Transfusion,
Muhimbili University College of Health Sciences, Dar-es-
Salaam, Tanzania
D. Mark Layton MB BS, FRCP, FRCPCH
Center for Hematology, Imperial College London, London,
UK
Anthony Letai MD, PhD
Department of Medical Oncology, Dana-Farber Cancer
Institute, Harvard Medical School, Boston, MA, USA
Lucio Luzzatto MD
Department of Haematology and Blood Transfusion,
Muhimbili University College of Health Sciences, Dar-es-
Salaam, Tanzania
William M. McKillop PhD
Department of Pediatrics, Medical College of Wisconsin,
Milwaukee, WI, USA
Jeffrey A. Medin PhD
Departments of Pediatrics and Biochemistry, Medical Col-
lege of Wisconsin, Milwaukee, WI, USA
Paul Moss MD, PhD
School of Cancer Sciences, University of Birmingham, Birm-
ingham, UK
Jyoti Nangalia PhD, MRCP, FRCPath
Welcome Sanger Institute, Hinxton; Department of Haema-
tology, University of Cambridge; Wellcome-Medical
Research Council Cambridge Stem Cell Institute, Cam-
bridge, UK
Cristina Navarrete PhD, FRCPath
Histocompatibility and Immunogenetics Service Develop-
ment Department, NHS Blood and Transplant; Department
of Immunology and Molecular Pathology, University College
London, London, UK
Susan O’Brien MD
Department of Leukemia, University of Texas MD Anderson
Cancer Center, Houston, TX, USA
Steven Okoli MBChB, FRCP, FRCPath
Center for Hematology, Imperial College London, London,
UK
Carolyn J. Owen MD, MDres(UK), FRCPC
Division of Hematology and Hematological Malignancies,
University of Calgary, Foothills Medical Centre, Calgary,
Canada
Arnab Pain PhD
Biological and Environmental Sciences and Engineering
(BESE) Division, King Abdullah University of Science and
Technology, Jeddah, Saudi Arabia; Nuffield Division of Clin-
ical Laboratory Sciences (NDCLS), The John Radcliffe Hos-
pital, University of Oxford, Headington, Oxford, UK

Bela Patel MD, FRCPath, MD(res)
Centre of Haemato-Oncology, Barts Cancer Institute, Queen
Mary University of London, London, UK
Drew Provan MD, FRCP, FRCPath
Blizard Institute, Barts and The London School of Medicine
and Dentistry, Queen Mary University of London, London,
UK
David J. Roberts DPhil, MRCP, FRCPath
National Health Service Blood and Transplant (Oxford), The
John Radcliffe Hospital, Oxford, UK
David T. Scadden MD
Department of Stem Cell and Regenerative Biology, Harvard
Stem Cell Institute, Harvard University; Center for Regenera-
tive Medicine, Massachusetts General Hospital, Boston, MA,
USA
Christian Scharenberg MD, PhD
Department of Hematology, Skaraborgs Hospital, Skövde;
Department of Cell and Molecular Biology, Karolinska Insti-
tute, Stockholm, Sweden
John W. Semple PhD
Division of Hematology and Transfusion Medicine, Lund
University, Lund, Sweden
Contributors xi
Jonathan S. Stamler MD
Harrington Discovery Institute and Institute of Transfor-
mative Molecular Medicine, University Hospitals Cleveland
Medical Center and Case Western Reserve University, Cleve-
land, OH, USA
Marilyn J. Telen MD
Department of Medicine, Division of Hematology, Duke
University Medical Center, Durham, NC, USA
Louise Tilley PhD
International Blood Group Reference Laboratory, NHS
Blood and Transplant, Bristol, UK
Amy M. Trottier MSc, MD, FRCPC
Division of Hematology and Hematological Malignancies,
University of Calgary, Foothills Medical Centre, Calgary,
Canada
David Weatherall MD, FRCP, FRS
Formerly of Weatherall Institute of Molecular Medicine, The
John Radcliffe Hospital, Oxford, UK
Preface to the fourth edition
Hematology is a fast-moving discipline with innovation both
diagnostically and therapeutically. In the 19 years since the
first edition of Molecular Hematology was published, many
advances have been made. Molecular techniques have helped
explain the basis of many diseases, starting initially with red
cell disorders and hemostasis. However, thanks to the use of
molecular biology we can now diagnose and stratify patients
with diseases such as leukemia, myeloma, myeloproliferative
neoplasms, and others. Such advances in technology have not
only helped explain the underlying basis of the diseases, but
have also provided targets for treatment.
The world of red cells started the whole specialty of molec-
ular medicine and, using molecular biology techniques,
many of the phenotypic features of red cell disorders have
been explained. This is discussed eloquently by the late Pro-
fessor Sir David Weatherall at the beginning of the book.
Other non-malignant areas which have been updated include
the Functions of Blood Group Antigens, von Willebrand Dis-
ease, and Platelet Disorders.
Undoubtedly, hemato-oncology has seen the biggest
explosion in terms of understanding the molecular basis of
diseases such as leukemias, lymphomas, the myeloprolifer-
ative neoplasms, myeloma, and myelodysplastic syndromes.
The huge array of biological markers makes stratification and
treatment much more sophisticated than ever before. All the
chapters dealing with malignant blood diseases have been
thoroughly revised and brought up to date.
However, despite the growing complexity in terms of
pathogenesis, diagnosis, and management of patients with
blood diseases, the ethos of the book remains the same –
namely, to provide a succinct account of the molecular biol-
ogy of hematological disease written at a level where it should
be of benefit to both the seasoned molecular biologist and
the practicing clinician alike. We have retained the original
structure for the chapters, high-quality artwork, and Further
Reading sections in order to make the book visually appealing
and relevant to modern hematology practice.
We very much hope you enjoy this edition and, as always,
we welcome any comments or suggestions from readers,
which we will attempt to incorporate into the next edition.
Drew Provan
John Gribben
xiii
Further reading
Anderson, K.C. and Ness, P.M. (eds.) (2000). Scientific Basis of Transfu-
sion Medicine: Implications for Clinical Practice, 2e. Philadelphia, PA:
WB Saunders.
Beutler, E. and Lichtman, M.A. (eds.). Williams’ Hematology, 6e. New
York: McGraw-Hill.
Cooper, G.M. (1997). The Cell: A Molecular Approach. Washington, DC:
ASM Press.
Cox, T.M. and Sinclair, J. (1997). Molecular Biology in Medicine. Oxford:
Blackwell Science.
Jameson, J.L. (ed.) (1998). Principles of Molecular Medicine. New York:
Humana Press.
Mullis, K.B. (1990). The unusual origin of the polymerase chain reac-
tion. Scientific American 262: 56–65.
Roitt, I. (2001). Roitt’s Essential Immunology, 10e. Oxford: Blackwell Sci-
ence.
Stamatoyannopoulos, G., Nienhuis, A.W., Majerus, P.W., and Varmus,
H. (eds.) (2000). The Molecular Basise of Blood Diseases, 2e. Philadel-
phia, PA: W.B. Saunders.
Watson, J.D., Gilman, M., Witkowski, J., and Zoller, M. (eds.) (1992).
Recombinant DNA, 2e. New York: Scientific American Books.
xv
Acknowledgments

We would like to express thanks to Claire Bonnett, Publisher, and Deirdre Barry, Senior Editorial Assistant, for their help with
this work.

xvi
Chapter 1 Beginnings: the molecular
pathology of hemoglobin
David Weatherall
Weatherall Institute of Molecular Medicine, John Radcliffe Hospital, Oxford, UK

Historical background, 1 Molecular aspects of the high frequency of the hemoglobin variants, 17
The structure, genetic control, and synthesis of normal hemoglobin, 2 Molecular aspects of the prevention and management of the
The molecular pathology of hemoglobin, 6 hemoglobin disorders, 18
Genotype–phenotype relationships in the thalassemias, 12 Postscript, 18
Structural hemoglobin variants, 16 Further reading, 18

within bacteria. The steady improvement in the proper-

Historical background ties of cloning vectors made it possible to generate libraries
of human DNA growing in bacterial cultures. Ingenious
Linus Pauling first used the term “molecular disease” in 1949,
approaches were developed to scan the libraries to detect
after the discovery that the structure of sickle cell hemoglobin
genes of interest; once pinpointed, the appropriate bacte-
differed from that of normal hemoglobin. Indeed, it was this
rial colonies could be grown to generate larger quantities of
seminal observation that led to the concept of molecular
DNA carrying a particular gene. Later it became possible
medicine, the description of disease mechanisms at the level
to sequence these genes, persuade them to synthesize their
of cells and molecules. However, until the development of
products in microorganisms, cultured cells, or even other
recombinant DNA technology in the mid-1970s, knowledge
species, and hence to define their key regulatory regions.
of events inside the cell nucleus, notably how genes function,
The early work in the field of human molecular genet-
could only be the subject of guesswork based on the structure
ics focused on diseases in which there was some knowl-
and function of their protein products. However, as soon as
edge of the genetic defect at the protein or biochemical level.
it became possible to isolate human genes and to study their
However, once linkage maps of the human genome became
properties, the picture changed dramatically.
available, following the identification of highly polymorphic
Progress over the last 30 years has been driven by techno-
regions of DNA, it was possible to search for any gene for a
logical advances in molecular biology. At first it was possible
disease, even where the cause was completely unknown. This
only to obtain indirect information about the structure and
approach, first called reverse genetics and later rechristened
function of genes by DNA/DNA and DNA/RNA hybridiza-
positional cloning, led to the discovery of genes for many
tion; that is, by probing the quantity or structure of RNA
important diseases.
or DNA by annealing reactions with molecular probes. The
As methods for sequencing were improved and automated,
next major advance was the ability to fractionate DNA into
thoughts turned to the next major goal in this field, which was
pieces of predictable size with bacterial restriction enzymes.
to determine the complete sequence of the bases that con-
This led to the invention of a technique that played a central
stitute our genes and all that lies between them: the Human
role in the early development of human molecular genetics,
Genome Project. This remarkable endeavor was finally com-
called Southern blotting after the name of its developer, Edwin
pleted in 2006. The further understanding of the functions
Southern. This method allowed the structure and organiza-
and regulation of our genes will require multidisciplinary
tion of genes to be studied directly for the first time and led
research encompassing many different fields. The next stage
to the definition of a number of different forms of molecular
in the Human Genome Project, called genome annotation,
pathology.
entails analyzing the raw DNA sequence in order to deter-
Once it was possible to fractionate DNA, it soon became
mine its biological significance. One of the main ventures
feasible to insert the pieces into vectors able to divide
in the era of functional genomics will be in what is termed
proteomics, the large-scale analysis of the protein products of
Molecular Hematology, 4th Edition. Edited by Drew Provan and John G. Gribben.
genes. The ultimate goal will be to try to define the protein
© 2020 John Wiley & Sons Ltd. Published 2020 by John Wiley & Sons Ltd.

1
2 Molecular Hematology

complement, or proteome, of cells and how the many dif-

ferent proteins interact with one another. To this end, large-
The structure, genetic control, and
scale facilities are being established for isolating and purify-
synthesis of normal hemoglobin
ing the protein products of genes that have been expressed
in bacteria. Their structure can then be studied by a vari- Structure and function
ety of different techniques, notably X-ray crystallography
The varying oxygen requirements during embryonic, fetal,
and nuclear magnetic resonance spectroscopy. The crys-
and adult life are reflected in the synthesis of different
tallographic analysis of proteins is being greatly facilitated
structural hemoglobins at each stage of human development.
by the use of X-ray beams from a synchrotron radiation
However, they all have the same general tetrameric struc-
source.
ture, consisting of two different pairs of globin chains, each
In the last few years both the utility and extreme com-
attached to one heme molecule. Adult and fetal hemoglobins
plexity of the fruits of the genome project have become
have α chains combined with β chains (Hb A, α2 β2 ),
apparent. The existence of thousands of single-nucleotide
δ chains (Hb A2 , α2 δ2 ), and γ chains (Hb F, α2 γ2 ). In
polymorphisms (SNPs) has made it possible to search for
embryos, α-like chains called ζ chains combine with γ
genes of biological or medical significance. The discovery of
chains to produce Hb Portland (ζ2 γ2 ), or with ε chains to
families of regulatory RNAs and proteins is starting to shed
make Hb Gower 1 (ζ2 ε2 ), while α and ε chains form Hb
light on how the functions of the genome are controlled,
Gower 2 (α2 ε2 ). Fetal hemoglobin is heterogeneous; there
and studies of acquired changes in its structure, epigenetics,
are two varieties of γ chain that differ only in their amino
promise to provide similar information. Recent develop-
acid composition at position 136, which may be occupied by
ments in new-generation sequencing of DNA and RNA are
either glycine or alanine; γ chains containing glycine at this
also providing invaluable information about many aspects
position are called G γ chains, those with alanine A γ chains
of gene regulation.
(Figure 1.1).
During this remarkable period of technical advance, con-
The synthesis of hemoglobin tetramers consisting of two
siderable progress has been made toward an understanding
unlike pairs of globin chains is absolutely essential for the
of the pathology of disease at the molecular level. This has
effective function of hemoglobin as an oxygen carrier. The
had a particular impact on hematology, leading to advances
classical sigmoid shape of the oxygen dissociation curve,
in the understanding of gene function and disease mecha-
which reflects the allosteric properties of the hemoglobin
nisms in almost every aspect of the field.
molecule, ensures that, at high oxygen tensions in the lungs,
The inherited disorders of hemoglobin – the thalassemias
oxygen is readily taken up and later released effectively at the
and structural hemoglobin variants, the commonest human
lower tensions encountered in the tissues. The shape of the
monogenic diseases – were the first to be studied systemati-
curve is quite different to that of myoglobin, a molecule that
cally at the molecular level and a great deal is known about
consists of a single globin chain with heme attached to it,
their genotype–phenotype relationships. This field led the
which, like abnormal hemoglobins that consist of homote-
way to molecular hematology and, indeed, to the develop-
tramers of like chains, has a hyperbolic oxygen dissociation
ment of molecular medicine. Thus, even though the genet-
curve.
ics of hemoglobin is complicated by the fact that different
The transition from a hyperbolic to a sigmoid oxygen
varieties are produced at particular stages of human devel-
dissociation curve, which is absolutely critical for normal
opment, the molecular pathology of the hemoglobinopathies
oxygen delivery, reflects cooperativity between the four
provides an excellent model system for understanding any
heme molecules and their globin subunits. When one of
monogenic disease and the complex interactions between
them takes on oxygen, the affinity of the remaining three
genotype and environment that underlie many multigenic
increases markedly; this happens because hemoglobin can
disorders.
exist in two configurations, deoxy(T) and oxy(R), where T
In this chapter I consider the structure, synthesis, and
and R represent the tight and relaxed states, respectively.
genetic control of the human hemoglobins, describe the
The T configuration has a lower affinity than the R for
molecular pathology of the thalassemias, and discuss briefly
ligands such as oxygen. At some point during the addition
how the complex interactions of their different genotypes
of oxygen to the hemes, the transition from the T to the R
produce a remarkably diverse family of clinical phenotypes;
configuration occurs and the oxygen affinity of the partially
the structural hemoglobin variants are discussed in more
liganded molecule increases dramatically. These allosteric
detail in Chapter 14. Readers who wish to learn more about
changes result from interactions between the iron of the
the methods of molecular genetics, particularly as applied to
heme groups and various bonds within the hemoglobin
the study of hemoglobin disorders, are referred to the reviews
tetramer, which lead to subtle spatial changes as oxygen is
cited at the end of this chapter.
taken on or given up.

31 32 99 100
ζ ψζ ψα2 ψα1 α2 α1 θ1
16
ζ2ε2 ζ2γ2 α2ε2
Hb Gower 1 Hb Portland Hb Gower 2
Embryo
Fig. 1.1 The genetic control of human hemoglobin production in embryonic, fetal, and adult life. The standard names for these genes
are as follows: Alpha genes HBA1 and HBA2, Beta gene HBB, Gamma genes HBG1 and HBG2, Delta gene HBD, and the embryonic genes HBE1
and HBZ.
The precise tetrameric structures of the different human
hemoglobins, which reflect the primary amino acid
sequences of their individual globin chains, are also vital
for the various adaptive changes that are required to ensure
adequate tissue oxygenation. The position of the oxygen
dissociation curve can be modified in several ways. For
example, oxygen affinity decreases with increasing CO2 ten-
sion (the Bohr effect). This facilitates oxygen loading to the
tissues, where a drop in pH due to CO2 influx lowers oxygen
affinity; the opposite effect occurs in the lungs. Oxygen affin-
ity is also modified by the level of 2,3-diphosphoglycerate
(2,3-BPG) in the red cell. Increasing concentrations shift
the oxygen dissociation curve to the right (i.e. they reduce
oxygen affinity), while diminishing concentrations have the
opposite effect. 2,3-BPG fits into the gap between the two β
chains when it widens during deoxygenation, and interacts
with several specific binding sites in the central cavity of
the molecule. In the deoxy configuration the gap between
the two β chains narrows and the molecule cannot be
accommodated. With increasing concentrations of 2,3-BPG,
which are found in various hypoxic and anemic states,
more hemoglobin molecules tend to be held in the deoxy
configuration and the oxygen dissociation curve is therefore
shifted to the right, with more effective release of oxygen.
Fetal red cells have greater oxygen affinity than adult
red cells, although, interestingly, purified fetal hemoglobin
has an oxygen dissociation curve similar to that of adult
hemoglobin. These differences, which are adapted to the oxy-
gen requirements of fetal life, reflect the relative inability of
Hb F to interact with 2,3-BPG compared with Hb A. This is
because the γ chains of Hb F lack specific binding sites for
2,3-BPG.
Beginnings: the molecular pathology of hemoglobin 3
1 Kb
30 31 104 105
ε Gγ Aγ ψβ δ β
11
α2γ2 α2β2 α2δ2
HbF HbA HbA2
Fetus Adult
In short, oxygen transport can be modified by a variety
of adaptive features in the red cell that include interactions
between the different heme molecules, the effects of CO2 , and
differential affinities for 2,3-BPG. These changes, together
with more general mechanisms involving the cardiorespira-
tory system, provide the main basis for physiological adapta-
tion to anemia.
Genetic control of hemoglobin
The α- and β-like globin chains are the products of two dif-
ferent gene families which are found on different chromo-
somes (Figure 1.1). The β-like globin genes form a linked
cluster on chromosome 11, spread over approximately 60 kb
(kilobase or 1000 nucleotide bases). The different genes that
form this cluster are arranged in the order 5′ –ε–G γ–A γ–
ψβ–δ–β–3′ . The α-like genes also form a linked cluster, in
this case on chromosome 16, in the order 5′ –ζ–ψζ–ψα1–
α2–α1–3′ . The ψβ, ψζ, and ψα genes are pseudogenes; that
is, they have strong sequence homology with the β, ζ, and
α genes, but contain a number of differences that prevent
them from directing the synthesis of any products. They may
reflect remnants of genes that were functional at an earlier
stage of human evolution.
The structure of the human globin genes is, in essence,
similar to that of all mammalian genes. They consist of long
strings of nucleotides that are divided into coding regions,
or exons, and non-coding inserts called intervening sequences
(IVSs) or introns. The β-like globin genes contain two
introns, one of 122–130 base pairs between codons 30 and 31
and one of 850–900 base pairs between codons 104 and 105
(the exon codons are numbered sequentially from the 5′ to
4 Molecular Hematology

the 3′ end of the gene, i.e. from left to right). Similar, though
smaller, introns are found in the α and ζ globin genes. These
introns and exons, together with short non-coding sequences
at the 5′ and 3′ ends of the genes, represent the major func-
tional regions of the particular genes. However, there are also
extremely important regulatory sequences which subserve
these functions that lie outside the genes themselves.
At the 5′ non-coding (flanking) regions of the globin
genes, as in all mammalian genes, there are blocks of
nucleotide homology. The first, the ATA box, is about 30
bases upstream (to the left) of the initiation codon; that is, the
start word for the beginning of protein synthesis (see later).
The second, the CCAAT box, is about 70 base pairs upstream
from the 5′ end of the genes. About 80–100 bases further
upstream there is the sequence GGGGTG, or CACCC, which
may be inverted or duplicated. These three highly conserved
DNA sequences, called promoter elements, are involved in
the initiation of transcription of the individual genes. Finally,
in the 3′ non-coding region of all the globin genes there
is the sequence AATAAA, which is the signal for cleavage
and polyA addition to RNA transcripts (see Gene Action and
Globin Synthesis).
The globin gene clusters also contain several sequences
that constitute regulatory elements, which interact to pro-
mote erythroid-specific gene expression and coordination
of the changes in globin gene activity during develop-
ment. These include the globin genes themselves and their
promoter elements: enhancers (regulatory sequences that
increase gene expression despite being located at a con-
siderable distance from the genes) and “master” regulatory
sequences called, in the case of the β globin gene cluster, the
locus control region (LCR) and, in the case of the α genes,
HS40 (a nuclease-hypersensitive site in DNA 40 kb from the
α globin genes). Each of these sequences has a modular struc-
ture made up of an array of short motifs that represent the
binding sites for transcriptional activators or repressors.
Gene action and globin synthesis
The flow of information between DNA and protein is sum-
marized in Figure 1.2. When a globin gene is transcribed,
messenger RNA (mRNA) is synthesized from one of its
strands, a process which begins with the formation of a tran-
scription complex consisting of a variety of regulatory pro-
teins together with an enzyme called RNA polymerase (see
later). The primary transcript is a large mRNA precursor
which contains both intron and exon sequences. While in
the nucleus, this molecule undergoes a variety of modifica-
tions. First, the introns are removed and the exons are spliced
together. The intron/exon junctions always have the same
sequence: GT at their 5′ end, and AG at their 3′ end. This
appears to be essential for accurate splicing; if there is a muta-
tion at these sites this process does not occur. Splicing reflects
a complex series of intermediary stages and the interaction

A
C C A
A C T T
C A A A T A
C A T T A A
C T A G A A
Flanking IVS 1 IVS 2 Flanking Gene
NC GT AG GT AG NC
5' 3' mRNA precursor

5' CAP Excision of introns

AAAA-A Splicing of exons
AAAA-A Processed mRNA
Nucleus
Cytoplasm

AAAA-A Translation
AUG UG
Ribosome ACC UUC UAA
U AC G AAG

Transfer RNA

Amino
acid Growing
chain
Finished
chain
Processed Fig. 1.2 The mechanisms of globin gene
chain transcription and translation.

of a number of different nuclear proteins. After the exons are
joined, the mRNAs are modified and stabilized; at their 5′ end
a complex CAP structure is formed, while at their 3′ end a
string of adenylic acid residues (polyA) is added. The mRNA
processed in this way moves into the cytoplasm, where it
acts as a template for globin chain production. Because of
the rules of base pairing – that is, cytosine always pairs with
thymine, and guanine with adenine – the structure of the
mRNA reflects a faithful copy of the DNA codons from which
it is synthesized; the only difference is that, in RNA, uracil (U)
replaces thymine (T).
Amino acids are transported to the mRNA template on
carriers called transfer RNAs (tRNAs); there are specific
tRNAs for each amino acid. Furthermore, because the genetic
code is redundant (i.e. more than one codon can encode a
particular amino acid), for some of the amino acids there are
several different individual tRNAs. Their order in the globin
chain is determined by the order of codons in the mRNA.
The tRNAs contain three bases, which together constitute an
anticodon; these anticodons are complementary to mRNA
codons for particular amino acids. They carry amino acids
to the template, where they find the appropriate positioning
by codon–anticodon base pairing. When the first tRNA is
in position, an initiation complex is formed between several
protein initiation factors together with the two subunits that
constitute the ribosomes. A second tRNA moves in along-
side and the two amino acids that they are carrying form
a peptide bond between them; the globin chain is now two
amino acid residues long. This process is continued along
–20 0 20
5' HS
4 3 2 1
ε Gγ
Chromosome 11
LCR
–40 –20
HS–40
Chromosome 16
Fig. 1.3 The positions of the major regulatory regions in the β and α globin gene clusters. The arrows indicate the position of the
erythroid lineage-specific nuclease-hypersensitive sites. HS, hypersensitive.
Beginnings: the molecular pathology of hemoglobin 5
the mRNA from left to right, and the growing peptide chain
is transferred from one incoming tRNA to the next; that is,
the mRNA is translated from 5′ to 3′ . During this time the
tRNAs are held in appropriate steric configuration with the
mRNA by the two ribosomal subunits. There are specific
initiation (AUG) and termination (UAA, UAG, and UGA)
codons. When the ribosomes reach the termination codon,
translation ceases, the completed globin chains are released,
and the ribosomal subunits are recycled. Individual globin
chains combine with heme, which has been synthesized
through a separate pathway, and then interact with one like
chain and two unlike chains to form a complete hemoglobin
tetramer.
Regulation of hemoglobin synthesis
The regulation of globin gene expression is mediated mainly
at the transcriptional level, with some fine tuning during
translation and post-translational modification of the gene
products. DNA that is not involved in transcription is held
tightly packaged in a compact, chemically modified form that
is inaccessible to transcription factors and polymerases and is
heavily methylated. Activation of a particular gene is reflected
by changes in the structure of the surrounding chromatin,
which can be identified by enhanced sensitivity to nucleases.
Erythroid lineage-specific nuclease-hypersensitive sites are
found at several locations in the β globin gene cluster. Four
are distributed over 20 kb upstream from the ε globin gene
in the region of the β globin LCR (Figure 1.3). This vital
40 60 kb
3' HS
1
Aγ δ β
0 20 40 kb
ζ α2 α1
6 Molecular Hematology
regulatory region is able to establish a transcriptionally active
domain spanning the entire β globin gene cluster. Several
enhancer sequences have been identified in this cluster. A
variety of regulatory proteins bind to the LCR, and to the
promoter regions of the globin genes and to the enhancer
sequences. It is thought that the LCR and other enhancer
regions become opposed to the promoters to increase
the rate of transcription of the genes to which they are
related.
These regulatory regions contain sequence motifs for
various ubiquitous and erythroid-restricted transcription
factors. Binding sites for these factors have been identified in
each of the globin gene promoters and at the hypersensitive
site regions of the various regulatory elements. A number
of the factors which bind to these areas are found in all cell
types. They include Sp1, Yy1, and Usf. In contrast, a number
of transcription factors have been identified, including
GATA-1, EKLF, and NF-E2, which are restricted in their
distribution to erythroid cells and, in some cases, megakary-
ocytes, and mast cells. The overlapping of erythroid-specific
and ubiquitous-factor binding sites in several cases suggests
that competitive binding may play an important part in
the regulation of erythroid-specific genes. Another binding
factor, SSP, the stage selector protein, appears to interact
specifically with ε and γ genes. Several elements involving
the chromatin and histone acetylation required for access of
these regulatory proteins have been identified.
The binding of hematopoietic-specific factors activates the
LCR, which renders the entire β globin gene cluster tran-
scriptionally active. These factors also bind to the enhancer
and promoter sequences, which work in tandem to regu-
late the expression of the individual genes in the clusters. It
is likely that some of the transcriptional factors are devel-
opmental stage specific, and hence may be responsible for
the differential expression of the embryonic, fetal, and adult
globin genes. The α globin gene cluster also contains an ele-
ment, HS40, which has some structural features in common
with the β LCR, although it is different in aspects of its struc-
ture. A number of enhancer-like sequences have been identi-
fied, although it is becoming clear that there are fundamental
differences in the pattern of regulation of the two globin gene
clusters.
In addition to the different regulatory sequences outlined,
there are also sequences which may be involved specifically
with “silencing” of genes, notably those for the embryonic
hemoglobins, during development.
Some degree of regulation is mediated by differences in
the rates of initiation and translation of the different mRNAs,
and at the post-transcriptional level by differential affinity
for different protein subunits. However, this kind of post-
transcriptional fine tuning probably plays a relatively small
role in determining the overall output of the globin gene
products.
Regulation of developmental changes
in globin gene expression
During development, the site of red cell production moves
from the yolk sac to the fetal liver and spleen, and thence
to bone marrow in the adult. Embryonic, fetal, and adult
hemoglobin synthesis is approximately related in time to
these changes in the site of erythropoiesis, although it is quite
clear that the various switches, between embryonic and fetal
and between fetal and adult hemoglobin synthesis, are beau-
tifully synchronized throughout these different sites. Fetal
hemoglobin synthesis declines during the later months of
gestation, and Hb F is replaced by Hb A and Hb A2 by the
end of the first year of life.
Although the exact mechanism of the switch from fetal to
adult hemoglobin is still not understood, recent studies of
patients with unusually high levels of HbF and genome-wide
association studies (GWAS) have yielded extremely promis-
ing information about some of the regulatory genes involved.
They include BCL11A, MYB, and KLF1. It is clear from stud-
ies of these and related genes that they are involved directly in
the regulation of hemoglobin switching, work which is yield-
ing great promise for the development of future technology
for increasing HbF synthesis to modify the phenotype of tha-
lassemias and sickle cell anemia.
The molecular pathology
of hemoglobin
As is the case for many monogenic diseases, the inherited dis-
orders of hemoglobin fall into two major classes. First, there
are those that result from reduced output of one or other
globin genes, the thalassemias. Second, there is a wide range
of conditions that result from the production of structurally
abnormal globin chains; the type of disease depends on how
the particular alteration in protein structure interferes with
its stability or function. Of course, no biological classifica-
tion is entirely satisfactory and those which attempt to define
the hemoglobin disorders are no exception. There are some
structural hemoglobin variants which happen to be synthe-
sized at a reduced rate and hence are associated with a clini-
cal picture similar to thalassemia. And there are other classes
of mutations which simply interfere with the normal tran-
sition from fetal to adult hemoglobin synthesis, a family of
conditions given the general title hereditary persistence of fetal
hemoglobin (HPFH). Furthermore, because these diseases are
all so common and occur together in particular populations,
it is not uncommon for an individual to inherit a gene for
one or other form of thalassemia and a structural hemoglobin
variant. The heterogeneous group of conditions that results
from these different mutations and interactions is summa-
rized in Table 1.1.
 Beginnings: the molecular pathology of hemoglobin 7

Table 1.1 The thalassemias and related disorders

Defective β globin gene transcription
There are a variety of mechanisms that interfere with nor-
α Thalassemia γ Thalassemia mal transcription of the β globin genes. First, the genes may
α0 be either completely or partially deleted. Overall, deletions
α+ δ Thalassemia
of the β globin genes are not commonly found in patients
Deletion (−α)
with β thalassemia, with one exception: a 619-bp deletion
Non-deletion (αT ) εγδβ Thalassemia
involving the 3′ end of the gene is found frequently in the
β Thalassemia Hereditary persistence of Sind populations of India and Pakistan, where it constitutes
β0 fetal hemoglobin about 30% of the β thalassemia alleles. Other deletions are
β+ Deletion
extremely rare.
Normal Hb A2 (δβ)0
A much more common group of mutations, which results
“Silent” Non-deletion
Dominant Linked to β globin genes
in a moderate decrease in the rate of transcription of the
β globin genes, involves single-nucleotide substitutions in
δβ Thalassemia G γβ+
or near the TATA box at about −30 nucleotides (nt) from
(δβ)+ A γβ+
the transcription start site, or in the proximal or distal pro-
(δβ)0 Unlinked to β globin genes
moter elements at −90 and −105 nt. These mutations result
(A γδβ)0
in decreased β globin mRNA production, ranging from 10%
to 25% of the normal output. Thus, they are usually associ-
Over recent years, determination of the molecular pathol- ated with the mild forms of β+ thalassemia. They are particu-
ogy of the two common forms of thalassemia, α and β, has larly common in African populations, an observation which
provided a remarkable picture of the repertoire of mutations explains the unusual mildness of β thalassemia in this racial
that can underlie human monogenic disease. In the sections group. One particular mutation, C → T at position −101 nt
that follow I describe, in outline, the different forms of molec- to the β globin gene, causes an extremely mild deficit of β
ular pathology that underlie these conditions. globin mRNA. Indeed, this allele is so mild that it is com-
pletely silent in carriers and can only be identified by its inter-
The β thalassemias action with more severe β thalassemia alleles in compound
heterozygotes.
There are two main classes of β thalassemia, β0 thalassemia,
in which there is an absence of β globin chain production,
Mutations that cause abnormal processing
and β+ thalassemia, in which there is a variable reduction
of mRNA
in the output of β globin chains. As shown in Figure 1.4,
mutations of the β globin genes may cause a reduced output As mentioned earlier, the boundaries between exons and
of gene product at the level of transcription or mRNA introns are marked by the invariant dinucleotides GT at the
processing, or translation, or through the stability of the donor (5′ ) site and AG at the acceptor (3′ ) site. Mutations that
globin gene product. affect either of these sites completely abolish normal splicing

Deletions

I IVSI 2 IVS 2 3

PR C I FS SPL SPL FS SPL SPL FS Poly A

NS NS NS

100 bp
Point mutations
Fig. 1.4 The mutations of the β globin gene that underlie β thalassemia. The heavy black lines indicate the length of the deletions. The
point mutations are designated as follows: PR, promoter; C, CAP site; I, initiation codon; FS, frameshift and nonsense mutations; SPL, splice
mutations; Poly A, poly A addition site mutations.
8 Molecular Hematology
and produce the phenotype of β0 thalassemia. The transcrip-
tion of genes carrying these mutations appears to be normal,
but there is complete inactivation of splicing at the altered
junction.
Another family of mutations involves what are called splice
site consensus sequences. Although only the GT dinucleotide
is invariant at the donor splice site, there is conservation
of adjacent nucleotides and a common, or consensus,
sequence of these regions can be identified. Mutations
within this sequence can reduce the efficiency of splicing
to varying degrees, because they lead to alternate splicing
at the surrounding cryptic sites. For example, mutations of
the nucleotide at position 5 of IVS-1 (the first intervening
sequence), G → C or T, result in a marked reduction of
β chain production and in the phenotype of severe β+
thalassemia. On the other hand, the substitution of C for T
at position 6 in IVS-1 leads to only a mild reduction in the
output of β chains.
Another mechanism that leads to abnormal splicing
involves cryptic splice sites. These are regions of DNA which,
if mutated, assume the function of a splice site at an inappro-
priate region of the mRNA precursor. For example, a variety
of mutations activate a cryptic site which spans codons 24–
27 of exon 1 of the β globin gene. This site contains a GT
dinucleotide, and adjacent substitutions that alter it so that it
more closely resembles the consensus donor splice site result
in its activation, even though the normal splice site is intact. A
mutation at codon 24 GGT → GGA, though it does not alter
the amino acid which is normally found in this position in
the β globin chain (glycine), allows some splicing to occur
at this site instead of the exon–intron boundary. This results
in the production of both normal and abnormally spliced β
Normal splicing
GT IVS1 AG GT IVS2
β gene
TTGGTCT
β+ thalassemia
globin mRNA and hence in the clinical phenotype of severe
β thalassemia. Interestingly, mutations at codons 19, 26, and
27 result in both reduced production of normal mRNA (due
to abnormal splicing) and an amino acid substitution when
the mRNA which is spliced normally is translated into pro-
tein. The abnormal hemoglobins produced are Hb Malay, Hb
E, and Hb Knossos, respectively. All these variants are associ-
ated with a mild β+ thalassemia-like phenotype. These muta-
tions illustrate how sequence changes in coding rather than
IVS influence RNA processing, and underline the impor-
tance of competition between potential splice site sequences
in generating both normal and abnormal varieties of β globin
mRNA.
Cryptic splice sites in introns may also carry mutations
that activate them even though the normal splice sites remain
intact. A common mutation of this kind in Mediterranean
populations involves a base substitution at position 110 in
IVS-1. This region contains a sequence similar to a 3′ accep-
tor site, though it lacks the invariant AG dinucleotide. The
change of the G to A at position 110 creates this dinucleotide.
The result is that about 90% of the RNA transcript splices to
this particular site and only 10% to the normal site, again pro-
ducing the phenotype of severe β+ thalassemia (Figure 1.5).
Several other β thalassemia mutations have been described
which generate new donor sites within IVS-2 of the β globin
gene.
Another family of mutations that interferes with β globin
gene processing involves the sequence AAUAAA in the 3′
untranslated regions, which is the signal for cleavage and
polyadenylation of the β globin gene transcript. Somehow,
these mutations destabilize the transcript. For example, a
T → C substitution in this sequence leads to only one-tenth of
AG
10%
Fig. 1.5 The generation of a new splice site in
an intron as the mechanism for a form of β+
90% thalassemia. For details see text.

the normal amount of β globin mRNA transcript, and hence
to the phenotype of a moderately severe β+ thalassemia.
Another example of a mutation which probably leads to
defective processing of the function of β globin mRNA is the
single-base substitution A → C in the CAP site. It is not yet
understood how this mutation causes a reduced rate of tran-
scription of the β globin gene.
There is another small subset of rare mutations that involve
the 3′ untranslated region of the β globin gene, and these are
associated with relatively mild forms of β thalassemia. It is
thought that these interfere in some way with transcription,
but the mechanism is unknown.
Mutations that result in abnormal
translation of β globin mRNA
There are three main classes of mutations of this kind.
Base substitutions that change an amino acid codon to a
chain termination codon prevent the translation of β globin
mRNA and result in the phenotype of β0 thalassemia. Several
mutations of this kind have been described; the common-
est, involving codon 17, occurs widely throughout South-
east Asia. Similarly, a codon 39 mutation is encountered fre-
quently in the Mediterranean region.
The second class involves the insertion or deletion of
one, two, or four nucleotides in the coding region of the β
globin gene. These disrupt the normal reading frame, cause
a frameshift, and hence interfere with the translation of β
globin mRNA. The end result is the insertion of anomalous
amino acids after the frameshift until a termination codon
is reached in the new reading frame. This type of mutation
always leads to the phenotype of β0 thalassemia.
Finally, there are several mutations which involve the β
globin gene initiation codon and which, presumably, reduce
the efficiency of translation.
Unstable β globin chain variants
Some forms of β thalassemia result from the synthesis of
highly unstable β globin chains which are incapable of form-
ing hemoglobin tetramers and which are rapidly degraded,
leading to the phenotype of β0 thalassemia. Indeed, in many
of these conditions no abnormal globin chain product can be
demonstrated by protein analysis, and the molecular pathol-
ogy has to be interpreted simply on the basis of a derived
sequence of the variant β chain obtained by DNA analysis.
Recent studies have provided some interesting insights
into how complex clinical phenotypes may result from the
synthesis of unstable β globin products. For example, there
is a spectrum of disorders that result from mutations in exon
3 which give rise to a moderately severe form of β tha-
lassemia in heterozygotes. It has been found that nonsense or
frameshift mutations in exons I and II are associated with the
Beginnings: the molecular pathology of hemoglobin 9
absence of mRNA from the cytoplasm of red cell precursors.
This appears to be an adaptive mechanism, called nonsense-
mediated decay, whereby abnormal mRNA of this type is not
transported to the cytoplasm, where it would act as a tem-
plate for the production of truncated gene products. How-
ever, in the case of exon III mutations, apparently because
this process requires the presence of an intact upstream exon,
the abnormal mRNA is transported into the cytoplasm and
hence can act as a template for the production of unstable
β globin chains. The latter precipitate in the red cell pre-
cursors together with excess α chains to form large inclu-
sion bodies, and hence there is enough globin chain imbal-
ance in heterozygotes to produce a moderately severe degree
of anemia.
The α thalassemias
The molecular pathology of the α thalassemias is more com-
plicated than that of the β thalassemias, simply because there
are two α globin genes per haploid genome. Thus, the normal
α globin genotype can be written αα/αα. As in the case of β
thalassemia, there are two major varieties of α thalassemia,
α+ and α0 thalassemia. In α+ thalassemia one of the linked α
globin genes is lost, either by deletion (−) or mutation (T); the
heterozygous genotype can be written –α/αα or αT α/αα.
In α0 thalassemia the loss of both α globin genes nearly
always results from a deletion; the heterozygous genotype is
therefore written − −/αα. In populations where specific dele-
tions are particularly common, Southeast Asia (SEA) or the
Mediterranean region (MED), it is useful to add the appropri-
ate superscript as follows: – –SEA /αα or – –MED /αα. It follows
that when we speak of an “α thalassemia gene,” what we are
really referring to is a haplotype; that is, the state and function
of both of the linked α globin genes.
α0 thalassemia
Three main molecular pathologies, all involving deletions,
have been found to underlie the α0 thalassemia phenotype.
The majority of cases result from deletions that remove both
α globin genes and a varying length of the α globin gene
cluster (Figure 1.6). Occasionally, however, the α globin gene
cluster is intact, but is inactivated by a deletion which involves
the major regulatory region HS40, 40 kb upstream from the
α globin genes, or the α globin genes may be lost as part of a
truncation of the tip of the short arm of chromosome 16.
As well as providing us with an understanding of the
molecular basis for α0 thalassemia, detailed studies of these
deletions have yielded more general information about the
mechanisms that underlie this form of molecular pathology.
For example, it has been found that the 5′ breakpoints of a
number of deletions of the α globin gene cluster are located
approximately the same distance apart and in the same order
10 Molecular Hematology
–50 –10 0 10 20
ζ2 ψζ1 ψα2 ψα1 α2 α1
Inter-ζHVR
along the chromosome as their respective 3′ breakpoints;
similar findings have been observed in deletions of the β
globin gene cluster. These deletions seem to have resulted
from illegitimate recombination events, which have led to the
deletion of an integral number of chromatin loops as they
pass through their nuclear attachment points during chro-
mosomal replication. Another long deletion has been charac-
terized in which a new piece of DNA bridges the two break-
points in the α globin gene cluster. The inserted sequence
originates upstream from the α globin gene cluster, where
normally it is found in an inverted orientation with respect
to that found between the breakpoints of the deletion. Thus
it appears to have been incorporated into the junction in a
way that reflects its close proximity to the deletion breakpoint
region during replication. Other deletions seem to be related
to the family of Alu-repeats, simple repeat sequences that
are widely dispersed throughout the genome; one deletion
appears to have resulted from a simple homologous recombi-
nation between two repeats of this kind that are usually 62 kb
apart.
A number of forms of α0 thalassemia result from terminal
truncations of the short arm of chromosome 16 to a site about
50 kb distal to the α globin genes. The telomeric consensus
sequence TTAGGGn has been added directly to the site of the
break. Since these mutations are stably inherited, it appears
that telomeric DNA alone is sufficient to stabilize the ends of
broken chromosomes.
Quite recently, two other molecular mechanisms have
been identified as the cause of α0 thalassemia which, though
rare, may have important implications for an understand-
ing of the molecular pathology of other genetic diseases. In
one case, a deletion in the α globin gene cluster resulted
in a widely expressed gene (LUC7L) becoming juxtaposed
to a structurally normal α globin gene. Although the latter
retained all its important regulatory elements, its expression
was silenced. It was found in a transgenic mouse model that
transcription of antisense RNA mediated the silencing of the
30
θ1
3'HVR
Fig. 1.6 Some of the deletions that underlie α0
and α+ thalassemia. The colored rectangles
beneath the α globin gene cluster indicate the
lengths of the deletions. The unshaded regions
indicate uncertainty about the precise breakpoints.
The three small deletions at the bottom of the figure
represent the common α+ thalassemia deletions.
HVR, highly variable regions.
α globin gene region, a finding that provides a completely
new mechanism for genetic disease. In another case of α0
thalassemia, in which no molecular defects could be detected
in the α globin gene cluster, a gain-of-function regulatory
polymorphism was found in the region between the α globin
genes and their upstream regulatory elements. This alteration
creates a new promoter-like element that interferes with the
normal activation of all downstream α-like globin genes.
In short, detailed analysis of the molecular pathology of
the α0 thalassemias has provided valuable evidence not only
about how large deletions of gene clusters are caused, but also
about some of the complex mechanisms that may underlie
cases in which the α gene clusters remain intact, but in which
their function is completely suppressed.
α+ thalassemia
As mentioned earlier, the α+ thalassemias result from the
inactivation of one of the duplicated α globin genes, by either
deletion or point mutation.
𝛼 + Thalassemia due to gene deletions. There are two com-
mon forms of α+ thalassemia that are due to loss of one
or other of the duplicated α globin genes, −α3.7 and −α4.2 ,
where 3.7 and 4.2 indicate the size of the deletions. The way
in which these deletions have been generated, approximately
4 kb long, was probably generated by an ancient duplication
event. The homologous regions, which are divided by small
inserts, are designated X, Y, and Z. The duplicated Z boxes
are 3.7 kb apart and the X boxes are 4.2 kb apart. The result
of misalignment reflects the underlying structure of the α
globin gene complex (Figure 1.7). Each α gene lies within
a boundary of homology and reciprocal crossover between
these segments at meiosis, a chromosome is produced with
either a single (−α) or triplicated (ααα) α globin gene. As
shown in Figure 1.7, if a crossover occurs between homol-
ogous Z boxes 3.7 kb of DNA are lost, an event which is
described as a rightward deletion, −α3.7 . A similar crossover

ψα1
(a)
ψα1
(b) Rightward crossover
Fig. 1.7 Mechanisms of the generation of the
common deletion forms of α+ thalassemia. (a)
The normal arrangement of the α globin genes, ψα1
with the regions of homology X, Y, and Z. (b) The
crossover that generates the −α3.7 deletion. (c) The
crossover that generates the −α4.2 deletion. (c) Leftward crossover
between the two X boxes deletes 4.2 kb, the leftward deletion
−α4.2 . The corresponding triplicated α gene arrangements
are called αααanti 3.7 and αααanti 4.2 . A variety of different
points of crossing over within the Z boxes give rise to differ-
ent length deletions, still involving 3.7 kb.
Non-deletion types of 𝛼 + thalassemia. These disorders
result from single or oligonucleotide mutations of the par-
ticular α globin gene. Most of them involve the α2 gene but,
since the output from this locus is two to three times greater
than that from the α1 gene, this may simply reflect ascertain-
ment bias due to the greater phenotypic effect and, possibly,
a greater selective advantage.
Overall, these mutations interfere with α globin gene
function in a similar way to those that affect the β globin
genes. They affect the transcription, translation, or post-
translational stability of the gene product. Since the prin-
ciples are the same as for β thalassemia, we do not need
to describe them in detail, with one exception, a mutation
which has not been observed in the β globin gene cluster.
It turns out that there is a family of mutations that involves
the α2 globin gene termination codon, TAA. Each specifi-
cally changes this codon so that an amino acid is inserted
instead of the chain terminating. This is followed by “read-
through” of α globin mRNA, which is not normally trans-
lated until another in-phase termination codon is reached.
The result is an elongated α chain with 31 additional residues
at the C-terminal end. Five hemoglobin variants of this type
have been identified. The commonest, Hb Constant Spring,
occurs at a high frequency in many parts of Southeast Asia.
It is not absolutely clear why the read-through of normally
untranslated mRNAs leads to a reduced output from the α2
Beginnings: the molecular pathology of hemoglobin 11
α2 α1
X Y Z X Y Z
ψα1 α2 α1
3.7
αααanti
α2 α1
3.7
–α
ψα1 α2 α1
4.2
αααanti
α2 α1
4.2
–α
gene, although there is considerable evidence that it in some
way destabilizes the mRNA.
α thalassemia/mental retardation syndromes
There is a family of mild forms of α thalassemia which is
quite different to that described in the previous section and
which is associated with varying degrees of mental retarda-
tion. Recent studies indicate that there are two quite different
varieties of this condition, one encoded on chromosome 16
(ATR-16) and the other on the X chromosome (ATR-X).
The ATR-16 syndrome is characterized by relatively mild
mental disability with a variable constellation of facial and
skeletal dysmorphisms. These individuals have long dele-
tions involving the α globin gene cluster, but removing
at least 1–2 Mb. This condition can arise in several ways,
including unbalanced translocation involving chromosome
16, truncation of the tip of chromosome 16, and the loss of
the α globin gene cluster and parts of its flanking regions by
other mechanisms.
The ATR-X syndrome results from mutations in a gene on
the X chromosome, Xq13.1–q21.1. The product of this gene
is one of a family of proteins involved in chromatin-mediated
transcriptional regulation. It is expressed ubiquitously dur-
ing development and at interphase it is found entirely within
the nucleus in association with pericentromeric heterochro-
matin. In metaphase, it is similarly found close to the cen-
tromeres of many chromosomes but, in addition, occurs at
the stalks of acrocentric chromosomes, where the sequences
for ribosomal RNA are located. These locations provide
important clues to the potential role of this protein in the
12 Molecular Hematology
establishment and/or maintenance of methylation of the
genome. Although it is clear that ATR-X is involved in α
globin transcription, it also must be an important player in
early fetal development, particularly of the urogenital system
and brain. Many different mutations of this gene have been
discovered in association with the widespread morphologi-
cal and developmental abnormalities which characterize the
ATR-X syndrome.
α thalassemia and the myelodysplastic syndrome
Since the first description of Hb H (see later section) in the red
cells of a patient with leukemia, many examples of this asso-
ciation have been reported. The condition usually is reflected
in a mild form of Hb H disease, with typical Hb H inclusions
in a proportion of the red cells and varying amounts of Hb H
demonstrable by hemoglobin electrophoresis. The hemato-
logical findings are usually those of one or other form of the
myelodysplastic syndrome. The condition occurs predomi-
nantly in males in older age groups. Very recently it has been
found that some patients with this condition have mutations
involving ATR-X. The relationship of these mutations to the
associated myelodysplasia remains to be determined.
Rarer forms of thalassemia and related
disorders
There are a variety of other conditions that involve the β
globin gene cluster which, although less common than the
β thalassemias, provide some important information about
mechanisms of molecular pathology and therefore should be
mentioned briefly.
The δβ thalassemias
Like the β thalassemias, the δβ thalassemias, which result
from defective δ and β chain synthesis, are subdivided into
the (δβ)+ and (δβ)0 forms.
The (δβ)+ thalassemias result from unequal crossover
between the δ and β globin gene loci at meiosis with the
production of δβ fusion genes. The resulting δβ fusion
chain products combine with α chains to form a family of
hemoglobin variants called the hemoglobin Lepores, after the
family name of the first patient of this kind to be discovered.
Because the synthesis of these variants is directed by genes
with the 5′ sequences of the δ globin genes, which have defec-
tive promoters, they are synthesized at a reduced rate and
result in the phenotype of a moderately severe form of δβ
thalassemia.
The (δβ)0 thalassemias nearly all result from long dele-
tions involving the β globin gene complex. Sometimes they
involve the A γ globin chains and hence the only active locus
remaining is the G γ locus. In other cases the G γ and A γ loci
are left intact and the deletion simply removes the δ and β
globin genes; in these cases both the G γ and the A γ globin
gene remain functional. For some reason, these long dele-
tions allow persistent synthesis of the γ globin genes at a
relatively high level during adult life, which helps to com-
pensate for the absence of β and δ globin chain production.
They are classified according to the kind of fetal hemoglobin
that is produced, and hence into two varieties, G γ(A γδβ)0
and G γA γ(δβ)0 thalassemia; in line with other forms of tha-
lassemia, they are best described by what is not produced:
(A γδβ)0 and (δβ)0 thalassemia, respectively. Homozygotes
produce only fetal hemoglobin, while heterozygotes have a
thalassemic blood picture together with about 5–15% Hb F.
Hereditary persistence of fetal hemoglobin
Genetically determined persistent fetal hemoglobin synthe-
sis in adult life is of no clinical importance, except that its
genetic determinants can interact with the β thalassemias or
structural hemoglobin variants; the resulting high level of Hb
F production often ameliorates these conditions. The differ-
ent forms of HPFH result from either long deletions involv-
ing the δβ globin gene cluster, similar to those that cause
(δβ)0 thalassemia, or point mutations that involve the pro-
moters of the G γ or A γ globin gene. In the former case there
is no β globin chain synthesis and therefore these condi-
tions are classified as (δβ)0 HPFH. In cases in which there
are promoter mutations involving the γ globin genes, there is
increased γ globin chain production in adult life associated
with some β and δ chain synthesis in cis (i.e. directed by the
same chromosome) to the HPFH mutations. Thus, depend-
ing on whether the point mutations involve the promoter of
the G γ or A γ globin gene, these conditions are called G γ β+
HPFH and A γ β+ HPFH, respectively.
There is another family of HPFH-like disorders in which
the genetic determinant is not encoded in the β chain clus-
ter. In one case the determinant encodes on chromosome 6,
although its nature has not yet been determined.
It should be pointed out that all these conditions are very
heterogeneous and that many different deletions or point
mutations have been discovered that produce the rather sim-
ilar phenotypes of (δβ)0 or G γ or A γ β+ HPFH.
Genotype–phenotype relationships in
the thalassemias
It is now necessary briefly to relate the remarkably diverse
molecular pathology described in the previous sections to
the phenotypes observed in patients with these diseases.
It is not possible to describe all these complex issues here.
Rather, I shall focus on those aspects that illustrate the more
general principles of how abnormal gene action is reflected
 Beginnings: the molecular pathology of hemoglobin 13

in a particular clinical picture. Perhaps the most important
question that I will address is why patients with apparently
identical genetic lesions have widely differing disorders,
a problem that still bedevils the whole field of medical
genetics, even in the molecular era.
The β thalassemias
As we have seen, the basic defect that results from the 200
or more different mutations that underlie these conditions is
reduced β globin chain production. Synthesis of the α globin
chain proceeds normally and hence there is imbalanced
globin chain output with an excess of α chains (Figure 1.8).
Unpaired α chains precipitate in both red cell precursors and
their progeny with the production of inclusion bodies. These
interfere with normal red cell maturation and survival in a
variety of complex ways. Their attachment to the red cell
membrane causes alterations in its structure, and their degra-
dation products, notably heme, hemin (oxidized heme), and
iron, result in oxidative damage to the red cell contents
and membrane. These interactions result in intramedullary
destruction of red cell precursors and in shortened survival
of such cells as they reach the peripheral blood. The end
result is anemia of varying severity. This, in turn, causes tis-
sue hypoxia and the production of relatively large amounts of
erythropoietin; this leads to a massive expansion of the inef-
fective bone marrow, resulting in bone deformity, a hyperme-
tabolic state with wasting and malaise, and bone fragility.
A large proportion of hemoglobin in the blood of β tha-
lassemics is of the fetal variety. Normal individuals produce
about 1% of Hb F, unevenly distributed among their red cells.
In the bone marrow of β thalassemics, any red cell precursors
that synthesize γ chains come under strong selection because
they combine with α chains to produce fetal hemoglobin, and

α
Ex
ce
γ ss
β

α2γ2 Denaturation
HbF Degradation

Selective survival of
HbF-containing Hemolysis Destruction of RBC
precursors precursors

Increased levels of Splenomegaly Ineffective

HbF in red cells (pooling, plasma erythropoiesis
volume expansion)

High
oxygen affinity of Anemia
red cells
R
O ed u e
c su a
2 d
eli ed Tis poxi
ve h y
ry
Erythropoietin

Transfusion
Marrow
expansion
Inc
r
ab ease
so d
rp iro
tio n
Skeletal deformity n Iron loading
Increased metabolic rate
Wasting
Gout
Folate deficiency
Endocrine deficiencies
Cirrhosis
Fig. 1.8 The pathophysiology of β thalassemia. Cardiac failure
14 Molecular Hematology
therefore the degree of globin chain imbalance is reduced.
Furthermore, the likelihood of γ chain production seems to
be increased in a highly stimulated erythroid bone marrow.
It seems likely that these two factors combine to increase the
relative output of Hb F in this disorder. However, it has a
higher oxygen affinity than Hb A, and hence patients with
β thalassemia are not able to adapt to low hemoglobin levels
as well as those who have adult hemoglobin.
The greatly expanded, ineffective erythron leads to an
increased rate of iron absorption; this, combined with iron
received by blood transfusion, leads to progressive iron
loading of the tissues, with subsequent liver, cardiac, and
endocrine damage.
The constant bombardment of the spleen with abnormal
red cells leads to its hypertrophy. Hence there is progressive
splenomegaly with an increased plasma volume and trapping
of part of the circulating red cell mass in the spleen. This
leads to worsening of the anemia. All these pathophysiolog-
ical mechanisms, except for iron loading, can be reversed by
regular blood transfusion which, in effect, shuts off the inef-
fective bone marrow and its consequences.
Thus it is possible to relate nearly all the important fea-
tures of the severe forms of β thalassemia to the primary
defect in globin gene action. However, can we also explain
their remarkable clinical diversity?
Phenotypic diversity
Although the bulk of patients who are homozygous for β tha-
lassemia mutations or compound heterozygotes for two dif-
ferent mutations have a severe transfusion-dependent phe-
notype, there are many exceptions. Some patients of this type
have a milder course, requiring few or even no transfusions,
a condition called β thalassemia intermedia. A particularly
important example of this condition is illustrated by the clin-
ical findings in those who inherit β thalassemia from one
parent and Hb E from the other, a disorder called Hb E/β
thalassemia. Because the mutation that produces Hb E also
opens up an alternative splice site in the first exon of the β
globin gene, it is synthesized at a reduced rate and therefore
behaves like a mild form of β thalassemia. It is the common-
est hemoglobin variant globally and Hb E/β thalassemia is
the commonest form of severe thalassemia in many Asian
countries. It has an extraordinarily variable phenotype, rang-
ing from a condition indistinguishable from β thalassemia
major to one of such mildness that patients grow and develop
quite normally and never require transfusion.
Over recent years a great deal has been learned about some
of the mechanisms involved in this remarkable phenotypic
variability. In short, it reflects both the action of modifying
genes and variability in adaptation to anemia and, almost cer-
tainly, the effects of the environment. Given the complexity of
these interactions, it is helpful to divide the genetic modifiers
Table 1.2 Mechanisms for the phenotypic diversity of the β
thalassemias
Genetic modifiers
Primary: alleles of varying severity
Secondary: modifiers of globin chain imbalance
α Thalassemia
Increased α globin genes: ααα or αααα
Genes involved in unusually high Hb F response
Tertiary: modifiers of complications
Iron absorption, bone disease, jaundice, infection
Adaptation to anemiaa
Variation in oxygen affinity (P50 ) of hemoglobin
Variation in erythropoietin response to anemia
Environmental
Nutrition
Infection
Others
a There may be genetic variation in the adaptive mechanisms.
of the β thalassemia phenotype into primary, secondary, and
tertiary classes (Table 1.2).
The primary modifiers are the different β thalassemia alle-
les that can interact together. For example, compound het-
erozygotes for a severe β0 thalassemia mutation and a milder
one may have an intermediate form of β thalassemia of
varying severity, depending on the degree of reduction in β
globin synthesis under the action of the milder allele. This is
undoubtedly one mechanism for the varying severity of Hb
E/β thalassemia; it simply reflects the variable action of the
β thalassemia mutation that is inherited together with Hb E.
However, this explanation is not relevant in cases in which
patients with identical β thalassemia mutations have widely
disparate phenotypes.
The secondary modifiers are those which directly affect
the degree of globin chain imbalance. Patients with β tha-
lassemia who also inherit one or other form of α thalassemia
tend to have a milder phenotype because of the reduction
in the excess of α globin genes caused by the coexistent α
thalassemia allele. Similarly, patients with severe forms of
thalassemia who inherit more α genes than normal because
their parents have triplicated or quadruplicated α gene
arrangements tend to have more severe phenotypes. Other
patients with severe thalassemia alleles appear to run a
milder course because of a genetically determined ability to
produce more γ globin chains and hence fetal hemoglobin,
a mechanism that also results in a reduced degree of globin
chain imbalance. It is now clear that several gene loci are
involved in this mechanism; the best characterized is a
polymorphism in the promoter region of the G γ globin gene
that appears to increase the output from this locus under

conditions of hemopoietic stress. However, there are clearly
other genes involved in increasing the output of Hb F. Recent
genome-wide linkage studies have shown clear evidence that
there are determinants on chromosomes 6 and 8 and a par-
ticularly strong association has been found with BCL11A, a
transcription factor known to be involved in hematopoiesis.
The exact mechanism for the associated increase in Hb F
in β thalassemia and in sickle cell anemia remains to be
determined.
The tertiary modifiers are those that have no effect on
hemoglobin synthesis, but modify the many different com-
plications of the β thalassemias, including osteoporosis, iron
absorption, jaundice, and susceptibility to infection.
Although neglected until recently, it is also becoming
apparent that variation in adaptation to anemia and the envi-
ronment may also play a role in phenotypic modification
of the β thalassemias. For example, patients with Hb E/β
thalassemia have relatively low levels of Hb F and hence
their oxygen dissociation curves are more right-shifted than
patients with other forms of β thalassemia intermedia with
significantly higher levels of Hb F. Very recent studies also
suggest that the erythropoietin response to severe anemia
for a given hemoglobin level varies considerably with age;
patients during the first years of life have significantly higher
responses to the same hemoglobin level than those who
are older. This observation may go some way to explaining
the variation in phenotype at different ages that has been
observed in children with Hb E/β thalassemia. Finally, it is
clear that further studies are required to dissociate the effects
on the phenotype of genetic modifiers and environmental
factors.
Thus, the phenotypic variability of the β thalassemias
reflects several layers of complex interactions involving
genetic modifiers together with variation in adaptation and,
almost certainly, the environment. These complex interac-
tions are summarized in Table 1.2.
The α thalassemias
The pathophysiology of the α thalassemias differs from that
of the β thalassemias mainly because of the properties of the
excess globin chains that are produced as a result of defective
α chain synthesis. While the excess α chains produced in β
thalassemia are unstable and precipitate, this is not the case
in the α thalassemias, in which excess γ chains or β chains
are able to form the soluble homotetramers γ4 (Hb Bart’s)
and β4 (Hb H) (Figure 1.9). Although these variants, par-
ticularly Hb H, are unstable and precipitate in older red cell
populations, they remain soluble sufficiently long for the red
cells to mature and develop relatively normally. Hence there
is far less ineffective erythropoiesis in the α thalassemias and
the main cause of the anemia is hemolysis associated with
the precipitation of Hb H in older red cells. In addition, of
Beginnings: the molecular pathology of hemoglobin 15
Fetus Adult
γ β
α
α2γ2 α2β2
Excess Excess
γ4 β4
Hb Bart's Hb H
High oxygen affinity—hypoxia
Instability of homotetramers
Inclusion bodies. Membrane damage
Shortened red cell survival—hemolysis
Splenomegaly—hypersplenism
Fig. 1.9 The pathophysiology of α thalassemia.
course, there is a reduction in normal hemoglobin synthe-
sis, which results in hypochromic, microcytic erythrocytes.
Another important factor in the pathophysiology of the α
thalassemias is the fact that Hb Bart’s and Hb H are useless
oxygen carriers, having an oxygen dissociation curve similar
to that of myoglobin. Thus the circulating hemoglobin level
may give a false impression of the oxygen-delivering capacity
of the blood, and patients may be symptomatic at relatively
high hemoglobin levels.
The different clinical phenotypes of the α thalassemias are
an elegant example of the effects of gene dosage (Figure 1.10).
The heterozygous state for α+ thalassemia is associated with
minimal hematological changes. That for α0 thalassemia (the
loss of two α globin genes) is characterized by moderate
hypochromia and microcytosis, similar to that of the β tha-
lassemia trait. It does not matter whether the α genes are lost
on the same chromosome or on opposite pairs of homolo-
gous chromosomes. Hence the homozygous state for α+ tha-
lassemia, −α/−α, has a similar phenotype to the heterozy-
gous state for α0 thalassemia (− −/αα).
The loss of three α globin genes, which usually results
from the compound heterozygous states for α0 and α+ tha-
lassemia, is associated with a moderately severe anemia with
the production of varying levels of Hb H. This condition,
hemoglobin H disease, is characterized by varying anemia
and splenomegaly with a marked shortening of red cell
survival.
Finally, the homozygous state for α0 thalassemia (− −/
− −) is characterized by death in utero or just after birth, with
the clinical picture of hydrops fetalis. These babies produce
no α chains and their hemoglobin consists mainly of Hb
Bart’s with variable persistence of embryonic hemoglobin.
This is reflected in gross intrauterine hypoxia; although these
babies may have hemoglobin values as high as 8–9 g/dL,
16 Molecular Hematology

α0 Thal. trait α0 Thal. trait different amino acid in the affected globin chain. Rarely,
these variants result from more subtle alterations in the
structure of the α/β globin chains. For example, shortened
X chains may result from internal deletions of their particular
genes, while elongated chains result from either duplications
within genes or frameshift mutations, which allow the
chain termination codon to be read through and in which
additional amino acids are added to the C-terminal end. The
majority of the 700 or more structural hemoglobin variants
are of no clinical significance but a few, because they interfere
with the stability or functions of the hemoglobin molecule,
are associated with a clinical phenotype of varying severity.
Normal α0 Thal. α0 Thal. Hb Bart's
trait trait hydrops

Genotype–phenotype relationships
α0 Thal. trait α+ Thal. trait The sickling disorders

The sickling disorders represent the homozygous state for the

X sickle cell gene, sickle cell anemia, and the compound het-
Normal α0 Thal. α+ Thal. Hb H
trait trait disease
Fig. 1.10 The genetics of the common forms of α thalassemia.
The open boxes represent normal α genes and the green boxes
deleted α genes. The mating shown at the top shows how two α0
thalassemia heterozygotes can produce a baby with the Hb Bart’s
hydrops syndrome. In the mating at the bottom, between individuals
with α0 and α+ thalassemia, one in four of the offspring will have Hb
H disease.
most of it is unable to release its oxygen. This is reflected in
the hydropic changes, a massive outpouring of nucleated red
cells, and hepatosplenomegaly with persistent hematopoiesis
in the liver and spleen.
Structural hemoglobin variants
The structural hemoglobin variants are described in detail in
Chapter 14. Here, their molecular pathology and genotype–
phenotype relationships are briefly outlined.
Molecular pathology
The molecular pathology of the structural hemoglobin
variants is much less complex than that of the thalassemias.
The majority result from missense mutations – base sub-
stitutions that produce a codon change which encodes a
erozygous state for the sickle cell gene and various structural
hemoglobin variants, or β thalassemia. The chronic hemol-
ysis and episodes of vascular occlusion and red cell seques-
tration that characterize sickle cell anemia can all be related
to the replacement of the normal β6 glutamic acid by valine
in Hb S. This causes a hydrophobic interaction with another
hemoglobin molecule, triggering aggregation into large poly-
mers. It is this change that causes the sickling distortion of the
red blood cell and hence a marked decrease in its deformabil-
ity. The resulting rigidity of the red cells is responsible for the
vaso-occlusive changes that lead to many of the most serious
aspects of all the sickling disorders.
The different conformations of sickle cells (banana shaped
or resembling a holly leaf) reflect different orientations of
bundles of fibers along the long axis of the cell, the three-
dimensional structure of which is constituted by a rope-like
polymer composed of 14 strands. The rate and extent of poly-
mer formation depend on the degree of oxygenation, the cel-
lular hemoglobin concentration, and the presence or absence
of Hb F. The latter inhibits polymerization and hence tends
to ameliorate sickling. Polymerization of Hb S causes dam-
age to the red cell membrane, the result of which is an irre-
versibly sickled cell. Probably the most important mechanism
is cellular dehydration resulting from abnormalities of potas-
sium/chloride cotransport and Ca2+ -activated potassium
efflux. This is sufficient to trigger the Ca2+ -dependent (Gar-
dos) potassium channel, providing a mechanism for the loss
of potassium and water and leading to cellular dehydration.
However, the vascular pathology of the sickling disorders
is not entirely related to the rigidity of sickled red cells.
There is now a wealth of evidence that abnormal interactions
between sickled cells and the vascular endothelium play a
major role in the pathophysiology of the sickling disorders.
Recently it has been demonstrated that nitric oxide may
also play a role in some of the vascular complications of

this disease. It has been found that nitric oxide reacts much
more rapidly with free hemoglobin than with hemoglobin
in erythrocytes, and therefore it is possible that such decom-
partmentalization of hemoglobin into plasma, as occurs in
sickle cell disease and other hemolytic anemias, diverts nitric
oxide from its homeostatic vascular function.
The sickling disorders are discussed in detail in Chap-
ter 14, Hemoglobinopathies due to Structural Mutations.
Unstable hemoglobin variants
There is a variety of different mechanisms underlying
hemoglobin stability resulting from amino acid substitu-
tions in different parts of the molecule. The first is typi-
fied by amino acid substitutions in the vicinity of the heme
pocket, all of which lead to a decrease in stability of the
binding of heme to globin. A second group of unstable vari-
ants results from amino acids that simply disrupt the sec-
ondary structure of the globin chains. About 75% of globin
is in the form of α helix, in which proline cannot partici-
pate except as part of one of the initial three residues. At least
11 unstable hemoglobin variants have been described that
result from the substitution of proline for leucine, five that are
caused by the substitution of alanine by proline, and three in
which proline is substituted for histidine. Another group of
variants that causes disruption of the normal configuration
of the hemoglobin molecule involves internal substitutions
that somehow interfere with its stabilization by hydropho-
bic interactions. Finally, there are two groups of unstable
hemoglobins that result from gross structural abnormalities
of the globin subunits; many are due to deletions involving
regions at or near interhelical corners. A few of the elongated
globin chain variants are also unstable.
Abnormal oxygen transport
There is a family of hemoglobin variants associated with high
oxygen affinity and hereditary polycythemia. Most result
from amino acid substitutions that affect the equilibrium
between the R and T states (see Structure and Function). Thus,
many of them result from amino acid substitutions at the α1 –
β2 interface, the C-terminal end of the β chain, and at the
2,3-BPG binding sites.
Congenital cyanosis due to hemoglobin variants
There is a family of structural hemoglobin variants that is
designated Hb M, to indicate congenital methemoglobine-
mia, and is further defined by their place of discovery. The
iron atom of heme is normally linked to the imidazole group
of the proximal histidine residue of the α and β chains.
There is another histidine residue on the opposite side,
near the sixth coordination position of the heme iron; this,
the so-called distal histidine residue, is the normal site of
binding of oxygen. Several M hemoglobins result from the
Beginnings: the molecular pathology of hemoglobin 17
substitution of a tyrosine for either the proximal or distal
histidine residue in the α or β chain.
Molecular aspects of the high
frequency of the hemoglobin variants
The inherited disorders of hemoglobin are by far the com-
monest monogenic diseases. It is estimated that between
300 000 and 400 000 babies are born with these conditions,
mainly across the tropical belt or in countries with large num-
bers of immigrants from this region. Work over many years
has shown that the major reason for this high frequency is
heterozygote protection against malaria infection, particu-
larly that due to Plasmodium falciparum. Heterozygote pro-
tection raises the frequency of births of heterozygotes until an
equilibrium is achieved by the early loss of the homozygotes
for many of these diseases. In the case of α+ thalassemia, pre-
sumably because homozygotes have an extremely mild phe-
notype, the gene frequency is by far the commonest for any
monogenic disorder, in some parts of the world almost reach-
ing fixation.
The mechanism for the protective effect of heterozygotes
against malaria is still not completely understood, although
there is extensive evidence that in most cases it may be
multifactorial. In the case of the sickle cell trait, there is
evidence for impairment of invasion and growth of the
malarial parasites under conditions of low oxygen tension
and enhanced removal of the parasite from infected red
cells. There is also evidence for a reduced display of the
parasite-encoded protein P. falciparum Erythrocyte Protein-
1 (PfEMP1), which is involved in the sequestration of mature
parasites in small blood vessels and hence damage to tissues,
particularly the brain. There is also evidence for improved
acquisition of malaria-specific immunity in heterozygotes.
Rather similar findings have been observed in the case of
carriers for HbC and for impairment of P. falciparum red
cell invasion and growth in heterozygotes for HbE. In the
case of the α thalassemias, there appears to be reduced
pathogenicity due to reduced cytoadherence or rosetting, a
process whereby uninfected red cells aggregate round cells
that are infected with P. falciparum. There is also evidence
of immunological priming, whereby babies are more prone
to infection by Plasmodium vivax early in life, which may
result in later protection because of cross immunity between
P. falciparum and P. vivax.
There are some remarkable epistatic interactions between
the hemoglobin disorders in response to malaria. For
example, while heterozygotes for HbS or α+ thalassemia are
protected in those who inherit the genes for both conditions,
this protection is completely canceled out. Although the
exact mechanism is not understood, this interaction reduces
the amount of HbS in heterozygotes, suggesting that a
critical mass of this variant is required for protection against
18 Molecular Hematology
malaria. Several other epistatic interactions of this type have
been described recently and they are providing invaluable
information about the variable distribution of these variants
in different populations.
Molecular aspects of the prevention
and management of the hemoglobin
disorders
Over the years there have been increasingly sophisticated
approaches applied for screening and prenatal diagnosis of
the hemoglobin variants, particularly the thalassemias, in an
attempt to reduce the numbers of births of homozygotes or
compound heterozygotes. At first this was carried out by fetal
blood sampling in the mid-trimester of pregnancy, followed
by hemoglobin analysis by globin chain synthesis. When
DNA analysis became possible, this approach was replaced
by DNA studies of material obtained by chorion villous sam-
pling. Both these approaches were invasive, of course, but
they have led to a dramatic reduction in the births of affected
babies in many high-frequency populations. Very recently
a non-invasive approach has been established by attempt-
ing to detect homozygotes for monogenic diseases by next-
generation sequencing of fetal blood in maternal plasma.
This approach is still in its early days, but some promising
results have already been obtained.
Although for many years there has been a gradual
improvement in the symptomatic management of the
hemoglobin disorders, the only approach for a cure has been
the application of bone marrow transplantation, which of
course has the disadvantage of requiring matched donors.
More recently there has been a major effort directed at
the development of somatic-cell gene therapy, particularly
for the thalassemias. There has been considerable progress
toward this process using retroviral vectors and related
approaches. Although progress has been slow, successes have
been reported and by the use of new gene-editing techniques
further progress in the near future seems very likely. The
other area of molecular therapy – that is, attempts to increase
the level of fetal hemoglobin in patients with different forms
of β thalassemia or sickle cell anemia – was discussed earlier
in this chapter.
Postscript
In this short account of the molecular pathology of
hemoglobin, we have considered how mutations at or close
to the α or β globin genes result in a diverse family of clinical
disorders due to the defective synthesis of hemoglobin or its
abnormal structure. Work in this field over the last 30 years
has given us a fairly good idea of the repertoire of differ-
ent mutations that underlie single-gene disorders and how
these are expressed as discrete clinical phenotypes. Perhaps
more importantly, however, the globin field has taught us how
the interaction of a limited number of genes can produce a
remarkably diverse series of clinical pictures, and something
of the basis for how monogenic diseases due to the same
mutation may vary widely in their clinical expression.
Further reading
General background
Peltonen, L. and McKusick, V.A. (2001). Genomics and medicine. Dis-
secting human disease in the postgenomic era. Science 291: 1224–
1229.
Weatherall, D.J. (2013). The role of the inherited disorders of
hemoglobin, the first “molecular diseases,” in the future of human
genetics. Annu. Rev. Genomics Hum. Genet. 14: 1–24.
Weatherall, D.J. and Clegg, J.B. (2001). The Thalassaemia Syndromes, 4e.
Oxford: Blackwell Science.
Weatherall, D.J., Schechter, A.N., and Nathan, D.G. (eds.) (2013).
Hemoglobin and Its Diseases. Cold Spring Harbor, NY: Cold Spring
Harbor Laboratory Press.
Hemoglobin genetics and structural
variants
Steinberg, M.H., Forget, B.G., Higgs, D.R., and Weatherall, D.J. (eds.)
(2009). Disorders of Hemoglobin, 2e. New York: Cambridge University
Press.
Hemoglobin switching
Forget, B.G. (2011). Progress in understanding the hemoglobin switch.
N. Engl. J. Med. 365: 852–854.
Sankaran, V.G. and Nathan, D.G. (2010). Reversing the hemoglobin
switch. N. Engl. J. Med. 363: 2258–2260.
Sankaran, V.G. and Orkin, S.H. (2013). The switch from fetal to
adult hemoglobin. Cold Spring Harb. Perspect. Med. https://doi.org/
10.1101/cshperspect.a011643.
Thein, S.L. and Menzel, S. (2009). Discovering the genetics underlying
foetal haemoglobin production in adults. Br. J. Haematol. 145: 455–
467.
The β thalassemias
Fucharoen, S. and Weatherall, D.J. (2013). The hemoglobin E tha-
lassemias. Cold Spring Harb. Perspect. Med. https://doi.org/10.1101/
cshperspect.a011734.
Lettre, G. (2013). The Search for Genetic Modifiers of Disease Sever-
ity in the beta-Hemoglobinopathies. Cold Spring Harb. Perspect. Med.
https://doi.org/10.1101/cshperspect.a015032.
Nienhuis, A.W. and Nathan, D.G. (2013). Pathophysiology and clinical
manifestations of the beta thalassemias. Cold Spring Harb. Perspect.
Med. https://doi.org/10.1101/cshperspect.a011726.
Olivieri, N.F., Muraca, G.M., O’Donnell, A. et al. (2008). Studies in
haemoglobin E beta-thalassaemia. Br. J. Haematol. 141: 388–397.

Quek, L. and Thein, S.L. (2007). Molecular therapies in beta-
thalassaemia. Br. J. Haematol. 136: 353–365.
Thein, S.L. (2008). Genetic modifiers of the beta-haemoglobinopathies.
Br. J. Haematol. 141: 357–366.
The α thalassemias
Higgs, D.R. (2013). The molecular basis of alpha thalassemia. Cold
Spring Harb. Perspect. Med. https://doi.org/10.1101/cshperspect.
a011718.
Higgs, D.R. and Gibbons, R.J. (2010). The molecular basis of alpha-
thalassemia: a model for understanding human molecular genetics.
Hematol. Oncol. Clin. North Am. 24: 1033–1054.
Vichinsky, E. (2013). Natural history and clinical manifestations
of the alpha thalassemias. Cold Spring Harb. Perspect. Med.
https://doi.org/10.1101/cshperspect.a011742.
Evolutionary background of hemoglobin
disorders
Penman, B.S., Pybus, O.G., Weatherall, D.J., and Gupta, S. (2009).
Epistatic interactions between genetic disorders of hemoglobin can
explain why the sickle-cell gene is uncommon in the Mediterranean.
Proc. Natl. Acad. Sci. U.S.A. 106: 21242–21246.
Beginnings: the molecular pathology of hemoglobin 19
Weatherall, D.J., Williams, T.N., Allen, S.J., and O’Donnell, A. (2010).
The population genetics and dynamics of the thalassemias. Hematol.
Oncol. Clin. North Am. 24: 1021–1031.
Williams, T.N., Mwangi, T.W., Wambua, S. et al. (2005). Negative epis-
tasis between the malaria-protective effects of alpha+ − thalassemia
and the sickle cell trait. Nat. Genet. 37: 1253–1257.
Molecular basis of prevention and
management of hemoglobin disorders
Cao, A. and Kan, Y.W. (2013). The prevention of thalassemia.
Cold Spring Harb. Perspect. Med. https://doi.org/10.1101/cshperspect.
a011775.
Lo, Y.M. and Chiu, R.W. (2010). Noninvasive approaches to prena-
tal diagnosis of hemoglobinopathies using fetal DNA in maternal
plasma. Hematol. Oncol. Clin. North Am. 24: 1179–1186.
Nienhuis, A.W. and Persons, D.A. (2013). Development of gene ther-
apy for thalassemia. Cold Spring Harb. Perspect. Med. https://doi.org/
10.1101/cshperspect.a011833.
Orkin, S.H. and Reilly, P. (2016). MEDICINE. Paying for future success
in gene therapy. Science 352: 1059–1061.
Chapter 2 Stem cells
David T Scadden1,2
1 Department of Stem Cell and Regenerative Biology, Harvard Stem Cell Institute, Harvard University, Boston MA, USA
2
Center for Regenerative Medicine, Massachusetts General Hospital, Boston MA, USA
Introduction, 21
Stem cell definitions and distinctions, 21
Hematopoietic stem cell concepts and their origin, 21
Molecular regulation of hematopoiesis, 24
Introduction
The generation of sufficient numbers of blood cells to main-
tain homeostasis requires the sustained production of mature
cells. This process, called hematopoiesis, yields approxi-
mately 1011 blood cells daily, with the capability for dra-
matic increases in the number and subsets of cells in response
to physiological stress. Hematopoiesis is therefore a highly
dynamic process dependent upon numerous modulating fac-
tors. Its prodigious production capability derives from the
sustained presence of a cell type which is generally quiescent,
but the descendants of which proliferate vigorously. This cell
is the hematopoietic stem cell (HSC).
Stem cell definitions and distinctions
Stem cells derive their name from their ability to produce
daughter cells of different types. Stem cells are defined by a
combination of the traits of self-maintenance and the ability
to produce multiple, varied offspring. Putting this in more
biological terms, stem cells have the unique and defining
characteristics of self-renewal and differentiation into multi-
ple cell types. Thus, with each cell division there is an inher-
ent asymmetry in stem cells that is generally not found with
other cell types.
While their name implies that stem cells have specific
intrinsic characteristics, there are multiple different types of
stem cells, each defined by their production ability. Totipo-
tent stem cells are capable of generating any type of cell in the
body, including those of the extra-embryonic tissues, such as
Molecular Hematology, 4th Edition. Edited by Drew Provan and John G. Gribben.
© 2020 John Wiley & Sons Ltd. Published 2020 by John Wiley & Sons Ltd.
Trafficking of primitive hematopoietic cells, 29
Manipulating HSCs for clinical use, 31
Summary, 34
Further reading, 34
the placental tissues (Figure 2.1). Pluripotent stem cells may
give rise to any type of cell found in the body except those
of the extra-embryonic membranes. They can produce ecto-
derm, mesoderm, or endoderm cells. It has also become pos-
sible to create pluripotent cells by “reprogramming” mature
cells. This allows for pluripotent cells to be made from any
individual, a powerful tool for basic biology, disease mod-
eling, and possible future cell therapies. These are called
induced pluripotent stem cells (iPS). Pluripotent stem cells
not made by reprogramming include embryonic stem cells,
isolated from the inner cell mass of the blastocyst; embry-
onic germ cells, isolated from embryonic gonad precursors;
and embryonic carcinoma cells, isolated from teratocarcino-
mas. Pluripotent stem cells may be maintained indefinitely in
culture under specialized conditions that prevent differenti-
ation. In particular, embryonic stem cells have been used to
generate “knockout” mice, animals harboring targeted gene
disruptions via homologous recombination that permit the
in vivo study of individual gene function. Lastly, multipo-
tent stem cells, such as the HSCs of the bone marrow, are
capable of giving rise to multiple mature cell types, but only
those of a particular tissue, such as blood. Multipotent stem
cells are found in adults, perhaps in all tissue, and function
to replace dead or damaged tissue. Such stem cells are com-
monly referred to as “adult” stem cells.
Hematopoietic stem cell concepts and
their origin
The cellular compartment model
The short-lived nature of most blood cells was first deduced
in the 1960s using thymidine labeling of reinfused blood.
These studies demonstrated that the maintenance of normal
21
22 Molecular Hematology

Inner cell mass

Fertilization Totipotent cells Blastocyst Fetal tissues Adult tissues

Pluripotent cells Multipotent cells

Embryonic stem “Tissue” or “adult”

(ES) cells stem cells

Fig. 2.1 Sources and types of stem cells.

Source: Adapted with gratitude from the National Institutes of Health Stem Cell Information website.

numbers of blood cells in the adult requires a process with
the capacity to briskly generate large numbers of mature
cells along multiple blood lineages. The early history of
HSC research was largely shaped by cellular biology and
animal transplantation experiments. It was advanced by
experiments in the early 1960s demonstrating that injection
of marrow cells could generate large hematopoietic colonies
in the spleen of irradiated mice. Such colonies were the
clonal progeny of single initiating cells, termed colony-
forming units, spleen (CFU-S), and contained hematopoietic
populations of multiple lineages. CFU-S were further
transplantable, demonstrating their self-renewing nature.
HSCs are a minor component of marrow cells, able both
to generate large numbers of progeny differentiated along
multiple lines and to renew themselves.
The field was further advanced by the use of in vitro cell
culture techniques; in particular, solid-state cultures of mar-
row and spleen cells furthered understanding of the colony-
forming capacity of individual hematopoietic cells. The orig-
inal technique demonstrated clonal colonies of granulocytes
and/or macrophages, termed in vitro colony-forming cells
(CFCs), which are now considered lineage-committed pro-
genitor cells. These cells could be separated from whole mar-
row cells and from CFU-S, were more numerous than CFU-S,
and could be detected in splenic colonies as the progeny
of CFU-S. These observations gave rise to the concept of
the three-compartment model of hematopoiesis, the compart-
ments being stem cells, progenitor cells, and dividing mature
cells in increasing numbers; each compartment consists of
the amplified progeny of cells in the preceding compartment.
Subsequent analyses have added further complexity to
the compartment model of hematopoiesis. The term CFU-S
describes at least two groups of precursor cells. One group,
arising from committed progenitors with little capacity
for self-renewal, gives rise to colonies that peak in size by
day 8, while a second, arising from a more primitive cell
that is capable of self-renewal, yields colonies that peak in
size at day 12. To further highlight the complexity of the
hematopoietic hierarchy, a rarer population of hematopoietic
cells provides longer-term repopulation of an irradiated host
than CFU-S. These long-term repopulating cells have the
capacity for sustained self-renewal and were considered the
true adult stem cells. The presence of stromal cells in the
cultures is important for the long-term culture of CFU-S
and repopulating cells. Cells capable of long-term survival in
culture on stroma were termed long-term culture-initiating
cells (LTC-ICs) and cobblestone area-forming cells (CAFCs).
These multipotential cell types were considered more
primitive than lineage-committed progenitor cells, but more
mature than long-term repopulating cells.
Thus, a more complex version of the compartmental
model has emerged. This provides a model with two pop-
ulations of stem cells, the most immature group consisting
of long-term repopulating cells and a more mature group
of short-term repopulating cells. These groupings are simple
models for what is likely to be a continuum of cells with self-
renewal and multipotent differentiation capacity. The long-
term repopulating cells also tend to be more quiescent, serv-
ing as a deep reserve for blood cell production in times of
physiological stress. More mature progenitors (also called
colony-forming cells or CFCs) have the capacity to give rise to
colonies of clonal origin in semisolid media containing fully
mature cells, permitting their analysis. A more mature set of
precursor cells constitutes the bulk of bone marrow cells and
has unique, identifiable features by light microscopy. Rapid
division of precursor cells culminates in the production of

Stem cells
Microenvironment
Long-term
LTC-IC/CAFC
Fig. 2.2 Schematic view of hematopoiesis. See text
for definition of abbreviations.
Source: Modified from Figure 12.1 in Hematology: Basic
Principles and Practice, 3rd edn (ed. R. Hoffman), 2000,
with permission from Elsevier. CFU-F/RF
mature cells. Although hematopoiesis proceeds according to
this orderly scheme (Figure 2.2), special consideration must
be given to the development of T and B lymphocytes. These
cells are generated in the thymus and bone marrow, respec-
tively, by a similar hierarchical process. Mature T and B
lymphocytes enter peripheral lymphoid organs, where they
encounter relevant antigens, leading to the production of new
cells from reactivated mature cells. This process amplifies the
de novo bone marrow formation of T and B lymphocytes. In
addition, some members of this type of cell, memory T or B
lymphocytes, are capable of sustained self-renewal, serving,
in effect, as unipotent stem cells. Their inability to produce
multiple different types of daughter cells distinguishes them
from multipotent HSCs.
In summary, the compartment model has given rise to
terms that are generally applied to cells of hematopoietic ori-
gin. HSCs are those that are multipotent and self-renewing.
Progenitor cells have limited ability to self-renew and are
likely to be unipotential or of very limited multipotential.
Precursor cells are restricted to a single lineage, such as neu-
trophil precursors, and are the immediate precursors of the
mature cells found in the blood. The mature cells are generally
short-lived and preprogrammed to be highly responsive to
cytokines, while the stem cells are long-lived, cytokine resis-
tant, and generally quiescent.
Models of lineage commitment
Several theories have emerged to describe the manner by
which HSCs undergo lineage commitment and differentiate.
Some studies support a deterministic theory whereby the
stem cell compartment encompasses a series of closely
related cells maturing in a stepwise process. Other studies
Stem cells 23
Hematopoiesis
Progenitor cells Precursor cells Mature cells
Short-term
CFU-S
HPP-CFC
CFU-C
CFU-Mix
suggest that hematopoiesis is a random, stochastic process.
The stochastic theory is based on in vitro observations that
multilineage colonies develop variable combinations of
lineages, and that such lineage choices occur independently
of external influences.
Similar controversy exists regarding the role of cytokines
in cell lineage determination. An instructive model suggests
that cytokine signaling forces the commitment of prim-
itive cells along a particular lineage. Ectopic expression
of the granulocyte macrophage colony-stimulating factor
(GM-CSF) receptor in a common lymphoid progenitor
(CLP) population was capable of converting the cells from a
lymphoid to a myeloid lineage. The influence of the GM-CSF
receptor was sufficiently dominant to change the entire
differentiation program of cells, but only the CLP stage of
development. A permissive model postulates that decisions
about cell fate occur independently of extracellular signals.
This model suggests that cytokines serve only to allow certain
lineages to survive and proliferate. Evidence supporting this
model is provided by the ectopic expression of growth recep-
tors in progenitor cells. Expression of the erythropoietin
receptor in a macrophage progenitor results in macrophage
colony formation, whereas expression of the macrophage
colony-stimulating factor (M-CSF) receptor in an erythroid
progenitor results in erythroid rather than macrophage
colony formation. Replacing the thrombopoietin receptor
(c-mpl) with a chimeric receptor consisting of the extra-
cellular domain of c-mpl with the cytoplasmic domain of
the granulocyte colony-stimulating factor (G-CSF) receptor
results in normal platelet counts in homozygous “knock-in”
mice. Therefore, the instructive and permissive models may
both be correct, but at different stages of hematopoietic
differentiation. Cells at earlier points in the differentiation
24 Molecular Hematology
cascade may be more susceptible to fate-altering stimuli,
while more committed cells may be irreversibly deter-
mined, with only proliferation, cell death, or the rate of
differentiation susceptible to influence by external signals.
Molecular regulation of
hematopoiesis
The molecular nature of stem cell regulatory pathways has
been determined using a variety of genetic approaches,
including genetic loss-of-function and gain-of-function stu-
dies. These have provided several important concepts regard-
ing the molecular control of hematopoiesis. First, some genes
have binary functions and exert an effect by being above a
threshold of expression, rendering their targets “on.” Other
genes function in a continuum and have different effects at
different levels. Secondly, while perturbations in single genes
may have dramatic cellular effects, gene products often func-
tion in complexes and, in the case of transcription regulators,
are only active if they have access to sites of open chromatin.
Finally, signal integration often depends on the assembly
of large signaling complexes and the spatial proximity of
molecules to facilitate interaction is therefore important.
Cell-intrinsic regulators of hematopoiesis
Cell cycle control
The quiescent nature of HSCs is supported by their low level
of staining with DNA and RNA nucleic acid dyes, which is
consistent with low metabolic activity. These studies have
indicated a heterogeneity among stem cells with a subgroup
that is deeply quiescent. Various studies have sought to deter-
mine the cell-intrinsic regulators of hematopoiesis involved
in HSC cycle control.
RNA analysis has been used to profile pertinent tran-
scription factors and other molecules in HSCs induced
to differentiate along various lineages by the application
of cytokines. Elevated levels of cyclin-dependent kinase
inhibitors (CDKIs) have been observed, suggesting that
CDKIs present in HSCs function to exert a dominant
inhibitory tone on HSC cell cycling. The bone marrow
of some mouse strains deficient in CDKI p57(CDKN1C),
p21(CDKN1A), or p18(CDKN2C) have increased HSC cell
cycling, suggesting that these CDKIs function as domi-
nant negative regulators of HSC proliferation. Other CDKIs,
such as p27(CDKN1B), may serve as negative regulators of
hematopoietic progenitor cells.
Self-renewal, commitment, and lineage determination
Experimental results involving transcription factors have
demonstrated cell-intrinsic roles in both global and lineage-
specific hematopoietic development. Mice engineered
to lose the function of specific genes have been used to
define the impact of those genes on hematopoiesis. Loss-
of-function studies involving the transcription factors
c-Myb, AML1 (CBF2), SCL (tal-1), LMO2 (Rbtn2), GATA-2,
and TEL/ETV6 have demonstrated global effects on all
hematopoietic lineages. Stem cells in animals deficient in
these molecules fail to establish definitive hematopoiesis.
Similar methods have indicated that some regulators have
different roles at different times in development. For example,
the gene product of transcription factor SCL is absolutely
required for establishing HSCs. Unexpectedly, there is not
a requirement for SCL once the stem cell pool is present
in the adult. Rather, SCL is required only for erythroid
and megakaryocytic homeostasis. Therefore, transcription
factor regulation of the stem cell compartment can be highly
dependent on the stage of development of the organism.
Loss-of-function studies have also proved useful in identi-
fying lineage-specific transcription factors. Mice genetically
deficient in the transcription factor Ikaros lack T and B lym-
phocytes and natural killer cells, but maintain erythropoiesis
and myelopoiesis. Notch is required for T lineage induction
and shifts differentiation away from B cell development.
Further, losing expression of some genes can enable cells to
revert in differentiation. For example the Pax-5 transcription
factor is essential for B cell maturation, and loss of it results
in cells that have already rearranged their B cell receptor
or immunoglobulin locus showing up in the T and NK cell
compartment. The critical roles for specific transcription
factors in establishing and maintaining cell fate are shown in
Figure 2.3. The ability of transcription factors to affect gene
expression is also highly dependent upon epigenetic features
of a particular gene locus. The histone marks and chromatin
accessibility regulators that participate in the epigenome are
beyond the scope of this chapter but should be recognized
as critical aspects of gene expression and therefore, cell
state.
Cell-extrinsic regulators
Ultimately, hematopoietic stem and progenitor cell decisions
are regulated by the coordinated action of transcription fac-
tors as modified by extracellular signals. Extracellular sig-
nals in the form of hematopoietic growth factors are medi-
ated via cell surface hematopoietic growth factor receptors.
Hematopoietic growth factors exert specific effects when act-
ing alone and may have different effects when combined
with other cytokines. There are at least six receptor super-
families, and most growth factors are members of the type
I cytokine receptor family. The effects of various cytokines
during myelopoiesis are illustrated in Figure 2.4.
Type I cytokine receptors
Type I receptors do not possess intrinsic kinase activity, but
lead to phosphorylation of cellular substrates by serving
 Stem cells 25

NK cell
Pro-T

T cell
Lymphoid pathway IL-7R+
CLP SCL (–) C/EBPα (–)
GATA-2 (–) PU.1 (–)
NF-E2 (–) Aiolos (++)
IL-7R+ GATA-1 (–) GATA-3 (++)
c-mpl–
Pro-B
SCL (–) C/EBPα (–)
GATA-2 (–) PU.1 (+)
HSC B cell
NF-E2 (–) Aiolos (+)
IL-7R+
GATA-1 (–) GATA-3 (+)
SCL (–) C/EBPα (–)
GATA-2 (–) PU.1 (+)
NF-E2 (–) Aiolos (++)
LT-HSC ST-HSC GATA-1 (–) GATA-3 (–)
SCL (++) C/EBPα (±) GMP
Myeloid pathway
GATA-2 (++) PU.1 (±) Monocyte
NF-E2 (–) Aiolos (±) CMP
IL-7R–
GATA-1 (±) GATA-3 (±) Epo-R– Granulocyte
c-mpl+
SCL (+) C/EBPα (++)
GATA-2 (–) PU.1 (±)
SCL (++) C/EBPα (±) NF-E2 (–) Aiolos (–)
GATA-2 (+) PU.1 (±) GATA-1 (–) GATA-3 (–)
NF-E2 (+) Aiolos (±) MEP
GATA-1 (+) GATA-3 (–) Megakaryocyte
Epo-R+
Erythrocyte
SCL (++) C/EBPα (–)
GATA-2 (++) PU.1 (±)
NF-E2 (++) Aiolos (–)
GATA-1 (++) GATA-3 (–)

Fig. 2.3 Transcription factors active at various stages of hematopoiesis. CLP, common lymphoid progenitor; CMP, common myeloid
progenitor; GMP, granulocyte monocyte progenitor; MEP, megakaryocyte erythrocyte progenitor; NK, natural killer.
Source: Redrawn from Akashi, K., Traver, D., Miyamoto, T., Weissman, I.L. (2000). A clonogenic common myeloid progenitor that gives rise to all
myeloid lineages. Nature 404: 193–197, with permission of Spring Nature.

as docking sites for adapter molecules with kinase activity. Receptor tyrosine kinases
Examples of receptors in this family include leukemia
Receptors with intrinsic kinase activity are of considerable
inhibitory factor (LIF), interleukin (IL)-1, IL-2, IL-3, IL-4,
relevance to hematology because their ligands are growth fac-
IL-5, IL-6, IL-7, IL-9, IL-13, IL-18, GM-CSF, G-CSF, erythro-
tors, but also because abnormalities of them can result in
poietin, prolactin, growth hormone, ciliary neurotrophic
unregulated activation. For example, internal tandem dupli-
factor, and c-mpl. These receptors share several features,
cations of the Flt3 receptor are associated with acute myeloid
including enhanced binding and/or signal transduction
leukemia (AML) and activating mutations of c-kit play a
when expressed as heterodimers or homodimers, four cys-
key role in systemic mastocytosis and core binding factor
teine residues and fibronectin type III domains in the extra-
AMLs. Loss of c-fms, the receptor for M-CSF, has been asso-
cellular domain, WSXWS ligand-binding sequence in the
ciated with myelodysplasia and a predisposition to AML. The
extracellular cytokine receptor domains, and lack of a known
availability of agents targeting tyrosine kinases also makes
catalytic domain in the cytoplasmic portion. Another shared
these receptors of particular interest in hematologic disease.
feature of receptors in this family is the ability to transduce
Details of receptor–ligand pairs are provided in Table 2.1.
signals that prevent programmed cell death (apoptosis).
Protein serine–threonine kinase receptors
Type II cytokine receptors
This family includes the 30 members of the transforming
This class includes the receptors for tissue factor, IL-10, and growth factor (TGF)-β superfamily, which bind to their
interferon (IFN)-γ. This family contains a type III fibronectin receptors as homodimers. Members of this family include
domain in the extracellular domain, like the type I family. the three TGF-β receptors: type I (TbRI, 53 kDa), type II
26 Molecular Hematology

G-CSF, IL-1, IL-6, IL-10, IL-11, IL-12, IL-13

HPP-CFC
b-FGF, HGF, LIF, SCF/KL, FLT3 Ligand, TPO
Mixed progenitor Pluripotent
CFU-GEMM stem cell
cell
FLT3, SCF, IL-3, IL-6, GM-CSF, G-CSF SCF IL-4
Myelomonocytic IL-3 IL-5
progenitor G-CSF
CFU-GM GM-CSF
SCF, IL-3, IL-6, G-CSF
GM-CSF, CSF-1 IL-5

CFU-M CFU-G CFU-Eos CFU-Baso

NGF

CSF-1, IL-3 G-CSF IL-5 IL-10, IL-9,

GM-CSF GM-CSF GM-CSF CSF
IL-4
Monoblast Myeloblast

CSF-1 G-CSF
GM-CSF GM-CSF
Promonocyte Myelocyte

IL-5, IL-3, IL-4

GM-CSF
CSF-1 G-CSF
Basophil,
Monocyte Neutrophil Eosinophil mast cell
NGF

Chemokines
MCP-1-3, IP-10, Rantes, MIP-1α IL-8, NAP-2, Gro-α Eotaxin, MCP-4 Rantes IL-8, MCP-1, 3
Fig. 2.4 Cytokines active at various stages of hematopoiesis. See text for definition of abbreviations.
Source: Modified from Figure 16.3 in Hematology: Basic Principles and Practice, 3rd edn (ed. R. Hoffman), 2000, with permission from Elsevier.

(TbRII, 75 kDa), and type III (TbRIII, 200 kDa). Members
of this family have a profound inhibitory effect on the
growth and differentiation of hematopoietic cells and on
auxiliary hematopoietic cells. Binding of TFG-β requires
TbRII. After binding, signal transduction occurs via acti-
vation of serine–threonine kinase cytoplasmic domains of
the receptor chains, which results in the phosphorylation of
Smad molecules on serines. Phosphorylated Smad complexes
translocate to the nucleus, where they induce or repress
gene transcription. TGF-β is the best-characterized negative
regulator of hematopoiesis. It inhibits mitosis by inducing
cell cycle inhibitors such as p21(CDKN1A), p27(CDKN1C),
and p16(CDKN2A), inhibiting the cyclin-dependent kinases
Cdk4 and Cdk6, and inducing phosphorylation of the
retinoblastoma protein. The TGF-β receptor family and its
downstream mediators act as braking factors for a number of
cell types and are frequently inactivated by somatic mutation
in a number of cancers.
Chemokine receptors
This family comprises seven transmembrane-spanning
G-protein-coupled receptors that influence both cell cycle
and cellular movement, or chemotaxis. These receptors are
divided into three families, α or CXC, β or CC, and γ or C, on
the basis of variability in cysteine residues. The best charac-
terized is CXCR4, which mediates homing and engraftment
of HSCs in bone marrow and is critical to hematopoietic
development. IL-8 and macrophage inflammatory protein
(MIP)-1α act as inhibitors of progenitor cell proliferation.
 Stem cells 27

Table 2.1 Factors affecting hematopoietic control

Growth
factor
Growth factor receptor Produced by Bioactivity Deficient states

Erythropoiesis
EPO (erythropoietin) EPO-R Adult kidney Stimulates clonal growth of CFU-E and BFU-E subsets Anemia
Liver during Suppresses erythroid progenitor cell apoptosis
development Induces bone marrow release of reticulocytes
Induces erythroid globin synthesis
SF (steel factor), kit c-kit (CD117) Bone marrow Promotes proliferation and differentiation of pre-CFC cells Anemia
ligand, mast cell perivascular Acts synergistically with IL-3, GM-CSF, and TPO to support Mast cell deficiency
growth factor mesenchymal growth of CFU-GEMM, BFU-E, and CFU-Mk
cells Expansion of committed progenitor cells in vivo
Endothelial cells Stimulates mast cell hyperplasia, degranulation, and
IgE-dependent mediator release
IGF-1 (insulin-like IGF-1R Liver Induces DNA synthesis and has anti-apoptotic effects in Growth retardation,
growth factor, erythroid progenitors neurological
somatomedin C) Simulates erythroid colony growth in the absence of EPO defects,
at high doses homozygous
deficiency lethal
Granulopoiesis
G-CSF (granulocyte G-CSFR Monocytes, Stimulates growth of progenitors committed to neutrophil Neutropenia, failure to
colony-stimulating macrophages, differentiation develop neutrophilic
factor) endothelial cells, Activates neutrophil phagocytosis leukocytosis in
fibroblasts Stimulates quiescent HPCs to enter G1 /S response to
Stimulates mobilization of HSCs and HPCs from bone infection
marrow to periphery
GM-CSF GM-CSFR Mast cells, T Stimulates multilineage hematopoietic progenitor cells Susceptibility to
(granulocyte lymphocytes, Stimulates BFU-E and granulocyte, macrophage, and infections caused by
macrophage endothelial cells, eosinophil colony growth obligate
colony-stimulating fibroblasts, thymic intracellular
factor) epithelial cells organisms
M-CSF (macrophage c-fms Monocytes, Induces monocyte/macrophage growth and differentiation Macrophage and
colony-stimulating macrophages, and activation osteoclast
factor) fibroblasts, deficiency,
epithelial cells, hematopoietic
vascular failure
endothelium,
osteoblasts
Thrombopoietin c-mpl Bone marrow Stimulates in vitro growth of CFU-Mk, megakaryocytes, Thrombocytopenia
stroma, spleen, and platelets
renal tubule, liver, Stimulates clonal growth of individual CD34+ CD38− cells
muscle, brain Synergizes with SF, IL-3, and FL
Primes response to platelet activators ADP, epinephrine,
and thrombin, but no effect on aggregation
IL-5 IL-5R T lymphocytes Stimulates eosinophil production and activation Inability to mount
Activates cytotoxic T cells eosinophilic
Induces immunoglobulin secretion response
IL-11 IL-11R Fibroblasts, bone Acts synergistically with IL-3 or SF to stimulate the clonal No hematological
marrow stroma growth of erythroid (BFU-E and CFU-E) and primitive defect
megakaryocytic (BFU-Mk) progenitors
Shortens duration of G0 of HPCs
Quickens hematopoietic recovery after chemotherapy and
radiation

(continued)
28 Molecular Hematology

Table 2.1 (Continued)

Growth
factor
Growth factor receptor Produced by Bioactivity Deficient states

Lymphopoiesis
IL-7 IL-7R Bone marrow Induces clonal growth of pre-B cells B- and T-cell
stroma, spleen, Induces growth of pre-T cells lymphopenia
thymus
IL-2 IL-2R T lymphocytes Induces proliferation and activation of T cells, B cells, and Fatal
NK cells immunoproliferative
disorder, loss of
self-tolerance
IL-15 IL-15R Monocytes, Induces proliferation and activation of T cells, B cells, and
macrophages, NK cells
epithelial cells,
skeletal muscle
cells, bone
marrow and
thymic stroma
IL-4 IL-4R T lymphocytes Induces proliferation of activated B cells Defective T helper cell
Inhibits IL-2-stimulated proliferation of B cells responses
Induces T-cell proliferation
IL-10 Inhibits monocyte/macrophage-dependent synthesis of
Th1- and Th2-derived cytokines
Early-acting factors
IL-3 IL-3R T lymphocytes, mast Stimulates multilineage colony growth and growth of No hematopoietic
cells primitive cell lines with multilineage potential defect in steady
Stimulates BFU-E proliferation state, deficient
delayed-type
hypersensitivity
FLT3-ligand (FL) FLT-3R, flk2 Most tissues, Weak colony-stimulating activity alone, but synergizes Reduction in pro-B
including spleen, with IL-3, GM-CSF, SF, IL-11, IL-6, G-CSF, IL-7, and others cells, pre-B cells,
lung, stromal Augments retroviral transduction of HSCs when added to B-cell
cells, peripheral cytokine cocktails colony-forming
blood Mobilizes HSCs to periphery weakly alone, but adds potential, reduced
mononuclear cells greatly to G-CSF repopulating
capacity of stem
cells
IL-9 (T-cell growth IL-9R T lymphocytes Stimulates growth of BFU-E when combined with EPO
factor) Stimulates clonal growth of fetal CFU-Mix and CFU-GM
IL-6 IL-6R Macrophages, Synergistic with IL-3 for CFU-GEMM colony growth Reduced HSC and
endothelial cells, Synergistic with IL-4 in inducing T-cell proliferation and progenitor cell
fibroblasts, T colony growth survival, reduced
lymphocytes Synergistic with M-CSF in macrophage colony growth T-cell numbers,
Synergistic with GM-CSF in granulocyte colony growth reduced
Co-induces differentiation of B cells proliferation and
maturation of
erythroid and
myeloid cells

BFU-E, burst-forming unit, erythroid; CFU-mix, colony-forming unit, mix; CFU-Mk, colony-forming unit, megakaryocyte; CFU-GM, colony-forming
unit, granulocyte/macrophage; CFU-GEMM, colony-forming unit, granulocyte, erythroid, monocyte, megakaryocyte.

Members of this receptor family have also been implicated
in cancer metastasis and the entry of HIV-1 into cells.
Tumor necrosis factor receptor family
Members of the tumor necrosis factor receptor (TNFR) fam-
ily have varied effects, some having the ability to induce pro-
grammed cell death and others stimulating mesenchymal
cells to secrete hematopoietic growth factors. These recep-
tors contain Cys-rich extracellular domains and 80-amino
acid cytoplasmic “death domains,” which are required for
transducing the apoptotic signal and inducing NF-κB acti-
vation. Members of this family include TNFR1, TNFR2, fas,
CD40, nerve growth factor (NGF) receptor, CD27, CD30,
and OX40, each with at least one distinct biological effect.
Components of the hematopoietic
microenvironmental niche
While soluble factors influence stem cell fate, these factors
are seen by the cell in the context of the cell–cell contact
among heterologous cell types and cell–matrix contact that
comprise the three-dimensional setting of the bone marrow.
What actually constitutes the critical microenvironment for
hematopoiesis is surprisingly poorly defined. The ability of
primitive cells to mature in vitro in complex stromal cul-
tures suggests that at least some elements of the regulatory
milieu of the bone marrow can be recapitulated ex vivo. Stud-
ies based solely on ex vivo systems are suspect, however, as no
fully satisfactory re-creation of stem cell expansion or self-
renewal has been defined. Recognizing this limitation, it has
been determined that mesodermal cells of multiple types are
needed to enable hematopoietic support. These include mul-
tiple mesenchymal stromal cells, adipocytes, hematopoietic
cells, and endothelium. HSCs tend to be perivascular in loca-
tion endogenously and in periendosteal regions when trans-
planted. The most immature hematopoietic stem/progenitor
cells (HSPCs) have been reported to be adjacent to periar-
teriolar mesenchymal cells, while more mature cells of lym-
phoid lineage are nearby to more mature osteolineage cells.
There appears to be spatial organization of the bone marrow
with multiple niches for particular subsets of HSPC, though
the precise details of the marrow microenvironment are still
being defined.
Trafficking of primitive
hematopoietic cells
The migratory behavior characteristic of primitive
hematopoietic cells is an area of intense research because of
its relationship to bone marrow transplantation. Trafficking
of HSCs can be divided into the components of homing,
retention, and engraftment. Homing describes the tendency
Stem cells 29
of cells to arrive at a particular environment, while retention
is their ability to remain in such an environment after arrival.
Lastly, engraftment reflects the ability of cells to divide and
form functional progeny in a given microenvironment.
Much has been learned about trafficking from the ontogeny
of mouse and human HSCs.
Hematopoietic ontogeny
In both humans and mice, hematopoiesis occurs sequen-
tially in distinct anatomical locations during development.
These shifts in location are accompanied by changes in the
functional status of the stem cells and reflect the changing
needs of the developing organism. These are relevant for
adult hematopoiesis, since they offer insight into how the
blood production process can be located in different places
with distinct regulation.
There are essentially five sites of blood cell forma-
tion recognized in mammalian development, and these
are best defined in the mouse. At about embryonic day
7.5 (E7.5), blood and endothelial progenitors emerge in
the extra-embryonic yolk sac blood islands. The yolk sac
supports the generation of primitive hematopoietic cells,
which are primarily composed of nucleated erythrocytes and
macrophages. The macrophages generated in the yolk sac are
generally tissue-resident cells that durably populate tissues
throughout adult life. Examples are the microglia of the brain
and the Kupffer cells of the liver. More sustained or definitive
hematopoiesis derives from the aorta–gonad–mesonephros
(AGM) region, as has been defined in both the mouse (E8.5)
and the human. Current data suggest that yolk sac cells do
not seed the AGM region, but rather hematopoietic cells arise
there de novo. The placenta may also be a site of de novo HSC
generation. By E10 in the mouse, the fetal liver assumes the
primary role of cell production. By E14 in the mouse and
the second trimester of human gestation, the bone marrow
becomes populated with HSCs and it takes over blood cell
production, along with the spleen and thymus. The spleen
remains a more active hematopoietic organ in the mouse than
in the human. Stem cell proliferation is very active in the
AGM region and fetal liver. While it is also initially robust
in the bone marrow, there is a dramatic shift to quiescence
shortly after residence in the bone marrow, experimentally
defined as after four weeks in the mouse.
Homing and engraftment of HSCs
following infusion
Despite the use of HSC transplantation for over three
decades, the exact mechanisms whereby bone marrow cells
home to the bone marrow are not fully understood. Other
than lectins, no adhesion receptors have been identified that
are exclusively present on HSCs. Furthermore, no adhesion
30 Molecular Hematology
ligands, other than hemonectin, have been identified that are
exclusively present in the bone marrow microenvironment.
When first infused, HSCs lodge in the microvasculature
of the lung and liver; they then colonize the bone marrow,
first passing through marrow sinusoids, migrating through
the extracellular space of the bone marrow, and ultimately
settling in stem cell niches. Passage through the endothelial
barriers at first requires tethering, through endothelium-
expressed addressins that bind hematopoietic cell selectins,
and this is followed by firm attachment mediated by
integrins.
Selectins are receptors expressed on hematopoietic cells
(L- and P-selectins) and endothelium (E- and P-selectins).
They have long extracellular domains containing an
aminoterminal Ca2+ -binding domain, an epidermal growth
factor domain, and a series of consensus repeats similar to
those present in complement regulatory molecules. Ligands
for selectins are sialylated fucosyl glucoconjugates present on
endothelium. L-selectin is present on CD34+ hematopoietic
progenitors, while L-selectin and P-selectin are present on
more mature myeloid and lymphoid cells. Tethering by
selectins allows integrin-mediated adhesion to the endothe-
lium. Integrins, a family of glycoproteins composed of α
and β chains responsible for cell–extracellular matrix and
cell–cell adhesion, not only provide firm attachment, but also
allow migration of hematopoietic cells through the endothe-
lium and bone marrow extracellular space. The functional
state of integrins is only loosely tied to their expression level
and depends on ligand affinity modulation, regulated by the
β subunit in response to cytokines and other stimuli.
The process of migration depends on the establishment
of adhesion at the leading edge of the cell and simultaneous
release at the trailing edge. The rate of migration depends
on dynamic changes in the strength of the cell–ligand inter-
actions, which is dictated by the number of receptors and
their affinity state and the strength of the adhesion receptor–
cytoskeleton interactions. Cell–ligand interaction strength
may also be modulated by chemokines that provide direc-
tional cues for cell migration based on ligand gradients. Thus,
successful engraftment relies not only on the presence of sev-
eral different adhesion receptors, but also on the presence of
chemokine gradients that encourage directional movement
of the cells. Chemokines such as CXCL12, the ligand for the
receptor CXCR4, are localized discontinuously in the bone
marrow microvasculature and provide site-specific guidance
for HSPC migration and localization.
Egress of HSCs from bone marrow under
physiological conditions
The majority of primitive HSCs are resident within the bone
marrow space under steady-state physiological conditions.
However, a population of CD34+ cells capable of forming
CFCs and LTC-ICs and of long-term repopulation may be
found circulating in the peripheral blood, and these may
increase after physiological stressors such as exercise, stress,
and infection. Animal studies have suggested that a rela-
tively large number of bone marrow–derived stem cells cir-
culates during the course of a day, and that these cells peri-
odically transit back into an engraftable niche to establish
hematopoiesis. Defining the processes involved is important
in guiding new approaches to peripheral blood stem cell
mobilization for transplantation.
Examining mice in which specific adhesion molecules
have been deleted has revealed several key molecular deter-
minants of stem cell localization in the bone marrow. Among
these, the chemokine receptor CXCR4 has perhaps the most
striking phenotype. In the absence of this receptor, stem
cells fail to traffic from the fetal liver to the bone marrow
in development. Partly because of these studies, others have
defined that CXCR4 is relevant for the engraftment of trans-
planted stem cells, and that the modulation of CXCR4 sig-
naling can affect adult stem cell localization in the bone
marrow versus peripheral blood. As described, the integrin
and selectin families are also important molecular partici-
pants in stem cell location. For example, HSCs from animals
that are heterozygous deficient for β1 integrin cannot com-
pete with wild-type cells for the colonization of hematopoi-
etic organs. Pre-incubation of HSCs with α4 integrin anti-
bodies prior to transplantation results in decreased bone
marrow and increased peripheral recovery of cells, while
the continued presence of α4 antibodies prevents engraft-
ment. Evidence for selectin involvement has been demon-
strated in animals deficient of single selectins or combi-
nations of selectins. Endothelial P-selectin mediates leuko-
cyte rolling in the absence of inflammation, while L-, P-,
and E-selectins contribute to leukocyte rolling in the set-
ting of inflammation. L-selectin is important in lymphocyte
homing. Transplantation studies performed in animals defi-
cient in P- and E-selectins demonstrate severely decreased
engraftment due to impaired homing, an effect that is further
compromised by blocking vascular cell adhesion molecule
(VCAM)-1.
Mature hematopoietic cells are thought to migrate from
the marrow to the blood by similar mechanisms, though
these are not well defined. One purported mechanism is a
shift in expression from molecules thought to interact with
stromal proteins to those that interact with endothelium.
For example, myeloid progenitors express functional α4β1
and α5β1 integrins that act to ensure that these progenitors
are retained in the bone marrow through interactions
with VCAM and fibronectin. Mature neutrophils, in con-
trast, express β2 integrins that permit interaction with
ligands, such as intercellular adhesion molecule, expressed

by endothelial cells. Mature neutrophils also express β1
integrins that permit interaction with collagen and laminin
present in basal membranes, perhaps regulating a progres-
sive shift in cell affinities for specific microenvironmental
determinants that ultimately results in cell egress into the
blood. Mobilization of murine HSCs induced by cyclophos-
phamide or G-CSF is accompanied by changes in integrin
expression levels and functional changes in homing, thus
linking cellular localization with adhesion molecule receptor
expression.
Manipulating HSCs for clinical use
Mobilization of HSCs
Mobilization of HSCs in response to chemotherapy or
cytokines was first documented in the 1970s and 1980s.
This process may be induced by a variety of molecules,
including cytokines such as G-CSF, GM-CSF, IL-7, IL-3,
IL-12, SCF, and flt-3 ligand; and chemokines such as IL-8,
MIP-1α, Gro-β, and SDF-1. The one that is most often
used clinically is G-CSF, which may be combined with
chemotherapeutic agents for added benefit. This mobilizing
capability has resulted in a dramatic change in the manner
by which HSCs are harvested for transplantation. Up to 25%
of candidates for autologous transplantation are unable to
mobilize sufficient cells to enable the procedure to be safely
performed. The study of mobilization and its counterpart,
engraftment, has implications of great significance for
patient care. The ability of G-CSF to mobilize bone marrow
HSCs has several apparent mechanisms. The first is reported
to be the activation of neutrophils, causing the release of
neutrophil elastases capable of cleaving CXCR4 on HSCs,
thus reducing HSC–bone marrow interaction. Other recep-
tors that undergo cleavage are VCAM-1 and c-kit. A second
mechanism of G-CSF-induced mobilization is via CD26,
an extracellular dipeptidase present on primitive HSCs that
is able to cleave SDF-1 to an inactive form. Other options
for improving mobilization include co-administration of
G-CSF and inhibitors directed against the CXCR4–CXCL12
interaction. The inhibition of the CXCR4 receptor by a
small molecule has been shown to effectively mobilize
HSCs into the blood of patients, including those with poor
G-CSF-induced mobilization. Adding this compound to
G-CSF enhances HSPC mobilization and leukopheresis
yields. G-CSF does invoke an inflammatory response and
alternatives for its use are actively sought. Recent pre-clinical
data suggest that CXCR4 antagonism plus CXCR2 agonism
with agents such as Gro-b or CXCR4 plus VLA-4 antagonists
may provide efficient mobilization alternatives.
Stem cells 31
Isolating stem cells for manipulation
Characteristics of HSCs used for isolation
Physical Early attempts to isolate HSCs were based on cell
size and density. In order to clarify whether the heterogene-
ity of CFU-S was due to differences in the input cells used,
velocity sedimentation was performed to separate cells by
size, demonstrating that smaller cells were more likely to
produce secondary CFU-S than larger cells. HSCs are similar
in size to mature lymphocytes and, when flow cytometry
is performed, overlap the lymphocyte region on plots of
forward and side scatter.
Using cell-cycle-active drugs Because HSCs are largely in a
quiescent portion of the cell cycle (G0 or G1 ), investigators
have used cell-cycle-active drugs to deplete bone marrow
populations of cycling cells and thereby enrich for primitive
HSCs. Treatment of mice with nitrogen mustard resulted in
a 30-fold enrichment in CFU-S. HSCs may be isolated by in
vitro treatment with 5-fluorouracil, and this remains the most
commonly used agent. In addition, HSC populations may be
further enriched by first stimulating cells to enter the cell
cycle with the early-acting cytokines c-kit ligand and IL-3,
before forcing them to metabolic death. This strategy is use-
ful for human cells but not murine cells, probably because of
different cycling characteristics. It should be noted that these
techniques may affect the quality of HSCs obtained.
Markers of primitive HSCs A variety of strategies have been
used to identify HSC surface markers. In general, these have
depended on excluding cells that express antigens known to
be on more mature cells. For example, B cell, T cell, mono-
cyte, granulocyte, and erythroid markers are combined in
a so-called lineage cocktail to remove cells that bind any
of these antibodies. The remainder of the cells have been
sequentially tested for binding of particular antibodies and
then transplanted to test the bound cells’ function as stem
cells. This iterative process has enabled the definition of cells
enriched for reconstituting function in irradiated mice. In the
case of human cells, the mice are multi-immune deficient,
so they are tolerant of human hematopoietic cell engraft-
ment. Using such strategies, antibody combinations are now
defined that enrich for long-term reconstituting HSCs. The
panels of markers used are provided in Table 2.2.
Supravital stains Since HSCs are inherently quiescent, spend
most of their time in inactive portions of the cell cycle, and
are resistant to toxins, exclusion of dyes has been used as a
method of isolation. The DNA dye Hoechst 33342 was first
used to separate quiescent cells from the bone marrow. Cells
with low-intensity staining were enriched for high prolifer-
ative potential (HPP)-CFC and day-12 CFU-S. The red and
32 Molecular Hematology
Table 2.2 Proposed surface markers of primitive hematopoietic
stem cells
Mouse Human
CD34low/− CD34+
Sca-1+ CD49f+
C-kit+ Thy1+ (CD90)+
CD150+ CD38low/−
CD48− CD45RA−/low
lin− lin−
blue emissions from this dye have been used to define a small
subset of bone marrow cells known as the side population
(SP). SP cells have extremely low fluorescence emission in
these channels, resulting from efflux of Hoechst 33342 by
multidrug resistance pumps that are highly expressed on
HSCs. SP cells constitute approximately 0.1% of the bone
marrow and are highly enriched in reconstitution potential.
The mitochondrial dye rhodamine-123 (Rh-123) has also
been used to subdivide primitive stem cells. Mitochondria in
quiescent cells bind low levels of Rh-123 and fluorescence-
activated cell sorting (FACS) can be used to separate Rh-
123low cells. These cells were enriched for day-13 CFU-S and
multilineage reconstituting potential.
While supravital stains have been useful, the simplicity of
analysis with fluorescent antibodies against cell surface mark-
ers has generally supplanted them. Using a combination of
antibody staining features, it is possible to enrich for HSCs
such that fewer than 10 cells are required to reconstitute
hematopoiesis.
Methods of isolation of HSCs
Fluorescence-activated cell sorting
While the flow cytometer may be used for the analysis of cells,
the apparatus may also physically sort cells of desired fluo-
rescence or fluorescence pattern, size, and granularity char-
acteristics. Sorting is both expensive and labor intensive, as it
requires costly machines, a high degree of expertise, and time
to sort samples consisting of single-cell suspensions. How-
ever, FACS is the most commonly used method to isolate
highly purified HSCs using both positive and negative selec-
tion strategies, with fluorescence-labeled antibodies directed
against primitive hematopoietic cell antigens, as described
earlier.
Magnetic bead columns Large-volume isolation of HSC
subsets has been facilitated by the use of magnetic bead
columns. Using this system, cells are incubated with anti-
bodies directed against primitive hematopoietic cells. These
antibodies are typically coupled to a hapten. A second-step
incubation is then performed using a magnetic microbead
conjugated to a hapten that is able to bind the first-step
hapten. The effect is to label HSCs with a magnetic bead.
Cells are then passed through a column mounted adjacent to
a magnet. Labeled cells are retained within the column and
unbound cells can be washed through. Then, the column is
removed from the magnet and the desired cells may be eluted.
Alternatively, negative selection may be performed by cap-
turing only the cells that pass through the column. For exam-
ple, a sample may be depleted of mature cells by labeling
with antibodies directed against mature blood cell antigens
(Linpos ). Cells can then be passed over a column in which
the mature cells adhere and immature cells pass through and
may be isolated.
Systems of these types permit rapid isolation of large num-
bers of primitive cells of relatively high purity. Bead-based
methods sacrifice cell purity for higher cell yields compared
with FACS.
Ex Vivo expansion
Given the possible clinical applications of HSCs for such uses
as bone marrow transplantation, there is increasing interest
in strategies that both result in an increase in the quantity of
HSCs and the ability to manipulate HSCs ex vivo. Thus, ex
vivo expansion of HSCs represents a highly prioritized goal
of clinically oriented HSC research.
The first benefit of expanding HSCs is to provide sufficient
cells for transplantation when insufficient numbers exist. For
example, cord blood represents a rich source of primitive
CD34+ cells that are less immunocompetent and are there-
fore transplantable across partial human leukocyte antigen
(HLA) disparity barriers. However, the absolute quantity of
HSCs within a single cord blood is low and transplantation
is followed by periods of aplasia. Ex vivo expansion would
thereby facilitate cord blood transplantation. Similarly, selec-
tive expansion of HSC subsets would permit the extension
of tumor-free cells from patients with limited quantities
of normal bone marrow due to bone marrow–infiltrating
diseases, such as leukemia, for the purpose of autologous
transplantation.
The second benefit of ex vivo manipulation is that HSCs
have a relative growth advantage over other cell types, such
as tumor cells. Therefore, ex vivo growth provides a purging
effect. Furthermore, specific tumor cell purging may be
achieved via the application of certain cytokines (IL-2, IFN-
γ), antitumor agents such as 5-fluorouracil or cyclophos-
phamide, tumor-specific antibodies combined with
complement-mediated lysis, and oncogene-specific tyrosine
kinase inhibitors, in addition to other targeted therapies,
such as antisense oligonucleotides, prior to use of the graft.

The third benefit is the support of gene transfer into HSCs
for the purpose of gene therapy or gene editing. A variety
of gene transfer and gene editing mechanisms are improved
with cell cycling. Further, low efficiency events in gene mod-
ification may be made clinically relevant if HSC expansion
can be achieved.
Strategies to expand HSCs ex vivo have used a wide variety
of cytokine cocktails. These have often resulted in more cells,
but not more stem cells. Progress has been made with small
molecules added to cytokine combinations. The cytokines
generally involve kit ligand, thrombopoietin, IL-6, and flt-3
ligand. Adding aryl hydrocarbon receptor antagonist led to
improved outcome in umbilical cord blood transplant in one
clinical study. Also, early results look encouraging for nicoti-
naminde or the small molecule UM171 currently in clini-
cal testing. Other results indicate that culture of cells with
Notch ligands can reduce intervals of myeloid cell cytopenia
and brief exposure to prostaglandinE2 analogues may affect
cord blood transplant. This area of research has historically
been one of frustration, but these developments indicate that
progress is being made. (Note to reader: the chapter author
does have involvement in companies focused on this area, so
interpret with caution.)
Functional analysis of HSCs
Functional assays for HSCs are critical, as immunopheno-
typic and other isolation methods depend on characteris-
tics that are not necessarily consistent with stem cell func-
tion. This is particularly true when assessing cells after in
vitro manipulation or in older subjects. Immunophenotypes
have been defined by transplantation, generally using young
donors. Only functional validation should be considered a
bona fide measure of stem cells. However, even in the case
of most assays, the setting is transplantation, which is an
extreme state and does not necessarily reflect stem cells under
homeostasis.
In Vitro assays The CFU-C measures hematopoietic
progenitor function and is performed by plating cells in
semisolid media containing methylcellulose and one or more
cytokines. After 5–14 days, colonies comprising mature cell
populations committed to either myeloid or lymphoid
lineages may be observed. While most colonies obtained
using this assay are composed of cells of a single lineage,
less frequently multipotent progenitors can yield colonies
containing multiple lineages. Another type of primitive cell,
known as the HPP-CFC, which possesses a high degree of
proliferative and multilineage potential, may be detected
in this culture system. Formation of HPP-CFC colonies,
characterized by size greater than 0.5 mm and multilineage
composition, requires the use of multiple cytokines in order
Stem cells 33
to proliferate. These are more primitive progenitors, but they
cannot be equated with HSCs.
The LTC-IC assay correlates more closely to HSCs. Here,
hematopoietic cells are plated on top of stromal cell lines or
irradiated primary bone marrow stroma. Primitive HSCs are
able to initiate growth and generate progeny in vitro for up
to 12 weeks. Progenitor cells and mature myeloid cells are
removed weekly to prevent overgrowth. Ultimately, HSCs,
characterized by high proliferative and self-renewal capabil-
ities, are able to sustain long-term culture and may be enu-
merated at the conclusion of the assay.
The CAFC assay represents a type of LTC-IC that similarly
measures the ability of cells to initiate growth and generate
progeny in vitro for up to 12 weeks. However, the readout
is slightly different. Hematopoietic cells are plated at limit-
ing dilution on top of a monolayer consisting of irradiated
bone marrow stroma or a stromal cell line. The growth of
colonies consisting of at least five small non-refractile cells
reminiscent of cobblestones, found underneath the stromal
layer, is counted. Such cultures are maintained using weekly
half-media changes until up to 5 weeks after seeding. In this
assay, more primitive cells appear later, and day-35 CAFCs
represent a close correlate of a cell with in vivo long-term
multilineage repopulating potential. LTC-ICs may be enu-
merated after day 35 by completely removing the CAFC
medium, overlaying methylcellulose, and counting the num-
ber of colonies produced after 8–10 days.
In Vivo assays The CFU-S assay, first developed by Till and
McCulloch in 1961, is described earlier in this chapter (see
Hematopoietic stem cell concepts and their origin). Bone mar-
row or spleen cells are transplanted to irradiated recipients
and animals are killed after 8 or 12 days for analysis of spleen
colonies, termed CFU-S8 and CFU-S12 , respectively. Cells
that give rise to CFU-S8 are predominantly unipotential and
produce erythroid colonies. CFU-S12 colonies consist of sev-
eral types of myeloid cells, including erythrocytes, megakary-
ocytes, macrophages, and granulocytes. Cells giving rise to
CFU-S12 represent a more primitive population of multipo-
tent cells than those that result in CFU-S8 . However, these
assays measure primitive progenitors, not HSCs.
The long-term repopulation assay is a more accurate
measure of HSC activity. Whole collections of hematopoi-
etic cells or fractionated subpopulations are transplanted to
lethally irradiated syngeneic mice, typically by tail vein injec-
tion. Recipients are screened for ongoing hematopoiesis 8–
10 weeks after transplantation. By this time, hematopoiesis
is firmly established and donor-derived blood is produced
by transplanted HSCs. This assay requires that cells ful-
fill the two central features of HSCs: multilineage recon-
stitution, consistent with multipotentiality, and indefinite
hematopoiesis, indicative of self-renewal.
34 Molecular Hematology

Assess percentage of
multilineage bone
marrow engraftment
Lethal irradiation
using flow cytometry
with 9-10Gy
with antibodies directed
against Ly5.1 and Ly5.2
Ly5.2 test cells 104
>16 weeks

103

Ly5.1 PE
Ly5.1 mouse 102

101

Ly5.1 whole bone 100 101 102 103 104

marrow competitor Ly5.2 FITC
cells Fig. 2.5 Competitive repopulation assay.

Tracking of transplanted cells was originally conducted termed competitive repopulation units. The competitive
using radiation-induced chromosomal abnormalities or by repopulation assay using congenic mouse strains is depicted
retrovirally marking donor cells. However, a major advance in Figure 2.5.
in the ability to track transplanted cells has been the develop-
ment of congenic mice with minor allelic differences in the
leukocyte common antigen Ly5, which is expressed on all
Summary
nucleated blood cells. The C57/BL6 (“black-6”) strain con-
tains the Ly5.2 antigen, while the BL6/SJL strain contains a
Investigation of HSCs has been facilitated by the develop-
separate allele, Ly5.1. However, these syngeneic strains may
ment of in vitro and in vivo assays of hematopoietic cell func-
be transplanted interchangeably. Both antibodies are avail-
tion, followed by the identification of molecular cell surface
able with distinct fluorescent labels. FACS analysis using
markers that permit the isolation of purified subsets of cells
these antibodies permits measurement of donor-derived
with defined characteristics. Studies in this field have con-
reconstitution of the nucleated blood lineages. However, ery-
tributed greatly to the understanding of both general stem
throcytes and platelets do not express the Ly5 antigen and
cell biology and hematopoiesis. Further investigation of cell-
cannot be tracked using this technique. Instead, investigators
intrinsic and cell-extrinsic regulators of hematopoiesis will
use congenic strains with allelic variants of hemoglobin and
enable rational manipulation of HSCs and thereby extend the
glucose phosphate isomerase to track erythroid and platelet
current uses of stem cells in clinical practice.
engraftment, respectively.
A modification of this assay permits quantitation of HSCs
within the graft. Here, HSCs are quantified by transplanting
limiting-dilution numbers of bone marrow into lethally Further reading
irradiated recipients. Each recipient also receives 1 × 105
cells of the host’s marrow to ensure survival during the Regulation of hematopoiesis
period of pancytopenia immediately after irradiation. At
Chai-Ho, W. and Chute, J.P. (2017). Paracrine regulation of normal and
10–12 weeks, host peripheral blood is assessed to determine
malignant hematopoiesis. Curr. Opin. Hematol. 24 (4): 329–335.
whether donor-derived reconstitution has occurred. Donor Gottgens, B. (2015). Regulatory network control of blood stem cells.
cells must constitute at least 1% of the peripheral blood to Blood 125 (17): 2614–2620.
contend that at least one HSC was present in the donor Hu, D. and Shilatifard, A. (2016). Epigenetics of hematopoiesis and
population. Also, both lymphoid and myeloid lineages must hematological malignancies. Genes Dev. 30 (18): 2021–2041.
demonstrate at least 1% donor derivations. The percentage
of reconstituted animals in each group may be plotted
against the number of input cells to determine a limiting- Hematopoietic stem cell niche
dilution estimate of the frequency of HSCs within the donor Gao, X., Xu, C., Asada, N., and Frenette, P.S. (2018). The hematopoi-
population. This assay is termed a competitive repopulation etic stem cell niche: from embryo to adult. Development 145 (2),
assay, as transplanted HSCs compete with the host’s HSCs dev139691.
that survive irradiation-induced death, in addition to host Hoggatt, J., Kfoury, Y., and Scadden, D.T. (2016). Hematopoietic stem
cells transplanted with the graft. The HSCs detected are cell niche in health and disease. Annu. Rev. Pathol. 11: 555–581.

Hematopoietic stem cell quiescence
Foudi, A., Hochedlinger, K., Van Buren, D. et al. (2009). Analysis of
histone 2B-GFP retention reveals slowly cycling hematopoietic stem
cells. Nat. Biotechnol. 27 (1): 84–90.
Wilson, A., Laurenti, E., and Trumpp, A. (2009). Balancing dormant and
self-renewing hematopoietic stem cells. Curr. Opin. Genet. Dev. 19
(5): 461–468.
Isolation of hematopoietic stem cells
Majeti, R., Park, C.Y., and Weissman, I.L. (2007). Identification of a
hierarchy of multipotent hematopoietic progenitors in human cord
blood. Cell Stem Cell 1 (6): 635–645.
Oguro, H., Ding, L., and Morrison, S.J. (2013). SLAM family mark-
ers resolve functionally distinct subpopulations of hematopoietic
stem cells and multipotent progenitors. Cell Stem Cell 13 (1): 102–
116.
Stem cells 35
Hematopoietic stem/progenitor cell
expansion
Boitano, A.E., Wang, J., Romeo, R. et al. (2010). Aryl hydrocarbon recep-
tor antagonists promote the expansion of human hematopoietic stem
cells. Science 329 (5997): 1345–1348.
Lund, T.C., Boitano, A.E., Delaney, C.S. et al. (2015). Advances in umbil-
ical cord blood manipulation – from niche to bedside. Nat. Rev. Clin.
Oncol. 12 (3): 163–174.
Hematopoietic stem cell engraftment
Broxmeyer, H.E. (2016). Enhancing the efficacy of engraftment of cord
blood for hematopoietic cell transplantation. Transfus. Apher. Sci. 54
(3): 364–372.
Hoggatt, J., Speth, J.M., and Pelus, L.M. (2013). Concise review: sowing
the seeds of a fruitful harvest: hematopoietic stem cell mobilization.
Stem Cells 31 (12): 2599–2606.
Chapter 3 The genetics of acute
myeloid leukemias
Amy M. Trottier & Carolyn J. Owen
Division of Hematology and Hematological Malignancies, University of Calgary, Foothills Medical Centre, Calgary, Canada

Introduction, 37 AML therapies and MRD monitoring targeted by genetics, 45

AML with recurrent cytogenetic abnormalities, 37 Summary, 46
Molecular genetic aberrations not detectable by conventional Further reading, 46
cytogenetics, 40

biallelic mutations of CEBPA, as well as the addition of a

Introduction new provisional entity, AML with mutated RUNX1. Molecu-
lar analysis for mutations of CEBPA, NPM1, and FLT3 is now
Acute myeloid leukemia (AML) is a heterogeneous dis-
standard of care in newly diagnosed AML and the impor-
ease with respect to clinical and morphological features
tance of many other gene mutations is being increasingly
as well as acquired genetic aberrations. Traditionally AML
recognized. These insights have led to the incorporation of
patients were divided into three broad risk groups based
molecular aberrations into validated risk stratification mod-
on cytogenetic abnormalities: favorable, intermediate, and
els, such as that of the European LeukemiaNet (ELN), and
adverse, with each having different cure rates. However,
recommendations for risk-adapted treatment strategies, aim-
approximately 50% of patients have a normal karyotype
ing to improve treatment outcomes and minimize toxicity.
(normal-karyotype AML). These patients have varied clini-
Unfortunately, the overall outcomes in AML remain poor,
cal outcomes, with some attaining a good response to con-
with most patients succumbing to their disease. Outcomes
ventional chemotherapy and others having high rates of
are particularly poor for older adults (>65 years).
relapse and disease-related mortality. This led to the search
This chapter reviews the prognostically important recur-
for recurrent genetic aberrations that are not detectable
rent genetic aberrations observed in AML. As most cur-
by conventional cytogenetics. With the advent and wider
rently available data on the impact of genetic abnormalities in
availability of next-generation sequencing methods such as
AML are derived from large trials enrolling younger patients
whole-genome sequencing and gene expression profiling,
(<65 years), the implications for older adults, who constitute
significant progress has been made in the identification of
the majority of incident AML cases (median age at diagnosis
recurrent genetic mutations at the molecular level. These
is between 65 and 70 years), remain unclear.
“molecular markers” include mutations in specific genes as
well as altered gene expression.
The importance of these new molecular markers was man-
ifest in the 2008 update of the World Health Organization AML with recurrent cytogenetic
(WHO) classification with the addition of the provisional abnormalities
entities AML with mutated NPM1 and AML with mutated
CEBPA. Verification of the prognostic significance of these Cytogenetic abnormalities in AML have long been known to
molecular aberrations and the discovery of numerous other have prognostic relevance. Several recurrent translocations
leukemogenic driver mutations via whole-genome sequenc- are associated with a favorable prognosis when treated with
ing are reflected in the recent 2016 revision to the WHO appropriate therapeutic agents, while particular numerical
classification with the formal incorporation (no longer pro- chromosomal abnormalities, such as monosomies of chro-
visional entities) of AML with mutated NPM1 and AML with mosomes 5 and/or 7, are associated with a poor prognosis.
Advances in molecular genetic research have enabled a better
understanding of the mechanisms by which these transloca-
Molecular Hematology, 4th Edition. Edited by Drew Provan and John G. Gribben. tions cause leukemia. Importantly, these studies have demon-
© 2020 John Wiley & Sons Ltd. Published 2020 by John Wiley & Sons Ltd. strated that the pathogenesis of AML is one of a sequential

37
38 Molecular Hematology
acquisition of genetic aberrations, with a single aberration
being insufficient alone to cause overt leukemia.
Core-binding factor leukemias
The core-binding factor (CBF) is a key regulator of
hematopoiesis and the most frequent target of chromo-
somal translocations associated with leukemia. This tran-
scription factor is composed of two heterodimeric com-
ponents, the α-subunit encoded by RUNX1 (also called
AML1 or CBFA) and the β subunit encoded by CBFB.
Homozygous loss of function of either RUNX1 or CBFB
in genetically engineered mice results in a complete lack
of definitive hematopoiesis and embryonic death, indicat-
ing that both components of CBF are necessary for nor-
mal hematopoietic development. CBF AML is a relatively
frequent subtype of adult AML, classified in the favorable
risk category, accounting for ∼15% of adult AML. CBF
AML includes the balanced translocations t(8;21)(q22;q22),
in which the RUNX1 gene is fused to RUNX1T1 (ETO), and
inv(16)(p13.1q22) or t(16;16)(p13.1q22), in which CBFB is
fused to MYH11. The CBF partner gene in each case has
constitutive activity leading to persistent expression of the
RUNX1–RUNX1T1 (AML-ETO) or CBFβ–MYH11 fusion
protein, causing recruitment of the corepressor complex and
underexpression of genes regulated by the CBF complex.
Data from in vitro studies and transgenic animal models
suggest a dominant-negative role for these fusion genes.
RUNX1–RUNX1T1 acts as a dominant-negative inhibitor
of the wild-type CBF, with the fusion transcript retaining
RUNX1’s DNA binding and heterodimerization domains,
but lacking the C-terminal transcription activation domain.
The mechanism of dominant-negative activity is less clear
for CBFβ–MYH11, but both prevent transactivation of CBF
targets. Despite this understanding, the precise mechanism
by which these fusion proteins exert their leukemogenic
effect remains to be elucidated. However, it is clear that
the translocations alone are not sufficient to result in overt
leukemia. For example, conditional alleles of RUNX1–
RUNX1T1 expressed in adult hematopoietic progenitors are
not sufficient to cause AML, but result instead in a myelo-
proliferative phenotype. Experience from mouse models has
shown that expression of these fusion proteins resulted in
aberrant self-renewal but, by themselves, were insufficient
to result in leukemia, with exposure to chemical mutagens
being necessary to induce overt AML.
Although CBF AML is classified in the favorable risk
category, there is heterogeneity within this group and ∼40%
of patients will relapse after standard treatment. Recent
studies have found additional genetic aberrations in ∼90%
of CBF AML. Several of these cooperating mutations may
contribute to leukemogenesis as well as impact survival.
For example, c-KIT mutations – found in 20–25% of t(8;21)
and ∼30% of inv(16)/t(16;16) – negatively impact prognosis
by leading to a higher risk of relapse. Also, a recent study
using high-throughput sequencing in CBF AML found
that cooperative mutations in chromatin modifiers and
cohesin were common in t(8,21) patients and associated
with a particularly high risk of relapse when combined
with mutations in the tyrosine kinase signaling pathway.
Similar mutations in chromatin modifiers and cohesin were
uncommon in inv(16)/t(16;16) AML patients.
Acute promyelocytic leukemia
As with the CBF genes, there are also multiple chromoso-
mal translocations that involve the retinoic acid receptor
RAR𝛼 locus on chromosome 17. The subgrouping of patients
with RAR𝛼 translocations is particularly important because
of the distinctly better prognosis that most of these translo-
cations confer. In t(15;17)(q22;q12), the RAR𝛼 gene fuses
with a nuclear regulatory factor called the promyelocytic
leukemia gene (PML), giving rise to the PML–RARα gene
fusion product. Rarely, the PML–RARα fusion can result
from cryptic or complex cytogenetic rearrangements and
lack t(15;17)(q22;q12) on routine cytogenetic testing. Sev-
eral infrequent variant translocations with RAR𝛼 also occur
involving the ZBTB16 (previously known as PLZF), NPM1,
NuMA, STAT5B, or PRKAR1A genes. Each of these is asso-
ciated with an acute promyelocytic leukemia (APL) char-
acterized by a block in differentiation at the promyelocyte
stage of hematopoietic development. The t(15;17)(q22;q12)
is the most frequent translocation causing APL and has
been extensively studied. PML–RARα expression is associ-
ated with a block in differentiation due to aberrant recruit-
ment of the nuclear co-repressor complex, similar to obser-
vations in the context of the RUNX1–RUNX1T1 and CBFβ–
MYH11 fusions.
APL with PML-RARα is considered to have an excel-
lent prognosis (with a complete remission rate greater than
90%). It has a particular sensitivity to all-trans-retinoic acid
(ATRA), a ligand for RARα, which has proven to be very
effective therapy for this subtype of AML. The efficacy of
ATRA in the treatment of APL is related to the ability of
ATRA to bind to the fusion protein, with resultant disso-
ciation of the nuclear co-repressor complex. Promyelocytes
are then able to engage normal hematopoietic differentia-
tion programs that ultimately result in apoptotic cell death.
However, some variant RAR𝛼 translocations, including cases
with the ZBTB16-RARα and STAT5B-RARα fusion prod-
ucts, are resistant to ATRA, suggesting they have different
functional effects than the PML–RARα translocation. To
stress the importance of the PML-RARα fusion, including
cryptic rearrangements, the 2016 revision to the WHO clas-
sification has renamed APL with t(15;17)(q22;q12); PML-
RARα to APL with PML-RARα.

Similar to the CBF leukemias, expression of PML–RARα
is not sufficient to induce AML. There are several lines
of evidence supporting this assertion, including transgenic
murine models of PML–RARα-induced AML. In this model,
although the fusion gene is present in the germline and is
expressed during embryonic and adult development, the ani-
mals do not develop AML until three to six months after
birth, and even then with a modest penetrance of only 15–
30%, and often with the acquisition of secondary cytogenetic
abnormalities such as FLT3 mutations, which occur in 34–
45% of APL.
KMT2A gene rearrangements (11q23)
Abnormalities of 11q23 containing the KMT2A gene (for-
merly known as the mixed lineage leukemia, MLL, gene)
are found in 3–7% of AMLs. These abnormalities can occur
at any age, but are particularly common in children and
in therapy-related AML after treatment with DNA topoiso-
merase II inhibitors. The KMT2A gene is the most promiscu-
ous of genes, involved in over 120 different translocations in
AML with more than 70 different partner genes now identi-
fied. These translocations lead to fusion genes containing the
N-terminus of KMT2A fused in-frame to the partner gene,
most of which have subsequently been noted to have roles
in normal hematopoiesis. Molecular studies have also shown
that KMT2A is frequently rearranged by cryptic transloca-
tions that are not detectable by conventional cytogenetics.
KMT2A encodes for a histone methyltransferase involved
in the regulation of gene transcription via chromatin remod-
eling. It is important for embryonic development and regula-
tion of hematopoietic differentiation. Similar to RUNX1 and
CBFB, homozygous disruption of KMT2A in mice is lethal to
the embryos because of disordered hematopoiesis. KMT2A is
thought to function as both a transcriptional activator and
a transcriptional repressor, using different domains within
the protein. Importantly, the repression domain is more
N-terminal to the activation domain, such that oncogenic
KMT2A fusion proteins are thought to retain their repres-
sion, while losing their transcriptional activation capability.
The prognostic impact of KMT2A translocations is vari-
able and depends on the particular fusion partner gene. The
most frequent KMT2A translocation detectable by cytoge-
netics in AML is t(9;11)(p22;q23), involving MLLT3 (AF9),
which has an intermediate prognosis. In general, 11q23
abnormalities confer an intermediate to poor prognosis.
Other rare chromosomal translocations
Other rare recurrent chromosomal translocations, including
t(6;9)(p23;q34.1), inv(3)(q21.3q26.2) or t(3;3)(q21.3;q26.2),
and t(1;22)(p13.3;q13.3), are found in AML and are
listed as distinct entities in the WHO 2016 classification.
The genetics of acute myeloid leukemias 39
t(6;9)(p23;q34.1) results in the formation of a chimeric
fusion gene DEK-NUP214. The DEK-NUP214 protein alters
nuclear transport and acts as an aberrant transcription factor.
t(6;9)(p23;q34.1) is a rare adverse risk cytogenetic abnormal-
ity, found in 0.7–1.8% of AML. It typically occurs in children
and young adults, is often associated with basophilia and
multilineage dysplasia, is usually the sole karyotypic abnor-
mality, is commonly associated with FLT3-ITD mutations,
and is associated with a poor prognosis.
Inv(3)(q21.3q26.2) or t(3;3)(q21.3;q26.2) is the most com-
mon abnormality of chromosome 3q in AML, results in
MECOM activation by repositioning of a GATA2 enhancer,
and leads to GATA2 haploinsufficiency. Inv(3)/t(3;3) is found
in 1–2% of AML, is seldom seen in pediatric AML, and is
an adverse cytogenetic risk factor conferring a short over-
all survival (OS) and low complete remission rate. AML
(megakaryoblastic) with t(1;22)(p13.3;q13.3); RBM15-MKL1
occurs in <1% of AML, is found in infants and young chil-
dren (≤3 years) with acute megakaryoblastic leukemia with-
out Down syndrome, and is suggested to have a poor progno-
sis, though this is uncertain due to the low number of cases
reported.
Numerical chromosomal abnormalities
In addition to translocations, recurrent numeric chromo-
somal abnormalities are noted frequently in AML. These
numerical aberrations are non-random, with particular chro-
mosomes being more likely to exhibit losses or gains.
Acquired trisomy of chromosome 8 or 21 is frequently seen
and is associated with an intermediate prognosis. Partial or
complete loss of chromosome 5 and/or 7 confers a very poor
prognosis. Abnormalities of chromosome 7, usually in the
form of monosomy or partial deletion of the long arm, con-
stitute the second most frequent numerical chromosomal
abnormality in de novo AML, after trisomy 8.
Many investigators have attempted to narrow the region
of interest on chromosomes 5 and 7 in order to determine
the deleted genes that result in leukemia, but no single
culprit gene has been conclusively identified. However,
investigations in myelodysplastic syndromes (MDS) involv-
ing deletion of the long arm of chromosome 5 (5q–) were
more successful. A novel investigational technique using
RNA interference identified loss of the ribosomal subunit
encoding gene RPS14 as the probable cause of the erythroid
defect in the 5q– syndrome. In models using genetically
engineered mice, haploinsufficiency of ERG1 and APC genes
(both located on 5q) in combination with loss of TP53
activity increased the rate of development of AML, but did
not necessarily cause overt leukemia. These results suggest
that deletion of a number of different genes on chromosome
5 along with cooperating mutations in tumor suppressor
genes may be required to result in leukemia.
40 Molecular Hematology

Monosomy 7 is also interesting, because it is the most com-
monly acquired abnormality in children with syndromes that
predispose to AML, such as Fanconi anemia and other bone
marrow failure syndromes. This suggests that monosomy 7 is
a secondary genetic aberration and likely requires the pres-
ence of other mutations to cause AML.
Molecular genetic aberrations not
detectable by conventional
cytogenetics
The development of next-generation sequencing technology,
also known as high-throughput sequencing, has allowed for
rapid and relatively inexpensive DNA and RNA sequencing,
including whole-genome sequencing, and has revolutionized
the understanding of the complex genetic landscape in AML.
These technologies have uncovered hundreds of novel recur-
rent genetic mutations at the molecular level; however, only a
small subset of these have been found to contribute to leuke-
mogenesis and/or to clearly impact prognosis.
mutation alone is not sufficient to transform a normal cell
into a leukemic blast.
Using whole-genome and whole-exome sequencing of 200
cases of de novo AML, a landmark study by the Cancer
Genome Atlas Research Network published in 2013 gener-
ated a genomic database and found that nearly all cases had at
least one driver mutation in one of eight functional categories
relevant in the pathogenesis of AML, including the nucle-
ophosmin gene (NPM1) (27% of cases), activated signaling
(59%), myeloid transcription factors (22%), tumor suppres-
sor genes (16%), spliceosome genes (14%), cohesin-complex
genes (13%), DNA methylation-related genes (44%), and
chromatin-modifying genes (30%). The relative frequency of
mutations in these functional categories in de novo AML is
shown in Figure 3.1.
A summary of the now well-established molecular mark-
ers that have been incorporated into routine care of AML
(e.g. NPM1, FLT3, and CEBPA) as well as recently discov-
ered mutations in AML is presented in what follows, focusing
on driver gene mutations associated with possible prognostic
impact in patients with AML.

Driver gene mutations NPM1

The subset of recurrent gene mutations that have been impli-
cated in leukemogenesis are known as driver mutations.
The majority of mutations found in AML are not involved
directly in leukemogenesis, but rather are believed to occur
in unstable leukemic cells or in hematopoietic stem cells
prior to an AML-initiating event. These are called passenger
mutations and do not impact the course of disease. Some
driver mutations are thought to initiate the leukemogenic
process (the initiating or founding clone), whereas others
act in a cooperating fashion, accelerating or encouraging the
leukemogenic process, and in some cases contributing to
chemotherapy resistance and/or risk of relapse. Various pat-
terns of cooperativity as well as mutual exclusivity between
different genes have also been found. It is thus clear that one
NPM1 mutations were first reported in 2005 by Falini and
colleagues as the most frequent genetic alteration in patients
with normal-karyotype AML. NPM1 is a nucleolar phospho-
protein that continuously shuttles between the nucleus and
the cytoplasm, but with predominant nucleolar localization.
NPM1 acts as a molecular chaperone in the nucleolus and is
also involved in cell cycle progression and regulation of the
alternate reading frame (ARF)/TP53 tumor suppressor path-
way. Mutations in NPM1 occur in exon 12, with more than
95% consisting of a 4-bp insertion at position 960. This 4-bp
insertion results in a frameshift within the C-terminal region
of the NPM1 protein, leading to loss of nucleolar localization
and gain of a nuclear export signal. The mutant NPM1
protein (called NPM1c+) therefore acquires aberrant

NPM1 (27%) Activated signaling (59%)

Myeloid transcription factors (22%)
Spliceosome genes (14%)
DNA methylation-related genes (44%)
Tumor suppressor genes (16%)
Cohesin-complex genes (13%)
Chromatin-modifying genes (30%)
Fig. 3.1 Frequency of driver mutations within eight distinct
functional categories in de novo acute myeloid leukemia, as
found by the Cancer Genome Atlas Research Network using
whole-genome sequencing.

cytoplasmic localization, which is readily detected by
immunohistochemistry. This cytoplasmic localization is
thought to be necessary for the leukemogenic activity of
NPM1c+, although details of the exact mechanism are still
not fully elucidated.
NPM1 mutations occur predominantly in the normal-
karyotype AML subgroup, with about 50% of this group
affected. A recent landmark study from the cancer genome
project used cytogenetics and gene sequencing to evaluate
driver mutations in 1540 AML patients. They examined pat-
terns of co-mutation and mutually exclusive mutations. In
this study NPM1 mutations were almost never found in
isolation. Commonly co-occurring mutations in DNMT3A,
FLT3, NRAS, and TET2 were found. Of these patterns of co-
mutation, the most extensively studied has been AML with
mutated NPM1 also carrying the FLT3-ITD mutation, which
coexists in ∼40% of NPM1-mutated cases.
The prognostic relevance of NPM1 mutations has been
clarified in several large studies, with clear evidence that
patients with mutated NPM1/wild-type FLT3 have an
improved prognosis over other normal-karyotype AML
patients. With conventional chemotherapy these individuals
have an event-free survival (EFS) and OS similar to those
with CBF leukemias. In most studies, the favorable out-
come predicted by NPM1 mutations is restricted to patients
who lack FLT3-ITD. Interestingly, initial studies suggest
that NPM1 mutations are very stable between diagnosis
and relapse, and thus are likely to constitute an initiating
rather than cooperating event in leukemogenesis. Subse-
quent acquisition of FLT3 or other cooperating mutations is
thought to be necessary to trigger overt AML development.
Given the stability of NPM1 mutations and their high fre-
quency, quantitative reverse transcription polymerase chain
reaction (RT-qPCR) assays for NPM1 mutations have been
developed for minimal residual disease (MRD) monitoring
and found, in research studies, to be predictive for relapse;
however, they are not yet used in routine clinical practice and
require validation.
Activated signaling
FLT3
FLT3 (FMS-like tyrosine kinase 3) encodes a membrane-
bound receptor tyrosine kinase with important roles in
hematopoietic stem/progenitor cell survival, differentiation,
and proliferation. FLT3 is constitutively activated by acquired
mutation in approximately 20–40% of cases of AML. In 20–
25% of cases, internal tandem duplications (ITDs) are noted
in the juxtamembrane domain, ranging in size from 3 to
400 nucleotides, with the resultant transcripts always being
in-frame and adding to the length of the protein. FLT3 is nor-
mally expressed in myeloid and lymphoid progenitor cells
The genetics of acute myeloid leukemias 41
and expression is lost as hematopoietic cells differentiate. In
comparison, ITDs of FLT3 result in constitutive activation of
the tyrosine kinase. This results in subsequent activation of
downstream signaling targets, including the phosphatidyli-
nositol 3-kinase (PI3K)/AKT, Ras/MAPK, and JAK2/STAT
pathways. Mutations in FLT3 also occur in the activation
loop of the tyrosine kinase domain (TKD) in about 5–10%
of AML cases. These also result in constitutive kinase acti-
vation, but there is increasing evidence that the downstream
signaling events are different in the setting of TKD versus
ITD, such that their clinical implications are not equivalent.
AML with FLT3-ITD constitutes an important subset of
AML cases that has a poor prognosis in most studies of chil-
dren and adults. The clinical implications of TKD muta-
tions remain controversial and larger studies are required
to define if these also confer a poor prognosis, though they
are probably not independently prognostic. The frequency
of FLT3 mutations varies considerably in different cytoge-
netic subgroups of AML, with a particularly high frequency
of 30–35% in normal-karyotype and t(15;17) AML cases. The
poor prognosis of FLT3-ITD in normal-karyotype AML is
conferred via increased relapse rates (RRs), with subsequent
reduced EFS and OS. The complete remission (CR) rate is
not affected, but several studies have shown that complete
remission duration (CRD) is significantly reduced in these
patients. While the significance of FLT3-ITD in t(15;17) cases
is unclear, it does not appear to be an independent prognostic
factor in APL patients.
The dosage of FLT3-ITD is also important. Most FLT3-
ITDs are found in the heterozygous state, although some
cases are noted to have homozygous FLT3-ITD with com-
plete loss of the wild-type allele. The homozygous FLT3
mutation is thought to arise as a consequence of mitotic
recombination resulting in segmental uniparental disomy.
Prognosis appears to be worse for these patients and for other
patients with high expression ratios of the mutated to wild-
type FLT3. Finally, the size of the duplicated segment may
also affect EFS, with a suggestion that longer ITDs result in
worsened outcomes.
The relatively high frequency of FLT3-ITD initially stimu-
lated interest for its use as a marker of MRD. However, studies
have shown that the mutational status of FLT3 may change
between diagnosis and relapse, with about 9% of patients
losing their FLT3-ITD status. The length of the FLT3-ITD
has also been noted to vary between diagnosis and relapse.
This discordance between diagnosis and relapse suggests
that FLT3 mutations are not sufficient for the development
of overt AML, but may be secondary, cooperating events
in leukemogenesis. This is further demonstrated by murine
bone marrow transplantation and “knock-in” models, in
which FLT3-ITD induces a myeloproliferative neoplasm
(MPN), but does not cause AML. The FLT3-ITD phenotype
is similar to that reported in the murine bone marrow
42 Molecular Hematology
transplantation assay for other constitutively activated tyro-
sine kinases associated with myeloproliferative phenotypes
in humans, including BCR-ABL, TEL-PDGFβR, TEL-ABL,
and TEL-JAK2. Taken together, these data indicate that
constitutive activation of tyrosine kinases is sufficient to
induce a myeloproliferative phenotype, but not AML.
c-KIT
The c-KIT gene, on chromosome 4q, encodes c-KIT, a
member of the type III receptor tyrosine kinase family,
the same family as FLT3. c-KIT’s ligand, stem cell factor
(SCF), promotes c-KIT dimerization and phosphorylation,
activating downstream signaling pathways important for
proliferation, differentiation, and migration of hematopoi-
etic stem cells. Gain-of-function mutations in c-KIT cause
ligand-independent constitutive kinase activation and are
found in 5–10% of all AML. However, a much higher
frequency of 12–45% is observed in the CBF leukemias.
Interestingly, c-KIT mutations are not seen in APL. Thus, a
significant functional redundancy is noted between c-KIT
and FLT3 mutations, with c-KIT mutations arising in CBF
leukemias in which FLT3 aberrations are absent, and FLT3
mutations arising in APL in which c-KIT mutations are
absent. The occurrence of c-KIT and FLT3 mutations within
the same patient is also very rare, suggesting mutual exclu-
sivity. Like FLT3-ITD in normal-karyotype AML, c-KIT
mutations, particularly in exons 8 and 17, confer a higher
risk of relapse and possibly inferior OS in CBF leukemia.
The National Comprehensive Cancer Network guidelines
now incorporate c-KIT, with c-KIT mutations moving CBF
leukemia from the favorable to intermediate risk category.
Due to its recognized adverse prognostic impact, molecular
testing for c-KIT mutations in CBF leukemia has become
standard of care in many major centers.
Myeloid transcription factors
RUNX1
RUNX1 is commonly dysregulated in AML, by translocations
in t(8;21) CBF leukemia, copy number variations, and point
mutations. Although point mutations occur in only 6–10% of
de novo AMLs, they are much more frequent in specific sub-
groups, arising in about 20% of AML with minimal differ-
entiation (FAB M0) and about 25% of AML associated with
MDS, including refractory anemia with excess blasts, AML
with multilineage dysplasia, and AML following MDS. These
point mutations are yet more frequent in therapy-related
MDS/AML and in radiation-associated MDS/AML (tested in
survivors of the atomic bomb in Hiroshima).
Further evidence of RUNX1’s involvement in the patho-
genesis of AML has come from the analysis of pedigrees with
an inherited predisposition to MDS/AML. Familial platelet
disorder with a propensity to develop myeloid malignancy
(FPD/AML syndrome) is an autosomal dominant disorder
caused by germline mutations in the RUNX1 gene. Individ-
uals with FPD/AML have a variable risk of developing overt
MDS/AML (20–60% in different pedigrees) and a large range
in age of presentation of MDS/AML (age 6–75 years). This
latency suggests that other cooperating driver mutations are
necessary to trigger overt MDS/AML. RUNX1 mutations are
frequently associated with trisomy 21 (Down syndrome) with
an increased incidence of AML, and AML with trisomy 13
and FLT-ITD. RUNX1 mutations are mutually exclusive of
NPM1 and CEBPA mutations.
Patients with RUNX1 mutations have significantly lower
CR rates, shorter EFS, and shorter OS than patients with
AML with wild-type RUNX1. The adverse prognostic signifi-
cance of RUNX1 mutations has been recognized by the WHO
and in the 2016 revision to the WHO classification a provi-
sional category of AML with mutated RUNX1 (for cases of
de novo AML not associated with MDS-related cytogenetic
abnormalities) has been added.
CEBPA
CEBPA (CCAAT/enhancer-binding protein α) is a transcrip-
tion factor that regulates genes involved in myeloid differenti-
ation, particularly by inducing granulocytic development of
bipotential myeloid progenitors. The protein consists of N-
terminal transactivating domains, a DNA-binding domain,
and a C-terminal leucine-zipper region (bZIP), necessary for
dimerization. The gene contains two translational start sites,
yielding a 42-kDa and a smaller 30-kDa isoform. Somatic
CEBPA mutations are noted in 5–14% of AML cases and
70% of these mutations are detected in patients with normal-
karyotype AML. Two classes of mutations commonly occur.
One involves N-terminal mutations, which prematurely stop
translation and abolish the production of the full-length
42-kDa isoform, while translation of the 30-kDa isoform
remains intact. The second class of mutations involves in-
frame mutations in the bZIP region and may impair DNA
binding, homodimerization, and heterodimerization. Most,
but not all, AML cases with mutated CEBPA are biallelic,
with a combination of an N-terminal and a bZIP mutation.
Others involve a homozygous CEPBA mutation and fewer
still are monoallelic, with a single heterozygous mutation and
retained expression of a wild-type allele.
Normal-karyotype AML with biallelic CEBPA mutations
represents a distinct molecular and clinical subtype with
favorable prognosis. CR rates are significantly higher and
EFS and OS are significantly longer for biallelic-mutated
CEBPA compared to wide-type CEBPA or monoallelic
CEBPA mutations. FLT3 and NPM1 mutations are common
in both wide-type and monoallelic CEBPA-mutated AML,
but are rare in biallelic variants.
Similar to RUNX1, CEBPA mutations have been observed
in very rare pedigrees with familial AML. The reported

pedigrees were all noted to exhibit germline mutations in
the N-terminal region of CEBPA. The nature and timing of
CEBPA mutations in familial AML provide further insight
into the sequence of molecular events in the development of
leukemia. This insight comes particularly from second muta-
tions targeting the bZIP region of the remaining wild-type
allele, which are commonly observed in familial and sporadic
CEBPA-associated AML. These appear necessary to cause
overt leukemia in familial CEBPA-associated AML. As with
FPD/AML, the age of presentation of disease is variable in
familial CEBPA-associated AML, ranging from 4 to 39 years.
However, the disease appears to have near-complete pene-
trance, suggesting that the combining two CEBPA mutations
are initiating rather than cooperating leukemogenic events.
Interestingly, genome sequencing has found that mutations
in GATA2 (another myeloid transcription factor) are com-
mon in cases of AML with biallelic CEBPA mutations (20–
40% of cases), but rare in monoallelic or wild-type CEBPA,
suggesting that GATA2 may act as a cooperating mutation
and contribute to leukemogenesis in a significant number
of AML with biallelic-mutated CEBPA cases. The prognostic
impact of this cooperating mutation is unclear and requires
further study.
Tumor suppressors
TP53
TP53 (Tumor protein 53), located on chromosome 17p,
encodes p53, a tumor suppressor protein that contains
DNA binding, transcriptional activation, and oligomeriza-
tion domains. p53 acts as a tumor suppressor through several
mechanisms including regulation of DNA repair, cell cycle
arrest, and initiation of apoptosis. TP53 mutations are asso-
ciated with a variety of human malignancies, are commonly
found in therapy-related AML (t-AML; 30–40% of cases) and
AML with complex cytogenetics (53–78%), and are associ-
ated with poor prognosis.
Until recently, it was believed that TP53 mutations in
t-AML were the result of DNA damage from cytotoxic
chemotherapy. However, recent studies using genome
sequencing of t-AML patients and chemotherapy-naı̈ve
adults have shown that TP53 mutations increase with
age, can occur several years (three to six years) before
the development of t-AML, and have been found prior
to the exposure of chemotherapy in patients who later
developed t-AML. This led to an alternative hypothesis for
the pathogenesis of t-AML, one in which chemotherapy
does not directly induce TP53 mutations, but instead rare,
age-related TP53 mutations occurring in hematopoietic
stem cells are relatively chemotherapy resistant and are
preferentially selected for after exposure to chemotherapy.
This hypothesis was supported by murine models. The
latency time between TP53 mutation and overt AML, as well
The genetics of acute myeloid leukemias 43
as the frequent co-occurrence of cytogenetic aberrations,
supports the necessity of additional cooperating mutations
for the development of AML.
Mutation of TP53, most commonly a missense mutation
in the DNA-binding domain, is the most frequent muta-
tion observed in AML with complex cytogenetics. It is
also often associated with specific copy number variants,
including −5/−5q, −7/−7q, and −17p. TP53 mutations are
mutually exclusive of FLT3, NPM1, and RUNX1 mutations.
TP53 mutations are independently associated with resistance
to chemotherapy, lower CR rates, and shorter OS. Their
association with complex cytogenetics is additive, with co-
occurrence having a particularly poor prognosis.
WT1
WT1 (Wilms tumor 1) gene, on chromosome 11p, encodes
a zinc-finger transcription factor. The precise role of WT1 in
normal and malignant hematopoiesis remains controversial,
but it has been implicated in the regulation of cell survival,
proliferation, and differentiation. Its role as a tumor suppres-
sor gene has been implicated as mutations in WT1, primarily
in exons 7 and 9, are observed in 10% of normal-karyotype
AMLs. However, WT1 is also overexpressed in various can-
cers, including the majority of AMLs, suggesting a role as
an oncogene. High expression of WT1 in AML made it an
attractive potential marker for MRD monitoring; however,
unlike NPM1, which shows stability during disease progres-
sion, WT1 mutational status has been found to change often
during disease progression and at relapse, decreasing its util-
ity for MRD monitoring. Mutations in WT1 act as an inde-
pendent negative prognostic indicator in AML by reducing
rates of CR, increasing rates of relapse, and shortening OS.
Mutations are noted in heterozygous and homozygous states
and show some association with FLT3-ITD, though larger
studies are required to confirm this.
Spliceosome genes
SRSF2
SRSF2 (Serine/Arginine-Rich Splicing Factor 2) encodes a
pre-mRNA splicing factor protein that is part of the spliceo-
some. SRSF2 is commonly mutated in MDS and CMML
and has been implicated in driving dysplasia and ineffec-
tive hematopoiesis. Co-mutations in ASXL1, RUNX1, and
IDH1/2 have been commonly described in SRSF2-mutated
MDS. Mutations in SRSF2 are most commonly heterozygous
missense mutations resulting in a gain of function. SRSF2
mutations are associated with older age at diagnosis and a
worse prognosis in MDS, with shorter OS and increased fre-
quency of progression to AML.
SRSF2 mutations in AML are activity being studied
and have been found in 6–7% of cases. They have been
44 Molecular Hematology
established to be highly specific for secondary AML (AML
arising from MDS or CMML). These mutations have been
found to occur early in leukemogenesis, are not sufficient
to induce overt leukemia as the sole mutation, and show
stability with disease evolution, often persisting in clonal
remissions. Studies have shown that SRSF2 driver mutations
in AML are associated with older age and inferior OS; how-
ever, data for their role as an independent prognostic marker
in AML have been conflicting and larger studies are needed.
Cohesin complex genes
Cohesin is a large multiprotein complex composed of
SMC1A, SMC3, RAD21, STAG1, and STAG2. The cohe-
sion complex’s main role is to maintain the polarity of sister
chromatids during mitosis, but it is also involved in double-
stranded DNA damage repair and transcriptional regula-
tion. Recurrent mutations, typically loss-of-function muta-
tions, in the cohesin complex have been found in 6–13% of
de novo and 20% of secondary AML cases, respectively. In
one large cohort study, 48% of cohesion-mutated AML cases
had a normal karyotype. Cohesin mutations have also been
reported in MDS and are especially prevalent in high-risk
MDS. Mutations of STAG2, located on chromosome X, were
found to be associated with shorter OS in MDS and were
more common in high-risk disease. Mutations of the cohesin
complex are mutually exclusive of one another and have
been found to co-occur with NPM1, RUNX1, and ASXL1
mutations.
Murine and in vitro models have shed light on the role of
cohesin mutations in leukemogenesis. In these studies, muta-
tion of cohesin in hematopoietic stem cells led to a differ-
entiation block with increased CD34+ progenitor cells and
a shift toward myeloid lineages, with cohesin knockdown
mice later developing features of MPNs. Cohesin mutations
appear to occur early in leukemogenesis and are insufficient
to cause overt leukemia as the sole mutation. The prognos-
tic significance of cohesin complex mutations in AML is
currently unknown, with contradictory reports that require
more investigation.
DNA methylation
DNMT3A
DNA methylation is an epigenetic process (a process that
alters gene expression without changing DNA sequences)
that is frequently altered in malignancies, including AML.
Cancer genomes often exhibit global DNA hypomethylation,
but have also been found to display DNA hypermethyla-
tion of the promotor regions of tumor suppressor genes. The
Cancer Genome Atlas Research Network found recurrent
mutations in genes involved in DNA methylation in 44% of
AML cases. Of these, mutation in DNA methyltransferase 3A
(DNMT3A) was the most common, seen in 26% of all AML
cases examined.
DNMT3A encodes a DNA methyltransferase enzyme that
functions independently of replication. Whole-genome and
whole-exome sequencing have found that DNMT3A is one of
the most frequently mutated genes in AML, occurring in 12–
35% of all cases, and is most often seen in normal-karyotype
AML. The most common mutations of DNMT3A involve
missense mutations at residue R882 near the carboxyl termi-
nus of the encoded protein. The precise biological and func-
tional effect of these mutations is not yet fully understood.
However, a recent study examining clonal relationships in
AML found that mutations in DNMT3A, along with muta-
tions in other genes that encode epigenetic modifiers, were
acquired early on in leukemogenesis, were usually present
in the founding clone, and were rarely the sole mutation.
DNMT3A is frequently co-mutated along with NPM1, FLT3,
and IDH1/2 (particularly IDH2). These data suggest that iso-
lated mutations in DNMT3A are not sufficient to result in
overt AML.
DNMT3A mutations have been found to confer a worse
prognosis, with reduced OS and relapse-free survival (RFS)
compared to wide-type DNMT3A in several studies, includ-
ing two meta-analyses with over 4000 patients. However,
other studies, stratified by ELN risk group, have found that
mutated DNMT3A is associated with worse OS and RFS only
for patients in the ELN Intermediate-1 risk group.
IDH1/2
IDH1 and IDH2 encode for isocitrate dehydrogenases
that catalyze the oxidative decarboxylation of isocitrate
to α-ketoglutarate. IDH1 is cytoplasmic, whereas IDH2 is
mitochondrial in location. IDH1/2 are mutated in several
different cancers, including AML, and share similar fea-
tures in these cases. IDH1/2 mutations are predominantly
somatic, are heterozygous, and occur at an early stage of
tumourigenesis. Mutations in IDH1 and IDH2 have been
found in 6–16% and 8–19% of AML, respectively. They
are most commonly found in normal-karyotype AML, are
mutually exclusive of one another, and are associated with
mutated NPM1. Epigenetic studies have found that IDH1/2
mutations result in global DNA hypermethylation and
impair hematopoietic differentiation.
The prognostic impact of IDH1 and IDH2 mutations is less
clear. Several studies have found no impact of IDH mutations
on OS, CR rate, or RFS for normal-karyotype AML; how-
ever, results from the literature are conflicting. Some studies
have found an adverse impact of particular IDH1 mutations,
whereas others found no difference. The findings for IDH2
mutations have ranged from a favorable impact with longer
OS for R140Q IDH2 mutation to an adverse prognosis with
worse OS and reduced CR rates for R172 IDH2 mutations,
and no significant impact in other cases.

TET2
TET2 (Tet Methylcytosine Dioxygenase 2) encodes a methyl-
cytosine dioxygenase, which is involved in myelopoiesis. In
2009, somatic mutations in TET2 were described for the first
time in various myeloid disorders, including MDS, MPNs,
and AML. Since that time the frequency of TET2 mutations
and its role in leukemogenesis have been examined. Mouse
models and in vitro studies have found that TET2 mutations,
which lead to loss of function, result in widespread DNA
hypermethylation of enhancer elements of tumor suppressor
genes and consequent decreased expression of these protec-
tive genes.
Mutated TET2 occurs in 12–17% and 24–32% of de novo
and secondary AML, respectively. Like the other genes
involved in DNA methylation (IDH1/2 and DNMT3A),
TET2 mutations occur early in leukemogenesis. TET2-
mutated AML is most frequently associated with a normal
karyotype and is often seen with co-mutations in NPM1 and
ASXL1. TET2 mutations and IDH1/2 mutations are mutually
exclusive.
As with IDH mutations, the independent prognostic sig-
nificance of TET2 mutations remains unclear, with conflict-
ing evidence. Some groups have reported inferior OS ver-
sus wide-type TET2 for young (age <60) patients with AML,
whereas others have reported no significant difference. Other
studies have found that mutated TET2 confers worse CR rates
and shorter EFS and OS only for normal-karyotype patients
in the ELN favorable risk category.
Chromatin modifiers
ASXL1
ASXL1 (Additional Sex Combs Like-1), located on chro-
mosome 20q, encodes a protein member of the Polycomb
group that is thought to disrupt chromatin in particular areas,
resulting in both epigenetic activation of certain genes and
repression of others. ASXL1 mutations have been found in
MPNs, MDS, CMML, and AML. The majority of mutations
are frameshift and nonsense mutations in exon 12. ASXL1
mutations are relatively specific for secondary AML, occur-
ring at a frequency of 6.5% of de novo AML but 30% of
secondary AML. ASXL1 mutations are also associated with
RUNX1, SRSF2, and IDH2 mutations as well as older age
at diagnosis, whereas they are almost mutually exclusive of
NPM1, FLT3-ITD, and DNMT3A mutations. Like other epi-
genetic modifiers, mutations in ASXL1 are believed to occur
early in the process of leukemogenesis and are not sufficient
alone to result in overt AML.
Mutated ASXL1 has been associated with a poor prognosis
in AML, with lower CR rates and shorter OS and EFS com-
pared to wide-type ASXL1. A recent study on the genomic
classification and prognosis of AML that enrolled over
1500 patients found that the negative prognostic impact of
The genetics of acute myeloid leukemias 45
mutated ASXL1 and SRSF2 was additive, with co-occurrence
having a worse prognosis than either mutation alone.
KMT2A (formerly MLL)
As with RUNX1, the KMT2A gene was first noted to be
aberrantly regulated in AML through translocations. Subse-
quently, KMT2A partial tandem duplications (PTDs) were
reported in 3–11% of normal-karyotype AML patients.
These mutations are more frequent in AML patients with
normal karyotype or with trisomy 11, and are often associ-
ated with FLT3 mutation. KMT2A-PTD was the first gene
mutation shown to negatively affect prognosis in normal-
karyotype AML patients, with most large studies confirming
these initial findings. Similar to the other genes involved in
DNA methylation and chromatin modification previously
discussed, KMT2A-PTDs are interesting because the pres-
ence of a heterozygous KMT2A-PTD has been associated
with silencing of the wild-type allele in AML blasts. The
mechanism for this silencing appears to involve epigenetic
modifications rather than direct mutational effects. Activity
of DOT1L, a histone methyltransferase protein, has recently
been found to cause aberrant hypermethylation in KMT2A-
rearranged AML, and appears to be vital in the pathogenesis
and maintenance of KMT2A-rearranged leukemia. This
suggests that epigenetic events may be sufficient to serve
as cooperating events in leukemogenesis. This mechanistic
understanding is also important in identifying a new avenue
for molecularly targeted therapy in AML through the use
of targeted therapy, hypomethylating agents, and histone
deacetylase inhibitors.
AML therapies and MRD monitoring
targeted by genetics
Advances in molecular technologies have led to the discov-
ery of numerous driver mutations in AML and have pro-
vided abundant new targets for directed therapies in AML
as well as for MRD monitoring. A comprehensive review of
targeted therapies and MRD monitoring is beyond the scope
of this chapter and readers are referred to recent review arti-
cles for further details. The best example of targeted therapy
in AML is the use of ATRA in APL, in which ATRA over-
comes the PML–RARα-induced differentiation block, lead-
ing to maturation of the malignant myeloblasts and promye-
locytes. Mutations in genes involved in DNA methylation,
including DNMT3A and TET2, are attractive targets for
hypomethylating agents such as azacitidine and decitabine.
Several other molecular targeted therapies are also being
investigated and are in various stages of development, rang-
ing from preclinical studies to phase III trials, and some with
recent FDA approval. These include first-generation broad-
spectrum kinase inhibitors (e.g. midostaurin and sorafenib),
46 Molecular Hematology
second-generation FLT3-specific inhibitors (Quizartinib,
Crenolanib, and Gilteritinib), IDH1 (Ivosidenib) and IDH2
inhibitors (Enasidenib), DOT1L inhibitors for KMT2A-
mutated AML (EPZ-5676), polyvalent WT1 peptide vaccine,
and several others. There is much hope that these targeted
therapies will improve survival for AML patients.
Assessment of treatment response currently relies on eval-
uation of bone marrow morphology and MRD detection by
flow cytometry. However, morphologic assessment is sub-
ject to limited sensitivity and interobserver variability, and
even the additional use of flow cytometry can be affected by
hemodilution. Though the presence of MRD by flow cytom-
etry is prognostically informative, studies have found that
about one-third of younger AML patients with no detectable
MRD by flow cytometry will still relapse. Due to these flaws,
there has been a push toward finding more sensitive mark-
ers of MRD. RT-qPCR is used for MRD monitoring in APL
and an undetectable PML-RARα transcript is a prerequi-
site for molecular remission. Other recurrent genetic aberra-
tions such as mutated NPM1, FLT3-ITD, and WT1 have been
investigated for RT-qPCR MRD monitoring. Of these, NPM1
seems the most promising given its high frequency and stabil-
ity in AML. RT-qPCR assays for MRD monitoring have also
been evaluated for CBF leukemia and found to be predictive
of relapse.
Summary
Recurrent cytogenetic abnormalities are frequent in AML
and many new molecular aberrations have also been discov-
ered that are undetectable by conventional karyotyping. The
advent of next-generation sequencing technologies has shed
much light on leukemogenesis and has uncovered numerous
driver mutations commonly found in AML. Some of these
driver mutations are initiating events, others are cooperating
mutations, and several have prognostic impact. Studies
suggest that at least two, and often more, driver mutations
are required for the development of overt AML. Various
patterns of mutual cooperativity as well as exclusivity have
been found among these mutations. The vast majority of
identified driver mutations fall within one of eight functional
categories, including NPM1, activated signaling, myeloid
transcription factors, tumor suppressor genes, spliceosome
genes, cohesion-complex genes, DNA methylation-related
genes, and chromatin-modifying genes.
Currently, AML patients who are at high risk of relapse
are recommended for allogeneic stem cell transplantation,
a treatment with a relatively high morbidity and mortality.
The decision of whether to proceed to transplantation is a
difficult one for both the patient and the physician. Many
of the newly discovered driver mutations have prognostic
implications and may help risk stratify patients to make treat-
ment decisions clearer. The recently published 2016 revision
to the WHO classification now incorporates several molecu-
lar categories, including AML with mutated NPM1 and AML
with biallelic mutations of CEBPA, and has a new provisional
entity of AML with mutated RUNX1. Molecular analysis for
mutations of CEBPA, NPM1, and FLT3 is now standard of
care in newly diagnosed AML and has been incorporated
in various risk stratification models. New sequencing tech-
nologies and the discovery of numerous driver mutations
have spurred research toward novel targeted therapies, use
of molecular markers for MRD monitoring, and develop-
ment of a fully genomic classification of AML. In the future,
it may be possible to use combinations of conventional and
molecularly targeted therapies, in selected clinical contexts,
to improve outcomes and/or to reduce toxicity. Given the
dismal prognosis for most patients with AML, these discov-
eries have brought new excitement to the field of leukemia
research and have identified new avenues for the prognosti-
cation, treatment, and monitoring of patients.
Further reading
Genetics of AML
Bullinger, L. and Valk, P.J. (2005). Gene expression profiling in acute
myeloid leukemia. J. Clin. Oncol. 23: 6296–6305.
Cancer Genome Atlas Research Network (2013). Genomic and epige-
nomic landscapes of adult de novo acute myeloid leukemia. N. Engl.
J. Med. 368: 2059–2074.
Dohner, H., Estey, E.H., Amadori, S. et al. (2010). Diagnosis and
management of acute myeloid leukemia in adults: recommenda-
tions from an international expert panel, on behalf of the European
LeukemiaNet. Blood 115: 453–474.
Jamani, K. and Owen, C. (2015). Update on recurrent genetic aber-
rations in acute myeloid leukemia. Int. J. Hematol. Oncol. 4: 179–
190.
Lindsley, R.C., Mar, B.G., Mazzola, E. et al. (2015). Acute myeloid
leukemia ontogeny is defined by distinct somatic mutations. Blood
125: 1367–1376.
Mrozek, K., Marcucci, G., Paschka, P. et al. (2007). Clinical relevance
of mutations and gene-expression changes in adult acute myeloid
leukemia with normal cytogenetics: are we ready for a prognostically
prioritized molecular classification? Blood 109: 431–448.
Papaemmanuil, E., Gerstung, M., Bullinger, L. et al. Genomic classifica-
tion and prognosis in acute myeloid leukemia. N. Engl. J. Med. 374:
2209–2221.
Raghavan, M., Lillington, D.M., Skoulakis, S. et al. (2005). Genome-
wide single nucleotide polymorphism analysis reveals frequent par-
tial uniparental disomy due to somatic recombination in acute
myeloid leukemias. Cancer Res. 65: 375–378.
Renneville, A., Roumier, C., Biggio, V. et al. (2008). Cooperating gene
mutations in acute myeloid leukemia: a review of the literature.
Leukemia 22: 915–931.

Core-binding factor leukemias
Duployez, N., Marceau-Renaut, A., Boissel, N. et al. (2016). Compre-
hensive mutational profiling of core binding factor acute myeloid
leukemia. Blood 127: 2451–2459.
Speck, N.A. and Gilliland, D.G. (2002). Core-binding factors in
haematopoiesis and leukemia. Nat. Rev. Cancer 2: 502–513.
NPM1
Falini, B., Mecucci, C., Tiacci, E. et al. (2005). Cytoplasmic nucleophos-
min in acute myelogenous leukemia with normal karyotype. N. Engl.
J. Med. 352: 254–266.
Schnittger, S., Schoch, C., Kern, W. et al. (2005). Nucleophosmin gene
mutations are predictors of favorable prognosis in acute myelogenous
leukemia with a normal karyotype. Blood 106: 3733–3739.
Activated signaling
Chen, W., Xie, H., Wang, H. et al. (2016). Prognostic significance of KIT
mutations in core-binding factor acute myeloid leukemia: a system-
atic review and meta-analysis. PLoS One 11: e0146614.
Gale, R.E., Green, C., Allen, C. et al. (2008). The impact of FLT3 internal
tandem duplication mutant level, number, size, and interaction with
NPM1 mutations in a large cohort of young adult patients with acute
myeloid leukemia. Blood 111: 2776–2784.
Thiede, C., Steudel, C., Mohr, B. et al. (2002). Analysis of FLT3-
activating mutations in 979 patients with acute myelogenous
leukemia: association with FAB subtypes and identification of sub-
groups with poor prognosis. Blood 99: 4326–4335.
Myeloid transcription factors
Dufour, A., Schneider, F., Metzeler, K.H. et al. (2010). Acute myeloid
leukemia with biallelic CEBPA gene mutations and normal karyotype
represents a distinct genetic entity associated with a favorable clinical
outcome. J. Clin. Oncol. 28: 570–577.
Osato, M. (2004). Point mutations in the RUNX1/AML1 gene: another
actor in RUNX leukemia. Oncogene 23: 4284–4296.
Wouters, B.J., Lowenberg, B., Erpelinck-Verschueren, C.A. et al. (2009).
Double CEBPA mutations, but not single CEBPA mutations, define a
subgroup of acute myeloid leukemia with a distinctive gene expres-
sion profile that is uniquely associated with a favorable outcome.
Blood 113: 3088–3091.
Tumor suppressors
Wong, T.N., Ramsingh, G., Young, A.L. et al. (2015). Role of TP53 muta-
tions in the origin and evolution of therapy-related acute myeloid
leukaemia. Nature 518: 552–555.
Spliceosome genes
Thol, F., Kade, S., Schlarmann, C. et al. (2012). Frequency and prognos-
tic impact of mutations in SRSF2, U2AF1, and ZRSR2 in patients with
myelodysplastic syndromes. Blood 119: 3578–3584.
The genetics of acute myeloid leukemias 47
Wu, S.J., Kuo, Y.Y., Hou, H.A. et al. (2012). The clinical implication of
SRSF2 mutation in patients with myelodysplastic syndrome and its
stability during disease evolution. Blood 120: 3106–3111.
Cohesin complex genes
Thota, S., Viny, A.D., Makishima, H. et al. (2014). Genetic alterations
of the cohesin complex genes in myeloid malignancies. Blood 124:
1790–1798.
DNA methylation
Figueroa, M.E., Abdel-Wahab, O., Lu, C. et al. (2010). Leukemic IDH1
and IDH2 mutations result in a hypermethylation phenotype, disrupt
TET2 function, and impair hematopoietic differentiation. Cancer Cell
18: 553–567.
Im, A.P., Sehgal, A.R., Carroll, M.P. et al. (2014). DNMT3A and IDH
mutations in acute myeloid leukemia and other myeloid malignan-
cies: associations with prognosis and potential treatment strategies.
Leukemia 28: 1774–1783.
Chromatin modifiers
Schnittger, S., Eder, C., Jeromin, S. et al. (2013). ASXL1 exon 12 muta-
tions are frequent in AML with intermediate risk karyotype and are
independently associated with an adverse outcome. Leukemia 27:
82–91.
AML therapies and MRD monitoring
targeted by genetics
Grimwade, D. and Freeman, S.D. (2014). Defining minimal residual
disease in acute myeloid leukemia: which platforms are ready for
“prime time”? Hematol. Am. Soc. Hematol. Educ. Program 2014: 222–
233.
Perl, A. and Carroll, M. (2007). Exploiting signal transduction pathways
in acute myelogenous leukemia. Curr. Treat. Options Oncol. 8: 265–
276.
Stein, E.M. and Tallman, M.S. (2016). Emerging therapeutic drugs for
AML. Blood 127: 71–78.
Familial AML
Michaud, J., Wu, F., Osato, M. et al. (2002). In vitro analyses of
known and novel RUNX1/AML1 mutations in dominant familial
platelet disorder with predisposition to acute myelogenous leukemia:
implications for mechanisms of pathogenesis. Blood 99: 1364–
1372.
Owen, C., Barnett, M., and Fitzgibbon, J. (2008). Familial myelodyspla-
sia and acute myeloid leukemia: a review. Br. J. Haematol. 140: 123–
132.
Song, W.-J., Sullivan, M.G., Legare, R.D. et al. (1999). Haploinsufficiency
of CBFA2 (AML1) causes familial thrombocytopenia with propensity
to develop acute myelogenous leukemia (FPD/AML). Nat. Genet. 23:
166–175.
Chapter 4 Molecular diagnostics and risk
assessment in myeloid malignancies
Christian Scharenberg1,2 & Torsten Haferlach3
1
Department of Hematology, Skaraborgs Hospital, Skövde, Sweden
2
Department of Cell and Molecular Biology, Karolinska Institute, Stockholm, Sweden
3 MLL Munich Leukemia Laboratory, Munich, Germany

Introduction, 49 Clinical implications, 52

General methodology of cytogenetic analysis in hematological Treatment selection, 52
malignancies, 50 Therapy-related chromosomal aberrations in secondary AML, 54
Methods of detection, 50 Recurrently mutated genes in MDS, 54
Genetic and clinical consequences of chromosomal translocation, 51 Recurrently mutated genes in AML, 56
Karyotypic evolution, 52 Further reading, 57

20% in patients over the age of 65 and 90, respectively. To

Introduction
The recent progress in high-throughput genome sequencing
technologies has enabled the discovery of genetic lesions that
drive the pathogenesis of the majority of human cancers. This
progress allowed for the identification of recurring chromo-
somal abnormalities and gene alterations that has tremen-
dously advanced our understanding of the pathobiology of
myeloid neoplasms.
While previously considered related albeit distinct enti-
ties, acute myeloid leukemia (AML) and myelodysplastic
syndromes (MDS) are now thought of as part of a dis-
ease continuum with a recognizable pre-malignant state. In
addition to the usefulness of molecular analyses in diag-
nosis and prognosis, the discovery of such mutations has
offered genetic tools to study clonal diversity and disease
evolution.
Clonal hematopoiesis
While cancer is now known to result from the stepwise accu-
mulation of somatic mutations, it has remained difficult to
define the initial stages preceding the development of overt
hematological malignancies.
Using existing datasets of exome sequencing on periph-
eral blood samples from more than 30 000 patients with-
out known hematological cancers, three studies found evi-
dence for somatic mutations in blood DNA. Despite being
rare (less than 1%) in people under 40 years of age, somatic
mutations increase with age, with frequencies up to 10% and
Molecular Hematology, 4th Edition. Edited by Drew Provan and John G. Gribben.
© 2020 John Wiley & Sons Ltd. Published 2020 by John Wiley & Sons Ltd.
describe this phenomenon, the acronym CHIP for clonal
hematopoiesis of indeterminate potential has been adopted.
Individuals in whom mutations were detected had usually
only a single mutation. Remarkably, mutations in only three
genes, DNMT3A, TET2, and ASXL1, explained the majority
of driver mutations sustaining clonal hematopoiesis.
The initial hit occurs in a single stem cell, and it is believed
that the subsequent expansion of the formed clone is quite
substantial (median clone size measured in the peripheral
blood was approximately 18%). In contrast to previous stud-
ies that could detect certain chromosomal translocations
only transiently in healthy individuals, CHIP appears to per-
sist for years with no indication of spontaneous resolution.
Clonal hematopoiesis seems to be a strong predictor for
the development of hematological malignancy (hazard ratio
of ∼12), confirming that the initial mutation constitutes a
pre-malignant state. Direct evidence for the evolution from
clonal hematopoiesis to hematologic malignancy was possi-
ble in a small number of patients who developed AML, as the
leukemic samples were found to contain the somatic muta-
tions at high allele fractions. Therefore, clonal hematopoiesis
is now viewed as part of the natural evolution from a pre-
malignant state to overt leukemia.
Myelodysplastic syndromes
The MDS comprise a series of hematologic conditions of
abnormal cellular maturation leading to chronic cytopenias
(i.e. anemia, neutropenia, thrombocytopenia). As a result,
patients with MDS are at risk for symptomatic anemia, infec-
tion, and bleeding, as well as progression to AML, which is
often refractory to standard treatment.

49
50 Molecular Hematology

The complexities of the pathobiology of MDS are begin- understanding of the pathobiology of AML. In addition
ning to be elucidated. The development of MDS occurs via a to having considerable prognostic significance for patients
series of genetic changes in a hematopoietic stem cell (HSC). with AML, certain cytogenetic abnormalities identified by
These changes alter normal hematopoietic growth and differ- karyotype analysis affect the planning of treatment. Abnor-
entiation, resulting in an accumulation of abnormal, imma- malities in certain genes (e.g. mutations in FLT3, NPM1,
ture myeloid cells in the bone marrow, leading to impairment KIT) as well as gene expression profiles confer prognostic
of normal hematopoiesis. Advances in the identification of significance in adult patients with AML.
recurring chromosomal abnormalities and gene alterations
have provided insight into the pathobiology of MDS.
It is well established that specific cytogenetic abnormal-
General methodology of cytogenetic
ities identified by karyotype analysis or fluorescence in situ
analysis in hematological
hybridization (FISH) analysis bear prognostic significance
malignancies
for patients with primary MDS and affect the planning
of treatment. While certain gene mutations also confer
The malignant cells in many patients with leukemia, lym-
independent prognostic significance in adult patients with
phoma, or another malignant hematologic disease have
MDS, it is not yet clear how to incorporate these changes into
acquired clonal chromosomal abnormalities. Some spe-
treatment planning. Even those patients without obvious
cific cytogenetic abnormalities are closely, and sometimes
abnormalities detected by karyotypic analysis, FISH, or
uniquely, associated with morphologically and clinically
gene mutation analyses likely have acquired copy number
distinct subsets of leukemia or lymphoma, as well as with
alterations or abnormalities in gene expression profiles,
their prognosis.
which may help to identify genes with important roles for
Compared to other malignancies and given the ease of
the pathogenesis of MDS.
obtaining and processing bone marrow or peripheral blood
The development of secondary, therapy-related MDS or
samples, significantly more information is available about
AML is associated with characteristic chromosomal abnor-
the relationship between cytogenetic abnormalities and the
malities in patients who developed MDS or AML after radi-
pathogenesis and natural history of the leukemias.
ation therapy and/or chemotherapy for an earlier disease, for
The detection of clonal cytogenetic abnormalities allows
example lymphoma or a solid tumor, or even non-malignant
the differentiation between benign reactive lymphoid or
disorders, such as rheumatoid arthritis, or following organ
myeloid hyperplasia and a monoclonal malignant prolifer-
transplantation.
ation, and in some cases the establishment of specific diag-
noses, for example the Philadelphia chromosome in chronic
Acute myeloid leukemia myeloid leukemia (CML). In addition, the study of cytoge-
netic abnormalities has provided insight into disease patho-
At first glance the development of AML resembles that of
genesis (by identifying genes controlling cell growth and
MDS. However, after a series of genetic changes generate
leukemogenesis), prognosis (clonal evolution can signify a
pre-AML stem cells, the final transformative event leading to
more aggressive disease course), and treatment planning
overt development of AML occurs in a hematopoietic pre-
(certain chromosomal changes predict for response to spe-
cursor rather than in HSCs. By altering normal hematopoi-
cific therapies, or for the choice of targeted therapy).
etic growth and differentiation, large numbers of abnormal,
immature myeloid cells accumulate in the bone marrow and
peripheral blood. While capable of dividing and proliferat-
ing, these cells cannot differentiate into mature hematopoi- Methods of detection
etic cells.
Not unlike other malignant diseases, the genetic alter- The following methods are used for the detection of cytoge-
ations in AML include mutation of oncogenes as well as the netic aberrations: conventional metaphase cytogenetic anal-
loss of tumor suppressor genes. Compared to solid tumors, ysis, FISH analysis, reverse transcription-polymerase chain
however, many hematologic malignancies are associated reaction (RT-PCR), microarray-based genomic copy number
with a single characteristic cytogenetic abnormality, for analysis, and DNA or RNA sequencing.
instance the Philadelphia chromosome t(9;22) in chronic Though technically difficult, the entire tumor genome can
myeloid leukemia and t(15;17) in acute promyelocytic be examined by conventional metaphase cytogenetic analy-
leukemia (APL). sis. This method is therefore of particular importance in the
The identification of recurring chromosomal abnor- evaluation of hematologic malignancies with many potential
malities and translocations has largely advanced our genetic abnormalities (e.g. AML). Conventional metaphase

cytogenetic analysis is performed by culturing the tumor
sample in vitro to stimulate cellular division. Once the cells
divide readily, mitotic division is arrested in metaphase by
adding a chemical (e.g. colchicin). Subsequently, chromoso-
mal banding is performed using various enzymes and histo-
logical stains.
FISH analysis relies on the ability of single-stranded DNA
to anneal to complementary DNA. The technique employs
fluorescently labeled DNA probes that are hybridized to
either metaphase chromosomes or interphase nuclei affixed
to a glass slide. This method can be employed with bone
marrow or peripheral blood smears, cytospin slides, as well
as fixed and sectioned tissue. It is both rapid and sensitive,
allowing the detection of recurring numerical and structural
abnormalities. FISH is most useful when certain abnormali-
ties are suspected or seem likely based on other investigations
(e.g. based on distinctive morphology). A significant short-
coming is that additional abnormalities or multiple clones
will not be detected by this method alone. Hence, FISH serves
as a complement to conventional cytogenetic analysis in the
majority of clinical cases.
PCR-based methods markedly increase the sensitivity with
which chromosomal aberrations can be detected. Transloca-
tions resulting in fusion genes are ideally suited for detection
by RT-PCR, a technique where messenger RNA (mRNA) is
first copied into complementary DNA (cDNA) and subse-
quently the fusion gene is amplified by PCR using specific
primers from each gene. In certain cases, DNA sequencing
is then used to confirm the identity of the amplified prod-
uct. Given the high sensitivity of RT-PCR (down to detection
of one abnormal in a million normal cells), this method is
ideally suitable for the detection of minimal residual disease
(MRD) in patients who are in clinical remission.
Genomic microarray technology using single-nucleotide
polymorphisms (SNPs) allows for the identification of dis-
ease susceptibility loci in genome-wide association studies
and the identification of other acquired abnormalities, such
as genomic imbalances (e.g. duplications or cryptic deletions)
and loss of heterozygosity (LOH), which occurs without
concomitant changes in the gene copy number via somatic
mitotic recombinations known as uniparental disomy.
Next-generation sequencing (NGS) techniques are increas-
ingly being incorporated into the routine diagnostic eval-
uation of leukemia and lymphoma. Highly sensitive at the
single-nucleotide level, they allow the detection of recur-
ring gene fusion transcripts and gene mutations. The anal-
ysis of DNA using gene panels of targeted exon mutations
allows the concurrent detection of multiple genes in a panel
reaction (targeted sequencing). RNA sequencing (performed
using either targeted panels or whole RNA) enables detection
of recurring leukemia-specific gene fusion transcripts, and
can identify novel gene fusions that define specific leukemia
Molecular diagnostics and risk assessment in myeloid malignancies 51
entities (e.g. multiple tyrosine kinase gene fusions in Ph-like
B-ALL).
While the detection of specific chromosome translocations
and gene fusions allows a precise diagnosis of leukemia, anal-
ysis of recurrent somatic mutations in AML and MDS adds
important genetic information, offering better risk stratifica-
tion and selection of targeted therapies.
Importantly, each technique has its advantages and dis-
advantages – while molecular methods, such as RT-PCR or
NGS techniques, assess the entire cell population, FISH pro-
vides analysis at the single-cell level. A comprehensive picture
of the genetic landscape of leukemia or lymphoma emerges
from the combination of conventional cytogenetic analysis,
FISH tests, microarray, and NGS studies, which enables per-
sonalized genetic-based treatment.
Genetic and clinical consequences of
chromosomal translocation
An important role in the process of malignant transforma-
tion is that of alterations in the expression of genes that lie at
the breakpoints of the chromosomal translocations or within
functional domains of the encoded proteins. The genes with
a role in transformation can be grouped into the follow-
ing functional classes: (i) tyrosine or serine protein kinases;
(ii) cell surface receptors and growth factors; (iii) proteins
that regulate transcription; and (iv) regulators of apoptosis.
Among these, genes that encode proteins regulating tran-
scription are the most commonly involved. Gene transcrip-
tion is the process by which the information in genomic DNA
is transcribed to RNA. Transcription factors are central to
this process, recognizing and binding to target sequences in
the regulatory elements of genes or to other DNA-binding
proteins in a tissue-specific fashion. Chromosomal translo-
cations can alter gene function by aberrant regulation of gene
expression or via expression of a novel fusion protein.
The latter result from the juxtaposition of coding
sequences from two genes that are normally located on
different chromosomes. This produces a fusion mRNA and a
chimeric protein with altered gene function. There are three
well-characterized fusion proteins: BCR-ABL1, PML-RARA,
and RUNX-RUNXT1.
As a result of the t(9;22) in CML, a shortened chromo-
some 22 (i.e. Philadelphia chromosome) leads to the BCR-
ABL1 fusion protein. BRC-ABL1 contains the entire catalytic
domain from the tyrosine kinase ABL1, leading to constitu-
tively increased tyrosine kinase activity and explaining the
abnormal cellular proliferation observed in CML. Imatinib
and other tyrosine kinase inhibitors specifically target this
tyrosine kinase activity and exhibit remarkable activity in all
phases of CML.
52 Molecular Hematology
In APL, another fusion protein is formed by the t(15;17)
producing a fusion protein involving the promyelo-
cytic leukemia and retinoic acid receptor-alpha proteins
(PML-RARA). The APL fusion protein is responsible for
arresting the growth of promyelocytes, presumably by
aberrant repression of retinoic acid receptor-mediated
gene transcription through histone deacetylase-dependent
chromatin remodeling. Treatment with one of two com-
pounds, all-trans-retinoic acid (ATRA) or arsenic trioxide,
can bypass the transcriptional repression and activate genes
leading to the terminal differentiation of promyelocytes.
AML with t(8;21) constitutes a clinically distinct sub-
set of AML with a characteristic morphology, portend-
ing a favorable prognosis. Molecularly, the t(8;21) forms
a chimeric protein RUNX1-RUNX1T1. The gene RUNX1
encodes one subunit of a heterodimeric transcription fac-
tor termed core-binding factor (CBF), shown to be essential
for hematopoiesis. The oncogenic potential of the RUNX1-
RUNX1T1 fusion protein is believed to be due to aber-
rant recruitment of nuclear transcriptional co-repressor
complexes, leading to transcriptional repression of normal
RUNX1 target genes.
All of the fusion proteins mentioned are specific for the
disease. Thus, they can be used for the initial diagnosis as well
as for detecting MRD.
Karyotypic evolution
Additional cytogenetic abnormalities may help to detect
transformation to a more advanced disease stage or to a more
aggressive malignancy. As evidenced by subclones on cyto-
genetic analysis, clonal heterogeneity and evolution are fre-
quent in AML and constitute an independent adverse prog-
nostic marker in cytogenetically adverse-risk AML.
Clinical implications
Both the clinical course as well as the likelihood of respond-
ing to particular treatments can be predicted by specific
genetic abnormalities (e.g. retinoic acid or arsenic trioxide
in APL). The prognostic information derived from cyto-
genetic analysis is often independent of other clinical fea-
tures. Patients who have less favorable clinical or cytogenetic
characteristics may be better treated with more intensive or
investigational therapies, while those with more favorable
prognostic features may be treated with standard regimens.
Moreover, the disappearance of a chromosomal abnormality
present at diagnosis signifies complete remission following
treatment, while its reappearance invariably indicates relapse
of the disease.
Treatment selection
The optimal timing of hematopoietic cell transplantation
(HCT) in the treatment of AML and ALL is still unclear.
Cytogenetic analysis before commencement of treatment can
guide the choice among post-remission therapies that differ
in effectiveness, acute and chronic morbidity, as well as cost.
The majority of AML patients with favorable cytogenetic fea-
tures – such as t(8;21) and inv(16) or t(16;16) – can be cured
solely with intensive consolidation chemotherapy. Patients
who relapse can be salvaged with HCT in early relapse or
second remission. On the other hand, patients with a chro-
mosome abnormality that portends a bad prognosis – for
instance complex karyotype (≥3 abnormalities) in AML –
are rarely cured with standard therapy and can have drug-
resistant disease. These patients may benefit from transplan-
tation in first remission or investigational drugs.
Recurrent cytogenetic abnormalities
in MDS
A diagnosis of MDS is usually made by evaluation of the
bone marrow and peripheral smear in the appropriate clin-
ical context. In patients with otherwise unexplained refrac-
tory cytopenia and no morphological evidence of dysplasia,
certain cytogenetic abnormalities are sufficient for the diag-
nosis of MDS.
At the time of diagnosis, approximately half of patients
with MDS have clonal chromosomal abnormalities. These
chromosomal changes can manifest as structural abnormal-
ities involving one chromosome (e.g. inversion, interstitial
deletion), numerical changes (e.g. monosomy or trisomy), or
a balanced translocation between two chromosomes. About
15% of patients have multiple abnormalities and thus exhibit
complex karyotypes (≥3 or more abnormalities). During the
course of the disease, additional chromosomal aberrations
may develop on top of previous ones or emerge in patients
with no previous karyotypic abnormality. These changes are
invariably associated with poor prognosis and progression to
acute leukemia. In contrast to AML, none of the chromoso-
mal abnormalities is specific for MDS and they are also found
in other myeloid diseases.
The karyotypic abnormalities that are most commonly
found in MDS are del(5q), del(7q) or monosomy 7 (−7),
trisomy 8 (+8), and del(20q), as well as loss of the Y chromo-
some. Although the same cytogenetic lesions also occur in
de novo AML, the balanced translocations typically found in
AML are very uncommon in MDS. Rather, the chromosomal
abnormalities in MDS are usually unbalanced, indicating
gain or loss of chromosomal material. Cytogenetic abnormal-
ities are even more prevalent in therapy-related MDS/AML,
as up to 70% of cases have chromosomal abnormalities,

particularly deletions of chromosomes 5 and 7 or chromo-
some 17p. Deletion of the long arm of chromosome 5 with or
without additional karyotypic abnormalities is not only the
most frequent chromosomal aberration in MDS, it is the only
one constituting its own entity in the WHO classification.
The term “5q– syndrome” refers to a distinctive type
of MDS that is defined by isolated del(5q) or del(5q) plus
one other abnormality, with the exception of −7/del(7q). A
characteristic hallmark on morphology examination is the
finding of megakaryocytes with hypolobated nuclei. The
deletions on chromosome 5 are large interstitial deletions
resulting in the retention of only one normal allele of all genes
contained within the deleted segment – this is referred to as
a haploinsufficient state. Numerous genes on the long arm of
chromosome 5 have been implicated in MDS pathogenesis
and its response to specific therapies.
MDS cases with del(5q) as the sole chromosomal abnor-
mality carry a relatively good prognosis and frequently
respond to treatment with lenalidomide. In contrast, mono-
somy 5 or del(5q) with more than one additional chromoso-
mal change is associated with advanced MDS and portends a
worse prognosis.
With the notable exception of MDS with isolated del(5q),
other chromosomal aberrations in MDS have not correlated
with specific morphological or clinical entities. However,
they do play a substantial role in pathogenesis and progno-
sis and have a major impact in the clinical management of
patients.
Certain cytogenetic abnormalities in MDS are used in the
prediction of survival and progression to AML. The presence
or absence of specific chromosomal aberrations is incorpo-
rated into several prognostic scoring systems for MDS, such
as the original and revised International Prognostic Scoring
Systems (IPSS and IPSS-R), the World Health Organization
(WHO) Prognostic Scoring System, and the MD Anderson
Cancer Center MDS model. The most widely used system,
IPSS-R, is shown in Figure 4.1.
Importantly, cytogenetic analysis cannot predict individ-
ual MDS patient outcome due to the fact that many patients
succumb to persistent pancytopenia, irrespective of progres-
sion to AML.
Recurrent cytogenetic aberrations in AML
A key component in the evaluation of all patients with newly
diagnosed or suspected AML is the cytogenetic analysis of
metaphase cells. In some cases, specific cytogenetic abnor-
malities correspond to morphologically and clinically dis-
tinct subsets of the disease. Because of this, the WHO classi-
fication of tumors of the hematopoietic and lymphoid tissues
incorporates genetic findings in addition to morphological,
immunophenotypic, and clinical features to define distinct
subtypes of AML.
Molecular diagnostics and risk assessment in myeloid malignancies 53
About 50–60% of patients with de novo AML have
abnormal karyotypes if standard banding techniques are
used. The most common abnormalities seen on karyotype
are t(15;17)(q24.1;q21.1), trisomy 8, t(8;21)(q22;q22.1),
rearrangements of 11q, and inv(16)(p13.1q22)/t(16;16)
(p13.1;q22). Regardless of blast count, AML is diagnosed if
any of the following cytogenetic abnormalities are present:
r AML with t(8;21)(q22;q22.1); RUNX1-RUNX1T1 (previ-
ously AML1-ETO). Found in approximately 7% of adults
with newly diagnosed AML, it is the most frequent abnor-
mality in children with AML. This cytogenetic abnormal-
ity has a distinct morphological phenotype and heralds a
favorable prognosis in adults, but a poor prognosis in chil-
dren.
r APL with t(15;17)(q24.1;q21.2); PML-RARA. Highly spe-
cific for APL, the translocation results in a novel PML-
RARA fusion protein that is thought to arrest differenti-
ation at the promyelocyte stage. APL constitutes a medi-
cal emergency carrying a high rate of early mortality, often
due to hemorrhage from disseminated intravascular coag-
ulation. Treatment should be started as soon as the diag-
nosis is suggested by cytological criteria. Cytogenetic and
molecular genetic analysis can then confirm the diagno-
sis. If prompt treatment with ATRA or arsenic trioxide is
started, these patients have an excellent prognosis.
r AML with inv(16)(p13.1q22) or t(16;16)(p13.1;q22);
CBFB-MYH11. Representing approximately 5% of newly
diagnosed AML, its classical bone marrow morphology
typically shows monocytic and granulocytic differenti-
ation with abnormal eosinophils. Patients with inv(16)
generally respond well to intensive chemotherapy.
While these cytogenetic aberrations constitute a diagno-
sis of AML irrespective of blast count, there are additional
subtypes of AML where certain genetic abnormalities bear
prognostic significance. However, these require bone marrow
blast counts of greater than 20% for diagnosis of AML:
r AML with t(9;11)(p21.3;q23.3); KMT2A-MLLT3. By far
the most promiscuous translocation partner in AML
involves 11q23.3. More than 100 different recurring rear-
rangements involving 11q23.3 have been described in
AML, especially among the monoblastic and myelomono-
cytic types. Of these, t(9;11) is the most common abnor-
mality and involves the KMT2A gene and MLLT3 (AF9).
Patients generally have an intermediate response to stan-
dard therapy.
r AML with t(6;9)(p23;q34.1); DEK-NUP214. Seen in about
1% of patients with newly diagnosed AML, this is mor-
phologically associated with basophilia, pancytopenia, and
dysplasia and frequently associated with FLT3-ITD muta-
tions. Patients generally have a poor response to standard
therapy, but it is currently unclear whether the association
with FLT3-ITD or the lesion itself is responsible for this.
54 Molecular Hematology

Calculation of risk score

Cytogenetic risk group

Very good 0 del(11q), -Y

Good 1 normal, del(5q) alone or with one other anomaly, del(12p), del(20)
Intermediate 2 +8, del(7q), i(17q), +19, +21, other abnormalities in two or more independent clones
Poor 3 der(3q), –7, double abnormality with –7 or del(7q), complex with 3 abnormalities
Very poor 4 complex with >3 abnormalities

Bone marrow blast percentage

≤2% 0
2–5% 1
5–10% 2 Assigning IPSS-R risk group
>10% 3
Time until
Median 25%
%of survival develop
Total score IPSS-R risk group patients (years) AML (years)
Hemoglobin (g/dL)
0–1.5 Very low 19 8.8 Not reached
≥10 0 1.5–3 Low 38 5.3 10.8
8–10 1 3–4.5 Intermediate 20 3 3.2
<8 1.5 4.5–6 High 13 1.6 1.4
>6 Very high 10 0.8 0.7
Platelet count (G/L)

≥100 0
50–100 0.5
<50 1

Absolute neutrophil count (G/L)

≥0.8 0
<0.8 0.5

Fig. 4.1 International Prognostic Scoring System Revised (IPSS-R). Karyotypic abnormalities, bone marrow blast percentage, and severity of
peripheral blood cytopenias are scored and used to assign myelodysplastic syndrome (MDS) patients to one of five risk groups exhibiting significant
differences in median survival and probability of developing acute myeloid leukemia (AML).
Source: Based on Greenberg, P.L., Tuechler, H., Schanz, J. et al. (2012). Revised international prognostic scoring system for myelodysplastic
syndromes. Blood 120(12): 2454–2465.

r AML with inv(3)(q21.3q26.2) or t(3;3)(q21.3;q26.2);
MECOM/EVI1. Accounting for approximately 1% of AML
cases, inv(3) aberrations are associated with increased
atypical megakaryocytes in the bone marrow and a poor
response to therapy. As a consequence of the cytogenetic
abnormality, the MECOM gene is overexpressed, likely
explaining the peripheral blood thrombocytosis and bone
marrow morphology.
Therapy-related chromosomal
aberrations in secondary AML
The most common subtype of therapy-related MDS/AML
constitutes patients who were being treated with alkylating
agents. These patients often have abnormalities in chromo-
somes 5 and/or 7.
Patients being treated with DNA topoisomerase II
inhibitors (e.g. etoposide, doxorubicin, mitoxantrone,
epirubicin) account for the second most frequent subtype
of therapy-related AML. Frequently, balanced translocations
involving the KMT2A gene at 11q23.3, or the RUNX1 gene
at 21q22.1, are found in these AML cases.
Recurrently mutated genes in MDS

Patients who develop therapy-related secondary MDS or The recent advances in less expensive high-throughput
AML after chemotherapy and/or radiation therapy for an genome sequencing technologies have enabled the discovery
earlier disease have also been found to carry characteristic of genetic lesions that drive the pathogenesis of the majority
chromosomal abnormalities. of human cancers, including the MDS.

Based on panels of genes that were previously found to
be mutated in other myeloid diseases, a number of studies
have described the mutational landscape of somatic point
mutations in MDS. According to current data, up to 90% of
patients bear mutations in at least one recurrently mutated
gene, and it is believed that the remaining 10% of patients
have mutations in genes yet to be identified.
MDS is the result of the sequential accrual of somatic
mutations in HSCs. Based on calculations, these stem cells
amass approximately 1.3 mutations in exons every decade,
but the majority of these are innocent bystanders with no
functional alteration. However, this calculation presumes
that the rate of mutation is constant – in reality, the process of
mutational accrual could be dynamic, as one mutation might
affect the acquisition of subsequent mutations.
The spectrum of recurrent somatic mutations in MDS and
other myeloid diseases can be grouped according to their bio-
logical pathways, as in the following.
Mutations in splicing factors
A hallmark of eukaryotes, the alternative splicing of pre-
mRNA enables the generation of different proteins from the
same gene. Interestingly, one of the hallmarks of cancer is
aberrant alternative splicing. Recurrent mutations in genes
that are part of the spliceosome have been found in up to 60%
of cases of MDS.
The mutations in SF3B1, U2AF1, and SRSF2 are invariably
found within defined hotspots of these genes, leading to
loss of function and altering the operation of the splicing
machinery. Mutations within the splicing componentry
most commonly affect SF3B1, occurring in up to 90% of
MDS patients with ring sideroblasts (RARS). In this context,
SF3B1 mutations are associated with a good prognosis. It
is believed that the ring sideroblast phenotype is the result
of aberrant splicing of genes central to iron homeostasis.
The second most frequent splicing factor mutation in MDS
affects the SRSF2 gene; this gene is commonly mutated in the
overlap syndromes of MDS/myeloproliferative neoplasms
(MPN). Mutations in SRSF2 were demonstrated to alter
binding of the protein to splice enhancers, with resulting
mis-splicing of several important downstream genes, one
of which is EZH2. In 10–15% of MDS cases mutations are
found in U2AF1. Resulting loss of function is reported to
lead to increased exon skipping. SRSF2 and U2AF1 have
not been associated with a similar favorable prognosis to
SF3B1. Despite these findings, no single mis-spliced isoform
has hitherto been implicated in the pathogenesis of MDS.
Moreover, studies using mouse models of splicing genes
have not yielded similar patterns of altered splicing, either
between mice and humans or between individual mice. The
role that these mutations play in MDS pathobiology is per-
haps unrelated to their known function within the splicing
machinery.
Molecular diagnostics and risk assessment in myeloid malignancies 55
Epigenetic regulators
Another set of genes found to be recurrently mutated in MDS
are genes that are involved in epigenetic regulation. Epige-
netic regulation is the result of post-translational modifica-
tion of either histones or the DNA itself.
Histone modification
Enzymes can covalently modify the tails of histones, leading
to changes in the structure of chromatin and affecting the
binding of protein involved in gene regulation. Among the
central regulators of transcriptional silencing in development
are the Polycomb repressive complexes (PRCs), composed
of two separate protein complexes, PRC1 and PRC2. Their
action on histones H2A and H3 results in the compaction
of chromatin. Genes that are part of both protein complexes
have been found to be mutated in MDS. Mutation in two
components of PRC1 – that is, BCOR and BCORL1 – has been
shown to be recurrently mutated in 5% of MDS cases and to
portend an unfavorable prognosis. EZH2 is mutated in about
5% of MDS patients; the resulting protein is part of the cat-
alytic subunit of PRC2 and studies in transgenic mice have
confirmed that loss of Ezh2 results in MDS.
While in itself ASXL1 is not part of the PRC, it has been
shown to indirectly affect PRC2, leading to deubiquitylation
of histone H2A. Approximately 20% of MDS patients carry
recurrent mutations in ASXL1. Mutations in BCOR, EZH2,
and ASXL1 have been shown to disrupt normal hematopoi-
etic differentiation, likely explaining their putative role in
dysplasia and cytopenia.
DNA methylation
The regulatory regions of the DNA can either be accessible
or inaccessible. Which state is favored is largely controlled
by the methylation of cytosines in repetitive CpG elements.
DNA methyltransferases (DNMTs) add methyl groups,
while the TET family of proteins demethylate in a multistep
process. Mutations in DNMT3A leading to loss of function
of the protein are found in approximately 15% of MDS cases.
TET2 is one of the most frequently mutated genes, found in
up to 30% of MDS cases. The state of hypermethylation of
cytosines at enhancer sites is believed to repress several genes
with important roles in myeloid differentiation. Of note is
that DNMT3A and TET2 are both expressed in HSCs – in
a complementary, interconnected, yin-and-yang kind of
fashion – having essential roles in self-renewal and myeloid
differentiation.
An initially puzzling observation was how mutations
in enzymes involved in the citric acid cycle play a role
in MDS and leukemia pathogenesis. It was found that
cells with mutations in IDH1 and IDH2 accumulated a
novel oncometabolite, 2-hydroxyglutarate (2-HG), instead
56 Molecular Hematology
of alpha-ketoglutarate. 2-HG was shown to diffuse to the
nucleus leading to inhibition of, among others, TET2, and
thereby also affecting DNA methylation. Mutations in IDH1
or IDH2 occur in ∼5% of patients with MDS and portend an
unfavorable clinical outcome. While the oncometabolite was
shown to be sufficient to promote leukemogenesis, the effects
were found to be reversible. This has led to the development
of 2-HG inhibitors, which are currently being evaluated in
clinical trials.
Cohesin complex
The formation of a specific protein complex that holds
together sister chromatids preventing the collapse of the
replication fork is an important part of DNA repair. This pro-
tein complex is made up of the so-called cohesins; that is,
STAG2, SMC3, SMC1A, and RAD21. In aggregate, inactivat-
ing mutations in cohesin genes are found in 15% of MDS
cases. Perhaps somewhat counterintuitive given their func-
tion, cohesin mutations have not been associated with chro-
mosomal abnormalities. Rather, cohesin mutations are now
believed to lead to altered gene expression due to a failure to
stabilize DNA loops that normally facilitate the interaction of
promoters and distant enhancers.
Transcription factors
A small group of hematopoietic transcription factors have
been found to be recurrently mutated in MDS. Since
germline mutations in RUNX1, GATA2, and ETV6 had been
linked to the development of MDS and AML in inher-
ited bone marrow failure disorders, these genes were also
included in targeted sequencing panels for the study of MDS.
Found in approximately 10% of MDS cases, mutations in
RUNX1 are often associated with severe thrombocytopenia.
The protein forms the subunit of the CBF and is responsi-
ble for DNA binding. RUNX1 has several target genes with
important roles in hematopoiesis.
GATA2 is a zinc finger protein highly expressed in
HSCs and essential for normal hematopoietic differentiation.
Somatic mutations in GATA2 account for only 1% of MDS
cases. The transcription factor Wilms tumor protein WT1
has been shown to recruit TET2 to specific loci and is mutated
in fewer than 5% of cases of MDS. There are additional tran-
scription factors with mutation frequencies below 1%.
Cell signaling genes
Genes for components of cell signaling networks were found
to be mutated in a host of other myeloid malignancies
and were thus obvious candidates to study in MDS. How-
ever, compared to AML, chronic myelomonocytic leukemia
(CMML), or polycythemia vera/essential thrombocythemia
(PV/ET), the frequencies with which these genes are mutated
in MDS is quite low. Mutations in components of the MAPK
pathway, for example NRAS, KRAS, NF1, and PTPN11, are
the most prevalent in MDS, and together account for approx-
imately 10% of MDS cases. In general, most mutations in sig-
naling pathway components lead to the constitutive activa-
tion of the protein product. Of note is that these mutations
usually occur late in the course of the disease in more aggres-
sive subclones that signify the transition to secondary AML.
TP53
The single most frequently mutated tumor suppressor gene
among all human cancers is tumor protein 53 (TP53). In the
clinical condition known as Li–Fraumeni syndrome, humans
are born with a single mutant allele of TP53 and these indi-
viduals have a drastically increased risk for many types of
cancer, including MDS and AML. In patients with MDS,
somatic mutations of TP53 are usually associated with high
blast counts, low platelet counts, a complex karyotype, and
previous exposure to chemotherapy. Dubbed “guardian of
the genome,” p53 functions as a mediator of cellular stress.
In the appropriate context, p53 increases pro-apoptotic genes
and mediates cell cycle arrest. Accumulation of additional
genomic aberrations and chromosomal instability together
with chemotherapy resistance observed in patients with inac-
tivation of the p53 pathway are likely due to inadequate acti-
vation of the DNA damage response and loss of cell cycle
arrest measures, respectively.
The negative prognostic impact of TP53 mutations had
previously been demonstrated in a number of hematolog-
ical malignancies, including MDS. Mainly seen in high-
risk and secondary (therapy-related) MDS, TP53 mutations
occurred frequently in the context of complex karyotypes,
and have also been associated with disease progression and
poor response to therapy. In contrast, TP53 mutations were
rarely observed in lower-risk MDS. However, these studies
suffered from small sample size and the inherent low sensi-
tivity associated with the use of conventional Sanger sequenc-
ing techniques. When using sensitive deep sequencing tech-
niques, TP53 mutations could be found in approximately
20% of low-risk del(5q) MDS patients. These mutations can
be detected years before disease progression and confer an
increased risk of evolution to overt leukemia.
Recurrently mutated genes in AML
The observation that the mutational landscapes of MDS and
AML show a significant overlap underlines the notion that
MDS and AML represent stages of a disease continuum. As
with MDS, factors that contribute to splicing, transcription,
epigenetic regulation, cohesion, and cell signaling are

frequently found mutated in AML, as is the gene TP53.
However, there are some mutations representing late events
in AML pathogenesis which are mostly not reported in MDS
cases, notably mutations in NPM1, CEBPA, and FLT3.
Nucleophosmin 1 (NPM 1) participates in many different
cellular processes. It acts as a histone chaperonin and inter-
acts with BRCA2 to regulate centromere duplication. It is also
implicated in proliferation and cell survival, due to its regula-
tion of the tumor suppressor factors alternate reading frame
(ARF) and TP53, as well as the proapoptotic protein kinase
R. Mutations usually occur in exon 12 of NPM1.
CEBPA is a myeloid transcription factor. Mutations in
CEBPA result in inhibition of granulocytic maturation. Since
the introduction of “AML with mutated CEBPA” as a provi-
sional entity in the 2008 edition of the WHO classification,
it has become evident that cases with monoallelic and cases
with biallelic CEBPA mutation differ in terms of gene expres-
sion and prognosis. The overall favorable outcome was only
observed when both CEBPA alleles were mutated, hence the
disease category was adapted in the current WHO classifica-
tion in 2017.
Although FLT3 mutations do not define an AML disease
category, it is strongly recommended to investigate the
FLT3 mutation status at diagnosis. FLT3 encodes a receptor
tyrosine kinase, which acts as a receptor for the cytokine
FLT3LG. Upon ligand activation, FLT3 regulates factors of
both JAK-STAT and RAS signaling pathways. Mutations of
FLT3 can occur within the tyrosine kinase domain (TKD)
and internal tandem duplications (ITDs) of FLT3. Both
render FLT3 constitutively active, which promotes prolif-
eration and cell survival. The presence of a FLT3 mutation
is generally associated with an adverse outcome. Aside to
its prognostic value, early detection of a FLT3 mutation has
clinical implications, as patients may benefit from FLT3
inhibitors, such as midostaurin.
Mutations in AML are more prevalent than in MDS. In
almost every case (96% to >99%) at least one mutation is
readily detected. The vast majority of cases bear two or more
mutations, so that gene–gene interactions have to be taken
into account, especially for prognostic evaluation.
The mutational landscape in AML displays a highly spe-
cific co-mutation pattern: while some mutations are fre-
quently found co-mutated, others are mutually exclusive. A
good example is the recurrently mutated gene NPM1. Co-
mutations of NRAS are frequently detected in amino acid
residues G12 and G13, but never for Q61. The same holds
true for the co-mutation pattern of NPM1 and IDH2: co-
mutations are restricted to IDH2-R140 and not observed for
IDH2-R172.
The importance of the genetic context for prognosis
is increasingly recognized. Depending on occurring co-
mutations, a NPM1 mutation can be associated with a favor-
able or an inferior prognosis. The presence of concomitant
Molecular diagnostics and risk assessment in myeloid malignancies 57
DNMT3A and FLT3-ITD mutations confers an adverse out-
come. Most interestingly, a strong prognostic impact was
only detectable when all three genes were mutated. In con-
trast, a NPM1 mutation in the presence of mutated DNMT3A
and NRAS-G12 or NRAS-G13 as well as FLT3 wild type is a
positive prognostic indicator.
Three recurrently mutated genes define AML disease cate-
gories among the group “AML with recurrent genetic abnor-
malities.” Aside from the respective mutation, blast counts
greater than 20% are required to meet the diagnostic criteria
for AML. There are no specific morphological features asso-
ciated with either of the three molecularly defined AML sub-
types:
r AML with mutated NPM1. This is found in 27–35% of all
adult AML cases and in 45–64% of adult cases with a nor-
mal karyotype. Multilineage dysplasia may be present, but
does not take precedence over the NPM1 mutation. While
the individual prognosis also depends on the genomic con-
text, as already discussed, patients with this subtype gener-
ally correspond well to induction therapy and have a good
prognosis.
r AML with biallelic mutation of CEBPA. As is the case for
AML with mutated NPM1, this subtype is also associated
with a normal karyotype and a favorable outcome. The
presence of multilineage dysplasia does not influence prog-
nosis. FLT3-ITD and GATA2 zinc finger 1 are recurrently
found co-mutated in this AML subtype; however, their
prognostic impact is still unclear.
r AML with mutated RUNX1. This subtype was included
as a provisional entity in the revised 2016 edition of
the WHO classification and should only be diagnosed
if AML with other genetic abnormalities, therapy-related
myeloid neoplasms, or AML with myelodysplasia-related
changes can be ruled out. AML with mutated RUNX1 may
display karyotypic aberrations and co-mutations are fre-
quently found. Among the frequently co-mutated genes
are ASXL1, KMT2A-PTD, FLT3-ITD, IDH1-R132, and
IDH2-R140 or IDH2-R172. Some studies report an adverse
outcome for this AML subtype. Additionally, a concomi-
tant ASXL1 mutation worsens the prognosis.
Further reading
Abelson, S., Collord, G., Ng, S.W.K. et al. (2018). Prediction of acute
myeloid leukaemia risk in healthy individuals. Nature 559: 400–404.
Arber, D.A., Orazi, A., Hasserjian, R. et al. (2016). The 2016 revision
to the World Health Organization (WHO) classification of myeloid
neoplasms and acute leukemia. Blood 127: 2391–2405.
Bacher, U., Schnittger, S., Macijewski, K. et al. (2012). Multilineage dys-
plasia does not influence prognosis in CEBPA-mutated AML, sup-
porting the WHO proposal to classify these patients as a unique
entity. Blood 119: 4719–4722.
58 Molecular Hematology
Brandts, C.H., Sargin, B., Rode, M. et al. (2005). Constitutive activation
of Akt by Flt3 internal tandem duplications is necessary for increased
survival, proliferation, and myeloid transformation. Cancer Res. 65:
9643–9650.
Cancer Genome Atlas Research Network (2013). Genomic and epige-
nomic landscapes of adult de novo acute myeloid leukemia. N. Engl.
J. Med. 368: 2059–2074.
Chen, W., Drakos, E., Grammatikakis, I. et al. (2010). mTOR signaling
is activated by FLT3 kinase and promotes survival of FLT3-mutated
acute myeloid leukemia cells. Mol. Cancer 9: 292.
Choudhary, C., Schwable, J., Brandts, C. et al. (2005). AML-associated
Flt3 kinase domain mutations show signal transduction differences
compared with Flt3 ITD mutations. Blood 106: 265–273.
Dohner, H., Estey, E., Grimwade, D. et al. (2017). Diagnosis and man-
agement of AML in adults: 2017 ELN recommendations from an
international expert panel. Blood 129: 424–447.
Dohner, H., Weisdorf, D.J., and Bloomfield, C.D. (2015). Acute myeloid
leukemia. N. Engl. J. Med. 373: 1136–1152.
Falini, B., Macijewski, K., Weiss, T. et al. (2010). Multilineage dysplasia
has no impact on biologic, clinicopathologic, and prognostic features
of AML with mutated nucleophosmin (NPM1). Blood 115: 3776–
3786.
Gaidzik, V.I., Bullinger, L., Schlenk, R.F. et al. (2011). RUNX1 mutations
in acute myeloid leukemia: results from a comprehensive genetic and
clinical analysis from the AML Study Group. J. Clin. Oncol. 29: 1364–
1372.
Greenberg, P.L., Tuechler, H., Schanz, J. et al. (2012). Revised inter-
national prognostic scoring system for myelodysplastic syndromes.
Blood 120 (12): 2454–2465.
Greif, P.A., Dufour, A., Konstandin, N.P. et al. (2012). GATA2 zinc fin-
ger 1 mutations associated with biallelic CEBPA mutations define a
unique genetic entity of acute myeloid leukemia. Blood 120: 395–403.
Grimwade, D., Ivey, A., and Huntly, B.J. (2016). Molecular landscape of
acute myeloid leukemia in younger adults and its clinical relevance.
Blood 127: 29–41.
Jongen-Lavrencic, M., Grob, T., Hanekamp, D. et al. (2018). Molecular
minimal residual disease in acute myeloid leukemia. N. Engl. J. Med.
378: 1189–1199.
Kunchala, P., Kuravi, S., Jensen, R. et al. (2018). When the good go
bad: mutant NPM1 in acute myeloid leukemia. Blood Rev. 32: 167–
183.
Pabst, T., Mueller, B.U., Zhang, P. et al. (2001). Dominant-negative
mutations of CEBPA, encoding CCAAT/enhancer binding protein-
alpha (C/EBPalpha), in acute myeloid leukemia. Nat. Genet. 27: 263–
270.
Pang, Q., Christianson, T.A., Koretsky, T. et al. (2003). Nucleophosmin
interacts with and inhibits the catalytic function of eukaryotic initia-
tion factor 2 kinase PKR. J. Biol. Chem. 278: 41709–41717.
Papaemmanuil, E., Gerstung, M., Bullinger, L. et al. (2016). Genomic
classification and prognosis in acute myeloid leukemia. N. Engl. J.
Med. 374: 2209–2221.
Paschka, P., Schlenk, R.F., Gaidzik, V.I. et al. (2015). ASXL1 mutations
in younger adult patients with acute myeloid leukemia: a study by the
German-Austrian Acute Myeloid Leukemia Study Group. Haemato-
logica 100: 324–330.
Schnittger, S., Dicker, F., Kern, W. et al. (2011). RUNX1 mutations are
frequent in de novo AML with noncomplex karyotype and confer an
unfavorable prognosis. Blood 117: 2348–2357.
Schuurhuis, G.J., Heuser, M., Freeman, S. et al. (2018). Mini-
mal/measurable residual disease in AML: consensus document from
ELN MRD Working Party. Blood 131: 1275–1291.
Stone, R.M., Mandrekar, S.J., Sanford, B.L. et al. (2017). Midostaurin
plus chemotherapy for acute myeloid leukemia with a FLT3 mutation.
N. Engl. J. Med. 377: 454–464.
Swerdlow, S.H., Campo, E., Pileri, S.A. et al. (2016). The 2016 revision
of the World Health Organization (WHO) classification of lymphoid
neoplasms. Blood 127: 2375–2390.
Taskesen, E., Bullinger, L., Corbacioglu, A. et al. (2011). Prognostic
impact, concurrent genetic mutations, and gene expression features
of AML with CEBPA mutations in a cohort of 1182 cytogenetically
normal AML patients: further evidence for CEBPA double mutant
AML as a distinctive disease entity. Blood 117: 2469–2475.
Wang, H.F., Takenaka, K., Nakanishi, A., and Miki, Y. (2011). BRCA2
and nucleophosmin coregulate centrosome amplification and form
a complex with the Rho effector kinase ROCK2. Cancer Res. 71:
68–77.
Chapter 5 Molecular basis of acute
lymphoblastic leukemia
Bela Patel & Fiona Fernando
Centre of Haemato-Oncology, Barts Cancer Institute, Queen Mary University of London, London, UK
Introduction, 59
Molecular methodologies, 59
Primary cytogenetic subgroups, 60
Biological consequences of chromosomal alterations, 60
Additional cooperating genetic events, 60
Introduction
Cure rates for childhood acute lymphoblastic leukemia
(ALL) have markedly increased over the last four decades
and now reach >80%. Despite these undoubtedly excellent
results, challenges remain, particularly for relapsed ALL,
which is a major contributor to childhood mortality. In adult
ALL, survival gains have been less pronounced despite the
use of optimized approaches, and currently only 50% achieve
long-term survival. In both groups, treatment toxicity is a sig-
nificant limitation leading to recurring short- and long-term
morbidity. Thus, along with a clear need to improve treat-
ment efficacy, there is a requirement also to mitigate treat-
ment toxicity. The need for more rationalized approaches to
treatment is therefore pressing.
Recent efforts to molecularly characterize ALL have greatly
informed this priority area. Novel molecular-oncogenetic
groups have emerged providing new prognostic and pre-
dictive biomarkers for informing novel and optimized
treatments. Moreover, molecular profiling has paved the way
for the discovery of key pathways essential for the develop-
ment and progression of ALL, thereby contributing essential
knowledge not only of disease biology, but also of therapy
failure. Current advances in our understanding of the genetic
basis of ALL and therapy failure are summarized in this chap-
ter, with particular emphasis given to how such insights are
evolving opportunities for clinical translation.
Molecular Hematology, 4th Edition. Edited by Drew Provan and John G. Gribben.
© 2020 John Wiley & Sons Ltd. Published 2020 by John Wiley & Sons Ltd.
Developmental origins, 62
Molecular genetic subsets, 62
Molecular evaluation of Minimal Residual Disease, 69
Conclusion, 70
Further reading, 70
Molecular methodologies
Tools for interrogating the molecular pathophysiology of
ALL encompass genomic, transcriptomic, and epigenomic
methods, which synergistically contribute to in-depth molec-
ular profiling. The two time-honored methods for accurately
determining recurrent chromosomal abnormalities routinely
applied in clinical practice are chromosomal G banding
on metaphase nuclei and fluorescence in situ hybridization
(FISH). These methods evaluate macroscopic genetic change
and will discriminate major molecular subgroups in ALL.
However, they have limited resolution.
Over the last decade, several technologies have evolved
that are capable of mapping genetic change at both the
submicroscopic and genome-wide levels. These method-
ologies include microarray profiling of single-nucleotide
polymorphisms and comparative genomic hybridization to
detect DNA copy number change, corresponding to ∼kilo
mega base regions of genomic amplification and/or deletions
in potential candidate oncogenes and tumor suppressors.
The development of next-generation high-throughput
sequencing technologies has paved the way for massively
parallel sequencing of targeted gene loci, the protein-coding
exome and whole genome itself contributing an in-depth
insight into the global cancer genome. Complementing these
DNA approaches are methods that evaluate the transcrip-
tome and epigenome, including gene expression microarray
analysis together with direct mRNA sequencing. Insight into
the epigenomic mechanisms regulating gene expression in
leukemia has proceeded through high-density DNA methy-
lation arrays detecting CpG nucleotide status at transcrip-
tional start sites, as well as through chromatin immunopre-
cipitation assays (CHIP–CHIP array or CHIP sequencing),
59
60 Molecular Hematology

where DNA-associated proteins regulating locus-specific
gene expression can be directly profiled. Collectively, these
methods contribute an unprecedented depth of knowledge
about the acquired molecular defects and genomic landscape
of ALL.
Primary cytogenetic subgroups
Altogether, approximately ∼30% of children with ALL and
a further 40% of adults demonstrate a gross chromosomal
alteration as a defining feature of their disease. The detection
of a gross chromosomal anomaly relies on microscopic eval-
uation of G-banded metaphase chromosomes, or the appli-
cation of locus-specific and centromeric FISH gene probes,
together with targeted molecular techniques for determina-
tion of cytogenetically cryptic lesions such as t(12;21). These
tools are integral to the primary diagnostic evaluation of ALL,
as they assist in formalizing the diagnosis by determining the
presence of an ALL-specific chromosomal aberration, which
in turn also informs the underlying prognosis.
A list of the recurring cytogenetic abnormalities reported
in ALL, their prevalence, and prognostic association is pro-
vided in Figure 5.1. The nature of the chromosomal lesion
varies according to whether B- or T-lineage disease is present.
There are also striking differences in the distribution of cyto-
genetic abnormalities by age. The recurring hyperdiploid and
t(12;21) karyotypes both associate with excellent prognosis
and are almost exclusively observed in childhood ALL, where
they account for the two largest cytogenetic subsets. By con-
trast, both these good-risk subtypes are rarely observed in
adult ALL. Conversely, the single largest cytogenetic subset
in adult ALL is formed by t(9;22), which confers poor out-
come and is seen in up to 25% of adults. The t(9;22) is rarely
observed in childhood ALL, accounting for only 2–3% of
cases. Hence, good-risk cytogenetic features highly predom-
inate in childhood ALL, whereas there is an excess of poor-
risk cytogenetic subtypes in adult ALL. Cases of pre-B ALL
lacking a well-defined cytogenetic abnormality are now col-
lectively referred to as B-other ALL.
Biological consequences of
chromosomal alterations
The recurring cytogenetic aberrations reported in ALL
reflect either numerical changes in chromosome complement
or an abnormality in chromosome structure, both leading to
a gross structural anomaly usually identifiable through con-
ventional cytogenetics. Both hyper- and hypodiploid ALL
are defined by an altered chromosomal count, whereas most
other cytogenetic subsets have arisen through an acquired
chromosomal mutation. The single most common cause of
chromosomal mutation is chromosomal translocation, for
example t(12;21), t(9;22). Deletions, inversions, and isochro-
mosomes represent further examples of structural chromo-
somal change; however, these alterations do not give rise to a
recurrent entity in ALL and therefore are not currently rec-
ognized as a major cytogenetic subset.
The process of chromosomal translocation arises from
chromosomal breakage followed by recombination of sepa-
rated fragments without any loss or duplication of genetic
material (i.e. balanced rearrangements). The biological con-
sequences of chromosomal displacement depends on the
individual gene targets affected and are therefore consid-
ered separately, but as a broad principle the net effect is dis-
ruption in the homeostatic function of one of the affected
genes (Figure 5.2). It is also important to note that while the
pathogenic consequences of chromosomal rearrangements
have undoubtedly provided insight into the early initiation
of leukemogenesis, recently completed genome-wide stud-
ies have supplemented this understanding by defining addi-
tional mutational targets, which collaborate with primary
acquired chromosomal defects (reviewed later) to establish
the full disease phenotype.
Additional cooperating genetic events
Several lines of evidence suggest that primary chromosomal
lesions alone are not the full explanatory cause of ALL.
The first and most compelling is the low concordance of
ALL (10–15%) in monozygotic twins sharing a common
chromosomal translocation. The second is the failure of ALL
chromosomal lesions to induce malignant transformation
in a variety of experimental models, and the last the finding
of ALL-associated gene rearrangements in the healthy pop-
ulation. Consequently, researchers have sought to identify
additional complementary genetic lesions that complete
disease development.
Higher-resolution genetic methods utilizing DNA
microarray and next-generation sequencing have provided
the critical insight here, by revealing an additional landscape
of submicroscopic genome-wide change accompanying ALL
development. Critical findings are that compared to many
epithelial cancers the overall mutation burden in ALL is low,
with ∼6–8 copy number alterations per individual tumor,
suggesting that gross genomic instability is not a major
mechanism of malignancy. Among the alterations identified,
focal gene deletions dominate over gene amplifications and
sequence mutations. The most common targets of genetic
alterations are biological pathways controlling lymphoid
development (PAX5, IKZF1 EBF1, LMO2); cell cycle
(CDKN2A/CDKN2B, TP53); cytokine signaling (CD200,
 Molecular basis of acute lymphoblastic leukemia 61

Cytogenetic Molecular Children Adults Major pathogenic

abnormality alteration % % consequence

Pre-B ALL
Chimeric transcription factor with
t(12:21) (p13:q22) ETV6-RUNX1 22 <1 altered properties

High hyperdipoidy Associated with multiple 25-30 10

submicroscopic gene Determined by individual gene
alterations

Deregulated tyrosine kinase

t(9:22) (q34:q11) BCR-ABL1 2 19

t(4:11) (q21:q23) MLL-AF4 3 ~7–10

Other MLLq23

Epigenetic dysregulation
MLL-ENL: (t11:19) (q23:p13) <2 <2
MLL-AF9:
t(9:11) (p22:q23)
t(10:11)q13-14:q14-21)

IgH@–CRLF2,
P2RY8-CRLF2 Dysregulated CRLF2 expression
CRLF2 deregulation ~5–7 6
Associated with multiple
Determined by individual gene
Low hypodiploidy/Near submicroscopic gene
triploidy alterations ~1 4
Chimeric transcription factor with
t(1:19) (q23:p13) TCF3-PBX1 ~4 ~2 altered properties

RUNX1 amplification, miR-

802, and genes mapping to the Not well known
iAMP21 Down syndrome critical region 2 <1
ID4 6p22
CEBP Oncogene activation
IgH@-t EPOR 19p13 <5 8

None “B other” - 30 40 –

Pre-T ALL
Oncogene TF upregulation
1p32 deletion (SIL/TAL) ~15–25 Lower
t(1:14) (p34:q11)
t(1:7) (p32:q35) TAL1
(SIL/TAL 25%) (SIL/TAL ~5–10%)
5’ super-enhancer mutations

t(11:14) (p15:q11)
t(7:11) (q34:p15) LMO1
t(11:14) (p13:q11) 10–14 –
Oncogene TF upregulation
t(7:11) (q34:p13)
11p13 deletion LMO2

t(10:14) (q24:q11)
t(7:10) (q35:q24) Oncogene TF upregulation
del(10)(q24q26) HOX11/TLX1 3–7 30

t(5:14) (q35:q32) TLX3/HOX11L2/-BCL11B 20–25 5 Oncogene TF dysregulation

Intermediate risk
PICALM-MLLT10 (CALM-
t(10:11) (p13:q14) AF10) ~10 – Fusion TF oncogene
Adverse risk Fusion TF oncogene
t(11:19)(q23:p13) KMT2A–MLLT1 (MLL-ENL) 1% –

t(7:9)(q34:q34) Notch1 rearrangement <1 – Oncogene activation

good risk
t(8:14)(q24:q11) MYC ~1 – Proto-oncogene activation

NUP214-ABL1 ABL1 3-6 – Deregulated tyrosine kinase

t(6:7)(q23:q34) MYB 3 – Oncogene TF dysregulation

t(8:14)(q24:q11) MYC 1 – Oncogene activation

None - ~50 – –

Fig. 5.1 Cytogenetic subsets in acute lymphoblastic leukemia (ALL) and major molecular abnormalities. TF, transcription factor.
62 Molecular Hematology

Phosphorylation of
tyrosine residues on
multiple substrates

ATP

BCR-ABL
t(9;22)
kinase
BCR-ABL

• Cellular proliferation
• Cell survival
• Reduced stromal
A B dependance
Hybrid fusion gene

A B
Regulatory/enhancer Constitutive activation
Chromosomal Balanced
rearrangement translocation

T-cell
TAL1 target transformation
Molecular alterations (s) t(1;14) +++ genes
TAL1-TCRD
TAL1

Molecular pathogenesis

Fig. 5.2 Recurring targets of submicroscopic gene alterations in pre-B acute lymphoblastic leukemia (ALL) and association with disease subgroups.
ped, pediatric.

BTLA, and BLNK); drug response, such as NR3C gluco-
corticoid receptor; as well as tumor suppressors (PTEN and
RB1), transcriptional activators of hematopoiesis (ETV6 and
ERG), and epigenetic modifiers. While certain lesions are
enriched in specific cytogenetic groups and are considered
separately, a prevalent finding in pre-B ALL is inactivating
mutations in genes encoding key transcription factors
involved in B cell development. The two most commonly
mutated B cell transcription factors are PAX5, observed in
>30% (deletions > point mutations > cryptic transloca-
tions), and IKAROS1, (IKZF1) deletions, found in >10%.
Both transcription factors control cell transition from the
pro to mature B cell stage, with gene inactivation predicted
to impair B cell differentiation.
Developmental origins
Collectively, these newly characterized molecular lesions not
only provide greater insight into the etiological cause of
ALL, but also significantly advance our understanding of the
natural history of this disease. In this respect, identical twin
leukemias have served as an ideal model for studying the
temporal sequence of genetic events leading to ALL develop-
ment. Seminal studies here have identified a frequent prena-
tal origin of chromosomal lesions associated with prolonged
latency to ALL development consistent with these defects,
representing an early event in leukemogenesis. Additional
oncogenetic lesions are acquired postnatally, indicating that
these are secondary “hits” in ALL development. Altogether
these findings, together with substantial evidence support-
ing the requirement for additional events other than primary
chromosomal lesions, strongly support a multistep etiology
underlying ALL development, wherein primary cytogenetic
alterations are founder events occurring very early in ALL
evolution, leading to the establishment of a pre-leukemic
clone with an altered propensity for survival, self-renewal,
or transcription. The disruption in normal cellular behavior
endows a selective advantage to pre-leukemic clones which
predisposes for further genetic events affecting multiple inde-
pendent cellular pathways (e.g. B cell development), leading
to clone evolution and eventual full-blown clinical disease
development. Thus, the recent expanded knowledge of ALL
oncogenomics, derived in particular through high-resolution
profiling, has led to a fundamentally enhanced understand-
ing of the evolutionary basis of the disease.
Molecular genetic subsets
B-precursor all
Increasing knowledge of chromosomal aberrations and
their molecular counterparts, together with an improved
 Molecular basis of acute lymphoblastic leukemia 63

Pathway Gene(s) Frequency

B-cell transcription factor Pax-5 ~30%

>60% Ph-pos ALL
-Inactivating mutations IKZF1 ~30% Ph-neg adult ALL
~15% Ph-neg paed ALL
~30% Ph-like ALL

IKZF2 ~50% low hypodiploid ALL

IKZF3 ~13% near haploid ALL

Cell cycle regulation CDKN2A/B ~25%

~10% Ph-like ALL

Cytokine receptor/kinase
JAK 1/2 >95% CRLF2 mutated
signalling 25% Down syndrome
~5–16% Kinase-activating
-Activating mutations ~50% of Ph-like ALL mutations in Ph-like ALL
CRLF2
>50% Down syndrome
~50% JAK/2 mutated CRLF2, tyrosine kinase
genes (ABL1/2, CSFR1,
Kinase-activating PDGFRB), other JAK-
mutations >90% of Ph-like
STAT targets (e.g. JAK2,
EPOR) and RAS

RTK/RAS ~70% near haploid ALL

Tumour suppressor 90% low hypodiploid ALL

TP53
Inactivating mutations

RB1 ~40% low hypodiploid ALL

Transcriptional regulator
~7%
Inactivating deletions ERG –15% of B other

Fig. 5.3 Mechanisms of aberrant gene activities by chromosomal translocations.

genome-wide understanding of secondary oncogenetic
events in ALL, has defined new molecular entities while also
enhancing our understanding of disease pathogenesis in
well-established cytogenetic groups. These are now further
discussed and summarized in Figure 5.3.
Philadelphia chromosome positive t(9;22) (q34;q11) ALL
(Ph-pos)
The pathogenic consequence of balanced translocation
between chromosomes 9 and 22, t(9;22)(q34;q11), leads to a
shortened chromosome 22, also known as the Philadelphia
chromosome. Ph-pos ALL is perhaps the best-elucidated
molecular entity of ALL, owing partly to the enhanced
molecular pathophysiological understanding of chronic
myeloid leukemia (CML), which similarly is characterized
by t(9;22) genetic rearrangement. Creation of the Philadel-
phia chromosome (Ph) 22q leads to fusion of the breakpoint
cluster region gene (BCR) at chromosome 22q11 to the
Abelson gene (ABL) protein-tyrosine kinase gene at chro-
mosome 9q23 during t(9;22) recombination, with resultant
BCR-ABL1 oncogene creation. Depending on the breakpoint,
either the p210 or p190 BCR/ABL isoform results, which
can readily be detected ty reverse transcription-polymerase
chain reaction (RT-PCR). The resulting BCR/ABL1 fusion
gene encodes a chimeric BCR/ABL1 oncoprotein that
has constitutively activated ABL1 tyrosine kinase enzyme
activity, an activator of cellular signaling. Tyrosine kinases
initiate cellular signal transduction cascades by catalyzing
the transfer of the phosphate of adenosine triphosphate
(ATP) to tyrosine residues on target protein (signaling)
substrates, which drives a signal transduction program to
enhance cell proliferation, self-renewal, and reduce apop-
tosis – behaviors which typically contribute to malignant
transformation.
Molecular elucidation of BCR-ABL1-mediated oncoge-
nesis has blazed the trail for molecularly targeted therapies
in the form of tyrosine kinase inhibitors (TKIs). Imatinib, a
first-generation BCR/ABL1 TKI, blocks the binding of ATP
to the kinase domain of the BCR-ABL1 oncoprotein, which
in turn blocks the phosphorylation activity of BCR/ABL1
oncoprotein, thereby extinguishing oncogenic cell signaling.
64 Molecular Hematology
Subsequent-generation TKIs nilotinib, dasatinib and pona-
tinib have increased potency for kinase inhibition. Although
TKIs broadly share the same mechanism of action, namely
competing with ATP, the phosphorylating entity, at the
catalytic binding site of tyrosine kinase, they differ from
each other most notably in potency and in their interac-
tion with either the closed/non-functional (imatinib and
nilotinib) or open/functional conformation of the kinase
domain (dasatinib). Despite the proven efficacy of TKIs in
Ph ALL, resistance can occur by a multitude of mechanisms,
including genetic mutations caused by acquired mutations
in the ABL-Kinase domain, which prevent drug binding.
The most intractable ABL1 kinase mutation is the T315I,
which confers pan-TKI resistance with sensitivity remaining
to ponatinib alone. Overall, the molecular elucidation of the
BCR-ABL oncogene is a shining example of how molecular
scrutiny can directly inform underlying ALL biology, yield-
ing novel predictors of response to targeted treatments and
facilitate individualization of treatment.
However, unlike CML, where TKI inhibition effects a
functional cure, Ph-pos ALL cannot be cured through
TKI inhibition alone, pointing to important biological
differences between these diseases. One key difference is
the target cell for transformation. ALL represents malignant
transformation of a lymphoid progenitor, as opposed to the
hematopoietic stem cell (HSC), which is the cellular target of
malignant transformation in CML. Furthermore, additional
acquired defects in several tumor suppressor genes typify
Ph-associated ALL versus CML, strongly implicating the
disruption of these genes as contributory to Ph ALL patho-
genesis and aggressiveness. One of the key recurring genetic
inactivations associated with Ph ALL involves the IKZF1
gene encoding the B cell-specific transcription factor Ikaros,
leading to severely reduced expression of the functional
protein. Over 60% of Ph ALL exhibit focal deletions of
IKZF1, making this the most prevalent co-occurring genetic
abnormality associated with Ph ALL; point mutations also
contribute to IKZF1 haploinsufficiency, although these
are less frequently observed. Recurring loss of function
in CDKN2A/B cell cycle control together with PAX5 and
EBF1 B cell development genes are also prevalent (∼50%)
in the Ph ALL subtype and likely contribute to BCR-ABL1
leukemogenesis. Loss of Ikaros has been shown by some
studies to confer a poor prognosis, yielding this as a potential
marker for clinical risk stratification.
MLL-rearranged ALL
Chromosomal translocations involving the KMT2 gene (also
known as MLL) represent one of the most aggressive forms of
ALL. A large number of partner genes, AFF1 (AF4), MLLT1
(ENL), MLLT4 (AF6), MLLT3 (AF9), and MLLT10 (AF10),
have been shown to recombine with mixed-lineage leukemia
(MLL), but by far the most common is AF4, located on chro-
mosome 4, which undergoes rearrangement with MLL on
chromosome 11, resulting in t(4;11)(q21;q23) and MLL-AF4
gene fusion. A typical MLL fusion protein contains the N ter-
minus of MLL encoded by the first 8–13 exons and the C ter-
minus of one of over 50 fusion partner genes. MLL rearrange-
ment mediated by t(4;11) (q21;q23) is frequently present in
infant leukemia, accounting for up to 50% of cases, with
a diminishing incidence (2% in children, 7–10% in adults)
thereafter.
A number of features define MLL-rearranged ALL as a
distinct clinicobiological subtype, namely poor outcome,
unique association with a pro-B immunophenotype, and
excessive use of VH6 immunoglobulin gene – an early
event in immunoglobulin gene rearrangement as well as
a distint gene expression profile. The major mechanisms
underpinning MLL-rearranged leukemogenesis have now
been clarified and point to disordered epigenetics as a major
mechanism of malignancy. The MLL gene encodes a large
multifunctional epigenetic regulator (DNA methyltrans-
ferase and histone H3K4 methyltransferase – a positive global
regulator of gene transcription providing essential epigenetic
maintenance of the HOX family of genes, master regulators
of hematopoiesis. MLL fusion proteins corrupt the intrinsic
epigenetic regulatory activity, leading to global changes in
gene transcription and in particular inappropriate expression
of MLL target genes such as HOXA9 and MEIS1, which act
downstream to suppress differentiation and induce aberrant
self-renewal.
Extensive studies have characterized the mediators of tran-
scriptional and epigenetic perturbations caused by MLL
mutations. These studies have identified co-factor epigenetic
enzymes (LEDGF, DOT1L, CBX8, KDM1A, BRD4, P-TEFb,
BET) that assemble with MLL fusion proteins to form mul-
tiprotein complexes to mediate the epigenetic disturbance.
The elucidation of these protein associates of MLL fusions
has opened up several therapeutic opportunities. For exam-
ple, identification of the DOT1L mediator of H3K79 histone
modification, which is significantly enriched at MLL target
genes, has led to the targeting of histone methylation induced
by DOT1L. Similarly, the BET family of chromatin adapter
proteins, which associate with oncogenic MLL fusions, can
be targeted by novel small-molecule inhibitors that displace
BET proteins from chromatin. Both strategies are currently in
early-phase clinical trials and results are eagerly awaited. Tar-
geting of epigenetic enzymes that associate with MLL fusions
to induce leukemogenic suppression is therefore a promising
therapeutic avenue.
High hyperdiploidy
The hyperdiploid karyotype is characterized by mas-
sive aneuploidy. The formal definition is an increase in

chromosomal copy numbers, with gains of between 51
and 65 chromosomes. Hyperdiploid karyotypes constitute
25–30% of pediatric ALL, making this one of the largest
cytogenetic subsets in children. The prognosis of high-
hyperdiploid ALL in children is excellent, with overall
survival exceeding 90%. In contrast, high hyperdiploidy
occurs less frequently in adult ALL and does not confer
a similar magnitude of excellent prognosis, although cer-
tain chromosomal amplifications may be associated with
increased survival gain. Chromosomal gains in hyperdiploid
ALL typically include +X, +4, +6, +10, +14, +17, +18, and
+21, and can also include trisomies (e.g. +4, +10, +17,
+18), which may be specified as additional biomarkers
for improved survival. The pathogenic consequences of
chromosomal hyperdiploidy are poorly understood, as the
specific target genes affected are not known, but it is generally
believed that gene dosage effects contribute to the oncogenic
role of a hyperdiploid karyotype. Secondary genetic events
(IKZF1 deletions, ETV6, and RAS pathway mutation) have
been identified using high-resolution techniques, suggesting
that high hyperdiploidy also arises through a multistep
process. The presence of these secondary abnormalities has
not been shown to confer prognostic significance.
t(12;21) (p13;q22) ALL
The t(12;21)(p13;q22) translocation is the most common
chromosomal abnormality in pediatric ALL. The rearrange-
ment creates a fusion gene encompassing the non-DNA bind-
ing region of TEL (ETV6) and almost the entire locus of
AML1 (RUNX1), which retains its coding sequence. ETV6-
RUNX1 resulting from t(12;21)(p13;q22) is cytogenetically
cryptic and cannot be discerned reliably in G-banded prepa-
rations, therefore FISH or RT-PCR is required for its accu-
rate detection. TEL and AML1 are two well-characterized
transcription factors which are required for development and
definitive hematopoiesis. The TEL-AML1 fusion causes per-
turbations in normal hematopoiesis, enhancing self-renewal
of B cell progenitors and altering HSC homeostasis, but is
insufficient on its own to induce the full leukemia pheno-
type. The multistep model of leukemogenesis which premises
a pre-leukemia initiating clone has been best demonstrated
for this unique ALL subtype.
Cytokine receptor-like factor 2 deregulated ALL
Deregulated expression of CRLF2, resulting from juxtapo-
sitioning of the CRLF2 gene to either the Immunoglobu-
lin heavy gene enhancer (IGH@-CRLF2) or the pseudo-
autosomal region or the P2RY8 promoter (P2RY8-CRLF2)
of chromosomes X or Y, occurs in up 5–7% of B-ALL,
with an increased incidence in B-other (30%) and in
Down syndrome–associated ALL (>50%). The biochemical
Molecular basis of acute lymphoblastic leukemia 65
consequences of chromosomal activation of CRLF2 appear
to be activation of the JAK-STAT, ERK, and mTOR/PI3K cel-
lular signaling pathways. Furthermore, JAK2 gene mutations
are concomitantly present in ∼50% of CRLF2-rearranged
ALL, which contributes to hyperactivation of this signal-
ing axis. Although an association of CRLF2-deregulated
ALL with clinical outcome has not been reproducibly con-
firmed, some studies in adults demonstrate an association
with higher rates of treatment failure. More significantly, the
discovery of CRLF2 molecular alteration provides a rationale
target for signal transduction inhibitors of the JAK/STAT and
PI3K pathways in CRLF2-overexpressing ALL.
Hypodiploid ALL
Karyotypes with <45 chromosomes are collectively classi-
fied as hypodiploid. Further dissection by the degree of
hypodiploidy generates two major subgroups: near haploid
(NH), with 24–31 chromosomes, which is more common,
and low hypodiploid, with 32–39 chromosomes. While there
is no distinction in prognosis – both are uniformly poor
– recent genetic studies reveal variance in the molecular
pathologies of these groups. NH ALL is characterized by
a very high frequency of mutations, deletions, amplifica-
tions, and or/sequence mutations that activate RAS signaling
and several tyrosine receptor kinases (FLT3, NF1, MAPK1,
PTPN11), in addition to a high prevalence of inactivating
alterations of the lymphoid transcription factor IKZF3 gene
(∼13%), which is almost exclusive to the NH subtype. By con-
trast, LH is characterized in >90% of cases by TP53 alter-
ations, with further abnormalities of CDKN2A/B, RB1, and
inactivation of IKZF2 (∼50%). Interestingly, TP53 mutations
are constitutional in a significant proportion of childhood LH
cases, indicative of a possible underlying Li–Fraumeni syn-
drome. The finding of hypodiploid cells harboring activating
mutations in RAS signaling pathways provides a rationale for
blockade therapies targeting MAPK/MEK/ERK signaling in
this poor-risk ALL subset.
T ALL
T ALL accounts for ∼20% of adult and 8–15% of child-
hood ALL. Gross chromosomal alteration is less frequently
observed in T- versus B-lineage ALL, with just 50% of cases
harboring a primary lesion. A further contrast is that T ALL–
specific abnormalities are not clearly prognostic for relapse
or overall survival, and are therefore not clinically applied
for risk assignment. The nature of chromosomal aberrancy
in T ALL is dominated by balanced translocations, which
occur in up to 20–25% of cases, with numerical changes being
exceedingly rare. Although the participating chromosome
regions undergoing rearrangement can vary, the unifying
outcome of these events is induction of aberrant expression
66 Molecular Hematology
of hematopoietic transcription factor oncogenes, due to inap-
propriate positioning of these genes near constitutively active
T-cell-specific enhancers/regulatory regions within the TCR
beta (7q32-q36) or TCRA–TCRD (14q11) 7q32-q36 loci.
These oncogenic transcription factors include basic helix-
loop-helix (bHLH) family members such as TAL1, LYL1,
TAL2, and BHLHB1, LIM domain only (LMO) genes (LMO1
and LMO2), and HOX transcription factor oncogenes TLX1,
TLX3, and HOXA. Activation of these transcription factor
genes or transcription factor accessory proteins (LMO1 and
2) leads to enhanced activity of the transcriptional regulatory
protein at the promoter or enhancer elements of a specific set
of target genes involved in the development of hematopoietic
lineages.
A prototypical example of chromosomal translocation/
transcription factor dysregulation in T ALL is the disruption
of TAL1 (T-cell acute leukemia), also known as the SCL or
TCL5 gene. The genetic events leading to TAL1 overexpres-
sion occur by a variety of mechanisms. Most commonly,
TAL1 is activated by a cryptic interstitial deletion, TALd,
which results in the deletion of a ∼100 kb DNA fragment
next to the TAL1 locus and fusion with the 5′ part of the SIL
gene, leading to formation of the SIL-TAL1 chimeric gene.
This deletion occurs in up to 25% of T ALL in children, but
is much less common (∼5–10%) in adults. TAL1 overex-
pression is also brought about by t(1;14)(p33;q11), which
leads to inappropriate TAL1 activation, due to its relocation
close to the TCRAD locus. A recently discovered genetic
mechanism delineating a novel basis for TAL1 dysregulation
in T ALL independent of chromosomal translocation is
the occurrence of somatic mutation in non-coding regions
downstream of TAL1, leading to the formation of an onco-
genic “super-enhancer” that drives transcription and over-
expression of TAL1. Thus, multiple genetic defects in T ALL
converge on TAL-1 oncogene overexpression.
The second largest cytogenetic subgroup in T ALL, rep-
resenting ∼20% of cases in childhood, is characterized by
t(5;14)(q35;q32) which juxtapositions the TLX3/HOX11L2
gene to BCL11B, leading to TLX3 transcription factor dysreg-
ulation via a non-TCR enhancer-driven mechanism. Occa-
sionally, chromosomal translocations in T ALL are seen that
produce fusion genes encoding chimeric transcription factor
oncogenes, such as t(11;19)(q23;p13) leading to MLL-ENL,
and also t(10;11)(p13;q14) resulting in PICALM-MLLT10
(CALM-AF10). A common feature of leukemias harbor-
ing these fusion transcription factor oncogenes is the aber-
rant expression of developmentally important HOXA genes.
Recently, a rare cryptic 9q34 deletion generating the SET-
NUP214 fusion gene has been described in T ALL which,
similar to BCR-AB1, has constitutively activated tyrosine
kinase activity and is associated with sensitivity to inhibi-
tion with tyrosine kinase inhibitors, especially nilotinib and
dasatinib.
Somatic mutations involving the NOTCH1 gene lead to
constitutive activation of Notch1 signaling, which is a promi-
nent oncogenic pathway in T ALL. In both childhood and
adult ALL, the incidence of activating NOTCH1 mutations
is between 50% and 70%. In addition, FBXW7 mutations,
present in about 15% of T ALL cases, contribute to Notch
activation by impairing the proteasomal degradation of acti-
vated Notch1 in the nucleus. The NOTCH1 gene encodes a
type I transmembrane heterodimeric receptor, which func-
tions as a ligand-activated transcription factor. Upon lig-
and binding, a series of proteolytic cleavages of the recep-
tor, the last of which is catalyzed by gamma secretase,
releases intracellular Notch1, which acts downstream to
form a transcriptional activator complex for Notch1 tar-
get genes influencing differentiation, proliferation, apoptotic
events, and in-cell fate determination/choices. Crucial effec-
tors of the oncogenic program downstream of NOTCH1-
mutated T ALL include MYC and the PI3K–AKT signal-
ing pathway. A rare translocation t(7;9) (q34;q34.3) leads
to aberrant expression of a number of truncated Notch1
isoforms, referred to as TAN (for translocation-associated
Notch homolog). These truncated Notch1 isoforms (for
example TAN1) have a dominant oncoprotein effect, possess-
ing constitutive, ligand-independent activity. However, the
major cause of Notch1 dysregulation in T ALL is somatic
mutations of the extracellular Notch1 heterodimerization
domain or C terminal PEST domain, which confer consti-
tutive, ligand-independent activity to the Notch1 receptor.
Discovery of molecularly altered Notch1 has led to tar-
geted therapies that therapeutically terminate Notch1 sig-
naling using γ-secretase inhibitors. Studies have shown that
combined with steroids, γ-secretase inhibitors can lead to
clinically significant leukemia regressions, demonstrating the
promising potential for Notch1 blockade therapies in T ALL.
A further frequent target of recurrent mutation in T ALL is
CDKN2A deletions, occurring in 70% of cases. Multiple other
components of cell cycle control are also targets of damaging
mutations. These include chromosomal deletion of RB1 and
CDKN1B genes and CCND, all seen in up to 15% of cases.
Thus cell cycle dysregulation plays a prominent role T ALL
pathogenesis.
The MYC oncogene is one of the most frequently acti-
vated oncogenes in T ALL. The mechanism of MYC upreg-
ulation is mostly post-translational, although it can result
from t(8;14)(q24;q11) chromosomal translocation in 1% of
cases. Altered signal transduction programs also contribute
to T ALL oncogenesis. Gain-of-function mutations within
the JAK-STAT signaling axis are also a typical feature of T
ALL. Activating mutations target the IL7R gene (10%) encod-
ing the IL-7 receptor, JAK1, and JAK3 (10%) genes, as well
as more downstream effectors STAT5B (5–10%). Activating
mutations that predict hyperactivity of the MAPK pathway
are also reported and result from PTEN tumor suppressor
 Molecular basis of acute lymphoblastic leukemia 67

T-cell maturity

Early T precursor cell Early cortical Late cortical

Immunophenotype

sCD3 – +/– +
sCD4 – + +
sCD8 – + –
sCD1a – + –
sCD5 Weak
Myeloid marker +
Stem cell marker +
Genetic alterations

Notch1/FBXW7 + +++ ++
+/– +++ ++
CDKN2A

Myeloid tumor suppressors NUP214-ABL1 TAL1 activation

Hematopoietic genes Homebox genes PTEN deletion
Epigenetic regulators PHD finger protein 6
Cell signaling Wilms tumor 1
Protein tyrosine phosphatase
Non-receptor type 2
Fig. 5.4 Genetic abnormalities across T acute lymphoblastic leukemia (ALL) subtypes.

loss, as well as activation of HRAS, KRAS, and PTPN11 onco-
genes. The finding of these molecular alterations has clearly
opened up opportunities for signal blockade therapies which,
although still in the early phase of pre-clinical and clinical
testing, hold tremendous promise.
Further insights into the molecular pathogenesis of T ALL
have been garnered from recent genome-wide and transcrip-
tomic studies. These efforts have led to the recognition of
novel molecular oncogenic subtypes corresponding to devel-
opmentally distinct stages of T ALL which are underpinned
by unique biology (Figure 5.4). A major T ALL entity iden-
tified is the early T-cell precursor (ETP) subgroup, which
exhibits a unique constellation of clinical, cellular, and molec-
ular features. ETP ALL comprises 10–15% of childhood T
ALL and approximately 40–50% of adult T ALL. ETP ALL
is defined by characteristic immunophenotypic features typ-
ical of early immature (early cortical) thymocytes: absence of
CD1a, CD4, and CD8 expression, weak CD5, and aberrant
expression of at least one myeloid or stem-cell marker such
as CD34 and myeloid surface antigens (CD13 and CD33).
Compared with more developmentally mature T ALL (e.g.
cortical T ALL), the clinical outcome of this group is poor.
Large-scale sequencing data have provided major insights
into the molecular mechanisms underlying this unique clini-
cobiological subgroup. Transcriptionally, ETP T ALLs closely
relate to HSCs and myeloid progenitors. The spectrum of
acquired somatic alterations are distinct from classical T ALL
genetic alterations of transcription factor oncogene activa-
tion. The mutations identified affect three principal path-
ways, namely loss-of-function mutations in transcription
factor–encoding genes involved in hematopoietic develop-
ment (e.g. RUNX1 IKZF1, ETV6 GATA3 EP300; ∼50%), acti-
vating mutations in factors mediating cytokine receptor and
cellular signaling axis (NRAS, KRAS JAK1, JAK3, IL7R FLT3),
and inactivating mutations in several epigenetic modifiers
(PRC2, a H3K27 trimethylase, and loss of function of EZH2,
IDH1, IDH2, DNMT3A). Notably, some of the reported
genetic alterations correspond to mutations in myeloid-
specific oncogenes and tumor suppressor genes, supporting
the demonstration of a myeloid gene expression program in
ETP ALL. A particular prominence of aberrant activation
of JAK-STAT signaling in ETP tumors due to IL7 receptor
and JAK1 and JAK3 as well as NRAS/KRAS pathway muta-
tions suggests that these are rational targets for therapeutic
intervention.
Crucially, genome-wide studies of T ALL have high-
lighted the increasing importance of the non-protein cod-
ing regions in leukemogenesis. Several T ALL–specific non-
coding RNAs which are under the direct control of NOTCH1
have been identified which can contribute to the oncogenic
state of T ALL. In addition, recurring chromosomal duplica-
tions distal to MYC have also been reported. Such alterations
create an enhancer for oncogenic NOTCH1, which in turns
drives MYC expression. Therefore, there is a broadening
68 Molecular Hematology
genomic basis for the core oncogenic programs implicated
in T ALL oncogenesis, namely NOTCH1 and MYC, which
extends beyond just an intergenic cause. Recently, genomic
profiling studies have identified recurrent mutations in ribo-
somal protein genes, in particular RPL10, RPL5, and RPL11,
and CCR4–NOT transcription complex subunit 3 (CNOT3),
which are thought to contribute to T ALL development by
increasing basal levels of the protein synthesis that facilitates
downstream malignant transformation.
Ph-like ALL
Ph-like ALL is a recently characterized molecular subgroup
in which the underlying gene expression signature is simi-
lar to that of Ph-positive ALL, but where a detectable t(9;22)
chromosomal rearrangement or BCRABL1 oncogene is lack-
ing. The incidence of Ph-like ALL increases with age, from
10% in children to up to 27% in young adults. The common
features of this uniquely identified subgroup are poor clinical
outcome and enrichment of CRLF2 rearrangement and over-
expression (∼50%), plus IKZF1 gene deletion (∼30%). The
identification of Ph-like ALL has led to fervent studies into
the underlying signaling pathway alterations mediating the
Ph-like gene expression signature.
Apart from identification of JAK1/2 mutations and rear-
rangement of CRLF2, both of which are highly preva-
lent in Ph-like ALL, a number of novel kinase-activating
mutations have been identified. These include kinase-
activating gene fusions (∼12%) involving PDGFRB, ABL1,
ABL2, and CSF1R; gene rearrangement of JAK2 (∼7%)
and EPOR (∼3%); other mutations in JAK-STAT signal-
ing (∼12%, FLT3, IL7R, SH2B3, JAK1/3, TYK2, IL2RB,
and TSLP); in addition to Ras pathway mutations (5%,
KRAS, NRAS, NF1, PTPN11, and BRAF). The importance
of delineating the underlying activating mutations of Ph-
like ALL is underscored by the potential tractability of
these pathways to treatment intervention. For instance, Ph-
like ALL would theoretically be amenable to inhibitors
of the JAK pathway (CRLF2, JAK2, and EPOR) or tyro-
sine kinases such as dasatinib (ABL1, ABL2, PDGFRB, and
CSF1R). Although large-scale data demonstrating the effec-
tiveness of genetics-based treatment intervention in Ph-
like ALL are currently lacking, the identification of Ph-
like ALL offers real potential for novel treatments in this
highly distinct poor-risk group. Standardized diagnostics to
facilitate Ph-like ALL identification will be a necessary devel-
opment to realize the potential for optimizing therapy in
these patients.
Biological basis of relapsed ALL
Recurrence of ALL during or following primary ther-
apy carries an extremely dismal prognosis, with long-term
survival estimated at just ∼7%. The biological mechanisms
giving rise to treatment resistance and eventual therapy
failure have been greatly informed by increasing in-depth
molecular characterization of progressive ALL. In particu-
lar, the analysis of matched diagnosis and relapse pairs and
of xenografted primary ALL tumors has defined key causal
mechanisms contributing to leukemia clone evolution and
the specific genomic drivers of relapse (Figure 5.5). These
studies reveal that in the majority of cases, relapses are not
direct descendants of the original tumor, but are rather the
evolved products of a pre-ancestral clone that was concur-
rently present (at a low level) in the primary tumor. This
is consistent with a model of intratumoral clonal diversity,
where, rather than a single genome or clone, there are multi-
ple tumor genomes/clones present within a tumor.
Detailed evaluation of genetic patterning in sequential
samples together with xenografted primary ALL tumors
reveals that instead of a linear sequence of mutational events,
ALL clones appear to have evolved by a more branching
subclonal diversification, resulting in a highly variegated
subclonal architecture comprising multiple variant subclones
which may share only minimal genetic relatedness. This
branching clonal architecture, realized from the molecular
genetics of ALL, has now become a major tenet across the
wider cancer field. This model predicts that under certain
selective pressures – that is, those imposed for instance
by cytotoxic therapy – variant subclones possessing a par-
ticular fitness advantage will survive to further diversify
and/or cause relapsed disease. The precise determinants of
“fitness” – that is, whether it reflects an advantage con-
ferred by a particular mutant genotype versus epigenetic
or microenvironmental interaction, or indeed whether it is
entirely stochastic – continue to be debated, with evidence
currently for all of these.
From the point of view of the key genomic drivers of
disease relapse, several have so far been identified. The main
driver mutations emerging at relapse include deletions of
cell cycle regulator CDKN2A/B and lymphoid transcription
factors ETV-6, IZKF1 ERG, SPI1, TCF4, and TCF7L2,
tumor suppressors (TP53), RAS signaling, plus chromatin
modifiers SETD2 (histone methyltransferase) and CREBBP
(histone acetyltransferase), the latter impairing response to
glucocorticoid response genes as well as nucleoside response
genes, NCOR1. These lesions can be subclonal at diagnosis,
in keeping with the evolutionary theory of cancer, where
mutant-bearing fitter clones survive chemotherapeutic tar-
geting. Potentially, the a priori knowledge of relapse-specific
mutations at diagnosis may in the future inform the applica-
tion of upfront therapies (cytokine blockade and drugs that
modulate histone marks, e.g. histone demethylases, histone
deacetylases, and epigenetic readers targeting bromodomain
proteins) aimed at eliminating relapse-initiating clones
which may mitigate against ALL recurrence.
 Molecular basis of acute lymphoblastic leukemia 69

Therapy

Branching
clonal
architecture

Emergence
of minor
(ancestral)
subclone

Diagnosis Relapse

Gene alterations

Major Ancestral CREBBP

Minor subclones Drug
clone clone NTC52 resistance
NCOR1
TP53
IKZF1

Fig. 5.5 Molecular mechanisms underlying acute lymphoblastic leukemia relapse.

sensitive quantitative technology for enumeration of MRD

Molecular evaluation of Minimal titers.
Residual Disease The major modality for measuring MRD in ALL is
polymerase chain reaction (PCR) analysis of the Ig/TCR
A principal feature of ALL reflecting the cellular origin of
rearrangements or chimeric fusion gene targets resulting
the disease – a B- or T-cell precursor – is rearrangement of
from chromosomal translocations (e.g. BCR-ABL), although
genes within the Ig and TCR locus. This is a normal physi-
immunological methods utilizing flow cytometry are
ological process active in developing lymphocytes and leads
increasingly being applied particularly outside Europe. Since
to antibody diversity. While Ig/TCR gene rearrangements in
Ig/TCR rearrangements occur in the majority (>90%) of pre-
ALL carry no pathophysiological consequence, their occur-
cursor B ALL (pre-B ALL) and T ALL patients, this approach
rence can be exploited for diagnostic purposes to measure
is widely applicable. Ig/TCR MRD quantification relies on
for the presence of minimal residual disease (MRD). The
the rearrangement and subsequent joining of gene compo-
concept of MRD arises from known limitations in the con-
nents within the V, D, and J regions of the immunoglobulin
ventional assessment of disease response, where despite <5%
and T-cell receptor locus. The joining of component V, D,
blasts by cytomorphology, signifying attainment of com-
and J genes results in loss of original nucleotides as well as
plete remission, there can remain up to 1010 submicroscopic
the de novo insertion of nucleotides (N region insertion)
tumor cells present, defined hereafter as the MRD. As the
at the junctional regions. The co-occurring insertion and
majority of treatment failures occur in patients achieving
loss of nucleotides at the junctional sites are unique to each
complete remission (CR), this points to the limited utility of
lymphocyte or lymphocyte clone. In the case of a leukemic
cytomorphology-defined CR for predicting disease relapse,
clone, these junctional regions serve as leukemic signatures
and highlights the potential clinical relevance of MRD.
or unique fingerprints of the patient’s leukemia cell, which
A very large body of evidence confirms the clinical impact
can be amplified by PCR using breakpoint-specific primers
of MRD in ALL and its role as a prognostic indicator of
for MRD assessment. The aim is to identify at least two MRD
relapse risk. Consequently, MRD evaluation is now routinely
PCR targets per patient. This is achieved by first determining
adopted for risk stratification in frontline protocols to delin-
the presence of clonal Ig and TCR gene rearrangements in a
eate the potential for either treatment success or treatment
diagnostic sample by heteroduplex analysis of PCR products.
failure. Measurement of MRD requires a trackable marker
This method relies on denaturation of PCR products, fol-
that is highly specific to the leukemia clone, combined with
lowed by rapid cooling to induce homoduplex (bearing PCR
70 Molecular Hematology
products with identical junctional regions) and heterodu-
plex (bearing PCR products with heterogeneous junctional
regions) formation, corresponding to clonal and polyclonal
cell populations, respectively. Homoduplexes and heterodu-
plexes are separated on a polyacrylamide gel, where the
former migrate rapidly and the latter more slowly, forming a
background smear. Homoduplexes corresponding to clonal
IG/TCR rearrangements are then sequenced to identify
junctional regions. Thereafter, PCR primers can be designed
complementary to the junctional regions and tested for sensi-
tivity and specificity for clone amplification in serially diluted
diagnostic DNA. Detection and quantitation of MRD after
treatment can then be performed by standard curve gener-
ation from the diagnostic dilution series using a real-time
fluorescence reporter PCR system. The main drawback of
Ig/TCR-based MRD quantification is the occurrence of con-
tinuing rearrangements of the original clone during the dis-
ease course, which can lead to false negative results. This can
be reduced by careful primer design, which avoids amplifica-
tion of gene segments involved in continuing gene rearrange-
ments and the use of at least two MRD targets per patient.
In the case of chromosomal aberrations, the resultant
fusion genes can be used as reverse transcriptase PCR (RT
PCR) targets for MRD monitoring. The Philadelphia chro-
mosome, t(9;22), and resultant BCR/ABL fusion gene is by
far the most commonly assessed target for MRD monitoring
in this category. In contrast to antigen gene receptor targets,
these targets are not subject to clonal evolution and remain
stable throughout disease course. Excellent sensitivities of
10−4 –10−6 can be reached. A disadvantage of this method is
false positive results due to contamination. In addition, only
a minority (<40%) of patients carry these targets, thus appli-
cation is limited. Furthermore, RNA and therefore fresh sam-
ples (<48 hours) are required for reliable results.
Conclusion
Insights into the genetic landscape of ALL have paved the
way for defining new and important clinical subgroups where
the genetics not only inform the clinical behavior, but also
importantly reveal opportunities for targeted treatments.
Novel molecular subgroups in ALL defined by dysregu-
lated cellular signaling and epigenetic perturbation are key
findings that have informed the field both scientifically and
clinically. Despite clear advances in our understanding, a
notable area for further development is adult ALL, where
large-scale characterizations have yet to be performed.
Such endeavors should provide a comprehensive under-
standing of the landscapes defining a highly heterogeneous
disease.
Further reading
Hunger, S.P. and Mullighan, C.G. (2015). Redefining ALL classifica-
tion: toward detecting high-risk ALL and implementing precision
medicine. Blood 125: 3977–3987.
Iacobucci, I. and Mullighan, C.G. (2017). Genetic basis of acute lym-
phoblastic leukemia. J. Clin. Oncol. 35 (9): 975–983.
Chapter 6 Chronic myeloid leukemia
Hagop Kantarjian, Jorge Cortes, Elias Jabbour & Susan O’Brien
Department of Leukemia, University of Texas MD Anderson Cancer Center, Houston TX, USA
Introduction, 71
Epidemiology, 71
Clinical presentation and natural history, 71
Prognostic models, 73
Introduction
Chronic myeloid leukemia (CML) is a clonal myeloprolif-
erative neoplasm that arises from a pluripotent stem cell.
The Philadelphia (Ph) chromosome results from a recipro-
cal translocation between chromosomes 9 and 22, and con-
stitutes the cytogenetic hallmark of CML. It can be detected
in myeloid, erythroid, megakaryocytic, B, and sometimes
T lymphoid cells, but not in marrow fibroblasts. A criti-
cal milestone in CML research was the demonstration that
this translocation involved the ABL1 (v-abl Abelson murine
leukemia viral oncogene homolog 1) gene on chromosome
9 and the BCR (breakpoint cluster region) gene on chromo-
some 22, and resulted in the formation of the chimeric BCR-
ABL1 fusion transcript that encodes the constitutively active
BCR-ABL1 tyrosine kinase. The discovery that BCR-ABL1
plays a pivotal role in the pathogenesis of CML set the stage
for the development of therapeutic strategies aimed specifi-
cally at inhibiting this kinase and its downstream signals. In
this chapter we summarize the current knowledge regarding
the molecular biology of CML and the treatment modalities,
including novel BCR-ABL1 tyrosine kinase inhibitors (TKIs).
Epidemiology
CML has an annual incidence of 1.6 per 100 000 adults,
and is slightly more frequent in men (male to female ratio
1.4 : 1). CML represents approximately 15% of all leukemias
and accounts for up to 20% of all cases of adult leukemia
in Western societies. According to existing epidemiological
databases, the median age of onset of CML is 60–65 years
Molecular Hematology, 4th Edition. Edited by Drew Provan and John G. Gribben.
© 2020 John Wiley & Sons Ltd. Published 2020 by John Wiley & Sons Ltd.
Molecular biology, 73
Therapy, 74
Summary, 83
Further reading, 83
and the incidence increases with age. There are no known
causal etiologies; therefore, the disease is neither preventable
nor inherited. However, ionizing radiation is leukemogenic
and CML has been observed in individuals exposed to the
radiation of atomic bomb explosions in Japan in 1945. The
incidence of CML was 50-fold higher than in non-exposed
subjects and peaked approximately 10 years after the explo-
sion, although patients younger than 15 years of age devel-
oped CML earlier than those 30 years or older. In most cases
of CML no antecedent radiation exposure is discernible.
Clinical presentation and natural
history
CML evolves typically in three phases. Approximately 90%
of patients with CML are diagnosed in the chronic phase
(CML-CP), characterized by high numbers of immature
myeloid cells and mature granulocytes in the bone marrow
and peripheral blood. Patients diagnosed in CML-CP are
often asymptomatic. If symptoms are present, they usually
relate to the presence of splenomegaly (e.g. abdominal full-
ness, early satiety, pain). Other signs and symptoms that
may occur are anorexia, weight loss, fever, fatigue, or ane-
mia. Leukocytosis, with white blood cell (WBC) counts fre-
quently exceeding 100 × 109 l−1 , is frequently seen, but only
occasionally leads to signs and symptoms of hyperviscos-
ity (priapism, cerebrovascular accidents, dizziness, confu-
sion) or retinal hemorrhage. In CML-CP, CML cells retain
the ability to differentiate and produce morphologically nor-
mal blood elements capable of carrying out the physiologi-
cal functions of normal counterparts. In about 50% of cases,
CML-CP is diagnosed after a routine blood test for unre-
lated reasons. The peripheral blood smear in CML-CP is
characterized by a left shift (i.e. increasing numbers of
71
72 Molecular Hematology
younger progenitors), although blasts typically comprise less
than 5% of all WBCs. Basophilia is often present. The leuko-
cyte alkaline phosphatase activity is reduced both in inten-
sity and in the number of neutrophil band forms that stain
positively for this enzyme. The bone marrow aspirate and
biopsy are hypercellular and demonstrate granulocytic and
megakaryocytic hyperplasia, basophilia, and a blast per-
centage of 5% or less. Historically, the median survival of
patients in CML-CP was three to five years. Untreated, most
patients progressed to a chronic myeloid leukemia blastic
phase (CML-BP), characterized by a peripheral or bone mar-
row blasts percentage of 30% or more, and frequently pre-
ceded by an chronic myeloid leukemia accelerated phase
(CML-AP). Before the introduction of TKI therapy, the esti-
mated annual risk of transformation from CML-CP to CML-
BP was approximately 10% in the first two years after diag-
nosis and 15–20% thereafter. With TKI therapy, this annual
incidence was reduced to 2–3% in the first two years, and to
1% or less thereafter. Both CML-AP and CML-BP are char-
acterized by increasing arrest of maturation.
A variety of classifications have been used to define the
different CML phases. One classification defines CML-AP
as the presence of any of the following features: cytogenetic
clonal evolution, 15% or more blasts, 30% or more blasts
plus promyelocytes, 20% or more basophils, or platelets
lower than 100 × 109 l−1 unrelated to therapy (Table 6.1).
Although other classification systems have been proposed
for defining CML-AP and CML-BP, they have not been clin-
ically validated. Patients in CML-AP may present with fever,
night sweats, weight loss, or bleeding associated with throm-
bocytopenia. However, the transition from CML-CP to
CML-AP is usually subclinical and laboratory monitoring is
necessary to detect disease progression. The average survival
Table 6.1 Diagnostic criteria of accelerated phase according to MD Anderson Cancer Center (MDACC), International Bone Marrow Transplant
Registry (IBMTR), and World Health Organization (WHO)
MDACC IBMTR
Blasts 15–29% 10–29%
Blasts + promyelocytes ≥30% ≥20%
Basophils ≥20% ≥20%b
Platelets (×109 l−1 ) <100 Unresponsively high or persistently low
Cytogenetics CE CE
WBC NA Difficult to control, or doubling <5 days
Anemia NA Unresponsive
Splenomegaly NA Increasing
Other NA Chloromas, myelofibrosis
CE, clonal evolution; NA, not applicable; WBC, white blood cell count.
a In the WHO criteria, blastic phase is defined as a blast percentage of 20% or higher. In the MDACC and IBMTR, a definition of blastic phase
requires the presence of at least 30% blasts.
b Basophils plus eosinophils.
of patients in CML-AP was one to two years before the TKI
era, but with TKI therapy the estimated four-year survival
has increased to 60–70%. In contrast to CML-AP following
CML-CP, de novo CML-AP has a better prognosis with front-
line TKI therapy, with an estimated eight-year survival rate
of 80%.
A diagnosis of CML-BP requires the demonstration of at
least 30% blasts in the peripheral blood and/or the bone
marrow, or the presence of extramedullary blastic foci. The
World Health Organization (WHO) classification has pro-
posed that the blast percentage that defines BP be changed
from ≥30% to ≥20%, but this proposal has not been validated
in large studies using TKI. Immunophenotypically, CML-
BP can display either myeloid or lymphoid features. In lym-
phoid CML-PB, blast cells often express both lymphoid and
myeloid cell surface markers (40–50%). The CML-BP is lym-
phoid in 20–30% of patients, myeloid in 50%, and undiffer-
entiated in 20%; megakaryocytic (M7) or erythroblastic (M6)
in <5%. Cells from patients with lymphoid CML-BP exhibit
high levels of the enzyme terminal deoxynucleotidyl trans-
ferase (TdT). Most cases of lymphoid CML-BP arise from
progenitors of the B cell lineage, and express CD10, CD19,
and CD22. T-cell CML-BP has been rarely described. The
phenotype of myeloid CML-BP cells resembles that of acute
myeloid leukemia cells, with blasts staining with myeloper-
oxidase and expressing CD13, CD33, and CD117. Clini-
cally, CML-BP is characterized by prominent hypercatabolic
symptoms, related to increasing tumor burden, and resis-
tance to conventional chemotherapeutic agents. The median
survival of patients in CML-BP prior to the introduction of
TKI therapy was two to six months; with TKI- based therapy
it improved to a median of one to three years, particularly
when TKI is combined with chemotherapy.
WHO
10–19%a
NA
≥20%
<100 or >1000 Unresponsive
CE not at diagnosis
NA
NA
NA
Megakaryocyte proliferation, fibrosis

Prognostic models
The different CML phases predict for very different survival
probabilities. However, the prognosis is also variable among
patients in the same phase of the disease. Several patient and
disease characteristics have been found to be prognostically
useful and have been used to generate prognostic models. In
1984, Sokal and colleagues studied 813 patients with CML
collected from six European and American series published
in the 1960s and 1970s. The following hazard ratio function
for death was derived from baseline patient and disease
characteristics:
𝜆i (+) 𝜆0 (t) = Exp 0.0116 (age − 43.4)
+ 0.0345 (spleen − 7.51) + 0.188 [(platelets∕700)2
− 0.563] + 0.0887 (blasts − 2.10)
This function segregated patients into three prognostic
groups with hazard ratios of <0.8, 0.8–1.2, and >1.2, and
median survivals of 2.5, 3.5, and 4.5 years, respectively. The
Sokal prognostic score system was developed in the era
before interferon alpha therapy. The Hasford score, which
takes into account age, platelet count, peripheral blast count,
spleen size, eosinophils, and basophils, was developed to
identify risk groups in patients with CML treated with
interferon. Both risk models have been validated in several
studies of patients treated with TKIs, but the prognostic
significance of several clinical factors has changed. Clinical
factors that maintain prognostic significance in the era of
TKIs include high blast (≥10–15%) and basophil (≥20%)
percentages, massive splenomegaly, and chromosome abnor-
malities involving 17p-/isochromosome 17, chromosome 7,
or 3q26.2.
Genome-wide analyses of gene expression profiles have
been used as prognostication tools in multiple malignancies.
e1
e1’
e2’
m-bcr
e12
M-bcr
e16
e19
μ-bcr
Fig. 6.1 The Philadelphia chromosome and
associated BCR-ABL1 molecular abnormalities.
Chronic myeloid leukemia 73
In a study of DNA microarrays used to analyze gene
expression in 91 patients with CML, patients in CML-CP
or CML-BP revealed marked differences in gene expression.
There was a strong correlation in expression levels between
CML-AP and CML-BP (r = 0.81). Gene expression profiling
may facilitate a more comprehensive stratification of CML
that segregates genetically defined risk subgroups.
Molecular biology
BCR-ABL1 oncogene
The fusion of the ABL1 and BCR genes resulting from
the reciprocal translocation t(9;22)(q34;q11.2) gives rise to
the BCR-ABL1 oncogene that encodes for the constitutively
active BCR-ABL1 tyrosine kinase. Several experimental
models, such as BCR-ABL1-expressing CD34+ cells in culture
or retrovirally transduced BCR-ABL1-positive mouse cells,
have established a causal relationship between BCR-ABL1
and human leukemia. BCR-ABL1 can transform hematopoi-
etic stem cells, but not committed progenitors lacking self-
renewal capacity.
The breakpoint within ABL1 at 9q34 occurs in nearly all
patients upstream of exon 2, with some variability in the
specific site either upstream of exon Ib, downstream of exon
Ia, or more frequently between the two (Figure 6.1). Break-
points within BCR localize to three main breakpoint cluster
regions (bcr). In most patients with CML and in one-third of
those with Ph-positive acute lymphoblastic leukemia (ALL),
the breakpoint maps to the major breakpoint cluster region
(M-bcr), which spans BCR exons 12–16 (formerly called b1–
b5), giving rise to a fusion transcript with either e13a2(b2a2)
or e14a2(b3a2) junctions that translates into a 210-kDa
protein (p210BCR-ABL1 ). In two-thirds of patients with
Chromosome 22 Chromosome 9
1b
5’
BCR 1a
3’
a2
a3
5’
ABL
3’
a11
e1a2 p190bcr-abl
e13a2
p210bcr-abl
e14a2
e19a2 p230bcr-abl
16
74 Molecular Hematology
Ph-positive ALL and rarely in CML, the breakpoints within
BCR localize to an area of 54.4 kb between exons e2′ and e2,
termed the minor breakpoint cluster region (m-bcr), which
translates to a 190-kDa protein (p190BCR-ABL1 ). A third
breakpoint cluster region (μ-bcr) has been identified giving
rise to a 230-kDa fusion protein (p230BCR-ABL1 ), associated
with a more indolent CML course and with a phenotype
more akin to chronic neutrophilic leukemia (Figure 6.1).
BCR–ABL1 kinase-signaling pathways
BCR-ABL1 signals through an intricate network of molecular
pathways that promote diminished growth factor/adhesion
dependence, decrease apoptosis, and enhance proliferation.
Phosphorylation of BCR Tyr177 provides a high-affinity
docking site for the SH2 domain of growth factor receptor-
bound protein 2 (GRB2), which in turn recruits SOS (a
guanine-nucleotide exchanger of RAS), thus activating
RAS and the adapter GRB2-associated binding protein 2
(GAB2). BCR-ABL1-induced GAB2 phosphorylation acti-
vates the phosphatidylinositol 3-kinase (PI3K)/AKT and the
RAS/ERK pathways. BCR-ABL1 also phosphorylates the SRC
family of kinases (SFKs) HCK, LYN, and FGR, which leads
to activation of STAT5 (signal transducer and activation of
transcription 5). Upon dimerization, STAT5 translocates
to the nucleus and binds to cognate DNA sequences to
modulate gene transcription. STAT5 also upregulates the
anti-apoptotic protein BCLXL , which is repressed by the
tumor suppressor and negative regulator of granulocyte dif-
ferentiation ICSBP (transcription factor interferon consensus
sequence binding protein). The RAC subfamily of guanosine
triphosphatases (GTPases) RAC1, RAC2, and RAC3, as well
as the enzyme 12/15lipoxygenase (12/15-LO), have also been
identified as important elements in BCR-ABL1 downstream
signaling. Although BCR-ABL1 signals through a multitude
of downstream components, the kinase activity of this
oncogenic enzyme is central to the pathogenesis of CML.
This provided the rationale to develop therapeutic strategies
that inhibit the kinase activity of BCR-ABL1.
Therapy
General
Since 2000, the treatment of CML has changed dramatically.
Conventional chemotherapeutic agents such as busulfan
or hydroxycarbamide (hydroxyurea) are no longer used,
except to achieve initial or transient cytoreduction. The
maturing experience of imatinib mesylate and other TKIs
has brought about a change in the therapeutic algorithm
for CML. Today, five TKIs have been approved by the Food
and Drug Administration (FDA) to treat CML: imatinib,
dasatinib, nilotinib, bosutinib, and ponatinib (Figure 6.2).
Imatinib mesylate (Gleevec) is an orally bioavailable
2-phenylaminopyrimidine, relatively selective for the con-
stitutively active tyrosine kinase of the BCR-ABL1 fusion
protein. It also inhibits other kinases such as KIT, platelet-
derived growth factor receptor (PDGFR)α and PDGFRβ,
and ABL-related gene (ARG). Dasatinib (Sprycel) is a sub-
nanomolar inhibitor of BCR-ABL1 and SFKs, with potent
activity against KIT (IC50 13 nmol l−1 ), PDGFRβ (IC50
28 nmol l−1 ), and ephrin receptor EPHA2 (IC50 17 nmol l−1 )
tyrosine kinases. It is 350 times more potent than imatinib
in vitro. Dasatinib inhibits multiple imatinib-resistant BCR-
ABL1 mutant isoforms, except for T315I. Nilotinib (Tasigna)
is a phenylaminopyrimidine derived from the crystal struc-
ture of imatinib in complex with ABL1 kinase. Nilotinib has
30–50 times improved affinity and inhibitory activity against
unmutated BCR-ABL1 and inhibits the tyrosine kinase activ-
ity of 32 of 33 BCR-ABL1 mutants tested, the exception being
T315I. Bosutinib (Bosulif) is a potent SFK and ABL1 kinase
inhibitor with negligible activity against KIT or PDGFR,
which may result in an improved toxicity profile, with less
myelosuppression (no inhibition of KIT) and fewer pleural
effusions (no inhibition of PDGFR). Ponatinib (Iclusig) is a
third-generation BCR-ABL1 multikinase inhibitor, and the
only one which is effective against T315I, the gatekeeper
mutation. Ponatinib inhibits FLT3, FGFR, VEGFR, PDGFR,
and c-KIT. Omacetaxine mepesuccinate (Synribo) is a non-
TKI protein synthesis inhibitor approved by the FDA for the
treatment of CML following failure of two TKIs. Allogeneic
stem cell transplantation (SCT) remains a curative modality
and a valid option for patients who fail TKI therapy. The
structures of the five TKIs are shown in Figure 6.2.
Several general principles are emerging in CML therapy.
One is the importance of adherence to treatment, which is
critical for the success of therapy. Patients who are less than
90% adherent to TKI therapy as prescribed have a signifi-
cantly lower probability of achieving the deepest molecular
responses compared with more compliant patients.
A second principle is the importance of daily continuous
uninterrupted TKI therapy in both the chronic phase (single-
agent TKI; possible addition of other agents if hematologic
resistance and lack of better treatment alternatives), or in
transformation (CML-AP–TKI alone or in combination with
other standard therapies, e.g. hydroxyurea, cytarabine, azaci-
tidine, or decitabine; CML-BP–TKI combinations with acute
leukemia chemotherapies). TKI therapy is superior to non-
TKIs (e.g. hydroxyurea), even when responses less than com-
plete cytogenetic response (CCyR) are achieved. In CML-BP
the use of TKI in combination with chemotherapy is superior
to single-agent TKI or chemotherapy alone, even after failure
of TKI or chemotherapy. The only exceptions to not using
TKIs in chronic or advanced CML phases are low-risk CML
as maintenance after allogeneic SCT; CML with sustained
undetectable BCR-ABL1 transcript levels beyond two to
three years, where TKI treatment discontinuation may be
 Chronic myeloid leukemia 75

N CF3
H H
N N N N
CH3 O
H
N O N N N N
H N
N

N
Imatinib Nilotinib (AMN107)
N

CI CI
H
N N S O CI
HN OCH3
N N
HN
H3CO CN
N
N O N
N N
H3C
HO Dasatinib (BMS-354825) Bosutinib (SKI-606)

N CH3
N H N
N O N

N CF3
Ponatinib
Fig. 6.2 Chemical structures of imatinib and second/third-generation tyrosine kinase inhibitors used to treat chronic myeloid leukemia.

considered (treatment-free remission or molecular cures,
discussed later); or persistent and intractable cytopenias
related to TKIs not alleviated by optimization of the TKI
dose schedules.
A third general principle is to avoid discarding or changing
a TKI unless there is evidence of loss of CCyR (rather than
loss of major molecular response, MMR), or unless there is
severe toxicity or chronic moderate toxicity affecting qual-
ity of life not alleviated with optimization of the TKI dose
schedule (dose reduction) and symptom management.
Because of the availability of five TKIs and the ease of chang-
ing therapy, there is a tendency to change TKIs in situa-
tions of perceived suboptimal response (e.g. increased BCR-
ABL1 transcripts to more than 0.1%, termed loss of MMR) or
because of toxicities which could be minimized or eliminated
with TKI dose reductions and proper medical management.
This could result in discarding TKIs which could still offer
clinical benefit to patients.
Frontline CML therapy
Based on promising preclinical data and efficacy in phase
I and phase II trials in CML salvage (after interferon fail-
ure), the phase III randomized multinational IRIS (Interna-
tional Randomized study of Interferon-alpha plus cytarabine
versus STI571) study compared imatinib 400 mg orally daily
with interferon-alpha in combination with cytarabine in 1106
newly diagnosed patients with CML-CP. After a follow-up of
eight years, the CCyR rate with imatinib was 87%, the MMR
(BCR-ABL1 transcript levels by International Standard [IS]
<0.1%) was 70%, and the rate of undetectable BCR-ABL1
transcripts was 40%. The estimated 10-year survival rate was
83%. The annual rate of progression from CML-CP to CML-
BP was 1.5–2% in the first two years and less than 0.5% there-
after. In the IRIS trial, the depth of response achieved on ima-
tinib therapy was the most important prognostic factor. The
achievement of CCyR at 12 months and beyond was associ-
ated with a significant survival benefit. Achieving MMR was
associated with improved event-free survival (EFS) among
patients who had achieved CCyR, but not with improved sur-
vival, because of the availability of effective salvage thera-
pies if implemented optimally as soon as there is evidence
of cytogenetic relapse. More recently, achievement of unde-
tectable BCR-ABL1 transcripts sustained for at least two
years has offered the possibility of treatment discontinuation
(discussed later).
The DASISION study was a phase III randomized trial
that compared dasatinib 100 mg daily with imatinib 400 mg
daily in 519 patients with newly diagnosed CML-CP. The
confirmed CCyR rate at 12 months (primary study endpoint)
significantly favored dasatinib. In the final five-year study
results, the cumulative rate of MMR by five years was 76%
with dasatinib and 64% with imatinib (p = 0.002); the cumu-
lative rate of MR4.5 (equivalent to a 4.5 log reduction in
76 Molecular Hematology
BCR-ABL1 transcripts from baseline; BCR-ABL1 transcripts
[IS] less than 0.0032%) was 42% versus 33% (p = 0.025). The
estimated five-year survival rate was 91% with dasatinib and
90% with imatinib. The rate of transformation to CML-AP
or CML-BP by five years was 4.6% with dasatinib and 7.3%
with imatinib.
In another multicenter trial, the North American Coop-
erative Groups randomized patients with newly diagnosed
CML-CP to dasatinib 100 mg once daily or imatinib 400 mg
once daily. Similar to the results of the DASISION study,
dasatinib resulted in a higher rate of CCyR compared with
imatinib (84% vs. 69%, p = 0.04).
A third phase III randomized study, SPIRIT 2, also com-
pared dasatinib 100 mg daily to imatinib 400 mg daily. The
12-month MMR rate was 58% with dasatinib and 43% with
imatinib (p < 0.001). The 12-month CCyR rates were 51%
versus 40% (p < 0.002). The progression rate to CML-AP was
0.5% with dasatinib and 0.7% with imatinib and to CML-BP
1% and 1.7%, respectively.
The ENEST-nd was a large phase III international ran-
domized study that compared two doses of nilotinib, 300 mg
twice daily and 400 mg twice daily (the dose approved for
CML salvage), to imatinib 400 mg once daily. The rate of
MMR by 12 months was 43–44% with nilotinib and 22%
with imatinib (p < 001). The cumulative incidence of CCyR
by 24 months was 85–87% with nilotinib and 77% with
imatinib (p < 0.001). With a minimum follow-up of five
years, the cumulative incidence of MMR by five years was
77% with both nilotinib schedules and 60% with imatinib
(p < 0.0001%). The incidence of BCR-ABL1 transcripts
≤0.0032% IS (MR4.5) by six years were 55–56% with nilo-
tinib and 33% with imatinib (p < 0.0001%). The incidence
of transformation to CML-AP/BP was 3.9% with nilotinib
300 mg twice daily, 2.1% with nilotinib 400 mg twice daily,
versus 7.4% with imatinib (p = 0.06 and 0.003, respectively).
However the definition of CML-AP used the pre-TKI
criteria, which are now known to be less relevant. The inci-
dence of CML-BP (a more relevant clinical endpoint) with
nilotinib versus imatinib was not reported. The estimated
five-year EFS rates were 95%, 97%, and 93%, respectively.
When considering Sokal-risk category, nilotinib was more
effective in reducing the transformation rate in higher-risk
Sokal disease. The rates of transformations were 1%, 1%,
and 0% among patients with low Sokal risk treated with
nilotinib 300 mg orally twice daily, nilotinib 400 mg orally
twice daily, or imatinib 400 mg orally once daily. The rates
were 2%, 1%, and 10% among patients with intermediate-
risk Sokal, and 9%, 5%, and 11% among patients with
high-risk Sokal. The estimated five-year survival rates were
94% with nilotinib 300 mg twice daily, 96% with nilotinib
400 mg twice daily, and 92% with imatinib once daily. The
survival rates in patients with low-risk Sokal scores were
97%, 99%, and 100% respectively. In intermediate-risk
Sokal group they were 94%, 97%, and 89%, respectively. In
the high-risk Sokal cohort they were 89%, 91%, and 84%,
respectively.
A second trial from China used the same design and
enrolled 267 patients. The MMR rate at 12 months was 52%
with nilotinib and 28% with imatinib. The rates of CCyR
(84% versus 87%) and progression-free survival (PFS; 95%
each) were similar at 24 months. The rates of progression to
CML-AP/BP and of survival were similar with nilotinib and
imatinib.
While nilotinib therapy was well tolerated, there was a
cumulative increased risk of cardiovascular events. In the
ENEST-nd trial, the six-year cumulative risk of a cardiovas-
cular event was 9.9% with nilotinib 300 mg twice daily, 15.9%
with nilotinib 400 mg twice daily, and 2.5% with imatinib
400 mg once daily. Therefore, any potential added benefit of
using nilotinib as initial therapy in the intermediate- and
high-risk Sokal groups should be weighed against the cumu-
lative risk of cardiovascular complications.
At the MD Anderson Cancer Center (MDACC), several
frontline studies ran sequentially and in parallel, and eval-
uated the outcomes of patients with newly diagnosed CML
treated with imatinib 400 mg daily (n = 68), imatinib 800 mg
daily (n = 200), dasatinib 50 mg twice daily or 100 mg daily
(n = 106), and nilotinib 400 mg twice daily (n = 108). An
analysis of the total experience of 482 patients is summarized
in Table 6.2. The experience showed similar benefits with
high-dose imatinib and second-generation TKIs in relation
to early surrogate endpoints of long-term outcome (CCyR,
MMR, MR4.5). However overall survival was not signifi-
cantly different by treatment strategy. The table shows the
cumulative incidence of outcomes (usually reported in the lit-
erature as outcome by a certain time), as well as the incidence
of outcomes at particular times, a more accurate reflection of
the benefit of TKIs.
Selection of frontline TKI therapy
Current CML treatment guidelines offer imatinib 400 mg
daily, dasatinib 100 mg daily, or nilotinib 300 mg twice daily
(on an empty stomach) as frontline treatment options in
CML. With the availability of generic imatinib, the choice of
frontline TKI therapy may depend on several pre-treatment
considerations, as well as the endpoints of therapy. In 2019,
the annual cost of generic imatinib is about $4000 for two
(among nine) least costly generics in the United States,
$3000–8000 in Canada, and $400 in India. Survival is sim-
ilar with all three TKIs, therefore generic imatinib may
be the preferred treatment of choice for many patients. If
second-generation TKIs are chosen over generic imatinib
(estimated annual cost about 30% of patented imatinib), the
incremental cost-effectiveness ratio (ICER = annual ratio of
additional cost over incremental benefit) is about $800 000/

Table 6.2 The MD Anderson Cancer Center long-term experience of chronic myeloid leukemia frontline therapy with tyrosine kinase inhibitors
% Response parameter Imatinib 400 mg
%Cumulative CCyR by 5 years 85
%CCyR at 5 years 87
% MMR at 5 years 76
% Cumulative MR4.5 by 5 years 56
% MR4.5 at 5 years 57
Compared with imatinib 400 mg daily, the 5-year event-free survival rate was significantly higher with imatinib 800 mg (p = 0.029), dasatinib
(p = 0.003), and nilotinib (p = 0.031). However, there were no differences in the rates of 5-year failure-free survival (p = 0.32, p = 0.075,
p = 0.332), transformation-free survival (p = 0.053, p = 0.038, p = 0.493), or overall survival (p = 0.563, p = 0.162, p = 0.981).
CCyR, complete cytogenetic response; MMR, major molecular response; MR4.5, see text.
quality-adjusted life-years (QALY), significantly higher than
the accepted ICER of $50 000/QALY.
Beyond survival, several factors may be considered in
deciding on TKI therapy: (i) the incidence of durable
undetectable BCR-ABL1 transcripts for longer than two to
three years (which may offer the option of treatment discon-
tinuation and treatment-free remissions); (ii) the CML Sokal
risk; (iii) the out-of-pocket expenses to the patient; (iv) the
patient’s age; and (v) the pre-treatment medical comorbidi-
ties (e.g. hypertension, diabetes, chronic lung disease, pan-
creatitis) and the associated risks for serious adverse events
(such as arterio-thrombotic events, pleural effusions, pancre-
atitis, pulmonary hypertension).
The incidence of treatment-free remissions, discussed
later in the chapter, is estimated at 20–25% with second-
generation TKIs and 13–15% with imatinib. This therapeutic
endpoint may be more important among young patients (e.g.
age ≤50–55 years or younger) in whom treatment discontin-
uation after 5–10 years is relevant (to avoid long-term treat-
ment costs and side effects) compared with older patients
(e.g. age ≥65–70 years or older). Second-generation TKIs
may potentially improve outcome in patients with high-risk
Sokal CML (10–20% of CML) and might be chosen for these
patients as frontline therapy.
Out-of-pocket expenses are an increasing burden on
patients, at least in the US. With 25% out-of-pocket expenses
and a second-generation TKI annual cost of $150 000+, out-
of-pocket expenses could exceed $25 000 annually, close
to half the annual income of an average family. This may
decrease patient compliance, which would be associated
with a worse outcome. In such instances, generic imatinib
may reduce costs, improve compliance, and even provide
better results (better compliance) than second-generation
TKIs.
Some relative contraindications or caution should be
exerted based on prior medical history. Dasatinib should be
avoided in patients with a history of chronic lung disease or
pleural effusions. Nilotinib should be avoided in patients with
Chronic myeloid leukemia 77
Imatinib 800 mg Nilotinib Dasatinib
90 99 97
90 93 96
86 91 90
74 80 78
74 71 71
a previous history of cardiovascular event, pancreatitis, or
severe diabetes.
Discontinuation of TKI therapy and
treatment-free remissions
Several studies have shown that TKI discontinuation in
patients with CML-CP on TKI therapy who have sustained
undetectable BCR-ABL1 transcripts for at least two years
results in treatment-free remissions in 40–60%. The inci-
dence of MR4.5 (two years or more) on frontline TKI ther-
apy is about 42% with second-generation TKIs, and 26%
with imatinib therapy (this increases to 40% with sequential
second-generation TKIs). Therefore, the rates of treatment-
free remissions range from 13% to 21%. Treatment discon-
tinuation, when attempted in community practice, should
adhere to strict criteria to avoid harm to patients. Patient
must have low- or intermediate-risk disease (not high-risk),
a documented history of only chronic phase disease (no evi-
dence of transformation), and an optimal response to front-
line TKI therapy. The combined duration of TKI treatments
should be at least five or more years and undetectable BCR-
ABL1 transcripts should be documented for at least two
to three years, with assessments every six months during
that period. Patients must have quantifiable BCR-ABL1 tran-
scripts (e13a2, e14a2) and comply with frequent monitoring
by polymerase chain reaction (PCR) every one to two months
in the first two years after TKI discontinuation, and every
three to six months thereafter. A recent analysis indicated
that, at current prices, second-generation TKIs do not offer
a good treatment value for the endpoint of treatment-free
remission.
Treatment endpoints in CML and other
endpoints of evolving “treatment value”
The achievement of CCyR (Ph-positive metaphases 0%;
BCR-ABL1 transcripts [IS] ≤1%) by/at 12 months and later
78 Molecular Hematology
on TKI therapy is associated with a significant survival
benefit compared with achievement of lesser degrees of
response. Therefore, achieving CCyR is the primary end-
point of TKI therapy. Achievement of MMR (BCR-ABL1
≤0.1% IS) is associated with modest improvements in EFS,
but not with a survival benefit. Achievement of sustained
undetectable BCR-ABL1 transcripts offers the possibility of
treatment discontinuation. Lack of achievement of MMR or
of MR4.5 should not be interpreted as a need to change TKI
therapy or to consider allogeneic SCT. Other measurements
of outcome include EFS, transformation-free survival (TFS),
and PFS. Event is defined differently in various studies and
could include instances of treatment discontinuation that
have no relevance to survival outcome (e.g. discontinuation
for toxicity or non-compliance; death from causes unrelated
to CML, etc.). Transformation encompasses CML-AP, and
many of the AP criteria have lost their prognostic relevance
on TKI therapy.
At MDACC, the major treatment milestones are at six
and twelve months. Patients with BCR-ABL1 transcripts (IS)
>10% at six months, or not achieving CCyR with BCR-ABL1
transcripts (IS) ≤1% at one year, or with loss of response at
any later time, are offered a change of TKI therapy if possi-
ble. Patients on second TKIs with BCR-ABL1 transcripts (IS)
>10% at six months should be observed closely for signs of
progression, but cannot be changed to “better” TKIs in gen-
eral, and have a better survival with continued TKI therapy
compared with proceeding to allogeneic SCT (five-year sur-
vival rates 85–90% vs. 65–70%). TKI therapy is not changed
in patients in CCyR but not in MMR.
All key randomized trials of imatinib compared with
second-generation TKIs as initial therapy for CML-CP have
shown a significant benefit of surrogate response endpoints,
but similar survivals with second-generation TKIs versus
imatinib. Generic imatinib is now available at a low annual
cost. Using second-generation TKIs as frontline therapy over
imatinib results in an ICER value of $800 000/QALY, an
excessive “treatment value.”
There has been a recent emphasis on treatment endpoints
other than survival. The three key ones include (i) achieve-
ment of MMR; (ii) achievement of early treatment response
(BCR-ABL1 transcripts <10% by three months of TKI ther-
apy); and (iii) achievement of durable undetectable BCR-
ABL1 transcripts for at least two years, as a condition for
treatment-free remission. These endpoints have been empha-
sized in independent continuing medical education meetings
and symposia, and by CML experts. Therefore, it is important
to explore in depth the supportive data.
Achievement of “deeper molecular responses” (MMR) was
suggested as an important treatment endpoint that may cor-
relate with improved survival. This implied that since MMR
rates are higher with second-generation TKI compared with
imatinib, they should be used as a better frontline therapy.
However, several long-term follow-up studies have shown
that, among patients achieving CCyR, survival was similar
whether they also achieved MMR or lesser degrees of molec-
ular response.
Early response after three months of therapy was also
reported to be an important treatment landmark associated
with differences in survival (discussed later). This implied
that at least 30% of patients on frontline imatinib would be
switched to second-generation TKIs after only three months
of imatinib. As discussed shortly, the potential change of ther-
apy after six months for lack of early response is more appro-
priate and would apply to only 8% of patients.
More recently there has been an emphasis on TKI treat-
ment discontinuation and treatment-free survival, an impor-
tant therapeutic endpoint. However, the cost-effectiveness
analysis of such an approach, assuming treatment-free remis-
sions of 20–25% with second-generation TKIs and 13–15%
with imatinib, did not show a good treatment value for
second-generation TKIs at their current prices. A more rea-
sonable annual price for second-generation TKIs is needed
(perhaps $10 000–20 000 per year rather than $150 000+) to
offer a good treatment value for this endpoint.
Monitoring and determining treatment
failure
At baseline, a marrow examination is needed to establish the
diagnosis, determine the percentage of blasts and basophils
(10% of patients may have higher percentages in the marrow
vs. blood that could change their CML phase), and perform
cytogenetic analysis (confirm the presence of the Philadel-
phia chromosome and assess for additional chromosomal
abnormalities, or clonal evolution). A pre-treatment fluores-
cence in-situ hybridization (FISH) analysis (assess Ph pos-
itivity) and PCR (measure BCR-ABL1 transcripts) help to
avoid later false negative tests if alternative transcripts were
present initially (2–3% of cases).
The current recommendation that follow-up marrow stud-
ies be done at 3, 6, and 12 months after starting therapy may
not be necessary. An alternative is to evaluate response by
FISH and PCR on peripheral blood. If a patient is responding
optimally, and the FISH study is negative at 6 or 12 months
and/or BCR-ABL1 transcripts (IS) <1%, it may be reasonable
to omit marrow examinations, as the patient is likely to be in
CCyR.
For patients in CCyR receiving TKI therapy, periodic
molecular monitoring using FISH and PCR every six months
is acceptable. FISH and PCR are used as complementary
diagnostic tests to assess for response concordance (to detect
possible false positive/high or negative/low values generated
by either assay). For patients in CCyR and MMR, a PCR
test every six months is sufficient to monitor response. For
patients with BCR-ABL1 transcripts between 0.1% and 1%
and negative FISH analysis, more frequent monitoring every
three months may be of value.

Early molecular response has been shown to have strong
prognostic value. In a study from the UK, patients with newly
diagnosed CML on imatinib therapy who achieved BCR-
ABL1 transcript levels >10% at three months had an eight-
year survival rate of 54%, compared with 93% for those with
lower levels. However, patients did not have the option of
an early change to second-generation TKIs. At MDACC, a
similar analysis showed 10-year survival rates of 94% ver-
sus 98%; these patients accessed second-generation TKIs at
the first evidence of loss of CCyR. In the ENEST-nd and
DASISION trials, the differences in survival rates were 86–
87% versus 97%. These data suggested the need for a change
from imatinib to second-generation TKIs at three months
among patients with BCR-ABL1 transcripts (IS) >10%. This
would require a change of therapy in 30% of patients on
frontline imatinib therapy, and may not be appropriate con-
sidering that it is a one-time test which has a high rate of
variability around the PCR 10% value, and may be influ-
enced by patient compliance and treatment interruptions
over a short period of three months. Analysis from Aus-
tralia and MDACC showed that an evaluation at six months
might be more appropriate. In these analyses, patients with
BCR-ABL1 transcripts (IS) >10% at six months had a worse
long-term outcome (although still approximately 60% were
alive and free of events long term), regardless of the BCR-
ABL1 transcript levels at three months. This requires a
change of therapy in only 8% of patients, the ones who really
need it.
A similar situation in a patient on second-generation TKIs
(dasatinib, nilotinib) does not necessarily imply considera-
tion of a change of TKIs (no better one available) or of allo-
geneic SCT (five-year survival is better with continuation of
TKI versus implementing allogeneic SCT).
Treatment of CML following frontline
failure
Imatinib resistance can be categorized as primary (failure to
respond to TKI from the onset of therapy) and secondary
(after initial achievement of response). Mutations within the
kinase domain of BCR-ABL1 are a frequent mechanism of
resistance to TKIs. The frequency of BCR-ABL1 mutations
in CML-CP after cytogenetic relapse on imatinib is 30% and
after hematological relapse 50%. Mutation detection is rare
in patients responding to frontline therapy (less than 5%).
Mutation rates are higher (50–90%) in patients experienc-
ing CML transformation. More than 100 different BCR-ABL1
point mutations encoding for single amino acid substitutions
have been reported, conferring different degrees of imatinib
resistance (Table 6.3). The most frequently reported muta-
tions map to the P-loop region of the kinase domain, which
serves as a docking site for phosphate moieties of adenosine
triphosphate (ATP). Mutations also frequently map to the
activation loop, which impair the achievement of the inactive
Chronic myeloid leukemia 79
conformation of the kinase to which imatinib binds, the cat-
alytic domain, and the gatekeeper T315 residue. The T315I
mutation causes steric hindrance to TKI binding and con-
fers resistance to all TKIs, except ponatinib. Some patients
with CML failing sequential TKI therapies carry more than
one mutation within the same BCR-ABL1 molecule (“com-
pound” mutations or “polymutants”), which are associated
with increased oncogenic potency compared with each single
mutation.
BCR-ABL1 mutations do not explain all cases of clinical
resistance to TKI therapy. Other possible mechanisms of
resistance include BCR-ABL1-independent mechanisms
(most common in patients with primary resistance); ima-
tinib plasma levels; excessive binding of imatinib to the
plasma protein α1 -acid glycoprotein 1 (AGP1); overexpres-
sion of ABCB1 (MDR-1) transmembrane protein, which
regulates imatinib efflux from the cell; polymorphisms of the
human organic cation transporter (hOCT1), which regulates
imatinib influx; clonal evolution; SFK or BCR-ABL1 over-
expression; and intrinsic low TKI sensitivity of quiescent
CML stem cells (Lin− CD34+ BCRABL1-positive cells). Each
of the mechanisms of resistance suggests different possible
therapeutic interventions. The long-term updates of TKIs
in CML salvage are showing continued favorable results in
CML-CP–resistant disease. In a phase III study investigating
different dose schedules of dasatinib in 670 patients with
CML-CP following imatinib resistance or intolerance, the
cumulative CCyR rate was 50%, the MMR rate 42%, and the
estimated seven-year survival rate 65%. In a phase II trial of
bosutinib 500 mg orally daily in 288 patients with CML after
imatinib resistance or intolerance, the CCyR was 48%, MMR
35%, and the estimated two-year survival rate 91%. The four-
year follow-up study of nilotinib 400 mg twice daily in 321
patients with CML after imatinib failure resulted in a CCyR
rate of 45% and an estimated four-year survival rate of 78%.
In the phase II trial of ponatinib 45 mg orally daily in 267
patients with heavily pretreated CML-CP, the CCyR rate was
54%, the MMR rate 39%, and the four-year rates of PFS and
survival were 60% and 81%, respectively. Among the subset
of 64 patients with a T315I mutation, the CCyR rate was 70%,
MMR rate 58%, and estimated four-year survival rate 77%.
In a pivotal study of omacetaxine, 122 patients with CML-
CP in chronic (n = 81) or CML-AP (n = 41) who had received
two or more prior TKIs were treated. In the chronic phase,
the major cytogenetic response rate was 20% (complete 10%),
the median duration of response was 17.7 months, and the
median survival was 34 months.
Drug dose schedules
All five BCR-ABL1 TKIs are given orally. Imatinib 400 mg
daily, nilotinib 300 mg twice daily (on an empty stomach),
and dasatinib 100 mg daily are FDA-approved frontline ther-
apies for the treatment of CML. They are also approved for
80 Molecular Hematology

Table 6.3 Relative activity profile of BCR-ABL1 tyrosine kinase inhibitors in imatinib-resistant mutants

IC50 -fold increase relative to wild type (W = 1)

Mutation Imatinib Bosutinib Dasatinib Nilotinib Ponatinib

M244V 0.9 0.9 2 1.2 3.2

L248R 14.6 22.9 12.5 30.2 6.2
L248V 3.5 3.5 5.1 2.8 3.4
G250E 6.9 4.3 4.4 4.6 6
Q252H 1.4 0.8 3.1 2.6 6.1
Y253F 3.6 1 1.6 3.2 3.7
Y253H 8.7 0.6 2.6 36.8 2.6
E255K 6 9.5 5.6 6.7 8.7
E255V 17 5.5 3.4 10.3 12.9
V299L 1.5 26.1 8.7 1.3 0.6
T315A 1.7 6 58.9 2.7 0.4
T315I 17.5 45.4 75 39.4 3
T315V 12.2 29.3 738.8 57 2.1
F317L 2.6 2.4 4.5 2.2 0.7
F317R 2.3 33.5 114.8 2.3 4.9
F317V 0.4 11.5 21.3 0.5 2.3
M351T 1.8 0.7 0.9 0.4 1.2
F359I 6 2.9 3 16.3 2.9
F359V 2.9 0.9 1.5 5.2 4.4
H396R 3.9 0.8 1.6 3.1 5.9
F486S 8.1 2.3 3 1.9 2.1

White = very sensitive (IC50-fold increase <2); yellow = moderately sensitive (IC50 2.1–4); orange = resistant (IC50 4.1–10); red = highly resistant
(IC50 > 10).
Source: Adapted from Redaelli S., Mologni L., Rostagno R., et al. (2012). Three novel patient-derived BCR/ABL mutants show different sensitivity
to second and third generation tyrosine kinase inhibitors. Am. J. Hematol. 87: E125–E128.

the treatment of CML following failure of frontline therapy.
Nilotinib 400 mg orally daily is the approved dose for CML
after failure of frontline therapy. Dasatinib 100 mg daily is
the approved dose for CML after failure of frontline therapy;
dasatinib 140 mg daily is indicated for patients with CML
in transformation. In addition, bosutinib 500 mg daily and
ponatinib 45 mg daily are approved for the treatment of CML
following failure of frontline therapy. Omacetaxine is given
at the dose of 1.25 mg m−2 twice daily subcutaneously for
14 days during induction, and for 7 days every 1–2 months
during maintenance. This schedule may be too myelosup-
pressive. Omacetaxine for five to seven days during induc-
tion, and for two to five days every one to two months during
maintenance (alone or in combination with a TKI), may be
less myelosuppressive and equally effective.
As long-term experience is gained with the TKIs, it is
clear that drug dose modifications can eliminate or reduce
significant toxicities and allow continuation of therapy. For
example, in a case of pleural effusions, dasatinib can be
held, steroids initiated for a short period (e.g. prednisone
40–60 mg daily × 3–5 days), and dasatinib restarted at a
lower dose schedule of 20–70 mg daily, which is still effective
and significantly less toxic. Nilotinib can also be used at
a lower dose schedule of even 150–200 mg orally twice
daily. Bosutinib can be used in a dose schedule range of
300–500 mg daily. While the approved ponatinib dose is
45 mg daily, dose schedules of 15–30 mg daily appear to be
effective and significantly less toxic.
Choice of TKIs
The FDA-approved indications for the five TKIs are sum-
marized in Table 6.4. The choice of TKI depends on several
factors in both the frontline and salvage settings. In frontline
therapy, TKI selection may be based on the cost of the
TKI (generic imatinib vs. more expensive patented second-
generation TKIs), CML Sokal risk (higher vs. lower), patient
age and treatment aim (survival vs. treatment-free remis-
sion), and prior medical history (e.g. diabetes, hypertension,
cardiac or other vasospastic or vaso-occlusive problem,
pancreatitis, pulmonary problems including chronic
obstructive pulmonary disease or pleural effusions). This
 Chronic myeloid leukemia 81

Table 6.4 Food and Drug Administration-approved indications of tyrosine kinase inhibitors (TKIs) and omacetaxine in chronic myeloid leukemia

Frontline Salvage Accelerated Blastic

Imatinib 400 mg QD 400 mg QD 600 mg QD 600 mg QD

(IFN failure) (IFN failure) (IFN failure)
Dasatinib 100 mg QD 100 mg QD 140 mg QD 140 mg QD
R1 R1 R1
Nilotinib 300 mg BID 400 mg BID 400 mg BID —
R1 R1
Bosutinib — 500 mg QD 500 mg QD 500 mg QD
R2 R2 R2
Ponatinib — 45 mg QD 45 m QD 45 mg QD
R2 R2 R2
Omacetaxine — 1.25 mg m−2 SQx14d 1.25 mg m−2 SQx14d —
Q28d (induction), Q28d (induction),
then x7d Q28d then x7d Q28d
R3 R3

QD, daily; BID, twice daily; IFN, interferon alpha.

R1 = resistance or intolerance to prior therapies including imatinib.
R2 = resistance or intolerance to prior TKIs.
R3 = resistance or intolerance to two prior TKIs.

was detailed earlier. In CML salvage therapy, two additional then be restarted on lower doses of dasatinib 20–70 mg
factors are considered: prior TKI exposures, and whether daily. Myelosuppression (particularly thrombocytopenia) is
there are any BCR-ABL1-guiding mutations. more common with dasatinib compared with imatinib. Pul-
monary hypertension can occur occasionally and manifests
clinically as shortness of breath, normal chest radiograph,
Side effects of TKIs and right-side heart failure. It is reversible with dasatinib
discontinuation.
Imatinib is associated with mild to moderate toxicities,
Notable side effects with nilotinib include headache and
including fatigue, nausea, vomiting, diarrhea, skin rashes,
skin rashes (common in 20–30%, but mild to moderate; alle-
muscle cramps and bone aches, periorbital or leg edema,
viated by dose reduction), self-limited elevation of indirect
and weight gain. Serious but uncommon toxicities (<2%)
bilirubin (10%), elevations of blood sugar (10–20%), and
include a decreased glomerular filtration rate, which is
rarely pancreatitis (1–2%). The six-year cumulative incidence
reversible with treatment discontinuation or interruption
of vaso-spastic/vaso-occlusive disease is 10% with nilotinib
and resuming the drug at a lower dose. Hepatic or car-
300 mg twice daily and 15% with nilotinib 400 mg twice daily.
diopulmonary adverse events are rare, as well as peripheral
The most significant toxicities of bosutinib include
neuropathies and central neurological toxicities, which are
myelosuppression (30%), liver function abnormalities (17%,
at times misdiagnosed as dementia or Alzheimer’s disease.
mostly mild to moderate; severe in 7%), and gastrointestinal
These reverse slowly after TKI discontinuation. Drug-
problems (nausea 45%; diarrhea 70–80% – early and self-
related myelosuppression occurs in 10–30% of patients and
limited, but requiring dose reductions to 300–400 mg daily,
is managed with brief treatment interruptions and/or dose
severe in 8%).
modifications, or with growth factors (e.g. erythropoietin
Significant ponatinib toxicities include pancreatitis in
for anemia, filgrastim for neutropenia).
10–15%, severe thrombocytopenia in 35%, severe skin rashes
On average, the chronic mild to moderate toxicities affect-
in 5–10%, vaso-spastic and vaso-occlusive disease in 15–20%,
ing quality of life are lower with second-generation TKIs
and hypertension in 35% (severe in 10–15%).
compared with imatinib (fatigue, fluid retention and perior-
bital edema, muscle cramps and bone aches, weight gain).
Dasatinib is associated with pleural effusions in up to 30%
Other therapies
of patients, which may be severe, requiring thoracentesis in
2–3%. Most resolve with dasatinib treatment interruptions, Interferon alpha was developed as a CML therapy in
diuretics, and short courses of corticosteroids. Patients can the 1980s and became a standard of care in 1990–2000.
82 Molecular Hematology
Interferon therapy induced survival benefits, but was toxic.
A CCyR was obtained in 10–35% of patients and was
associated with a 10-year survival rate of 78%, establishing
the concept of achievement of CCyR to improve survival.
Of interest is that 30% of patients in CCyR had undetectable
BCR-ABL1 transcripts and most had not relapsed after more
than 10 years of follow-up, establishing the uncommon but
possibly curative effect of interferon alpha in CML. Today,
interferon alpha is rarely used as a standard of care; it can
be used in women with CML who become pregnant and
cannot receive TKIs, in combination with TKIs to overcome
CML resistance, in investigational TKI combination studies
to improve therapeutic results, or in investigational studies
aimed at eradication of CML minimal residual disease.
Hydroxyurea, a ribonucleotide reductase inhibitor, is a
well-tolerated oral agent that controls CML counts rapidly.
Rare side effects include nausea, rashes, mouth ulcers, and
hand or leg ulcers. The dose range is 1–10 g daily × 1–5 days
depending on the degree of leukocytosis. The dose is adjusted
to keep the WBC count around 3–10 × 109 cells l−1 . Hydrox-
yurea can be used for initial cytoreduction, or as part of a
combination approach with TKIs. It should not be used alone
as a treatment in CML.
Busulfan (1,4-dimethane-sulfonyl-oxybutane) is the first
alkylating agent to demonstrate activity in CML. It is asso-
ciated with significant toxicities: severe prolonged myelo-
suppression, myelofibrosis (“spent phase of CML”), and
Addison-like syndrome. Today, the role of busulfan is only
as part of conditioning regimens in allogeneic SCT.
Pregnancy, CML, and TKIs
If a pregnant woman is diagnosed as having CML, she can
be managed with leukapheresis as indicated during the first
trimester of pregnancy, and with hydroxyurea therapy sub-
sequently until delivery, then with more definitive therapy.
Some experts recommend interferon alpha therapy during
pregnancy, but this may have adverse effects on the fetus. If a
woman with CML becomes pregnant while on TKI therapy,
the drug should be discontinued. Among 125 infants born to
women with CML on imatinib (drug discontinued after preg-
nancy documented), 12 had birth defects, including 3 with a
syndrome of ocular, renal, and skeletal abnormalities. Chil-
dren of women whose male partners have CML and are on
imatinib have all been normal. The incidence of fetal mal-
formations after dasatinib exposure in a similar scenario is
higher, about 10%.
Allogeneic stem cell transplantation
Allogeneic SCT is curative, but is associated with signifi-
cant morbidities and with a one-year mortality of 5–30%
depending on several factors: patient age, source of cells,
degree of matching, CML phase, and others. It is most
effective in CML-CP. Among patients in first CML-CP who
undergo matched related sibling transplant, the 20-year sur-
vival rate is 40–50%. The International Bone Marrow Trans-
plant Registry (IBMTR) data of >6000 patients showed a 5-
year survival rate of 60% in CML-CP/sibling donor SCT, but
a 20-year survival rate of 40–45%. About 10% of patients
still die in years 5–20 from transplant-related complications
rather than CML relapse. Chronic morbidities following
SCT include graft-versus-host disease (GVHD), second neo-
plasm, cataracts, infertility, hip necrosis, and GVHD-caused
organ damage (pulmonary, hepatic) or immune-mediated
complications. Disease-free survival (DFS) rates are 60–80%
among patients <30–40 years of age, and 30–40% in patients
>50 years. DFS rates are 30–50% in patients with CML-AP
and 5–30% in patients with CML-BP. Patients with clonal
evolution as the only CML-AP criterion have DFS rates of
60%. Patients receiving SCT in second chronic-phase CML
have DFS rates of 40%. Non-myeloablative preparative regi-
mens are safer in older patients and have reduced transplant-
associated complications and mortality. Haplo-identical and
cord blood transplants are showing promising results.
The role and timing of allogeneic SCT in patients with
CML are changing based on the maturing experiences
with TKIs and SCT, as well as the cost of therapies and
regional considerations. Patients who present with or evolve
into CML-AP/BP should receive TKI combinations with
chemotherapy to reduce the CML burden and proceed to
allogeneic SCT. The exception may be de novo CML-AP,
which may still respond well to long-term TKI therapy,
particularly if a CCyR is achieved early. Among patients
whose CML manifests resistance to TKI therapy in CML-
CP, the treatment choice depends on whether mutations
are detected, and whether there is clonal evolution. Patients
with T315I mutation should receive ponatinib, and proceed
to allogeneic SCT (pending longer-term results with pona-
tinib). Patients with particular mutations may be treated with
mutation-selective TKI (nilotinib, dasatinib, bosutinib, or
ponatinib), based on the mutation sensitivity to particular
TKIs. Patients with clonal evolution or with mutations unre-
sponsive to TKIs have short response durations; allogeneic
SCT should be considered as a definitive therapy, particularly
in patients who do not achieve a cytogenetic response after
6–12 months of TKI therapy. Older patients (≥65–70 years)
and patients who lack a donor may forgo the option of cura-
tive allogeneic SCT in favor of years of good disease control
in CML-CP with TKI-based therapies.
Most relapses after SCT occur in the first three to
five years. Patients usually exhibit molecular relapse prior
to cytogenetic-hematologic relapse. CML relapse usually
responds to TKI therapies (response rates 50–60% if a molec-
ular or cytogenetic relapse in CML-CP). Other options
include donor lymphocyte infusions, interferon alpha, or a

second SCT. Donor lymphocyte infusions may exacerbate
GVHD (which can be fatal) and cause severe myelosuppres-
sion (20%).
Therapy for CML in accelerated and blastic
phases
All five TKIs and omacetaxine have shown efficacy in CML-
AP/BP. The single-agent TKI activity is modest in CML-AP
and even lower in CML-BP. In CML-AP, TKIs are associated
with CCyR rates of 20–30% and four-year survival rates of
40–50%, depending on the CML-AP definition and extent
of prior TKI therapies. The second-generation TKIs produce
better results than imatinib, either alone or in combination.
All TKIs have shown fewer responses, which are also less
deep and of shorter duration in CML-BP, and combination
modalities should be pursued. These include acute myeloid
leukemia regimens (e.g. idarubicin + cytarabine) plus TKIs
or azacitidine/decitabine plus a TKI (if intensive chemother-
apy is not indicated) in myeloid-undifferentiated CML-BP,
and acute lymphoid leukemia regimens plus TKIs (e.g. hyper-
CVAD + dasatinib or ponatinib) in lymphoid CML-BP. Cen-
tral nervous system prophylaxis should also be administered
in lymphoid CML-BP, since 30% of patients may develop cen-
tral nervous system disease. In all cases, patients should be
referred for allogeneic SCT as soon as possible. They may also
benefit from post-SCT TKI maintenance.
Philadelphia chromosome-negative CML
Some patients present with a morphological picture of CML
without detectable Ph-positive disease by cytogenetic anal-
ysis. This condition encompasses diverse entities. A third
of patients with typical CML morphology have the BCR-
ABL1 molecular abnormality detected by FISH or PCR (Ph-
negative, BCR-ABL1-positive CML) and have similar clinical
features, response to TKI therapy, and prognosis to patients
with Ph-positive CML.
Patients who have atypical CML morphologies (ane-
mia, thrombocytopenia, absent basophils or eosinophils,
presence of monocytosis) and who do not have the
BCR-ABL1 molecular abnormality (Ph-negative, BCR-
ABL1-negative) are categorized by the WHO classification
into (i) chronic myelomonocytic leukemia (CMML); (ii)
atypical CML, BCR/ABL1-negative; (iii) myelodysplastic/
myeloproliferative neoplasm, unclassifiable (MDS-MPN-
U); and (iv) myelodysplastic/myeloproliferative neoplasm
with ringed sideroblasts and marked thrombocytosis
(MDS-MPD-RS-T).
Patients with CMML are diagnosed based on periph-
eral monocytosis greater than 109 l−1 and ≥10% periph-
eral monocytes. Patients with CMML can be treated with
Chronic myeloid leukemia 83
hypomethylating agents (azacitidine, decitabine), hydrox-
yurea, cytarabine or AML-type therapy, and considered for
allogeneic SCT, depending on the clinical condition and
prognosis.
Atypical CML is a diagnosis of exclusion, an entity that
does not meet the WHO criteria for BCR-ABL1-positive
CML, primary myelofibrosis, polycythemia vera, or essential
thrombocythemia. There are SETBP1 mutations in 15–30%.
Patients with chronic neutrophilic leukemia have signif-
icant neutrophilia without dysplasia. There are mutations of
CSF3R in 90% and mutations of SETBP1 in 50%. Patients with
chronic neutrophilic leukemia and CSF3R mutations may
respond to ruxolitinib (JAK2 inhibitor) therapy (response
rate 50–55%).
MDS/MPN-RS-T manifests as anemia with 15% or more
ring sideroblasts in the erythroid precursors and thrombocy-
tosis ≥450 × 109 . There are mutations of SF3B1 in 50–70%,
which is associated with a better prognosis (median survival
6.9 years versus 3.3 years with wild-type SF3B1).
Patients with MDS-MPD and translocations involving
5q33 – t(5q33; other); PDGFR-beta rearrangement – may
respond to imatinib therapy.
Summary
BCR-ABL1 TKIs have significantly improved prognosis in
CML. Today, TKI frontline therapy in CML-CP is associ-
ated with an estimated 10-year survival rate of 85%. With
TKI therapy, good patient compliance, and optimal monitor-
ing, the relative survival in patients with CML is similar to
that of normal individuals. Frontline therapy with imatinib
and new-generation TKIs results in the durable disappear-
ance of BCR-ABL1 transcripts, which may lend itself to possi-
ble treatment discontinuation and treatment-free remission.
Combinations of TKIs with other active agents that target
the downstream signaling events or are directed against the
“dormant” CML stem cells may increase the proportion of
patients with durable complete molecular responses, and the
rate of treatment-free remissions.
Further reading
Molecular biology of CML
Daley, G.Q., Van Etten, R.A., and Baltimore, D. (1990). Induction of
chronic myelogenous leukemia in mice by the P210bcr/abl gene of
the Philadelphia chromosome. Science 247: 824–830.
Deininger, M.W., Goldman, J.M., and Melo, J.V. (2000). The molecular
biology of chronic myeloid leukemia. Blood 96: 3343–3356.
Groffen, J., Stephenson, J.R., Heisterkamp, N. et al. (1984). Philadelphia
chromosomal breakpoints are clustered within a limited region, bcr,
on chromosome 22. Cell 36: 93–99.
84 Molecular Hematology
Heisterkamp, N., Jenster, G., ten Hoeve, J. et al. (1990). Acute leukemia
in bcr/abl transgenic mice. Nature 344: 251–253.
Melo, J.V. (1996). The diversity of BCR-ABL fusion proteins and their
relationship to leukemia phenotype. Blood 88: 2375–2384.
Prognostic factors and models
Cortes, J., Talpaz, M., O’Brien, S. et al. (2006). Staging of chronic myeloid
leukemia in the imatinib era. Cancer 106: 1306–1315.
Gratwohl, A., Hermans, J., Goldman, J.M. et al. (1998). Risk assessment
for patients with chronic myeloid leukemia before allogeneic blood
or marrow transplantation. Chronic Leukemia Working Party of the
European Group for Blood and Marrow Transplantation. Lancet 352:
1087–1092.
Hasford, J., Pfirrmann, M., Hehlmann, R. et al. (1998). A new prognostic
score for survival of patients with chronic myeloid leukemia treated
with interferon alfa. Writing Committee for the Collaborative CML
Prognostic Factors Project Group. J. Natl. Cancer Inst. 90: 850–858.
Kantarjian, H.M., Dixon, D., Keating, M.J. et al. (1988). Characteristics
of accelerated disease in chronic myelogenous leukemia. Cancer 61:
1441–1446.
Sokal, J.E., Cox, E.B., Baccarani, M. et al. (1984). Prognostic discrimi-
nation in “good-risk” chronic granulocytic leukemia. Blood 63: 789–
799.
Interferon alpha
Guilhot, F., Chastang, C., Michallet, M. et al. (1997). Interferon alfa-2b
combined with cytarabine versus interferon alone in chronic myel-
ogenous leukemia. N. Engl. J. Med. 337: 223–229.
Kantarjian, H.M., Smith, T.L., O’Brien, S. et al. (1995). Prolonged sur-
vival in chronic myelogenous leukemia after cytogenetic response to
interferon-alpha therapy. Ann. Intern. Med. 122: 254–261.
Talpaz, M., Kantarjian, H.M., McCredie, K. et al. (1986). Hematologic
remission and cytogenetic improvement induced by recombinant
human interferon alpha A in chronic myelogenous leukemia. N. Engl.
J. Med. 314: 1065–1069.
Allogeneic stem cell transplantation
Barrett, J. and Ito, S. (2015). The role of stem cell transplantation for
chronic myelogenous leukemia in the 21st century. Blood 125: 3230–
3235.
Crawley, C., Szydlo, R., Lalancette, M. et al. (2005). Outcomes of
reduced intensity transplantation for chronic myeloid leukemia: an
analysis of prognostic factors from the Chronic Leukemia Working
Party of the EBMT. Blood 106: 2969–2976.
Gratwohl, A., Hermans, J., Niederwieser, D. et al. (1993). Bone mar-
row transplantation for chronic myeloid leukemia: long-term results.
Chronic Leukemia Working Party of the European Group for Bone
Marrow Transplantation. Bone Marrow Transplant. 12: 509–516.
Hansen, J.A., Gooley, T.A., Martin, P.J. et al. (1998). Bone marrow
transplants from unrelated donors for patients with chronic myeloid
leukemia. N. Engl. J. Med. 338: 962–968.
Oehler, V.G., Gooley, T., Snyder, D.S. et al. (2007). The effects of ima-
tinib mesylate treatment before allogeneic transplantation for chronic
myeloid leukemia. Blood 109: 1782–1789.
Weisdorf, D.J., Anasetti, C., Antin, J.H. et al. (2002). Allogeneic bone
marrow transplantation for chronic myelogenous leukemia: compar-
ative analysis of unrelated versus matched sibling donor transplanta-
tion. Blood 99: 1971–1977.
Imatinib
Björkholm, M., Ohm, L., Eloranta, S. et al. (2011). Success story of tar-
geted therapy in chronic myeloid leukemia: a population-based study
of patients diagnosed in Sweden from 1973 to 2008. J. Clin. Oncol. 29:
2514–2520.
Cortes, J. and Kantarjian, H. (2012). How I treat newly diagnosed
chronic phase CML. Blood 120: 1390–1397.
Druker, B.J., Guilhot, F., O’Brien, S.G. et al. (2006). Five-year follow-up
of patients receiving imatinib for chronic myeloid leukemia. N. Engl.
J. Med. 355: 2408–2417.
Druker, B.J., Sawyers, C.L., Kantarjian, H. et al. (2001). Activity of a spe-
cific inhibitor of the BCR-ABL tyrosine kinase in the blast crisis of
chronic myeloid leukemia and acute lymphoblastic leukemia with the
Philadelphia chromosome. N. Engl. J. Med. 344: 1038–1042.
Druker, B.J., Tamura, S., Buchdunger, E. et al. (1996). Effects of a selec-
tive inhibitor of the Abl tyrosine kinase on the growth of Bcr-Abl pos-
itive cells. Nat. Med. 2: 561–566.
Kantarjian, H., O’Brien, S., Jabbour, E. et al. (2012). Improved survival in
chronic myeloid leukemia since the introduction of imatinib therapy:
a single-institution historical experience. Blood 119: 1981.
Kantarjian, H., Sawyers, C., Hochhaus, A. et al. (2002). Hematologic and
cytogenetic responses to imatinib mesylate in chronic myelogenous
leukemia. N. Engl. J. Med. 346: 645–652.
O’Brien, S.G., Guilhot, F., Larson, R.A. et al. (2003). Imatinib compared
with interferon and low-dose cytarabine for newly diagnosed chron-
icphase chronic myeloid leukemia. N. Engl. J. Med. 348: 994–1004.
Second- and third-generation TKIs
Cortes, J.E., Kim, D.W., Pinilla-Ibarz, J. et al. (2013). A phase 2 trial of
ponatinib in Philadelphia chromosome-positive leukemias. N. Engl.
J. Med. 369: 1783–1796.
Cortes, J., Nicolini, F., Wetzler, M. et al. (2013). Subcutaneous omacetax-
ine mepesuccinate in patients with chronic-phase chronic myeloid
leukemia previously treated with 2 or more tyrosine kinase inhibitors
including Imatinib. Clin. Lymphoma Myeloma Leuk. 13 (5): 584–591.
Cortes, J.E., Saglio, G., Kantarjian, H.M. et al. (2016). Final 5-year
study results of DASISION. The dasatinib versus imatinib study
in treatment-naı̈ve chronic myeloid leukemia patients trial. J. Clin.
Oncol. 34: 2333–2340.
Gambacorti-Passerini, C., Brummendorf, T., Kim, D. et al. (2014). Bosu-
tinib efficacy and safety in chronic phase chronic myeloid leukemia
after imatinib resistance or intolerance: minimum 24-month follow-
up. Am. J. Hematol. 89: 732–742.
Hochhaus, A., Saglio, G., Hughes, T.P. et al. (2016). Long-term
benefits and risks of frontline nilotinib vs imatinib for chronic
myeloid leukemia in chronic phase: 5-year update of the randomized
ENESTnd trial. Leukemia 30: 1044–1054.
Kantarjian, H.M., Giles, F.J., Bhalla, K.N. et al. (2011). Nilotinib is effec-
tive in patients with chronic myeloid leukemia in chronic phase after
imatinib resistance or intolerance: 24-month follow-up results. Blood
117: 1141–1145.

Kantarjian, H., Giles, F., Wunderle, L. et al. (2006). Nilotinib in imatinib
resistant CML and Philadelphia chromosome-positive ALL. N. Engl.
J. Med. 354: 2542–2551.
Shah, N.P., Rousselot, P., Schiffer, C. et al. (2016). Dasatinib in imatinib-
resistant or-intolerant chronic-phase, chronic myeloid leukemia
patients: 7-year follow-up of study CA180-034. Am. J. Hematol. (9):
869–874.
Talpaz, M., Shah, N.P., Kantarjian, H. et al. (2006). Dasatinib in ima-
tinib resistant Philadelphia chromosome-positive leukemias. N. Engl.
J. Med. 354: 2531–2541.
Weisberg, E., Manley, P.W., Breitenstein, W. et al. (2005). Characteriza-
tion of AMN107, a selective inhibitor of native and mutant Bcr-Abl.
Cancer Cell 7: 129–141.
Chronic myeloid leukemia 85
Mechanisms of resistance to TKIs
Gorre, M.E., Mohammed, M., Ellwood, K. et al. (2001). Clinical resis-
tance to STI-571 cancer therapy caused by BCR-ABL gene mutation
or amplification. Science 293: 876–880.
Shah, N.P., Nicoll, J.M., Nagar, B. et al. (2002). Multiple BCR-ABL kinase
domain mutations confer polyclonal resistance to the tyrosine kinase
inhibitor imatinib (STI571) in chronic phase and blast crisis chronic
myeloid leukemia. Cancer Cell 2: 117–125.
Shah, N.P., Skaggs, B.J., Branford, S. et al. (2007). Sequential ABL
kinase inhibitor therapy selects for compound drug-resistant BCR-
ABL mutations with altered oncogenic potency. J. Clin. Invest. 117:
2562–2569.
Chapter 7 Myeloproliferative neoplasms
Jyoti Nangalia1,2,3 , Anthony J. Bench4 , Anthony R. Green3,5 & Anna L.
Godfrey6
1 Wellcome Sanger Institute, Hinxton, UK
2
Department of Haematology, University of Cambridge, Cambridge, UK
3
Wellcome-Medical Research Council Cambridge Stem Cell Institute, Cambridge, UK
4 Laboratory Medicine, NHS Lothian, Edinburgh, UK
5
Department of Haematology, Cambridge Institute for Medical Research, Cambridge, UK
6 Department of Haematology, Cambridge University Hospitals NHS Foundation Trust, Cambridge, UK

Introduction, 87 Chronic neutrophilic leukemia and chronic eosinophilic leukemia, 95

The JAK2V617F mutation in Philadelphia-negative MPNs, 87 Integration of molecular analysis into diagnostic and management
Other JAK2 mutations in MPNs, 91 pathways, 95
MPL mutations in ET and PMF, 92 Conclusions and future directions, 96
CALR mutations in ET and PMF, 93 Further reading, 97
Other somatic mutations in PV, ET, and PMF, 94

diagnostic tool in PV, were an early demonstration of the

Introduction hypersensitivity of hematopoietic progenitors from MPN
patients to cytokines.
The classical Philadelphia-negative myeloproliferative neo-
The biological behavior of PV, ET, and PMF, and the
plasms (MPNs) comprise polycythemia vera (PV), essential
clinical similarities between them, became clearer with
thrombocythemia (ET), and primary myelofibrosis (PMF).
the identification in 2005 of the JAK2V617F mutation in
These disorders share features including altered stem cell
the majority of PV patients and approximately half of
behavior, overproduction of myeloid lineages, propensities to
patients with ET and PMF. Subsequently, the findings of
thrombosis, hemorrhage and splenomegaly, and, in a minor-
mutations in exon 12 of the JAK2 gene in JAK2V617F-
ity of patients, transformation to acute myeloid leukemia
negative PV, and of CALR and MPL mutations in patients
(AML). They were originally grouped together by Dameshek
with JAK2V617F-negative ET and PMF, have further clar-
along with chronic myeloid leukemia (CML) in 1951. The
ified the biology of MPNs and provided major tools for
World Health Organization (WHO) categorizes the MPNs
molecular diagnosis that are now integrated into interna-
along with CML and rarer clonal disorders, including chronic
tional diagnostic criteria. However, it has also become clear
neutrophilic leukemia (CNL). Clonal mast cell diseases are
not only that the pathogenesis of these disorders depends on
now classified separately and will not be discussed in this
these “phenotypic” mutations that are relatively specific to
chapter.
MPNs, but that a range of other genetic lesions may also have
Evidence that the MPNs represent clonal disorders was
roles in initiation of disease and in driving transformation.
originally derived from the presence of recurrent abnor-
mal karyotypes in approximately one-third of PV and PMF
patients, together with the finding that X-chromosome inac-
The JAK2V617F mutation in
tivation patterns in the majority of PV and PMF patients, and
Philadelphia-negative MPNs
a proportion of ET patients, demonstrate clonally derived
blood cells. Biological studies of hematopoietic progenitor
The JAK2V617F mutation is predominantly found in
cells by Prchal and Axelrad in 1974 showed that in semisolid
patients with PV (>95% of patients), ET, and PMF (50–
media, erythroid progenitors from the bone marrow of PV
60%). Among other myeloid neoplasms it can be found
patients could be grown without the addition of erythro-
in association with SF3B1 mutations in about half of
poietin (EPO), in contrast to those of normal individuals
patients with the uncommon overlap entity myelodysplas-
which required the presence of EPO. These “endogenous ery-
tic/myeloproliferative neoplasm with ring sideroblasts and
throid colonies” (EECs), which were subsequently used as a
thrombocytosis (MDS/MPN-RS-T), but at much lower fre-
quencies in AML and other chronic myeloid disorders.
Molecular Hematology, 4th Edition. Edited by Drew Provan and John G. Gribben. JAK2 is crucial for effective signaling through a number
© 2020 John Wiley & Sons Ltd. Published 2020 by John Wiley & Sons Ltd. of cytokine receptors, including those for erythropoietin

87
88 Molecular Hematology

(a)
pseudokinase kinase

JAK2 FERM SH2 JH2 JH1

617
Val Phe

(b)

EPO

P P

V617F
V617F
JAK2

JAK2

JAK2

JAK2

JAK2

JAK2
EPOR

EPOR

EPOR

EPOR

EPOR

EPOR
P P
P P P P
P P P P
P P P P

No signal JAK/STAT JAK/STAT

PI3K/AKT PI3K/AKT
ERK/MAPK ERK/MAPK
Fig. 7.1 The JAK2V617F mutation. (a) JAK2 protein showing the valine to phenylalanine substitution at codon 617. (b) Role of JAK2 in cytokine
signaling. The cytokine receptor EPOR binds JAK2 as a homodimer (left). On ligand binding, a conformational change within EPOR brings the two
JAK2 molecules into close proximity, initiating a cascade of phosphorylation events and activation of downstream pathways (middle). The JAK2
V617F molecule is constitutively active, leading to erythropoietin (EPO)-independent activation of signaling pathways (right).

(EPOR) and thrombopoietin (TPOR). Homozygous knock-
out of murine Jak2 results in embryonic lethality due to the
complete absence of erythropoiesis. In the unstimulated
state, EPOR forms a transmembrane homodimer, the cyto-
plasmic domain of which is bound to two JAK2 molecules
(Figure 7.1). Ligand binding to the extracellular domain
results in transphosphorylation and activation of the JAK2
molecules (Figure 7.1). Activated JAK2 phosphorylates
tyrosine residues within EPOR which, in turn, promotes
recruitment and JAK2-mediated phosphorylation of other
JAK2 substrates such as STAT5 and phosphatidylinositol
3-kinase (PI3K). Activation of various intracellular signaling
pathways ensues (Figure 7.1). JAK2 possesses two JAK
homology (JH) domains (Figure 7.1). The JH1 domain pos-
sesses tyrosine kinase activity, while the JH2 or pseudokinase
domain is homologous to JH1 and predominantly plays an
auto-inhibitory role, by directly interacting with and
stabilizing the activation loop within JH1 in the inactive
form.
Biochemical and cellular effects of the
JAK2V617F mutation
The valine at codon 617 of JAK2 lies at an evolutionarily
conserved position within the JH2 domain (Figure 7.1) and
is important for the interaction between the JH1 and JH2
domains. The V617F mutation results in the substitution
of valine by a bulkier phenylalanine residue, which appears
to interfere with JH2-mediated auto-inhibition by a num-
ber of mechanisms and leads to constitutive activation of
the tyrosine kinase (Figure 7.1). The downstream effects of
JAK2 V617F may depend on its association with homod-
imeric cytokine receptors such as EPOR and TPOR, but also
with heterodimeric receptors (e.g. interferon-γ receptors)

and with related tyrosine kinases (e.g. JAK1 and TYK2). It
also has a scaffold role in some cytokine receptor complexes
and is required for effective trafficking of receptors to the
cell surface. The cellular effects of the mutation will there-
fore depend on the other signaling molecules expressed by a
specific cell type.
The biochemical effects of the JAK2V617F mutation reflect
both canonical and non-canonical targets. A variety of
experimental systems have confirmed that the mutation is
associated with activation of canonical pathways including
JAK/STAT, PI3K/AKT, and MAPK/ERK signaling. Constitu-
tive activation of STAT5 in erythroid progenitors can induce
the formation of EECs and, consistent with this, STAT5
activation appears particularly important in the erythrocy-
tosis associated with JAK2-mutant MPNs: impairment of
STAT5 signaling can abrogate JAK2V617F-induced cytokine
independence in cell lines, as well as the erythrocytosis
observed in JAK2V617F-driven mouse models of MPNs.
Altered expression of many downstream genes has been
implicated in disease pathogenesis. For example, the anti-
apoptotic protein Bcl-xL is required for the growth of ery-
throid progenitors and is upregulated in PV erythroid pro-
genitors. Interestingly, JAK2 V617F may not only upregulate
Bcl-xL through STAT5-driven transcription, but may also
inhibit the normal DNA damage response by interfering with
changes in intracellular pH and Bcl-xL deamidation. Many
other apoptotic mediators have also been implicated down-
stream of JAK2 V617F, including the anti-apoptotic protein
FLIPshort, ID1, Pim1/2 kinases, the BH3-only proteins Bad
and Bim, and the Bcl2-like protein Mcl-1, downstream of the
STAT3, STAT5, PI3K/AKT, and MAPK/ERK pathways.
Non-canonical effects of JAK2 were identified following
the observation that JAK2 was not confined to the cyto-
plasm but was also present in the nucleus. The kinase has
histone phosphorylation activity and induces a specific chro-
matin modification, phosphorylation of H3Y41, which alters
gene expression through impaired binding of the protein
HP1α. Moreover, JAK2 V617F can also phosphorylate and
alter the activity of a histone arginine methyltransferase,
PRMT5. It therefore seems that canonical activation of gene
transcription (e.g. downstream of activated STATs) and per-
turbations of chromatin modification may be complemen-
tary mechanisms through which mutant JAK2 alters cellular
signaling.
At the cellular level, studies of JAK2V617F in different
hematopoietic cell compartments suggested that while there
may be preferential expansion of the mutant cells, this may
occur predominantly in the later stages of erythroid and
granulocytic differentiation. For example, when quantitative
assays are used to estimate the relative proportions of cells
that are mutant and wild type for JAK2 in the bone mar-
row of patients with PV and ET, the V617F allele burden
Myeloproliferative neoplasms 89
is typically higher in granulocytes and erythroblasts com-
pared to more primitive (CD34+ ) cell fractions. Consistent
with this, knock-in mouse models carrying JAK2V617F show
increased numbers of lineage-restricted hematopoietic pro-
genitors, including those of the megakaryocytic/erythroid
and granulocyte/macrophage lineages.
Conversely, the effects of JAK2V617F on hematopoietic
stem cells (HSCs) have been much less clear. Xenotransplan-
tation experiments, in which CD34+ cells from PV patients
were transplanted into non-obese diabetic/severe combined
immunodeficiency (NOD/SCID) mice, found that the pro-
portion of JAK2V617F-positive cells was reduced in reconsti-
tuted marrow compared to the original donor cells, arguing
against a selective advantage for the mutant cells. Studies of
the number and function of HSCs in JAK2V617F knock-in
mice have yielded different conclusions between the differ-
ent models, perhaps due to variation in whether the human
or protein mutant gene was used and how this was intro-
duced into the Jak2 gene locus. Studies of MPN patients are
further complicated by the potential presence of additional
mutations in JAK2-mutant or wild-type cells, and uncertainty
about the processes that determine HSC number and regen-
eration, particularly during aging. At present there is no con-
clusive evidence that JAK2V617F confers an advantage on
human HSCs, and additional genetic factors, either consti-
tutional or acquired, may well be required to promote the
expansion and persistence of JAK2-mutant clones in MPNs.
Factors influencing disease presentation in
JAK2V617F-positive MPNs
Following the identification of the JAK2V617F mutation,
many studies have attempted to determine the additional fac-
tors that influence which specific MPN phenotype develops
in a patient carrying the JAK2V617F mutation. For exam-
ple, patients with JAK2V617F-positive ET, as compared to
JAK2V617F-negative ET, show certain clinical characteris-
tics reminiscent of PV, including higher hemoglobin levels
and neutrophil counts, lower platelet counts, greater bone
marrow cellularity, erythropoiesis, and granulopoiesis, lower
serum ferritin and mean corpuscular volume. These find-
ings suggested that JAK2-mutated ET could represent a
“milder” form of PV and prompted further studies investi-
gating which factors might lead to a phenotype of ET or PV
in the context of JAK2V617F. Homozygosity for JAK2V617F
is acquired by many patients, but is more frequent in PV
than ET, with homozygous-mutant precursors detectable in
approximately 80% and 50% of patients, respectively. More-
over, the homozygous-mutant subclones are much larger in
patients with PV than those in ET. These findings suggested
that homozygosity for the mutation might have a role in
driving a PV phenotype. Quantitative polymerase chain
90 Molecular Hematology
(a)
Wild-type allele
P1 P2
P3
G
Mutant allele
P1 P2
T P3
(b)
Wild-type allele R
R Q
G G
Mutant allele
R Q
R Q
T T
Fig. 7.2 Detection of the JAK2V617F mutation. (a) Allele-specific polymerase chain reaction (PCR). Amplification with primers P1 and P3 yields
a product of 364 bp (control), whereas amplification with primers P1 and P2 yields a 203 bp PCR product from the mutant allele only. (b) Real-time
PCR using a dual-labeled probe specific for the mutant allele. Amplification from the wild-type allele results in displacement but not destruction of
the probe, resulting in no release of fluorescence. In contrast, amplification from the mutant allele results in Taq-dependent destruction of the
probe, releasing the reporter (R) whose fluorescence can then be detected.
reaction (PCR) assays can be used in patient samples not
only to detect the presence of the V617F mutation, but
also to accurately quantify the proportion of mutated alleles
(Figure 7.2). Among patients with PV, higher V617F allele
burdens (measured in peripheral blood granulocytes) were
associated with more extreme PV features, such as higher
hemoglobin levels and white cell counts and lower platelet
counts at diagnosis, again consistent with the concept that
JAK2V617F homozygosity might be important in the PV
phenotype.
Mouse models have also been informative in investigat-
ing the factors determining phenotype in the context of
JAK2V617F. These models have been generated by a num-
ber of approaches: the earliest models used transduction of
murine bone marrow with retroviruses carrying JAK2V617F;
several transgenic models were then developed; and most
recently knock-in mouse models have been generated with
mutant JAK2 in a heterozygous or homozygous state. In
general these mice have developed convincing myeloprolif-
erative phenotypes with features including erythrocytosis,
leukocytosis, splenomegaly, and increased marrow fibrosis,
but the specific features have varied substantially between
models. With respect to PV and ET, a knock-in model car-
rying a mutant human JAK2 gene showed a phenotype of
al F+
7
or 61
V
er
K2
m
at
JA
W
N
Control
Mutant specific
Q JA
JAK2
0.1
V617F+
0.01
∆Rm
Normal
0.001
0.0001
20 22 24 26 28 30 32 34 36 38 40
Cycle
mild thrombocytosis with a heterozygous mutation, whereas
homozygosity was associated with marked erythrocytosis,
confirming that the acquisition of JAK2V617F homozygos-
ity can be associated with PV-like, rather than ET-like, dis-
ease. However, other knock-in models have shown significant
erythrocytosis even with a heterozygous mutation, and the
explanations for these discrepant phenotypes are not entirely
clear.
Early mouse models also demonstrated that in the con-
text of different genetic backgrounds, the JAK2V617F muta-
tion may be associated with variable hematological features.
This supported the concept that constitutional genetic differ-
ences may affect the presentation of human MPNs, which has
been further confirmed by genome-wide association stud-
ies (GWAS) examining the prevalence of single-nucleotide
polymorphisms (SNPs) in MPN patients. A specific constitu-
tional haplotype at the JAK2 locus (“46/1”) is strongly associ-
ated with the development of JAK2V617F-positive MPNs, as
well as with JAK2 exon 12-mutated and MPL-mutated MPNs.
It remains uncertain whether the mechanism for this reflects
an increased risk of developing a JAK2 mutation, or whether
the presence of the 46/1 haplotype facilitates the development
of clinical disease following the acquisition of JAK2V617F.
Additional GWAS have identified other SNPs that are
 Myeloproliferative neoplasms 91

associated with MPNs, including at the TERT, TET2, and (a)

SH2B3 gene loci. Interestingly, an SNP at the HBS1L/MYB
locus was not only associated with MPNs, but within JAK2-
mutated patients was associated with an increased risk of
ET and not PV, further supporting the concept that consti-
tutional genetic differences may influence not only disease
development but also phenotype.
Additional mutations may coexist with JAK2V617F and
there is evidence that interactions between mutations may
also influence phenotype. For example, mutations in the
gene for the transcription factor NF-E2 may coexist with the
JAK2V617F mutation in PV patients, and may cooperate in
driving erythrocytosis and other myeloproliferative features.
TET2 mutations were first identified in patients with very
large JAK2V617F-heterozygous clones, in keeping with the
concept that additional mutations could confer an advantage
that promotes the growth of JAK2-mutant progenitors. In
such patients with multiple driver mutations, the precise
order in which these are acquired leads to the establishment
of a particular clonal substructure within HSC and progeni-
tor compartments, and may therefore determine the relative
expansion of the myeloid lineages. Additional mutations are (b)
likely to be particularly important in JAK2V617F-positive
PMF and in patients who transform from PV or ET to 0 Normal
myelofibrosis. Both primary and post-PV/ET myelofibrosis K539L
are characterized by a greater average number of somatic
F537-K539delinsL
mutations than PV or ET, and the spectrum of mutations –5
found in these disorders shows significant overlap with N542-E543del
AML and other chronic myeloid malignancies, as discussed E543-D544del
shortly. F537-I546dupF547L
–10
HRM Difference Plot

Other JAK2 mutations in MPNs 70 75 80

Approximately 2% of well-characterized PV patients do not
carry a JAK2V617F mutation. In 2007 it was shown that
many of these patients carry a mutation within JAK2 exon
12. Over 20 mutations have now been described, clustering
within a small region spanning codons 537–547 and most
often affecting Leu539 or Glu543 (Figure 7.3). This SH2–
JH2 linker region lies just upstream of the JH2 pseudoki-
nase domain (Figure 7.3) and is known to undergo a con-
formational change in the presence of the V617F mutation.
Like JAK2V617F, JAK2 exon 12 mutations lead to cytokine
independence and increased phosphorylation of JAK2 sub-
strates. Murine models carrying exon 12 mutations show ele-
vated hematocrit, but normal or mildly elevated neutrophil
and platelet counts. Similarly, patients with PV who carry a
JAK2 exon 12 mutation often display an isolated erythrocy-
tosis, together with isolated bone marrow erythroid hyper-
plasia without granulocytic or megakaryocytic expansion, in
contrast to JAK2V617F-positive patients who characteristi-
cally show hyperplasia of all three myeloid lineages. However,
Fig. 7.3 Types and detection of JAK2 exon 12 mutations. (a)
Three main types of mutation can be described: lysine to leucine
substitution at codon 539 (shaded); deletion of glutamic acid at codon
543 (shaded); duplication affecting codons 537–546 (underlined). (b)
Detection of JAK2 exon 12 mutations by high resolution melting
analysis.
Source: Frequency of mutations taken from Passamonti F., Elena C.,
Schnittger S., et al. (2011). Molecular and clinical features of the
myeloproliferative neoplasm associated with JAK2 exon 12 mutations.
Blood 117: 2813–6.
the incidence of thrombosis, myelofibrosis, acute leukemia,
and death appears to be similar between PV patients with
JAK2V617F and exon 12 mutations. JAK2 exon 12 mutations
have not been detected within JAK2V617F-negative ET or
PMF patients and are therefore specific to PV. Hence, muta-
tions within JAK2 exon 12 lead to increased kinase activ-
ity that drives the HSC toward erythropoiesis, but not other
hematopoietic lineages.
92 Molecular Hematology
A number of germline JAK2 mutations have been reported
in familial MPNs, including V617I and other mutations
within and outside of the pseudokinase domain. These muta-
tions have most often been associated with hereditary throm-
bocytosis and have shown to have weaker functional effects
on JAK2 compared to JAK2V617F. More recently, non-
canonical, activating JAK2 mutations were also identified in
a minority of individuals with JAK2/MPL/CALR-unmutated
ET and in a rare case of isolated erythrocytosis, and again
these were mostly confirmed to be germline.
MPL mutations in ET and PMF
Mutations in the gene for the thrombopoietin receptor, MPL,
were first reported in 2006 and have since been found in
4–9% of patients with PMF and 1–11% of those with ET.
The commonest acquired mutations are W515L and W515K
within exon 10 of the protein, but other mutations are found
at the W515 residue and elsewhere (Figure 7.4). The S505N
variant was originally described in hereditary thrombocyto-
sis, but has also been found as an acquired mutation in ET
and PMF. While a number of other variants cause heredi-
tary thrombocytosis, additional rare mutations outside exon
(a) S505
Extracellular TM JM
W515
(b)
2 HRM Difference Plot
1 Normal
0 W515L
–1
–2
–3
–4
–5
–6
–7
–8
–9
79.5 80.0 80.5 81.0 81.5 82.0 82.5 83.0 83.5 84.0 84.5 85.0 85.5 86.0 86.5 87.0 87.5 88.0
10 have also been identified in JAK2/CALR-unmutated ET,
and recently some of these have been confirmed to have
functional effects. Many patients acquire homozygous MPL
mutations, especially those with PMF and/or with a W515K
mutation, while others have compound mutations (on the
same or different chromosomes) or a coexistent JAK2V617F
mutation.
MPL mutations can confer cytokine independence in cell
lines, together with activation of targets downstream of the
thrombopoietin receptor, including JAK2, STAT3, STAT5,
Akt, and ERK. Introduction of the W515K and W515A
mutations into murine bone marrow was associated with
a myeloproliferative phenotype of thrombocytosis, leukocy-
tosis, extramedullary hematopoiesis, megakaryocytic hyper-
plasia, and fibrosis of the bone marrow and spleen. These
mice did not develop erythrocytosis, in keeping with a selec-
tive effect on megakaryocytic precursors and with the obser-
vation that these mutations are not associated with PV in
humans. Patients with MPL-mutated PMF are more likely to
be female, older, and have more severe anemia than those
with MPL-unmutated PMF, while those with MPL-mutated
ET tend to present with lower hemoglobin levels, higher
platelet counts, and lower bone marrow cellularity compared
to those with JAK2V617F-positive ET.
Cytoplasmic
W515A
S505N
W515K
W515R
Fig. 7.4 Detection of MPL exon 10 mutations.
(a) The S505 and W515 amino acid residues lie
within the transmembrane domain (TM) and
juxtamembrane domain (JM), respectively.
(b) Detection of MPL exon 10 mutations by
high-resolution melting analysis.
 Myeloproliferative neoplasms 93

with the remainder of mutations being alternative insertions

CALR mutations in ET and PMF or deletions affecting the same DNA region (Figure 7.5).
Mutations in CALR result in a 1 bp shift of the codon-reading
Despite the landmark discoveries of mutations in JAK2 and
frame of the DNA, resulting in a neomorphic CALR pro-
MPL in MPNs, just under half of patients with ET and
tein with a novel carboxyl (C)-terminus, which suggests a
PMF still lacked a phenotypic driver mutation that accounted
gain of function. This is in contrast to other genes affected by
for their disease. In 2013, next-generation sequencing tech-
insertion or deletion mutations, where there is often prema-
niques enabled the identification of mutations in the gene
ture truncation of the protein sequence and resultant loss of
calreticulin in around 80% of such JAK2/MPL-unmutated
function.
patients. CALR mutations are insertions or deletions affect-
Unlike JAK2 and MPL, which are proteins involved in
ing the terminal exon 9 of the gene, and 85% of mutations are
cytokine receptor signaling in hematopoietic cells, CALR
either a 52-bp deletion (type 1; c.1099_1150del; L367fs*46) or
resides within the lumen of the endoplasmic reticulum (ER),
a 5-bp insertion (type 2; c.1154_1155insTTGTC; K385fs*47),

c.1154_1155insTTGTC
(a) p.K385fs*47 (Type 2)

N domain P domain C domain KDEL

c.1092_1143del
p.L367fs*46 (Type 1)

(b)
Normal

50 75 100 125 150 175 200

Size (nt)

c.1092_1143del
(Type 1)

50 75 100 125 150 175 200

c.1154_1155insTTGTC
(Type 2)
Wild-type allele

Fig. 7.5 Detection of CALR exon 9 mutations.

(a) CALR mutations lie within the C-terminal Mutant allele
portion of the CALR protein. (b) Detection of CALR
exon 9 mutations by fluorescent polymerase chain
reaction followed by high-resolution capillary
electrophoresis. 50 75 100 125 150 175 200
94 Molecular Hematology
where it functions as a chaperone protein that ensures the
proper folding of newly synthesized glycoproteins prior to
their exportation to the cell surface or extracellular excretion
via the Golgi apparatus. Mutant CALR loses its C-terminal–
KDEL amino acid ER retention motif, as well as numer-
ous calcium binding sites. Studies exploring the mechanism
by which mutant CALR results in myeloproliferation have
revealed a novel paradigm for MPN pathogenesis. Mutant
CALR is able to transform Ba/F3 cells expressing the throm-
bopoietin receptor MPL and render cells cytokine indepen-
dent. Mutant but not wild-type CALR has been shown to
activate MPL and result in downstream STAT, MAPK, and
PI3K activation in cell lines in the absence of thrombopoietin.
Mutant CALR is now known to directly interact with MPL
within the ER. This interaction has been shown to be depen-
dent on both the extracellular N-glycosylation residues of
MPL and the mutant carboxyl-terminus of CALR. It is possi-
ble that the loss of –KDEL allows mutant CALR, coupled with
MPL, to leave the ER. Activation of MPL is believed to occur
while in complex with CALR in a TPO-independent manner,
which subsequently results in downstream constitutive JAK-
STAT signaling. Mice that are recipients of bone marrow cells
retrovirally transduced with either CALR 52-bp deletion or
CALR 5-bp insertion develop PMF and ET, respectively, reca-
pitulating the disease seen in patients and demonstrating that
mutant CALR is sufficient to cause sustained megakaryocyte
proliferation and increased platelet production.
Given this novel interaction between mutant CALR and
MPL, it is unsurprising that mutations in CALR result in an
MPN phenotype that is remarkably similar to that caused by
MPL mutations, with both being associated with ET or PMF.
Both MPL- and CALR-mutated ET are characterized by an
isolated thrombocytosis, unlike JAK2-mutated ET, which is
associated with an increase in trilineage hematopoiesis and
modest increases in hemoglobin and white cell counts, in
addition to thrombocytosis. Clinically, patients with CALR
mutations are younger and suffer fewer thromboses com-
pared to JAK2-mutated counterparts. Elevated white blood
cell counts correlate with an increased risk of thrombosis
in ET. Therefore, it is possible that the differences in the
incidence of thrombosis between JAK2V617F- and CALR-
mutated patients reflect the association between JAK2V617F
and leukocytosis in ET patients, or this may be due to
mutant JAK2-related platelet and leucocyte activation. The
thrombotic tendency of JAK2V617F-mutated ET may also be
related to the increased erythrocytosis that is evident in this
subgroup compared to JAK2-unmutated ET patients. Within
PMF, molecular status also affects clinical outcome, with
CALR-mutated patients carrying a better overall survival
compared to patients with either JAK2 mutations or those
that are triple negative (JAK2, CALR, and MPL unmutated).
The identification of mutations in JAK2, MPL, and CALR
has dramatically improved diagnosis for MPN patients and
molecular testing is now used routinely in the clinical setting.
However, 10–15% of patients with ET and PMF still lack a
recurrent genetic driver of their disease. It is possible that
these patients are a heterogeneous group, with a subset still
carrying as yet unidentified somatic drivers of their disease,
and others harboring either reactive myeloproliferation or
other myeloid disorders.
Other somatic mutations in
PV, ET, and PMF
In addition to mutations in JAK2, MPL, and CALR, a num-
ber of other somatically acquired mutations are found in
MPNs. In general, these additional mutations are not exclu-
sive to MPNs and are found across myeloid malignancies, and
in some cases are also found in lymphoid malignancies and
solid tumors. Exome sequencing of MPNs has shown that PV
and ET have on average six to seven mutations per case, and
PMF has a significantly higher number of mutations, in keep-
ing with its being a more advanced phase of disease. Over-
all, the burden of somatic mutations in MPNs is similar to
that observed in other chronic and acute hematological neo-
plasms, and is roughly one log-fold lower than that seen in
common solid tumors, such as epithelial cancers.
Up to one-third of MPN patients harbor additional
mutations in genes that regulate the epigenome, either via
DNA methylation, such as TET2, DNMT3A, and IDH1/2,
or chromatin remodeling, such as EZH2 and ASXL1.
TET (Ten-eleven translocation methylcytosine dioxyge-
nase) proteins are involved in the demethylation of DNA
through the conversion of 5-methylcytosine (5-mc) to
5-hydroxymethylcytosine (5-hmc). Loss-of-function muta-
tions are found in 10% of MPNs and are associated with HSC
expansion and myeloid progenitor expansion in murine
models and TET2-mutated MPN patients. DNMT3A is a de
novo DNA methyltransferase that is also mutated in around
10% of MPNs. Mutations are also believed to be loss of
function, and the commonest single-nucleotide mutation at
R882 has been shown to confer a dominant negative effect
over wild-type DNMT3A. Loss of DNMT3A also results in
marked HSC compartment expansion in mouse models, and
in human cells has been shown to confer a stem cell advan-
tage over DNMT3A-wild-type cells. Clinical studies have
not demonstrated consistent associations between DNMT3A
or TET2 mutations and alterations to MPN phenotype or
outcome. However, these mutations are some of the most
common somatically acquired genetic aberrations to be
found in the blood of elderly individuals who have normal
blood count parameters, and their presence is associated
with clonal hematopoiesis. Mutations affecting histone
methylation are common in advanced MPNs. Affected genes
code for proteins of the Polycomb repressive complex 2
(PRC2) that normally trimethylates histone H3 at lysine
23 to repress transcription. Loss-of-function mutations

in ASXL1 (Additional sex combs like 1) are present in a
quarter of PMF patients and have been shown to cause
myeloproliferation and anemia in murine models. Mutations
in EZH2 (PcG Enhancer of Zeste Homolog 2) are also seen
in <5% of MPNs, and are more frequent in advanced disease.
A further distinct category of mutations observed in MPNs
reside in genes that code for proteins of the intracellular
spliceosome machinery, such as SF3B1, U2AF1, SRSF2, and
ZRSR2. Mutations are present in 2–5% of MPNs, less frequent
than in myelodysplastic syndromes. In some cases, muta-
tions (such as those affecting U2AF1 or SRSF2) are asso-
ciated with a poorer prognosis when associated with PMF.
Patients with splicing factor mutations often have myelodys-
plastic features on bone marrow examination, and in some
cases distinct morphological features are seen, such as the
presence of ringed sideroblasts in association with mutated
SF3B1. Mutations result in distinct alterations to mRNA
motif recognition sites and mis-spliced target genes are often
important players in myeloid leukemogenesis. Interestingly,
mutations in splicing factors are generally mutually exclu-
sive. This suggests either that cells cannot tolerate more than
one aberration to the splicing machinery, or that one mutated
component is sufficient to alter overall function, such
that additional mutations do not provide a further clonal
advantage.
Additional mutations in MPNs are found more frequently
in the context of disease progression to post-MPN AML, such
as RUNX1, IDH1/2, and TP53. In some cases monoallelic loss
of TP53, either through del17p or genetic mutation, can be
detected during chronic-phase disease, with acquisition of
biallelic loss associated with the progression of the disease.
Chronic neutrophilic leukemia and
chronic eosinophilic leukemia
CNL is a rare MPN characterized by expansion of seg-
mented neutrophils in the blood and bone marrow granulo-
cytic hyperplasia, without an increase in immature blast cells
or morphological dysplasia. Mutations in the gene for the
colony-stimulating factor 3 receptor, CSF3R, had previously
been found in individuals with severe congenital neutrope-
nia, but in 2013 were identified in a significant proportion of
patients with CNL. Although mutations were also reported
in the related MDS/MPN atypical CML, subsequent stud-
ies have suggested that CSF3R mutations are relatively spe-
cific for CNL and can be found in most patients with CNL
defined strictly according to WHO criteria. There are two
classes of activating CSF3R mutations. Membrane-proximal
mutations, the commonest of which is T618I, cause ligand-
independent signaling through the JAK-STAT pathway and
may be amenable to pharmacological inhibition with JAK
inhibitors such as ruxolitinib. Truncation mutations, by con-
trast, cause dysregulation of SRC family and TNK2 kinases.
Myeloproliferative neoplasms 95
Other mutations reported in patients with CNL include those
in ASXL1 (50–60% of patients), SETBP1 (14–43%), TET2
(30%), and CALR (<10%), with ASXL1 mutations appearing
to confer an adverse prognosis.
Among the clonal stem cell disorders associated with
eosinophilia, the most common molecular abnormalities are
fusion genes that involve the PDGFRA, PDGFRB, or FGFR1
genes and result in constitutive activation of a tyrosine
kinase. This group includes both myeloid and lymphoid
diseases, and these are now categorized separately within the
WHO classification as “Myeloid/lymphoid neoplasms with
eosinophilia and rearrangement of PDGFRA, PDGFRB, or
FGFR1, or with PCM1-JAK2.” The commonest lesion is a
submicroscopic (800 kb) deletion of chromosome 4q12 that
leads to the generation of a FIP1L1–PDGFRA fusion gene,
which is not visible with standard G banding but can be
detected by nested reverse transcriptase-polymerase chain
reaction (RT-PCR) or by fluorescence in situ hybridization
(FISH) using a probe (CHIC2) within the deleted region.
Other lesions can be detected by FISH or as rearrangements
within an abnormal karyotype. Patients who lack these
fusion genes but demonstrate a significant unexplained
eosinophilia can however be diagnosed with chronic
eosinophilia leukemia, not otherwise specified (CEL, NOS)
if an alternative clonal marker is identified. For example, the
JAK2V617F mutation and KIT D816V (typically associated
with mastocytosis) have been identified in 3–4% of patients
with hypereosinophilia of unknown significance and may
coexist with a variety of other mutations, such as in TET2,
SRSF2, EZH2, and ASXL1.
Integration of molecular analysis into
diagnostic and management pathways
In recent years, clinical classification schemes have been
greatly enhanced in their ability to aid diagnosis, prognosis,
and therapy by the incorporation of their relevant molecular
features. In the case of the MPNs, the impact of newly identi-
fied mutations has been so significant that schemes have had
to change dramatically to accommodate them. Classification
of the MPNs, first attempted by the Polycythemia Vera
Study Group (PVSG), was a tool for distinguishing these
disorders from phenotypically similar clonal and non-clonal
blood disorders. The identification and characterization
of JAK2, MPL, and CALR mutations have offered a set of
highly specific diagnostic tests for the MPNs. New simplified
classification systems have been devised that incorporate
the molecular changes underlying the MPNs (Tables 7.1 and
7.2). As shown in these tables, both approaches empha-
size the importance of a JAK2, MPL, or CALR mutation,
although they differ in the significance attached to these
markers as compared to other clinical and histological
criteria.
96 Molecular Hematology

Table 7.1 The 2016 World Health Organization (WHO) diagnostic criteria for polycythemia vera, essential thrombocythemia, and primary
myelofibrosisa

Polycythemia vera
Major

1. Hb >165 g l−1 (men), >160 g l−1 (women) or Hct >0.49 (men), >0.48 (women) or Elevated red cell mass >25% above mean normal
predicted value
2. BM biopsy features: hypercellularity; trilineage proliferation (erythroid, granulocytic, and megakaryocytic); pleomorphic, mature
megakaryocytes
3. JAK2V617F or JAK2 exon 12 mutation

Minor

Subnormal serum erythropoietin level

Essential thrombocythemia
Major
1. Platelet count ≥450 × 109 l−1
2. BM biopsy features: predominantly megakaryocytic hyperplasia; large, mature megakaryocytes
3. Not meeting criteria for another myeloid neoplasm (e.g. CML, PV, PMF, MDS)
4. JAK2, CALR, or MPL mutation
Minor

Another clonal marker; or absence of reactive thrombocytosis

Primary myelofibrosis (overt)

Major

1. Megakaryocytic hyperplasia and atypia; reticulin and/or collagen fibrosis

2. Not meeting criteria for another myeloid neoplasm (e.g. CML, ET, PV, MDS)
3. JAK2, CALR, or MPL mutation; in the absence of these mutations, another clonal marker (e.g. ASXL1, EZH2, TET2, IDH1/2 mutation) or
absence of reactive myelofibrosis

Minor
1. Unexplained anemia
2. Leukocytosis ≥11 × 109 l−1
3. Splenomegaly (palpable)
4. Increased serum LDH
5. Leucoerythroblastic blood film

BM, bone marrow; CML, chronic myeloid leukemia; Hb, hemoglobin; Hct, hematocrit; LDH, lactate dehydrogenase; MDS, myelodysplastic
syndrome.
a Diagnosis of polycythemia vera (PV) requires either all three major criteria, or the first two major criteria and the minor criterion.

Diagnosis of essential thrombocythemia requires all four major criteria or the first three major criteria and the minor criterion.
Diagnosis of overt primary myelofibrosis (PMF) requires all three major criteria and at least one minor criterion. Separate criteria exist for the
diagnosis of pre-fibrotic primary myelofibrosis, the principal difference being the absence of reticulin fibrosis >grade 1.

Molecular markers do not only offer a simplified strat-
egy for diagnosis, but also add prognostic information that
can guide management. For example, mutations in ASXL1,
SRSF2, EZH2, and IDH1/2 are associated with an adverse
prognosis in PMF, and in the case of ASXL1 this is inde-
pendent of the International Prognostic Scoring System.
Mutational analysis may therefore aid in decisions such as
the appropriateness of allogeneic stem cell transplantation.
In the future, molecular information also may increasingly
inform the potential benefit of targeted therapies, an exam-
ple being the use of JAK inhibitors in patients with CNL and
a membrane-proximal CSF3R mutation.
Conclusions and future directions
Identification of JAK2, MPL, and CALR mutations that
are relatively specific to PV, ET, and PMF has reinforced
 Myeloproliferative neoplasms 97

Table 7.2 British Committee for Standards in Haematology (BCSH) diagnostic criteria for polycythemia vera, essential thrombocythemia, and
primary myelofibrosisa

Polycythemia vera
A1 High hematocrit (men >52%, women >48%) or an increased red cell mass (>25% above predicted value)
A2 Mutation in JAK2

Essential thrombocythemia
A1 Platelet count >450 × 109 l−1
A2 Presence of an acquired pathogenetic mutation (e.g. in JAK2, CALR, or MPL genes)
A3 No other myeloid malignancy, especially JAK2-positive PV, PMF, CML, or MDS
A4 No reactive cause for thrombocytosis and normal iron stores
A5 Bone marrow aspirate and trephine biopsy showing increased megakaryocytes with a spectrum of morphology, predominantly large
with hyperlobated nuclei and abundant cytoplasm. Reticulin generally not increased.

Primary myelofibrosis
A1 Reticulin grade 3 or higher (on a 0–4 scale)
A2 Pathogenetic mutation or absence of both BCR-ABL1 and reactive causes of bone marrow fibrosis
B1 Palpable splenomegaly
B2 Unexplained anemia
B3 Leucoerythroblastosis
B4 Teardrop red cells
B5 Constitutional symptoms (drenching night sweats, weight loss >10% over six months, unexplained fever, or diffuse bone pain)
B6 Histological evidence of extramedullary hematopoiesis

Diagnosis of essential thrombocythemia requires A1 − A3, or A1 + A3 − A5.

Diagnosis of primary myelofibrosis (PMF) requires A1 + A2 and any two of the B criteria.
CML, chronic myeloid leukemia; MDS, myelodysplastic syndrome.
a Diagnosis of polycythemia vera (PV) requires both criteria to be present. Separate criteria exist for the diagnosis of JAK2-negative PV.

Source: Adapted from McMullin M.F., Harrison C.N., Ali S. et al. (2019). A guideline for the diagnosis and management of polycythaemia vera. A
British Society for Haematology Guideline. Br. J. Haematol. 184: 176–191. Harrison C.N., Butt N., Campbell P., et al. (2014). Modification of British
Committee for Standards in Haematology diagnostic criteria for essential thrombocythaemia. Br. J. Haematol. 167: 421–3; Reilly J.T., McMullin
M.F., Beer P., et al. (2012). Guideline for the diagnosis and management of myelofibrosis. Br. J. Haematol. 158: 453–71, with permission.

Dameshek’s dictum that these disorders are closely linked.
It has also revolutionized all aspects of the MPNs, including
our understanding of pathogenetic mechanisms, clarification
of diagnostic criteria, and, importantly, development of new
therapies. The discovery of the JAK2V617F mutation led to
the rapid development of JAK inhibitors, the first of which
was approved for use in PMF just seven years later in 2012,
and was subsequently also licensed for some patients with
PV. A greater understanding of the additional mutations in
MPNs, especially those affecting histone modifications, has
prompted clinical trials of other novel drugs such as histone
deacetylase inhibitors.
In spite of these advances, important challenges remain in
the understanding of MPN disease biology and its translation
into improving clinical outcomes. Studies of mutational allele
burdens in patients taking pharmaceutical agents demon-
strate that none of these have a reliable, selective, inhibitory
effect on the HSCs that drive disease. This issue must be tack-
led if therapy is to reverse the underlying disease process
and reduce the risk of disease transformation, particularly to
AML with its poor prognosis. It is also increasingly evident
that clinical heterogeneity in all of the chronic myeloid dis-
orders reflects significant underlying mutational heterogene-
ity, and in future more comprehensive mutational analysis
might be able to facilitate a more “personalized” approach to
therapy.
Further reading
Introduction
Arber, D.A., Orazi, A., Hasserjian, R. et al. (2016). The 2016 revision to
the World Health Organization classification of myeloid neoplasms
and acute leukemia. Blood 127: 2391–2405.
Bench, A.J. and Pahl, H.L. (2005). Chromosomal abnormalities and
molecular markers in myeloproliferative disorders. Semin. Hematol.
42: 196–205.
Campbell, P.J. and Green, A.R. (2006). The myeloproliferative disorders.
N. Engl. J. Med. 355: 2452–2466.
98 Molecular Hematology
Dameshek, W. (1951). Some speculations on the myeloproliferative syn-
dromes. Blood 6: 372–375.
Prchal, J.F. and Axelrad, A.A. (1974). Letter: Bone-marrow responses in
polycythemia vera. N. Engl. J. Med. 290: 1382.
The JAK2V617F mutation in
Philadelphia-negative MPNs
Anand, S. and Huntly, B.J. (2012). Disordered signaling in myelopro-
liferative neoplasms. Hematol. Oncol. Clin. North Am. 26: 1017–
1035.
Baxter, E.J., Scott, L.M., Campbell, P.J. et al. (2005). Acquired mutation
of the tyrosine kinase JAK2 in human myeloproliferative disorders.
Lancet 365: 1054–1061.
Godfrey, A.L., Chen, E., Pagano, F. et al. (2012). JAK2V617F homozy-
gosity arises commonly and recurrently in PV and ET, but PV is char-
acterized by expansion of a dominant homozygous subclone. Blood
120: 2704–2707.
Godfrey, A.L. and Green, A.R. (2012). Genotype–phenotype inter-
actions in the myeloproliferative neoplasms. Hematol. Oncol. Clin.
North Am. 26: 993–1015.
James, C., Mazurier, F., Dupont, S. et al. (2008). The hematopoietic
stem cell compartment of JAK2V617F-positive myeloproliferative
disorders is a reflection of disease heterogeneity. Blood 112: 2429–
2438.
James, C., Ugo, V., Le Couedic, J.P. et al. (2005). A unique clonal JAK2
mutation leading to constitutive signalling causes polycythaemia
vera. Nature 434: 1144–1148.
Jones, A.V. and Cross, N.C. (2013). Inherited predisposition to myelo-
proliferative neoplasms. Ther. Adv. Hematol. 4: 237–253.
Jutzi, J.S., Bogeska, R., Nikoloski, G. et al. (2013). MPN patients harbor
recurrent truncating mutations in transcription factor NF-E2. J. Exp.
Med. 210: 1003–1019.
Keil, E., Finkenstadt, D., Wufka, C. et al. (2014). Important scaffold
function of the Janus kinase 2 uncovered by a novel mouse model
harboring a Jak2 activation-loop mutation. Blood 123: 520–529.
Kralovics, R., Passamonti, F., Buser, A.S. et al. (2005). A gain-of-function
mutation of JAK2 in myeloproliferative disorders. N. Engl. J. Med.
352: 1779–1790.
Levine, R.L., Wadleigh, M., Cools, J. et al. (2005). Activating mutation
in the tyrosine kinase JAK2 in polycythemia vera, essential thrombo-
cythemia, and myeloid metaplasia with myelofibrosis. Cancer Cell 7:
387–397.
Li, J., Kent, D.G., Chen, E., and Green, A.R. (2011). Mouse models of
myeloproliferative neoplasms: JAK of all grades. Dis. Model Mech. 4:
311–317.
Ortmann, C.A., Kent, D.G., Nangalia, J. et al. (2015). Effect of mutation
order on myeloproliferative neoplasms. N. Engl. J. Med. 372: 601–612.
Tapper, W., Jones, A.V., Kralovics, R. et al. (2015). Genetic variation at
MECOM, TERT, JAK2 and HBS1L-MYB predisposes to myeloprolif-
erative neoplasms. Nat. Commun. 6: 6691.
Vainchenker, W., Delhommeau, F., Constantinescu, S.N., and Bernard,
O.A. (2011). New mutations and pathogenesis of myeloproliferative
neoplasms. Blood 118: 1723–1735.
Zhao, R., Follows, G.A., Beer, P.A. et al. (2008). Inhibition of the Bcl-xL
deamidation pathway in myeloproliferative disorders. N. Engl. J. Med.
359: 2778–2789.
Other JAK2 mutations in MPNs
Grisouard, J., Li, S., Kubovcakova, L. et al. (2016). JAK2 exon 12 mutant
mice display isolated erythrocytosis and changes in iron metabolism
favoring increased erythropoiesis. Blood 128: 839–851.
Kapralova, K., Horvathova, M., Pecquet, C. et al. (2016). Cooperation
of germ line JAK2 mutations E846D and R1063H in hereditary ery-
throcytosis with megakaryocytic atypia. Blood 128: 1418–1423.
Mead, A.J., Rugless, M.J., Jacobsen, S.E., and Schuh, A. (2012). Germline
JAK2 mutation in a family with hereditary thrombocytosis. N. Engl.
J. Med. 366: 967–969.
Milosevic Feenstra, J.D., Nivarthi, H., Gisslinger, H. et al. (2016).
Whole-exome sequencing identifies novel MPL and JAK2 mutations
in triple-negative myeloproliferative neoplasms. Blood 127: 325–332.
Passamonti, F., Elena, C., Schnittger, S. et al. (2011). Molecular and clin-
ical features of the myeloproliferative neoplasm associated with JAK2
exon 12 mutations. Blood 117: 2813–2816.
Scott, L.M., Tong, W., Levine, R.L. et al. (2007). JAK2 exon 12 mutations
in polycythemia vera and idiopathic erythrocytosis. N. Engl. J. Med.
356: 459–468.
MPL mutations in ET and PMF
Beer, P.A., Campbell, P.J., Scott, L.M. et al. (2008). MPL mutations in
myeloproliferative disorders: analysis of the PT-1 cohort. Blood 112:
141–149.
Guglielmelli, P., Pancrazzi, A., Bergamaschi, G. et al. (2007). Anaemia
characterises patients with myelofibrosis harbouring Mpl mutation.
Br. J. Haematol. 137: 244–247.
Pecquet, C., Staerk, J., Chaligne, R. et al. (2010). Induction of myelo-
proliferative disorder and myelofibrosis by thrombopoietin receptor
W515 mutants is mediated by cytosolic tyrosine 112 of the receptor.
Blood 115: 1037–1048.
Pikman, Y., Lee, B.H., Mercher, T. et al. (2006). MPLW515L is a novel
somatic activating mutation in myelofibrosis with myeloid metapla-
sia. PLoS Med. 3: e270.
Vannucchi, A.M., Antonioli, E., Guglielmelli, P. et al. (2008). Character-
istics and clinical correlates of MPL 515W>L/K mutation in essential
thrombocythemia. Blood 112: 844–847.
CALR mutations in ET and PMF
Araki, M., Yang, Y., Masubuchi, N. et al. (2016). Activation of the throm-
bopoietin receptor by mutant calreticulin in CALR-mutant myelo-
proliferative neoplasms. Blood 127: 1307–1316.
Chachoua, I., Pecquet, C., El-Khoury, M. et al. (2016). Thrombopoietin
receptor activation by myeloproliferative neoplasm associated calreti-
culin mutants. Blood 127: 1325–1335.
Klampfl, T., Gisslinger, H., Harutyunyan, A.S. et al. (2013). Somatic
mutations of calreticulin in myeloproliferative neoplasms. N. Engl. J.
Med. 369: 2379–2390.
Kollmann, K., Warsch, W., Gonzalez-Arias, C. et al. (2017). A novel
signalling screen demonstrates that CALR mutations activate essen-
tial MAPK signalling and facilitate megakaryocyte differentiation.
Leukemia 31: 934–944.
Marty, C., Harini, N., Pecquet, C. et al. (2014). Calr mutants retroviral
mouse models lead to a myeloproliferative neoplasm mimicking an

essential thrombocythemia progressing to a myelofibrosis. Blood 124:
157–157.
Nangalia, J., Massie, C.E., Baxter, E.J. et al. (2013). Somatic CALR muta-
tions in myeloproliferative neoplasms with nonmutated JAK2. N.
Engl. J. Med. 369: 2391–2405.
Other somatic mutations in PV, ET, and PM
Grinfeld, J., Nangalia, J., and Green, A.R. (2017). Molecular determi-
nants of pathogenesis and clinical phenotype in myeloproliferative
neoplasms. Haematologica 102: 7–17.
Guglielmelli, P., Lasho, T.L., Rotunno, G. et al. (2014). The number
of prognostically detrimental mutations and prognosis in primary
myelofibrosis: an international study of 797 patients. Leukemia 28:
1804–1810.
Vannucchi, A.M., Lasho, T.L., Guglielmelli, P. et al. (2013). Mutations
and prognosis in primary myelofibrosis. Leukemia 27: 1861–1869.
Chronic neutrophilic leukemia and chronic
eosinophilic leukemia
Cools, J., DeAngelo, D.J., Gotlib, J. et al. (2003). A tyrosine kinase
created by fusion of the PDGFRA and FIP1L1 genes as a thera-
peutic target of imatinib in idiopathic hypereosinophilic syndrome.
https://doi.org/10.1056/NEJMoa025217. N. Engl. J. Med. 348: 1201–
1214.
Maxson, J.E., Gotlib, J., Pollyea, D.A. et al. (2013). Oncogenic CSF3R
mutations in chronic neutrophilic leukemia and atypical CML. N.
Engl. J. Med. 368: 1781–1790.
Meggendorfer, M., Haferlach, T., Alpermann, T. et al. (2014).
Specific molecular mutation patterns delineate chronic neu-
trophilic leukemia, atypical chronic myeloid leukemia, and chronic
myelomonocytic leukemia. Haematologica 99: e244–e246.
Myeloproliferative neoplasms 99
Pardanani, A., Lasho, T.L., Laborde, R.R. et al. (2013). CSF3R T618I
is a highly prevalent and specific mutation in chronic neutrophilic
leukemia. Leukemia 27: 1870–1873.
Schwaab, J., Umbach, R., Metzgeroth, G. et al. (2015). KIT D816V and
JAK2 V617F mutations are seen recurrently in hypereosinophilia of
unknown significance. Am. J. Hematol. 90: 774–777.
Integration of molecular analysis into
diagnostic and management pathways
Harrison, C.N., Butt, N., Campbell, P. et al. (2014). Modification of
British Committee for Standards in Haematology diagnostic criteria
for essential thrombocythaemia. Br. J. Haematol. 167: 421–423.
McMullin M.F., Harrison C.N., Ali S. et al. (2019). A guideline for the
diagnosis and management of polycythaemia vera. A British Society
for Haematology Guideline. Br. J. Haematol. 184: 176–191.
Reilly, J.T., McMullin, M.F., Beer, P.A. et al. (2012). Guideline for the
diagnosis and management of myelofibrosis. Br. J. Haematol. 158:
453–471.
Vannucchi, A.M., Lasho, T.L., Guglielmelli, P. et al. (2013). Mutations
and prognosis in primary myelofibrosis. Leukemia 27: 1861–1869.
Conclusions and future directions
Andersen, C.L., McMullin, M.F., Ejerblad, E. et al. (2013). A phase II
study of vorinostat (MK-0683) in patients with polycythaemia vera
and essential thrombocythaemia. Br. J. Haematol. 162: 498–508.
Harrison, C., Kiladjian, J.J., Al-Ali, H.K. et al. (2012). JAK inhibi-
tion with ruxolitinib versus best available therapy for myelofibrosis.
N. Engl. J. Med. 366: 787–798.
Verstovsek, S., Mesa, R.A., Gotlib, J. et al. (2012). A double-blind,
placebo-controlled trial of ruxolitinib for myelofibrosis. N. Engl. J.
Med. 366: 799–807.
Chapter 8 Lymphoma genetics
Jennifer L. Crombie1 , Anthony Letai1 & John G. Gribben2
1 Department of Medical Oncology, Dana-Farber Cancer Institute, Harvard Medical School, Boston MA, USA
2
Barts Cancer Institute, Barts and The London School of Medicine and Dentistry, Queen Mary University of London, London, UK
Introduction, 101
Techniques, 101
Commonly occurring translocations in lymphoma, 103
Burkitt lymphoma, 104
Diffuse large B-cell lymphoma, 105
High-grade B-cell lymphomas with MYC and BCL2 or BCL6
translocations, 105
Mantle cell lymphoma, 106
Introduction
As with all cancers, lymphomas were originally categorized
based primarily on morphology and clinical behavior. The
use of antibodies against cell surface markers allowed the
study of lymphoma specimens with antibody panels that
could, along with morphological criteria, usually place a
given lymphoma into a diagnostic category. Even within a
given lymphoma category, however, there is considerable
heterogeneity of clinical behavior. A prominent example is
diffuse large B-cell lymphoma (DLBCL), in which over two-
thirds of patients are cured with conventional chemotherapy,
whereas those with relapsed/refractory disease have dismal
outcomes. In this case, as with many of the lymphomas,
a prognostic index (the International Prognostic Index
or IPI) based on pre-treatment clinical criteria is able to
provide very useful prognostic information. However, even
among patients with the same IPI score, significant clinical
heterogeneity persists. Furthermore, it is likely that the IPI
defines subclasses of lymphomas based upon biological
differences among these lymphomas. Studies of genetic
abnormalities have proved to be important tools to further
classify and prognosticate lymphomas. In addition, a better
understanding of the molecular pathophysiology has led to
improvements in therapeutic options.
Molecular Hematology, 4th Edition. Edited by Drew Provan and John G. Gribben.
© 2020 John Wiley & Sons Ltd. Published 2020 by John Wiley & Sons Ltd.
Follicular lymphoma, 106
Lymphoplasmacytic lymphoma, 106
Mucosa-associated lymphoid tissue lymphomas, 107
Anaplastic large cell lymphoma, 107
Chronic lymphocytic leukemia, 107
Hodgkin lymphoma, 108
Conclusions, 108
Further reading, 109
Techniques
Techniques for studying genetic abnormalities in tumor spec-
imens have undergone a revolution in the past two decades
(Table 8.1). Initial genetic analyses were based on the tech-
nique of chromosomal study by Giemsa-trypsin banding. In
these studies, cells are grown in short-term culture, usually
in the presence of mitogens. Colcemid treatment results in
cell accumulation in metaphase, at which point the cells are
fixed and dropped on glass slides. The slides are treated with
trypsin followed by Giemsa to give a banding pattern. An
experienced cytogenetic technician can then identify normal
chromosomes, translocations, numerical abnormalities, and
sometimes more subtle deletions. The technique can iden-
tify only genetic changes large enough to disrupt a Giemsa-
stained band, requiring a change of many megabases.
More modern techniques are able to detect abnormalities
with greater sensitivity. Southern hybridization starts with
the electrophoretic separation of tumor DNA on a gel, fol-
lowed by transfer to a membrane. This membrane is then
probed with radioactively labeled polynucleotide probes spe-
cific for certain genes of interest. Changes in the expected
size or intensity of the band of interest can indicate muta-
tion, translocation, amplification, or deletion of the gene of
interest. It can also be used to evaluate the presence of clonal
101
102 Molecular Hematology
Table 8.1 Techniques to study lymphoma genetics
r Cytogenetic analysis
r Southern blot analysis
r Polymerase chain reaction (PCR) analysis
r Fluorescent in situ hybridization (FISH)
r Comparative genomic hybridization (CGH)
r CGH microarray
r Gene expression profiling
r Single-nucleotide polymorphism (SNP) array
r microRNA expression profiling
r Next-generation sequencing (NGS): whole-genome sequencing,
whole-exome sequencing, RNA sequencing, and ChIP
sequencing
r Clustered regularly interspersed short palindromic repeats
(CRISPR)-Cas9 functional genetic screening
rearrangements of the immunoglobulin loci in B cells or the
T-cell receptor locus in T cells.
Polymerase chain reaction (PCR) technology has allowed
for the detection of genetic abnormalities using only a small
amount of tumor DNA. Using primers designed to flank
the genomic region of interest, repetitive cycles of anneal-
ing, DNA polymerization, and thermal melting eventually
yield a PCR product. The presence and size of this product
may be analyzed by gel electrophoresis to determine the pres-
ence of a translocation. Furthermore, a PCR product may be
sequenced to look for point mutations. Like Southern blot-
ting, PCR can also be used to evaluate the presence of clonal
rearrangements of the immunoglobulin loci in B cells or the
T-cell receptor locus in T cells.
Fluorescent in situ hybridization (FISH) uses fluores-
cently labeled DNA probes to bind to specific regions of
genomic DNA. Images are then analyzed under a fluorescent
microscope. Numerical chromosomal abnormalities may be
detected by simply counting the number of signals per cell:
more than two indicates the addition of a chromosome,
whereas less than two indicates a deletion. To investigate a
potential translocation, two probes are used, one to detect
the genomic DNA on each side of the known translocation. If
the two probes are consistently approximated, this indicates
the presence of a translocation. FISH may be performed on
interphase cells, so that growth in culture is not a requirement
for this type of analysis as it is for conventional cytogenetics.
Tests for small deletions or other subtle abnormalities may be
better performed on metaphase cells. FISH requires knowl-
edge of the area to be labeled.
Since they rely on the annealing of a labeled specific DNA
probe or primer, Southern hybridization, PCR, and FISH are
techniques to determine the presence or absence of a known
genetic abnormality. Modern techniques which provide a
genome-wide scan for abnormalities, requiring no prior sus-
picion of a particular abnormality which can have a clinical
impact, are comparative genomic hybridization (CGH), gene
expression profiling (GEP), single-nucleotide polymorphism
(SNP) arrays, micro-RNA arrays, next-generation sequenc-
ing (NGS), and clustered regularly interspersed short palin-
dromic repeats (CRISPR)-Cas9 based lethality screens.
In original CGH techniques, DNA is isolated from the
tumor sample and a normal control sample. The DNA in
each is labeled with a different fluorescent dye, for example
green for the tumor DNA and red for the normal DNA. These
samples are mixed and hybridized onto slides of metaphase
spreads of normal cells. Images of metaphase spreads are
then analyzed for the green-to-red color ratio. Regions of
chromosomes that have a high green-to-red ratio contain
a putative area of amplification. Regions that have a low
green-to-red ratio contain a putative deletion. In this way,
the entire genome may be examined for abnormalities. Other
techniques, generally beyond what is performed in clinical
laboratories, are required to determine the critical genes
involved in areas of amplification and deletion. Small abnor-
malities and balanced translocations cannot be observed
using this technique. Currently in more common use is a
variation of the CGH technique (known as “array CGH”)
which hybridizes the DNA to defined arrays of genomic
DNA fragments. These arrays can cover the genome and
improve resolution to the size of the DNA fragments within
the microarray, currently down to the level of 1 MB.
GEP, which allows for a comprehensive, quantitative
examination of the mRNA transcripts of a tumor sample, a
group of molecules that has been termed the “transcriptome,”
was one of the original tools to provide a deeper insight
into the molecular underpinnings of lymphoma. In this tech-
nique, mRNA is purified from a fresh or frozen tumor sam-
ple, amplified, and fluorescently labeled. The labeled RNA
is then hybridized to oligonucleotide probes, which can
be quantitated by microscopy and image analysis software.
Comparison of different tumor samples, and comparison
with wild type, can then allow the determination of tran-
scripts that are over- or underrepresented in certain con-
ditions or tumors. Tumors may be categorized using these
profiles, and subgroups of messages may be used to create
predictors of clinical behavior. For example, GEP has been
used use to characterize two molecular subgroups of DLBCL,
germinal center B-cell (GCB) subtype and activated B-cell
(ABC) subtype, which have both prognostic and therapeu-
tic implications. In recent years, RNA sequencing using NGS,
which is discussed in greater detail shortly, has been replac-
ing microarray assays for evaluation of RNA expression.
Whole-genome analyses of SNPs have also been facili-
tated by the development of SNP arrays. DNA fragments
are digested and annealed to oligonucleotide linkers, ampli-
fied, and hybridized to an SNP array plate. Hundreds of

thousands of oligonucleotides, representing all known SNPs,
can be assessed for using this method. This can identify
which SNPs are present in a given tumor sample. Simi-
larly, consistent loss of all SNPs along a region indicates
a region of chromosomal loss, whereas consistent increase
in SNP signal at a particular region indicates amplification.
As with GEP, newer NGS-based methods are replacing this
technology.
miRNAs are small non-coding RNA transcripts that act
primarily by modulating the expression of other genes. miR-
NAs exert this function by annealing to mRNAs, the con-
sequence of which can be shortened mRNA half-life or
decreased translation. Any single miRNA can modulate the
expression at many different genes. It has been difficult, how-
ever, to use the primary sequence of the miRNA to predict
the genes at which it acts. Several studies have demonstrated
that by measuring the levels of the hundreds of known miR-
NAs in the genome, one can segregate cancers into different
groups. Furthermore, such segregation may provide infor-
mation about prognosis, progression, and response to ther-
apy. Supporting the concept that miRNAs play an important
role in determining cancer behavior, non-coding regions of
the genome that are frequently deleted in cancer often con-
tain miRNA genes. Chronic lymphocytic leukemia (CLL) is
an example of a lymphoid cancer for which miRNA biology
has been informative. For instance, it has been found that in
many cases of CLL, particularly those of indolent behavior,
there is downregulation or deletion of mir-15-A and mir-16-
1, located at 13q14.3. Both of these miRNAs have the ability
to decrease BCL-2 levels, so their deletion may explain the
high levels of BCL-2 found in nearly all CLL cells. In addi-
tion, expression levels of a limited number of miRNAs may
distinguish between indolent and aggressive clinical subtypes
in CLL.
More recently, NGS has revolutionized DNA and RNA
analysis. Whole-exome sequencing (WES), for example, has
been utilized on paired tumor and germline DNA sam-
ples to identify recurrently, somatically mutated protein-
coding genes in lymphoma, allowing for the identification of
genes not previously recognized as drivers of cancer. Whole-
genome sequencing (WGS), which as the name implies
results in sequencing of the entire genome, has the ability
also to interrogate variants that may be important for con-
trolling gene transcriptional regulation or splicing. Results
from these sequencing tools have allowed for the develop-
ment of clinical gene panels able to identify frequently occur-
ring mutations with potential therapeutic impact. These
panels are transforming the management of hematological
malignancies, such as leukemia, providing physicians with
rapid knowledge of mutations that can guide the use of tar-
geted therapy and risk stratify patients to various treatment
approaches. While these panels are not yet in clinical use for
lymphoma, they will likely be incorporated in the near future,
Lymphoma genetics 103
allowing physicians access to the genomic landscape and the
ability to personalize management.
Similarly, high-throughput RNA sequencing (RNA-Seq),
which involves NGS of cDNA, is able to measure the pres-
ence and quantity of RNA in a biological sample. This tech-
nique can identify alternatively spliced gene transcripts, gene
fusions, and post-transcriptional modifications. RNA-Seq
can also identify different populations of RNA in a sample,
including total RNA, miRNA, and tRNA.
Lastly, technologies such as chromatin immunoprecipita-
tion (ChIP) assays can be combined with NGS in a pro-
cess known as chromatin immunoprecipitation–sequencing
(ChIP-Seq). This is a powerful method to determine genome-
wide binding sites for transcription factors and other pro-
teins. Together, these technologies allow for high-throughput
sequencing of thousands of genes from small clinical sam-
ples, readily providing massive amounts of data on lym-
phoma genomics and transcriptomics.
Genome editing technology based on the prokaryotic
CRISPR–Cas9 system has also revolutionized genome engi-
neering and our ability to identify genes that are critical for
survival. In this technique, libraries of small-guide RNAs
(sgRNAs) can guide the Cas9 enzyme to target a host of
genes in lymphoma models. Next-generation DNA sequenc-
ing can subsequently be used to identify genes of interest.
This functional genomics approach identifies mutations with
functional consequence and thus has the potential to identify
optimal therapeutic targets.
Commonly occurring translocations
in lymphoma
Commonly occurring translocations in lymphoma may
be crudely divided into two main classes: those that foster
increased proliferation, and those that inhibit programmed
cell death, or apoptosis. The classical gene in lymphomagen-
esis that induces proliferation is c-MYC. Burkitt’s lymphoma
(BL), one of the most rapidly dividing lymphomas, is the
archetype of a lymphoma that overexpresses c-MYC by
the t(8;14). c-MYC is a helix-loop-helix leucine zipper
transcription factor, which requires heterodimerization
with the protein MAX to activate transcription and induce
proliferation. Targets of this dimer include genes controlling
cell cycle progression, cell growth, metabolism, differenti-
ation, and apoptosis. The net effect of c-MYC expression is
generally an increase in proliferation; however, this effect is
context specific. In some cells, c-MYC overexpression can
induce cell cycle arrest or apoptosis via p53. Therefore, it
may require a concurrent apoptotic defect to permit c-MYC
overexpression.
BCL-2 is an oncogene that does not directly foster
increased proliferation, but rather opposes apoptosis. It does
104 Molecular Hematology

Table 8.2 Chromosomal translocations in non-Hodgkin lymphomas (NHL)

% of cases Proto- Mechanism of activation

NHL histological type Translocation involved oncogene Function of oncogene

Burkitt lymphoma t(8;14) 80% c-MYC Cell proliferation and Transcriptional deregulation
t(2;8) 15% growth
t(8;22) 5%
Diffuse large cell lymphoma der(3) 35% BCL-6 Transcriptional repressor, Transcriptional deregulation
required for germinal
center formation
Mantle cell lymphoma t(11;14) >70% Cyclin D1 Cell cycle regulator Transcriptional deregulation
(BCL-1)
Follicular lymphoma t(14;18) 90% BCL-2 Anti-apoptotic Transcriptional deregulation
Lymphoplasmacytic lymphoma t(9;14) 50% PAX-5 Transcription factor Transcriptional deregulation
regulating B cell
proliferation
Mucosa-associated lymphoid t(11;18) 50% API-2-MLT API-2 is anti-apoptotic Fusion protein
tissue (MALT) lymphoma t(1;14) rare BCL-10 ? Anti-apoptotic Transcriptional deregulation
Anaplastic large cell t(2;5) 60% in adults NPM/ALK Anaplastic lymphoma kinase Fusion protein
lymphoma 85% in children (ALK) is a tyrosine kinase

this at least in part by binding and sequestering pro-apoptotic
BCL-2 family members, preventing them from communicat-
ing or executing death signals, especially at the mitochon-
drion. It is classically overexpressed in follicular lymphoma
due to the t(14;18). This is also the most commonly occurring
translocation to occur in conjunction with t(8;14) in double-
hit lymphomas, an aggressive subtype of lymphoma that is
often refractory to conventional chemotherapy, occurring in
approximately 80% of cases.
A frequent hallmark of translocations found in B-cell
lymphomas is their exploitation of immunoglobulin gene
regulatory elements to drive expression of an oncogene in
a malignant B-cell or B-cell precursor. Burkitt’s lymphoma
is an example of a lymphoma characterized by the overex-
pression of c-MYC. While the most common translocation
is t(8;14), which puts c-MYC under the control of the
immunoglobulin heavy chain (IgH) transcription elements,
the less common t(2;8) and t(8;22) are also found, putting
c-MYC transcription under the control of the light chain
κ and λ transcription elements, respectively. The t(14;18)
found in follicular lymphoma drives BCL-2 expression using
IgH transcription elements. The BCL-6 expression found
in many DLCBL cases is often driven by IgH, Igκ, and Igλ
elements in the t(3;14), t(2;3), and t(3;22), respectively.
PAX5 expression in lymphoplasmacytic lymphoma and
cyclin D1 expression in mantle cell lymphoma are likewise
driven by the t(9;14) and t(11;14) which exploit the IgH
locus. Improved techniques of genetic study have allowed
for the identification of a large number of chromosomal
translocations, the most common of which are shown in
Table 8.2. Those abnormalities which are the most common,
or which have been demonstrated to have the greatest impact
on prognosis or treatment, are described there.
While these translocations are commonly seen in lym-
phoma, and can represent the defining feature of subtypes
such as follicular lymphoma, other lymphomas are less com-
monly associated with chromosomal translocations. GEP
and other sequencing techniques have been used to iden-
tify the drivers of lymphogenesis in many of these cases.
For example, while lymphoplasmacytic lymphoma (LPL) was
previously thought to have no clear translocation of onco-
gene, we now know that over 90% of LPLs have MYD88
L265P mutations. Similarly, BRAF mutations have been
found in nearly all cases of hair cell lymphoma.
Burkitt lymphoma
Burkitt lymphoma (BL) is a very high-grade B-cell malig-
nancy. Pathologically it is characterized by small, non-
cleaved cells. The presence of many apoptotic malignant
cells gives rise to tingible body macrophages and the “starry
sky” appearance characteristic of this and other very rapidly
dividing tumors. Frequent mitotic figures also demonstrate
the rapid cell division typical of this tumor. Though rapidly
dividing, it is one of the most curable lymphomas, with
>90% of adults enjoying long-term survival when treated
with a regimen similar to that proposed by MacGrath. As the
MacGrath regimen is quite different, and yields much
improved results when compared to the R-CHOP regimen
that is used for other aggressive B-cell lymphomas, it is

important to make the diagnostic distinction between BL
and DLBCL.
Genetic testing plays a key role in making the diag-
nosis. The genetic hallmark of BL is overexpression of
the c-MYC oncogene due to a translocation which places
c-MYC transcription under the control of elements at an
immunoglobulin locus. The most common translocation,
t(8;14), is a chromosomal rearrangement involving c-MYC
and the IgH locus. Other translocations involve c-MYC
with the κ(t(2;8)) or λ(t(8;22)) light chain loci. It is difficult
to make the diagnosis of BL in the absence of a c-MYC
translocation by either cytogenetics, FISH, or PCR. The
c-MYC (myelocytomatosis) oncogene is a helix-loop-helix,
zinc finger containing transcription factor that is associated
with a proliferative phenotype. There have been reports
of lymphomas that resemble BL, though they lack MYC
rearrangements. These lymphomas are associated by 11q
alterations, and are now recognized as a new provisional
entity in the revised WHO classification of 2016.
NGS has further improved our understanding of the
underlying biology of BL. For example, 70% of sporadic and
40% of endemic BLs harbor mutations in TCF3, a transcrip-
tion factor that regulates the B-cell receptor and phospho-
inositide 3-kinase (PI3K), or their negative regulator ID3.
The mechanism through which these mutations cooperate
with MYC and the potential to target these mutations ther-
apeutically remain to be explored.
Evidence for latent Epstein–Barr virus (EBV) infection is
also found in nearly all of the African endemic BL, but in only
20% of the sporadic form found outside Africa. It has been
suggested that EBV plays a causative role by opposing apop-
tosis, though the exact mechanism is unknown.
Diffuse large B-cell lymphoma
DLBCLs, the most common subtype of non-Hodgkin
lymphoma (NHL), are a heterogeneous group of lymphomas
of aggressive clinical behavior. The majority likely derive
from follicular center cells, and roughly one-fifth of DLBCLs
derive from transformation of a pre-existing follicular
lymphoma. As their name suggests, DLBCLs have a diffuse
histological pattern of large lymphoid cells. Approximately
two-thirds of patients with this disease will be cured
with combination chemotherapy and the CD20 antibody
rituximab. While approximately half of patients with
relapsed/refractory disease can be rescued by autologous
stem cell transplant following high-dose therapy, the remain-
ing patients have limited therapeutic options and dismal
outcomes.
In an attempt to better divide the heterogeneous group
of diseases encompassed by the label DLCBL, an IPI was
created. The IPI uses just five pieces of clinical and laboratory
Lymphoma genetics 105
data to further subclassify DLBCLs into four groups. While
this formulation does provide a useful refinement of prog-
nosis, it still falls short of the ideal predictor: one that would
definitively determine, prior to a particular therapy, whether
that therapy would work. As this scoring system was devel-
oped in the pre-rituximab era, its ability to predict outcomes
with novel therapy remains unknown.
While the ideal predictor of response to therapy may
not yet be available in practice, progress has been made in
improving prognosis using the wealth of molecular data pro-
vided by GEP. Two groups, one based at the US National Can-
cer Institute and one based at Dana-Farber Cancer Institute,
have published results of applying GEP to lymphoma samples
for which clinical data were available. In both cases, predic-
tors generated by GEP were able to identify new subclasses of
lymphomas and also to further refine prognosis even within
IPI subgroups.
The two main molecular subgroups that define cell of
origin (COO) by GEP are GCB-like and ABC-like lym-
phomas. There are somatic mutations common to both
subtypes, including activating mutations of TP53, genes
involved in immunosurveillance (B2M, CD58), epigenetic
regulators (CREBBP/EP300, KMT2D/C, MEF2B), and BCL6.
Other mutations appear to cluster by COO. In particular,
mutations affecting genes involved in histone modification
(EZH2) and translocations involving BCL-2 are more com-
mon in the GCB subtype, while genes involved in B-cell
receptor signaling (MYD88, CD79A, CARD11, TNFAIP3)
and the NF–κB pathway are common in the ABC sub-
type. These classifications also have prognostic significance,
with patients with ABC subtype DLBCL having worse out-
comes. The Hans algorithm can be used to identify these sub-
groups immunohistochemically, using antibodies to CD10,
BCL6, and IRF4/MUM1, though categorization does not
exactly mirror the findings detected by GEP. Newer meth-
ods in which RNA is extracted from formalin-fixed paraffin-
embedded tissue have the potential to improve classification
of COO, though they are not routinely available.
As the molecular signature of DLBCL become more clearly
defined, the molecules involved in that signature can be
immediately identified as potential targets of anti-cancer
therapy, a feat not possible when prognosis is determined by
purely clinical criteria. Although the clinical significance of
many of the mutations remains unknown, there are ongoing
efforts to exploit these mutations with targeted therapies and
to tailor therapy by COO.
High-grade B-cell lymphomas with
MYC and BCL2 or BCL6 translocations
While there was previously an entity known as DLBCL
unclassifiable, with features intermediate between DLBCL
106 Molecular Hematology
and BL, the criteria was imprecise and not uniformly used. As
DLBCL unclassifiable was found to be overlap with DLBCL
with MYC and BCL-2 and/or BCL-6 translocations by mor-
phology and GEP, a new category known as high-grade B-cell
lymphoma with MYC and BCL2 or BCL6 translocations, also
known as “double-hit” and “triple-hit” lymphomas, has been
incorporated into the WHO criteria. These patients often
present at an advanced stage and have rapidly progressive dis-
ease that is refractory to therapy. GEP has identified a host of
mutations in epigenetic mediators, genes involved in signal-
ing pathways, and tumor suppressors.
Mantle cell lymphoma
Mantle cell lymphoma is a B-cell lymphoma thought to be
the malignant counterpart to the memory B cells found in
the mantle zone of lymphoid follicles. It has characteristic
cell surface markings of CD5+, CD10−, and CD23−. While
it is classically thought of as an aggressive lymphoma, clin-
ically indolent variants with low proliferation indexes have
been identified. These indolent variants are often character-
ized by a mutated IGHV heavy chain and negative staining
for SOX11.
While mantle cell lymphoma often responds to cyto-
toxic chemotherapy, it has frustrated attempts at cure with
chemotherapy, though there are reports of long-term sur-
vivors following allogeneic stem cell transplantation. As in
DLBCL, a mantle cell IPI score has been developed, known
as the MIPI score, which places patients into one of three risk
groups, with low-risk scores correlating with a five-year over-
all survival (OS) of 60% and high-risk scores correlating with
an OS of approximately two years.
Mantle cell lymphoma is almost uniformly characterized
via classical cytogenetics or PCR by a t(11;14) which puts
the cyclin D1 (also known as BCL-1, or B-cell leukemia/
lymphoma 1) gene under control of the igH transcription
control elements. Cyclin D1 binds to and activates cyclin-
dependent kinases (CDKs). An important target of this acti-
vated CDK complex is the retinoblastoma (RB) gene product.
In its hypophosphorylated state, RB inhibits entry into the S
phase of the cell cycle by binding the transcription factor E2F.
When RB is phosphorylated, E2F is freed to activate the tran-
scription of genes which propel the cell into S phase. There-
fore, overexpression of cyclin D1 acts to overcome this late
G1 phase checkpoint and maintain continuous proliferation.
Follicular lymphoma
Follicular lymphoma is an indolent lymphoma. The COO
is thought to be the follicular center B cell. Histologically it
is characterized by a follicular pattern in the lymph node.
The appearance can be similar to that of non-malignant
follicular hyperplasia. Light chain restriction can be useful
in suggesting the clonality of the tumor, which distinguishes
it from benign hyperplasia.
A t(14;18) translocation is found in >85% of follicular
lymphomas. This rearrangement puts the BCL-2 gene under
the transcriptional control of elements from the igH locus.
The BCL-2 protein functions to oppose programmed cell
death. It is presumed that BCL-2 expression in malignancies
such as follicular lymphoma permits survival of cancer cells
under conditions (cell cycle checkpoint violation, metastatic
location, genomic instability) which would otherwise trigger
programmed cell death. The cloning of BCL-2 led to the
identification of a family of related proteins. While some are
anti-apoptotic like BCL-2, many are pro-apoptotic, but all
function in the control of apoptosis.
While follicular lymphoma can be cured by local ther-
apy in very localized disease, it is more usually diagnosed in
an advanced stage where cure is exceedingly rare. It is gen-
erally quite responsive to chemotherapy, but almost always
relapses. The clinical course is commonly marked by a series
of chemotherapy-induced remissions followed by relapses,
with the interval between these decreasing over time. The end
stage of the disease may be characterized by insuperable resis-
tance to chemotherapy or by transformation to an aggressive
large B-cell phenotype. Despite the very low cure rate, many
patients nonetheless survive for more than a decade due to
the indolent nature of the disease. Follicular lymphoma can
transform into a higher-grade lymphoma with DLBCL mor-
phology. Numerous genetic changes have been associated
with this transformation, including trisomy 7, loss of p53, and
c-MYC rearrangements.
As outcomes in follicular lymphoma are heterogeneous, a
variety of prognostic scoring systems have been developed.
The FLIPI score and the more recent FLIPI2 score incor-
porate clinical criteria to predict OS and progression-free
survival (PFS), respectively. There has also been integration
of gene mutations into a prognostic model known as the
m7-FLIPI score. This scoring system incorporates the muta-
tional status of seven genes (EZH2, ARID1A, MEF2B, EP300,
FOXO1, CREBBP, and CARD11), along with the FLIPI score
and ECOG performance status, to create a more predictive
prognostic tool.
Lymphoplasmacytic lymphoma
Lymphoplasmacytic lymphoma (LPL) is an indolent lym-
phoma. The cells of this lymphoma have a phenotype that lies
midway between mature lymphocytes and plasma cells, and
for this reason they are often nicknamed “lymphocytes.” This

lymphoma commonly expresses IgM, which can lead to the
syndrome of Waldenstrom macroglobulinemia (WM). WM
is characterized by IgM expression, hyperviscosity, bleeding,
Raynaud’s phenomenon, visual disturbances, and other neu-
rological symptoms.
Roughly half of all LPL cases will demonstrate the t(9;14)
which juxtaposes the PAX5 gene and the IgH locus. PAX5
encodes the BSAP (B-cell specific activator protein), which
is a transcription factor. Its expression is associated with
increased expression of genes important in early B cell
development and decreased expression of the p53 tumor
suppressor.
WGS of patients with WM has also identified the somatic
mutation MYD88 L265P to occur in more than 90% of
patients. Mutated MYD88 triggers tumor growth through
the activation of NF-κB by Bruton’s tyrosine kinase (BTK).
Mutations in MYD88 suggest susceptibility to ibrutinib, a tar-
geted inhibitor of BTK, while CXCR4 mutations alternatively
convey resistance.
Mucosa-associated lymphoid tissue
lymphomas
The mucosa-associated lymphoid tissue (MALT) lymphomas
are thought to arise from the extranodal counterpart to post-
follicular memory B cells found in the marginal zone of
lymph node follicles. These tumors are often localized, and
their behavior is generally indolent. At least some have a
dependence on continued antigen stimulation for survival,
as demonstrated by the prolonged complete responses that
are seen when early-stage gastric MALT lymphoma is treated
with an antibiotic regimen to eradicate chronic helicobacter
pylori infection.
t(11;18)(q21;q21) is found in more than half of all low-
grade MALT lymphomas, with a preference for gastric lym-
phomas. The translocation is not typically found in high-
grade MALT lymphomas. API-2-MALT 1 fusion protein is
expressed from the mutant locus. API-2 (also known as
IAP-2) belongs to a family of inhibitors of apoptosis that pre-
vent death likely due to their direct interaction with caspases,
the proteases activated by programmed cell death. The phys-
iological function of the MALT 1 protein is less well under-
stood, though it possesses a caspase-like domain at its C ter-
minus. The function of the fusion protein is unclear, though
there is some evidence that it activates NF-κB, perhaps lead-
ing to inhibition of apoptosis.
BCL-10 is overexpressed in a minority of MALT lym-
phoma cases via the t(1;14)(p22;q32), putting the coding
region of BCL-10 under the influence of the IgH enhancer.
The function of this protein is unclear, but some researchers
have suggested an interaction between BCL-10 and MALT-1,
Lymphoma genetics 107
leading to NF-κB activation. Others have shown that API2-
MALT 1 correlates with the nuclear location of BCL-10.
These findings suggest that these two translocations may be
involved in activating the same pathway.
Trisomy 3 is observed in 20–60% of all MALT lymphomas.
The oncogenic properties of this numerical chromosomal
abnormality are not understood.
Anaplastic large cell lymphoma
Anaplastic large cell lymphoma (ALCL) is characterized by
strong surface expression of the CD30 (Ki-1) antigen, a
cytokine receptor in the tumor necrosis factor receptor fam-
ily. The majority of ALCLs demonstrate T cell surface mark-
ers and/or clonal rearrangements of the T-cell receptor locus.
There are two main clinical forms, systemic and cutaneous.
The cutaneous form is particularly indolent. While it is clin-
ically aggressive, systemic ALCL is generally sensitive to
chemotherapy.
Approximately 50% of systemic ALCLs carry the t(2;5),
which confers good prognosis. The t(2;5)(p23;q35) results
in a chimeric gene encoding a fusion of the nucleophosmin
(NPM) and anaplastic lymphoma kinase (ALK) proteins.
NPM is a multifunctional protein which has been implicated
in ribosome assembly, control of centrosome duplication,
and nuclear transport as a shuttle protein; it also possesses
chaperonin and ribonuclease activities. ALK is a member
of the insulin family of receptor tyrosine kinases. Its natural
ligand is unknown. The NPM–ALK fusion contains the
oligomerization domain of NPM and the tyrosine kinase
domain of ALK. It results in a self-oligomerizing, constitu-
tively active tyrosine kinase with transforming properties.
NPM–ALK can activate numerous downstream effectors,
including phospholipase C-γ PI3K, and RAS. Patients with
ALK mutations may also derive benefit from the targeted
inhibitor of ALK, crizotinib.
Chronic lymphocytic leukemia
CLL is a low-grade lymphoma marked by a peripheral lym-
phocytosis of CD5+, CD20+, and CD23+ small lymphocytes,
similar in morphology to normal lymphocytes. Staging based
on the presence of lymphadenopathy, organomegaly, anemia,
or thrombocytopenia can provide prognostic information,
with those cases in the best prognostic groups enjoying near
normal mean survival times.
Prognosis can also be estimated by purely molecular
criteria. In about half of CLL cases, VH genes exhibit
somatic hypermutation, while the others lack VH mutations.
CLL with IgVH genes that exhibit more than 2% somatic
108 Molecular Hematology

hypermutation, or mutated IGHV status, have significantly
better responses to chemoimmunotherapy and OS than the
latter.
Results of FISH cytogenetics, which detect genomic
aberrations in over 80% of cases, also have prognostic
implications. As FISH is a directed rather than screen-
ing technique, these abnormalities have previously been
demonstrated by conventional banding techniques. The
abnormalities include del 13q (50%), del 11q (18%), +12q
(16%), del 17p (7%), and del 6q (7%). Regression analysis
allowed the assignment of 90% of these cases to one of
five prognostic classes based on genetic abnormalities. The
best prognostic group included those that had 13q deletion
as their sole abnormality, whereas the worst prognostic
group contained those with a 17p deletion. CLL with 17p
and 11q deletions was more likely to have extensive lym-
phadenopathy, splenomegaly, cytopenias, and B symptoms.
These data raise the question of whether the classical clinical
staging, which can be used to predict survival, is partly just
a surrogate for particular genetic abnormalities, and it is the
genetic abnormalities and resulting expression patterns that
are more important in determining prognosis.
While there are no disease-defining mutations in CLL,
sequencing has revealed a number of mutations that occur
at low frequency. Mutations in TP53, NOTCH1, SF3B1, and
BIRC3, for example, are of clinical interest because of their
prognostic and therapeutic implications. There are ongoing
efforts to further understand how these mutations and oth-
ers impact pathogenesis and whether they can be targeted
therapeutically.
While genetic changes in CLL have yet to result in major
therapeutic advances, further understanding of the signaling
pathways implicated in CLL has recently yielded a phar-
macological revolution, with the approval of several new
targeted agents. B-cell receptor signaling, for example, has
emerged as a driving factor in CLL survival. Downstream of
the B-cell receptor is BTK, which is essential for activating
additional downstream targets such as Akt, extracellular
signal–regulated kinase (ERK), and NF-κB. BTK has also
been implicated in B cell homing and adhesion. The targeted
inhibitor of BTK, ibrutinib, has resulted in high rates of
durable responses in patients with relapsed/refractory CLL
and those with high-risk features, such as deletion 17p, lead-
ing to its Food and Drug Administration (FDA) approval
in both the frontline and relapsed/refractory settings. PI3K,
another key component of the breakpoint cluster region
(BCR) signaling pathway, has also been identified as a
therapeutic target, with idelalisib, a PI3Kδ, being FDA
approved for relapsed/refractory disease in combination
with rituximab.
BCL-2, which expressed at high levels in more than
70% of CLL cases, is also a target in CLL. While BCL-2
overexpression in CLL is rarely if ever due to a t(14;18), it is
thought that the deletion of two miRNA genes, mir-15-A and
mir-16-1, which downregulate BCL-2 levels, are responsible
for overexpression. Nearly all CLL cases are dependent on
BCL-2 for survival, which is largely due to the requirement
for BCL-2 to tonically sequester the large amounts of the
pro-death molecule BIM that are generated in CLL cells.
When BCL-2 function is abrogated, BIM is released, the
mitochondria are permeabilized, and the cell dies. Veneto-
clax, a targeted BCL-2 inhibitor, has had dramatic clinical
activity in patients with relapsed/refractory CLL, including
those with high-risk biology, and is now a fundamental
therapeutic option for patients with CLL.
Hodgkin lymphoma
Classical Hodgkin lymphoma (cHL), which arises from
germinal center or post-germinal center B cells, is char-
acterized histologically by small numbers of malignant
Reed–Sternberg (RS) cells surrounded by an extensive but
ineffective inflammatory cell infiltrate. It has a bimodal
distribution, with one peak occurring in young adults and
another in adults of older age. While HL is curable in
majority of patients treated with conventional chemotherapy
with or without radiation, there is a subset of patients with
relapsed/refractory disease who are more difficult to treat.
cHL lack surface immunoglobulin expression and B-cell
receptor signaling, and rely on alternative survival pathways,
such as constitutive activation of NF-κB. Despite advances
in understanding the genomics of cHL, there had been few
therapies aimed at targeting known mutations. More recently,
however, high-density SNP arrays with paired transcriptional
profiles have identified 9p24.1 amplifications, which result in
increased expression of PD-1 ligands, PDL1 and PDL-2, and
subsequently increased immune evasion. Based on this dis-
covery, checkpoint blockade has since emerged as an effec-
tive therapy for cHL, with excellent responses in patients with
relapsed/refractory disease.
Conclusions
Identification of the genes involved in lymphoma pathogen-
esis has allowed for better characterization of the disease.
This has also led the discovery of novel, targeted agents and
immunotherapies which have dramatically improved out-
comes for patients. While great progress has been made, there
are many questions remaining about the wealth of genetic
information we now possess, and further investigation on
how best to utilize this information is required.

Further reading
Introduction
Sehn, L.H., Donaldson, J., Chhanabhai, M. et al. (2005). Introduction
of combined CHOP plus rituximab therapy dramatically improved
outcome of diffuse large B-cell lymphoma in British Columbia. J. Clin.
Oncol. 23 (22): 5027–5033.
Techniques
Alizadeh, A.A., Eisen, M.B., Davis, R.E. et al. (2000). Distinct types of
diffuse large B-cell lymphoma identified by gene expression profiling.
Nature 403 (6769): 503–511.
Barrans, S.L., Crouch, S., Care, M.A. et al. (2012). Whole genome
expression profiling based on paraffin embedded tissue can be used to
classify diffuse large B-cell lymphoma and predict clinical outcome.
Br. J. Haematol. 159 (4): 441–453.
Dubois, S., Viailly, P.J., Mareschal, S. et al. (2016). Next-generation
sequencing in diffuse large B-cell lymphoma highlights molecular
divergence and therapeutic opportunities: a LYSA study. Clin. Cancer
Res. 22 (12): 2919–2928.
Lohr, J.G., Stojanov, P., Lawrence, M.S. et al. (2012). Discovery and pri-
oritization of somatic mutations in diffuse large B-cell lymphoma
(DLBCL) by whole-exome sequencing. Proc. Natl. Acad. Sci. U S A
109 (10): 3879–3884.
Commonly occurring translocations in
lymphoma
Treon, S.P., Xu, L., Yang, G. et al. (2012). MYD88 L265P somatic muta-
tion in Waldenstrom’s macroglobulinemia. N. Engl. J. Med. 367 (9):
826–833.
Burkitt lymphoma
Love, C., Sun, Z., Jima, D. et al. (2012). The genetic landscape of muta-
tions in Burkitt lymphoma. Nat. Genet. 44 (12): 1321–1325.
Magrath, I., Adde, M., Shad, A. et al. (1996). Adults and children with
small non-cleaved-cell lymphoma have a similar excellent outcome
when treated with the same chemotherapy regimen. J. Clin. Oncol. 14
(3): 925–934.
Richter, J., Schlesner, M., Hoffmann, S. et al. (2012). Recurrent muta-
tion of the ID3 gene in Burkitt lymphoma identified by integrated
genome, exome and transcriptome sequencing. Nat. Genet. 44 (12):
1316–1320.
Salaverria, I., Martin-Guerrero, I., Wagener, R. et al. (2014). A recur-
rent 11q aberration pattern characterizes a subset of MYC-negative
high-grade B-cell lymphomas resembling Burkitt lymphoma. Blood
123 (8): 1187–1198.
Schmitz, R., Young, R.M., Ceribelli, M. et al. (2012). Burkitt lymphoma
pathogenesis and therapeutic targets from structural and functional
genomics. Nature 490 (7418): 116–120.
Swerdlow, S.H., Campo, E., Pileri, S.A. et al. (2016). The 2016 revision of
the World Health Organization classification of lymphoid neoplasms.
Blood 127 (20): 2375–2390.
Lymphoma genetics 109
Diffuse large B-cell lymphoma
Davis, R.E., Ngo, V.N., Lenz, G. et al. (2010). Chronic active B-cell-
receptor signalling in diffuse large B-cell lymphoma. Nature 463
(7277): 88–92.
International Non-Hodgkin’s Lymphoma Prognostic Factors Project
(1993). A predictive model for aggressive non-Hodgkin’s lymphoma.
N. Engl. J. Med. 329 (14): 987–994.
Morin, R.D., Johnson, N.A., Severson, T.M. et al. (2010). Somatic muta-
tions altering EZH2 (Tyr641) in follicular and diffuse large B-cell
lymphomas of germinal-center origin. Nat. Genet. 42 (2): 181–185.
Morin, R.D., Mendez-Lago, M., Mungall, A.J. et al. (2011). Frequent
mutation of histone-modifying genes in non-Hodgkin lymphoma.
Nature 476 (7360): 298–303.
Scott, D.W., Mottok, A., Ennishi, D. et al. (2015). Prognostic significance
of diffuse large B-cell lymphoma cell of origin determined by digital
gene expression in formalin-fixed paraffin-embedded tissue biopsies.
J. Clin. Oncol. 33 (26): 2848–2856.
Sehn, L.H., Berry, B., Chhanabhai, M. et al. (2007). The revised inter-
national prognostic index (R-IPI) is a better predictor of outcome
than the standard IPI for patients with diffuse large B-cell lymphoma
treated with R-CHOP. Blood 109 (5): 1857–1861.
Shipp, M.A., Ross, K.N., Tamayo, P. et al. (2002). Diffuse large B-
cell lymphoma outcome prediction by gene-expression profiling and
supervised machine learning. Nat. Med. 8 (1): 68–74.
Teras, L.R., DeSantis, C.E., Cerhan, J.R. et al. (2016). 2016 US lymphoid
malignancy statistics by World Health Organization subtypes. CA
Cancer J. Clin. 66 (6): 443–459.
Vose, J.M., Bierman, P.J., Anderson, J.R. et al. (1992). Progressive dis-
ease after high-dose therapy and autologous transplantation for lym-
phoid malignancy: clinical course and patient follow-up. Blood 80 (8):
2142–2148.
Mantle cell lymphoma
Hoster, E., Dreyling, M., Klapper, W. et al. (2008). A new prognostic
index (MIPI) for patients with advanced-stage mantle cell lymphoma.
Blood 111 (2): 558–565.
Follicular lymphoma
Bernell, P., Jacobsson, B., Liliemark, J. et al. (1998). Gain of chromo-
some 7 marks the progression from indolent to aggressive follicle
centre lymphoma and is a common finding in patients with diffuse
large B-cell lymphoma: a study by FISH. Br. J. Haematol. 101 (3): 487–
491.
Certo, M., Del Gaizo Moore, V., Nishino, M. et al. (2006). Mitochondria
primed by death signals determine cellular addiction to antiapoptotic
BCL-2 family members. Cancer Cell 9 (5): 351–365.
Federico, M., Bellei, M., Marcheselli, L. et al. (2009). Follicular lym-
phoma international prognostic index 2: a new prognostic index
for follicular lymphoma developed by the international follicular
lymphoma prognostic factor project. J. Clin. Oncol. 27 (27): 4555–
4562.
Lo Coco, F., Gaidano, G., Louie, D.C. et al. (1993). p53 mutations are
associated with histologic transformation of follicular lymphoma.
Blood 82 (8): 2289–2295.
110 Molecular Hematology
McDonnell, T.J., Deane, N., Platt, F.M. et al. (1989). bcl-2-
immunoglobulin transgenic mice demonstrate extended B cell
survival and follicular lymphoproliferation. Cell 57 (1): 79–88.
Pastore, A., Jurinovic, V., Kridel, R. et al. (2015). Integration of gene
mutations in risk prognostication for patients receiving first-line
immunochemotherapy for follicular lymphoma: a retrospective anal-
ysis of a prospective clinical trial and validation in a population-based
registry. Lancet Oncol. 16 (9): 1111–1122.
Solal-Celigny, P., Roy, P., Colombat, P. et al. (2004). Follicular lymphoma
international prognostic index. Blood 104 (5): 1258–1265.
Tsujimoto, Y., Yunis, J., Onorato-Showe, L. et al. (1984). Molecular
cloning of the chromosomal breakpoint of B-cell lymphomas and
leukemias with the t(11;14) chromosome translocation. Science 224
(4656): 1403–1406.
Lymphoplasmacytic lymphoma
Iida, S., Rao, P.H., Nallasivam, P. et al. (1996). The t(9;14)(p13;q32) chro-
mosomal translocation associated with lymphoplasmacytoid lym-
phoma involves the PAX-5 gene. Blood 88 (11): 4110–4117.
Treon, S.P., Tripsas, C.K., Meid, K. et al. (2015). Ibrutinib in previously
treated Waldenstrom’s macroglobulinemia. N. Engl. J. Med. 372 (15):
1430–1440.
Mucosa-associated lymphoid tissue
lymphomas
Lucas, P.C., Yonezumi, M., Inohara, N. et al. (2001). Bcl10 and MALT1,
independent targets of chromosomal translocation in MALT lym-
phoma, cooperate in a novel NF-kappa B signaling pathway.
J. Biolumin. Chemilumin. 276 (22): 19012–19019.
Maes, B., Demunter, A., Peeters, B., and De Wolf-Peeters, C. (2002).
BCL10 mutation does not represent an important pathogenic mecha-
nism in gastric MALT-type lymphoma, and the presence of the API2-
MLT fusion is associated with aberrant nuclear BCL10 expression.
Blood 99 (4): 1398–1404.
Uren, A.G., O’Rourke, K., Aravind, L.A. et al. (2000). Identification of
paracaspases and metacaspases: two ancient families of caspase-like
proteins, one of which plays a key role in MALT lymphoma. Mol. Cell
6 (4): 961–967.
Anaplastic large cell lymphoma
Gambacorti-Passerini, C., Messa, C., and Pogliani, E.M. (2011). Crizo-
tinib in anaplastic large-cell lymphoma. N. Engl. J. Med. 364 (8): 775–
776.
Kutok, J.L. and Aster, J.C. (2002). Molecular biology of anaplastic lym-
phoma kinase-positive anaplastic large-cell lymphoma. J. Clin. Oncol.
20 (17): 3691–3702.
Morris, S.W., Kirstein, M.N., Valentine, M.B. et al. (1994). Fusion of
a kinase gene, ALK, to a nucleolar protein gene, NPM, in non-
Hodgkin’s lymphoma. Science 263 (5151): 1281–1284.
Vose, J., Armitage, J., Weisenburger, D., and International TCLP (2008).
International peripheral T-cell and natural killer/T-cell lymphoma
study: pathology findings and clinical outcomes. J. Clin. Oncol. 26
(25): 4124–4130.
Chronic lymphocytic leukemia
Byrd, J.C., O’Brien, S., and James, D.F. (2013). Ibrutinib in relapsed
chronic lymphocytic leukemia. N. Engl. J. Med. 369 (13): 1278–1279.
Chiaretti, S., Marinelli, M., Del Giudice, I. et al. (2014). NOTCH1,
SF3B1, BIRC3 and TP53 mutations in patients with chronic lympho-
cytic leukemia undergoing first-line treatment: correlation with bio-
logical parameters and response to treatment. Leuk. Lymphoma 55
(12): 2785–2792.
Cimmino, A., Calin, G.A., Fabbri, M. et al. (2005). miR-15 and miR-16
induce apoptosis by targeting BCL2. Proc. Natl. Acad. Sci. U S A 102
(39): 13944–13949.
Crombie, J. and Davids, M.S. (2017). IGHV mutational status testing in
chronic lymphocytic leukemia. Am. J. Hematol. 92 (12): 1393–1397.
Damle, R.N., Wasil, T., Fais, F. et al. (1999). Ig V gene mutation status
and CD38 expression as novel prognostic indicators in chronic lym-
phocytic leukemia. Blood 94 (6): 1840–1847.
Del Gaizo Moore, V., Brown, J.R., Certo, M. et al. (2007). Chronic
lymphocytic leukemia requires BCL2 to sequester prodeath BIM,
explaining sensitivity to BCL2 antagonist ABT-737. J. Clin. Invest. 117
(1): 112–121.
Dohner, H., Stilgenbauer, S., Benner, A. et al. (2000). Genomic aberra-
tions and survival in chronic lymphocytic leukemia. N. Engl. J. Med.
343 (26): 1910–1916.
Furman, R.R., Sharman, J.P., Coutre, S.E. et al. (2014). Idelalisib and rit-
uximab in relapsed chronic lymphocytic leukemia. N. Engl. J. Med.
370 (11): 997–1007.
Hamblin, T.J., Davis, Z., Gardiner, A. et al. (1999). Unmutated Ig V(H)
genes are associated with a more aggressive form of chronic lympho-
cytic leukemia. Blood 94 (6): 1848–1854.
Roberts, A.W., Davids, M.S., Pagel, J.M. et al. (2016). Targeting BCL2
with venetoclax in relapsed chronic lymphocytic leukemia. N. Engl.
J. Med. 374 (4): 311–322.
Wang, L., Lawrence, M.S., Wan, Y. et al. (2011). SF3B1 and other novel
cancer genes in chronic lymphocytic leukemia. N. Engl. J. Med. 365
(26): 2497–2506.
Hodgkin lymphoma
Ansell, S.M., Lesokhin, A.M., Borrello, I. et al. (2015). PD-1 block-
ade with nivolumab in relapsed or refractory Hodgkin’s lymphoma.
N. Engl. J. Med. 372 (4): 311–319.
Green, M.R., Monti, S., Rodig, S.J. et al. (2010). Integrative analysis
reveals selective 9p24.1 amplification, increased PD-1 ligand expres-
sion, and further induction via JAK2 in nodular sclerosing Hodgkin
lymphoma and primary mediastinal large B-cell lymphoma. Blood
116 (17): 3268–3277.
Chapter 9 The molecular biology of chronic
lymphocytic leukemia
John G. Gribben
Barts Cancer Institute, Barts and The London School of Medicine and Dentistry, Queen Mary University of London, London, UK

Introduction and definition, 111 BCL2 family member expression in CLL, 116
CLL pathogenesis, 111 The CLL microenvironment, 116
Epidemiology, 111 CLL and impaired immune responses, 116
CLL cell of origin, 112 Prognostic factors and their clinical utility, 117
Stereotype receptors, 113 Richter transformation, 118
Molecular genetics of CLL, 113 Conclusions, 119
The B-cell receptor and B-cell signaling, 115 Further reading, 119

signaling and the BCL2 inhibitor venetoclax. The activity of

Introduction and definition agents targeting these pathways provides strong evidence for
their importance in the pathogenesis and maintenance of this
Chronic lymphocytic leukemia (CLL) is characterized by
disease.
the clonal expansion of small mature-appearing malignant B
cells with a distinct immunophenotype, with co-expression
of CD19, CD5, and CD23, coupled with low levels of CD20 CLL pathogenesis
and surface immunoglobulins. This immunophenotype is
different from any normal B cell, making it more difficult The molecular pathogenesis of CLL is complex and incom-
to determine the cell of origin of this malignancy. Although pletely understood. It appears to be a complex interaction of
CLL often has an indolent clinical course, there is great het- genetic predisposition, poorly understood cancer-initiating
erogeneity in its clinical behavior, with some patients rapidly events in mature B cells, and the subsequent acquisition of a
progressing to need for therapy, while other patients live long variety of genetic and epigenetic changes, shaped by interac-
term with the disease with few symptoms and might never tion with the host microenvironment (Figure 9.1).
require treatment.
Research in CLL has identified a large number of prognos-
Epidemiology
tic factors that can be used to help identify those patients who
are at high risk for disease progression. Because of its indo-
CLL is the most common leukemia in the Western world,
lent course, it was previously felt that CLL cells resembled
but its incidence varies between individuals in different geo-
and arose from long-lived resting lymphocytes, but this has
graphical regions, and ranges from 0.06% of individuals in
been challenged by heavy water labeling experiments demon-
Europe and the USA to less than 0.01% of individuals in
strating that CLL often contains fractions of cells with rapid
eastern Asia. The risk of developing CLL is almost twice
turnover. This disease has therefore become a paradigm for
as high for men than for women. The risk of developing
the concept of a cancer living in a near-equilibrium of cells
CLL increases with age, with the median age at diagnosis of
driven by proliferation through the B-cell receptor (BCR)
around 72 years.
and ongoing cell death, driven by the expression of both
pro- and anti-apoptotic members of the BCL2 family of pro-
Hereditary factors
teins. An understanding of the pathophysiology of this dis-
ease has led to great advances in its treatment by the approval The finding that strong geographical variability does not alter
of agents such as ibrutinib and acalabrutinib targeting BCR in patients of Asian origin who live in Europe or the USA

Molecular Hematology, 4th Edition. Edited by Drew Provan and John G. Gribben.
© 2020 John Wiley & Sons Ltd. Published 2020 by John Wiley & Sons Ltd.

111
112 Molecular Hematology

Leukemia
initiating Microenvironment Secondary
event interactions lesion

Disease progression
Genetic Initiation Promotion/accumulation/ Chemorefractoriness
predisposition loss of immune control Richter transformation
Polygenic del13q Signaling pathways TP53
IRF4 +12 BCR NOTCH1
IRF8 NF-kB SF3B1
BCL2 TLR BIRC3
miR-15/16 CD38 ATM
VLA-4 Integrins MYC
NOTCH1 CDKN2A
CXCR4
Fig. 9.1 Chronic lymphocytic leukemia (CLL) pathogenesis and progression. CLL arises in the setting of polygenic predisposition and this
differs among geographical locations. Leukemia-initiating genetic events permit longer survival of B cells, which leads to subsequent accumulation,
resulting in subclonal genetic complexity.

suggests that genetic and not environmental factors are can increase the risk of CLL, and no evidence that dietary or
important. This is further supported by the familial prepon- lifestyle factors increase risk.
derance also seen in CLL. First-degree relatives of patients
with CLL have an 8.5-fold elevated risk of developing this dis-
ease, and the concordance of CLL is higher among monozy-
gotic than dizygotic twins. Genome-wide association studies
CLL cell of origin
in familial CLL have shown that this susceptibility is poly-
CLL can be divided into two major subsets, distinguished
genic, and single-nucleotide polymorphisms (SNPs) have
by whether CLL cells express immunoglobulin heavy chain
been identified in over 30 loci associated with familial CLL,
variable (IGHV) genes which have undergone somatic hyper-
suggesting that common genetic variations contribute to her-
mutation in their variable regions. Those cases that have less
itable risk and that these genetic variations differ in diverse
then 98% homology with germline are designated mutated
populations around the world. A number of these loci occur
chronic lymphocytic leukemia (M-CLL), and those that have
at sites of genes implicated in CLL pathogenesis. For exam-
98% or higher are designated unmutated chronic lympho-
ple, CLL-associated SNPs have been found at the BCL2 locus
cytic leukemia (UM-CLL). These are felt to reflect the stage
and also at IRF4, leading to reduced expression of interferon
of normal B cell differentiation from which the CLL cells
regulatory factor 4, implicated in Notch signaling. An SNP
originate.
associated with reduced expression of miR-15a and miR16-
At diagnosis, some 65% of CLL cases have mutated
1 is associated with familial CLL, and reduced expression of
IGHV, which are therefore proposed to arise from post–
these microRNA allows for increased expression of BCL2 and
germinal center B cells expressing immunoglobulin that has
ZAP70, proteins that confer increased resistance to cell death
undergone somatic hypermutation and, in some cases, also
or enhanced BCR-mediated signaling.
immunoglobulin isotype switching (Figure 9.2). The other 35
of cases express unmutated IGHV, suggesting origin from a
Environmental factors
B cell that has not undergone germinal center–driven differ-
Pesticide exposure might be a risk factor for CLL, and the US entiation. Patients with CLL cells with mutated IGHV typi-
Department of Veterans Affairs has accepted that exposure cally have a more indolent disease course than CLL patients
to Agent Orange is a risk factor for the disease. There is little whose cells express unmutated IGHV. Some CLL cases have
evidence to suggest that ionizing radiation or viral infections been described that are similar to unmutated IGHV CLL, but

Germinal center
Selection
Low/lost affinity
CD5+ Proliferation
B Cell and SHM Improved affinity
Dark Zone
Constrained
SHM
IGHV-unmutated CLL
Fig. 9.2 Chronic lymphocytic leukemia (CLL) cell of origin. A CD5+ B cell enters the germinal center and undergoes a process of affinity
maturation and class switching. Leukemogenesis events arising in cells at different stages within this pathway explain the mutated and unmutated
subtypes of CLL seen. SHM, somatic hypermutation.
originate from B cells with limited somatic mutation, such as
those encoded by mutated IGHV3-21.
Stereotype receptors
The repertoire of immunoglobulin (Ig) molecules produced
by CLL cells is considerably more limited than the reper-
toire of Ig molecules in normal B cells, reflecting biased
use of IGHV genes that have restricted somatic mutation
and limited junctional and heavy–light chain combinatorial
diversity. As many as 30% of patients have CLL cells that
express Ig “stereotypes,” and more than 1% of cases have
near-identity in their primary structure in the variable region
among patients. This limited Ig diversity provides compelling
evidence that CLL B cells are selected by the binding activity
of their expressed surface Ig, suggesting that B-cell receptor
signaling plays a critical role in CLL pathogenesis.
Molecular genetics of CLL
Unlike other B-cell malignancies, chromosomal translo-
cations are rare in CLL, but there are a number of
The molecular biology of chronic lymphocytic leukemia 113
Apoptosis
Class-switched
FDC Memory B cell
Class
switching
TH Cell
Light Zone IgM+IgD+
Memory
? B cell
IGHV-mutated
Ig+ CLL
IGHV-mutated
IgM+IgD+ CLL
CLL with limited SHM
(e.g. IGHV3-21/IGLV3-21)
well-characterized genetic alterations, including chromoso-
mal alterations, mutations, alterations in miRNA expression,
and epigenetic modifications.
Chromosomal alterations
More than 80% of CLL patients carry at least one of four
common chromosomal alterations. The most common is a
deletion in chromosome 13q14.3 (del13q), occurring in more
than 50% of patients and associated with a good progno-
sis. Deletion in chromosome 11 (del11q) is found in 18%
of patients, often associated with alterations in ATM, which
encodes a protein involved in DNA repair. Del 11q is asso-
ciated with more rapid disease progression, but has a good
response to modern therapy. Trisomy 12 is found in 16%
of patients with CLL and is associated with an intermediate
prognosis. Deletion in chromosome 17 (del17p) is found in
5–7% of patients, is associated with loss of the tumor suppres-
sor gene TP53, and has the worst prognostic impact.
Somatic mutations
The application of whole-exome sequencing (WES) and
whole-genome sequencing (WGS) has transformed our
114 Molecular Hematology

* Q < 0.05 • Q < 0.10 Subclonal-low Subclonal-high Clonal % cases based on mutations:
25
* 36%
86 14% (unmutated)
20
18%
(clonal +
22% subclonal)
15 •
% cases

10%
51 *
43
10 36 •
*
27 27 27 *
22 20 *
19 18 17 17
5 15 15 14 *
11 *
7 6 6 6 5 5 3 3 2 2 2
0
1

AF

F1
53

M 4
M 5
R2

X1

AS
ZN E

AS
92

X
6

1
M

O1

3
7
T1

GA

ZM 2
OR

2
88

3
1
3B

F
I

S1
HL

AF

M
RC
XW

1
X3

ND

YM
KB
CH

AT

NX

IR
TP

F2

EG

BR
PO

DT

NR
KR

ED
YD
XP

BC

PI
SF

RP

TR
KL
BI

DD
NF

CC
FB
T
NO

Fig. 9.3 Frequency of genetic events seen in chronic lymphocytic leukemia and demonstrated in a word cloud.
Source: Modified from Landau, D.A., Tausch, E., Taylor-Weiner, A.N. et al. (2015). Mutations driving CLL and their evolution in progression and
relapse. Nature 526(7574): 525–530.

understanding of CLL pathophysiology, and sequences of
more than 500 cases have been reported to date. CLL is one
of the tumors with the lowest background mutational load
(0.6 per Mb). It is genetically heterogeneous and there are
no unifying or diagnostic gene mutations (Figure 9.3). The
most frequently identified mutations were found in SF3B1
(21%), ATM (15%), TP53 (7%), NOTCH1 (6%), and BIRC3
(4%). Recurrent somatic mutations have been observed in
genes that have a role in mRNA processing (SF3B1 and
XPO1), DNA damage (TP53 and ATM), Notch signaling
(NOTCH1), B-cell signaling (EGR2 or BRAF), chromatin
modification (HIST1H1E, CHD2 and ZMYM3), Wnt signal-
ing, and inflammatory pathways (MYD88).
Next-generation sequencing has also revealed consider-
able intratumoral heterogeneity in CLL. Some somatic muta-
tions, such as those in MYD88, or chromosomal abnormal-
ities, such as trisomy 12 or del13q, are usually found in
almost all CLL cells in any one patient, suggesting that these
genetic alterations occurred early in disease evolution. Other
mutations, notable those found in SF3B1 or NOTCH1, TP53,
or del17p, are found in only a fraction of the CLL cells and
must therefore represent subclonal events which occur later
in disease evolution. These subclonal driver mutations are
associated with more aggressive disease, particularly when
two or more mutations are found concurrently. Single-cell
analysis is being used to determine whether these events
are occurring in the same cells or whether there are dif-
ferent clonal subpopulations, and whether mutations occur-
ring within the same cell are required to affect prognosis.
Of considerable interest are studies examining the impact of
chemotherapy and how this can induce large clonal shifts due
to increases in the proportions of CLL cells that have muta-
tions in genes conferring resistance to therapy, notable TP53
mutation or del17p.
MicroRNA alterations
CLL was the first human disease found to be associated
with alterations in microRNA (miRNA), with the identifi-
cation that miR-15a and miR-16-1 are deleted, altered, or

downregulated in up to 60% of cases of CLL and are dysfunc-
tional in some cases of familial CLL. Reduced expression or
loss of miR-15a and miR-16-1 can enhance the expression
of target genes such as BCL2 and MCL1, which encode
anti-apoptotic proteins of the BCL2 family. A number of
other miRNAs are dysregulated or differentially expressed
in subgroups with distinctive biological features, including
miR-29a/b, miR-29c, miR-34b, miR-181b, and miR-3676,
which target TCL1A. Notably, there is transgenic expression
of TCL1 in B cells which promotes the development of
CLL in mice. Increased expression of miR-155 is associated
with enhanced BCR signaling, B cell proliferation, and
lymphomagenesis.
Epigenetic changes
CLL cells exhibit global hypomethylation combined with
local hypermethylation. Methylation profiling has demon-
strated substantial heterogeneity of intratumoral methyla-
tion. Increasing methylation heterogeneity is associated with
increased genetic complexity, likely due to the acquisition
of subclonal mutations. There is therefore a link between
genetic and epigenetic evolution in CLL. UM-CLL and M-
CLL cells have distinctive methylation patterns, similar to
those of pre–germinal center versus post–germinal center
memory B cells, providing further evidence for the cell of ori-
gin, as shown in Figure 9.2. However, when methylation sig-
natures are used to classify distinct clinical CLL subgroups,
Antigen
BCR
CD79
A B
P P
P S P S P
Y Y
LYN K LYN K
P P
P PLC γ2
Fig. 9.4 Simplified B-cell receptor (BCR) signaling pathways and inhibitors of the pathway available to treat chronic lymphocytic leukemia.
The molecular biology of chronic lymphocytic leukemia 115
the data suggest that CLL cells derive from a continuum of B
cell maturation states.
The B-cell receptor and B-cell signaling
CLL cells typically co-express low levels of expression of sur-
face IgM and IgD, and relatively few cases of CLL express sur-
face IgG. The BCR in CLL comprises a heterodimer of the
transmembrane Ig molecule of IgM or IgD and the signal-
ing components CD79A (Igα) and CD79B (Igβ). Functional
BCR is required for CLL cell survival. Crosslinking of surface
Ig phosphorylates the immunoreceptor tyrosine-based acti-
vation motifs of CD79A and CD79B molecules and triggers
BCR-mediated signaling. The Ig of CLL B cells most likely
engages with auto-antigens, leading to constitutive BCR sig-
naling in vivo (Figure 9.4).
Enhanced BCR-mediated signaling is more commonly
observed in UM-CLL, whereas an anergic response pre-
dominates in M-CLL, perhaps accounting for the more
indolent behavior of M-CLL cases. BCR signaling also has
impacts on other signaling pathways, including integrins and
chemokines. The chemokine receptor CXCR4 is downmodu-
lated by BCR signaling, altering CLL cell adhesion and migra-
tion. This intrinsic dependence on BCR-mediated signaling
explains the great sensitivity of CLL to BCR inhibitors such as
Bruton’s tyrosine kinase (BTK) or phosphoinositide 3-kinase
(PI3K) inhibitors. Indeed, the success of treatment with the
Extracellular
PIP 2 PIP 3 Intracellular
AKT
PI3K δ
P mTOR
BTK P
PKC β
Idelalisib
IKK
Ibrutinib NF- K B
Acalabrutinib
116 Molecular Hematology
BTK inhibitor ibrutinib provides perhaps the strongest evi-
dence for the importance of BCR-mediated signaling in the
survival of CLL cells, and has activity in CLL cells with
TP53 abnormalities that do not respond well to conventional
chemotherapy.
BCL2 family member expression in CLL
The BCL2 protein family consists of at least 30 related pro-
teins characterized by the expression of BCL2 homology
domains. BCL2 family members are divided into three dif-
ferent subclasses based on structural and functional fea-
tures. Pro-survival or anti-apoptotic proteins include BCL2,
BCL-XL, and MCL1 proteins, which are all overexpressed
in CLL and are upregulated by contact of CLL cells with
the tumor microenvironment. The mechanism of expression
of BCL2 in CLL likely involves miRNA-mediated changes
in the malignant cells, and the promoter region of CBCL2
is hypomethylated in CLL, leading to increased expression.
The pro-apoptotic family members include BAX and BAD,
and the BH3-only protein members include BID, BIM, BAD,
NOVA, and PUMA. CLL cells express increased levels of both
pro- and anti-apoptotic proteins, so that these cells exist in a
complex relationship which is now able to be exploited in the
clinic (Figure 9.5).
Venetoclax is a small-molecule BH3 mimetic that inhibits
BCL-2. This drug is highly potent in inducing apoptosis in
CLL cells, by diminishing the capacity of BCL-2 to sequester
BIM. Venetoclax is effective in patients with relapsed or
refractory disease and in patients with relapsed disease and
TP53 mutations or loss. Studies are examining the use of
Pro-apoptotic
BCL-2 protein
BCL-2
Cancer cell survival
BCL-2 overexpression allows
cancer cells to evade apoptosis by
sequestering pro-apoptotic proteins.
Fig. 9.5 BCL2 family member expression in chronic lymphocytic leukemia and impact of sequestering BCL2 by venetoclax.
venetoclax with or without anti-CD20 monoclonal antibody,
and with or without ibrutinib, targeting both the BCR and
BCL2 pathways to provide perhaps the most effective treat-
ments currently available.
The CLL microenvironment
CLL depends upon survival signals that the cells receive
in the tumor microenvironment. Within lymph nodes,
CLL form “proliferation centers,” where they interact with
non-malignant stromal cells, CLL-associated macrophages
(Nurse-like cells), T cells, and mesenchymal-derived stromal
cells, among others (Figure 9.6). Within proliferation centers,
all CLL cells are exposed to chemokines, integrins, cytokines,
and survival factors including BAFF and APRIL, which acti-
vate canonical NF-κB, and induce the expression of miR-155,
thereby enhancing BCR-mediated signaling. Cytokines such
as IL-4 upregulate surface IgM, potentially facilitating inter-
actions of the CLL cell with auto-antigens. Notch or Hedge-
hog signaling provides pro-survival stimulation, particularly
in those cases with trisomy 12.
CLL and impaired immune responses
CLL is associated with significant immune suppression and
infectious complications are a major cause of morbidity
and mortality in this disease. An important aspect of CLL
is the development of hypogammaglobulinemia, with the
consequent increased risk of infection and poor response to
vaccination. The mechanism involved is unclear, but likely
Venetoclax
Apoptosis
initiation
Mi
toc
Pro-apoptotic
h
BIM
on
BAX
protein
dri
on
BAK
BAX
Cancer cell death
Activation
of caspases
Cytochrome c
Venetoclax binds selectively to BCL-2,
freeing pro-apoptotic proteins that initiate
programmed cell death (apoptosis).
 The molecular biology of chronic lymphocytic leukemia 117

Antigen

IL-10 CCR1/3
CCL3/4

CCR4
IL-4 Nurse-like cell
T cell CD267 (TACI) LAM
CD269 (BCMA)

CC
CD268 (BAFF-R) FF)

L
(BA L)

17
CCL slg 7
5 I
CD2 6 (APR

/22
3/4
2 5 1
CCR1/3 CD CD3
CD40L
CD40 IL-4

CCL19 CCR7 Cell CD38

B-cell CXCL13
CCL21 5
CXCR
CD38
CD44 CXCL12

4
CR
HA VLA-4
CX
Frz
HEV ROR2 ROR1
12
CXCL
VCA Wnt
M-1 Wnt5a

Strom
al cell

Fig. 9.6 Interactions with chronic lymphocytic leukemia (CLL) cells within the microenvironment. HEV, high endothelial venules; LAM,
lymphoma-associated macrophage.

represents replacement of CLL cells for normal B cells within
the immune microenvironment, as well as impairment of
T cell help. CLL cells express high levels of programmed
cell death 1 ligand 1 (PD-L1) and programmed cell death 1
ligand 2 (PD-L2), which suppress the effector responses of
T cells expressing programmed cell death protein 1 (PD-1)
and other immune checkpoint inhibitors, leading to an
“exhausted” T cell and NK cell phenotype and impaired
cellular immune function.
Prognostic factors and their
clinical utility
The clinical course of newly diagnosed CLL is extremely vari-
able, with some patients being rapidly symptomatic or devel-
oping high-risk disease, which requires treatment soon after
diagnosis, while others remain free of symptoms for decades.
Prognostic factors that can be used to identify patients
who require earlier treatment include clinical features, and
genetic, molecular, and biochemical characteristics of the
CLL cells. Multivariable models, prognostic indices, and
nomograms have been developed to consolidate the prog-
nostic factors. Commonly used parameters associated with
poorer outcome are late-stage disease at initial presen-
tation, male sex, increased age, poor performance status
(largely due to co-morbidities), and CLL cell characteristics,
including unmutated IGHV, increased expression of ZAP-70,
CD49d, or CD38, high serum beta-2-microglobulin, and/or
the presence of complex karyotype, del11q, del17p, or TP53
mutations.
The most reliable prognostic models are those developed
to predict the interval from diagnosis to time of first treat-
ment. There is also increasing interest in predictive models
to define outcome with specific types of therapy. Prognos-
tic markers that help predict response to specific treatments
are termed predictive biomarkers. TP53 and IGHV muta-
tional status are powerful predictive biomarkers for response
to chemoimmunotherapy. These markers are often specific
for the type of treatment being investigated, and both of them
118 Molecular Hematology
Diagnosis Relapse Refractoriness
Chemotherapy Chemotherapy
TP53 mutated
CLL cell
Small TP53 mutated Chemotherapy Subsequent expansion
subclone mixed with removes TP53 wt of the TP53 mutated clone
TP53 wt clones clones and selects
for the TP53 mutated
subclone
lose their predictive value when treatment with ibrutinib
or venetoclax is used. One of the most powerful predictive
biomarkers has been the emergence of del17p to TP53 muta-
tions (Figure 9.7). These changes are relatively rare at diagno-
sis, but recent WES data has suggested that small subclones
harboring these changes might more frequently be present
at low frequency. Subsequent treatment with chemotherapy
removes the sensitive CLL cells, leaving behind and enriching
for cells with mutations conferring resistance to chemother-
apy, so that over time the disease becomes more refractory to
treatment.
Treatment options are changing, and novel agents, notably
BCR-signaling inhibitors and BCL2 inhibitors, are highly
effective agents that have activity among patients who would
have been considered high risk when the only option was
conventional chemotherapy. These agents both have activity
in TP53 mutated cells. Ongoing clinical research is investigat-
ing whether earlier use of these agents will prevent the emer-
gence of chemorefractory clones.
Richter transformation
Richter transformation (RT) represents the clinicopatholog-
ical transformation of CLL into aggressive lymphoma, most
commonly diffuse large B-cell lymphoma (DLBCL), or more
rarely Hodgkin lymphoma (Figure 9.8). About 7% of CLL
patients develop RT, with an incidence rate of approximately
0.5% per year. In over half the patients with RT, the lym-
phoma cells with Richter transformation do not express CD5
or CD23, which are almost invariably expressed by CLL cells.
The DLBCL-like lymphomas in RT most often share the same
IGHV-D-J rearrangement as the original CLL clone and as
such may express unmutated IGHV, unlike de novo DLBCLs,
Fig. 9.7 Molecular mechanism of chemoresistance in
chronic lymphocytic leukemia. Chemotherapy selects
for expansion of subclones, such as TP53 mutant cells,
with change in subclonal architecture over time and after
chemotherapy.
which virtually always express IGHV with somatic mutations
as a post–germinal center malignancy.
RT cells have other distinctive genetic differences from de
novo DLBCL. About 60% of RT lymphomas have inactivating
mutations and/or deletions in TP53, often with deregulation
of MYC, which is observed in around 40% of cases. MYC
deregulation is usually caused by translocations juxtaposing
MYC to Ig loci, gene amplification of MYC at 8q24, or
somatic mutations affecting MYC transregulatory factors,
such as NOTCH1, which is mutated in around 30% of cases.
NOTCH1 mutations are largely mutually exclusive with MYC
oncogenic activation, consistent with the observation that
NOTCH1 directly stimulates MYC transcription and sug-
gesting that activation of oncogenic MYC may be a common
pathway in RS transformation. CDKN2A encodes the cell
cycle progression inhibitor p16, and is mutated or deleted
in roughly 30–50% of cases, but this rarely occurs in CLL or
de novo DLBCL. The high prevalence of TP53 disruption in
RS reflects the selection of a chemorefractory clone under
the pressure of previous CLL treatments, thus suggesting an
explanation for the poor outcome and the limited sensitivity
to conventional drugs. On the other hand, RT lymphomas
typically do not have mutations in a number of genes which
are commonly seen in de novo DLBCL, including those
encoding proteins involved in nuclear factor-κB signaling or
in transcriptional repressors such PRDM1/BLIMP1 or BCL6.
Up to 20% of the DLBCL-type Richter transformation
and ∼50% of Hodgkin lymphoma–type Richter transforma-
tion have IGHV rearrangements that differ from that of the
original CLL clone, suggesting that these lymphomas repre-
sent a de novo secondary malignancy; some of these seem
to be associated with Epstein–Barr virus infection, particu-
larly in patients with severe disease-related immune dysfunc-
tion and/or treatment-related immune suppression. Cases
 The molecular biology of chronic lymphocytic leukemia 119

(a) DLBCL
NOTCH1
MYC
mutations
translocation
amplification
MGA
mutations
MYC
activation
CDKN2A
loss

Transformation

CLL

TP53 Richter
disruption transformation

(b)
Clonally related
Richter
80%
CLL V4-39 D6 J4

V4-39 D6 J4

Clonally unrelated
Richter
V4-34 D2-2J3

20%

Fig. 9.8 Molecular mechanisms of Richter transformation in chronic lymphocytic leukemia (CLL). (a) Changes in TP53, activation of MYC,
and loss of CDKN2A are among the most frequent changes seen with diffuse large B-cell lymphoma (DLBCL) transformation in CLL. (b)
Representation of clonally related and de novo clonally unrelated DLBCL arising in CLL.

with unrelated IGHV rearrangements have better response to
chemotherapy and improved outcome.
Conclusions
Advances in our understanding of CLL pathophysiology help
explain the clinical heterogeneity that we see in this disease.
With the identification of pathways involved in the develop-
ment of CLL, an increasing array of prognostic and predic-
tive biomarkers are increasingly used in clinical practice. The
importance of BCR signaling in driving CLL cell prolifera-
tion and BCL2 overexpression to prevent apoptosis explains
the long survival of CLL cells. Identification of the impor-
tance of BCR signaling and BCL2 pathways to CLL survival
has led to the development of the most effective agents that
we have in this disease, and these novel agents are able to
overcome the poor prognostic markers that predict poor
response to conventional chemotherapy. These novel agents
targeting these pathways are very well tolerated and suit-
able for more elderly patients with CLL. With increased sur-
vival in CLL with these new agents, Richter transformation is
emerging as a major unmet need in CLL, and identification
of agents that will improve the outcome of these patients is
the focus of much ongoing research.
Further reading
Burger, J.A., Tedeschi, A., Barr, P.M. et al. (2015). Ibrutinib as initial
therapy for patients with chronic lymphocytic leukemia. N. Engl. J.
Med. 373 (25): 2425–2437.
Byrd, J.C., Brown, J.R., O’Brien, S. et al. (2014). Ibrutinib versus ofatu-
mumab in previously treated chronic lymphoid leukemia. N. Engl. J.
Med. 371 (3): 213–223.
Fischer, K., Bahlo, J., Fink, A.M. et al. (2016). Long-term remissions
after FCR chemoimmunotherapy in previously untreated patients
120 Molecular Hematology
with CLL: updated results of the CLL8 trial. Blood 127 (2): 208–
215.
Hallek, M., Cheson, B.D., Catovsky, D. et al. (2018). iwCLL guidelines
for diagnosis, indications for treatment, response assessment, and
supportive management of CLL. Blood 131 (25): 2745–2760.
Landau, D.A., Sun, C., Rosebrock, D. et al. (2017). The evolutionary
landscape of chronic lymphocytic leukemia treated with ibrutinib
targeted therapy. Nat. Commun. 8: 2185.
Landau, D.A., Tausch, E., Taylor-Weiner, A.N. et al. (2015). Mutations
driving CLL and their evolution in progression and relapse. Nature
526 (7574): 525–530.
Stilgenbauer, S., Eichhorst, B., Schetelig, J. et al. (2018). Venetoclax for
patients with chronic lymphocytic leukemia with 17p deletion: results
from the full population of a phase II pivotal trial. J. Clin. Oncol. 36
(19): 1973–1980.
Chapter 10 The molecular biology of
multiple myeloma
Wee Joo Chng1,2 & P. Leif Bergsagel3
1
National University Cancer Institute, National University Health System of Singapore, Singapore
2
University of Singapore, National University Hospital, Singapore
3 Division of Hematology-Oncology, Comprehensive Cancer Center, Mayo Clinic Arizona, Scottsdale AZ, USA
Introduction, 121
A plasmablast/plasma cell tumor of post–germinal center B cells, 121
Stages of plasma cell neoplasms, 121
Immunoglobulin translocations present in the majority of MM
tumors, 122
Marked karyotypic complexity in MM, 122
Chromosome content associated with at least two different pathogenic
pathways, 122
Recurrent IgH translocations and recurrent trisomies representing
primary oncogenic events, 123
Introduction
Multiple myeloma (MM) is an incurable post–germinal cen-
ter B-cell malignancy. In 2018, it is estimated that in the
USA 30 770 new cases will have been diagnosed, with 12 770
patients succumbing to the disease. It is almost always pre-
ceded by a pre-malignant tumor called monoclonal gam-
mopathy of undetermined significance (MGUS), which is the
most common lymphoid tumor in humans, occurring in
approximately 4% of individuals over the age of 50. The
prevalence of both MM and MGUS increases with age,
and is about twofold higher in African Americans than in
Caucasians, although the rate of progression from MGUS to
MM is similar in these two populations.
A plasmablast/plasma cell tumor of
post–germinal center B cells
Post–germinal center B cells that have undergone productive
somatic hypermutation, antigen selection, and immunoglob-
ulin heavy chain (IgH) switching can generate plasmablasts,
which typically migrate to the bone marrow, where the
microenvironment enables differentiation into long-lived
plasma cells. Importantly, MGUS and MM are monoclonal
tumors that are phenotypically similar to plasmablasts/
Molecular Hematology, 4th Edition. Edited by Drew Provan and John G. Gribben.
© 2020 John Wiley & Sons Ltd. Published 2020 by John Wiley & Sons Ltd.
Universal cyclin D dysregulation, 123
Molecular classification of MM, 124
Prognostic and therapeutic implications of molecular classifications, 124
Possible events mediating transformation of MGUS to MM, 125
Model of molecular pathogenesis of MM, 126
Critical role of the bone marrow microenvironment and novel
therapeutic strategies targeting the bone marrow milieu, 127
Conclusion, 128
Further reading, 128
long-lived plasma cells, including a strong dependence
on the bone marrow microenvironment for survival and
growth. In contrast to normal long-lived plasma cells, MGUS
and MM tumors retain some potential for an extremely low
rate of proliferation, usually with no more than a few percent
of cycling cells until advanced stages of MM.
Stages of plasma cell neoplasms
Based on the current diagnostic criteria proposed by the
International Myeloma Working Group, four stages of
plasma cell neoplasms can be identified based on the
presence of myeloma-defining events (hypercalcemia, renal
impairment, anemia, bone lesions, >60% bone marrow
plasma cells, involved/uninvolved free light chain ratio
>100, more than one focal lesion on magnetic resonance
imaging [MRI]), level of bone marrow plasma cell infil-
tration and serum or urine monoclonal immunoglobu-
lins, and extramedullary involvement. The stages include
MGUS, smoldering multiple myeloma (SMM), symptomatic
myeloma, and plasmacytoma (PCT) (Table 10.1). MGUS can
progress sporadically to MM expressing the same mono-
clonal immunoglobulins with a probability of about 0.6–3%
per year.
Through the analysis of several large prospective cohort
studies, three predictive factors for progression of MGUS
to MM were identified, including M-protein greater than
15 g l−1 , IgM or IgA M-protein, and presence of an abnormal
free light chain ratio. In the presence of all three risk factors,
121
122 Molecular Hematology
Table 10.1 Features used to diagnose monoclonal gammopathy of undetermined significance (MGUS), smoldering myeloma (SMM), and mul-
tiple myeloma (MM)
Serum Bone marrow plasma
M-protein cells (BMPC) Myeloma-defining events
MGUS <30 g l−1 <10% None
SMM ≥30 g l−1 10–60% None
MM Any ≥10% or PCT Hypercalcemia, anemia, renal impairment, bone lesions, >60% BMPC, involved/
uninvolved free light chain ratio >100, >1 focal lesion on magnetic resonance imaging
PCT Any <10% None, with biopsy-proven plasmacytoma (PCT)
the risk of progression at 20 years is 58%, compared with 5%
in the absence of any of these risk factors. The risk of pro-
gression to MM is higher for SMM than for MGUS, with a
median time to progression ranging from one to five years.
In one study, M-protein in excess of 30 g l−1 , presence of IgA
subtype, and urinary M-protein excretion above 30 g l−1 were
factors associated with early progression to MM.
Extramedullary MM is a more aggressive tumor that
can present as secondary or primary plasma cell leukemia,
depending on whether or not preceding intramedullary MM
has been recognized. Human multiple myeloma cell lines
(HMCLs), which are presumed to include most oncogenic
events involved in tumor initiation and progression of the
corresponding tumor, have been generated mainly from a
subset of extramedullary MM tumors.
Immunoglobulin translocations
present in the majority of MM tumors
Like other post–germinal center B cell tumors, translocations
involving the IgH locus (14q32) or one of the immunoglob-
ulin lambda (IgL) loci (κ, 2p12 or λ, 22q11) are common.
In general, these events are mediated by errors in one of the
three B-cell-specific DNA modification mechanisms: VDJ
recombination, IgH switch recombination, or somatic hyper-
mutation. With rare exceptions, these translocations result
in dysregulated or increased expression of an oncogene that
is positioned near one or more of the strong immunoglob-
ulin enhancers on the translocated derivative chromosome
14. However, translocations involving an IgH switch region
uniquely dissociate the intronic from one or both 3′ IgH
enhancers, so that an oncogene might be juxtaposed to an
IgH enhancer on either or both of the derivative chromo-
somes, as first demonstrated for FGFR3 on der(14) and
WHSC1 on der(4) in MM.
These IgH translocations are efficiently detected by fluo-
rescence in situ hybridization (FISH) analyses. Large stud-
ies from several groups show that the prevalence of IgH
translocations increases with disease stage: about 50% in
MGUS or SMM, 55–70% for intramedullary MM, 85% in
plasma cell leukemia (PCL), and above 90% in HMCLs. Lim-
ited studies indicate that IgL translocations are present in
about 10% of MGUS/SMM tumors, and in about 15–20%
of intramedullary MM tumors and HMCLs. Translocations
involving an Igκ locus are rare, occurring in only 1–2% of
MM tumors and HMCLs.
Marked karyotypic complexity in MM
The karyotypes of MM are characterized by complex abnor-
malities, both structural and numerical. Numerical chro-
mosomal abnormalities are present in virtually all MM
tumors and most, if not all, MGUS tumors. There is non-
random involvement of different chromosomes in different
myeloma tumors, and often heterogeneity among cells within
a tumor. The mechanism underlying this karyotypic instabil-
ity is not fully understood. Centrosome abnormalities, one of
the mechanisms mediating chromosomal instability in solid
tumors, have been identified in MGUS and about one-third
of MM tumors. However, mutations of genes involved in the
mitotic spindle checkpoint, another mechanism leading to
genomic instability in solid tumors, have not been identified
in MM. It is thought that karyotypic complexity increases
during tumor progression, although karyotypic progression
has not been well documented.
Chromosome content associated with
at least two different pathogenic
pathways
There is a clear consensus that chromosome content reflects
at least two pathways of pathogenesis. Approximately half of
tumors are hyperdiploid (HRD; 48–75 chromosomes), and
typically have multiple trisomies involving chromosomes 3,
5, 7, 9, 11, 15, 19, and 21, but only infrequently (<10%) have
one of the recurrent IgH translocations. Non-hyperdiploid
(NHRD) tumors (<48 and/or >75 chromosomes) usually
 The molecular biology of multiple myeloma 123

(∼70%) have one of the recurrent IgH translocations. These

Table 10.2 Recurrent primary immunoglobulin heavy chain (IgH)
patterns of genetic aberration are already present in MGUS, translocation partners in MM
suggesting that the HRD/non-HRD dichotomy is estab-
lished early during disease pathogenesis, probably at dis- Genomic Class of gene
ease initiation, and is dictated by initial oncogenic events. Gene partner locus dysregulated Frequency
Tumors that have a t(11;14) translocation may represent
a distinct category of NHRD tumors, as they are often CCND1 11q13 CCND 15%
diploid or pseudodiploid, sometimes with this transloca- CCND2 12p13 CCND <1%
CCND3 6p21 CCND 2%
tion as the only karyotypic abnormality detected by conven-
MAF 16q23 MAF 5%
tional cytogenetics. This turns out to be true for a propor-
MAFB 21q12 MAF 2%
tion of t(11;14) MM even when assessed by high-resolution MAFC 8q24.3 MAF <1%
aCGH. MMSET/FGFR3 4p16 MMSET and 15%
In contrast to the selective occurrence of recurrent IgH usually FGFR3
translocations in NHRD tumors, other genetic events –
17p loss or TP53 (gene product p53) mutations, RAS
mutations, secondary immunoglobulin translocations, MYC have not been adequately confirmed as essential for mainte-
translocations – often occur with a similar prevalence in nance of the tumor and/or as therapeutic targets.
HRD and NHRD tumors. Extramedullary MM tumors and In HRD MM, the recurrent trisomies of chromosomes 3,
HMCLs nearly always have an NHRD phenotype, consistent 5, 7, 9, 11, 15, 19, and 21 appear to be the earliest events.
with the hypothesis that HRD tumors are more stromal In a karyotypic analysis of a large number of HRD myelo-
cell dependent than NHRD tumors. We have virtually no mas, the presence of these trisomies is associated with the
information about the timing, mechanism, or molecular simplest karyotypes, often occurring as the only abnormal-
consequences of hyperdiploidy. We do not know if the extra ities, suggesting that they are the earliest acquired genetic
chromosomes are accumulated one at a time in sequential abnormalities in these tumors. The pattern of acquisition of
steps, or as one catastrophic event. For tumors that are HRD these chromosomal trisomies seems to be random. Further-
but have one of the recurrent translocations – most often more, FISH studies have shown that this pattern of trisomies
t(4;14) – we do not know if hyperdiploidy occurred before already exists in MGUS, therefore representing early events.
or after the translocation.

Universal cyclin D dysregulation

Recurrent IgH translocations and
recurrent trisomies representing
primary oncogenic events
There are seven recurrent chromosomal partners and onco-
genes that are involved in IgH translocations in approxi-
mately 40% of MM tumors. These translocations lead to the
overexpression of three classes of genes: cyclin D, MAF, and
MMSET/FGFR3 (Table 10.2). The recurrent translocation
breakpoints usually occur within or near switch regions, but
sometimes within or near VDJ sequences, suggesting that
these translocations are mediated by errors in IgH switch
recombination or somatic hypermutation. Since there is no
evidence that IgH switch recombination or somatic hyper-
mutation mechanisms are active in normal plasma cells or
plasma cell tumors, it is presumed that these translocations
usually represent primary, perhaps initiating, oncogenic
events as normal B cells pass through germinal centers. In
addition, the most common of these translocations, t(4;14),
t(11;14), and t(14;16), are already detected in MGUS, further
highlighting their early role in disease pathogenesis. With the
exception of FGFR3 (especially with an activating mutation)
and possibly MAF, the consequences of these translocations
Analysis of gene expression data suggests that almost all
cases of plasma cell neoplasm starting from the MGUS stage
express one or more of the cyclin D genes in an aberrant fash-
ion. Therefore, it has been proposed that dysregulation of a
cyclin D gene provides a unifying, early oncogenic event in
MGUS and MM. MGUS and MM appear closer to normal,
non-proliferating plasma cells than to normal proliferating
plasmablasts, for which 30% or more of the cells can be in
S phase, yet the expression level of cyclin D1, cyclin D2, or
cyclin D3 mRNA in MM and MGUS is distinctly higher than
in normal plasma cells, in fact at a level comparable to that
of cyclin D2 mRNA expressed in normal proliferating plas-
mablasts. About 25% of MM tumors have an IgH translo-
cation that directly dysregulates a cyclin D gene or a MAF
gene encoding a transcription factor that markedly upregu-
lates cyclin D2.
Although MM tumors with t(4;14) express moderately
high levels of cyclin D2, the cause of increased cyclin
D2 expression remains unknown. Whereas normal bone
marrow plasma cells express little or no detectable cyclin D1,
the majority of HRD tumors express cyclin D1 biallelically,
while most other tumors express increased levels of cyclin
124 Molecular Hematology
D2 compared with normal bone marrow plasma cells,
both by unknown mechanism. Only a few percent of MM
tumors do not express increased levels of a cyclin D gene
compared with normal plasma cells, but many of these
tumors appear to represent samples that are substantially
contaminated by normal cells, and another large fraction
of these tumors express little or no RB1, eliminating the
necessity of expressing a cyclin D gene.
Molecular classification of MM
The advent of genomic technologies has facilitated the global
examination of copy number, translocation, expression, and
mutation in a highly parallel fashion. The use of such high-
dimensional data has allowed the identification of novel
disease subtypes. In myeloma, the use of supervised and
unsupervised methods has led to the development of dif-
ferent molecular classifications. Interestingly, although the
two approaches are different, there are significant overlaps
between them, with the subcategories anchored by known
recurrent primary genetic abnormalities, providing strong
validation that the primary genetic events drive most of the
subsequent changes in cellular transcription in MM.
One of the first of these classifications, the translo-
cation/cyclin D (TC) classification, takes the supervised
approach and is based on the spiked expression of genes
deregulated by primary IgH translocations, and the univer-
sal overexpression of cyclin D genes either by these translo-
cations or via other mechanisms. The resultant classification
identifies eight groups of tumors: those with primary translo-
cations (designated 4p16, 11q13, 6p21, Maf), those that over-
express CCND1 and CCND2 either alone or in combination
(designated D1, D1 + D2, D2), and the rare cases that do not
overexpress any cyclin D genes (designated “None”). Most of
the patients with HRD MM fall within the D1 and D1 + D2
groups. This classification system therefore focuses on the
different kinds of mechanism that dysregulate a cyclin D gene
as an early and unifying event in pathogenesis.
There are several different unsupervised approaches, with
the best known developed at the University of Arkansas for
Medical Sciences (UAMS), which identified seven tumor
groups characterized by the co-expression of unique gene
clusters. Interestingly, these clusters also identify tumors
with t(4;14), MAF translocations, t(11;14), and t(6;14),
corresponding to the MS, MF, and CD-1 and/or CD-2
groups, respectively. In this analysis, t(11;14) and t(6;14) can
belong to either the CD-1 or CD-2 group, depending on the
expression of CD20 and other B-cell-related genes. This is
consistent with the finding that t(11;14) and t(6;14) have very
similar expression profiles, clinical profiles, and outcome. In
contrast to the TC classification, the UAMS classification
identifies HRD MM as a distinct group, HY. However, this
may be somewhat misleading, since the HY group, which is
about 28% of MM tumors, includes only about 60% of HRD
tumors. The distribution of the remaining HRD tumors
among the other six groups has not been clarified, although
most are probably in the PR group, defined by increased
expression of proliferation-related genes; and the LB group,
defined by low bone disease and lower expression of genes
associated with bone disease in MM, such as FRZB and
DKK1.
There is significant overlap between the TC classification
and the UAMS molecular classification. When TC classes are
assigned to the 414 MM cases assigned a UAMS molecular
classification in the original dataset, the MS and MF groups
correspond to the 4p16 and Maf groups, respectively, with
100% concordance. The CD-1 and CD-2 groups together cor-
respond to the 11q13 and 6p21 groups (88% concordance).
The HY group contains most of the TC D1 and D1 + D2 cases
(96% concordance). The bulk of the TC D2 cases (67%) fall
within the LB group. The PR group contains a mix of patients
from the different TC classes: 13% 4p16, 4% Maf, 9% 11q13,
21% D1, 19% D1 + D2, 28% D2, and 6% None. The PR group
is therefore enriched for CCND2-expressing tumors, consis-
tent with the hypothesis that these tumors are more aggres-
sive. The additional insights gained from the incorporation of
next-generation sequencing has been primarily to fill in the
details of the basic outline provided by the TC and UAMS
classifications, and are outlined below.
Prognostic and therapeutic
implications of molecular
classifications
The intrinsic properties of the tumor cell are informative in
predicting the prognosis and the response to existing ther-
apies. In particular, studies over recent years have clearly
established the importance of genetic abnormalities in prog-
nosis. For example, it has been well documented that an
unfavorable outcome is associated with each of the follow-
ing: hypodiploidy compared with hyperdiploidy; 17p13 dele-
tion, p53 mutation, t(4;14), and t(14;16). A recent large study
from the Intergroupe Francophone du Myélome group has
confirmed that t(4;14) and 17p13 deletion are independent
prognostic factors in newly diagnosed MM patients under-
going high-dose therapy with stem cell transplantation, and
can further dissect each of the International Staging System
(ISS) categories. Furthermore, it established that most of the
prognostic impact of chromosome 13 deletion is due to the
concurrent presence of t(4;14), as patients with chromosome
13 deletion without t(4;14) do not have adverse prognosis.
In patients treated with stem cell transplantation, the poor
prognosis associated with t(4;14) appears to be due to early
disease relapse. These patients derive little benefit from stem

cell transplantation and alternative therapeutic strategies are
needed. With the addition of sequencing information it has
now become clear that the poor prognosis associated with
del17p is only seen in patients who have concomitant inacti-
vation of p53, either through mutation or homozygous dele-
tion of the other allele. Biallelic p53 inactivation is seen at
diagnosis in 4% of patients, and identifies an ultra-high-risk
population with a median overall survival of less than two
years.
Recent studies have shown that treatment with bortezomib
overcomes the deleterious impact of chromosome 13 dele-
tion on prognosis. This has led to a proposal for risk-stratified
management akin to the management of acute leukemias,
where genetically defined high-risk patients will be treated
with an upfront bortezomib- or lenalidomide-based regimen,
whereas standard-risk patients will receive upfront high-dose
therapy, with stem cell transplantation after induction ther-
apy. In addition, the genes or pathways deregulated by the
specific genetic events could be potential novel therapeutic
targets. This is best exemplified by the clinical development
of FGFR3 inhibitor therapy for MM with t(4;14).
As the gene expression–based classifications have a strong
genetic basis, they could easily identify these high-risk
genetic groups, for example the 4p16 and MAF groups in
the TC classification and the PR, MS, and MF groups in the
UAMS classification, and identify the clinically important
molecular subtypes of myeloma. In addition, new insights
provided by these classifications could be exploited thera-
peutically. For example, the underlying cyclin D deregula-
tion potentially has an important therapeutic implication, as
differential targeting of cyclin D may be very useful and add
specificity to treatment. Indeed, some potential agents target-
ing cyclin D2 have been identified in a drug library screen.
Possible events mediating
transformation of MGUS to MM
Very few differences between MGUS and MM have been
identified at the genetic and molecular levels. All the pri-
mary genetic events, the recurrent IgH translocations and tri-
somies of HRD MM, are already present in MGUS. By impli-
cation, additional genetic events are therefore required for a
transformation to MM.
Loss of cell cycle control
Chromosome 13 deletion is one of the most common genetic
abnormalities in MM and is also found in MGUS at appar-
ently similar frequency. There does not appear to be any
role for haploinsufficiency of RB1, and inactivation of the
remaining allele is uncommon (1–2%), but when it develops
it is associated with poor outcome. Other components of the
The molecular biology of multiple myeloma 125
RB pathway can also be dysregulated in MM. Inactivation of
CDKN2C (p18), a critical gene for normal plasma cell devel-
opment, is likely to contribute to increased proliferation.
There is biallelic deletion of p18 in 30% of HMCLs and in
nearly 10% of tumors in the highest quintile of proliferation,
as determined by an expression-based proliferation index.
Forced expression of CDKN2C by retroviral infection of
HMCLs that express little or no endogenous p18 substan-
tially inhibits proliferation. Paradoxically, about 60% of
HMCLs and 60% of the more proliferative MM tumors have
increased expression of p18 compared with normal plasma
cells. There is evidence that the E2F transcription factor,
which is upregulated in association with increased prolif-
eration, increases the expression of p18, presumably as a
feedback mechanism. Apart from the lack of a functional RB1
protein in approximately 10% of HMCLs, the mechanism(s)
by which most HMCLs and proliferative tumors become
insensitive to increased p18 levels is not yet understood.
Activating RAS pathway mutations
An emerging genetic difference between MGUS and MM is
activating RAS pathway mutation. There is a high prevalence
of activating mutations of N- or K-RAS (40%), BRAF (7%),
and FGFR3 (2% of all patients, but 17% of t(4;14)) in newly
diagnosed MM tumors, with only a small increase occur-
ring during tumor progression. Sequencing reveals that the
mutations are often subclonal, and subclones with different
mutations of the pathway frequently coexist. The prevalence
is similar in HMCLs. Importantly, less than 5% of MGUS
tumors have RAS mutations, consistent with the hypothesis
that RAS mutations may mark, if not mediate, the MGUS to
MM transition.
MYC activation
One of the major differences in gene expression that has been
identified between MGUS and MM is the increased expres-
sion of MYC and its target genes in MM. Genomic stud-
ies have shown that there are frequent chromosome translo-
cations involving the MYC locus, resulting in increased
expression of MYC, in about 40% of newly diagnosed MM.
There are diverse partner loci involved in the translocations,
with one-third involving an immunoglobulin locus (most
frequently the heavy chain locus). The other two-thirds
involve plasma cell-specific super enhancers (e.g. FAM46C,
TXNDC5, FOXO3) from elsewhere in the genome. Many
of the rearrangements are cryptic, explaining why they were
not detected previously. They are also unevenly distributed
across primary genetic subtypes, with a low of 25% in t(11;14)
MM, and a high of 55% in HRD MM.
The biological validity of MYC in the progression of
MGUS to MM is provided by an MM mouse model,
126 Molecular Hematology
which utilizes activation-induced cytidine deaminase (AID)-
dependent somatic hypermutation to activate the MYC onco-
gene sporadically. In mice with a genetic predisposition to
developing monoclonal gammopathy (but not a control stain
without this predisposition), the sporadic activation of MYC
led to the development of MM. MYC has been shown to be
directly involved in DNA replication, binding and activat-
ing DNA replicative origins and regulating S-phase transition
during the cell cycle through a non-transcriptional mech-
anism. Therefore, in MM, activation of MYC may cooper-
ate with cyclin D/RB deregulation – which negates the early
G1 S cell cycle checkpoint that allows the establishment of
the initial limited clonal plasma cell expansion (MGUS) – to
overcome remaining cell cycle constraints, leading to further
clonal expansion and transformation to MM.
NFkB pathway
It has long been felt that activation of the NFκB pathway is
important in the pathogenesis of MM, but little is known
about the prevalence of NFκB activation or mechanisms that
cause it. An array of mutations that result in the constitu-
tive activation of the NFκB pathway have been identified in
about 20% of patient samples and about 50% of HMCLs.
The most common event is inactivating mutations of TRAF3
in 13% of patients. In addition, inactivating mutations of
TRAF2, BIRC2/BIR3 (gene product cIAP1/2), and CYLD
were identified. Chromosome translocations and amplifi-
cations resulting in activation of NFκB-inducing kinase
(NIK), CD40, LTBR, TACI, NFκB1, and NFκB2 were also
reported.
Although activation of both the canonical and non-
canonical pathways is seen, the preponderance of mutations
result most directly in increased processing of NFκB2 p100 to
p52 (i.e. activation of the non-canonical pathway). Depletion
of NIK with shRNAs directed against NIK results in inhibi-
tion of both the classical and alternative NFκB pathways, and
also growth inhibition. Half of primary MM tumors have an
expression signature of NFκB target genes, with activating
mutations identified in less than half of these patients. Pre-
sumably either other mutations or ligand-dependent interac-
tions in the bone marrow microenvironment are responsible
for the NFκB activation in the remaining patients.
Abnormalities of TP53 gene and
chromosome 17p loss
Mutations of TP53 are relatively rare in newly diagnosed
MM, occurring in approximately 5% of tumors. However,
the frequency of mutations appears to increase with disease
stage, and is about 30% in PCL and 65% in HMCLs. Deletion
(mainly monoallelic) of the TP53 locus, as detected by inter-
phase FISH, occurs in about 10% of MM and approximately
40% of PCL and HMCLs. As noted above, the poor prognosis
associated with del17 is entirely explained by a concomitant
inactivation of p53. There is no prognostic impact of del17p
in patients with intact p53.
Gain of chromosome 1q and loss of
chromosome 1p
A number of laboratories have determined, using a com-
bination of FISH, array comparative genomic hybridization
(aCGH), and gene expression profiling, that there is a gain
of sequences, and corresponding increased gene expression,
at 1q21 in 30–40% of tumors. These gains are concentrated
substantially in those tumors that have t(4;14) or t(14;16),
or have a high proliferation expression index. It has been
proposed that the increased proliferation in tumors with a
gain of 1q21 sequences is due to the increased expression
of CKS1B as a result of increased copy number. One might
expect to find other mechanisms, such as localized amplifi-
cation or a translocation, if increased CKS1B expression is a
cause of increased proliferation, but there is no evidence for
other mechanisms to increase CKS1B expression. Further-
more, CKS1B expression correlates closely with the expres-
sion of a number of proliferation genes, and therefore appears
to be a consequence rather than a cause of the proliferation.
So, it seems prudent to remain skeptical that CKS1B is the
gene targeted by the gain of 1q21 sequences. Furthermore,
amplification of chromosome 1q tends to involve the whole q
arm, and on aCGH the minimally deleted areas do not always
include CKS1B.
Chromosome 1p loss is also common, occurring in about
46% of patients when analyzed by aCGH and in 7–36% in
previous karyotype studies. Unlike 1q gain, 1p loss is usually
interstitial and involves fragments of varying length. Chro-
mosome 1p loss has been shown to be a marker of poor prog-
nosis in cytogenetic studies.
Model of molecular pathogenesis
of MM
Based on the results just summarized, a model for the molec-
ular pathogenesis of MM has been proposed. Chromosome
content appears to identify two different, but perhaps over-
lapping, pathways of pathogenesis: NHRD tumors and HRD
tumors. In about 40% of tumors, a primary chromosome
translocation results in the dysregulated expression of an
oncogene. These primary genetic events lead to cyclin D
dysregulation, either directly – cyclin D1 in t(11;14) and
cyclin D3 in t(6;14) – or indirectly – cyclin D2 in t(4;14) and
maf translocations – or by an as yet unknown mechanism –
cyclin D1 and/or D2 in HRD MM. The dysregulation of one
of three cyclin D genes may render the cells more susceptible
 The molecular biology of multiple myeloma 127

Table 10.3 Well-characterized genetic subtypes of multiple myeloma

Heavy Light RAS 13 TRAF3

Genetics Ploidy Prognosis chain chain Morphology CD20 mutation deletion mutation Bone CCND

t(11;14) NHRD/ Good G κ Lymphoplasmacytoid ++ ++ −/+ + ++ D1

t(6;14) diploid D3
t(4;14) NHRD Poor A λ Plasmablastic − − +++ ++ −/+ D2
t(14;16) NHRD Poor A λ Plasmablastic + − ++ −/+ −/+ D2
t(8;14)
t(14;20)
Trisomies HRD Good G κ Normal − ++ −/+ −/+ ++ D1

HRD, hyperdiploid; NHRD, non-hyperdiploid.

to proliferative stimuli, resulting in selective expansion as
a result of interaction with bone marrow stromal cells that
produce Interleukin 6 (IL-6) and other cytokines. These
primary genetic events, primary IgH translocations and
trisomies of chromosomes 3, 5, 7, 9, 11, 15, 19, and 21, are
already present at the earliest identified stage of tumorigen-
esis, and define subtypes of MM with unique clinical (bone
disease, heavy chain subtypes, prognosis), molecular (types
of cyclin D expressed, associated mutations), pathological
(morphology, CD expression), and cytogenetic (ploidy)
features (Table 10.3).
A second genetic “hit” leading to subsequent transfor-
mation from MGUS to MM may be mediated by activation
of MYC or further deregulation of the RB pathway. The
MYC pathway may be activated by RAS mutation, IgH-MYC
translocations, or other as yet unidentified mechanisms.
The RB pathway may be further deregulated by RB hap-
loinsufficiency or p18 loss. These mechanisms are not
mutually exclusive and the pattern of cooperation is still not
yet understood. There appears to be enrichment for some
pathway deregulation, depending on primary genetic abnor-
malities. For example, tumors with cyclin D2 overexpression
are significantly enriched for chromosome 13 deletion and
hence RB haploinsufficiency, whereas RAS mutations are
significantly associated with t(11;14) tumors. NFκB appears
to be an alternate pathway, probably important for providing
the growth and survival signals for maintenance of tumor
cell survival, growth, and proliferation. It can be activated
through external signals from the tumor microenvironment,
such as B cell activating factor (BAFF), or signaling triggered
by contact between the malignant plasma cells and the
accessory cells in the bone marrow microenvironment.
Alternative mechanisms intrinsic to the tumor cells such
as gene mutations may also activate these pathways (e.g.
mutations in the NFκB pathway).
Later events that may lead to more aggressive tumor char-
acteristics and emancipation from dependence on the bone
marrow microenvironment include mutation of TP53 or
deletion of the TP53 locus on chromosome 17p13, and com-
plex secondary translocations. In the future, it would be
important to further interrogate how these pathways inter-
act with each other, and differentiate the critical from non-
critical pathways during tumorigenesis that should be tar-
geted therapeutically.
Critical role of the bone marrow
microenvironment and novel
therapeutic strategies targeting the
bone marrow milieu
As already mentioned, different molecular pathways can be
activated in the malignant plasma cells by intrinsic or extrin-
sic mechanisms, either through microenvironmental sig-
nals derived from cytokines or via interaction between the
tumor cells and accessory cells in the bone marrow microen-
vironment, such as stromal cells, endothelial cells, osteo-
clasts, and osteoblasts. Direct contact between MM cells
and cells of the bone marrow milieu induce paracrine or
autocrine release of cytokines and growth factors such as IL-
6, insulin-like growth factor, hepatocyte growth factor, and
vascular endothelial growth factor, which provide further
proliferative and survival signals by activating intracellular
pathways such as PI3K/AKT/mTOR/p70S6K, IKK-α/NFκB,
RAS/RAF/MAPK, and JAK/STAT3.
These interactions and the activated pathways not only
provide growth and survival signals, but also result in the
uncoupling of new bone formation from bone resorption,
leading to the formation of lytic bone lesions and resis-
tance of the tumor cells to different chemotherapeutic agents.
The importance of the bone marrow microenvironment
in MM biology is highlighted by the spate of new drugs
being developed to target these cytokines/growth factors or
contact-mediated pathways (Table 10.4). The Food and Drug
128 Molecular Hematology

by the multistage accumulation of genetic abnormalities

Table 10.4 Therapeutic compounds targeting the bone marrow
microenvironment
deregulating different pathways. Much of this knowledge
is already being utilized for diagnosis, prognosis, and risk
stratification of patients. Importantly, from a clinical stand-
Target pathway-activated therapy
point, this knowledge has led to the development of novel
therapeutic strategies, some of which are already in clinical
Targeting growth factors and cytokines
use, and many others showing promise in pre-clinical and
IL-6 Ras/Raf/Mek/Erk IL-6 antibodies
early clinical studies.
JAK2/STAT3 IL-6R antibodies
PI3K/AKT IL-6 superantagonist
IGF-1 Ras/Raf/Mek/Erk IGF1R kinase inhibitor
NFκB IGF1R antibodies Further reading
PI3K/AKT IGF-1 ligand antibodies
VEGF Ras/Raf/Mek/Erk VEGFR tyrosine kinase
General
inhibitor
β-Catenin Wnt PKF115-584 Chng, W.J., Glebov, O., Bergsagel, P.L. et al. (2007). Genetic events in the
DKK1 Wnt Anti-DKK1 neutralizing pathogenesis of multiple myeloma. Best Pract. Res. Clin. Haematol.
antibodies 20: 571–596.
Targeting interactions between multiple myeloma plasma cells Dispenzieri, A., Katzmann, J.A., Kyle, R.A. et al. (2010). Prevalence
and accessory cells and risk of progression of light-chain monoclonal gammopathy of
Adhesion molecules Proteasome inhibitors undetermined significance: a retrospective population-based cohort
Cell adherence Proteasome inhibitors, study. Lancet 375: 1721–1728.
immunomodulatory drugs Kawano, Y., Moschetta, M., Manier, S. et al. (2015). Targeting the bone
(IMiDs) marrow microenvironment in multiple myeloma. Immunol. Rev. 263:
Intracellular signaling molecules 160–172.
PI3K PI3 kinase inhibitor Kuehl, W.M. and Bergsagel, P.L. (2012). Molecular pathogenesis of mul-
AKT Perifosine tiple myeloma and its premalignant precursor. J. Clin. Invest. 122:
mTOR Rapamycin, CCI-779 3456–3463.
MEK/ERK MEK1/2 inhibitor Kyle, R.A., Larson, D.R., Therneau, T.M. et al. (2018). Long-term follow-
BCL2/ BCL2 inhibitor (venetoclax) up of monoclonal gammopathy of undetermined significance. N.
MCL1 In clinical trials Engl. J. Med. 378: 241–249.
Morgan, G.J., Walker, B.A., and Davies, F.E. (2012). The genetic archi-
Surface molecules
tecture of multiple myeloma. Nat. Rev. Cancer 12: 335–348.
CD38 Daratumumab, isatuximab,
Nair, S., Branagan, A.R., Liu, J. et al. (2016). Clonal immunoglobulin
CAR-T, bispecifics
against lysolipids in the origin of myeloma. N. Engl. J. Med. 374: 555–
SLAMF7 Elotuzumab
561.
BCMA GSK2857915, multiple
Siegel, R.L., Miller, K.D., and Jemal, A. (2018). Cancer statistics, 2018.
CAR-T, and bispecific
CA Cancer J. Clin. 68: 7–30.
antibodies
Weiss, B.M., Abadie, J., Verma, P. et al. (2009). A monoclonal gammopa-
FcRH5 Monoclonal antibodies,
thy precedes multiple myeloma in most patients. Blood 113: 5418–
bispecifics, CAR-T
5422.
IGF, insulin-like growth factor; PI3K, phosphatidylinositol 3-kinase;
VEGF, vascular endothelial growth factor.
Genetics
Bolli, N., Avet-Loiseau, H., Wedge, D.C. et al. (2014). Heterogeneity of
Administration–approved drugs thalidomide, bortezomib, genomic evolution and mutational profiles in multiple myeloma. Nat.
and lenalidomide are all thought to exert their therapeutic Commun. 5: 2997.
effect at least partly through targeting the microenvironment. Fonseca, R., Debes-Marun, C.S., Picken, E.B. et al. (2003). The recur-
rent IgH translocations are highly associated with nonhyperdiploid
variant multiple myeloma. Blood 102: 2562–2567.
Hurt, E.M., Wiestner, A., Rosenwald, A. et al. (2004). Overexpression
Conclusion of c-maf is a frequent oncogenic event in multiple myeloma that pro-
motes proliferation and pathological interactions with bone marrow
Significant progress has been made in understanding the stroma. Cancer Cell 5: 191–199.
molecular pathogenesis and biology of MM. Oncogenic Zingone, A., Cultraro, C.M., Shin, D.-M. et al. (2010). Ectopic expres-
pathways can be activated through cell-intrinsic or -extrinsic sion of wild-type FGFR3 cooperates with MYC to accelerate devel-
mechanisms. Similar to other cancers, MM is characterized opment of B-cell lineage neoplasms. Leukemia 24: 1171–1178.

Secondary genetic events
Annunziata, C.M., Davis, R.E., Demchenko, Y. et al. (2007). Frequent
engagement of the classical and alternative NF-kappaB pathways by
diverse genetic abnormalities in multiple myeloma. Cancer Cell 12:
115–130.
Boyd, K.D., Ross, F.M., Walker, B.A. et al. (2011). Mapping of chro-
mosome 1p deletions in myeloma identifies FAM46C at 1p12 and
CDKN2C at 1p32.3 as being genes in regions associated with adverse
survival. Clin. Cancer Res. 17: 7776–7784.
Chiecchio, L., Dagrada, G.P., Ibrahim, A.H. et al. (2009). Timing of
acquisition of deletion 13 in plasma cell dyscrasias is dependent on
genetic context. Haematologica 94: 1708–1713.
Flores-Montero, J., de Tute, R., Paiva, B. et al. (2016). Immunophenotype
of normal vs. myeloma plasma cells: toward antibody panel specifi-
cations for MRD detection in multiple myeloma. Cytometry B Clin.
Cytom. 90: 61–72.
Halliley, J.L., Tipton, C.M., Liesveld, J. et al. (2015). Long-lived plasma
cells are contained within the CD19(−)CD38(hi)CD138(+) subset in
human bone marrow. Immunity 43: 132–145.
Jourdan, M., Cren, M., Robert, N. et al. (2014). IL-6 supports the
generation of human long-lived plasma cells in combination with
either APRIL or stromal cell-soluble factors. Leukemia 28: 1647–
1656.
Keats, J.J., Fonseca, R., Chesi, M. et al. (2007). Promiscuous mutations
activate the noncanonical NF-kappaB pathway in multiple myeloma.
Cancer Cell 12: 131–144.
Kuo, A.J., Cheung, P., Chen, K. et al. (2011). NSD2 links dimethylation
of histone H3 at lysine 36 to oncogenic programming. Mol. Cell 44:
609–620.
Lode, L., Eveillard, M., Trichet, V. et al. (2010). Mutations in TP53 are
exclusively associated with del(17p) in multiple myeloma. Haemato-
logica 95: 1973–1976.
Martinez-Garcia, E., Popovic, R., Min, D.-J. et al. (2011). The MMSET
histone methyl transferase switches global histone methylation and
alters gene expression in t(4;14) multiple myeloma cells. Blood 117:
211–220.
Pei, H., Wu, X., Liu, T. et al. (2013). The histone methyltransferase
MMSET regulates class switch recombination. J. Immunol. 190: 756–
763.
Pei, H., Zhang, L., Luo, K. et al. (2011). MMSET regulates histone
H4K20 methylation and 53BP1 accumulation at DNA damage sites.
Nature 470: 124–128.
Salhia, B., Baker, A., Ahmann, G. et al. (2010). DNA methylation analy-
sis determines the high frequency of genic hypomethylation and low
frequency of hypermethylation events in plasma cell tumors. Cancer
Res. 70: 6934–6944.
Shah, M.Y., Martinez-Garcia, E., Phillip, J.M. et al. (2016).
MMSET/WHSC1 enhances DNA damage repair leading to an
increase in resistance to chemotherapeutic agents. Oncogene 35:
5905–5915.
Steinbrunn, T., Stuhmer, T., Gattenlohner, S. et al. (2011). Mutated RAS
and constitutively activated Akt delineate distinct oncogenic path-
ways, which independently contribute to multiple myeloma cell sur-
vival. Blood 117: 1998–2004.
Walker, B.A., Wardell, C.P., Chiecchio, L. et al. (2011). Aberrant global
methylation patterns affect the molecular pathogenesis and prognosis
of multiple myeloma. Blood 117: 553–562.
The molecular biology of multiple myeloma 129
MYC
Affer, M., Chesi, M., Chen, W.D. et al. (2014). Promiscuous MYC locus
rearrangements hijack enhancers but mostly super-enhancers to dys-
regulate MYC expression in multiple myeloma. Leukemia 28: 1725–
1735.
Chesi, M., Robbiani, D.F., Sebag, M. et al. (2008). AID-dependent acti-
vation of a MYC transgene induces multiple myeloma in a conditional
mouse model of post-germinal center malignancies. Cancer Cell 13:
167–180.
Chng, W.J., Huang, G.F., Chung, T.H. et al. (2011). Clinical and biolog-
ical implications of MYC activation: a common difference between
MGUS and newly diagnosed multiple myeloma. Leukemia 25: 1026–
1035.
Dib, A., Gabrea, A., Glebov, O.K. et al. (2008). Characterization of MYC
translocations in multiple myeloma cell lines. J. Natl. Cancer Inst. 39:
25–31.
Gabrea, A., Martelli, M.L., Qi, Y. et al. (2008). Secondary genomic rear-
rangements involving immunoglobulin or MYC loci show similar
prevalences in hyperdiploid and nonhyperdiploid myeloma tumors.
Genes Chromosomes Cancer 47: 573–590.
Molecular classification
Bergsagel, P.L., Kuehl, W.M., Zhan, F. et al. (2005). Cyclin D dysregu-
lation: an early and unifying pathogenic event in multiple myeloma.
Blood 106: 296–303.
Broyl, A., Hose, D., Lokhorst, H. et al. (2010). Gene expression profiling
for molecular classification of multiple myeloma in newly diagnosed
patients. Blood 116: 2543–2553.
Keats, J.J., Chesi, M., Egan, J.B. et al. (2012). Clonal competition with
alternating dominance in multiple myeloma. Blood 120: 1067–1076.
Lohr, J.G., Stojanov, P., Carter, S.L. et al. (2014). Widespread genetic
heterogeneity in multiple myeloma: implications for targeted therapy.
Cancer Cell 25: 91–101.
Mikulasova, A., Wardell, C.P., Murison, A. et al. (2017). The spectrum
of somatic mutations in monoclonal gammopathy of undetermined
significance indicates a less complex genomic landscape than that in
multiple myeloma. Haematologica 102: 1617–1625.
Rasche, L., Chavan, S.S., Stephens, O.W. et al. (2017). Spatial
genomic heterogeneity in multiple myeloma revealed by multi-region
sequencing. Nat. Commun. 8: 268.
Walker, B.A., Boyle, E.M., Wardell, C.P. et al. (2015). Mutational spec-
trum, copy number changes, and outcome: results of a sequencing
study of patients with newly diagnosed myeloma. J. Clin. Oncol. 33:
3911–3920.
Walker, B.A., Mavrommatis, K., Wardell, C.P. et al. (2018). A high-risk,
double-hit, group of newly diagnosed myeloma identified by genomic
analysis. Leukemia 2019; 33: 159–170.
Walker, B.A., Mavrommatis, K., Wardell, C.P. et al. (2018). Identifica-
tion of novel mutational drivers reveals oncogene dependencies in
multiple myeloma. Blood 132: 587–597.
Weinhold, N., Ashby, C., Rasche, L. et al. (2016). Clonal selection and
double-hit events involving tumor suppressor genes underlie relapse
in myeloma. Blood 128: 1735–1744.
Zhan, F., Huang, Y., Colla, S. et al. (2006). The molecular classification
of multiple myeloma. Blood 108: 2020–2028.
130 Molecular Hematology
Prognosis
Avet-Loiseau, H., Attal, M., Campion, L. et al. (2012). Long-term anal-
ysis of the IFM 99 trials for myeloma: cytogenetic abnormalities
[t(4;14), del(17p), 1q gains] play a major role in defining long-term
survival. J. Clin. Oncol. 30: 1949–1952.
Brioli, A., Kaiser, M.F., Pawlyn, C. et al. (2013). Biologically defined risk
groups can be used to define the impact of thalidomide maintenance
therapy in newly diagnosed multiple myeloma. Leuk. Lymphoma 54:
1975–1981.
Kumar, S., Fonseca, R., Ketterling, R.P. et al. (2012). Trisomies in mul-
tiple myeloma: impact on survival in patients with high-risk cytoge-
netics. Blood 119: 2100–2105.
Nair, B., van Rhee, F., Shaughnessy, J.D. et al. (2010). Superior results of
total therapy 3 (2003–33) in gene expression profiling-defined low-
risk multiple myeloma confirmed in subsequent trial 2006–66 with
VRD maintenance. Blood 115: 4168–4173.
Thanendrarajan, S., Tian, E., Qu, P. et al. (2017). The level of dele-
tion 17p and bi-allelic inactivation of TP53 has a significant impact
on clinical outcome in multiple myeloma. Haematologica 102: e364–
e367.
Targeted therapy
Andrulis, M., Lehners, N., Capper, D. et al. (2013). Targeting the BRAF
V600E mutation in multiple myeloma. Cancer Discov. 3: 862–869.
Bjorklund, C.C., Lu, L., Kang, J. et al. (2015). Rate of CRL4(CRBN) sub-
strate Ikaros and Aiolos degradation underlies differential activity of
lenalidomide and pomalidomide in multiple myeloma cells by regu-
lation of c-Myc and IRF4. Blood Cancer J. 5: e354.
Chesi, M., Matthews, G.M., Garbitt, V.M. et al. (2012). Drug response
in a genetically engineered mouse model of multiple myeloma is pre-
dictive of clinical efficacy. Blood 120: 376–385.
Delmore, J.E., Issa, G.C., Lemieux, M.E. et al. (2011). BET bromod-
omain inhibition as a therapeutic strategy to target c-Myc. Cell 146:
904–917.
Demchenko, Y.N., Brents, L.A., Li, Z. et al. (2014). Novel inhibitors are
cytotoxic for myeloma cells with NFkB inducing kinase-dependent
activation of NFkB. Oncotarget 5: 4554–4566.
Heuck, C.J., Jethava, Y., Khan, R. et al. (2016). Inhibiting MEK in MAPK
pathway-activated myeloma. Leukemia 30: 976–980.
Lee, L., Draper, B., Chaplin, N. et al. (2017, 2018). An APRIL based
chimeric antigen receptor for dual targeting of BCMA and TACI in
multiple myeloma. Blood 2018; 131: 746–758.
Mulligan, G., Lichter, D.I., Di Bacco, A. et al. (2014). Mutation of NRAS
but not KRAS significantly reduces myeloma sensitivity to single-
agent bortezomib therapy. Blood 123: 632–639.
Qing, J., Du, X., Chen, Y. et al. (2009). Antibody-based targeting of
FGFR3 in bladder carcinoma and t(4;14)-positive multiple myeloma
in mice. J. Clin. Invest. 119: 1216–1229.
Seckinger, A., Delgado, J.A., Moser, S. et al. (2017). Target expression,
generation, preclinical activity, and pharmacokinetics of the BCMA-
T cell bispecific antibody EM801 for multiple myeloma treatment.
Cancer Cell 31: 396–410.
Tai, Y.T., Mayes, P.A., Acharya, C. et al. (2014). Novel anti-B-cell mat-
uration antigen antibody-drug conjugate (GSK2857916) selectively
induces killing of multiple myeloma. Blood 123: 3128–3138.
Trudel, S., Li, Z.H., Wei, E. et al. (2005). CHIR-258, a novel, multitar-
geted tyrosine kinase inhibitor for the potential treatment of t(4;14)
multiple myeloma. Blood 105: 2941–2948.
Weinhold, N., Heuck, C.J., Rosenthal, A. et al. (2016). Clinical value of
molecular subtyping multiple myeloma using gene expression profil-
ing. Leukemia 30: 423–430.
Current therapy
Attal, M., Lauwers-Cances, V., Hulin, C. et al. (2017). Lenalidomide,
bortezomib, and dexamethasone with transplantation for myeloma.
N. Engl. J. Med. 376: 1311–1320.
Attal, M., Lauwers-Cances, V., Marit, G. et al. (2012). Lenalidomide
maintenance after stem-cell transplantation for multiple myeloma. N.
Engl. J. Med. 366: 1782–1791.
Dimopoulos, M.A., Oriol, A., Nahi, H. et al. (2016). Daratumumab,
lenalidomide, and dexamethasone for multiple myeloma. N. Engl. J.
Med. 375: 1319–1331.
Lokhorst, H.M., Plesner, T., Laubach, J.P. et al. (2015). Targeting CD38
with daratumumab monotherapy in multiple myeloma. N. Engl. J.
Med. 373: 1207–1219.
Lonial, S., Dimopoulos, M., Palumbo, A. et al. (2015). Elotuzumab ther-
apy for relapsed or refractory multiple myeloma. N. Engl. J. Med. 373:
621–631.
Mateos, M.V., Dimopoulos, M.A., Cavo, M. et al. (2018). Daratumumab
plus bortezomib, melphalan, and prednisone for untreated myeloma.
N. Engl. J. Med. 378: 518–528.
Mateos, M.V., Hernandez, M.T., Giraldo, P. et al. (2013). Lenalidomide
plus dexamethasone for high-risk smoldering multiple myeloma. N.
Engl. J. Med. 369: 438–447.
McCarthy, P.L., Owzar, K., Hofmeister, C.C. et al. (2012). Lenalidomide
after stem-cell transplantation for multiple myeloma. N. Engl. J. Med.
366: 1770–1781.
Mikhael, J.R., Dingli, D., Roy, V. et al. (2013). Management of newly
diagnosed symptomatic multiple myeloma: updated Mayo Stratifica-
tion of Myeloma and Risk-Adapted Therapy (mSMART) consensus
guidelines 2013. Mayo Clin. Proc. 88: 360–376.
Moreau, P., Masszi, T., Grzasko, N. et al. (2016). Oral ixazomib, lenalido-
mide, and dexamethasone for multiple myeloma. N. Engl. J. Med. 374:
1621–1634.
Palumbo, A., Cavallo, F., Gay, F. et al. (2014). Autologous transplanta-
tion and maintenance therapy in multiple myeloma. N. Engl. J. Med.
371: 895–905.
Palumbo, A., Chanan-Khan, A., Weisel, K. et al. (2016). Daratumumab,
bortezomib, and dexamethasone for multiple myeloma. N. Engl. J.
Med. 375: 754–766.
San Miguel, J.F., Schlag, R., Khuageva, N.K. et al. (2008). Bortezomib
plus melphalan and prednisone for initial treatment of multiple
myeloma. N. Engl. J. Med. 359: 906–917.
Stewart, A.K., Rajkumar, S.V., Dimopoulos, M.A. et al. (2015). Carfil-
zomib, lenalidomide, and dexamethasone for relapsed multiple
myeloma. N. Engl. J. Med. 372: 142–152.
Weber, D.M., Chen, C., Niesvizky, R. et al. (2007). Lenalidomide plus
dexamethasone for relapsed multiple myeloma in North America. N.
Engl. J. Med. 357: 2133–2142.
Chapter 11 The molecular basis of bone
marrow failure syndromes and red cell
enzymopathies
Deena Iskander1 , Lucio Luzzatto2 & Anastasios Karadimitris2
1
Centre for Haematology, Imperial College London, Hammersmith Hospital, London, UK
2 Department of Haematology and Blood Transfusion, Muhimbili University College of Health Sciences, Dar-es-Salaam, Tanzania
Introduction, 131
Conditions associated with bone marrow failure, 131
Red cell enzyme deficiencies, 144
Introduction
Since the last edition of this book the pace of advances in
molecular hematology has been such that, after due consider-
ation, we had to decide that, within the space allowed, it has
become impossible to do justice to the original title of this
chapter, “The Molecular Basis of Anemia.” We have opted
instead to focus on anemias resulting from bone marrow fail-
ure (BMF) syndromes, and on anemias caused by heritable
defects in the glycolytic pathway and in the pentose phos-
phate pathways. In these areas, discoveries spurred in good
part by new technologies such as high-throughput sequenc-
ing have led to new insights into pathophysiology and also
to new therapeutic developments that we aim to highlight
here.
Conditions associated with bone
marrow failure
The disease entities falling under this heading are quite
diverse, but they have in common a crucial pathogenetic
mechanism: the loss of hematopoietic stem cells and/or pro-
genitor cells (HSPCs). The pace of stem cell depletion varies
widely, from weeks (e.g. in idiopathic aplastic anemia, IAA)
to years (e.g. in Fanconi anemia, FA). These two condi-
tions exemplify well two broad categories of BMF syndromes:
those that are acquired and those that are inherited (see
Table 11.1).
Molecular Hematology, 4th Edition. Edited by Drew Provan and John G. Gribben.
© 2020 John Wiley & Sons Ltd. Published 2020 by John Wiley & Sons Ltd.
Concluding remarks, 151
Further reading, 151
Acquired BMF syndromes
Idiopathic aplastic anemia
IAA accounts for the majority (about 80–90%) of cases of
acquired BMF syndromes, with an incidence estimated at
2 per million per year (two- to threefold higher in the Far
East and possibly in Africa).
Immunopathogenesis The most direct evidence that IAA may
be an autoimmune disorder came from the clinical observa-
tion that patients with IAA have complete or partial rever-
sion of their pancytopenia when they are treated with anti-
lymphocyte globulin (ALG). Subsequently, it was shown that
patients with IAA often have increased numbers of “acti-
vated” CD8+ CD25+ T cells in their blood and bone marrow.
In addition, T cells from IAA patients can inhibit the growth
of autologous in vitro hematopoietic colonies, and the growth
of colonies from human leukocyte antigen (HLA)-identical
siblings. Based on these observations, a current model of the
pathogenesis of IAA posits that auto-reactive T cells attack
hematopoietic stem cells (HSCs), causing their depletion –
hence the reduction of HSCs in severe IAA to about 1% of
normal.
The primary event that triggers this aberrant immune
response remains elusive; a possible viral cause has long been
sought, but never proven. The identity of the putative auto-
antigen on HSCs also remains unknown. There is evidence,
however, that the inhibitory effect on HSCs of auto-reactive T
cells is mediated, at least in part, through interferon-gamma
(IFN-γ). In addition, IFN-γ upregulates the Fas receptor
on the surface of HSPCs, thus facilitating activation of the
Fas-dependent apoptotic pathways.
As in other autoimmune diseases, there is overrepresen-
tation of certain HLA alleles in IAA patients compared with
population controls: this is true for the HLA-DR2 in patients
131
132 Molecular Hematology

Table 11.1 Classification of the bone marrow failure (BMF) syndromes

Mode of Chromosomal
Disease inheritance locus Gene Clinical manifestations

Inherited
Fanconi anemia AR/X linked See Table 11.2 See Table 11.2 See text
Dyskeratosis congenita X-linked recessive Xq28 DKC1 See text
AD 3q26 TERC
AD or AR 5p15 TERT
AR 17p13 CTC1
AD 14q12 TINF2
AR 15q14 NOP10
AR 17p13 WRAP53/TCAB1
AD or AR 16q2 ACD/TPP1
AD or AR 20q13 RTEL1
AR 16p13 PARN
AR 5q35 NHP2
Diamond–Blackfan anemia AD See text See text See text
X-linked recessive Xp11 TSR2
X-linked recessive Xp11 GATA-1
Shwachman–Diamond syndrome AR 7q11 SBDS Neutropenia, exocrine pancreatic
insufficiency, metaphyseal dysostosis
Amegakaryocytic thrombocytopenia AR 1p34 c-mpl (thrombopoietin Absent megakaryocytes in bone marrow,
receptor) late BMF
Thrombocytopenia with absent radii AR 1q21 RBM8A Bilateral radial aplasia, lower limb
syndrome anomalies, cow’s milk intolerance, renal
anomalies, and cardiac anomalies
Congenital thrombocytopenia and AD 7p15–p14 HOXA11 Aplastic anemia, proximal radius–ulna
radius–ulna synostosis De novo 3q26 MECOM synostosis, clinodactyly, syndactyly, hip
dysplasia, and sensorineural hearing loss
Cartilage–hair hypoplasia AR 9p13 RMRP Short tubular bones, sparse hair, severe
immunodeficiency, macrocytic anemia
Pearson marrow–pancreas syndrome Mitochondrial Usually from nt Contiguous genes Pancreatic exocrine dysfunction,
8.469 to nt deleted sideroblastic anemia (see also text)
13447
Acquired
Idiopathic
Transient erythroblastopenia of childhood (TEC)
Radiation
Drugs and chemicals
Regular: cytotoxic, benzene
Idiosyncratic: chloramphenicol, NSAIDs, anti-epileptics, gold
Viruses
Epstein–Barr virus
Hepatitis
Parvovirus
Human immunodeficiency virus
Cytomegalovirus
Human T-lymphotropic virus-1
Human herpes virus-6
Immune diseases
Thymoma (5–10% develop pure red cell aplasia)
Myasthenia gravis
Systemic lupus erythematous
Pregnancy
Paroxysmal nocturnal hemoglobinuria

AD, autosomal dominant; AR, autosomal recessive; NSAID, BMF, bone marrow failure; non-steroidal anti-inflammatory drug; nt, nucleotide.
 The molecular basis of bone marrow failure syndromes and red cell enzymopathies 133
with IAA of European ancestry, and for others in Japanese
patients with IAA.
Clinical findings have signaled an overlap in pathogene-
sis between IAA and myelodysplastic syndrome (MDS), in
particular with the hypoplastic form of MDS; indeed, the
differential diagnosis between the two is often difficult. It is
now recognized that 20–30% of patients with MDS, especially
those conventionally classified as having refractory anemia
(RA) with hypocellular marrow, respond to immunosuppres-
sive therapy (IST) with alleviation of their cytopenias, sug-
gesting that an immune process, similar to that operating in
IAA, is also involved in the pathogenesis of a subset of MDS
patients. In addition, in 20% of patients with RA, small pop-
ulations of paroxysmal nocturnal hemoglobinuria (PNH)-
like blood cells are detected, as they are commonly seen in
IAA (see later). What is more, their presence is predictive of
response to IST, providing further evidence of an immune
process in the pathogenesis of MDS in this select group of
patients.
The role of mutations in genes involved in telomere main-
tenance in the pathogenesis of IAA is discussed later (see
Dyskeratosis congenita).
Clinical aspects and treatment The clinical picture of IAA
generally reflects the extent of HSPC loss and the subsequent
cytopenias. Typically, a patient with severe IAA presents with
bruising and mucosal bleeding, anemia and septic episodes
(bacterial or fungal), but without hepatosplenomegaly. The
differential diagnosis of IAA, as well as that of hypoplastic
MDS, includes inherited BMF syndromes, the aplastic form
of childhood acute lymphoblastic leukemia, and infectious
and malignant processes that may infiltrate the bone marrow.
Thus the diagnosis of IAA is made eventually by exclusion of
bone marrow infiltration/fibrosis and of major granulocytic
or megakaryocytic dysplasia.
The treatment of IAA involves supportive care for all
patients. Further management is dictated by its severity (as
determined by the degree of pancytopenia, reticulocytope-
nia, and bone marrow cellularity) and by the patient’s age.
Hematopoietic stem cell transplantation (HSCT) from an
HLA-identical sibling or from an alternative donor is the
treatment of choice for younger patients <40 years with
severe IAA, and offers better than 75% 6-year survival. In
the absence of an appropriate donor, or when the patient is
older or the disease milder, immunosuppressive treatment,
in particular the combination of ALG or antithymocyte
globulin (ATG) with cyclosporin, results in a complete or
partial response in the majority of cases. More recently, the
thrombopoietin receptor agonist (TpoR) eltrombopag was
approved for the treatment of patients with severe aplastic
anemia refractory to IST. Eltrombopag is thought to
stimulate stem cell self-renewal and “awaken” HSCs from
their dormant state by non-competitive binding to the TpoR
on the few residual HSCs in IAA. In one study, eltrombopag
improved one or more cytopenias in 40% of patients after
three to four months of treatment. Furthermore, hematolog-
ical responses were sustained after drug discontinuation in
five patients who had been on eltrombopag for months to
years.
Clonal hematopoiesis in IAA A significant proportion of
patients with IAA harbor PNH clones and approximately
20% of patients treated with IST have clinical manifestations
of PNH. IAA, especially after successful IST, also bears a sig-
nificant risk of late clonal disorders: the risk of MDS, AML,
and tumors of other organs is 18.8%, compared with 3.1%
after bone marrow transplantation. Cytogenetic abnormal-
ities observed in IAA include +8, −7 (usually indicative of
MDS if present at diagnosis), del(13q), and del(5q). Another
common genetic lesion, identified by single-nucleotide poly-
morphism (SNP) array-based karyotyping, is acquired copy
number and loss of heterozygosity of the short arm of chro-
mosome 6 (6pLOH), which leads to the loss of the respec-
tive HLA-A haplotype. Given that in IAA CTLs may target
HSPCs that present autoantigens expressed by class I HLA
molecules, 6pLOH might result in immune escape and clonal
dominance of a subset of HSPCs. SM have also been reported
in PIGA (characteristic of PNH) and STAT3 (characteristic of
40% of patients with T-LGL).
In the last few years, high-throughput sequencing has
expanded the repertoire of genetic events underlying clonal
hematopoiesis in IAA and identified some of their clinical
sequelae. These studies demonstrate that SMs in IAA occur
in approximately 50% of patients and are often the same
as observed in myeloid malignancies. For example, in one
cohort, 20% of patients had mutations in ASXL1, DNMT3,
and BCOR. In those with IAA for more than six months, the
presence of these mutations was associated with a 40% risk of
progression to MDS. While ASXL1 and DNMT3A mutations
are predictive of poor response to immunosuppression and
poor survival, PIGA and BCOR/BCOR1 mutations correlate
positively with disease response and survival. Importantly,
out of 43 patients successfully treated with eltrombopag, 8
developed clonal cytogenetic abnormalities, most frequently
monosomy 7, which is known to predate transformation to
MDS and AML.
It will be pertinent to address when, in the natural history
of the disease, these clones arise and how they are selected
for. We can hypothesize that, analogous to PNH and LOH6p
clones, clones with SMs are propagated by immune destruc-
tion, both inherent to the disease and resulting from IST
(Figure 11.1). The interpretation of SMs in IAA must also
take into account the recent observation of CHIP. In this
context, the presence of SMs is associated with an increased
risk of hematological malignancy, in particular in cases with
a high mean allele burden. Further clarification of these
134 Molecular Hematology

CD8+CD25+
CD4+THI CTLS TREGS

Recovery of normal
Immune-mediated BM hematopoiesis
destruction via IFNy
and TNFα
HSC pool
Intrinsic
to AA

OR
IST
Emergence of clones by
Aging Eltrambopag
immune escape and/or
proliferative advantage PNH
- occurs in patients with
PIGA mutant HSC and
GPI-specific T cells
CHIP whereby PIGA mutant
DNMT3A clone repopulates bone
TET2 Clone size marrow
ASXL1 Stable clone
TP53 size MDS/AML
Trisomy 8 Monosomy 7
Deletion 13q ASXL
LOH6P DNMT3A
BCOR/BCOR1
Fig. 11.1 Somatic mutations (SM) in idiopathic aplastic anemia (IAA). IAA results from immune destruction of hematopoietic stem cells
(HSCs) by CD8+ cytotoxic T lymphocytes (CTLs). Increased CD4+ T-helper-1 (TH1) cells offer assistance to CTLs and insufficient T-regulatory (Tregs)
cells are postulated to maintain autoreactive clones. Treatment with IST or eltrombopag enables some stem cells to survive/recover. In this scenario,
any clone with a somatic mutation (SM) that confers resistance to CTLs and/or has a proliferative advantage will either remain stable in size or will
expand. This entails a risk of transformation to myelodysplastic syndromes (MDS) and/or acute myeloid leukemia (AML). SMs also take place in the
context of normal aging and, in view of the presence of the resulting clones, the term “clonal hemopoiesis of indeterminate potential” (CHIP) has
been coined. These SMs are usually distinct from those that arise in IAA, but may also predispose to MDS/AML.

distinct physio- and pathological entities should allow for
more tailored treatment, for instance earlier consideration
of HSCT in patients with IAA and unfavorable SMs. At the
same time, it is important to note that there is no evidence
that IAA originates from abnormal clones. The current
phrase “clonal hematopoiesis in aplastic anemia” means
exactly what it says: namely, that in the bone marrow envi-
ronment of IAA a variety of clones – which would otherwise
remain unnoticed – may instead become detectable or
even prosper, whether because they have a relative selective
advantage or through sheer opportunism as discussed in
detail elsewhere (Luzzatto & Risistano, 2018).
Paroxysmal nocturnal hemoglobinuria
PNH is a rare acquired hematological disorder with three
main clinical features: intravascular hemolysis, tendency to
thrombosis, and BMF of variable severity. As in IAA, the
precise cause of BMF remains unclear; in contrast, the molec-
ular mechanism of hemolysis is well explained. For this rea-
son, the space devoted here to this condition is out of propor-
tion to its prevalence.
Molecular pathogenesis The initiating event in the pathogen-
esis of PNH consists of SM(s) in the X-linked gene PIGA in a
multipotent HSC (sometimes more than one). PIGA encodes
the enzymatically active subunit of an N-acetylglucosamine
transferase. This enzyme catalyzes an early step in the for-
mation of a complex glycolipid molecule called glycosylphos-
phatidylinositol (GPI; Figure 11.2). The synthesis of GPI
takes place initially on the cytoplasmic surface of the endo-
plasmic reticulum, and is then completed on its luminal sur-
face. Once formed, the GPI molecules (anchors) are cova-
lently linked through a trans-peptidation reaction to the
carboxy-terminus of a variety of proteins. The GPI-linked
proteins, after post-translational modifications in the Golgi
apparatus, emerge eventually on the cell surface, to which
they remain attached through the GPI anchor. As a result of
the impaired synthesis of GPI, blood cells are either com-
pletely (PNH III) or partially (PNH II) deficient in GPI-
linked proteins. X-linkage of PIGA is highly relevant to the
pathogenesis of PNH, because there is only one allele in
males, and only one active allele in female somatic cells, and
therefore just one inactivating mutation is sufficient to cause
the consequences just outlined.
 The molecular basis of bone marrow failure syndromes and red cell enzymopathies 135
PIGT*
GPI
EtNP protein maturation in
Golgi
PIGO
PGAP2
M PGAP3
EtNP M
PIGN PIGV
M
EtNP
PIGM
GlcN PIGL
PIGA
Inositol
P
Cell membrane
Fig. 11.2 The glycosylphosphatidylinositol (GPI) anchor and its
genetic disorders. The structure of GPI–protein (protein–EtNP-
6Manα1-2Manα1-6Manα1-4GlcNα1-6myoinositol-phospholipid) is
shown. GPI biosynthesis takes place mostly in the endoplasmic
reticulum. The first step involves the addition of acetylglucosamine
(GlcN) to phosphatidylinositol (inositol-P). Synthesis of the GPI anchor
proceeds with the serial addition of a glycan moiety consisting of three
mannose (M) molecules each modified by phosphoethanolamine
(EtNP). Through a transpeptidation reaction, proteins with the
appropriate carboxy-terminal amino acid motif are attached covalently
to the terminal EtNP. The GPI–protein complex subsequently travels to
the cell surface, where the protein becomes anchored to the lipid
bilayer through GPI. In PNH, PIG-A, a protein encoded by the X-linked
gene PIGA and a member of a multi-subunit enzymatic complex that
catalyzes the first step (i.e. addition of GlcN to inositol-P), is
somatically mutated in one or few HSCs. As a result, very little GPI is
synthesized, or none at all, with consequent severe deficiency of
GPI-anchored proteins on the surface of the mutated HSCs and their
progeny. In Inherited GPI deficiency, germline hypomorphic mutations
in PIGL, PIGM, PIGN, PIGV, PIGO, PGAP2, or PGAP3 result in variable
GPI deficiency. Germline mutations in PIGA are embryonically lethal in
males, with the exception of two families. Here, monoallelic PIGA
mutations resulted in a truncated PIGA protein with sufficient residual
function to allow survival of the affected individuals, but causing
severe congenital abnormalities and intellectual disability. For PIGT, a
case of a germline mutation in one allele with a tissue-specific somatic
mutation expressed in blood cells has been reported. EtNP,
phosphoethanolamine; GlcN, glucosamine; M, mannose; P,
Phosphatidyl.
Complement activation and pathophysiology of red cell
destruction For the majority of GPI-linked proteins the func-
tional consequences of their cell surface deficiency are not
known; however, for CD55 and especially CD59, we do know
they are directly implicated in the destruction of red cells in
PNH. CD59 is a critical negative regulator of complement
action at the surface of erythrocytes, as it interferes with
the formation of the membrane attack complex (MAC). The
MAC is formed first by the assembly of the terminal comple-
ment pathway components C6, C7, C8; then, the large multi-
protein structure is completed through homopolymerization
of C9, which produces physical pores in the cell membrane;
it is this final step that CD59 hinders. Activation of the ter-
minal pathway depends in turn on cleavage activation of C5
by upstream components of the alternative or classical prox-
imal pathways of complement activation. When, as a result
of a PIGA mutation, CD59 is low or absent, red cells become
susceptible to baseline or stress-induced complement activa-
tion, resulting in red cell lysis (intravascular hemolysis). This
phenomenon is reproduced in vitro in the Ham test, whereby
the patient’s red cells are lysed by either autologous or ABO-
compatible donor acidified serum (pH about 6.5), as comple-
ment binds to red blood cell (RBC) at a weakly acidic pH.
As for thrombosis, this may result in part from
complement-mediated activation of platelets deficient
in CD55 and CD59. Other, as yet unidentified, genetic or
acquired factors affecting coagulation and/or fibrinolysis
may have an additive or synergistic effect in producing
thrombosis, which may be devastating, especially as it tends
to affect abdominal and cerebral veins. Given the central
role of C5 in complement activation, we could expect that
inhibiting its cleavage would prevent formation of the MAC
and thus destruction of red cells deficient in surface CD59.
Indeed, eculizumab, a humanized anti-C5 monoclonal
antibody, is a very potent inhibitor of C5 cleavage with
spectacular clinical effects (see later). By abrogating intravas-
cular hemolysis, eculizumab also reduces the free plasma
hemoglobin that binds and depletes plasma nitric oxide,
a natural vasodilator; thus, it also reduces the incidence
of smooth muscle dystonias not uncommon in PNH (i.e.
abdominal pain, retrosternal pain, and erectile dysfunction).
Cellular pathogenesis Since PIGA-mutant PNH HSCs
lack GPI-linked proteins, one might have thought that
they would compete poorly with normal HSCs. Instead,
PNH HSCs can expand until they largely supplant normal
hematopoiesis. In the first instance, this paradox may be
explained by invoking an intrinsic proliferative advantage of
PNH HSCs over normal HSCs, as is the case with leukemic
cells; however, the fact that patients with PNH can live for
decades with normal and PNH hematopoiesis coexisting
in their bone marrow goes against this notion. Lack of
a competitive growth advantage by PNH clones has also
136 Molecular Hematology
been demonstrated experimentally in pig-a null mouse
models. It remains possible that secondary mutations in the
PIG-A mutant clone may confer a proliferative advantage.
In fact, in two patients molecular studies revealed that, as
a result of the same (12;12) chromosomal translocation,
the transcription factor HMGA2 had a truncation of its 3′
untranslated region. This may remove negative regulatory
elements, resulting in increased expression of HMGA2, but
how increased HMGA2 could impart the cell with apparently
non-malignant growth advantage is not known. Recently,
mutations in several genes other than PIGA, essentially the
same as in IAA, have been found in the blood cells of patients
with PNH. However, the same study has shown that clonal
expansion depended on the PIGA mutation itself.
The most widely accepted pathogenetic model, the escape
model, was suggested by the long-known association between
PNH and IAA. In this model, the focus is on the notion that
in PNH HSC-specific T cells would selectively target nor-
mal HSCs, but not PNH HSCs. Under these circumstances,
PNH HSCs would expand and sustain hematopoiesis, in
some cases for years, to the tune of 90% or more. Several
lines of evidence support the escape model. First, small PNH
clones are present in as many as 50% of patients with bona
fide IAA; and minute PNH clones (∼1 in 105 granulocytes)
are seen even in normal individuals. Second, expanded T
cell clones have been demonstrated in the blood of PNH
patients at a frequency three times higher than in appro-
priate controls. Further characterization of T cells in PNH
has revealed (i) increased frequency of CD8+ CD57+ T cells
expressing predominantly activating killer immunoglobulin
and NKG2/CD94 receptors; (ii) the presence of exactly the
same, or very similar, T-cell receptor (TCR)-β CDR3 clono-
typic sequences in the CD8+ CD57+ T cell subsets from differ-
ent patients; and (iii) there is increased representation of the
HLA-DR2 allele in PNH patients compared with population
controls (as in IAA).
The molecular target of the postulated autoreactive T cells
is not yet known. Potential targets are a surface GPI-linked
protein or immunogenic peptides derived therefrom and
presented by the HLA complex; or the GPI molecule itself,
presented by CD1d. Experimental evidence has ruled out the
first mechanism, while the second mechanism is supported
by the identification of GPI-specific, CD1d-restricted T cells
in the peripheral blood of patients with PNH, and recently
also with IAA.
Molecular pathology All types of mutations have been
observed in the PIGA gene in patients with PNH. The
majority (∼75%) are small insertions or deletions caus-
ing frameshifts, and the rest are nonsense and missense
point mutations. The nonsense and frameshift mutations are
spread throughout the coding sequence (exons 2–6), presum-
ably because they cause complete inactivation of the gene
product wherever they fall, whereas interestingly, missense
mutations are clustered mainly within exon 2, where it is pre-
sumed that amino acid residues critical for catalytic activity
are located.
Clinical aspects and treatment Although in the prototypi-
cal PNH patient hemoglobinuria is picturesquely qualified
as paroxysmal, intravascular hemolysis is in fact continuous.
Against this background of chronic hemolysis, the parox-
ysms are acute exacerbations experienced by PNH patients
during intercurrent illnesses such as infections (presumably
because this is associated with activation of complement),
other stressful events, or for no obvious reason.
Hemolytic anemia with macrocytosis (partly due to reticu-
locytosis and partly due to BMF), different degrees of throm-
bocytopenia and leukopenia, iron deficiency, and hemosider-
inuria should raise the suspicion of PNH. The occurrence of
venous thrombosis at an unusual site, such as the brain or the
abdomen, ought to be an even more compelling pointer to
the possible diagnosis of PNH. Venous thrombosis of small
or large vessels is a serious and potentially life-threatening
complication of PNH, occurring in 40–50% of patients.
Historically, the diagnosis was confirmed by the Ham
test (which only detects the PNH abnormality in erythro-
cytes), but this assay has largely been supplanted by flow
cytometric analysis. The latter is the most sensitive diagnos-
tic tool for PNH (capable of detecting <1 in 10 000 PNH
cells): it will demonstrate that there is a population of ery-
throcytes and leukocytes with absent or markedly reduced
expression of GPI-linked proteins such as CD55 and CD59.
Detection of GPI-deficient monocyte and granulocyte clones
is also indicated by the absence of fluorescent aerolysin
(FLAER), which binds directly to the cell surface GPI anchor
itself.
Since its introduction in 2006, eculizumab has been the
main treatment for PNH. This humanized mAb inhibits
cleavage of the complement component C5, resulting in (i)
almost complete abrogation of hemolysis in the majority of
but not all patients; (ii) reduction in transfusion require-
ments, resulting in hemoglobin stabilization or transfusion
independence in up to two-thirds of patients; (iii) significant
improvement in quality of life for most patients, in particular
due to the mitigation of fatigue; and (iv) a reduction in the
risk of thrombosis. Importantly, patients at risk of thrombo-
sis (i.e. those with large PNH clones or an additional throm-
bophilia) should still be considered for prophylactic long-
term anticoagulation. Patients on eculizumab should also be
vaccinated against encapsulated bacteria, especially Neisseria
meningitides, to which they are susceptible due to comple-
ment blockade.
Unfortunately, worldwide use of eculizumab is precluded
by its high cost. Furthermore, the clinical response to
eculizumab is not uniform in all patients. There are several
 The molecular basis of bone marrow failure syndromes and red cell enzymopathies 137
reasons for this variability. First of all, in some cases the
anemia is due not only to hemolysis, but also to BMF, with
eculizumab acting on the former, but not on the latter.
Second, in all patients on eculizumab there is extravascular
hemolysis, which occurs because, in the absence of the drug,
C3 bound to PNH RBC membranes leads to C5 activation
and complement-mediated cell lysis. On eculizumab C5 is
blocked, and therefore cell lysis does not occur; however,
RBCs become opsonized by C3 molecules and are in turn
phagocytosed in the reticuloendothelial system. As a result
of this opsonization process, the Coombs test becomes
positive (typically it is negative in untreated PNH).
Of note is that the extent of C3 binding to PNH RBCs
has been shown to correlate with homozygosity for the rare
L allele at the complement 1 receptor locus. Thus, the L/L
(as opposed to L/H or hereditary hemochromatosis, HH)
genotype is associated with a lesser response to treatment
with eculizumab in patients with PNH. In a small number of
Japanese and Chinese patients, complete failure to respond
to eculizumab is caused by an SNP in C5 (c.2654G → A,
p.Arg885His), which abrogates binding of eculizumab to its
target peptide in C5; the last two phenomena highlight a
classical paradigm of pharmacogenomics. Thus, there is still
(extravascular) hemolysis on eculizumab and, as a result,
some patients still require blood transfusion; because they
no longer lose iron through hemoglobinuria, they are now
at risk of iron overload and may require iron chelation. How-
ever, it is important to note that these patients still benefit
from the abrogation of intravascular hemolysis and the con-
sequent symptoms already mentioned.
Long-term therapeutic options for PNH include immuno-
suppressive agents (e.g. a combination of ALG/ATG and
cyclosporin, especially for patients with moderate to severe
cytopenias). In the era of eculizumab, allogeneic HSCT may
still be offered to patients as the only definitive treatment, but
in practice it will be carried out only in select patients (such
as those developing MDS; see later).
PNH and other clonal disorders
Patients with PNH have a small risk (<4%) of developing
MDS (the reverse, i.e. patients with MDS developing PNH, is
discussed earlier) and AML. Because a similar risk exists in
IAA and in the inherited BMFs, it is probably the perturbed
inflammatory marrow environment in conjunction with the
SMs described earlier, rather than the PIGA mutations per se,
that allows the emergence of premalignant (MDS) or malig-
nant (AML) clones, predisposing to MDS and AML.
Inherited GPI deficiency
Inherited glycosylphosphatidylinositol deficiency (IGD)
encompasses a group of diseases caused by heritable
autosomal recessive mutations in genes of the GPI biosyn-
thetic pathway. In the first example of IGD, in two unrelated
consanguineous families, affected children had spontaneous
splanchnic vein thrombosis in the first year of life and
absence seizures that became refractory to treatment later
on. In contrast to PNH, there is no significant intravascular
hemolysis nor BMF. In all three affected children, the same
promoter mutation was identified: a C>G substitution at
position −270 from the start codon of the PIGM gene, which
encodes a mannosyltransferase responsible for the first
and essential mannosylation reaction in the biosynthesis
of GPI.
The hypomorphic nature of the C>G mutation allows
for some transcriptional output, the extent of which varies
from tissue to tissue, hence the variable degree of GPI
deficiency. For example, in red cells, GPI expression is
almost normal, explaining the lack of significant hemolysis;
by contrast, granulocytes and B cells display almost com-
plete, and fibroblasts partial, GPI deficiency. The precise
molecular mechanism of this disease has recently been
delineated: C>G substitution disrupts binding of the generic
transcription factor Sp1 to a glucocorticoids (GC)-rich box
in the core promoter of PIGM. Unlike in red cells, where
PIGM transcription is independent of Sp1 binding to its core
promoter, in B cells PIGM transcription is dependent upon
Sp1 binding, hence the divergent GPI expression across cell
lineages. Impaired Sp1 binding leads to reduced histone
acetylation in the promoter of PIGM and in turn transcrip-
tional repression by nucleosomal compaction, recruitment
of the Polycomb repressor complex, and establishment of
a bivalent chromatin state. As such, PIGM-associated IGD
represents the first Mendelian disease whose pathogenesis
is attributed to Polycomb-mediated bivalent chromatin
domain transcriptional repression. Importantly, histone
hypoacetylation is reversible. Specifically, in the presence
of histone deacetylase inhibitors (HDACis) such as sodium
butyrate, histone acetylation, PIGM transcriptional activity,
GPI biosynthesis, and surface GPI-linked protein expression
can all be restored in vivo as well as in vitro. This remarkable
effect was highlighted by the complete abrogation of compli-
cated, life-long, and treatment-refractory epilepsy in a child
with IGD treated with sodium butyrate.
In a small number of patients with variable clinical
phenotypes, whole-exome sequencing analysis has identi-
fied several other germline mutations in genes of the GPI
biosynthetic pathway (Figure 11.2). As with PIGM, these
patients have homozygous or biallelic mutations that result
in partial deficiency of the corresponding protein. Clinical
manifestations comprise epilepsy, learning difficulties,
and multiple organ abnormalities, but not BMF, with the
exception of one reported case of PNH caused by a germline
mutation in PIGT accompanied by an SM in neutrophils
(Figure 11.2).
138 Molecular Hematology

Inherited BMF syndromes commonly of the radius and thumb), skin lesions (hyperpig-
mentation, café-au-lait spots), renal and urinary tract mal-
Fanconi anemia
formations, and gonadal dysfunction. However, the clinical
The incidence of FA is estimated at 1 : 360 000 in the gen- spectrum is even wider, as it includes congenital defects of the
eral population. Early studies indicated that FA was a geneti- gastrointestinal system, heart, and central nervous system.
cally heterogeneous disease, a notion confirmed by the iden- Patients with FA have an unusually high risk of develop-
tification, through the use of somatic cell hybridization and ing treatment-resistant MDS and AML, estimated at 52% by
latterly genome sequencing, of 21 complementation groups. the age of 40 years. Furthermore, the risk of a variety of solid
All 21 FA genes have now been cloned and characterized tumors, especially squamous cell carcinomas of the skin,
(Table 11.2). The overall frequency of heterozygotes for FA head, neck, and anogenital region, is several times higher
mutant genes in the general population is estimated at 1 in than in the general population. Patients with mutations in
300. In Ashkenazi Jews and in the Afrikaans population of BRCA2 or PALB2 usually develop AML or embryonal tumors
South Africa it is much higher (1 in 100 and 1 in 89, respec- in the first few years of life. While FA results from bial-
tively), most likely as a result of founder effects. lelic mutations, heterozygous mutations in BRCA1, BRCA2,
PALB2, and RAD51C are associated with increased risk of
Clinical aspects The most common clinical manifestation is solid tumors, in particular breast and ovarian cancer. This has
gradual onset of BMF. BMF typically appears by the age of implications for first-degree relatives of an individual with FA
10 years (median age 7 years; range birth to 31 years) and caused by one of these specific complementation groups.
is often heralded by thrombocytopenia, macrocytosis, and
increased hemoglobin F levels (the latter indicative of stress Cellular phenotype and function of FA proteins FA pro-
erythropoiesis). Congenital abnormalities are present in two- teins constitute the “FA pathway,” a DNA damage-response
thirds of patients and comprise skeletal abnormalities (most pathway responsible for repair of inter- and intrastrand
DNA cross-links (ICLs), either occurring spontaneously or
induced by agents like mitomycin C or diepoxybutane (Fig-
ure 11.3). ICLs prevent the separation of the two DNA
Table 11.2 Identity of the Fanconi anemia (FA) genes and proteins strands, leading to replication fork arrest and cell cycle arrest
in the late S phase. Completion of the cell cycle is not possi-
Approximate ble unless ICLs are repaired. FA proteins, organized in several
FA genes Chromosome frequency of FA (%) groups functioning in a linear fashion, are essential for suc-
FANCA 16q24.3 65 cessful resolution of the stalled DNA replication at the site of
FANCB Xp22.2 2 the ILC. The Fanconi anemia core complex (FCC), compris-
FANCC 9q22.3 12 ing nine proteins (FANCA, B, C, E, F, G, L, M, and T), receives
FANCD1/BRCA2 13q12.3 2 signals of DNA damage through phosphorylation by ATR, a
FANCD2 3p25.3 4 kinase sensor of DNA damage. This promotes binding of the
FANCE 6p21.3 1 FCC to the damaged DNA through FANCM and FAAP24.
FANCF 11p15 2 FANCL, through its ubiquitin ligase activity, monoubiquiti-
FANCG 9p13 8 nates FANCD2 and FANCI, the two proteins constituting the
FANCI 15q26.1 1
ID complex. FANCL interacts with UBE2T, a ubiquitin con-
FANCJ/BRIP1 17q23.1 2
jugating enzyme that is required for monoubiquitination of
FANCL 2p16.1 <1
FANCM 14q21.3 <1
FANCD2. Ubiquitination of the ID proteins along with their
FANCN/PALB2 16p12.1 <1 phosphorylation by ATR are required for their specific local-
FANCO/RAD51Ca 17q23 <1 ization to the site of DNA damage, where they interact with
FANCP/SLX4 16p13.3 <1 a further family of proteins.
FANCQ/ERCC4/XPF 16p13.12 <1 The finding that these proteins, corresponding to
FANCR/RAD51a 15q15.1 <1 FANCD1, FANCN, and FANCJ complementation groups,
FANCS/BRCA1a 17q21.31 <1 were in fact identical to BRCA2 and its associated PALB2
FANCT/UBE2Ta 1q32.1 <1 and BRIP1 proteins, respectively, was an exciting discovery
FANCU/XRCC2a 7q36.1 <1 linking the FA pathway to an already established process
FANCV/REV7/MAD2L2 1p36.22 <1
of DNA repair. In concert with a number of other proteins
a Loss
of the DNA repair machinery (e.g. RAD51, RAD51C,
of function mutations in these genes have been associated
XRCC2), BRCA1 and BRCA2 are involved in different
with Fanconi-like congenital abnormalities, but not with bone mar-
row failure.
repair processes that include nucleotide excision repair,
translesional synthesis, and homologous recombination, all
 The molecular basis of bone marrow failure syndromes and red cell enzymopathies 139

F C
B
FC complex
E
M G
Downstream effectors of DNA repair
T L A +Ub
ID complex
DNA Polζ
damage ATR Ub N
Ub P Q S
(ICLs) V J
+P I
D2 Ub Ub
P Ub Ub
P R
I I
D2 D2 D1 U
P P O
P P

Fig. 11.3 Molecular pathogenesis of Fanconi anemia (FA). The FA pathway of DNA repair is activated in response to intra- or interstrand DNA
cross-links. First the DNA sensor ATR phosphorylates the Fanconi core complex (FCC) and ID complexes. Subsequently, the FCC monoubiquitinates
FANCD2 and FANCI. The ID complex is recruited to the site of DNA damage and acts in concert with DNA repair proteins. Specifically, FANCP/SLX4
and FANCQ/ERCC4 catalyze unlooping of the interstrand cross-links (ICL), allowing translesional synthesis by Polζ, one component of which is
FANCV/REV7. The final phase of ICL repair is mediated by the remaining DNA repair proteins, including FANCD1/BRCA2, FANCN/PALB2,
FANCJ/BRIP2, and FANCS.BRCA1 along with FANCR/RAD51 and its paralogs: FANCO/RAD51c and FANCU/XRCC2. Ultimately the coordinated
functions of these proteins leads to the repair of the DNA cross-link through nucleotide excision repair, translesional synthesis, and homologous
recombination.

of which are required for removal and repair of ICLs and
restarting of DNA replication. Nucleotide excision repair
is also performed by two proteins recently found to be
mutated in dyskeratosis congenita (DC): SLX4 and ERCC4.
These form a complex that facilitates nucleolytic incisions
at the ICL by the endonuclease ERCC4, paving the way for
homologous recombination. Taken together, a mutation
affecting any one of the FANC proteins impairs DNA repair,
resulting in genomic instability in HSCs and, in turn, BMF.
Molecular genetics FA is an autosomal recessive disorder,
with the exception of FANCB, which is X-linked, and
FANCO/FANCR, which are autosomal dominant. The most
characteristic cellular feature of FA cells is the formation of
DNA double-strand breaks on exposure to DNA inter- and
intrastrand adducting agents (clastogens), such as mitomycin
C and diepoxybutane. The in vitro response to clastogens
has made it possible to test cells from different patients for
their ability to cross-correct each other’s defect by somatic
cell fusion (hence the 21 complementation groups).
Mutations of FANCA account for about 60% of FA cases
and are spread throughout the gene. None of the mutant alle-
les is common and few have been encountered more than
once. Thus, identifying mutations in newly diagnosed cases
of FA where the complementation group is unknown involves
comprehensive sequencing of multiple FANC genes in par-
allel, an approach that has been facilitated by the recent
access to rapid and economical high-throughput sequencing,
which allows the simultaneous, targeted sequencing of tens of
genes.
Mutations in the FANCC gene account for about 10–15%
of FA cases. The IVS4 + A → T and del322G mutations com-
prise more than 75% of FANCC mutations. The IVS4 + A → T
allele is found in Ashkenazi Jews at a polymorphic frequency
(1 in 80), and it is responsible for 85% of the FA cases in this
population. Patients with IVS4 or exon 14 mutations tend
to have earlier onset of hematological complications (BMF
and MDS/AML), and to have a shorter survival compared
with patients with exon 1 mutations or patients with FANCA
or FANCG-related FA. Mutations in the FANCG gene are
found in about 10% of FA cases. The stop codon mutation
E105X accounts for 44% of mutations in German FANCG
patients. It is interesting that although biallelic null muta-
tions are very frequent in FANC genes of FCC, only hypo-
morphic mutations have been identified in FANCD2 (3–6%
of all cases of FA), highlighting a more important role of the
ID complex that is not restricted to ICL correction only, but
also encompasses the repair of ionizing irradiation-induced
double-strand DNA damage. Consistent with this, FANCD2
patients, despite the hypomorphic nature of the mutations,
appear to develop the most severe form of FA, with multiple
developmental abnormalities and early development of BMF
and malignancy. The prevalence of mutations in other FANC
genes is very low (<2%).
FA and cancer An important role of the FA pathway in
the biology of sporadic cancers is increasingly being recog-
nized. Epigenetic silencing of FANC genes has been iden-
tified in various solid tumors and in acute leukemias; in
some cases this correlates with their sensitivity to chemother-
apeutic agents, causing DNA cross-links such as cisplatin,
cyclophosphamide, and melphalan. In a small proportion of
patients with AML, SMs in FANCA have been identified, but
their functional significance is unknown. Therefore, studying
the functional status of the FA pathway in different tumors
may offer valuable leads to therapeutic choices.
140 Molecular Hematology
Diagnosis The clastogen test (outlined earlier) remains the
gold standard clinical diagnostic test; however, screening for
pathogenic mutations in the FANC genes has now entered
routine clinical practice. As a deficiency in any of the eight
proteins of the FCC impairs its ubiquitin ligase activity,
immunoblotting for ubiquitinated FANCD2 can also be used
as a screening test to focus the search within the FCC genes or
in those acting downstream of the FCC. With this approach,
genetic counselling, pre-natal, and pre-implantation diagno-
sis are now feasible for most families with affected children.
In areas with a large Ashkenazi Jewish population (e.g. New
York City), it is realistic to carry out screening for polymor-
phic FANC alleles on a wider population basis, so that it can
be offered to all couples at risk.
Treatment The conventional management of FA focuses on
the consequences of BMF and includes hematopoietic growth
factors, blood product support, and androgens. About half of
patients respond to androgens initially, but often suffer from
significant side effects, including androgen-induced hepatic
adenomas. Eventually all patients become refractory.
HSCT from an HLA-identical unaffected sibling or from
alternative sources is currently the only therapeutic approach
that can successfully achieve long-term correction of BMF
and possible prevention of MDS and AML. The best results,
with reduced short- and long-term treatment-related mortal-
ity, are obtained with fludarabine-based, reduced-intensity,
non-myeloablative conditioning HSCT regimens that do not
include irradiation (to reduce the risk of solid tumors). T
cell depletion of the graft is employed to minimize graft-
versus-host disease. Unfortunately, even subsequently HSCT
patients remain at increased risk of developing solid tumors,
and thus require close long-term follow-up as well as pre-
ventative strategies such as human papilloma virus (HPV)
vaccination.
The cloning of FA genes has paved the way for gene ther-
apy for patients with FA. There is clinical and experimental
evidence that HSCs with corrected phenotype have a survival
and growth advantage over uncorrected cells and can support
long-term hematopoiesis. Successful transfer of FA genes to
a small number of autologous HSCs should therefore be ade-
quate to ameliorate the severity of BMF.
Dyskeratosis congenita
DC is another rare (incidence approximately 1 in 1 million),
genetically heterogeneous, inherited disorder with BMF as a
major feature. DC is primarily a disorder of telomeres – it
could be said to belong to the telomeropathies. Telomeres,
the ends of chromosomes, are made of several megabases of
DNA consisting of identical nucleotide repeats (TTAGGG),
and of a protein complex essential for chromosome integrity
(Figure 11.4).
The clinical hallmarks of DC are lacy reticulated skin, nail
dystrophy, and mucosal leukoplakia in the first years of life;
and later the development of BMF. Other clinical manifes-
tations include developmental delay, short stature, ocular,
dental, and skeletal abnormalities, hyperhidrosis, hyperkera-
tinization of the palms and soles, bullae on minimal trauma,
hair loss, nail loss, sometimes gonadal failure, and features
of premature aging. Patients with DC are also at risk of pul-
monary fibrosis and hepatic failure. The clinical spectrum of
DC is variable among patients and dynamic with age: some
patients manifest only minimal physical features and normal
hematopoiesis, while others have the classical triad of skin,
nail, and mucosal abnormalities coupled with early-onset
BMF. Like FA, DC predisposes to malignancy, in particular
MDS/AML and squamous cell carcinomas of the head, neck,
and anogenital tract. The median age of onset is 35 years for
MDS and 37 years for solid cancers.
Molecular genetics and pathogenesis Eleven genes have been
identified to cause DC when mutated (Table 11.1). Mutations
occur de novo, and the diseases they cause are inherited as X-
linked recessive, autosomal dominant, or autosomal recessive
traits.
The X-linked form of DC was the first to be identified
and characterized: it results mostly from missense mutations
in the housekeeping gene DKC1. Mutations in DKC1, which
maps to Xq28 (like G6PD and hemophilia A), account for
about 40% of all DC cases. Hoyerall–Hreidarsson syndrome,
a rare disorder characterized by severe growth failure, severe
immunodeficiency, cerebellar abnormalities, and BMF, is in
most cases allelic to DKC1.
Dyskerin, the protein product of DKC1, is homologous
to the yeast protein Cbf5p, which, in conjunction with the
H/ACA class of small nucleolar RNAs, is required for the
pseudouridylation (i.e. conversion of uridine to pseudouri-
dine) of pre-ribosomal RNA (rRNA), a modification essential
for ribosome biogenesis. Although defective pseudouridyla-
tion and ribosome biogenesis have been observed in a hypo-
morphic DKC1 mouse model, other studies failed to demon-
strate any qualitative or quantitative defect of rRNA in DC
patients. Instead, it was shown that dyskerin binds, via its
domain called PUA, to an H/ACA domain of the telom-
erase RNA component (TERC); that is, the template used by
its protein component, the telomerase reverse transcriptase
(TERT). Thus, dyskerin is intimately linked to telomerase in
its crucial function of telomere maintenance during cell divi-
sion. Most DKC1 mutations map to the PUA domain, linking
DC to abnormal RNA binding.
Telomerase, as just outlined, is a ribonucleoprotein com-
plex, and dyskerin is one of the several proteins which, by
interacting directly or indirectly with TERC, preserve its sta-
bility and its ability to function as a template for TERT.
Telomerase activity is highest in rapidly dividing cells (e.g.
 The molecular basis of bone marrow failure syndromes and red cell enzymopathies 141

Shelterin complex Telomerase complex

protection of telomere ends telomere elongation

Telomere trafficking
PARN TERC processing
WRAP53
0 R
P1 GA
ACD NO
P2
TIN2 NH
POT1 TERT DKC1
G strand T T 3’
R R TTAGGG
F F AAUCCC
C strand 2 1
5’
5’
P1 TERC
RA STNI
RNA template
CTC1
TENI
Telomere replication
RTELI
+ stability
CST complex
C strand synthesis

Telomeric DNA

Telomere

Centromere

Telomere
Fig. 11.4 Schematic representation of human proteins involved in telomere maintenance. TERC encodes a non-coding RNA and no
protein. This RNA template interacts with TERT and other members of the telomerase complex to elongate telomeres by adding hexanucleotide
tandem repeats to the ends of chromosomes. The allied roles of the components of the CST and shelterin complexes are shown. Shaded proteins
are those that, when affected by germline mutations, cause dyskeratosis congenita (DC).

in hematopoietic progenitors, in tumor cells, during organo-
genesis, and in early life). The critical evidence that DC is a
disease of telomere dysfunction was provided by the find-
ing that heterozygous mutations in the TERT and TERC
genes themselves cause a disease virtually indistinguishable
from that caused by DKC1 mutations. Once again, not unlike
FA, the multitude of genes whose mutations can give rise
to a dyskeratosis phenotype relates to the complexity of
the telomere maintenance/servicing machinery (see Table
11.1 and Figure 11.4). NOP10 and NHP2, members of the
H/ACA snoRNPs family, localize like DKC1 itself to nucleoli
and to Cajal bodies in the nucleus. WRAP53 assists with the
transportation of TERT to the Cajal bodies, and PARN facili-
tates the interaction between TERC and the telomere. RTEL1
encodes a DNA helicase required to unwind the T loop struc-
ture, a prerequisite for telomere lengthening, whereas CTC1
forms part of the CST complex that caps and protects telom-
eres. Finally, TINF2 (encoding TIN2) and ACD encode pro-
teins that form part of the so-called shelterin complex, which
binds to telomeric DNA and provides stability. Mutations in
TINF2 have been associated with the Revesz syndrome, a
form of DC characterized by bilateral exudative retinopathy.
Patients with monoallelic mutations are affected, and there-
fore the condition is designated as autosomal dominant; how-
ever, those with biallelic mutations have far more severe dis-
ease. Thus, all types of DC are related to a common molecular
pathogenetic pathway, namely the maintenance of telomere
length. In keeping with this notion, all types of blood cells
in DC patients have significantly shorter telomeres than age-
matched controls. Shortening of telomeres to a critical length
can result in either (i) growth arrest and apoptosis – hence
the development of BMF; or (ii) neoplastic transformation –
hence the increased propensity to epithelial tumors.
The study of the genetics of DC has revealed at least three
additional interesting features. First, in families with TERC,
TERT, and TINF2 mutations, DC has tended to show earlier
onset and more severe manifestations in successive genera-
tions, a phenomenon known as genetic anticipation. In keep-
ing with this, the rate of loss of telomere length was higher
in successive generations within the same DC family (just as
142 Molecular Hematology
had been observed in Terc−/− mice). Second, heterozygous
germline mutations of TERT or TERC have been detected in
a small minority (<5%) of patients who had been initially
diagnosed with IAA. In many cases these mutations were
not obviously deleterious for the function of telomerase in
vitro; in some cases they were not detected in the parents of
index cases and in others they did not segregate with the dis-
ease in the respective families. A possible interpretation of
these findings is that mild TERC and TERT mutations may
predispose to BMF; however, this develops only upon expo-
sure to a trigger, such as an immunological attack. Third,
in patients with pathogenic mutations of TERT (and in one
case of DKC1), somatic mosaicism has been demonstrated,
whereby the mutation is found in DNA from skin fibroblasts
but not in DNA from peripheral blood cells. The presumed
mechanism is genetic reversion through back-mutation in a
single hematopoietic cell. The self-corrected cell would have
a significant selective advantage, in some cases sufficient to
repopulate the bone marrow, an example of nature’s gene
therapy.
Diagnosis DC is characterized by very short telomeres,
defined as <1% of age-matched normal controls, in the
majority of leukocyte subsets. Measurements are performed
by multiparameter flow cytometry fluorescence in situ
hybridization (flow-FISH). In lymphocytes, sensitivity and
specificity for DC are 97% and 91%, respectively. Genetic
testing is desirable to confirm diagnosis, but pathogenic
germline mutations are detected in only about 70% of
patients with a clinical diagnosis of DC, either because the
panel of genes is incomplete or because some causative genes
are still unknown. TERT and TERC mutations correlate with
IAA, pulmonary fibrosis, and adult-onset disease.
Treatment As for FA, there is no definitive treatment for DC,
and allogeneic HSCT is the only curative option for severe
BMF or MDS/leukemia. HSCT in DC has an increased risk
of graft failure, graft-versus-host disease, pulmonary fibro-
sis, hepatic cirrhosis, and veno-occlusive disease. If a related
donor is unavailable, one can try therapy with androgens.
The mechanism of action of these agents is thought to be
direct upregulation of TERT gene expression with conse-
quent increased telomerase activity. Most patients with DC
die before the age of 30 years, predominantly from complica-
tions of BMF, opportunistic infections, pulmonary complica-
tions, and malignancy.
In a recent study, human induced pluripotent stem cells
(iPSCs) were generated from skin fibroblasts of a DC patient
and differentiated to intestinal cells. Correction of the disease
allele was performed with clustered regularly interspaced
short palindromic repeats (CRISPR)-Cas9-mediated homol-
ogous recombination. Examination of the resultant isogenic
DC and control cell lines showed that the Wnt signaling
pathway (linked to the shelterin telomere capping complex)
was essential for intestinal stem cell renewal, and that treat-
ment with Wnt agonists corrected the DC phenotype and
restored telomere length to normal. Thus, the recent remark-
able advances in genome editing approaches may lead to the
future development of gene therapy for DC and possibly for
other BMF syndromes.
Diamond–Blackfan anemia
Diamond–Blackfan anemia (DBA) is a rare (1–2 in 100 000
live births) inherited syndrome with isolated anemia and ery-
throblastopenia as the main features, although with disease
progression patients may develop bi- or trilineage hypopla-
sia. Clinically DBA manifests itself in the first months of
life with macrocytic anemia, reticulocytopenia, increased
hemoglobin F, and elevated erythrocyte adenosine deami-
nase (eADA) levels. It is not known why the anemia asso-
ciated with DBA rarely manifests during fetal life. The bone
marrow is normocellular, with a characteristic paucity of
erythroblasts (<5% of nucleated cells). Growth retardation
is seen in about 30% of patients, and other developmen-
tal abnormalities (including skeletal, cardiac, and urogenital
defects) in 50% of patients. There is also an increased risk
(∼20% by age 46 years) of MDS/AML or solid tumors, mainly
osteogenic sarcoma.
Molecular genetics and pathogenesis In most cases, DBA is
caused by a monoallelic, loss-of-function ribosomal protein
(RP) gene mutation. DBA is therefore one of the “riboso-
mopathies,” a group of inherited BMF syndromes also com-
prising Shwachman–Diamond syndrome, X-linked DC, and
cartilage hair hypoplasia (Table 11.1). In 55–60% of cases
mutations are de novo, whereas in the remainder they are
transmitted in an autosomal dominant fashion.
The first pathogenic genetic lesion identified was a het-
erozygous mutation in RPS19, the most common (25% of
cases) gene involved in DBA. The finding that the whole
RPS19 gene is deleted in some cases of DBA strongly argued
that haploinsufficiency, rather than a dominant negative
effect, is the underlying pathogenetic mechanism in DBA.
As well as being a structural component of the mature small
ribosomal subunit (40S; the large subunit is 60S), RPS19 is
also required in the nucleolus for the correct processing and
cleavage from the precursor rRNA of the 18S fragment, which
is the main rRNA constituent of 40S. In the same process,
which is RPS19 independent, precursor rRNA is also cleaved
to 5.8S and 28S rRNA, which, along with the independently
transcribed 5S rRNA, are the constituents of the large ribo-
somal subunit. Haploinsufficiency of RPS19 and of the other
RP genes linked to DBA results in defective formation of
the ribosomal subunits and eventually defective formation
of the mature ribosome. How impaired ribosome function
 The molecular basis of bone marrow failure syndromes and red cell enzymopathies 143

might lead to the relatively selective phenotype of erythroid underlying the development of DBA, we are yet to elucidate
hypoplasia is not clear. It is possible that this is related to the the molecular basis of the clinical heterogeneity observed in
extraordinary demand that hemoglobin production imposes this disease, including the remarkably variable penetrance
on the translational apparatus. In addition, many RPs, includ- within families.
ing RPS19, have been found to participate in extra-ribosomal
functions. For example, RPS19 might affect cell cycle progres- Diagnosis Diagnostic criteria for DBA include the presence
sion through its interaction with the serine/threonine kinase of classical hematological and clinical features and identifica-
Pim-1; whereas RPL11 and RPL5 stabilize and activate p53 tion of a causative gene mutation. The presence of congeni-
through an inhibitory physical association with the p53 reg- tal anomalies, age <1 year, raised eADA, and macroytosis all
ulator MDM2. Clarification of the mechanisms by which per- point toward a diagnosis of DBA, rather than transient ery-
turbed ribosome biogenesis leads to fewer erythroid progen- throblastopenia of childhood (TEC), the main differential.
itors with impaired function will be facilitated by the recent This is an important distinction, as the latter condition usu-
immunophenotypic characterization of early and late human ally resolves spontaneously within a few months.
erythroid progenitors. In specialized DBA centers, genetic diagnosis using
Since the identification of RPS19, several other RP genes targeted next-generation sequencing of RP genes has
have been implicated in DBA, including RPS24, RPS26, replaced traditional Sanger sequencing. This approach,
RPS10, RPS17, RPL5, RPL35A, and RPL11 (Table 11.3). A complemented by additional techniques such as multiplex
wide variety of mutations have been described, including ligation-dependent probe amplification for validation of
missense, nonsense, splicing, and whole allele deletions. Of computationally identified deletions, has led to an improve-
note is that DBA resulting from an RP gene deletion can ment in the pick-up rate of mutations from 40% to around
occur in the context of other syndromes, for example the 70% of patients, and to the identification of new candidate
microdeletion 3q29 syndrome, in which the deleted region RP genes (Table 11.3).
encompasses RPL35A. Mutations in the promoter region of Interestingly, a recent study discovered that 8 of 173
RPS19 have also been observed, and it is likely that more vari- patients with presumed DBA but no pathogenic RP gene
ants in regulatory areas will be implicated in DBA as knowl- mutation were in fact harboring mitochondrial DNA dele-
edge of the function of the non-coding genome expands. In tions similar to those seen in the Pearson marrow-pancreas
the last few years two X-linked genes associated with DBA syndrome. Therefore, while we have addressed each of the
have been discovered: GATA-1 (an erythroid-specific tran- inherited BMF syndromes separately, it is important to rec-
scription factor) and TSR2 (a RP chaperone). While great ognize that they share a triad of clinical features: anemia,
progress has been made in characterizing the genetic lesions physical abnormalities, and predisposition to cancer. Given
the overlap between these conditions and the accessibility of
genetic testing in routine clinical practice, a child presenting
with congenital anemia should be ideally tested for multi-
Table 11.3 Genes mutated in Diamond–Blackfan anemia (DBA)
ple conditions, using a panel encompassing multiple genes.
Gene name Genetic locus % of cases of DBA Furthermore, cases in which a mutation is not identified by
targeted multigene assays are candidates for whole-genome
RPS19 19q13.2 25 sequencing with the aim of novel gene discovery, as exempli-
RPL5 1p22.1 6.6 fied by the Genomics England 100 000 Genomes Project.
RPS26 12q13.2 6.4
RPL11 1p36.11 4.8
Treatment Blood transfusion, combined with iron chelation,
RPL35A 3q29 3
RPS10 6p21.31 2.6
is the mainstay of treatment for moderate to severe anemia
RPS24 10q22.3 2 in DBA. After regular immunizations with live vaccines
RPS17 15q25.2 1 are completed during infancy, most patients are started on
RPS7 2p25.3 <1 high-dose glucocorticoids (GC). Initially 80% of patients
RPS29 14q21.3 <1 respond, with a surge in reticulocytes followed by a rise
RPL26 17p13.1 <1 in hemoglobin, to the extent that the patient may become
RPL15 3p24.2 <1 transfusion independent. However, not infrequently GC
RPL31 2q11.2 <1 treatment is discontinued, either because the response is lost
RPS27 1q21.3 <1 (this can happen with puberty or pregnancy), or because at
RPS28 19p13.2 <1
the dose required to maintain therapeutic benefit the side
RPL27 17q21.1-21.2 <1
effects are unacceptable. In the long term only 20–40% of
GATA-1 Xp11.23 <1
TSR2 Xp11.22 <1
patients remain on a very low dose of GC; a further 10–20%
seem to go on to spontaneous remission by the age of 25.
144 Molecular Hematology
Patients who are steroid resistant, and those with severe iron
overload or progressive BMF, must be considered for allo-
geneic HSCT. Ideally this should be performed before the age
of 10 years. At present there is convincing evidence to recom-
mend sibling donors, and promising results from unrelated
donors (either haploidentical or umbilical cord blood).
Recent experimental evidence indicated that, through acti-
vation of the rapamycin-sensitive mTOR/RPS6 pathway, the
amino acid L-leucine enhances ribosomal translational activ-
ity in DBA cells with RPS19 mutations. As yet, a clinically
effective treatment for patients with DBA has failed to evolve
from these findings. One of the most clinically important
unsolved riddles in DBA is the mechanism of action of
GC. A recent study using chromatin immunoprecipitation
sequencing in normal murine fetal liver erythroid progeni-
tors showed a physical interaction between PPARalpha and
glucocorticoid receptor (GR), the transcription factor that
mediates GC effects on gene transcription. It seems reason-
able to expect that as a follow-up to this and other studies,
the riddle of the GC effect will be resolved in the near future;
this will facilitate the development of novel therapies directly
targeted to erythropoiesis, which will be effective in steroid-
resistant patients and/or synergize with GC, so that their dose
can be reduced. The experiment just mentioned suggests a
possible therapeutic role for PPARα agonists in DBA.
As supportive therapy improves and patients with DBA
can expect longer lives, it becomes increasingly important to
understand the drivers of malignant transformation in a sub-
set of patients. The relevance of ribosome biology to diseases
far more common than DBA has been highlighted recently by
the discovery of RP gene mutations in hematological malig-
nancies (specifically in T-cell acute lymphoblastic leukemia
and in chronic lymphocytic leukemia). In many cases these
somatically acquired mutations interfere with the function of
the tumor suppressor p53.
Red cell enzyme deficiencies
Inherited abnormalities of red cell enzymes, red cell enzy-
mopathies, are a distinct set of genetic disorders with one
important clinical manifestation in common, namely chronic
hemolytic anemia (acute haemolytic anaemia in the case
of G6PD deficiency). This section deals with those enzy-
mopathies affecting red cell metabolism for which the molec-
ular basis has been elucidated (Table 11.4). We do not discuss
conditions in which an enzyme abnormality is also expressed
in red cells but the main clinical manifestations are elsewhere
(e.g. the porphyrias, galactosemia, Lesch–Nyhan syndrome).
All of these defects, except G6PD deficiency, are rare to very
rare. For the sake of brevity, we will give a brief overview
of enzymopathies, then focus on deficiencies of pyruvate
kinase (PK), glucose-6-phosphate dehydrogenase (G6PD),
and pyrimidine 5′ -nucleotidase –representative of disorders
of glycolysis, the pentose phosphate pathway, and nucleotide
metabolism, respectively.
Clinical features
In these disorders hemolytic anemia occurs with varying
degrees of severity. It is not unusual for the presentation to
be in the guise of severe neonatal jaundice that may require
exchange transfusion; if the anemia is less severe it may
present later in life, or it may even remain asymptomatic and
be detected incidentally when a blood count is performed for
unrelated reasons. The spleen is often enlarged. When other
systemic manifestations occur, they involve the central ner-
vous system, sometimes entailing severe mental retardation,
or the neuromuscular system, or both.
Diagnosis
The diagnosis of hemolytic anemia is usually not diffi-
cult, thanks to the triad of normo-macrocytic anemia,
reticulocytosis, and hyperbilirubinemia. Enzymopathies
should be considered in the differential diagnosis of any
chronic Coombs-negative hemolytic anemia. A definitive
diagnosis can be made only by demonstrating the deficiency
of an individual enzyme by a quantitative assay. For the sake
of economy, it is sensible to carry out these rather laborious
tests in order of frequency of occurrence of the various
enzymopathies: first glucose-6 phosphate dehydrogenase
(G6PD), then PK, then glucose 6-phosphate isomerase, and
so on (Table 11.4). If a particular molecular abnormality is
already known in the family, then of course one could test
directly for that at the DNA level, bypassing the need for
enzyme assays. As for other multigenic heritable disorders,
the advent of high-throughput sequencing technologies
allows simultaneous analysis of multiple genes implicated in
glycolytic enzyme deficiencies, thus simplifying the genetic
diagnosis of these disorders. Nevertheless, enzymatic assays
should still be performed to validate candidate pathogenic
mutations.
Molecular pathophysiology
Most of the glycolytic enzymes involved are housekeeping
enzymes present, by definition, in all cells. Therefore, one
might expect that a severe reduction in activity of any
of these might have generalized clinical manifestations.
However, we can identify at least two reasons why red cells
are more severely affected. First, red cells have a much more
limited metabolic machinery than most other somatic cells;
if a particular enzyme is deficient, other cells may cope by the
 The molecular basis of bone marrow failure syndromes and red cell enzymopathies 145

Table 11.4 Synopsis of red cell enzymopathiesa

Isoenzymeb Prevalence of Main clinical features

characteristic enzyme associated with Benefit from Chromosomal
Enzyme (abbreviation) of red cells deficiency enzyme deficiencyc splenectomyd localization

Adenylate kinase (AK) 1 Very rare HA, CNS Partial 9q34.1

Aldolase A Very rare HA, myopathy 16q22–q24
Cytochrome b5 reductase Rare Pseudocyanosis, CNS 22q13.31–qter
2, 3-Diphosphoglycerate Very rare Polycythemia 7q31–q34
mutase (DPGM)
Enolasee 1 (α) Very rare HA 1pter–p36.13
Glucose 6-phosphate B Common HA None Xq28
dehydrogenase (G6PD)
Glucose 6-phosphate isomerase Rare HA, NM, CNS Partial 19q13.1
(GPI)
γ-Glutamylcysteine synthetase Very rare HA, CNS 6p12
(GLCLC)f
Glutathione peroxidase Very rareg ?g 3q11–q12
(GSH-Px)
Glutathione reductase (GSR) Very rare HA 8p21
Glutathione synthetase (GSS) Very rare HA, CNS 20q11.2
Glyceraldehyde 3-phosphate Very rare HA 12p13.31–p13.1
dehydrogenase (GAPD)e
Hexokinase (HK) R (I) Very rare HA Partial 10q22
Lactate Dehydrogenase-A Very rare Myopathy 11p15
(LDHA)
Monophosphoglycerate mutase B Very rare Myopathy and 10q25.3
(PGAM-B) myoglobinuria
Phosphofructokinase (PFK)h M Very rare HA, myopathy 12q13
L 21q22.3
Phosphoglycerate kinase (PGK) 1 Very rare HA, CNS, NM Partial Xq13
Pyrimidine 5′ -nucleotidase Rare HA Partial 7p15–p14
(P5′ N1)
Pyruvate kinase (PK) Ri Rare HA Partial 1q21
Triosephosphate isomerase (TPI) Very rare HA, CNS, NM, None 12p13

a We have listed all enzymes in the intermediary metabolism of red cells for which, to the best of our knowledge, the corresponding cDNA/gene
has been cloned.
b No entry in this column means that there are no known isoenzymes; therefore it is assumed that the same enzyme type is present in all tissues.
c CNS, central nervous system involvement; HA, hemolytic anemia; NM, neuromuscular manifestations.
d Data available only on some patients.
e Since no mutations have yet been reported, there is no formal proof that HA associated with this enzyme deficiency is due to mutation of the

corresponding gene.
f γ-Glutamylcysteine synthetase consist of two subunits, a catalytic subunit and a regulatory subunit. The data concerning the catalytic subunit

are shown here.

g Enzyme reported to be low in some nutritional deficiencies, but clear evidence that inherited deficiency of glutathione peroxidase exists is

lacking.
h PFK in normal red cells consists of a mixture of the five tetrameric species that can be formed from random association of the M (muscle) and

L (liver) highly homologous subunits (i.e. M4, M3L, M2L2, ML3, L4).
i The red cell form of PK called R is produced by the gene encoding the L (liver) subunit. Because a different promoter is used, the size of liver PK

is 543 amino acids.

Source: Modified from Luzzatto L., Notaro R. (1998) Red cell enzymopathies. In Jameson J.L. (ed.), Principles of Molecular Medicine. Clifton, NJ:
Humana Press, pp. 197–207, with permission.
146 Molecular Hematology
use of alternative or surrogate metabolic pathways. Second,
mature red cells lack nuclei and organelles, thus are not
capable of protein synthesis; therefore, if a particular enzyme
is made highly unstable by a mutation, other cell lineages
can compensate by increased enzyme synthesis, but red cells
cannot.
Management
There is no specific treatment for these conditions. Patients
with moderate anemia may require occasional blood transfu-
sion when they experience exacerbations of the anemia due
to increased rate of hemolysis or to decreased red cell pro-
duction secondary to infection (the most extreme example
being aplastic crisis from parvovirus infection). Patients with
chronic severe anemia may require regular blood transfusion
therapy, and in some patients splenectomy has been benefi-
cial (Table 11.4).
Pyruvate kinase deficiency
PK deficiency is the most common defect of the glycolytic
pathway causing chronic non-spherocytic hemolytic anemia
(CNSHA): it has an estimated frequency of 1 in 20 000 in
people of European ancestry, and a higher frequency in the
Old Order Amish population of Pennsylvania. The clinical
phenotype is variable, from hydrops fetalis (non-immune) to
mild anemia presenting in the second or third decade. There-
fore, underdiagnosis of PK deficiency occurs at both ends of
the clinical spectrum.
Pathophysiology
PK catalyzes the second adenosine triphosphate (ATP)-
producing step of the glycolytic pathway; thus, it accounts
for nearly 50% of the RBCs’ chemical energy in the form
of ATP (see Figure 11.5 for an overview of glycolysis). The
activity of PK, the last reaction in the anaerobic glycolytic
pathway, is allosterically regulated by the product of one
of the first reactions of the pathway, namely fructose-1,6-
bisphosphate (FBP). There are four PK isoforms (M1, M2,
L, and R) encoded by two separate genes (PK-M and PK-LR).
Alternative splicing of the PK-LR transcript encodes PKL and
PKR in the liver and RBCs, respectively.
In PK deficiency, like in other glycolytic enzymopathies,
a shortage of energy supply leads to shortened RBC lifespan
and RBC destruction, mainly in the spleen. Glycolytic
enzymes are normally present in cells in considerable
excess, and therefore the 50% loss of enzyme activity seen
in heterozygotes does not become limiting for red cell
survival; thus, heterozygotes for PK deficiency do not have
hemolytic anemia. Indeed, the disease shows an autosomal
recessive pattern of inheritance: affected patients are either
homozygotes or compound heterozygotes with biallelic
mutations. From the hematologic point of view, it is inter-
esting that PK-deficient patients, after splenectomy, have
reticulocytosis to a degree (up to 50%) very seldom seen in
any other human condition; this has led to the suggestion
that there may be a specific mechanism for the spleen to
remove PK-deficient reticulocytes from the circulation.
Another interesting feature of PK deficiency is that 2,3-
bisphosphoglycerate (BPG) levels are high (up to three times
normal): this produces a right shift in the hemoglobin–
oxygen (O2 ) dissociation curve, thus increasing O2 delivery
to tissues, which partially compensates for anemia.
Molecular genetics
Over 200 mutations have been implicated in PK deficiency
and, as is the case for other glycolytic enzymes, most muta-
tions are of the missense type, causing single amino acid
replacements. This is important, because the low level of
residual enzyme activity can still support some metabolic
flow through the glycolytic pathway, which explains how red
cells survive in the circulation, even though their lifespan is
reduced. With respect to how precisely individual missense
mutations reduce enzyme activity, we must consider at the
protein level several possible mechanisms:
1 In the majority of cases, loss of activity is probably due to
decreased stability of the protein. In such cases we would pre-
dict that, compared to erythrocytes, other cells, e.g. hepato-
cytes, might be much less affected, because they can compen-
sate for decreased stability through increased synthesis of the
enzyme.
2 In some cases the amino acid replacement, being within
or near the active center of the enzyme, may affect substrate
binding (Km ), or catalytic rate (Kcat ), or both.
3 In some cases the allosteric regulation by FBP may be
affected (altered Ka ).
In cases (2) and (3) not only red cells but other cells, in
which the rate of glycolysis is critical, will be affected as
well. Among severe missense mutations of this type, one
of the most common in Caucasians is 994A (Gly332Ser),
which can, when homozygous, cause intrauterine death.
Cloning of the PKLR gene in an expression vector and
purification of recombinant normal and mutant human PKR
enzyme has facilitated study of the biochemical impact of
single amino acid substitutions. Of course, as with most
inherited disorders, the correlation between genotype and
phenotype based on the molecular properties of the enzyme
is only a first approximation, because other inherited and
acquired factors regularly come into play. Whereas most PK
mutations are sporadic, it was found in sub-Saharan Africa
that one polymorphic mutation (E277K) has heterozygote
frequencies up to 7%, and it has been suggested it may have
been biologically selected by malaria.
 The molecular basis of bone marrow failure syndromes and red cell enzymopathies 147

Glucose
ATP
HK
ADP G6PD
Glucose 6-phosphate 6-phosphogluconate

GPI NADP NADPH

Fructose 6-phosphate
ATP
GSH GSSG Pentose
PFK
phosphates
ADP ADP
GSS
Fructose 1,6-bisphosphate
ATP Gly
Aldolase γGluCys

TPI
Cytochrome b5
Dihydroxyacetone Glyceraldehyde
reductase
phosphate 3-phosphate
NAD + Pi HbFe2+
GAPD
NADH HbFe3+
1,3-diphosphoglycerate
ADP
PGK DPGM
2,3-DPG
ATP
Fig. 11.5 The Embden–Meyerhof pathway for
anaerobic glycolysis and the pentose phosphate 3-phosphoglycerate
pathway for oxidative glycolysis, as well as related
reactions (not the complete metabolic machinery PGAM
of the red cell). Enzymes are enclosed in rounded
2-phosphoglycerate
boxes. Abbreviations as in Table 11.4. Additional
abbreviations: DPG, diphosphoglycerate; GSH, Enolase
reduced glutathione; GSSG, glutathione; HbFe2+, H2O
hemoglobin; HbFe3+, methemoglobin; γGluCys, Phosphoenolpyruvate
γ-glutamylcysteine.
ATP
Source: Luzzatto L., Notaro R., (1998). Red cell PK
enzymopathies. In Jameson J.L. (ed.), Principles of
ADP
Molecular Medicine. Clifton, NJ: Humana Press, pp.
197–207. Reproduced with permission. Pyruvate Lactate

Diagnosis may be required in equivocal cases. Prenatal diagnosis of

PK deficiency is suggested by the constellation of clinical
features and hematological indices; unlike in other glycolytic
enzymopathies, red cell morphology can be also suggestive,
with a multitude of cells displaying spiculae on their surface.
As with all autosomal recessive disorders, there may be no
family history. Diagnosis is confirmed by finding in red cells
reduced PK enzymatic activity, which is usually 10–25% of
normal in patients with biallelic mutations. Of note is that,
since routine enzyme tests measure Vmax without regard to
Km or Ka , enzymatic activity correlates poorly with the sever-
ity of hemolysis, and may even be normal. Mutation analysis
at-risk fetuses is also feasible, by amniotic fluid or chorionic
villi sampling, followed by molecular testing of the PKLR
gene.
Management
The mainstay of therapy is supportive, with phototherapy
and/or exchange transfusion in the newborn period, and
regular or intermittent red cell transfusions in children and
adults. Iron chelation must be sometimes recommended
even in patients who are not transfusion dependent, because
with PK deficiency there is a propensity for iron overload.
148 Molecular Hematology
Splenectomy (invariably total) can help to reduce transfusion
requirement. In severe cases, provided there are no systemic
manifestations other than hemolytic anemia, bone marrow
transplantation can be considered; of course, it should be car-
ried out, if at all, before there is organ damage (e.g. from iron
overload).
The development of more specific therapies has been
guided by structural analysis of the PKR enzyme. AG-348, a
novel first-in-class small molecule, has been designed to bind
to the allosteric FBP site of PKR so as to simulate activation by
FBP, and thus functionally restoring the activity of a mutant
enzyme to a normal or near-normal level. In an early clini-
cal trial, this orally administered small molecule resulted in
clinically meaningful improvement in nearly 50% of patients
with PK deficiency, in some cases with normalization of the
hemoglobin level.
Glucose 6-phosphate dehydrogenase
deficiency
G6PD deficiency is by far the most common abnormality
occurring in the pentose phosphate pathway. It affects 200–
300 million individuals worldwide, with a high prevalence
in populations of Africa, southern Europe, the Middle East,
Southeast Asia, and parts of Oceania, as well as in regions
to which migrations from these areas have taken place. The
overall geographical distribution of G6PD deficiency and its
heterogeneity, together with clinical field studies and in vitro
culture experiments, strongly support the view that this com-
mon genetic trait has been selected by Plasmodium falci-
parum malaria, by virtue of the fact that it confers relative
resistance against this highly lethal infection. Delineation of
the molecular mechanism underlying this resistance could
be used to develop new approaches to protect G6PD-normal
individuals against malaria.
Clinical features
Three types of clinical presentation are well characterized:
1 The vast majority of G6PD-deficient people are asymp-
tomatic most of the time, but they are at risk of developing
acute hemolytic anemia (AHA), which may be triggered by
drugs (e.g. primaquine, rasburicase), infections, or ingestion
of fava beans.
2 The risk of developing neonatal jaundice is much greater
in G6PD-deficient than in G6PD-normal newborns. This
is of great public health importance, because untreated
severe neonatal jaundice can lead to permanent neurological
damage.
3 CNSHA: in contrast to the first two, this clinical pre-
sentation is very rare. The clinical picture is rather simi-
lar to CNSHA associated with glycolytic enzymopathies (see
earlier) and again it is of variable severity. However, the
hemolysis is characteristically exacerbated by the same agents
that can cause AHA in people with the ordinary type of G6PD
deficiency.
Diagnosis
In individuals with clinically apparent G6PD-deficiency
(type 3 in the list), the anemia is usually normocytic and
normochromic, and its severity ranges from moderate to
extremely severe. The morphology of red cells is not char-
acteristic (non-spherocytic). AHA (in G6PD deficiency of
type 1) is due to intravascular as well as extravascular hemol-
ysis, and hence is associated with hemoglobinemia and
hemoglobinuria. Here the blood film shows instead rather
spectacular evidence of hemolysis in the guise of anisocy-
tosis, polychromasia, spherocytes, bite cells, blister cells, and
hemighosts. Supravital staining reveals the presence of Heinz
bodies, consisting of precipitates of denatured hemoglobin.
Definitive diagnosis relies on the direct demonstration of
decreased activity of G6PD in red cells by an appropriate
enzyme assay. Of note is that during an acute crisis the level
of enzyme may be overestimated due to reticulocytosis, while
at the same time the oldest (most G6PD-deficient) red cells
have been destroyed.
Molecular genetics
G6PD is a homodimeric or homotetrameric molecule, and its
single subunit is encoded by an X-linked gene that maps to
Xq28. The tetramer, a dimer or the dimer, is in equilibrium
with the dimer itself. Of the 187 mutations reported in the
G6PD gene, most are point mutations causing single amino
acid substitutions. It is therefore likely that complete absence
of G6PD would be lethal; this has been proven in the mouse
(see later). As a result of the phenomenon of X-chromosome
inactivation in somatic cells (lyonization), female heterozy-
gotes are genetic mosaics, in whom approximately one-half
of the red cells are normal and approximately one-half are
G6PD deficient. However, there is a wide distribution around
this mean, and in some cases the skewing is extreme. There-
fore clinical manifestations, such as favism, can occur in both
hemizygous males and heterozygous females, but they tend
to be milder in the latter, roughly in proportion to the frac-
tion of red cells that are G6PD deficient. This correlation
between enzyme activity and clinical phenotype has been for-
mally demonstrated in a study of 200 heterozygous children
exposed to dapsone.
Function of G6PD
In intermediary metabolism G6PD is aptly depicted as
the first step in the pentose phosphate pathway. However,
several lines of evidence indicate that its most essential role
is not to produce pentose, but rather to provide reductive
 The molecular basis of bone marrow failure syndromes and red cell enzymopathies 149

potential in the form of NADPH. NADPH is necessary for Molecular pathophysiology

the replenishment of GSH when it is oxidized by reactive
AHA is seen with most of the variants of G6PD whereby
oxygen species (ROS). These are generated endogenously,
red cells retain some 10% of normal G6PD activity,
all the time in the course of deoxy hemoglobin cycling, at a
resulting in a limited capacity of these cells to withstand
low rate; and at a much higher rate from the metabolism of
the oxidative action of an exogenous factor. In contrast,
pro-oxidant compounds, such as primaquine or substituted
in a subset of variants (referred to as class I variants), the
pyrimidines present in fava beans. In G6PD-normal red cells
steady-state level of G6PD is so low that it becomes limiting
NADPH production is stepped up in response to ROS; in
for red cell survival, even in the absence of any oxidant
G6PD-deficient red cells ROS result in GSH depletion and
challenge; the result is CNSHA. Numerous point mutations
subsequent red cell damage (oxidative hemolyis). G6PD-null
in the G6PD gene causing CNSHA have been identified
mouse embryos have been obtained through targeted inac-
(Figure 11.6); interestingly, some individual mutations
tivation of G6PD in embryonic stem cells. From a detailed
appear to have arisen independently more than once, since
analysis of hemizygous mutant embryos, who die by day
they have been observed in cases from distinct geographical
10.5, it was inferred that the cause of death is precisely the
areas. With the majority of mutations, G6PD deficiency is
onset of aerobic metabolism, confirming the non-redundant
caused, at the biochemical level, by decreased stability of the
role of G6PD in protection against oxidative stress. Interest-
enzyme; that is, a marked acceleration of the physiological
ingly, heterozygous embryos also die, somewhat later, only if
decay of G6PD and of many other cytoplasmic proteins as
their G6PD-null gene is of maternal origin; in this case the
erythrocytes age. With a few mutations, G6PD deficiency
cause of death is a defective placenta, as a consequence of
is caused instead by decreased catalytic activity or altered
the selective inactivation of the paternal X chromosome in
substrate affinity. Although we cannot explain the reason
extra-embryonic tissues.

1 kb

1 2 3 4 5 6 7 8 9 10 11 12 13
(a)

100bp
m U
A S A h k
a t i S v C
M z
2 3 4 5 6 7 8 9 10 11 12 13
X

(b)
U Union; C Canton; M Mediterranean; A A-(202A); k Kaiping; t Taipei; v Viangchan;

m Mahidol; h Chatham; i Coimbra; S Seattle; s Santamaria; a Aures; z Cosenza; A A-(968C).

Fig. 11.6 Distribution of mutations along the human G6PD gene (Xq28). (a) Genomic structure of the human G6PD gene. Exons 1–13
are shown as numbered rectangles (black rectangles represent coding sequences; shaded rectangles represent non-coding sequences). (b) The
locations of amino acid substitutions are shown along the coding sequence of the gene, in which the exons are shown as open boxes.
Substitutions giving rise to the more severe sporadic (class I) variants are shown as filled circles below. Small deletions, a nonsense mutation, and a
splice site mutation are shown as filled rectangles, a cross and ∫, respectively. The milder class II and class III variants are shown as open circles
above; proven polymorphic variants are shown as a letter in a colored circle. Class IV variants are shown as open ellipses. All variants shown result
in G6PD deficiency, except G6PDA.
Source: Luzzatto L., Mehta A., Vulliamy T. (2001). Glucose-6-phosphate dehydrogenase deficiency. In Scriver C.R., Beaudet A.L., Sly W.S., Valle D.
(eds.), The Metabolic and Molecular Basis of Inherited Disease, 8th edn. New York: McGraw-Hill, pp. 4517–4553, with permission.
150 Molecular Hematology
for a severe clinical phenotype in every case, a cluster of
mutations causing CNSHA in exons 10 and 11 corresponds
closely to the region of the molecule where the two subunits
interface. Dimerization is essential for enzyme activity, and
it is not surprising that amino acid replacements in this
region will interfere with dimer formation or cause marked
instability of the dimer (Figure 11.6).
Enzyme testing versus DNA testing
Given that nowadays many laboratories are more geared to
polymerase chain reaction (PCR)-based technologies than to
enzyme assays, it is tempting to diagnose G6PD deficiency
by simply testing for underlying mutations. This is perfectly
feasible, especially in areas where prevalent mutations are
known. However, when no mutation is found this approach is
not definitive, because a known mutation unexpected in that
area, or a novel mutation, would be missed. DNA testing is
certainly the best way to detect all heterozygotes; however, it
will tell us nothing about skewing of X-chromosome inacti-
vation, and therefore it will not predict the risk and severity
of AHA in a heterozygote. For these reasons, measurement
of red cell G6PD activity is still necessary to inform clin-
ical decision-making. The need that has emerged in many
parts of the world to test for G6PD deficiency before admin-
istering primaquine for definitive treatment of Plasmodium
vivax malaria, has prompted very recently the development
of “point of care” G6PD assays requiring practically no labo-
ratory facilities.
Management
The commonest manifestations of G6PD deficiency, neonatal
jaundice and AHA, are largely preventable or controllable by
screening, surveillance, and avoidance of triggering factors,
particularly fava beans, by G6PD-deficient subjects. One of
the current challenges is to develop a good animal model
that will allow pre-clinical testing of new drugs to assess
whether they are likely to cause AHA in G6PD-deficient
individuals. When a patient presents with AHA, and once
the cause is diagnosed, no specific treatment may be needed
if the episode is mild. At the other end of the spectrum, and
especially in children, AHA may be a medical emergency
requiring immediate blood transfusion. The management
of neonatal jaundice does not differ from that of neonatal
jaundice due to causes other than G6PD deficiency and,
in order to prevent neurological damage, treatment with
phototherapy and/or exchange blood transfusion may be
required. The management of CNSHA is similar to that of
CNSHA due to glycolytic enzymopathies, but in addition
it is important to avoid exposure to potentially hemolytic
drugs. Again, although there is no evidence of selective red
cell destruction in the spleen (as seen in hereditary sphero-
cytosis), splenectomy has proven beneficial in severe cases.
Recent advances in molecular biology offer the possibility
of novel therapies for this common disease. First, lifelong
expression of human G6PD at therapeutic levels has been
obtained in RBCs and in white blood cells of mice through
retroviral-mediated transfer into HSCs. Second, in human
G6PD-deficient cells (class I–III variants) restoration of
G6PD activity has been achieved in vitro using HDACi. This
effect is mediated by histone hyperacetylation at the G6PD
promoter, resulting in increased transcription, protein level,
and enzymatic activity of the mutant protein. Interestingly,
HDACi do not exert this effect on all genes within the
glycolytic pathway; rather, transcription of G6PD is selec-
tively upregulated. Further understanding of the mechanism
of action of HDACi and other epigenetic modifiers could
therefore open up therapeutic opportunities in other dis-
eases in which missense mutations result in an enzymatic
deficiency.
Pyrimidine 5′ -nucleotidase deficiency
Nucleotide metabolism in RBCs is necessary for the main-
tenance of ATP and guanosine triphosphate (GTP) levels.
Deficiency of pyrimidine 5′ -nucleotidase, involved in pyrim-
idine metabolism, is of interest for several reasons. First, it
is probably the third most common red cell enzymopathy
(trailing G6PD and PK deficiencies). Second, the diagnosis
can be suspected from red cell morphology, because it is
associated with basophilic stippling (accounted for by the
accumulation of RNA in mature red cells). Third, although
we do not really understand the precise mechanism, 5′ -
nucleotidase deficiency is a good example of how the red
cell has virtually only one way to manifest it is suffering –
almost any metabolic abnormality will lead to accelerated
destruction (i.e. hemolysis). Fourth, the anemia of chronic
lead poisoning (which had been known for a long time to
be associated with basophilic stippling) turns out to be the
consequence of the fact that lead is a powerful inhibitor of
5′ -nucleotidase; thus in terms of its hematological effects,
lead poisoning is a phenocopy of 5′ -nucleotidase deficiency.
Since mapping of the pyrimidine-5′ -nucleotidase gene
(NT5C3) to chromosome 7 in 2001, 24 different mutations,
most commonly in homozygosity, have been identified in just
over 30 unrelated families with 5′ -nucleotidase deficiency.
Mutations include missense, nonsense, alterations of splicing
sites, insertions, and deletions, and diagnosis is by molecular
testing. 5′ -nucleotidase deficiency has also been associated
with hemoglobin E, hemoglobin D Punjab, and spectrin defi-
ciency. In the first case, concurrent 5′ -nucleotidase deficiency
made the clinical expression of HbE disease more severe.
Investigation of a father and daughter pair affected by
5′ -nucleotidase deficiency, in which the father had a more
severe disease phenotype, revealed that the father was
homozygous for a polymorphism in the TATAA element of
 The molecular basis of bone marrow failure syndromes and red cell enzymopathies 151
the promoter region in the uridine diphosphoglucurono-
syl transferase 1A (UGT1A) gene (TA7/TA7), while the
daughter was heterozygous (TA6/TA7). This polymor-
phism, usually associated with benign Gilbert syndrome,
contributed to more severe hemolysis in the father. Such a
case illustrates the value of molecular characterization of
multiplex families affected by a disease primarily regarded as
monogenic, with the aim of identifying disease-modifying
epistatic factors that contribute to clinical heterogeneity and
that could be the target of therapeutic intervention.
Concluding remarks
Since the last edition of this volume, advances in different
areas of biology have already had impacts on the diagnosis
and treatment of BMF and hemolytic anemias. In IAA the
impressive therapeutic success of the TpoR agonist eltrom-
bopag promises to offer a simple and effective therapeutic
option for patients who are not candidates for allogeneic stem
cell transplantation.
In inherited BMF syndromes, the ease of genetic diag-
nosis by next-generation sequencing of multigene panels
already offers accurate and swifter diagnosis for patients
and their families, allowing for appropriate management
decisions, such as selection of disease-free familial donors
for hematopoietic stem cell transplantation and in many
instances pre-natal diagnosis and counseling.
In PNH, the success of eculizumab has spurred the devel-
opment of a new generation of complement inhibitors.
Because these molecules target the C3 component of com-
plement, they are expected to be more effective in con-
trolling extravascular hemolysis (regularly seen in patients
on eculizumab) as well as intravascular hemolysis; and
some could be administered orally. In the meantime, an
eculizumab analog, ravulizumab, with a much longer half-
life, has become available.
Gene editing technologies, in particular CRISPR/Cas9-
based gene correction, offer a more realistic hope than ever
before for gene therapy, in principle for any monogenic dis-
order, such as those covered in this chapter.
Increasing knowledge of epigenetic mechanisms of how
gene transcription is regulated and relevant mechanistic
studies focused on genes associated with heritable BMF and
hemolytic anemias also promise to lead to new therapies.
The case of restoring G6PD activity in G6PD-deficient cells
by simply enhancing transcription of the mutant allele using
HDAC inhibitors provides the proof of concept for a novel
and simple notion: transcriptional upregulation of mutant
gene alleles has significant therapeutic potential, both in the
case of missense mutation-associated enzymopathies with
respect to the wild-type allele, and in the case of diseases
due to haploinsufficiency. Exploration of this approach will
be greatly facilitated by the advent of a plethora of epigenetic
drugs, beyond HDAC inhibitors.
An emerging and highly promising field of therapeutics
for glycolytic enzymopathies is the development of small
molecules acting as allosteric activators.
Therefore, it is realistic to expect that when this chapter is
due to be updated again, more patients with BMF syndromes
and hemolytic anemias will have benefited from novel and
effective treatments based on a deeper understanding of the
molecular pathogenesis of the respective diseases.
Further reading
Aplastic anemia
Desmond, R., Townsley, D.M., Dumitriu, B. et al. (2014). Eltrom-
bopag restores trilineage haematopoiesis in refractory severe aplastic
anaemia that can be sustained on discontinuation of drug. Blood 123
(12): 1818–1825.
Luzzatto L. and Risitano A.M. (2018). Advances in understanding the
pathogenesis of acquired aplastic anaemia. Br J Haematol. 182(6):
758–776.
Marsh, J.C. and Mufti, G.J. (2016). Clinical significance of acquired
somatic mutations in aplastic anaemia. Int. J. Hematol. 104 (2): 159–
167.
Nishimura, J., Yamamoto, M., Hayashi, S. et al. (2014). Genetic variants
in C5 and poor response to eculizumab. N. Engl. J. Med. 370 (7): 632–
639.
Young, N.S. (2002). Acquired aplastic anaemia. Ann. Intern. Med. 136:
534–546.
Young, N.S. and Barrett, A.J. (1995). The treatment of severe acquired
aplastic anaemia. Blood 85: 3367–3377.
Young, N.S. and Maciejewski, J. (1997). The pathophysiology of
acquired aplastic anaemia. N. Engl. J. Med. 336: 1365–1372.
Young, N.S., Scheinberg, P., and Calado, R.T. (2008). Aplastic anaemia.
Curr. Opin. Hematol. 15: 162–168.
Paroxysmal nocturnal hemoglobinuria
Almeida, A.M., Murakami, Y., Layton, D.M. et al. (2006). Hypomorphic
promoter mutation in PIGM causes inherited glycosylphosphatidyli-
nositol deficiency. Nat. Med. 12: 846–851.
Araten, D.J., Nafa, K., Pakdeesuwan, K., and Luzzatto, L. (1999).
Clonal populations of haematopoietic cells with paroxysmal noctur-
nal hemoglobinuria genotype and phenotype are present in normal
individuals. Proc. Natl. Acad. Sci. U.S.A. 96: 5209–5214.
Belet, S., Fieremans, N., Yuan, X. et al. (2014). Early frameshift mutation
in PIGA identified in a large XLID family without neonatal lethality.
Hum. Mutat. 35 (3): 350–355.
Gargiulo, L., Papaioannou, M., Sica, M. et al. (2013).
Glycosylphosphatidylinositol-specific, CD1d-restricted T cells
in paroxysmal nocturnal hemoglobinuria. Blood 121 (14): 2753–
2761.
Hillmen, P., Young, N.S., Schubert, J. et al. (2006). The complement
inhibitor eculizumab in paroxysmal nocturnal hemoglobinuria. N.
Engl. J. Med. 355: 1233–1243.
152 Molecular Hematology
Krawitz, P.M., Höchsmann, B., Murakami, Y. et al. (2013). A case of
paroxysmal nocturnal hemoglobinuria caused by a germline muta-
tion and a somatic mutation in PIGT. Blood 122 (7): 1312–1315.
Luzzatto, L., Bessler, M., and Rotoli, B. (1997). Somatic mutations in
paroxysmal nocturnal hemoglobinuria: a blessing in disguise? Cell 88:
1–4.
Takeda, J., Miyata, T., Kawagoe, K. et al. (1993). Deficiency of the GPI
anchor caused by a somatic mutation of the PIG-A gene in paroxys-
mal nocturnal hemoglobinuria. Cell 73: 703–711.
Fanconi anemia
Alter, B. and Young, N.S. (1998). The bone marrow failure syndromes.
In: Nathan and Oski’s Hematology of Infancy and Childhood, 5e (eds.
D.G. Nathan and S.H. Orkin), 237–335. Philadelphia: WB Saunders.
Ameziane, N., May, P., Haitjema, A. et al. (2015). A novel Fanconi
anaemia subtype associated with a dominant-negative mutation in
RAD51. Nat. Commun. 6: 8829.
Andreassen, P.R., D’Andrea, A.D., and Taniguchi, T. (2004). ATR cou-
ples FANCD2 monoubiquitination to the DNA-damage response.
Genes Dev. 18: 1958–1963.
Cohn, M.A. and D’Andrea, A.D. (2008). Chromatin recruitment of DNA
repair proteins: lessons from the Fanconi anaemia and double-strand
break repair pathways. Mol. Cell 32: 306–312.
Garcia-Higuera, I., Taniguchi, T., Ganesan, S. et al. (2001). Interaction
of the Fanconi anaemia proteins and BRCA1 in a common pathway.
Mol. Cell 7: 249–262.
Howlett, N.G., Taniguchi, T., Olson, S. et al. (2002). Bi-allelic inactiva-
tion of BRCA2 in Fanconi anaemia. Science 297: 606–609.
Joenje, H. and Patel, K.J. (2001). The emerging genetic and molecular
basis of Fanconi anaemia. Nat. Rev. Genet. 2: 446–459.
Levitus, M., Waisfisz, Q., Godthelp, B.C. et al. (2005). The DNA heli-
case BRIP1 is defective in Fanconi anaemia complementation group
J. Nat. Genet. 37: 934–935.
Litman, R., Peng, M., Jin, Z. et al. (2005). BACH1 is critical for homol-
ogous recombination and appears to be the Fanconi anaemia gene
product FANCJ. Cancer Cell 8: 255–265.
Meetei, A.R., Medhurst, A.L., Ling, C. et al. (2005). A human ortholog
of archaeal DNA repair protein HEF is defective in Fanconi anaemia
complementation group M. Nat. Genet. 37: 958–963.
Mosedale, G., Niedzwiedz, W., Alpi, A. et al. (2005). The vertebrate Hef
ortholog is a component of the Fanconi anaemia tumour-suppressor
pathway. Nat. Struct. Mol. Biol. 12: 763–771.
Niedzwiedz, W., Mosedale, G., Johnson, M. et al. (2004). The Fan-
coni anaemia gene FANCC promotes homologous recombination
and error-prone DNA repair. Mol. Cell 15: 607–620.
Pichierri, P. and Rosselli, F. (2004). The DNA crosslink-induced S-phase
checkpoint depends on ATR–CHK1 and ATR–NBS1–FANCD2
pathways. EMBO J. 23: 1178–1187.
Strathdee, C.A., Gavish, H., Shannon, W.R., and Buchwald, M. (1992).
Cloning of cDNAs for Fanconi’s anaemia by functional complemen-
tation. Nature 356: 763–767.
Venkitaraman, A.R. (2004). Tracing the network connecting BRCA and
Fanconi anaemia proteins. Nat. Rev. Cancer 4: 266–276.
Wang, W. (2007). Emergence of a DNA-damage response network con-
sisting of Fanconi anaemia and BRCA proteins. Nat. Rev. Genet. 8:
735–748.
Wang, X., Andreassen, P.R., and D’Andrea, A.D. (2004). Functional
interaction of monoubiquitinated FANCD2 and BRCA2/FANCD1 in
chromatin. Mol. Cell. Biol. 24: 5850–5862.
Wang, A.T. and Smogorzewska, A. (2015). SnapShot: Fanconi anaemia
and associated proteins. Cell 160 (1–2): 354–354.e1. 186.
Xia, B., Dorsman, J.C., Ameziane, N. et al. (2007). Fanconi anaemia is
associated with a defect in the BRCA2 partner PALB2. Nat. Genet. 39:
159–161.
Yamamoto, K., Ishiai, M., Matsushita, N. et al. (2003). Fanconi anaemia
FANCG protein in mitigating radiation- and enzyme-induced DNA
double-strand breaks by homologous recombination in vertebrate
cells. Mol. Cell. Biol. 23: 5421–5430.
Dyskeratosis congenita
Autexier, C. and Lue, N.F. (2006). The structure and function of
telomerase reverse transcriptase. Annu. Rev. Biochem. 75: 493–
517.
de la Fuente, J. and Dokal, I. (2007). Dyskeratosis congenita: advances
in the understanding of the telomerase defect and the role of stem cell
transplantation. Pediatr. Transplant. 11 (6): 584–594.
Heiss, N.S., Knight, S.W., Vulliamy, J.T. et al. (1998). X-linked dysker-
atosis congenita is caused by mutations in a highly conserved gene
with putative nucleolar functions. Nat. Genet. 19: 32–38.
Jones, M., Bisht, K., Savage, S.A. et al. (2009). The shelterin complex and
haematopoiesis. J. Clin. Invest. 126 (5): 1621–1629.
Jongmans, M.C., Verwiel, E.T., Heijdra, Y. et al. (2012). Revertant
somatic mosaicism by mitotic recombination in dyskeratosis con-
genita. Am. J. Hum. Genet. 90: 426–433.
Kirwan, M. and Dokal, I. (2008). Dyskeratosis congenita: a genetic dis-
order of many faces. Clin. Genet. 73: 103–112.
Mitchell, J.R., Wood, E., and Collins, K. (1999). A telomerase compo-
nent is defective in the human disease dyskeratosis congenita. Nature
402: 551–555.
Vulliamy, T., Marrone, A., Goldman, F. et al. (2001). The RNA compo-
nent of telomerase is mutated in autosomal dominant dyskeratosis
congenita. Nature 413: 432–435.
Woo, D.H., Chen, Q., Yang, T.L. et al. (2016). Enhancing a Wnt-telomere
feedback loop restores intestinal stem cell function in a human organ-
otypic model of dyskeratosis congenita. Cell Stem Cell 19 (3): 397–
405.
Diamond–Blackfan anaemia
Clinton, C. and Gazda, H.T. (2009). Diamond-Blackfan anaemia. In:
GeneReviews (eds. R.A. Pagon, M.P. Adam, H.H. Ardinger, et al.),
1993–2016. Seattle, WA: University of Washington, Seattle.
De Keersmaecker, K., Atak, Z.K., Li, N. et al. (2013). Exome sequencing
identifies mutation in CNOT3 and ribosomal genes RPL5 and RPL10
in T-cell acute lymphoblastic leukaemia. Nat. Genet. 45 (2): 186–190.
Draptchinskaia, N., Gustavsson, P., Andersson, B. et al. (1999). The gene
encoding ribosomal protein S19 is mutated in Diamond–Blackfan
anaemia. Nat. Genet. 21: 169–175.
Gerrard, G., Valgañón, M., Foong, H.E. et al. (2013). Target enrichment
and high-throughput sequencing of 80 ribosomal protein genes to
identify mutations associated with Diamond-Blackfan anaemia. Br. J.
Haematol. 162 (4): 530–536.
 The molecular basis of bone marrow failure syndromes and red cell enzymopathies 153
Iskander, D., Psaila, B., Gerrard, G. et al. (2015). Elucidation of the
EP defect in Diamond-Blackfan anaemia by characterisation and
prospective isolation of human EPs. Blood 125 (16): 2553–2557.
Sankaran, V.G., Ghazvinian, R., Do, R. et al. (2012). Exome sequenc-
ing identifies GATA1 mutations resulting in Diamond-Blackfan
anaemia. J. Clin. Invest. 122: 2439–2443.
Vlachos, A., Ball, S., Dahl, N. et al. (2008). Diagnosing and treating Dia-
mond Blackfan anaemia: results of an international clinical consensus
conference. Br. J. Haematol. 142: 859–876.
Red cell enzyme deficiencies
Beutler, E. (1994). Glucose-6-phosphate dehydrogenase deficiency. N.
Engl. J. Med. 324: 169–174.
Chiarelli, L.R., Morera, S.M., Galizzi, A. et al. (2008). Molecular basis
of pyrimidine 5′ -nucleotidase deficiency caused by 3 newly identi-
fied missense mutations (c.187T>C, c.469G>C and c.740T>C) and
a tabulation of known mutations. Blood Cells Mol. Dis. 40 (3): 295–
301.
Grace, R.F., Zanella, A., Neufeld, E.J. et al. (2015). Erythrocyte pyruvate
kinase deficiency: 2015 status report. Am. J. Hematol. 90 (9): 825–830.
Hirono, A., Kanno, H., Miwa, S., and Beutler, E. (2001). Pyruvate defi-
ciency and other enzymopathies of the erythrocyte. In: The Metabolic
and Molecular Basis of Inherited Disease, 8e (eds. C.R. Scriver, A.L.
Beaudet, W.S. Sly and D. Valle), 4637–4664. New York: McGraw-Hill.
Luzzatto, L. (2015). G6PD deficiency: a polymorphism balanced by
heterozygote advantage against malaria. Lancet Haematol. 2: e400–
e401.
Luzzatto, L., Mehta, A., and Vulliamy, T. (2001). Glucose-6-phosphate
dehydrogenase deficiency. In: The Metabolic and Molecular Basis of
Inherited Disease, 8e (eds. C.R. Scriver, A.L. Beaudet, W.S. Sly and D.
Valle), 4517–4553. New York: McGraw-Hill.
Luzzatto, L., Nannelli, C., and Notaro, R. (2016). Glucose-6-phosphate
dehydrogenase deficiency. Hematol. Oncol. Clin. North Am. 30 (2):
373–393.
Makarona, K., Caputo, V.S., Costa, J.R. et al. (2014). Transcriptional and
epigenetic basis for restoration of G6PD enzymatic activity in human
G6PD-deficient cells. Blood 124 (1): 134–141.
Mentzer, W.C. (1998). Pyruvate kinase deficiency and disorders of gly-
colysis. In: Nathan and Oski’s Hematology of Infancy and Childhood,
5e (eds. D.G. Nathan and S.H. Orkin), 665–703. Philadelphia: WB
Saunders.
Pamba, A., Richardson, N.D., Carter, N. et al. (2012). Clinical spec-
trum and severity of hemolytic anaemia in glucose 6-phosphate
dehydrogenase-deficient children receiving dapsone. Blood 120 (20):
4123–4133.
Chapter 12 Anemia of chronic disease
Tomas Ganz
Department of Medicine, David Geffen School of Medicine at UCLA, Los Angeles CA, USA

Introduction and definition, 155 Principles of treatment, 158

General principles, 155 Future treatments, 159
Diagnosis, 155 Further reading, 159
Pathogenesis, 156

compounded by direct suppression of the erythrocyte pre-

Introduction and definition cursor population by inflammatory substances, blood loss,
decreased erythrocyte survival, decreased production of ery-
Anemia of chronic disease (more specifically referred to as
thropoietin, and true iron deficiency. Several of these factors
anemia of inflammation) is the second most common form
may contribute to the pathogenesis of the extreme variant of
of anemia (after iron deficiency) and is seen in the set-
the anemia that develops acutely in the critical care setting.
ting of chronic infections, inflammatory disorders includ-
ing rheumatological conditions and inflammatory bowel dis-
eases, certain malignancies, and chronic kidney diseases. A
related form of anemia develops within days in critically ill
Diagnosis
patients with sepsis or other acute conditions of systemic
inflammation. While the prevalence of iron deficiency in Iron parameters
the industrialized countries is now decreasing, anemia of
Hypoferremia, a decrease in serum iron concentration, is
inflammation would be expected to become more prevalent
the defining feature of anemia of inflammation. It develops
as the number of elderly with chronic inflammatory con-
within hours of the onset of infection or severe inflamma-
ditions continues to rise. The disorder usually presents as
tion. Synthesis of the iron-binding protein transferrin (mea-
a mild-to-moderate anemia, with erythrocyte morphology
sured directly or as total iron-binding capacity) is decreased,
ranging from normocytic and normochromic to microcytic
unlike iron-deficiency anemia where it is often increased.
and hypochromic, depending on the severity and duration of
The decrease in transferrin concentrations develops more
the underlying disease. Anemia of inflammation is defined
slowly than the decrease in serum iron levels because of the
by inadequate erythrocyte production in the setting of low
longer half-life of transferrin (8–12 days) compared with that
serum iron, despite preserved or even increased macrophage
of iron (about 90 minutes).
iron stores.

General principles
The essential pathogenic feature of this anemia is the iron
restriction of hemoglobin synthesis. Unlike iron-deficiency
anemia, the restriction is due to the slow release of iron from
stores rather than the absolute deficiency of total body iron.
Depending on the underlying disease, the anemia is variably
Molecular Hematology, 4th Edition. Edited by Drew Provan and John G. Gribben.
© 2020 John Wiley & Sons Ltd. Published 2020 by John Wiley & Sons Ltd.
Serum ferritin
Ferritin is found in serum as a large multimer consisting
mostly of glycosylated L-ferritin subunits. Ferritin is secreted
by macrophages and hepatocytes, and its serum levels are
increased in response to iron loading of these cells. Systemic
inflammation also increases ferritin secretion. Serum ferritin
concentrations, reflecting inflammation and iron stores,
are increased in anemia of inflammation and decreased
in iron deficiency. Serum ferritin is thus very useful in
the differential diagnosis of patients with low serum iron
concentrations. However, depleted iron stores in patients

155
156 Molecular Hematology
with coexisting inflammation may yield intermediate ferritin
values, and this test becomes less useful. In this situation,
iron deficiency should be suspected if ferritin unexpectedly
decreases below the patient’s usual baseline.
It has been suggested that serum transferrin receptor
assay may be helpful in differentiating iron-deficiency ane-
mia from anemia of inflammation. Soluble transferrin recep-
tor is increased in iron deficiency, reflecting the post-
transcriptional stimulation of the synthesis of transferrin
receptor by iron-regulatory proteins 1 and 2 during cellular
iron deficiency, but it can also be increased whenever ery-
throblast numbers increase as well as in inflammatory dis-
orders, where it may reflect iron restriction affecting ery-
throcyte precursors. It is possible that the quantitative effects
of inflammation and iron deficiency on serum transferrin
receptor assay differ due to the differential effects of the
two processes on the erythroblasts, but this remains to be
explored experimentally and in larger human studies.
Bone marrow studies
Bone marrow aspiration or biopsy is now rarely necessary
for the diagnosis of anemia of inflammation. In general, the
bone marrow appears normal, unless the underlying disease
directly affects it. The most important information obtained
from marrow examination is the content and distribution of
iron, found as storage iron in the cytoplasm of macrophages
or as functional iron in nucleated red cells. In normal indi-
viduals, several Prussian blue-staining particles are seen in
many macrophages, and about one-third of nucleated red
cells, called sideroblasts, contain blue inclusions. In iron defi-
ciency, both sideroblasts and macrophage iron are absent. In
contrast, in anemia of inflammation, sideroblasts are fewer
but macrophage iron is increased. The increase in storage
iron in the face of decreased plasma iron and fewer siderob-
lasts is characteristic of anemia of inflammation. Although
the bone marrow iron stain could be considered the gold
standard for the differential diagnosis of anemia of inflam-
mation and iron deficiency, the procedure causes patient dis-
comfort, can be confounded by differences in interpreter
skills and by the presence of poorly mobilizable iron deposits
from intravenous iron preparations, and the serum ferritin
assay has largely replaced it.
Pathogenesis
Hypoferremia
Within hours of the onset of systemic infection or inflam-
mation, plasma iron concentrations markedly decrease in
both humans and experimental animal models. This defen-
sive response, hypoferremia of inflammation, is thought to
restrict the availability of iron to extracellular microbes,
limiting their proliferation during early phases of infec-
tion. Recent studies indicate that non-transferrin bound iron
(NTBI) is a particularly potent simulator of pathogenesis by
gram-negative bacteria, and may be the most important tar-
get of the hypoferremic response.
Hypoferremia of inflammation is mediated by inflamma-
tory cytokines, including prominently interleukin (IL)-6,
which increase rapidly in the first few hours of infection
and induce the synthesis of the iron-regulatory hormone
hepcidin. Hepcidin binds to the iron efflux channel ferro-
portin on iron-exporting cells (macrophages, enterocytes,
and hepatocytes), causing the internalization and degra-
dation of ferroportin. Iron efflux is diminished and the
continuing consumption of iron depletes the extracellular
iron pool (Figure 12.1). By recycling senescent erythrocytes,
macrophages in the liver and the spleen normally supply
most of the extracellular iron. During infection and inflam-
mation, macrophages retain iron, restricting its delivery to
plasma and to erythrocyte precursors (Figure 12.2).
Iron restriction of erythropoiesis
Although hypoferremia develops within hours of the onset
of infection or inflammation, the effect of decreased iron
and decreased hemoglobin synthesis on circulating ery-
throcytes is not immediately reflected in average erythro-
cyte hemoglobin content or size. This is because the iron
restriction affects only erythrocyte precursors engaged in
hemoglobin synthesis and not mature erythrocytes. New ery-
throcytes normally replace less than 1% of the circulating ery-
throcytes per day, so the average erythrocyte indices (mean
corpuscular hemoglobin concentration, mean corpuscular
volume) change slowly. However, some newer blood analyz-
ers, which can measure the hemoglobin content of reticu-
locytes, detect the effect of iron restriction as a decrease in
reticulocyte hemoglobin before there is a significant change
in total hemoglobin.
Iron restriction in anemia of inflammation is also detected
as an increase in zinc protoporphyrin. Normally, when iron
is sufficient during an intermediate step in the synthesis of
heme, iron becomes incorporated into protoporphyrin IX. In
iron deficiency, zinc partially replaces iron and the amount
of zinc incorporated into protoporphyrin IX is increased.
In anemia of inflammation, zinc protoporphyrin is also
increased, indicating that insufficient iron is reaching the
sites of heme synthesis in developing erythrocytes. Over the
course of several weeks, iron restriction results in anemia
that can eventually become microcytic and hypochromic and
resemble iron-deficiency anemia. The development of ane-
mia of inflammation is accelerated by other effects of inflam-
mation or the underlying disease that may act to shorten ery-
throcyte lifespan or cause blood loss.
 Anemia of chronic disease 157

RBC
Ferroportin
Ferroportin RBC Hepcidin
RBC
RBC

Increased
hepcidin
DMT1 DMT1
Ferroportin
RBC
RBC
Hepcidin

Ferroportin Macrophage Ferroportin Macrophage

Hepcidin
Hepcidin Hepcidin

Hepcidin

Fig. 12.1 Hepcidin–ferroportin interaction causes iron sequestration in macrophages. (Left) Macrophages in the spleen and liver ingest
senescent erythrocytes and extract their iron. When hepcidin is low, ferroportin molecules on macrophage cell membranes rapidly export the
recycled iron (gray tint). (Right) During inflammation, hepcidin synthesis is increased; hepcidin binds to macrophage ferroportin and induces its
internalization and degradation. Export of iron from macrophages slows, and iron remains in macrophage cytoplasm bound to ferritin. DMT,
divalent metal transporter; RBC, red blood cells.

Inflammation

Spleen

↓Fpn
↑hepcidin
↑hepcidin
e
↓F
↑hepc

↓Fpn
Liver ↓Fe
idin

Plasma
↓Fe-Tf
↓Fpn ↓Fe-
Tf
e
↓F

↓Erythropoiesis

Duodenum
Fig. 12.2 Iron restriction in anemia of inflammation. Under the influence of increased hepcidin concentrations, the release of stored and
recycled iron (Fe) from splenic and hepatic macrophages and hepatocytes is decreased, as is the absorption of dietary iron in the duodenum.
Continued iron utilization depletes plasma iron, causing hypoferremia and limiting iron delivery to erythropoietic tissue. Fpn, ferroportin; Tf,
transferrin.
158 Molecular Hematology
Effects of inflammation on iron absorption
and release from stores
Inflammation-induced hepcidin also decreases the absorp-
tion of dietary iron by acting on ferroportin displayed on
the basolateral membranes of enterocytes. Decreased iron
absorption usually does not contribute much to the develop-
ment of anemia, because iron stores are generally sufficient
to supply erythropoiesis for many months. However, the
inflammatory blockade of iron absorption may contribute to
anemia in situations where iron demand is high and inflam-
mation is long-lasting, as is the case in growing children with
inflammatory bowel diseases or juvenile rheumatological
disorders. Hepcidin would also be expected to inhibit the
release of iron from hepatocyte stores, but the contribution
of this effect to anemia of inflammation has not yet been
established.
Cytokines involved in anemia of
inflammation
The contribution of specific cytokines to hypoferremia and
anemia of inflammation has been difficult to delineate,
mainly because cytokines interact in complex networks reg-
ulating each other’s synthesis. In humans, depending on the
disease, therapies targeting IL-1, IL-6, and tumor necro-
sis factor (TNF)-α have all been shown to improve anemia
of inflammation, along with other inflammatory manifesta-
tions. The association of IL-6 with anemia of inflammation
appears to be the strongest. In both humans and mouse mod-
els, IL-6 induces hepcidin synthesis within hours by a tran-
scriptional mechanism causing hypoferremia.
Cytokines also have direct effects on erythropoiesis. Stud-
ies in erythropoietic culture systems have demonstrated that
TNF-α, IL-1, and interferon (IFN)-β and IFN-γ can inter-
fere with erythropoietin stimulation of erythroid progeni-
tors. This effect is seen in mice engineered to lack hepcidin,
where treatment with potent inflammatory stimuli does not
cause hypoferremia but anemia still develops, albeit in a sub-
stantially milder form than in the presence of hepcidin and
hypoferremia.
Inflammation and erythropoietin
resistance
The normal response to anemia is an increase in erythro-
poietin production and subsequent compensatory increase in
erythropoiesis. It has been suggested that for the same sever-
ity of anemia, erythropoietin production in anemia of inflam-
mation is lower than it would be in other types of anemia.
Indeed, in some studies of patients with anemia associated
with rheumatoid arthritis or cancer, erythropoietin levels
were not as high as in iron-deficiency anemia, although other
studies have not observed a significant difference in erythro-
poietin concentrations between iron deficiency and anemia
of inflammation. Erythropoietin suppression by cytokines
or lipopolysaccharide has been seen in some animal mod-
els of sepsis as well as in erythropoietin-producing cell lines.
However, such suppression of erythropoietin production is
not a major pathogenic mechanism of anemia of inflamma-
tion. If it were, the administration of relatively small amounts
of erythropoietin should be sufficient to reverse the anemia
of inflammation. Clinical experience and studies in animal
models suggest the opposite: severe inflammation decreases
the effectiveness of erythropoietin, a common clinical situa-
tion described as erythropoietin resistance. Both anemia and
erythropoietin resistance have been seen in mice moderately
overexpressing hepcidin, supporting the importance of the
hepcidin–ferroportin axis and inflammatory iron restriction
in the pathogenesis of anemia of inflammation, as well as in
erythropoietin resistance.
Effects of inflammation on erythrocyte
lifespan
In early studies, a small decrease in erythrocyte lifespan was
noted in some animal models of inflammation, and it was
suggested that inflammation activates reticuloendothelial
macrophages to increase the removal of damaged or senes-
cent erythrocytes. Other potential disease-specific mecha-
nisms that could contribute to the decreased lifespan include
the deposition of opsonic antibodies on erythrocytes, and
mechanical damage to erythrocytes from microvascular fib-
rin strands or injured endothelia. In most cases, the effects on
erythrocyte lifespan are relatively small, and would be com-
pensated for by a minor increase in erythropoiesis.
Principles of treatment
Most often, the presenting symptoms of anemia of inflam-
mation, fatigue, and exercise intolerance are difficult to
distinguish from those of the underlying disease, be it an
infection, inflammatory disease, or malignancy. Whenever
possible, the treatment should be aimed at the underlying
disease. Effective treatment of the underlying disease will
also lead to the resolution of anemia. Unfortunately, this is
not always feasible.
Infrequently, anemia of inflammation can be severe
enough to exacerbate underlying cardiac or pulmonary
disease and cause cardiac ischemia. In this acute setting,
erythrocyte transfusion can rapidly improve oxygen delivery
and the manifestations of ischemia. More commonly, anemia
is moderate and its treatment is less urgent. Erythropoietin

with or without high-dose parenteral iron can be useful
for anemia symptoms that cannot be reversed by treating
the causative disease. High doses of parenteral iron may be
required to overwhelm the hepcidin-induced retention of
iron in macrophages, and possibly to expand erythrocyte
precursor populations that produce a hepcidin-suppressing
factor. In each patient, the potential side effects of erythro-
poietin and iron must be weighed against the anticipated
benefits, and the clinical effectiveness of the drugs. The con-
tinued need for therapy should be evaluated regularly and
critically.
Future treatments
Improved understanding of the pathogenesis of anemia of
inflammation should lead to the development of therapies
targeting the cytokine pathways that mediate increased
hepcidin production, hepcidin itself, its interaction with
ferroportin, or the ferroportin internalization pathway. If
successful, such interventions should reverse the iron block
and provide sufficient iron for normal erythropoiesis.
Further reading
General
Nemeth, E. and Ganz, T. (2014). Anemia of inflammation. Hematol.
Oncol. Clin. North Am. 28: 671–681.
Diagnosis
Braga, F., Infusino, I., Dolci, A., and Panteghini, M. (2014). Soluble
transferrin receptor in complicated anemia. Clin. Chim. Acta 431:
143–147.
Hastka, J., Lasserre, J.J., Schwarzbeck, A. et al. (1993). Zinc protopor-
phyrin in anemia of chronic disorders. Blood 81: 1200–1204.
Anemia of chronic disease 159
Mechanisms of hypoferremia
Kawabata, H., Tomosugi, N., Kanda, J. et al. (2007). Anti-interleukin 6
receptor antibody tocilizumab reduces the level of serum hepcidin
in patients with multicentric Castleman’s disease. Haematologica 92:
857–858.
Kemna, E., Pickkers, P., Nemeth, E. et al. (2005). Time-course analysis of
hepcidin, serum iron, and plasma cytokine levels in humans injected
with LPS. Blood 106: 1864–1866.
Nemeth, E., Rivera, S., Gabayan, V. et al. (2004). IL-6 mediates hypofer-
remia of inflammation by inducing the synthesis of the iron regula-
tory hormone hepcidin. J. Clin. Invest. 113: 1271–1276.
Rivera, S., Nemeth, E., Gabayan, V. et al. (2005). Synthetic hepcidin
causes rapid dose-dependent hypoferremia and is concentrated in
ferroportin-containing organs. Blood 106: 2196–2199.
Hepcidin and anemia
Gardenghi, S., Renaud, T.M., Meloni, A. et al. (2014). Distinct roles
for hepcidin and interleukin-6 in the recovery from anemia in mice
injected with heat-killed Brucella abortus. Blood 123: 137–145.
Kim, A., Fung, E., Parikh, S.G. et al. (2014). A mouse model of anemia
of inflammation: complex pathogenesis with partial dependence on
hepcidin. Blood 123: 1129–1136.
Nicolas, G., Bennoun, M., Porteu, A. et al. (2002). Severe iron deficiency
anemia in transgenic mice expressing liver hepcidin. Proc. Natl. Acad.
Sci. U.S.A. 99: 4596–4601.
Roy, C.N., Mak, H.H., Akpan, I. et al. (2007). Hepcidin antimicrobial
peptide transgenic mice exhibit features of the anemia of inflamma-
tion. Blood 109: 4038–4044.
Clinical studies
Cash, J.M. and Sears, D.A. (1989). The anemia of chronic disease: spec-
trum of associated diseases in a series of unselected hospitalized
patients. Am. J. Med. 87: 638–644.
Cazzola, M., Ponchio, L., de Benedetti, F. et al. (1996). Defective iron
supply for erythropoiesis and adequate endogenous erythropoietin
production in the anemia associated with systemic-onset juvenile
chronic arthritis. Blood 87: 4824–4830.
Corwin, H.L. and Krantz, S.B. (2000). Anemia of the critically ill: “acute”
anemia of chronic disease. Crit. Care Med. 28: 3098–3099.
Chapter 13 The molecular basis of
iron metabolism
Nancy C. Andrews1 & Tomas Ganz2
1
Duke University School of Medicine, Durham NC, USA
2
Department of Medicine, David Geffen School of Medicine at UCLA, Los Angeles CA, USA

Introduction, 161 Iron disorders, 166

Mechanisms of iron transport, 161 Conclusions, 170
Regulation of iron homeostasis, 164 Further reading, 171

Our understanding of the molecular processes of iron

Introduction metabolism has advanced considerably over the last two
decades as new techniques in genetics and molecular biology
Iron is one of several metals that are essential for nor-
have been applied to problems in this field. Much of what we
mal cellular processes. Its primary role in mammalian biol-
have learned has come from the study of patients with dis-
ogy is to bind oxygen in hemoglobin and myoglobin, and
eases of iron metabolism, and from investigation of animals
to catalyze the enzymatic transfer of electrons by iron-
with spontaneous and induced mutations in genes important
dependent enzymes, including cytochromes, peroxidases,
for the transport and storage of iron. Most features of iron
ribonucleotide reductases, and catalases. When iron is defi-
metabolism are very similar among mammalian species, val-
cient, the synthesis of iron-containing proteins is impaired,
idating the use of rodent models. More recently, important
with adverse consequences for oxygen delivery and cellular
insights into iron disorders in humans have also come from
metabolism. However, the same properties that make iron
the study of iron metabolism in zebrafish and even yeast.
useful for these functions can also lead to cellular damage
when iron is present in excess. Normally, several binding
proteins constrain the activity of iron, but when their iron-
binding capacity is exceeded, iron promotes the formation
Mechanisms of iron transport
of reactive oxygen species that attack cellular lipids, proteins,
and nucleic acids. Thus, iron balance must be carefully main- General principles
tained to avoid the deleterious effects of iron deficiency and
Iron is a large charged ion that cannot freely diffuse across
iron overload. All known disorders of iron metabolism can
cellular membranes. Transmembrane transfer requires
be considered abnormalities of iron balance.
specific carrier proteins. There are two general ways in
There is no physiological excretion mechanism for iron:
which cells transport iron. Some cells, such as intestinal
iron losses result only from bleeding and exfoliation of skin
epithelial cells, hepatocytes, macrophages, and placental
and mucosal cells. Under normal conditions, iron enters the
trophoblast, are equipped both to take in (import) iron and
body exclusively by dietary absorption, and absorption is
to release (export) it. These cell types are involved in the
closely regulated to balance the small losses. Iron balance is
acquisition, storage, and mobilization of iron for the entire
disrupted when intake and losses are not matched. Iron defi-
organism. Many other cells import iron but do not release it
ciency occurs when the dietary iron supply is inadequate,
unless the cells are destroyed. An intermediate group of cells,
when losses are increased (primarily because of bleeding), or
including erythrocyte precursors and cardiac myocytes,
when both of these circumstances are present. Iron overload
export iron primarily to maintain cellular homeostasis, but
results when iron absorption is inappropriately increased
the released iron contributes little to systemic iron needs.
due to genetic defects in iron-regulatory proteins, or when
Approximately 25 mg of iron are needed every day to sup-
repeated blood transfusions create a substantial iron burden.
port hemoglobin production in maturing erythrocytes. This
amount is much greater than the 1–2 mg entering the body
Molecular Hematology, 4th Edition. Edited by Drew Provan and John G. Gribben. each day through the intestine. The iron for erythropoiesis is
© 2020 John Wiley & Sons Ltd. Published 2020 by John Wiley & Sons Ltd. largely provided by splenic and hepatic (reticuloendothelial)

161
162 Molecular Hematology

macrophages recycling iron from old erythrocytes and
making it available to developing erythroid precursors.
Cells use at least three mechanisms to take up iron. Intesti-
nal absorptive cells have cell surface transporters that carry
ferrous (Fe2+ ) ions directly across the membrane. Erythroid
precursors use membrane receptors to take up iron bound to
a protein carrier, concentrate the iron in a subcellular com-
partment, and then transfer it across the membrane of that
compartment into the cytoplasm. Hepatocytes use both of
these mechanisms. Finally, recycling macrophages acquire
iron through the phagocytosis of aged or damaged erythro-
cytes, lysing the cells and extracting the iron from their
hemoglobin.
Intestinal iron transport
Our current understanding of intestinal non-heme iron
transport is illustrated in Figure 13.1. Iron absorption takes
place in an acidic environment in the proximal small intes-
tine, just distal to the gastric outlet. Most non-heme dietary
iron is in the ferric (Fe3+ ) form, whose very limited solubility
is enhanced by the low pH. It is reduced to Fe2+ by a brush
border ferrireductases, including the duodenal cytochrome b
(DCYTB). The Fe2+ ions pass through divalent metal trans-
porter 1 (DMT1, formerly called Nramp2, DCT1), a mem-
brane protein that allows iron to traverse the apical bilayer.
DMT1 requires an acidic environment for its activity, because
it cotransports protons with iron atoms. The requisite pro-
tons are generated by apical Na+ /H+ exchangers, predomi-
nantly NHE3. Once it crosses the membrane, some of the iron
is retained within the absorptive intestinal cells (enterocytes),
and some is exported through the basolateral membrane
through the action of a distinct transporter, ferroportin. A
multicopper oxidase protein, hephaestin, facilitates basolat-
eral transport, perhaps by oxidizing Fe2+ to Fe3+ to allow it
to exit from ferroportin and to bind to the plasma iron carrier
protein, transferrin.
The expression of ferroportin on the basolateral mem-
brane of enterocytes is the principal mechanism control-
ling the rate of iron flux through this transport system.
The more ferroportin on the basolateral membranes, the
more iron from enterocytes is passed into the plasma. Iron
retained within the enterocytes is lost from the body when
these cells finish their short lifespan and slough into the gut
lumen. The partitioning of iron (i.e. the process that gov-
erns how much enters the plasma and how much is retained
within cells) is regulated by the hormone hepcidin through
post-translational degradation of ferroportin. The hepcidin–
ferroportin interaction plays an important role in determin-
ing the overall efficiency of iron absorption.
In humans and other meat-eating organisms, direct
absorption of heme contributes importantly to total iron
absorption. The pathways involved in heme absorption are
not well understood and may not be present in mice.
Other iron uptake mechanisms
Upon entry into the plasma, iron attaches to transferrin,
an abundant plasma protein that binds iron with extremely
high affinity. Transferrin serves three important functions.
First, it keeps iron in solution. In an aqueous, neutral
pH environment, iron exists as Fe3+ ion, which is almost

? e– Fe3+
Fe Hephaestin Fe2Tf
Lumen
Heme
Blood
Ferroportin
Fe Fe Fe2+

HO
Fe3+ Fe2+ Fe3+
Ferrireductases
Ferritin
Fe2+
H+ DMT1

H+ NHE3
+
Na Enterocyte

Fig. 13.1 Intestinal non-heme iron absorption. The cartoon shows an absorptive enterocyte from the duodenal epithelium, joined to adjacent
cells by iron-impermeable tight junctions. The apical brush border is on the luminal side (left) and the basolateral surface is on the blood side
(right). Dietary Fe3+ iron is reduced by the ferrireductase DCYTB and other mechanisms to produce Fe2+ ion for transport. Fe2+ crosses the apical
membrane through the action of DMT1 to enter the cell. There, iron is partitioned between storage and export; stored iron is ultimately lost from
the body when the epithelial cells senesce and exfoliate into the gut lumen. Meanwhile, a fraction of the iron is exported across the basolateral
membrane by ferroportin. The iron hormone hepcidin binds to ferroportin and causes its degradation, thereby inhibiting iron efflux, proportionally
to hepcidin concentration. Hephaestin is a ferroxidase-like protein that aids in iron export, probably by oxidizing the Fe2+ iron leaving ferroportin to
the Fe3+ form, which binds to plasma transferrin (TF).

insoluble. Second, it renders iron non-reactive, and allows
it to circulate in a safe, non-toxic form. Third, transferrin
facilitates the delivery of iron to cells bearing transferrin
receptors (TFRs) on their surfaces.
Most differentiated cell types express few, if any, trans-
ferrin receptors, but there are three important exceptions:
tumor cells, activated lymphocytes, and erythroid precursors.
Tumor cells presumably use transferrin receptors to opti-
mize iron uptake to support rapid proliferation. The same
may be true for activated lymphocytes. However, the great-
est demand for iron is by erythroid precursors, to support
the large-scale production of hemoglobin. In normal adults,
about two-thirds of the total body iron endowment is found
in hemoglobin, distributed among erythroid precursor cells
and circulating erythrocytes.
Iron-loaded transferrin binds with high affinity to trans-
ferrin receptors. Of the two forms of the receptor, transfer-
rin receptor 1 (TFR1) appears to be critical for cellular iron
uptake, but both TFR1 and transferrin receptor 2 (TFR2) are
involved in systemic iron homeostasis. Transferrin receptors,
carrying their iron–transferrin cargo, undergo endocytosis
(Figure 13.2). Portions of the cell membrane bearing trans-
ferrin receptors invaginate into the cytoplasm and bud off
as intracellular vesicles (endosomes). Protons are pumped
Transferrin receptor
DMTI
H+
H+
Erythroid precursor
Fe2.transferrin
–Apo-transferrin
Fig. 13.2 The transferrin cycle. The transferrin cycle of
receptor-mediated endocytosis is initiated by binding of diferric
(Fe2 )-transferrin to a cell surface transferrin receptor. The
ligand–receptor complex is internalized by invagination of
clathrin-coated pits to form specialized endosomes. Influx of protons
into the endosome decreases its pH to approximately 5.5, facilitating
release of iron from transferrin. The iron is then transferred to the
cytoplasm by DMT1. Apo-transferrin and transferrin receptor return to
the cell surface for further cycles of iron uptake.
The molecular basis of iron metabolism 163
into the endosomes to lower their internal pH, leading to the
release of iron from transferrin. The liberated iron then leaves
the endosome to enter the cytoplasm. This also requires a
transmembrane transport step, which is probably mediated
by DMT1 and facilitated by the low endosomal pH. The pro-
cess is assisted by one or more essential ferrireductases, in
erythroid cells principally Steap3 (six transmembrane epithe-
lial antigen of the prostate 3), which convert ferric to ferrous
iron. Within the cytoplasm, ferrous iron is shuttled to sites
of use and storage. Two iron chaperones, poly(rC)-binding
proteins (PCRB) 1 and 2, have been shown to escort iron to
ferritin and several other proteins that require or transport
iron. Meanwhile, transferrin and transferrin receptor pro-
teins return to the cell surface, where transferrin is released
and the receptors become available for further cycles of iron
delivery.
Why should cells have evolved the complicated trans-
ferrin cycle when it is possible to take up iron directly?
There are at least two likely answers. First, tight binding
of iron to transferrin is advantageous while iron is in the
circulation, but it complicates the transport of iron into
cells. The pH-dependent release of iron, occurring in a con-
trolled intracellular environment, solves the problem of lib-
erating the iron. Second, binding of iron-loaded transfer-
rin to transferrin receptors serves to concentrate iron in the
vicinity of DMT1, probably achieving much higher local
iron concentrations than would be possible without such a
mechanism. This allows more efficient iron uptake by cells
with large needs (erythroid precursors, tumor cells, activated
lymphocytes), without exposing other cells to unnecessary
iron.
Undoubtedly, other cell types use these and other schemes
for assimilating iron. Hepatocytes and macrophages are par-
ticularly important in iron homeostasis and their iron uptake
mechanisms, though not well understood, deserve men-
tion. Hepatocytes express both types of transferrin recep-
tor and likely take up iron via the transferrin cycle. Hepato-
cytes also avidly take up non-transferrin-bound iron (NTBI)
when the plasma iron concentration exceeds the binding
capacity of transferrin. This is an abnormal situation, except
perhaps in the portal circulation after an iron-rich meal,
because there are usually about three times as many trans-
ferrin iron-binding sites as are needed (i.e. transferrin is
normally about 30% saturated with iron). However, patients
with iron overload may have more iron than transferrin can
accommodate, and this excess iron appears to be rapidly
removed from the circulation by hepatocytes. The molecu-
lar mechanism for hepatic NTBI uptake uses the SLC39A14
(also called ZIP14) transporter, and in the absence of this
transporter in mice, NTBI is no longer taken up by the
liver. Hepatocytes express ferroportin and release iron at
least in part by a hepcidin-controlled ferroportin-dependent
mechanism.
164 Molecular Hematology
As discussed earlier, reticuloendothelial macrophages
obtain iron by phagocytosing and breaking down ery-
throcytes. This process takes place in discrete phagocytic
vesicles within the cells, and likely involves the action of
heme oxygenase, an enzyme that catalyzes the degrada-
tion of heme. After crossing the phagosomal membranes,
cytoplasmic iron is stored in ferritin. Macrophages export
ferrous iron through their abundant ferroportin, assisted
by the plasma ferroxidase ceruloplasmin so that ferric iron
is delivered to plasma transferrin. Similar to intestinal cells
and hepatocytes, reticuloendothelial macrophages partition
their iron content into retained and released portions. This
process is regulated in response to the iron needs of the
body by the hepatic hormone hepcidin, which induces the
internalization and degradation of macrophage ferroportin.
While there is currently no direct way to measure how
much iron is retained and how much is released, com-
monly used laboratory tests provide some information.
The concentration of serum iron (and hence transferrin
saturation) is determined by two factors: macrophage iron
release and erythroid iron utilization. When erythropoiesis
occurs at a steady rate, transferrin saturation is determined
primarily by the rate of macrophage iron release, increasing
when there is increased iron export and decreasing when
iron is retained or when less iron is being recycled from
erythrocytes. In contrast, the concentration of serum ferritin
roughly correlates with the amount of storage iron in the
body. Serum ferritin appears to be derived primarily from
reticuloendothelial macrophages and hepatocytes. However,
serum ferritin is not a very accurate indicator, because levels
are increased by inflammation, tissue damage, and rare
congenital hyperferritinemia disorders. Nonetheless, a low
serum ferritin value invariably indicates depleted iron stores.
Sites of iron storage
Iron is stored within cells in the cavities of ferritin protein
multimers. There are two types of ferritin subunit, L and H,
both approximately 20 kDa in size. These subunits assemble
in varying proportions into 24-subunit cage-like structures.
Up to several thousand iron atoms can be stored in each fer-
ritin multimer. Like transferrin, ferritin’s function is to pre-
vent iron from reacting with other cellular constituents, and
allows controlled iron release in response to increased cellu-
lar needs. The molecular details of iron incorporation into
and release from ferritin are not fully understood, but it now
appears that iron is released by ferritinophagy, a regulated
process in which nuclear receptor coactivator 4 (NCOA4)
binds to ferritin under low iron conditions and delivers fer-
ritin to the autophagy pathway, culminating in the degrada-
tion of ferritin in lysosomes and the release of iron into the
cytoplasm. Under some circumstances, ferritin and other cel-
lular components are partially degraded and conglomerated
to form hemosiderin, a heterogeneous iron-containing sub-
stance that probably serves little purpose but to keep iron
from causing harm. Both ferritin and hemosiderin accumu-
late in iron-overloaded tissues.
The liver serves as the primary depot for iron in excess of
immediate needs. It has a very large capacity for storing iron,
though this capacity is ultimately exceeded in iron overload
disorders. While other tissues (myocardium, pancreas) also
fill up with iron in iron overload, the liver is frequently the
first site where damage from iron overload becomes apparent.
Hepatocytes avidly take up NTBI from the plasma.
Reticuloendothelial macrophages are also important for
iron storage, but their iron comes from degraded erythro-
cytes. Patients treated with frequent transfusions typically
accumulate excess iron in macrophages first, and only later
in other tissues. This pattern of iron accumulation has been
referred to as “siderosis” to distinguish it from hemochro-
matosis, which is primary iron loading of parenchymal cells.
Regulation of iron homeostasis
Iron homeostasis requires the coordinated regulation of iron
transport and iron storage so that tissues will have adequate
amounts to meet their needs, but will not become overloaded
with iron. Regulation must involve the control of cellular iron
import, export, and partitioning. Over the last decade, a bet-
ter understanding of regulation at each of these steps has
emerged.
There are at least four known regulators of intestinal iron
absorption (Figure 13.3): iron stores, erythropoietic demand,
hypoxia, and inflammation. The stores regulator modulates
absorption several-fold, increasing absorption in iron defi-
ciency and decreasing absorption in iron overload. The ery-
throid regulator is more potent: it can increase iron absorp-
tion many-fold when erythropoiesis becomes iron restricted.
The hypoxia regulator is not well characterized, but its effects
may be at least partly distinct from those of the erythroid reg-
ulator. This regulator increases iron absorption in response
to hypoxia. Finally, an inflammation regulator decreases iron
absorption in response to inflammation from a variety of
causes.
Similar influences also regulate the release of recycled
iron from reticuloendothelial macrophages. During inflam-
mation or iron sufficiency, iron is retained in macrophages,
but is released in response to iron deficiency or erythroid
demand. The regulation of iron storage in hepatocytes is less
well understood, but it is clearly responsive to systemic iron
requirements.
The iron-regulatory hormone hepcidin is likely to be
the common effector of the stores, erythroid, hypoxia, and
inflammation regulators (Figure 13.3). Hepcidin (also called
LEAP, HAMP) is produced by hepatocytes as a 25-amino acid

Hypoxia Inflammation
Liver Erythropoietic signal
he
pc
Plasma
id
Fe-Tf
in
(a) Iron signal
Fig. 13.3 Regulators of iron absorption, storage, and tissue distribution. (a) Iron stores, erythropoietic demand, hypoxia, and inflammation
modulate the hepatic production of hepcidin, an iron-regulatory hormone. Increased iron stores and inflammation both increase hepcidin
expression (↑), whereas increased erythropoietic demand and hypoxia decrease hepcidin expression (⊥). (b) Hepcidin controls the major flows of
iron in the body, including the efflux of iron into plasma from absorptive enterocytes, from splenic and hepatic macrophages involved in recycling
of iron from erythrocytes, and from hepatocytes that store iron. Plasma iron is destined for hemoglobin synthesis by erythrocyte precursors in the
bone marrow, and for other sites of iron utilization. Fe, iron; Fpn, ferroportin; Tf, transferrin.
peptide with four disulfide bonds from an 84-amino acid pre-
cursor molecule, with the ultimate cleavage mediated by the
prohormone convertase furin. Administration of synthetic
hepcidin to mice results in profound and prolonged hypofer-
remia. The hypoferremic effect of hepcidin is due to its ability
to inhibit the major mechanisms that deliver iron to plasma:
intestinal iron absorption and the release of iron from recy-
cling compartments and stores in the spleen and the liver.
Continuing consumption of iron by erythropoiesis and other
iron-requiring processes then rapidly depletes the relatively
small plasma and extracellular iron compartment.
On the cellular and molecular levels, hepcidin inhibits
the efflux of cellular iron into extracellular fluid or plasma,
exerting its effect on the major cell types involved in iron
transport: enterocytes, macrophages, hepatocytes, and pla-
cental syncytiotrophoblast. Unlike other cell types, iron-
exporting cells express ferroportin, the sole known cellu-
lar iron exporter. Hepcidin acts by binding to ferroportin,
inducing its internalization and lysosomal degradation (Fig-
ure 13.4). The ferroportin degradation pathway is similar
to that of other receptors undergoing ligand-induced endo-
cytosis, and is triggered by ubiquitination of a string of
lysine residues in the cytoplasmic loop connecting the two
halves of the molecule. The loss of ferroportin from cell
membranes proportionately reduces the efflux of iron into
extracellular fluid. Alternatively, at higher concentrations,
hepcidin may directly occlude the channel through which
iron moves.
The molecular basis of iron metabolism 165
he
pc
idi Spleen
n
Fpn
Liver
hepcidin
Fe
Fpn
Fe
Plasma
he Fe-Tf
pc Fe-T
idi f
n
Fpn
Bone marrow
Fe
and other sites
of iron usage
Duodenum (b)
The production of hepcidin is increased in response to
dietary or parenteral iron loading, presumably as part of
a compensatory mechanism to decrease iron absorption
and decrease plasma iron (primarily derived from recycling
macrophages). Hepcidin thus acts as the stores regulator. The
production of hepcidin is decreased by iron deficiency, allow-
ing more iron to enter the body through the intestine and
more iron to enter the plasma from recycling macrophages.
In addition, hepcidin is suppressed by increased erythro-
poietic activity, sometimes even in the face of systemic iron
overload, confirming that hepcidin is the ultimate erythroid
regulator. How enhanced erythropoiesis suppresses hepcidin
production is not yet fully understood. Recent studies indi-
cate that erythropoietin-stimulated erythroblasts secrete ery-
throferrone (also called Fam132b), a member of the TNFα
superfamily, which acts on hepatocytes through an unknown
pathway to suppress hepcidin transcription. The production
of hepcidin is also decreased in hypoxia, suggesting that
hepcidin is an ultimate effector of the hypoxia regulator. The
effect of hypoxia on hepcidin could be partly mediated by
hypoxia-induced erythropoietin acting on erythroblasts to
increase erythroferrone secretion. Finally, hepcidin expres-
sion is induced by inflammation, probably through a direct
action of the cytokine interleukin (IL)-6 and other inflamma-
tory mediators on hepatocytes, where hepcidin transcription
is then modulated by the JAK2-STAT3 regulatory pathway.
In this case, induced hepcidin expression leads to decreased
intestinal iron absorption and decreased macrophage iron
166 Molecular Hematology
DMT1
Hepcidin
Ferritin
Ferroportin
Fig. 13.4 Regulation of cellular iron transport by hepcidin. Hepcidin regulates the release of iron from enterocytes (shown in the figure) and
macrophages by binding to ferroportin and inducing its internalization and degradation in lysosomes. Under the influence of hepcidin, iron export
is decreased and iron accumulates in cytoplasmic ferritin. In enterocytes, regulatory effects of increased cellular iron may result in decreased iron
uptake.
release, acting as an inflammation regulator. There is grow-
ing evidence that, in response to the inflammatory regulator,
increased hepcidin expression contributes to the abnormal
iron homeostasis observed in the anemia of chronic disease
(also known as the anemia of inflammation). Useful clinical
assays for hepcidin levels in serum are available.
Based on recent studies in cellular models, genetically
altered mice, and patients with genetic iron disorders, hep-
cidin synthesis in hepatocytes is regulated by iron sensors
and signaling molecules assembled around the scaffold of the
bone morphogenetic protein receptor (BMPR) and its sig-
naling pathway. Two molecules that may function as iron
sensors are TFR1 and TFR2. The transmembrane molecule
HFE can bind to both transferrin receptors, but is displaced
from TFR1 when the receptor binds its main ligand, iron-
transferrin. The glycosylphosphatidylinositol (GPI)-linked
membrane protein hemojuvelin binds to BMPR and its bone
morphogenetic protein (BMP) ligands, and may act as a co-
receptor that converts the generic BMPR pathway into an
DMT1
Ferritin
Ferroportin
Hepcidin
iron-regulatory pathway. Although multiple BMPs, including
BMP-2, BMP-4, BMP-6, and BMP-9, can stimulate hepcidin
synthesis in hepatocytes and hepatocyte-derived cell lines,
BMP-6 and BMP-2 appear to be uniquely specialized for sig-
naling in iron homeostasis. Remarkably, the BMPs impor-
tant for iron homeostasis are not produced by hepatocytes
but by adjacent sinusoidal endothelial cells, yet their secre-
tion reflects closely the amounts of iron stored in hepatocytes.
How all these cells and molecules interact to regulate hep-
cidin is still an unfinished story.
Iron disorders
In the most general terms, iron disorders represent a failure of
homeostatic mechanisms that match intestinal iron absorp-
tion and systemic iron distribution to the iron requirements
of erythropoiesis and other iron-consuming processes. Iron
homeostasis results in the maintenance of normal plasma

and extracellular iron concentrations (10–30 μmol l−1 in
humans) and normal iron stores in macrophages and hep-
atocytes that help buffer transient disturbances in the supply
or demand for iron. Either genetic or environmental causes,
and often both, can contribute to iron disorders.
Iron-deficiency disorders
Iron deficiency is a major public health problem and the most
common nutritional cause of anemia. Very rarely, iron defi-
ciency is genetic and is manifested as hypoferremia, hypo-
ferritinemia, and iron-restricted microcytic anemia refrac-
tory to oral iron administration. Recent studies have shown
that homozygous or compound mutations in a gene encod-
ing TMPRSS6 (Transmembrane protease, serine 6, also called
matriptase 2) are responsible for most cases of iron-refractory
iron-deficiency anemia. TMPRSS6 negatively regulates hep-
cidin, by cleaving membrane hemojuvelin.
Most genetic disorders of iron metabolism result in iron
overload rather than iron deficiency. They are attributable to
mutations that affect the regulation of intestinal iron absorp-
tion and body iron distribution.
Iron overload disorders
It is now clear that there are many genetic iron overload dis-
orders, affecting distinct components of the iron-regulatory
circuitry. They can be generally classified as hemochro-
matosis disorders (iron deposition in parenchymal cells) or
siderosis disorders (deposition of iron in reticuloendothelial
macrophages). The known disorders and their genetic causes
are listed in Table 13.1. Over the last few years, it has become
clear that the final common mechanism for most forms of
hereditary hemochromatosis is relative or absolute hepcidin
deficiency or, rarely, resistance to hepcidin. The lack of hep-
cidin or the loss of its effect on ferroportin results in exces-
sive or unrestrained absorption of dietary iron, as well as the
depletion of the macrophage iron storage compartment. The
genetically defective proteins in hereditary iron overload dis-
ease are implicated as regulators of hepcidin (HFE, TFR2,
hemojuvelin), targets of hepcidin (ferroportin), or molecules
that convey iron to the iron-sensing mechanism in hepato-
cytes (transferrin).
HFE-associated hemochromatosis
The most common form of hemochromatosis was classi-
cally described as the triad of cirrhosis, diabetes, and skin
melanosis (“bronze diabetes”), but fortunately such advanced
disease is now rare. It is a late-onset disorder, inherited in
an autosomal recessive pattern and characterized by iron
deposition in parenchymal cells of the liver, pancreas, and
The molecular basis of iron metabolism 167
heart. Macrophages of the reticuloendothelial system are iron
depleted or at least relatively spared. This disorder results
from a small but chronic increase in intestinal iron absorp-
tion, averaging about two- to threefold the normal level.
Over time, the presence of iron causes damage by promot-
ing the formation of toxic oxygen radicals, which attack
cellular structures and thereby cause reactive fibrosis. The
earliest manifestation of HFE-associated hemochromatosis
is increased transferrin saturation, often approaching 100%
before tissue iron deposition is noted. The treatment for
hemochromatosis is phlebotomy, and this has been used
effectively for more than half a century. Initially, blood is
removed frequently to rapidly decrease storage iron. Later,
iron balance is maintained by periodic phlebotomy, titrated
to meet the needs of the individual patient. This treat-
ment apparently produces no significant morbidity and has
been shown to normalize the life expectancy of affected
patients.
Hemochromatosis has been recognized as an inborn error
of iron metabolism since the 1930s. In 1976 a French physi-
cian, Marcel Simon, made the important observation that the
genetic predisposition to hemochromatosis was linked to the
human major histocompatibility complex on chromosome
6p, and was most frequently associated with an HLA-A3 hap-
lotype. This insight laid the groundwork for the discovery of
causative mutations in the HFE gene 20 years later. It is now
known that most patients with classical hemochromatosis are
homozygous for a unique mutation (cysteine 282 to tyrosine,
or C282Y) in HFE, perhaps originating in a northern Euro-
pean ancestor.
HFE is an atypical HLA class I molecule, similar to its chro-
mosomal neighbors. Although most members of this family
are involved in immune regulation, HFE has no known func-
tion in the immune system. It interacts with TFR1 on the cell
surface, where its binding site overlaps with that of transfer-
rin. Recent studies indicate that the binding of transferrin to
TFR1 displaces HFE, making it active as a stimulus for hep-
cidin synthesis.
The C282Y mutation is highly prevalent. In typical popula-
tions of northern European descent, the carrier frequency has
been estimated to be between about 1 in 8 and 1 in 10. This
indicates that about 1 in 200 individuals are homozygous,
and at risk of iron loading. However, not all C282Y homozy-
gotes will develop clinical hemochromatosis. There is a wide
range in iron loading and its complications. Some individ-
uals will have severe manifestations by the third decade
of life, whereas others may never have signs or symptoms
of hemochromatosis. This variability is probably explained
by gender (iron loss through menstruation), other genetic
factors (modifying genes), and environmental factors (e.g.
alcohol intake, dietary iron consumption). Although several
modifying genes have been proposed, their identification and
characterization are still incomplete.
168 Molecular Hematology

Table 13.1 Iron overload disorders

Chromosomal
location of Gene (types of
Disorder defective genes mutations) Phenotype Hepcidin

Hemochromatosis disorders
HFE-associated 6p, near the HLA HFE (missense and Iron accumulation in the parenchymal Low or
hemochromatosis complex splicing mutations; cells of the liver, heart, pancreas; inappropriately
(also called type 1 C282Y is by far the elevated transferrin saturation; relative normal for iron
hemochromatosis) most important) paucity of iron in macrophages. overload
Clinical manifestations include liver
fibrosis, cirrhosis, markedly increased
incidence of hepatocellular carcinoma,
cardiomyopathy, and diabetes
Juvenile 1q Hemojuvelin hepcidin Similar to HFE-associated Very low to
hemochromatosis 19q (all known mutations hemochromatosis, but greatly absent
(also called type 2 prevent production accelerated, leading to severe cardiac
hemochromatosis) of any hepcidin and endocrine complications in the
protein) second decade of life
TFR2-associated 7q TFR2 (missense and Similar to HFE-associated Low
hemochromatosis nonsense mutations) hemochromatosis
(also called type 3
hemochromatosis)
Autosomal dominant 2q Ferroportin (rare Iron accumulation in parenchymal cells ?
hemochromatosis missense mutations
(also called type 4 causing resistance to
hemochromatosis) hepcidin). Most
other missense
ferroportin
mutations cause loss
of function and
siderosis
Siderosis disorders
Autosomal dominant 2q Ferroportin (only Macrophage-predominant iron loading; Normal or high
siderosis (also called missense mutations) in some patients parenchymal iron (few patients
type 4 loading can occur later. Some patients studied)
hemochromatosis) have anemia early in their course
(particularly women). Ferritin levels are
markedly elevated, but serum
transferrin saturation generally is not
African siderosis ? Unknown Similar to autosomal dominant siderosis, ?
but ferroportin mutations have not
been reported. Thought to be a
combination of genetic and
environmental factors
Disorders of iron balance
Atransferrinemia 3q Transferrin (missense Deficiency in serum transferrin leading to Low
mutations) tissue iron overload and severe
iron-deficiency anemia
Aceruloplasminemia 3q Ceruloplasmin Deposition of iron in the brain, liver, and ?
(missense and null pancreas. Late-onset
mutations) neurodegenerative disease, dementia.
and diabetes

In addition to C282Y, other mutations and polymorphisms
have been identified in the HFE gene. The most common of
these is a histidine-to-aspartic acid substitution at amino acid
63 of the protein (H63D). This polymorphism is found in
about one-fifth of the world’s population. Although it may
occasionally be associated with iron overload, particularly
when it is found in individuals heterozygous for the C282Y
mutation, its clinical significance is probably minor, though
clinical laboratories test for it. Most other HFE mutations are
quite rare and are not identified by routine screening tests.
In the past, the proportion of at-risk C282Y homozygous
individuals who develop clinical hemochromatosis was esti-
mated to be about 20–40%. However, Beutler and colleagues
published the results of a large questionnaire-based study
of patients seen by a health maintenance organization, in
which they concluded that less than 1% of C282Y homozy-
gotes would have severe disease. There is no consensus yet
on laboratory parameters that identify C282Y homozygotes
at risk for clinically significant organ damage, but recent data
indicate that few if any patients with serum ferritin levels
below 1000 ng ml−1 develop cirrhosis, the main complication
of hemochromatosis.
Juvenile hemochromatosis
Juvenile hemochromatosis is similar to HFE-associated
hemochromatosis, but very rare, and characterized by earlier
onset of iron loading and its complications. The target organs
are the same as those affected in HFE hemochromatosis,
but cardiac and endocrine dysfunction are more severe, and
untreated patients typically die from cardiomyopathy by age
30 years. Liver cirrhosis and failure are uncommon. There are
several possible explanations for this pattern. First, experi-
ence with patients who develop siderosis from chronic trans-
fusion therapy suggests that rapid iron loading is especially
toxic for the heart and endocrine tissues. Secondly, patho-
logical iron deposition in the adolescent years may be partic-
ularly bad for young hearts which are growing to meet the
demands of a larger body mass; this is analogous to the prob-
lems noted with doxorubicin cardiotoxicity in this age group.
Furthermore, endocrine problems are probably more appar-
ent in adolescents because they fail to go through normal
pubertal development. Juvenile hemochromatosis is a partic-
ularly lethal disorder, but it can be effectively treated by phle-
botomy.
Studies of families with juvenile hemochromatosis have
shown that there are at least two genetic loci responsible.
Some individuals are homozygous for ablative mutations in
the hepcidin gene on human chromosome 19q. However,
most juvenile hemochromatosis families have homozygous
or compound heterozygous mutations in hemojuvelin (also
called HFE2, or repulsive guidance molecule RgmC) on
The molecular basis of iron metabolism 169
chromosome 1q. Hemojuvelin is a GPI-linked membrane
protein that functions as an essential regulator of hepcidin.
Disruption of both copies of hemojuvelin has been shown to
cause profound hepcidin deficiency. It thus appears that the
greater severity of the juvenile form of the disease is due to
the lower levels of hepcidin compared with the adult form of
hemochromatosis.
TFR2-associated hemochromatosis
A third form of autosomal recessive hemochromatosis
(type 3) has been reported to be indistinguishable from HFE-
associated hemochromatosis in its clinical manifestations,
but patients have no mutations in the HFE gene. Instead, their
disease is caused by mutations in the TFR2 gene on chro-
mosome 7q. TFR2 is a protein that is highly homologous to
the transferrin receptor (also known as TFR1), and highly
expressed by hepatocytes and hematopoietic cells. TFR2 is
capable of binding transferrin and transporting it into the
cell, but it does so at much lower efficiency than does trans-
ferrin receptor. TFR2 is stabilized by interaction with iron–
transferrin and is also able to bind HFE. Patients and mice
with homozygous TFR2 mutations have low hepcidin expres-
sion, implicating this protein in the regulation of hepcidin
synthesis, possibly as a sensor of transferrin saturation. Some
patients have been shown to have nonsense mutations, indi-
cating that TFR2 is not an essential protein for survival. Type
3 hemochromatosis is effectively treated by phlebotomy.
Two siblings with combined HFE and TFR2 mutations
were reported to have the severe juvenile form of the disease,
suggesting that HFE and TFR2 have additive effects on hep-
cidin expression.
Ferroportin-associated hemochromatosis
Most reported ferroportin mutations cause the loss of ferro-
portin function and the clinical picture of macrophage iron
overload (siderosis) with few if any clinical consequences
(see later). A rare form of hereditary hemochromatosis is
an autosomal dominant disorder (type 4 hemochromato-
sis) due to ferroportin mutations that confer resistance to
hepcidin-induced internalization (gain of function). Such a
lesion would be expected to act dominantly and mimic hep-
cidin deficiency. Its most severe form involves serine or tyro-
sine substitutions in C326 in the hepcidin-binding site of fer-
roportin. Affected patients presented with early onset of liver
disease and arthritis.
Other forms of hemochromatosis
Patients have been reported with relatively mild hemochro-
matosis who have no apparent mutations in the HFE, TFR2,
170 Molecular Hematology
hemojuvelin, or hepcidin genes. Recent studies implicated
dominantly acting mutations in BMP6 propeptide, which
may interfere with BMP6 secretion.
Autosomal dominant siderosis
Autosomal dominant siderosis, formerly also called auto-
somal dominant (or type 4) hemochromatosis, has a dis-
tinct clinical picture. Patients with this disorder may have
iron-deficiency anemia early in life, but later present with
(sometimes massively) increased serum ferritin concentra-
tion and macrophage iron accumulation. Some may even-
tually develop parenchymal iron deposition in addition to
macrophage iron loading, probably because their specific
mutations may also cause resistance to hepcidin.
This disorder has a very interesting pathogenesis. It is
due to missense mutations in ferroportin, the cellular iron
exporter. This seems paradoxical at first, because the muta-
tions alter ferroportin function, and ferroportin acts as the
basolateral enterocyte transporter involved in intestinal iron
absorption (see Figure 13.1). However, it is important to con-
sider that ferroportin also plays a major role in macrophage
iron release. The lack of nonsense mutations in this disor-
der, and evidence from cellular and animal models of this
disease, suggest that the mutated ferroportin exerts a dom-
inant negative effect on the normal version of the molecule.
The resulting loss of functional ferroportin is severe enough
to impair macrophage iron release, resulting in accumula-
tion of iron in macrophages and a decrease in the amount
of plasma iron available to developing erythroid precursors.
Iron-restricted erythropoiesis probably signals for a compen-
satory increase in intestinal iron absorption, overcoming the
genetic deficiency of ferroportin in the enterocytes. It appears
that iron loading of macrophages has few if any clinical
consequences.
African siderosis
The pathology of African siderosis is strikingly similar to
that of autosomal dominant siderosis due to ferroportin gene
mutations, although the genetic basis of African siderosis
has not been described. Once called Bantu siderosis because
of the population affected, this disorder was originally
attributed to excessive dietary iron intake. It is common in
sub-Saharan Africans, many of whom drink a traditional
alcoholic beverage brewed in non-galvanized steel drums.
The iron content of the brew is substantial, resulting in
massive iron ingestion. However, the observations that not
all drinkers develop iron overload and that some individuals
develop similar iron overload without drinking the beverage
support the notion that there is a genetic component to
this disorder. The ferroportin mutation Q248H is relatively
common (2–13%) in southern Africa and may cause mild
resistance to hepcidin and tendency to iron overload. It is
not yet known whether European and American individuals
of African descent are also more susceptible to iron overload
as a result of the same iron-loading gene.
Abnormal iron distribution
There are two well-characterized (but very rare) disorders
due to mutations in plasma proteins important in iron
metabolism. These mutations do not directly affect intestinal
iron absorption. Rather, they perturb tissue iron distribution.
Atransferrinemia
Atransferrinemia is a severe deficiency of the plasma iron-
binding protein transferrin, due to mutations that truncate
or alter the coding sequence of the transferrin gene. As a
result, erythroid precursors are iron starved and severe ane-
mia results. Paradoxically, all non-hematopoietic tissues are
iron loaded, probably because intestinal iron absorption is
enhanced to try to provide more iron to erythroid precursors,
and because NTBI is avidly taken up by many parenchymal
cell types. Mice and humans with this disorder have low hep-
cidin, indicating that iron–transferrin is an important regu-
lator of hepcidin synthesis. This disorder can be treated by
transfusion of packed red blood cells or, more appropriately,
by infusion of human transferrin.
Aceruloplasminemia
Aceruloplasminemia is a deficiency or absence of plasma
ceruloplasmin. Ceruloplasmin was once thought to be a
plasma copper carrier, but it is now clear that its primary
role is as a ferroxidase, aiding in the release of iron from
macrophages, hepatocytes, and cells of the central nervous
system. Patients with this disorder are generally well early in
life, but gradually develop tissue iron deposition in the liver,
pancreas, and brain. They typically present in middle age
with retinal degeneration, dementia, hepatic iron deposition,
and diabetes. Treatment with deferoxamine is ineffectual;
treatment with normal plasma may provide some benefit.
Conclusions
Iron disorders are among the most common of human
afflictions. They invariably result from abnormalities of iron
balance, most often due to defects in the iron-regulatory
hormone hepcidin, its receptor/iron channel ferroportin, or
their interaction. Iron overload is underdiagnosed because
it produces signs and symptoms that are common in adult

populations. However, iron overload disorders are usually
easy to treat and clinicians should be vigilant in considering
them.
Further reading
General
Coffey, R. and Ganz, T. (2017). Iron homeostasis: an anthropocentric
perspective. J. Biol. Chem. 292: 12727–12734.
Hentze, M.W., Muckenthaler, M.U., Galy, B., and Camaschella, C.
(2010). Two to tango: regulation of mammalian iron metabolism. Cell
142: 24–38.
Regulation of iron homeostasis
Babitt, J.L., Huang, F.W., Wrighting, D.M. et al. (2006). Bone morpho-
genetic protein signaling by hemojuvelin regulates hepcidin expres-
sion. Nat. Genet. 38: 531–539.
Canali, S., Zumbrennen-Bullough, K.B., Core, A.B. et al. (2017).
Endothelial cells produce bone morphogenetic protein 6 required for
iron homeostasis in mice. Blood 129: 405–414.
Nemeth, E., Tuttle, M.S., Powelson, J. et al. (2004). Hepcidin regulates
cellular iron efflux by binding to ferroportin and inducing its inter-
nalization. Science 306: 2090–2093.
The molecular basis of iron metabolism 171
Nicolas, G., Chauvet, C., Viatte, L. et al. (2002). The gene encoding the
iron regulatory peptide hepcidin is regulated by anemia, hypoxia, and
inflammation. J. Clin. Invest. 100: 1037–1044.
Iron overload disorders
Bardou-Jacquet, E. and Brissot, P. (2014). Diagnostic evaluation of
hereditary hemochromatosis (HFE and non-HFE). Hematol. Oncol.
Clin. North Am. 28: 625–635.
Beutler, E. (2006). Hemochromatosis: genetics and pathophysiology.
Annu. Rev. Med. 57: 331–347.
Kanwar, P. and Kowdley, K.V. (2013). Diagnosis and treatment of heredi-
tary hemochromatosis: an update. Expert Rev. Gastroenterol. Hepatol.
7: 517–530.
Pietrangelo, A. (2007). Hemochromatosis: an endocrine liver disease.
Hepatology 46: 1291–1301.
Abnormal iron distribution
Bartnikas, T.B. (2012). Known and potential roles of transferrin in iron
biology. Biometals 25: 677–686.
Xu, X., Pin, S., Gathinji, M. et al. (2004). Aceruloplasminemia: an inher-
ited neurodegenerative disease with impairment of iron homeostasis.
Ann. N.Y. Acad. Sci. 1012: 299–305.
Chapter 14 Hemoglobinopathies due
to structural mutations
D. Mark Layton & Steven Okoli
Center for Hematology, Imperial College London, London, UK
Introduction, 173
Normal hemoglobin structure and function, 173
Sickle cell anemia, 175
Sickle/β thalassemia, 184
Hemoglobin C disease, 185
Hemoglobin SC disease, 185
Other sickle hemoglobinopathies, 186
Introduction
This chapter discusses those hemoglobin disorders that are
caused by mutations in the exon (i.e. coding) portion of the
α or β globin genes. Alterations of globin gene expression
(thalassemias) are reviewed in Chapter 1. We confine our-
selves mainly to the structural variants that are responsible
for clinically significant hemoglobinopathies.
Normal hemoglobin structure and
function
The adult major hemoglobin molecule, HbA, is a tetramer
formed by four polypeptide chains: two α chains of 141
amino acids each and two β chains of 146 amino acids each.
Each globin chain harbors a prosthetic group (heme) formed
by protoporphyrin IX in a complex with a single iron atom
(Figure 14.1). The heme, embedded in globin, is surrounded
by a hydrophobic niche which favors maintenance of iron in
the ferrous state. The heme pocket is large enough for oxygen
to penetrate, but larger ligands capable of binding to the iron,
such as carbon monoxide, have difficulty gaining access.
Oxygen transport to tissues, the principal function of
hemoglobin, depends on blood flow, which in turn is
governed by cardiac output and microcirculatory resistance
and distribution, and by hemoglobin concentration. Oxy-
gen extraction by the tissues depends on the shape of the
oxygen dissociation curve of hemoglobin and on tissue Po2 .
Molecular Hematology, 4th Edition. Edited by Drew Provan and John G. Gribben.
© 2020 John Wiley & Sons Ltd. Published 2020 by John Wiley & Sons Ltd.
Homozygous HbE and Hb E/β thalassemia, 187
Unstable hemoglobins, 188
High oxygen affinity hemoglobins, 188
Low oxygen affinity hemoglobins, 189
M hemoglobins, 189
Conclusions, 190
Further reading, 191
The shape of the oxygen equilibrium curve of hemoglobin
is sigmoid. This shape is determined by the extent of
cooperativity, a phenomenon whereby the hemoglobin
molecule modulates its affinity for oxygen when partially
saturated. The first portion of the curve has a gentle slope,
reflecting hemoglobin’s low affinity for oxygen at initiation
of the loading process. In other words, when hemoglobin is
totally deoxygenated, it has a rather poor avidity for oxygen
(Figure 14.2). As loading proceeds and hemoglobin binds
more oxygen molecules, the slope of the curve steepens,
indicating that the affinity for oxygen has increased. This
occurs after two molecules of oxygen have bound to the
heme of deoxyhemoglobin tetramers, a property which
helps hemoglobin to rapidly become fully oxygenated. If red
cells are exposed to sufficient oxygen to saturate only half of
the hemes available, most molecules will either not be oxy-
genated at all or will be entirely oxygenated, with a very small
component of partially oxygenated molecules. The sigmoid
shape of the oxygen dissociation curve allows hemoglobin
to release oxygen efficiently. Abnormal hemoglobins such as
Hb Bart’s (γ4 ) which lack cooperativity are unable to deliver
oxygen to the tissues efficiently.
At the molecular level, cooperativity is accounted for by
the fact that hemoglobin exists stably in only two structural
forms, the R (relaxed) and T (tense) states, which facilitate
the switch between the oxy and deoxy states. Transition from
the deoxy (T) to the oxy (R) state occurs when a molecule
of hemoglobin has bound two or three molecules of oxygen.
While this concept has been challenged, postulated interme-
diate stable states have not been generally recognized.
The heme-triggering mechanism for conformational
change has been resolved. The iron in deoxyhemoglobin is
173
174 Molecular Hematology

Hemoglobin tetramer

β2 β1
F'

A F A' F
E'
β2 β1
A
H'
H H'
B E'
E G G' H H' E
C
E'
GH C'
B
G
C' G' C
D' B D
F' E
H'
H E
H'
H'
E' F A
A' E' F'
A
A' α1 α2

α2 α1

(a) Front view (b) Side view

Fig. 14.1 Hemoglobin molecule. (a) The front view depicts the hemoglobin tetramer and the three axes of symmetry. The vertical line (marked
by a solid ellipse) tracks the true twofold axis of symmetry (if you look down this axis you will see first the central cavity constituted by the β
chains). The dashed ellipses and dashed lines mark the two pseudoaxes of symmetry, since the symmetry is only approximate. Only the 21 carbons
are shown, with none of the side chains. Numbers in bold type depict the residues in direct contact between the α1 and β2 chains. The α1 β2
dimer (green) never dissociates and interacts with the α2 β1 dimer (light shading) to change the conformation from T to R. Alterations in this area
can produce high- or low-affinity hemoglobins. (b) The side view of the tetramer depicts the stable dimer, the one that does not move or dissociate.

Horse methemoglobin Horse deoxyhemoglobin

α2 α2 β1
β1

β2 α1 α1
β2

Fig. 14.2 The allosteric transition of hemoglobin: R → T. Right: Deoxygenated (T, tense) conformation of hemoglobin. Notice that the center
cavity (space between the two β chains, indicated in green) in the hemoglobin molecule is larger than in the oxygenated tetramer (left). This space
is occupied by 2,3-diphosphoglycerate (2,3-DPG) when the tetramer is in the blue cell. The allosteric effector, 2,3-DPG, is in a little over equimolar
concentration with the tetramer. The T form has low affinity for oxygen. Left: Oxygenated (R, relaxed) conformational state of hemoglobin. The
central cavity has been reduced in size by the movement of the β chains toward each other. This tetramer cannot bind 2,3-DPG. The R form has
high affinity for oxygen. The change in conformation forms and breaks bonds between the β chains (green) and the α chains (light shading) on
each side of the central cavity. The title says “methemoglobin” but this hemoglobin form is identical to oxyhemoglobin.
 Hemoglobinopathies due to structural mutations 175

slightly out of the plane of the heme (domed configuration)

because the pyrrole rings are also slightly pyramidal. When
the ligand binds the sixth coordinating position of the iron,
significant steric stresses are introduced, and to relieve this
strain the distal histidine moves 8° to become perpendicular
to the heme, significantly decreasing the doming of the
iron (the angle between iron and the heme decreases to 4°).
There is also the displacement of FG5 in the same direction
of the histidine F8 (a conserved residue to which heme iron
is covalently linked in all hemoglobins). The configuration
around the heme has now changed to the oxygenated
(R state), and a chain of events takes place involving the
critical interactions that change the conformation of the
hemoglobin tetramer.
Several heterotropic ligands within the red cell shift
the oxygen dissociation curve to the right, toward lower
oxygen affinity. These include CO2 , hydrogen ions, and
2,3-diphosphoglycerate (2,3-DPG). Hemoglobin binds CO2
while it is offloading oxygen and releases CO2 when it binds
oxygen, serving to dissipate the increase in concentration of
CO2 in the tissues and conveniently delivering this metabolic
Fig. 14.3 Sickled cell. This is an HbS homozygous (SS) red cell that
end product to the alveoli of the lungs. It accomplishes this
has been deoxygenated. Notice the digitations stemming in all
particular task with ease, because CO2 inhibits hemoglobin’s directions. These digitations are the product of the presence of
oxygen-carrying capacity by decreasing the oxygen affinity of fascicles of fibers, which are the polymerized form of deoxy HbS.
the molecule.
Hemoglobin binds hydrogen ions more efficiently in a low-
pH environment. The Bohr effect describes the changes in proposed that sickling reflected an abnormality of the
oxygen affinity secondary to pH changes within a certain hemoglobin molecule, based on the observation of Irving
range: the lower the pH, the lower the oxygen affinity or the Sherman, then a medical student, that under the polarizing
higher the P50 . This means that an increased concentration of microscope sickle cells, induced by deoxygenation, exhibited
protons favors a low-affinity state in hemoglobin. The phys- birefringence. Birefringence signified to Pauling that some
iological advantage this confers is twofold: in the tissues the form of molecular alignment or orientation existed in these
lower pH consequent on higher CO2 tension favors oxygen cells, and since the predominant molecule in the red cell
unloading, whereas in the lungs there is an increase in pH is hemoglobin, it had to be this that was involved in the
with CO2 release which enhances oxygen uptake. pathology. Electrophoretic studies, which demonstrated
the altered mobility of sickle hemoglobin (HbS), confirmed
this interpretation and the concept of molecular disease
Sickle cell anemia was born.
The chemical structure of HbS was defined by Vernon
Ingram, who had developed a technique for probing the pri-
Genetics
mary sequence of a protein based on its fingerprint after
The genetic basis of sickle cell anemia is central to the history cleavage by trypsin. This revealed that the structure of HbS
of medical genetics. Pre-dated for generations by vernacular differed from normal hemoglobin by only one tryptic pep-
descriptions among West African tribes, the first account tide, subsequently found to have a single amino acid change
of sickle cell disease (SCD) in Western medical literature is at position 6 of the β chain, where a polar glutamic acid
attributed to James Herrick, the Chicago physician, who in residue was replaced by a non-polar valine. At the DNA
1910 observed sickle-shaped red cells (Figure 14.3) in the level this corresponds to an A → T transversion at codon 6
blood of a dental student from Grenada who suffered from (GAG → GTG) of the β chain.
chronic hemolytic anemia. In 1947 James Neel concluded The globin genes were among the first in the human
that with respect to the mode of inheritance, sickle cell genome to be cloned. The β and β-like globin genes are
anemia was a homozygous state and sickle cell trait (the located on the short arm of chromosome 11 at 11p15.5 and
asymptomatic carrier state) the heterozygous state for a the α and α-like genes at chromosome 16p13.3. Utilizing
genetic character that had yet to be defined. Linus Pauling polymorphic restriction sites distributed across the β globin
176 Molecular Hematology

gene cluster, it has been established that in Africa the βS gene
is associated with four distinct chromosomal haplotypes,
Benin, Bantu or Central African Republic, Senegal, and
Cameroon, named after the geographical areas in which the
haplotypes are found (Figure 14.4). The geographical reach
of the sickle mutation spread further with the discovery of
a different haplotype linked to the βS gene in the eastern
oasis of Saudi Arabia and among tribal population groups
in central India. It was postulated this Indo-European sickle
mutation originated in the Harappa people of the Indus
Valley, and then dispersed, probably during the Sassanian
Empire, to regions in the Middle East (now eastern Saudi
Arabia, Bahrain, Kuwait, and Oman). Based on these haplo-
types a multicentric model has been proposed in which the
sickle gene arose independently around the world on at least
five occasions. This has recently been challenged by analysis
of whole-genome sequencing data which suggest a single
origin for the sickle mutation approximately 7300 years
ago. The present-day gene frequency, which equates to
carrier rates that reach 25% in sub-Saharan Africa, reflects
a heterozygote advantage that outweighs the reduced
fitness in the homozygous state (balanced polymorphism).
This selection pressure in favor of HbS heterozygotes is
attributable to partial protection against the life-threatening
complications of Plasmodium falciparum malaria, with a
consequent reduction in mortality. The exact mechanism(s)
of protection has yet to be fully elucidated.
The sickle gene has also spread through gene flow; for
example, the βS allele linked to the Benin haplotype has
spread to North Africa, Sicily, Greece, Turkey, most of the
Arab world, and the Americas, through the vicissitudes of
war (e.g. Sudanese troops in Sicily during the Arab conquest),
commerce, and the horrors of the Atlantic slave trade. It
is estimated that worldwide approximately 300 000 children
with sickle cell anemia are born annually, and that the figure
could rise to 400 000 by 2050.
Despite its simple Mendelian inheritance, the clinical
course of SCD is extremely heterogeneous. The basis for
the striking variability in the clinical course and severity
of sickle cell anemia has been the subject of intense study.
Both genetic and non-genetic factors are important, though
their relative contribution to disease severity is still not

Senegal Arab-
India
2
Benin

3 Bantu

60 50 40 30 20 10 0 Kb

ε Gγ Aγ ψβ δ β
5ʹ 3ʹ
Fig. 14.4 The β-globin gene cluster
haplotypes linked to βS in Africa and the
Hinc Xmn Hind Pvu Hinc Hinf Ava Hpa Bam HI Middle East/Pakistan/India. At the top of the
II I III II II I II I figure are shown the geographical distributions
Hgi
AI Gγ corresponding to the haplotypes described
underneath. A haplotype is a particular array of
SS Benin – – – – + – + – ++ – + polymorphic sites (i.e. sites that vary among
SS Bantu – – + – + – – – ++ ++
individuals), defined here by the capacity of
SS Senegal – + + – + + + + ++ ++ endonuclease enzymes to recognize the short
SS Arab-India + + + – + + + + –
sequence and cut the DNA.

well defined. A pilot study of identical twin pairs revealed
concordance in growth, hematological parameters, and
certain complications, such as priapism and splenomegaly,
but variance with respect to other clinical manifestations.
Socioeconomic status and access to medical care as well as
the prevalence of infections that carry increased morbidity
and mortality in SCD are important variables. Increased fetal
hemoglobin (HbF) production is a major ameliorating factor
in SCD. The switch from predominant expression of HbF to
adult hemoglobin (HbA) after birth coincides with the stage
at which children with SCD first develop symptoms, as noted
by Janet Watson in 1948. Data from the large prospective
Cooperative Study of Sickle Cell Disease (CSSCD) showed
higher levels of fetal hemoglobin were independently associ-
ated with improved survival and reduced complication rates.
This beneficial effect was also observed in the Jamaican
SCD cohort. Patients with a steady-state Hb F level of 8–9%
or more have a more favorable course. A further striking
example is found in compound heterozygotes for the βS
allele and deletional forms of hereditary persistence of
fetal hemoglobin, whose red cells contain 20–30% HbF.
Despite most of their remaining hemoglobin being HbS,
they are largely asymptomatic. The protective effect of HbF
is attributable to differences in the amino acid sequence of
γ-globin and β-globin that affect the biophysical properties
of the corresponding hemoglobin tetramer. The γ-globin
chain has a glutamine rather than a threonine residue at
position 87, which weakens hydrophobic interactions impor-
tant for stabilization of the HbS polymer. Consequently, HbF
as well as the hybrid tetramer HbFS (α2 γβS ) have a low
probability of copolymerization with HbS.
To date, three quantitative trait loci associated with
variation in HbF level in patients with SCD have been
identified. The first of these, a single-nucleotide polymor-
phism (SNP) in the promoter of HBG2, the gene which
encodes G γ-globin, identified by the restriction enzyme
Xmn I, is linked to the Senegal and Saudi Arabian-Indian
βS haplotypes associated with higher HbF levels. Only part
of the difference in HbF levels observed in SCD is accounted
for by genetic variation at the β globin locus. Genome-wide
association studies have identified two further quantitative
trait loci, the intergenic region between HBS1L and MYB
on chromosome 6q23 and BCL11A on chromosome 2p16,
which account for variability in HbF in normal non-anemic
populations and up to 50% of that seen in patients with
sickle cell anemia. The HBS1L-MYB intergenic region has
been shown to regulate erythropoiesis and HbF production,
though the functional variant at this locus has yet to be
defined. BCL11A encodes a zinc finger protein that represses
transcription of the HBG1 and HBG2 genes. SNPs associated
with HbF level co-locate with an erythroid-specific enhancer
in intron 2 of BCL11A that is essential for gene expression.
The mechanisms that underlie these quantitative traits have
Hemoglobinopathies due to structural mutations 177
important implications for the development of strategies for
HbF induction to treat β hemoglobinopathies.
Alpha thalassemia, an extremely common genetic
variant in populations where the βS allele is prevalent,
exerts a beneficial effect through lowering intracellular
HbS concentration, a key determinant of the rate of poly-
merization. Individuals who co-inherit sickle cell anemia
and α-thalassemia have higher hemoglobin levels and a
lower risk of complications linked to hemolysis, including
stroke, priapism, leg ulcers, nephropathy, and cholelithiasis.
Unfortunately, the favorable impact of α-thalassemia is
offset by an increased risk of vaso-occlusive crises and
avascular necrosis, possibly due to the higher hematocrit.
There are conflicting data on the effect of α-thalassemia
on mortality in SCD, though several studies describe an
increase in prevalence of α-thalassemia with age, implying
an association with improved survival.
It is inevitable genetic factors beyond effects on
hemoglobin synthesis contribute to the clinical hetero-
geneity of SCD given its complex pathophysiology. A
substantial number of genetic association studies have been
conducted, but with few exceptions the discovery of genetic
modifiers has proved challenging (reviewed in Lettre 2012).
Convincing associations have been shown for polymor-
phism of the UGT1A1 gene and risk of cholelithiasis, as
well as the increased risk and progression of nephropathy
among individuals with SCD who are homozygous or
compound heterozygous for the G1 or G2 variant alleles of
the APOL1 gene, found at high frequency in some African
populations. Causality is less clear in other cases. An SNP
located upstream of COMMD7, a gene expressed in the lung
implicated in NF-ΚB signaling, and HMOX1 gene promoter
polymorphisms that alter heme oxygenase activity have
been associated with acute chest syndrome. A genome-wide
association study in a large group of children with SCD
enrolled in the CSSCD and Silent Infarct Transfusion
(SIT) trial revealed preliminary evidence of an association
between an SNP at a locus on chromosome 4q27 previously
associated with autoinflammatory disorders and acute,
severe vaso-occlusive pain. Recently, differences in blood
mononuclear transcriptomic profile have been found to
correlate with clinical severity of SCD and, in conjunction
with key clinical markers, to be superior in prediction of
mortality than clinical criteria alone. This represents a step
toward the discovery of much-needed biomarkers to aid
disease and treatment stratification.
Clinical features
Sickle cell anemia refers to the homozygous state for the βS
gene, in which the majority of the hemoglobin in the red cells
is HbS. Polymerization of HbS when ambient oxygen ten-
sion is reduced induces sickling (marked changes in red cell
178 Molecular Hematology
shape and deformability with the formation of irreversibly
sickled cells, ISCs), leading to decreased pliability, lowering
of rheological competence, and hemolysis. In addition, there
are myriad pleiotropic effects, including increased adherence
of sickle and other blood cells to endothelium, induction of
red cell dehydration, pro-coagulability, and perturbance of
vasoregulation due to reduced nitric oxide (NO) bioavailabil-
ity. As discussed, the primary pathophysiological events as
well as the pleiotropic effects are subject to modification by
epistatic genes, which may be polymorphic. The combined
impact of all these factors accounts for the great interindi-
vidual variability of the clinical phenotype: some patients are
severely affected, while others have only mild disease, and the
majority span these two extremes.
Anemia is present in all cases, with a mean hemoglobin
concentration of around 8 g dl−1 (higher levels occur in
Indian and some Middle Eastern populations), reticu-
locytosis (typically 5–20%), dense cell fraction – mean
corpuscular hemoglobin concentration (MCHC) >38 g dl−1
or density >1.1009 – of up to 40%, an elevated white cell
count (mean 18 × 109 l−1 ), and commonly thrombocytosis.
Anemia is tolerated relatively well due to the reduced oxygen
affinity of HbS and increased 2,3-DPG concentration within
the red cell. Several hematological variables, including
total hemoglobin, are modulated by the level of HbF and
co-inheritance of α thalassemia (−α/−α genotype more so
than −α/αα), which occurs at a significant frequency among
sickle cell anemia patients (20–50% of cases). Higher levels
of HbF are associated with the Senegal and Arab-Indian
β globin haplotypes. Gender also exerts an influence; for
example, females with sickle cell anemia have a significantly
higher HbF level. In males hemoglobin concentration
increases between 10 and 24 years, contributing to a higher
frequency of vaso-occlusive crises.
The cardinal manifestation of sickle cell anemia is the
painful crisis. This commonly involves multiple sites in the
extremities, joints, spine, abdomen, ribs, and sternum. The
onset of pain can be insidious or rapid, and may affect sev-
eral sites simultaneously or extend over time from one loca-
tion to others. The pathological basis of the painful crisis
is infarction of the bone marrow. Systemic embolization of
bone marrow or fat to the lung, brain, or kidneys may occur,
and is an ominous development that may herald acute chest
syndrome or multiorgan failure. The site first affected in
infancy is often the hands or feet (dactylitis). Necrosis of
the epiphysis may lead to premature fusion with shorten-
ing of the digits. Paradoxically, painful crises are more fre-
quent in patients with greater red cell deformability. Serial
observations indicate that during painful crises a decrease
in low-density cells of greater deformability precedes that of
dense cells. This may be reconciled with a model whereby
cells that are less dense but more adherent to the endothe-
lium bedeck the vessels, facilitating entrapment of dense cells
and ultimately vaso-occlusion. This may be an oversimpli-
fication and other mechanisms, for example reflex shunt-
ing of blood away from the bone marrow and perturbation
of vasoregulation, have been invoked. Precipitating factors
include cold, infection, dehydration, altitude, excessive exer-
cise, emotional stress, sleep apnea, obstruction to the periph-
eral circulation (e.g. tourniquet use), and possibly cocaine
abuse.
The clinical complications of sickle cell anemia are pro-
tean (Figure 14.5) and involve almost every organ system.
The lungs and kidneys are particularly affected. There is
an increase in glomerular filtration rate as well as hypos-
thenuria from early childhood. Papillary necrosis with hema-
turia, renal insufficiency, and nephrotic syndrome may occur
(Figure 14.6). The lungs are the target of both acute and
chronic complications. Acute chest syndrome may be asso-
ciated with infection (viral, bacterial, or atypical organisms)
and fat embolism. The latter is associated with a more severe
clinical course. Repeated episodes of acute chest syndrome
predispose to the development of chronic sickle lung dis-
ease. Cerebral vasculopathy may lead to stroke, silent cerebral
infarction, and neurocognitive defects. From infancy there
is progressive splenic atrophy, against which co-inheritance
of α thalassemia affords protection. The resulting loss of
splenic reticuloendothelial function leads to susceptibility to
infection with capsulated bacteria. Before the advent of new-
born screening and prophylaxis, overwhelming Streptococcus
pneumoniae infection was the leading cause of death in chil-
dren with SCD.
The prevalence of avascular necrosis of the femoral or
humeral head reaches 20% in adulthood and is greater
among patients whose hematocrit is higher, including those
with coexistent thalassemia. Peripheral retinopathy, charac-
terized by microvascular occlusion and choroidal infarcts, is
common in SCD. Abnormal arteriovenous communications
develop at the border of vascular and ischemic retina and are
the precursor of proliferative sickle retinopathy. In patients
with sickle cell anemia, and even heterozygotes, glaucoma
may occur after traumatic hyphemas, since the hypoxic
milieu of the anterior chamber favors sickling of red cells,
which are unable to exit through the canal of Schlemm. As in
other chronic hemolytic anemias, cholelithiasis is common.
The prevalence of gallstones increases with age and is also
correlated directly with the number of TA repeats in the
promoter of the hepatic uridine diphosphate glucuronosyl-
tranferase (UGT1A1) gene, and inversely with the presence
of α-thalassemia. Hepatopathy in sickle cell anemia encom-
passes several entities that include intrahepatic cholestasis
and hepatic sequestration. In the latter, rapid enlargement of
the liver is associated with a precipitous fall in hemoglobin
akin to that in acute splenic sequestration.
Congestive cardiac failure may develop in the absence of
other recognized causes in sickle cell anemia. This implies
 Hemoglobinopathies due to structural mutations 179

Pain
Dactylitis
Long bones
Trunk

Sequestration
Splenic
Hepatic
Chest syndrome
Mesenteric syndrome

Infection
Pneumococcal
Parvovirus
Salmonella

Priapism

Upper airway obstruction

Stroke
Subarachnoid hemorrhage
Retinopathy

Gallstones
Avascular necrosis

Hyposthenuria

Delayed growth and development

Leg ulcers

Fig. 14.5 Age dependency of complications in Chronic renal failure

sickle cell anemia. Chronic sickle lung
Source: Davies S.C., Oni L. (1997). Management of
patients with sickle cell disease. BMJ, 315:
656–660. Reproduced with permission of British 0 5 10 15 20 25 30
Medical Journal. Age (years)

that the myocardium may be the target of vaso-occlusive

Fig. 14.6 Renal medulla of transgenic mice expressing βS and
βS-Antilles . Note the extensive obstructive aggregation of sickle red
cells in the renal medullary vasculature.
damage, although overt myocardial infarction is not com-
mon. The infrequency of vaso-occlusion in the heart may
be the consequence of the very high perfusion pressure of
coronary vessels, as well as the squeezing effect of ventricular
contraction. Pulmonary arterial hypertension has emerged as
a common and serious complication. Estimated pulmonary
artery pressure is increased in up to 30% of sickle cell anemia
patients based on echocardiographic findings. Multiple stud-
ies have shown this to be associated with an increased risk of
death, though the magnitude of risk varies between studies.
In a study conducted at the National Institutes of Health with
a median follow-up of four years, the mortality in patients
with a tricuspid regurgitant velocity ≥2.5 m s−1 was 36%.
In sickle cell anemia, pregnancy poses a significant risk to
the mother and fetus. The role of prophylactic transfusion
during pregnancy remains controversial and many centers
adopt a selective approach, allied to close monitoring of
180 Molecular Hematology

mother and fetus in the context of a high-risk obstetric
program, with generally favorable outcomes.
Pathophysiology
There are five major contributors to the pathophysiology of
sickle cell anemia.
Polymerization of HbS
The substitution of valine for glutamic acid at β6 sub-
verts an essential property of hemoglobin, that it should
maintain solubility at high concentration within the red
cell. The initial step in the polymerization of HbS, known
as nucleation, involves the assembly of small aggregates
of deoxy HbS molecules. Once nucleation has occurred,
propagation of the polymer ensues at an exponential rate.
The self-association of deoxy HbS molecules to form a
nucleus is a stochastic process, hence the reaction has a
delay time. This is shortened by an increase in intracellular
hemoglobin concentration, lowering the pH, and an increase
in temperature, which may be one reason dehydration and
infection trigger vaso-occlusive events in sickle cell disor-
ders. The final product of polymerization is a rope-like fiber
composed of seven double strands twisted helically around a
vertical axis. Depending on their orientation within the cell,
these polymers deform the shape of the red cell to produce
one of several morphological forms (Figure 14.7).
The intermolecular contacts within and between the
double strands have been extensively characterized. Lateral
contact between adjacent tetramers in the double strand
involves the mutant β6 Val, which is buried in a hydrophobic

Strand 2

Strand 1

(a)

(c) 1β2 2β1

1α2
Axial
1β1
(b)

2β2
(d)
1β1

2α2
Lateral 2β1 Axial

Fig. 14.7 HbS polymer structure. This is

the assembly unit of the HbS polymer: the
Werner–Love double strand. This double strand is
part of the HbS crystal, but in the crystal it
propagates in all directions to form a lattice. The
sickle fiber wraps around to form a helical fiber. The
squares depict the four different areas of contact
that form the double strand.

acceptor pocket formed by β88 Leu and β85 Phe, allowing
ingress of water molecules which form bridging hydrogen
bonds that stabilize the contact. The failure of HbF and HbA2
(and the corresponding hybrid tetramers) to copolymerize
with HbS has been attributed to specific amino acid residues,
but how these disrupt axial and lateral contacts in the HbS
polymer is still not fully resolved. The reduced oxygen
affinity of HbS favors the formation of deoxy-HbS and this is
exacerbated by the high 2,3-DPG content of sickle red cells.
Elevated sphingosine kinase activity may also contribute
through the generation of sphingosine-1-phosphate, which
also lowers hemoglobin oxygen affinity.
Adhesion of sickle red cells and other blood cells
to the endothelium
Based purely on the kinetics of HbS polymerization, it is pre-
dicted that the majority of red cells will not undergo sickling
in vivo, because the delay time for polymerization exceeds the
time taken to traverse the microvascular and venous circula-
tion and reach the lungs, where reoxygenation occurs. From
this has emerged the concept that sickle red cells are more
adherent to the vascular endothelium, which has shaped our
understanding of the pathophysiology of sickle cell anemia.
Adhesion of sickle red cells to the vascular endothelium plays
a key role in vaso-occlusion by retarding transit, thereby
favoring HbS polymerization and contributing to mechani-
cal obstruction.
In early studies Hebbel and colleagues, originally using
human umbilical vein endothelial cell monolayers, estab-
lished that sickle red cells were abnormally adherent to
endothelial cells, and postulated that this was linked to dis-
ease severity. The enhanced adherence of sickle red cells has
been reproduced in more physiological systems. The mech-
anisms proposed for these adhesive interactions are complex
and full discussion is beyond the scope of this chapter. Not
all have been validated in microcirculatory preparations or
proven to be relevant to the microvascular bed in which vaso-
occlusion occurs.
Sickle reticulocytes, which express the integrin α4β1 very
late antigen 4 (VLA-4) that binds to the vascular cell adhe-
sion molecule (VCAM)-1 on the surface of endothelial cells
and the subendothelial protein fibronectin, show increased
adherence to the endothelium of post-capillary venules, the
principal site of vaso-occlusion. Increased expression of other
adhesion molecules, including platelet glycoprotein (CD36)
and basal cell adhesion molecule (BCAM), by sickle red cells
promotes their adherence to the endothelium. Obstruction
of the microcirculation is preceded by the adhesion of young
sickle red cells, which trap dense cells and ISCs in those areas
to which they are adherent. Leukocytes also participate in
vaso-occlusion. Circulating activated leukocytes are present
in sickle cell anemia and show increased expression of αMβ2
Hemoglobinopathies due to structural mutations 181
integrin, which mediates adhesion to the endothelium. The
expression of P-selectin by activated endothelium mediates
abnormal rolling and adhesion of sickle red cells and initi-
ates neutrophil binding. Intravital microscopy in a murine
sickle model has shown that following an inflammatory stim-
ulus, neutrophils are recruited to the endothelium of post-
capillary venules, where they capture sickle red cells, possibly
through recognition of surface IgG, complement, or inter-
cellular adhesion molecule (ICAM)-4 (Landsteiner–Weiner
blood group antigen).
Platelets also contribute to the vaso-occlusive process. The
platelets of patients with sickle cell anemia show increased
expression of activation and adhesion markers, includ-
ing activated αIIbβ3 integrin (glycoprotein IIb/IIIa) and
P-selectin (CD62P), which mediate the formation of circu-
lating heterocellular aggregates with red cells, monocytes,
and neutrophils. Thrombospondin released from activated
platelets and shown to be elevated in plasma during painful
crisis may bridge CD36 or αVβ3 integrin expressed on
microvascular endothelial cells and CD36 or sulfated gly-
cans on sickle red cells. Increased plasma levels of platelet-
derived soluble CD40 ligand, which potentially contribute to
endothelial activation and pro-coagulant activity, are found
in patients with sickle cell anemia and correlate with the fre-
quency of acute pain episodes.
Adhesive interactions represent a promising target for
therapeutic intervention in SCD. Transgenic sickle mice that
lack E-selectin and P-selectin are protected against vaso-
occlusion. Several clinical trials of selectin blockade are now
underway (see later). Von Willebrand factor and αVβ3 inte-
grin appear to be other good candidates, since antibodies
against both inhibit cell adhesion in living circulatory prepa-
rations. Endothelial injury, which may be mediated by sickle
cells or NO depletion, leads to activation of endothelial cells,
with upregulation of the transcription factor nuclear factor
(NF)-κB and enhanced expression of multiple cell adhesion
molecules, and establishes a vicious circle that potentiates the
adherence of sickle red cells and vaso-occlusion.
Red cell dehydration
Red cell dehydration represents an integral component of
the pathophysiology of sickle cell anemia, because poly-
merization of HbS is critically dependent on intracellular
hemoglobin concentration. Although the MCHC is close to
normal, sickle red cells have a wide density distribution.
After they enter the circulation, a proportion of young sickle
red cells undergo rapid dehydration, which accelerates HbS
polymerization with conversion to dense cells and ultimately
ISCs. Cell dehydration is due to loss of potassium accompa-
nied by water efflux.
Three membrane transport pathways have been implicated
in the dehydration of sickle red cells. The best characterized
182 Molecular Hematology
are the Ca2+ -dependent K+ (Gardos) channel and K+ /Cl−
cotransport. In addition, perturbance of cation homeostasis
in sickle cell anemia is due to a less well-defined conduc-
tance pathway, referred to as Psickle, that may reflect abnor-
mal function of the mechanosensitive ion channel PIEZO1,
defects of which cause hereditary xerocytosis (stomatocyto-
sis). Activation of K+ /Cl− cotransport occurs in sickle cell
anemia as well as other hemoglobinopathies due to posi-
tively charged hemoglobin variants such as HbC. It has been
proposed that this is mediated by binding of the abnormal
hemoglobin to the inner surface of the cell membrane. In
normal red cells K+ /Cl− cotransport is confined to reticu-
locytes, but in sickle cell anemia activity persists in mature
red cells. The Ca2+ -dependent K+ (Gardos) channel is acti-
vated by intracellular accumulation of Ca2+ , which occurs on
deoxygenation in sickle red cells. Cells containing HbF are
less susceptible to this effect, suggesting it is due to polymer-
ization. When red cells undergo sickling, there is increased
cation permeability due to distortion of the membrane suffi-
cient to activate the Ca2+ -dependent K+ (Gardos) channel,
leading to K+ efflux and cell dehydration. Dense cells and
ISCs have markedly shortened survival and reduced oxygen
affinity.
Anemia
Anemia is not simply a product of hemolysis in sickle cell
anemia. The extent of marrow expansion is less than in
thalassemia major (five as opposed to ten times normal).
One explanation for this is the right shift in the oxygen equi-
librium curve, to which high red cell 2,3-DPG contributes,
as described earlier. This results in increased oxygen delivery
to the tissues, and explains why erythropoietin levels do
not increase until the hemoglobin falls below 9 g dl−1 .
Even below this threshold, however, the erythropoietin
response is blunted (i.e. inappropriately low for the degree
of anemia). The reason for this is not fully understood,
but the fact that this relative deficiency of erythropoietin
increases with age suggests that occult renal damage might be
responsible.
Hemolysis results from damage to and dysfunction of
the red cell membrane caused by polymerization of HbS.
Oxidative damage to membrane proteins and exposure of
phosphatidylserine on the outer surface of the red cell pro-
mote phagocytosis by macrophages. Sickle red cells are under
increased oxidative stress, the mechanism of which is com-
plex and includes the generation of superoxide and hydrogen
peroxide from the autoxidation of intracellular and extracel-
lular HbS (the latter being amplified in the absence of cellular
antioxidant systems), generation of hydroxyl radicals via the
Fenton reaction, activation of the xanthine–xanthine oxidase
pathway through recurrent ischemia–reperfusion injury, and
the depletion of reduced glutathione and enzymatic antiox-
idants, including superoxide dismutase and peroxiredoxins.
The hemolysis is also complex because of the heterocellular
nature of red cells in sickle cell anemia. There are actually
three red cell populations: the most dense red cells, which
survive only 4–5 days; F cells (which contain predominantly
HbF), which have close to a normal lifespan; and the remain-
ing majority of red cells, which have an intermediate survival.
Coexistent α thalassemia or high HbF levels reduce
hemolysis and consequently anemia. Interestingly, the
level of HbF mirrors the pattern of circulating progenitors
and hematopoietic cytokines seen in sickle cell anemia.
Low-HbF sickle cell anemia patients have an increased
number of peripheral blood burst-forming units-erythroid
(BFU-E), which are in active cycle. In addition, these
patients have constitutively elevated levels of erythropoietin,
granulocyte/macrophage colony-stimulating factor, and
stem cell factor, and lower levels of the inhibitory cytokine
transforming growth factor (TGF)-β, presumably reflecting
the need to maintain a more intense erythropoietic drive.
Sickle cell anemia patients are susceptible to an acute anemia
superimposed on that resulting from chronic hemolysis,
due to infection with parvovirus B19. This can produce
a life-threatening fall in hemoglobin, which is generally
self-limiting (after 7–15 days). This results from the ability
of parvovirus B19 to infect BFU-E through the blood group
P system that serves as its receptor.
Nitric oxide effects
In recent years, a new paradigm linking vascular dysfunction
and hemolysis in the pathophysiology of sickle cell anemia
has emerged. This is underpinned by evidence that in sickle
cell anemia, even in the steady state, the bioavailability of NO
is decreased, leading to perturbance of vascular homeostasis.
This arises because free hemoglobin, released into the plasma
as a consequence of intravascular hemolysis at a rate which
exceeds haptoglobin binding capacity, reacts rapidly with NO
produced by the endothelium to form methemoglobin and
nitrate. This scavenging mechanism does not occur under
physiological conditions because of the diffusion barriers
presented by the red cell membrane and the cell-free zone
adjacent to the endothelium.
The release of arginase-1 contained in red cells compounds
the effect on NO bioavailability through the conversion of l-
arginine, the substrate for NO synthase, to l-ornithine. Renal
dysfunction in sickle cell anemia may further compromise
the supply of arginine, as the kidneys are the primary site of
de novo synthesis. Low plasma arginine levels are found in
adults with SCD and predict early mortality. Hemolysis also
leads to release of asymmetric dimethylarginine (ADMA)
from red cells. ADMA inhibits NO synthase, which may
further limit NO production in SCD. The consequences of
reduced NO bioavailability include vasoconstriction, acti-
vation of endothelial cells with production of endothelin-1
(a potent vasoconstrictor and inflammatory mediator), and

upregulation of adhesion molecules, platelet activation, tis-
sue factor expression, and thrombin generation, all of which
may contribute to the complex pathophysiology of sickle
cell anemia. This model is supported by identification of a
clinical subphenotype of sickle cell anemia, associated with
impaired NO bioavailability, characterized by more severe
hemolysis and increased risk of vascular complications,
including ischemic stroke, pulmonary hypertension, leg
ulcers, priapism, and possibly nephropathy.
Treatment
Despite the in-depth understanding of the molecular basis
of sickle cell anemia, prefigured by Pauling’s landmark study
almost 70 years ago, improvement in clinical outcome has
been slow and remains elusive in low-income settings, where
childhood mortality is high. Progress in the development
of disease-modifying treatments has been disappointing and
lagged behind advances in pathophysiology. The concept that
adhesive interactions between circulating sickle red cells, leu-
cocytes, and endothelial cells play a central role in the patho-
physiology of SCD first evolved over 30 years ago. Despite
this, to date no therapy that targets cell adhesion in SCD has
been licensed.
In high- and middle-income countries, the introduction
of newborn screening combined with preventive measures
that include penicillin prophylaxis, vaccination, and par-
ent/carer education have proved highly effective in reducing
childhood mortality. The judicious use of red blood cell
transfusion in devastating complications such as acute
chest syndrome, stroke, splenic sequestration, and aplastic
crisis has also dramatically improved outcomes. With these
measures, mortality rates have improved to 0.5 per 100 000
affected children. The picture in adults is less favorable, with
average life expectancy still well below that of the general
population. Pivotal randomized controlled trials conducted
since the 1990s provide an evidence base for the use of
transfusion for primary and secondary prevention of stroke
and a rationale for the adoption of chronic transfusion pro-
grams to reduce the frequency of vaso-occlusive crises and
acute chest syndrome. In the acute setting, current therapies
are supportive, with emphasis on rapid assessment of the
intensity, quality, and distribution of pain and the presence
of comorbidities, coupled with prompt administration of
opioid analgesia, antibiotic therapy where indicated, and
hydration. Individually tailored opioid analgesia regimens
with the use of adjunctive non-steroidal anti-inflammatory
drugs are favored by most centers.
Hydroxycarbamide (hydroxyurea) was approved by the
Food and Drug Administration (FDA) in 1998, and is the
first licensed drug that has been shown to modify the natu-
ral history of SCD. Hydroxycarbamide therapy reduces the
incidence of painful crises and acute chest syndrome in
adult and pediatric sickle cell anemia patients, and leads to
Hemoglobinopathies due to structural mutations 183
a decrease in hospitalization and transfusion requirements.
Predictors of response to hydroxycarbamide include high
white cell count and absence of the Bantu haplotype. The
mechanism by which hydroxycarbamide, a ribonucleotide
reductase inhibitor, acts involves the recruitment of erythroid
progenitors programmed to produce HbF. This appears to
be mediated through NO-dependent activation of soluble
guanylyl cyclase. Although predicated on its action as an
inducer of HbF, it is apparent that hydroxycarbamide exerts
other effects of potential benefit in SCD. These include
reduced expression of adhesion molecules, enhanced NO
production and red cell hydration, and reduction in the
leukocyte, platelet, and reticulocyte counts.
Several targeted drug therapies for SCD have emerged
recently. Agents that block red cell and leukocyte adhe-
sion are attractive candidates. Crizanlizumab, the first
drug designed to target a specific cell adhesion molecule
implicated in vaso-occlusion, is a humanized monoclonal
antibody against P-selectin which blocks interaction with
its cognate ligand, P-selectin glycoprotein ligand 1. The
rationale for its use builds on extensive pre-clinical studies
(see earlier). The SUSTAIN study, a phase 2, multicenter,
placebo-controlled, double-blind trial, assessed the safety
and efficacy of this novel therapy, with or without hydrox-
ycarbamide, in patients with SCD. Over the course of the
12-month trial, crizanlizumab significantly reduced the rate
of pain crises compared to placebo. The results of a phase
3 clinical trial of rivipansel (GMI-1070), a pan-selectin
inhibitor that in earlier studies reduced the duration of pain
crises, length of hospital stay, and opioid requirement, are
awaited. In 2017 the FDA approved oral L-glutamine for use
in SCD based on the results of a phase 3 clinical trial which
over 48 weeks showed a significant reduction in crisis and
hospitalization rate in the intervention arm. The proposed
mechanism of action of L-glutamine protects red cells against
oxidative damage through promoting NADH generation.
Another promising avenue for drug development is inhibi-
tion of HbS polymerization, the initiating pathophysiological
event. Previous efforts toward this goal have been thwarted
by unfavorable pharmacological properties and off-target
effects. Aromatic aldehydes such as 5-hydroxymethylfurfural
(5-HMF) and its derivative GBT440 are allosteric effectors of
HbS which increase its oxygen affinity, maintain the R state,
and consequently inhibit polymerization. GBT440, which is
active orally, binds to the N-terminal valine of the α-chain
of hemoglobin, forming a reversible Schiff base. Based on
pre-clinical and early-phase clinical trial data, GBT440 has
received Breakthrough Therapy and Priority Medicines des-
ignation from the FDA and European Medicines Agency,
respectively. A phase 3, double-blind, randomized, placebo-
controlled, multicenter study of GBT440, the HOPE study, is
in progress.
The options for curative therapy in SCD are expanding.
Allogeneic bone marrow transplantation (BMT) for SCD
184 Molecular Hematology
has been performed since the 1980s, though access is
limited, even in developed countries. Current consensus
guidelines recommend this option for young patients,
ideally pre-school age, with symptomatically severe SCD
and a human leukocyte antigen (HLA)-matched sibling
donor. A recent international survey of the results of 1000
HLA-identical sibling hematopoietic stem cell transplants
showed the five-year event-free and overall survival, using
predominantly conventional myeloablative conditioning,
were 91.4% and 92.9%, respectively. Transplant-related
mortality was 7% and the risk of graft rejection 2.3%. Both
increase with age. Overall survival was higher in patients
younger than 16 years (95%) than those aged 16 years or
older (81%). Reduced intensity conditioning has produced
encouraging results and holds promise as a safer option
in older patients. Only a minority of eligible patients have
an HLA-matched related donor. This has prompted con-
sideration of alternative stem cell sources in the form of
unrelated umbilical cord blood or marrow and haploiden-
tical donors, which opens the possibility of cure to a wider
group, though at the cost of increased graft rejection and
mortality.
The long road to gene therapy for SCD reached a turning
point in 2017 with the report by Ribeil et al. of the success-
ful treatment of a 13-year-old boy with a severe sickle cell
phenotype, on a prophylactic red cell transfusion program,
using a lentiglobin vector carrying a β-globin gene modified
to enhance the anti-sickling effect (LentiGlobin BB305 vec-
tor βA-T87Q ). At 15 months, the patient was asymptomatic
and transfusion independent, with sustained production of
the anti-sickling β-globin and correction of disease mark-
ers. No safety concerns were reported. Additional results
will follow from the HGB-205 and HGB-206 clinical trials
of LentiGlobin BB305 in SCD. Several additional gene ther-
apy trials using alternative gene cassettes are ongoing. In the
long term gene therapy holds considerable promise for the
treatment of SCD, though to confirm the durability of clini-
cal benefit and safety, greater patient numbers and extended
follow-up are needed.
Since the discovery of the clustered regularly interspersed
short palindromic repeats (CRISPR)/Cas9 complex, gene
editing has emerged as a potential tool for the treatment
of SCD. The complex consists of a Cas9 endonuclease
and a sequence-specific single-guide RNA, which directs
the endonuclease to the target DNA sequence. Following
cleavage by the Cas9 endonuclease, one of two mechanisms
may be employed to repair the double-strand break: non-
homologous end joining (NHEJ) or homology-directed
recombination (HDR). HDR enables genetic correction,
but currently this is difficult to achieve at therapeutic
efficiency in quiescent hematopoietic stem cells in which
NHEJ is the dominant repair pathway. At present, genetic
disruption which relies on NHEJ to produce targeted
insertions/deletions seems a more realistic prospect for the
treatment of SCD and other β hemoglobinopathies.
The manipulation by gene editing of factors which silence
γ-globin expression offers an attractive approach. BCL11A
is a potent repressor of fetal hemoglobin, but it also has key
non-erythroid functions in hematopoietic stem cells, neu-
ron development, B-cell lymphopoiesis, and dendritic cells.
A solution to the problem of selectivity might be to tar-
get only one of the two BCL11A alleles, as in the mouse
model BCL11A repression of γ-globin is exquisitely sensi-
tive to gene dosage. SNPs affecting BCL11A which result
in a modest decrease in gene expression, preserving around
65% of normal activity, are associated with an approxi-
mately threefold increase in fetal hemoglobin in patients
with sickle cell disease. Individuals who are haploinsuffi-
cient for BCL11A have an elevated fetal hemoglobin level.
An elegant alternative is disruption by genome editing of an
erythroid-specific enhancer for BCL11A to achieve selective
reduction of expression in erythroid precursors. Strategies
which take advantage of the competitive expression of globin
genes at the β locus to replace HbA with a non-sickling
form such as HbF hold particular promise in the treatment
of SCD.
Sickle/β thalassemia
This syndrome is observed in geographical regions in which
both HbS and β thalassemia are frequent, such as Africa,
Sicily, Greece, Turkey, the Arab countries, and regions of
the Americas with African and southern Mediterranean
admixture.
There are two phenotypic forms: S/β+ thalassemia, in
which the red cells contain between 20 and 40% HbA, the
remainder comprising HbS, HbF, and HbA2 ; and S/β0 tha-
lassemia, in which the red cells contain only HbS, HbF,
and HbA2 . The latter may sometimes be difficult to dis-
tinguish from the homozygous state for the βS gene and
this may require pedigree or genotypic analysis. The rel-
ative frequency of S/β0 and S/β+ thalassemia reflects the
prevalence of individual β thalassemia mutations, which
varies from one geographical region to another. In peo-
ple of African descent S/β+ thalassemia predominates,
whereas in the Mediterranean S/β0 thalassemia is more
common.
Though similar to that of sickle cell anemia, the clin-
ical spectrum of S/β thalassemia is broader and in S/β+
thalassemia significantly milder. However, certain complica-
tions, including proliferative retinopathy, osteonecrosis, and
hypersplenism, are more common.
The hemoglobin level is higher and mean cell volume
(MCV) as well as mean corpuscular hemoglobin (MCH) typ-
ically lower than in sickle cell anemia.

Hemoglobin C disease
Worldwide, hemoglobin C is the third most common struc-
tural hemoglobin variant (after HbS and HbE). Individuals
who are homozygous for HbC display a mild hemolytic ane-
mia that is generally asymptomatic. Splenomegaly is com-
mon, but usually clinically benign. The anemia is mild to
moderate in degree and accompanied by a reduced MCV. The
peripheral blood film is fairly characteristic and, with some
experience, diagnostic. The red cells appear hypochromic
due to their flatness (rather than low MCH or MCHC; in fact,
the red cells have a normal MCH and an increased MCHC).
Folded and target cells are prominent. Red cells may, though
infrequently in the absence of splenectomy, show intracellu-
lar tetragonal crystals of HbC, best detected in reticulocyte
preparations, where they retain their red color.
The βC gene (β6 Glu → Lys) has a frequency one-quarter
that of the βS gene among African Americans. The βC
mutation most likely originated in Burkina Faso, where
the highest gene frequency (up to 0.5) is found, decreasing
concentrically, and encompassing Mali, Ivory Coast, and
Ghana, all of which lie east of the Niger river.
The pathophysiology of HbC disease is dominated by
the reduced solubility of hemoglobin, which results from
replacement of β6 glutamic acid by lysine and its effect on
membrane transport. The former is manifest as a tendency
for oxy HbC to form tetragonal crystals (Figure 14.8).
Although less deformable, the red cells in HbC disease do
not produce vaso-occlusion, even when present in signifi-
cant numbers after splenectomy. This may be explained by
dissolution of oxy HbC crystals under conditions of low Po2
in the microvascular circulation. The red cells are uniformly
denser. This characteristic is associated with stimulation
of K+ /Cl− cotransport (higher than in SS cells, despite a
lower reticulocyte count), a transport system which, through
efflux of K+ and accompanying water loss, leads to red cell
Fig. 14.8 Hb C tetragonal crystals.
Hemoglobinopathies due to structural mutations 185
dehydration. The kinetics of K+ /Cl− cotransport are altered
in HbC disease red cells, with delayed switch-off when the
stimulus for activity (cell volume increase or low pH) is
removed. Whether this alteration is sufficient to explain
the pathophysiology of HbC disease and its relation to
interaction of HbC with the red cell membrane remains to
be determined.
Epidemiological data from Burkina Faso reveals that indi-
viduals homozygous for HbC are particularly resistant to
death from malaria. This is congruent with previous obser-
vations that the red cells of HbC homozygotes are less able
to release merozoites than normal or heterozygote (HbAC)
red cells. By virtue of this greater protective effect, HbC may
eventually supersede the HbS gene with respect to preva-
lence in those populations where the two mutants coexist.
The multiplication rate of P. falciparum is lower in the red
cells of HbC homozygotes than in normal red cells, proba-
bly due to disintegration of ring forms and trophozoites. It
appears that only a subset of red cells of HbC homozygotes
supports normal parasite replication.
The structural basis for HbC crystallization has been elu-
cidated. HbC crystals form by nucleation and grow by the
attachment of individual HbC molecules from the solution.
HbC crystallization is possible because of the huge entropy
gain, likely stemming from the release of up to 10 water
molecules per protein intermolecular contact–hydrophobic
interaction. The propensity of oxy HbC to form crystals is
attributable to increased hydrophobicity resulting from the
conformational changes that accompany the β6 Lys muta-
tion. The oxy ligand state is thermodynamically driven to a
limited number of aggregation pathways, with a high propen-
sity to form the tetragonal crystal structure. This contrasts
with deoxy HbC, which energetically favors equally multiple
pathways of aggregation, not all of which will culminate in
crystal formation, and which has a different crystal form.
Hemoglobin SC disease
The compound heterozygous disorder hemoglobin SC dis-
ease combines two structural variants, HbS and HbC, neither
of which is associated with a clinical phenotype in the simple
carrier state. The clinical basis of Hb SC disease, a significant
sickling disorder, therefore demands an explanation. The
pathophysiology of Hb SC disease derives from two mech-
anisms. The relative proportion of HbS to HbC is 50 : 50, in
contrast to sickle cell trait in which the ratio of HbS to HbA
is 40 : 60. This reflects the variable electrostatic affinity of
structurally distinct β chains for α globin, which governs the
proportion of different αβ subunits and hemoglobin assem-
bly within the red cell. The red cells of Hb SC individuals
have a higher MCHC, reflecting dehydration due to increased
K+ /Cl− cotransport. Since polymerization is dependent on
186 Molecular Hematology
AA
Reticulocytes
Dense SC cells
ISC
AA AC SC SC CC SS Beads
intracellular HbS concentration, both these factors favor
polymer formation. In keeping with this, in vitro studies have
shown that increasing the proportion of HbS from 40% to
50% increased the rate of polymerization almost 15-fold. Hb
SC red cells are denser than normal or the majority of sickle
cell anemia red cells (Figure 14.9). Interestingly, HbS favors
HbC crystallization; this explains, in addition to the loss of
splenic function, why crystals are generally more prominent
in Hb SC than HbC disease. However, for the reasons already
given, this does not contribute to vaso-occlusion.
Hb SC disease is milder than sickle cell anemia. Virtually
all patients survive to adulthood and average life expectancy
is close to that of matched controls. The clinical spectrum
resembles that of sickle cell anemia, but the frequency and
severity of most specific complications are generally less.
Two exceptions show a disproportionate prevalence or sever-
ity in Hb SC disease – osteonecrosis and proliferative sickle
retinopathy the frequency of which is similar and higher
respectively to that in sickle cell anemia. Splenomegaly is
more frequent and functional hyposplenism delayed. Mean
red cell survival is around 27 days compared with 17 days in
sickle cell anemia. Hematologically, Hb SC disease is char-
acterized by a higher hemoglobin, lower MCV, lower retic-
ulocyte count, fewer dense red cells, and lower HbF level
compared with sickle cell anemia. The blood film in Hb
SC disease may be diagnostic. In addition to the red cell
forms seen in HbC disease, there are frequent dense cells,
some of which resemble a broad-beamed canoe and occa-
sionally ISCs. Intraerythrocytic crystals are best visualized
with supravital stains (Figure 14.10) or after incubation of
blood with 3% sodium chloride (NaCl).
Other sickle hemoglobinopathies
Many compound heterozygous states have been described in
which HbS is co-inherited with a second β globin variant,
most of which are of no clinical significance. Exceptions
are Hb SO-Arab, Hb SD-Punjab (Los Angeles), and Hb
SE. Hb O-Arab (β121 Glu → Lys) and Hb D-Punjab (β121
Density MCHC
g/ml g/dl
1.065 26.0
Fig. 14.9 Percoll density gradients of AA, CC,
1.085 34.0 and SC red cells. This shows the separation of red
cells by density. Light-density cells are on top,
high-density cells on the bottom. Note that SS
1.095 38.0
blood has normal-density red cells as the largest
1.107 40.5 compartment, then some very dense cells, and
1.125 49.0 some intermediate-density cells between the two.
1.143 55.0
In contrast, all the cells of the CC patients are
denser than normal red cells. The SC red cells are
intermediate.
Glu → Gln) interact with HbS to produce a severe clinical
phenotype akin to sickle cell anemia. This is explained by the
fact that these substitutions at the β121 residue, a contact
point in the HbS polymer, increase the nucleation rate for
polymer formation. Only a small number of cases of Hb SE
have been described, all of which displayed a mild pheno-
type. Other compound heterozygous sickling disorders due
to a second structural variant are rare. Co-inheritance of Hb
Quebec-Chori (β87 Thr → Ile) and HbS is an interesting
example. Since Hb Quebec-Chori is electrophoretically
silent, the compound heterozygous state with HbS mimics
sickle cell trait; however, its presence can be distinguished
by mass spectrometry. The proposed mechanism for the
interaction between the two variant globin chains is that
substitution of a bulky hydrophobic isoleucine residue at β87
creates a more favorable environment in the acceptor pocket
for β6 Val, which stabilizes intermolecular contacts in the
polymer.
Several dominant sickle mutations in which there is a sec-
ond substitution in addition to β6 Val in the same β chain
have been described. This results in assembly of a “super-
Hb S,” which manifests with symptoms in the heterozygote.
Several examples have been well characterized. Hb S-Antilles
(β6 Glu → Val; 23 Val → Ile) is expressed in the heterozygote
at a level of around 40% and produces a syndrome resem-
bling a mild sickle cell anemia phenotype. The mechanism
is complex: the second mutation reduces the solubility from
a Csat of 18 g dl−1 for Hb S to 12 g dl−1 , with the result that
polymerization is promoted. Hb S-Antilles also has reduced
oxygen affinity. This oxygen equilibrium shift favors sickling.
A similar effect is seen with Hb Jamaica Plain (β6 Glu → Val;
68 Leu → Phe), in which the amino acid substitution at β68
reduces oxygen affinity, producing symptoms of sickling in
the heterozygous state. The other well-documented dom-
inant sickle mutation is Hb S-Oman (β6 Glu → Val; 121
Glu → Lys). This generates an even more powerful “super-
Hb S.” Even at expression levels as low as 20%, resulting
from concomitant inheritance of the α thalassemia geno-
type −α/−α, this is associated with a phenotype similar to
Hb S-Antilles. Since the two dominant forms of Hb S have

Fig. 14.10 SC red cells from a double-stained
smear. This blood smear is from a patient with Hb
SC and is stained with Wright’s and reticulocyte
stain. Note the fl at thin cells (which generate
target cells), folded cells, and those with
intracellular tetragonal crystals (arrows).
similar solubility in the deoxy state, the more severe pheno-
type of Hb S-Oman implies that the β121 mutation (identical
to that in Hb O-Arab) possesses intrinsic pathophysiologi-
cal properties. This hypothesis is supported by the hemol-
ysis and perturbance of red cell cation transport evident in
individuals homozygous for Hb O-Arab. Rare symptomatic
HbS heterozygotes in whom deficiency of pyruvate kinase
has been identified have been described. The mechanism in
these cases is attributed to the increased 2,3-DPG content of
red cells, with a block in glycolysis distal to the Rapoport–
Luebering shuttle that shifts the hemoglobin oxygen equilib-
rium in favor of deoxy Hb S.
Homozygous HbE and Hb E/β
thalassemia
Hemoglobin E (β26 Glu → Lys), the most common struc-
tural hemoglobin variant worldwide, is prevalent across a
wide geographical area, from the eastern provinces of India to
the Philippines. Populations living near the common border
of Cambodia, Laos, and Thailand (the Khmer people) have
the highest incidence (gene frequency reaching 0.5) of HbE,
the selection pressure for which is resistance to infection by
P. falciparum malaria.
In heterozygotes HbE constitutes 25–30% of the
hemoglobin. In common with the other widespread β
Hemoglobinopathies due to structural mutations 187
chain variants, HbS and HbC, this proportion is reduced in
the presence of α thalassemia. Homozygotes exhibit a benign
thalassemic phenotype because the mutation responsible
produces both a structurally altered β chain and reduced
expression due to the generation of an alternative splice site.
The clinical syndrome of homozygous HbE disease is very
mild, with no or minimal anemia (once nutritional causes
of anemia are treated) and low MCV, with target cells in the
smear and hypochromia. Density gradient separation reveals
that although the red cells are small they have a normal
MCHC, as a consequence of a diminished MCH, due to the
thalassemic component of this disease. People who inherit
HbE and β thalassemia trait present with the features of
thalassemia major or intermedia (see Chapter 1).
The protective effect of HbE against P. falciparum malaria
has been assessed in a mixed erythrocyte invasion assay. The
parasite preferentially invaded normal (AA) red cells com-
pared with AE, EE, or E/β thalassemia red cells. Heterozy-
gote (AE) red cells were the poorest target, showing inva-
sion restricted to approximately 25% of the red cells. It has
been postulated that AE red cells might have an uniden-
tified membrane abnormality that renders the majority of
the red cell population relatively resistant to invasion by P.
falciparum, consequently reducing parasitemia and hence
the lethality of infection. This is consistent with the Hal-
dane hypothesis of heterozygote protection against severe
malaria.
188 Molecular Hematology
Unstable hemoglobins
Cases of non-spherocytic hemolytic anemia due to a struc-
turally unstable hemoglobin were first recognized because
precipitation occurred on thermal incubation of hemolysate,
a phenomenon which forms the basis of the heat stability test.
In many patients Heinz bodies were present. There are over
100 mutations that render the hemoglobin molecule unsta-
ble, of which only a minority are associated with significant
clinical symptoms. About 75% are β chain variants. Unstable
hemoglobins are generally inherited in an autosomal dom-
inant manner, although rare homozygous cases, for exam-
ple Hb Bushwick (β74 Gly → Val) and Hb Taybe (α1 38/39
Thr deletion), have been described. In a significant number
of cases there is no antecedent family history, indicating that
the mutation has arisen spontaneously. The unstable variant
usually accounts for 10–30% of the total hemoglobin.
Structural alterations may render hemoglobin unstable by
five mechanisms:
r destabilization of the attachment of heme to globin;
r interference with stability of the α β contact area, which
1 1
does not normally dissociate;
r introduction of either a bulky or charged side chain to the
hydrophobic interior of the hemoglobin molecule;
r disruption of the secondary structure of the molecule by
a side chain that is not α-helix-friendly (commonly as a
result of proline substitution);
r elongation of a globin subunit.
Hb Köln (β98 Val → Met) is the most common of the
unstable hemoglobins and shows no ethnic predilection.
Many unstable hemoglobin variants do not alter the sur-
face charge of the hemoglobin molecule and therefore have
an electrophoretic mobility identical to that of HbA. Light
bands representing dissociated globin subunits, in some cases
devoid of heme, may be observed.
Clinically, unstable hemoglobins are characterized by
hemolytic anemia, which ranges from mild to severe depend-
ing on the causative mutation, and the presence of pigmen-
turia (due mostly to dipyrroles). Unstable hemoglobins due
to γ chain variants, such as Hb Poole (G γ130 Trp → Gly), are
a rare cause of transient neonatal hemolysis. The mechanism
of red cell destruction, much of which occurs in the bone
marrow, involves displacement of heme, which destabilizes
the hemoglobin tetramer and its composite globin subunits
and leads to the formation of hemichrome. Globin subunits
aggregate and precipitate, leading to reduced deformability
and removal or entrapment of red cells, which contain Heinz
bodies. Febrile episodes and ingestion of oxidant drugs may
accelerate hemolysis. Hemoglobin level may be modulated
by altered oxygen affinity, being near normal in those
unstable variants with increased oxygen affinity (e.g. Hb
Köln) and much lower in those with reduced oxygen affinity
(e.g. Hb Hammersmith). Hb Zurich (β63 His → Arg) is
another interesting variant that affects multiple hemoglobin
functions. The structural change in Hb Zurich alters the
heme pocket, resulting in reduced oxygen affinity but little
effect on carbon monoxide (CO) binding. The higher levels
of carboxyhemoglobin (CO-Hb) found in patients with Hb
Zurich stabilize hemoglobin, with the irony that smoking
affords protection against hemolysis.
In most cases treatment revolves around supportive mea-
sures that include folic acid supplementation, prompt treat-
ment of infection and fever, and avoidance of oxidant drugs.
In severe cases splenectomy has been performed and is gen-
erally, though not universally, beneficial. Hydroxycarbamide
has been used to stimulate Hb F in unstable β chain variants.
Some unstable hemoglobin variants result in low hemoglobin
oxygen saturation measurements by pulse oximetry due to
their altered absorbance spectra.
High oxygen affinity hemoglobins
The sigmoid curve of the oxygen binding of hemoglobin
reflects the transition from the low-affinity deoxy (T) to
the oxy (R) state. With this molecular switch, hemoglobin
acquires a high affinity for oxygen. Mutations that stabilize
the R state or destabilize the T state produce hemoglobins
with increased oxygen affinity. In response, the oxygen-
sensing mechanism in the kidney triggers an increased
output of erythropoietin, which in turn stimulates red cell
production. This results in erythrocytosis and an elevated
hematocrit.
There are approximately 100 high-affinity hemoglobin
variants, most of which are rare. Almost all show a domi-
nant mode of inheritance. Variants associated with a marked
increase in hemoglobin oxygen affinity are likely to be lethal
in homozygotes, and the few examples of homozygous high-
affinity hemoglobins reported are confined to less severe phe-
notypes. The first high-affinity variant to be described was
Hb Chesapeake (α2 98 Arg → Leu), which has a P50 on whole
blood of 19 mmHg (normal 26 mmHg) and exhibits reduced
cooperativity. This mutation results in substitution at a con-
served residue of the α1 β2 interface between dimer subunits
that is important for stability of the T conformer. The substi-
tution of leucine stabilizes the R state.
The molecular defects responsible for high-affinity
hemoglobins may be divided into those that occur at sites
directly involved in the R → T transition (the α1 β2 contact
area or the switch region of the molecule, particularly the
C and N terminals), such as Hb Chesapeake; those which
disrupt the 2,3-DPG binding site in the central cavity, such
as Hb Old Dominion (β143 Val → Met); mutations at the
α1 β1 interface that produce a structural change favoring the
R state, such as Hb San Diego (β109 His → Tyr); and those

which constrain quaternary conformational change through
polymerization, such as Hb Porto Alegro (β9 Ser → Cys).
The clinical course of high-affinity hemoglobin variants is
benign. The reduction in tissue oxygen delivery is compen-
sated for by an increased red cell mass and treatment is not
generally required. Maternal high-affinity hemoglobin vari-
ants do not compromise fetal well-being, arguing against a
physiologically important role for the greater affinity for oxy-
gen of Hb F than Hb A in normal pregnancy. Curiously, the
degree of erythrocytosis may differ among individuals who
inherit high-affinity variants with similar P50 values, suggest-
ing other factors influence the adaptation to hypoxia.
Low oxygen affinity hemoglobins
Hemoglobins with low oxygen affinity deliver oxygen to
the tissues more efficiently and are associated with ane-
mia due to reduced erythropoietic drive. Relatively few
hemoglobin variants with low oxygen affinity have been
described. Cyanosis will be evident if the deoxyhemoglobin
level exceeds 5 g dl−1 . This may be present at birth in low-
affinity α chain variants, but its onset is delayed in low-
affinity β chain variants because of the latency in β globin
expression.
The first low-affinity variant to be described was Hb
Kansas (β102 Asn → Thr). Two other substitutions at β102
are associated with low affinity, Hb Beth Israel (β102
Asn → Ser) and Hb St Mande (β102 Asn → Tyr). This is
explained by the fact that β102 Asn, a highly conserved
residue, is required to stabilize the hemoglobin molecule at
the α1 β2 interface in the R state. If cyanosis in the absence
100
ob
gl
yo
80
M
Oxygen saturation (%)
Hig
60
Fig. 14.11 Oxygen-binding curves of 40
hemoglobins with high and low oxygen
affinity. Notice that the extraction of oxygen by
the tissues, which is the difference between
20
pulmonary oxygen pressure (100 mmHg) and
capillary oxygen pressure (40 mmHg), is lower than
normal in high-affinity hemoglobins (low P50 ) and
higher than normal in low-affinity hemoglobins.
0 20
This is why the former have erythrocytosis and the
latter (most of the time) anemia.
Hemoglobinopathies due to structural mutations 189
of overt cardiopulmonary causes and anemia are evident, the
diagnosis of a low-affinity hemoglobin variant should be con-
sidered, partly to avoid unnecessary investigation. However,
neither cyanosis nor anemia is universal. In patients in whom
the P50 is markedly increased, anemia is not usually present.
Hemoglobin separation techniques are helpful only if the
amino acid substitution alters the charge of the molecule and
measurement of the oxygen dissociation curve (P50 ) is the
gold standard. The oxygen delivery properties of high- and
low-affinity hemoglobins are illustrated in Figure 14.11.
M hemoglobins
M hemoglobins result from mutations which stabilize heme
iron in the oxidized ferric (Fe3+ ) state. Inheritance is auto-
somal dominant. Nine M hemoglobin variants have been
described, most of which are the product of substitution, in
all but one case by tyrosine, of the proximal or distal histidine
interacting with the heme group. All are rare, with the excep-
tion of Hb Iwate (α2 87 His → Tyr), originally described in
the Iwate prefecture of Japan. The interactions between the
tyrosine-substituted histidine residues and heme iron have
been resolved for some M hemoglobins by X-ray crystallog-
raphy and nuclear magnetic resonance (NMR) spectroscopy,
and suggest a model whereby the side group of tyrosine
acts as an internal ligand preventing interaction with the
residual histidine residue. One other molecular defect, Hb
M-Milwaukee 1 (β67 Val → Glu), has been described in
which the carboxyl group of the substituted glutamic acid
residue binds and stabilizes oxidized heme iron. Predictably,
some M hemoglobins have reduced oxygen affinity.
20
High
in
Normal
y
nit
15
ffi
ha
O2 content (vol %)
al
rm
No
10
nity Low
affi
Low
5
40 60 80 100
Oxygen tension (mmHg)
190 Molecular Hematology
Hemoglobin A Hemoglobin M Boston
G He
lix
E Helix
87 His E Helix
58 His 58 Try
F Helix
Hemoglobin A Hemoglobin M Milwaukee
lix
G He
67 Val
92 His 92 His
E Helix
F Helix F Helix
M hemoglobins resulting from mutation of the α, β,
or G γ genes have been described. The latter were iden-
tified in newborns with “cyanosis” that resolved sponta-
neously in the first few months of life. Affected individuals
appear slate gray in color due to pseudocyanosis (there may
in addition be an element of true cyanosis in low-affinity
M hemoglobin variants), but are otherwise asymptomatic
and have a normal life expectancy. Pulse oximetry mea-
surements are unreliable. Although standard hemoglobin
separation techniques may be helpful, definitive diagno-
sis requires examination of the visible absorbance spec-
trum between 450 and 650 nm of a hemolysate, which
reveals individual spectral patterns that differ from that
of methemoglobin (Figure 14.12). Differential diagnosis
includes enzymopathic or acquired methemoglobinemia,
sulfhemoglobinemia, and unstable hemoglobin variants such
as Hb Chile (β28 Leu → Met) associated with accelerated
methemoglobin formation. The main value in identification
of M hemoglobins lies in the avoidance of unnecessary med-
ical intervention.
Conclusions
Structurally abnormal hemoglobins may present as one of
several well-defined clinical syndromes:
r homozygous or compound heterozygous inheritance of
HbS produces a picture of chronic anemia and cumulative
organ damage, punctuated by intermittent painful crises
and other acute manifestations.
G He
lix
87 Tyr
F Helix
lix
G He
Fig. 14.12 Structure of M hemoglobins. Heme
environment of two M hemoglobins compared
with HbA. Note the diverse changes in the distal
67 Glu
and proximal tyrosines (which have replaced the
normal histidines) in each of the M hemoglobins.
E Helix These hemoglobins have a characteristic visible
spectrum around 610 nm.
r HbC in the homozygous state is associated with mild
chronic hemolysis, but when co-inherited with HbS pro-
duces a syndrome similar to but generally milder than
sickle cell anemia.
r homozygous HbE results in hemolysis and a mild micro-
cytic anemia, but in combination with β thalassemia pro-
duces a phenotype of remarkable variability, ranging from
mild thalassemia intermedia to transfusion-dependent
thalassemia major.
r unstable hemoglobins lead to a hemolytic anemia of vari-
able severity.
r low oxygen affinity variants and M hemoglobins are char-
acteristically associated with cyanosis or pseudocyanosis,
though if unstable may also be associated with hemolysis
of mild degree.
r high oxygen affinity mutant hemoglobins present with ery-
throcytosis.
This panoply of clinical syndromes can be attributed to
the following alterations of the hemoglobin molecule: (i)
modification of the biophysical properties of the hemoglobin
molecule (e.g. polymerization of deoxy HbS); (ii) altered
ligand binding affinity; (iii) changes in the heme environ-
ment (e.g. M hemoglobins); (iv) instability of the functional
tetramer; and (v) reduced globin synthesis (e.g. HbE). Over
1000 genetic variants of hemoglobin have been described.
These are predominantly due to missense mutation and most
are rare and clinically silent. Only HbS, HbC, and HbE have
reached a high frequency in human populations as a result
of the selective pressure of malaria. These together with less
common variants have provided a wealth of insight into the
structure and function of human hemoglobin.
 Hemoglobinopathies due to structural mutations 191

Nagel, R.L., Johnson, J., Bookchin, R.M. et al. (1980). Beta-chain contact
Further reading sites in the haemoglobin S polymer. Nature 283: 832–834.
Porteus, M. (2016). Genome editing: a new approach to human thera-
General peutics. Annu. Rev. Pharmacol. Toxicol. 56: 163–190.
Prengler, M., Pavlakis, S.G., Prohovnik, I. et al. (2002). Sickle cell
Eaton, W.A. and Hofrichter, J. (1990). Sickle cell hemoglobin polymer-
disease: the neurological complications. Ann. Neurol. 51: 543–
ization. Adv. Protein Chem. 40: 63–279.
552.
Henry, E.R., Bettati, S., Hofrichter, J. et al. (2002). A tertiary two-state
Ribeil, J.A., Hacein-Bey-Abina, S., Payen, E. et al. (2017). Gene ther-
allosteric model of hemoglobin. Biophys. Chem. 98: 198–164.
apy in a patient with sickle cell disease. N. Engl. J. Med. 376: 848–
Nagel, R.L. (2002). Disorders of hemoglobin function and stability. In:
855.
Blood: Principles and Practices of Hematology (eds. R.I. Handon, S.E.
Serjeant, G.R. and Serjeant, B.E. (2001). Sickle Cell Disease. Oxford:
Lux and T.P. Stossel). Philadelphia, PA: J.B. Lippincott.
Oxford University Press.
Steinberg, M., Forget, B., Higgs, D. et al. (eds.) (2009). Disorders of
Ware, R.E., de Montalambert, M., Tshilolo, L. et al. (2017). Sickle cell
Haemoglobin: Genetics, Pathophysiology and Clinical Management.
disease. Lancet 390: 311–323.
Cambridge: Cambridge University Press.
Thom, C.S., Dickson, C.F., Gell, D.A. et al. (2013). Hemoglobin variants:
biochemical properties and clinical correlates. Cold Spring Harb. Per- Hemoglobin C
spect. Med. 3: a011858.
Dewan, J.C., Feeling-Taylor, A., Puius, Y.A. et al. (2002). Structure of
mutant human carbonmonoxyhemoglobin C (betaE6K) at 2.0 Å res-
Sickle cell disease olution. Acta Crystallogr. D Biol. Crystallogr. 58: 2038–2042.
Fabry, M.E., Romero, J.R., Suzuka, S.M. et al. (2000). Hemoglobin C
Chaturvedi, S., Bhatnagar, P., Bean, C.J. et al. (2017). Genome-wide asso-
in transgenic mice: effect of HbC expression from founders to full
ciation study to identify variants associated with acute severe vaso-
mouse globin knockouts. Blood Cell Mol. Dis. 26: 331–347.
occlusive pain in sickle cell anemia. Blood 130: 686–688.
Fairhurst, R.M., Fujioka, H., Hayton, K. et al. (2003). Aberrant develop-
Desai, A.A., Lei, Z., Bahroos, N. et al. (2017). Association of circulating
ment of Plasmodium falciparum in hemoglobin CC red cells: implica-
transcriptomic profiles with mortality in sickle cell disease. Blood 129:
tions for the malaria protective effect of the homozygous state. Blood
3009–3016.
101: 3309–3315.
Dever, D.P., Bak, R.O., Reinisch, A. et al. (2016). CRISPR/Cas9 beta-
Modiano, D., Luoni, G., Sirima, B.S. et al. (2001). Haemoglobin C pro-
globin gene targeting in human haematopoietic stem cells. Nature
tects against clinical Plasmodium falciparum malaria. Nature 414:
539: 384–389.
305–308.
Gluckman, E., Cappelli, B., Bernaudin, F. et al. (2017). Sickle cell dis-
Nagel, R.L. and Lawrence, C. (1991). The distinct pathobiology of SC
ease: an international survey of results of HLA-identical sibling
disease: therapeutic implications. Hematol. Oncol. Clin. North Am. 5:
hematopoietic stem cell transplantation. Blood 129: 1548–1556.
433–451.
Halsey, C. and Roberts, I.A. (2003). The role of hydroxyurea in sickle
Vekilov, P.G., Feeling-Taylor, A.R., Petsev, D.N. et al. (2002). Inter-
cell disease. Br. J. Haematol. 120: 177–186.
molecular interactions, nucleation, and thermodynamics of crystal-
Herrick, J.B. (1910). Peculiar elongated and sickle-shaped red blood cor-
lization of hemoglobin C. Biophys. J. 83: 1147–1156.
puscles in a case of severe anemia. Arch. Intern. Med. 6: 517.
Kato, G.J., Steinberg, M.H., and Gladwin, M.T. (2017). Intravascular
hemolysis and the pathophysiology of sickle cell disease. J. Clin. Hemoglobin E
Invest. 127: 750–760.
Chotivanich, K., Udomsangpetch, R., Pattanapanyasat, K. et al. (2002).
Kaul, D.K., Fabry, M.E., and Nagel, R.L. (1996). The pathophysiology of
Hemoglobin E: a balanced polymorphism protective against high
vascular obstruction in the sickle syndromes. Blood Rev. 10: 29–44.
parasitemias and thus severe P falciparum malaria. Blood 100: 1172–
Lettre, G. (2012). The search for genetic modifiers of diseases sever-
1176.
ity in the β-hemoglobinopathies. Cold Spring Harb. Perspect. Med.
Fucharoen, S. and Weatherall, D.J. (2012). The hemoglobin E tha-
2: a015032.
lassemias. Cold Spring Harb. Perspect. Med. 2: a011734.
Lettre, G. and Bauer, D.E. (2016). Fetal haemoglobin in sickle-cell dis-
Schrier, S.L. (2002). Pathophysiology of thalassemia. Curr. Opin. Hema-
ease: from genetic epidemiology to new therapeutic strategies. Lancet
tol. 9: 1231–1226.
387: 2554–2564.
Chapter 15 Molecular pathogenesis of malaria
David J. Roberts1 , Arnab Pain2,3 & Chetan E. Chitnis4
1 National Health Service Blood and Transplant (Oxford), John Radcliffe Hospital, Oxford, UK
2
Biological and Environmental Sciences and Engineering (BESE) Division, King Abdullah University of Science and Technology, Jeddah, Saudi Arabia
3 Nuffield Division of Clinical Laboratory Sciences (NDCLS), The John Radcliffe Hospital, University of Oxford Headington, Oxford, UK
4
Malaria Group, Pasteur Institute, Paris, France
Introduction, 193
Epidemiology of malaria, 193
Life cycle of malaria, 195
Invasion of red blood cells, 196
Structural basis for the interaction of EBPs with host receptors, 198
Reticulocyte-binding proteins, 199
Apical membrane antigen 1 and rhoptry proteins, 199
Introduction
Malaria is a major global public health problem and it
is estimated that it is responsible for over 400 000 deaths
annually and 160 million infections. The vast majority of
morbidity and mortality from malaria is caused by infection
with Plasmodium falciparum, although Plasmodium vivax,
Plasmodium ovale curtisi, Plasmodium ovale wallikeri and
Plasmodium malariae also cause infection and illness. In a
few areas, Plasmodium knowlesi, Plasmodium cynomolgi and
Plasmodium simium can cause zoonotic infection and illness.
The pathology of malaria is entirely related to the growth and
development of the parasite in red blood cells, and indeed a
significant proportion of the mortality and morbidity is due
to anemia. Malaria could therefore quite reasonably, if some-
what unconventionally, be considered the greatest single
cause of years of lost life due to a hematological disease. A
combination of insecticide-treated bed nets, highly effective
artemisinin-based drug regimens, and better medical care
has greatly reduced deaths from malaria in the last decade.
The malarial parasite has a complex life cycle, alternating
between humans and the female Anopheles mosquito, a liver
and blood stage of growth, complex eukaryotic metabolic
systems, and a panoply of mechanisms to evade protective
host responses. The vast landscape of interactions between
humans, parasites, and mosquitoes has been illuminated by
genomics and genetic epidemiology. A short review can only
highlight some of the many key molecular processes involved
in the growth, pathology, and prevention or treatment of
Molecular Hematology, 4th Edition. Edited by Drew Provan and John G. Gribben.
© 2020 John Wiley & Sons Ltd. Published 2020 by John Wiley & Sons Ltd.
Merozoite surface protein 1 and invasion, 199
Adhesion of infected red blood cells to host cells, 200
Malarial anemia, 202
Ineffective erythropoiesis in malaria, 204
Future prospects, 205
Further reading, 206
malaria. In this chapter we examine the basic features of
malaria the parasite and malaria the disease, the molecular
aspects of the invasion of red blood cells, and the adhesion of
infected red blood cells to host cells and receptors and their
role in pathogenesis. We also consider the pathology of ane-
mia, which is a prominent feature of the disease, and outline
drug treatment and vaccine development for malaria, as well
as genomic approaches to the study of the parasite, pathology,
and prevention of the disease.
Epidemiology of malaria
Approximately 1000 million people live in areas where
malaria parasites are endemic. According to the World
Health Organization (WHO), 161 million clinical cases of
malaria were reported worldwide in 2016, giving rise to
430 000 deaths per year; this represents a 45% decrease in
mortality since 2000. These spectacular statistics are partly
due to the massive use of artemisinin-based combination
therapies (ACTs) as a first-line treatment (Figure 15.1).
There is substantial evidence that the incidence of severe
disease, particularly in East and West Africa, is falling, due
to the use of artemisinin combination treatment, leading to
the recent award of the Nobel Prize to Youyou Tu for the
discovery of artemisinin, and impregnated bed nets as part
of the Gates Foundation and WHO initiatives (available
at https://www.gatesfoundation.org/What-We-Do/Global-
Health/Malaria and https://rollbackmalaria.com). Because
of these recent advances, the WHO announced support
of a more ambitious goal: progressive malaria elimination
in endemic countries, leading to its complete eradication.
Although there is understandable and appropriate interest
193
194 Molecular Hematology

Fig. 15.1 Global malaria endemicity. Areas are colored according to malaria endemicity (prevalence): light blue, hypoendemic (areas in which
childhood infection prevalence is less than 10%); medium blue, mesoendemic (areas with infection prevalence between 11% and 50%); dark
blue, hyperendemic and holoendemic (areas with an infection prevalence of 50% or more). Gray areas are a combined mask of areas outside of
the transmission limits and areas of population density less than 1 person per km2 .
Source: Snow R.W., Guerra C.A., Noor A.M., Myint H.Y., Hay S.I. (2005). The global distribution of clinical episodes of Plasmodium falciparum
malaria. Nature 434: 214–217, with permission.

in the molecular aspects of the disease and new therapies, it
remains true that in endemic areas the disease is associated
with poverty and poor access to established treatments or
simple preventive measures.
Malaria occurs when the mean temperature is above 20 °C
for P. falciparum and 15 °C for other human malarias, per-
mitting development of the mosquito stages of development
(sporogony). Transmission rarely occurs above 1500 m, in
arid regions, or in the Central and South Pacific, where suit-
able vectors are absent. Moreover, P. vivax malaria is rare in
Africa, where the population frequency of the blood group
Duffy negative (Fy(a)- Fy(b)-) is high. In malaria-endemic
areas, the pattern of clinical malaria is determined by the
number and timing of infective mosquito bites. Mosquito
numbers increase after rainfall, leading to an increase in new
infections; in naı̈ve individuals, parasites can cause chronic
infection lasting many months, either as a result of long blood
stage infection or repeated waves of blood stage parasitemia
from activation of dormant liver-stage hypnozoites in P. vivax
and P. ovale.
The intensity of transmission determines the distribution
of clinical symptoms in different age groups. In general,
in areas of high transmission younger children suffer from
severe disease; as transmission decreases, older children suf-
fer from severe disease. When transmission is sporadic, the
population can lose much of its natural immunity, and a sud-
den increase in mosquitoes can result in epidemics where
large numbers fall ill in all age groups.
The clinical presentation of malaria infection is also wide
and influenced by age, immune status, and pregnancy and
by the species, genotype, and perhaps the geographical ori-
gin of the parasite. In endemic areas, many infections present
as an uncomplicated febrile illness. In more severe forms of
the disease, children may present with prostration or inabil-
ity to take oral fluids, or in younger children an inability to
suckle. Alternatively, they may exhibit several syndromes of
severe disease, including anemia, coma, respiratory distress,
and hypoglycemia, and may also have a high rate of bac-
teremia, due to susceptibility to coincident bacterial infec-
tion. In most age groups, anemia is frequently accompanied
by more than one syndrome of severe disease, and the already
substantial case fatality rate of 15–20% for severe malaria in
African children rises significantly when multiple syndromes
of severe disease are present.
The age distribution of anemia and other syndromes of
severe disease is a consistent, but puzzling feature, of the
epidemiology of clinical malaria. Children born in endemic
areas are protected from severe malaria in the first six months
of life by the passive transfer of maternal immunoglobulins
and by fetal hemoglobin. Beyond infancy, the most com-
mon presentation of severe disease changes from anemia
in children aged one to three years, in areas of high trans-
mission, to cerebral malaria in older children, in areas of
lower transmission. As transmission intensity declines fur-
ther, severe malaria is most frequently found in older age
groups.

There is scant evidence to explain the relationships
between age, transmission, and predominant clinical syn-
drome of severe malaria. One possible explanation is the age-
dependent increase in expression of complement receptor 1
(CR1) on red blood cells, and so a parallel age-dependent
increase in the capacity to inactivate complement compo-
nents absorbed or deposited directly onto the surface of the
red blood cell. Another explanation, by no means mutually
exclusive, is that erythropoietin (EPO) has a powerful cyto-
protective effect on neural, cardiac, and other tissues and,
indeed, increased levels of EPO are found in younger chil-
dren with anemia and with malaria. A clinical study showed
that there was a significant protective effect of high levels of
endogenous EPO on survival after malaria-induced coma,
particularly in children under two years old.
In endemic areas, even today, it is believed that selection
for malaria-protective alleles has taken place over the last
5000 years, after the widespread introduction of agriculture
allowed significant increases in population density, transmis-
sion, and parasite virulence. Sickle cell trait and thalassemia
traits and allele for glucose-6-phosphate dehydrogenase
(G6PD) protect from infection and are truly polymorphic
characteristics in many parts of the world. Understanding
the genetic epidemiology has provided classic examples of
the principle of genetic selection in vivo, for example bal-
ancing selection for sickle cell trait and negative epistasis for
sickle cell trait and α thalassemia. The homozygous forms
of these disorders cause significant clinical disease, such as
sickle cell disease, β thalassemia, and G6PD deficiency. In
endemic areas these genetic diseases represent major public
health problems.
Life cycle of malaria
In P. falciparum malaria, the infective sporozoite forms are
inoculated from the salivary glands of a female Anophe-
les mosquito into the subcutaneous tissues and sometimes
the bloodstream (Figure 15.2). These thin, needle-shaped
cells, 10–12 μm in length, circulate briefly with a half-life
of approximately 30 min, traversing sinusoidal Kupffer cells
and several hepatocytes before residing in a single hepato-
cyte, requiring sporozoite microneme protein essential for
cell traversal (PfSPECT) or perforin-like protein 1 (PfPLP1).
Invasion depends, at least partly, on the thrombospondin
domains on the circumsporozoite protein (CSP) and the
thrombospondin-related adhesive protein (TRAP). CSP and
TRAP bind to heparan sulfate proteoglycans on the surface
of hepatocytes. Within these cells rapid multiplication takes
place over 4–10 days to produce a liver schizont containing
about 30 000 merozoites. When ready to leave, the parasite
Molecular pathogenesis of malaria 195
A E
Liver
B D
Asexual
Growth
C
Adhesion to endothelial
and other host cells
Key:
RBC Merozoites Ring Trophozoites/
Schizonts
Fig. 15.2 Life cycle of Plasmodium. (a) The asexual life cycle begins
when sporozoites from a female mosquito taking a blood meal enter
the circulation and invade hepatocytes. (b) Up to 30 000 merozoites
are formed. The hepatocyte becomes detached and migrates to the
liver sinusoid, where budding of parasite-filled vesicles or
“merosomes” occurs. Following rupture of the merosome, infective
merozoites are released and invade erythrocytes (RBC). (c) Within
RBCs, the parasite develops through the stages of rings, trophozoites,
and schizonts. Mature schizonts burst to release erythrocytic
merozoites, which invade new RBCs. (d) A small proportion of
merozoites in RBCs transform into male and female gametocytes,
which are ingested by the mosquito. (e) The male and female gametes
fuse and transform into an oocyst, which divides asexually into many
sporozoites that migrate to the salivary gland, from where they are
released during the next blood meal.
induces hepatocyte cell death, causing the release of mero-
zoites in membrane-enclosed structures or “merosomes” that
are extruded from the infected cell, so avoiding host-cell
defense mechanisms.
The merozoites bind and then invade red blood cells
(see later). The host plasma membrane is invaginated to
form the parasitophorous vacuole. Immediately after inva-
sion, the developing parasites appear as fine “ring forms.”
Between 10 and 15 hours, the cytoplasm appears to thicken,
and 16 hours after invasion granules of the black pigment
hemozoin, the product of hemoglobin digestion, begin to
appear. At this time receptors are expressed at the surface
196 Molecular Hematology
of the infected red blood cell that mediate adhesion to host
ligands not only on venular endothelium, but also on pla-
centa, uninfected red blood cells, platelets, and dendritic
cells, and many of these interactions have been associated
with pathology. These mature infected red blood cells or
trophozoites are sequestered in tissue beds as they adhere to
post-capillary venular endothelium. Nuclear division begins,
at about 30 hours, to form schizonts containing up to 32
merozoites. At 48 hours the red blood cell is ruptured to
release the merozoites into the circulation to continue fur-
ther cycles of asexual multiplication.
The erythrocytic cycle of growth may yield a 10-fold
increase in parasitemia in vivo and a patent or microscop-
ically detectable infection six days after the liver stage is
completed. After two or more cycles, the infection is often
manifested by the paroxysms of fever that accompany the
release of merozoites. Growth continues until parasite multi-
plication is reduced by chemotherapy, specific or non-specific
defense mechanisms, or occasionally the demise of the host.
When fully grown, some merozoites become committed
during the previous erythrocytic cycle to form male or female
gametocytes, which are distinguished by dispersed pigment
in a single nucleus. They are sequestered for the first five
days of their development in the peripheral circulation, but
later appear as circulating, mature, crescent-shaped gameto-
cytes. The sexual phase of the parasite life cycle begins after
male and female gametocytes are ingested by a feeding female
Anopheles mosquito. In the midgut of the mosquito the game-
tocytes shed the red blood cell membrane, which appears to
be precipitated by a drop in ambient temperature. A female
gametocyte forms a single macrogamete, but male gameto-
cytes undergo several rounds of nuclear division to produce
flagellated motile microgametes, which migrate to fertilize a
macrogamete. The resulting zygote forms a mobile ookinete
and migrates through the epithelial wall of the midgut to
rest on the external surface. The oocyst divides repeatedly to
form about 10 000 sporozoites, which enter the acinar cells
of the salivary glands and are infective when injected into the
human host.
In P. vivax and P. ovale infections, some sporozoites enter-
ing the liver form dormant forms or hypnozoites, which only
begin to divide some months later to cause further relapses.
P. ovale is now officially two parasites – P. ovale wallikeri and
P. ovale curtisi. In P. malariae, erythrocytic development takes
72 hours and thus causes quartan fever (i.e. on days 1 and 4).
A fifth human malaria parasite has been identified in
Southeast Asia. P. knowlesi, whose normal hosts are long-
tailed and pig-tailed macaques, has been shown to be a major
cause of malaria in Malaysia – particularly on the island of
Borneo. This example of a zoonotic malaria should not be
surprising. Malaria is a historic zoonosis: both P. falciparum
and P. vivax became human parasites after independent
host switches from African primates. Further examples of
malaria infection from non-human primates may follow.
Indeed, a zoonotic outbreak of Plasmodium simium parasites
in humans in Brazil and a first case of naturally infected Plas-
modium cynomolgi in human have been described recently.
Invasion of red blood cells
The invasion of erythrocytes by malaria parasites is a com-
plex process and requires many specific molecular interac-
tions. These interactions are self-evidently required for the
survival of the parasite, and it would seem likely that the par-
asite proteins involved in these interactions would be promis-
ing candidates for malaria vaccines.
Erythrocyte invasion by malaria parasites is a multistep
process. After initial binding with an erythrocyte, the mero-
zoite reorients itself so that its apical end faces the erythro-
cyte. A junction forms between the apical surface of the
merozoite and the erythrocyte, and several apical organelles
such as the rhoptries, micronemes, and dense granules dis-
charge their contents to create an invagination in the erythro-
cyte surface. As the merozoite moves into this invagination,
the junction at the apical end transforms into a circumferen-
tial ring that moves around the merozoite and pinches closed
behind the merozoite when invasion is complete.
The receptors that mediate the invasion of red blood cells
are found in the rhoptries, micronemes, and on the cell sur-
face of merozoites. The intracellular location of many of
these parasite receptors and a rapid invasion process may, at
least in part, allow the parasite to evade protective immune
responses.
Use of host ligands during red blood
cell invasion
Plasmodium vivax and knowlesi are completely dependent
on interaction with the Duffy blood group antigen for the
invasion of human red blood cells. Electron microscopy
studies have shown that the interaction of P. knowlesi, and
by analogy P. vivax, merozoites with the Duffy antigen as
ligand is necessary for junction formation during invasion.
This blood group antigen is a 38-kDa glycoprotein with
7 putative transmembrane segments and 66 extracellular
amino acids at the N-terminus (Figure 15.3). The Duffy
antigen serves as a receptor for a family of proinflammatory
cytokines that includes the chemokines interleukin (IL)-8,
melanoma growth-stimulating activity, and monocyte
chemotactic peptide 1. Indeed, IL-8 and melanoma growth-
stimulating activity inhibit the invasion of Duffy-positive
human erythrocytes by P. knowlesi, with 50% inhibition
achieved at nanomolar concentrations in vitro, suggesting
that it may be possible to block the critical step of junction
formation during invasion. The binding site for P. vivax and

Parasite (protein)
P. vivax DBL
(PvDBP)
P. falciparum DBL DBL
(EBA-175)
P. falciparum
M1/2
(AMA-1)
P. vivax
(RBP-1)
P. falciparum
(Pf-NBP1)
P. vivax
(Pv-RBP2b)
P. falciparum
(Pf-RH4)
P. falciparum
(Pf-RH5)
Fig. 15.3 Apical protein families involved in erythrocyte invasion. Bar diagrams represent the various protein families found in apical
organelles of Plasmodium species merozoites that mediate erythrocyte invasion. Multiple P. falciparum PvRBP-2 orthologs, referred to as PfRH1,
PfRH2a, PfRH3, PfRH4, and PfHR5, have also been identified.
P. knowlesi has been mapped to a 35-amino-acid segment
of the extracellular region at the N-terminus of the Duffy
antigen. Sulfation of tyrosine 41 within this region of the
Duffy antigen is critical for recognition by P. vivax and
P. knowlesi.
Unlike P. vivax and P. knowlesi, P. falciparum does not use
the Duffy antigen as a ligand during invasion. Instead, stud-
ies have shown that sialic acid residues of glycophorin A are
ligands for P. falciparum during invasion. A 175-kDa P. fal-
ciparum protein, also known as erythrocyte binding antigen
(EBA)-175, binds sialic acid residues on glycophorin A dur-
ing invasion. EBA-175 does not bind En(a)- erythrocytes,
which lack glycophorin A, but retain glycophorin B and other
sialoglycoproteins, suggesting that sialic acid residues and
the peptides from glycophorin A are required for binding.
A 63-amino-acid tryptic glycopeptide from the extracellular
region of glycophorin A, which blocks binding of EBA-175 to
human erythrocytes, serves as the binding site. It is possible
that EBA-175 makes contact directly with the peptide back-
bone of glycophorin A, or that the peptide backbone presents
sialic acid residues in the correct conformation for binding.
A novel malaria resistance locus was identified close to the
cluster of genes encoding glycophorins, and a haplotype at
this locus provides 33% protection against severe malaria.
The protective allele was not immediately obvious, and a
series of careful genetic analyses demonstrated complex copy
number variants at this locus. Genome sequencing defined 8
deletions and 8 duplications, as well as 11 singleton variants.
The combined allele frequency of glycophorin copy number
variants in African populations was 11% compared to 1.1%
in non-African populations. One of the variants, DUP4, was
associated with a 40% decreased risk of both cerebral malaria
Molecular pathogenesis of malaria 197
Host receptors
Duffy blood group
Glycophorin A
Unknown
Reticulocytes
Trypsin-resistant and neuraminidase-
sensitive receptor on red cells
Transferrin receptor
CR1
Basigin
and severe malarial anemia. The allele encodes a protein
where the extracellular domain of GYPB joins the transmem-
brane and intracellular domains of GYPA, creating a peptide
sequence at their junction that defines the Dantu antigen in
the MNS blood group system. Parasite growth has previously
been shown to be impaired in Dantu+ cells, consistent with
the new epidemiological findings that the Dantu blood group
antigen is associated with protection from malaria.
While P. vivax is completely dependent on a single path-
way for invasion, P. falciparum uses multiple invasion path-
ways and is not confined to using sialic acid residues of
glycophorin A. Some P. falciparum laboratory strains use
sialic acid residues on glycophorin B as ligands. Other strains
are capable of invading neuraminidase-treated erythrocytes
using sialic acid–independent ligands.
Field isolates collected from patients with malaria com-
monly use alternative invasion pathways independent of
sialic acid residues of glycophorin A, and recently the Ok
blood group antigen, basigin, has been identified as a recep-
tor for PfRh5, a parasite ligand that is essential for blood
stage growth. These multiple invasion pathways may provide
P. falciparum with a survival advantage when faced with host
immune responses or receptor heterogeneity in natural infec-
tions, and pose a daunting challenge for vaccine design.
Parasite receptors used during invasion
Parasite proteins that bind erythrocyte receptors during
invasion belong to the erythrocyte-binding protein (EBP)
family (Figure 15.3). The EBP family includes multiple
P. vivax and P. knowlesi Duffy-binding proteins (PvDBP
and PkDBP), P. knowlesi β and γ proteins, which bind
198 Molecular Hematology
Duffy-independent receptors on rhesus erythrocytes, and
P. falciparum EBA-175 and EBA-140.
The extracellular region of each EBP contains two con-
served cysteine-rich domains, regions II and VI, which
contain conserved cysteines and hydrophobic amino acid
residues. P. falciparum EBA-175 contains a tandem dupli-
cation in region II of the N-terminal cysteine-rich region.
The functional binding domain of each EBP maps to region
II. Region II of PvDBP binds the human Duffy antigen,
region II of PkDBP binds both human and rhesus Duffy
antigens, P. knowlesi β region II binds sialic acid residues
on rhesus erythrocytes, and P. knowlesi γ region II binds
as yet unidentified Duffy-independent receptors on rhesus
erythrocytes. A member of the reticulocyte binding protein
(RBP) family, normocyte binding protein Xa (NBPXa), is
responsible for human red cell infectivity in P. knowlesi. In
the case of P. falciparum EBA-175, region II binds sialic acid
residues on human glycophorin A.
The EBA-140 ligand is a paralogue of P. falciparum
EBA-175 protein. They share homology of domain struc-
ture, including region II, with two homologous F1 and F2
domains, and are responsible for ligand–erythrocyte receptor
interaction during invasion. Recombinant EBA-140 region
II binds to the endogenous and recombinant glycophorin C,
but does not bind to Gerbich-type glycophorin C nor to gly-
cophorin D, the truncated form of glycophorin C which lacks
the N-glycan, or to desialylated GPC.
Structural basis for the interaction of
EBPs with host receptors
Regions containing binding residues have been mapped to
domains of PvDBP and P. knowlesi β protein, which contain
approximately 350 amino acids with 12 conserved cysteines.
Chimeric binding domains containing stretches of P. vivax
region II fused to stretches of P. knowelsi β region II have
been expressed and used to define binding residues for
the Duffy antigen; these residues map to a 170-amino-acid
stretch between the fourth and seventh cysteines of P. vivax
region II, and binding residues for sialic acid map to a
53-amino-acid stretch between the fourth and fifth cysteines
of P. knowlesi β region II. EBPs are localized in micronemes,
not on the surface of merozoites. Invasion studies using
another Apicomplexan parasite, Toxoplasma gondii, have
shown that discharge of microneme proteins such as MIC2
is regulated by cytoplasmic-free calcium. It is possible that
cytoplasmic calcium levels may serve as a second messenger
to trigger release of EBPs to the merozoite surface from
micronemes during invasion. The external signals that
trigger microneme release in Apicomplexan parasites during
host cell invasion are not yet known.
The ability to transfect malaria parasites has allowed func-
tional analysis of parasite proteins. Replacement of the EBA-
175 gene with a mutant copy encoding a truncated EBA-175
protein without region VI and the transmembrane and cyto-
plasmic segments has no effect on localization of EBA-175 in
micronemes. Disruption of the EBA-175 gene in the P. falci-
parum strain Dd2/NM, which uses sialic acid–independent
invasion pathways, has no effect on invasion or growth rates.
However, when the EBA-175 gene is disrupted in the P. fal-
ciparum strain W2Mef, which is dependent on sialic acid
residues on glycophorin A, invasion is reduced. P. falci-
parum W2Mef transfectants expressing truncated EBA-175
use a sialic acid–independent invasion pathway, demonstrat-
ing that P. falciparum can switch to alternative invasion path-
ways. The ability to switch to alternative invasion pathways
may allow P. falciparum to escape intervention strategies that
target EBA-175.
Region II of EBA-175 and PvDBP have been expressed as
secreted functional proteins that bind erythrocytes with the
correct specificity. Antibodies directed against region II of
PvDBP block the binding of PvDBP to erythrocytes, and anti-
bodies to region II of EBA-175 inhibit the invasion of ery-
throcytes by P. falciparum in vitro, providing support for their
development as malaria vaccines. Somewhat surprisingly,
high-affinity antibodies to EBA-175 inhibit invasion by P. fal-
ciparum strains that use alternate sialic acid– glycophorin A–
independent invasion pathways. Antibodies directed against
EBA-175 region II may inhibit invasion via such other path-
ways by steric hindrance after binding to EBA-175 at the api-
cal surface. Given the limited homology between region II
of EBP homologs, it is unlikely that antibodies raised against
EBA-175 will cross-react with its homologs to block receptor
binding.
The crystal structures of PvDBP region II (PvRII) and
region II (F1–F2) of EBA-175 (EBA175-RII), together with
the identification of binding residues, reveal interesting fea-
tures of receptor recognition by these diverse DBL domains.
While the three-dimensional structures of these domains
are similar, EBA175-RII forms dimers, whereas PvRII is
monomeric. The receptor recognition site on PvRII is fully
exposed, whereas four sialic acid–binding sites lie in chan-
nels at the dimer interface in the case of EBA175-RII. Signif-
icant conformational changes are predicted to be necessary
for receptor engagement by EBA175-RII. Regions of poly-
morphic clusters map to the face opposite the binding site,
suggesting that the receptor recognition site is not under
selective immune pressure upon natural infection. Indeed,
in a field study conducted in children residing in a malaria
endemic region of Papua New Guinea, high-titer binding
inhibitory antibodies against PvRII were rare. However, once
children developed anti-PvRII binding inhibitory antibodies,
they were strain transcending and protected against P. vivax
infection. These observations support the development of a
 Molecular pathogenesis of malaria 199

vaccine for P. vivax malaria based on region II of P. vivax binding, and provide a molecular basis for the restriction of
Duffy-binding protein. P. falciparum to its human host among other primates.

Reticulocyte-binding proteins
Apical membrane antigen 1 and
rhoptry proteins
Plasmodium vivax preferentially invades reticulocytes, which
account for only about 1–4% of circulating red blood cells.
Apical membrane antigen 1 (AMA-1) is a well-characterized
Two P. vivax reticulocyte-binding proteins (PvRBP-1 and
rhoptry protein and a leading malaria vaccine candidate.
PvRBP-2) have been identified and are localized to the apical
Sequence analysis of AMA-1 from different primate and
surface of P. vivax merozoites (Figure 15.3). They may there-
rodent malaria parasites reveals the presence of 16 conserved
fore be responsible for the recognition of reticulocytes by P.
cysteine residues. The pattern of disulfide linkages formed
vivax during invasion. The Duffy antigen is expressed on the
by these cysteine residues in P. falciparum AMA-1 has
surface of mature erythrocytes as well as reticulocytes, but an
been experimentally determined. Based on the pattern of
irreversible junction forms only when the P. vivax merozoite
disulfide linkages, AMA-1 can be divided into three domains
encounters a reticulocyte.
(Figure 15.3). Regions M1 and M2 from the N-terminal
Now, transferrin receptor 1 (TfR1) has been identified
cysteine-rich regions of MAEBL share homology with the
as the receptor for P. vivax reticulocyte-binding protein 2b
first two domains of AMA-1. AMA-1 is detected in the
(PvRBP2b). TfR1 mutant cells deficient in PvRBP2b bind-
rhoptries of mature merozoites in late-stage schizonts, is
ing were refractory to the invasion of P. vivax, but not to
proteolytically processed at the time of schizont rupture,
the invasion of P. falciparum. PvRBP2b monoclonal antibod-
and is subsequently evenly distributed over the merozoite
ies that can inhibit reticulocyte binding also block P. vivax
surface. Anti-AMA-1 antibodies inhibit erythrocyte inva-
entry into reticulocytes, providing concrete evidence that the
sion, and immunization with purified AMA-1 provides
TfR1–PvRBP2b invasion pathway is critical for the recogni-
protection against parasite challenge. Furthermore, the
tion of reticulocytes during P. vivax invasion.
gene for P. falciparum AMA-1 cannot be knocked out using
Multiple P. falciparum PvRBP-2 orthologs, referred to as
gene-replacement methods. This evidence suggested that
PfRH1, PfRH2a, PfRH3, PfRH4, and PfHR5, have also been
AMA-1 plays a critical, non-redundant role in invasion.
identified (Figure 15.3). Like PvRBP-2, the P. falciparum
Invasion of merozoites into a host red cell requires
orthologs are localized at the apical surface of merozoites and
formation of an electron dense ring at the parasite–host cell
bind erythrocyte receptors to mediate invasion. Erythrocyte
interface called the moving junction (MJ), through which
binding has been established for PfRH1, PfRH2, and also for
the parasite inserts itself into the cell in an active motile
PfRH4, which binds to CR1.
process, akin to the migration of cells over a surface. In
However, the most exciting development has been the
the MJ, AMA1 binds to its receptor, the parasite-encoded
identification of the host receptor for the P. falciparum
rhoptry neck protein (RON) complex, which is targeted to
ligand PfHR5. A series of elegant experiments revealed
the host cell membrane from merozoite organelles early in
an unexpected pathway for red cell invasion. Erythrocyte
invasion. Crystal structures of AMA1 have shown that a par-
invasion was inhibited by soluble basigin or by knocking
tially mobile loop, termed the DII loop, forms part of a deep
down basigin, and invasion could be completely blocked
groove in domain I and overlaps with the RON2 binding site.
using low concentrations of anti-basigin antibodies across all
Rhoptry proteins also play an important role in the forma-
laboratory strains. Furthermore, Ok(a-) erythrocytes, which
tion of new permeability pathways (NPPs) in Plasmodium-
express a basigin variant that has a weaker binding affinity
infected erythrocytes, which allow nutrient uptake. RhopH2
for PfRh5, had reduced invasion efficiencies. However, it is
interacts with RhopH1, RhopH3, the erythrocyte cytoskele-
puzzling that Ok blood group antigens are not obviously
ton, and exported proteins involved in host cell remodeling. If
polymorphic in malaria endemic regions. Nevertheless,
RhopH2 expression is reduced, uptake of nutrients and vita-
these findings have provoked renewed interest in further
mins and parasite growth are reduced.
explorations of PfRh5 as a vaccine candidate antigen. A
series of papers has demonstrated that recombinant PfRh5
can induce antibodies in humans that block the invasion of
red blood cells by merozoites, and a recombinant chimeric Merozoite surface protein 1 and
antibody (Ab-1) against basigin inhibited the PfRH5–basigin invasion
interaction and blocked erythrocyte invasion across multiple
parasite strains. Intriguingly, binding of PFRh5 to basigin Merozoite surface protein (MSP)-1 was the first merozoite
may be responsible for the species specificity of PfRH5 surface protein to be identified and has been extensively
200 Molecular Hematology

investigated. Homologs of MSP-1 have been identified from

a number of primate malaria parasites, as well as several
Adhesion of infected red blood cells to
rodent malaria parasites. Sequence analysis shows that,
host cells
within a species, MSP-1 contains conserved and more poly-
Plasmodium falciparum is distinguished from the other
morphic regions. Interestingly, MSP-1 undergoes extensive
human malarias by the adherence of a very high propor-
proteolytic processing, either during schizogeny or soon after
tion of the more mature infected red blood cells, containing
merozoite release. Proteolysis results in four proteolytic frag-
asexual parasites, and the first stages of gametocytes to post-
ments of 83 kDa (N-terminus), 30 kDa, 38 kDa, and 42 kDa
capillary venular endothelium. The host ligands and parasite
(C-terminus). These polypeptides remain non-covalently
receptors that mediate the sequestration of ring-stage para-
associated at the merozoite surface. The C-terminal
sites have been widely studied, as there is considerable evi-
42-kDa fragment is attached to the plasma membrane of the
dence that the adhesion of infected red blood cells to some
merozoite by a glycosylphosphatidylinositol (GPI) anchor.
host ligands is associated with the development of severe
Final processing steps cleave the 42-kDa fragment into a
clinical disease and evasion and indeed modulation of the
33-kDa fragment and then into a 19-kDa fragment. Because
immune system.
MSP-1 is evenly distributed over the merozoite surface, it
Sequestration of infected red blood cells begins with a
is suggested that MSP-1 may mediate the initial interaction
similar sequence of events to that seen during the adhesion
with the erythrocyte. Full-length P. falciparum MSP-1 binds
of leukocytes to endothelium. Using a flow cell chamber to
erythrocytes in a sialic acid–dependent manner.
mimic the shear forces believed to be present in vivo, infected
Antibodies directed against MSP-119 block erythrocyte
red blood cells can be seen to roll across the substrate (indi-
invasion, and immunization with recombinant MSP-119 pro-
cating receptor–ligand interactions with very rapid on and off
tects mice as well as monkeys against parasite challenge, pro-
rates). Most host receptors are involved with tethering and
viding support for a blood-stage vaccine based on MSP-1.
rolling, but only two receptors, CD36 and chondroitin sul-
Invasion-inhibitory antibodies directed against P. falciparum
fate A (CSA), support the adhesion of infected red blood cells
MSP-119 block proteolytic processing of MSP-1. Following
under flow.
natural exposure to P. falciparum infections, individuals in
Electron microscopy of post-mortem tissues has shown
endemic areas acquire antibodies that can block the bind-
that parasites are sequestered in various organs, including
ing of processing-inhibitory antibodies to MSP-1. It may be
the heart, lung, brain, liver, kidney, subcutaneous tissues, and
important to design a subunit vaccine based on MSP-119
placenta. Post-capillary venular endothelium and placental
that elicits processing-inhibitory antibodies without eliciting
syncytiotrophoblasts in placenta express a variety of indu-
antibodies that block their binding to MSP-1.
cible and non-inducible receptors. Parasites isolated from
It has not been possible to knock out the P. falciparum
natural infections or laboratory strains can bind in variable
MSP-1 gene using gene-targeting methods, indicating that
numbers to many different receptors. The total number of
MSP-1 is essential for invasion and survival. Moreover, P.
parasites, the specificity and degree of binding to host ligands,
falciparum MSP-119 has limited sequence diversity, possibly
and the expression of these receptors are thought to deter-
because of functional constraints, and this provides a fur-
mine the tissue and organ distribution of parasites, and quite
ther advantage for vaccine development. However, it has been
possibly the clinical syndromes observed in severe malaria.
shown that Plasmodium chabaudi MSP-119 can function-
Adhesion of P. falciparum-infected red blood cells to
ally replace the P. falciparum domain, even though the two
endothelium and other host receptors is mediated by a par-
sequences are highly divergent, suggesting that the sequence
asite protein expressed at the cell surface, the aptly named
of MSP-119 may not be tightly constrained by function.
P. falciparum erythrocyte membrane protein 1 (PfEMP1).
This observation raises the possibility that introduction of
PfEMP1 is encoded by a large var gene family. As the name
a vaccine based on MSP-119 may result in the selection of
suggests, these proteins are highly variable, not only antigeni-
P. falciparum parasites with mutant MSP-119 domains that
cally but also in their ability to bind host ligands. Each par-
are functional, but antigenically divergent, allowing parasites
asite cell contains 60 or so different var genes, but only one
to escape host immune responses. Several trials have used
is expressed at any one time and the cell can rapidly switch
MSP-1 as a candidate vaccine antigen, but immunogenicity
expression from one gene to another at a rate of 2–4% per
has been poor, highlighting some of the problems faced in
cycle, a phenomenon known as clonal antigenic variation.
designed a complex subunit vaccine for a protozoal pathogen.
Isolates of parasites from patients or laboratory lines there-
Erythrocyte invasion by malaria parasites is a multistep
fore include infected red blood cells that contain heteroge-
process that is mediated by highly specific molecular interac-
neous antigenic and adhesive phenotypes.
tions that, once understood, may help in the rational design
PfEMP1 has a complex domain structure (Figure 15.4).
of vaccines that attempt to block erythrocyte invasion and
The extracellular region of PfEMP1 is composed of variable
prevent malaria.

Parasite (protein)
P. falciparum
DBL1α CIDR1α DBL2β C2 DBL3γ DBL4ε DBL5δ
(PfEMP1)
Fig. 15.4 Plasmodium falciparum erythrocyte membrane protein (PfEMP)-1. The extracellular region of PfEMP-1, which is expressed on the
infected red blood cell surface, is composed of variable numbers of five possible DBL domains, named after their homology to the DBL domains
involved in red blood cell invasion, and one or two cysteine-rich interdomain regions (CIDRs).
numbers of different DBL domains, named after their
homology to the DBL domains involved in red blood cell
invasion, and one or two cysteine-rich interdomain regions
(CIDRs). DBL domains have been classified into five different
classes, and the binding domains for several host receptors
have been mapped to various DBL and CIDR domains.
PfEMP1 is involved in many pathogenic processes (Fig-
ure 15.5). Almost all infected red blood cells can bind
to thrombospondin and the platelet glycoprotein CD36,
while a subset of infected cells expressing different mem-
bers of the PfEMP1 family can bind to intercellular adhe-
sion molecule (ICAM)-1, platelet endothelial cell adhesion
molecule (PECAM)-1, P-selectin, αvβ3 integrin, vascular
cell adhesion molecule (VCAM)-1, and, perhaps of relevance
for cerebral malaria, endothelial protein C receptor (EPCR).
Adhesion of infected red blood cells or “rosetting” has been
associated with severe disease, and in different strains may be
mediated by CR1, ABO blood group antigens, glycosamino-
glycans, and/or IgM. Similarly, the ability of infected red
blood cells to adhere to platelets and form large clumps
has also been associated with severe disease and is medi-
ated by the adhesion of infected red blood cells to CD36 or
to the complement receptor gC1qR. The exact relationship
between these molecular adhesive phenotypes and disease is
not entirely clear, but it seems possible that the formation
of multicellular aggregates and/or binding of infected cells
to a ligand expressed on a particular endothelial bed could
contribute to microvascular obstruction and tissue pathology
and organ dysfunction (Figure 15.5).
For example, sequestration of parasites in the brain may
be related to cerebral malaria or coma and may involve
adhesion of infected red blood cells to the ICAM-1 receptor
and/or EPCR (Figure 15.5). Severe childhood malaria is
associated with expression of specific PfEMP1 subtypes
containing domain cassettes (DCs) 8 and 13. EPCR, which
mediates the cytoprotective effects of activated protein C,
has been identified as the endothelial receptor for DC8
and DC13 PfEMP1. EPCR binding is mediated through the
amino-terminal cysteine-rich interdomain region (CIDR1α)
of DC8 and group A PfEMP1 subfamilies, and this CIDRα1
interferes with protein C binding to EPCR. This PfEMP1
adhesive property links P. falciparum cytoadhesion to a host
receptor, EPCR, involved in anticoagulation and endothelial
Molecular pathogenesis of malaria 201
Host receptors
CD36, ICAM-1,
CIDR2β ATS
CSA, etc.
cytoprotective pathways, and may have a significant role on
the pathology of cerebral malaria and aid the development
of new malaria interventions.
However, the closest linkage between tissue-specific adhe-
sion and pathology is for malaria during pregnancy (Fig-
ure 15.5). Pregnant women are more susceptible to malaria
and not only suffer severe acute malaria and chronic anemia,
but also experience premature delivery, low birthweights, and
increased neonatal mortality. Parasitized red blood cells are
sequestered in the placenta in pregnant women and, unlike
isolates from other groups of patients, can bind to CSA but
not to CD36. This adhesive phenotype is linked to expression
of a single PfEMP1 with a DBL domain that binds CSA and a
non-CD36-binding CIDR1, while CD36-adherent parasites
express a PfEMP1 with a CD36-binding CIDR1. The appar-
ently restricted molecular structures associated with placen-
tal malaria suggest that the CSA-binding PfEMP1 may be a
good candidate antigen to prevent malaria in pregnancy.
Immunity to malaria has a major role in controlling
disease. Intriguingly, epidemiological evidence suggests that
immunity to severe forms of malaria may be gained after
one or two episodes of infection, and experimentally this has
been achieved in the absence of an antibody response. Even
after many tens of exposures, people may develop immunity
to disease, but are not refractory to asymptomatic infection.
Again, the immune responses that underlie this form of
immunity are not clear. However, anti-PfEMP1 antibodies
do play some role in protecting against pathogenic infec-
tions. Exposure to parasites that sequester in the placenta
during pregnancy induces strain-transcending immunity
that stops infected red blood cells adhering to CSA and
may protect a multigravid mother and fetus from placental
malaria. During the development of clinical immunity,
particularly during early childhood, strain-specific anti-
bodies to PfEMP1 are important in preventing infection
with previously encountered isolates. It also appears that
virulent isolates from children with severe malaria are not
rare, but, on the contrary, are more commonly recognized
by children’s sera than isolates from cases with mild malaria.
It remains possible that the functionally significant regions
of PfEMP1 that mediate binding to CD36, CSA, or EPCR
have conserved epitopes that can be targeted for vaccine
development.
202 Molecular Hematology

HA CSA TSP ICAM-1 ELAM-1 VCAM-1 CD36 PECAM

P-Sel (CD31)

Placenta Brain Microvasculature

CR1
HS-like GAGs, IgM,
blood group A

Rosetting

Placental malaria Cerebral malaria Microvascular adhesion

CD36
Sequestration

PfEMP1 variant antigens Clumping

CD36

A B C D E F

Dendritic cell
Antigenic distinct waves of parasitaemia
Fig. 15.5 Plasmodium falciparum erythrocyte membrane protein (PfEMP)1 and pathology. PfEMP1 expressed on the surface of mature
red blood cells infected with P. falciparum undergoes clonal antigenic variation and can bind to many host receptors through its multiple adhesion
domains. The different properties of PfEMP1 – sequestration for evading clearance in the spleen and antigenic variation for evading
antibody-dependent killing – contribute to the virulence and pathogenesis of P. falciparum and are essential for survival of the parasite. Parasite
sequestration in the brain and placenta contribute to the complications of cerebral malaria and placental malaria, respectively. Simultaneous
binding to several receptors, binding of uninfected erythrocytes (rosetting), and clumping of infected erythrocytes through platelets are associated
with the pathogenesis of malaria. Parasite-infected red blood cells binding to dendritic cells downregulate the host immune response. CR1,
complement receptor 1; ELAM-1, endothelial/leukocyte adhesion molecule 1; HA, hyaluronic acid; HS-like GAGs, heparin sulfate-like
glycosaminoglycans; P-Sel, P-selectin; PECAM (CD31), platelet endothelial cell adhesion molecule 1; TSP, thrombospondin; VCAM-1, vascular cell
adhesion molecule 1.
Source: Miller L.H., Baruch D.I., Marsh K., Doumbo O.K. (2002). The pathogenic basis of malaria. Nature 415: 673–679, with permission.

The exact relationship between the adhesion of infected

red blood cells and pathology may involve obstruction
Malarial anemia
of the microcirculation, initiation of a local inflammatory
In endemic areas, the etiology of anemia is complex. In
response, and/or endothelial activation and immune mod-
children, acute or chronic malaria infection is a major pre-
ulation. Understanding such processes may define possible
cipitating factor of severe anemia causing admission to hos-
routes for therapy of severe disease.
pital. Not infrequently, patients present with a hemoglobin
 Molecular pathogenesis of malaria 203

level of less than 5 g dl−1 with or without respiratory distress
secondary to metabolic acidosis. The sudden appearance
of hemoglobin in the urine, indicating severe intravascular
hemolysis leading to hemoglobinemia and hemoglobinuria
(so-called blackwater fever), was described in early studies of
anemia in expatriates living in endemic areas, but has been
more common in Southeast Asia and Papua New Guinea,
where some cases are associated with G6PD deficiency
and treatment with a variety of drugs including quinine,
mefloquine, and artesunate. However, some recent cases
of severe intravascular hemolysis have been identified in
Africa, prompting speculation that hemolysis may be linked
to the use of artemisinin derivatives.
The anemia of P. falciparum malaria is typically nor-
mocytic and normochromic, with a notable absence of
reticulocytes. The anemia of malaria may be accompanied
by modest leukocytosis, leukopenia, leukemoid reaction,
monocytosis, lymphocytosis, and thrombocytopenia, but
these changes are neither in themselves diagnostic, nor do
they help guide management. However, malarial pigment
is often seen in neutrophils and monocytes and has been
associated with severe disease and an unfavorable outcome.
Malaria provides ample reasons for both increased red
cell destruction and reduced red cell production (see Fig-
ure 15.6 for an overview). Destruction of red blood cells
is inevitable as parasites complete their 48-hour growth

Clearance of non-iRBC
RSP-2 and specific antibody
RSP-2 Immune complexes Spleen
Loss of deformability
MØ
MØ
Clearance of iRBC

MØ
Clearance of debris
iRBC and debris
Activation of MØ
Pigment and GPI-
containing monocyte
Direct effect of
parasite
Indirect effect by
cytokines and other
– mediators e.g. TNF-αα,
+ MIF, IL-12, and IL-10

+
Anemia +
EPO
Malaria infection
Erythropoiesis

KEY: Reticulocyte RBC iRBC iRBC

Ring stage Schizont stage

Fig. 15.6 Pathogenesis of malarial anemia. Severe malarial anemia is characterized by destruction of infected red blood cells (iRBC) following
schizogony and clearance of both iRBCs and uninfected RBCs. During malarial infection, ring surface protein (RSP)-2 may be deposited on the
membrane, immune complexes may be absorbed, and senescent-type changes may occur. The resultant immune complexes of RBCs, antigen, and
immunoglobulin, e.g. RBC : RSP2 : Ig, are cleared by splenic macrophages (MØ). Pigment-containing macrophages may release inflammatory
cytokines and other biologically active mediators such as 4-hydroxynonenal. Macrophage inhibitory factor (MIF) may be released by macrophages
or a plasmodial homolog may suppress erythropoiesis. Malarial pigment or other parasite products may have a direct inhibitory effect on
erythropoiesis. Inhibition of erythropoiesis may be at one or more sites in the growth and differentiation of hematopoietic progenitors. Both
indirect and direct effects may cause suppression of the bone marrow and spleen, resulting in inadequate reticulocyte counts for the degree of
anemia. EPO, erythropoietin; GPI, glycosylphosphatidylinositol anchors of merozoite proteins; Hz, hemozoin.
204 Molecular Hematology
cycle and lyse their temporary host cell. Some parasites may
be removed from erythrocytes as immature ring forms by
phagocytic cells, leaving the red blood cells with residual par-
asite antigens to continue to circulate, albeit with reduced
survival. Infected erythrocytes may also be phagocytosed by
macrophages following opsonization by immunoglobulins
and/or complement components. Other signals for recogni-
tion of infected erythrocytes by macrophages include abnor-
mally rigid membranes, exposure of phosphatidylserine, and
other altered host antigens.
Changes to uninfected red blood cells also contribute to
their own enhanced clearance by phagocytes. The activ-
ity and number of macrophages are increased in malarial
infection, and increased removal of uninfected cells may
occur. Moreover, the molecular signals for recognition of
uninfected erythrocytes for removal by macrophages are
enhanced. Uninfected erythrocytes bind increased amounts
of immunoglobulin and/or complement, as detected by the
direct antiglobulin test (Coombs test). These antibodies do
not have a specificity, but may instead represent immune
complexes adsorbed onto the surface of red blood cells.
Moreover, red cells from malaria patients are not only more
susceptible to phagocytosis, but also show increased surface
IgG and deficiencies in CR1 and CD55 compared with con-
trols. Uninfected red cells in children and adults with severe
disease are less deformable and this is a significant predictor
of the severity of anemia and indeed outcome, consistent with
the notion that these cells are being removed by the spleen.
Active erythrophagocytosis is a conspicuous feature within
the bone marrow during P. vivax and P. falciparum malaria,
and it is highly probable that this also occurs within the
spleen. Children with acute P. falciparum malaria have high
circulating levels of interferon (IFN)-γ and tumor necrosis
factor (TNF)-α, a synergistic combination of cytokines that
activates macrophages.
Parasite proteins may also contribute to the clearance of
uninfected red blood cells. The ring surface protein (RSP)-2
is expressed shortly after merozoite invasion of red blood
cells and is widely deposited on uninfected red blood cells
by contact with merozoites in the circulation. Opsonization
of these RSP-2-bearing uninfected red blood cells could
contribute to erythrophagocytosis. Indeed, high levels
of antibodies that could facilitate complement-mediated
phagocytosis of cells expressing RSP-2 are found in sera from
immune adults and children with severe anemia. The RSP-2
antigen is also present on the surface of erythroblasts in the
bone marrow of P. falciparum-infected patients and it has
been suggested that damage to developing erythroid cells by
RSP-2 and anti-RSP-2 antibody could also contribute to the
development of anemia.
Thus, all the available evidence points to increased retic-
uloendothelial clearance in P. falciparum malaria, persisting
long after recovery. These changes are presumably a host
defense mechanism, maximizing the clearance of parasitized
erythrocytes.
Ineffective erythropoiesis in malaria
Reticulocytopenia has been confirmed in numerous clinical
studies of malarial anemia. The histopathological study of
the bone marrow of children with malarial anemia shows
erythroid hyperplasia with increased numbers of erythroid
precursors. However, maturation is abnormal as observed
by light and electron microscopy. Abdalla and Weatherall
described the hallmark characteristics of such abnormal
maturation, namely cytoplasmic and nuclear bridging and
irregular nuclear outline. They later confirmed that the
distribution of the erythroid progenitors through the cell
cycle is abnormal in malarial anemia, with an increased
proportion of cells in G2 phase compared with normal con-
trols. Ineffective erythropoiesis also contributes to anemia
in animal models of malaria. A recent study has shown that
vaccinated Aotus monkeys, after a challenge infection, may
develop moderate to severe anemia following rapid clearance
of uninfected erythrocytes but with low reticulocyte counts,
indicating bone marrow dysfunction.
Given the importance of EPO to erythropoiesis, attention
has focused on the levels of this crucial cytokine in malarial
infection. Serum EPO was appropriately raised in a single
study of African children suffering from malarial anemia.
However, other studies in adults from Thailand and Sudan
have suggested that EPO concentration, although raised,
was inappropriate for the degree of anemia. The most recent
studies of EPO in African children have shown a supra-
physiological rise in EPO levels compared with age-matched
community controls with non-malarial anemia, and there
appears to be a defective response to EPO in malaria, at least
in these children.
The prime candidates for the host factors mediating
dyserythropoiesis are imbalances in TNF-α, IFN-γ, and
IL-10. The concentrations of TNF-α and IFN-γ have been
correlated with the severity of the disease. While low con-
centrations of TNF-α (<1 ng ml−1 ) stimulate erythropoiesis,
higher levels of TNF-α have been shown to suppress ery-
thropoiesis. Furthermore, it is possible that high levels of
these inflammatory cytokines may contribute to reduced and
abnormal production of erythrocytes, and also to increased
erythrophagocytosis.
High levels of the Th2 cytokine IL-10 might prevent the
development of severe malarial anemia. Low levels of IL-10
have been described in African children with severe malar-
ial anemia. It has been suggested that IL-10 may induce
heme oxygenase and so reduce oxidative damage to red blood
cells and/or developing erythroid cells. Similarly, IL-12 may
be associated with a protective innate immune response to

malaria, and low IL-12 levels have been associated with severe
malaria in African children.
There is also substantial evidence that the lysate of infected
erythrocytes may directly modulate the function of host
cells. During its blood stage, the malaria parasite proteolyzes
host hemoglobin in an acidic vacuole to obtain amino acids,
releasing heme as a byproduct, which is autoxidized to
potentially toxic hematin: aquaferriprotoporphyrin IX or
H2 O-Fe(III)PPIX. β-hematin forms as a crystalline cyclic
dimer of Fe(III)PPIX and is complexed with protein and
lipid products as malarial pigment or hemozoin. Arese and
colleagues showed that the function of monocytes and of
monocyte-derived macrophages is severely inhibited after
ingestion of malaria pigment or hemozoin. These cells are
unable to repeat phagocytosis and to generate oxidative
burst when appropriately stimulated. Furthermore, after
phagocytosis of hemozoin, myeloid cells are unable to kill
ingested fungi, bacteria, and tumor cells and to respond to
IFN-γ stimulation, but instead respond by increased release
of IL-1β, TNF-α, and macrophage inflammatory protein
(MIP)-1α and MIP-1β.
The hemozoin polymer of heme moieties may be com-
plexed with biologically active compounds. The oxidation
of membrane lipids catalyzed by ferric heme produces
lipoperoxides. There is accumulating evidence that 4-
hydroxynonenal and other lipoperoxides, including 15-
hydroxy-arachidonic acid (15-HETE), may play a role in
the pathophysiology of malaria. Endoperoxides produced
in pigment-containing monocytes or macrophages may
impair erythroid growth. Hemozoin may also directly
inhibit erythroid development in vitro and increased levels
of plasma hemozoin and pigment in monocytes have been
associated with anemia.
Hemozoin may have a direct effect as a proinflammatory
danger signal that activates the NALP3 inflammasome, caus-
ing the release of IL-1β. Similar to other NALP3-activating
particles, hemozoin activity is blocked by inhibiting phago-
cytosis, K(+) efflux, and NADPH oxidase.
A further class of parasite products that may contribute
to malarial anemia are the GPI anchors of merozoite
proteins, such as MSP-1, MSP-2, and MSP-4, which are
found in plasma during infections. GPIs are likely to con-
tribute to anemia, since their injection into mice results
in a transient decrease in the number of circulating red
blood cells, probably through induction of TNF-α from
macrophages. More recently, it has been demonstrated that
the proinflammatory response from human monocytes is
through interaction of GPIs with TLR2 and, to a lesser
extent, TLR4. Antibodies specific to GPIs were present in
the sera of adults from endemic regions in Kenya, but
the level of these antibodies was less in younger children
who, in general, have more severe disease and malarial
anemia.
Molecular pathogenesis of malaria 205
Many possible molecular mechanisms of red blood cell
clearance have been described in a variety of experimental
and clinical systems. The challenges now are to determine
the relative contribution of these mechanisms to anemia in
children with malaria, and to seek ways to prevent anemia
without increasing susceptibility to disease.
Future prospects
Molecular methods and the application of modern pro-
teomic and genomic technologies are likely to transform
our approach to the diagnosis, treatment, and prevention of
malaria. The diagnosis of malaria is based on the identifi-
cation of circulating blood-stage parasites in thick and thin
films. In endemic areas, laboratory staff are skilled at the
examination of thick films and are routinely able to detect 1
parasite in 100 high-power fields of a thick film, which corre-
sponds to a sensitivity of approximately 5–50 parasites μl−1 .
Nevertheless, the diagnosis of malaria by microscopy in
non-endemic countries has proven problematic. Routine
laboratories may only achieve sensitivities of the order of
500 parasites μl−1 using thick films. Detection of circulating
malarial antigens is another potentially attractive, but
ultimately limited, alternative to the laborious method of
screening blood films. The widely available tests detect Plas-
modium histidine-rich protein 2 and Plasmodium-specific
lactate dehydrogenase by immunochromatography. The
formulation of the tests using dipstick antigens allows rapid
testing to be performed by laboratory staff. However, the
sensitivity is 100–1000 parasites μl−1 , which is comparable
to the sensitivity achieved by inexperienced microscopists.
The current recommendations for malaria diagnosis in the
UK emphasize clearly that these tests only have a place
when experienced staff are not available. Amplification of
circulating parasite DNA using the repeated rRNA genes
is an extremely sensitive method of malaria diagnosis.
The sensitivity may be as low as 0.005 parasites μl−1 or
5 parasites ml−1 . This approach may be developed as a future
method for routine diagnostic use.
Malaria requires urgent effective chemotherapy to prevent
the progression of disease and this may be the most crucial
public health intervention to reduce the global mortality
from malaria. Resistance to the first-line treatment of chloro-
quine and to dihydrofolate reductase inhibitors has led to
artemisinin-based combination treatments. This group of
drugs has a remarkable history and represents a novel class
of chemical compounds known as the artemisinins. Chinese
scientists rediscovered the activity of extracts of the plant
Artemisia annua from careful rereading of medieval pharma-
copeias. The active substance was crystallized and identified
as artemisinin in 1973, and derivatives dihydroartemisinin
(DHA), artemether, artesunate, and arteether were first
206 Molecular Hematology
prepared in China in the 1970s. The mechanism of action of
these drugs was proposed to be within the “food vacuole,”
where the drug could inhibit digestion of hemoglobin, but
no single mechanism of action has been defined. In any
event, artemisinin acts very early in the parasite life cycle,
killing ring-stage parasites and so preventing the maturation
and sequestration of malarial-infected erythrocytes. There
has been increased drug resistance, mainly in Southeast Asia,
where drug resistance has been strongly associated with
mutation in the falciparum gene encoding a K13 propeller
mutation, but this mutation has arisen against a background
of a complex genetic phenotype that may have allowed the
parasite to adapt to many different drug regimens and/or
grow in erythrocytes with hemoglobin variants. There are
now isolated case reports of emerging resistance or molecu-
lar markers of resistance to artemisinin-based combination
therapy in Africa, where widespread resistance could reverse
the recent advances in the burden of malaria.
The search for a malaria vaccine could almost be a case
study in the triumph of hope over experience. However,
there have been some recent trials showing significant effi-
cacy. The RTS,S/AS01 vaccine was developed by Joe Cohen
at GlaxoSmithKline and uses a recombinant P. falciparum
circumsprozoite protein, which is naturally expressed in
high levels on the surface of the infective forms injected
by mosquitoes, fused to the hepatitis B surface protein. A
series of trials have shown a reduction in episodes of severe
malaria by 50% in phase 3 studies and Ghana, Kenya, and
Malawi partnered with the WHO in the Malaria Vaccine
Implementation Program (MVIP) to make the RTS,S/ASO1
vaccine available in selected areas of the three countries,
beginning in 2018. This program will assess the feasibility
and safety of the vaccine in real-life settings. There are also
over 40 subunit vaccines in different stages of development,
many of which are based on the antigens described in this
chapter (see www.malariavaccine.org).
This chapter has described the essential features of the epi-
demiology, clinical presentations, and life cycle of the parasite
and the main features of our present molecular understand-
ing of key features of the invasion of red blood cells, adhe-
sion of infected erythrocytes to host ligands, and how these
and other host–parasite interactions lead to cerebral and pla-
cental malaria and anemia. The data revealed by sequencing
the parasite, human, and mosquito genomes are still being
applied, increasing our understanding not only of the patho-
genesis of disease but of the biology of the parasite.
These new approaches to advance the biology of malaria
have been accompanied by a large international effort to
develop new drugs for the disease, which are likely to be
needed, not only if artemisinin drug resistance emerges on
a wide scale, but also to treat the dormant liver stages (or
hypnozoites) of P. ovale and P. vivax. Novel high-throughput
screening platforms, large chemical libraries, and coordi-
nated international effort has allowed the Medicines for
Malaria Venture to assemble 400 unique compounds with
blood-stage antimalarial activity for continued future devel-
opment (see https://www.mmv.org/research-development).
There are considerable scientific, technical, organizational,
and political problems faced in the development and imple-
mentation of new methods and programs for diagnosing,
treating, and preventing malaria. However, the huge pro-
grams now underway would be expected to continue to roll
back the disease. In a remarkable public health achievement,
Sri Lanka was certified as malaria free by the WHO in 2016,
which gives some hope that the eradication of malaria may be
achieved one day, at least in areas where transmission is low
and there is the necessary infrastructure to sustain complex
public health programs.
Further reading
Cowman, A.F., Healer, J., Marapana, D., and Marsh, K. (2006). Malaria:
biology and disease. Cell 167: 610–624.
Howes, R.E., Battle, K.E., Mendis, K.N. et al. (2016). Global epidemiol-
ogy of Plasmodium vivax. Am. J. Trop. Med. Hyg. 95 (6): 15–34.
Kirchner, S., Power, B.J., and Waters, A.P. (2016). Recent advances in
malaria genomics and epigenomics. Genome Med. 8: 92.
Lelliott, P.M., McMorran, B.J., Foote, S.J., and Burgio, G. (2015). The
influence of host genetics on erythrocytes and malaria infection: is
there therapeutic potential? Malar. J. 14: 289.
Rabinovich, R.N., Drakeley, C., Djimde, A.A. et al. (2017). malERA:
an updated research agenda for malaria elimination and eradication.
PLoS Med. 14 (11): e1002456.
Woodrow, C.J. and White, N.J. (2017). The clinical impact of
artemisinin resistance in Southeast Asia and the potential for future
spread. FEMS Microbiol. Rev. 41 (1): 34–48.
Chapter 16 Molecular coagulation
and thrombophilia
Björn Dahlbäck1 & Andreas Hillarp2
1
Department of Translational Medicine, Section of Clinical Chemistry, Lund University, University Hospital, Malmö, Sweden
2
Department of Clinical Chemistry and Transfusion Medicine, Halland County Hospital, Halmstad, Sweden
Introduction, 207
Blood coagulation, 207
Molecular genetics of venous thromboembolism, 211
Introduction
Venous thrombosis is a major medical problem affecting
millions of individuals worldwide each year. It is a typical
multifactorial disease, the pathogenesis involving both
environmental and genetic risk factors. A mutation in the
factor V (FV) gene (Arg506Gln or FV Leiden) is the most
common genetic risk factor known to date. Activated protein
C (APC) regulates the activity of FVa by cleaving several sites
in FVa, and Arg506 is one of them. APC resistance, which
is the consequence of the FV mutation, results in a lifelong
hypercoagulable state. A point mutation in the prothrombin
gene is another relatively common risk factor, whereas
deficiencies of the anticoagulant proteins antithrombin
(AT), protein C, or protein S are less common. Owing to
the high prevalence of the FV and prothrombin mutations,
combinations of genetic defects are relatively common in the
general population. Such individuals have a highly increased
risk of thrombosis.
Blood coagulation
At sites of vascular damage, circulating platelets adhere to
subendothelial structures and undergo a series of reactions
that lead to primary hemostasis due to the formation of a
platelet plug. Concomitant to these events, the subendothe-
lial membrane protein tissue factor (TF) is exposed to blood.
A small amount of activated factor VII (FVIIa) present in cir-
culating blood binds to TF and triggers a series of proteolytic
reactions that culminate in the formation of thrombin and
the conversion of fibrinogen to insoluble fibrin.
Molecular Hematology, 4th Edition. Edited by Drew Provan and John G. Gribben.
© 2020 John Wiley & Sons Ltd. Published 2020 by John Wiley & Sons Ltd.
Management of thrombophilia, 217
Conclusions, 218
Further reading, 218
FVIIa bound to TF specifically cleaves and activates the
two vitamin K-dependent plasma proteins, factor IX (FIX)
and factor X (FX) (Figure 16.1). Activated FX (FXa) activates
prothrombin to thrombin, whereas activated FIX (FIXa) acti-
vates FX. Both FIXa and FXa are poor enzymes that require
protein cofactors, calcium ions, and negatively charged phos-
pholipid surfaces for expression of full biological activity.
The protein cofactors for FIXa and FXa are the activated
forms of factor VIII (FVIIIa) and factor V (FVa), respectively
(Figure 16.2). As a result of multiple protein–protein and
protein–phospholipid interactions, enzymatically highly effi-
cient complexes are assembled on the phospholipid surface.
The initiation of blood coagulation by TF is usually
referred to as the extrinsic pathway or the TF pathway.
In association with injury, this is the physiologically most
important mechanism of blood coagulation. However, coag-
ulation can also be activated through the intrinsic pathway,
triggered by activation of the contact phase proteins (FXII,
FXI, prekallikrein, and high-molecular-weight kininogen)
that follows exposure of blood to certain negatively charged
surfaces. In recent years, it has been discovered that platelets
contain long-chain negatively charged polyphosphates that
may serve as a natural activator for the intrinsic system.
The intrinsic pathway does not appear to be physiologically
important for injury-related coagulation in vivo, illustrated
by the lack of bleeding problems in individuals with a defi-
ciency of FXII. However, the interest in the intrinsic pathway
has increased dramatically in the last decade after the dis-
covery that it may be important for the formation of arterial
thrombosis, and efforts are being made to develop inhibitors
to the system for safe anticoagulation.
Thrombin generated at sites of vascular injury expresses a
number of procoagulant properties. It amplifies the coagu-
lation process by activating FXI, and in addition it activates
platelets and converts fibrinogen to fibrin. Moreover, in a pos-
itive feedback reaction, thrombin converts the procofactors
207
208 Molecular Hematology

VWF

IXa VIII
VIIa IX/X
T
IXa T
IXa T Xa
Xa
VIIIa Va
PT
Xa X Xa
IXa
TF

Extravascular cell Activated platelet

Fig. 16.1 Initiation and propagation of blood coagulation. Blood coagulation takes place on the surface of cell membranes where enzymes
and cofactors form complexes that efficiently convert their respective proenzyme substrates to active enzymes. The exposure of tissue factor (TF) to
blood, with subsequent binding of FVII/FVIIa and activation of FIX and FX, initiates the reaction sequence. The subsequent assembly of tenase
(FIXa/FVIIIa) and prothrombinase (FXa/FVa) complexes on the surface of negatively charged phospholipid membranes (mainly provided by platelets)
results in amplification, propagation of the process, and the generation of high concentrations of thrombin (T). Thrombin feedback activates FVIII
(circulates in complex with von Willebrand factor, VWF) and FV.

(a) (b)
A2 A2

R306 R506 R306

A1 A1
R506

C2 FXa ~70 A C2
~85 A
A3 A3
APC
(SP domain)
C1 C1

Membrane Membrane

Fig. 16.2 Molecular model of FVa, highlighting its interactions with FXa and APC. FVa is shown as a solid surface (domains A1, green; A2,
cyan; A3, brown; C1, pink; C2, purple) and FXa as an orange ribbon. The virtual membrane is represented as a gray box. (a) FVa/FXa complex, with
FVa residues probed experimentally to be important for FXa binding shown in white. (b) APC approaching the cleavage site at Arg506. The two
main cleavage sites of APC (Arg306 and Arg506) are indicated in red. The serine protease domain of APC is shown as a blue ribbon.
Source: Segers K., Dahlbäck B., Nicolaes G.A. (2007). Coagulation factor V and thrombophilia: background and mechanisms. Thromb Haemost
98: 530–542, with permission.

FV and FVIII into their biologically active counterparts (FVa
and FVIIIa).
Regulation of blood coagulation
The efficient reactions of the coagulation system have
considerable biological potential and strict regulation
is required. For this purpose, several plasma proteins and
protein–cell interactions are involved in constant monitoring
of the circulation. At each level of the coagulation pathway,
membrane-bound molecules expressed on the surface of
intact endothelial cells, circulating inhibitors, and negative
feedback mechanisms provide efficient control.
Antithrombin (AT) is the most important serine pro-
tease inhibitor (serpin) involved in the regulation of blood
coagulation. AT inhibits thrombin as well as FXIa, FIXa,
and FXa and, under certain conditions, also FVIIa. AT
forms a highly stable complex with the protease and, as a
consequence, the protease is trapped and eliminated from
the circulation. The activity of AT is stimulated by heparin,
which accelerates the rate of formation of the AT–protease
complexes. Under normal physiological conditions, heparan
sulfate proteoglycans present on the endothelial cell surface
stimulate the activity of AT, whereas heparin injections
are used in clinical situations. During the inhibition of
thrombin, an important role of heparin is to function as
a bridge between thrombin and AT. In addition, heparin
induces conformational changes in AT that are associated
with the generation of a more efficient inhibitor. In the
inhibition of FXa, the conformational change appears to be
more important than the bridging mechanism.
The tissue factor pathway inhibitor (TFPI) regulates the
TF pathway. The full-length TFPI (TFPIα) is composed of
three protease inhibitory domains belonging to the Kunitz
type of inhibitors and a highly positively charged C-terminal
tail. There are also shorter isoforms containing two or three
Kunitz domains in circulation. TFPI has the unique capacity
to inhibit the FVIIa–TF–FXa complex and is therefore
highly efficient in turning off the TF pathway. The inhibition
mediated by TFPI occurs in two steps: the first involves
inhibition of FXa by the middle Kunitz domain; the first
Kunitz domain then binds and inhibits FVIIa. TFPI can also
directly inhibit FXa, in a phospholipid-bound reaction that
is stimulated by the vitamin K–dependent protein S. Protein
S interacts with the third Kunitz domain. The majority of
TFPIα is bound to glycosaminoglycans on endothelial cells
(approximately 80%) and only a minor fraction of TFPIα is
present in plasma, where it is mainly associated with coag-
ulation FV, which interacts with the C-terminus of TFPI.
Recently, a minor splice variant of FV having a truncated
B-domain (702 amino acid residues deleted), which exposes
a high-affinity binding site for TFPIα, is shown to be a carrier
for TFPIα in circulation. The relative distribution of TFPIα
between full-length FV and the FV splice variant remains
Molecular coagulation and thrombophilia 209
to be determined. The endothelial surface also contains
shorter TFPI isoforms that are linked with glycosylphos-
phatidylinositol (GPI) anchors. They presumably serve to
inhibit premature TF-dependent activation of coagulation
on the vascular wall. Shorter isoforms are also present in
plasma circulating bound to lipoproteins. The function of
the lipoprotein-associated isoforms is not understood.
The highly efficient procoagulant reactions of thrombin
are physiologically adequate at sites of vascular injury and
are instrumental in efficient hemostasis. However, the same
reactions pose a threat to the organism, as uncontrolled
coagulation leads to thrombus formation. Nature has solved
this dilemma in intricate and fascinating ways, one of which
is the transformation of thrombin into an efficient initiator
of a natural anticoagulant pathway, the protein C system.
The conversion of thrombin from a procoagulant into an
anticoagulant enzyme depends on the presence of intact
endothelium. Thus, thrombin generated at sites of intact
vasculature binds to the endothelial membrane protein
thrombomodulin, which is a potent modulator of thrombin
activity and a cofactor to thrombin in the activation of pro-
tein C (Figure 16.3). A receptor for protein C, the endothelial
protein C receptor (EPCR), has been shown to stimulate the
activation of protein C by the thrombin–thrombomodulin
complex. APC inactivates membrane-bound FVa and FVIIIa
by limited proteolysis in reactions that are potentiated by
a cofactor protein designated protein S and, in the case of
FVIIIa degradation, also by the non-activated form of FV
(Figure 16.4).
Under physiological conditions, procoagulant and antico-
agulant mechanisms are balanced in favor of anticoagulation,
whereas the anticoagulant system is downregulated and pro-
coagulant forces prevail at sites of vascular damage. Defects
in this ingenious system are associated with increased throm-
bin generation and a hypercoagulable state, leading to an
increased risk of thrombosis.
Protein C anticoagulant system
Protein C is a vitamin K–dependent plasma protein that is
synthesized mainly in the liver. It is homologous to FVII, FIX,
and FX and shares with them a common modular organi-
zation. From the N-terminus, these proteins contain a vita-
min K–dependent γ-carboxyglutamic acid (Gla)-rich mod-
ule, two epidermal growth factor (EGF)-like modules, and
a serine protease module. The Gla domains bind calcium
ions and provide the vitamin K–dependent clotting proteins
with phospholipid-binding properties. Upon activation by
the thrombin–thrombomodulin complex, the serine protease
module is converted to an active enzyme. APC is highly spe-
cific in its proteolytic activity, cleaving a limited number of
peptide bonds in FVa and FVIIIa.
Intact FV is a high-molecular-weight protein and
shares with the homologous FVIII molecule the modular
210 Molecular Hematology

APC
PC APC
PC
EPCR TM EPCR
TM
T TM
PC TM PC
T T
T PC T

T
T

Fig. 16.3 Activation of protein C by thrombin–thrombomodulin. Thrombomodulin (TM) serves as a cofactor to thrombin (T) in the activation
of protein C. It is present on all endothelial cells. Endothelium also contains the endothelial protein C receptor (EPCR) that binds the Gla domain of
protein C and helps present protein C to the T/TM complex. Activated protein C (APC) remains in the bloodstream to control coagulation reactions.

arrangement A1, A2, B, A3, C1, and C2 (Figure 16.5).
On activation of FV by thrombin or FXa, peptide bonds
surrounding the B-module (positions 709 and 1545) are
cleaved, releasing the B-module, which is thus not part of
FVa. FVIII is activated by thrombin in a similar fashion,
leading to release of the B-module. In circulation, FVIII is
bound to von Willebrand factor (VWF) and the thrombin-
mediated activation of FVIII results in the release of FVIIIa
from VWF. APC cleaves three peptide bonds in FVa at
Arg306, Arg506, and Arg679, whereas FVIIIa is cleaved
at Arg336 and Arg526. As a consequence of the APC-
mediated cleavages, FVa and FVIIIa lose their procoagulant
properties.
APC alone has poor anticoagulant activity and it is only
in the presence of its cofactors protein S and FV that efficient
anticoagulant function is expressed. This was demonstrated
in an experimental system based on the degradation of FVI-
IIa. In this system, it was found that full anticoagulant activity
of APC was obtained in the presence of the combination of
FV and protein S. The synergistic APC cofactor activity of
FV requires APC-mediated proteolysis of at least the Arg506
cleavage site (Figure 16.5). Thus, the Arg506 → Gln muta-
tion is important for understanding the mechanism of APC
resistance (FV Leiden; see later). The APC cofactor activity
of FV appears specific for the degradation of FVIIIa, whereas
the FVa degradation is unaffected by this FV activity. FV

S
V

APC
V S

APC VIIIa APC

T
APC
VIIIi
S
V
Va
APC Vi
APC

Fig. 16.4 Degradation of FVa and FVIIIa by APC. Both FVa and FVIIIa are cleaved and inhibited by APC in phospholipid membrane-bound
reactions that also involve cofactors to APC. Protein S and APC are sufficient to inhibit FVa, whereas the regulation of FVIIIa additionally includes
FV, which in this situation serves as cofactor to APC.

FV A2
A1 B
A1 A2 B A3 C1C2 A3
C2 C1
Thrombin APC
FXa
Procoagulant FVa Anticoagulant FVac
A1 A2
Ca2+ A1 A2 B A3 C1C2
A3 C1C2
APC
Thrombin
FXa
Inactive FVi
A1 A2
Ca2+
A3 C1C2
Fig. 16.5 Procoagulant and anticoagulant properties of factor
V. Proteolytic modification of single-chain FV results in the expression
of either procoagulant or anticoagulant properties. Thrombin and FXa
cleave peptide bonds in and surrounding the B domain (positions 709,
1018, 1545) and thus activate FV to procoagulant FVa, which
functions as a cofactor to FXa in the activation of prothrombin. Intact
FV is sensitive to cleavage by APC (at positions 306, 506, and 679),
which increases the anticoagulant activity of FV. FV modified by APC
(FVac) functions as a synergistic APC cofactor with protein S in the
degradation of FVIIIa. The anticoagulant properties of FV are lost on
further proteolysis by thrombin or FXa. Likewise, the procoagulant
effects of FVa are lost as a result of cleavage by APC. Thus, FV plays a
crucial and central role, balancing procoagulant and anticoagulant
forces. Arrowheads denote the three APC cleavage sites at positions
306, 506, and 679.
loses its APC cofactor activity upon proteolysis by thrombin,
but it gains procoagulant properties as a cofactor to FXa.
Thus, FV is similar to thrombin in being able to express
both procoagulant and anticoagulant effects. However,
whereas the anticoagulant effects of thrombin depend on its
binding to thrombomodulin, the anticoagulant properties
of FV are potentiated by APC-mediated proteolysis of the
non-activated form of FV.
Protein S is also a vitamin K–dependent plasma protein,
but, unlike the other vitamin K–dependent coagulation
proteins, it is not a serine protease. It is a multimodular
protein containing a Gla module, a thrombin-sensitive
module, four EGF-like modules, and a large module homol-
ogous to sex hormone-binding globulin (SHBG; see later
Figure 16.8). Protein S also has functions outside the protein
C system and 60–70% of protein S in plasma circulates
Molecular coagulation and thrombophilia 211
bound to C4b-binding protein (C4BP), a regulator of the
complement system. The Gla module of protein S provides
both free protein S and the protein S–C4BP complex with
phospholipid-binding ability. This is important for the
localization of coagulation and complement regulatory
activities to certain cell membranes, for example to the phos-
phatidylserine that is exposed on apoptotic cells. Protein
S binding to such cells has been shown to be involved in
stimulation of phagocytosis of these cells.
During the degradation of free FVa (i.e. not bound to
FXa) by APC, the cleavage at Arg506 is faster than that at
Arg306. The cleavage at Arg506 leads only to partial loss of
FVa activity, whereas the cleavage at Arg306 leads to efficient
inactivation of FVa. Protein S serves as a cofactor for the
cleavage at Arg306, but has minor effects on the Arg506
cleavage. This, together with a specific protection of the
Arg506 site exerted by FXa, indicates that the Arg306 site
is the most important site for regulation of FVa activity in
the prothrombinase complex. On the other hand, FVa that
is not part of a prothrombinase complex is first cleaved at
the Arg506 site, because the kinetics of this cleavage are
more favorable than those for the cleavage site at Arg306. In
experiments in vitro, protein S has been shown to express an
anticoagulant activity that is independent of the presence of
APC. The exact mechanism is still unclear, despite consid-
erable research. An anticoagulant activity of protein S that is
potentially physiologically important is due to its interaction
with the third Kunitz domain of TFPIα and the direct
stimulation of the TFPIα-mediated inhibition of FXa. Other
proposals suggest that protein S inhibits prothrombin activa-
tion through direct interactions of protein S with FVa, FXa,
and the phospholipid membrane. The in vivo physiological
significance of this APC-independent anticoagulant activity
is unclear. Regardless of its mode of action, protein S is an
important anticoagulant protein in vivo, as demonstrated by
protein S knockout studies and by the association between
protein S deficiency and venous thrombosis.
Molecular genetics of venous
thromboembolism
The annual incidence of venous thrombosis in Western soci-
eties is approximately 1–2 per 1000. Thrombotic episodes
tend to occur in conjunction with surgery, fractures, preg-
nancy, the use of oral contraceptives, and immobilization. In
addition, genetic defects are frequently involved and many
patients report positive family histories. Genetic defects
known to predispose to thrombosis include inherited APC
resistance due to FV mutations (mainly FV Leiden), a
point mutation in the prothrombin gene (G20210A), and
deficiencies of anticoagulant protein C, protein S, or AT.
These inherited causes of thrombophilia are autosomal
212 Molecular Hematology
dominant disorders and the prothrombotic mutations result
in a lifelong increased risk of thrombosis. However, the
penetrance of the disease may be variable and the individual
risk profile is still difficult to predict.
Factor V gene mutations causing APC
resistance
In 1993, APC resistance was described as a cause of inher-
ited thrombophilia and it was soon demonstrated to be
highly prevalent (20–60%) among thrombosis patients. In
APC resistance, APC does not give a normal prolongation of
the clotting time. In more than 95% of cases, the molecular
defect associated with APC resistance is a single point muta-
tion in the FV gene, F5 (GeneID 2153, OMIM ID 227400,
188055, 612309). The mutation is a G → A substitution at
nucleotide position 1691 in the FV gene, which predicts the
replacement of Arg506 with a Gln (Figure 16.5). The mutant
FV is known as FVR506Q, FV Leiden, or FV:Q506 (R and Q
are one-letter codes for Arg and Gln, respectively). Recently,
additional rare FV mutations have been described in indi-
viduals with venous thrombosis and APC resistance. In a
Japanese family, FVNara (W1920R) is demonstrated to be
the cause of the APC resistance phenotype. In six European
patients, APC resistance is caused by FVBonn (A512V). In
both FVNara and FVBonn, the FV mutations were proven
to be the cause of the APC resistance phenotype. In several
other venous thrombosis patients with APC resistance, FV
mutations have been found, but not experimentally proven to
be the cause of the APC resistance phenotype. These muta-
tions are FV E666D in a Chinese family, and three mutations
in venous thrombosis patients with Chilean Amerindian eth-
nic background – FV M443L, FV E461Q, and FV G493E. Two
different mutations affecting the Arg306 site have recently
been found in thrombosis cases, FV Cambridge and FV Hong
Kong, but such mutations appear to be rare. They do not
result in APC resistance and are not major risk factors for
thrombosis.
The FV Leiden allele is found only in Caucasians, and
its prevalence varies considerably in the general population
of Western societies. High prevalence (up to 15%) is found
in southern Sweden, Germany, Greece, Arab countries, and
Israel. In the Netherlands, the UK, and the USA, around 3–
5% of the population carry the mutant allele. A lower preva-
lence (around 2%) is found in Hispanics. The high prevalence
of the FV Leiden allele in certain populations suggests a pos-
sible survival advantage, and there is a reduced risk of bleed-
ing after delivery in women carrying the mutation. In the his-
tory of humankind, the slightly increased risk of thrombosis
associated with the FV Leiden allele has presumably not been
a negative survival factor, because thrombosis develops rela-
tively late in life and does not influence fertility. In addition,
many of the circumstantial risk factors for thrombosis, such
as a sedentary lifestyle, surgery, and the use of oral contra-
ceptives, did not affect our ancestors.
The high prevalence of the FV Leiden allele in Western
societies is the result of a founder effect. It has been estimated
that the mutation event occurred around 21 000 years ago,
after the “Out of Africa Exodus” that took place 100 000 years
ago and also after the separation of Asians from Europeans.
This explains why the mutant FV allele is common among
European populations, but is not present among Japanese,
Chinese, or in the original populations of Africa, Australia,
or America.
A large number of studies have demonstrated a relation-
ship between the presence of APC resistance (FV Leiden)
and an increased risk of venous thrombosis. Differences in
selection criteria of patients and in the prevalence of the
mutant allele in the general population explain the wide vari-
ation in results obtained from different studies. However, the
general consensus is that the FV Leiden allele is the most
common genetic risk factor for venous thrombosis in West-
ern societies. The odds ratio, describing the increased risk
of thrombosis in affected individuals, has been calculated to
be two- to threefold for those carrying the defect in a het-
erozygous form, whereas homozygous individuals are at an
even higher risk of thrombosis. The FV Leiden allele does
not appear to be a strong risk factor for arterial thrombo-
sis, such as myocardial infarction. The FV Leiden allele is
associated with a hypercoagulable state, which is reflected
by increased levels of prothrombin activation fragments in
plasma of individuals with inherited APC resistance. Two
molecular mechanisms are involved (Figures 16.5 and 16.4).
One is that an APC cleavage site in FVa is lost, which impairs
the normal degradation of FVa by APC. The other surpris-
ing observation is that FV Leiden is a poor APC cofactor in
the degradation of FVIIIa, because the cleavage at Arg506
is required for the expression of the APC cofactor activity
of FV.
In the degradation of normal FVa, the APC cleavage at
Arg506 has favorable kinetics compared with cleavages at
other sites. The Arg506 cleavage is approximately 10-fold
faster than the cleavage at Arg306, and the activity of
FVa:Q506 (FVa Leiden) is therefore inhibited at an approx-
imately 10-fold lower rate than FVa:R506. Generated FVa
Leiden persists longer than normal FVa and can form active
prothrombinase complexes with FXa. However, degradation
of free FVa (i.e. FVa not bound to FXa) is different from
that of FVa, which is part of the prothrombinase complex.
In the prothrombinase complex, the Arg506 site is protected
from degradation by APC by both FXa and prothrombin. In
addition, protein S functions as an APC cofactor primarily
for the Arg306 cleavage. As a consequence, APC-mediated
degradation of FVa, which is part of the prothrombinase
complex, follows a different pathway compared with that of
free FVa. Therefore, when FVa:R506 and FVa:Q506 are part
 Molecular coagulation and thrombophilia 213

0 2 4 6 8 10 12 14 kb

Exon 1 2 3A 3B 4 5 6

mRNA

200 bp

Type II
Mature polypeptide chain

Type I

100 aa
Fig. 16.6 Structure of the human antithrombin gene and location of detrimental missense mutations in the antithrombin molecule.
The gene for antithrombin (SERPINC1) is localized to chromosome 1q23–q25 and spans 13.4 kb of DNA (upper part). It comprises seven exons and
results in an mRNA of 1.7 kb (middle part). The antithrombin molecule (lower part) is synthesized as a single polypeptide chain comprising a
mature protein of 432 amino acid residues and a signal peptide (shaded) of 32 amino acid residues. Many mutations of different types causing
antithrombin deficiency have been described. Shown here are examples of missense mutations leading to amino acid substitutions associated with
type I deficiency (indicated by open circles below the polypeptide chain) or type II deficiency (open, shaded, and filled circles denote type II HBS,
type II RS, and type II PE variants, respectively).

of assembled prothrombinase complexes, the rates of their According to the Human Gene Mutation Database
degradation by APC plus protein S are similar. (HGMD; www.hgmd.cf.ac.uk), there are 353 distinct muta-
Laboratory investigation of inherited APC resistance due tions reported in the SERPINC1 gene (Table 16.1). In a major-
to the FV Leiden allele can be done using both a functional ity of cases, the genetic defect is a point mutation leading
APC-resistance test and molecular biology assays. A mod- to missense, nonsense, or splice-site mutations. Other muta-
ified APC-resistance test involving dilution of the patient’s tions involve small deletions or insertions, which result in
plasma in FV-deficient plasma is highly sensitive and specific frameshift alterations that are often deleterious. Partial or
for the presence of the FV Leiden allele. whole gene deletions are relatively uncommon causes of AT
deficiency (<10% of the reported mutations). Type II RS vari-
ants are defective in protease inactivation and mutations in
Deficiency of antithrombin
Subjects with inherited AT deficiency are, with few excep-
tions, heterozygotes. It is caused by mutations in the AT gene, Table 16.1 Mutations in genes of anticoagulant proteins retrieved
SERPINC1 (GeneID 462, OMIM ID 107300, 613118), which from the HGMD database (www.hgmd.cf.ac.uk), November 2016
is located at chromosome 1q23–q25. Heterozygous AT defi-
ciency is found in 0.02–0.05% of the general population and Mutation type SERPINC1 PROC PROS1
in 1–2% of thrombosis patients, suggesting that the genetic
defect is associated with at least a 10-fold increased risk of Missense/nonsense 196 (56%) 276 (74%) 249 (61%)
thrombosis – somewhat higher than estimated for APC resis- Splicing 25 (7%) 31 (8%) 44 (11%)
tance. AT deficiency may be of either type I or type II. Type Regulatory 1 (0.3%) 14 (4%) 4 (1%)
I deficiency is characterized by low levels of both immuno- Small deletions 65 (18%) 29 (8%) 49 (12%)
Small insertions 29 (8%) 16 (4%) 24 (6%)
logical and functional AT, whereas type II denotes functional
Small indels 3 (0.8%) 4 (1%) 5 (1%)
defects. Type II cases can be further divided into three sub-
Gross deletions 28 (8%) 3 (0.8%) 25 (6%)
types: RS (reactive site mutants), HBS (heparin binding site Gross insertions/duplications 2 (0.6%) 0 (0%) 7 (2%)
mutants), or PE (mutants giving pleiotropic effects). A large Complex rearrangements 4 (1%) 0 (0%) 2 (0.5%)
number of AT deficiencies are caused by missense mutations, Repeat variations 0 (0%) 0 (0%) 0 (0%)
which alter key amino acid residues in the primary structure Total no. of mutations 353 373 409
of the protein (Figure 16.6).
214 Molecular Hematology

the vicinity of the reactive site have been found. The type II
HBS deficiency carries mutations in the heparin-binding site,
and type II PE variants are caused by a limited number of
mutations between amino acids 402 and 429.
Protein C deficiency
Protein C deficiency is caused by mutations in the protein
C gene, PROC (GeneID 5624, OMIM ID 176860, 612304),
which is located on chromosome 2q13–q21. Clinically, there
are two types of protein C deficiency. Type I is characterized
by a parallel reduction in protein C antigen and functional
activity. Type II is characterized by a functional defect in the
protein, and its plasma concentration may be normal. The
majority of reported cases of protein C deficiency are of type
I. However, plasma assays can sometimes have limited dis-
criminative power between the different types. There is also
an overlap between plasma concentrations at the low end of
the normal distribution and at the upper end of protein C
deficiency, which can complicate correct phenotyping.
Heterozygous deficiency of protein C is identified in
2–5% of thrombosis patients. The prevalence of protein C
deficiency in the population is estimated to be approximately
0.3%. The relative risk of venous thrombosis associated with
protein C deficiency is similar to or slightly higher than that
of FV Leiden. Protein C deficiency is not a risk factor for
arterial thrombosis, although there are studies claiming that
protein C deficiency may have an impact on the onset of arte-
rial occlusive diseases. Homozygous or compound heterozy-
gous protein C deficiency is a rare condition (1 in 200 000 to
1 in 400 000) that often leads to severe and fatal thrombosis
in the neonatal period. The clinical picture is that of purpura
fulminans and the symptoms include necrotic skin lesions
due to microvascular thrombosis. Other major symptoms
are thrombosis in the brain and disseminated intravascular
coagulation. Several cases have been successfully treated
with fresh frozen plasma or with protein C concentrates.
Genetic analysis has been performed in a large number
of cases with protein C deficiency and there are 373 muta-
tions in the HGMD associated with protein C deficiency
(Table 16.1). Most genetic defects are missense mutations,
which lead to single amino acid substitutions, or nonsense
mutations located within the coding region of the gene (Fig-
ure 16.7). Missense mutations causing type I deficiency are

0 2 4 6 8 10 kb

Exon 1 2 3 45 6 7 8 9

mRNA

200 bp

Type II
GLA EGF1 EGF2 AP Serine protease domain

Type I

100 aa
Fig. 16.7 Structure of the human protein C gene and location of detrimental missense mutations in the protein C molecule. Human
protein C is encoded by the PROC gene, localized to chromosome 2q13–q14, which spans approximately 11 kb of DNA (upper part). The gene
comprises nine exons, which yield about a 1.8-kb mRNA transcript (middle part). The protein C mRNA encodes a pre-pro-protein C sequence of
461 amino acid residues (lower part). The pre-sequence (dark shading) serves as a signal peptide and the pro-sequence (light shading) functions as
a signal for proper γ-carboxylation of the protein. The mature protein consists of 419 residues and can be divided into a γ-carboxyglutamic acid
(GLA) domain, two epidermal growth factor (EGF) domains, and a serine protease domain. During processing of the protein, an internal dipeptide
is removed from the protein and the mature protein circulates as a covalently linked two-chain molecule. Between the second EGF domain and the
protease part of the molecule is an activation peptide (AP) region, which is released on protein C activation. Shown here are examples of missense
mutations resulting in amino acid substitutions associated with type I deficiency (indicated below the polypeptide chain) or type II deficiency (above
the polypeptide chain).

scattered over the entire polypeptide chain. Mutations lead-
ing to type II deficiency are less common, but have also been
found in almost all the modules of protein C, including the
propeptide, the Gla module, EGF1, the activation peptide
(AP), and the serine protease domain. Mutations in the pro-
moter region of the gene, which affect the plasma protein
concentration, and mutations affecting RNA splicing have
also been found.
Protein S deficiency
Protein S deficiency is linked to the protein S gene, PROS1
(GeneID 5627, OMIM ID 176880, 612336, 614514). The
PROS1 gene is located close to the centromere at chro-
mosome 3p11.1–q11.2. A homologous pseudogene (PROSP,
GeneID 5628) is located in the near vicinity and shares more
than 95% similarity with PROS1 exon and intron sequences,
which must be taken into consideration when designing
genetic investigations of PROS1.
Phenotyping for protein S deficiency is dependent on the
correct measurement of protein S plasma concentration in
the laboratory. This is not a simple task to perform and will
be further complicated by the presence of pools of free and
C4BP-bound protein S. However, the level of free protein S
discriminates better between those with and without protein
S deficiency than the level of total protein S. This is because
the concentrations of protein S and C4BPβ+ , which is the
protein S-binding isoform of C4BP, are equimolar in protein
S–deficient individuals and most of the protein S is bound
to C4BPβ+ . Protein S deficiency with low levels of both free
and total protein S is called type I, whereas protein S defi-
ciency with low free protein S and normal total protein S has
been believed to constitute a separate genetic type (type III).
However, coexistence of the two types in many protein S–
deficient families demonstrates that they represent different
phenotypic variants of the same genetic disease. Mutations
that cause functional defective protein S are referred to as
type II deficiency. To date, very few type II deficiencies have
been found, presumably related to the poor diagnostic per-
formance of available functional protein S assays.
Heterozygous protein S deficiency is present in 1–10%
of thrombosis patients; the wide variety may be explained
by several factors such as different inclusion and exclusion
criteria and geographical/ethnic differences between studies.
The prevalence of protein S deficiency in the general pop-
ulation has been estimated to be 0.03–0.13% in European
populations, but higher figures have been reported from
Japanese and Thai populations. Family studies suggest that
heterozygous carriers have at least a fivefold increased risk
of thrombosis compared with their healthy relatives, which
is similar to the rates in thrombophilic families with protein
C deficiency and APC resistance (FV Leiden). Homozygous
protein S deficiency is extremely rare, but appears to give
Molecular coagulation and thrombophilia 215
a similar picture to homozygous protein C deficiency, with
purpura fulminans in the neonatal period.
The PROS1 database, published by the International
Society on Thrombosis and Haemostasis in 2000, contained
more than 200 entries, in which 131 different mutations were
considered to be detrimental. In the HGMD there are 409 dif-
ferent PROS1 mutations that may be associated with a protein
S deficiency phenotype. Most of the gene defects are missense
or nonsense mutations, and mutations affecting splicing
or small insertion/deletion defects are less common (Fig-
ure 16.8). Due to the large size of the protein S gene and the
presence of a closely linked and highly similar pseudogene,
identification of mutations is not easy. Furthermore, in many
families with phenotypically established protein S deficiency,
PROS1 gene mutations are not found, although linkage
between the PROS1 locus and protein S deficiency has been
established. The reason for the difficulties in identifying pro-
tein S gene mutations in some families may be explained by
the fact that common genetic screening techniques may miss
large deletions as a cause for quantitative protein S deficiency.
This was also shown to be the case by extensive segregation
analysis using a dense set of genetic single-nucleotide
polymorphism (SNP) and microsatellite markers in families
where mutations in the PROS1 gene have not been found. By
this approach, three of eight investigated families could be
explained by large and unique deletions. If this finding, of an
unusually high frequency of large deletions, is reproduced
in other family materials, then such mutations must be con-
sidered a major explanatory factor for protein S deficiency.
Prothrombin gain-of-function mutations
A point mutation in the prothrombin gene, F2 (GeneID
2147, OMIM ID 176930.0009), has been identified as the
second most common independent risk factor for venous
thrombosis. The mutation involves a G → A base substitu-
tion at nucleotide position 20210, which is located in the 3′
untranslated regions of the F2 gene (Figure 16.9). Thus, the
point mutation does not lead to protein structure alterations.
Instead, the mutation is associated with increased plasma
levels of prothrombin through a gain-of-function phenotype
that has been explained by increased efficiency of the F2 gene
mRNA. The prevalence of the mutation in the general pop-
ulation is 0–5%, depending on the ethnicity of the studied
population, and the mutation is associated with an approx-
imately twofold increased risk of thrombosis. Similar to the
FV Leiden mutation, the F2 20210A variant originated from
a single mutational event and the allele is only found in Cau-
casian populations.
Another gain-of-function mutation was originally found
in a Japanese family with hereditary thrombosis. The muta-
tion is located in the last exon of the F2 gene and leads to a
substitution of arginine for leucine at position 596 (OMIM
216 Molecular Hematology

0 50 60 70 80 90 100 kb

Exon 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

mRNA

200 bp

Type II

GLA TSR EGF1EGF2 EGF3 EGF4 SHBG

Type I

100 aa
Fig. 16.8 Structure of the human protein S gene and location of detrimental missense mutations in the protein S molecule. The gene
for human protein S (PROS1) comprises 15 exons (upper part) and spans 100 kb of DNA and is localized to chromosome 3p11.1–q11.2. Exons are
denoted by open bars and introns by lines. Introns denoted by dashed lines between exons indicate gaps and are not drawn to scale. The PROS1
mRNA is approximately 3.5 kb in size (middle part). The mRNA is translated into a pre-pro-protein S of 676 amino acid residues (lower part). The
polypeptide chain can be divided into a signal peptide (dark shading), a pro-peptide (light shading), a thrombin sensitive region (TSR), a
γ-carboxyglutamic acid (GLA) domain, four epidermal growth factor (EGF)-like domains, and a large carboxy-terminal domain homologous to sex
hormone–binding globulin (SHBG). Shown here are examples of missense mutations resulting in amino acid substitutions associated with type I
deficiency (indicated below the polypeptide chain) or type II deficiency (above the polypeptide chain).

1 kb

2 4 6 8 10 12 14
Exon 1 3 5 7 9 11 13

13 14

3’ UT

A
ATGGTTCCCAATAAAAGTGACTCTCAGC AGCCTC
G
Fig. 16.9 Structure of the human prothrombin gene and location of the prothrombin 20210 G → A mutation. The human gene for
prothrombin (F2) comprises 14 exons and spans approximately 20 kb of DNA on chromosome 11p11–q12. The nucleotide sequence flanking the
G → A transition at nucleotide 20210 (indicated in bold) in the 3′ untranslated region of the F2 gene is shown at the bottom. The putative
polyadenylation signal is boxed. The 20210A allele represents a gain-of-function mutation associated with elevated levels of plasma prothrombin
and an increased risk of venous thrombosis.
 Molecular coagulation and thrombophilia 217

100

80 FV Leiden
60

Free of thrombosis (%)

Fig. 16.10 Thrombosis-free survival curves for individuals 40 Protein S Protein S
with different FV genotypes and coinherited protein S deficiency deficiency
20
deficiency. (a) The increased risk of thrombosis with combined +FV Leiden
0 (a)
defects is illustrated with thrombosis-free survival curves for 21
individuals with single defects, FV Leiden or protein S deficiency, 100
FV: normal
and 18 individuals with both defects. There was no significant 80
difference between those with single defects, whereas differences
60 FV Leiden:
between individuals with the FV Leiden allele or protein S heterozygous
deficiency and those with combined defects were significant. (b) 40
Probability of being free of thrombotic events at a certain age for FV Leiden: homozygous
146 normal individuals, 144 heterozygotes and 18 homozygotes 20
for the FV Leiden allele (Kaplan–Meier analysis). Highly significant 0
differences were observed between normals and heterozygotes 0 20 40 60 80 100
and between heterozygotes and homozygotes. Age (years) (b)

ID 176930.0015). This arginine residue 596 is located in
a highly conserved region important for the coordination
of sodium and contains one of the antithrombin-binding
sites in the prothrombin molecule. It is anticipated that the
mutation conveys antithrombin resistance by impairment of
the thrombin–antithrombin complex formation and thereby
leads to an increased risk of thrombosis. In recent years, other
mutations affecting the same amino acid residue (Arg596Gln
and Arg596Trp) have been associated with venous throm-
boembolism and a laboratory phenotype of antithrombin
resistance. The different mutations, at the same codon, in
apparently unrelated families indicate that this may represent
a hotspot for prothrombin mutations. However, there are so
far only a handful of reports of antithrombin resistance and
the relative frequency of this hereditary risk factor has yet to
be determined.
Severe thrombophilia is a multigenic
disease
Venous thrombosis is a typical multifactorial disease involv-
ing one or more environmental and/or genetic risk factors.
In Western societies, many individuals carry more than one
genetic risk factor because the FV Leiden allele is so common.
In contrast, in countries where the FV Leiden allele is rare,
few individuals carry more than one genetic defect. This may
explain the difference in incidence of thromboembolic dis-
ease between Japan and China on the one hand and Europe
and the USA on the other. The frequency of individuals car-
rying two or more genetic defects can be calculated on the
basis of the prevalence of the individual genetic defects in the
general population. In a country where the prevalence of FV
Leiden is 10%, combinations of protein C deficiency and FV
Leiden are expected to be present in between 1 in 3000 and 1
in 10 000 individuals. A similar calculation for the combina-
tion of the prothrombin mutation and FV Leiden allele sug-
gests the prevalence of combined defects to be 1–2 per 1000
individuals. Thus, a large number of people carry more than
one genetic defect and such individuals have a considerably
increased risk of thrombosis. The FV Leiden allele is thus
found to be an additional genetic risk factor in certain throm-
bophilic individuals with a deficiency of protein C, protein S,
or AT as well as in cases with the prothrombin mutation (Fig-
ure 16.10).
The thrombotic tendency in individuals with inherited
genetic defects is highly variable and some individuals never
develop thrombosis, whereas others develop recurrent severe
thrombotic events at an early age. This depends on the par-
ticular genotype, the coexistence of other genetic defects, and
the presence of environmental risk factors such as oral con-
traceptives, trauma, surgery, and pregnancy. Thus, women
with hetero- or homozygosity for the FV Leiden allele who
also use oral contraceptives have a considerably increased
risk of thrombosis. In this context it is noteworthy that the
World Health Organization (WHO) recommends that com-
bined hormonal contraceptives should be avoided in women
with inherited thrombophilia.
Management of thrombophilia
Decisions about medical intervention due to the presence
of one or more genetic defects should be based on careful
consideration of the clinical picture, including the patient’s
family history. The impact of the medical history on the
use of oral anticoagulants is perhaps even more important
218 Molecular Hematology
in individuals with APC resistance (FV Leiden) or the
prothrombin mutation than in those with the rarer defi-
ciencies of protein C, protein S, or AT. The risk of bleeding
complications due to anticoagulant therapy must always be
weighed against the benefits of anticoagulation, especially if
an oral anticoagulant is used for periods exceeding three to
six months, when the risk of thrombotic recurrence probably
declines. New clinical data are continually emerging and no
general consensus regarding screening, prophylaxis, and the
treatment of symptomatic patients has yet been established.
When the FV Leiden allele is present in homozygous form,
or heterozygosity is combined with a second genetic defect,
prophylactic treatment with heparin or oral anticoagulants
is recommended in situations known to be associated with a
high risk of thromboembolic complications, such as surgery
or pregnancy, even if the patient has never experienced any
thrombosis or has no family history of such complications.
For heterozygous asymptomatic carriers lacking a family his-
tory of thrombosis, short-term prophylaxis has been recom-
mended in high-risk situations, but it remains to be estab-
lished whether prophylaxis should be given in all situations
associated with a risk of thrombosis.
Symptomatic heterozygous patients should be managed in
the same way as any other patient with thrombotic events
until more specific recommendations are established. It is not
known whether the presence of a genetic defect is associ-
ated with an increased risk of recurrence, even though most
studies on APC resistance tend to suggest that this is indeed
the case. Patients with combined defects, and probably also
patients with single-gene defects, may be at increased risk of
recurrence and should accordingly be given long-term anti-
coagulation therapy beyond six months, even after an iso-
lated thromboembolic event. However, more data are needed
before these recommendations can be considered generally
applicable.
The potential benefits of general screening for APC resis-
tance and/or the FV Leiden allele prior to thrombotic events
or in the presence of such circumstantial factors as oral con-
traceptive use, pregnancy, and surgery are obvious, but more
prospective data are needed, not least in terms of cost–benefit
ratios, before any general recommendations can be made.
Conclusions
Inherited APC resistance, caused by the Arg506Gln muta-
tion in the FV gene (FV Leiden), is the most common genetic
risk factor for thrombosis identified to date. The mutated
FV has normal procoagulant properties, but the loss of the
APC cleavage site at position 506 in FV results in impaired
regulation of coagulation and a hypercoagulable state. The
prevalence of FV Leiden in Caucasian populations varies
between 2% and 15%. A genetic variant in the prothrombin
gene (G20210A) is another common prothrombotic risk
factor, with a prevalence of approximately 2% in the general
population. Other less common independent genetic risk
factors include abnormalities in the genes for AT, protein
C, and protein S. Families with thrombophilia present with
variable penetrance of thrombosis explained by different
combinations of genetic defects and environmental risk
factors. Patients with combined genetic defects are at higher
risk of thrombosis than those with single-gene defects. Thus,
evaluation of patients with thrombosis must be performed in
order to fully estimate the risk for thrombosis in each case.
Further reading
Blood coagulation: introduction and
regulation
Camire, R.M. (2016). Rethinking events in the haemostatic process: role
of factor V and TFPI. Haemophilia 22 (Suppl 5): 3–8.
Dahlbäck, B. (2005). Blood coagulation and its regulation by anticoagu-
lant pathways: genetic pathogenesis of bleeding and thrombotic dis-
eases. J. Int. Med. 257: 209–223.
Dahlbäck, B. (2016). Pro- and anticoagulant properties of factor V in
pathogenesis of thrombosis and bleeding disorders. Int. J. Lab. Hema-
tol. 38 (Suppl 1): 4–11.
Dahlbäck, B. and Villoutreix, B.O. (2005). Regulation of blood coagu-
lation by the protein C anticoagulant pathway: novel insights into
structure–function relationships and molecular recognition. Arte-
rioscler. Thromb. Vasc. Biol. 25: 1311–1320.
Esmon, C.T. (2014). Targeting factor Xa and thrombin: impact on coag-
ulation and beyond. Thromb. Haemost. 111: 625–633.
Furie, B. and Furie, B.C. (2005). Thrombus formation in vivo. J. Clin.
Invest. 115: 3355–3362.
Long, A.T., Kenne, E., Jung, R. et al. (2016). Contact system revisited: an
interface between inflammation, coagulation, and innate immunity.
J. Thromb. Haemost. 14: 427–437.
Mann, K.G., Brummel-Ziedins, K., Orfeo, T., and Butenas, S. (2006).
Models of blood coagulation. Blood Cells Mol. Dis. 36: 108–117.
Morrissey, J.H. and Smith, S.A. (2015). Polyphosphate as modulator of
hemostasis, thrombosis and inflammation. J. Thromb. Haemost. 13
(Suppl 1): S92–S97.
Rau, J.C., Beaulieu, L.M., Huntington, J.A., and Church, F.C. (2007). Ser-
pins in thrombosis, hemostasis and fibrinolysis. J. Thromb. Haemost.
5 (Suppl 1): 102–115.
Vincent, L.M., Tran, S., Livaja, R. et al. (2013). Coagulation factor V
(A2440G) causes East Texas bleeding disorder via TFPIα. J. Clin.
Invest. 123: 3777–3787.
Molecular genetics of venous
thromboembolism, APC resistance,
and FV leiden
Bertina, R.M., Koeleman, B.P., Koster, T. et al. (1994). Mutation in blood
coagulation factor V associated with resistance to activated protein C.
Nature 369: 64–67.

Bezemer, I.D., Bare, L.A., Doggen, C.J.M. et al. (2008). Gene variants
associated with deep venous thrombosis. J. Am. Med. Assoc. 299:
1306–1314.
Castoldi, E. and Rosing, J. (2010). APC resistance: biological basis and
acquired influences. J. Thromb. Haemost. 8: 445–453.
Dahlbäck, B., Carlsson, M., and Svensson, P.J. (1993). Familial throm-
bophilia due to a previously unrecognized mechanism characterized
by poor anticoagulant response to activated protein C: prediction of a
cofactor to activated protein C. Proc. Natl. Acad. Sci. U.S.A. 90: 1004–
1008.
Greengard, J.S., Sun, X., Xu, X. et al. (1994). Activated protein C resis-
tance caused by Arg506Gln mutation in factor Va. Lancet 343: 1361–
1362.
Griffin, J.H., Evatt, B., Wideman, C., and Fernandez, J.A. (1993). Anti-
coagulant protein C pathway defective in majority of thrombophilic
patients. Blood 82: 1989–1993.
Heit, J.A. (2008). The epidemiology of venous thromboembolism in the
community. Arterioscler. Thromb. Vasc. Biol. 28: 370–372.
Koster, T., Rosendaal, F.R., de Ronde, H. et al. (1993). Venous thrombo-
sis due to poor anticoagulant response to activated protein C: Leiden
Thrombophilia Study. Lancet 342: 1503–1506.
Lindqvist, P.G., Svensson, P.J., Dahlbäck, B., and Marsal, K. (1998). Fac-
tor V R506Q mutation (activated protein C resistance) associated
with reduced intrapartum blood loss: a possible evolutionary selec-
tion mechanism. Thromb. Haemost. 79: 69–73.
Nogami, K., Shinozawa, K., Ogiwara, K. et al. (2014). Novel FV nutation
(W1920R, FVNara) associated with serious deep vein thrombosis and
more potent APC resistance relative to FVLeiden. Blood 123: 2420–
2428.
Pezeshkpoor, B., Castoldi, E., Mahler, A. et al. (2016). Identification
and functional characterization of a novel F5 mutation (Ala512Val,
FVBonn) associated with activated protein C resistance. J. Thromb.
Haemost. 14: 1353–1363.
Svensson, P.J. and Dahlbäck, B. (1994). Resistance to activated protein
C as a basis for venous thrombosis. N. Engl. J. Med. 330: 517–522.
Vandenbroucke, J.P., Koster, T., Briet, E. et al. (1994). Increased risk of
venous thrombosis in oral-contraceptive users who are carriers of
factor V Leiden mutation. Lancet 344: 1453–1457.
Voorberg, J., Roelse, J., Koopman, R. et al. (1994). Association of
idiopathic venous thromboembolism with single point-mutation at
Arg506 of factor V. Lancet 343: 1535–1536.
Antithrombin deficiency
Lane, D.A., Bayston, T., Olds, R.J. et al. (1997). Antithrombin mutation
database: 2nd (1997) update. Thromb. Haemost. 77: 197–211.
Perry, D.J. and Carrell, R.W. (1996). Molecular genetics of antithrombin
deficiency. Hum. Mut. 7: 7–22.
Picard, V., Nowak-Göttl, U., Biron-Andreani, C. et al. (2006). Molecular
bases of antithrombin deficiency: twenty-two novel mutations in the
antithrombin gene. Hum. Mut. 27: 600. [Erratum in Human Mutation
2006, 27, 1160.].
Van Boven, H.H., Vandenbroucke, J.P., Briet, E., and Rosendaal, F.R.
(1999). Gene–gene and gene–environment interactions determine
Molecular coagulation and thrombophilia 219
the risk of thrombosis in families with inherited antithrombin defi-
ciency. Blood 94: 2590–2594.
Protein C system and protein C deficiency
Dahlbäck, B. and Villoutreix, B.O. (2005). Regulation of blood coagu-
lation by the protein C anticoagulant pathway: novel insights into
structure–function relationships and molecular recognition. Arte-
rioscler. Thromb. Vasc. Biol. 25: 1311–1320.
Esmon, C.T. (2012). Protein C anticoagulant system – anti-
inflammatory effects. Semin. Immunopathol. 34: 127–132.
Griffin, J.H., Zlokovic, B.V., and Mosnier, L.O. (2015). Activated protein
C: biased for translation. Blood 127: 2898–2907.
Segers, K., Dahlbäck, B., and Nicolaes, G.A. (2007). Coagulation fac-
tor V and thrombophilia: background and mechanisms. Thromb.
Haemost. 98: 530–542.
Protein S and protein S deficiency
Dahlbäck, B. (2007). The tale of protein S and C4b-binding protein, a
story of affection. Thromb. Haemost. 98: 90–96.
Gandrille, S., Borgel, D., Sala, N. et al. (2000). Protein S deficiency: a data
base of mutations. Summary of the first update. Thromb. Haemost. 84:
918–934.
Garcia de Frutos, G., Fuentes-Prior, P., Hurtado, B., and Sala, N. (2007).
Molecular basis of protein S deficiency. Thromb. Haemost. 98: 543–
556.
Johansson, A.M., Hillarp, A., Säll, T. et al. (2005). Large deletions of the
PROS1 gene in a large fraction of mutation negative patients with
protein S deficiency. Thromb. Haemost. 94: 951–957.
Peraramelli, S., Rosing, J., and Hackeng, T.M. (2012). TFPI-dependent
activities of protein S. Thromb. Res. 129 (Suppl 2): S23–S26.
Prothrombin gene mutations
Bulato, C., Radu, C.M., Campello, E. et al. (2016). New prothrombin
mutation (Arg596Trp, prothrombin Padua 2) associated with venous
thromboembolism. Arterioscler. Thromb. Vasc. Biol. 36: 1022–1029.
Danckwardt, S., Hartmann, K., Gehring, N.H. et al. (2006). 3′ End pro-
cessing of the prothrombin mRNA in thrombophilia. Acta Haematol.
115: 192–197.
Dziadosz, M. and Baxi, L.V. (2016). Global prevalence of prothrombin
gene mutation G20210A and implications in women’s health: a sys-
tematic review. Blood Coagul. Fibrinolysis 27: 481–489.
Gehring, N.H., Frede, U., Neu-Yilik, G. et al. (2001). Increased efficiency
of mRNA 3′ end formation: a new genetic mechanism contributing
to hereditary thrombophilia. Nat. Genet. 28: 389–392.
Miyawaki, Y., Suzuki, A., Fujita, J. et al. (2012). Thrombosis from a
prothrombin mutation conveying antithrombin resistance. N. Engl.
J. Med. 366: 2390–2396.
Poort, S.R., Rosendaal, F.R., Reitsma, P.H., and Bertina, R.M. (1996). A
common genetic variation in the 3′ -untranslated region of the pro-
thrombin gene is associated with elevated plasma prothrombin levels
and an increase in venous thrombosis. Blood 88: 3698–3703.
Chapter 17 The molecular basis of hemophilia
Daniel P. Hart1 & Paul L.F. Giangrande2
1 The Royal London Hospital Haemophilia Centre, Barts and The London School of Medicine and Dentistry, QMUL London, UK
2
Oxford Haemophilia and Thrombosis Centre, Churchill Hospital, Oxford, UK

Introduction: clinical features of hemophilia, 221 Inhibitor formation: etiology and clinical implications, 226
Inheritance of hemophilia, 223 Therapeutic applications of molecular biology to patient care, 228
Molecular basis of hemophilia A, 223 Conclusions, 232
Molecular basis of hemophilia B, 225 Further reading, 232

on the anterior thigh, weakness, and wasting of leg muscles

Introduction: clinical features (Figure 17.3). Bleeding from the gastrointestinal tract
of hemophilia (melena) and bleeding into the urinary tract (hematuria)
may also occur. There is also a significant risk of intracranial
The clinical severity (bleed phenotype) is critically deter-
hemorrhage in severe hemophilia, which in the past was a sig-
mined by the concentration of circulating factor VIII (or IX)
nificant cause of mortality when treatment was not so readily
in the blood, and severe hemophilia is defined by a clotting
available. Higher levels of factor VIII (or IX) above 5 IU dl−1
factor concentration below 1 IU dl−1 (Table 17.1). The
are associated with a milder form of the disease, with no
hallmark of severe hemophilia, in the absence of treatment,
spontaneous joint bleeds, but a definite risk of bleeding
is recurrent and spontaneous hemarthrosis. Typically, hinge
after even relatively minor injury or procedures (e.g. dental
joints such as the knees, elbows, and ankles are affected, but
extraction).
bleeds may also occur in the wrist or shoulder. Bleeding into
Treatment of bleeding episodes involves the intravenous
the hip joint is unusual. The acutely affected joint becomes
injection of coagulation factor concentrates (CFCs); the total
swollen and warm, and is held in a position of flexion (Fig-
dose and frequency of treatment will also be determined by
ure 17.1), with no external discoloration or bruising around
the severity and site of bleeding. Preventive infusion of CFC
the joint. It is unusual for an infant to suffer spontaneous
is the gold standard of care (prophylaxis therapy). Patients
hemarthroses in the first few months of life, and the first
regularly inject CFC intravenously (most commonly two to
joint to be affected tends to be the ankle as the child learns
four times a week) to prevent bleeds, rather than just treating
to crawl. The first sign of a hemarthrosis in an infant will
on demand after bleeds or an injury has occurred. The dose
often be obvious discomfort and distress, accompanied by
and frequency of prophylaxis infusions are determined by the
limping or reluctance to use a limb. Recurrent bleeds into a
half-life of the chosen CFC and increasingly after identify-
joint lead to synovitis and joint damage, ultimately resulting
ing the pharmacokinetics (PK) of that CFC in the candidate
in crippling arthritis (Figure 17.2). Bleeding into muscles is
patient. Prophylaxis aims to maintain a measurable plasma
also a feature of hemophilia, but this is usually a consequence
level of clotting factor activity above the patient’s baseline
of direct injury, albeit often minor (Figure 17.3). Bleeds into
(trough level) just prior to the next CFC prophylaxis dose.
certain areas are particularly dangerous because of the risk of
Patients on prophylactic therapy experience few or even no
compression of neighboring structures, manifesting then as a
spontaneous bleeds, and thus progressive joint damage and
compartment syndrome. Patients with inhibitory antibodies
arthritis can be minimized or avoided. The quality of life of
are particularly at risk in this regard, as bleeds may be more
patients on prophylaxis may be greatly enhanced, allowing
difficult to control. Bleeds in the tongue can obstruct the air-
them to lead much more independent lives.
way, and retroperitoneal bleeding within the iliopsoas muscle
Approximately 30% of patients with severe hemophilia A
may result in femoral nerve compression, causing paresthesia
can be expected to develop inhibitory antibodies to factor
VIII CFC at some stage. In contrast, inhibitor development
Molecular Hematology, 4th Edition. Edited by Drew Provan and John G. Gribben.
in hemophilia B is very rare and encountered in less than
© 2020 John Wiley & Sons Ltd. Published 2020 by John Wiley & Sons Ltd. 5% of patients. The development of such antibodies poses

221
222 Molecular Hematology

Table 17.1 The relation of blood levels of factor VIII (or IX) to the severity of hemorrhagic manifestations

Level (IU dl−1 ) Hemorrhagic manifestations

50–150 Normal level, no bleeding problems

25–50 No problems in day-to-day life. Tendency to bleed after major surgery
5–25 Mild hemophilia. Bleeding typically occurs only after significant injuries
2–5 Moderately severe hemophilia: spectrum of bleed phenotype – some with spontaneous bleeds and early arthropathy,
others experiencing bleeds associated with injury, albeit often relatively minor
<1 Severe hemophilia with spontaneous and recurrent bleeding into muscles and joints. Risk of intracranial hemorrhage.

Fig. 17.2 Radiograph of the knee of a patient with severe

hemophilic arthropathy. Joint replacement surgery was
Fig. 17.1 Acute hemarthrosis in severe hemophilia. This usually subsequently carried out in this case.
arises in the absence of injury. The joints most frequently involved are
the knees, elbows, and ankles. The joint is swollen, warm, and tender,
but there is no external bruising or discoloration.

considerable problems in treatment, as these immunoglob-

ulins (IgG) are capable of rapidly inactivating infused factor
VIII, and furthermore the antibody titer may rise dramat-
ically after a course of factor VIII (anamnesis). Very occa-
sionally, acquired hemophilia A may arise in a previously
normal individual due to the formation of autoantibodies
directed against factor VIII, and both males and females may
be affected. Hemarthrosis is unusual in acquired hemophilia,
and the principal manifestations are usually extensive super-
ficial purpura and muscle bleeds. Acquired hemophilia Fig. 17.3 Magnetic resonance imaging (MRI) scan showing
arises most often in the elderly, associated with underly- bilateral iliopsoas hemorrhage. This bleed was associated with a
ing malignant or autoimmune diseases in approximately complete but transient paralysis in both legs, as the femoral nerve is
half of cases, with a small, younger cohort associated with located on the anterior surface of the muscle and may be compressed
pregnancy. in such cases.
 The molecular basis of hemophilia 223

Turner syndrome (XO karyotype) and androgen insensitivity

Inheritance of hemophilia syndrome (XY karyotype). A proportion of females carrying
a hemophilia gene will have factor activity levels, from that
The genes for factors VIII and IX (F8 and F9, respectively) are
corresponding gene, below recognized laboratory normal
both located at the telomeric end of the X chromosome and
ranges through a process called skewed lyonization. These
thus hemophilia is inherited as an X-linked recessive condi-
girls/women effectively have mild hemophilia and should be
tion (Figure 17.4). The daughters of affected males are obli-
identified and offered care commensurate with their levels,
gate carriers, but the sons are normal. The phenotype remains
as would be offered to boys/men with equivalent factor lev-
constant within a family, so the daughter of a man with only
els. The remaining carrier women have normal laboratory
mild hemophilia may be reassured that she will not pass on a
range factor activity in their plasma. Importantly, normal
severe form of the condition. However, up to a half of all cases
plasma factor activity cannot be used as a surrogate marker
of hemophilia arise in the absence of a previous family history
of whether a daughter of a carrier mother is a carrier her-
and are due to a new mutation. The most famous example is
self. Genetic testing will always be required to clarify true
that of Queen Victoria, who had a hemophilic son (Leopold)
carriership.
and two daughters (Alice and Beatrice) who turned out to be
carriers. Forensic genetic testing on exhumed remains of the
assassinated Russian royal family has since proven the royal
hemophilia to be severe hemophilia B. There are instances of Molecular basis of hemophilia A
hemophilia affecting females due to inheritance of the defec-
tive gene from both parents (see Figure 17.4, generation IV), Factor VIII is an essential cofactor for the activation of fac-
and there are also case reports of hemophilia in females with tor X by activated factor IX (see Chapter 16). Factor VIII

1 2
I

II 1 2 3 4 5 6

III 1 2 3 4 5 6 7 8 9

IV 1 2 3 4

Normal Female Homozygous Normal Hemophilic

female carrier female male male

Fig. 17.4 The inheritance of hemophilia. The genes for factors VIII and IX are both encoded on the X chromosome, and inheritance is thus sex
linked. The daughter of a man with hemophilia is an obligate carrier, but the son of a father living with hemophilia will not be affected.
Hemophilia may thus be transmitted to a grandson via a carrier daughter. Color blindness and Duchenne muscular dystrophy are other examples
of conditions that are X linked.
224 Molecular Hematology
X chromosome
q28
G6PD
qter 100 kb Factor VIII
10 kb
1 2 3 4 5 6 7 8 9 101112 13 14 15 16 17 18 19 20 21 22 23 24 25
5'
Exon 22
F8A F8B
must itself undergo proteolytic cleavage at two distinct sites
through the action of thrombin before it becomes physiolog-
ically active. It circulates in plasma as a large glycoprotein
bound non-covalently to the larger protein, von Willebrand
factor (VWF). The factor VIII gene (F8) was first cloned in
1984. It is 186 kb in length and is situated on the long arm
of the X chromosome at Xq28 (Figure 17.5). The factor VIII
gene consists of 26 exons, which range in size from 69 bp
(exon 5) to 3.1 kb (exon 14). The factor VIII message is nearly
9 kb in size and encodes a mature protein of 2332 amino
acids. Approximately half of all cases of severe hemophilia
and all cases of mild and moderate hemophilia result from
heterogeneous mutations that occur throughout the F8 gene.
By far the commonest single genetic defect causing
severe hemophilia is an inversion in intron 22, which is
FVIII
cen
q28
tel
Fig. 17.6 “Flip tip inversion.” This unique inversion of the tip of the X chromosome is now recognized to be responsible for about half of all
cases of severe hemophilia A in all ethnic groups. Crossover occurs between a copy of F8A within the F8 gene and one of the two telomeric
copies. Crossover with the distal copy is more common, occurring in approximately 80% of cases where an inversion is identified.
Fig. 17.5 The factor VIII gene. The factor VIII gene
(F8) was cloned in 1984 and is located toward the
telomeric end of the long arm of the X chromosome
(Xq28). It maps distal to the gene encoding glucose
6-phosphate dehydrogenase (G6PD), about 1 Mb
from the Xq telomere. The gene is composed of 26
26
exons, of which exon 14 is the largest. The large
intron 22 contains a nested intronless gene termed
3' F8A (the function of the gene and its transcript are
unknown). There are two further copies of F8A
located about 400 kb telomeric to the factor VIII
gene. The spliced factor VIII mRNA is ∼9 kb in length
Exon 23
and the mature factor VIII molecule is composed of
2332 amino acids.
encountered in as many as 45% of people with severe
hemophilia A in all ethnic groups (Figure 17.6). The inver-
sion mechanism involves an intronless gene of unknown
function, designated F8A. Two copies of this gene are located
near the tip of the X chromosome and there is another copy
within intron 22 of the factor VIII gene itself. During meio-
sis, either of the two telomeric copies may cross over with
the intronic copy, resulting in a division of the gene into
two halves facing in opposite directions and separated by
approximately 500 kb. Crossover with the distal copy is much
more common than crossover with the proximal copy, and
accounts for approximately 80% of all inversions. It is now
recognized that inversion almost always forms during a male
meiosis. It is believed that the presence of a large region of
non-homology between the X and Y chromosomes during
3' tel 3'
FVIII mRNA3'
F8B
F8A
F8B
5' F8A
F8A
5' mRNA5'
F8A FVIII
F8A F8A
tel tel

meiotic pairing may favor a misalignment, and the presence
of a second X chromosome with a complementary region
may act as a stabilizing factor. An important clinical conse-
quence of this observation is that when an apparently new
and spontaneous case of hemophilia is diagnosed in which
the gene inversion is identified, it is likely that the defect arose
in the maternal grandfather’s allele (during spermatogene-
sis), and thus the mother can generally be assumed to be a
carrier and at risk of having another affected male child. The
resulting truncated protein product is presumably unstable,
resulting in severe hemophilia. The inversion is not found in
individuals with non-severe forms of hemophilia. The inver-
sion is easily detected and the identification of this defect
as the commonest cause of severe hemophilia has simpli-
fied both screening for carriers and antenatal diagnosis of
hemophilia. The inversion may be identified with Southern
blotting, although many laboratories now use either inverse
polymerase chain reaction (PCR) or long-range PCR for
identification of this inversion. More recently, inversions in
intron 1 of the F8 gene have been identified as a cause of
severe hemophilia and this abnormality appears to be respon-
sible for approximately 5% of all cases of severe hemophilia A.
Since approximately half of all cases of severe hemophilia are
associated with these two inversions, it is usual practice to
screen samples from new cases for these two abnormalities
first.
Developments in molecular biology have permitted more
rapid identification of defects in hemophilia. Southern blot-
ting has been superseded by methods involving PCR ampli-
fication of either patient DNA or material derived from
the reverse transcription of mRNA (RT-PCR). Automated
sequence analyzers enabling sequencing of the entire F8
gene are now more affordable. Where not available, pre-
vious methods may still be relevant to identify restricted
areas of abnormal DNA in patients with hemophilia, which
may then be targeted for specific attention. These meth-
ods include amplification and mismatch detection (AMD),
conformation-sensitive gel electrophoresis (CSGE), denatur-
ing gradient gel electrophoresis (DGGE), high-resolution
melting analysis (HRM), and pyrosequencing. Approxi-
mately 4% of cases of hemophilia are the consequence of gene
deletions, which have been reported throughout the gene
and which are very variable in size. As with the intron 22
inversion, most deletions are associated with a severe clin-
ical phenotype. To date, around 650 different single base
substitutions have been described, of which approximately
75% predict a single amino acid change from the wild-type
sequence (missense). Such mutations affect RNA processing,
mRNA translation, or the fine structure of factor VIII itself.
The majority of these missense F8 mutations result in non-
severe forms of hemophilia A. A further 100 lead to the cre-
ation of preliminary peptide chain termination or prema-
ture stop codons and the production of truncated factor VIII
The molecular basis of hemophilia 225
molecules devoid of any functional activity. Frameshift muta-
tions resulting from insertions or small deletions have also
been identified as a cause of severe hemophilia. Duplications
within one or more exons of the F8 gene have recently been
identified in a small minority of patients with hemophilia
A using multiplex ligation-dependent probe amplification
(MLPA). A full list of mutations described in association with
hemophilia is outside the scope of this chapter, but addi-
tional information is provided in the Further Reading section
or via the F8 variant database (http://www.factorviii-db.org).
Knowledge of an individual’s F8 mutation has relevance to
predicting inhibitor risk: the more severe the mutation and
the resultant absence of translated protein, the higher the
risk. However, even single amino acid differences of mis-
sense mutations can still provoke an alloimmune response,
resulting in clinically impactful rejection antibody formation
(inhibitor).
Approximately 40% of all missense mutations arise at
CG dinucleotide sites, resulting in a change to TG or
CA sequences. It is generally believed that CG nucleotides
represent genomic hotspots. Cytosine is predominantly
methylated in human DNA, but this is relatively unstable
and 5-methylcytosine is prone to spontaneous deamination
to yield a GT mismatch that is inefficiently repaired. It is also
of interest that a missense mutation may be associated with
varying degrees of clinical severity. Thus, a C → T mutation
at nucleotide 1689 within exon 14 resulting in replacement
of arginine by cysteine has been reported in association with
both severe and mild clinical phenotypes.
Molecular basis of hemophilia B
The factor IX gene (F9) is also located on the long arm
of the X chromosome at band Xq27, and is encoded by a
stretch of DNA spanning 33.5 kb that contains eight exons
(Figure 17.7). The basic structure of the gene is similar in
organization to that of protein C and coagulation factors
VII and X, and it is likely that they all originated in the
distant past from a common ancestral gene by duplication.
Factor IX mRNA comprises 2.8 kb and encodes a mature
protein of 415 amino acids. This is made up of a glutamic
acid–rich sequence (Gla domain) and two epidermal growth
factor (EGF)-like domains separated from the serine pro-
tease domain by an activation region. The 12 glutamic acid
residues in the Gla domain undergo post-translational γ-
carboxylation, which is necessary for binding of calcium,
and exon 2 encodes a recognition site for the carboxylase.
Exon 1 encodes the signal peptide necessary for transport
into the endoplasmic reticulum. Exon 6 encodes the activa-
tion peptide that is cleaved off during the activation of fac-
tor IX by either factor XI or a complex of tissue factor and
factor VII. Exons 7 and 8 encode the catalytic regions of
226 Molecular Hematology
0 10 20
1 2 3 4 5 6 7
Exons
–46 –18 +1 39 47 85 128 145 180
PRE PRO GLA H EGF–type B EGF–type A Activation
Fig. 17.7 The factor IX gene (F9). The factor IX gene (F9) was first cloned in 1982. It is located on the long arm of the X chromosome at band
Xq27. The gene spans 34 kb and encodes eight exons (exons are shown as colored boxes, and dotted lines between the gene and protein indicate
protein domains encoded by each exon). The signal peptide and propeptide sequences are cleaved during processing and activation of factor IX.
EGF, epidermal growth factor domains; GLA, γ-carboxyglutamic acid domain.
factor IX, which are responsible for subsequent activation of
factor X in the coagulation cascade. The gene is controlled by
a promoter.
The F9 gene, cloned in 1982, is considerably smaller than
F8. The first defects identified in hemophilia B were gross
deletions, detected by Southern blotting. However, it is now
recognized that gene deletions account for only approxi-
mately 3% of all cases of hemophilia B. No equivalent of
the F8 gene inversion has been encountered in hemophilia
B and it is now clear that point mutations account for the
vast majority of cases of hemophilia B; over 500 have been
described from families around the world. The great major-
ity involve single base changes, which have been identified in
all domains of the protein. In contrast to missense F8 muta-
tions causing non-severe hemophilia A, inhibitor formation
in F9-missense, non-severe hemophilia B is unheard of, the
explanation for which is unknown. The unusually high fre-
quency of mutations at CG dinucleotide sites in hemophilia
B probably reflects the high number of CG dinucleotides at
critical sites in the F9 gene. Exon 8 is the largest at 1.9 kb in
length and half of all mutations are found in this exon. How-
ever, about 20–30% of cases of mild hemophilia B are due
to a small number of founder mutations. The original case
of Christmas disease (named after the first patient, Stephen
Christmas, described in 1952) has been identified as a G → C
mutation at nucleotide 31170, resulting in substitution of cys-
teine by serine within exon 8.
Mutations in the promoter region of the F9 gene (e.g.
T → A at −20 and G → A at −6) are relatively rare and
account for around 2% of all cases. However, they are of par-
ticular interest as they can give rise to the unique hemophilia
B Leyden phenotype, where the factor IX level rises signifi-
cantly after puberty, with loss of the bleeding tendency. Most
of these mutations have been shown to be located in regions
30 kb
8
F.IX gene
415
Catalytic F.IX protein
that contain binding sequences for liver-enriched transcrip-
tion factors, which are presumably influenced by androgenic
steroids.
A gain-of-function F9 mutation R338L, F9 Padua (substi-
tution of leucine for arginine at position 338), described in an
Italian family after a case of juvenile thrombophilia, is para-
doxically of particular interest in gene therapy strategies for
severe hemophilia B. The R338L gene sequence is being used
in early-phase clinical trials to increase the detectable levels
of FIX protein function to close to the normal FIX activity
range after in vivo gene therapy.
Full details of the many genetic abnormalities associated
with hemophilia B can be found at the websites listed at the
end of this chapter.
Inhibitor formation: etiology and
clinical implications
A proportion of patients with hemophilia will develop
alloimmune immunoglobulins directed against infused fac-
tor VIII (or IX) CFC after exposure to these therapeutics for
treatment of bleeding episodes. The individual’s immune sys-
tem perceives the infused, wild-type CFC as “foreign,” given
that the individual was not born with and is thus not toler-
ant to the “normal,” wild-type protein sequence when first
exposed to the infused CFC. This is potentially very serious,
as individuals become refractory to conventional doses of
CFCs and bleeding can be very difficult to control. Inhibitory
antibodies interfere with the normal function of factor VIII
in a number of different ways. The most frequent site of
inhibitor binding occurs within the A2 and C2 domains.
Inhibitors may thereby block the ability of activated factor
VIIIa to bind and activate factors IXa and X, or inhibit the

binding of factor VIII to VWF or negatively charged phos-
pholipid surfaces. Inhibitors may also hinder the activation
of factor VIII by thrombin, or the subsequent release of factor
VIII from VWF. Proteolysis of factor VIII has recently been
identified as a novel additional mechanism of inactivation in
some cases.
Activated prothrombin complex concentrates (e.g.
FEIBA®) and recombinant activated factor VIIa (Novo-
Seven®) are valuable therapeutic materials in controlling
bleeding in those patients with inhibitory antibodies. Recom-
binant porcine factor VIII (susoctocog alpha, Obizur®) has
also been developed and is emerging as an additional,
licensed therapeutic agent, but only in acquired hemophilia
A currently. The rationale is that the porcine molecule is
sufficiently similar (homologous) to human factor VIII to
exert a hemostatic effect, but at the same time sufficiently
different to avoid rapid neutralization by antibodies. An
alternative novel therapeutic has harnessed a biphenotypic
monoclonal antibody (Emicizumab, Hemlibra®) to mimic
the function of the FVIII molecule, bringing activated FIX
in proximity to FX to activate it. This has recently been
licensed for prophylaxis in chronic inhibitor hemophilia A
patients, but is not sufficiently efficacious to treat serious,
breakthrough bleeds or cover major surgery, at which time
care is needed in choice of additional hemostatic agents
to treat bleeds, due to potential thrombotic complications.
This molecule may also become available for prophylaxis
in those living with severe hemophilia A without a chronic
inhibitor.
Another important strategy in the management of these
patients who develop inhibitory antibodies is immune toler-
ance induction (ITI), which involves the daily administration
of conventional CFC over a period of some months. This usu-
ally results in the eventual disappearance of the antibody, as
the body becomes tolerant of the protein and inhibitor for-
mation is suppressed.
Following the elimination of the risk of transmission of
viruses by CFCs, the risk of inhibitor development remains
the principal danger faced by people with hemophilia nowa-
days. Data from the UK registry suggest that approximately
30% of all patients with severe hemophilia A will develop
antibodies at some time, half of which are transient and
low-titer inhibitors. The majority of inhibitors in severe
hemophilia A occur in previously untreated patients (PUPs)
early in life, during the first 50 exposure days (EDs) to the
therapeutic CFC, with median occurrence at approximately
10–15 days in severe hemophilia A. However, in non-severe
hemophilia A, the risk is now thought to be lifelong and per-
sists after 50 exposure days. In this context, the alloantibody
has the potential to cross-react with the individual’s endoge-
nous baseline FVIII activity level, decreasing their baseline to
potential severe levels (<1%), with a resultant deterioration
in bleed phenotype. There is also a low-level occurrence of
The molecular basis of hemophilia 227
HIGH RISK
75% multidomain
Large
deletions light chain
Nonsense
mutations
single domain Intron 22
heavy chain inversions
Non A-run C1–C2 junction
Small Missense
deletions
0% Non C1–C2 junction
A-run
Splice-site
LOW RISK mutations
Fig. 17.8 Mutation types and risk of inhibitor development.
inhibitor detection in previously treated patients (PTPs) with
severe hemophilia A, despite years, often decades, of regular
exposure to FVIII CFC.
It is now clear that the major factor which determines
the predisposition to inhibitor development is the underlying
molecular defect. Certain types of gene defects in hemophilia
are undoubtedly associated with a significantly increased risk
of inhibitor development (Figure 17.8). The risk of inhibitor
development in patients with severe molecular defects, such
as large deletions, nonsense mutations, and the intron-22
inversion, is 7–10 times higher than in patients with other
defects such as missense mutations, small deletions, and
splice-site mutations. The overall risk of inhibitor develop-
ment in patients with the common intron-22 inversion is
approximately 30%.
However, there is additional evidence from family and
twin studies that other subtle genetic factors play a role,
although no associations with specific human leukocyte
antigens (HLAs) or other linkages have been conclusively
identified. Race may also influence the risk of inhibitor
development, and several studies have shown that people of
Afro-Caribbean origin are more susceptible to inhibitor for-
mation. Links between structural variants (polymorphisms)
in immune-response genes have been reported. Such variants
might have an impact on inhibitor development by increas-
ing the production and secretion of chemicals that ultimately
enhance the production of antibodies directed against factor
VIII. In the case of interleukin (IL)-10, one particular poly-
morphism (134-bp variant of a CA repeat microsatellite in
the promoter region of the IL10 gene) was found to be associ-
ated with a 4.4-fold increased risk of inhibitor development.
This same polymorphism has already been shown to influ-
ence the level of antibody production in such diverse diseases
as myasthenia gravis, multiple myeloma, and systemic lupus
erythematosus. More recently, the same group reported that
a particular polymorphism (C/T at −318) in the cytotoxic
228 Molecular Hematology
T-lymphocyte-associated protein (CTLA)-4 receptor of
certain immune cells actually confers protection against
inhibitor development. Other than knowledge of a family
history of inhibitor formation which confers increased risk,
it continues to be challenging to risk stratify inhibitor risk by
other genetic markers. It has become clear that CFC product
choice may be contributory to inhibitor risk in PUPs, either
between recombinant products or between a recombinant
(rFVIII) and plasma-derived, VWF-containing product
(pdFVIII). The SIPPET group described the benefit of the
latter choice of pdFVIII over rFVIII CFC to be restricted
to the minority of individuals with less disruptive F8
mutations.
Inhibitor development in hemophilia B is a rare event,
occurring in probably less than 3% of patients, and these
patients often have an underlying large gene deletion. In
contrast with the immunoglobulin inhibitors in patients
with hemophilia A, inhibitory antibodies in patients with
hemophilia B are often capable of fixing complement pro-
teins. The administration of CFC in such cases may there-
fore trigger severe, often anaphylactic, allergic reactions. The
development of nephrotic syndrome has also been reported
in patients with hemophilia B and inhibitors undergoing
immune tolerance induction treatment with factor IX con-
centrates.
Therapeutic applications of molecular
biology to patient care
Carrier testing
Ideally, carriers of hemophilia should be identified before
a pregnancy, and offered counseling. The inheritance of
hemophilia is X-linked, as with other disorders such as color
blindness and Duchenne muscular dystrophy. The daughters
of men with hemophilia are thus obligate carriers of the
condition, with a subsequent 50 : 50 chance of passing on
the condition to their son, and there is a similar chance that
a daughter of a carrier will also herself be a carrier of the
condition. No special genetic tests are therefore required
to determine the carrier status of daughters of men with
hemophilia (unless paternity is uncertain), although the
results of DNA-based studies are likely to be useful for
subsequent antenatal diagnostic procedures. The phenotype
of hemophilia remains constant within a family, so that the
daughter of a man with only mild hemophilia may be reas-
sured that she can only transmit a similarly mild form of the
condition. However, a more common problem is to be con-
fronted with a woman with only a vague history of a bleeding
disorder in a distant relative. National patient registers can be
very helpful in establishing the type and severity of bleeding
disorder of an affected relative as a first step in determining
which tests need to be carried out. It may seem logical to
initiate carrier testing to determine carrier status as soon
as possible in girls with a family history of the condition,
as this would facilitate management in the case of an early
and possibly unexpected pregnancy. There are significant
differences in legislation as well as clinical practice among
healthcare professionals in various countries with regard to
the timing of testing of children for genetic disorders. Some
take the view that it is unethical to test very young children
in order to determine carrier status for inherited disorders
for conditions which have no immediate implications for
their own health. In the UK, most hematologists would
generally be prepared to offer genetic confirmatory carrier
testing to girls in their early teens, with the proviso that
the issues must be discussed with the family and the child
deemed able to understand the implications of such testing.
It is important to check FVIII or FIX plasma activity lev-
els in obligate and possible carriers (see Figure 17.4) soon
after birth. It should be emphasized that these plasma lev-
els should not be used to determine whether a female is a
carrier of hemophilia, and that only DNA-based tests should
be used. Indirect testing of carrier status using the tracking
of restriction fragment-length polymorphisms (RFLPs) has
now been superseded by direct identification of the underly-
ing causative genetic abnormality, such as mutation or inver-
sion. Once the molecular defect has been identified in an
individual with either hemophilia A or B, direct screening
for that defect could be applied in subsequent generations
for both carrier testing and antenatal diagnosis. It is recom-
mended that hemophilia centers collate information on the
family pedigree (“family tree”) to identify potential carriers
and facilitate screening.
RFLP analysis may still be employed in the few cases
where, for example, a mutation has not been identified or ver-
ified in a potential carrier. This technique is also still widely
used in countries around the world that do not yet have
access to more sophisticated technology. The method is based
on the fact that there are some genetic polymorphisms that
represent natural variations of the genome sequence, with-
out any adverse impact on the function of the molecule.
The polymorphisms are detected by cleavage of patient DNA
with restriction enzymes, which generate fragments of vary-
ing size according to the presence or absence of the poly-
morphism. Only intragenic markers should be used, because
the small possibility of recombination may result in erro-
neous diagnoses when extragenic probes are used. The most
widely intragenic markers used are dinucleotide repeats in
introns 13 and 22; XbaI dimorphism in intron 22; and diges-
tion products using BclI in intron 18. The limitations of
this approach include the fact that samples have to be taken
from several members to permit the tracking of the mutant
gene responsible for hemophilia in the family. Blood from an
affected family member will be required, and this may not be

possible in some cases. It should also be appreciated that there
is ethnic variation of the allelic frequencies of the various
polymorphisms, more so with factor IX than factor VIII, and
this may influence the choice of probes used in some family
studies. Furthermore, non-paternity may confound attempts
to track the gene in families.
Germline and somatic mosaicism may complicate the pic-
ture. This needs particular consideration in cases of sporadic
hemophilia where the mother of a child with hemophilia does
not appear to carry the mutation in her own leukocyte DNA.
This has been reported to be a particular problem when the
apparently de novo mutation is a point mutation. For this rea-
son, it is recommended not to state that the mother of a child
with hemophilia is not a carrier, even when the mutation is
not identified in her somatic DNA.
Antenatal diagnosis of hemophilia
As a general rule, it is the practice in the UK to perform
antenatal procedures to determine whether or not a fetus
has hemophilia only where a termination is being contem-
plated. The general experience in the UK has been that only
a minority of women subsequently take up the offer of ante-
natal diagnosis with a view to termination if an affected fetus
is identified. This reflects the fact that many women with
affected relatives appreciate the major treatment advances in
recent years, such as recombinant products for lifelong pro-
phylaxis, the expectation of an essentially normal life quality
and life expectancy for the younger generation of people with
hemophilia, and the imminence of potentially curative gene
therapy.
Amniocentesis was the first technique employed for ante-
natal diagnosis of hemophilia and other X-linked disorders
such as muscular dystrophy. While amniocentesis is both
technically simple and safe, an important limitation is the
fact that it may only be employed in the second trimester of
pregnancy, after approximately 15 weeks’ gestation. Chori-
onic villus sampling (CVS) was first applied to antenatal
diagnosis of a number of genetic disorders in the early
1980s, but is now the principal method used for antenatal
diagnosis of hemophilia and several other single-gene
disorders. The main advantage is that the method may be
applied for antenatal diagnosis during the first trimester,
so that if termination of the pregnancy is required this is
easier to carry out. Furthermore, the results of the test are
often available within only a few days of the procedure, as
(in contrast to amniocentesis) there is no need to culture
cells before genetic analysis. A sample is obtained by either
a transabdominal or transvaginal route, under ultrasound
guidance (Figure 17.9). CVS should not be undertaken
before 11 weeks of pregnancy in order to minimize the risk
of inducing congenital limb abnormalities. It is now common
for there to be non-invasive, prenatal determination of the
The molecular basis of hemophilia 229
Uterine wall
Amniotic fluid
Placenta
Chorionic villi
Fig. 17.9 Chorionic villus sampling. A sample of trophoblastic
tissue from the placental area is aspirated with a fine needle under
general anesthesia. The procedure is usually carried out after
11 weeks’ gestation. DNA isolated from the fetal tissue can then be
analyzed to determine fetal sex and status with regard to hemophilia.
fetal gender early in the first trimester (as early as five weeks’
gestation) by free-fetal DNA analysis (detecting a Y chro-
mosome in fetal cells identified in the maternal circulation).
This result then contributes to the discussion on whether to
pursue more invasive confirmatory testing for the presence
of the mutated F8 or F9 gene, and generally obviates the
need for CVS in the presence of a female fetus. An imminent
extension to this non-invasive assay will be for full antenatal
diagnostics.
Direct fetal blood sampling may also be used for antena-
tal diagnosis of hemophilia, but this method is usually only
offered as a last resort, either because it was not possible
to carry out DNA-based family studies in time or because
such studies were carried out but did not yield results. In
this technique, fetal blood is taken from fetal umbilical ves-
sels under ultrasound guidance. The procedure requires con-
siderable expertise and will thus not be available in all hos-
pitals. It is usually carried out at a minimum of 18 weeks’
gestation. The levels of factor VIII and IX in a normal
fetus at around 19 weeks’ gestation are significantly lower
than those in an adult, approximately 40 and 10 IU dl−1 ,
respectively.
The need to abort an implanted fetus is a formidable
ethical concern for many female carriers of hemophilia. Pre-
implantation diagnosis is another technique that has been
developed, and this may prove to be particularly attractive to
women who would not be prepared to undergo conventional
termination of a well-established pregnancy. The female
partner must undergo an in vitro fertilization cycle and the
230 Molecular Hematology
fertilized ova are then biopsied. Embryos shown to be unaf-
fected by hemophilia can then be transferred to the uterus.
The initial approach was to offer determination of embry-
onic sex alone using dual fluorescence in situ hybridization
of blastomere cells with labeled probes specific for the sex
chromosomes. More recently, specific mutation analysis
has become possible, although it must be emphasized that
this has only been performed in a very limited number of
cases and is still far from being a service that is available
on a routine basis even in major hospitals. Sperm sorting
has been developed as a cheaper and easier method for
pre-implantation sex determination, using flow cytometry to
sort spermatozoa labeled with a fluorescent dye that binds to
DNA. The resulting higher fluorescence of X chromosomes,
which contain more DNA than Y chromosomes, facilitates
separation.
Recombinant blood products
The development of plasma-derived CFCs in the early 1970s
dramatically improved both the longevity and quality of life
of patients with hemophilia, and the demand for factor VIII
and IX has risen steadily. The burgeoning global demand for
factor VIII can no longer be met by products derived from
volunteer blood donors. The manufacture of recombinant
coagulation factor proteins offers the promise of unlimited
supplies, albeit at increased cost. However, the most impor-
tant advantage of recombinant products is safety with regard
to the transmission of human pathogens. Many patients with
hemophilia were infected with HIV and/or hepatitis C before
the introduction of physical methods of viral inactivation
of plasma-derived CFCs in 1985. More recently, there has
been concern about the possibility of transmission of vari-
ant Creutzfeldt–Jakob disease (vCJD) via blood products, as
the prions believed to be the cause of this neurological dis-
order are extremely resistant to the usual viral inactivation
procedures, such as heat treatment and exposure to a sol-
vent/detergent mixture. Many clinicians now regard recom-
binant products as the treatment of choice for all patients with
hemophilia, as they offer the best possible protection from
the transmission of blood-borne pathogens. However, the
increased cost with regard to conventional plasma-derived
products has limited the availability of these products in
many parts of the world.
Recombinant CFCs are manufactured by insertion of the
most common H1 or H2 F8 human haplotype gene into
mammalian cell lines – such as Chinese hamster ovary
(CHO) cells or baby hamster kidney (BHK) cells – which
are then grown in culture on an industrial scale. Factor
VIII (or IX) is secreted into the growth medium, from
which it is subsequently extracted by monoclonal or other
immunoaffinity chromatography (Figure 17.10). The origi-
nal recombinant factor VIII products all contained added
human albumin as a stabilizer. However, third- and fourth-
generation products are now available in which alternative
stabilizers are used. Human albumin and all bovine pro-
teins have also been eliminated from the culture media
of these modern products, and the incorporation of spe-
cific viral elimination/inactivation steps further increases the
margin of safety. Standard half-life recombinant factor VIII
has an essentially identical amino acid structure to natu-
ral plasma factor VIII, although increasingly the apparently
functionless B domain of the F8 gene is deleted or truncated
to optimize production efficiency. Human cell (HEK293)-
derived recombinant factor VIII became available for the
first time recently as a standard or extended half-life prod-
uct (see later). Evidence exists to demonstrate different pro-
tein post-translational modifications between human- and
CHO- or BHK-derived CFC products. It remains unclear
whether this translates into human cell-derived rFVIII prod-
ucts having a reduced immunogenicity compared to CHO-
or BHK-derived products. Additional developments include
further structural modifications to increase the half-life of
the infused FVIII CFC by fusing the FVIII protein to an
immunoglobulin Fc stem, enabling apparent recycling via the
endothelial surface neonatal Fc receptor (FcR). Glycopegyla-
tion is an alternative strategy for half-life prolongation. Both
strategies have increased the half-life of FVIII comparably by
1.5-fold.
Recombinant factor IX is also available. Recombinant
factor IX is identical in amino acid sequence to the Ala148
(as opposed to the less common Thr148) human polymor-
phic variant. Plasma factor IX is synthesized in the liver and
undergoes post-translational glycosylation of a number of
glutamic acid residues. Vitamin K is a vital cofactor in this
process, which is essential for its activity, but recombinant
factor IX is not as effectively carboxylated. The post-infusion
recovery does appear to be reduced when compared with
plasma-derived products, although the plasma half-life
is identical. It is a smaller molecule than factor VIII and
requires no albumin, or other material, to be added to
the final product as a stabilizer. The cell line is grown in
media that contain no animal or human-derived proteins,
but the product is subjected to nanofiltration to enhance
its safety profile. There is no suggestion of an increased
risk of inhibitor development associated with the use of
recombinant factor IX.
Factor IX has more complex pharmacokinetics than FVIII,
as a smaller protein that is distributed extravascularly and
also binds to collagen IV. Consensus on the true half-life of
FIX is debatable, but it is likely longer than the previously esti-
mated 18 hours due to this apparent complex multicompart-
mental pharmacokinetics. Strategies to prolong FIX half-life
in the intravascular space have proved successful with either
glycopegylation, fusion to albumin or the immunoglobulin
Fc molecule. Although prolonging detectable FIX activity in
 The molecular basis of hemophilia 231

FVIII
vWF

(Circle of bacterial DNA that replicates)

(a) Plasmid vector (b) Isolate human gene sequence and
splice into a suitable vector
FVIII
vWF

Nucleus

Cells in
nutrient media

Vial of cells is “cultured” in

media, cells multiply into
CHO cell Insert recombinant
millions of cells, each
plasmid into host cell
expressing FVIII into the
media
(c) Chinese hamster ovary (CHO) cell (d) Bioreactor
FVIII
Impurities

Impurities pass
Column is washed
through column
and rFVIII protein
and rFVIII binds
is released
to antibodies

Impurities
rFVIII
Protein
(e) FVIII is purified from media using a monoclonal
Ab column and ion-exchange columns
Fig. 17.10 Manufacture of recombinant factor VIII. The gene for human factor VIII is incorporated into a bacterial vector (a, b). Inclusion of
the von Willebrand factor gene also enhances production of factor VIII. The vector is then inserted into a mammalian host cell (c). The cells grow
and multiply in a nutrient medium, and then secrete factor VIII (d). Factor VIII can be extracted and purified by immunoaffinity chromatography (e).
The final product also contains no von Willebrand factor.

the intravascular space three- to fivefold, the impact of each Gene therapy for hemophilia
of these novel technologies on extravascular distribution of
Gene therapy (see also Chapter 23) poses a number of ethi-
FIX is incompletely understood.
cal problems, particularly since effective and safe treatment
Another useful recombinant product is recombinant acti-
with recombinant coagulation factors is already available for
vated factor VII (NovoSeven; Novo Nordisk). It is now recog-
patients with both hemophilia A and B. Gene therapy does
nized that factor VII plays a key role in initiation of the coag-
offer the prospect of a cure for hemophilia in the long term,
ulation cascade through contact with tissue factor released
but it must be emphasized that although clinical trials in
from damaged tissues, to form activated factor VII (see also
both hemophilia A and B are progressing well into later
Chapter 16). Recombinant activated factor VII is very useful
phases, there is still a need for long-term surveillance for
in the clinical management of patients with either hemophilia
potential side effects. Trials are all currently in adults and it
A or B and inhibitory antibodies, as well as those with
will be some time before incrementally younger children are
acquired hemophilia.
232 Molecular Hematology
included in trials. It should also be noted that gene therapy
does not correct the germ line F8 or F9 mutation.
There are two basic approaches to gene delivery into cells.
The first technique involves the direct injection of transduc-
ing vector into the bloodstream or target tissue, with sub-
sequent in vivo transformation of the cells which take up
the gene, namely hepatocytes in hemophilia. Alternatively,
target cells may be modified by removal of cells from a
patient, with subsequent modification ex vivo of these cells
followed by reinfusion. The first strategy has been favored in
hemophilia trials, utilizing non-integrating, non-replicative,
hepatotropic adeno-associated viruses (AAVs). Wild-type
AAVs are communally acquired, intercurrent viral infections
of little consequence to an immunocompetent host. AAVs
exist in multiple serotypes, some more homologous than oth-
ers, and are less immunogenic than adenovirus vectors in the
gene therapy setting. Patients may have pre-existing immu-
nity to a given AAV serotype, currently making them ineligi-
ble for gene therapy. A dose of AAV gene therapy does gen-
erate high-titer antibodies against the specific AAV serotype
vector used which may cross-react with other serotypes,
making AAV gene therapy a single opportunity. Some, but
not all, serotypes trialed to date provoke a viral capsid-
specific T-cell response thought, at least in part, to be contrib-
utory to a complicating rise in liver-specific enzymes requir-
ing intervention with steroids.
Despite the recent success of non-integrative AAV strate-
gies, other gene therapy, gene editing, or base editing plat-
forms are also in development. Integration events provoke
concerns of oncogenesis. However, “safe-haven” gene edit-
ing may circumvent this issue, as is now being trialed in
hemophilia B, precisely inserting the F9 gene downstream of
the albumin promoter in hepatocytes using zinc finger nucle-
ase (ZFN).
The smaller F9 gene made it easier to package into the
modified AAV vectors for early study, yielding durable low-
level yet effective FIX expression in humans. Implementa-
tion of the aforementioned R338L gain of function F9 Padua
polymorphism has further improved expression in subse-
quent hemophilia B trials. Despite the larger F8 gene with
the anticipated problems of vector packaging, the first-in-
man hemophilia A gene therapy trial has normalized FVIII
activity levels in early participants, durable for at least a year.
For the first time, curative gene therapy seems to be a distinct
possibility.
Conclusions
Hemophilia is an inherited disorder of coagulation, asso-
ciated with congenital deficiency of factor VIII (or IX).
It is inherited in a X-linked fashion, so that males are
predominantly affected, but some female carriers will
also be symptomatic. Approximately half of cases arise in
families with no previous family history, and represent new
mutations. The typical features of severe hemophilia include
spontaneous bleeding into joints and resultant disability,
but in the absence of treatment more serious complica-
tions (such as intracranial hemorrhage) will lead to early
death.
The first products used for the treatment of hemophilia
were derived from human plasma, but unfortunately the use
of pooled plasma products before 1985 resulted in the trans-
mission of serious viral infections such as HIV and hepatitis
C to many patients. The development of recombinant blood
products has eliminated the risk of transmission of these
infections, and also offers the prospect of a secure supply
chain and half-life–extended products. The life expectancy
of the younger generation of hemophiliacs now approaches
that of the normal population.
The commonest molecular defect in severe hemophilia
A is an inversion in intron 22 of the factor VIII gene
on the X chromosome, which accounts for approximately
half of all cases. Genetic testing documents the genetic
defect in each family, and identifies carriers within fam-
ilies. Antenatal diagnosis is readily available, although in
countries with sustainable hemophilia care, recourse to ter-
mination of an affected pregnancy is rare. Recent trials
of gene therapy for hemophilia have yielded encouraging
results.
Hemophilia resources on the internet
Factor VIII Variant Database. Available at http://www.
factorviii-db.org
World Federation of Haemophilia. Available at www.wfh
.org
United Kingdom Haemophilia Centre Doctors’ Organisa-
tion (UKHCDO). Available at www.ukhcdo.org
European Association for Haemophilia and Allied Disor-
ders (EAHAD). Available at www.eahad.org
Further reading
General
Darby, S.C., Kan, S.W., Spooner, R.J. et al. (2007). Mortality rates, life
expectancy, and causes of death in people with hemophilia A or B
in the United Kingdom who were not infected with HIV. Blood 110:
815–825.
Lee, C.A., Berntorp, E., and Hoots, W.K. (eds.) (2005). Textbook of
Hemophilia. Oxford: Blackwell Publishing.
Zeepvat, C. (1998). Prince Leopold: The Untold Story of Queen Victoria’s
Youngest Son. Stroud: Sutton Publishing.

Hemophilia A
Antonarakis, S.E., Rossiter, J.P., Young, M. et al. (1995). Factor VIII gene
inversions in severe hemophilia A: results of an international consor-
tium study. Blood 86: 2206–2212.
Bagnall, R.D., Waseem, N., Green, P.M., and Giannelli, F. (2002). Recur-
rent inversion breaking intron 1 of the factor VIIII gene is a frequent
cause of severe hemophilia A. Blood 99: 168–174.
Davidson, C.J., Tuddenham, E.G., and McVey, J.H. (2003). 450 million
years of hemostasis. J. Thromb. Haemost. 1 (7): 1487–1494.
Keeney, S., Mitchell, M., and Goodeve, A. (2005). The molecular anal-
ysis of haemophilia A: a guideline from the UK Haemophilia Centre
Doctors’ Organization Haemophilia Genetics Laboratory Network.
Haemophilia 11: 387–397.
Leuer, M., Oldenburg, J., Lavergne, J.M. et al. (2001). Somatic mosaicism
in hemophilia A: a fairly common event. Am. J. Hum. Genet. 69: 75–
87.
Rost, S., Löffler, S., Pavlova, A. et al. (2008). Detection of large dupli-
cations within the factor VIII gene by MLPA. J. Thromb. Haemost. 6:
1996–1999.
Wood, W.I., Capon, D.J., Simonsen, C.C. et al. (1984). Expression of
active human factor VIII from recombinant DNA clones. Nature 312
(5992): 330–337.
Hemophilia B
Choo, K.H., Gould, K.G., Rees, D.J., and Brownlee, G.G. (1982). Molec-
ular cloning of the gene for human anti-haemophilic factor IX. Nature
299 (5879): 178–180.
Giangrande, P.L.F. (2005). Haemophilia B: Christmas disease. Expert
Opin. Pharmacother. 6: 1517–1524.
Mitchell, M., Keeney, S., and Goodeve, A. (2005). The molecular anal-
ysis of haemophilia B: a guideline from the UK Haemophilia Centre
Doctors’ Organization Haemophilia Genetics Laboratory Network.
Haemophilia 11: 398–404.
Reijnen, M.J., Peerlinck, K., Maasdam, D. et al. (1993). Hemophilia B
Leyden: substitution of thymine for guanine at position −21 results
in a disruption of a hepatocyte nuclear factor 4 binding site in the
factor IX promoter. Blood 82: 151–158.
Rogaev, E.I., Grigorenko, A.P., Faskhutdinova, G. et al. (2009). Geno-
type analysis identifies the cause of the “royal disease”. Science
326 (5954): 817.
Simioni, P., Tormene, D., Tognin, G. et al. (2009). X-linked throm-
bophilia with a mutant factor IX (factor IX Padua). N. Engl. J. Med.
361: 1671–1675.
Yoshitake, S., Schach, B., Foster, D.C. et al. (1985). Nucleotide sequenc-
ing of the gene for human factor IX (antihemophilic factor B). Bio-
chemistry 24: 3736–3750.
Inhibitors
Astermark, J., Donfield, S.M., Gomperts, E.D. et al., Hemophilia
Inhibitor Genetics Study (HIGS) Combined Cohort. (2013). The
polygenic nature of inhibitors in hemophilia A: results from the
Hemophilia Inhibitor Genetics Study (HIGS) Combined Cohort.
Blood 121 (8): 1446–1454. https://doi.org/10.1182/blood-2012-06-
434803.
The molecular basis of hemophilia 233
Astermark, J., Oldenburg, J., Escobar, M. et al. (2005). The Malmö Inter-
national Brother Study (MIBS). Genetic defects and inhibitor devel-
opment in siblings with severe hemophilia A. Haematologica 90: 924–
931.
Astermark, J., Oldenburg, J., Pavlova, A. et al. (2006). Polymorphisms
in the IL10 but not in the IL1beta and IL4 genes are associated with
inhibitor development in patients with hemophilia A. Blood 107:
3167–3172.
Collins, P.W., Chalmers, E., Hart, D.P. et al., United Kingdom
Haemophilia Centre Doctors. (2013). Diagnosis and treatment of fac-
tor VIII and IX inhibitors in congenital haemophilia (4th edition).
UK Haemophilia Centre Doctors Organization. Br. J. Haematol.
160 (2): 153–170.
Collins, P.W., Palmer, B.P., Chalmers, E.A. et al., United Kingdom
Haemophilia Centre Doctors’ Organization. (2014). Factor VIII
brand and the incidence of factor VIII inhibitors in previously
untreated UK children with severe hemophilia A, 2000-2011. Blood
124 (23): 3389–3397.
Eckhardt, C.L., Loomans, J.I., van Velzen, A.S. et al., INSIGHT Study
Group. (2015). Inhibitor development and mortality in non-severe
hemophilia A. J. Thromb. Haemost. 13 (7): 1217–1225.
Eckhardt, C.L., van Velzen, A.S., Peters, M. et al., INSIGHT Study
Group. (2013). Factor VIII gene (F8) mutation and risk of inhibitor
development in nonsevere hemophilia A. Blood 122 (11): 1954–1962.
Huth-Kühne, A., Baudo, F., Collins, P. et al. (2009). International recom-
mendations on the diagnosis and treatment of patients with acquired
hemophilia A. Haematologica 94: 566–575.
Oldenburg, J. and Pavlova, A. (2006). Genetic risk factors for inhibitors
to factor VIII and IX. Haemophilia 12 (Suppl 6): 15–22.
Peyvandi, F., Mannucci, P.M., Garagiola, I. et al. (2016). A randomized
trial of factor VIII and neutralizing antibodies in hemophilia A. N.
Engl. J. Med. 374 (21): 2054–2064.
Saint Remy, J.M. (2008). How to get rid of inhibitors. Haemophilia
14 (Suppl 3): 33–35.
Saint Remy, J.M., Lacroix-Desmazes, S., and Oldenburg, J. (2004).
Inhibitors in haemophilia: pathophysiology. Haemophilia 10 (Suppl
4): 146–151.
Shepherd, A.J., Skelton, S., Sansom, C.E. et al. (2015). A large-scale com-
putational study of inhibitor risk in non-severe haemophilia A. Br. J.
Haematol. 168 (3): 413–420.
van Velzen, A.S., Eckhardt, C.L., Hart, D.P. et al., INSIGHT study group.
(2015). Inhibitors in nonsevere haemophilia A: outcome and eradi-
cation strategies. Thromb. Haemost. 114 (1): 46–55.
Carrier testing and antenatal diagnosis
Dunn, N.F., Miller, R., Griffioen, A., and Lee, C.A. (2008). Carrier test-
ing in haemophilia A and B: adult carriers’ and their partners’ expe-
riences and their views on the testing of young females. Haemophilia
14: 584–592.
Finning, K.M. and Chitty, L.S. (2008). Non-invasive fetal sex determi-
nation: impact on clinical practice. Semin. Fetal Neonatal. Med. 13:
69–75.
Lavery, S. (2009). Preimplantation genetic diagnosis of haemophilia. Br.
J. Haematol. 144: 303–307.
Ludlam, C.A., Pasi, K.J., Bolton-Maggs, P. et al. (2005). A framework for
genetic service provision for haemophilia and other inherited bleed-
ing disorders. Haemophilia 11: 145–163.
234 Molecular Hematology
Street, A.M., Ljung, R., and Lavery, S. (2008). Management of carriers
and babies with haemophilia. Haemophilia 14 (Suppl 3): 181–187.
Hemophilia therapeutics: current
and future
Oldenburg, J., Mahlangu, J.N., Kim, B. et al. (2017). Emicizumab pro-
phylaxis in hemophilia A with inhibitors. N. Engl. J. Med. 377 (9):
809–818.
Pasi, K.J., Rangarajan, S., Georgiev, P. et al. (2017). Targeting of
antithrombin in hemophilia A or B with RNAi therapy. N. Engl. J.
Med. 377 (9): 819–828.
Peters, R. and Harris, T. (2018). Advances and innovations in
haemophilia treatment. Nat. Rev. Drug Discov. 17 (7): 493–
508.
Chapter 18 The molecular basis of von
Willebrand disease
Luciano Baronciani
Fondazione IRCCS Ca’ Granda Ospedale Maggiore Policlinico, Angelo Bianchi Bonomi Hemophilia and Thrombosis Center, Milan, Italy
Introduction, 235
Function of VWF in primary hemostasis, 235
Function of VWF in blood coagulation, 236
Gene organization, synthesis, and multimeric structure of VWF, 236
von Willebrand disease and its classification, 236
Introduction
von Willebrand disease (VWD) is a common inherited bleed-
ing disorder. Precise data regarding its prevalence are not
available due to the extreme variability in clinical symptoms
of mild VWD. However, population-based studies give an
estimate of clinically significant VWD with a prevalence of
at least 100 per million. VWD is caused by the deficiency
or dysfunction of a multimeric plasma glycoprotein, von
Willebrand factor (VWF). Because of its ability to bind to
a number of ligands, VWF is involved in hemostasis via
a number of mechanisms, but can be considered as hav-
ing two main roles: in primary hemostasis and in intrin-
sic blood coagulation. VWF is directly involved in platelet
binding to the subendothelium and in platelet–platelet inter-
actions, and also acts as the carrier of procoagulant fac-
tor VIII (FVIII). Mutations at the VWF locus can affect
VWF synthesis, its complex biosynthetic assembly, stability
in the circulation, and its binding interactions with specific
ligands.
Function of VWF in primary
hemostasis
At the time of a hemostatic challenge, VWF acts as a bridge
between platelets and the subendothelium of blood vessels,
and is involved in the formation of the platelet plug. The role
Molecular Hematology, 4th Edition. Edited by Drew Provan and John G. Gribben.
© 2020 John Wiley & Sons Ltd. Published 2020 by John Wiley & Sons Ltd.
Genetic defects in VWD, 238
Treatment of VWD, 244
Conclusion, 246
Further reading, 247
of platelets in hemostasis is to become irreversibly attached at
sites of injury. The primary physical factor that affects platelet
binding to the vessel wall is the rate of blood flow in the
vessel, which is faster at the center and slower close to the
wall. These variations in velocity create a shearing effect, or
shear stress, between layers of fluid. Disruption of the vascu-
lar endothelial surface leads to exposure of the subendothe-
lium and results in an alteration of the rate of blood flow
and an increase in shear stress. Plasma VWF binds rapidly
and tightly to subendothelial collagen. VWF does not con-
stitutively bind platelets, but in high blood-flow conditions
is capable of tethering platelets and exposing thrombogenic
surfaces through the interaction of its A1 domain with the
platelet receptor glycoprotein (GP)Ib. However, the VWF–
GPIb interaction does not provide irreversible platelet adhe-
sion because of the fast dissociation rate, and platelets teth-
ered to the vessel wall still move constantly in the direction
of the flow, but at a much slower rate.
A second molecule on the platelet surface is required
to obtain firm platelet adhesion: the integrin GPIIb-IIIa
(αIIbβ3). This molecule is responsible for the platelet–
platelet interaction, which is mediated by VWF and, under
slow-flow conditions, by fibrinogen. αIIbβ3 does not appear
to be involved in the first events of platelet adhesion, proba-
bly because its rate of binding to VWF is too slow to mediate
the initial platelet attachment to the vessel wall under high-
flow conditions. However, when platelets become activated
as a consequence of the VWF–GPIb interaction, αIIbβ3
increases its affinity for its ligand VWF. This event, together
with the slow motion of platelets due to the VWF–GPIb inter-
action, allows αIIbβ3 to bind platelets irreversibly to the ves-
sel wall (Figure 18.1). In all vessels with a high shear rate,
VWF is the primary mediator of platelet binding to the vessel
wall and of platelet aggregation.
235
236 Molecular Hematology

Initial tethering Translocation Firm adhesion

Non-activated Activated
Fluid flow Activation
Torque

GPlbα A1 domain RGDS

Non-activated αllbβ3
Activated αllbβ3 von Willebrand factor

Fig. 18.1 Schematic representation of platelet adhesion to immobilized von Willebrand factor (VWF). At first, platelets are tethered to
the subendothelium through the interaction of their glycoprotein (GP)Ibα with VWF (A1 domain), while the inactivated αIIbβ3 do not bind to the
VWF’s RGDS sequence. Due to the torque imposed by the flowing fluid, platelets begin to roll. New bonds are formed as different regions of the
membrane of the rolling platelets come in contact with the surface, and the translocation continues until the αIIbβ3 becomes activated and binds
firmly to the RGDS of the VWF.
Source: Adapted with permission from Ruggeri Z.M. (1997). von Willebrand Factor. Journal of Clinical Investigation 99: 559–564.

The biosynthesis of VWF is a complex process, involv-

Function of VWF in blood
coagulation
VWF circulates in plasma as a complex associated with FVIII.
The binding of VWF to FVIII is required to stabilize FVIII in
the circulation, preventing cleavage by activated protein C or
factor Xa. This interaction with VWF is crucial in prolong-
ing the half-life of FVIII and concentrating FVIII at the point
of bleeding. When patients with severe VWD (type 3) are
treated with purified FVIII, this is cleared with a half-life of
less than 2 hours, whereas the VWF–FVIII complex infused
in the same patient has a half-life of about 12–14 hours. VWF
binds to FVIII via regions within the first 272 residues of the
mature polypeptide, in the D′ and D3 domains of VWF.
Gene organization, synthesis, and
multimeric structure of VWF
The gene encoding VWF has been mapped in chromosome
12, has a length of approximately 178 kb with 52 exons, and
transcribes a messenger RNA of about 8.2 kb. Analysis of
the VWF gene is complicated by the existence of a partial
unprocessed pseudogene in chromosome 22. The pseudo-
gene extends from exon 23 to exon 34 and shares a high
degree of homology with the gene (97%). VWF is synthe-
sized in megakaryocytes and endothelial cells as a precur-
sor of 2813 amino acids, the pre-pro-VWF. It is composed
of a 22-residue signal peptide, a 741-residue propeptide and a
2050-residue mature subunit. More than 95% of the sequence
accounts for structural domains arranged in the following
order: D1-D2-D′ -D3-A1-A2-A3-D4-C1-C2-C3-C4-C5-C6-
CK (Figure 18.2).
ing post-translational processing of the protein prior to stor-
age or release into the circulation. The initial dimeriza-
tion occurs by disulfide bonding between cysteine residues
in the cystine knot (CK) carboxyl-terminal region of the
monomer (tail-to-tail dimerization). The signal peptide is
cleaved before it enters the Golgi apparatus. The tail-to-tail
glycosylated dimers are then transported to the Golgi appara-
tus, where multimerization and further glycosylation occur.
The propeptide of VWF mediates the assembly of VWF mul-
timers. This process requires the presence of the D1 and
D2 domains (propeptide) and the D′ and D3 domains of
the mature polypeptide. This is followed by cleavage of the
propeptide sequence (Figure 18.3) and secretion of both the
mature VWF and the propeptide into the circulation, or stor-
age within the Weibel–Palade bodies of endothelial cells or
the α-granules of megakaryocytes/platelets. VWF in these
storage organelles presents very large multimers (ultra-large
VWF) that exceed the size of multimers found in plasma.
However, when these molecules are released into the circu-
lation, these forms persist only for a limited time, because
a metalloprotease, ADAMTS-13 (a disintegrin and metal-
loprotease with a thrombospondin type 1 motif, member
13), physiologically cleaves the ultra-large VWF, reducing its
molecular weight.
von Willebrand disease and its
classification
The most common symptoms of mild VWD are mucosal
bleeding (epistaxis, gingival bleeding, menorrhagia) and pro-
longed bleeding after surgical procedures and dental extrac-
tions. Hemarthroses and soft-tissue hematomas are rare, but
 The molecular basis of von Willebrand disease 237

Pseudogene chromosome 22
Gene chromosome 12 23 34
51 introns 178 kb

1 11 28 38 52
5’ 3’

2A* 2A* 2A 2A 2A
(IIC) (IIC) 2N*2N*(IIE) 2B 2M (IIA) 2M (IID)
SP

NH2 D1 D2 D’ D3 A1 A2 A3 D4 C1 C2 C3 C4 C5 C6 CK COOH
ADAMTS
GPlbα 13
multimer collagen GPIIb/IIIa dimer
FVIII heparin 2813 a.a
S-S type I & III (αIIbβ3) S-S
collagen
pro-peptide type IV VWF mature polypeptide
Fig. 18.2 Structure of the von Willebrand factor (VWF) gene, pseudogene, and protein precursor. The schematic structure of the
pre-pro-VWF is reported along with the homologous repeated domain. Locations of intersubunit disulfide bonds involved in dimerization and
multimerization are shown together with the binding sites for several ligands. Arrows point out the position of von Willebrand disease (VWD) type
2 mutations. Recessive VWD type 2 variants are indicated by an *.

mRNA
5’ 3’ 8.5 Kb
Translation
pre-pro-VWF glycosylation
N C 2813 a.a. ( 260 Kda )
sp VW Ag II VWF

C-terminal Rough endoplasmic

SS
dimerization reticulum
520 kDa
CC

N N

Carbohydrate N-terminal
processing multimerization
pro-VWF multimer SS SS

SS SS SS
CC CC

Golgi apparatus

N N N N N N

up to 20000 kDa

Propeptide cleavage
N
VWF multimer N
N N

SS SS SS SS

SS
Golgi storage body
SS SS SS SS
CC CC CC CC or secretion
N N N N N N N N N N

Fig. 18.3 Processing steps of the von Willebrand factor (VWF) multimer synthesis and subcellular localization of the
post-translational modification events. The propeptide is shown in blue, the mature subunit in red. The N- and C- prefixes represent,
respectively, the amino and carboxyl terminal ends of the protein. Monomers are linked together at the C-termini by disulfide bonds to form
dimers (endoplasmic reticulum), which further multimerize in the Golgi apparatus. The cleavage of the propeptide occurs prior to secretion.
238 Molecular Hematology
they occur in severely affected individuals (type 3). The diag-
nosis of VWD is suspected in individuals with these symp-
toms and a family history of bleeding.
Several VWF assays are used in the diagnosis and classi-
fication of VWD. The initial laboratory assessments evalu-
ate plasma levels of FVIII coagulant activity (FVIII:C), VWF
antigen (VWF:Ag), and VWF-GPIb binding activity, mea-
sured usually using the antibiotic ristocetin and formalin
fix or lyophilized platelets (VWF ristocetin cofactor activity,
VWF:RCo). Recently, several new assays, not using platelets
with or without ristocetin, have been developed to measure
the VWF–GPIb binding activity, allowing many laboratories
to evaluate this important VWF function. Additional tests are
necessary for the classification of VWD types, most of which
are not yet standardized and are performed only in special-
ized laboratories.
VWF collagen-binding activity (VWF:CB) measures the
capacity of VWF to bind collagen (typically type 1 or type 3).
A VWF:CB value that is disproportionately low compared
with VWF:Ag reflects the loss of high-molecular-weight
VWF multimers, or can reflect a specific VWF collagen-
binding deficiency, as in the case of mutations in the
A3 domain. VWF multimer analysis evaluates the dis-
tribution of VWF oligomers in the plasma through an
SDS-agarose electrophoresis (western blot technique). Using
a non-reducing low-resolution (1.2% agarose/0.1% sodium
dodecylsulfate) gel, it is possible to appreciate the lack
of high- and intermediate-molecular-weight multimers
or the presence of ultra-large VWF multimers. Using a
non-reducing intermediate-resolution (1.6% agarose/0.1%
sodium dodecylsulfate) gel, individual VWF oligomers from
normal plasma appear as a triplet, consisting of a main
band flanked by two satellite bands, being the products of
proteolysis by ADAMTS-13. Measurement of VWF binding
capacity to FVIII (VWF:FVIIIB) is crucial to diagnose type
2N VWD patients. Ristocetin-induced platelet agglutination
(RIPA) assay evaluates the VWF affinity to platelet GPIb. The
assay is performed using an aggregometer, evaluating patient
platelet-rich plasma agglutination at different concentrations
of ristocetin. This test is important to identify type 2B
variants and in particularly to differentiate type 2B from
type 2A VWD. The evaluation of von Willebrand factor
propeptide (VWFpp) plasma level is useful to establish the
VWF clearance rate by determining the steady-state VWFpp
to VWF:Ag ratio. Evaluation of VWF platelet content, using
VWF:Ag and VWF:RCo, might be useful to discriminate
among different VWD types.
This large number of measurements reflects the fact that
none of them is by itself sensitive and specific enough for a
diagnosis of VWD. VWD is a highly heterogeneous disease in
which there are quantitative and qualitative abnormalities of
VWF, caused so far only by mutations at the VWF locus. The
classification identifies three basic types (Table 18.1). Type 1
Table 18.1 Current classification of von Willebrand disease
Type 1 Partial quantitative deficiency of von Willebrand factor
(VWF)
Type 2 Qualitative deficiency of VWF
2A Decreased VWF-dependent platelet adhesion with lack
of high-molecular-weight VWF multimers
2B Increased affinity for platelet glycoprotein Ib
2M Decreased VWF-dependent platelet adhesion with
normal multimeric structure
2N Markedly decreased VWF binding affinity for factor VIII
Type 3 Complete deficiency of VWF
is characterized by the partial quantitative deficiency of VWF.
Type 2 is characterized by qualitative abnormalities of VWF
and is further categorized into subgroups (2A, 2B, 2M, and
2N) depending on the nature of the qualitative defect (see
later). Type 3 is the most severe form of the disease, character-
ized by the complete absence of VWF in plasma and platelets.
Transmission of type 1 VWD is usually dominant, type 2 is
dominant in the majority of cases, whereas transmission of
type 3 is recessive (Figure 18.4).
Genetic defects in VWD
In the last two decades, many molecular defects of the VWF
gene have been identified in VWD patients, with more than
900 different mutations reported. At first, most of the muta-
tions were identified in patients with the functional variants
(type 2A, 2B, 2M, and 2N), then in the rarer type 3, and
last in the most common type 1 VWD. The unique pheno-
types of the type 2 variants made it possible to restrict the
genetic analysis to the exons encoding for specific structural
domains, such as the A1 domain for the variants 2B and most
2M, A2 for the more common variant 2A, and D′ –D3 for the
variant 2N. On the other hand, characterization of molecular
defects in type 1 and 3 VWD requires extensive screening,
since mutations are not restricted to specific regions and are
usually scattered throughout the VWF gene.
In addition, whereas the identification of a nonsense muta-
tion or a small deletion, more common in type 3 VWD, can
undoubtedly be considered the cause of the disease, the iden-
tification of missense mutations, more common in type 1
VWD, requires supplementary studies in order to confirm
their relationship with the disease, causing further complex-
ity in the molecular characterization of the most common
variant. In most type 2 cases the mutations consist in amino
acid substitutions (missense mutations), although small in-
frame deletions or insertions have been reported. In type 3
VWD the majority of molecular defects identified are respon-
sible for null alleles (nonsense, splice-site mutations, small
 The molecular basis of von Willebrand disease 239

(a)

Ch12 Ch12 Ch12 Ch12

Ch12 Ch12 Ch12 Ch12 Ch12 Ch12 Ch12 Ch12 Ch12 Ch12 Ch12 Ch12

Ch12 Ch12 Ch12 Ch12 Ch12 Ch12 Ch12 Ch12 Ch12 Ch12 Ch12 Ch12

(b)

Ch12 Ch12 Ch12 Ch12

Ch12 Ch12 Ch12 Ch12 Ch12 Ch12 Ch12 Ch12 Ch12 Ch12

Ch12 Ch12 Ch12 Ch12 Ch12 Ch12 Ch12 Ch12 Ch12 Ch12
Fig. 18.4 (a) A family pedigree with an autosomal dominant segregation, as in von Willebrand disease (VWD) type 1, 2A(IIA), 2A(IID), 2A(IIE), 2B,
and 2M. (b) A family pedigree with an autosomal recessive segregation, as in VWD type 2A(IIC), 2N, and 3.
240 Molecular Hematology
deletions, small insertions, and, more rarely, large gene dele-
tions). Nevertheless, numerous missense mutations have also
been identified. In type 1 the opposite seems to be the case,
with a majority of missense mutations and a minority of
splice-site mutations, small deletion and insertion (some-
times in-frame), and more rarely nonsense mutations.
Nowadays, the diffusion of new molecular techniques, like
multiplex ligation-dependent probe amplification (MLPA)
and in particular next-generation sequencing (NGS), along
with the use of online databases and in silico analysis, has
improved patients’ molecular characterization. MLPA anal-
ysis allows the identification of large deletions or duplica-
tions in the whole VWF coding region in a single assess-
ment. NGS enables investigation of the entire VWF coding
region in just few days, making the molecular characteriza-
tion of VWD types 1 and 3 as straightforward as type 2. The
use of online databases reporting previously identified VWD
mutations, along with databases containing single nucleotide
variants and their frequency in different population groups,
helps to confirm, sustain, or exclude the pathogenicity of an
identified variant. In addition, different kinds of software are
now available online to evaluate (in silico analysis) the poten-
tial negative impact of candidate missense, synonymous, or
splicing mutations.
Type 1 VWD
This is the most common form, comprising approximately
70% of all cases of VWD. Patients have mild to moderate
bleeding symptoms, a normal or variably prolonged bleed-
ing time, decreased levels of VWF in plasma, and a normal
multimeric structure, although some cases show a slightly
decreased level of high-molecular-weight multimers. Type 1
VWD may be difficult to diagnose due to natural variations
in circulating VWF levels related to the influence of environ-
mental factors, such as exercise, thyroid hormone, estrogens,
and ABO blood type. VWF levels are lower by as much as
30% in individuals with blood group O compared with indi-
viduals with other blood groups.
In the last decade, a few molecular epidemiological studies
of about 350 index cases with type 1 VWD have significantly
advanced our knowledge, in particular of the genetic basis
of the disorder. Direct sequencing analysis of the whole cod-
ing region and exon/intron boundaries, along with the pro-
moter region of the VWF gene, has allowed the identification
of mutations in about 60–65% of VWD type 1 patients. Most
of these were missense mutations (70%), with the remain-
der being splice-site mutations (12%), small insertions and
deletions (8%; one quarter in-frame), nonsense mutations
(2%), and promoter changes (8%). Only 16% of the muta-
tions cause null alleles. Therefore, haploinsufficiency result-
ing from a heterozygous null allele results in reduced VWF
expression in a small proportion of cases. At variance with
VWD type 2, the missense mutations are found throughout
the entire VWF gene.
Patients with type 1 VWD could be divided roughly into
three groups: (i) patients with a fully penetrant dominant
mutation with low VWF levels; (ii) patients with an incom-
plete penetrant mutation, where not all the family members
with the mutation have low VWF levels; and (iii) patients
with no apparent mutations in the VWF gene. The first
group is characterized by penetrant missense mutations like
p.Cys1130Phe and p.Cys1149Arg. An in vitro study has clar-
ified the molecular mechanism of these dominant negative
mutations found in the D3 domain, showing that intracellu-
lar retention of mutant VWF is the predominant responsi-
ble mechanism. In this group also, accelerated clearance can
cause type 1 VWD, like the case of mutation p.Arg1205His
(Vicenza) being in the D3 domain. This variant has been
found in several populations (all large type 1 VWD stud-
ies) and is always characterized by a low level of VWF:Ag,
normal platelet VWF, and ultra-large plasma VWF multi-
mers. Compared with healthy controls, the half-life of VWF
Vicenza is reduced 4.4-fold after desmopressin, suggesting
that rapid clearance accounts for the moderately severe VWF
deficiency. In addition, rapid clearance decreases the time
during which a large, circulating VWF multimer can be
cleaved by ADAMTS-13, resulting in partial preservation of
ultra-large multimers. Mutation p.Ser2179Phe is also charac-
terized by a short VWF half-life after desmopressin, although
not to the same extent as mutation p.Arg1205His. Also this
mutation has been reported by several population studies on
type 1 VWD patients.
In the second group, incomplete penetrant mutations, like
p.Tyr1584Cys, have been identified. In the heterozygous state
this mutation does not always result in low VWF levels and
bleeding symptoms. Apparently, only the coinheritance of
blood group O, which has been linked with increased VWF
proteolysis, results in type 1 VWD in individuals with muta-
tion p.Tyr1584Cys. Although p.Tyr1584Cys is the mutation
that has been found more often among the VWD type 1
patients investigated, there is still some debate regarding the
pathological role of this variant.
The third group of patients, with no identified mutations,
remains to be clarified. It cannot be excluded that mutations
in genes other than VWF engaged in VWF biosynthesis and
processing could be the cause of the reduced plasma levels of
VWF. However, in this group there is the highest percent-
age of subjects with blood group O, and these are usually
patients with low borderline VWF levels, who should per-
haps not be classified as type 1 VWD patients. Indeed, in
the National Heart, Lung, and Blood Institute, NIH guide-
lines, patients are classified as type 1 VWD only if their
VWF:Ag and VWF:RCo values are lower than 30 IU dl−1 .
Quite recently, three molecular epidemiological studies have
been published, each reporting a cohort of VWD patients
 The molecular basis of von Willebrand disease 241

(type 1, 2, and 3) from a different country (Spain, France, and
Portugal). In all three studies only type 1 VWD patients with
VWF levels <30 IU dL−1 were included. Due to this selec-
tion, the percentage of VWF mutations identified in type 1
patients was higher (89.2%, 96.4%, and 100%, respectively, in
the three populations) than that reported in previous studies
(about 65%). The VWD patients from the French study were
molecularly characterized using Sanger sequence and MLPA
analysis, whereas both Spanish and Portuguese patients were
characterized also using NGS analysis. Among the 161 type
1 patients molecularly characterized in the French study, half
were found to carry null alleles (49% truncating sequence
variations versus 51% of missense mutations), a percentage
much higher than that reported by previous population stud-
ies in type 1 VWD. For the authors, this finding might sug-
gest that the penetrance of some truncating mutations qual-
ified as “recessive” may be variable, like in the case of the
exon 4–5 deletion identified recurrently in patients of British
white origin with either type 3 (in the homozygous or com-
pound heterozygous state) or type 1 (in the heterozygous
state) phenotype. However, further evaluations of these trun-
cating sequence variations (one-fourth of which are poten-
tial splice-site mutations) are needed, along with the investi-
gation of patients’ family members, to better understand the
role of these mutations in type 1 VWD.
The term “severe” VWD has been used occasionally for
VWD type 1 characterized by very low VWF levels. However,
VWD type 1 due to dominant heterozygous mutations, with
few exceptions (e.g. p.Arg1205His), is rarely associated with
VWF levels as low as 10 IU dl−1 . More often these “severe”
VWD type 1 are due to two different mutations (homozygous
or compound heterozygous state), similar to VWD type
3. In these cases the distinction between VWD type 3 (see
later) and type 1 might be difficult, and the evaluation of the
VWFpp level, detectable in type 1 but not in type 3 patients,
could be helpful to establish the correct diagnosis.
Type 2 VWD
Type 2 VWD accounts for approximately 20–25% of all VWD
cases, is caused by a variety of mechanisms, and is associated
with a variety of specific defects, reflecting the multifunc-
tional role of VWF and the complicated post-translational
processing of this moiety.
Type 2A
Type 2A is characterized by decreased VWF-dependent
platelet adhesion due to the reduction or absence of high- and
intermediate-molecular-weight multimers (Figure 18.5a).
The bleeding diathesis is caused by a lack of the most bio-
logically active high-molecular-weight multimers. In the cur-
rent classification of VWD, several different phenotypes, with
these features, are designated VWD type 2A. However, these
distinct phenotypes, described in the previous classification
of VWD as IIA, IIC, IID, and IIE, are due to different molec-
ular pathomechanisms.
The most common form of type 2A (originally designated
VWD type IIA) is due to mutations in the A2 domain.
This variant is characterized by a lack of the VWF high-
and intermediate-molecular-weight multimers, markedly
pronounced satellite bands (Figure 18.5b), and dominant
inheritance. A similar pattern can also be seen in type 2B
VWD patients, and the use of RIPA assay is required to

(a) (b)

HMW
multimer

N 2M 2A 2B N N 2A 2B 2A 2A N
(IIA) (IIC) (IIE)
Fig. 18.5 (a) Comparison, in non-reducing low-resolution gel, of plasma von Willebrand factor (VWF) multimeric structure from von Willebrand
disease (VWD) patients and normal subjects. Lanes 3 (VWD type 2A/IIA) and 4 (VWD type 2B) show the lack of high-molecular-weight (HMW)
multimers. Lane 2 (VWD type 2M) shows a normal multimeric pattern. Lanes 1 and 5 show a normal multimeric structure from healthy controls.
(b) Comparison, in non-reducing intermediate-resolution gel, of plasma VWF multimeric structure from VWD patients and normal subjects. Lanes
1 and 6 show a normal multimeric structure from healthy controls, with the typical triplet structure (brackets); the satellite bands are indicated by
the arrows. Lanes 2 (VWD type 2A/IIA) and 3 (VWD type 2B) show a markedly increased triplet structure, with enhanced satellite bands due to the
increased susceptibility of these variants to ADAMTS-13 proteolysis. Lane 4 (VWD type 2A/IIC) shows the absence of a triplet structure (only the
main band is present), whereas lane 5 (VWD type 2A/IIE) shows markedly reduced satellite bands.
242 Molecular Hematology
discriminate between these two variants. In type 2A patients,
platelet-rich plasma agglutinates at a higher concentration
of ristocetin (>1.2 mg ml−1 ; normal range 0.8–1.2 mg ml−1 ),
whereas in type 2B patients platelet-rich plasma agglutinates
at a lower concentration of ristocetin (<0.7 mg ml−1 ).
Two mechanisms have been report to explain phenotype
2A, subclassified as group 1 and group 2. Group 1 mutations
lead to a VWF subunit that is improperly folded, retained
in the endoplasmic reticulum, and subsequently degraded.
Multimers formed by the interactions between normal and
mutant subunits are retained in the cell. Therefore, the largest
multimers are more efficiently retained, accounting for the
characteristic type 2A multimer pattern. The defect of group
2 does not interfere with VWF secretion, but instead renders
the mutant subunit more susceptible to proteolytic cleavage
(between amino acids Y1605 and M1606 of the A2 domain)
by ADAMTS-13. This enhanced plasma proteolysis results
in the selective loss of the largest VWF multimers typical of
type 2A. However, more recently, in vitro expression stud-
ies have shown that group 1 mutations in the A2 domain
also confer increased susceptibility to ADAMTS-13 cleavage,
as confirmed by the pronounced satellite bands that these
patients shown in their plasma VWF multimers. Over 50 dis-
tinct mutations have been reported in the A2 VWF struc-
tural domain, the most frequent being p.Arg1597Trp and
p.Ile1628Thr.
The second most common type 2A variant (pheno-
type IIE) is caused by mutations in the D3 domain. This
variant is characterized by a decreased level, more rarely
a lack, of high-molecular-weight multimers, reduced or
absent satellite bands (Figure 18.5b), and dominant inher-
itance. A large group of VWD type 2A (IIE) patients from
Germany has been characterized on the basis of their plasma
VWF multimeric structure, and most of the mutations identi-
fied involve a cysteine residue (p.Cys1130Arg, p.Cys1130Trp,
p.Tyr1146Cys, p.Cys1153Tyr, p.Cys1173Phe). The majority
of these mutations correlate with a moderate to severe secre-
tion defect and compromised the VWF multimerization pro-
cess, hence the lack/decreased level of high-molecular-weight
multimers. In addition, due to the position of the muta-
tions in the D3 domain, some of these variants were also
found associated with modestly reduced VWF:FVIIIB. The
important proportion of type 2A (IIE) variants among type
2A patients has been recently underlined also by the inves-
tigation of the French VWD cohort (42 type 2A/IIE cases
out of 122 type 2A patients). The mutation spectrum of
type 2A (IIE) partly overlapped with VWD type 1 (see ear-
lier), which points to the problem of interindividual phe-
notypic differences in spite of the same VWF genotype, or
problems in the performance of non-standardized multimer
analysis.
A rarer type 2A variant (phenotype IIC) is due to muta-
tions in the VWF propeptide (D1 and D2 domains). This
variant is characterized by a lack of high-molecular-weight
multimers, absence of the triplet structure (only the main
band is present; Figure 18.5b), and recessive inheritance.
The VWF propeptide is required for the formation of VWF
multimers and the sorting of VWF to storage granules. The
presence of these mutations in the propeptide compromises
the multimerization process, and hence the lack of high-
molecular-weight multimers. Only a few mutations have
been identified, mainly in the D2 domain (e.g. p.Gly550Arg,
p.Asn528Ser, and p.Cys623Trp). In the case of mutation
p.Asn528Ser, in vitro studies have also shown its role in pre-
venting the intracellular transport of VWF multimer into the
storage organelles.
A rare type 2A variant (phenotype IID) is caused by muta-
tions localized in the CK domain at the carboxy-terminal
of VWF. This variant is characterized by a decreased level
of high-molecular-weight multimers, the presence of extra
intervening bands between individual VWF oligomers, and
dominant inheritance. These mutations compromise the
pro-VWF dimerization process, therefore multimer analy-
sis shows additional bands between the individual VWF
oligomers, representing multimers with an odd number of
monomers. The mutations mainly involve cysteine residue
(p.Cys2771Tyr, p.Cys2771Ser, and p.Cys2773Arg), although
an out-of-frame deletion has also been identified in exon 51.
Type 2B
The VWF 2B variant shows increased affinity for platelet
GPIb, therefore the diagnosis of type 2B VWD patients
requires the use of RIPA assay for these patients’ results to
be enhanced (see earlier). Type 2B VWF is usually associ-
ated with the absence of high-molecular-weight multimers in
circulating plasma but not in platelets (Figure 18.5a). There-
fore, type 2B mutations do not impair the assembly of large
VWF multimers, but after secretion the VWF binds spon-
taneously to platelets, leading to low plasma levels. In addi-
tion, the spontaneous interaction of type 2B VWF with GPIb
accelerates the cleavage of VWF by ADAMTS-13, resulting
in the further depletion of the high-molecular-weight mul-
timers along with the enhancement of satellite bands (Fig-
ure 18.5b). The variable thrombocytopenia which may be
present in these patients is believed to be secondary to the for-
mation of platelet–VWF aggregates. However, in vitro stud-
ies have shown that the presence of type 2B VWF influences
megakaryocyte maturation, causing the formation of giant
platelets, platelets with altered morphology, and reducing the
number of platelet produced, suggesting also a direct role of
type 2B VWF in megakaryocytopoiesis.
Type 2B is inherited as a dominant trait. Mutations respon-
sible for this subtype are mainly in the A1 domain (or at the
D3–A1 junction), which binds to the GPIb receptor. With

a few exceptions, all mutations result in amino acid substi-
tutions, and a few of them (p.Arg1306Trp, p.Arg1308Cys,
p.Val1316Met, and p.Arg1341Gln) account for about 80% of
the reported cases.
Type 2B VWD has sometimes been misdiagnosed and
treated as autoimmune thrombocytopenia. Thrombocytope-
nia and the absence of the larger multimers are not con-
stantly present in type 2B VWD, however. For instance,
mutation p.Pro1266Leu causes an increased GPIb binding
with no apparent loss of large multimers and an absence
of thrombocytopenia before and after desmopressin, lead-
ing to earlier classification among type 1 variants (type 1
New York or type 1 Malmo). Patients carrying this muta-
tion have usually mild bleeding symptoms and in some cases
may have none. Pseudo-VWD, also known as platelet-type
VWD, is a rare platelet disorder due to gain-of-function
mutations in the platelet receptor GPIb (e.g. p.Gly233Val and
p.Met239Val). These mutations cause platelet GPIb to bind
spontaneously to VWF, resulting, as for VWD type 2B, in
enhanced RIPA. Either platelet/plasma mixing studies (per-
forming RIPA assay also using patient plasma with normal
platelets and vice versa) or genetic analysis of VWF and
GPIBA genes are necessary to perform the differential diag-
nosis.
Type 2M
Type 2M VWD refers to qualitative variants with decreased
VWF-dependent platelet adhesion not caused by the absence
of the high-molecular-weight multimers (Figure 18.5a). In
type 2M VWF a functional defect is caused by mutations that
compromise VWF binding to platelets or to subendothelium.
In the first case, type 2M is characterized by a VWF:RCo
that is disproportionately low compared with VWF:Ag,
in spite of the presence of a normal multimer distribution
(hence M for multimer). This variant is inherited as a domi-
nant trait and the mutations are identified in the A1 domain.
These defects, which in most cases are missense mutations
(e.g. p.Arg1315Cys, p.Tyr1321Cys, and p.Gly1324Ser), seem
to downregulate the binding of the A1 domain to the GPIb
platelet receptor. Therefore, in RIPA assay, platelet-rich
plasma of these patients agglutinates at a concentration of
ristocetin >1.2 mg ml−1 (normal range 0.8–1.2 mg ml−1 ),
similar to type 2A patients. Mutations p.Arg1374Cys and
p.Arg1374His have been reported often as type 2M variants,
but their classification as type 2M has been controversial.
In the second case, type 2M VWD is characterized by a dis-
proportionately low VWF:CB compared with VWF:Ag and a
normal multimer distribution. Also this variant is inherited
as a dominant trait and the mutations causing VWF colla-
gen binding defect have been found mainly in the A3 domain
(e.g. p.Ser1731Thr and p.Met1761Lys). Patients with these
mutations have been reported more rarely than patients with
The molecular basis of von Willebrand disease 243
type 2M VWD due to mutations in the A1 domain, and they
seem also to have milder bleeding symptoms.
Type 2N
Type 2N VWD refers to all the qualitative variants character-
ized by decreased affinity for FVIII. The first description of
this type of variant was of a patient from Normandy (hence
the N). This subtype is characterized by low levels of FVIII;
however, VWF levels are usually normal and have an intact
VWF multimeric pattern. It is difficult to distinguish type 2N
from mild hemophilia A. Hemophilia is an X-linked disease,
whereas type 2N VWD is an autosomal recessive disease. A
definite diagnosis can be made by demonstrating the reduced
binding of FVIII to VWF with assays exploring this VWF
property (i.e. VWF:FVIIIB). Since the majority of mutations
are located in the N-terminus of the mature VWF subunit
(D′ –D3), diagnosis can be confirmed by screening for muta-
tions from exons 18–25. Three mutations (p.Thr791Met,
p.Arg816Trp, and p.Arg854Gln) account for 90% of the type
2N mutations.
Type 3 VWD
Type 3 VWD is an autosomal recessive, clinically severe
disorder characterized by the complete or nearly complete
absence of VWF (<5 IU dl−1 ) in both plasma and platelets.
As in type 2N VWD, there is a secondary deficiency of FVIII,
and patients therefore present with a double defect of pri-
mary hemostasis and intrinsic coagulation. Type 3 VWD is
rare and accounts for less than 5% of all VWD cases, with
a prevalence in the general population of 0.5–1 per million.
The parents of type 3 patients are obligatory heterozygotes
and in most cases are asymptomatic.
Because of the large size of the gene, the presence of
the pseudogene, the low prevalence of the disease, and the
absence of a specific location of the mutations, the charac-
terization of the molecular defects in type 3 has been rela-
tively slow. At first, only large gene deletions were identified
using Southern blot analysis. Subsequently, the use of screen-
ing methods such as single-stranded conformational poly-
morphisms, chemical cleavage mismatch analysis, and con-
formational sensitive gel electrophoresis, along with DNA
sequence analysis, allowed the identification of mutations in
most patients.
The majority of the defects identified in type 3 VWD cause
null alleles. Among these, nonsense mutations are the most
common, followed by small deletions, splice-site mutations,
small insertions, and large deletions. Nevertheless, almost
a quarter of the mutations identified in these patients were
missense. Among the nonsense mutations, a few hotspot
mutations at the arginine codons (p.Arg365*, p.Arg1659*,
p.Arg1853*, and p.Arg2535*) have been found repeatedly in
244 Molecular Hematology
different populations. A single cytosine deletion, in a stretch
of 6 cytosines in exon 18, appears to be particularly com-
mon in patients from Sweden and Germany. In the Hungar-
ian population, a large deletion, including exons 1, 2, and
3, was found in about 25% of the unrelated type 3 VWD
patients investigated. Among a group of 42 type 3 patients
from Spain, the mutation p.Gln1311*, derived from a poten-
tial gene conversion, was significantly recurrent (6 patients
out 42). The same mutation was also identified among type 3
patients from Portugal (5 out of 15 families). In both studies
this mutation was identified among Spanish Romani fami-
lies. Patients with type 3 VWD may develop allo-antibodies
to VWF, which render replacement therapy ineffective. This
complication is strongly associated with the presence of
homozygous large gene deletions, although a few cases have
been reported due to nonsense mutations.
Treatment of VWD
The aim of therapy in VWD is to correct both hemosta-
sis defects: the decreased VWF-dependent platelet adhesion
(primary hemostasis) and the reduced FVIII level due to
the VWF deficiency or dysfunction (intrinsic coagulation).
The main treatment options in order to stop spontaneous
bleeding or to prevent bleeding at the time of surgical pro-
cedures are autologous replacement therapy with desmo-
pressin (1-deamino-8-D-arginine vasopressin, DDAVP) and
allogeneic replacement therapy with plasma-derived concen-
trates FVIII/VWF or recombinant VWF. Ancillary forms
of treatment are platelet concentrates, synthetic fibrinoly-
sis inhibitors, and combined estrogen/progestogen prepa-
rations, which in some clinical situations are adjunctive or
sometimes alternative to the main treatments.
Desmopressin
Desmopressin is a synthetic analogue of the antidiuretic
hormone vasopressin that, when administered to healthy
volunteers or patients with mild hemophilia and VWD, tran-
siently increases FVIII and VWF by releasing these moieties
from storage sites into plasma. Endothelial cell Weibel–
Palade bodies appear to be the source of both VWF and
FVIII. However, VWF is constitutively present in Weibel–
Palade bodies (the generation of these organelles is strictly
VWF dependent), whereas FVIII (which is mainly synthe-
sized in the liver) seems to be present only in a select subset of
endothelial cells (lung, heart, intestine, skin, and pulmonary
artery). Desmopressin induces VWF release into plasma by
binding to the vasopressin V2 receptor and thereby activat-
ing cyclic AMP-mediated signaling in vascular endothelial
cells.
The advantage of this compound is that it is relatively
inexpensive and carries no risk of transmitting blood-
borne infectious agents. When infused intravenously over
30 minutes at a dose of 0.2–0.3 μg kg−1 , desmopressin is
expected to increase plasma FVIII and VWF two- to four-
fold above basal levels. In general, high FVIII/VWF con-
centrations last in plasma for at least 8–10 hours and infu-
sions can be repeated every 12–24 hours, depending on the
type and severity of the bleeding episode. At the time of
diagnosis, a DDAVP infusion test should be performed to
evaluate individual response patterns. Patients with baseline
plasma levels of FVIII/VWF measurements in the range of
10–20 IU dl−1 or more are those who are more likely to reach
post-desmopressin levels sufficient to attain hemostasis, tak-
ing into account variables such as the type and severity of the
bleeding episode and the levels of FVIII/VWF that must be
attained and maintained to secure hemostasis. Even though
most patients with mild hemophilia A treated repeatedly with
desmopressin become less responsive to therapy, this prob-
lem is less frequent and prominent in patients with type
1 VWD. The drug is also available in concentrated forms
for subcutaneous and intranasal administration (at doses of
0.3 μg kg−1 and 150–300 μg, respectively), which can be con-
venient for home treatment.
Side effects of desmopressin are usually mild tachycar-
dia, headache, and facial flushing. Hyponatremia and volume
overload due to the antidiuretic effect of desmopressin are
relatively rare if fluid intake is not excessive during treatment.
Even though no thrombotic episodes have been reported in
VWD patients treated with desmopressin, this compound
should be used with caution in elderly patients with cardio-
vascular disease, because a few cases of myocardial infarction
and stroke have occurred in treated patients with hemophilia
and uremia.
Desmopressin is particularly effective in patients with type
1 VWD. A prospective clinical study reported that in 77
patients with type 1, a complete response was observed in
83%, partial response in 13%, and no response in only 4% of
patients. In other VWD types, responsiveness is less certain
(Table 18.2). In fact, in a different prospective trial only 18%
of the type 2 VWD patients were responsive. In this study,
Table 18.2 Indications for desmopressin in different types of von
Willebrand disease
Type Response
1 Usually effective
2A Usually ineffective
2B May be contraindicated
2M Usually ineffective
2N Rarely effective
3 Ineffective
 The molecular basis of von Willebrand disease 245

only one out of 15 type 2A patients was responsive to DDAVP. pasteurization. Another largely diffused product has
A similar result was obtained in type 2M, where DDAVP was the same VWF:RCo/VWF:Ag ratio (0.8–0.9), but is different
effective in only 3 patients out of 21. In type 2N FVIII:C lev- because it also contains similar relative amounts of VWF and
els increase following desmopressin, but released FVIII cir- FVIII (VWF:RCo/FVIII:C ratio 1.2). Two virucidal meth-
culates for a relatively short time in patients’ plasma, because ods, solvent/detergent and heating at high temperatures,
the stabilizing effect of VWF on FVIII is impaired. Therefore, are included in the manufacturing stage of this product.
plasma concentrates containing FVIII and VWF are usually Indeed, most of the plasma-derived products adopted these
preferable. In type 2B patients, desmopressin releases VWF methods with the goal of inactivating both enveloped and
from the endothelial cells. However, there is only a transient non-enveloped viruses.
increased of VWF plasma level, because the ultra-large VWF One unique product is a chromatography-purified VWF
multimers of this gain-of-function variant spontaneously concentrate with a very low FVIII content (VWF:RCo/
agglutinate platelets, worsening or causing patients’ throm- FVIII:C ratio 60). As many as three viral inactivation meth-
bocytopenia. Therefore, it is unclear if the use of desmo- ods are used in the production of this concentrate (sol-
pressin can provide clinical benefit in type 2B patients. How- vent/detergent, heating, and nanofiltration). The recently
ever, mild type 2B patients who present a full set of multi- commercialized product containing human recombinant
mers, such as those carrying mutation p.Pro1266Leu, have VWF is obtained by a genetically engineered cell line,
been treated with desmopressin without any complications. therefore there is virtually no risk of blood-borne virus
Patients with type 3 VWD are unresponsive to desmopressin, transmission. Recombinant VWF, similar to endothelial
because they lack releasable stores of VWF. cell VWF, presents ultra-large multimers, whereas most
plasma-derived products are deficient in high-molecular-
weight multimers. The presence of ultra-large multimers,
Transfusional therapies
along with its high purity, confer a higher specific activity
If desmopressin therapy is not effective or its use is con- (VWF:RCo/VWF:Ag >1) and a stronger hemostatic poten-
traindicated, transfusion therapy with FVIII/VWF concen- tial to this product in comparison to plasma-derived concen-
trates is the treatment of choice to prevent or stop bleed- trates, so it could be used in crucial bleeding situations (e.g.
ing. FVIII and VWF may be infused as fresh frozen plasma VWD associated with gastrointestinal angiodysplasia) where
(FFP), but the large volumes required severely limit its traditional FVIII/VWF products seem to be less effective.
use. Cryoprecipitate contains 5–10 times more FVIII and The dosages recommended for the control or prevention
VWF than FFP (each bag contains approximately 80–100 of bleeding are summarized in Table 18.3. Since in VWD
IU). Early studies indicated that cryoprecipitate adminis- patients FVIII has a longer half-life than in patients with
tered every 12–24 hours normalized plasma FVIII levels hemophilia A (20–24 vs. 12–14 hours), the infusion of one
and stopped or prevented bleeding in VWD. On the basis daily dose is sufficient to reach and maintain adequate plasma
of these observations, cryoprecipitate has been the main-
stay of treatment for many years. However, virucidal meth-
ods cannot be applied to cryoprecipitate produced by blood Table 18.3 Dosages of VWF:RCo/FVIII:C recommended in patients
banks, so that this product carries a small risk of transmitting with von Willebrand disease treated with FVIII/VWF concentrates
blood-borne infectious agents. Therefore, virus-inactivated
FVIII/VWF concentrates, originally developed for the treat- Type of Dose Number of Target plasma
ment of hemophilia A, were perceived as safer and were bleeding (IU kg−1 ) infusions VWF:RCo/FVIII:C level
preferred in the management of VWD patients. Nowadays,
several plasma-derived products are licensed for the treat- Major surgery 40–60 Every 12 hours 50–100 IU dl−1 ,
initially then maintain level until
ment of VWD. Moreover, quite recently, a recombinant VWF
once a day healing is complete
product has been licensed in the USA for the treatment of
(3–10 days)
adult VWD patients. All products licensed for VWD treat- Minor surgery 30–50 Once a day >30 IU dl−1 until
ment have been evaluated for safety and efficacy in prospec- healing is complete
tive clinical trials. However, some of these products have been (may require
more extensively evaluated than others and some present 1–3 days)
unique characteristics. Dental 20–30 Single dose >30 IU dl−1 for
One of the most widely used concentrates in the treatment extractions prior to >12 hours
of VWD contains relatively larger amounts of VWF than procedure
of FVIII (VWF:RCo/FVIII:C ratio 2.4) and has been used Spontaneous 20–60 Once a day > 30 IU dl−1 until
in treatment for the last 25 years. The VWF:RCo/VWF:Ag or traumatic bleeding stops
bleeding
ratio is 0.8 and the virucidal method adopted is
246 Molecular Hematology

levels for the treatment of spontaneous bleeding episodes

Table 18.4 Summary of management of different types and sub-
and to prevent excessive bleeding. The doses of this concen- types of von Willebrand disease (VWD)
trate recommended for their demonstrated efficacy in large,
prospective clinical trials are 40–60 IU kg−1 of VWF:RCo Treatment of Alternative or
(50–75 IU kg−1 in children because of the lower in vivo recov- choice adjunctive therapy
ery), which usually results in VWF:RCo plasma levels of 80–
120 IU dl−1 or higher. Even though the plasma half-life of Type 1 Desmopressin Antifibrinolytic amino
VWF:RCo is much shorter than that of FVIII:C (8–10 vs. (factor VIII/VWF acids
concentrates)
20–24 hours), usually these doses do not need to be repeated
Type 2A Factor VIII/VWF As above
more often than every 24 hours, although sometimes treat-
concentrates
ment intervals must be tailored to the clinical situation. (Desmopressina )
It is usually not necessary to carry out laboratory tests to Type 2B Factor VIII/VWF As above
monitor replacement therapy in patients with spontaneous concentrates
bleeding episodes. For surgical procedures we recommend Type 2M Factor VIII/VWF As above
measuring FVIII:C and VWF:RCo every 12 hours on the concentrates
operation day and then every 24 hours. Major surgical pro- (Desmopressina )
cedures are successfully carried out and spontaneous bleed- Type 2N Factor VIII/VWF As above
ing episodes controlled following the infusion of FVIII/VWF concentrates
concentrates. In the relatively rare instances when bleeding is (Desmopressina )
Type 3 Factor VIII/VWF Antifibrinolytic amino
not controlled, platelet concentrates (given immediately after
concentrates acids, platelet
FVIII/VWF-containing preparations, at doses of 4–5 × 1011
concentrates
platelets) are effective, particularly in patients with type 3 Type 3 Recombinant Recombinant activated
VWD. Platelets from type 3 VWD patients completely lack complicated by factor VIII factor VII
VWF and there is no uptake of the protein from plasma allo-antibodies
after the infusion of concentrates. The hemostatic effective-
a 1-deamino-8-D-arginine vasopressin (DDAVP) should also be trialed
ness of the transfusion of normal platelets is likely to be due
to the fact that these cells transport and localize ultra-large for types 2A, 2M, and 2N VWD patients, even though DDAVP will
likely be insufficiently effective for most of these cases and therefore
VWF at sites of vascular injury. From a practical standpoint,
FVIII/VWF concentrates will be required.
it must be emphasized that in one of the largest prospec-
tive study carried out in VWD patients, platelet concen-
trates became necessary to prevent or stop bleeding in one
case only. phenotypically, because the choice of treatment must be
To avoid the risk of a life-threatening anaphylactic reac- tailored to the different types of the disease. Such character-
tion, type 3 VWD patients who develop allo-antibodies ization is not simple, so in the majority of clinical centers it
against VWF must not be treated with products contain- is probably not worth setting up relatively complicated tests
ing VWF. Instead, treatment can be done using continuous such as multimer analysis assay, when samples can be sent
intravenous infusion of recombinant FVIII or using recom- for analysis to more expert laboratories that have become
binant activated FVII. Venous thromboembolic episodes sel- proficient during the study of large series of patients.
dom occur during repeated FVIII/VWF concentrate infu-
sions. As mentioned earlier, therapy should be monitored
von Willebrand disease resources on the
daily, also to avoid reaching very high FVIII plasma levels,
internet
known to be a risk factor for venous thromboembolism.
International Society on Thrombosis and Haemostasis-
Scientific and Standardisation Committee of von Willebrand
Conclusion Factor (ISTH-SSC). von Willebrand Factor Online Database
(VWFdb). Available at: http://www.vwf.group.shef.ac.uk/
The different options currently available for the management vwd.html
of VWD are summarized in Table 18.4. Treatment of spon- European Association for Haemophilia and Allied Dis-
taneous bleeding episodes and their prevention at the time orders (EAHAD) Coagulation Factor Variant Databases.
of invasive procedures are relatively simple and can certainly Available at: https://grenada.lumc.nl/LOVD2/VWF/home.
be tackled by the average clinical hematologist with access php?select_db=VWF
to a minimum of laboratory testing (FVIII:C and VWF:RCo OMIMTM Online Mendelian Inheritance in Man. Avail-
assays). However, patients need to be well characterized able at: http://omim.org/entry/193400

Further reading
General
Nichols, W.L., Hultin, M.B., James, A.H. et al. (2008). von Willebrand
disease (VWD): evidence-based diagnosis and management guide-
lines, the National, Heart, Lung and Blood Institute (NHLBI) Expert
Panel report (USA). Haemophilia 14: 171–232.
Sadler, J.E., Mannucci, P.M., Berntorp, E. et al. (2000). Impact, diagnosis
and treatment of von Willebrand disease. Thromb. Haemost. 84: 160–
174.
von Willebrand factor
Dong, Z., Thoma, R.S., Crimmins, D.L. et al. (1994). Disulfide bonds
required to assemble functional von Willebrand factor multimers. J.
Biol. Chem. 269: 6753–6758.
Fujikawa, K., Suzuki, H., McMullen, B., and Chung, D. (2001). Purifica-
tion of human von Willebrand factor cleaving protease and its iden-
tification as a new member of the metalloproteinase family. Blood 98:
1662–1666.
Ginsburg, D., Handin, R.I., Bonthron, D.T. et al. (1985). Human von
Willebrand factor (vWF): isolation of complementary DNA (cDNA)
clones and chromosomal localization. Science 228: 1401–1406.
Kroll, M.H., Hellums, J.D., McIntire, L.V. et al. (1996). Platelets and
shear stress. Blood 88: 1525–1541.
Lenting, P.J., Christophe, O.D., and Denis, C.V. (2015). von Willebrand
factor biosynthesis, secretion, and clearance: connecting the far ends.
Blood 125: 2019–2028.
Mancuso, D.J., Tuley, E.A., Westfield, L.A. et al. (1989). Structure of the
gene for human von Willebrand factor. J. Biol. Chem. 264: 19514–
19527.
Mancuso, D.J., Tuley, E.A., Westfield, L.A. et al. (1991). Human von
Willebrand factor gene and pseudogene: structural analysis and dif-
ferentiation by polymerase chain reaction. Biochemistry 30: 253–269.
Shelton-Inloes, B.B., Titani, K., and Sadler, J.E. (1986). cDNA sequences
for human von Willebrand factor reveal five types of repeated
domains and five possible protein sequence polymorphisms. Bio-
chemistry 25: 3164–3171.
Wagner, D.D. and Marder, V.J. (1984). Biosynthesis of von Willebrand
protein by human endothelial cells: processing steps and their intra-
cellular localization. J. Cell Biol. 99: 2123–2130.
Weiss, H.J., Sussman, I.I., and Hoyer, L.W. (1977). Stabilization of factor
VIII in plasma by the von Willebrand factor. Studies on posttransfu-
sion and dissociated factor VIII and in patients with von Willebrand’s
disease. J. Clin. Invest. 60: 390–404.
Zhou, Y.F., Eng, E.T., Zhu, J. et al. (2012). Sequence and structure rela-
tionships within von Willebrand factor. Blood 120: 449–458.
von Willebrand disease and its
classification
De Jong, A. and Eikenboom, J. (2016). Developments in the diagnostic
procedures for von Willebrand disease. J. Thromb. Haemost. 14: 449–
460.
Federici, A.B. (1998). Diagnosis of von Willebrand disease. Haemophilia
4: 654–660.
The molecular basis of von Willebrand disease 247
Sadler, J.E. (1994). A revised classification of von Willebrand disease.
Thromb. Haemost. 71: 520–525.
Sadler, J.E., Budde, U., Eikenboom, J.C.J. et al. (2006). Update on
the pathophysiology and classification of von Willebrand disease: a
report of the subcommittee on von Willebrand factor. J. Thromb.
Haemost. 4: 1–12.
Genetic defects in VWD
Acquila, M., Bottini, F., DI Duca, M. et al. (2009). Multiplex ligation-
dependent probe amplification to detect a large deletion within the
von Willebrand gene. Haemophilia 15: 1346–1348.
Baronciani, L., Federici, A.B., Castaman, G. et al. (2008). Prevalence
of type 2B “Malmo/New York” von Willebrand disease in Italy: the
role of von Willebrand factor gene conversion. J. Thromb. Haemost.
6: 887–890.
Baronciani, L., Federici, A.B., and Eikenboom, J.C.J. (2011). Clinical,
laboratory, and molecular markers of type 3 von Willebrand disease.
In: Von Willebrand disease: basic and clinical aspects (eds. A.B. Fed-
erici, C.A. Lee, E. Berntop, et al.), 148–165. Wiley-Blackwell.
Batlle, J., Pérez-Rodrı́guez, A., Corrales, I. et al. (2015). Molecular and
clinical profile of von Willebrand disease in Spain (PCM–EVW–ES):
proposal for a new diagnostic paradigm. Thromb. Haemost. 2: 1–11.
Bodo, I., Katsumi, A., Tuley, E.A. et al. (2001). Type 1 von Wille-
brand disease mutation Cys 1149Arg causes intracellular retention
and degradation of heterodimers: a possible general mechanism for
dominant mutations of oligomeric proteins. Blood 98: 2973–2979.
Bowen, D.J. and Collins, P.W. (2006). Insights into von Willebrand factor
proteolysis: clinical implications. Br. J. Haematol. 133: 457–467.
Corrales, I., Catarino, S., Ayats, J. et al. (2012). High-throughput molec-
ular diagnosis of von Willebrand disease by next generation sequenc-
ing methods. Haematologica 97: 1003–1007.
Cumming, A., Grundy, P., Keeney, S. et al. (2006). An investigation of
the von Willebrand factor genotype in UK patients diagnosed to have
type 1 von Willebrand disease. Thromb. Haemost. 96: 630–641.
Eikenboom, J., Hilbert, L., Ribba, A.S. et al. (2009). Expression of 14
von Willebrand factor mutations identified in patients with type 1
von Willebrand disease from the MCMDM-1VWD study. J. Thromb.
Haemost. 7: 1304–1312.
Eikenboom, J.C.J., Matsushita, T., Reitsma, P.H. et al. (1996). Dominant
type I von Willebrand disease caused by mutated cysteine residues in
the D3 domain of von Willebrand factor. Blood 88: 2433–2441.
Enayat, M.S., Guilliatt, A.M., Surdhar, G.K. et al. (2001). Aberrant
dimerization of von Willebrand factor as the result of mutations in
the carboxy-terminal region: identification of 3 mutations in mem-
bers of 3 different families with type 2A (phenotype IID) von Wille-
brand disease. Blood 98: 674–680.
Federici, A.B., Mannucci, P.M., Castaman, G. et al. (2009). Clinical and
molecular predictors of thrombocytopenia and risk of bleeding in
patients with von Willebrand disease type 2B: a cohort study of 67
patients. Blood 113: 526–534.
Fidalgo, T., Salvado, R., Corrales, I. et al. (2016). Genotype–phenotype
correlation in a cohort of Portuguese patients comprising the entire
spectrum of VWD types: impact of NGS. Thromb. Haemost. 116: 17–
31.
Gadisseur, A., van der Planken, M., Schroyens, W. et al. (2009). Domi-
nant von Willebrand disease type 2M and 2U are variable expressions
248 Molecular Hematology
of one distinct disease entity caused by loss-of-function mutations in
the A1 domain of the von Willebrand factor gene. Acta Haematol.
121: 145–153.
Gezsi, A., Budde, U., Deak, I. et al. (2010). Accelerated clearance alone
explains ultra-large multimers in von Willebrand disease Vicenza. J.
Thromb. Haemost. 8: 1273–1280.
Goodeve, A., Eikenboom, J.C.J., Castaman, G. et al. (2007). Phenotype
and genotype of a cohort of families historically diagnosed with type
1 von Willebrand disease in the European study, Molecular and Clin-
ical Markers for the Diagnosis and Management of Type 1 von Wille-
brand disease (MCMDM-1VWD). Blood 109: 112–121.
Haberichter, S.L., Balistreri, M., Christopherson, P. et al. (2006). Assay
of the von Willebrand factor (VWF) propeptide to identify patients
with type 1 von Willebrand disease with decreased VWF survival.
Blood 108: 3344–3351.
Haberichter, S.L., Budde, U., Obser, T. et al. (2010). The mutation N528S
in the von Willebrand factor (VWF) propeptide causes defective mul-
timerization and storage of VWF. Blood 115: 4580–4587.
Hassenpflug, W.A., Budde, U., Obser, T. et al. (2006). Impact of muta-
tions in the von Willebrand factor A2 domain on ADAMTS13-
dependent proteolysis. Blood 107: 2339–2345.
James, P.D., Notley, C., Hegadom, C. et al. (2007). The mutational spec-
trum of type 1 von Willebrand disease: results from a Canadian
cohort study. Blood 109: 145–154.
Keeling, D., Beavis, J., Marr, R. et al. (2012). A family with type 2M VWD
with normal VWF:RCo but reduced VWF:CB and a M1761K muta-
tion in the A3 domain. Haemophilia 18: e33–e2516.
Lyons, S.E., Bruck, M.E., Bowie, E.J.W., and Ginsburg, D. (1992).
Impaired intracellular transport produced by a subset of type IIA von
Willebrand disease mutations. J. Biolumin. Chemilumin. 267: 4424–
4430.
Mannucci, P.M., Bloom, A.L., Larrieu, M.J. et al. (1984). Atherosclerosis
and von Willebrand factor. I. Prevalence of severe von Willebrand’s
disease in western Europe and Israel. Br. J. Haematol. 57: 163–169.
Mannucci, P.M., Lombardi, R., Castaman, G. et al. (1988). von Wille-
brand disease “Vicenza” with larger-than-normal (supranormal) von
Willebrand factor multimers. Blood 71: 65–70.
Mazurier, C. and Meyer, D. (1996). Factor VIII binding assay of von
Willebrand factor and the diagnosis of type 2N von Willebrand dis-
ease – results of an international survey. On behalf of the subcom-
mittee on von Willebrand Factor of the Scientific and Standardization
Committee of the ISTH. Thromb. Haemost. 76: 270–274.
Meyer, D., Fressinaud, E., Gaucher, C. et al. (1997). Gene defects in
150 unrelated French cases with Type 2 von Willebrand disease: from
patient to the gene. Thromb. Haemost. 78: 451–456.
Nurden, P., Gobbi, G., Nurden, A. et al. (2010). Abnormal VWF mod-
ifies megakaryocytopoiesis: studies of platelets and megakaryocyte
cultures from patients with von Willebrand disease type 2B. Blood
115: 2649–2656.
O’Brien, L.A., James, P.D., Othman, M. et al. (2003). Founder von Wille-
brand factor haplotype associated with type 1 von Willebrand disease.
Blood 102: 549–557.
Othman, M. and Emsley, J. (2014). Platelet-type von Willebrand disease:
toward an improved understanding of the “sticky situation.”. Semin.
Thromb. Hemost. 40: 146–150.
Penas, N., Perez-Rodriguez, A., Torea, J.H. et al. (2005). Von Wille-
brand disease R1374C: type 2A or 2M? A challenge to the revised
classification. High frequency in the northwest of Spain (Galicia).
Am. J. Hematol. 80: 188–196.
Richards, S., Aziz, N., Bale, S. et al. (2015). ACMG Standards and Guide-
lines. Standards and guidelines for the interpretation of sequence
variants: a joint consensus recommendation of the American College
of Medical Genetics and Genomics and the Association for Molecular
Pathology. Genet. Med. 17: 405–424.
Ruggeri, Z.M., Pareti, F.I., Mannucci, P.M. et al. (1980). Heightened
interaction between platelets and factor VIII/von Willebrand factor
in a new subtype of von Willebrand disease. N. Engl. J. Med. 302 (19):
1047–1051.
Schneppenheim, R., Michiels, J.J., Obser, T. et al. (2010). A cluster of
mutations in the D3 domain of von Willebrand factor correlates with
a distinct subgroup of von Willebrand disease: type 2A/IIE. Blood
115: 4894–4901.
Sutherland, M.S., Cumming, A.M., Bowman, M. et al. (2009). A novel
deletion mutation is recurrent in von Willebrand disease types 1 and
3. Blood 114: 1091–1098.
Veyradier, A., Boisseau, P., Fressinaud, E. et al. (2016). A laboratory
phenotype/genotype correlation of 1167 French patients from 670
families with von Willebrand disease: a new epidemiologic picture.
Medicine (Baltimore) 95: e3038.
Yadegari, H., Driesen, J., Pavlova, A. et al. (2012). Mutation distribution
in the von Willebrand factor gene related to the different von Wille-
brand disease (VWD) types in a cohort of VWD patients. Thromb.
Haemost. 108: 662–671.
Zimmerman, T.S., Dent, J.A., Ruggeri, Z.M., and Nannini, L.H. (1986).
Subunit composition of plasma von Willebrand factor. Cleavage is
present in normal individuals, increased in IIA and IIB von Wille-
brand disease, but minimal in variants with aberrant structure of
individual oligomers (types IIC, IID, and IIE). J. Clin. Invest. 77 (3):
947–951.
Treatment of VWD
Berntorp, E., Archey, W., Auerswald, G. et al. (2008). A systematic
overview of the first pasteurised VWF/FVIII medicinal product, Hae-
mate P/Humate-P: history and clinical performance. Eur. J. Haema-
tol. Suppl. 80 (Suppl.70): 3–35.
Berntorp, E., Windyga, J., and European Wilate Study Group (2009).
Treatment and prevention of acute bleedings in von Willebrand dis-
ease – efficacy and safety of Wilate, a new generation von Willebrand
factor/factor VIII concentrate. Haemophilia 15: 122–130.
Bond, L. and Bevan, D. (1988). Myocardial infarction in a patient with
hemophilia treated with DDAVP (letter). N. Engl. J. Med. 318: 121.
Borel-Derlon, A., Federici, A.B., Roussel-Robert, V. et al. (2007). Treat-
ment of severe von Willebrand disease with a high-purity von
Willebrand factor concentrate (Wilfactin): a prospective study of 50
patients. J. Thromb. Haemost. 5: 1115–1124.
Brooker, M. (2012). Registry of clotting factor concentrates, 9e. Montréal:
World Federation of Hemophilia available from: www.wfh.org.
Castaman, G., Coppola, A., Zanon, E. et al. (2013). Efficacy and safety
during formulation switch of a pasteurized VWF/FVIII concentrate:
results from an Italian prospective observational study in patients
with von Willebrand disease. Haemophilia 19: 82–88.
Castaman, G., Lethagen, S., Federici, A.B. et al. (2008). Response to
desmopressin is influenced by the genotype and phenotype in type

1 von Willebrand disease (VWD): results from the European Study
MCMDM-1VWD. Blood 111: 3531–3539.
Castillo, R., Monteagudo, J., Escolar, G. et al. (1991). Hemostatic effect of
normal platelet transfusion in severe von Willebrand disease patients.
Blood 77: 1901–1905.
Chang, A.C., Rick, M.E., Ross, P.L., and Weinstein, M.J. (1998). Sum-
mary of a workshop on potency and dosage of von Willebrand factor
concentrates. Haemophilia 4 (Suppl 3): 1–6.
Coppola, A., Franchini, M., Makris, M. et al. (2012). Thrombotic
adverse events to coagulation factor concentrates for treatment of
patients with haemophilia and von Willebrand disease: a systematic
review of prospective studies. Haemophilia 18: e173–e187.
Favaloro, E.J., Franchini, M., and Lippi, G. (2012). Biological therapies
for von Willebrand disease. Expert Opin. Biol. Ther. 12 (5): 551–564.
Federici, A.B. (2008). The use of desmopressin in von Willebrand dis-
ease: the experience of the first 30 years (1977–2007). Haemophilia 14
(Suppl 1): 5–14.
Federici, A.B., Baudo, F., Caracciolo, C. et al. (2002). Clinical efficacy of
highly purified, doubly virus-inactivated factor VIII/von Willebrand
factor concentrate (Fanhdi) in the treatment of von Willebrand dis-
ease: a retrospective clinical study. Haemophilia 8: 761–767.
Federici, A.B., Mazurier, C., Berntorp, E. et al. (2004). Biologic response
to desmopressin in patients with severe type 1 and type 2 von Wille-
brand disease: results of a multicenter European study. Blood 103:
2032–2038.
Franchini, M. and Mannucci, P.M. (2014). Gastrointestinal angiodyspla-
sia and bleeding in von Willebrand disease. Thromb. Haemost. 112:
427–431.
Franchini, M. and Mannucci, P.M. (2016). Von Willebrand factor (Von-
vendi®): the first recombinant product licensed for the treatment of
von Willebrand disease. Expert Rev. Hematol. 9: 825–830.
Goudemand, J., Scharrer, I., Berntorp, E. et al. (2005). Pharmacoki-
netic studies on Wilfactin, a von Willebrand factor concentrate with a
low factor VIII content treated with three virus-inactivation/removal
methods. J. Thromb. Haemost. 3: 2219–2227.
Holmberg, L., Nilsson, I.M., Borge, L. et al. (1983). Platelet aggregation
induced by 1-desamino-8-D-arginine vasopressin (DDAVP) in type
IIB von Willebrand’s disease. N. Engl. J. Med. 309: 816–821.
James, P.D., Lillicrap, D., and Mannucci, P.M. (2013). Alloantibodies in
von Willebrand disease. Blood 122: 636–640.
Kaufmann, J.E., Oksche, A., Wollheim, C.B. et al. (2000). Vasopressin-
induced von Willebrand factor secretion from endothelial cells
involves V2 receptors and cAMP. J. Clin. Invest. 106: 107–116.
Kohler, M., Hellstern, P., and Wenzel, E. (1985). The use of heat-treated
Factor VIII-concentrates in von Willebrand’s disease. Blut 50: 25–27.
Mannucci, P.M. (2002). Venous thromboembolism in von Willebrand
disease. Thromb. Haemost. 88: 378–379.
The molecular basis of von Willebrand disease 249
Mannucci, P.M. (2004). Treatment of von Willebrand disease. N. Engl.
J. Med. 351: 683–694.
Mannucci, P.M., Bettega, D., and Cattaneo, M. (1992). Patterns of devel-
opment of tachyphylaxis in patients with haemophilia and von Wille-
brand disease after repeated doses of desmopressin (DDAVP). Br. J.
Haematol. 82: 87–93.
Mannucci, P.M., Chediak, J., Hanna, W. et al. (2002). Treatment of von
Willebrand’s disease with a high-purity factor VIII/von Willebrand
factor concentrate: a prospective, multicenter study. Blood 99: 450–
456.
Mannucci, P.M., Franchini, M., and Castaman, G., Federici AB; Italian
Association of Hemophilia Centers (2009). Evidence-based recom-
mendations on the treatment of von Willebrand disease in Italy. Blood
Transfus. 7 (2): 117–126.
Mannucci, P.M., Ruggeri, Z.M., Pareti, F.I., and Capitanio, A. (1977).
1-deamino-8-D-arginine vasopressin: a new pharmacological
approach to the management of haemophilia and von Willebrand
disease. Lancet 1: 869–872.
Mannucci, P.M., Tenconi, P.M., Castaman, G., and Rodeghiero, F.
(1992). Comparison of four virus-inactivated plasma concentrates for
treatment of severe von Willebrand disease: a cross-over randomized
trial. Blood 79: 3130–3137.
Mazurier, C., De Romeuf, C., Parquet-Gernez, A., and Goudemand,
M. (1989). In vitro and in vivo characterization of a high-purity,
solvent/detergent-treated factor VIII concentrate: evidence for its
therapeutic efficacy in von Willebrand’s disease. Eur. J. Haematol. 43:
7–14.
Mazurier, C., Gaucher, C., Jorieux, S., and Goudeman, M. (1994).
Biological effect of demopressin in eight patients with type 2N
(“Normandy”) von Willebrand disease. Br. J. Haematol. 88: 849–
854.
Pasi, K.J., Williams, M.D., Enayat, M.S., and Hill, F.G.H. (1990). Clin-
ical and laboratory evaluation of the treatment of von Willebrand’s
disease patients with heat-treated factor VIII concentrate (BPL8Y).
Br. J. Haematol. 75: 228–233.
Perkins, H.A. (1967). Correction of the hemostatic defects in von Wille-
brand disease. Blood 30: 375–380.
Rivard, G.E., Aledort, L., and Alphanate Surgical Investigators (2008).
Efficacy of factor VIII/von Willebrand factor concentrate Alphanate
in preventing excessive bleeding during surgery in subjects with von
Willebrand disease. Haemophilia 14: 271–275.
Shahani, T., Lavend’homme, R., Luttun, A. et al. (2010). Activation
of human endothelial cells from specific vascular beds induces the
release of a FVIII storage pool. Blood 115 (23): 4902–4909.
Smith, T.J., Gill, J.C., Ambroso, D.R., and Hathaway, W.E. (1989).
Hyponatremia and seizures in young children given DDAVP. Am. J.
Hematol. 31: 199–202.
Chapter 19 Platelet disorders
Kenneth J. Clemetson
Department of Haematology, Inselspital, University of Berne, Berne, Switzerland
Introduction, 251
Normal platelet function, 251
Diagnosis of platelet defects in bleeding disorders, 252
Platelet adhesion disorders, 253
Platelet aggregation defects: Glanzmann thrombasthenia, 256
Agonist receptor defects, 257
Defects in intracellular signaling pathways, 257
Introduction
Over the past few decades a large number of disorders
caused by genetic mutations in platelet components have
been diagnosed at a molecular level. These include defects in
platelet receptors, signaling pathways, and cytoskeletal pro-
teins. Generally, these were originally detected in bleeding
disorders, varying from severe through moderate to mild.
Platelets have a major role in hemostasis, particularly pri-
mary hemostasis, where platelets adhere to sites of vascular
damage, become activated, and bind further platelets by
aggregation to form a thrombus. The platelet plug formed
is then stabilized by secondary hemostasis, which involves
coagulation factors that assemble on the surface of the acti-
vated platelets, becoming activated in turn and leading to for-
mation of thrombin and cleavage of fibrinogen to form fibrin.
Many components have active roles in the overall process,
from first platelet contact to stable platelet plug. Most large
hospitals see over 100 patients annually, each with mild to
moderate bleeding disorders that are not caused by coagula-
tion factor problems and can be ascribed to uncharacterized
congenital platelet defects. While major defects in known
receptors may lead to severe bleeding and are relatively easy
to diagnose, others are often very difficult to diagnose at a
molecular level, and each represents a challenging research
project; most remain effectively undiagnosed. However, new
methods are beginning to become available which could
make such diagnoses more straightforward. This chapter
deals with genetic defects in platelet components that cause
Molecular Hematology, 4th Edition. Edited by Drew Provan and John G. Gribben.
© 2020 John Wiley & Sons Ltd. Published 2020 by John Wiley & Sons Ltd.
Platelet secretion defects (storage pool disease), 258
Procoagulant regulation defects, 259
Giant platelet syndromes and cytoskeletal defects, 260
Transcription factor defects, 260
Platelet protein polymorphisms and tendency to thrombosis, 260
Further reading, 261
human disease. Although these are mostly bleeding disor-
ders, genetic differences leading to an enhanced tendency to
thrombosis are also increasingly suspected.
Normal platelet function
Following damage to the endothelium that protects the vas-
cular system, which may simply be caused by the loss of “old”
endothelial cells due to the high shear stress in the arterial
system, platelets adhere to the subendothelium via glyco-
protein (GP)Ib on the platelet and von Willebrand factor
(VWF) attached to extracellular matrix proteins, particularly
collagen (Figure 19.1). This interaction alone is sufficient to
start platelet activation, but following initial platelet adhe-
sion other receptors such as the collagen receptor GPVI are
recruited. GPVI comes into contact with collagen and signals
via its Fcγ subunit to enhance platelet activation and recruit
integrins, which increase the binding of adherent platelets
and provide attachment for further aggregating platelets.
Activation of αIIbβ3 allows platelet–platelet aggregation via
ligands such as fibrinogen. Activated platelets release adeno-
sine diphosphate (ADP), adenosine triphosphate (ATP),
serotonin, and Ca2+ from dense granules and a wide range of
proteins from α-granules that contribute to platelet activa-
tion and aggregation, including fibrinogen, high-multimer
VWF, and thrombospondin, as well as growth factors and
chemokines that affect endothelial cells and leukocytes
involved in wound repair. The major collagen adhesive
receptor α2β1 is also activated, increasing platelet adhesion
to the exposed collagen of the subendothelium. Both these
integrins use disulfide bond-reshuffling mechanisms via
251
252 Molecular Hematology

Tethering Adhesion Activation Aggregation

= VWF = GPIb-V-IX = Activated α2β1

= Collagen = GPVI = Activated αIIbβ3
= Fibrinogen
= αIIbβ3
Fig. 19.1 Platelet thrombus formation. Following vessel wall injury, the matrix of the subendothelium is exposed. Under high-shear conditions,
platelets adhere and become tethered via GPIb–V–IX on platelets binding to von Willebrand factor (VWF) in the matrix. This interaction starts
platelet activation, but also brings platelets into contact with the subendothelium, allowing other receptors, such as GPVI, to interact with collagen
and enhance activation. Under static or low-shear conditions, the GPVI–collagen interaction is mainly responsible for the initial activation. Platelet
activation brings additional receptor–ligand interactions into play, such as α2β1–collagen, since this receptor does not bind in the resting state.
Platelet activation also leads to shape change, release of granule contents, activation of other receptors, such as αIIbβ3, and exposure of
negatively charged phospholipids, critical for procoagulant activity. Binding of VWF and fibrinogen to GPIb–V–IX and αIIbβ3 on adjacent platelets
leads to platelet aggregation and the formation of a thrombus.

thiol isomerase to change disulfide bond patterns in the β
subunits to control the resting or active state.
Activated platelets also show surface changes, with
exposure of negatively charged phospholipids that act,
together with surface receptors such as GPIb, to assemble
and activate coagulation factors, leading to conversion of
prothrombin to thrombin. The major effect of thrombin is
to convert fibrinogen to fibrin, stabilizing and solidifying the
thrombus. However, thrombin also feeds back to platelets
via receptors including GPIb and the seven-transmembrane
PAR1 and PAR4, activating them further. When platelets
are strongly activated by a combination of agonists such as
thrombin/collagen, that part of the activated platelet popu-
lation with strong Ca2+ fluxes shows enhanced procoagulant
properties though ballooning via water intake to present an
increased surface and by covalent binding of coagulation
factors. When platelets are treated with Ca2+ -ionophores,
all platelets fall in this hyperactive category. It is not yet
clear why with thrombin/collagen activation only part of the
platelet population is hyperactivated.
Diagnosis of platelet defects in
bleeding disorders
Most patients with clinically relevant platelet defects seek
medical assistance because of bleeding problems, ranging
from easy or excessive bruising, epistaxis, gingival bleed-
ing, and menorrhagia to excessive bleeding following tooth
extraction or other surgery. Such patients are normally
screened for coagulation defects; if none is detected, exami-
nation for platelet problems is the next logical step. Normally,
a complete blood count is done to exclude thrombocytopenia
and a peripheral blood smear is examined to check platelet
morphology.
A number of whole-blood techniques are now avail-
able to check platelet function, including PFA-100 and
Impact technologies. These should indicate the presence of a
platelet defect and its likely origin. Platelet aggregation tests
with a battery of agonists, including ADP, collagen, throm-
bin receptor activation peptide (TRAP), arachidonic acid,
epinephrine, and ristocetin, should also help to narrow down
which responses are defective. Flow cytometry or Western
blotting with specific monoclonal antibodies may be used
to check for the absence of receptors or to determine the
amounts present. Flow cytometry may also be used to exam-
ine the expression of P-selectin, the activation of αIIbβ3, or
the exposure of negatively charged phospholipids in response
to the classic agonists. All these techniques can provide valu-
able information to lead to a molecular diagnosis.
If a defect can be ascribed to a given platelet molecule,
coding regions can be amplified from genomic DNA and
sequenced. In rare cases it may also be necessary to sequence
non-coding regions to establish a genetic reason for a par-
ticular platelet protein deficiency. Regardless of whether a
disorder is homozygous or compound heterozygous in reces-
sive disease, or present in only one allele in dominant disease,
it is extremely useful to establish the familial inheritance
by analyzing DNA from other family members. Modern
“next-generation” sequencing methods are increasingly
 Platelet disorders 253

being used routinely for molecular diagnosis of bleeding
disorders of platelet origin. These are much more effective
when applied to patients with a clear phenotype where
likely genes can be targeted. However, problems arise where
the phenotype is simply mild bleeding and it is difficult
to ascribe the problem to poorly defined mutations/single
nucleotide polymorphisms (SNPs) in genes for unclearly
related proteins. In those cases where this has been effective,
there have generally been sufficient family members with
and without the phenotype to show a clear relationship. In
some cases it has been possible to establish mouse models of
these disorders, confirming and extending the diagnosis.
Platelet adhesion disorders
Bernard–soulier syndrome
Bernard–Soulier syndrome (BSS) is a rare bleeding disorder
caused by defective expression or function of the GPIb–V–
IX receptor complex (Figure 19.2). With a few still unclear
exceptions, it is inherited in an autosomal recessive way, and
homozygous cases are often associated with consanguinity.
BSS is characterized by thrombocytopenia, giant platelets
(up to 20 μm diameter), decreased platelet adhesion, reduced
platelet survival, and abnormal prothrombin consumption.
BSS platelets do not aggregate in response to ristocetin or
botrocetin and show weaker and slower responses to throm-
bin due to the lack/deficiency of GPIb as a thrombin receptor
that accelerates responses. As well as interacting with VWF
and thrombin, GPIb is a receptor for a wide range of lig-
ands with various physiological roles, including P-selectin,
thrombospondin 1, factors XI and XII, αMβ2, and high-
molecular-weight kininogen; its absence in BSS may there-
fore contribute to other aspects of this disorder.
Four separate genes, GPIBA (chromosome 17), GPIBB
(chromosome 22), and GP5 and GP9 (chromosome 3), code
for the subunits of the GPIb complex, which are expressed
relatively late in megakaryocyte maturation. All four subunits
belong to the leucine-rich repeat (LRR) family of proteins,
with 8, 2, 16, and 2 repeats present, respectively. GPIbα and
GPIbβ are linked covalently via disulfide bonds, probably in
a 1 : 2 ratio, while GPIX and GPV associate non-covalently
with GPIb, probably in 1 : 1 and 2 : 1 ratios, respectively (Fig-
ure 19.3). While GPIbβ and GPIX are certainly critical for the
function of the complex, with structural and signaling roles,

GPlb complex
Bernard-Soulier syndrome
Platelet-type von Willibrand disease

Collagen-receptor defects
αllbβ3
Glanzmann thrombasthenia
α2β1
GPVl

7 Transmembrane
receptor defects
e.g. TXA2 receptor

CLEC2 α5β1
α6β1
Fig. 19.2 Diagram of a resting platelet indicating the major receptors implicated in inherited bleeding disorders.
254 Molecular Hematology
VWF-binding
site Thrombin-binding
sites
GPIbα
GPIb–V–IX GPV
complex
Highly O-glycosylated
2GPIbβ
domain
GPIX
Calmodulin
14-3-3ζ
Actin Filamin
filaments
Myosin heavy chain IIA
the function of GPV remains controversial; it has been sug-
gested that it has a role in the thrombin response (it contains
a thrombin-binding and -cleavage site) as well as in platelet
collagen binding.
Mutations and other genetic defects causing BSS have now
been collected on a website (www.bernardsoulier.org) and
will therefore not be listed in detail here. Many of the muta-
tions found to cause BSS lie in the LRR domains and appear
to destabilize their folding. The subunits also contain typi-
cal disulfide bridge patterns, and thus mutation of the con-
stituent cysteines or of other amino acids close to cysteine
has deleterious effects on folding, leading to lack of expres-
sion of the affected subunit and hence decrease in expression
of the rest of the complex.
No cases of BSS due to defects in GPV have been described
and mice lacking GPV express GPIb–IX normally. On the
other hand, expression of GPIb–IX appears to be essential
for stable expression of GPV on platelets, possibly by protect-
ing it from proteolysis. Most mutations in GPIbα either cause
folding problems or are nonsense mutations leading to pre-
mature termination. Rare BSS variants have been described
as having dominant inheritance and lead to expression of
non-functional GPIb, although only one allele is affected.
While the molecular mechanism involved is not understood,
it has been suggested that the defective GPIbα molecule
prevents the normal one from functioning. This class of
mutation includes Leu57Phe, Leu129Pro, Ala156Val (known
as BSS Bolzano), and Leu179del (known as BSS Nancy).
Recently, a monoallelic Val31Leu mutation was described in
an Indian patient with BSS symptoms. Mutations in GPIbβ
generally have the same effect as those in GPIbα concern-
ing expression. Mutations in GPIX, while causing many of
the classic symptoms of BSS, generally do not produce such
a severe phenotype because trace amounts (up to 10%) of
Fig. 19.3 Structure of the GPIb–V–IX complex.
GPIbα is linked via disulfide bridges to two GPIbβ
molecules and this complex is non-covalently
associated with GPIX and GPV in the ratio of
2 : 4 : 2 : 1. Mutations and deletions leading to
Bernard–Soulier syndrome and platelet-type von
Willebrand disease are described in the text.
Mutations affecting cytoskeletal molecules such as
filamin-1 and myosin heavy chain IIA, which
interact directly or indirectly with GPIb complex,
can also lead to giant platelet syndromes.
functional GPIb are often still expressed, enough to support
basic levels of hemostasis if other aspects, such as coagula-
tion factors, are normal. The Asn45Ser mutation in GPIX
is the commonest cause of BSS among northern Europeans,
accounting for the majority of cases among these popula-
tions. Because this mutation generally does not cause such a
marked phenotype as many of the GPIbα and GPIbβ muta-
tions, patients have not been easily detected and diagnosis is
only now catching up. This mutation has also been found in
patients of Turkish origin. Other mutations in GPIX, leading
to its failure to express, produce similar phenotypes.
Mice models of BSS have been produced through targeted
knockout of GPIbα or GPIbβ, but so far not GPIX (most
likely because the rest of the complex is partially expressed).
The phenotype is essentially the same as in humans. Research
on megakaryocytes from these mice indicates that proplatelet
formation and microtubule coil assembly are defective. As
already noted, GPV knockout mice do not show major differ-
ences in phenotype using the present selection criteria and, in
particular, do not have characteristics of BSS.
Platelet-type von Willebrand disease
Platelet-type von Willebrand disease (VWD) is an inherited
dominant bleeding disorder resembling VWD type IIB, but
caused by mutations in GPIbα rather than in the A1 domain
of VWF (see Figures 19.2 and 19.3). Unlike BSS, platelet-type
VWD shows a gain of function, with spontaneous binding
of VWF to platelets. In platelet-type VWD, platelets are
often enlarged like those in BSS, although the reasons for
this are not yet clear. It may be due to enhanced binding
of VWF to GPIb, which blocks other GPIb functions. The
platelet-type VWD mutations that have been described
include Gly233Val/Ser and Met239Val, and a 27 bp deletion

in the region of the gene coding for the macroglycopeptide
domain of GPIbα has been reported to produce a similar
effect. Recently, two related patients were diagnosed with
platelet-type VWD due to a heterozygous 27 bp inframe
deletion in GP1BA that removes amino acids 459–467 and
therefore overlaps the deletion already mentioned. Studies
on mutant proteins suggest that mutations in other amino
acids of the flexible β-loop structure of GPIbα can also
produce this phenotype.
Collagen receptor defects
α2β1 integrin
A few patients have been reported with bleeding problems
related to an α2β1 integrin deficiency, showing defective
platelet adhesion to collagen while responses to other ago-
nists were normal. Both patients were female and became
normal following the menopause, suggesting a hormonal role
in the disorder. Neither α2−/− nor β1−/− mice have major
hemostatic problems, but defects were reported in aggre-
gation to fibrillar collagen and adhesion to soluble colla-
gen. Differences in occlusion times in thrombosis models in
α2−/− or β1−/− mice seem to be dependent on the model
used and remain controversial.
GPVI
Several patients have been described with mild bleed-
ing disorders related to low levels or deficiency of GPVI
Collagen
GPVI
C2-2
Polymorphism sites
Metalloprotease
cleavage site
R
DD
Fig. 19.4 Structure of the GPVI complex. GPVI is linked via a salt bridge to FcγR, which provides most of the signaling to the platelet
cytoplasm. The positions of mutations leading to decreased expression and of polymorphisms thought to affect function are indicated. Activation
of GPVI by autoantibodies can lead to coactivation of matrix metalloproteases and loss of GPVI by cleavage.
Platelet disorders 255
(Figure 19.4). Most of the defects have not been diagnosed
on a molecular level. There are two recent reports of bleed-
ing disorders caused by compound heterozygous mutations
in GPVI. One patient with a lifelong history of bleeding
problems had structurally normal platelets, but a functional
platelet defect. Platelet aggregation was normal except for
an absent response to collagen, convulxin, and the collagen-
related peptide. ATP-dense granule secretion was normal
with ADP, but defective with collagen. Thrombus forma-
tion on a collagen surface in flowing blood was reduced,
but more single platelets were attached. PFA-100 analysis
showed a shortened collagen/ADP closure time. Flow cytom-
etry showed absence of GPVI expression, while immunoblot-
ting showed strongly reduced levels of GPVI. The patient
is compound heterozygous for an out-of-frame 16 bp dele-
tion and a missense mutation S175 N in a highly conserved
residue of the second immunoglobulin-like GPVI domain.
The parents, who do not have clinical bleeding problems, are
heterozygous carriers. The mother carries the S175 N muta-
tion and has a mild platelet functional defect. In vitro studies
showed reduced membrane expression and convulxin bind-
ing in the S175 N mutant compared with the wild-type GPVI
receptor.
The other patient, a 10-year-old girl, had a tendency to
bruising since infancy, a prolonged bleeding time despite
a normal platelet count, and no antiplatelet antibodies.
Collagen-induced platelet activation was null, although
there was an incomplete deficiency of GPVI detected by
flow cytometry. Immunoblotting showed abnormal residual
R38C
C2-1
N-glycosylation
S175N
O-glycosylation
Salt bridge
ITAM-domains
256 Molecular Hematology
GPVI, and no FcγR defect. DNA sequencing revealed an
R38C mutation in exon 3 of one allele of GPVI and an inser-
tion of five nucleotides in exon 4 of the other allele, leading
to a premature nonsense codon and absence of the corre-
sponding mRNA. Expression of the R38C mutation gave
an abnormal protein migration and loss of collagen bind-
ing. This composite genetic GPVI defect leads to absence of
platelet responses to collagen and a mild bleeding phenotype.
Platelets from GPVI−/− mice do not adhere to colla-
gen under static conditions, probably because the α2β1
integrin is not activated. GPVI is probably important for
platelet activation on subendothelium under low-shear con-
ditions. Under high-shear conditions in GPVI−/− mice,
α2β1 is probably activated via GPIb/VWF interactions,
with VWF bound to collagen. Stable adhesion and spread-
ing are strongly affected in GPVI−/− platelets, indicating
that GPVI has an important role in downstream signaling
to molecules critical for thrombus formation. Tail bleeding
times in GPVI−/− mice are only slightly prolonged compared
with wild type. Thrombosis models using both GPVI−/− and
Fcγ−/− (the GPVI signaling subunit) mice are controversial,
because the type of vascular wall damage involved is vari-
able. Thus, laser-induced injuries did not implicate GPVI as a
major factor in thrombosis, whereas FeCl3 -induced injuries
suggested a more substantial role. Other research pointed
to thrombin production as a major factor in overcoming
GPVI/Fcγ deficiencies in mice thrombosis models. There is
a clear need for research involving mouse thrombosis mod-
els that are more closely related to human pathology, such as
in the presence of fragile vascular plaque, before any defini-
tive conclusions can be reached about collagen receptors in
human thrombosis. Increased amounts of soluble GPVI in
plasma seem to be indicative of raised platelet activation in
vivo, for whatever reason. This may include vascular prob-
lems, inflammation, cirrhosis, or latent cancer.
Platelet aggregation defects:
Glanzmann thrombasthenia
Glanzmann thrombasthenia (GT) is an autosomal recessive
bleeding disorder caused by defects in the αIIbβ3 integrin
or its signaling (see Figures 19.2 and 19.5). If this receptor
is absent or defective, it is unable to interact with its ligands,
including fibrinogen and VWF, but also fibronectin and vit-
ronectin. Thus, platelet aggregates are not formed or are lim-
ited in extent, preventing efficient recruitment of platelets to a
damaged vessel site. In addition, αIIbβ3– fibrinogen binding
has an important role in clot retraction, the process by which
wound edges are drawn together to help seal the wound.
Hence, GT patients form more scar tissue than normal
because wounds tend to reopen. Patients with platelets lack-
ing, or with a severe deficiency (<5% of normal) in αIIbβ3,
Ligand-
αIIb β3
binding site
β-Propeller βA
Ca2+-binding Disulfide-rich
sites domains
Hybrid
Thigh PSI
EGF-1
EGF-2
Calf-1
EGF-3
EGF-4
Calf-2
βTD
Membrane
Talin
Kindlin-3
Fig. 19.5 Structure of the αIIbβ3 complex. This major platelet
integrin is shown in the activated state. The various structural domains
are labeled. Cytoskeletal proteins associated with the cytoplasmic
domain of β3 in the activated state are shown.
are traditionally designated as type I, those with a moderate
deficiency (10–50% of normal) as type II, and others as vari-
ants. The latter includes patients who express non-functional
αIIbβ3 on their platelets. There is a very wide range of molec-
ular defects in αIIb or β3 and as a consequence the clini-
cal symptoms are highly variable. However, even the same
defect may show some variability in symptoms in different
patients, depending on other factors such as levels of coagu-
lation factors. Typical symptoms include purpura, epistaxis,
gum bleeding, and menorrhagia, but gastrointestinal and
interjoint bleeding and hematuria are rare. Problems gener-
ally follow trauma and are rarely spontaneous. Diagnostic cri-
teria are prolonged bleeding time and defective clot retrac-
tion. Platelet aggregation to all agonists except ristocetin is
absent or defective. Molecular diagnosis of GT includes flow
cytometry and/or Western blotting to establish whether both
subunits are absent or present in reduced amounts.
Mutations and other genetic defects causing GT have
now been collected on a database (http://sinaicentral.mssm.
edu/intranet/research/glanzmann) and are therefore not
listed in detail here. GT can be caused by mutations in either
subunit αIIb or β3 and may show some distinct differences.
Since fibrinogen is not synthesized in the megakaryocyte
but in the liver and transported to the α-granules from
the plasma by αIIbβ3-dependent endocytosis, GT patients

often lack or are deficient in platelet α-granule fibrinogen,
a contributory factor to their poor hemostatic function.
In patients lacking β3, who as a consequence lack the
vitronectin receptor αvβ3 as well as αIIbβ3, the platelet
α-granules, in addition to fibrinogen deficiency, also contain
up to five times the normal amounts of vitronectin. Vit-
ronectin is synthesized in megakaryocytes, so these results
suggest a possible role for αvβ3 in transporting vitronectin
from α-granules out of platelets. Both αIIb and β3 have
a complex gene structure, with the αIIb gene (ITGA2B)
composed of 30 exons and spanning 17 kb, while the β3
gene (ITGB3) consists of 15 exons and spans 46 kb. Both
are located on chromosome 17q21–23. Genetic defects are
distributed over the entire region.
Mutations/deletions may prevent subunit biosynthesis
or transport of precursors from the endoplasmic reticulum
to the Golgi or plasma membrane. When mutations do not
affect folding too seriously, expression of lower levels of the
complex may occur with some functions retained. Several
GT mutations lead to expressed dysfunctional receptor.
Thus, an Asp119Tyr mutation in β3 affects the key RGD
binding site, while Ser752Pro removes a key phosphory-
lation site involved in outside–in signaling. A stop codon
truncating β3 (Arg724tTer), so that only 8 of the normal 47
amino acids of the cytoplasmic domain are present, removes
binding sites for the cytoskeletal proteins talin and kindlin-2.
Cys560Arg and Cys598Tyr mutations in β3 produce platelets
that spontaneously bind fibrinogen, because the β3 in their
αIIbβ3 has its disulfide bridges blocked in the activated
pattern, and because the disulfide isomerase activity that
normally switches between resting and activated and vice
versa is unable to do so. A heterozygous cytoplasmic domain
mutation in the αIIb subunit Arg995Gln within the critical
GFFKR sequence reduces expression of the complex and
produces mild “thrombasthenia-like” symptoms. There is
some evidence for cases of GT caused by replacement of
DNA segments lying outside coding regions and affect-
ing mRNA stability. A more general syndrome affecting
inside–out activation of several classes of integrins present
in platelets, neutrophils, and lymphocytes has been shown to
be due to mutations in the CalDAG-GEFI gene, preventing
signaling to the small GTPase Rap1.
Recent studies on the linkage between ITGA2B and ITGB3
gene variants in patients with GT show disequilibrium, indi-
cating that most disease-causing mutations are recent.
Agonist receptor defects
ADP and ATP receptor defects
Platelets interact with ADP via the purinergic seven-
transmembrane receptors P2Y1 and P2Y12 . P2Y1 signals to
Platelet disorders 257
release Ca2+ from the dense tubular system and cytoskele-
tal changes lead to shape change. P2Y12 signals to αIIbβ3,
leading to platelet aggregation. Thus, platelet defects in P2Y12
were first thought to be GT variants. Platelets from rare
patients in both France and Italy showed decreased and
reversible platelet aggregation to ADP, but normal shape
change and calcium mobilization. This was due to an auto-
somal recessive hereditary disease affecting one allele of
the P2Y12 gene. No patients with P2Y1 defects have been
described so far, but knockout mice lacking this receptor have
been prepared. P2X1 is the platelet receptor for ATP. Three
molecules form a calcium channel that is regulated when
platelets are activated, to control intracellular calcium levels.
One case of a young girl with a bleeding syndrome caused by
deletion of a single amino acid in P2X1 has been described.
That this is a dominant inherited disease is presumably due to
the defective molecule preventing the active calcium channel
from forming.
Other primary agonist receptor defects
Rare examples of an Arg60Leu mutation in the thromboxane
(TX)A2 receptor have been reported in some Japanese fami-
lies that result in defective signaling. Mutations in other ago-
nist receptors have been reported, but it is not clear that these
affect platelet function. Common variants in the PAR4 gene
(F2RL3) have been described, but only recently were effects
of these on platelet activation observed, particularly in the
presence of clinically used PAR1 inhibitors. So far there have
been no reports of bleeding disorders linked to defects in the
PAR1 thrombin receptors.
Defects in intracellular signaling
pathways
A wide range of defects in signaling molecules are known,
often affecting other cells as well as platelets (Figure 19.6).
These include defects in phospholipase C (PLC) activation,
calcium mobilization, and plekstrin phosphorylation. Stud-
ies in one of these patients with impaired PLC activation
indicated a decrease only in PLC-β2, while other isoforms
were normal. Coding sequence was normal, but mRNA for
PLCβ2 was decreased, suggesting sequence changes in the
gene outside the coding region affecting mRNA stability.
Eight patients were described with bleeding problems and
abnormal aggregation and secretion in response to several
different agonists. Receptor-mediated calcium mobiliza-
tion and/or plekstrin phosphorylation were abnormal in
seven of these patients. Several patients have been reported
with platelet cyclooxygenase deficiency, a milder bleeding
disorder, and impaired platelet aggregation responses.
Thromboxane synthase deficiency has been described in two
258 Molecular Hematology

Giant platelet syndromes

Myosin heavy chain llA
Filamin 1

α-granule syndromes
Metabolic defects Gray platelet syndrome
ATP synthesis α, δ-deficiency
Quebec syndrome

Enzyme defects
Cyclooxygenase
Thromboxane synthase
Signaling defects Dense granule syndromes
Lipoxygenase
Kinases/phosphatases Chédiak–Higashi syndrome
G-proteins, Ca2+ fluxes Hermansky–Pudlak syndrome
Phospholipases/PKC Griscelli syndrome
Fig. 19.6 Diagram of a resting platelet indicating the major internal organelles and other structures implicated in inherited bleeding disorders.

patients. In other patients a deficiency of the PLC-γ2 isoform
or a specific decrease in platelet Gαq has been reported.
Defects in signaling molecules probably account for many
of the undiagnosed patients with mild to moderate bleeding
problems. Techniques to diagnose these on a molecular basis
are only starting to be developed, and even proteomics by
itself cannot be expected to detect all of these. There is still
a large unmet need here.
Platelet secretion defects (storage
pool disease)
This is a large and heterogeneous group of inherited disor-
ders caused by intracellular defects of platelets (Figure 19.6).
Since secretion is a function of many cell types, it is clear that
the phenotype may extend beyond hemostasis, depending on
which molecule is affected.
Defects of α-granules
Proteins in α-granules are either synthesized in megakary-
ocytes or endocytosed from plasma. Membranes of
α-granules also contain specific receptors, such as P-selectin
and CD63, synthesized in megakaryocytes that translocate
to the plasma membrane during platelet activation and
secretion and are good markers for these processes. Inher-
ited deficiencies of plasma proteins that are taken up into
α-granules will be reflected in their content.
Gray platelet syndrome
Gray platelet syndrome is characterized by the absence
of platelet contents and has a mostly autosomal recessive
inheritance. Packaging or storage of proteins during platelet
production in megakaryocytes is defective and the pro-
teins are directly released from the megakaryocytes. The
platelets are often larger than normal and the patients are
somewhat thrombocytopenic. Platelet aggregation responses
are variously affected and differ between patients. Tissue
inhibitors of metalloproteinases (TIMPs) are still present
in gray platelet syndrome and may account for some of this
variability. In 2011, three groups applied next-generation
sequencing to study the genetic defect underlying gray
platelet syndrome and showed mutations in NBEAL2 in
all patients. NBEAL2 is a member of the LYST (mutated in
Chesiak–Higashi syndrome) gene family. Following these
results, the production of NBEAL2−/− mice has extended

phenotype studies in gray platelet syndrome. After injury,
tissue regeneration was affected. Megakaryocyte maturation
was delayed and few α-granules observed. Unusual direct
secretion of VWF was also seen.
Quebec platelet disorder
As the name implies, this autosomal dominant bleeding
disorder was described in this region of Canada. Many
α-granule proteins are degraded proteolytically, including
membrane receptors such as P-selectin. Defective platelet
aggregation is particularly strong, with epinephrine as
agonist. The problem is caused by overexpression of uroki-
nase plasminogen activator, a protease normally stored in
α-granules and released on platelet activation.
Dense (δ) granule defects
Platelet-dense granules are storage sites for a number
of small molecules, including ADP, ATP, serotonin, and
calcium. Defects in these granules therefore affect platelet
aggregation responses (Figure 19.6). Granule deficiencies
are variable and disorders are known in which α-granules
are also affected, in which common pathway molecules are
defective (αδ-storage pool deficiency). Since similar types of
organelles exist in other cells, these are often disorders that
affect pigmentation of the skin and hair, such as Hermansky–
Pudlak, Chédiak–Higashi, and Griscelli syndromes. Mouse
models exist for several of these.
Hermansky–pudlak syndrome
Hermansky–Pudlak syndrome manifests as albinism and
absence of ceroid-lipofuscin storage in the reticuloendothe-
lial system. The syndrome is common in Puerto Rico and is
caused by a frameshift due to a 16 bp duplication in exon 15
of the HPS-1 gene, which normally encodes a 79 kDa protein
with two membrane-spanning domains. Defects in at least
eight genes cause distinct subtypes of Hermansky–Pudlak
syndrome in humans.
Chédiak–higashi syndrome
Severe immunological defects and progressive neurological
dysfunction accompany bleeding problems. The major diag-
nostic criterion is the presence of giant inclusion bodies in
a variety of cells with granules, including platelets. The gene
responsible (LYST) has been cloned and a series of frameshift
and nonsense mutations identified that produce a truncated
LYST protein. A milder form of the disease is thought to be
due to rare missense mutations. LYST protein is large and
complex, with domains suggesting that it regulates organelle
protein trafficking and membrane–membrane interactions.
Platelet disorders 259
Griscelli syndrome
Patients with Griscelli syndrome are partially albino with
silver hair. Neurological defects and/or immunodeficiency
are linked to lymphocyte cytotoxicity defects. Clinical man-
ifestations include fatal complications caused by lymphoid
cell activation and cytokine release. Griscelli syndrome
is associated with mutations in the genes for myosin Va,
Rab27a (a small GTPase), or melanophilin. A mouse
model for this disease has recently been characterized with a
mutation in Rab27a, a variant of the ashen mouse with a phe-
notype resembling Hermansky–Pudlak syndrome. Rab27a
deficiency in humans was identified in a young patient with
psychomotor retardation. The patient’s platelets lacked dense
granules, but had normal α-granules; epinephrine-induced
aggregation was defective.
Wiskott–aldrich syndrome
Platelets in Wiskott–Aldrich syndrome (WAS) aggregate
poorly and have few granules. This is an X-linked recessive
disease characterized by thrombocytopenia, small platelets,
eczema, immunodeficiency, and increased autoimmunity
and malignancy problems. Hereditary X-linked thrombocy-
topenia is a milder form lacking the immune problems. T
lymphocytes are the other major blood cells affected. The
502-amino-acid WAS protein is encoded by the large WAS
gene, which consists of 12 exons. Mutations in exons 1 and
2 mostly lead to the mild form. WAS protein regulates actin
polymerization in hematopoietic cells and its deficiency leads
to premature proplatelet formation in the bone marrow. In
mice, profilin-1 deficiency in megakaryocytes was shown to
lead to microtubule instability and a large increase in acety-
lated tubulin, as well as affecting platelet production and
causing a WAS-like platelet defect.
Procoagulant regulation defects
Scott syndrome
Scott syndrome is a very rare disorder found in a few humans
and some dogs. So far there is no mouse model. When nor-
mal platelets are activated, one of the changes produced is
the exposure of negatively charged phospholipids on the
surface, critical for procoagulant activity. This is absent in
Scott syndrome platelets and as a consequence less thrombin
and fibrin are formed in the platelet plug, leading to bleeding
complications. This defect also affects membrane vesicu-
lation, suggesting a common mechanism. Under in vitro
conditions, this phenomenon requires platelet activation by
thrombin and collagen (or a GPVI agonist such as convulxin
or collagen-related peptide). The molecular defect in Scott
260 Molecular Hematology
syndrome was recently attributed to mutations in the gene
for anoctamin 6 (TMEM16F), a Ca2+-activated Cl− channel.
Stormorken syndrome
Stormorken syndrome is an even rarer disorder confined so
far to one family, implying a dominant inheritance. Unlike
Scott syndrome, the platelets are constitutively activated,
with surface exposure of negatively charged phospholipids
and raised levels of microvesicles even under resting con-
ditions. The molecular basis of Stormorken syndrome was
recently shown to be a dominant mutation in STIM1. Gain-
of-function mutations in STIM1 are responsible for a CRAC
channel defect in York platelet syndrome characterized by
thrombocytopenia, ultrastructural granule abnormalities
including giant dense granules and massive, multilayered
target bodies, and lack of Ca2+ -storage in δ-granules.
Giant platelet syndromes and
cytoskeletal defects
Although this group is diverse, many of the patients have
mutations or other defects in the MYH9 gene coding for non-
muscle myosin heavy chain IIA (see Figures 19.3 and 19.6).
Since this is expressed in a variety of cells, not only platelets
are affected. Macrothrombocytopenia (variable, up to leuko-
cyte size) is accompanied by variable phenotypes in other
cells/tissues. Döhle bodies in leukocytes with distinctive
distribution of myosin IIA detected by immunofluorescence
is a major diagnostic criterion. Hearing loss caused by defec-
tive cilia, nephritis, and cataracts are all possible symptoms
in later life. Characteristic mutation zones in the protein
seem linked to particular symptoms: mutations in the head
region can affect folding of the entire molecule and can
yield more serious symptoms, whereas mutations toward
the C-terminus have less dramatic effects, possibly because
function is partly maintained and there is less tendency to
form denatured protein clusters. While non-muscle myosin
has an important role in many cells/tissues, the presence of
the IIB and IIC forms may compensate in some of these.
Giant platelets have also been reported in patients with
mutations in the FLNA gene coding for filamin A. As
well as platelets, neuronal migration was abnormal in these
patients. In platelets, filamin A is a major attachment site
for the GPIb complex, stabilizing the connection between
the membrane and the cytoskeleton, and anchoring GPIb
in platelets when they attach during adhesion under high
shear. Myosin IIA may play a role in this process, as well
providing a rationale for the giant platelets in these disor-
ders as well as BSS. Recently, a number of patients have been
described with mutations in the β-tubulin gene, also leading
to macrothrombocytopenia, and similar defects were found
in dogs. It is important to distinguish these large platelet syn-
dromes (including BSS), particularly when accompanied by
thrombocytopenia, from idiopathic thrombocytopenic pur-
pura, and to avoid treatments such as splenectomy, which are
not useful in these disorders.
Transcription factor defects
Autosomal dominant Paris–Trousseau syndrome is a typi-
cal example of this class of disorder, with decreased platelet
production and a mild bleeding tendency associated with a
deletion at 11q23. Platelets are often larger than normal with
giant α-granules. The defect leads to a hemizygous loss of the
FLI1 gene coding for Fli1 transcription factor. A subpopula-
tion of megakaryocyte precursors fails to mature adequately
to produce platelets. Other disorders including thrombo-
cytopenia are caused by mutations in the GATA-1 gene.
Platelets may be larger than normal, with poor response to
collagen. The phenotype depends on whether the mutations
affect GATA-1 interactions with DNA or FOG-1. GATA-1-
dependent platelet molecules include members of the GPIb
complex and α-granule components. Defects in RUNX1 lead
to platelets with a deficiency in protein kinase C-θ and lack
of phosphorylation of its downstream targets, but also an
increased tendency to develop leukemia.
Platelet protein polymorphisms and
tendency to thrombosis
Platelet activation responses vary widely in the population,
but are reproducible for individuals and show a strong genetic
component. Following the demonstration of major polymor-
phisms in coagulation factors affecting thrombotic tenden-
cies, there has been a major effort to identify polymorphisms
in platelet proteins such as receptors that could explain the
interindividual variation. There is also considerable interest
in polymorphisms that affect patient responses to antithrom-
botic treatments, because this could be an important way to
improve treatment and reduce mortality.
Common polymorphisms in αIIbβ3 were among the
first targets for statistical investigation. Each of the sub-
units contains a common amino acid dimorphism: β3 has
two isoforms, Leu33/Pro33, that lead to conformational
differences in the proteins; the Ile843/Ser843 dimorphism
in αIIb has also been intensively investigated. Most of the
studies were controversial, with only minor effects detected.
The situation in GPIbα is more complicated, with two
major polymorphic systems present: (i) a Thr145/Met145
dimorphism and (ii) several versions of a tandem repeat
structure of 13 amino acids within the macroglycopeptide

domain; these have the nomenclature VLTR-A, -B, -C, and
-D with four, three, two, and one copies, respectively, of the
tandem repeat. The incidence varies widely depending on
ethnic origins between Europe and East Asia. Again, there
has been no consensus reached about a role in thrombosis
susceptibility and many studies have given contradictory
results. Polymorphisms in collagen receptors α2β1 and
GPVI have also been implicated in thrombosis susceptibility,
but these studies remain controversial.
Perhaps if one considers that many individual genetic
differences may contribute to the overall thrombosis sus-
ceptibility, it is not so surprising that these single differences
do not have more pronounced effects. However, several
newly discovered polymorphisms, particularly in signaling
molecules, are still under investigation. One area which is
less controversial is the effect of cytochrome P450 polymor-
phisms on activation of clopidogrel, and therefore on its
ability to inhibit the ADP receptor P2Y12 . Among persons
treated with clopidogrel, carriers of a reduced-function
CYP2C19 allele had significantly lower levels of the active
metabolite of clopidogrel, diminished platelet inhibition,
and a higher rate of major adverse cardiovascular events,
including stent thrombosis, than did non-carriers.
Further reading
Overview of platelet function
Clemetson, K.J. (2007). A short history of platelet glycoprotein Ib com-
plex. Thromb. Haemost. 98: 63–68.
Clemetson, K.J. (2012). Platelets and primary haemostasis. Thromb. Res.
129: 220–224.
Clemetson, K.J. and Clemetson, J.M. (2001). Platelet collagen receptors.
Thromb. Haemost. 86: 189–197.
Coller, B.S. and Shattil, S.J. (2008). The GPIIb/IIIa (integrin alphaIIb-
beta3) odyssey: a technology-driven saga of a receptor with twists,
turns, and even a bend. Blood 112: 3011–3025.
Farndale, R.W., Slatter, D.A., Siljander, P.R., and Jarvis, G.E. (2007).
Platelet receptor recognition and cross-talk in collagen-induced acti-
vation of platelets. J. Thromb. Haemost. 5: 220–229.
Hartwig, J.H. and Italiano, J.E. Jr. (2006). Cytoskeletal mechanisms for
platelet production. Blood Cell Mol. Dis. 36: 99–103.
Jackson, S.P. (2007). The growing complexity of platelet aggregation.
Blood 109: 5087–5095.
Lau, T.L., Kim, C., Ginsberg, M.H., and Ulmer, T.S. (2009). The struc-
ture of the integrin αIIbβ3 transmembrane complex explains inte-
grin transmembrane signalling. EMBO J. 28: 1351–1361.
Ozaki, Y., Asazuma, N., Suzuki-Inoue, K., and Berndt, M.C. (2005).
Platelet GPIb-IX-V-dependent signaling. J. Thromb. Haemost. 3:
1745–1751.
Plow, E.F., Qin, J., and Byzova, T. (2009). Kindling the flame of integrin
activation and function with kindlins. Curr. Opin. Hematol. 16: 323–
328.
Platelet disorders 261
Prevost, N., Woulfe, D., Tognolini, M., and Brass, L.F. (2003). Contact-
dependent signaling during the late events of platelet activation. J.
Thromb. Haemost. 1: 1613–1627.
Ren, Q., Ye, S., and Whiteheart, S.W. (2008). The platelet release reac-
tion: just when you thought platelet secretion was simple. Curr. Opin.
Hematol. 15: 537–541.
Rivera, J., Lozano, M.L., Navarro-Nunez, L., and Vicente, V. (2009).
Platelet receptors and signaling in the dynamics of thrombus forma-
tion. Haematologica 94: 700–711.
Ruggeri, Z.M. and Mendolicchio, G.L. (2007). Adhesion mechanisms in
platelet function. Circ. Res. 100: 1673–1685.
Laboratory assessment of platelet
disorders
Harrison, P. (2009). Assessment of platelet function in the laboratory.
Hamostaseologie 29: 25–31.
Hayward, C.P. and Favaloro, E.J. (2009). Diagnostic evaluation of
platelet disorders: the past, the present, and the future. Semin.
Thromb. Hemost. 35: 127–130.
Simeoni, I., Stephens, J.C., Hu, F. et al. (2016). A high-throughput
sequencing test for diagnosing inherited bleeding, thrombotic, and
platelet disorders. Blood 127: 2791–2803.
Overview of inherited platelet disorders
Johnson, B., Lowe, G.C., Futterer, J. et al. (2016). Whole exome sequenc-
ing identifies genetic variants in inherited thrombocytopenia with
secondary qualitative function defects. Haematologica 101 (10):
1170–1179.
Kunicki, T.J., Williams, S.A., and Nugent, D.J. (2012). Genetic variants
that affect platelet function. Curr. Opin. Hematol. 19: 371–379.
Lanza, F. (2007). Murine models of platelet diseases. Transfus. Clin. Biol.
14: 35–40.
Nurden, P. and Nurden, A.T. (2008). Congenital disorders associated
with platelet dysfunctions. Thromb. Haemost. 99: 253–263.
Nurden, A.T. and Nurden, P. (2015). Inherited disorders of platelet func-
tion: selected updates. J. Thromb. Haemost. 13 (Suppl. 1): S2–S9.
Rao, A.K., Jalagadugula, G., and Sun, L. (2004). Inherited defects in
platelet signaling mechanisms. Semin. Thromb. Hemost. 30: 525–535.
Salles, I.I., Feys, H.B., Iserbyt, B.F. et al. (2008). Inherited traits affecting
platelet function. Blood Rev. 22: 155–172.
Bernard–soulier syndrome overview
Dumas, J.J., Kumar, R., McDonagh, T. et al. (2004). Crystal structure of
the wild-type von Willebrand factor A1–glycoprotein Ibα complex
reveals conformation differences with a complex bearing von Wille-
brand disease mutations. J. Biol. Chem. 279: 23327–23334.
Lanza, F. (2006). Bernard–Soulier syndrome (hemorrhagiparous
thrombocytic dystrophy). Orphanet J. Rare Dis. 1: 46.
Pham, A. and Wang, J. (2007). Bernard–Soulier syndrome: an inherited
platelet disorder. Arch. Pathol. Lab. Med. 131: 1834–1836.
Savoia, A., Kunishima, S., De Rocco, D. et al. (2014). Spectrum of
the mutations in Bernard-Soulier syndrome. Hum. Mutat. 35: 1033–
1045.
262 Molecular Hematology
Uff, S., Clemetson, J.M., Harrison, T. et al. (2002). Crystal structure of
the platelet glycoprotein Ibα N-terminal domain reveals an unmask-
ing mechanism for receptor activation. J. Biol. Chem. 277: 35657–
35663.
New GPIbα mutations
Imai, C., Kunishima, S., Takachi, T. et al. (2009). A novel homozygous 8-
base pair deletion mutation in the glycoprotein Ibα gene in a patient
with Bernard–Soulier syndrome. Blood Coagul. Fibrinolysis 20: 470–
474.
Noris, P., Perrotta, S., Bottega, R. et al. (2012). Clinical and laboratory
features of 103 patients from 42 Italian families with inherited throm-
bocytopenia derived from the monoallelic Ala156Val mutation of
GPIbα (Bolzano mutation). Haematologica 97: 82–88.
Rosenberg, N., Lalezari, S., Landau, M. et al. (2007). Trp207Gly in
platelet glycoprotein Ibα is a novel mutation that disrupts the connec-
tion between the leucine-rich repeat domain and the disulfide loop
structure and causes Bernard–Soulier syndrome. J. Thromb. Haemost.
5: 378–386.
Vettore, S., Scandellari, R., Moro, S. et al. (2008). Novel point mutation
in a leucine-rich repeat of the GPIbα chain of the platelet von Wille-
brand factor receptor, GPIb/IX/V, resulting in an inherited dominant
form of Bernard–Soulier syndrome affecting two unrelated families:
the N41H variant. Haematologica 93: 1743–1747.
New GPIbβ mutations
Hadjkacem, B., Elleuch, H., Gargouri, J., and Gargouri, A. (2009).
Bernard–Soulier syndrome: novel nonsense mutation in GPIbβ gene
affecting GPIb–IX complex expression. Ann. Hematol. 88: 465–472.
Velocardiofacial syndrome
Liang, H.P., Morel-Kopp, M.C., Curtin, J. et al. (2007). Heterozygous loss
of platelet glycoprotein (GP)Ib-V-IX variably affects platelet function
in velocardiofacial syndrome (VCFS) patients. Thromb. Haemost. 98:
1298–1308.
New GPIX mutations
Kunishima, S., Yamada, T., Hamaguchi, M., and Saito, H. (2006).
Bernard–Soulier syndrome due to GPIX W127X mutation in Japan
is frequently misdiagnosed as idiopathic thrombocytopenic purpura.
Int. J. Hematol. 83: 366–367.
Variant forms
Balduini, A., Malara, A., Pecci, A. et al. (2009). Proplatelet formation
in heterozygous Bernard–Soulier syndrome type Bolzano. J. Thromb.
Haemost. 7: 478–484.
Vettore, S., Scandellari, R., Scapin, M. et al. (2008). A case of Bernard–
Soulier syndrome due to a homozygous four bases deletion (TGAG)
of GPIbα gene: lack of GPIbα but absence of bleeding. Platelets 19:
388–391.
Platelet-type von Willebrand disease
Matsubara, Y., Murata, M., Sugita, K., and Ikeda, Y. (2003). Identification
of a novel point mutation in platelet glycoprotein Ibα Gly to Ser at
residue 233, in a Japanese family with platelet-type von Willebrand
disease. J. Thromb. Haemost. 1: 2198–2205.
Nurden, P., Lanza, F., Bonnafous-Faurie, C., and Nurden, A. (2007).
A second report of platelet-type von Willebrand disease with a
Gly233Ser mutation in the GPIBA gene. Thromb. Haemost. 97: 319–
321.
Othman, M., Notley, C., Lavender, F.L. et al. (2005). Identification
and functional characterization of a novel 27-bp deletion in the
macroglycopeptide-coding region of the GPIBA gene resulting in
platelet-type von Willebrand disease. Blood 105: 4330–4336.
Suva, L.J., Hartman, E., Dilley, J.D. et al. (2008). Platelet dysfunction and
a high bone mass phenotype in a murine model of platelet-type von
Willebrand disease. Am. J. Pathol. 172: 430–439.
Collagen receptor defects α2β1 integrin
Joutsi-Korhonen, L., Smethurst, P.A., Rankin, A. et al. (2003). The low-
frequency allele of the platelet collagen signaling receptor glycopro-
tein VI is associated with reduced functional responses and expres-
sion. Blood 101: 4372–4379.
Sarratt, K.L., Chen, H., Zutter, M.M. et al. (2005). GPVI and α2β1 play
independent critical roles during platelet adhesion and aggregate for-
mation to collagen under flow. Blood 106: 1268–1277.
GPVI
Arthur, J.F., Dunkley, S., and Andrews, R.K. (2007). Platelet glycopro-
tein VI-related clinical defects. Br. J. Haematol. 139: 363–372.
Dumont, B., Lasne, D., Rothschild, C. et al. (2009). Absence of collagen-
induced platelet activation caused by compound heterozygous GPVI
mutations. Blood 114: 1900–1903.
Hermans, C., Wittevrongel, C., Thys, C. et al. (2009). A compound het-
erozygous mutation in glycoprotein VI in a patient with a bleeding
disorder. J. Thromb. Haemost. 7: 1356–1363.
Hughan, S.C., Senis, Y., Best, D. et al. (2005). Selective impairment of
platelet activation to collagen in the absence of GATA1. Blood 105:
4369–4376.
Glanzmann thrombasthenia overview
Fang, J., Hodivala-Dilke, K., Johnson, B.D. et al. (2005). Therapeu-
tic expression of the platelet-specific integrin, αIIbβ3, in a murine
model for Glanzmann thrombasthenia. Blood 106: 2671–2679.
French, D.L. (1998). The molecular genetics of Glanzmann’s thrombas-
thenia. Platelets 9: 5–20.
Jayandharan, G., Nelson, E.J., Baidya, S. et al. (2007). A new multi-
plex PCR and conformation-sensitive gel electrophoresis strategy for
mutation detection in the platelet glycoprotein αIIb and β3 genes. J.
Thromb. Haemost. 5: 206–209.
Manickam, N., Sun, X., Li, M. et al. (2008). Protein disulphide isomerase
in platelet function. Br. J. Haematol. 140: 223–229.
Nurden, A.T., Pillois, X., Fiore, M. et al. (2015). Expanding the mutation
spectrum affecting αIIbβ3 Integrin in Glanzmann thrombasthenia:
screening of the ITGA2B and ITGB3 genes in a large international
cohort. Hum. Mutat. 36: 548–561.

αIIb defects
Losonczy, G., Rosenberg, N., Boda, Z. et al. (2007). Three novel muta-
tions in the glycoprotein IIb gene in a patient with type II Glanzmann
thrombasthenia. Haematologica 92: 698–701.
Vijapurkar, M., Ghosh, K., and Shetty, S. (2009). Novel mutations in
GPIIb gene in Glanzmann’s thrombasthenia from India. Platelets 20:
35–40.
β3 defects
Chen, P., Melchior, C., Brons, N.H. et al. (2001). Probing conformational
changes in the I-like domain and the cysteine-rich repeat of human
α3 integrins following disulfide bond disruption by cysteine muta-
tions: identification of cysteine 598 involved in αIIbβ3 activation. J.
Biol. Chem. 276: 38628–38635.
D’Andrea, G., Bafunno, V., Del Vecchio, L. et al. (2008). A α3
Asp217αVal substitution in a patient with variant Glanzmann throm-
basthenia severely affects integrin αIIbβ3 functions. Blood Coagul.
Fibrinolysis 19: 657–662.
Gonzalez-Manchon, C., Butta, N., Larrucea, S. et al. (2004). A
variant thrombasthenic phenotype associated with compound
heterozygosity of integrin α3-subunit: (Met124Val)α3 alters the
subunit dimerization rendering a decreased number of con-
stitutive active αIIbβ3 receptors. Thromb. Haemost. 92: 1377–
1386.
Liu, J., Jackson, C.W., Gruppo, R.A. et al. (2005). The α3 subunit of the
integrin αIIbβ3 regulates αIIb-mediated outsidein signaling. Blood
105: 4345–4352.
Mor-Cohen, R., Rosenberg, N., Peretz, H. et al. (2007). Disulfide bond
disruption by a α3-Cys549Arg mutation in six Jordanian fami-
lies with Glanzmann thrombasthenia causes diminished produc-
tion of constitutively active αIIbβ3. Thromb. Haemost. 98: 1257–
1265.
Nurden, P., Poujol, C., Winckler, J. et al. (2002). A Ser752βPro substi-
tution in the cytoplasmic domain of α3 in a Glanzmann thrombas-
thenia variant fails to prevent interactions between the αIIbβ3 inte-
grin and the platelet granule pool of fibrinogen. Br. J. Haematol. 118:
1143–1151.
Ruiz, C., Liu, C.Y., Sun, Q.H. et al. (2001). A point mutation in the
cysteine-rich domain of glycoprotein (GP) IIIa results in the expres-
sion of a GPIIb-IIIa (αIIbβ3) integrin receptor locked in a high-
affinity state and a Glanzmann thrombasthenia-like phenotype. Blood
98: 2432–2441.
Wang, R., Shattil, S.J., Ambruso, D.R., and Newman, P.J. (1997). Trunca-
tion of the cytoplasmic domain of α3 in a variant form of Glanzmann
thrombasthenia abrogates signaling through the integrin αIIbβ3
complex. J. Clin. Investig. 100: 2393–2403.
Kindlins and leukocyte adhesion defect
Pasvolsky, R., Feigelson, S.W., Kilic, S.S. et al. (2007). A LAD-III
syndrome is associated with defective expression of the Rap-1
activator CalDAG-GEFI in lymphocytes, neutrophils, and platelets.
J. Exp. Med. 204: 1571–1582.
Rognoni, E., Ruppert, R., and Fässler, R. (2016). The kindlin family:
functions, signaling properties and implications for human disease.
J. Cell Sci. 129: 17–27.
Platelet disorders 263
ADP and ATP receptor defects
Cattaneo, M. (2006). Disorders of platelet function. Abnormalities of
the platelet P2 receptors. Pathophysiol. Haemost. Thromb. 35: 10–14.
Daly, M.E., Dawood, B.B., Lester, W.A. et al. (2009). Identification and
characterization of a novel P2Y12 variant in a patient diagnosed with
type 1 von Willebrand disease in the European MCMDM-1VWD
study. Blood 113: 4110–4113.
Leon, C., Hechler, B., Freund, M. et al. (1999). Defective platelet aggre-
gation and increased resistance to thrombosis in purinergic P2Y1
receptor-null mice. J. Clin. Investig. 104: 1731–1737.
Oury, C., Toth-Zsamboki, E., Vermylen, J., and Hoylaerts, M.F. (2002).
Does the P2X1del variant lacking 17 amino acids in its extracellular
domain represent a relevant functional ion channel in platelets? Blood
99: 2275–2277.
Oury, C., Toth-Zsamboki, E., Vermylen, J., and Hoylaerts, M.F. (2006).
The platelet ATP and ADP receptors. Curr. Pharm. Des. 12: 859–875.
Remijn, J.A., Ijsseldijk, M.J., Strunk, A.L. et al. (2007). Novel molecular
defect in the platelet ADP receptor P2Y12 of a patient with haemor-
rhagic diathesis. Clin. Chem. Lab. Med. 45: 187–189.
Other primary agonist receptor defects
Edelstein, L.C., Simon, L.M., Lindsay, C.R. et al. (2014). Common vari-
ants in the human platelet PAR4 thrombin receptor alter platelet
function and differ by race. Blood 124: 3450–3458.
Hirata, T., Kakizuka, A., Ushikubi, F. et al. (1994). Arg60 to Leu mutation
of the human thromboxane A2 receptor in a dominantly inherited
bleeding disorder. J. Clin. Investig. 94: 1662–1667.
Intracellular signaling pathway defects
Bellucci, S., Huisse, M.G., Boval, B. et al. (2005). Defective collagen-
induced platelet activation in two patients with malignant
haemopathies is related to a defect in the GPVI-coupled signalling
pathway. Thromb. Haemost. 93: 130–138.
Dunkley, S., Arthur, J.F., Evans, S. et al. (2007). A familial platelet func-
tion disorder associated with abnormal signalling through the glyco-
protein VI pathway. Br. J. Haematol. 137: 569–577.
Mao, G.F., Vaidyula, V.R., Kunapuli, S.P., and Rao, A.K. (2002). Lineage-
specific defect in gene expression in human platelet phospholipase
C-α2 deficiency. Blood 99: 905–911.
Patel, Y.M., Patel, K., Rahman, S. et al. (2003). Evidence for a role
for Gαi1 in mediating weak agonist-induced platelet aggregation in
human platelets: reduced Gαi1 expression and defective Gi signaling
in the platelets of a patient with a chronic bleeding disorder. Blood
101: 4828–4835.
Platelet secretion defects (storage pool
disease): gray platelet syndrome
Albers, C.A., Cvejic, A., Favier, R. et al. (2011). Exome sequencing iden-
tifies NBEAL2 as the causative gene for gray platelet syndrome. Nat.
Genet. 43: 735–737.
Deppermann, C., Cherpokova, D., Nurden, P. et al. (2013). Gray platelet
syndrome and defective thrombo-inflammation in Nbeal2-deficient
mice. J. Clin. Investig. 123: 3331–3342.
264 Molecular Hematology
Drouin, A., Favier, R., Masse, J.M. et al. (2001). Newly recognized cellu-
lar abnormalities in the gray platelet syndrome. Blood 98: 1382–1391.
Gunay-Aygun, M., Falik-Zaccai, T.C., Vilboux, T. et al. (2011). NBEAL2
is mutated in gray platelet syndrome and is required for biogenesis of
platelet α-granules. Nat. Genet. 43: 732–734.
Kahr, W.H., Hinckley, J., Li, L. et al. (2011). Mutations in NBEAL2,
encoding a BEACH protein, cause gray platelet syndrome. Nat. Genet.
43: 738–740.
Nurden, A.T. and Nurden, P. (2007). The gray platelet syndrome: clinical
spectrum of the disease. Blood Rev. 21: 21–36.
Nurden, P., Jandrot-Perrus, M., Combrie, R. et al. (2004). Severe defi-
ciency of glycoprotein VI in a patient with gray platelet syndrome.
Blood 104: 107–114.
Quebec platelet disorder
Diamandis, M., Veljkovic, D.K., Maurer-Spurej, E. et al. (2008). Quebec
platelet disorder: features, pathogenesis and treatment. Blood Coagul.
Fibrinolysis 19: 109–119.
Veljkovic, D.K., Rivard, G.E., Diamandis, M. et al. (2009). Increased
expression of urokinase plasminogen activator in Quebec platelet dis-
order is linked to megakaryocyte differentiation. Blood 113: 1535–
1542.
Hermansky–pudlak syndrome
Gautam, R., Novak, E.K., Tan, J. et al. (2006). Interaction of Hermansky–
Pudlak syndrome genes in the regulation of lysosome-related
organelles. Traffic 7: 779–792.
Huizing, M., Parkes, J.M., Helip-Wooley, A. et al. (2007). Platelet α-
granules in BLOC-2 and BLOC-3 subtypes of Hermansky–Pudlak
syndrome. Platelets 18: 150–157.
Spritz, R.A. (1998). Molecular genetics of the Hermansky–Pudlak and
Chediak–Higashi syndromes. Platelets 9: 21–29.
Walker, M., Payne, J., Wagner, B., and Vora, A. (2007). Hermansky–
Pudlak syndrome. Br. J. Haematol. 138: 671.
Chédiak–higashi syndrome
Kaplan, J., De Domenico, I., and Ward, D.M. (2008). Chediak–Higashi
syndrome. Curr. Opin. Hematol. 15: 22–29.
Nagle, D.L., Karim, M.A., Woolf, E.A. et al. (1996). Identification and
mutation analysis of the complete gene for Chediak–Higashi syn-
drome. Nat. Genet. 14: 307–311.
Westbroek, W., Adams, D., Huizing, M. et al. (2007). Cellular defects in
Chediak–Higashi syndrome correlate with the molecular genotype
and clinical phenotype. J. Investig. Dermatol. 127: 2674–2677.
Griscelli syndrome
Barral, D.C., Ramalho, J.S., Anders, R. et al. (2002). Functional redun-
dancy of Rab27 proteins and the pathogenesis of Griscelli syndrome.
J. Clin. Investig. 110: 247–257.
Novak, E.K., Gautam, R., Reddington, M. et al. (2002). The regulation
of platelet-dense granules by Rab27a in the ashen mouse, a model of
Hermansky–Pudlak and Griscelli syndromes, is granule-specific and
dependent on genetic background. Blood 100: 128–135.
Wiskott–aldrich syndrome
Imai, K., Nonoyama, S., and Ochs, H.D. (2003). WASP (Wiskott–
Aldrich syndrome protein) gene mutations and phenotype. Curr.
Opin. Allergy Clin. Immunol. 3: 427–436.
Lutskiy, M.I., Shcherbina, A., Bachli, E.T. et al. (2007). WASP localizes
to the membrane skeleton of platelets. Br. J. Haematol. 139: 98–105.
Ochs, H.D., Filipovich, A.H., Veys, P. et al. (2008). Wiskott–Aldrich syn-
drome: diagnosis, clinical and laboratory manifestations, and treat-
ment. Biol. Blood Marrow Transplant. 15 (1 Suppl): 84–90.
Sabri, S., Foudi, A., Boukour, S. et al. (2006). Deficiency in the Wiskott–
Aldrich protein induces premature proplatelet formation and platelet
production in the bone marrow compartment. Blood 108: 134–
140.
Procoagulant regulation defects
Solum, N.O. (1999). Procoagulant expression in platelets and defects
leading to clinical disorders. Arterioscler. Thromb. Vasc. Biol. 19:
2841–2846.
Weiss, H.J. (2009). Impaired platelet procoagulant mechanisms in
patients with bleeding disorders. Semin. Thromb. Hemost. 35: 233–
241.
Scott syndrome
Bettache, N., Gaffet, P., Allegre, N. et al. (1998). Impaired redistribu-
tion of aminophospholipids with distinctive cell shape change dur-
ing Ca2 + −induced activation of platelets from a patient with Scott
syndrome. Br. J. Haematol. 101: 50–58.
Castoldi, E., Collins, P.W., Williamson, P.L., and Bevers, E.M. (2011).
Compound heterozygosity for 2 novel TMEM16F mutations in a
patient with Scott syndrome. Blood 117: 4399–4400.
van Geffen, J.P., Swieringa, F., and Heemskerk, J.W. (2016). Platelets
and coagulation in thrombus formation: aberrations in the Scott syn-
drome. Thromb. Res. 141 (Suppl 2): S12–S16.
Kunzelmann, K., Nilius, B., Owsianik, G. et al. (2014). Molecular func-
tions of anoctamin 6 (TMEM16F): a chloride channel, cation chan-
nel, or phospholipid scramblase? Pflugers Arch. 466: 407–414.
Suzuki, J., Umeda, M., Sims, P.J., and Nagata, S. (2010). Calcium-
dependent phospholipid scrambling by TMEM16F. Nature 468: 834–
838.
Stormorken syndrome
Misceo, D., Holmgren, A., Louch, W.E. et al. (2014). A dominant STIM1
mutation causes Stormorken syndrome. Hum. Mutat. 35: 556–564.
Stormorken, H., Holmsen, H., Sund, R. et al. (1995). Studies on the
haemostatic defect in a complicated syndrome. An inverse Scott syn-
drome platelet membrane abnormality? Thromb. Haemost. 74: 1244–
1251.
Giant platelet syndromes and cytoskeletal
defects
Althaus, K. and Greinacher, A. (2009). MYH9-related platelet disorders.
Semin. Thromb. Hemost. 35: 189–203.

Chen, Z. and Shivdasani, R.A. (2009). Regulation of platelet biogene-
sis: insights from the May–Hegglin anomaly and other MYH9-related
disorders. J. Thromb. Haemost. 7: 272–276.
Leon, C., Eckly, A., Hechler, B. et al. (2007). Megakaryocyte-restricted
MYH9 inactivation dramatically affects hemostasis while preserving
platelet aggregation and secretion. Blood 110: 3183–3191.
Nurden, A.T., Federici, A.B., and Nurden, P. (2009). Altered megakary-
ocytopoiesis in von Willebrand type 2B disease. J. Thromb. Haemost.
7: 277–281.
Provaznikova, D., Geierova, V., Kumstyrova, T. et al. (2009). Clinical
manifestation and molecular genetic characterization of MYH9 dis-
orders. Platelets 26: 1–8.
Saito, H., Matsushita, T., Yamamoto, K. et al. (2005). Giant platelet syn-
drome. Hematology 10: 41–46.
Stächele, J., Bakchoul, T., Najm, J. et al. (2015). Congenital macrothrom-
bocytopenia associated with a combination of functional poly-
morphisms in the TUBB1 gene. Hamostaseologie 35 (Suppl 1):
S18–S21.
Toren, A., Rozenfeld-Granot, G., Heath, K.E. et al. (2003). MYH9 spec-
trum of autosomal-dominant giant platelet syndromes: unexpected
association with fibulin-1 variant-D inactivation. Am. J. Hematol. 74:
254–262.
White, J.G. (2007). Platelet pathology in carriers of the X-linked GATA-
1 macrothrombocytopenia. Platelets 18: 620–627.
Transcription factor defects
Breton-Gorius, J., Favier, R., Guichard, J. et al. (1995). A new congeni-
tal dysmegakaryopoietic thrombocytopenia (Paris–Trousseau) asso-
ciated with giant platelet α-granules and chromosome 11 deletion at
11q23. Blood 85: 1805–1814.
Escher, R., Wilson, P., Carmichael, C. et al. (2007). A pedigree with auto-
somal dominant thrombocytopenia, red cell macrocytosis, and an
Platelet disorders 265
occurrence of t(12 : 21) positive pre-B acute lymphoblastic leukemia.
Blood Cell Mol. Dis. 39: 107–114.
Freson, K., Devriendt, K., Matthijs, G. et al. (2001). Platelet characteris-
tics in patients with X-linked macrothrombocytopenia because of a
novel GATA1 mutation. Blood 98: 85–92.
Sun, L., Gorospe, J.R., Hoffman, E.P., and Rao, A.K. (2007). Decreased
platelet expression of myosin regulatory light chain polypep-
tide (MYL9) and other genes with platelet dysfunction and
CBFA2/RUNX1 mutation: insights from platelet expression profiling.
J. Thromb. Haemost. 5: 146–154.
White, J.G. (2007). Platelet pathology in carriers of the X-linked GATA-
1 macrothrombocytopenia. Platelets 18: 620–627.
Platelet protein polymorphisms and
tendency to thrombosis
Jones, C.I., Bray, S., Garner, S.F. et al. (2009). A functional genomics
approach reveals novel quantitative trait loci associated with platelet
signalling pathways. Blood 114: 1405–1416.
Martini, C.H., Doggen, C.J., Cavallini, C. et al. (2005). No effect of poly-
morphisms in prothrombotic genes on the risk of myocardial infarc-
tion in young adults without cardiovascular risk factors. J. Thromb.
Haemost. 3: 177–179.
Mega, J.L., Close, S.L., Wiviott, S.D. et al. (2009). Cytochrome P-450
polymorphisms and response to clopidogrel. N. Engl. J. Med. 360:
354–362.
Meisel, C., Lopez, J.A., and Stangl, K. (2004). Role of platelet glycopro-
tein polymorphisms in cardiovascular diseases. Naunyn Schmiede-
bergs Arch Pharmacol. 369: 38–54.
Ouwehand, W.H. (2007). Platelet genomics and the risk of atherothrom-
bosis. J. Thromb. Haemost. 5: 188–195.
Ouwehand, W.H. (2009). The discovery of genes implicated in myocar-
dial infarction. J. Thromb. Haemost. 7: 305–307.
Chapter 20 The molecular basis of blood
cell alloantigens
Cristina Navarrete1,2 , Louise Tilley3 , Winnie Chong4 & Colin J. Brown5,6
1
Histocompatibility and Immunogenetics Service Development Department, NHS Blood and Transplant, London, UK
2
Department of Immunology and Molecular Pathology, University College London, London, UK
3 International Blood Group Reference Laboratory, NHS Blood and Transplant, Bristol, UK
4
Histocompatibility and Immunogenetics, Service Development Laboratory, NHS Blood and Transplant, London, UK
5 Histocompatibility and Immunogenetics Laboratory, NHS Blood and Transplant, London, UK
6
Faculty of Life Sciences & Medicine, King’s College London, London, UK

Introduction, 267 Neonatal alloimmune thrombocytopenia, 277

Identification of HLA gene polymorphism, 268 Neonatal alloimmune neutropenia, 281
HLA antigens, 271 Conclusions, 283
Hemolytic disease of the fetus and newborn, 273 Further reading, 283

ited to one type of blood cell, destruction of cells in the

Introduction newborn by maternal blood cell alloantibodies may lead
to the development of immune-mediated anemia, throm-
The transfusion of blood and fetal–maternal hemorrhage
bocytopenia, or neutropenia of the newborn. In contrast,
during pregnancy have provided unique models for study-
HLA class I alloantibodies do not cause cytopenias in the
ing the immune response against a plethora of polymorphic
newborn, but may compromise the effectiveness of platelet
blood cell surface markers, including the alloantigens of the
transfusions, complicate organ transplantation, cause febrile
human leukocyte antigen (HLA) system. Many blood cell
non-hemolytic transfusion reactions, or, on rare occasions,
membrane determinants show allelic variation, which can
precipitate transfusion-related acute lung injury (TRALI)
elicit the formation of alloantibodies. In nearly all transfusion
and occasionally delay engraftment in hemopoietic stem cell
situations and pregnancies, the recipient’s immune system is
transplantation.
challenged by blood cells mismatched for multiple alloanti-
The formation of alloantibodies after an incompatible
gen systems. As the difference between self and non-self is
challenge in the form of blood transfusion is more the excep-
limited, alloantibodies are only formed by a subset of recipi-
tion than the rule. In contrast to our detailed understand-
ents. Red cell alloantibodies are detected in 1–1.5% of preg-
ing of the molecular basis of blood cell alloantigens, we
nant women and in 2–3% of transfused individuals, and can
remain relatively ignorant about the mechanism of non-
increase significantly in multitransfused patients. The HLA
responsiveness. We have learned from animal experiments
alloantigens are more immunogenic than those of the red
that restriction in the ability to mount an immune response
cells, and 15–25% of multiparous women and 30–40% of
is largely controlled by genes of the major histocompati-
patients on long-term prophylactic platelet transfusions are
bility complex (MHC) or HLA. However, the reason why,
positive for HLA class I antibodies.
for example, some 25% of RhD-negative individuals fail to
Alloantigens were initially defined as polymorphic mem-
mount an anti-D response on repeated challenge with RhD-
brane determinants identified by polyclonal alloantibodies
positive red cells remains elusive. An exception to this is our
in serum samples from alloimmunized patients or pregnant
detailed understanding of the immune response against the
women, but the molecular basis of most alloantigens has
HPA-1a alloantigen on platelets. There is a near-complete
now been resolved and it is possible to use DNA-based tech-
restriction on the ability to form HPA-1a antibodies by the
niques to further characterize their polymorphism. Alloanti-
HLA class II allele DRB3*01:01. However, except for this
gens can be categorized as those shared by all blood cells,
example, our ability to identify the genes controlling the risk
for example HLA class I, and those unique to one blood
of alloimmunization remains limited, and further research
cell type, such as Rh on red cells and human platelet anti-
is needed to identify the genetic basis of this variability in
gen (HPA) on platelets (Table 20.1). When expression is lim-
responsiveness.
This chapter reviews the recent developments in the
Molecular Hematology, 4th Edition. Edited by Drew Provan and John G. Gribben. molecular aspects of blood cell alloantigens and highlights
© 2020 John Wiley & Sons Ltd. Published 2020 by John Wiley & Sons Ltd. their impact on clinical management. Recognizing the wide

267
268 Molecular Hematology

Table 20.1 Antigen expression on peripheral blood cells

Antigens Erythrocytes Platelets Neutrophils B lymphocytes T lymphocytes Monocytes

A, B, H +++ + +/(+) − − − −
I +++ ++ ++ − − −
Rh* +++ − − − − −
K +++ − − − − −
HLA class I −/(+) +++ +++ +++ +++ +++
HLA class II − − −/+ + +† +++ −/+ + +† +++
GPIIb/IIIa − +++ (+)‡ − − −
GPIa/IIa − +++ − − ++ −
GPIb/IX/V − +++ − − − −
CD109 (+)/+ +† − − (−)/+ +† (+)
FcγRIIIB (CD16b) − − +++ − − −/+ + + §*%
CD177 − − + + +# − − −
CTL2 − ? +++ + +$ + +$ ?
CD11b − − ++ − − + +&
CD11a − − ++ − − ++

+, + +, + + + Level of antigen expression in increasing order. (+) Weak expression. ? Not known. * Non-glycosylated. † On activated cells.
‡ GPIIIa in association with an alternative α chain αv. § When differentiated to macrophages expressing FcγRIIIA. # Expressed on subpopulation
of neutrophils. & Also expressed on natural killer cells. $ B and T lymphocytes not separated. % Expressed on subpopulation of monocytes. GP,
glycoprotein; HLA, human leukocyte antigen

variety of clinical conditions in which the HLA alloantigens
play a role, we have placed the main emphasis on:
r antibody-mediated cytopenias in the newborn by maternal
blood cell–specific alloantibodies;
r the complication of HLA class I alloimmunization in
patients receiving prophylactic platelet transfusions. The
HLA antigens have also been used to introduce the molec-
ular techniques currently utilized to identify alleles of
genes.
Identification of HLA gene
polymorphism
The impact of molecular biological techniques on our ability
to scan and identify allelic variation of human genes is best
exemplified by the HLA system. For decades the enormous
diversity of the alloantigens of the HLA system has presented
both technical and clinical challenges. It is well documented
that the use of molecular techniques, including sequencing-
based high-resolution typing, is contributing to improved
outcome in transplant patients. Improved matching allows
graft maintenance at lower levels of immunosuppression,
which is of great importance with the emerging evidence
that long-term use of potent immunosuppressive drugs is not
without side effects. The following sections review a number
of current molecular techniques used to define alleles of the
HLA genes. The same techniques may be used to define allelic
variation in genes encoding other blood cell alloantigens.
Molecular typing techniques
Initially, the detection and characterization of the HLA
molecules and polymorphisms were carried out using sero-
logical and cellular techniques. With the development of gene
cloning and DNA-based molecular techniques, it is now pos-
sible to perform a detailed analysis of these molecules at
the nucleotide level. The result of this analysis has shown
the existence of shared nucleotide sequence motifs between
alleles of the same and/or different loci. Similarly, it has
been shown that there are certain locus-specific nucleotide
sequences in both the coding regions (exons) and the non-
coding regions (introns) of the various genes.
DNA sequencing of many HLA alleles has demonstrated
that most of the polymorphisms are located in exons 2 and 3
of HLA class I and exon 2 of HLA class II molecules. These
exons code for the distal membrane domains of the HLA
molecules, which form the peptide-binding groove (Fig-
ure 20.1). Based on this information, a number of techniques
have been developed to identify this polymorphism using the
polymerase chain reaction (PCR). These include HLA typ-
ing using PCR-SSOP (polymerase chain reaction–sequence-
specific oligonucleotide probes), PCR-SSP (polymerase chain
reaction–sequence-specific primers), sequencing-based typ-
ing (SBT), and next-generation sequencing (NGS).
PCR-SSOP
In this technique, the gene of interest is amplified by PCR
using generic primers complementary to highly conserved
 The molecular basis of blood cell alloantigens 269

α1 α2

Peptide-
binding
region S
S
N
β2–m α3
Immunoglobulin- S S
like region
S S

Transmembrane
region
P
Cytoplasmic P
P
region C
3'UT
Class IA

Regulatory Leader Exon Exon Exon Transmembrane and

(a) sequences sequence 2 3 4 cytoplasmic

α1 β1

Peptide-
binding
region NN S

α2 β2
Immunoglobulin- S S
like region
S S
Fig. 20.1 (a) Schematic presentation of human
Papain cleavage Papain cleavage
leukocyte antigen (HLA) class I. The sites sites
non-covalent association between the HLA class I
protein (with three immunoglobulin-like domains, Transmembrane
region
α1 , α2 , and α3 ) and β2 -microglobulin (β2 -m) is
shown. The three α domains are encoded by three Cytoplasmic
exons of the HLA class IA gene on chromosome 6. region C C 3'UT
(b) Schematic presentation of HLA class II. The
α and β chains of the HLA class II protein (each Class IIA
with two distinct immunoglobulin-like domains, α1 Regulatory Leader Exon 2 Exon 3 Transmembrane and
and α2 , and β1 and β2 ) are non-covalently sequences sequence cytoplasmic
associated. Both domains of each chain are
Class IIB
encoded by their respective exons of the α and β
class II genes on chromosome 6. (b) Exon 2 Exon 3

gene segments. The PCR products are then immobilized onto
support membranes (e.g. nylon membranes) and analyzed
by probing the membranes with labeled oligonucleotides
designed to anneal with polymorphic sequences present in
the HLA alleles. By scoring the probes that bind to specific
regions, it is possible to assign an HLA type.
A modification called the reverse SSOP technique is more
widely used, whereby the relevant PCR product is labeled
with biotin and, following the denaturation and neutraliza-
tion steps, the product is hybridized with oligonucleotide-
specific probes coupled to fluorescent beads. Following
the addition of r-phycoerythrin-conjugated streptavidin
270 Molecular Hematology

(SAPE) to the reaction, the fluorescence is then measured
using a flow cytometer-like instrument, the Luminex.
Reverse SSOP is particularly amenable to batch testing
large number of samples rather than single sample testing,
and was a popular technique used by stem cell registries.
PCR-SSP
This technique involves the use of sequence-specific primers
in the PCR. The amplified DNA product is detected by gel
electrophoresis and this allows the rapid identification of the
HLA alleles in individual samples, since the read-out of this
method is the presence or absence of an amplicon for which
a specific primer was used. PCR-SSP was first developed to
define the various HLA-DRB3 alleles (Figure 20.2). Although
this is a rapid technique, many PCR reactions have to be
set up for each sample, for example 24 reactions for low-
resolution DR typing. An obvious advantage of PCR-SSP is
that, as two sequences are detected in cis, ambiguities which
may arise from PCR-SSOP typing can be resolved. For PCR-
SSP typing, however, the target sequence of the alleles must
be known, and novel unknown sequences may not always be
detected. PCR-SSP is also the technique of choice for HPA
typing (see The Molecular Basis of Platelet-Specific Antigens or

Allele-specific PCR (PCR–SSP)

Complete complementarity of PCR primers

Amplification
A
PCR primer mismatch 3'-end

No amplification
B

A B

Control: non-polymorphic PCR

product
Fig. 20.2 (a) Polymerase chain
reaction–sequence-specific primers (PCR-SSP).
With allele-specific PCR, specificity is obtained
Allele-specific PCR product
during the PCR. A single nucleotide mismatch at
the 3′ -end of the allele-specific primer (B primer in
the figure) will prevent the polymerase from
(a) commencing DNA amplification. Therefore, no
amplification of template DNA will occur with the B
HLA–DR/DQ allele primer, whereas with the A allele primer a
PCR–SSP
product is obtained. Ethidium bromide is used to
reveal amplified DNA with ultraviolet light after
DNA gel electrophoresis. Allele-specific DNA is
HLA–DR1, DR13 obtained in the A reaction (lower band) but not in
DRB3 the B reaction. In both reactions a control PCR
product is generated by amplification of a segment
of the human growth hormone gene (upper two
bands). (b) PCR-SSP for DR and DQ alleles.
Results of PCR-SSP with template DNA from a
single donor. Each lane represents the result of gel
HLA–DQ5, DQ6 electrophoresis of a single PCR reaction with one of
the allele-specific DR or DQ primers. From the
pattern of positive results, a DR (upper panel) and
DQ (lower panel) type can be concluded, in this
(b) case DR1, DR13, DRB3, and DQ5, DQ6.

HPA). Using the PCR-SSP technique, it is possible to deter-
mine whether the detected sequences are in cis or in trans.
Sequencing-based typing
The principle of DNA sequencing is relatively straight-
forward. It involves the denaturation of the DNA to be
analyzed to provide a single-strand template; a sequenc-
ing primer is then added and the extension is performed
by the addition of polymerase in the presence of excess
nucleotides. The sequencing mixture is divided into four
tubes, each of which contains a specific dideoxyribonucleo-
side triphosphate (ddATP). When this is incorporated into
the DNA strand, elongation is interrupted, leading to chain
termination. In each reaction there is random incorpora-
tion of the chain terminators and therefore products of
all sizes are generated. The products of the four reactions
are then analyzed by electrophoresis in parallel lanes of a
polyacrylamide–urea gel, and the sequence is read by com-
bining the results of each lane using an automated DNA
sequencer.
Next-generation sequencing
NGS involves PCR amplification of the required region, gene,
or exon, and pooling of the resulting amplicons. Two main
approaches have been used in the application of NGS for the
definition of HLA alleles. The first involves the amplifica-
tion and sequencing of the exons containing the polymor-
phic positions of the required genes (exons 2 and 3 for HLA
class I, A, B, and C; and exon 2 for HLA class II DR, DQ,
and DP). The second approach involves the amplification and
sequencing of the whole gene of interest. The main advantage
of this latter approach is that it allows the detection of poly-
morphisms in all exons and introns of the relevant gene, thus
removing HLA typing ambiguities obtained with the exon-
based approach. The amplification step is followed by the
fragmentation of these amplicons and the ligation of adapters
and unique indexes or identifiers, a step also known as library
preparation. The fragmented and ligated products are then
clonally amplified and sequenced. NGS generates hundreds
of thousands of sequences, and sorting and aligning of these
sequences to the reference genome can be extremely chal-
lenging; to achieve this, complex analysis algorithms have
been developed.
HLA antigens
These are a group of highly polymorphic cell surface
molecules that play a central role in the induction and reg-
ulation of immune responses, and as such they are involved
in self/non-self-recognition, tolerance, rejection of allografts,
The molecular basis of blood cell alloantigens 271
and graft-versus-host disease. The genes coding for these
molecules form part of a complex genetic system called the
MHC, located on the short arm of chromosome 6. This
region spans a distance of approximately 4000 kb and is
divided into HLA class I, class II, and class III genes. Class
III includes a group of non-MHC genes coding for proteins
with various immunological functions, such as the comple-
ment factor 4 and tumor necrosis factor (TNF)-α.
The development of recombinant DNA technology has led
to increased understanding of the genetic complexity, struc-
ture, and function of the HLA genes and molecules.
HLA class I genes
The HLA class I genes have been classified according to their
structure and function as classical and non-classical, or class
Ib genes. The classical HLA class I genes, HLA-A, HLA-B,
and HLA-C, code for heterodimers formed by a heavy (α)
chain of approximately 43 kDa, non-covalently linked to the
β2 -microglobulin light chain of 12 kDa. The latter is coded
for by a gene located outside the HLA region on chromo-
some 15. The extracellular portion of the α chain has three
domains (α1 , α2 , and α3 ) encoded by exons 2, 3, and 4,
respectively. Each domain is approximately 90 amino acids
in length. The transmembrane and cytoplasmic domains
are encoded by exons 5, 6, and 7, respectively. The β2 -
microglobulin, which confers stability on the molecule, is
non-covalently linked to the α3 domain (see Figure 20.1a).
The α1 and α2 domains are the most polymorphic regions
of these molecules and form a groove consisting of two α
helices, with an antiparallel-running β-pleated sheet forming
the floor of the groove. This groove, which is approximately
2.5 nm long and 1 nm wide, can accommodate a variety of
antigen-derived peptides of about 8–10 amino acids to be
presented to T cells.
In addition to the classical HLA class I genes, the
non-classical HLA class I genes are also located in this
region. They include HLA-E, HLA-F, and HLA-G and their
exon/intron organization is similar to that of the classical
class I genes, but they have a more restricted polymorphism.
The HLA class I genes are expressed on most tissues and
blood cells, including T and B lymphocytes and platelets
(Table 20.1). The non-classical class I genes HLA-E and HLA-
F are expressed on most tissues tested so far, whereas HLA-G
has to date only been detected on trophoblasts and mono-
cytes.
Two MHC class I chain-related genes (MIC-A and MIC-B),
located centromeric to HLA-B, have been described. Unlike
the classical and non-classical HLA class I, MIC genes do not
require binding to the β2 -microglobulin or peptide in order
to be expressed on the cell surface. So far, MIC expression has
been detected on freshly isolated endothelial cells, fibroblasts,
keratinocytes, and monocytes. These genes have also been
272 Molecular Hematology
found to be expressed on intestinal epithelial cells as a result
of stress, and on a variety of tumors of epithelial origin.
HFE is another non-classical class I gene, located 4 Mb
telomeric of HLA-A. This gene has been found to be asso-
ciated with the development of hereditary hemochromatosis
(HH). A single point mutation, 845A, replacing cysteine
with tyrosine at position 282 (C282Y) is found in over
90% of HH patients in the UK. The other two mutations,
replacing histidine by aspartate at amino acid position 63
(H63D) and serine by cysteine at amino acid 65 (S65C),
appear to be associated with milder forms of HH. This
gene does not have a direct immune function, as it has lost
the ability to bind antigenic peptides due to closure of the
antigen-binding groove. However, since HFE can bind to the
transferrin receptor, and in this way regulate iron uptake and
availability, it is possible that HFE may indirectly be involved
in the regulation of immune responses.
HLA class II genes
The HLA class II genes DR, DQ, and DP are all located within
the HLA class II region. There is one non-polymorphic DRA
and nine highly polymorphic DRB genes, of which DRB2, B6,
and B9 are pseudogenes. These genes code for heterodimers
formed by an α and a β chain, both encoded by genes within
the MHC. The extracellular portion of these molecules has
two domains (α1 and α2 and β1 and β2 ) encoded by exon
2 and 3 of each gene, respectively. The α1 and β1 domains
form the peptide-binding groove (see Figure 20.1b).
The number of DRB genes expressed in each haplotype
varies depending on the DRB1 allele expressed; for example,
HLA DRB1*01, DRB1*01:03, DRB1*08, and DRB1*10 hap-
lotypes only express the DRB1 gene. DRB1*15 and DRB1*16
haplotypes additionally express the DRB5 gene, which codes
for the DR51 product. The HLA DRB1*03:01 (DR17),
DRB1*03:02 (DR18), DRB1*11, DRB1*12, DRB1*13, and
DRB1*14 haplotypes also express the DRB3 genes, which
code for the DR52 specificity, while the HLA DRB1*04,
DRB1*07, and DRB1*09 alleles also express the DRB4 gene,
which encodes the DR53 product. There are a few exceptions
to this gene distribution; for example, a DRB5 gene has been
found linked to a DRB1*01 haplotype, and null DRB5 and
DRB4 genes have been identified.
In contrast, there are two DQA and three DQB genes; of
these, only A1 and B1 are expressed and both are polymor-
phic. Similarly, there are two DPA and two DPB genes, of
which only DPA1 and DPB1 are expressed and are polymor-
phic. More than 23 000 HLA alleles had been named by June
2019 (http://hla.alleles.org/nomenclature/index.html).
Other HLA-related genes located within the MHC class
II region include LMP2, LMP7, TAP1, and TAP2, which
are involved in the transport and processing of peptides
presented by class I molecules, and the HLA DMA, DMB,
DOA, and DOB genes, which participate in the loading of
peptides in the HLA class II molecules.
The HLA class II genes (DR, DQ, and DP) are consti-
tutively expressed on B lymphocytes, monocytes, and den-
dritic cells, and on activated T lymphocytes and granulocytes
(Table 20.1). HLA class II expression can also be induced
on non-hematopoietic cells such as fibroblasts and endothe-
lial cells, as the result of activation or by the effect of cer-
tain inflammatory cytokines, such as interferon (IFN)-γ and
TNF-α.
Function
The HLA molecules play a pivotal role in the induction and
regulation of the immune response. Both the phenomenon of
MHC restriction and the development of tolerance, learned
as T cells pass through the thymus, result in the selection of
a T-cell repertoire that will form the basis of an individual’s
capacity to respond to antigens. HLA class I molecules are
primarily but not exclusively involved in the presentation of
endogenous antigens, such as viral peptides, to CD8+ cyto-
toxic T cells, whereas HLA class II molecules present primar-
ily, but not exclusively, exogenous, such as bacterial, antigenic
peptides to CD4+ -helper T cells. These cells, once activated,
can initiate and regulate a variety of processes leading to the
maturation and differentiation of cellular and humoral effec-
tor cells, including the secretion of cytokines. The presenta-
tion of antigenic peptides is a highly regulated process and
requires fine interaction between the antigenic peptide, the
antigen-binding groove of the HLA molecules, and the T-
cell receptor. Allelic variation of the HLA molecules can pro-
foundly affect the ability to present certain peptides because
of the presence or absence of critical contact residues in the
peptide-binding groove.
HLA molecules on donor cells loaded with donor-derived
peptides can also be recognized directly by T cells of the
host by a mechanism called allorecognition. Two pathways
of allorecognition, direct and indirect, have been identified,
both of which lead to the strong alloimmunization seen in
patients receiving blood transfusions or a solid organ or bone
marrow/stem cell transplantation.
More recently, it has been shown that both classical and
non-classical HLA class I molecules interact with two func-
tionally distinct types of receptors, inhibitory and activat-
ing, present on natural killer (NK) cells. These receptors
belong to two families, the immunoglobulin superfamily, also
called killer immunoglobulin receptors (KIRs), and the C-
type lectin superfamily CD94, which can covalently assemble
with several members of the NKG2 family. The KIRs inter-
act with products of the HLA-A, -B, -C, and -G loci, whereas
CD94–NKG2 recognize the non-classical HLA-E molecule

presenting peptides derived from several HLA class I, A,
B, or C alleles and from HLA-G. MIC-A and MIC-B gene
products are recognized by receptors present on both NK and
γδ T cells.
Thus, HLA molecules have become increasingly relevant
in a variety of clinical situations, such as susceptibility to cer-
tain autoimmune and infectious diseases and in solid organ
and stem cell transplantation and blood cell alloimmuniza-
tion. With regard to the latter, two examples will be discussed:
refractoriness for prophylactic platelet transfusion by HLA
alloimmunization, and HLA class II restriction of the forma-
tion of anti-HPA-1a antibodies (see later).
Prophylactic platelet transfusions
Prophylactic platelet transfusions are essential for preventing
bleeding during intensive chemotherapy or other myeloab-
lative therapies. Increments in the platelet count after the
infusion of an adult dose of donor platelets (>250 × 109 l−1 )
are frequently disappointing. Non-immune factors, such
as splenomegaly, bleeding, sepsis, fever, and certain drugs
(e.g. amphotericin), can compromise the beneficial effect
of donor platelet infusions. In 10–20% of patients the
problem of poor increments is further compounded by
antibody-mediated destruction of donor platelets. Despite
this, the clinical definition of refractoriness remains much
disputed; clinically, the picture of an increased frequency of
platelet transfusions to maintain satisfactory platelet counts
and effective hemostasis requires further laboratory investi-
gations for HLA class I and for HPA alloantibodies, platelet
autoantibodies, or high-titer anti-A or anti-B antibodies.
The ability to type donors and patients for HLA class I A
and B and HPA by molecular techniques using genomic DNA
(see Molecular Typing Techniques; The Molecular Basis of
Platelet-Specific Antigens or HPA) has resulted in more accu-
rate matching of donors with patients, with improved platelet
recovery. The algorithm currently used for the management
of the alloimmunized patient with poor increments is shown
in Figure 20.3. Traditionally, HLA matching for the provi-
sion of compatible platelets for patients who have become
immunologically refractory to random platelet transfusion
has been based on the serological definition of these antigens
at the specificity level. More recently, a new approach based
on the identification of conformational epitopes present in
each HLA allele has been described. According to this strat-
egy, each HLA antigen is converted into a string of potentially
immunogenic epitopes, which are represented by amino acid
triplets (called eplets) on exposed parts of the HLA chains
accessible to alloantibodies. These eplet epitopes are amino
acid residues located within small clusters (with diameters
of about 0.3–0.35 nm) around a non-self-residue and are
formed by amino acids in linear sequences as described in the
The molecular basis of blood cell alloantigens 273
triplet model, but also include amino acids from discontinu-
ous regions of the sequence brought together by the folding
of the molecule. This is therefore considered to be a more
accurate representation of the epitope than that derived from
the triplet HLAMatchmaker model developed by Duques-
noy http://www.epitopes.net/. Consequently, it is possible to
determine the number of epitopes which are either shared
or differ between the donor and the recipient. The algorithm
also performs intra- and interlocus comparisons of poly-
morphic eplets to determine the spectrum of non-shared
eplets between HLA antigens of the donor and the patient.
Hemolytic disease of the fetus and
newborn
Rh system
The Rh system is the most complex red cell blood group
system, and the most clinically important after ABO. While
ABO antigens are carbohydrate in nature, with a wide tissue
distribution, the Rh antigens are protein based and specific
to red cells. Rh antigens are carried on two homologous
non-glycosylated proteins: RhD, carrying the D antigen, and
RhCE, carrying the C, c, E, and e antigens, encoded by RHD
and RHCE genes, respectively.
RH genes
The RHD and RHCE genes are located on chromosome 1, and
are tightly linked but have opposite orientations. Each gene
comprises 10 exons and each encodes a 417 amino acid non-
glycosylated protein (although the N-terminal methionine is
cleaved from the mature protein). The RHD and RHCE genes
are highly homologous (93.8% over all introns and coding
regions) and the encoded proteins differ only by between
32 and 35 amino acids (Table 20.2). The RhD and RhCE
proteins are both highly hydrophobic and are predicted to
span the red cell membrane 12 times, with 6 extracellular
loops (Figure 20.4). The function of the Rh proteins remains
unknown.
In D-negative individuals (around 15% of the Caucasian
population) the entire RhD protein is missing from the red
cell membrane, usually arising from homozygosity for a com-
plete deletion of the RHD gene. The pairs of antithetical anti-
gens C/c and E/e are encoded by polymorphisms within the
RHCE gene. RhC and Rhc proteins differ by four amino acids,
of which Ser103Pro, located on the second extracellular loop,
appears to be the most critical. RhE/e antigens are defined
by a single amino acid substitution, Pro226Ala, in the fourth
extracellular loop of RhCE (Table 20.2 and Figure 20.4).
274 Molecular Hematology

Patients likely to receive multiple platelet transfusion

Assess transfusion response

Poor response to random donor platelets on 2 or more occasions

HLA type
and test for
HLA-specific
antibodies

HLA antibody
test result

Ab positive
Provide HLA- Ab negative
selected platelets

Good response to Poor response to

Factors associated with non-
HLA-selected HLA-selected
immune platelet destruction
platelets platelets

Provide ABO
Continue Absent Present
compatible, “A”
transfusing HLA-
grade matches if
selected platelets
possible
Consider trial of Treat cause
HLA-selected Consider further platelet
Test for HLA platelets transfusion based on clinical
Test for HPA-
antibodies at status of the patient, e.g.
specific antibodies
regular intervals increase dose of platelets or
discontinue prophylactic
Poor response Good response transfusions

HPA antibody
test result Continue
transfusing HLA-
selected platelets
Positive Negative

Provide HLA- and Consider

HPA-selected 1. Non-immune
platelets consumption
2. ABO antibodies
Fig. 20.3 Platelet transfusions in alloimmunized patients. An algorithm outlining the decision process for the management of alloimmunized
patients refractory for random donor platelets. After confirmation of refractoriness for random donor platelets, patients are screened for human
leukocyte antigen (HLA) class I alloantibodies and, if positive, HLA class I-matched platelets are transfused. In 20–30% of patients, increments with
HLA class I-matched platelets are poor, and screening for human platelet antigen (HPA) antibodies should follow. Also, the possible presence of
potent anti-A or anti-B should be excluded, since platelets do carry ABO blood group antigens. If there are no detectable HLA class I antibodies, a
trial of HLA-matched platelets and screening for HPA antibodies should be considered (right arm of algorithm).
 The molecular basis of blood cell alloantigens 275

Table 20.2 Amino acid sequence of the RhD and RhCE proteins

RhCE (C) MSSKYPRSVR RCLPLCALTL EAALILLFYF FTHYDASLED QKGLVASYQV GQDLTVMAAI

RhCE (c) ---------- -----W---- ---------- ---------- ---------- ---------L
RhD ---------- -----W---- ---------- ---------- ---------- ---------I
RhCE (C) GLGFLTSSFR RHSWSSVAFN LFMLALGVQW AILLDGFLSQ FPSGKVVITL FSIRLATMSA
RhCE (c) -------N-- ---------- ---------- ---------- --P------- ----------
RhD -------S-- ---------- ---------- ---------- --S------- ----------
RhCE (C) MSVLISAGAV LGKVNLAQLV VMVLVEVTAL GTLRMVISNI FNTDYHMNLR HFYVFAAYFG
RhD L-----VD-- ---------- ---------- -N-------- --------MM -I--------
RhCE (E) LTVAWCLPKP LPKGTEDNDQ RATIPSLSAM LGALFLWMFW PSVNSPLLRS PIQRKNAMFN
RhCE (e) ---------- ---------- ---------- ---------- -----A---- ----------
RhD -S-------- --E----K-- T--------- ---------- --F--A---- --E----V--
RhCE TYYALAVSVV TAISGSSLAH PQRKISMTYV HSAVLAGGVA VGTSCHLIPS PWLAMVLGLV
RhD ----V----- ---------- --G---K--- ---------- ---------- ----------
RhCE AGLISIGGAK CLPVCCNRVL GIHHISVMHS IFSLLGLLGE ITYIVLLVLH TVWNGNGMIG
RhD -----V---- Y--G------ --P-S-I-GY N--------- -I-------D --GA------
RhCE FQVLLSIGEL SLAIVIALTS GLLTGLLLNL KIWKAPHVAK YFDDQVFWKF PHLAVGF
RhD ---------- ---------- ---------- -------E-- ---------- -------

Amino acid sequences (single letter code) showing differences between RhD and RhCE proteins, in addition to C/c (boxed) and E/e (shaded)
polymorphisms. Sequence identity is indicated by dashes.

S103P (C/c) P226A (E/e)

NH2
49 112
162 358

Rh CcEe 211
protein 384
267
313

408
49 112
162 358

Rh D 211
protein 384
267
313 408 COOH
417
NH2

Rh D
mRNA 1 2 3 4 5 6 7 8 9 10
4.5 1 2 3 4 5 6
Thr32 Ser99 Phe161 Ser230 Ile288 Ala354

–4.5
Fig. 20.4 RhD and RhCE proteins. Schematic representation of the topography of the RhCE (upper figure) and RhD (lower figure) proteins. Both
30 kDa non-glycosylated proteins traverse the red cell membrane 12 times. The difference between RhD and RhCE is defined by 36 of the 417
amino acids (colored circles indicate RhD-specific residues; shaded circles indicate C/c and E/e polymorphisms). The rectangular boxes (1–10)
indicate the positions of exon–exon boundaries in the RhD mRNA, and a hydropathy plot of the RhD protein is shown at the bottom.
276 Molecular Hematology
Immunogenicity of RhD and prevention of
immunization
The complete absence of the RhD protein in D-negative
individuals explains its high immunogenicity, and anti-D
can cause severe hemolytic transfusion reactions (HTRs) and
hemolytic disease of the fetus and newborn (HDFN). Pro-
duction of alloanti-D, stimulated by a D-positive fetus carried
by a D-negative mother, may result in severe HDFN, associ-
ated with hydrops, kernicterus, and possible fetal death. It has
been shown that this maternal anti-D production can be pre-
vented by the administration of IgG anti-D. The introduction
of anti-D prophylaxis for D-negative women after delivery,
or following sensitizing events during pregnancy, resulted in
a steep decline in HDFN-related mortality. Before 1969, 46
deaths per 100 000 births were attributed to RhD alloimmu-
nization, falling to 1.6/100 000 births in 1990. More recently,
the introduction of routine antenatal anti-D prophylaxis
(RAADP) for all D-negative pregnant women in the UK
has further reduced alloimmunization rates.
Knowledge of the molecular basis of blood group anti-
gens, together with the discovery of cell-free fetal DNA in
the maternal circulation, has allowed the development of
non-invasive fetal blood group genotyping from maternal
blood. Determination of fetal RHD genotype in pregnancies
at risk of HDFN allows appropriate monitoring and clinical
intervention to be directed at pregnancies where the fetus is
predicted to be D-positive. In the UK, approximately 38% of
D-negative pregnant women will carry a D-negative fetus and
are therefore not at risk of RhD alloimmunization. Recently,
large-scale non-invasive fetal RHD screening has been intro-
duced into the UK for all non-sensitized D-negative pregnant
women, to determine the individual requirement for anti-D
prophylaxis. Targeting of RAADP to those women confirmed
to be carrying a D-positive fetus prevents unnecessary expo-
sure to anti-D in the remainder, reducing the use of anti-D
and resulting in significant cost savings.
In addition to anti-D, antibodies to other Rh antigens are
also clinically significant, and associated with both HTR and
HDFN. Anti-c is the most clinically important after anti-D
and has been associated with severe HDFN, whereas anti-C,
anti-E, and anti-e have rarely been implicated in severe
HDFN. Non-invasive fetal genotyping to predict RhC/c and
E/e antigen status is useful in the management of at-risk preg-
nancies, although no prophylaxis is available.
Kell
The K antigen of the Kell blood group system is the most clin-
ically important red cell antigen outside of the ABO and Rh
systems. The antithetical antigens K and k are carried on Kell
glycoprotein (CD238), a 93 kDa type II transmembrane gly-
coprotein. The KEL gene consists of 19 exons, spanning about
21.5 Kb, encoding a 732 amino acid protein homologous to
a family of zinc-binding endopeptidases, including CD10. K
and k antigens differ by a single amino acid in the Kell gly-
coprotein, Met193Thr, encoded by a single nucleotide sub-
stitution in exon 6 of KEL. It is of interest to note that despite
this apparently minor difference between the two forms, K is
one of the most immunogenic blood group antigens. This is
possibly explained by the presence of an N-glycosylation site
at Asn191 in k, which is absent in the K protein.
Immunogenicity of K and the mechanism
of fetal anemia
Anti-K can cause both severe HTR and HDFN, and for
this reason it is recommended that girls and women of
childbearing age be transfused only with K-negative red
cells. K antigen is present on red cells of approximately 9% of
white people and 2% of black people. In HDFN caused by K
alloimmunization, fetal anemia appears to result mainly from
antibody-mediated inhibition of erythropoiesis. This sug-
gests a possible role for the Kell glycoprotein in the regulation
and growth of erythroid progenitor cells. It is more difficult
to predict the severity of anti-K HDFN than that caused by
anti-D, as there is little correlation between anti-K levels
and disease severity. Fetal genotyping to predict Kell antigen
status in at-risk pregnancies allows close monitoring, and
treatment where necessary, of those where the fetus is shown
to be K-positive.
Other red cell antigens and molecular
typing
The molecular bases of the majority of blood group alloanti-
gens have now been elucidated, and the genes encoding them
cloned and sequenced. This allows the use of molecular typ-
ing for the prediction of blood group phenotypes, and such
genotyping is routinely used for the most clinically relevant
blood group systems (Rh, Kell, Kidd, Duffy, and MNS).
Molecular typing is of particular use when serological typing
is not possible, such as for multitransfused patients (e.g.
those with hemoglobinopathies) and in prediction of fetal
blood group phenotype by non-invasive methods. Accurate
interpretation of blood group genotype into predicted phe-
notype requires in-depth knowledge of the genes involved
and their many allelic variants, which may be of particular
relevance in certain ethnic groups. Homology between RHD
and RHCE genes in the Rh system (also relevant to the
homologous genes of the MNS system) and a high degree of
polymorphism in these genes lead to challenges in designing
and correctly interpreting genotyping tests. However, geno-
typing offers a highly accurate method for the prediction of
blood group phenotype, and is invaluable in situations where
serological-based typing is not possible. Fetal genotyping for
clinically significant blood group antigens allows improved

clinical care of alloimmunized women and pregnancies at
risk of HDFN.
Pathology of HDFN
Maternal alloantibodies can be formed against a blood group
alloantigen present on fetal red cells but absent from mater-
nal cells. The main antigens implicated in HDFN are RhD,
K, and Rhc. Red cell alloantibodies of the IgG class (mostly
IgG1 and IgG3) can cross the placenta, bind to fetal red cells,
and shorten their survival, giving rise to hyperbilirubinemia
and anemia. In K immunization, hyperbilirubinemia is a less
reliable parameter for the prediction of the severity of fetal
anemia, and these pregnancies can be complicated by early
fetal loss. In England and Wales, approximately 500 fetuses
develop HDFN annually, with 25–30 perinatal deaths, and
an estimated further 20 spontaneous abortions prior to
28 weeks, attributed to HDFN.
Monitoring and treatment of HDFN
Screening for clinically significant red cell alloantibodies in
pregnant women is the standard of care in most European
countries. Of all pregnant women, 1–2% screen positive
for red cell alloantibodies, although in the vast majority
of cases the antibodies are not considered to be of clinical
significance. Determination of antibody specificity, and mea-
surement of antibody levels, allows assessment of the risk of
fetal anemia in order to identify cases of possible severe dis-
ease early in pregnancy, reducing the associated morbidity or
mortality. Once a fetus is assessed to be at moderate to severe
risk of developing fetal anemia, regular ultrasound exami-
nation and serial Doppler measurement of middle cerebral
artery peak systolic velocity (MCA PSV) are required.
Fetal anemia can be treated antenatally by intrauterine
transfusion (IUT), significantly reducing perinatal mortality,
with reported fetal survival rates of greater than 90% in IUT-
treated cases. In neonates, phototherapy has been shown
to effectively reduce the need for exchange transfusion in
the treatment of hyperbilirubinemia, in order to prevent
kernicterus. However, severe neonatal hemolytic anemia and
hyperbilirubinemia may be treated by exchange transfusion
with compatible donor red cells, correcting anemia and
removing both bilirubin and maternal alloantibodies from
the circulation.
Prevention of HDFN
Before the introduction of anti-D prophylaxis, most HDFN
cases were caused by RhD immunization. The routine RhD
testing of all pregnant women, combined with RAADP for
D-negative pregnant women, has been extremely successful
in lowering HDFN-associated morbidity and mortality. This
The molecular basis of blood cell alloantigens 277
can be facilitated by the use of non-invasive fetal geno-
typing to identify antigen-positive fetuses being carried by
antigen-negative mothers. Previous blood transfusion is the
most important risk factor for non-D immunization, partic-
ularly in the case of K immunization. For this reason, girls
and women of childbearing age are transfused only with
K-negative, RhD-matched blood. Further matching of Rh
antigens (in particular c and E) in this patient group may have
the potential to reduce alloimmunization, and consequently
the risk of subsequent HDFN.
Neonatal alloimmune
thrombocytopenia
HPA alloantigen systems
The HPA alloantigens are expressed on the surface of
platelets, but are not truly platelet-specific since some are also
found on the surface of other blood cell types (see Table 20.1).
Alloimmunization against HPA is associated with three clin-
ical syndromes:
1 neonatal/fetal alloimmune thrombocytopenia (NAITP);
2 post-transfusion purpura;
3 refractoriness for platelet transfusions.
NAITP was first described by van Loghem in 1959 and was
initially thought to be a rare disorder. Prospective screening
studies in pregnant (Caucasian) women have shown that 1 in
1100 neonates have severe thrombocytopenia (<50 × 109 l−1 )
due to maternal anti-HPA-1a, confirming the notion that
the most frequent cause of severe thrombocytopenia in the
term newborn is maternal alloantibodies against a fetal HPA
alloantigen.
This serious clinical condition is caused by the destruction
of fetal/neonatal platelets by maternal HPA alloantibodies of
the IgG class. Cerebral bleeds in the perinatal period are the
most concerning complication, which either occur in utero
or during delivery. In cases of severe thrombocytopenia
(<20 × 109 l−1 ), there remains a small but definite risk of
this serious complication in the first days of life, warranting
treatment. For proper clinical management, the cause of
severe thrombocytopenia in an otherwise healthy term
neonate should be determined with urgency, and prompt
correction of a count less than 20 × 109 l−1 by platelet
transfusion is of the utmost importance. This should precede
the outcome of platelet antibody investigation, which can be
a time-consuming process.
The molecular basis of platelet-specific
antigens or HPA
To date, 35 HPAs have been defined by immune sera, which
are grouped into 29 platelet-specific alloantigen systems. The
278 Molecular Hematology

Table 20.3 Platelet-specific alloantigen systems

System Antigen Alternative names Glycoprotein Nucleotide change Amino acid change

HPA-1 HPA-1a Zwa , PlA1 GPIIIa 176T Leu33

HPA-1b Zwb , PlA2 176C Pro33
HPA-2 HPA-2a Kob GPIbα 482C Thr145
HPA-2b Koa , Siba 482C Met145
HPA-3 HPA-3a Baka , Leka GPIIb 2621T Ile843
HPA-3b Bakb 2621G Ser843
HPA-4 HPA-4a Yukb , Pena GPIIIa 506G Arg143
HPA-4b Yuka , Penb 506A Gln143
HPA-5 HPA-5a Brb , Zavb GPIa 1600G Glu505
HPA-5b Bra , Zava 1600A Lys505
HPA-6bw Caa , Tua GPIIIa 1544G>A Arg>Gln489
HPA-7bw Moa GPIIIa 1297C>G Pro>Ala407
HPA-8bw Sra GPIIIa 1984T>C Arg>Cys636
HPA-9bw Maxa GPIIb 2602G>A Val>Met837
HPA-10bw Laa GPIIIa 263G>A Arg>Glu62
HPA-11bw Groa GPIIIa 1976G>A Arg>His633
HPA-12bw Iya GPIbβ 119G>A Gly>Glu15
HPA-13bw Sita GPIa 2483C>T Thr>Met799
HPA-14bw Oea GPIIIa 1109-1911delAAG Lys611del
HPA-15 HPA-15a Govb CD109 2108C Ser682
HPA-15b Gova 2108A Tyr682
HPA-16bw Duva GPIIIa 497C>T Thr>Ile140
HPA-17bw Vaa GPIIb/IIIa 662C>T Thr>Met195
HPA-18bw Caba GPIa 2235G>T Gln>His716
HPA-19bw Sta GPIIIa 487A>C Lys>Gln137
HPA-20bw Kno GPIIb 1947C>T Thr>Met619
HPA-21bw Nos GPIIIa 1960G>A Glu>Lys628
HPA-22bw Sey GPIIb 584A>C Lys>Thr164
HPA-23bw Hug GPIIIa 1942C>T Arg>Trp622
HPA-24bw Cab2a+ GPIIb 1508G>A Ser>Asn472
HPA-25bw Swia GPIa 3347C>T Thr>Met1087
HPA-26bw Seca GPIIIa 1818G>T Lys>Asn580
HPA-27bw Cab3a+ GPIIb 2614C>A Leu>Met841
HPA-28bw War GPIIb 2311G>T Val>Leu740
HPA-29bw Khab GPIIIa 98C>T Thr>Met7

HPA, human platelet antigen.

molecular basis of these have been determined and are all
mapped to certain platelet membrane glycoproteins (GPs).
Twelve of the most clinically relevant HPAs are grouped into
six biallelic systems. Of the 29 HPA systems, 22 are on the
integrin heterodimer αIIbβ3 or glycoprotein (GP)IIb/IIIa
(Table 20.3 and Figure 20.5); of the remaining seven, two
are on GPIb–IX–V, four on the integrin α2β1 or GPIa/IIa,
and one on CD109. The difference between the two alle-
les is a single-nucleotide substitution in the gene encod-
ing the relevant glycoprotein, with the exception of HPA-
14bw (Table 20.3). Amplification of genomic DNA by PCR-
SSP (Figure 20.6) and SBT can be used to type donors
and patients, even when the latter are thrombocytopenic.
High-throughput donor HPA typing can also be performed
using the 5′ -nuclease (TaqMan®) assay.
Immunogenicity and immune
response restriction
As the difference between self and non-self is defined by a
single amino acid substitution, the immunogenicity of the
HPA alloantigens is relatively low when compared with that
of some of the other blood cell alloantigens (e.g. RhD and
HLA class I). The two most clinically relevant HPA antigens
 The molecular basis of blood cell alloantigens 279

GPIIa GPIIa GPVI GPIIb GPIIa CD109 GPIbα GPV

α2 β2 αIIB β3 HPA-19w
K137Q HPA-2
T145M
HPA-5 HPA-16w
E505K T140l
HPA-17w
HPA-18w T195M
TQ716H HPA-4
HPA-13w R143Q
HPA-10w HPA-7w
T799M R62Q HPA-15
P407A S703Y
HPA-20w HPA-1
1 T619M 1 GPIX
L33P
HPA-9w HPA-6w
2 V837M 2
R489Q
3 HPA-3 3 HPA-8w
18435 s R636C GPIbβ
4 4 s
s HPA-11w HPA-12w s
βTD βTD R633H G15Q
HPA-14w
Del K611
HPA-21w
E628K

Talin
Fig. 20.5 Schematic presentation of the platelet glycoproteins. Schematic presentation of the platelet glycoprotein from which the human
platelet antigens (HPAs) can be found. The positions of the amino acid substitutions arising from allelic variation of the various glycoprotein genes
are marked by black arrows and the name of the HPA system is noted. Amino acids are described with one letter abbreviations.
Source: Adapted from Lucas G. (2009). Practical Transfusion Medicine, 3rd edn. Chichester: Wiley-Blackwell. Reproduced by permission of John
Wiley & Sons.

are HPA-1a and HPA-5b on the β3 and α2 integrins, namely
those on platelet glycoproteins GPIIIa and GPIa, respectively.
Alloantibodies against other HPA alloantigens are observed
infrequently in pregnancy, but do occur, albeit at a low fre-
quency, in hemato-oncological patients and other patients on
long-term prophylactic platelet transfusion.
The HPA-1a and -1b alloantigens are based on a C → G
mutation in the GPIIIa gene, resulting in a Leu33Pro
mutation. Why the immunogenicity of HPA-1a (Leu33) is
1a 1b 2a 2b 3a 3b 4a 4b 5a 5b
Fig. 20.6 Polymerase chain reaction–sequence-specific primers
(PCR-SSP) for human platelet antigen (HPA) alleles. Results of
agarose gel electrophoresis of PCR products obtained by amplification
of segments of the GPIIIa, GPIb𝛼, GPIIb, and GPIa genes using
allele-specific primers for the “a” and “b” alleles of the HPA-1, HPA-2,
HPA-3, HPA-4, and HPA-5 systems. The results for this donor are
HPA-1b1b, 2a2a, 3a3b, 4a4a, and 5a5a. Products of the allele-specific
amplification are the lower bands. In all reactions, a set of control
primers has been included to produce an amplicon (upper band)
derived from the human growth hormone gene.
magnitudes higher than that of its antithetical antigen HPA-
1b (Pro33) was initially not well understood. In the early
1980s, it was discovered that the formation of anti-HPA-1a in
pregnancy was positively associated with the HLA haplotype
A*01-B*08-DRB1*03. Further studies revealed that nearly
all antibody formers were positive for the *01:01 allele of the
DRB3 gene (DRB3*01:01 or DR52a; see HLA Class II Genes;
Figure 20.7). A prospective study in 25 000 pregnant women
showed that this class II marker has an odds ratio of 140,
which makes it one of the most reliable HLA associations
reported to date, with negative predictive power equal to
that of HLA-B27 in ankylosing spondylitis. A difference in
the efficiency of presentation of the GPIIIa-Leu33 (HPA-
1a)-derived oligopeptide between DRB3*01:01-positive and
-negative antigen-presenting cells to CD4+ T cells is the most
likely explanation of this restriction in alloimmunization.
The frequency of the HLA DRB3*01:01 allele in Caucasians is
33%, and this marker therefore has a high negative predictive
value, but a low positive one for anti-HPA-1a formation.
Allele frequencies
In Caucasians, the allele frequency for most HPA systems
is skewed toward the “a” allele. The allele frequencies vary
between populations; for example, GPIIIa-Pro33 (HPA-1b)
is extremely rare or absent in East Asia, while the opposite is
the case for the GPIIIa-Gln143 form (HPA-4b).
280 Molecular Hematology
N between self and non-self and triggers the
Incidence of NAITP
Alloantibodies against the HPA-1a (PlA1 , Zwa ) alloanti-
gen occur in 1 in 365 pregnancies and cause severe
thrombocytopenia, with a neonatal platelet count of less than
50 × 109 l−1 in 1 in 1100 term neonates.
Pathology
Maternal IgG alloantibodies against a fetal HPA alloantigen
can cross the placenta and bind to the fetal platelets, thus
reducing platelet survival. The HPA-1 and HPA-5 systems are
the two most clinically relevant, with the majority of cases
caused by antibodies against HPA-1a (80%) and HPA-5b
(10–15%).
Characteristics
Neonatal alloimmune thrombocytopenia presents in the oth-
erwise healthy newborn as petechiae or ecchymoses, or is
found coincidentally by a whole blood count. Severe cases
can present neurological symptoms due to cerebral bleeds or
with hydrops fetalis or cerebral cysts. NAITP can affect the
first pregnancy and has a 10% risk of severe intracranial hem-
orrhage. Diagnosis is based on the detection of HPA alloanti-
bodies in the maternal serum combined with a parental HPA
incompatibility, as determined by molecular typing meth-
ods such as PCR-SSP, TaqMan® assay, or sequencing methods
such as SBT or NGS.
Inheritance
All systems described to date are biallelic with codominant
expression of both alleles. Homozygous women can produce
Fig. 20.7 Structure of human leukocyte
antigen (HLA) DRB3*01:01 molecule. A view of
CHO
the peptide-binding site of the HLA class II
DRB3*01:01 molecule, as seen by the T-cell
receptor of a T lymphocyte. Two α helices (see
Figure 20.1b) are resting on a base of
antiparallel-running β-pleated sheets. The
antigen-derived peptide is not shown, but in our
example of immune response restriction, an
oligopeptide derived from fetal GPIIIa-Leu33 will
be juxtaposed between the two helices. The
presence of this peptide defines the difference
proliferation of GPIIIa-Leu33 (HPA-1a)
alloantigen-specific helper T cells.
alloantibodies against a paternally inherited platelet alloanti-
gen present on the fetal platelets but absent from the mater-
nal ones. In cases where the partner is homozygous for the
relevant alloantigen, there is a 100% chance that his future
offspring will be at risk, whereas in the case of heterozygosity
50% will be affected.
Treatment
A neonatal platelet count of less than 20 × 109 l−1 should be
corrected immediately, preferably with HPA-1a- and HPA-
5b-negative donor platelets, as these will be compatible with
the maternal HPA alloantibody in over 95% of cases. In cases
of absence of HPA-compatible platelets, initial transfusion of
random ABO/RhD-compatible platelets should be consid-
ered, followed by the transfusion of compatible ones if the
platelet count decreases further. In a typical case, the platelet
count should recover to normal within a week, although a
more protracted recovery can occur.
Counseling
Counseling of couples with an index case about the risks
of severe fetal/neonatal thrombocytopenia in a subsequent
pregnancy needs to be based on the severity of disease in the
index case and the outcome of immunological investigations.
The following should be considered:
r thrombocytopenia in subsequent cases is as severe or
generally more severe.
r antibody specificity and titer have some correlation with
severity; for example, HPA-5b antibodies generally cause
mild thrombocytopenia, which rarely results in a cerebral
bleed. The latter is generally associated with HPA-1a anti-
bodies.
 The molecular basis of blood cell alloantigens 281

Table 20.4 Human neutrophil antigens (HNA)

Antigen system Alleles Antigens Former name Gene CD

HNA-1 FCGR3B*01 HNA-1a NA1 FCGR3B CD16

FCGR3B*02 HNA-1b, HNA-1d NA2 (HNA-1b)
FCGR3B*03 HNA-1c, HNA-1b SH (HNA-1c)
FCGR3B*04 HNA-1a
FCGR3B*05 HNA-1b var

HNA-2* n/a HNA-2 NB1 CD177 CD177

HNA-3 SLC44A2*01 HNA-3a 5b SLC44A2 Not known
SLC44A2*02 HNA-3b
SLC44A2*03 HNA-3a var
HNA-4 ITGAM*01 HNA-4a Mart ITGAM CD11b
ITGAM*02 HNA-4b
HNA-5 ITGAL*01 HNA-5a Ond ITGAL CD11a
ITGAL*02 HNA-5b

*HNA-2 is defined by isoantibodies.

r a high-titer HPA antibody is more likely to be associated
with severe thrombocytopenia, but cerebral bleeds also
occur with low titers.
Neonatal alloimmune neutropenia
Maternal alloimmunization against neutrophil-specific
alloantigens on fetal/neonatal neutrophils is rare, although
there are no precise prevalence figures. Clinical presenta-
tion is one of mainly bacterial infection with a selective
neutropenia on a whole blood cell count.
HNA system
Neutrophils, like all other cells present in blood, express
polymorphic molecules that can induce strong antibodies
when transfused or transplanted or as result of pregnancy.
These antigens are called human neutrophil antigens
(HNAs). HNA-specific antibodies have been implicated
in the development of neonatal alloimmune neutropenia,
TRALI, and other alloimmune or autoimmune neutrope-
nias. So far five HNA antigenic systems have been described
(HNA-1 to HNA-5), which are the result of polymorphisms,
except for HNA-2.
HNA-1
HNA-1 is located on a glycosylphosphatidylinositol (GPI)-
anchored glycoprotein that forms the low-affinity Fcγ recep-
tor. HNA-1 is encoded by the FcγRIIIB (FCGR3B) gene
located on chromosome 1 and mediates IgG-induced phago-
cytosis. Five alleles of the HNA-1 system have been described,
FCGR3B*01, FCGR3B*02, FCGR3B*03, FCGR3B*04 and
FCGR3B*05. Five amino acid substitutions at positions 36,
65, 78, 82 and 106 determine the four HNA-1 antigens
(Table 20.4). A single amino acid substitution at position
78 of FCGR3B causes the expression of HNA-1c instead of
HNA-1d next to HNA-1b. Thus, individuals homozygous for
the gene encoding HNA-1a lack expression of HNA-1b, 1c,
and 1d. Due to duplication of the FCGR3B gene, as found
in HNA-1c positive Caucasians, the FCGR3B*01 (HNA-1a)
allele and the FCGR3B*03 (HNA-1b and HNA-1c) allele
are expressed from the same chromosome. Individuals that
possess this haplotype together with the FCGR3B*02 allele
(HNA1-b and HNA-1d expression) on the other chromo-
some will express all four antigens of the HNA-1 system.
The single nucleotide polymorphisms (SNPs) responsible
for the expression of the different HNA-1 antigens are
shown in Table 20.5. Individuals genotyped positive for the
FCGR3B*01 and FCGR3B*04 alleles will express the HNA-1a
epitope, while individuals positive for the FCGR3B*05 will
express the HNA-1bvar epitope for HNA-1b antibodies.
Some individuals who do not express the FCGR3B gene and
carry a null phenotype can develop isoantibody of known
clinical significance in cases of NAITP. The FcγRIIIB null
phenotype is rare and is based on a double deletion of
the FCGR3B gene, and is in some cases associated with a
deletion of the FcγRIIC (FCGR2C) gene. Deficiency for the
most abundant Fc receptor on neutrophils does not seem to
be associated with an obvious clinical phenotype. This is in
contrast with a mutation in the FcγRIIIA (FCGR3A) gene,
which encodes a Leu48His substitution in the first extra-
cellular domain of the NA cell FcγRIIIA which, although
only described in one infant, was associated with recurrent
282 Molecular Hematology

Table 20.5 HNA-1 antigens and associated alleles, SNP and amino acid positions

Nucleotide at SNP position Amino acid positions

Antigen Allele 108 194 233 244 316 36 65 78 82 106

HNA-1a FCGR3B*01 G A C G G Arg Asn Ala Asp Val

HNA-1b, 1d FCGR3B*02 C G C A A Ser Ser Ala Asn Ile
HNA-1b, 1c FCGR3B*03 C G A A A Ser Ser Asp Asn Ile
HNA-1a FCGR3B*04 G A C G A Arg Asn Ala Asp Ile
HNA-1bvar FCGR3B*05 C G C G A Ser Ser Ala Asp Ile

HNA, human neutrophil antigens; SNP, single nucleotide polymorphism.

and serious respiratory tract viral infection from birth. The
FcγRIIIB null phenotype can cause immune neutropenia in
the newborn due to maternal anti-FcγRIIIB isoantibodies.
PCR techniques can be used to determine the HNA-1
and FcγRIIIB null genotypes, and panels of HNA-typed
granulocytes are useful probes for alloantibody detection.
HNA-1 alloantibodies have also been implicated in cases
of autoimmune and alloimmune neutropenia, including
neonatal alloimmune neutropenia and TRALI.
HNA-2
HNA-2 (previously known as NB1) is expressed on a
58–64 kDa glycoprotein (CD177) found on the plasma mem-
brane and secondary granules of neutrophils. This glycopro-
tein is linked to the plasma membrane by a GPI anchor and
is coded for by a gene located on chromosome 19. HNA-2 is
expressed in approximately 50% of the total neutrophils by
95% of individuals. Some individuals do not express HNA-2
due to a transcription defect in the CD177 gene. The HNA-2
alloantigen typing is based on the use of human immune anti-
sera and immunofluorescence. HNA-2 antibodies have been
found in patients with alloimmune and autoimmune neu-
tropenia.
HNA-3
HNA-3 (HNA-3a antigen, previously known as 5b) is car-
ried on the choline-transporter-like protein 2, which is a
transmembrane protein encoded by the SLC44A2 gene on
chromosome 19p13.1. There have been two HNAs iden-
tified which are encoded by three alleles (SLC44A02*01,
SLC44A02*02, SLC44A02*03). These are the result of two
SNPs at positions 451 and 455, giving rise to amino acid
changes at positions 151 and 152, respectively. A SNP
(455G > A) changing an arginine to glutamine at position
152 of the protein is responsible for the difference between
HNA-3a and HNA-3b. SLC44A02*03 encodes the HNA-3avar
epitope, which is recognized by HNA-3a antibodies but with
weaker affinity. HNA-3a allo-antibodies are implicated in
febrile transfusion reactions, neonatal immune neutropenia,
and severe, often fatal cases of TRALI.
HNA-4
HNA-4 (previously known as Marta ) is located on the αM
chain (CD11b) of the β2 integrin MAC-1. HNA-4 is encoded
by the ITGAM gene located at chromosome 16p11.2. This
system has two antigens, HNA-4a (ITGAM*01) and HNA-4b
(ITGAM*02), the result of a single amino acid substitution
at position 61. aM/b2-integrin is a transmembrane protein
expressed on the surface of many leukocytes and involved
in adhesion, transmigration, phagocytosis, and cell-mediated
cytotoxity. The difference between the antigens is caused by a
unique SNP at position 230 of the gene, changing an arginine
into a histidine at protein level.
HNA-5
HNA-5 (previously known as Onda ) is located on the αL inte-
grin unit of the leucocyte function-associated antigen (LFA)-
1, also known as CD11b. It is encoded by the ITGAL gene
which is located at chromosome 16p11.2. aL/b2-integrin is a
leukocyte-specific adhesion molecule involved in leukocyte
interactions and trafficking. The HNA-5a antigen is deter-
mined by a single amino acid substitution at position 766 and
HNA-5a alloantibodies have been involved in NAIN.
Detection of HNA antigens and antibodies
Some HNA antigens are uniquely expressed on neutrophils
(HNA-1 and HNA-2), whereas others are also present on
other cells or tissues (HNA-3, HNA-4, and HNA5). With
the exception of HNA-2, the polymorphisms that determine
the HNA genotype of an individual can be detected using
molecular technique including PCR-SSP, PCR-SSOP, and
DNA sequencing. The detection of HNA-specific antibodies
however is still difficult, unreliable, and time-consuming and

can involve the use of a number of tests such as the granu-
locyte agglutination test, granulocyte immunofluorescence
test, chemiluminescence test, and the monoclonal antibody
immobilization of granulocyte antigens (MAIGA) test.
Indeed, many of these techniques (except for MAIGA) are
not able to distinguish specific HNA from HLA antibodies.
There is a need to develop a new generation of techniques
for the detection and characterization of HNA-specific
antibodies.
In addition to neonatal neutropenia, neutrophil-specific
antibodies are implicated in non-hemolytic febrile transfu-
sion reactions, TRALI, and autoimmune neutropenia. Severe
but reversible neutropenia in the newborn may require treat-
ment with antibiotics to control bacterial infection. So far,
there is minimal or no evidence that the mutations in the
FcγRIIIB protein have any functional consequences, exem-
plified by the FcγRIIIB null phenotype, which is not linked
with an obvious pathological phenotype.
Conclusions
The molecular basis of most blood cell alloantigens, includ-
ing those of the HLA system, has been discovered with the
use of molecular techniques. This has enabled their reliable
definition at high resolution, replacing previous techniques
that were based on the use of polyclonal antibody reagents.
New molecular techniques such as NGS allow allelic resolu-
tion and the discovery of new alleles. This body of knowledge
has already made a significant contribution to current clin-
ical management. On the one hand, first-trimester fetal
typing for blood cell-specific alloantigens can now be used
in cases of severe maternal alloimmunization to prevent fetal
and neonatal morbidity, and improved selection of HLA-
and HPA-matched donor platelets aids the management
of hemato-oncological patients. On the other hand, it is
envisaged that bone marrow and solid organ transplants
across apparent mismatches will be able to progress because
of the identification of so-called permissive mismatches,
and better HLA matching will allow accommodation of the
graft at lower levels of immune suppression. Finally, more
complete understanding of the molecular rules defining a
subject’s immune response status should lead to the identi-
fication of high responders, in order to target these for more
specific manipulation of their immune system, with the aim
of establishing alloantigen-specific non-responsiveness.
Further reading
Brown, C.J. and Navarrete, C.V. (2011). Clinical relevance of the HLA
system in blood transfusion. Vox Sang. 101: 93–105.
The molecular basis of blood cell alloantigens 283
Creary, L.E., Guerra, S.G., Chong, W. et al. (2019). Next-generation
HLA typing of 382 International Histocompatibility Working Group
reference B-lymphoblastoid cell lines: report from the 17th Interna-
tional HLA and Immunogenetics workshop. Hum. Immunol. 80: 449–
460.
Curtis, B.R. and McFarland, J.G. (2014). Human platelet antigens – 2013.
Vox Sang. 106: 93–102. Review.
Dendrou, C.A., Petersen, J., Rossjohn, J., and Fuggar, L. (2018). HLA
variation and disease. Nat. Rev. Immunol. 18: 325–339.
Duquesnoy, R.J., Kamoun, M., Baxter-Lowe, L.A. et al. (2015). A per-
sonal viewpoint: should HLA mismatch acceptability for sensitized
transplant candidates be determined at the high-resolution rather
than the antigen level? Am. J. Transplant. 15: 923–930.
Flesch, B.K. for the International Society of Blood Transfusion (ISBT)
HNA Nomenclature Subcommittee. (2015). Human neutrophil anti-
gens: a nomenclature update based on new alleles and new antigens.
ISBT Sci. Ser. 10: 243–249. Review.
Flesch, B.K. and Reil, A. (2018). Molecular genetics of the human neu-
trophil antigens. Transfus. Med. Hemother. 45: 300–309. Review.
Gajjar, K. and Spencer, C. (2009). Diagnosis and management of non-
anti-D red cell antibodies in pregnancy. Obstet. Gynaecol. 11: 89–95.
Guerra, S.G., Chong, W., Brown, C.J., and Navarrete, C.V. (2018). Evalu-
ation of Ion Torrent sequencing technology for rapid clinical human
leucocyte antigen typing. Int. J. Immunogenet. 45: 230–235.
de Haas, M., Thurik, F.F., Koelewijn, J.M., and van der Schoot, C.E.
(2015). Haemolytic disease of the fetus and newborn. Vox Sang. 109:
99–113.
Horton, R., Wilming, L., Rand, V. et al. (2004). Gene map of the
extended human MHC. Nat. Rev. Genet. 5: 889–899.
Janeway, C. Jr., Travers, P., Walport, M. et al. (2001). Immunology: The
Immune System in Health and Disease, 5e. New York: Garland Science.
Le van Kim, C., Mouro, I., Cherif Zahar, B. et al. (1992). Molecu-
lar cloning and primary structure of the human blood group RhD
polypeptide. Proc. Natl. Acad. Sci. U.S.A. 89: 10925–10929.
Lieberman, L., Greinacher, A., Murphy, M.F. et al. (2019). Fetal
and neonatal alloimmune thrombocytopenia: recommendations for
evidence-based practice, an international approach. Br. J. Haematol.
185: 549–562.International Collaboration for Transfusion Medicine
Guidelines (ICTMG).
van Loghem, J.J., Dorfmeyer, H., van de Hart, M. et al. (1959). Serolog-
ical and genetical studies on the platelet antigen (zw). Vox Sang. 4:
161–169.
Muschter, S., Berthold, T., and Greinacher, A. (2011). Developments in
the definition and clinical impact of human neutrophil antigens. Curr.
Opin. Hematol. 18: 452–460. Review.
Peterson, J.A., McFarland, J.G., Curtis, B.R., and Aster, R.H. (2013).
Neonatal alloimmune thrombocytopenia: pathogenesis, diagnosis
and management. Br. J. Haematol. 161: 3–14. Review.
Porcelijn, L. and de Haas, M. (2018). Neonatal alloimmune neutropenia.
Transfus. Med. Hemother. 45: 311–316. Review.
Porto, G., Brissot, P., Swinkels, D.W. et al. (2016). EMQN best prac-
tice guidelines for the molecular genetic diagnostic of hereditary
hemochromatosis (HH). Eur. J. Hum. Genet. 4: 479–495.
Qureshi, H., Massey, E., Kirwan, D. et al. (2014). BCSH guideline for
the use of anti-D immunoglobulin for the prevention of haemolytic
disease of the fetus and newborn. Transfus. Med. 24: 8–20.
Reil, A. and Bux, J. (2015). Geno- and phenotyping of human neutrophil
antigens. Methods Mol. Biol. 1310: 193–203.
284 Molecular Hematology
Reil, A., Sachs, U.J., Siahanidou, T. et al. (2013). HNA-1d: a new human
neutrophil antigen located on Fcγ receptor IIIb associated with
neonatal immune neutropenia. Transfusion 53: 2145–2151.
Rozman, P. (2002). Platelet antigens. The role of human platelet
alloantigens (HPA) in blood transfusion and transplantation. Transpl.
Immunol. 10: 165–181. Review.
Soothill, P.W., Finning, K., Latham, T. et al. (2015). Use of cffDNA to
avoid administration of anti-D to pregnant women when the fetus is
RhD-negative: implementation in the NHS. BJOG 122: 1682–1686.
Suzuki, S., Ranade, S., Osaki, K. et al. (2018). Reference grade character-
ization of polymorphisms in full-length HLA class I and II genes with
short-read sequencing on the ION PGM system and long-reads gen-
erated by single molecule, real-time sequencing on the PacBio plat-
form. Front. Immunol. 9, article 2294.
Veldhuisen, B., Porcelijn, L., Ellen van der Schoot, C., and Haas, M.D.
(2014). Molecular typing of human platelet and neutrophil antigens
(HPA and HNA). Transfus. Apher. Sci. 50: 189–199. Review.
Wagner, F.F. and Flegel, W.A. (2004). Review: the molecular basis of the
Rh blood group phenotypes. Immunohematology 20: 23–36.
Wienzek-Lischka, S., König, I.R., Papenkort, E.M. et al. (2017). HLA-
DRB3*01:01 is a predictor of immunization against human platelet
antigen-1a but not of the severity of fetal and neonatal alloimmune
thrombocytopenia. Transfusion 57: 533–540.
Williamson, L.M., Hackett, G., and Rennie, J. (1998). The natural his-
tory of fetomaternal alloimmunization to the platelet-specific antigen
HPA-1a (PlA1, Zwa) as determined by antenatal screening. Blood 92:
2280–2287.
Winkelhorst, D. and Oepkes, D. (2019). Foetal and neonatal alloim-
mune thrombocytopenia. Best Pract. Res. Clin. Obstet. Gynaecol.
Epub ahead of print. Review.
Zdravic, D., Yougbare, I., Vadasz, B. et al. (2016). Fetal and neonatal
alloimmune thrombocytopenia. Semin. Fetal Neonatal. Med. 21: 19–
27. Review.
Chapter 21 Functions of blood group antigens
Jonathan S. Stamler1 & Marilyn J. Telen2
1 Harrington Discovery Institute and Institute of Transformative Molecular Medicine, University Hospitals Cleveland Medical Center
and Case Western Reserve University, Cleveland OH, USA
2 Department of Medicine, Division of Hematology, Duke University Medical Center, Durham NC, USA
Introduction, 285
Anion-exchanger protein 1 (band 3 protein), 285
Rh protein, 289
Glycophorins C and D, 290
Introduction
Erythrocyte blood group antigens are, by definition, poly-
morphic epitopes of red cell surface structures that are rec-
ognizable by antibodies. These epitopes generally include rel-
atively small changes, such as the substitution of one amino
acid, in molecules that are largely the same from one indi-
vidual to another. After the initial discovery of ABO blood
groups at the beginning of the twentieth century, red cell
serologists went on to discover hundreds of such polymor-
phic epitopes and gradually succeeded in organizing them
into blood groups, antigens that arise from polymorphisms of
a single parent molecule. During the late twentieth century,
most of these blood groups were localized to carrier struc-
tures (proteins or polysaccharides). Discovery of the bio-
chemical and genetic basis for blood group antigens has now
led to exploration of the function of these proteins beyond
their importance in transfusion medicine.
This chapter reviews current knowledge about erythro-
cyte membrane proteins that bear blood group antigens and
whose functional importance has been characterized. Pri-
mary attention is paid to two of the most interesting of
these proteins: the anion exchanger protein, which bears anti-
gens of the Diego blood group system, and the Rh proteins.
Functional information about other proteins is summarized
briefly.
Anion-exchanger protein 1
(band 3 protein)
Anion-exchanger protein 1 (AE1) is the most prevalent
integral protein of the red blood cell (RBC) membrane
Molecular Hematology, 4th Edition. Edited by Drew Provan and John G. Gribben.
© 2020 John Wiley & Sons Ltd. Published 2020 by John Wiley & Sons Ltd.
Blood group antigen proteins that function as adhesion molecules, 291
Blood group antigen proteins with other functions, 293
Summary, 293
Further reading, 293
(∼1 million copies/cell), constituting 20–30% of the protein
mass. Previously referred to as band 3 protein, on the basis
of its original electrophoretic migration profile, AE1 is a
95–102 kDa variably glycosylated protein constructed of
911 amino acids, yielding three distinct structural and
functional domains: the N-terminal 45-kDa cytoplasmic
domain, the 55-kDa C-terminal hydrophobic transmem-
brane domain, and a small 28-amino-acid acidic cytoplasmic
C-terminus (Figure 21.1). Extracellular domains of AE1
express the Diego blood group system, composed of two
sets of antithetical antigens, Diego (Dia /Dib ) and Wright
(Wra /Wrb ). AE1 also expresses ABH and Ii antigens on a
single multibranched N-linked oligosaccharide attached to
the fourth extracellular domain.
Transmembrane domain
The transmembrane segment of AE1 (amino acids 404–882)
is responsible for the electroneutral cotransport of HCO− 3 for
Cl− , and is also likely involved in the export of nitric oxide
(NO) equivalents. The membrane-spanning domain accom-
modates the tide of HCO− 3 generated by the activity of ery-
throcytic carbonic anhydrase on carbon dioxide (CO2 ) enter-
ing the RBC, thus increasing the capacity of the blood to carry
CO2 from the tissues to the lungs. In the peripheral tissues,
CO2 taken up by RBCs is converted to H+ and HCO− 3 by the
action of carbonic anhydrase. The transmembrane domain
of AE1 exports HCO− −
3 in exchange for Cl to maintain phys-
+
iological pH in the plasma, while H binds to hemoglobin
(Hb), right-shifting its oxygen (O2 ) affinity to facilitate O2
unloading. The reverse occurs in the lungs. The anion
−
exchange domain is not specific for HCO− 3 ∕Cl and can
2− −
also transport other anions (SO4 , H2 PO4 ), anionic phos-
pholipids, and even divalent cations. Of particular interest
is that AE1 also transports negatively charged NO metabo-
lites (NO− − −
2 , NO3 and OONO , see later), and more generally
285
286 Molecular Hematology
Transmembrane domain
–SH –SH
C-terminus
Carbonic anhydrase II
– – –
–
– – Polyanionic N-terminus
HS SH - Hemoglobin
HS SH
- Glyceraldehyde-3-phosphate
dehydrogenase
Cytoplasmic - Phosphofructokinase
domain - Aldolase
Proline-rich
hinge region
Fig. 21.1 Schematic representation of an anion-exchanger 1
protein dimer. The cytoplasmic domains (CDAE1) are enclosed by the
tinted circle. CDAE1 cysteine thiols are indicated by –SH, and ⊝
indicates the negative charges of acidic amino acids in the N-terminus.
Proteins known to bind specifically to the N-terminus are also listed.
plays a central role in the export of NO vasodilatory activ-
ity from RBCs that likely involves a Cys-NO relay, includ-
ing Cys843 within the membrane-spanning region, Cys317
within the N-terminus, and hemoglobin (Cysβ93), which
associates with the N-terminus of AE1. Importantly, trans-
membrane C843 communicates with both the intracellu-
lar N-terminus and the extracellular plasma to mediate NO
transport across the membrane. An N-terminal truncated
form of AE1 is expressed by the basolateral membrane of the
α-intercalated cells in the renal collecting ducts and mediates
acid secretion.
C-terminal cytoplasmic domain
The C-terminus of the transmembrane domain of AE1 con-
cludes in a 33-residue cytoplasmic protrusion consisting pre-
dominantly of acidic amino acids and has been shown to
possess a binding site for the carbonic anhydrase II iso-
form. Using glutathione S-transferase (GST) fusion proteins
with normal and mutated constructs, Vince and colleagues
demonstrated that the critical binding site for carbonic anhy-
drase II is the acidic residue 887–890 region of AE1. A GST
fusion protein containing the C-terminal 33 amino acids of
AE2 could also bind carbonic anhydrase II, suggesting that
this tethering of carbonic anhydrase II to anion exchang-
ers is a general mechanism promoting bicarbonate transport
across membranes. AE2 is the most widely expressed form of
AE, located in the basolateral membrane of nearly all epithe-
lial cells.
N-terminal cytoplasmic domain
The primary function of the N-terminal cytoplasmic domain
(amino acids 1–403) of AE1 (CDAE1) is presumably to
maintain erythrocyte integrity and stability by bracing the
cytoskeleton to the plasma membrane, and consequently it
may influence RBC rheology. This large hydrophilic domain
projects into the RBC cytoplasm with a spatial orientation
that is strongly pH dependent. Under alkaline conditions (pH
10) the appendage is fully extended, whereas a slightly acidic
milieu (pH 6.5) induces flexion, such that the N-terminus is
brought close to the cell membrane. Flexibility in this part
of AE1 is conferred by a proline-rich hinge region between
amino acids 175 and 190. The actual orientation of CDAE1
is likely to be a semi-flexed state, because the internal pH
of RBCs is closer to 7.0–7.2. However, given that carbonic
anhydrase II binds to the C-terminus of AE1 in a juxtamem-
brane location and that each AE1 channel can export 2000–
3000 molecules of HCO− 3 per second, the micromilieu imme-
diately adjacent to CDAE1 may experience much lower pH
values under certain conditions, inducing greater degrees of
CDAE1 flexion. These regions were once thought to be dis-
crete operational units, but it is now clear that remote alter-
ations in membrane domain structure can influence cyto-
plasmic domain activity, suggesting that these regions are
functionally linked.
CDAE1 contains high-affinity binding sites for three
peripheral membrane proteins that link AE1 to the under-
lying cytoskeletal meshwork. Ankyrin is a 206 kDa pyrami-
dal protein that interacts with several sites throughout the
length of CDAE1, hinting that ankyrin binds to a flexed
conformation of CDAE1 in vivo. Ankyrin also possesses
three structural–functional domains: an N-terminal AE1 and
tubulin-binding domain, a middle spectrin-binding domain,
and a C-terminal domain containing sequences that reg-
ulate ankyrin binding to AE1 and spectrin. CDAE1 also
binds protein 4.2, or pallidin, a 77 kDa protein that binds
to ankyrin as well, acting to reinforce the AE1–ankyrin
interaction. Pallidin shows striking homology to transglu-
taminases, but lacks enzymatic activity owing to loss of a nec-
essary conserved active site residue. Finally, CDAE1 inter-
acts (at least in vitro) with protein 4.1, a 66 kDa moiety that
firmly binds spectrin near the actin-binding site, thus inten-
sifying the spectrin–actin interaction. The spectrin-binding
activity of protein 4.1 is concurrently regulated by phos-
phorylation/dephosphorylation, Ca2+ –calmodulin associa-
tion, and phosphatidylinositol pathways. In addition, spec-
trin, ankyrin, and AEI are known to be S-nitrosylated and it is
thought that S-nitrosylation leads to improvements in mem-
brane deformability.
The N-terminus of CDAE1 also includes specific
high-affinity binding sites for the glycolytic enzymes
glyceraldehyde 3-phosphate dehydrogenase (GAPDH),

phosphofructokinase (PFK), and aldolase. Binding to
CDAE1 results in a decrease in enzymatic activity (90% for
aldolase and 100% for GAPDH), while at the same time
enzyme substrates, products, and cofactors displace the
bound enzymes, suggesting that specific CDAE1–enzyme
interactions take place at the catalytic site or a linked
allosteric site. Aldolase and PFK mutually inhibit each
other from binding CDAE1, but neither enzyme can dis-
place bound GAPDH, indicative of a unique binding site for
GAPDH. S-nitrosylation of GAPDH, which may entail mod-
ification of both active-site and allosteric Cys, regulates both
GAPDH activity and its binding to the RBC membrane. In
addition, phosphorylation of tyrosine 8 of AE1 regulates gly-
colytic enzyme binding, and thus activity. Binding affinities
of the glycolytic enzymes for CDAE1 are strongly dependent
on H+ and salt concentration, decreasing dramatically as
the pH and ionic strength are raised into the physiological
range, calling into question the extent of binding in vivo.
However, several studies support the notion of significant
enzyme–CDAE1 binding in the intact erythrocyte.
Nitric oxide export
The role of NO as a critical third gas in the human res-
piratory cycle (in addition to O2 and CO2 ) has been gain-
ing acceptance and has been recently confirmed by stringent
genetic criteria. Only three amino acids are strictly conserved
in all mammalian hemoglobins; two of them are required
for O2 binding and one (Cysβ93) for NO bioactivity. Mice
deficient in Cysβ93 (replaced by alanine) possess RBCs that
are essentially normal by all traditional measures (O2 bind-
ing, O2 affinity, O2 content, hematocrit, etc.), but cannot oxy-
genate tissues adequately. The classic physiological response
of hypoxic vasodilation through which blood flow is cou-
pled to O2 delivery is profoundly impaired in these animals
whose hearts and limbs are starved of O2 . RBCs from these
animals cannot export NO bioactivity and thus cannot dilate
blood vessels or stimulate brain centers to counteract hypoxia
(increase respiratory rate). Most recent data indicate that the
animals are very susceptible to heart attacks and heart fail-
ure, suggesting that hemoglobin S-nitrosylation is essential
for both normal cardiovascular function and cardioprotec-
tion against heart disease.
Vascular-derived NO is generated chiefly by endothelial
NO synthase and is a principal vasodilator and inhibitor of
platelet activation. Vascular NO regulates blood pressure,
but is not involved directly in tissue oxygenation. A por-
tion of NO – whether in the form of NO itself, nitrite, or
S-nitrosothiols – ultimately penetrates circulating RBCs,
where it interacts with hemoglobin (Hb) at the heme pros-
thetic groups or with β-chain cysteine thiols (Cysβ93) to
form biologically active S-nitrosohemoglobin (SNO-Hb).
Under low O2 tension (conditions that promote the T state in
Functions of blood group antigens 287
FeNO SH FeO2 SNO FeNO SH
O2 SNO
T R T
NO/SNO O2
Vein Lungs Arterioles/
Capillaries
Fig. 21.2 Simplified allosteric model of NO–hemoglobin
interactions. Nitric oxide (NO) (whether sourced from NO itself,
nitrite, or SNOs) binds to the heme groups of deoxygenated or T-state
hemoglobin (Hb). On oxygenation, Hb assumes the R conformation
and NO is transferred to the cysteine thiol in position 93 of the
β-chain (Cysβ93) to form S-nitrosohemoglobin (SNO-Hb). During
transit through the physiological oxygen gradient, partial pressure of
oxygen (PO2 ) falls, SNO-Hb converts to the T state, and NO is released
from Cysβ93 to evoke vasodilation (thus regulating blood flow).
Hb), SNO-Hb can release NO bioactivity to effect vasodila-
tion (and potentially inhibit platelet activation; Figure 21.2).
However, effective NO export from the RBC requires that
NO itself be shielded from scavenging by the hemes in Hb.
This is accomplished by complexing NO with cysteine thiols
to form S-nitrosothiols (SNOs). Bioactive NO is thus present
principally in the form of SNOs (rather than NO per se)
and, as discussed earlier, these SNOs participate in a relay
whereby NO is transferred between Cys in Hb and AE1 and
thereby exported from RBCs. Alternatively stated, RBCs
sense ambient O2 levels by regulating the R/T transition
in Hb and adjust NO bioavailability accordingly. RBCs can
thereby cause microvessels to dilate in proportion to the
degree of hypoxia. To the extent that tissue Po2 is determined
principally by blood flow, impairment in RBC-SNO function
would be expected to lower tissue Po2 . The implications for
disorders of RBCs such as sickle cell disease are considered
later.
Compartmentalization of SNO in the erythrocyte mem-
brane requires interaction between SNO-Hb and the inner
membrane surface. Hemoglobin binds to the cytoplasmic
face of the RBC membrane via both non-specific binding and
specific protein–protein interactions. The predominant site
of specific interaction between Hb and the RBC membrane
is through high-affinity binding to the N-terminal cytoplas-
mic domain of AE1. More specifically, the site of Hb bind-
ing within CDAE1 has been localized to the polyanionic
extreme N-terminus, which contains 18 acidic amino acids,
thus forming an anionic milieu at physiological pH. Analy-
sis of co-crystals of Hb and a CDAE1 N-terminal peptide has
shown that this stretch of acidic residues inserts deeply into
the 2,3-diphosphoglycerate-binding pocket formed between
β-globin subunits of tetrameric Hb. CDAE1 thus binds with
higher affinity to deoxyHb than to oxyHb (in which the β-
cleft is occluded) and, consistent with the requirements of
thermodynamic linkage, the O2 affinity of Hb is reduced in
the presence of isolated CDAE1.
288 Molecular Hematology
Interestingly, the AE1 of kidney intercalated cells is a
truncated isoform, lacking the N-terminal 65 residues of
the CDAE1, and does not bind Hb, suggesting that these
residues perform an AE1-related function specific to the
RBC. Binding of Hb to CDAE1 is non-cooperative but
causes a rightward shift in the O2 dissociation curve. Simply
stated, binding of Hb to CDAE1 stabilizes the “T” or deoxy
conformation of Hb. CDAE1 also binds oxidized Fe3+ or
metHb more avidly than oxyHb. Because AE1 is present in
insufficient quantity to bind a significant fraction of the total
Hb in an RBC, the influence of AE1 on overall Hb O2 delivery
is probably negligible. This raises the possibility that AE1-
bound Hb serves an alternate function sufficiently executed
by a small fraction of T-state Hb molecules. Indeed, SNO-Hb
binding to AE1 would promote the T structure and enhance
NO group release. Thus, the CDAE1/SNO-Hb interaction
would appear to be designed to accelerate SNO release from
SNO-Hb (Figure 21.3). In support of this assertion, seques-
tration of SNO in RBC membranes following treatment
with NO is markedly reduced by prior treatment with AE1
inhibitors, including DIDS (4,4′ -diisothiocyanostilbene-
2,2′ -disulfonic acid), and phenyglyoxal and niflumic acid.
Further, SNO-Hb can transnitrosylate inside-out vesicles
(IOVs) of RBC membranes, an effect blocked in IOVs
purified from RBCs treated with AE1 inhibitors. Finally,
S-nitrosylated AE1 can be immunoprecipitated from
the membranes of RBCs and from IOVs treated with
SNO-Hb.
CDAE1 contains two cysteine residues (Cys201 and
Cys317) with reactive thiols that are clustered at the subunit
interface of dimeric AE1. These thiols, which are surrounded
by amino acids that fit an S-nitrosylation motif, are removed
by chymotrypsin (which cleaves CDAE1 from the trans-
membrane domain). Accordingly, transnitrosylation of RBC
IOVs by SNO-Hb is blocked by thiol-alkylating agents or
O2
Fe(II)O2 Fe(II)
S NO S NO [NO]
R T
S
AE1
Cytosol Membrane
Fig. 21.3 Interaction between S-nitrosohemoglobin (SNO-Hb)
and the cytoplasmic domain of erythrocyte anion-exchanger 1
(CDAE1) to effect vasodilation. Binding to CDAE1 stabilizes the T
state of SNO-Hb and reduces affinity for nitric oxide (NO). SNO-Hb
transfers NO to vicinal thiols in CDAE1, which can then effect
vasodilation and platelet inhibition. In this model, vasodilation entails
export of SNO through sequential transnitrosylation reactions, [NO].
enzymatic digestion of the IOVs with chymotrypsin. In
cross-linking studies, Hb forms a disulfide bond (–S–S–)
between its Cysβ93 thiols and the cytoplasmic domain
cysteines of AE1, suggesting close spatial interplay in this
region. Studies on selenium transport across RBCs point
to communication between Hb Cysβ93 and Cys317 in
CDAE1 and between Cys 317 and Cys843 in the membrane-
spanning region of AE1. In addition, Cys843 communicates
with extracellular thiols in albumin. Thus coupled equilibria
between Cys-NO in Hb, AE1, and extracellular proteins
provides a potential route to relay NO from Hb into the
plasma and vessel wall.
Hemichromes, which are native Hb molecules with altered
conformation near the heme, have a very high affinity for
CDAE1. Moreover, hemichromes formed by the covalent
modification of the Cysβ93 thiol (an SNO-Hb mimetic)
bind preferentially to AE1. Therefore, transnitrosylation of
a vicinal thiol (presumably Cys 317) within the cytoplasmic
domain of AE1 is a strong candidate mechanism for trans-
fer of the NO group from Cysβ93 of SNO-hemichromes to
the RBC membrane (Figure 21.3). In fact, like Hb, SNO-Hb
forms disulfide links with AE1 under oxidative conditions
that predispose to hemichrome formation.
In sum, a role for the anion exchange function of AE1
in transmembrane movement of NO bioactivity is likely.
AE1 can import anionic NO metabolites such as nitrite,
which may participate in SNO-Hb formation. AE1-affiliated
SNO-Hb constitutes a major source of NO-based vasodila-
tory activity. NO derived from SNO-Hb is relayed to Cys
within the N-terminus of AE1. AE1-liganded (S)NO is trans-
ferred to Cys in plasma proteins (or the vascular wall), likely
through the intermediacy of SNO within the membrane-
spanning region of AE1.
SNO-Hb in RBCs is distributed equally between the
cytosol and plasma membrane. However, only the plasma
membrane fraction is bioactive. Deficiency of plasma mem-
brane SNO can result from defects in either SNO-Hb syn-
thesis or its transfer of NO groups into the membrane. In
either case, RBC vasoactivity is compromised. Vasodilation
by RBCs is impaired in several diseases, including sickle cell
disease, which is characterized by a triple defect in SNO pro-
cessing: SNO-Hb synthesis is reduced (an intrinsic property
of Hb S); the interaction between Hb S and AE1 is disrupted;
and the transfer of NO into the membrane is impaired. Most
notably, Cys residues of AE1 that would otherwise accept the
NO groups are oxidized in sickle RBCs. Membrane SNO is
therefore greatly diminished. Impaired vasodilation in sickle
patients corresponds with clinical symptoms (painful crises
and organ damage) and may therefore represent a molec-
ular corollary of vaso-occlusive disorders. A deficiency in
plasma membrane SNO-AE1 has also been found in banked
blood, which likewise exhibits impaired vasodilation. Deficits
in SNO-Hb and/or in RBC-mediated vasodilation have also

been found in subsets of patients with heart failure, pul-
monary hypertension, and diabetes.
Deficiencies and mutations of AE1
Clinical conditions resulting from mutations or deficiencies
in AE1 range from asymptomatic to severe, with manifesta-
tions predictably reflecting perturbation of one or both of
the two primary functions of AE1, exchange/transport and
cytoskeletal membrane stability. A significant segment of the
population (5–20%) are heterozygous for an AE1 polymor-
phism called “band 3 Memphis,” involving a lysine to glu-
tamic acid substitution in codon 56, resulting in a function-
ally normal AE1.
Southeast Asian ovalocytosis (SAO), characterized geno-
typically as a deletion of amino acids 400–408 in AE1, results
in oval RBCs with increased rigidity, decreased osmotic
fragility, and reduced expression of certain red cell antigens.
Natural selection dramatically increases the prevalence of
SAO in areas with endemic malarial infection, given that
this defect confers reduced parasitemia and disease sever-
ity. However, homozygosity for the SAO mutation appears to
be incompatible with life, as homozygotes have never been
observed (even in areas with 40% prevalence), and the inci-
dence of miscarriage in heterozygous parents is twice that for
normals (20% vs. 9.8%).
Over 40 mutations of the AE1 protein have been found
to be associated with a subset of individuals with domi-
nant hereditary spherocytosis, a condition characterized by
loss of erythrocyte surface area and a mild to severe hemol-
ysis, almost invariably improved by splenectomy. Despite
being scattered through the entire protein, many mutations
in the membrane-spanning domain involve substitutions of
highly conserved arginine residues, thought to be important
in maintaining proper membrane orientation. Other muta-
tions causing hereditary spherocytosis, including those in the
cytoplasmic domain, lead to decreased binding affinities for
the cytoskeletal bridging proteins ankyrin and pallidin.
Complete AE1 deletion in a murine model results in a
deadly hypercoagulable state in which only 5–10% of AE1
null mice survive the neonatal period. Almost uniformly, ani-
mals were found at necropsy to have large thrombotic lesions
of the heart, subcapsular liver necrosis suggesting arteriolar
ischemia, and large vein thrombi. This phenotype is also
consistent with a defect in SNO export. Inaba and colleagues
similarly described a total deficiency of AE1 in Japanese black
cattle homozygous for a nonsense mutation (CGA → TGA)
in codon 646, causing an early stop signal. These animals
exhibit moderate uncompensated anemias, spherocytosis,
and mild acidosis at rest that is exacerbated by exercise.
It should be noted in this regard that RBCs may export
vasodilatory NO bioactivity through AE1-independent
mechanisms, including S-nitrosoglutathione transport and
Functions of blood group antigens 289
a pathway involving protein disulfide isomerase. The only
reported human case of total absence of AE1 occurred in a
Portuguese female delivered by cesarean section at 36 weeks
of gestation. The neonate was hydropic and severely anemic
and had massive hepatosplenomegaly. Polymerase chain
reaction (PCR) demonstrated homozygosity for the Coimbra
mutation (V488M), complete absence of AE1 and protein
4.2, and reduction of spectrin and ankyrin by 43% and
57%, respectively. Spherocytes and erythroblasts dominated
the peripheral blood morphology, while hyperchloremic
acidosis was identified by age three months. The patient is
being treated with an intensive transfusion regimen, chelator
therapy, and oral base replacement to counteract the acidosis.
Other than slightly delayed psychomotor development and
stable nephrocalcinosis, her condition remains stable.
In addition, hereditary renal tubular acidosis can result
from mutations in the genes encoding AE1, carbonic anhy-
drase II, or different subunits of the H+ -ATPase. Mutations
of AE1 may account for up to 70% of cases in some areas of
the world. These mutations, which affect the renal isoform of
AE1, cause defects in urinary acidification and also lead to
hypokalemia.
Rh protein
The rhesus (Rh) blood group antigens constitute the most
polymorphic family of human erythrocyte antigens and
are also the most frequent inducers of transfusion- and
parturition-related alloantibodies. Three separate proteins
make up the Rh family of proteins, including RhD and RhCE,
which are unglycosylated proteins of 417 amino acids with
97% homology, and rhesus-associated glycoprotein (RhAG),
which shows about 40% homology to RhD/CE. While scores
of individual Rh antigens are recognized, only five epitopes
(D, C, c, E, and e) carried on the RhD or CE proteins
are routinely identified; RhAG is thought to carry only the
high-frequency Duclos antigen. RhCE expresses the antithet-
ical Cc/Ee antigens, while no antithetical antigen exists for
D. However, the presence (Rh-positive) and absence (Rh-
negative) of the D antigen have sometimes been designated
as D and d, respectively. Thus, the eight most common haplo-
types associated with the Rh system are DCe, DcE, Dce, dce,
DCE, dcE, dCe, and dCE. The D antigen is a collection of
conformation-dependent epitopes along the entire RhD pro-
tein. Deletion of the RHD gene characterizes most D-negative
Caucasians, while in other populations (mainly Japanese and
black Africans) the D-negative phenotype appears to arise
from a variety of missense and nonsense mutations in oth-
erwise intact RHD genes. In addition, there are multiple
“weak D,” “D category,” and other variant D phenotypes.
These are catalogued by several websites, including Rhesus-
Base (http://www.rhesusbase.info). RhAG is a glycosylated
290 Molecular Hematology
409-amino-acid 50 kDa protein that appears to serve as a
chaperone for RhD and RhCE to the RBC membrane; Rh
antigens are only expressed if RhAG is also present.
Rh proteins appear to serve both as structural proteins
necessary for appropriate membrane conformation and cell
shape and also as transport proteins. Using known amino
acid sequences, studies using both topology prediction and
immunochemical analysis suggest that Rh proteins span the
erythrocyte membrane 12 times, with both termini in the
cytoplasm. Many multispanning membrane proteins have
a transporter function. Rh proteins, including D, CE, and
RhAG, belong to the ammonium transport (Escherichia coli
AmtB)/methylammonium permease superfamily. RhAG, as
well as the related proteins RhBG and RhCG found in non-
erythroid tissues, shares a molecular architecture much more
similar to the E. coli AmtB channel than do RhD and RhCE.
Current evidence is that RhD and RhCE proteins do not
function as transporters. However, Marini and colleagues
showed that RhAG expressed in a yeast mutant with deletions
in the three endogenous ammonium transporters (mep1Δ
mep2Δ mep3Δ) could restore growth defects elicited by low-
ammonium medium. However, this study has been criticized
because the expressed RhAG was not glycosylated and the
authors failed to show actual NH+ 4 uptake. Westhoff and
colleagues showed that Xenopus oocytes expressing solely
human RhAG protein were able to specifically transport NH+ 4 Glycophorins C and D
in a saturable, non-electrogenic, and pH-dependent man-
ner, suggesting that the RhAG transport mechanism involves
exchange of NH+ +
4 for H , resulting in the net transport of
ammonia (NH3 ) into the cell. More recent work has sug-
gested that erythroid RhAG may serve to sequester NH3 in
red cells, since it has been noted that blood plasma NH3 and
NH+ 4 levels are three times lower than the levels found in ery-
throcytes. Furthermore, in non-erythroid tissues, RhBG and
RhCG are located at specific cellular surfaces where there is
a disparate concentration of NH3 , such as at the luminal sur-
face in the kidney.
Others have argued that Amt proteins act as gas channels
handling NH3 , rather than as transporters of NH+ 4 cations,
and that both erythrocytic and non-mammalian Rh proteins
represent CO2 transporters. Soupene and colleagues found
that Rh protein expression in the green alga Chlamydomonas
reinhardtii was significantly higher under elevated CO2
conditions (air +3% CO2 for three hours) than in air alone
(0.035% CO2 ). Further, Rh mRNA and protein expression
remained low under conditions in which methylammonium
uptake was high (nitrogen-limiting, with arginine as the sole
nitrogen source instead of ammonium chloride, NH4 Cl).
Studies of CO2 transport in erythrocytes have variably sug-
gested that Rh and/or RhAG may be functionally coupled
with either AE1 or aquaporin to regulate CO2 permeability.
In oocytes, RhCG expression also appears to facilitate
CO2 permeation. In zebrafish, knockdown of Rh protein
expression affects both NH3 and CO2 excretion. Thus, to
date, the roles of RhAG, BG, and CG proteins in NH3 ,
NH4 + transport, and whether transport is electrogenic or
electroneutral (with −H+ exchange) remain unclarified. At
this point, further studies in mammalian systems will be
needed to confirm the function(s) of Rh and RhAG proteins.
Rhnull disease is characterized by the complete absence of
Rh proteins in the RBCs of affected subjects. The resulting
erythrocytes exhibit stomatocytic and spherocytic changes,
increased osmotic fragility, and deviations in ionic fluxes and
cell volume. Two variations of Rhnull disease have been iden-
tified, historically termed “amorph” and “regulator,” based
on the underlying molecular defect from which they arise.
The amorph variety involves a genetic change in the RhCE
gene coupled with deletion or inactivation of RHD, while the
regulator phenotype results from two mutant RHAG genes
(homozygote or compound heterozygote). These null pheno-
types also display marked reductions in Rh-associated mem-
brane proteins, including the LW glycoprotein (ICAM-4),
integrin-associated protein (IAP), and glycophorin B (a type
I membrane glycoprotein that bears the Ss antigens and may
serve to couple Rh with larger complexes containing AE1 and
GPA).
Glycophorins C and D (GPC/GPD) are 32 and 23 kDa mem-
brane glycoproteins arising from the same four-exon gene
(GE, also referred to as GYPC, 2q14–q21) via alternative
in-frame mRNA translation initiation sites, giving rise to
the Gerbich erythrocyte antigens. The Gerbich blood group
system currently includes six high-prevalence and five low-
prevalence antigens, of which Ge:2 and Ge:3 are the most
commonly identified and are used to define the Gerbich-
null Leach phenotype (Ge:-2,-3). GPC and GPD are identi-
cal, except that GPD has been truncated by 21 amino acids
at the N-terminus. There are two or three copies per cell of
GPC for each copy of GPD. While being infrequent targets
of alloantibodies, GPC/GPD appear to be vital membrane-
stabilizing components serving as high-affinity binding sites
for the peripheral membrane protein 4.1, which acts to affix
actin and spectrin elements. In addition to AE1–ankyrin–
spectrin, the GPC/GPD–protein 4.1–spectrin/actin bridge
constitutes the other major connection site between the ery-
throcyte membrane and its spectrin skeleton. Evidence for
the importance of GPC/GPD in membrane/cytoskeletal sta-
bility originally arose from observations in cells contain-
ing natural defects in one or more members of this bridge.
For instance, complete absence of erythrocyte GPC/GPD
results in a rare form of hereditary elliptocytosis character-
ized by impaired red cell mechanical properties. However,
these same GPC/GPD-deficient elliptocytic cells were also

found to contain reduced amounts of protein 4.1. In addi-
tion, GPC interacts with p55, a member of the membrane-
associated guanylate kinase family. The C-terminal region of
GPC and GPD interacts with the PDZ domain in p55 as part
of the ternary complex.
GPC/GPD deficiency is relatively common in Southeast
Asia. One particular variant, the Melanesian Gerbich-
negative phenotype, has been shown to reach a high
frequency (46.5%) in coastal areas of Papua New Guinea,
where Plasmodium falciparum malaria is hyperendemic,
suggesting that it may confer protection against erythrocyte
invasion by the parasite. This polymorphism is character-
ized genotypically by deletion of exon 3 (thus GPCΔex3)
and phenotypically by ovalocytic RBCs. The P. falciparum
erythrocyte-binding antigen EBA-140 appears to bind with
high affinity to GPC, and this interaction mediates a principal
invasion pathway into human erythrocytes. EBA-140 does
not bind to GPC in Ge-negative erythrocytes from Papua
residents homozygous for GPCΔex3, nor can P. falciparum
infect such cells using this invasion pathway. Using recombi-
nant P. falciparum EBA-140 expressed in Baculovirus, inves-
tigators have demonstrated that EBA-140 binding is depen-
dent on both GPC amino acid residues 36–63 as well as the
N-glycan attached to the N-terminal of GPC (but not GPD).
Interestingly, while the invasion of GPCΔex3 erythrocytes
by P. falciparum parasites is less efficient in vitro, no dif-
ferences in infection rates for either P. falciparum or Plas-
modium vivax in Ge-negative subjects have been observed to
Table 21.1 Adhesion molecules of erythrocytes
Blood group antigen Alternative name(s)
Ina /Inb CD44, In(Lu)-related p80
JMH CD108, semaphorin K1 (SEMA7A)
Lutheran (LU) B-CAM/LU, CD239
LW CD242, ICAM-4, LW
Naka (platelets) CD36 (reticulocytes only), platelet GPIV,
Naka (platelets)
Oka CD147, neurothelin, EMMPRIN
None known VLA-4 (reticulocytes only), α4β1 integrin
(CD49d/CD29)
None known CD47, integrin-associated protein (IAP)
None known CD58, lymphocyte-associated antigen-3
(LFA-3)
RAPH1 CD151, MIC2 gene product
Scianna (SC) ERMAP (erythroid membrane-associated
protein)
Functions of blood group antigens 291
date. However, subjects with other RBC polymorphisms that
are found in high frequencies in endemic malaria regions,
notably sickle cell hemoglobin (Hb S), also demonstrate
infectivity rates similar to those of subjects lacking the sickle
trait phenotype. Nevertheless, individuals with Hb S trait
show reduced parasite density and experience fewer cases of
cerebral or severe malaria. Thus, it may be that the GPCΔex3
phenotype is selected for by reducing erythrocyte invasion
rates by malaria parasites, thus lessening disease severity.
Studies to determine whether there has been natural selection
of the GPCΔex3 allele as a safeguard against severe malaria
remain to be conducted.
Blood group antigen proteins that
function as adhesion molecules
A number of erythrocyte membrane proteins have been iden-
tified as adhesion molecules. As shown in Table 21.1, these
include the proteins that bear the Lutheran, LW, and Indian
blood group antigens.
CD44 bears the Indian antigens and was first described
on erythrocytes as In(Lu)-related p80, because expression
on erythrocytes is downregulated by a dominant In(Lu)
allele, which was eventually identified by Singleton and
colleagues as any of numerous single-copy mutations in
EKLF1. CD44 was the first erythrocyte membrane protein
to be characterized as an adhesion molecule and bears
Ligand/adhesive function
Hyaluronan, possibly also fibronectin/also involved in immune
stimulation and signaling between cells.
α1β1 integrin, plexin C1/possible role in adhesion of activated
lymphocytes
Laminin containing the α5 chain, α4β1 integrin
Leukocyte, endothelial, and platelet integrins, including αVβ3,
LFA-1 (αLβ2), Mac-1 (αMβ2), αVβ1, αVβ5, αIIbβ3
Thrombospondin (platelets), LDL
Type IV collagen, fibronectin/laminin in other tissues/required for
expression of monocarboxylate (lactate) transporters
Thrombospondin, VCAM-1, possibly fibronectin
Thrombospondin, SIRP-1α
CD2
Integrins and laminin/proposed role in development and
structure of kidney and skin basement membranes
Possible adhesion molecule that localizes to points of cell–cell
contact
292 Molecular Hematology
homology to the cartilage link and proteoglycan core pro-
teins, which are known to interact with hyaluronan. This
similarity led to the demonstration that CD44 also bound
to this component of extracellular matrix and basement
membranes. On leukocytes, CD44 appears to be involved in
the interaction of leukocytes with endothelial cells, including
the homing of lymphocytes to peripheral lymphoid organs
and sites of inflammation. Ligand binding to CD44 can
also induce cytokine release and T-cell activation. However,
on erythrocytes the functional significance of CD44 is less
clear. One report has indicated that CD44 contributes to the
ability of early erythroid precursors to bind to bone marrow
fibronectin, and CD44 also contributes to red cell–leukocyte
interactions in sickle cell disease. Further, CD44 plays an
important role in the metastatic behavior of tumor cells.
The molecule that bears Lutheran blood group antigens
has been identified as a laminin receptor. Lutheran antigens
are expressed by two proteins that arise from alternate splic-
ing of a single gene. These two proteins (sometimes called
B-CAM and LU) are identical except for their cytoplasmic
domains, and they both appear to function equally well
as laminin receptors. Interestingly, B-CAM/LU expression
and function are increased on red cells from patients
with sickle cell disease, and this function can be activated
further by exposure of sickle red cells to epinephrine and
other reagents that lead to increased intracellular cyclic
adenosine monophosphate (AMP). Increase in cyclic AMP
leads to activation of protein kinase A, which phosphorylates
Table 21.2 Diverse functions of blood group antigen proteins
Blood group Alternative name(s)
Cartwright (Yt) ACHE
Chido/Rodgers (CH/RG) C4B/C4A
Colton (Co) Aquaporin-1 (AQP-1)
Cromer (Cr) Decay accelerating factor, CD55
Dombrock (Do) DO (ADP-ribosyltransferase-4 (ART4))
Duffy (Fy) DARC
Gil AQP3
Jr CD338, JR, ABCG2
Kell (K) KEL
Kidd (Jk) UT1
Knops/McCoy (Kn/McC) C3b/C4b receptor (CD35),
complement receptor type 1
Kx XK
Lan ABCB6
B-CAM/LU, thereby activating its adhesive function. Genetic
studies of red cell adhesion in patients with sickle cell disease
have shown that higher adhesion to laminin is associated
with specific polymorphisms in the genes encoding the
β2 -adrenergic receptor and adenylyl cyclase 6, one of the
more ubiquitous isoforms of adenylyl cyclase. B-CAM/LU
has also been implicated in the abnormal behavior of red
cells from patients with polycythemia vera. In this disease,
red cells are also abnormally adherent to laminin and bear
constitutively phosphorylated B-CAM/LU, most likely due
to downstream effects of activated JAK2. Finally, B-CAM/LU
has also been shown to mediate adhesion to the α4β1
integrin, also known as very late antigen 4 (VLA-4), which is
expressed by reticulocytes and leukocytes, among other cells.
However, the contribution of B-CAM/LU-mediated red cell
adhesion to the pathophysiology of vaso-occlusion or to the
frequent thrombotic complications of polycythemia remains
unproven in vivo, although B-CAM/LU adhesive activity has
been proposed as a biomarker for JAK2 mutation.
The LW protein also appears to be an adhesion receptor,
although, unlike CD44 and B-CAM/LU, LW interacts
with cell surface proteins rather than extracellular matrix
components. LW has also been called intercellular adhesion
molecule (ICAM)-4 because it is highly homologous to
other members of the ICAM family of adhesion molecules.
Like other ICAMs, LW can bind to the leukocyte integrins
CD11a/CD18 (LFA-1, αLβ2), CD11b/CD18 (MAC-1,
αMβ2), and CD11c/CD18. Recently, it has been shown that
Function
Acetylcholinesterase
C4B and C4A complement components adsorbed from plasma
Water channel
Promotes the degradation of C3 and C5 convertases
ADP-ribosyltransferase ectoenzyme
Promiscuous chemokine receptor, clears proinflammatory cytokines,
and also affects circulating neutrophil count
Water channel that also transports other small molecules, including
glycerol
Adenosine triphosphate (ATP)-dependent transport protein with broad
substrate specificity
Zinc-binding neutral endopeptidase; endothelin-3 converting enzyme
that cleaves big endothelin-3 to bioactive endothelin-3
Urea transporter important in renal urea concentrating ability
Binds C3b and C4b and facilitates immune clearance
Possibly a neurotransmitter transporter; deficiency causes
neuroacanthocytosis or McLeod syndrome
Binds heme and porphyrins and has a role in ATP-dependent uptake
into mitochondria

LW can bind to αV integrins, as well as to the αIIbβ3 integrin
of platelets. LW contributes to the adhesion of sickle red cells
to the endothelium by binding to αVβ3. This interaction
appears to be the dominant high-affinity interaction between
sickle red cells both in vitro and in vivo. Interestingly, LW
adhesive function is also activated by exposure of red cells
to epinephrine, which leads to cyclic AMP- and protein
kinase A-dependent LW phosphorylation. In addition, some
investigators speculate that LW may be important in facili-
tating the adhesion of erythroid precursors to bone marrow
macrophages in erythroid islands during erythropoiesis.
Blood group antigen proteins
with other functions
Table 21.2 lists blood group antigens associated with func-
tions other than those already discussed. As indicated, pro-
teins bearing blood group antigens have a broad diversity
of functions. Some, such as the proteins that bear the Kidd
and Colton blood group antigens, are transporters. Others,
such as those that bear the Cartwright and Kell antigens,
are ectoenzymes. In addition, erythrocytes bear receptors
for complement components and chemokines. The degree to
which polymorphisms and deficiency of these proteins con-
tribute to human disease continues to be further explored.
Summary
Proteins that bear blood group antigens have diverse func-
tions, and some proteins, such as AE1, encompass several
functions within a single protein molecule. Abnormalities of
these proteins, in the form of either deficiencies or mutations,
can lead to red cell disorders, such as hemolytic anemia, or
have more far-reaching effects, as in the association of Kx
deficiency with neuroacanthocytosis and mutations of AE1
with renal tubular acidosis. Finally, these proteins undoubt-
edly contribute both to normal physiology and to the patho-
physiology of human diseases, including sickle cell anemia,
malaria, and perhaps others.
Further reading
AE1
Anderson, R.A. and Marchesi, V.T. (1985). Regulation of the association
of membrane skeletal protein 4.1 with glycophorin by polyphospho-
inositide. Nature 318: 295–298.
Bennett-Guerrero, E., Veldman, T.H., Doctor, A. et al. (2007). Evolution
of adverse changes in stored RBCs. Proc. Natl. Acad. Sci. U.S.A. 104:
17063–17068.
Functions of blood group antigens 293
Bruce, L.J., Kay, M.M.B., Lawrence, C. et al. (1993). Band 3 HT, a human
red-cell variant associated with acanthocytosis and increased anion
transport, carries the mutation pro 868 leu in the membrane domain
of band 3. Biochem. J. 293: 317–320.
Celedon, G., Gonzalez, G., Pino, J., and Lissi, E.A. (2007). Peroxynitrite
oxidizes erythrocyte membrane band 3 protein and diminishes its
anion transport capacity. Free Radical Res. 41: 316–323.
Chu, H., Breite, A., Ciraolo, P. et al. (2008). Characterization of the
deoxyhemoglobin binding site on human erythrocyte band 3: impli-
cations for O2 regulation of erythrocyte properties. Blood 111: 932–
938.
Chu, H. and Low, P.S. (2006). Mapping of glycolytic enzyme-binding
sites on human erythrocyte band 3. Biochem. J. 400: 143–151.
Cohen, C.M., Dotimas, E., and Korsgren, C. (1993). Human erythro-
cyte membrane protein 4.2 (pallidin). Semin. Hematol. 30: 119–
137.
Davis, L., Lux, S.E., and Bennett, V. (1989). Mapping the ankyrin-
binding site of the human erythrocyte anion exchanger. J. Biol. Chem.
264: 9665–9672.
Funder, J. and Wieth, J.O. (1966). Chloride and hydrogen ion distribu-
tion between human red cells and plasma. Acta Physiol. Scand. 68:
234–245.
Galli, F., Rovidati, S., Ghibelli, L., and Canestrari, F. (1998).
S-nitrosylation of glyceraldehyde-3-phosphate dehydrogenase
decreases the enzyme affinity to the erythrocyte membrane. Nitric
Oxide 2: 17–27.
Gaston, B., May, W.J., Sullivan, S. et al. (2014). Essential role of
hemoglobin beta-93-cysteine in posthypoxia facilitation of breathing
in conscious mice. J. Appl. Phys. 116 (10): 1290–1299.
Genton, B., Al-Yaman, F., Mgone, C.S. et al. (1995). Ovalocytosis and
cerebral malaria. Nature 378: 564–565.
Grau, M., Pauly, S., Ali, J. et al. (2013). RBC-NOS-dependent
S-nitrosylation of cytoskeletal proteins improves RBC deformability.
PLoS One 8: e56759.
Guizouarn, H., Martial, S., Gabillat, N., and Borgese, F. (2007). Point
mutations involved in red cell stomatocytosis convert the electroneu-
tral anion exchanger 1 to a nonselective cation conductance. Blood
110: 2158–2165.
Gutierrez, E. and Sung, L.A. (2007). Interactions of recombinant
mouse erythrocyte transglutaminase with membrane skeletal pro-
teins. J. Membr. Biol. 219: 93–104.
Hassoun, H., Wang, Y., Vassiliadis, J. et al. (1998). Targeted inactivation
of murine band 3 (AE1) gene produces a hypercoagulable state caus-
ing widespread thrombosis in vivo. Blood 92: 1785–1792.
Hess, D.T., Matsumoto, A., Kim, S.-O. et al. (2005). Protein S-
nitrosylation: purview and parameters. Nat. Rev. Mol. Cell Biol. 6:
150–166.
Hongo, M., Haratake, M., Fuchigami, T., and Nakayam, M. (2012). A
thiol-mediated active transport of selenium by the erythroid anion
exchange 1 protein. Dalton Trans. 41: 7340–7349.
Inaba, M., Yawata, A., Koshino, I. et al. (1996). Defective anion trans-
port and marked spherocytosis with membrane instability caused by
hereditary total deficiency of red cell band 3 in cattle due to a non-
sense mutation. J. Clin. Invest. 97: 1804–1817.
Jarolim, P., Palek, J., Amato, D. et al. (1991). Deletion in erythrocyte
band 3 gene in malaria-resistant Southeast Asian ovalocytosis. Proc.
Natl. Acad. Sci. U.S.A. 88: 11022–11026.
294 Molecular Hematology
Jarolim, P., Rubin, H.L., Brabec, V. et al. (1995). Mutations of conserved
arginines in the membrane domain of erythroid band 3 protein lead
to a decrease in membrane-associated band 3 and to the phenotype
of hereditary spherocytosis. Blood 85: 634–640.
Jia, L., Bonaventura, C., Bonaventura, J. et al. (1996).
S-Nitrosohaemoglobin: a dynamic activity of blood involved in
vascular control. Nature 380: 221–226.
Khositseth, S., Sirikanerat, A., Wongbenjarat, K. et al. (2007). Distal
renal tubular acidosis associated with anion exchanger 1 mutations
in children in Thailand. Am. J. Kidney Dis. 49: 841–850.
Kindt, J.T., Woods, A., Martin, B.M. et al. (1992). Covalent alteration of
the prosthetic heme of human hemoglobin by BrCCl3 cross-linking
of heme to cysteine residue 93. J. Biol. Chem. 267: 8739–8743.
Kittanakom, S., Cordat, E., and Reithmeier, R.A. (2008). Dominant-
negative effect of Southeast Asian ovalocytosis anion exchanger 1 in
compound heterozygous distal renal tubular acidosis. Biochem. J. 410:
271–281.
Kollert-Jons, A., Wagner, S., Hubner, S. et al. (1993). Anion exchanger 1
in human kidney and oncocytoma differs from erythroid AE1 in its
NH2 terminus. Am. J. Phys. 265: F813–F821.
Ling, E., Danilov, Y.N., and Cohen, C.M. (1988). Modulation of red cell
band 4.1 function by cAMP-dependent kinase and protein kinase C
phosphorylation. J. Biol. Chem. 263: 2209–2216.
Liu, S.C., Zhai, S., Palek, J. et al. (1990). Molecular defect of the band 3
protein in Southeast Asian ovalocytosis. N. Engl. J. Med. 323: 1558–
1560.
Low, P.S., Allen, D.P., Zioncheck, T.F. et al. (1987). Tyrosine phospho-
rylation of band 3 inhibits peripheral protein binding. J. Biol. Chem.
262: 4592–4596.
McMahon, T.J., Moon, R.E., Luschinger, B.P. et al. (2002). Nitric oxide
in the human respiratory cycle. Nat. Med. 8: 711–717.
McMahon, T.J., Stone, A.E., Bonaventura, J. et al. (2000). Functional
coupling of oxygen binding and vasoactivity in S-nitrosohemoglobin.
J. Biol. Chem. 275: 16738–16745.
Miki, Y., Tazawa, T., Hirano, K. et al. (2007). Clearance of oxidized ery-
throcytes by macrophages: involvement of caspases in the genera-
tion of clearance signal at band 3 glycoprotein. Biochem. Biophys. Res.
Commun. 363: 57–62.
Nagarajan, S., Raj, R.K., Saravanakumar, V. et al. (2016). Mechanical per-
turbations trigger endothelial nitric oxide synthase activity in human
red blood cells. Sci. Rep. 6: 26935.
Ohanian, V., Wolfe, L.C., John, K.M. et al. (1984). Analysis of the ternary
interaction of the red cell membrane skeletal proteins spectrin, actin,
and 4.1. Biochemistry 23: 4416–4420.
Pawloski, J.R., Hess, D.T., and Stamler, J.S. (2001). Export by red blood
cells of nitric oxide bioactivity. Nature 409: 622–626.
Pawloski, J.R., Hess, D.T., and Stamler, J.S. (2005). Impaired vasodilation
by red blood cells in sickle cell disease. Proc. Natl. Acad. Sci. U.S.A.
102: 2531–2536.
Pawloski, J.R. and Stamler, J.S. (2002). Nitric oxide in RBCs. Transfusion
42: 1603–1609.
Pawloski, J.R., Swaminathan, R.V., and Stamler, J.S. (1998). Cell-free and
erythrocyte S-nitrosohemoglobin inhibits human platelet aggrega-
tion. Circulation 97: 263–267.
Perrotta, S., Borriello, A., Scaloni, A. et al. (2005). The N-terminal 11
amino acids of human erythrocyte band 3 are critical for aldolase
binding and protein phosphorylation: implications for band 3 func-
tion. Blood 106: 4359–4366.
Rauenbuehler, P.B., Cordes, K.A., and Salhany, J.M. (1982). Identifica-
tion of the hemoglobin binding sites on the inner surface of the ery-
throcyte membrane. Biochim. Biophys. Acta 692: 361–370.
Reinhart, A.F.R., Casey, J.R., Kalli, A.C. et al. (2016). Band 3, the human
red cell chloride/bicarbonate anion exchanger (AE1, SLC4A1), in a
structural context. Biochim. Biophys. Acta 1858: 1507–1532.
Reynolds, J.D., Ahearn, G.S., Angelo, M. et al. (2007).
S-Nitrosohemoglobin deficiency: a mechanism for loss of physio-
logical activity in banked blood. Proc. Natl. Acad. Sci. U.S.A. 104:
17058–17062.
Ribeiro, M.L., Allosio, N., Almeida, H. et al. (2000). Severe hereditary
spherocytosis and distal renal tubular acidosis associated with the
total absence of band 3. Blood 96: 1602–1604.
Shingles, R., Roh, M.H., and McCarty, R.E. (1997). Direct measure-
ment of nitrite transport across erythrocyte membrane vesicles using
the fluorescent probe, 6-methoxy-n-(3-sulfopropyl) quinolinium.
J. Bioenerg. Biomembr. 29: 611–616.
Singel, D.J. and Stamler, J.S. (2005). Chemical physiology of blood
flow regulation by red blood cells: the role of nitric oxide and
S-nitrosohemoglobin. Annu. Rev. Physiol. 67: 99–145.
Stamler, J.S., Jia, L., Eu, J.P. et al. (1997). Blood flow regulation by
S-nitrosohemoglobin in the physiological oxygen gradient. Science
276: 2034–2037.
Stamler, J.S., Toone, E.J., Lipton, S.A. et al. (1997). (S)NO signals:
translocation, regulation, and a consensus motif. Neuron 18: 691–696.
Stehberger, P.A., Shmukler, B.E., Stuart-Tilley, A.K. et al. (2007). Distal
renal tubular acidosis in mice lacking the AE1 (band3) Cl−/HCO3−
exchanger (slc4a1). J. Am. Soc. Nephrol. 18: 1408–1418.
Thevenin, B.J.M., Bicknese, S.E., Park, J. et al. (1996). Distance between
cys-201 in erythrocyte band 3 and the bilayer measured by single-
photon radioluminescence. Biophys. J. 71: 2645–2655.
Tsuneshige, A., Imai, K., and Tyuma, I. (1987). The binding of
hemoglobin to red cell membrane lowers its oxygen affinity. J.
Biochem. 101: 695–704.
Vince, J.W. and Reithmeier, R.A.F. (2000). Identification of the carbonic
anhydrase II binding site in the Cl−HCO3−anion exchanger AE1.
Biochemistry 39: 5527–5533.
von Kalckreuth, V., Evans, J.A., Timmann, C. et al. (2006). Promoter
polymorphism of the anion-exchange protein 1 associated with
severe malarial anemia and fatality. J. Infect. Dis. 194: 949–957.
Weaver, D.C., Pasternack, G.R., and Marchesi, V.T. (1984). The struc-
tural basis of ankyrin function. II. Identification of two functional
domains. J. Biol. Chem. 259: 6170–6175.
Zhang, R.L., Hess, D.T., Qian, Z. et al. (2015). Hemoglobin βCys93
is essential for cardiovascular function and integrated response to
hypoxia. Proc. Natl. Acad. Sci. U.S.A. 112: 6425–6430.
Zhang RL Hess DT, Reynolds JD, Stamler JS. (2016) Hemoglobin S-
nitrosylation plays an essential role in cardioprotection. J. Clin. Invest.
126(12): 4654–4658.
Zhou, J. and Low, P.S. (2001). Characterization of the reversible confor-
mational equilibrium in the cytoplasmic domain of human erythro-
cyte membrane band 3. J. Biol. Chem. 276: 38147–38151.
Rh
Avent, N.D., Madgett, T.E., Lee, Z.E. et al. (2006). Molecular biology of
Rh proteins and relevance to molecular medicine. Expert Rev. Mol.
Med. 8: 1–20.

Callebaut, I., Dulin, F., Bertrand, O. et al. (2006). Hydrophobic cluster
analysis and modeling of the human Rh protein three-dimensional
structures. Transfus. Clin. Biol. 13: 70–84.
Conroy, M.J., Bullough, P.A., Merrick, M., and Avent, N.D. (2005). Mod-
elling the human rhesus proteins: implications for structure and func-
tion. Br. J. Haematol. 131: 543–551.
Endeward, V., Cartron, J.P., Ripoche, P., and Gros, G. (2008). RhAG pro-
tein of the rhesus complex is a CO2 channel in the human red cell
membrane. FASEB J. 22: 64–73.
Khademi, S. and Stroud, R.M. (2006). The Amt/MEP/Rh family: struc-
ture of AmtB and the mechanism of ammonia gas conduction. Phys-
iology (Bethesda) 21: 419–429.
Le Van Kim, C., Colin, Y., and Cartron, J.-P. (2006). Rh proteins: key
structural and functional components of the red cell membrane.
Blood Rev. 20: 93–110.
Marini, A.M., Urrestarazu, A., Beauwens, R. et al. (1997). The Rh (rhe-
sus) blood group polypeptides are related to NH4+ transporters.
Trends Biochem. Sci. 22: 460–461.
Nakhoul, N.L. and Lee Hamm, L. (2013). Characteristics of mammalian
Rh glycoproteins (SLC42 transporters) and their role in acid-base
transport. Mol. Aspects Med. 34: 629–637.
Nicolas, V., Mouro-Chanteloup, I., Lopez, C. et al. (2006). Functional
interaction between Rh proteins and the spectrin-based skeleton in
erythroid and epithelial cells. Transfus. Clin. Biol. 13: 23–28.
Perry, S.F., Braun, M.H., Noland, M. et al. (2010). Do zebrafish Rh pro-
teins act as dual ammonia-CO2 channels? J. Exp. Zool. Ecol. Genet.
Physiol. 313: 618–621.
Planelles, G. (2007). Ammonium homeostasis and human rhesus glyco-
proteins. Nephron Physiol. 105: 11–17.
Soupene, E., Inwood, W., and Kustu, S. (2004). Lack of the rhesus protein
Rh1 impairs growth of the green alga Chlamydomonas reinhardtii at
high CO2 . Proc. Natl. Acad. Sci. U.S.A. 101: 7787–7792.
Soupene, E., King, N., Feild, E. et al. (2002). Rhesus expression in a green
alga is regulated by CO2 . Proc. Natl. Acad. Sci. U.S.A. 99: 7769–7773.
Szekely, D., Chapman, B.E., Bubb, W.A., and Kuchel, P.W. (2006). Rapid
exchange of fluoroethylamine via the rhesus complex in human ery-
throcytes: 19F NMR magnetization transfer analysis showing compe-
tition by ammonia and ammonia analogues. Biochemistry 45: 9354–
9361.
Westhoff, C.M. (2007). The structure and function of the Rh antigen
complex. Semin. Hematol. 44: 42–50.
Westhoff, C.M., Ferreri-Jacobia, M., Mak, D.O., and Foskett, J.K. (2002).
Identification of the erythrocyte Rh blood group glycoprotein as
a mammalian ammonium transporter. J. Biol. Chem. 277: 12499–
12502.
Glycophorins C and D
Chang, S.H. and Low, P.S. (2001). Regulation of the glycophorin C–
protein 4.1 membrane-to-skeleton bridge and evaluation of its contri-
bution to erythrocyte membrane stability. J. Biol. Chem. 276: 22223–
22230.
Colin, Y., Le Van Kim, C., Tsapis, A. et al. (1989). Human erythrocyte
glycophorin C. Gene structure and rearrangement in genetic variants.
J. Biol. Chem. 264: 3773–3780.
Diakowski, W., Grzybek, M., and Sikorski, A.F. (2006). Protein 4.1, a
component of the erythrocyte membrane skeleton and its related
Functions of blood group antigens 295
homologue proteins forming the protein 4.1/FERM superfamily.
Folia Histochem. Cytobiol. 44: 231–248.
Gascard, P. and Cohen, C.M. (1994). Absence of high-affinity band 4.1
binding sites from membranes of glycophorin C- and D-deficient
(Leach phenotype) erythrocytes. Blood 83: 1102–1108.
Hemming, N.J., Anstee, D.J., Mawby, W.J. et al. (1994). Localization of
the protein 4.1-binding site on human erythrocyte glycophorins C
and D. Biochem. J. 299: 191–196.
Kusunoki, H. and Kohno, T. (2007). Structural insight into the inter-
action between the p55 PDZ domain and glycophorin C. Biochem.
Biophys. Res. Commun. 359: 972–978.
Lobo, C.A., Rodriguez, M., Reid, M., and Lustigman, S. (2003). Gly-
cophorin C is the receptor for the Plasmodium falciparum erythro-
cyte binding ligand PfEBP-2 (BAEBL). Blood 101: 4628–4631.
Maier, A.G., Duraisingh, M.T., Reeder, J.C. et al. (2003). Plasmodium
falciparum erythrocyte invasion through glycophorin C and selection
for Gerbich negativity in human populations. Nat. Med. 9: 87–92.
Mayer, D.C., Jiang, L., Achur, R.N. et al. (2006). The glycophorin C N-
linked glycan is a critical component of the ligand for the Plasmodium
falciparum erythrocyte receptor BAEBL. Proc. Natl. Acad. Sci. U.S.A.
103: 2358–2362.
Patel, S.S., Mehlotra, R.K., Kastens, W. et al. (2001). The association of
the glycophorin C exon 3 deletion with ovalocytosis and malaria sus-
ceptibility in the Wosera, Papua New Guinea. Blood 98: 3489–3491.
Pinder, J.C., Chung, A., Reid, M.E. et al. (1993). Membrane attachment
sites for the membrane cytoskeletal protein 4.1 of the red blood cell.
Blood 82: 3482–3488.
Rydzak, J., Kaczmarek, R., Czerwinski, M. et al. (2015). The baculovirus-
expressed binding region of Plasmodium falciparum EBA-140 ligand
and its glycophorin C binding specificity. PLoS One 10: e0115437.
Takakuwa, Y., Tchernia, G., Rossi, M. et al. (1986). Restoration of nor-
mal membrane stability to unstable protein 4.1-deficient erythrocyte
membranes by incorporation of purified protein 4.1. J. Clin. Invest.
78: 80–85.
Telen, M.J., Le Van Kim, C., Chung, A. et al. (1991). Molecular basis for
elliptocytosis associated with glycophorin C and D deficiency in the
Leach phenotype. Blood 78: 1603–1606.
Adhesion proteins
Bailly, P., Tontti, E., Hermand, P. et al. (1995). The red cell LW blood
group protein is an intercellular adhesion molecule which binds to
CD11/CD18 leukocyte integrins. Eur. J. Immunol. 25: 3316–3320.
Brittain, J.E., Han, J., Ataga, K.I. et al. (2004). Mechanism of CD47-
induced alpha4beta1 integrin activation and adhesion in sickle retic-
ulocytes. J. Biol. Chem. 279: 42393–42402.
De Grandis, M., Cambot, M., Wautier, M.P. et al. (2013). JAK2V617F
activates Lu/BCAM-mediated red cell adhesion in polycythemia vera
through an EpoR-independent Rap1/Akt pathway. Blood 121: 658–
665.
De Grandis, M., Cassinat, B., Kiladjian, J.J. et al. (2015). Lu/BCAM-
mediated cell adhesion as biological marker of JAK2V617F activity
in erythrocytes of polycythemia vera patients. Am. J. Hematol. 90:
E137–E138.
El Nemer, W., Rahuel, C., Colin, Y. et al. (1997). Organization of the
human LU gene and molecular basis of the Lu(a)/Lu(b) blood group
polymorphism. Blood 89: 4608–4616.
296 Molecular Hematology
El Nemer, W., Wautier, M.P., Rahuel, C. et al. (2007). Endothe-
lial Lu/BCAM glycoproteins are novel ligands for red blood cell
alpha4beta1 integrin: role in adhesion of sickle red blood cells to
endothelial cells. Blood 109: 3544–3551.
Eyler, C.E., Jackson, T., Elliott, L.E. et al. (2008). Beta2-adrenergic recep-
tor and adenylate cyclase gene polymorphisms affect sickle red cell
adhesion. Br. J. Haematol. 141: 105–108.
Hermand, P., Gane, P., Huet, M. et al. (2003). Red cell ICAM-4 is a novel
ligand for platelet-activated alpha IIbbeta 3 integrin. J. Biol. Chem.
278: 4892–4898.
Hines, P.C., Zen, Q., Burney, S.N. et al. (2003). Novel epinephrine and
cyclic AMP mediated activation of B-CAM/LU-dependent sickle (SS)
RBC adhesion. Blood 101: 3281–3287.
Lee, G., Lo, A., Short, S.A. et al. Targeted gene deletion demonstrates
that the cell adhesion molecule ICAM-4 is critical for erythroblastic
island formation. Blood 108: 2064–2071.
Mankelow, T.J., Burton, N., Stefansdottir, F.O. et al. (2007). The Laminin
511/521-binding site on the Lutheran blood group glycoprotein is
located at the flexible junction of Ig domains 2 and 3. Blood 110: 3398–
3406.
Murphy, M.M., Evans, A., Zayed, M.A. et al. (2005). Role of Rap1 in
promoting sickle red blood cell adhesion to laminin via BCAM/Lu.
Blood 105: 3322–3329.
Singleton, B.K., Burton, N.M., Green, C. et al. (2008). Mutations in
EKLF/KLF1 form the molecular basis of the rare blood group In(Lu)
phenotype. Blood 112: 2081–2088.
Telen, M.J., Eisenbarth, G.S., and Haynes, B.F. (1983). Human erythro-
cyte antigens. Regulation of expression of a novel erythrocyte sur-
face antigen by the inhibitor Lutheran In(Lu) gene. J. Clin. Invest. 71:
1878–1886.
Telen, M.J., Udani, M., Washington, M.K. et al. (1996). A blood group-
related polymorphism of CD44 abolishes a hyaluronan-binding con-
sensus sequence without preventing hyaluronan binding. J. Biol.
Chem. 271: 7147–7153.
Udani, M., Zen, Q., Cottman, M. et al. (1998). Basal cell adhesion
molecule/Lutheran protein: the receptor critical for sickle cell adhe-
sion to laminin. J. Clin. Invest. 101: 2550–2558.
Verfaillie, C.M., Benis, A., Iida, J. et al. (1994). Adhesion of commit-
ted human hematopoietic progenitors to synthetic peptides from
the C-terminal heparin-binding domain of fibronectin: cooperation
between the integrin alpha 4 beta 1 and the CD44 adhesion receptor.
Blood 84: 1802–1811.
Wautier, M.P., El Nemer, W., Gane, P. et al. (2007). Increased adhesion
to endothelial cells of erythrocytes from patients with polycythemia
vera is mediated by laminin alpha5 chain and Lu/BCAM. Blood 110:
894–901.
Zen, Q., Batchvarova, M., Twyman, C.A. et al. (2004). B-CAM/LU
expression and the role of B-CAM/LU activation in binding of low-
and high-density red cells to laminin in sickle cell disease. Am. J.
Hematol. 75: 63–72.
Zen, Q., Cottman, M., Truskey, G. et al. (1999). Critical factors in basal
cell adhesion molecule/Lutheran-mediated adhesion to laminin.
J. Biol. Chem. 274: 728–734.
Zennadi, R., Hines, P.C., De Castro, L.M. et al. (2004). Epinephrine
acts via erythroid signaling pathways to activate sickle cell adhe-
sion to endothelium via LW–αvβ3 interactions. Blood 104: 3774–
3781.
Zennadi, R., Moeller, B.J., Whalen, E.J. et al. (2007). Epinephrine-
induced activation of LW-mediated sickle cell adhesion and vaso-
occlusion in vivo. Blood 110: 2708–2717.
Other blood group antigen proteins
Afenyi-Annan, A., Kail, M., Combs, M.R. et al. (2008). Lack of Duffy
antigen expression is associated with organ damage in patients with
sickle cell disease. Transfusion 48: 917–924.
Andolfo, I., Russo, R., Gambale, A., and Iolascon, A. (2016). New
insights on hereditary erythrocyte membrane defects. Haematolog-
ica 101: 1284–1294.
Hara-Chikuma, M. and Verkman, A.S. (2005). Aquaporin-3 func-
tions as a glycerol transporter in mammalian skin. Biol. Cell 97:
479–486.
Karamatic Crew, V., Burton, N., Kagan, A. et al. (2004). CD151, the first
member of the tetraspanin (TM4) superfamily detected on erythro-
cytes, is essential for the correct assembly of human basement mem-
branes in kidney and skin. Blood 104: 2217–2223.
Ogasawara, K., Osabe, T., Suzuki, Y. et al. (2015). A new ABCG2 null
allele with a 27-kb deletion including the promoter region causing
the Jr(a-) phenotype. Transfusion 55: 1467–1471.
Peng, J., Redman, C.M., Wu, X. et al. (2007). Insights into extensive dele-
tions around the XK locus associated with McLeod phenotype and
characterization of two novel cases. Gene 392: 142–150.
Tournamille, C., Filipe, A., Badaut, C. et al. (2005). Fine mapping of the
Duffy antigen binding site for the Plasmodium vivax Duffy-binding
protein. Mol. Biochem. Parasitol. 144: 100–103.
Zhang, W., Zitron, E., Homme, M. et al. (2007). Aquaporin-1 channel
function is positively regulated by protein kinase C. J. Biol. Chem. 282:
20933–20940.
Chapter 22 Autoimmune hematological
disorders
Drew Provan1 & John W. Semple2
1
Blizard Institute, Barts and The London School of Medicine and Dentistry, Queen Mary University of London, London, UK
2
Division of Hematology and Transfusion Medicine, Lund University, Lund, Sweden
Introduction, 297
The immune system, 297
The spectrum of autoimmune diseases, 304
Role of genetic factors, 304
Mouse models of autoimmune disease, 305
Human studies, 306
Introduction
Autoimmune diseases are disorders where antibodies or cells
react against self antigens to cause disease, at which point an
adaptive immune response is mounted against the self anti-
gen or antigens. This results in clearance of the antigen from
the body. The normal adaptive response results in complete
removal of the non-self antigen, whereas in autoimmune
disease there is incomplete clearance of the antigen, which
leads to perpetuation of the immune response. Autoimmune
disorders occur in about 5% of the population, although
many individuals have no symptoms. In all, there are more
than 70 different disorders, most of which are uncommon,
apart from rheumatoid disease and autoimmune thyroiditis.
Autoimmune diseases are clinical syndromes mediated by
activation of T and B lymphocytes, in the absence of infection
or other discernible cause. Until recently, although we could
describe the pathological features of autoimmune disease, we
had little idea as to their actual cause. Through the develop-
ment of animal models and the identification of target genes,
we have gained considerable insight into the pathogenetic
basis of these complex diseases. Autoreactive cells may affect
virtually any body tissue, including blood, and blood disor-
ders in which autoantibodies are found include cytopenias
such as autoimmune hemolytic anemia, immune thrombo-
cytopenia (ITP), and autoimmune neutropenia, in addition
to coagulation disorders such as acquired hemophilia.
Although autoimmune diseases are clinically and patho-
logically diverse, the common end result is damage to
Molecular Hematology, 4th Edition. Edited by Drew Provan and John G. Gribben.
© 2020 John Wiley & Sons Ltd. Published 2020 by John Wiley & Sons Ltd.
Immune thrombocytopenia as a model of autoimmune blood disease, 308
Targeted versus untargeted therapies for autoimmune disease, 311
Novel therapies for the treatment of other autoimmune diseases, 314
Conclusions, 315
Further reading, 315
autoantigen-containing tissues, which may result in disease.
Factors that play a role in this process include immune dys-
regulation, genetic factors, and environmental factors as trig-
gering events. We discuss all of these after briefly reviewing
the structure and function of the immune system.
The immune system
The immune system comprises cells and molecules whose
main role is defense against invading pathogens. The two
principal components are the innate immune system, com-
prising skin, mucous membranes, neutrophils, macrophages/
dendritic cells that serve as antigen-presenting cells (APCs),
and other scavenging cells, in addition to the complement
system and natural killer (NK) cells; and the adaptive
immune system, which involves exclusively B and T lympho-
cytes (Figure 22.1). The B cells are responsible for differenti-
ating into plasma cells and secreting antibody, aided by T cells
that can also act as effector cells. Key features of the adaptive
system include antigen receptor diversity, antigen specificity,
and immunological memory. This is in sharp contrast to the
innate system, which lacks these features.
The innate immune system
Despite varied challenges by many antigens, the innate sys-
tem lacks the ability to develop immunological memory;
however, this system’s responses remain constitutive and the
same throughout life. In evolutionary terms, the innate sys-
tem probably developed before the adaptive system. The
innate immune system is composed of physical, chemical,
and cellular components that act together to mediate the
297
298 Molecular Hematology

Microbes
Innate immunity Adaptive immunity

Epithelial B lymphocytes Antibodies

barriers

Phagocytes Effector T cells

T lymphocytes

NK cells

0 6 12 1 3 5
Hours Days
Time after infection
Fig. 22.1 Innate and adaptive immune systems. The innate system comprises physical barriers (e.g. skin) along with scavenger cells, while the
adaptive system comprises B and T lymphocytes. Temporarily, the innate system is the immediate first line of defense, but lacks specificity. The
adaptive system comes into play later and possesses immunological specificity and memory. NK, natural killer.
Source: Abbas A.K., Lichtman A.H., Pober J.S. (2000). Cellular and Molecular Immunology, 4th edn. © 2000. Reproduced with permission of
Elsevier.

first line of defense against invading microorganisms. These
components are intimately linked with inflammatory pro-
cesses and ultimately lead to the removal of most of the infec-
tious organisms encountered by a host. The innate immune
response activates quickly (within seconds) after exposure
of foreign infectious agents and is antigen non-specific, in
that there is no specific memory associated with the immu-
nity. These characteristics distinguish the innate immune
system from the adaptive immune response, which is medi-
ated exclusively by B and T cells, is slower to activate, and is
exquisitely antigen specific, generating memory with subse-
quent exposure to the stimulating antigen.
APCs are key players of innate immunity and physically
link innate and adaptive immune responses by presenting
antigens to T cells. APCs additionally possess surface
receptors for antibody (immunoglobulin) and complement.
Microorganisms opsonized by antibody and/or complement
molecules are recognized by specific receptors (e.g. Fc
receptors) on phagocytes (APC) and are phagocytosed and
broken down within the interior of the APCs. Within cells
such as neutrophils, killing and digestion of the pathogen
involve the generation of superoxide and hydroxyl radicals,
nitric oxide, and proteolytic enzymes. In addition to the
removal of pathogens, the innate system also plays a role in
the removal of dead cells and remodeling of tissue during
healing. Cells undergoing programmed cell death (apopto-
sis) express molecules such as phosphatidylserine on their
surface, targeting their removal. Dendritic cells also play a
key role in innate immunity, and activation of dendritic cells
occurs following exposure to heat-shock proteins, interferon
(IFN)-α, and other stimuli. Dendritic cells are professional
APCs that are able to migrate to lymph nodes, process
antigens, and present them to T cells in conjunction with
major histocompatibility complex (MHC) molecules, of
which there are two classes, class I and class II.
Complement
Because of limitations of space, the role of complement is
only very briefly discussed here. Following infection, cells of
the innate immune system migrate toward the affected site.
Complement component C3b coats the pathogen surface,
whereas the molecules C3a, C4a, and C5a act as neutrophil
chemoattractants, which can then trigger mast cells to release
histamine. This enhances vasodilation and permeability and
thus increased blood flow and vessel permeability, allowing
neutrophils to pass through the blood vessel walls more eas-
ily, an essential requirement for an effective innate response.
The complement system is a part of the innate immune
system and has a major function in enhancing the ability of
antibodies and phagocytes to clear infectious agent and dam-
aged cells. In addition, complement promotes inflammation
and initiates the development of the membrane attack com-
plex on the cell membrane, which leads to pore formation and
osmotic lysis of the cell. The complement system consists of
a number of liver-derived small proteins which are inactive
zymogens (inactive proteases) that circulate in the plasma.
Over 30 proteins and protein fragments make up the com-
plement system and they account for approximately 10% of
the globulin fraction of blood (Figure 22.2).
 COMPLEMENT SYSTEM
In order to avoid excessive spontaneous amplification of the complement system, Decay
the body has some regulatory mechanisms in place (gray diamonds) Accelerating
Factor (DAF)
Spontaneous Acts as an
Spontaneous cleavage of anaphylatoxin
Factor B Factor D decay thioester bond
(free in plasma) (free in plasma) C3a Action of C3 (purple)
(smaller)
Increases
vascular
permeability
Alternative pathway (red) Factor B binds C3b
[preferentially to C3b bound to Activation of C3 found
Factor D ‘Complement fixation’ ‘Opsonisation’
cellular membrane of pathogen C3bBb freely in blood plasma
cleaves Factor B Binds firmly by covalent Tags pathogen for
(during infection) or C3b convertase by proteolytic
bound to C3b bonding to reactive phagocytosis as
bound to collagen or basement cleavage
membrane (during injury)] molecules on pathogen phagocytes have C3b
(usually carbohydrates) receptors
C3b
(larger)

C3bBbC3b
(C5 convertase)
Pathogen Lectin pathway (green)
surface presents
mannose Acts as an
Binding of MBL to C5a
anaphylatoxin
mannose on
surface of pathogen C4bC2a Cleaves
Mannose C-5
convertase
Binding Loctin
Acute Phase (MBL) Common pathway for classical and
Response C5b
leading lectin systems (blue)
to increase in
production of
acute phase C4b Opsonisation
proteins by Initiates Hydrophobic
up to x 1000 Membrane regions of
CRP binds
C-Reactive phosphocholine CRP then binds Attack Complex C6,7 and 8
protein component of C1 present in (MAC) become
C2a
(CRP) lipopolysaccharides in plasma formation exposed and
Proteolytic by binding bind to cell
cell wall of pathogen cleavage of with C6, membrane
C4 and C2 C7 and C8 of target cell
Classical pathway (orange) C2b

Immunoglobins C1 activated C9 is
from the adaptive C4a Acts as an recruited and Leakage of
immune response anaphylatoxin polymerises cytoplasm
(at a later stage) on the C5b678 leading
complex to to death
form a large of target
pore in the cell
Membrane Attack membrane
Antibody binding to Complex
target antigen on
pathogen surface
CD59 found in
human cell
membrane
Alternative pathway acts first. Classical and Lectin pathways rely on the activation of Acute Phase Resonse and hence take longer.

Fig. 22.2 The complement system.

Source: https://commons.wikimedia.org/wiki/File:Complement system.jpg#filelinks
Autoimmune hematological disorders 299
300 Molecular Hematology
The adaptive immune system
Two requirements of an effective immune system are (i) the
ability to recognize millions of potential antigens; and (ii)
the prevention of self-reacting lymphocytes from causing
tissue damage. The former is achieved through irreversible
somatic recombination of immunoglobulin and T-cell recep-
tor (TCR) genes, generating many millions of different anti-
body and TCR molecules.
B and T cells possess antigen receptors
on their surface
The antigen receptor of the B cell is an immunoglobulin and
that of the T cell is the TCR. These molecules are expressed
on their respective cell surfaces and interact with antigen,
either as native (immunoglobulin) or processed (TCR)
antigens. The TCR is a transmembrane protein and consists
of a heterodimer of either αβ or γδ subunits. TCRs, like
immunoglobulins, contain hypervariable regions, and in
evolutionary terms the two receptors are probably related.
Unlike TCRs, antibody molecules are both expressed on the
B-cell surface and secreted into body fluids. One feature that
both receptors share is the ability to generate enormous diver-
sity through recombination of germline variable (V), diver-
sity (D), and joining (J) region segments, in addition to ran-
dom mutations within the rearranged genes, although only
immunoglobulins undergo somatic mutation. Immunoglob-
ulin molecules possess two key regions: the hypervariable
region (antigen binding) and the Fc portion at the C-terminal
end which, as already outlined, is recognized by the Fc recep-
tor (e.g. FcγR) on macrophages. Antibodies may be of the
immunoglobulin (Ig)G, IgA, IgM, IgD, or IgE class, with
subclasses within some of the groups (e.g. there are four IgG
subclasses and two IgA subclasses). Having such enormous
diversity ensures that there is an antibody for almost every
potential antigen encountered, but the downside of this
extreme diversity is that antibodies are generated that can
recognize self antigens (autoantibodies), and it is likely that in
normal healthy subjects autoantibodies are generated against
a wide variety of antigenic targets. Since autoimmune disease
is not common, there must exist a mechanism for removing
self-reacting antibodies. In effect, an immunological lack of
responsiveness or tolerance must exist, whereby self-reactive
cells are prevented from causing damage. Recent research has
shown this to be the case.
The major histocompatibility complex
Class I MHC molecules comprise human leukocyte antigen
(HLA)-A, -B, and -C, and class II molecules consist of HLA-
DP, -DQ, and -DR. Class II molecules are responsible for the
presentation of peptide antigens to the TCR on CD4+ helper
T cells, whereas class I molecules present peptide antigens to
CD8+ T cells.
NK cells and others such as neutrophils have recep-
tors for the immunoglobulin Fc region and are responsible
for antibody-dependent cellular cytotoxicity following FcγR
linkage of NK cells and antibody-opsonized targets. In addi-
tion, NK cells can effect killing using killer-activating recep-
tors, which recognize specific molecules on nucleated cells.
An inhibitory molecule (killer inhibitory) recognizes MHC
class I on nucleated cells, preventing killing, but if MHC
class I is lost (e.g. during infection of the cell by virus or after
malignant transformation), the nucleated cell is recognized
as being abnormal and is therefore targeted for destruction.
Soluble molecules: cytokines orchestrate
the immune response
The innate system relies on a complex network of soluble
molecules, such as cytokines and complement components,
which coordinate the entire immune response. We discuss
these briefly here, since they are implicated in the pathogen-
esis of autoimmune disease.
Cytokines are mediators secreted by cells that influence the
behavior of other cells. Most cytokines are soluble, apart from
interleukin (IL)-1 and tumor necrosis factor (TNF)-α, which
also have membrane-bound forms. Cytokines are small pro-
tein molecules of around 15–25 kDa whose actions include
promotion of cell growth, cellular communication, inflam-
mation, immunity, and repair of tissues. These molecules
are responsible for the regulation and orchestration of the
entire immune response. Their effects are short-lived and
their actions are largely local. In order to exert their effect,
cytokines interact with specific receptors and promote signal
transduction. Their main routes of action are via the Janus
kinase (JAK)/STAT and Ras/MAP pathways. The cytokine
profile may be pro-inflammatory or anti-inflammatory and
the cytokine balance will dictate whether a helper T-cell
clone engages and differentiates into a Th1 (e.g. interferon-
mediated) or a Th2 (e.g. IL-4- mediated) response. In general,
Th1 responses are effective against intracellular pathogens
and Th2 responses aid B cells and type I hypersensitivity
reactions.
Cytokine profiles of Th1 and Th2 responses
APCs, and in particular dendritic cells, are responsible for
T-cell differentiation toward the Th1 or Th2 phenotype; the
cytokines interferon and IL-12 play key roles in the Th1
response, whereas IL-4 mediates Th2 responses (Table 22.1).
Since cytokines play such a key role in orchestrating an
effective immune response, it is likely that dysregulation
of cytokine levels may induce an autoimmune response in
some disorders. This has been shown to be the case in
 Autoimmune hematological disorders 301

experimental models and also in human disease. For exam-

Table 22.1 Cytokine profiles of Th1 and Th2 responses
ple, transfection of the gene for IFN-γ on the insulin pro-
Th1 response Th2 response moter has been shown to induce inflammation within the
(activates macrophages) (deactivates macrophages) pancreas, with aberrant expression of MHC class II and
the development of diabetes. In addition, proinflammatory
IL-2 – cytokines, such as IL-12, TNF, and IFN-γ, can induce organ-
IL-3 – specific autoimmunity.
– IL-4
– IL-5
T cells
– IL-6
– IL-10 T cells develop within the thymus (Figure 22.3) and T cells
– IL-13 bearing αβ TCRs recognize processed antigen presented to
TNF-α – them by APCs, including dendritic cells. Within the thymus,
TNF-β TNF-β T cells are selected in order to prevent autoreactivity; that is
IFN-γ –
to say, mechanisms exist whereby T cells are prevented from
GM-CSF GM-CSF
reacting with self antigens. Positive selection occurs when T
These are the principal cytokines involved in generation of Th1 and
cells express TCRs that are able to interact with MHC com-
Th2 responses. Imbalance in Th1 or Th2 cytokines may play a role plexes on thymic cortex epithelial cells. The effect of positive
in the development of autoimmune disease, allergy, and other dis- selection is to prevent apoptosis. T cells expressing TCRs with
orders. high or low affinity for self molecules are negatively selected
and undergo apoptosis. The end result is that the thymus
releases only those T cells whose TCRs have a medium reac-
tivity against peptide antigens, and are thus are quiescent
until they encounter their specific peptide.

Cortex Medulla
MHC Dendritic cell
CD8 or macrophage
Cortical
epithelial cell
T cell
T-cell
receptor T-cell MHC
receptor
CD4
Medulla Positive CD8 or CD4
T cell selection
Cortex T cell
Positive
CD8 and CD4 Apoptosis selection
Negative
selection CD8 or CD4
T cell
T cell

Surviving
Apoptosis cells leave the
thymus

Thymus gland
Fig. 22.3 T-cell selection within the thymus. T cells undergo positive and negative selection. CD4+ CD8+ T cells interact with major
histocompatibility complex (MHC)–peptide complexes. Depending on the strength of the interaction, the T cells either undergo apoptosis (the
majority) or survive and leave the thymus (the minority).
Source: Delves P.J., Roitt I.M. (2000). The immune system. First of two parts. New England Journal of Medicine 343: 37–49. Reproduced with
permission of New England Journal of Medicine.
302 Molecular Hematology
Immunological tolerance prevents damage
to self antigens
Tolerance defines the body’s ability to recognize, but not react
with, self antigens while retaining the ability to respond to
non-self antigens. This process involves the negative selec-
tion of T cells, as already outlined. In addition, the process
involves the control of autoreactive B cells. The process of
selecting T cells and B cells in the thymus and bone marrow
respectively is known as central tolerance. For autoreactive
lymphocytes that escape into the periphery, there are addi-
tional peripheral tolerance mechanisms to provide a safety net
for unwanted autoreactivity.
T-cell tolerance: central mechanisms
Immature T cells from the bone marrow migrate to the thy-
mus, where they complete their development. The T cells
within the thymus interact with MHC molecules in low- or
high-affinity interactions. If the TCRs have low affinity for
the peptide (e.g. self peptide), they receive apoptotic signals
and die within the thymus. T cells participating in high-
affinity interactions have a similar fate, and it is only when the
interaction is of intermediate affinity that the T cells survive
and migrate to the periphery, a process termed positive selec-
tion. In general, positive selection occurs when CD4+ CD8+
T cells interact with TCR–MHC–peptide complexes. For
most T cells the interaction is of low avidity and the T cells
die before leaving the thymus. A minority of CD4+ CD8+
T cells have intermediate avidity reactions and hence these
cells survive, after which they mature into either CD4+ CD8−
or CD4− CD8+ cells. CD4+ T cells are the main effectors of
autoimmune disease.
T-cell tolerance: peripheral mechanisms
From studies of animal models, it is apparent that self-
reactive lymphocytes are present peripherally, having
escaped central tolerance checkpoints. Furthermore, some
form of immunological ignorance is invoked whereby such
circulating autoreactive T cells fail to respond to the specific
self antigen. The mechanisms for such ignorance are not
fully characterized, but may involve low levels of circulating
antigen (i.e. below a critical threshold), or the antigen may
be in a separate compartment from the autoreactive cells
(e.g. the blood–brain barrier), or there may be an absence of
the costimulatory signals required for T-cell costimulation
and activation.
T-cell activation: role of costimulatory molecules
Antigen reaches the lymphoid tissue via the lymphatic system
or, in some cases, within dendritic cells that have ingested and
processed antigen. In general, antigens in blood are taken to
the spleen and tissue antigen is taken to the lymph nodes.
Antigen is then processed and presented on MHC molecules
by two major routes: (i) the antigen may be produced within
the cell itself, for example viral antigens may be expressed
within cells and complexed with MHC class I molecules; or
(ii) APCs may take up antigen exogenously and then phago-
cytose, process, and express the antigen on MHC class II
molecules.
Recognition of antigen by T cells
Antigen recognition differs between CD4+ and CD8+ T cells.
For example, CD4+ T cells can only recognize peptide anti-
gens on class II molecules, whereas CD8+ T cells recognize
antigen in association with MHC class I molecules. In this
way the MHC dictates the type of response generated.
TCR signaling
TCR molecules are found on the T-cell surface complexed
with CD3 molecules. CD3 transmits signals intracellularly,
with tyrosine phosphorylation of residues in the CD3 cyto-
plasmic tail. This transmits signals to the nucleus, resulting
in T-cell proliferation. Co-receptor molecules on T cells are
important, since they play a major role in T-cell activation
following engagement of the TCR. The generation of a
specific immune response to a pathogen requires the recog-
nition of the foreign agent by specialized cells in the body,
presentation of the foreign antigen to T cells, and orchestra-
tion of the subsequent immune response. T cells must receive
two signals from the APCs for a complete immune response
to occur. The first signal involves the presentation of antigens
in the context of the MHC on the surface of the APC to
the TCR complex on the T cells (antigen-specific signal 1).
This interaction provides the specificity of the immune
response. This signal alone is not sufficient to trigger an
optimal immune response and a second signal is required.
The second, or costimulatory, signal involves the interaction
of B7 on the APC with CD28 receptors on the T cells. The
interaction of B7 with CD28 is required for the T cells to
proliferate, produce cytokines, and provide help for the
induction of antibody- and cell-mediated immune responses
directed at the original antigen. The B7-induced CD28 signal
is delivered to the T-cell nucleus via several intermediate
steps, and results in the direct induction of the IL-2 gene and
the receptor for this cytokine. T cells stimulated via their
TCR (antigen-specific signal) and the costimulatory signal
proliferate, produce many cytokines, and direct the specific
immune response against the presented antigen.
The principal co-receptors on APCs are CD80 (B7-1),
CD86 (B7-2), and CD40, binding respectively to CD28, cyto-
toxic T-lymphocyte antigen (CTLA)-4, and CD40 ligand on
 Autoimmune hematological disorders 303

Antigen specific
CD4+ T cell recognition APC (DC or B cell)

CD4

1 TCR MHC II
Fig. 22.4 Interaction between antigen-presenting Protection from
cells (APCs; e.g. dendritic cells or B cells) and CD4+ apoptosis
T cells. The first signal to the T cell is via major Proliferation
Antibody
histocompatibility complex (MHC) and T-cell receptor IL-2 production CD40L CD40
production
(TCR). This signal is not sufficient for proliferation and
IL-2 production and a second signal, such as interaction 2 APC function
CD80
of B7-1 or B7-2 with CD28, is required. If no second CD28
CD86
signal is received, the T cells undergo apoptosis.
Courtesy of Dr Masataka Kuwana, Institute for Costimulation
Advanced Medical Research, Keio University School of
CD152 (CTLA-4) CD80 T cell anergy
Medicine, Tokyo.

T cells. Dendritic cells express large quantities of both B7
and CD40, are professional APCs, and are the most effective
stimulators of naı̈ve T cells. Following activation, the T cells
undergo clonal expansion, generating effector cells that move
toward the site of inflammation after leaving the lymphoid
tissue (Figure 22.4).
might otherwise cause. We now know that low-level autore-
activity is found in all healthy individuals and that autoanti-
gens play a role in generating our normal immune reper-
toire. Self and non-self antigens have few differences and
Generation of immune repertoires

Mechanisms for controlling autoreactive T cells Bone marrow

Thymus
Harmful autoreactive T cells may be eliminated by several
mechanisms, including apoptosis, anergy, inhibition, clonal Central tolerance
deletion, and suppression. Anergy refers to T cells that do not Anti-self lymphocytes deleted by apoptosis
produce IL-2 after their encounter with antigen. Such cells (negative selection)
are not activated and may produce IL-10, which further sup-
presses T-cell activation. Inhibition of T cells is mediated by
Leakage of anti-self lymphocytes to periphery
CTLA-4, also known as CD152. This molecule binds to the
B7 family of receptors (B7-1 and B7-2) on APCs, including
Controlled by
B cells and dendritic cells, with higher affinity than to CD28,
a counter-receptor for B7. CTLA-4 mediates suppression of
the T cell. Peripheral tolerance

Loss of immunological tolerance leads to autoimmunity
In order for autoimmune disease to occur, there must be loss
of tolerance to self antigens. This occurs despite central and
peripheral T-cell tolerance controls. Possible mechanisms
leading to loss of tolerance include variability in intrathymic
T-cell deletion and activation of harmful autoreactive T cells
in the periphery, for example by infection. The latter has been
postulated to cause type 1 diabetes (insulin-dependent dia-
betes mellitus) in some individuals (Figure 22.5).
Beneficial effects of autoimmunity
The concept of clonal deletion has been central to immunol-
ogy for many years. This process deletes self-reactive lym-
phocytes in order to prevent the harmful effects that these
Wrong environment Tolerance fails Wrong genes
Autoimmune disease
Global (present) Therapies Selective (new)
Fig. 22.5 Autoimmunity is multifactorial. Central tolerance is
achieved within the bone marrow or thymus, but some self-reacting
cells leak out into the periphery, where peripheral tolerance
checkpoints exist. Self-reactive lymphocytes escaping central and
peripheral tolerance may cause autoimmune disease in predisposed
individuals (the “wrong” genes) when they encounter a specific
trigger (the “wrong” environment).
Source: Mackay I.R. (2000). Science, medicine, and the future:
tolerance and autoimmunity. British Medical Journal 321: 93–96.
Reproduced with permission of British Medical Journal.
304 Molecular Hematology
lymphocytes have probably not evolved in order to distin-
guish between the two; their role is rather to respond to anti-
gen under specific circumstances, for example as directed by
the cytokine network. In addition, autoreactive cells are also
responsible for the remodeling of tissues, wound healing, and
other physiological processes. Autoimmunity therefore dif-
fers from autoimmune disease. In fact, transient autoimmune
attack of self antigens has been shown to occur when there is
tissue damage, but this type of autoimmune response is gen-
erally not sustained.
The spectrum of autoimmune diseases
For clinical convenience, autoimmune diseases are subdi-
vided arbitrarily into those that are organ specific, such as
Hashimoto thyroiditis, and those that are non-organ specific
(or systemic), such as systemic lupus erythematosus (SLE).
Many diseases lie between these two extremes. Conditions
in which there are circulating immune complexes tend to be
systemic, while conditions associated with autoantibodies or
autoreactive T-cell responses are organ specific. In general,
however, this classification tells us nothing about the causes
of disease, and it may be better to classify diseases into those
in which there is disturbed selection, regulation, or death of
T and B cells (e.g. mediated by Fas or Fas receptor abnor-
malities), and secondly those in which there is abnormal
expression of an antigen, for example the demyelination syn-
drome that follows gut infection with the bacterium Campy-
lobacter jejuni. This method of classification may help guide
treatment.
Thus, the factors that may play a role in the development
of autoimmune disease include:
r failure of tolerance to self antigens;
r infection;
r tissue injury;
r abnormalities of APCs;
r imbalance between proinflammatory and anti-
inflammatory cytokines;
r genetic factors;
r variable effector mechanisms, for example production
of immune complexes, autoantibodies, or autoreactive
T cells.
Role of genetic factors
It has long been recognized that genetic factors play a key role
in autoimmune disease, the strongest correlation being with
MHC genes, especially MHC class II. Support for a genetic
basis comes from a variety of observations. For example,
autoimmune diseases often cluster within families, and twin
studies show that there is much higher concordance between
monozygotic than dizygotic twins: in monozygotic twins the
chance of both twins having autoimmune disease is higher
than in dizygotic twins. Examples of diseases with high con-
cordance rates between monozygotic twins include type 1
diabetes, rheumatoid arthritis, and SLE; the concordance
rate for monozygotic twins is around 20% and for dizygotic
twins it is less than 5%. Some disorders are due to mutations
in a single gene, for example autoimmune polyglandular
endocrinopathy with candidiasis and ectodermal dyspla-
sia (APECED) and the autoimmune lymphoproliferative
disorder. In APECED there is a mutation in the gene for
an autoimmune regulator protein that is active within the
thymus. The mutation leads to autoimmunity and immune
deficiency. In the autoimmune lymphoproliferative disorder,
the autoreactivity is due to an inability to induce apoptosis
of activated lymphocytes following encounter with antigen.
The disorder is autosomal dominant and involves Fas or its
receptor, both of which are involved in the downregulation of
activated cells following antigen exposure.
However, most autoimmune diseases are not caused by
single gene mutations but rather are multigenic, several
genes acting together to cause the disease phenotype. Type 1
diabetes is a prime example of a disease in which multiple
genes are implicated. For example, 95% of Caucasians with
type 1 diabetes possess HLA-DR3, HLA-DR4, or both (only
found in 40% of normal subjects), and 40–50% of patients
are heterozygous for HLA-DR3/DR4 (found in 5% of normal
subjects). Other non-HLA genes are also implicated in type 1
diabetes, including IL-2 polymorphism and another region
that maps close to CTLA-4. Surprisingly, some of these
genetic polymorphisms occur in normal individuals who
have normal immune function with no evidence of autoim-
mune disease. So it would appear that there is a need for
specific genetic polymorphisms in conjunction with other
susceptibility polymorphisms. The MHC is one candidate for
this role.
Several autoimmune diseases described to date have link-
age with specific MHC class I or II polymorphisms, but again
these do not produce the disease phenotype in isolation;
rather, they require the presence of polymorphisms within
other genes, such as those for cytokines (e.g. TNF-α). For
diseases such as ankylosing spondylitis, type 1 diabetes, and
rheumatoid disease, certain HLA subtypes induce disease
susceptibility. The link between MHC genotype and disease
susceptibility is logical given the fact that autoimmune
diseases involve autoreactive T cells. The ability of a T cell to
respond to a particular antigen depends largely on the MHC,
since the MHC determines how the antigen is presented to
the autoreactive T cells. In addition, the MHC has a powerful
influence on determining the body’s T-cell repertoire. Other
HLA alleles appear to be able to offer protection from
autoimmune disease. For example, when the disease-causing
allele HLA-DQB1*0301 or 0302 is present together with

HLA-DQB1*0602, the latter appears to offer protection from
disease. The mechanism underlying such protection is not
known. Some HLA alleles cause disease only within specific
populations. For example, HLA-DRB1*0401 and *0402 are
associated with rheumatoid disease in Europeans, but not in
black or Hispanic individuals. Again, the mechanism here
is not clearly understood.
Some autoimmune diseases have strong
association with cytokine gene
polymorphisms
Since cytokines, through interaction with their ligands,
orchestrate the immune response, it is possible that dysreg-
ulation of cytokines may play a role in autoimmune disease.
Indeed, many single nucleotide polymorphisms (SNPs) have
been described within cytokine or cytokine receptor genes,
and such SNPs may alter cytokine structure or their expres-
sion. This may result in overexpression or underexpression.
In the blood disorder ITP and other autoimmune diseases,
abnormal cytokine profiles have been described. In chronic
ITP, IL-2, IFN-γ, and IL-10 levels are increased, suggesting a
Th1 activation profile (Table 22.2).
Not only do cytokine gene SNPs predispose to disease, but
data from animal models such as the rheumatoid rat indicate
that specific polymorphisms may determine disease chronic-
ity and severity. In addition, there are data to suggest that
specific genetic polymorphisms may correlate with responses
to specific treatments, but it is probably too early to draw firm
conclusions from such studies.
Antibody Fc receptor gene polymorphisms
For autoimmune blood diseases in which autoantibody-
opsonized cells are sequestered and destroyed within the
reticuloendothelial system, there are data from a number of
studies showing that polymorphisms within the Fcγ recep-
tors may influence disease. Antiplatelet antibodies, as with
Table 22.2 Autoimmune diseases in which strong associations with
cytokine single nucleotide polymorphisms have been described
Implicated cytokine
Disease polymorphism
Juvenile chronic arthritis IL-6
Myasthenia and multiple sclerosis TGF-β
Rheumatoid disease IL-10
Associations have now been reported for many autoimmune
disorders and a full database is provided at www.bris.ac.uk/
pathandmicro/services/GAI/cytokine4.htm.
Autoimmune hematological disorders 305
other antibodies, possess two distinct functional compo-
nents: a Fab end, which binds to the targeted antigen, and an
Fc end, which is recognized by scavenger cells (e.g. macro-
phages) in the spleen, liver, and bone marrow. The recep-
tors on the macrophages that sense the presence of antibodies
are termed Fc receptors. For example, how well the Fc recep-
tors recognize and bind antibody attached to platelets deter-
mines, in part, how aggressively the platelets are destroyed
in ITP. Polymorphisms of the FcγRIIA (FCGR2A) gene have
also been implicated in heparin-induced thrombocytopenia,
SLE, and childhood recurrent bacterial infections.
Intravenous immunoglobulin (IVIg) is effective in ITP,
although its mechanisms of action are not fully understood.
However, one of the actions of IVIg is to attach to Fc recep-
tors on the macrophages, thereby blocking the receptor.
Once the receptor is blocked it cannot bind antibody bound
to platelets, and the antibody-coated platelets are spared
destruction by the immune system.
Recent reports have highlighted genetic variations in FcγR
genes which alter their ability to recognize and bind antibody.
The genetics of FcγR genes is complex, but essentially there
are three major types: FcγRIA, FcγRIIA, and FcγRIIIA. A
recent study of FcγRIIA suggests that a polymorphism in the
FCGR2A gene may be responsible for causing refractory ITP
(ITP that responds poorly to treatment). The polymorphism
influences the efficiency of the receptor to bind with antibody
molecules. A similar polymorphism in the FCGR2A gene has
been shown to alter the function of the receptor and may pre-
dispose individuals to autoimmune disease such as ITP, and
in those individuals developing ITP the disease is more likely
to be chronic. Human FCGR2A-transgenic mice have been
shown to have a more severe form of ITP than normal mice,
which provides further evidence that FcγRIIA contributes to
platelet destruction.
Mouse models of autoimmune disease
These have provided much insight into autoimmune disease
and to date around 25 genes have been shown to play a role
in the development of disease, either when the genes are
deleted or when they are overexpressed. The genes encode
cytokines or their receptors, costimulatory molecules, or pro-
teins involved in apoptotic pathways. Whether a mutation
within one of these critical genes induces disease depends
on the genetic background of the animal, which would tend
to imply that other genes can influence the phenotype. One
other point worth noting is that mutations within a gene may
be involved in more than one clinical disorder.
With respect to hematological autoimmune diseases, there
have been several animal models of ITP, including the
secondary autoimmune model, where thrombocytopenia is
secondary to lupus nephritis; the passive transfer model,
306 Molecular Hematology
where continuous injection of platelet-specific antibodies is
required to maintain a steady state of thrombocytopenia; and
the viral acute ITP model, where thrombocytopenia develops
following a viral infection. Although these models have been
important in studying treatments such as IVIg, they are not
ideal for pathophysiological studies, since either the induced
thrombocytopenia is secondary, or there is no endogenous
antibody production. Recently, however, a platelet-specific
immune model of ITP was developed using CD61 (GPIIIa)
knockout mice. In this active ITP model, immune spleen
cells from immunized CD61 KO mice are transferred into
severe combined immunodeficient (SCID) mice, which then
develop both an antibody- and T-cell-mediated ITP. This
model has the advantage in that both antibody- and T-cell-
mediated ITP is endogenously produced and various thera-
pies can be tested on both effector systems. With the advent
of transgenic mice expressing human Duffy and glycophorin
antigens, there are also now murine models of autoimmune
hemolytic anemia.
Human studies
A number of candidate genes have been studied in human
autoimmune disease, including variants of CTLA-4. This
molecule is a natural downregulator of activated T cells. One
polymorphism within this gene causes a reduction in the
inhibitory signal normally induced by the CTLA-4 molecule,
and this polymorphism has been found to be associated with
human type 1 diabetes, thyroid disease, and primary biliary
cirrhosis.
Critical events in the generation
of autoreactivity
Since autoimmune diseases are multifactorial, there are
obviously triggering events within genetically predisposed
individuals that lead to the development of autoimmune
disease. Animal models show that if animals genetically
susceptible to autoimmune disease are injected with self anti-
gens from genetically identical animals with an appropriate
adjuvant (e.g. bacterial), the animal will mount an immune
response against the self antigen. In humans, autoimmune
disease usually arises spontaneously, although it is accepted
that there must be triggering events giving rise to the disease
phenotype. What factors might be responsible for inducing
autoimmune disease? Several have been proposed, including
infectious and non-infectious triggers.
Environmental factors in the development
of autoimmune disease
If the development of autoimmune disease were entirely
genetic, then we would expect complete or near-complete
concordance in monozygotic twins. However, the concor-
dance rate for the second twin if the first twin has autoim-
mune disease is less than 50%, which suggests that addi-
tional, possibly external factors are required for autoimmune
disease. Possible external influences might include infec-
tion, and evidence for a role of infection in the develop-
ment of autoimmune disease is provided by disorders such
as Lyme disease and its associated arthritis. In this condition,
there is cross-reactivity between the pathogen and host tis-
sue antigens. The self protein targeted is leukocyte function-
associated antigen (LFA)-1, also called CD11a and CD18.
LFA-1 shares determinants with protein antigens of Borrelia
burgdorferi. Another example is Epstein–Barr virus (EBV),
which appears to have a role in the development of SLE, from
case–control studies using stored serum from patients with
SLE. Whether significant or not, there is one antigenic deter-
minant on EBV that is shared with SLE. However, this evi-
dence is fairly circumstantial at present.
How might infection cause autoimmunity?
Pathogens may be able to induce autoimmunity through a
variety of mechanisms:
r production of local inflammation;
r change in reticuloendothelial function, e.g. phago-
cytosis;
r production of neoantigens;
r molecular mimicry.
Local inflammation can expose costimulatory molecules
on APCs and lead to the breakdown of T-cell anergy and
the development of autoimmune disease. Tissue injury may
result in the production of neoantigens for which an autoan-
tibody may have specificity. Lastly, in molecular mimicry,
antigens on microorganisms may resemble those on the host
tissues, such that antibodies produced against the pathogen
will cross-react with the host tissue. Examples may include
multiple sclerosis, type 1 diabetes, and childhood acute ITP,
if the antibody produced in a childhood viral infection by
chance cross-reacts with antigen(s) on the platelet surface.
Although elegant, there is little apart from circumstantial
evidence at present to support the existence of molecular
mimicry in humans.
Non-infectious triggers
Infection may play a role in autoimmune disease, but what
non-infectious triggers might there be which could induce
autoimmune disease in susceptible individuals? Hormones
may play a role, since we know that most autoimmune dis-
orders are commoner in women than men, and in mice with
SLE the administration of estrogens worsens the disease;
this effect is believed to be due to an alteration of the B-cell
repertoire. Also in SLE, complement deficiency (e.g. C1 or
C4 components) may worsen the disorder. Haptens such as
 Autoimmune hematological disorders 307

drugs can also induce autoimmune disease. For example,
penicillin acts as a hapten when it binds to the red cell
membrane, inducing an autoimmune response that leads to
autoimmune hemolytic anemia. Complexes formed between
two proteins may trigger disease. For instance, celiac disease
is an autoimmune disorder in which transglutaminase and its
substrate gliadin form a complex that induces autoantibody
generation against both gliadin and transglutaminase.
Epitope spread
This phenomenon may play a role in some autoimmune
diseases and involves the initial generation of autoreactivity,
with subsequent development of chronic autoimmune
disease. During the process there is a gradual increase in
the number of autoantigens targeted by T cells; hence there
is spread to epitopes beyond the initial triggering epitope
(Figure 22.6).
Hormonal influences
Many autoimmune diseases are more common in females,
particularly in the childbearing years, when hormone levels
are at their highest. The mechanism whereby the hormonal
profile of females influences the development of autoimmune
disease is unknown.
Autoimmune disease is complex
and multifactorial
From studies of autoimmune disease, it is clear that ITP and
other autoimmune disorders are multifactorial. It is likely
that loss of tolerance to a self antigen alone is insufficient to
generate the autoimmune disorder. Rather, patients prob-
ably require (i) a specific set of genetic determinants (e.g.
polymorphisms within MHC, CTLA-4, or other genes); (ii)
dysregulation of the immune response (involving dendritic
cells, T or B cells, or all three); and (iii) an environmental
trigger. Autoimmune disease arises only when all these
determinants are present in an individual at one particular
time. This is reinforced by the observation that self-reactive
lymphocytes are commonly found in healthy individuals. For
example, siblings of patients who have autoimmune disorders
are more likely to have autoantibodies themselves, albeit at
lower titers than their affected siblings, but no overt evidence

Antibody coated Fcγ Activated macrophage

platelet receptor

Glycoprotein
IIb/IIIa
Autoantibody
Glycoprotein
Ib/IX

CD4
CD154
Anti- CD154
Glycoprotein CD40
glycoprotein CD40
IIb/IIIa IIb/IIIa CD4
B-cell clone 1 T-cell clone 1

Glycoprotein
Ib/IX
Glycoprotein
Anti-
IIb/IIIa
glycoprotein
IIb/IIIa Glycoprotein
Ib/IX

B-cell clone 2 T-cell clone 2

Fig. 22.6 Epitope spread. In immune thrombocytopenia (ITP), for example, platelet glycoproteins are processed and presented to T cells by
antigen-presenting cells (APCs). A variety of different peptides may be presented by APCs, resulting in multiple T-cell clones, each targeting distinct
epitopes on different glycoprotein molecules.
Source: Cines D.B., Blanchette V.S. (2002). Immune thrombocytopenic purpura. New England Journal of Medicine 346: 995–1008. Reproduced by
permission of New England Journal of Medicine.
308 Molecular Hematology
of autoimmune disease per se, perhaps because they have not
been exposed to the environmental trigger required to tip the
balance toward autoimmune disease.
How do autoreactive cells induce tissue
damage in autoimmune disease?
This is a large topic which is discussed only briefly here. In
autoimmune disease, since the autoantigen is not cleared
completely from the affected individual, the process tends to
be prolonged, because there is always a supply of autoantigen
available to keep the process going. Autoantibodies, espe-
cially IgG and IgM, may attach to cell membrane antigens
on cells and cause local tissue damage. Where the antigens
are soluble, a more systemic disease profile may result. In
addition to autoantibody-mediated disease, T cells can them-
selves directly cause disease. When autoantibodies attach
to blood cells, premature destruction of the cells results
within the spleen. Cells involved include red cells, resulting
in autoimmune hemolytic anemia, white cells, causing
neutropenia (any white cell may be involved but neutropenia
is commonest), and platelets, inducing thrombocytopenia.
Red cell attack by IgG antibodies results in premature red
cell destruction by the macrophages of the reticuloendothe-
lial system, via Fcγ receptors. When the autoantibody is
IgM, complement activation also occurs and there may be
intravascular lysis, although more commonly the comple-
ment cascade does not reach the lytic stage, but generally
stops at the C3d stage because of the presence of complement
regulatory proteins. However, even with incomplete comple-
ment activation, the presence of molecules such as C5a gener-
ates local inflammation and tissue damage through activation
of cytokines. Autoimmune blood cell disorders are discussed
shortly.
Rather than cause direct cellular damage, some autoanti-
bodies may activate or block receptors. For example, Graves
disease is caused by an autoantibody that stimulates the
thyroid-stimulating receptor, leading to elevated levels of
thyroid hormones. In myasthenia gravis, the autoantibody
blocks acetylcholine receptors, preventing neuromuscular
transmission. Other mechanisms of autoantibody and T-cell
damage also occur, but there is insufficient space to discuss
these here.
Immune thrombocytopenia as a model
of autoimmune blood disease
ITP is usually an acquired disorder in which platelets
are coated (opsonized) with antiplatelet autoantibodies and
removed prematurely by the reticuloendothelial system, pre-
dominantly the spleen, leading to a reduced peripheral blood
platelet count. Of interest is that it has been estimated that
approximately 40% of patients with chronic ITP have no
detectible antibodies, yet are thrombocytopenic. CD8+ T
cells have been linked to the pathogenesis of many autoim-
mune diseases such as type 1 diabetes, and it was shown
that ITP patients have a CD8+ T-cell-mediated cytotoxic-
ity that induces platelet destruction. This finding was also
shown in a murine model of active ITP. At present, however, it
is unknown whether cell-mediated platelet destruction con-
tributes to the severity of disease or the difficulty of treatment
in some patients with ITP.
The etiology of ITP is unknown and the clinical course is
variable and unpredictable. ITP has an incidence of around
60 new cases per million population per year in the USA.
Childhood ITP is generally seasonal, typically follows a trivial
viral infection or vaccination, is transient in most cases, and
requires no treatment, with spontaneous recovery in 80% of
cases. One proposed mechanism invoked in childhood ITP is
molecular mimicry, in which the antibody directed against an
invading pathogen coincidentally cross-reacts with one of the
platelet glycoprotein (GP) epitopes. As discussed earlier, the
normal adaptive immune response ceases once the offend-
ing pathogenic antigen is destroyed, and this might account
for the acute nature of ITP in childhood. That is to say, once
the pathogen is eradicated, the source of non-self antigen is
removed and cross-reacting antibody levels fall.
In the adult form there is usually no obvious antecedent
illness and most patients have chronic thrombocytope-
nia; spontaneous recovery is uncommon. In most cases of
adult ITP the platelet glycoprotein (GP) antigen targets are
GPIIb/IIIa and GPIb/IX.
Etiology
It is believed that ITP is most likely due to an inappropriate
immune response to an environmental trigger; the nature
of this trigger has not yet been identified. The disorder may
represent an abnormality of APCs, with an increase in the
numbers of CD4+ and CD8+ cells. It has been increasingly
recognized that T cells play a significant role in autoantibody
production and recent data suggest that, in some instances,
T cells may actually mediate platelet and megakaryocyte
(MK) destruction in ITP. Furthermore, two new and excit-
ing possibilities have emerged in ITP research related to
T-cell abnormalities that may prove be key to unraveling
the pathogenesis of the disorder, and may also reveal new
therapeutic options for ITP. It appears that patients with ITP
have functionally reduced numbers of T-regulatory cells that
may relieve T-cell tolerance mechanisms and allow platelet
autoimmunity to proceed, whereas therapeutic removal
of B cells in the disorder by therapies such as rituximab
raises platelet counts by actually normalizing these T-cell
functional deficiencies. Figure 22.7 highlights the cellular
immune abnormalities found in patients with ITP.
 Autoimmune hematological disorders 309

Inflammatory
stimuli CD40L
GPllbllla

Platelet M-CSF CD40

A
autoantigens sIg CD20
Activated
macrophage
Autoreactive
IL-1β, IL-1α B cell
CD68 TGF-β
CD40 D

Treg Autoantigen B7.1/7.2

processing
peptides B
IL-2, IL-4, IL-15
CD40 CD40L
IFN-γ, TNF-α,
CTLA-4 GM-CSF
Class II
F CD28
Class I Autoantibody
CD40L
production
C
TCR
sIL 2R
Microparticles Autoreactive
TCR IL2
Th cell
E
sIL2R-IL2

TCR
CD8+T cell Cell-mediated Platelet
NK cell immunity destruction
Class I
Fig. 22.7 Potential stimulatory (solid arrows) and inhibitory (hatched arrows) pathways that may affect the pathogenesis of chronic
immune thrombocytopenia (ITP). This model assumes that a tolerance event has broken down, allowing autoreactive T and B cells to be
present. (a) Platelets are normally taken up by macrophages during senescence and are presumably destroyed intracellularly within lysosomes.
Inflammatory stimuli can activate macrophages that may alter the way they normally process platelet autoantigens, or induce the expression of
inflammatory cytokines (e.g. IL-1) that may support autoimmune responses. (b) In the activated antigen-presenting cells (APC), platelet
autoantigenic peptides can be loaded onto major histocompatibility complex (MHC) class II molecules for presentation to the T-cell receptor (TCR)
of autoreactive Th cells. (c) Once the TCR has been occupied by the MHC class II molecules and platelet autoantigen, it can initiate a series of
molecular costimulatory interactions within the T cell. First, CD40L upregulated on the T-cell surface interacts with CD40 on the APC that
stimulates B7.1 and B7.2 expression. The TCR–MHC interaction also enhances CD28 expression on the Th cells, which then interact with the B7.1,
culminating in a strong costimulation response and activation of the Th cell. (d) Th activation leads to the secretion of Th0/Th1 cytokines (e.g. IL-2,
IL-10, IFN-γ) that effectively drive autoreactive B cells to divide and differentiate into plasma cells and secrete antiplatelet autoantibodies. Also
shown are a number of potential regulatory events that could either enhance or suppress autoimmunity in ITP. For example, macrophage
colony-stimulating factor (M-CSF) can activate macrophages within the spleen to enhance their phagocytosis of platelets. Additionally, IFN-γ and
TNF-α produced by the autoreactive Th cells can feed back on macrophages, enhancing their expression of MHC class II molecules and potentially
aggravate the response. On the other hand, transforming growth factor (TGF)-β (produced by either Th3 cells or platelets) together with the Th
cell’s soluble IL-2 receptor can control the autoimmune response by inhibiting lymphocyte activation. This may also occur during costimulation via
the expression of CTLA-4 on the Th cell. Also shown are the potential interactions of MHC class II- and CD40L-positive platelets with Th and B cells
activation, respectively. These events may be responsible for initiating and/or aggravating autoimmunity in patients with autoimmune
thrombocytopenia. (e) Recent evidence suggests that CD8+ T cells are active and present in patients with ITP who do not demonstrate
autoantibodies. These cytotoxic cells bind to platelets and mediate cytolysis. (f) It appears that many of the defects shown in the figure may be the
result of defective or lack of CD4+ CD25+ T regulatory (Treg) cells due to a breakdown in self-tolerance.
Source: Adapted from Coopamah M.D., Garvey M.B., Freedman J., Semple J.W. (2003). Cellular immune mechanisms in autoimmune
thrombocytopenic purpura: an update. Transfusion Medicine Reviews 17: 69–80. Reproduced with permission of Elsevier.
310 Molecular Hematology
With regard to genetic abnormalities in ITP, most atten-
tion has focused on the identification of MHC susceptibil-
ity genes, given their role in determining the nature and
specificity of the adaptive immune response. However, HLA
association studies in ITP have not produced clear results.
A large analysis focusing on HLA-DR4 gene variations in
more than 100 Japanese patients with ITP reported that the
DRB1*0410 allele was significantly increased in ITP patients
compared with controls. Moreover, this allele was signifi-
cantly decreased in patients who showed a good response to
prednisolone. MHC may therefore play a role in some cases
of ITP, but there are clearly other genes implicated. These
include genetic polymorphisms within cytokine and other
immune regulatory genes.
Clinical features
The ITP phenotype is heterogeneous: some patients suffer
major bleeding from the outset, while others have few
problems apart from an increased tendency to bruise. This
may partly be explained by the acquired platelet dysfunction
that is seen in some patients with ITP, which in turn may
be related to the target antigen involved in the autoimmune
process (this is discussed later; see Antibodies and Their
Target Antigens). Autoantibodies reacting with GPIIb/IIIa
affect platelet aggregation, and anti-GPIb/IX autoantibodies
impair platelet adhesion to the subendothelial matrix,
causing unexpectedly severe bleeding for the level of platelet
count. In contrast, however, to the thrombocytopenia due to
marrow infiltration (e.g. leukemias, lymphomas, and other
malignancies) or aplasia, patients with ITP are able to tolerate
remarkably low platelet counts and to maintain an adequate
quality of life. The degree of bleeding is largely dependent on
the platelet count, and patients with counts below 10 × 109 l−1
(and usually below 5 × 109 l−1 ) are at greatest risk of bleeding,
including intracranial bleeding.
Diagnosis
The diagnosis of ITP remains clinical, and one of exclu-
sion. Secondary causes include SLE, lymphoproliferative dis-
ease, and HIV infection. Standard investigation includes a
full blood count (isolated thrombocytopenia), blood film (to
ensure no red cell fragments, leukemia, parasitic infections),
and autoimmune profile (to exclude secondary cause). A
bone marrow examination is often carried out in adults but
not usually in children, and will typically show normal or
increased MK in an otherwise normal marrow. Immunolog-
ical assays have been devised, including platelet-associated
IgG or IgM and monoclonal antibody immobilization of
platelet antigens, but these do not alter the management and
are of debatable value.
Antibodies and their target antigens
Antiplatelet antibodies
Many patients with ITP show elevated levels of platelet-
associated IgG, which is believed to be the autoantibody, but
for unknown reasons platelet-associated IgG may be elevated
in other non-immunological causes of thrombocytopenia.
The autoantibodies involved in ITP are generally IgG, but
IgA and IgM autoantibodies have been reported. Opsonized
platelets are removed prematurely by the reticuloendothe-
lial system through an Fc-dependent mechanism, primarily
within the spleen. However, many patients fail to respond to
therapies aimed at inactivation of the reticuloendothelial sys-
tem, suggesting that other mechanisms of platelet destruction
exist.
Antigenic targets
Using antigen-specific assays, such as the monoclonal
antibody-specific immobilization of platelet antigens,
platelet-associated IgG, and antigen capture assays, several
platelet autoantigens have been characterized. These include
epitopes on GPIIb/IIIa (αIIIbβ3, the fibrinogen receptor)
and GPIb/IX (the von Willebrand receptor), which appear to
be the most frequently involved. Less commonly, GPIa/IIa,
GPIV, and GPV are involved. Recent reports suggest that
possibly 80% of autoantibodies are reactive to both GPIIb/
IIIa and GPIb/IX, possibly due to the serum in some
patients with ITP containing two different IgG antibodies.
In terms of disease chronicity, GP-specific autoantibodies
may be important in the pathogenesis of chronic ITP; from
available data, GPIIb/IIIa appear to play a major role in the
development of chronic ITP in approximately 40% of cases.
Which epitopes are involved?
Previous investigators have looked for autoantigenic epi-
topes on the GPIIb/IIIa molecule using competitive binding
between human autoantibodies and mouse monoclonal anti-
bodies. In addition, enzyme-cleaved IIb or IIIa fragments and
synthesized peptides corresponding to different sequences of
GPIIIa have been used to localize epitopes on the respective
glycoprotein.
Kekomaki et al. have shown that the 33-kDa chymotryp-
tic core fragment of IIIa is a frequent target. Fujisawa et al.,
using synthetic peptides corresponding to IIIa sequences,
showed that in 5 of 13 sera from patients with chronic ITP,
binding was to residues 721–744 or 742–762, correspond-
ing to the carboxy terminal of IIIa. Recently, Nieswandt et al.
have examined the pathogenic effects of IgG monoclonal
antibodies of different IgG subclasses against murine GPIIb/
IIIa, Ibα, Ib/IX, V, and CD31. Their data suggest that, at
least in mice, the antigenic specificity of the antiplatelet

antibodies determines the pathogenic effects rather than
the IgG subclass. They also demonstrated that antibodies
against GPIb/IX caused thrombocytopenia through an Fc-
independent mechanism, while that from autoantibodies
against GPIIb/IIIa involved the Fc system. Further work
is clearly needed in order to determine the significance of
all these findings, which may translate into stratification of
patients into those in whom Fc receptor blockade or inacti-
vation is a useful option, and those in whom it is not.
At the T-cell level, it has been demonstrated that T cells
from Japanese patients with chronic ITP could proliferate
in vitro in response to disulfide-reduced GPIIb/IIIa or the
molecule’s tryptic peptides. This suggested that autoreactive
CD4+ Th cells in chronic ITP need to recognize a modified
GPIIb/IIIa molecule, implying that antigen-processing
mechanisms within recipient APCs may be required to
present GPIIb/IIIa autoantigens in the context of self HLA-
DR molecules. Subsequently, mapping studies using six large
(~200 amino acids) recombinant fragments encoding differ-
ent portions of the GPIIbα and GPIIIa chains showed that
the T cells from patients with ITP recognized primarily the
amino-terminal portion of the two chains (GPIIbα 18–259
and GPIIIa 22–262) and that these molecules also stimulated
the production of antiplatelet autoantibodies. Ultimately, the
GPIIIa molecule has been mapped for CD4+ T-cell specifici-
ties by using 15-mer peptides of the GPIIIa chain, and this has
revealed several immunodominant peptides spanning the
entire breadth of the molecule. What was particularly intrigu-
ing was that despite a lack of HLA association observed in
many studies of patients with ITP, some patients apparently
recognize common elements on the GPIIIa molecule. This
may suggest that either a host (e.g. antigen processing) or
environmental (e.g. infection) factor could be responsible for
generating a common autoepitope that is presented to T cells
across different HLA. If this were true, it would be an effica-
cious way of developing peptide antigen-specific T-cell ther-
apies for autoantibody production and subsequent platelet
destruction in ITP.
Standard treatment
There is a lack of clinical trial data to help guide treatment,
and our energies should now be focused on constructing
high-quality randomized trials in order to determine the
most effective therapy in this disorder. Therapy is seldom
necessary for patients whose platelet counts exceed 30 ×
109 l−1 and in whom there are few spontaneous bleeding
episodes, unless they are undergoing any procedure likely
to induce blood loss. Standard treatments, including oral
prednisolone, IVIg, and splenectomy, will elevate the platelet
count sufficiently in most adults (Figure 22.8). However,
some 20–25% of adults with ITP are refractory to first-line
therapy.
Autoimmune hematological disorders 311
Chronic refractory ITP
This defines those patients who fail to respond to first-line
treatment or require unacceptably high doses of corticos-
teroids to maintain a safe platelet count. A number of agents
have been used as second-line therapy for ITP, including
high-dose steroids, high-dose IVIg, intravenous anti-D, vinca
alkaloids, danazol, azathioprine, combination chemotherapy,
and dapsone.
Targeted versus untargeted therapies
for autoimmune disease
Until now, most of our treatments for autoimmune dis-
ease have been untargeted and unselective in their modes of
action. In disorders such as ITP, the therapeutic aim has been
to induce global immunosuppression in the hope that, as part
of this process, the ITP-related component of the immune
system will be suppressed and that this will help reduce the
quantity of autoantibody produced. For antibody-mediated
autoimmune diseases, what remains unclear is whether the
B-cell population that is generating the antiplatelet autoanti-
body is the primary problem, or whether events downstream,
such as those involving antigen presentation or T-cell regu-
lation, are disturbed, and simply driving the passive B cells,
resulting in the autoantibody phenotype.
Now that we have a clearer understanding of the immuno-
logical mechanisms involved in autoimmune disease, we
have started to develop more targeted therapies. We are now
developing treatments designed to target T cells, B cells, and
other effectors within the immune system. For ITP these
include Campath-1H and anti-CD20. Although these agents
are not entirely specific because they deplete the B-cell com-
partment, they should reduce the quantity of autoantibody
produced. Other therapies which may be of benefit in ITP
are mycophenolate mofetil and anti-CD40 ligand.
Campath-1H
Campath-1H is a humanized IgG monoclonal antibody that
targets the CD52 antigen, present on mature human lym-
phocytes (T and B cells) and monocytes. Campath-1H is
effective in the treatment of malignant B-cell disorders, espe-
cially B-cell chronic lymphocytic leukemia, in which it has
been shown to be effective in clearing lymphocytes from both
blood and bone marrow.
Campath-1H has been used in a variety of autoimmune
diseases, including rheumatoid arthritis, vasculitis, and
Wegener granulomatosis. There is ongoing interest in its
use for the treatment of autoimmune hematological disease
that is refractory to first- and second-line therapies. One
recent study of the use of Campath-1H in autoimmune
312 Molecular Hematology

Corticosteroids,
intravenous immune globulin,
anti-D immune globulin, Splenectomy
danazol, vinca alkaloids Macrophage

Platelet

FCγ
receptor

Platelet
transfusion
Plasmapheresis

Bone marrow

Antibody
against CD20,
intravenous
immune globulin(?)
staphylococcal
protein A(?)

Thrombopoietin
B-cell
corticosteroids

Antibody Azathioprine,
against CD154 cyclophosphamide,
T-cell cyclosporin,
corticosteroids,
danazol
Fig. 22.8 Standard and novel treatment strategies in immune thrombocytopenia (ITP). Standard treatments include corticosteroids,
intravenous immunoglobulin, danazol, and vinca alkaloids. Their sites and modes of action are illustrated. New therapies include monoclonal
antibodies against CD154 (CD40 ligand) and CD20 (on B cells, leading to transient B-cell depletion).
Source: Cines D.B., Blanchette V.S. (2002). Immune thrombocytopenic purpura. New England Journal of Medicine 346: 995–1008. Reproduced
with permission of New England Journal of Medicine.

neutropenia, autoimmune hemolytic anemia, pure red cell
aplasia, ITP, and combined hemolytic anemia and ITP (Evans
syndrome) has shown responses in 15 of 21 patients treated;
in 6 patients the response was sustained. Campath-1H
therefore appears to be an effective agent in severe refractory
autoimmune disease. The drug is well tolerated, but because
it can precipitate bleeding during administration, it should
not be given in the presence of active bleeding (or active
infection).
Anti-CD20 monoclonal antibody therapy
Rituximab, a genetically engineered chimeric human/mouse
anti-CD20 monoclonal antibody, has been developed
as a treatment for B-cell lymphoproliferative disease
(non-Hodgkin lymphoma). The antibody is an IgG κ
immunoglobulin comprising murine light- and heavy-chain
variable-region sequences and human constant-region
sequences. The antigen-binding domain binds to the CD20
antigen on B cells, while the Fc domain mediates B-cell lysis
through recruitment of immune effector cells. Because of its
specificity for B cells, rituximab has been viewed as a poten-
tial treatment for autoimmune disease, the rationale being
the reduction or elimination of autoantibody-producing B
cells, with concomitant improvement of the autoimmune
disease. A recent study by Stasi et al. reports on the efficacy
of rituximab in the treatment of 25 patients with chronic
refractory ITP. Patients were treated if their platelet counts
were below 20 × 109 l−1 irrespective of symptoms, or at
higher platelet counts if bleeding or bruising was problem-
atic. All patients had received between two and five previous
treatments; eight had failed splenectomy. Rituximab was
administered in the same manner and dose as that used
in non-Hodgkin lymphoma. After four courses, 40% of
patients achieved a platelet count of at least 50 × 109 l−1 ; five
achieved complete remission (platelets >100 × 109 l−1 ); and
 Autoimmune hematological disorders 313

Cytokine proteins orchestrate

Fc receptor

Y
Y XY Y
CTLA-4

Antigen-presenting T cell B cell Antibody

cell

Fig. 22.9 Costimulatory blockade may be beneficial in some autoimmune diseases. CTLA-4 linked to human immunoglobulin Fc
(CTLA-4-Ig) blocks the critical second signal between antigen-presenting cells and T cells, resulting in T-cell anergy. This should result in a reduction
in antibody production and amelioration of disease if the autoimmune disease is antibody mediated. CTLA-4-Ig treatment has been shown to be
of benefit to patients with psoriasis. Similarly, anti-CD40 ligand also blocks the second signal with similar results, and has been shown to be of
value in refractory immune thrombocytopenia (ITP).

five partial remission (platelets 50–100 × 109 l−1 ). Responses Costimulatory blockade
were seen during treatment with rituximab, with a peak
Therapies such as Campath-1H and anti-CD20 may not pro-
response up to four weeks after the end of treatment; 28%
duce lasting remission if the autoimmune B cells are driven by
had responses that lasted for more than six months.
dysregulated T cells, and a novel agent, CTLA-4-Ig, has been
The results suggest that the use of rituximab resulted in
evaluated in psoriasis in an attempt to block T-cell costimu-
responses similar to those given by other second-line agents
lation, thereby inducing anergy in the T-cell compartment.
used in ITP (including vinca alkaloids, cyclophosphamide,
CTLA-4-Ig, a fusion protein between CTLA-4 and the Fc
and azathioprine), around 40–50%, but sustained responses
portion of human immunoglobulin, binds to B7-1 and B7-2,
to these agents is usually seen in fewer than 20% of patients
blocking T-cell costimulation (Figure 22.9). This small trial
(i.e. lower than for rituximab). Rituximab would appear to be
showed that, at least within this group of patients, CTLA-4-Ig
useful for some patients with chronic symptomatic refractory
was able to improve the disorder and was shown to be safe.
ITP in whom there is a definite need to elevate the platelet
CTLA-4-Ig may have applications within other autoimmune
count to a safe level.
disorders, including ITP. If a drug such as CTLA-4-Ig were
The mechanism of action of rituximab in ITP has been
shown to be effective in ITP, not only would it provide an
assumed to be due to selective depletion of CD20+ B cells that
additional targeted treatment modality, it would also provide
subsequently affects autoantibody development. This con-
useful evidence of T-cell dysfunction in this disease. Interest-
cept was recently shattered when it was demonstrated, using a
ingly, the CTLA-4 gene has been mapped as a susceptibility
variety of sophisticated techniques to analyze T-cell parame-
gene in autoimmune thyroid disease and type 1 diabetes in
ters, that only when the abnormal T-cell subsets were normal-
humans.
ized was rituximab therapy effective. The reasons for these
results are not clear, but may relate to how B-cell popula-
Other options: Helicobacter pylori
tions may be either important in maintaining autoreactive
eradication
T-cell activation patterns or, by decreasing the total mass of B
cells, may cause a collapse of autoreactive T-cell stimulation Helicobacter pylori is the main cause of gastritis and peptic
and normalization of the T-cell repertoire, even as the B cells ulcer disease. It has also been implicated in the development
begin to return months after the therapy. What is perhaps of gastric adenocarcinoma and mucosa-associated lymphoid
more enlightening is the demonstration that the abnormal tumors, and in some autoimmune disorders. Previous stud-
T-regulatory (Treg) populations are indirectly targeted by rit- ies in ITP showed improvement in platelet counts after erad-
uximab therapy; the anti-CD20 treatment reverses the Treg ication of the bacterium in patients shown to be positive for
deficiency in patients with ITP and normalizes the autoim- H. pylori. More recently, Emilia et al. looked for the presence
munity. Taken together, these studies truly lend credence to of H. pylori in 30 patients with chronic refractory ITP. It was
the notion that attacking T cells in ITP, even indirectly by found in 13 of 30 patients (43.3%). Standard triple therapy
the destruction of B cells, is perhaps the only way to reduce for H. pylori eradication resulted in a complete response in 4
platelet destruction effectively. of 12 patients in whom the bacterium was eradicated, and a
314 Molecular Hematology

Thrombopoietin Nplate Promacta

O OH

TPO-R binding
Fc
domain
domain
(including spacer regions)

OH

N NH
H3C O
N N
CH3

(a) (b) (c) CH3

Fig. 22.10 Structures of thrombopoietin (TPO) and synthetic c-mpl ligands. (a) Native TPO is a 332-amino-acid glycoprotein with a
molecular mass of 60–70 kDa. It is the major humoral regulator of platelet production. (b) Nplate (romiplostim) is a 60 kDa synthetic “peptibody”
that does not share any amino acid homology to native TPO. It comprises a human immunoglobulin Fc domain linked via polyglycine to two
divalent mpl-binding peptide regions. The Fc component extends the half-life of the drug in the circulation, while the peptide “warhead” binds to
the TPO receptor, c-mpl, and activates signaling. (c) Promacta (eltrombopag) is an orally bioavailable hydrazone small molecule with a molecular
mass of 546 Da. Unlike TPO and Nplate, which bind the extracellular domain of c-mpl, Promacta is reported to bind to the transmembrane region
of c-mpl.
Source: Kuter D.J. (2007). New thrombopoietic growth factors. Blood 109: 4607–4616. © American Society of Hematology. Reproduced by
permission.

partial response in 2 of 12 (16.6%). The responses were main-
tained for a median of 8.33 months. In addition, there are
other anecdotal reports of improvements in platelet counts
in adults and children with ITP after eradication of H. pylori.
Larger studies are required to confirm these earlier findings,
but from the available data triple therapy appears to offer a
non-immunosuppressive therapy for patients with refractory
ITP and possibly other autoimmune diseases.
On the other hand, platelets express Toll-like receptor
(TLR)4, and this has been shown to be responsible for the
thrombocytopenia induced by lipopolysaccharide adminis-
tration in vivo. It now appears that bacterial products such
as lipopolysaccharide together with IgG bound to platelets
can significantly enhance Fc-mediated platelet phagocytosis
by mononuclear phagocytes. This suggests that infectious
agents, in combination with antiplatelet antibodies, could
affect platelet destruction in vivo, and may be at least one
explanation of why thrombocytopenia worsens in some
patients with ITP during infections and, alternatively, reso-
lves in other patients with ITP who are treated with bacterial
eradication therapy.
Thrombopoietin (TPO) mimetics
Patients who fail first-line therapy can be treated with
thrombopoietin (TPO)-receptor agonists. Both eltrombopag
(Novartis) and romiplostim (Amgen) activate TPO recep-
tors on MK and induce platelet production via the JAK2 and
STAT5 kinase pathways. These drugs stimulate MK prolifer-
ation and differentiation, with a resulting increase in platelet
counts. Romiplostim is classed as an Fc peptide fusion pep-
tibody and consists of two 14-amino-acid peptide dimers
covalently linked to a human Fc fragment. The Fc protein
prolongs the half-life of the drug and the drug stimulates
megakaryopoiesis after subcutaneous injection. In contrast,
eltrombopag is a small molecule naphthalenesulfonic acid of
564.6 Da molecular weight; it has similar potency to native
TPO and can be administered as an oral tablet (Figure 22.10).
The two drugs have proven to be highly effective in raising
platelet counts in patients with chronic ITP. Of interest is that
it now appears that approximately a third of romiplostim-
treated patients remain in remission even after 24 weeks off
TPO treatment. How this occurs is unknown, but it appears
that romiplostim rescues the Treg deficiency observed during
active disease. It was shown that Treg function was increased,
and the platelet counts correlated with circulating trans-
forming growth factor (TGF) levels, which may be due the
increase in the platelet mass. Thus, the TPO-receptor ago-
nists may not only directly induce thrombopoiesis, but also
modulate the immune system, perhaps by modulating Treg
patients with ITP.
Novel therapies for the treatment
of other autoimmune diseases
Human and animal studies have been helpful in learning how
the immune system works in both health and disease, but can
such information be translated into better patient care? Until
recently, the treatment strategy for autoimmune diseases
has been to induce global immunosuppression in the hope
that the autoimmune process may be abrogated or stopped.
In some cases treatment is effective, but it is clear from

longitudinal follow-up studies of patients with ITP that there
is very significant morbidity and mortality associated with
our current treatments. Infection plays a major role in the
death of patients with autoimmune disease, and such fatal-
ities are usually induced by immunosuppression. Now that
we have a better understanding of the components of the
immune system and how these interact in disease, we should
be able to develop targeted therapies that aim to modify spe-
cific components of the immune system, while leaving most
of the immune system intact and able to fight infection.
Such therapeutic advances are in fact being made, and
many of these have been developed through knowledge
concerning specific components of the immune system in
disease. We have now been able to develop targeted therapies
for rheumatoid disease, multiple sclerosis, psoriasis, diabetes,
and systemic lupus. We discuss each of these briefly in turn.
Rheumatoid arthritis and blockade
of TNF-α
Disease-modifying therapies, designed to reduce deformity
in rheumatoid disease, have been around for a number of
years. Methotrexate is one of the main agents in this category.
Bone destruction in rheumatoid disease is partly mediated
by macrophages through an inflammatory process. With our
increased understanding of cytokines and their interactions,
specific modulators of the immune response have been devel-
oped. Because TNF-α plays a key role in the inflammatory
response, antibodies against TNF-α have been developed
and shown to be of value. To date, two antibodies appear
promising: the first is a TNF-α receptor–IgG1 fusion protein,
etanercept; the second, infliximab, is a monoclonal antibody
directed against TNF-α itself. This form of therapy also
appears to have a place in the management of other autoim-
mune diseases, such as Crohn’s disease, psoriatic arthropathy,
and ankylosing spondylitis.
Another cytokine involved in the pathogenesis of rheuma-
toid disease is IL-1. Blockade of the IL-1 receptor using a
recombinant antagonist may slow down the development of
erosive bone disease, but from available data this appears to
be less effective than TNF-α.
Multiple sclerosis
IFN-β has recently been introduced for the treatment of mul-
tiple sclerosis. The available study data indicate that it may
delay the onset of the disease if started immediately after the
patient’s first attack of optic neuritis.
Psoriasis
This autoimmune skin disorder has been treated by TNF-α
blockade with and without methotrexate. Other agents
Autoimmune hematological disorders 315
used in small clinical trials include IL-10 and CTLA-4-Ig;
the latter molecule is a recombinant protein comprising
the extracellular domain of CTLA-4 linked to the constant
region of IgG1. CTLA-4-Ig downregulates activated T cells
and prevents activation of naı̈ve T cells. CTLA-4-Ig was well
tolerated in a small trial conducted by Bristol-Myers-Squibb
and the patients’ skin condition improved by more than 50%.
Further studies are required to confirm safety and efficacy.
There are studies of other agents used in the treatment of
autoimmune disease, but there is insufficient space to dis-
cuss these here. However, it would appear that our knowl-
edge of immunity and autoimmunity is now being used to
design more subtle and specific treatments for patients with
autoimmune disease. It is expected that these treatments will
reduce immunosuppression, morbidity, and mortality and
offer modern solutions to these otherwise intractable dis-
eases. No doubt the future will see even more designer drugs
being developed for this fascinating group of diseases.
Conclusions
Autoimmune diseases are complex immunological disorders
affecting 5% of the population. Until recently our under-
standing of the pathogenesis and treatment of these disorders
was severely limited. However, with greater understanding
of the immune system in health and autoimmune disease
we are able to identify underlying abnormalities leading to
the develop of autoimmunity. With this new knowledge we
have been able to modify our therapies by replacing non-
selective immunosuppressive treatment with more subtle
targeted therapies.
Further reading
General
Janeway, C.A., Travers, P., Walport, M., and Capra, J.D. (1999). Immuno-
biology: The Immune System in Health and Disease, 4e. New York:
Current Biology.
Roitt, I.M. and Delves, P.J. (2001). Roitt’s Essential Immunology, 10e.
Oxford: Blackwell Science.
Immune system and HLA
Chien, Y.H., Gascoigne, N.R., Kavaler, J. et al. (1984). Somatic recombi-
nation in a murine T-cell receptor gene. Nature 309: 322–326.
Delves, P.J. and Roitt, I.M. (2000a). The immune system. First of two
parts. N. Engl. J. Med. 343: 37–49.
Delves, P.J. and Roitt, I.M. (2000b). The immune system. Second of
two parts. N. Engl. J. Med. 343: 108–117.
Klein, J. and Sato, A. (2000a). The HLA system. First of two parts.
N. Engl. J. Med. 343: 702–709.
316 Molecular Hematology
Klein, J. and Sato, A. (2000b). The HLA system. Second of two parts.
N. Engl. J. Med. 343: 782–786.
Parkin, J. and Cohen, B. (2001). An overview of the immune system.
Lancet 357: 1777–1789.
Tonegawa, S. (1983). Somatic generation of antibody diversity. Nature
302: 575–581.
Tolerance
Dario, A., Vignali, A., Collison, L.W., and Workman, C.J. (2008). How
regulatory T cells work. Nat. Rev. Immunol. 8: 523–532.
Kamradt, T. and Mitchison, N.A. (2001). Tolerance and autoimmunity.
N. Engl. J. Med. 344: 655–664.
Mackay, I.R. (2000). Science, medicine, and the future: tolerance and
autoimmunity. Br. Med. J. 321: 93–96.
Sarzotti, M. (1997). Immunologic tolerance. Curr. Opin. Hematol. 4:
48–52.
Autoimmunity
Davidson, A. and Diamond, B. (2001). Autoimmune diseases. N. Engl.
J. Med. 345: 340–350.
Maloy, K.J. and Powrie, F. (2001). Regulatory T cells in the control of
immune pathology. Nat. Immunol. 2: 816–822.
Sinha, A.A., Lopez, M.T., and McDevitt, H.O. (1990). Autoimmune
diseases: the failure of self tolerance. Science 248: 1380–1388.
Taneja, V. and David, C.S. (2001). Lessons from animal models for
human autoimmune diseases. Nat. Immunol. 2: 781–784.
Role of genetic and other factors in
autoimmune disease
Albert, L.J. and Inman, R.D. (1999). Molecular mimicry and autoimmu-
nity. N. Engl. J. Med. 341: 2068–2074.
Benoist, C. and Mathis, D. (2001). Autoimmunity provoked by infec-
tion: how good is the case for T cell epitope mimicry? Nat. Immunol.
2: 797–801.
Choy, E.H. and Panayi, G.S. (2001). Cytokine pathways and joint inflam-
mation in rheumatoid arthritis. N. Engl. J. Med. 344: 907–916.
Criswell, L.A., Moser, K.L., Gaffney, P.M. et al. (2000). PARP alleles and
SLE: failure to confirm association with disease susceptibility. J. Clin.
Invest. 105: 1501–1502.
Ermann, J. and Fathman, C.G. (2001). Autoimmune diseases: genes,
bugs and failed regulation. Nat. Immunol. 2: 759–761.
Gianani, R. and Sarvetnick, N. (1996). Viruses, cytokines, antigens, and
autoimmunity. Proc. Natl. Acad. Sci. U.S.A. 93: 2257–2259.
Marron, M.P., Zeidler, A., Raffel, L.J. et al. (2000). Genetic and phys-
ical mapping of a type 1 diabetes susceptibility gene (IDDM12) to
a 100-kb phagemid artificial chromosome clone containing D2S72-
CTLA4-D2S105 on chromosome 2q33. Diabetes 49: 492–499.
Spielman, R.S. and Ewens, W.J. (1996). The TDT and other family-based
tests for linkage disequilibrium and association. Am. J. Hum. Genet.
59: 983–989.
Wanstrat, A. and Wakeland, E. (2001). The genetics of complex autoim-
mune diseases: non-MHC susceptibility genes. Nat. Immunol. 2: 802–
809.
Immune thrombocytopenia
Aslam, R., Speck, E.R., Kim, M. et al. (2006). Platelet Toll-like receptor
expression modulates lipopolysaccharide-induced thrombocytope-
nia and tumor necrosis factor-production in vivo. Blood 107: 637–
641.
Bussel, J.B. (2002). Novel approaches to refractory immune thrombocy-
topenic purpura. Blood Rev. 16: 31–36.
Chow, L., Speck, E.R., Kim, M. et al. (2010). A murine model of severe
immune thrombocytopenia is induced by antibody- and CD8+ T
cell–mediated responses that are differentially sensitive to therapy.
Blood 115: 1247–1253.
Cines, D.B. and Blanchette, V.S. (2002). Immune thrombocytopenic
purpura. N. Engl. J. Med. 346: 995–1008.
Coopamah, M.D., Garvey, M.B., Freedman, J., and Semple, J.W. (2003).
Cellular immune mechanisms in autoimmune thrombocytopenic
purpura: an update. Transfus. Med. Rev. 17: 69–80.
de Luis, D.A., Varela, C., de La Calle, H. et al. (1998). Helicobacter
pylori infection is markedly increased in patients with autoimmune
atrophic thyroiditis. J. Clin. Gastroenterol. 26: 259–263.
Emilia, G., Longo, G., Luppi, M. et al. (2001). Helicobacter pylori erad-
ication can induce platelet recovery in idiopathic thrombocytopenic
purpura. Blood 97: 812–814.
Fujisawa, K., Tani, P., O’Toole, T.E. et al. (1992). Different specificities of
platelet-associated and plasma autoantibodies to platelet GPIIb-IIIa
in patients with chronic immune thrombocytopenic purpura. Blood
79: 1441–1446.
Gasbarrini, A., Franceschi, F., Tartaglione, R. et al. (1998). Regression
of autoimmune thrombocytopenia after eradication of Helicobacter
pylori. Lancet 352: 878.
Kekomaki, R., Dawson, B., McFarland, J. et al. (1991). Localization of
human platelet autoantigens to the cysteine-rich region of glycopro-
tein IIIa. J. Clin. Invest. 88: 847–854.
Kuwana, M., Kaburaki, J., and Ikeda, Y. (1998). Autoreactive T cells to
platelet GPIIb-IIIa in immune thrombocytopenic purpura. Role in
production of anti-platelet autoantibody. J. Clin. Invest. 102: 1393–
1402.
McKenzie, C.G., Guo, L., Freedman, J., and Semple, J.W. (2013). Cellu-
lar immune dysfunction in immune thrombocytopenia (ITP). Br. J.
Haematol. 163: 10–23.
Nieswandt, B., Bergmeier, W., Rackebrandt, K. et al. (2000). Identifica-
tion of critical antigen-specific mechanisms in the development of
immune thrombocytopenic purpura in mice. Blood 96: 2520–2527.
Portielje, J.E., Westendorp, R.G., Kluin-Nelemans, H.C. et al. (2001).
Morbidity and mortality in adults with idiopathic thrombocytopenic
purpura. Blood 97: 2549–2554.
Provan, D. and Newland, A. (2002). Fifty years of idiopathic thrombo-
cytopenic purpura (ITP): management of refractory ITP in adults. Br.
J. Haematol. 118: 933–944.
Semple, J.W. and Freedman, J. (2006). Mechanisms underlying autoim-
munity in hematology. Drug Discovery Today: Dis. Mech. 3: 231–
235.
Stasi, R., Pagano, A., Stipa, E. et al. (2001). Rituximab chimeric anti-
CD20 monoclonal antibody treatment for adults with chronic idio-
pathic thrombocytopenic purpura. Blood 98: 952–957.
Sukati, H., Watson, H.G., Urbaniak, S.J. et al. (2007). Mapping helper
T-cell epitopes on platelet membrane glycoprotein IIIa in chronic
autoimmune thrombocytopenic purpura. Blood 109: 4528–4538.

Zentilin, P., Savarino, V., Garnero, A. et al. (1999). Is Helicobacter pylori
infection a risk factor for disease severity in rheumatoid arthritis?
Gastroenterology 116: 503–504.
Therapeutic advances
Abrams, J.R., Lebwohl, M.G., Guzzo, C.A. et al. (1999). CTLA4Ig-
mediated blockade of T-cell costimulation in patients with psoriasis
vulgaris. J. Clin. Invest. 103: 1243–1252.
Bell, S. and Kamm, M.A. (2000). Antibodies to tumour necrosis factor
alpha as treatment for Crohn’s disease. Lancet 355: 858–860.
Brandt, J., Haibel, H., Cornely, D. et al. (2000). Successful treatment
of active ankylosing spondylitis with the anti-tumor necrosis factor
alpha monoclonal antibody infliximab. Arthritis Rheum. 43: 1346–
1352.
Cines, D.B., Wasser, J., Rodeghiero, F. et al. (2017). Safety and efficacy
of romiplostim in splenectomized and nonsplenectomized patients
with primary immune thrombocytopenia. Haematologica 102: 1342–
1351.
Kapur, R., Aslam, R., Guo, L. et al. (2016). Thymic-derived tolerizing
dendritic cells are up-regulated in the spleen upon treatment with
intravenous immunoglobulin in a murine model of immune throm-
bocytopenia. Platelets 25: 1–4.
Autoimmune hematological disorders 317
Kremer, J.M. (2001). Rational use of new and existing disease-
modifying agents in rheumatoid arthritis. Ann. Intern. Med. 134:
695–706.
Kuwana, M., Kawakami, Y., and Ikeda, Y. (2003). Suppression of
autoreactive T-cell response to glycoprotein IIb/IIIa by blockade
of CD40/CD154 interaction: implications for treatment of immune
thrombocytopenic purpura. Blood 101: 621–623.
Maini, R.N. and Taylor, P.C. (2000). Anti-cytokine therapy for rheuma-
toid arthritis. Ann. Rev. Med. 51: 207–229.
Mease, P.J., Goffe, B.S., Metz, J. et al. (2000). Etanercept in the treat-
ment of psoriatic arthritis and psoriasis: a randomised trial. Lancet
356: 385–390.
Provan, D. and Newland, A. (2015). Current management of primary
immune thrombocytopenia. Adv. Ther. 32: 875–887.
Semple, J.W., Aslam, R., Kim, M. et al. (2007). Platelet-bound
lipopolysaccharide enhances Fc receptor-mediated phagocytosis of
IgG opsonized platelets. Blood 109: 4803–4805.
Stasi, R., Cooper, N., Del Poeta, G. et al. (2008). Analysis of regulatory
T cell changes in patients with idiopathic thrombocytopenic purpura
receiving B-cell depleting therapy with rituximab. Blood 112: 1147–
1150.
Zufferey, A., Kapur, R., and Semple, J.W. (2017). Pathogenesis and thera-
peutic mechanisms in immune thrombocytopenia (ITP). J. Clin. Med.
6: 16–37.
Chapter 23 Molecular therapeutics in
hematology: gene therapy
William M. McKillop1 & Jeffrey A. Medin2
1 Department of Pediatrics, Medical College of Wisconsin, Milwaukee, WI, USA
2
Departments of Pediatrics and Biochemistry, Medical College of Wisconsin, Milwaukee, WI, USA

Introduction, 319 Adeno-associated virus gene transfer, 330

General comments on gene transfer/therapy, 319 Genetic immunotherapy, 331
Viral vectors for gene transfer, 321 Methods to improve gene therapy safety and suicide gene therapy, 334
Oncoretroviral gene transfer, 321 Conclusions, 335
Lentiviral gene transfer, 327 Further reading, 336
Adenoviral gene transfer, 329

(ALL). Several other CAR-T therapies are following close

Introduction behind this trajectory in what has become a fast-evolving
field in modern medicine.
Approved clinical investigation of gene transfer into humans
Since hematology has contributed so much to the gene-
began with the seminal trial of Rosenberg and colleagues,
sis and progression of human gene therapy, it is the purpose
which involved transplantation of genetically altered lym-
of this chapter both to reiterate the inherent promise and to
phocytes into patients. This landmark trial was closely fol-
revisit some of the remaining obstacles posed by the applica-
lowed by studies of therapeutic gene transfer using other
tion of gene transfer into humans employing hematopoietic
hematopoietic cells. Studies involving the blood system have
cells.
therefore been central to the development of human gene
Pluripotent hematopoietic stem cells (HSCs) are attrac-
therapy. Gene marking or gene therapy protocols are under
tive targets for gene therapy in humans because of their
increasingly intensive investigation worldwide, with the field
capacity for self-renewal and the systemic multilineage dis-
having grown to over 2900 approved clinical trials by the
tribution of their progeny (Figure 23.1). Sustained expres-
beginning of 2019.
sion of transgenes at clinically relevant levels in the progeny
Importantly, some of the first successes in the entire
of HSCs would result in novel and potentially curative treat-
field of gene therapy have recently been realized. Shenzhen
ments for a wide range of blood diseases, including, for exam-
SiBiono GeneTech Co. Ltd. obtained a drug license from the
ple, hemophilia A and B, hemoglobinopathies, hereditary
State Food and Drug Administration of China for Gendicine,
immune deficiencies, and some lysosomal storage disorders
an adenovirus-based therapy designed to treat head and
(LSDs). Even the partial correction of such blood disorders
neck squamous cell carcinoma in 2003. However, it was not
would have a substantial impact on the transfusion needs of
until 2012 and the approval of UniQure’s Glybera, an adeno-
the affected populations. Other hematopoietic cell subpopu-
associated virus-based therapy designed to treat lipoprotein
lations are also important targets for gene therapy. This dis-
lipase deficiency, that a gene therapy treatment strategy
cussion examines some of the targets of gene therapy involv-
was approved in the West. In 2016 Strimvelis, a lentiviral
ing the hematopoietic system and outcomes mediated by a
vector used to treat severe combined immune deficiency,
variety of gene transfer mechanisms.
became the second gene therapy approved in Europe. In
August 2017, the US Food and Drug Administration (FDA)
approved Kymriah (CTL019, tisagenlecleucel), a chimeric
antigen receptor T cell (CAR-T) therapy targeting CD19, General comments on gene
for the treatment of pediatric and young adult patients with transfer/therapy
relapsed/refractory B-cell acute lymphoblastic leukemia
Gene therapy has the potential to revolutionize medical treat-
Molecular Hematology, 4th Edition. Edited by Drew Provan and John G. Gribben. ment by selectively repairing or augmenting the expression
© 2020 John Wiley & Sons Ltd. Published 2020 by John Wiley & Sons Ltd. of defective genes by the insertion of new genetic material.

319
320 Molecular Hematology

Remove
hematopoietic
stem cells

Add genetically Stem cells proliferate

modified virus Harvest and return and gene is distributed
to patient

Fig. 23.1 Ex vivo transduction of hematopoietic stem cells (HSCs). HSCs can be harvested from patients and vector transduced. The
transduced cells are returned to the patient, where all blood cells maturing from the gene-modified hematopoietic stem cells retain a copy of the
transgene. In theory, this gene transfer method could help distribute the gene product throughout the body at clinically relevant levels.

To date, selective repair of genetic defects within the host
genome has proven to be a very difficult endpoint; however
genome editing technologies capable of such repair are
now being investigated. Targetable DNA cleavage reagents
including zinc finger nucleases (ZFN) and transcription
activator-like effector nucleases (TALENs) cut specific
sequences of DNA, allowing for homologous recombination
with recombinant exogenous DNA and the introduction of
an edited sequence at the target site. However, ZFN-based
strategies appear to suffer from context-dependent recog-
nition site binding. There is also some degeneracy in the
TALEN-DNA binding specificity, both nucleases are highly
sensitive to DNA methylation, and they both require very
specific protein–DNA interactions. A third nuclease-based
strategy dependent solely on nucleic acid base pairing
is the clustered regularly interspaced short palindromic
repeats/CRISPR associated protein 9 (CRISPR/Cas9)
system. CRISPR/Cas9 offers an easy-to-engineer and partic-
ularly efficient system for gene editing, but off-target effects
are still possible, as the short recognition sequences used
in the system are rarely unique in the genome. Although
tremendously promising, protocols using these gene editing
strategies are just beginning to enter the clinic. Thus, the
great majority of gene therapy studies conducted to date have
been gene augmentation protocols, involving co-expression
or even amplified overexpression of wild-type cDNA
sequences.
Let us take a step back. An array of techniques has been
described to facilitate gene transfer into blood cells. A
discussion of some of those methods is warranted. Such
techniques can be broadly grouped as either physical methods
or viral vector methods. Physical methods of gene transfer
are generally of lower efficiency and usually provide only
transient gene expression in the absence of selection. The
advantages of physical methods of delivery are that genes
are transferred without viral sequences, which may affect
the biology of the target cell or the host. In addition, large
or multiple genes can be transferred, and gene transfer

is independent of the proliferative status and cell-surface
receptor profile of the target cell. Those methods of physical
gene transfer that have been used in the clinical setting
include electroporation (electrical fields that create channels
in cell membranes, allowing passage of DNA into the cell),
particle bombardment (microscopic gold beads labeled
with DNA, for example, which are forced through the cell
membrane by carbon dioxide (CO2 )-driven pressure), and
liposomal encapsulation of DNA. Liposomal-mediated gene
delivery has gained most acceptance in the clinical arena,
driven by the higher gene transfer efficiencies obtained with
newer liposomal formulations and by safety considerations
as an alternative to viral vector-mediated gene delivery.
Other non-viral gene delivery systems include the use of
plasmid DNA alone (so-called naked DNA), the use of
synthetic polymers, transposons, and the adaptation of
bacterial gene delivery systems. The use of plasmid DNA
alone is particularly efficacious in specific tissues such as
muscle. Thus, non-viral DNA-based gene delivery is gaining
in popularity for use in malignancy and infectious disease,
and in applications directed toward the vascular bed. In
cancer, DNA-based vaccines have been used in clinical trials
where the immunogenic antigen is encoded by the plasmid
itself, with or without an added immunostimulatory gene.
Yet, while this approach is promising, clearly more work on
modulating the biology of the immune response is needed.
Physical methods such as electroporation, direct DNA
transfer, and liposome-mediated DNA transfer have been
used with varying levels of success for transferring genes
into hematopoietic cells. However, transgenes rarely integrate
with physical methods of transfer, except transposons, and
the vectors are diluted in dividing daughter cells, and thus
are generally of little value to long-term stable gene transfer
applications. Nevertheless, when transient gene expression
is sufficient, such as in cancer immunotherapy applications,
physical methods of gene transfer may find a niche.
Viral vectors for gene transfer
As an alternative to the physical methods just described,
protocols requiring higher-efficiency gene transfer or stable
long-term gene expression make use of recombinant viral
vector transduction. The recombinant vectors used are mod-
ified from their wild-type state. The most commonly used
viral backbones in clinical gene therapy trials were previously
based on murine oncoretroviruses and human adenoviruses;
lately recombinant adeno-associated viruses (AAVs) and
lentiviruses (LVs) are preferred. The relative merits and dis-
advantages of some of these vector systems are outlined in
Table 23.1. The particular nuances of the acquired or inher-
ited disorder to be corrected determines which delivery sys-
tem is more appropriate.
Molecular therapeutics in hematology: gene therapy 321
Viral gene transfer methods take advantage of facets of
the normal virus life cycle to facilitate the transfer of genetic
material into target cells. For synthesis of recombinant LV-
based gene transfer systems, for example, the wild-type viral
genome is modified by deletions of most incumbent viral
genes and insertion of therapeutic or marker sequences in
their place. The viral gene products necessary to produce
recombinant virions are then provided in trans by trans-
fections and replication-incompetent virions are produced.
Replication-incompetent virions are deemed safe for clini-
cal applications because they transduce target cells only once;
they are unable to cause a subsequent infection because of the
absence of secondary expression of viral genes required for
replication and virion packaging (Figure 23.2).
Retrovirus-based vectors are attractive for the treatment
of conditions requiring stable long-term expression engen-
dered by genomic integration. Retroviruses also display a
fairly wide tropism, as well as minimal toxicity and minimal
adverse immune responses in human patients. Two retro-
viruses commonly used in clinical gene therapy protocols tar-
geting hematopoietic cells are oncoretroviruses and LVs.
Oncoretroviral gene transfer
Oncoretroviruses are double-stranded RNA viruses belong-
ing to the gammaretrovirus genera of the family Retroviridae.
The viral RNA genome is reverse-transcribed into double-
stranded DNA, which integrates into the genome of tar-
get cells. Currently, oncoretroviral vectors derived from the
Moloney murine leukemia virus (MMLV) and other murine
oncoretroviruses are used for clinical gene transfer protocols
targeting hematopoietic cells.
For gene transfer vectors, the wild-type oncoretroviral
genome is modified by deleting most of the gag, and all
pol and env sequences. Viral sequences that are retained
in the oncoretroviral vector include the packaging signal
and the long terminal repeats (LTRs), which are necessary
for viral integration and often drive transcription of the
marking or therapeutic transgene in the absence of an added
heterologous promoter. These deletions render the vectors
replication-incompetent while making approximately 6–8 kb
of space available for the insertion of desired transgene
sequences. Stable oncoretroviral packaging cell lines have
been engineered to minimize the chance of accidental
replication-competent retrovirus (RCR) production. Plas-
mids used to create these stable packaging cell lines engineer
expression of key components of the parental virus, but have
extensive deletions along with split genes and promoters.
These modifications have minimized recombination-prone
homologous sequences between the added gene transfer
vector encoding the transgene of interest and the stable
packaging cells, thereby significantly reducing the chance of
 322 Molecular Hematology

Table 23.1 Some viral vectors used in the application of gene therapy

Helper-dependent Adeno-associated
Oncoretrovirus Adenovirus adenovirus virus Herpes virus Lentivirus Pox viruses

Advantages r Well-studied r Very efficient r Large cloning r Cell-cycle r Very efficient r Integrates r Efficient gene
r Integrates gene transfer capacity independent gene transfer r Efficient gene transfer
r High transgene r Not immunogenic r Safe r Cell-cycle transfer
expression independent r Large capacity

Disadvantages r Cell-cycle r Immunogenic r Technically r Limited cloning r Transient r Production r Transient

dependent r Transient cumbersome capacity r Cytotoxic r Public expression

r Promoter expression r Helper-virus r Helper-virus perception r Immunogenic

silencing r Pre-existing contamination contamination

r Safety issues immunity r Transient r Production

expression

Potential r Many involving r Cancer r Gene r Cancer r Cancer r Many involving r Cancer

applications hematopoietic immunotherapy augmentation immunotherapy immunotherapy hematopoietic immunotherapy

cells therapy r Gene r Pain cells
augmentation management r Non-dividing

therapy via muscle targets

r Brain delivery
 Molecular therapeutics in hematology: gene therapy 323

Infectious Replication incompetent

HIV virus lentiviral vector

Transfer Packaging Envelope

Therapeutic plasmid plasmid plasmid
gene
Infectious
HIV virus HIV genes

HIV genes
Host nucleus
Host DNA

Replication
incompetent
lentiviral vector

Infectious
HIV virus

Target cell

Therapeutic
Therapeutic transgene
No virions
transgene released
No viral
proteins

Host nucleus

Host DNA
Fig. 23.2 Biology of the HIV retrovirus and the production of HIV-derived, replication-incompetent lentiviral vectors. For wild-type
retroviruses such as HIV, after binding and entry the provirus integrates into the genome of infected cells. Viral proteins and new mRNA molecules
are produced. Virions are then assembled for the next round of infection. For HIV-derived recombinant replication-incompetent lentiviral vectors,
the gag, pol, and env genes have been removed to allow space for subcloning of the transgene of interest. To make a recombinant lentiviral
vector, the transfer plasmid containing the therapeutic gene flanked by the HIV long terminal repeats (LTRs) is introduced into a production cell
line, along with separate plasmids engineering expression of gag, pol, and env genes in trans. The resulting replication-incompetent lentiviral
vector can then bind to, and be internalized by, a target cell wherein integration into the host genome occurs and the vector transgene product is
synthesized. Since the infected target cell lacks the genes necessary to form a new virion, the vector is not infectious secondarily.

RCR being generated. Another measure of safety is naturally
present in oncoretroviral gene transfer systems as these
vectors, when packaged in murine-based packaging cell lines
at least, are rapidly inactivated by human serum. Taking a
lesson from commonly used LV-based gene transfer systems
(see later), investigators have also generated recombinant
oncoretroviruses with self-inactivating (SIN) 3′ LTRs. Thus,
upon reverse transcription, this promoter unit is modified,
making it unable to drive transcription itself if productive
recombination should occur, and also making it unable
324 Molecular Hematology
to drive transcription in the presence of viruses that may
encode cross-promoting functions. Lastly, along with the
inherent safety features described, all viral supernatants and
patient cell samples that have been exposed to clinical-grade
oncoretroviral supernatants are extensively tested for the
presence of RCR and other possible contaminants prior to
infusion into patients.
Oncoretroviral vectors were one of the preferred gene
transfer vectors for early clinical gene therapy protocols.
Their main advantage over other delivery systems was sta-
ble integration of the vector into the host cell genome, allow-
ing long-term transgene expression in the target cell and
its progeny. These vectors also generated minimal immune
responses themselves due to the extensive deletions of wild-
type coding subunits that they have undergone. Disadvan-
tages of oncoretroviral vectors include the requirement of tar-
get cells to be in cycle for genomic integration, the specificity
of virion/receptor binding, which can reduce infection rates
into some target cells depending on the env pseudotyping
employed, the extensive periods of ex vivo culturing that may
be required to obtain efficient gene transfer into key target
cells such as human HSCs, and the randomness of the inte-
gration event itself.
Some of the earliest work with oncoretroviral vectors
involved gene-marking studies in hematopoietic cells. By
marking transduced cells, the long-term distribution and
survival of transplanted cells could be followed in vivo.
The reporter gene that was used most often in early clin-
ical studies was the bacterial neomycin phosphotransferase
(neoR ) gene which, when expressed, confers resistance to
the neomycin analog G418. Other marker genes have also
been used and include those sequences encoding for the
murine heat-stable antigen, the human CD24 and CD25 anti-
gens, the truncated nerve growth factor receptor, modified
CD4 or CD19 or CD34 antigens, and an array of fluores-
cent proteins. The stable and unique integration pattern of
proviral DNA in the genome of marked cells can provide a
permanent marker for individual hematopoietic or malig-
nant cells and their clonal descendants. This marking pat-
tern can be established using a polymerase chain reaction
(PCR)-based analysis that has been developed called ligation-
mediated PCR, which provides information on the actual site
of integration of the provirus. Clinical applications in which
gene marking has provided new and important information
include the infusion of oncoretrovirally marked, autologous,
tumor-infiltrating lymphocytes into patients with advanced
melanoma and the infusion of oncoretrovirally marked bone
marrow or peripheral blood into patients with myeloid
leukemia, myeloma, and neuroblastoma. The study of such
patients offers three important lines of investigation:
1 Is retroviral-mediated gene transfer relatively safe?
2 Do genetically altered bone marrow or blood stem cells
contribute to long-term hematopoiesis?
3 Do malignant cells or their precursors contribute to the
high relapse rates observed after myeloablative therapy and
autologous HSC transplantation?
A number of groups have reported the consequences of
infusing gene-marked bone marrow cells into humans and
the contribution of contaminating tumor cells in the graft
to disease recurrence. The first important observation from
these studies was that oncoretrovirus-mediated gene transfer
appears relatively safe. No detrimental effects, either on the
autograft or in patients, have been reported in these marking
studies. Replication-competent oncoretrovirus has also not
been detected at appreciable levels in patients participating in
clinical trials. That said, in two clinical gene therapy protocols
that will be further described, T-cell leukemias developed in
five patients and appear to have resulted from the integration
of the therapeutic provirus into the regulatory region of a cer-
tain oncogene (see later). The second important observation
from earlier gene-marking studies was that multiple HSCs
contributed to long-term hematopoiesis, albeit at relatively
low levels. Interestingly, rather than all daughter hematopoi-
etic cells being derived from a single HSC, it thus appears that
multiple stem cells contribute to the formation of the renew-
ing blood system.
In the first gene-marking studies in children reported by
Brenner and colleagues, 2–15% of clonogenic hematopoietic
progenitor cells were marked after autologous bone marrow
transplantation. The marker gene was detectable for up to
four years after transplant and was found in granulocytes,
B cells, and T cells, at least by PCR. In the earlier adult
gene-marking studies, however, oncoretroviral transduction
of marrow or peripheral blood HSCs has resulted in detec-
tion of integrated vector in only a very low percentage of
peripheral blood cells. For example, in one study, although
the marker gene persisted for up to two years, neoR-positive
cells could only be detected intermittently with analyses
employing PCR. Work by a number of groups suggested
that modification of the transduction protocols would
result in a significant improvement in the engraftment
of genetically modified HSCs. Along these lines, since
preclinical experience indicated that the use of bone marrow
stroma enhances gene transfer into HSCs, this approach was
evaluated in some early clinical trials. However, the results
of this adaptation proved to be no better than those just
discussed and, in general, are very similar to those observed
in the pediatric gene transfer studies.
In total, data on more than 40 patients enrolled in gene-
marking studies during bone marrow transplantation for
acute myeloid leukemia, neuroblastoma, chronic myeloid
leukemia, breast cancer, myeloma, acute leukemia, or non-
Hodgkin lymphoma have been presented. Of the relapsed
patients, gene-marked tumor cells have been detected in
a high percentage in many studies. In one patient with
acute myeloid leukemia, the simultaneous detection of a

Table 23.2 Strategies for optimization of stable gene transfer into hematopoietic stem cells
Strategy
Inducing recipient cells to cycle
Increased cell–virus contact
Increase viral receptor levels on target cells
Alternatively pseudotyped recombinant retroviruses
Modified retroviral vectors to increase efficiency and
infect non-cycling cells
Positive selection of transduced cells
cytogenetic marker along with the neoR gene confirmed that
gene-marked cells contributed to relapse. A second critical
observation was made a number of years ago when Rill and
colleagues reported that a multiplicity of neuroblastoma cells
in the graft contributed to relapse. The high frequency of
gene-marked relapse, despite the very low frequency of trans-
fused malignant cells, strongly suggests that a large percent-
age of tumor cells in the graft contribute to relapse or, alterna-
tively, that tumor cells susceptible to retroviral gene marking
are uniquely capable of engraftment and clonal expansion.
The gene-marking studies mentioned set the stage for
the investigation of multiple maneuvers designed to increase
long-term gene transfer efficiency into HSCs. Some of the
strategies being pursued are described in Table 23.2. These
include increasing the true direct target cell-to-virus contact
ratio by incorporating prior HSC enrichment, altering retro-
viral envelope utilization, or increasing co-localization of
vector and cell in gene transfer protocols. Newer vector sys-
tems improving on both oncoretroviral and lentiviral back-
bones and incorporating alternative envelope pseudotyping
may also increase gene transfer efficiency, while refinement
of growth factor combinations may more efficiently induce
HSC cycling in shorter time periods, thus improving retrovi-
ral integration without compromising engraftment. Many of
these approaches have now been incorporated into clinical
protocols and have likely contributed to some of the recent
successes in this field, as described shortly.
Oncoretroviral vectors have also been used in chemopro-
tection studies. Cancer chemotherapy often involves induced
hematopoietic system toxicity, leading to myelosuppression.
Gene therapy strategies aiming to protect hematopoietic cells
Molecular therapeutics in hematology: gene therapy 325
Method
Optimization of ex vivo cytokine stimulation
Collection of cells during recovery phase after myeloablation or mobilization
Culture on stromal layers
Centrifugation of cells and virus during transduction (spinoculation)
Viral supernatant flow-through systems
Coat dishes with fibronectin fragment
Higher viral titers and multiple exposures
Increase levels of amphotropic receptor by phosphate depletion
Transfer viral receptor into cell by adenovirus or adeno-associated virus
Target subpopulations of cells that have high levels of receptors
Exploiting GALV, RD114, 10A1 receptors for entry
VSV-G envelope to expand tropism and allow virion concentration
Lentivirus- and foamy virus-based vectors
Add positive selectable marker to vector: metabolic, fluorescent, cell-surface
against the toxicity of chemotherapeutic drugs, i.e. chemo-
protection studies, are under exploration. The gene-marking
studies just mentioned suggest that normal bone marrow
cells could be removed, transduced with a vector engineering
expression of a drug resistance gene, and returned to patients.
Theoretically, such protected cells would then expand clon-
ally after chemotherapy treatment and confer relative resis-
tance to a cytotoxic or cytostatic agent, allowing further dose
escalation and a potential cure of some patients. Further-
more, such drug resistance genes, if expressed in a multi-
cistronic format with a second therapeutic gene product, may
even serve as selectable markers allowing ex vivo or in vivo
enrichment or “preselection” of functionally gene-marked
cells, which may further enhance observed clinical outcomes.
The original chemoprotection studies made use of
the human multidrug resistance 1 (MDR-1) gene (P-
glycoprotein). The approach used for gene transfer there
was generally similar to that used for the human gene-
marking trials. MDR-1 transduced and transplanted cells
engrafted and conferred drug resistance to bone marrow
cells in vivo, and also allowed for positive selection of
MDR-1 transduced cells by chemotherapy. Infusion of
oncoretroviral MDR-1-modified HSCs was found to be safe,
with no deleterious effects of infusing such gene-modified
cells observed, but only minimal levels of gene marking
were noted following transplantation. Multiple studies have
investigated an alternative strategy aiming to overexpress
mutant methylguanine methyltransferase (MGMT) in HSCs
to allow them to survive multiple rounds of chemotherapy
targeting glioblastoma, for example. The mutant MGMT was
safe, allowed an increase in the mean number of tolerated
326 Molecular Hematology
chemotherapeutic cycles, and resulted in improved clinical
outcome for the seven patients treated.
The first human clinical gene transfer trials for inher-
ited single-gene disorders used oncoretroviral vectors and
focused on adenosine deaminase (ADA) deficiency, a con-
dition that leaves the body unable to produce leukocytes
and thus predisposed to infection. Recombinant oncoretro-
viruses containing the normal human ADA cDNA were
transferred into either peripheral blood T lymphocytes,
bone marrow cells, or cord blood cells from ADA-deficient
patients. T-lymphoid cells expressing the normal ADA gene
have a selective growth and survival advantage over ADA-
deficient cells, even though patients were maintained on
pegylated-ADA enzyme replacement therapy (ERT) for ethi-
cal reasons. In those early studies, patients who had received
multiple infusions of autologous ADA-transduced blood cells
had increased levels of enzyme in their serum, and up to 20%
of their peripheral blood T cells were found to carry provirus
for some period. Indeed, long-term follow-up has revealed
that one patient has maintained this level of marking for over
10 years. In two of the three initial trials in which patients
received autologous marrow or cord blood cells transduced
with ADA cDNA-containing oncoretroviruses, between 12%
and 40% of colony-forming units (CFUs) were transduced,
and genetically marked cells were found for greater than
one year after infusion. In the one study, by Hoogerbrugge
and colleagues, in which provirally marked cells were not
maintained for greater than six months, there was lower in
vitro gene transfer efficiency of 5–12% CFU prior to trans-
plant. In the cord blood study, there was evidence for a
selective growth advantage of T cells, as there were higher
levels of marked T cells than myeloid cells, even while the
patients were maintained on progressively lower doses of
PEG-ADA therapy. ERT was withdrawn from one patient
and the number of T cells carrying the provirus increased to
30%; however, the total number of B lymphocytes and natu-
ral killer cells dropped, and the patient had reduced immune
function. That patient subsequently resumed PEG-ADA
treatment.
Inherited immunodeficiencies, such as severe combined
immunodeficiency (SCID) of ADA (see earlier) and X-linked
SCID (see later), have become focal points for the potential
benefits and some of the potential hazards of gene therapy.
The studies in ADA-SCID likely represent the first tangible
correction of an inherited disorder by stable transfer of a
therapeutic gene into primitive hematopoietic cells and their
progeny, although it is difficult to state this unequivocally
due to the simultaneous administration of ERT. However,
those initial results have been surpassed recently for ADA-
SCID in a study published by Aiuti and colleagues, where
productive transfer of the ADA gene was effected into CD34+
cells of non-myeloablated recipients for whom ERT was not
available. Many of the incremental improvements in the gene
transfer protocols mentioned were adapted in that protocol,
and higher levels of gene marking (up to 25% of CFU-C ini-
tially) and functional correction were demonstrated. Long-
term multilineage hematopoietic cell marking was also found
in both patients. This study demonstrated, without the con-
founding implications of ERT, that stable long-term correc-
tion of this immune deficiency could be accomplished using
this therapeutic approach.
Impressive results have also been demonstrated in efforts
to correct another inherited immunodeficiency, X-linked
SCID (SCID-X1). This is caused by a deficiency of the com-
mon γ-chain subunit of the receptors for the cytokines inter-
leukin (IL)-2, IL-4, IL-7, IL-9, and IL-15, and is thereby not
a candidate for soluble factor augmentation therapy. Expres-
sion of the γ-chain was also expected to offer a growth advan-
tage to productively transduced cells. This was indeed the
case and in a first report in 2000, two patients were shown to
have fully corrected immune function as a result of the gene
therapy. Since this landmark first description of this benefi-
cial outcome, other SCID-X1 patients have also been treated
by this method. The original clinical trials treated 20 patients
and resulted in the majority of these individuals retaining
improved and stable immune function. Yet even with these
impressive results, this study highlights areas where the field
of gene therapy must still progress in its understanding. This
is because of reports that five of the patients receiving the
corrective 𝛾-chain gene in two separate clinical gene ther-
apy protocols for SCID-X1 employing similar vectors and
transduction protocols went on to develop acute T-cell lym-
phoblastic leukemia-like disease approximately three years
after the transplantation of oncoretrovirally transduced cells.
In at least two of these patients the development of this prolif-
erative disorder has debatably been ascribed to a deleterious
integration event that may have caused dysregulated expres-
sion of a proto-oncogene called LMO2. However, it should
be emphasized that since analogous vector backbones and
transduction conditions have been used for other studies,
such as the ADA-SCID trials mentioned earlier, without the
emergence of adverse events, the possibility exists that these
leukemia-like diseases are a specific consequence of over-
expression of the 𝛾-chain gene itself, since it impacts many
diverse signaling pathways that affect a number of cellular
functions in vivo.
Still, a second attempt at SCID-XI gene therapy was
launched in 2008, with results published in 2012 detailing
follow-up on the nine patients enrolled at sites in France
and the USA. The oncoretroviral vector was modified to be
a SIN vector and to contain deletions in the viral enhancer
sequences that may have activated LMO2. Of the eight
patients who received treatment, seven displayed immune
system reconstitution, and analysis of vector insertion dis-
played significantly reduced clustering within LMO2 or other
proto-oncogenes. Although very promising, these children

will be followed for 15 years and longer to further assess the
efficacy and safety following treatment.
Results from another study involving the hematopoietic
system have also generated some close scrutiny. Clonal dom-
inance of cells with specific integration events was observed
in a recombinant oncoretrovirus-based clinical gene therapy
trial for the X-linked form of chronic granulomatous dis-
ease. There patients have impaired immune function and
cannot sufficiently resist bacterial and fungal infections, due
to an inability of neutrophils to generate superoxide ions;
expression of the CYBB gene can abrogate this. Interest-
ingly, the amplified integration locus involved the MDS1
and EVI1 genes in that case, which differs from that men-
tioned in the SCID-X1 trials and may reflect alterations
in the activity of myeloid cells post-transduction with this
cDNA, rather than a specific sequence preference for vector
integration.
Another particular group of inherited disorders currently
undergoing gene therapy preclinical and clinical trials are the
LSDs. LSDs are genetic diseases that result from the loss of
metabolic genes that normally work in the lysosome to break
down particular cellular molecules. Over time, the result of
the disrupted enzyme activity is an accumulation of substrate,
leading to a broad spectrum of debilitating symptoms. As
most LSDs are the result of the loss or malfunction of a single
enzyme, and as even a small fraction of cells producing work-
ing copies of the malfunctioning enzymes may be sufficient
to cross-correct lysosomes and positively impact the disease,
LSDs are promising targets for gene therapy.
Gaucher disease is an LSD resulting from a deficiency
in the enzyme glucocerebrosidase that is manifested mainly
in macrophages, and it has been proposed that this defect
may be especially amenable to treatment by therapeutic gene
transfer into HSCs. Results from two earlier clinical gene
transfer studies targeting this disorder have been reported.
Mobilized peripheral blood or marrow CD34+ cells from
Gaucher patients were transduced with an oncoretroviral
vector that engineered expression of glucocerebrosidase and
then infused into non-myeloablated autologous recipients. In
both studies, transduced cells were detected at low levels in
blood and/or marrow leukocytes. One patient who received
cells transduced with an MFG-based oncoretroviral vector
manifested increased levels of enzyme activity corresponding
to 50% of normal, which was maintained for 12 months after
infusion. No therapeutic benefit or increased enzyme activity
was detected in other patients or in the other study. Develop-
ment of viable murine models for this LSD have been very
beneficial for the testing and implementation of novel gene
therapy strategies involving some of the conditional manip-
ulations already mentioned or employing newer vectors. In
addition to Gaucher disease, gene therapy clinical trials are
underway investigating a number of other LSDs, including
Fabry disease, mucolipidosis, mucopolysaccharidosis type I,
Molecular therapeutics in hematology: gene therapy 327
II, and III, metachromatic leukodystrophy, neuronal ceroid
lipofuscinoses, and Pompe disease.
Collectively, the results from the gene transfer studies for
genetic diseases described here (and others not directly men-
tioned) illustrate several points that will likely impact on the
clinical success of gene transfer protocols for other single
gene inherited disorders.
1 The presence in patients of cells carrying the provirus for
greater than one year has demonstrated the feasibility of gene
therapy and newer vector systems may further enhance this.
2 The transfer of genes that provide a growth or survival
advantage can provide long-term expression and mainte-
nance of transduced cells.
3 A selective advantage of transduced cells can compen-
sate for a modest gene transfer efficiency and yet, if too
strong, may lead to amplification of deleterious transforma-
tion events.
4 Using the incrementally optimized protocols, gene deliv-
ery using recombinant viruses along with their subsequent
transgene expression levels may presently be sufficient to cor-
rect a number of disorders that are directly or indirectly man-
ifested in the hematopoietic system.
Although still used in groundbreaking clinical stud-
ies tracking toward FDA approval, enthusiasm about
oncoretroviral vectors was somewhat reduced following the
leukemogenesis reports post-X-SCID treatment described
earlier. We now know that oncoretroviral vectors have a
predisposition to integrate at certain genetic “hotspots,”
including within genes, and in regions immediately sur-
rounding transcriptional start sites, leading to the potential
for dysregulation of nearby endogenous genes. In an effort to
overcome these obstacles and improve therapeutic safety and
efficiency, replication-incompetent complex retroviruses
based on lenti-retroviruses (LVs) and human foamy virus
have been developed. These vectors can integrate into
these and other key non-dividing target cells and are less
predisposed to integration near active genes.
Lentiviral gene transfer
LVs and human foamy virus are also members of the family
Retroviridae. They are sometimes termed “complex” retro-
viruses, in contrast to oncoretroviruses, which are termed
“simple” retroviruses. Foamy viruses have a large genome
(and hence a large transgene carrying capacity), do not cause
any known disease, and can be used for in vivo gene trans-
fer because they are not effectively inactivated by human
serum. LVs also show significant promise as gene transfer
vehicles and many laboratories around the world and subse-
quent clinical trials have adopted this alternative delivery sys-
tem. Current LVs contain less than 25% of the HIV-1 genome
and seem to be much less susceptible to “shutdown” effects
328 Molecular Hematology
compared with recombinant oncoretroviruses. To further
increase safety, many laboratories now use third-generation
recombinant LVs that have a SIN 3′ LTR. Much like SIN
oncoretroviruses, SIN LV minimize the risk that replication-
competent lentivirus will be created. SIN vectors also reduce
the chance that the expression of endogenous genes located
near the insertion site will be enhanced by viral LTR activity,
and reduce or eliminate transcriptional interference between
the LTR and the promoter driving the transduced transgene.
To further minimize the chance of recombination lead-
ing to the production of replication-competent lentivirus,
some laboratories even use up to seven separate plasmids
in transient transfections of packaging cell lines to gener-
ate recombinant virions (although three separate plasmids
is more the norm). LVs can be readily pseudotyped with
alternative envelope proteins, such as VSV-g, which expands
the tropism of the virions and also allows stable concentra-
tion of effective viral titer by ultracentrifugation. LVs can
also transduce non-dividing cells. There is thus considerable
optimism that recombinant LVs will overcome some of the
shortcomings of MMLV-based oncoretroviruses by facilitat-
ing delivery of transgenes of interest to a wide spectrum of
cell types, including HSCs. Furthermore, due to a marked
increase in LV-mediated transduction efficiencies over previ-
ous methods, it may be possible now to better manage other
secondary components of the ex vivo gene therapy proce-
dure, such as enhancing engraftment of transduced cells in
recipients by minimizing culture periods and exposure to
differentiation-inducing cytokines, for example. Alterations
have also been made to LV backbones such as the addi-
tion of a central polypurine tract and a woodchuck hepati-
tis virus post-transcriptional regulatory element, to possibly
enhance nuclear import or enhance transgene expression lev-
els, respectively.
As previously mentioned, LV genomic integration
patterns differ from oncoretroviral integration patterns.
Unlike oncoretroviral vectors, which show a propensity
to integrate into promoter-proximal regions, LVs tend to
distribute throughout open chromatin and thus present less
of an insertional mutagenesis risk. That said, due to their
increased efficiency, LVs tend to integrate into permissive
cells in higher copy numbers than oncoretrovirus-based
vectors. This increase can be good in order to generate
higher levels of transgene products for correction, but it
also necessitates careful planning concerning the effective
multiplicity of infections (MOIs) and the tolerance of
the infected cell population to multiple proviral copies.
Indeed, it may turn out that MOIs with LVs actually need
to be reduced to minimize this multiplicity effect in some
protocols. Importantly, in a landmark first LV study in
HIV-AIDS patients, no detectable recombination occurred
between the gene transfer vector and wild-type HIV-1, even
when the latter was present in very high copy numbers in
patient blood. Many gene therapy trials using LVs have been
completed to date and many more are underway.
As previously mentioned, Kymriah, the first gene therapy
approved in the USA, utilizes the LV delivery system. Kym-
riah is a CAR-T cell therapy based on infusion of autolo-
gous T cells transduced with an LV engineering expression
of a CD19-directed CAR for the treatment of ALL. Kym-
riah has been evaluated in a number of clinical trials. One
trial was based on the treatment of 30 children and adults;
27 patients experienced complete remission, including many
patients in whom stem cell transplantation had failed. The
CAR-T cells were detectable in the blood, bone marrow, and
cerebrospinal fluid of patients who displayed a response. All
patients experienced cytokine-release syndrome, but this was
controllable with tocilizumab. Positive outcomes in a subse-
quent global trial, ELIANA, helped secure FDA approval. A
phase II trial investigating Kymriah’s effect on diffuse large
B-cell lymphoma is underway (JULIET). Kymriah has also
been used in the clinic to treat chronic lymphocytic leukemia
(CLL). In one study, 14 patients were treated with autolo-
gous T cells transduced with Kymriah LV and infused at
varying doses. More than 50% of patients displayed a posi-
tive response. The in vivo expansion and persistence of the
CAR-T cells correlated with responses, no patient displaying
a complete remission relapsed, and minimal residual disease
was not detectable in patients displaying a complete remis-
sion. All responding patients developed B-cell aplasia and
experienced cytokine release syndrome.
Although the first LV-based gene therapy won FDA
approval in 2017, the first protocol in humans to employ
recombinant LVs was conducted in the early 2000s by
Levine and colleagues. That trial aimed to use the VRX496
(Lexgenleucel-T) LV to treat HIV infection. VRX496 is an
LV engineering expression of an antisense transgene against
the HIV envelope. The LV is delivered to autologous CD4
T cells, which are then reinfused into the patient. Infection
with HIV results in antisense expression and disruption of
HIV replication. In one clinical study, a single IV infusion
of VRX496-containing CD4 T cells into five patients was
found to be safe. Engraftment was observed. No evidence of
insertional mutagenesis was noted, nor was recombination
between the gene transfer vector and wild-type HIV. Patients
displayed improved CD4 counts, and in one patient a sig-
nificant decrease in viral load was seen. A subsequent study
investigated multiple VRX496 infusions into 17 patients
and noted a significant decrease in viral load. No evidence
of clonal selection of LV-transduced T cells or integration
enrichment near oncogenes was detected. Several other
clinical trials have used gene therapy to target HIV. One
study involved treatment of four HIV patients with an LV
delivering three RNAs: a shRNA targeting an exon shared by
the HIV genes tat and rev, an RNA hairpin transactivating
region decoy that antagonizes viral transactivation, and

an anti-CCR5-specific hammerhead ribozyme intended
to block viral entry. The therapy was well tolerated and
the shRNA was detected in primary blood mononuclear
cells and/or primary blood granulocytic cells at least six
months post-treatment in all four patients. Another ongoing
study is testing an LV engineering expression of an anti-
CCR5 shRNA and C46 peptide to block HIV entrance into
T cells.
LVs have also been used to treat X-linked adrenoleukodys-
trophy (ALD), a severe brain demyelinating disorder result-
ing from a deficiency in ALD protein. Autologous HSCs were
removed from two patients and engineered to express ALD
by LV vector transduction. Between 24 and 30 months post-
treatment, polyclonal reconstitution had occurred, with 9–
14% of granulocytes, monocytes, and T and B lymphocytes
expressing the ALD protein. Beginning 14–16 months after
treatment, cerebral demyelination had stopped.
Marina Cavazzana-Calvo’s group has published on their
first patient treated with a LV delivering an anti-sickling β-
globin gene into autologous HSCs. Fifteen months after treat-
ment, therapeutic anti-sickling β-globin was still detectable
and sickle crises, as well as other phenotypic hallmarks
of sickle cell disease, were absent in the patient. No
adverse events unrelated to busulfan conditioning were
reported. Cavazzana-Calvo’s group also used LVs to treat β-
thalassemia, a form of thalassemia resulting from reduced
or absent hemoglobin β chain synthesis that can result in
severe anemia. More than 30 months after LV β-globin gene
transfer, an adult patient with severe β-thalassemia who had
required monthly transfusions since childhood had been
transfusion independent for the previous 21 months.
Several trials have been conducted using LV-based gene
therapy treatments for Wiskott–Aldrich syndrome, a rare
immunodeficiency associated with severe microthrombocy-
topenia, eczema, recurrent infections, and susceptibility to
autoimmunity and lymphomas. In one study, autologous
HSCs corrected by LV transduction were infused in seven
patients; 24 months post-treatment six of the seven patients
were alive and showed sustained clinical benefit, with sub-
stantially reduced disease-related days of hospitalization. The
patient who died succumbed to a pre-existing drug-resistant
herpes virus infection. In a separate study, three patients
were treated analogously and showed stable engraftment with
improvements in platelet counts, immune functions, and
clinical scores. Importantly, in these three patients, the LV-
mediated therapy did not induce integrations near onco-
genes, and no aberrant clonal expansion was observed 20–
32 months following the study. In a second cohort of four
patients treated by this same protocol, LV-transduced HSCs
repopulated both the bone marrow and the periphery, with
a normal distribution of B-cell subsets, and led to normal
immunoglobulin and autoantibody expression in all treated
patients.
Molecular therapeutics in hematology: gene therapy 329
Furthering the LSD clinical work presented earlier, Luigi
Naldini and Alessandra Biffi and colleagues have targeted
metachromatic leukodystrophy, a demyelinating lysosomal
storage disease caused by arylsulfatase A deficiency. Patients
received autologous HSCs transduced with a LV-encoding
arylsulfatase A; 18 months post-treatment the great major-
ity of patients displayed stable engraftment of gene-corrected
cells, increased arylsulfatase A activity in circulating HSCs
and in cerebrospinal fluid, a reduction of the glycosphin-
golipid buildup normally associated with disease pathology,
and improved functional outcome.
Direct bilateral injection of an LV into the putamen
was used to restore dopamine production in patients with
advanced Parkinson’s disease; 15 patients received the LV, 3
at a low dose, 6 at a mid dose, and 6 at a high dose. During the
first 12 months of follow-up, 51 mild and 3 moderate adverse
events were reported; however, a significant improvement
in mean motor scores off medication was reported in all
patients.
Attempts to use vectors other than traditional HIV-1-
derived LVs have been made and are now entering the clinic.
One such strategy is based on equine infectious anemia virus
(EIAV) and has been used to treat neovascular age-related
macular degeneration (NVAMD). NVAMD is an eye disease
leading to vision loss. Endostatin and angiostatin were deliv-
ered by EIAV LV to patients with advanced NVAMD by sub-
retinal injection in three dosing cohorts. Each dose was well
tolerated with no toxicities. Long-term transgene expression
and reduced disease progression were noted.
Adenoviral gene transfer
Recombinant adenoviral gene transfer systems are com-
monly used for some gene therapy applications, especially
in immunotherapy protocols. Adenoviruses infect cells effi-
ciently in vitro and in vivo, express high levels of trans-
gene products, can infect both cycling and stationary cells,
and exhibit wide tissue tropisms. They can also be pro-
duced at very high titers and have a large carrying capac-
ity for foreign cDNAs. Adenoviral vectors have been used
to efficiently transfer genes into a variety of human tumors
by direct injection, and into circulating hematopoietic pro-
genitor and malignant cells in ex vivo transduction proce-
dures. Nevertheless, circulating B and T lymphocytes have
generally proven relatively resistant to adenoviral-mediated
gene transfer. So had CD34+ hematopoietic cells, until it was
found that serotype 35 fiber knob domains can help over-
come this blockade. Furthermore, many patients have pre-
existing immunity to common adenoviral serotypes used
in clinical protocols. This can lead to a strong immune
response to the vector and severely restricts readministration
efficacy.
330 Molecular Hematology
The receptor for most adenoviruses used in gene trans-
fer applications has been identified as the coxsackie-and-
adenovirus receptor protein, and has been useful in deter-
mining the mechanism of adenovirus binding and cell entry.
Adenoviral vectors do not integrate into the genome at an
appreciable frequency and are lost from most cycling tar-
get cells within weeks of transduction. Thus, in comparison
to integrating vectors, adenovirus may appear comparatively
safe; however, they are not without their own set of risks. In
1999 teenager Jesse Gelsinger died due to a massive immune
response generated to the adenovirus vector he had been
treated with in a trial at the University of Pennsylvania. Ade-
noviral constructs have since been refined, with many potent
immuno-stimulatory viral sequences having been eliminated
so as to leave only the necessary inverted terminal repeats,
and thus reduce the immunogenicity of the vector. Oncolytic
adenovirus vectors, which replicate only in permissive cancer
cells, have been developed and hold promise for the treat-
ment of solid tumors and leukemic cells. For example, one
such vector system only replicates in cells harboring defective
p53 tumor-suppressor gene function. With these modifica-
tions, adenoviruses may be ideal vectors if the generation of
an immune response and/or short-term high levels of trans-
gene expression are desired; however, they are not suitable for
therapies that require long-term expression or integration of
the transgene.
Adeno-associated virus gene transfer
Versions of the human parvovirus adeno-associated virus
(AAV) are also being used more frequently as vehicles
for therapeutic gene transfer. AAVs are non-pathogenic in
humans and establish a latent infection in the absence of ade-
novirus itself or helper functions provided by other viruses.
Wild-type AAV integrates at some frequency into a specific
site on human chromosome 19q13.3-qter, which appears to
be a relatively benign location. Only the AAV inverted termi-
nal repeats are required as transcriptional units for recombi-
nant AAV vectors, which allows about 4 kb for foreign inserts
given the size of the parental viral genome. However, both the
efficiency and specificity of integration are lost without the
wild-type genes; recombinant AAV integrates at a very low
frequency. Due to the risk of insertional mutagenesis posed
by integrating vectors, this, along with low immunogenicity,
has become a major selling point for AAV gene therapy.
With that said, there have been reports of AAV delivery
resulting in hepatocellular carcinoma formation in mice,
and wild-type AAV2 has been found to integrate near and
potentially dysregulate oncogenes in human hepatocellular
carcinomas. Thus, further study on recombinant AAV
gene therapy vector integration is warranted. AAV vectors
also carry a small genetic payload, leaving them clinically
useful only for the delivery of relatively small transgenes.
As usually produced, AAVs have titers of approximately
106 particles ml−1 and can be concentrated to greater than
109 particles ml−1 , although in each preparation many
non-infectious particles are generated. Human CD34+
hematopoietic stem/progenitor cells have been transduced
with recombinant AAV vectors, with up to 80% of CFU
carrying the transgene. Interestingly, the optimal use of
recombinant AAV-based gene transfer vectors to impact the
hematopoietic system may be in secondary manifestations.
In a landmark study, Kay and colleagues injected recom-
binant AAV vectors that engineered expression of factor IX
(F.IX) into skeletal muscles of severely afflicted hemophilia
B patients. Long-term vector persistence was observed in
that study, as were slight increases in the circulating levels
of the corrective proenzyme. Subsequent studies aimed to
improve on the limited F.IX expression seen in this early AAV
work. Using a new AAV serotype (AAV8) and an improved
transgene design, a dose-escalation study infusing F.IX AAV8
into peripheral veins was conducted. F.IX expression was
stable three years after infusion in most patients and aver-
aged 5.1% of normal in the high-dose group. The treatment
was well tolerated. However, it should be noted that some
of these patients were given an immunosuppressive corticos-
teroid four to eight weeks post-treatment. Importantly, most
of the high-dose patients were able to reduce their exogenous
F.IX treatment following gene therapy.
Considerable work has also gone into investigating the use
of AAV to treat type 2 Leber congenital amaurosis (LCA).
LCA is an inherited retinal dystrophy that causes loss of
vision at an early age. In three independent stage I/II trials
using Luxturna (voretigene neparvovec, hRPE65), an AAV2
construct, a total of 20 patients were treated by subretinal vec-
tor administration. In all three trials, all patients experienced
improvements in visual sensitivity, peaking roughly a year
post-treatment. In two of the three trials, the patients began
to lose the AAV-confirmed benefit after two to three years.
However, sustained improvement in their condition was
noted in the third trial. A following phase III study reported
that 65% of participants displayed maximal improvement in
the trial’s primary endpoint: bilateral multiluminance mobil-
ity testing, one year after treatment. On the back of this study,
Luxturna is a front-runner to be the first AAV gene therapy
approved by the FDA.
In α1-antitrypsin (AAT) deficiency, mutations in the SER-
PINA1 gene led to reduced liver secretion of AAT and
impaired anti-protease activity in the lung, leading to early-
onset pulmonary emphysema. In an early trial, AAV2 encod-
ing the AAT cDNA sequence was injected intramuscularly
into the upper arm of 12 patients at various doses, all of which
were below comparable levels used in rodent studies. Serum
AAT levels were found to be above background in only 1 of
the 12 subjects dosed, which was fairly unsurprising given

the relatively low dosage used. A follow-up trial made use of
an AAV1 vector, as it displayed 100-fold higher hAAT serum
levels than that seen with AAV2 constructs in murine mus-
cle transduction models. This follow-up AAV1 study resulted
in detectable AAT in the serum of almost all patients, and
they maintained subtherapeutic levels of this factor for at
least 90 days and up to one year post-treatment. A phase 2
trial followed aiming to employ a dose escalation to achieve
therapeutic transgene levels. To reach the maximum dose,
AAV was delivered in 100 separate intramuscular injections.
A linear relationship between dose and serum AAT levels was
noted; however, even at the maximum dosage serum AAT
levels were still below a therapeutically relevant level.
Another AAV-based gene therapy trial investigated treat-
ment of spinal muscular atrophy type 1 (SMA1) in 15 infants.
SMA1 is caused by low levels of the survival motor neuron
protein and affects all muscles in the body. That study was a
dose-escalation trial delivering an AAV9-based (AVXS-101)
construct intravenously through a peripheral limb vein. Thus
far, results appear positive, as this treatment schema has led
to improved survival as well as improved motor skills, espe-
cially in the higher-dose cohorts.
Attempts at using DNA-editing ZFNs (see earlier) have
begun in the clinic, with one study using AAV to deliver
an anti-CCR5 ZFN into 12 HIV patients. Years following
treatment, the gene-modified T cells could be detected in
circulation, but only at low levels. Many other AAV vectors
are currently being evaluated in both preclinical and clinical
proof-of-concept studies.
Genetic immunotherapy
A number of potential approaches using gene therapy for
hematological malignancies can be considered (Table 23.3).
Proof of concept for immunotherapy as a valid approach to
the treatment of hematological malignancy has been pro-
vided by earlier clinical studies that demonstrated that infu-
sion of allogeneic T cells can eradicate minimal disease in
patients relapsing after allogeneic transplant. Unfortunately,
Table 23.3 Approaches to gene therapy targeting hematological malignancies
Strategy Genes employed
Gene replacement p53, p16, Rb
Gene inhibition BCR-ABL, MYC, CCND1 (cyclin D1), BCL-2
Suicide genes Thymidine kinase, cytosine deaminase
Drug resistance genes MDR-1, DHFR, MGMT
Immunotherapy IL-2, IL-12, B7-1, CD40L, GM-CSF,
tumor-associated antigens (TAAs)
Molecular therapeutics in hematology: gene therapy 331
these encouraging allogeneic responses require a haploiden-
tical T-cell donor, which is not available to the vast majority
of patients. Another more widely applicable approach is the
use of autologous tumor cells that have been genetically engi-
neered to express immunostimulatory cytokines. Trials using
this strategy in myeloma, low-grade lymphoma, leukemia,
and CLL are being pursued.
Many tumor cells express unique antigens on their cell
surface, either alone or as proteolytically cleaved peptides in
association with major histocompatibility complex (MHC)
class I molecules. The specificity and sheer quantity of these
tumor-associated antigens (TAAs) may allow immune effec-
tor cells to distinguish tumor from normal tissue (Fig-
ure 23.3). Some of these TAAs have been isolated and shown
to be recognized by T-cell receptor (TCR) complexes on
human cytotoxic T lymphocytes. This has engendered gene
therapy strategies to augment T-cell-mediated eradication of
tumors. Indeed, while to date cancer has been the primary
target of the bulk of clinical gene therapy protocols, this area
has seen limited success. That is until the results of a land-
mark study were published in 2006 by Morgan and colleagues
demonstrating definitive success in this field. The cDNAs for
the α and β chains of a TCR against a melanoma antigen were
subcloned into an oncoretroviral vector and used to trans-
duce peripheral blood lymphocytes of melanoma patients.
Following infusions of the transduced cells into 17 patients,
two patients demonstrated actual sustained tumor regres-
sions as determined by standard criteria. Furthermore, one
year after infusions both of the responding patients had high
levels of gene-transduced cells in their circulation (20–70%
of peripheral blood mononuclear cells).
A further evolution of this immune augmentation strategy
has led to the creation of chimeric antigen receptors (CARs).
As described earlier, Kymriah, the first gene therapy approved
in the USA, is a CAR-based strategy. CARs use engineered
receptors endowing a patient’s own immune effector cells
with specificity against a particular TAA. A TCR specific to
the TAA is not required; rather, monoclonal antibodies raised
against the TAA are used as the platform for CAR approaches.
The single-chain variable fragment of a monoclonal antibody
Problems and merits
Gene delivery to every cell required
Specific to defective cancer cell
Gene delivery to every cell required
Immunogenicity and bystander effect contribute
Stem cell gene delivery required
Systemic response, autoimmunity a theoretical problem
For TAAs, requires that tumors express foreign antigen
332 Molecular Hematology

Antigen-presenting cell T cell Effect

Adhesion
Peptide
Adhesion receptor T-cell recruitment
and killing of
MHCI T-cell receptor IL–2 target cells
expressing
Costimulatory receptor peptide
Costimulation
γ interferon
IL–12

Tumor cell
Adhesion
Peptide
Signaling
Adhesion receptor
defective

MHCI T-cell receptor Results in anergy

Costimulatory receptor

Gene-engineered tumor cells

Insert Adhesion
gene of Adhesion receptor
interest Restoration
of cell killing
MHCI T-cell receptor IL–2
B7–1
Costimulatory receptor

IL–12

Fig. 23.3 Some facets of drug resistance immunotherapy. Efficient signaling of T cells occurs following adhesion, engagement of the major
histocompatibility complex (MHC) class I/T-cell receptor signal, and after a costimulatory signal is received. Under normal circumstances,
professional antigen-presenting cells (APCs) present processed peptides in the groove of the MHC class I to T cells. Tumor cells are often deficient
in one or all of the components required to function as APCs, because they lack an appropriate tumor antigen, cannot process the antigen, or are
deficient in the adhesion molecules, MHC class I, or costimulatory molecules required to generate a T-cell response. These missing components can
be provided or expression enhanced using gene transfer techniques. By overcoming the deficiencies of the tumor cell, the gene-engineered cells
can serve as autologous cancer vaccines presenting foreign antigen to the host T cells.

against a TAA is fused to the intracellular signal components
of a TCR and other costimulatory signaling domains. These
constructs can be transduced into the patient’s own effector
immune cells and, when reintroduced into the patient, pro-
vide transduced cells with the ability to recognize cancer cells
expressing that TAA. Although still in their infancy, CAR-T
cell clinical trials have shown considerable promise treating
blood cancers. Clinical trials are underway for ALL, CLL, dif-
fuse large B-cell lymphoma (DLBCL), follicular lymphoma,
and multiple myeloma, among other indications. CD-19, a
B-cell TAA, is the most well-studied CAR antigen. By
2015 over 200 patients had undergone CD19-targeted CAR-
modified T cell clinical trials for the treatment of ALL, with
reported positive response rates hovering over 80%.
As more CAR-T cell trials enter the clinic, variations of
the core strategy have emerged (as have playful adapta-
tions of the CAR acronym). Multiplexing TCR complexes to
broaden the immune response and overcoming antigen drift
and subsequent tumor evasion, adaptation of recombinant
LVs to increase gene transfer frequencies, and incorporation
of methods to overcome immune dampening mechanisms
driven by the tumors themselves may increase the potency

of these strategies in future protocols. Examples of the new
CARs undergoing preclinical work will be listed here.
T cells redirected for universal cytokine killings (TRUCKs)
co-express both a CAR and an antitumor cytokine. Cytokine
production can be induced by T-cell activation and thereby
recruit immune cells to tumor sites. Universal CAR-T cells
are also engineered from allogeneic T cells to eliminate
endogenous TCR and/or HLA molecules, and thus prevent
graft-versus-host disease (GVHD) and transplant rejection.
Self-driving CARs co-express both a CAR and a tumor lig-
and chemokine receptor to enhance tumor homing. Armored
CARs secrete active cytokines or express ligands that improve
CAR-T cell efficacy and persistence. The “armor” agent is
chosen based on knowledge of the tumor microenviron-
ment to further enhance CAR-T cell efficacy and persis-
tence. Self-destructing CARs can be engineered by transient
CAR expression obtained by electroporation of RNA into
the cell, or by inducing apoptosis with cell suicide systems.
Conditional CAR-T cells are inactive until binding an exoge-
nous small molecule that activates the CAR. Marked CAR-T
cells express both a CAR and a known tumor epitope that
existing therapies can clear. If adverse events result from
treatment, administration of the anti-epitope therapy will
clear the marked CAR-T cells and hopefully alleviate symp-
toms. Tandem CAR-T cells express a CAR consisting of two
linked recognition domains that both must interact with a
target on the tumor cell. A dual CAR-T cell expresses two
separate CARs with different targets: one CAR is fused to
only the CD3ζ intracellular domain, and the second CAR
is fused to only the costimulatory intracellular domain(s).
Thus, dual CAR-T cell activation also requires co-expression
of two targets on a tumor. A safety CAR contains both a
CAR and an antigen-binding domain linked to an intracel-
lular inhibitory domain. If the tumor expresses antigens for
both the CAR and the inhibitory domain scFv, the CAR
will be inactive. Thus, these CAR-T cells only become acti-
vated when tumors display the CAR target, but lack the
inhibitory receptor target. Bionic CARs combine CAR-T cells
with bispecific antibodies (antibodies featuring the benefits
of two monoclonal antibodies and differing binding specifici-
ties in one construct) to improve killing efficiency. Many of
these novel CAR-based systems may make their way into the
clinic soon.
To date, the completed CAR trials are encouraging, but
as these studies have occurred only recently, a considerably
longer follow-up period will be required to provide data on
long-term patient safety/efficacy responses.
Another facet of immunotherapy for cancer that has
received substantial interest is the application of gene
delivery techniques to induce specific immune responses
in patients by directing efficient antigen-presenting cells
(APCs) to present peptides from such TAAs to T cells. Where
Molecular therapeutics in hematology: gene therapy 333
patients have sufficient immune capacity for this kind of
manipulation, this can cause upregulation of the immune
response against that specific TAA. The most immuno-
logically powerful (so-called professional) APCs are bone
marrow–derived dendritic cells. Dendritic cells express
MHC class I and II, B7-1, B7-2, CD40, ICAM-1, and LFA-3.
They are capable of presenting processed antigen for days,
and are potent stimulators of immunity when administered
as vaccines to animals. Dendritic cells modulate immune
responses in part by secretion of IL-12 (hence the notion
of converting tumor cells into APCs by introduction of
IL-12 and B7-1 sequences). Dendritic cells can be readily
expanded from bone marrow progenitors in vitro using
cytokine-supplemented medium (useful cytokines include
Flt3L, TNF-α, GM-CSF, and IL-4). Such cells may also be
genetically engineered by a variety of methods to express
TAAs, thereby presenting peptides in a proper context; this
holds promise for the immunotherapy of hematological
malignancies and for some solid tumors. Clinical trials
have been performed in this area targeting a number of
TAAs. Results have indicated that immune responses can be
generated in patients. Current work is focused on enhancing
that immune outcome and capitalizing on that facet to
decrease tumor burdens and/or treat metastatic disease.
As alluded to earlier, tumor cells known to express poten-
tially antigenic peptides manage to evade host immuno-
surveillance and proliferate in vivo. Thus, tumor cells may
lack or downmodulate expression of the necessary acces-
sory signals required to induce expression by immune effec-
tor cells of cytokines that are necessary for activation and
directed in vivo expansion of cytotoxic T lymphocytes. The
end result is anergy, a failure of T cells to respond to the tumor
antigen. In addition to optimal presentation of antigen to the
TCR, efficient activation of naı̈ve T cells requires a second
costimulatory signal. It is now appreciated that molecules of
the B7 family (B7-1/CD80, B7-2/CD86) on APCs engaging
CD28/CTLA-4 receptors on T cells play a key role in this pro-
cess, inducing autocrine IL-2 production and T-cell prolifera-
tion. Murine models have demonstrated that T-cell-mediated
rejection of tumors can be induced by transduction of tumor
cells with such costimulatory molecules. In the absence of
costimulatory signals, it is possible to bypass this require-
ment by ectopic expression of cytokines, and thus overcome
or prevent anergy of the immune effector cells. Proof of prin-
ciple that cytokine gene-transduced tumor cells can prevent
tumor engraftment has been obtained. Such models have also
shown that transduction of the genes for various cytokines,
such as IL-2, IL-4, IL-6, IL-7, IL-12, IFN-γ, GM-CSF, and
TNF-α, into murine tumors not only led to primary rejec-
tion of the modified cells, but often elicited protective immu-
nity against subsequent tumor challenge with unmodified
tumor cells. Furthermore, in such models, synergy has been
334 Molecular Hematology
demonstrated between molecules with varying mechanisms
of action, for example IL-2, IL-12, and B7-1.
Methods to improve gene therapy
safety and suicide gene therapy
With the obvious importance of stably integrating vectors
in gene therapy protocols involving the hematopoietic sys-
tem, and given the deleterious outcomes in clinical trials for
one inherited disorder (see earlier), it is appropriate that the
field has directed attention to studying the safety of such gene
delivery agents. As already discussed, extensive sequenc-
ing analyses have revealed that LVs and oncoretroviral vec-
tors have different integration patterns. Yet, in both cases,
such integrations are still fairly random. Efforts are under-
way to tether the viral integrases to specific sequences in the
genome, thereby directing proviral integration into specific
areas of chromatin; however, to date these strategies have
yielded limited success, with the specificity of integration
being only minimally altered.
Another strategy is the incorporation of insulator elements
into the viral 3′ LTR. As the 3′ LTR of retroviral vectors is
copied to the 5′ LTR during the intracellular generation of
provirus, insulator elements provided in the 3′ LTR should
yield inserted retrovirus surrounded by insulator motifs, and
thereby reduce the likelihood of viral activation of host genes
near the site of insertion. A problem here is that such manip-
ulations usually reduce vector titer dramatically. Perhaps the
most studied safety system is the usage of gene transfer
to endow target cells of interest with “suicide” factors that
allow their selective eradication should deleterious outcomes
arise. This approach certainly has applications in hemato-
logical transplantation, and also in numerous other devel-
oping research fields employing different candidate popula-
tions, such as embryonic stem cells and induced pluripotent
stem cells.
At least three general applications of suicide gene therapy
can be envisioned. One is direct tumor therapy, where the
vector is injected into the tumor mass and patients are given
the prodrug, which is activated only in the tumor. The sec-
ond application is in reduction of GVHD after donor lym-
phocyte infusion (DLI). Here, if symptoms of GVHD appear,
productively transduced and transplanted cells can be selec-
tively removed by the addition of prodrug. Lastly, if a truly
portable system exists, such safety elements could conceiv-
ably be incorporated into any cell transplanted out of its nor-
mal context or into any gene therapy vector, in order to pro-
tect transduced cells from genotoxicity and the development
of leukemias should such outcomes occur.
One class of suicide genes makes use of antibody–
immunotoxin conjugates engineered to specifically bind
unique cell surface markers transduced into the target cell
population. The antibody–immunotoxin conjugate is admin-
istered in vivo to specifically eradicate transduced trans-
planted cells. As this strategy relies on the administration of
toxic substances in vivo, a significant understanding of the
physiological response to the toxin is required to minimize
toxicity and maximize bioavailability of exogenous effector
molecules.
Another suicide strategy is the use of molecular switches
that engage apoptotic pathways in transduced cells. One
such strategy is the inducible caspase 9 (iCasp9) system.
The iCasp9 transgene encodes a chimeric receptor engi-
neered to be activated by the addition of an exogenous small
molecule, referred to as a chemical inducer of dimeriza-
tion (CID). Induced dimerization of iCasp9 results in activa-
tion of the mitochondrial apoptosis pathway in transduced
cells. Clinically, the iCasp9 system has been used in five
patients who received stem cell transplantation for relapsed
leukemia and DLI. On addition of the CID, a rapid elim-
ination of transplanted T cells resulted. Thus, this promis-
ing suicide system is currently being evaluated in several
CAR-T-based clinical trials. However, it is important to note
that iCasp9-based cell fate control requires transgene expres-
sion above a certain threshold level to induce apoptosis, and
thus transduced cells remaining post-CID application may
prove difficult to eliminate, rendering this strategy less use-
ful for applications requiring a more complete elimination of
engineered cells.
Perhaps the best-studied suicide genes are enzymes which,
when expressed by transduced target cells, confer suscepti-
bility to drug-induced cell death by specifically converting
normally non-toxic prodrugs into potent cytolytic or cyto-
static molecules. The gene most commonly employed in clin-
ical trials in this context is the human herpes simplex virus
(HSV) type 1 thymidine kinase transgene, which confers sen-
sitivity to the drugs ganciclovir and aciclovir, among others.
Use of thymidine kinase is further enhanced for some direct
tumor applications by diffusion of the converted prodrug into
neighboring cells, and thus a bystander effect occurs (Fig-
ure 23.4). The obvious limitations of this treatment approach
are that not all cells targeted will be successfully gene mod-
ified and thus, even with the bystander effect, only a frac-
tion of malignant cells will be destroyed. Nevertheless, appli-
cations of this type are in clinical trials for treatment of
solid tumors. Another proven clinical application of this sui-
cide approach, which may actually have a higher proba-
bility of success, is in the prevention of GVHD following
transplantation. These trials were first performed by Bonini
and colleagues in patients who received T-cell-depleted allo-
geneic bone marrow transplantation, followed by infusions
of oncoretrovirally transduced lymphocytes. The transduced
cells demonstrated antitumor activity in five patients. Three
patients developed GVHD, which was controlled by the addi-
tion of ganciclovir.
 Molecular therapeutics in hematology: gene therapy 335

Tumor

Gap
junction
between
cells

TK TK
protein protein
TK
protein

Ganciclovir
Vector carrying prodrug
herpes simplex thymidine
kinase (TK) gene

TK
TK

TK

Death of TK
expressing cells
Fig. 23.4 Basis of suicide gene therapy. The metabolic product of the activity of the thymidine kinase suicide gene appears capable of diffusing
into neighboring cells via gap junctions, which join the cells together. This non-specific diffusion allows a greater effective cell killing percentage
than might be predicted using direct gene transfer efficiency alone. Because cells are dying and releasing tumor antigen into the local milieu of the
tumor, it is possible that such suicide gene therapies will synergize with immune-based treatment strategies.

The HSV thymidine kinase system relies on a slow and

fairly inefficient mechanism to convert ganciclovir into its
Conclusions
toxic form, and the enzyme itself is a foreign protein, which
While it is still a relatively young field, some successes have
has resulted in antithymidine kinase immune responses
been observed in clinical gene therapy trials. Some concerns
being reported in patients receiving transduced cells. This
have also been raised. Development and implementation
can lead to premature clearing of the transplanted cell popu-
of gene delivery systems have led to the accumulation
lation. It should also be noted that many transplant patients
of important secondary knowledge. For example, gene
are already on prophylactic ganciclovir to reduce the possi-
delivery methods are now routinely used in most basic
bility of cytomegalovirus infection.
biology laboratories to study many different processes.
For the reasons noted, Sato and colleagues have attempted
Viral entry mechanisms are much better understood and
to improve on the HSV thymidine kinase system. They use a
immune responses initiated following viral infections are
human enzyme (thymidylate kinase) that has been minimally
better characterized. As a result of the use of integrating
modified to efficiently convert azidothymidine monophos-
vectors, it has also been demonstrated that some T cells in
phate to azidothymidine diphosphate. This system allows
humans are actually very long-lived and that multiple HSCs
for killing of non-cycling cells through mitochondrial mem-
contribute to hematopoiesis. Much more has been learned
brane disruption. In vitro and in vivo testing of such a sui-
concerning mechanisms that contribute to the development
cide safety system has been completed, and recombinant LV
of leukemias. Some concepts and approaches to molecular
that engineers expression of this variant enzyme is being pro-
therapy of diseases that interface with the blood system
duced for eventual clinical trials.
336 Molecular Hematology
have been described in this chapter; many other therapeutic
strategies are being developed. Indeed, the advent and
success of human gene therapy trials combined with rapid
advances in gene transfer technology and implementation of
differential delivery systems, along with the full sequencing
of the human genome, are setting the stage for a new era in
human medicine. The field of human molecular medicine
will surely continue to expand and make inroads in a thera-
peutically relevant way. It seems likely that study of the blood
system, both in the laboratory and in patients, will continue
to play a pivotal role in the further development of this area.
Further reading
General
Brenner, M.K. (2001). Gene transfer and the treatment of hematological
malignancy. J. Intern. Med. 249: 345–358.
Jinek, M., Chylinski, K., Fonfara, I. et al. (2012). A programmable dual-
RNA–guided DNA endonuclease in adaptive bacterial immunity. Sci-
ence 337: 816–821.
June, C.H. (2007). Adoptive T cell therapy for cancer in the clinic. J. Clin.
Invest. 117: 1466–1476.
Melchiorri, D., Pani, L., Gasparini, P. et al. (2013). Regulatory evaluation
of Glybera in Europe – two committees, one mission. Nat. Rev. Drug
Discovery 12: 719–723.
Miller, A.D. (1992). Human gene therapy comes of age. Nature 357: 455–
460.
Rosenberg, S.A., Aebersold, P., Cornetta, K. et al. (1990). Gene transfer
into humans: immunotherapy of patients with advanced melanoma,
using tumor-infiltrating lymphocytes modified by retroviral gene
transduction. N. Engl. J. Med. 323: 570–578.
Rosenberg, S.A. and Restifo, N.P. (2015). Adoptive cell transfer as per-
sonalized immunotherapy for human cancer. Science 348: 62–68.
Williams, D.A. and Smith, F.O. (2000). Progress in the use of gene trans-
fer methods to treat genetic blood diseases. Hum. Gene Ther. 11:
2059–2066.
Viral vectors for gene transfer
Bergelson, J.M., Cunningham, J.A., Droguett, G. et al. (1997). Isolation
of a common receptor for Coxsackie B viruses and adenoviruses 2
and 5. Science 275: 1320–1323.
Crystal, R.G. (1995). Transfer of genes to humans: early lessons and
obstacles to success. Science 270: 404–410.
Dilloo, D., Rill, D., Entwistle, C. et al. (1997). A novel herpes vector
for the high-efficiency transduction of normal and malignant human
hematopoietic cells. Blood 89: 119–127.
Horn, P.A., Topp, M.S., Morris, J.C. et al. (2002). Highly efficient
gene transfer into baboon marrow repopulating cells using GALV-
pseudotype oncoretroviral vectors produced by human packaging
cells. Blood 100: 3960–3967.
Miller, A.D. and Buttimore, C. (1986). Redesign of retrovirus packaging
cell lines to avoid recombination leading to helper virus production.
Mol. Cell. Biol. 6: 2895–2902.
Miller, D.G., Adam, M.A., and Miller, A.D. (1990). Gene transfer by
retrovirus vectors occurs only in cells that are actively replicating at
the time of infection. Mol. Cell. Biol. 10: 4239–4242.
Mitchell, R.S., Beitzel, B.F., Schroder, A.R. et al. (2004). Retroviral DNA
integration: ASLV, HIV, and MLV show distinct target site prefer-
ences. PLoS Biol. 2: E234.
Mulligan, R.C. (1993). The basic science of gene therapy. Science 260:
926–932.
Naldini, L. (2015). Gene therapy returns to centre stage. Nature 526:
351–360.
Naldini, L., Blomer, U., Gallay, P. et al. (1996). In vivo gene delivery and
stable transduction of nondividing cells by a lentiviral vector. Science
272: 263–267.
Pawliuk, R., Bachelot, T., Wise, R.J. et al. (1999). Long-term cure of the
photosensitivity of murine erythropoietic protoporphyria by prese-
lective gene therapy. Nat. Med. 5: 768–773.
Qin, G., Takenaka, T., Telsch, K. et al. (2001). Preselective gene therapy
for Fabry disease. Proc. Natl. Acad. Sci. U.S.A. 98: 3428–3433.
Thornhill, S.I., Schambach, A., Howe, S.J. et al. (2008). Self-inactivating
gammaretroviral vectors for gene therapy of X-linked severe com-
bined immunodeficiency. Mol. Ther. 16: 590–598.
Oncoretroviral gene transfer
Abonour, R., Williams, D.A., Einhorn, L. et al. (2000). Efficient
retrovirus-mediated transfer of the multidrug resistance 1 gene into
autologous human long-term repopulating hematopoietic stem cells.
Nat. Med. 6: 652–658.
Aiuti, A., Slavin, S., Aker, M. et al. (2002). Correction of ADA-SCID by
stem cell gene therapy combined with nonmyeloablative condition-
ing. Science 296: 2410–2413.
Blaese, R.M., Culver, K.W., Miller, A.D. et al. (1995). T lymphocyte-
directed gene therapy for ADA-SCID: initial trial results after 4 years.
Science 270: 475–480.
Cavazzana-Calvo, M., Hacein-Bey, S., de Saint Basile, G. et al. (2000).
Gene therapy of human severe combined immunodeficiency (SCID)-
X1 disease. Science 288: 669–672.
Deisseroth, A.B., Zu, Z., Claxton, D. et al. (1994). Genetic marking
shows that Ph+ cells present in autologous transplants of chronic
myelogenous leukemia (CML) contribute to relapse after autologous
bone marrow transplant in CML. Blood 83: 3068–3076.
Dunbar, C.E., Cotter-Fox, M., O’Shaughnessy, J.A. et al. (1995). Retro-
virally marked CD34-enriched peripheral blood and bone marrow
cells contribute to long term engraftment after autologous transplan-
tation. Blood 85: 3048–3057.
Fischer, A., Hacein-Bey-Abina, S., and Cavazzana-Calvo, M. (2010). 20
years of gene therapy for SCID. Nat. Immunol. 11: 457–460.
Hesdorffer, C., Ayello, J., Ward, M. et al. (1998). Phase I trial of
retroviral-mediated transfer of the human MDR1 gene as marrow
chemoprotection in patients undergoing high-dose chemotherapy
and autologous stem-cell transplantation. J. Clin. Oncol. 16: 165–
172.
Hoogerbrugge, P.M., van Beusechem, V.W., Fischer, A. et al. (1996).
Bone marrow gene transfer in three patients with adenosine deam-
inase deficiency. Gene Ther. 3: 179–183.
Howe, S.J., Mansour, M.R., Schwarzwaelder, K. et al. (2008). Insertional
mutagenesis combined with acquired somatic mutations causes

leukemogenesis following gene therapy of SCID-X1 patients. J. Clin.
Invest. 118: 3143–3150.
Kohn, D.B., Weinberg, K.I., Nolta, J.A. et al. (1995). Engraftment of
genemodified umbilical cord blood cells in neonates with adenosine
deaminase deficiency. Nat. Med. 1: 1017–1023.
Mossoba, M.E. and Medin, J.A. (2006). Cancer immunotherapy using
virally transduced dendritic cells: animal studies and human clinical
trials. Expert Rev. Vaccines 5: 717–732.
Onodera, M., Ariga, T., Kawamura, N. et al. (1998). Successful periph-
eral T-lymphocyte-directed gene transfer for a patient with severe
combined immune deficiency caused by adenosine deaminase defi-
ciency. Blood 91: 30–36.
Ott, M.G., Schmidt, M., Schwarzwaelder, K. et al. (2006). Correction of
X-linked chronic granulomatous disease by gene therapy, augmented
by insertional activation of MDS1-EVI1, PRDM16 or SETBP1. Nat.
Med. 12: 401–409.
Rill, D.R., Santana, V.M., Roberts, W.M. et al. (1994). Direct demon-
stration that autologous bone marrow transplantation for solid
tumors can return a multiplicity of tumorigenic cells. Blood 84: 380–
383.
Sorrentino, B.P., Brandt, S.J., Bodine, D. et al. (1992). Selection of drug-
resistant bone marrow cells in vivo after retroviral transfer of human
MDR1. Science 257: 99–103.
Lentiviral gene transfer
Aiuti, A., Biasco, L., Scaramuzza, S. et al. (2013). Lentiviral hematopoi-
etic stem cell gene therapy in patients with Wiskott–Aldrich syn-
drome. Science 341: 1233151.
Biffi, A., Bartolomae, C.C., Cesana, B. et al. (2011). Lentiviral vector
common integration sites in preclinical models and a clinical trial
reflect a benign integration bias and not oncogenic selection. Blood
117: 5332–5339.
Biffi, A., Montini, E., Lorioli, L. et al. (2013). Lentiviral hematopoietic
stem cell gene therapy benefits metachromatic leukodystrophy. Sci-
ence 341: 1233158.
Brenner, M.K., Rill, D.R., Holladay, M.S. et al. (1993). Gene marking to
determine whether autologous marrow infusion restores long term
haemopoiesis in cancer patients. Lancet 342: 1134–1137.
Brenner, M.K., Rill, D.R., Moen, R.C. et al. (1993). Gene-marking to
trace origin of relapse after autologous bone-marrow transplantation.
Lancet 341: 85–86.
Levine, B.L., Humeau, L.M., Boyer, J. et al. (2006). Gene transfer in
humans using a conditionally replicating lentiviral vector. Proc. Natl.
Acad. Sci. U.S.A. 103: 17372–17377.
Maude, S.L., Frey, N., Shaw, P.A. et al. (2014). Chimeric antigen recep-
tor T cells for sustained remissions in leukemia. N. Engl. J. Med. 371:
1507–1517.
Molecular therapeutics in hematology: gene therapy 337
Saar, G. and June, C.H. (2014). Going viral: chimeric antigen receptor T-
cell therapy for hematological malignancies. Immunol. Rev. 263: 68–
89.
Zielske, S.P., Reese, J.S., Lingas, K.T. et al. (2003). In vivo selection of
MGMT (P140K) lentivirus-transduced human NOD/SCID repop-
ulating cells without pretransplant irradiation conditioning. J. Clin.
Invest. 112: 1561–1570.
Adeno-associated virus gene transfer
Kay, M.A., Manno, C.S., Ragni, M.V. et al. (2000). Evidence for gene
transfer and expression of factor IX in haemophilia B patients treated
with an AAV vector. Nat. Genet. 24: 257–261.
Manno, C.S., Pierce, G.F., Arruda, V.R. et al. (2006). Successful transduc-
tion of liver in hemophilia by AAV-factor IX and limitations imposed
by the host immune response. Nat. Med. 12: 342–347.
Nathwani, A.C., Reiss, U.M., Tuddenham, E.G. et al. (2014). Long-term
safety and efficacy of factor IX gene therapy in hemophilia B. N. Engl.
J. Med. 371: 1994–2004.
Russell, S., Bennett, J., Wellman, J.A. et al. (2017). Efficacy and safety of
voretigene neparvovec (AAV2-hRPE65v2) in patients with RPE65-
mediated inherited retinal dystrophy: a randomised, controlled,
open-label, phase 3 trial. Lancet 390: 849–860.
Tebas, P., Stein, D., Tang, W.W. et al. (2014). Gene editing of CCR5 in
autologous CD4 T cells of persons infected with HIV. N. Engl. J. Med.
370: 901–910.
Genetic immunotherapy
Fesnak, A.D., June, C.H., and Levine, B.L. (2016). Engineered T cells: the
promise and challenges of cancer immunotherapy. Nat. Rev. Cancer
16: 566–581.
Morgan, R.A., Dudley, M.E., Wunderlich, J.R. et al. (2006). Cancer
regression in patients after transfer of genetically engineered lympho-
cytes. Science 314: 126–129.
Methods to improve gene therapy safety,
and suicide gene therapy
Bonini, C., Ferrari, G., Verzeletti, S. et al. (1997). HSV-TK gene trans-
fer into donor lymphocytes for control of allogeneic graft-versus-
leukemia. Science 276: 1719–1724.
Montini, E., Cesana, D., Schmidt, M. et al. (2006). Hematopoietic stem
cell gene transfer in a tumor-prone mouse model uncovers low geno-
toxicity of lentiviral integration. Nat. Biotechnol. 24: 687–696.
Sato, T., Neschadim, A., Konrad, M. et al. (2007). Engineered human
tmpk/AZT as a novel enzyme/prodrug axis for suicide gene therapy.
Mol. Ther. 15: 962–970.
Chapter 24 Pharmacogenomics
Leo Kager1 & William E. Evans2
1 Department of Pediatrics, St. Anna Children’s Hospital, Medical University, Vienna, Austria
2
St Jude Children’s Research Hospital, Memphis TN, USA

Introduction, 339 Pharmacogenomics to optimize oral antithrombotic therapy, 348

Principles of pharmacogenomics, 340 Summary and challenges for the future, 350
Pharmacogenomics to improve childhood ALL therapy, 342 Further reading, 351

monogenic loci for drug-metabolizing enzymes that strongly

Introduction influence the effects of medications. Interestingly, two
of the most important clinical examples of Mendelian
It has long been recognized that there is great heterogene-
pharmacogenetics – polymorphisms in the thiopurine
ity in the way people respond to medications. For exam-
S-methyltransferase (TPMT) gene that affect the efficacy
ple, when a standard dose of a certain drug is given to a
and toxicity of thiopurines, and polymorphisms in the
cohort of patients, some will respond and some will not
cytochrome P4502C9 (CYP2C9) and vitamin K epoxide
respond, some will respond only partially, and some will
reductase complex 1 (VKORC1) that affect the efficacy and
experience adverse drug reactions (ADRs) that can be life-
toxicity of warfarin – were again discovered in the field of
threatening. This variation in both host toxicity and treat-
hematology, namely in the treatment of acute lymphoblastic
ment efficacy can have many different causes, including envi-
leukemia (ALL) and oral anticoagulant therapy, respectively.
ronmental (e.g. nutrition, drug interactions), physiological
However, it is well recognized that most pharmacologi-
(e.g. age, gender, nutritional status, organ functions), patho-
cal effects result from the interplay of numerous gene prod-
physiological (e.g. pathogenesis and severity of the disease
ucts, and since the human genome has been sequenced and
being treated), and genetic factors. Overall, genetic factors
the human haplotypes of the most common form of genetic
are estimated to account for 15–30% of interindividual dif-
variation, namely single-nucleotide polymorphisms (SNPs),
ferences in drug metabolism and response. For certain drugs,
have been mapped, genome-wide approaches (i.e. pharma-
however, genetic factors are of the utmost importance and
cogenomics) are often used to elucidate the genomic con-
can account for up to 95% of interindividual variability in
tributors of variability in drug effects. The terms “pharmaco-
drug disposition and effects.
genetics” and “pharmacogenomics” are synonymous for all
The science of pharmacogenomics, which aims to define
practical purposes, and we herein use the term pharmacoge-
the genomic determinants for drug disposition and effects,
nomics. Besides pharmacogenomics, also the field of phar-
has a long tradition in the field of hematology. In the 1950s,
macoepigenomics, which focuses on the identification of
the relationship between hemolysis after antimalarial ther-
pharmacologically relevant epigenetic variants, has evolved.
apy and the inherited glucose-6-phosphate dehydrogenase
Recent advances in genome, transcriptome, and
(G6PD) activity in erythrocytes was identified. This discov-
epigenome interrogation technologies (e.g. arrays for
ery explained why the ADR of hemolysis is observed mainly
genome-wide SNP, mRNA, and DNA-methylation anal-
in Africans, where up to 10% of individuals are deficient in
yses, “next-generation” DNA sequencing technologies
G6PD, but is rarely seen in other ethnic groups like Euro-
like whole-exome sequencing – WES, coding regions and
peans, in whom G6PD deficiency is uncommon. In 1959,
“untranslated regions” only – and whole-genome sequenc-
Friedrich Vogel defined pharmacogenetics as “the study of
ing – WGS, coding and non-coding regions) and in silico
the role of genetics in drug response.” Until 2000, efforts
analytical approaches have become sufficiently robust
were mainly concentrated on mapping highly penetrant
and cost-effective and allow essentially agnostic genome-
wide investigations to identify pharmacologically relevant
Molecular Hematology, 4th Edition. Edited by Drew Provan and John G. Gribben. relationships between genomic variants and well-defined
© 2020 John Wiley & Sons Ltd. Published 2020 by John Wiley & Sons Ltd. pharmacological endpoints, for example in genome-wide

339
340 Molecular Hematology
association studies (GWAS). Recent advances in single
cell RNA and DNA sequencing provide the technology to
interrogate the clonal nature of somatic variants, to identify
sub-clones that may be resistant to cancer chemotherapy.
Once discovered, such pharmacological relationships have
first to be validated and then translated into clinical practice,
the latter of which is the most difficult task.
In this short chapter, validated, clinically relevant exam-
ples are presented to illustrate how pharmacogenomics can
be used to rationally improve current drug therapy in hema-
tological diseases and to identify novel targets for developing
new therapeutic approaches in hematological diseases.
Principles of pharmacogenomics
The effects of drugs are determined by the interplay of
many gene products that influence the pharmacokinetics and
pharmacodynamics of medications. Whereas pharmacoki-
netics describes the absorption, distribution, metabolism,
and excretion of drugs (so-called ADME), pharmacodynam-
ics studies the relationship between the pharmacokinetic
properties of drugs and their pharmacological effects, which
can be desired or adverse. The ultimate goal of pharmacoge-
nomics is to elucidate functionally relevant genomic determi-
nants for drug disposition (germline variants) and response
(variants in drug targets) to select medications and dosage
of medications on the basis of each patient’s inherited ability
to metabolize, eliminate, and respond to specific drugs (Fig-
ure 24.1). In hematological malignancies, for example, the
genomic architecture differs between the normal host cells
(i.e. germline) and the malignant cells (i.e. the therapeutic
(germline)
Host cells
GENOTYPE
PK
Genomic variants
Single-nucleotide polymorphisms (SNPs)
insertions/deletions (Indels <50 bp)
Varying number of tandem repeats (VNTR)
Copy number variants (CNV, <50 bp)
Epigenetic variants
PD
Somatic variants
Target
cells
GENOTYPE Pharmacological effects
EFFICACY
target or targets), which harbor germline and also acquired
somatic variants. Many variants in the human genome and
in cancer cells have been identified that influence the expres-
sion and activity of pharmacologically relevant proteins.
Variation in the human genome
The human genome consists of about 3 × 109 bp (3 Gb).
Although any two humans are thought to be up to 99.9%
identical in their DNA sequence, the remaining small frac-
tion of the genome, which constitutes the genetic diversity
among individuals, contains many forms of variation, rang-
ing from large, microscopically visible chromosome anoma-
lies to single-nucleotide changes. Variants that are about
3 Mb or more in size are referred to as microscopic vari-
ants. The smaller, and much more abundant, variants include
SNPs, small (<50 bp) insertions and deletions (indels) of
nucleotides, variation in the number of repeats of a spe-
cific motif (i.e. variable-number tandem repeats, minisatel-
lites, and microsatellites), duplications, and variants of DNA
segments 50 bp or larger that are collectively defined as
copy number variations (CNVs). Common DNA variants,
often defined as greater than 1% in a given population,
are named polymorphisms, whereas rare DNA variants are
named mutations.
Single-nucleotide polymorphisms
The most common and most extensively studied inherited
genomic variations are SNPs, positions in the genome where
individuals have inherited a different nucleotide. SNPs are
found approximately every 300 bp in both the coding and
DRUGS
Delivery
Absorption
Distribution
Metabolism
Excretion
Target Fig. 24.1 General principles of
Mechanism of action
Drug responses
pharmacogenomics. The goal of
pharmacogenomics is to elucidate genomic
determinants that influence the pharmacokinetics
(PK) and pharmacodynamics (PD) of delivered
drugs, thereby influencing both efficacious and
toxic effects. Investigating these functionally
TOXICITY important variants in the human population can
help to explain some differences in drug effects.

non-coding regions of the human genome. About 660 million
SNPs have been identified (of which about 130 million are
currently validated) and are available in the public domain at
http://www.ncbi.nlm.nih.gov/projects/SNP/snp_summary.
cgi?view+summary=view+summary&build_id=151.
SNPs that cause amino acid changes in the encoded
protein are named non-synonymous coding SNPs, whereas
SNPs that do not change the amino acid composition are
named synonymous (or silent) SNPs. Amino acid substitu-
tions have the potential to change the function of a protein,
and have more relevance in functionally important domains
of the protein (e.g. the catalytic cleft of an enzyme). However,
it has been demonstrated for the adenosine triphosphate
(ATP)-binding cassette transporter ABCB1 that a silent
or synonymous SNP can also affect in vivo protein folding
and function. Moreover, SNPs in genomic regions that
influence the expression of functionally relevant genes
like transcription factors or microRNA binding sites can
also have pharmacological consequences. For example, a
so-called microRSNP (defined as a functional SNP that can
interfere with microRNA function and results in loss of the
microRNA-mediated regulation of a drug target gene) was
found to be associated with antifolate resistance by influ-
encing the expression of the antifolate target dihydrofolate
reductase (DHFR) gene. In addition, a promoter polymor-
phism in the gene encoding centrosomal protein 72kD
(CEP72), which creates a binding site for a transcriptional
repressor, leading to lower expression of CEP72 mRNA, was
found to be significantly associated with vincristine-induced
peripheral neuropathy in children with ALL; and these
results were recently confirmed in adults with ALL.
Haplotypes, linkage disequilibrium, and haplotype map
SNPs and other genomic variants are not inherited inde-
pendently, but rather belong to segments of DNA that are
inherited as units, with each unit referred to as a haplotype.
Genome-wide haplotypes can be constructed by linkage dis-
equilibrium (LD) analysis, a statistical measure of the extent
to which particular alleles or SNPs at two loci are associated
with each other in the population. LD occurs when haplotype
combinations of alleles or SNPs at different loci occur more
frequently than would be expected from random association.
SNPs and alleles of interest are presumably inherited together
if they are physically close to each other (typically 50 kb apart
or closer), producing strong LD.
The international HapMap consortium created a genome-
wide map of haplotypes, which revealed a block-like struc-
ture of LD, as well as the existence of areas of low or high
recombination rate, leading to the identification of so-called
tagging (tag) SNPs. Tag SNPs can be used to predict with high
probability the alleles at other cosegregating “tagged” SNPs.
It was also found that common SNPs are in LD with other
Pharmacogenomics 341
common variants in the human genome (e.g. structural vari-
ants), and the largest public catalogue of human variation and
genotype data is the database of the 1000 Genomes Project
(http://www.1000genomes.org/about).
Copy number variants
DNA segments 50 bp or larger and which are present at vari-
able copy number in comparison with a reference genome are
defined as CNVs. It is estimated that up to 9.5% of the genome
contributes to CNVs (losses > gains); any variation in copy
number will affect a wide spectrum of genomic sequences,
and possibly many pharmacologically relevant genes. CNVs
have been identified to influence the activity of major drug-
metabolizing enzymes like CYP isoenzymes and glutathione
S-transferases. For example, fully functional gene duplication
and multiplication of the cytochrome P450 2D6 (CYP2D6),
which is responsible for the metabolism of more than 30%
of all orally administered drugs (e.g. opioids, antidepressants,
etc.) can result in ultra-rapid drug metabolism and therapeu-
tic failure, or excessive response in patients.
One example, which illustrates the potential importance of
CYP2D6 copy numbers on drug effects, is severe intoxication
with codeine (which is often used to treat children suffering
from pain crisis in sickle cell disease, SCD), an analgesic pro-
drug that is metabolized by CYP2D6 to the active metabo-
lite morphine. High levels of morphine can be caused by
ultra-rapid activation due to three copies of CYP2D6, exem-
plified by a woman who had taken codeine for pain shortly
after she had given birth. Whereas the woman survived the
intoxication, her child died from a lethal dose of morphine
via her breast milk. Though very important, analytical test-
ing for CNVs is non-trivial and more laborious than for
SNPs, and current efforts focus on the development of clini-
cally useful assays for pharmacogenomic CNV analyses. The
major catalogues for CNVs are the Database of Genomic
Variants (http://dgv.tcag.ca/dgv/app/home) and the database
of genomic structural variation (dbVar, http://www.ncbi.nlm.
nih.gov/dbvar).
Somatic variations
Non-random genetic abnormalities, including gains and
losses of chromosomes, can be found in the majority
of hematological malignancies. This can create differences
between normal host cells (i.e. germline genotype) and can-
cer cells (acquired somatic variants), and such differences can
have pharmacologically relevant consequences. For example,
resistance mechanisms against thiopurines, which are key
components in the successful therapy of childhood ALL, have
recently been identified in ALL cells, and include somatic
deletions of genes encoding proteins that regulate the sta-
bility of the DNA mismatch repair enzyme MSH2, acquired
342 Molecular Hematology
activating mutations in the genes encoding the cytosolic
nucleotidase NT5C2, and the purine biosynthesis enzyme
phosphoribosyl pyrophosphate synthetase 1 (PRPS1). Col-
lectively, such findings can help to establish mechanistic
models for drug resistance in cancer cells, can help to pre-
dict relapse (and to refine stratification of patients into dif-
ferent treatment arms), and can be used to develop strategies
to circumvent drug resistance mechanisms in hematological
malignancies.
Epigenetic variations and epidrugs
Besides genome sequence variation, epigenetic regulation of
gene expression is increasingly recognized to contribute to
differences in the pharmacological effects of many medica-
tions, referred to as pharmacoepigenomics. Epigenetic signal-
ing is mediated through DNA methylation, DNA hydrox-
ymethylation, and post-translational histone modifications.
Epigenetic biomarkers for drug responses have been identi-
fied in the therapy of hematological malignancies. One exam-
ple is somatic hypomethylation of the promoters of caspase 1
(CASP1) and its activator NLR family, pyrin domain contain-
ing 3 (NLRP3) in ALL cells, leading to overexpression of both
genes, and higher caspase 1 activity in leukemia cells. This has
been shown to lead to glucocorticoid resistance in ALL cells,
due to caspase 1 cleavage of the glucocorticoid receptor.
Small molecules that interfere with the epigenetic control
of gene expression, for example inhibitors of DNA methyl-
transferase (DNMT) or histone deacetylase (HDAC), are
often referred to as “epidrugs.” Of note is that epidrugs were
initially developed to treat hematological malignancies, and
the first US Food and Drug Administration (FDA) approved
epidrugs were the DNMT1 inhibitor vidaza for the treat-
ment of myelodysplastic syndromes, and the HDAC inhibitor
vorinostat for the treatment of advanced cutaneous T-cell
lymphoma.
Genetic variation among ethnic groups
It is well known that differences in the frequency and nature
of genetic variants among ethnic groups must also be recog-
nized when attempting to extrapolate research from one pop-
ulation to another. For example, allele frequencies and types
of polymorphisms in the TPMT and NUDT5 genes, which
influence the efficacy and toxicity of thiopurine therapy, vary
greatly among different ethnic groups. Whereas pharmaco-
logically relevant variants in TPMT are frequent in Euro-
peans and Africans, these variants are rare in Asians. Con-
versely, actionable NUDT15 variants are frequently found in
East Asians and Hispanics, but they are rare in Europeans and
not observed in Africans. Therefore, pharmacogenomic rela-
tions must be validated for each therapeutic indication within
different ethnic groups.
Approaches for establishing
pharmacogenomic models
Pharmacogenomics is a broad strategy for establishing mod-
els by integrating information from functional genomics,
high-throughput molecular analyses, pharmacokinetics, and
pharmacodynamics. These models can be used to further
optimize existing drug therapy or identify novel therapeu-
tic targets. As already described, there is a bewildering
array of human genomic variation, and a central issue in
pharmacogenomics is to elucidate those genomic determi-
nants that are pharmacologically relevant, namely which
genomic variants are associated with certain drug-related
phenotypes. Pharmacogenomic models can be established
using two broad strategies: candidate gene or agnostic
genome-wide strategies. More details including the advan-
tages and disadvantages of these approaches are provided in
Figure 24.2.
We herein provide selected examples of how candidate
gene and genome-wide approaches have been used to opti-
mize therapies in hematological diseases, and focus on the
improvement of childhood ALL therapy and oral antithrom-
botic therapies.
Pharmacogenomics to improve
childhood ALL therapy
Childhood ALL is the most common cancer in children.
Insights from genome profiling and sequencing have
improved our understanding of ALL pathogenesis and have
further proven that ALL is a heterogeneous and often poly-
clonal disease. There are more and more ALL subtypes iden-
tified, each of which has distinctive somatic genomic alter-
ations, which propel tumorigenesis and influence response to
antileukemic medications. Although about 85% of children
with ALL can be cured with current risk-adapted treatment
protocols in industrialized countries – stratification in these
protocols is based for example on the patient’s age (infants
versus older), leukemia cell immunophenotype (B-lineage
versus T-lineage), and genetics (“low-risk” chromosomal
abnormalities like t(12/21)/ETV6-RUNX1, high hyper-
diploidy, t(1;19)/TCF3-PBX1, versus “high-risk” chromoso-
mal abnormalities like t(9;22)/BCR-ABL1, low hypodiploidy,
t(17;19)/TCF3-HLF, etc.), as well as response to therapy; and
patients are stratified based on their risk of relapse into differ-
ent regimens like “low-risk” or “high-risk” regimens – ALL
remains a leading cause of death from disease in children.
Major causes of death in childhood ALL are treatment fail-
ure (i.e. relapse or less frequent, progressive disease during
therapy) and ADRs, like bone marrow toxicity-associated
lethal infections. The significant reduction in treatment
failures during the last decades, which resulted in significant
 Pharmacogenomics 343

Single-gene approach Pathway-gene approach

Drug-related
• Analyzes the possible phenotype • Analyzes several
candidate gene functionally related
candidate genes
• Requires a large
cohort to test the • Requires a smaller
association cohort to test the
Genome-wide approach association
− Carries the risk of not
finding an association − Carries the risk of
missing important
• Analyzes the whole genome genes
+ Provides proof of
(expression and SNP)
principle if an
association is found + Provides a more
• Requires smaller cohorts to test biologically meaningful
associations association

− Difficult assessment of biological

meaning (i.e. risk of false positives)

+ May identify new associations

Fig. 24.2 Comparison of single-gene, candidate pathway-gene, and genome-wide pharmacogenomic approaches to the analysis of
drug-related phenotypes. The main characteristics (•), disadvantage (−), and advantage (+) of each approach are indicated. The arrows between
the three approaches indicate that any approach can lead to another and that a combination of approaches might be considered. SNP,
single-nucleotide polymorphism.
Source: Cheok M.H., Evans W.E. (2006). Acute lymphoblastic leukaemia: a model for the pharmacogenomics of cancer therapy. Nature Reviews.
Cancer 6: 117–129. Reproduced by permission of Springer Nature.

improvements in outcomes, was mainly achieved by shifting
a higher proportion of patients into more intensive treatment
arms (i.e. “high-risk” ALL therapies) based on novel iden-
tified “high-risk” profiles, such as minimal residual disease
(MRD) load on day 15 of induction therapy in recent Berlin–
Frankfurt–Münster (BFM) protocols. As a higher proportion
of patients had received more intensive therapies in recently
completed treatment trials, for example the Associazione
Italiana di Ematologia e Oncologia Pediatrica (AIOP) and
BFM group’s childhood AIOP-BFM ALL 2009 protocol, the
number of deaths due to ADRs have increased. Therefore,
there is an urgent need to identify those patients who are at
high risk for lethal ADRs or for death due to treatment fail-
ure, in order to tailor their supportive and/or antileukemia
therapy accordingly, or to develop novel therapeutic
strategies and design novel innovative treatment protocols.
We herein provide selected clinically relevant examples
to demonstrate the potential of pharmacogenomics in this
context. We describe how testing for G6PD deficiency can
help to prevent the occurrence of potentially lethal hemoly-
sis after administration of rasburicase (which is the standard
medication to treat hyperuricemia), and how results from
candidate gene (variants in TPMT) and genome-wide inves-
tigations (which identified functionally relevant variants in
NUDT15) have provided important insights that can be used
to develop strategies to avoid the potentially lethal ADR
leukopenia in “at-risk” patients treated with thiopurines. Per-
tinent information on other germline genetic variants, which
have been reported to be associated with ADRs of ALL
medicines, including the non-lethal ADRs glucocorticoid-
induced osteonecrosis and vincristine-induced neuropathy,
are summarized in Table 24.1. We furthermore provide
examples on how genome-wide interrogation investigations
have helped to identify thiopurine resistance mechanisms
(somatic variants in NT5C2 and PRPS1) in ALL cells from
relapse, and how this information can be used to develop
strategies to overcome these drug resistance mechanisms.
In addition, we describe how an integrated drug response
profiling approach has been successfully used to identify
novel therapies for patients with TCF3-HLF-positive ALL, a
rare lethal subtype when treated with current treatment pro-
tocols; and we have summarized in Table 24.2 the results
of pharmacogenomic investigations, which have identified
potential actionable somatic genomic variants in patients
with other “high-risk” ALL subtypes; that is, mixed-lineage
leukemia (MLL)-rearranged infant ALL, t(9;22)/BCR-ABL
ALL, and “BCR-ABL1-like” ALL (which is a subgroup of ALL
that has a similar gene expression signature and outcome
as t(9;22)/BCR-ABL1 ALL, but lacks the t(9;22)/BCR-ABL1
aberration).
344 Molecular Hematology

Table 24.1 Selected examples of germline genetic variants (for TPMT and NUDT15 see text) associated with adverse drug reactions (ADRs) of
acute lymphoblastic leukemia medications, identified via genome-wide association studies

Gene (variant SNP ID) and function(s)

of the encoded protein Drug(s) Possible mechanism(s) of action ADR

SLCO1B1 (rs11045879), solute carrier Methotrexate (MTX) MTX transporter, altered MTX Gastrointestinal
organic anion transporter family clearance toxicity
member 1B1, drug transporter
Near the GRIN3A locus (rs10989692), Drugs that contribute Glutamate is involved in osteoblast Osteonecrosis in
glutamate ionotropic receptor NMDA to osteonecrosis activation and impairs endothelial children and
type subunit 3A, glutamate regulated (dexamethasone, barrier function adults
ion channels asparaginase, MTX)
BMP7 (rs75161997), bone morphogenic Drugs that contribute Variants in BMP7 may alter bone Osteonecrosis in
protein 7, member of the to osteonecrosis metabolism and formation; BMP7 is children
transforming growth factor-beta (dexamethasone, also toxic to smooth muscles; may <10 years
family of proteins, plays a role in bone, asparaginase, MTX) induce vascular injury in bone
kidney, and brown adipose tissue vasculature
development and bone homeostasis
CEP72 (rs924607), centrosomal protein Vincristine (VCR) rs924607 T promotor variant creates VCR neuropathy
72kD, involved in the recruitment of a binding site for a transcription
key centrosomal proteins to the repressor, leading to lower CEP72
centrosome and microtubule mRNA expression
formation
HAS3, (rs2232228), hyaluron synthase 3, Anthracyclines Hyaluran is enriched in tissues Cardiotoxicity
enzyme that produces undergoing remodeling after
low-molecular-weight hyaluran injury – impaired remodeling after
cardiac injury
CPA2 (rs199695765), carboxypeptidase Asparaginase rs199695765, non-sense variant; Pancreatitis
A2, secreted pancreatic enzyme loss-of-function variants in CPA1
were found to be associated with a
risk for chronic pancreatitis

SNP, single-nucleotide polymorphism.

Table 24.2 “High-risk” acute lymphoblastic leukemia (ALL) subtypes and selected examples of druggable targets that have been identified via
pharmacogenomics investigations

ALL subtype Identified drug targets Drugs

MLL-rearranged ALL (affects mainly infants)
t(9;22)/BCR-ABL1 or Ph-ALL (more frequent
in adolescents and adults)
“BCR-ABL1-like” ALL, exhibits a gene
expression profile similar to that of
Ph-ALL but lacks the BCR-ABL1 fusion
protein
FLT3 was found to be overexpressed in MLL-rearranged ALL FLT3 inhibitor: lestaurtinib
Chimeric fusion protein BCR-ABL1 has tyrosine kinase TKIs (ABL1 inhibitors):
activity, and can be selectively blocked via TKIs imatinib, dasatinib
Genetic key subgroups with therapeutic implications have
been identified:
a. TKIs: imatinib, dasatinib
a. ABL-class rearrangements targeting ABL1, ABL2, CSF1R,
b. JAK inhibitor: ruxolitinib
and PDGFRB, which are sensitive to imatinib and
dasatinib
b. JAK2 fusions, EPOR rearrangements, and IL7R/SH2B3
alterations are sensitive to JAK inhibitors

ABL, ABL proto-oncogene 1; CSF1R, colony stimulating factor 1 receptor; EPOR, erythropoietin receptor; FLT3, FLT3 Fms-Related Tyrosine Kinase
3; IL7R, interleukin 7 receptor; JAK2, Janus kinase 2; MLL, mixed-lineage leukemia; PDGFRB, platelet-derived growth factor receptor beta; Ph,
Philadelphia; SH2B3, SH2B adaptor protein 3; TKI, tyrosine kinase inhibitor; VEGFR2, vascular endothelial growth factor receptor 2.
 Pharmacogenomics 345

Rasburicase and G6PD deficiency competition with direct drug inactivation (S-methylation)
via TPMT and (hydrolysis of the triphosphates TGTP
Tumor lysis syndrome (TLS) is the most common disease-
and TdGTP to monophosphates TGMP and TdGMP) via
related emergency in hematological cancers. TLS occurs
NUDT15. TPMT and NUDT15 activities determine how
when tumor cells release their contents into the blood-
much thiopurine is inactivated and how much remains for
stream (either spontaneously or due to therapy), and this
incorporation into DNA.
causes electrolyte and metabolic disturbances (i.e. hyper-
uricemia, hyperkalemia, etc.), which, untreated, will result
in renal insufficiency, cardiac arrhythmias, and ultimately Variants in TPMT
death due to multiorgan failure. Hyperuricemia, which con-
Variations in TPMT activity are regulated primarily by
tributes to TLS, can be prevented/treated via rasburicase, an
variants in the TPMT gene. Three non-synonymous SNPs
enzyme which catalyzes the cleavage of uric acid to hydro-
account for more than 95% of the relevant TPMT vari-
gen peroxide, and which is then further metabolized via glu-
ants, namely TPMT*2 (dbSNP identification number rs
tathione peroxidase (GSHPX). In patients with G6PD defi-
1800462, nucleotide change c.238G > C, hereafter referred to
ciency, however, the activity of GSHPX is markedly reduced
as amino acid change p.Ala80Pro), TPMT*3C (p.Tyr240Cys),
in red blood cells (RBCs), and treatment with rasburicase
and TPMT*3A (p.Ala154Thr + p.Tyr240Cys). One in 300
results in oxidative damage to RBCs (i.e. acute hemolytic ane-
persons carries two variant TPMT alleles and does not
mia, AHA). Rasburicase administration is an essential ele-
express functional TPMT activity (due to lesser stabil-
ment of supportive therapy in children with ALL at diagnosis
ity of the variant protein); about 5–10% are heterozygous
and during initial induction therapy.
and have intermediate levels of enzyme activity, whereas
However, fatalities have been reported after rasburicase
95% of individuals are homozygous for the wild-type allele
administration in ALL patients with G6PD deficiency. G6PD
(TPMT*1/TPMT*1) and have normal TPMT activity (Fig-
deficiency is a very common disorder, and about 5% of the
ure 24.3). Population studies have shown significant differ-
world population (highest incidence in the subtropical girdle
ences in TPMT pharmacogenetics among ethnic groups. For
with high malaria prevalence) is affected. As of August 2016,
example, TPMT*3A is the most common variant allele in
about 180 functionally relevant variants had been identified
Europeans, whereas TMPT*3C accounts for more than 50%
in the G6PD gene, most of which are missense mutations
of variants in Africans. TMPT*3C is also the major variant
that alter G6PD stability. Complete loss of G6PD is lethal;
allele in East Asian populations, which generally lack the
and only a few rare, complex variants that cause significant
TPMT*3A allele.
reduction in enzyme activity result in severe transfusion-
Patients who carry two non-functional TPMT alleles expe-
dependent chronic non-spherocytic hemolytic anemia. As
rience severe hematotoxicity if treated with conventional
rasburicase administration can cause lethal ADRs, the use
doses of thiopurines. Depending on the dose (e.g. 60 mg/m2
of this drug has been contraindicated by the FDA and the
daily) and the comedications, many patients with one non-
European Medicines Agency (EMA) in patients with G6PD
functional TPMT allele can tolerate MP therapy at full doses.
deficiency. More details, including information on G6PD
However, these patients might be at higher risk of dose-
genetic and activity tests, as well as genetic variant classifica-
limiting hematotoxicity with slightly higher mercaptopurine
tion (classes I–IV), are available at the Clinical Pharmacoge-
doses (e.g. 75 mg/m2 daily), but may experience better
netics Implementation Consortium (CPIC) guideline website
leukemia control than do those who have two wild-type
(https://www.pharmgkb.org/molecule/PA10176).
TPMT alleles. Moreover, patients with one non-functional
TPMT allele may be at increased risk of epipodophyllotoxin-
related acute myeloid leukemia and irradiation-induced
Optimization of thiopurine therapy
brain tumors as a result of thiopurine therapy, and of veno-
The thiopurine antimetabolites mercaptopurine (MP) and occlusive disease of the liver following thioguanine therapy.
thioguanine (TG) are essential components in childhood Most importantly, results from the St Jude Children’s
ALL treatment protocols. Thiopurines have narrow thera- Research Hospital (SJCRH) Total XIIIB childhood ALL
peutic indices, which explain the frequently observed toxici- treatment trial provide proof of principle that prospective
ties, including severe myelosuppression (mainly leukopenia) MP dose adjustment based on TPMT genotypes can decrease
with the risk of life-threatening infections. To exert cytotox- toxicity without a compromise in treatment efficacy. Based
icity, the prodrugs MP and TG have to undergo anabolism on an FDA advisory committee recommendation, a change
to form active cytotoxic thioguanine nucleotides (TGNs). in labeling for MP, with TPMT testing and dosage rec-
These multistep anabolic reactions, which include conver- ommendations provided for TPMT-deficient patients, was
sion into thioguanosine triphosphate (TGTP) and reduc- implemented in 2004, and more details on TPMT-guided
tion to deoxythioguanosine triphosphate (TdGTP), are in preemptive thiopurine dose adjustment, which is considered
346 Molecular Hematology

100 TGN
(HPRT) wt/wt
(active)
80 MP Inherited
Percent of population

MeMP differences in
(TPMT) (inactive) metabolism
60

40

20
wt/var
var/var
0
5 10 15 20 25 30
(a) Fig. 24.3 Thiopurine methyltransferase
TPMT activity
(TPMT). (a) Mercaptopurine (MP) can be activated
Protein to thioguanine nucleotides (TGN) by hypoxanthine
(b) level
phosphoribosyltransferase (HPRT) or inactivated to
methylmercaptopurine (MeMP) via TPMT. The
genetic polymorphism in TPMT results in a trimodal
population frequency distribution in TPMT activity,
TGN concentration

Inherited differences with deficient activity caused by inheritance of two

2000 in drug levels variant (var) alleles, intermediate activity caused by
heterozygosity, and high activity associated with
homozygote wild-type (wt) genotypes. (b) TPMT
activity is directly proportional to the amount of
1000
TPMT protein and (c) is inversely related to
intracellular concentrations of active TGN
metabolites in hematopoietic cells following MP
(c) var/var wt/var wt/wt
therapy. (d) The biochemical basis for low protein
Toxicity Risk of relapse conferred by the most common var. polymorphism
100 (719A → G) is illustrated by the longer half-life for
Percent remaining

t1/2 = 18 h in vitro expressed wt TPMT when compared with

719 wt the rapidly degraded var protein.
10 Source: Jones T.S., Yang W., Evans W.E., Relling
M.V. (2007). Using HapMap tools in
t1/2 = 0.25 h
719 var pharmacogenomic discovery: the thiopurine
1 methyltransferase polymorphism. Clinical
0 5 10 15 20 25 Pharmacology and Therapeutics 81: 729–734.
(d) Time (h) Reproduced by permission of John Wiley & Sons.

a prototype of pharmacogenomic precision medicine thiopurine-induced myelotoxicity, however, is considerably

approaches, is available at the CPIC guideline website
(https://www.pharmgkb.org/guideline/PA166104945).
Patients who share the same TPMT genotypes still exhibit
considerable variations in their response to thiopurines, and
other genetic variations that contribute to the toxicity and
response to thiopurines have recently been identified via
GWAS studies. The major finding was the identification
of four loss-of-function germline NUDT15 coding variants,
which play an important role in thiopurine intolerance, espe-
cially in East Asian populations.
Variants in NUDT15
It is well known that the frequency of TPMT deficiency
causing variants is lower in Asians compared to individuals
of European descent (~3% versus ~10%). The frequency of
higher in Asians. In order to identify other genetic variants
that contribute to thiopurine-induced myelotoxicity in
Asians, a case–control GWAS study was performed in
Korean patients who were treated with thiopurines, used as
an immunosuppressant, for chronic inflammatory bowel dis-
ease. This research identified a non-synonymous SNP in the
NUDT15 gene (rs116855232, c.415C>T, hereafter referred to
as p.Arg139Cys) that encodes nudix hydrolase 15 (or nucleo-
side diphosphate-linked moiety X-type motif 15), which was
strongly associated with thiopurine-induced leukopenia.
Researchers from the SJCRH and from the Children’s
Oncology Group (COG) confirmed the importance of this
variant to causing MP intolerance in a GWAS study on
more than 1000 children treated for ALL. In addition,
it was confirmed that NUDT15 deficiency (low- or
intermediate-activity diplotypes) was most common in

East Asians (22.6%) and Hispanics (for example, 21.2% in
Peruvians), rare in Europeans, and not observed in Africans.
Subsequent targeted sequencing of NUDT15 identified a
total of four functionally relevant loss-of-function cod-
ing variants (p.Arg139Cys, p.Arg139His, pVal18Ile, and
p.Val18_Val19insGlyVal) that were associated with thiop-
urine intolerance. Five NUDT15 haplotypes (*1–*5) were
inferred from distinct combinations of these four variants,
and functional analyses showed that NUDT15 deficiency
results in excessive levels of active thiopurine metabo-
lites and host toxicity. Of note is that ALL cells carrying
intermediate-activity NUDT15 diplotypes were shown to
be significantly more sensitive to thiopurines compared to
NUDT15 wild-type diplotypes. Therefore, thiopurine dose
reduction in NUDT15-deficient patients, in order to avoid
the ADR leukopenia, will most likely not result in a compro-
mise of therapeutic effect, because NUDT15-deficient ALL
cells are more sensitive to thiopurines. Prospective clinical
trials, however, are necessary to prove this hypothesis. To
further personalize thiopurine therapy in different ethnic
populations, a polygenic dosing algorithm that incorporates
both TPMT and NUDT15 variants should be implemented.
Relapsed ALL, thiopurine resistance, and
variants in NT5C2 and PRPS1
Up to 15% of children with ALL experience disease recur-
rence, and unfortunately most of them will die. Identification
of the mechanisms of drug failure in resistant ALL clones
is urgently needed, in order to develop strategies to over-
come these resistance mechanisms. Genomic profiling and
sequencing investigations of ALL samples obtained from
diagnosis, remission, and relapse have already helped to
identify different pathways (e.g. transcription regulation,
lymphoid development, Janus kinase and Ras signaling,
cell cycle regulation, tumor suppression, mismatch repair,
epigenetic regulation, and nucleotide metabolism), in which
somatic mutations are enriched at relapse. We herein focus
on mutations in the nucleotide metabolism pathway and
provide two examples of functionally relevant variants in
key genes of nucleotide metabolism (i.e. NT5C2 and PSPR1)
that have recently been identified to play a role in resistance
to thiopurines and development of relapse in a subset of
children with ALL.
Variants in NT5C2
Two groups of researchers found that up to 10% of B-cell
precursor (BCP)-ALL and up to 20% of T-ALL relapse
samples harbor somatic gain-of-function variants in the
gene NT5C2. NT5C2 encodes a 5′ nucleotidase, which can
selectively inactivate thiopurines, thereby protecting ALL
cells against thiopurine-induced apoptosis. NT5C2 enzyme
activity was measured to be increased up to 48-fold in the
Pharmacogenomics 347
variant proteins. In vitro experiments confirmed that variants
with high NT5C2 activity conferred resistance against thiop-
urines, but not against other antileukemic agents. Ultra-deep
sequencing revealed that in 2 out of 7 patients, the minor
mutant clone already existed at ALL diagnosis. It is assumed
that NT5C2 gain-of-function harboring clones exist at
diagnosis or evolve early during therapy, and can undergo
selective outgrowth in treatment phases, when thiopurines
play a major role, like in oral maintenance therapy, which is
given for >1 year as a final element of ALL therapy. Indeed,
there was a significant association between activating
NT5C2 variants and early ALL relapse (within 36 months
from diagnosis); and this early outgrowth of resistant clones
during or shortly after completion of maintenance therapy
supports this assumption. Strategies to overcome the drug
resistance somatic phenotype gain of function in NT5C2
can focus either on the inhibition of NT5C2 via small
molecules, or on the use of drugs that are not inactivated
via NT5C2.
Variants in PRPS1
Functionally relevant gain-of-function mutations in another
mediator of purine metabolism, PRPS1, were recently iden-
tified in almost 7% of relapsed BCP-ALL samples. PRPS1
encodes phosphoribosyl pyrophosphate synthetase 1, which
is a key enzyme in the de novo purine synthesis (DNPS) path-
way. In contrast to NT5C2 mutations, ultra-deep sequencing
of serial bone marrow samples revealed that PRPS1 muta-
tions were not detectable at ALL diagnosis, but increased
exponentially before clinical relapse, which occurred early in
treatment. Expression of gain-of-function variants in PRPS1
resulted in resistance to thiopurine-induced apoptosis. Func-
tional investigations revealed that the drug resistance PRPS1
variants showed defective feedback inhibition to adenosine-
and guanosine-diphosphates, which allowed for continuous
activation of DNPS, which in turn results in an increase
of intracellular hypoxanthine levels, which inhibit the acti-
vation of the thiopurine prodrugs (MP and TG) into their
active metabolites. These findings demonstrate that the rare
acquired somatic PRPS1 variants in ALL relapse clones can
drive thiopurine resistance via altering the feedback inhi-
bition of nucleotide synthesis and competitive inhibition
of bioactivation of the prodrugs MP and TG. One logi-
cal strategy to overcome this drug resistance mechanism
would be the inhibition of the continuously activated DNPS
pathway. Indeed, inhibition of GART (phosphoribosyl-
glycinamide formyltransferase), another important enzyme
of DNPS, via the small molecule inhibitor lometrexol,
reversed thiopurine drug resistance in PRPS1-mutant cells
in vitro.
Clearly, much work is needed to further elucidate the
genomic drivers for treatment failure in ALL. The examples
of NT5C2 and PRPS1, however, provide evidence that in
348 Molecular Hematology
the near future the routine use of genomic interrogation
techniques could help to identify patients with emergent
relapse at an early stage, and this may allow the tailoring
of therapy accordingly, e.g. by drugging DNPS in clones
that develop resistance to thiopurines via PRPS1-mediated
DNPS activation.
Optimization of therapy in children with
TCF3-HLF ALL
One typical feature of childhood ALL is the subtype-defining
presence of chromosomal aberrations like translocations
and gross aneuploidies. Translocations, such as t(12;21),
t(9;22), t(1;19), or t(17;19) and the corresponding gene
fusions ETV6-RUNX1, BCR-ABL1, TCF3-PBX1, or TCF3-
HLF, which mainly involve genes that play an important role
in hematopoietic development, are considered to be primary
ALL-initiating abnormalities, which comprise the leukemic
clone and define subtype biology. These initiating lesions
typically cooperate with somatic secondary abnormalities,
which often involve B-cell development genes like IKAROS
zinc finger 1 (IKZF1) or paired box 5 (PAX5), which are
present only in ALL subclones.
Two ALL subtypes have chimeric fusion genes that
involve transcription factor 3 (TCF3); namely, ALL with
t(1;19)/TCF3-PBX1 (~5% of ALL patients) and ALL with
t(17;19)/TCF3-HLF (~1% of ALL patients). Whereas chil-
dren with TCF3-PBX1-positive ALL have excellent outcomes
with contemporary BFM and SJCRH Total therapies, all so far
reported children with TCF3-HLF-positive ALL have experi-
enced early disease relapse and died. In an attempt to iden-
tify curative treatment options for children with TCF3-HLF-
positive ALL, an international collaborative study group
set up a series of genomic and functional studies, includ-
ing an integrated drug–response profiling approach. They
first identified that the chimeric fusion protein TCF3-HLF
promotes cellular transcriptional reprogramming toward a
drug-resistant immature state. Subsequently, they found that
TCF3-HLF xenografts are highly sensitive to the BCL2
(B-cell CLL/lymphoma 2)-targeting drug venetoclax. Com-
bination of venetoclax with conventional chemotherapy
induced durable remissions in patient-derived xenografts;
and these results from preclinical investigations hold great
promise to offer a chance for a cure for patients with TCF3-
HLF-positive ALL in future international clinical trials that
combine venetoclax and conventional chemotherapy.
Pharmacogenomics to optimize oral
antithrombotic therapy
Cardiovascular diseases (CVDs) are the most common cause
of death, accounting for ~17.5 million deaths in 2012 (stroke:
6.7 million deaths, coronary heart disease: 7.4 million deaths;
http://www.who.int/mediacentre/factsheets/fs317/en). Arte-
rial or venous thrombotic events are major fatal complica-
tions of CVDs, and can be prevented via antithrombotic med-
ications, like oral anticoagulants and antiplatelet therapies.
Coumarins and variants in CYP2C9
and VKORC1
Although treatment trials demonstrated a favorable effi-
cacy and safety profile of novel (non-vitamin K antagonist)
oral anticoagulants, which either directly inhibit thrombin
(e.g. dabigatran) or factor Xa (e.g. rivaroxaban, apixaban,
and edoxaban), millions of patients still take coumarins to
prevent thromboembolic events in chronic conditions such
as atrial fibrillation, deep venous thrombosis, pulmonary
emboli, acute myocardial infarction, stroke, and disease
and/or replacement of heart valves. The oral anticoagulants
of the coumarin type, warfarin (used in the UK and the USA),
acenocoumarol, and phenprocoumon (preferentially used in
continental Europe), have similar pharmacodynamic prop-
erties but differences in half-life, and act by inhibiting the
activation of vitamin K–associated clotting factors. A very
narrow therapeutic index, with risk of serious hemorrhage
if overcoagulated and thrombosis if undercoagulated and
interindividual variability in response to coumarins, neces-
sitates individualization of treatment, which is based primar-
ily on monitoring prothrombin time and calculation of the
International Normalized Ratio (INR). Whereas INR is help-
ful in tailoring coumarin maintenance therapy, prospective
studies have identified coumarin induction therapy as the
period when the INR is most likely to be out of range and
when the rate of iatrogenic ADRs is greatest.
Many factors have been identified as affecting the degree
of anticoagulation achieved by coumarins, including patient
age (lower dose requirement in the elderly), gender, body
size, ethnicity, diet (particularly vitamin K intake), cigarette
smoking, disease (e.g. liver diseases), and coadministration
of other drugs (particularly those which inhibit the activity
of CYP2C9). Polymorphisms in genes that affect the phar-
macokinetics (CYP2C9, and to a lesser extent other CYP
enzymes) and pharmacodynamics (VKORC1) of coumarins,
however, have been shown to act as major determinants of
coumarin dosage requirements.
Coumarins are a racemic mixture of r and s enantiomers
that differ in their patterns of metabolism and in their
potency of pharmacodynamic effect. For example, it has been
suggested that s-warfarin accounts for up to 70% of the over-
all anticoagulation response of warfarin. After oral adminis-
tration, warfarin is completely absorbed and bound to albu-
min (99%) in plasma. Free warfarin is taken up into liver cells,
where it is biologically active and either inhibits VKORC1 or
is catabolized by cytochrome P450 isoenzymes (Figure 24.4).
 Pharmacogenomics 349

Target Metabolism

Epoxide
Warfarin
reductase
R-warfarin
VKORC1
S-warfarin
CYP1A2
Fig. 24.4 Mechanism of action of warfarin. The Cytochrome
Reduced vitamin K Vitamin K epoxide CYP2C9
racemic mixture of r- and the more potent P450 CYP3A4
s-warfarin inhibits the reductase in the vitamin K
cycle, impairing the synthesis of active vitamin
K–dependent clotting factors in liver cells and S-OH-warfarin R-OH-warfarin
Carboxylase
causing bleeding. The cytochrome P450 isoenzyme
CYP2C9 (and to a lesser extent CYP3A4 and
CYP1A2) and vitamin K epoxide reductase complex
1 (VKORC1) genotypes influence warfarin dose Glutamic acid Gamma-Carboxyglutamic acid
requirement. Active blood clotting proteins

Variants in CYP2C9 VKORC1*1, the low-dose coumarin haplotype VKORC1*2,

and the high-dose coumarin haplotypes VKORC1*3 and
A number of CYP isoforms contribute to warfarin
VKORC1*4. The significantly higher average warfarin
metabolism; however, hydroxylation by CYP2C9 is the
requirement in Africans is in line with the significantly lower
most important inactivation pathway of the pharmacolog-
occurrence of the low-dose coumarin VKORC1*2 haplotype
ically more relevant s-warfarin. CYP2C9 is the principal
in Africans. Overall, the hereditary pharmacodynamic
CYP2C isoenzyme in the human liver, and it is involved
factor VKORC1 may explain about 25% of the variance in
in the oxidative metabolism of several clinically important
coumarin dose requirement, compared with 5–10% for the
medications, including oral anticoagulants, phenytoin, and
hereditary pharmacokinetic factor CYP2C9 alone.
various non-steroidal anti-inflammatory drugs. Numerous
CYP2C9 and VKORC1 genotypes have been incorporated
polymorphic alleles (CYP2C9*1 to CYP2C9*60) have been
into dosing algorithms in order to estimate the appropriate
identified for the known CYP2C9 gene, according to The
coumarin starting dose, and in 2010 the FDA updated the
Human Cytochrome P450 (CYP) Allele Nomenclature
warfarin drug label and suggested that VKORC1 and CYP2C9
Database (https://www.pharmvar.org/htdocs/archive/index
genotypes should be taken into consideration when the drug
original.htm), at least half of which are associated with
is prescribed. Dosing algorithms and more details are avail-
diminished enzyme activity. The two most common
able at https://www.pharmgkb.org/guideline/PA166104949
CYP2C9 variants are CYP2C9*2 and CYP2C9*3. As with
and www.warfarindosing.org. Randomized controlled
most polymorphisms, there are differences in the frequency
clinical trials, such as the European Pharmacogenetics
of polymorphic CYP2C9 alleles among different ethnic
of Anticoagulant Therapy (EU-PACT) and the US Clar-
groups. In Europeans, the overall allelic frequency of
ification of Optimal Anticoagulation through Genetics
CYP2C9*2 is about 10–20%, and that of CYP2C9*3 is about
(COAG) trials, have investigated the potential benefit of
5–10%. The *2 and *3 variants are very rare in African Amer-
genotype-based strategies for initiating coumarin therapy.
icans and Asians; 95% of these persons express the wild-type
Although the use of genotype-based algorithms has resulted
genotype *1/*1 (i.e. extensive metabolizers). Compared with
in a greater percentage of time in therapeutic range than
the wild-type enzyme activity of CYP2C9*1, the enzyme
standard-fixed dosing in the EU-PACT trial, no reductions
activity of the CYP2C9*2 variant is reduced by about 30–50%,
of severe ADRs like stroke or bleeding were reported with the
and the CYP2C9*3 variant activity is reduced by 90% in vitro.
pharmacogenomic dosing. These trials, however, were not
powered for the pharmacodynamics endpoints bleeding or
Variants in VKORC1
thromboembolic events, and the Genetics Informatics Trial
VKORC1 regenerates reduced vitamin K for another cycle (GIFT) of Warfarin to Prevent Deep Venous Thrombosis
of catalysis, essential for the γ-carboxylation of the vitamin is investigating this topic. Of note is that recently published
K–dependent clotting factors II, VII, IX, and X (Figure 24.4). results from the Effective Anticoagulation with Factor Xa
The identification of common variants in VKORC1 has Next Generation in Atrial Fibrillation–Thrombolysis in
emerged as one of the most important genetic factors Myocardial Infarction 48 (ENGAGE AF-TIMI 48) trial
determining coumarin dose requirements. Main VKORC1 (which enrolled more than 14 000 patients with atrial
haplotypes include the reference haplotype (wild-type) fibrillation) provided evidence that testing for VKORC1 and
350 Molecular Hematology
CYP2C9 was able to identify patients who are more likely
to bleed with warfarin therapy. This information could be
used to optimize choice of oral anticoagulant therapy; that
is, to use edoxaban in patients who are sensitive (typically
1–2 variant alleles) or highly sensitive (typically 3–4 variant
alleles) to warfarin.
Clopidogrel and variants in CYP2C19
Platelets play a crucial role in thrombosis and the devel-
opment of acute coronary syndromes (ACS) because a
platelet-rich thrombus forms at the site of the ruptured
atherosclerotic plaque. Thus, inhibition of platelet function
is an effective strategy in the treatment and prevention of
thrombosis, especially after percutaneous coronary inter-
ventions (PCIs). The main classes of antiplatelet agents
include aspirin, the thienopyridines (clopidogrel and pra-
sugrel), the nonthienopyridine P2Y purinergic receptor 12
(P2Y12) antagonists (ticagrelor), and intravenous GPIIb/IIIa
antagonists. Dual platelet inhibition via aspirin and P2Y12
receptor antagonists is the guideline-approved standard of
care in patients with ACS and PCI with stenting. The newer
P2Y12 receptor antagonists prasugrel and ticagrelor have
superior efficacy compared with clopidogrel, but have a
higher risk for bleeding and are more expensive.
Clopidogrel is an orally administered prodrug and the
response to it is heterogeneous (up to 20% of treated patients
do not respond to clopidogrel and are at risk for stent
thrombosis, which usually results in sudden death or heart
attack, and a few patients have strong responses with bleed-
ing). Once absorbed, 85% of clopidogrel is inactivated via
esterases, and only about 15% of the prodrug remains avail-
able for a multistep activation via hepatic CYP enzymes. The
active drug selectively and irreversibly binds to the ADP-
dependent P2Y12 receptor on thrombocytes and thereby
inhibits platelet activation and aggregation for the platelets’
lifespan, which is about 10 days. Candidate gene investiga-
tions identified loss-of-function variants in the activating
enzyme CYP2C19 to significantly affect drug response. As of
May 2019, a total of 35 variant CYP2C19 alleles (https://www.
pharmvar.org/htdocs/archive/cyp2c19.htm) have been iden-
tified; the most important poor metabolizer (PM) alleles are
*2 (~15% in Europeans and Africans, ~30% in Asians) and
the less frequent *3 (2–9% in Asians, less than 1% in Euro-
peans and Africans). The *17 gain-of-function allele results
in enhanced CYP2C19 enzyme activity, and can place these
ultra-metabolizing individuals at a higher risk for bleeding
because of increased drug activation.
Based on the results of clinical investigations, the FDA
has issued a “black box” warning for clopidogrel in regard
to reduced effectiveness in PM individuals (i.e. two loss-of-
function CYP2C19 alleles), and genotyping for the important
variants is widely available. The CPIC Dosing Guideline for
clopidogrel recommends an alternative antiplatelet therapy
(e.g. prasugrel, ticagrelor) for CYP2C19 poor (two loss-of-
function alleles) or intermediate (one loss-of-function allele)
metabolizers if there is no contraindication.
Summary and challenges for
the future
As there are numerous non-genetic factors that influence
drug effects (e.g. compliance, nutritional factors, concurrent
medications), it is clear that pharmacogenomics will never
explain all interindividual variability in drug effects. How-
ever, there is clear evidence that pharmacogenomic mod-
els can help to improve drug treatment in hematology by
facilitating appropriate dose individualization and optimal
treatment selection. For example, clinically actionable vari-
ants have been identified in the TPMT, NUDT15, CYP2C9,
CYP2C19, and VCORC1 genes, and the TPMT/NUDT15 and
CYP2C9/VCORC1 as well as the CYP2C19 models can be
used to individualize thiopurine and antithrombotic therapy
a priori, thereby reducing the risk of severe ADRs.
As of 2016, clinically actionable genomic variants had
been identified in approximately 20 genes that affect about 80
drugs. However, some barriers (e.g. lack of studies that prove
the cost/benefit, lack of financial reimbursement, etc.) still
prevent the widespread use of pre-emptive genetic testing
(ideally by the use of multigene panels) to guide drug therapy.
Initiatives like the European Pharmacogenetics Implemen-
tation Consortium (EU-PIC; http://www.eu-pic.net) and the
Implementing Genomics in Practice (IGNITE, https://www.
genome.gov/Funded-Programs-Projects/Implementing-
Genomics-in-Practice-IGNITE) Network are developing
and testing pharmacogenomics implementation strategies.
Moreover, the CPIC (https://cpicpgx.org) provides evidence-
based, peer-reviewed, and publicly available guidelines that
aim to support physicians to bring personalized medicine
into clinical practice, by using an individual’s genetic infor-
mation to prescribe medications. Electronic clinical decision
support systems (CDSSs) can make it feasible to utilize
genetic information to prescribe drugs, and CDSSs are con-
tinuously improving and becoming more widely available.
Collectively, it can be expected that pharmacogenomics will
be an important component of the announced Precision
Medicine Initiative (https://www.whitehouse.gov/precision-
medicine).
One example of the successful implementation of a CDSS
evidence-based pharmacogenomics precision medicine
approach in routine clinical hematology was recently
reported from the SJCRH. Before the administration of
codeine, which is given to treat pain crisis in SCD, more
than 600 patients with SCD were genotyped for CYP2D6
variants (CYP2D6 activates the prodrug codeine to the

active morphine). Interruptive alerts recommend against
codeine for patients with high-risk CYP2D6 status (i.e. 7.1%
ultra-rapid metabolizers who are at risk for life-threatening
ADRs; 1.4% poor metabolizers who are at risk for insufficient
pain control), and this example shows how safety concerns
can be reduced via preemptive genotyping. It should also
be noted that the frequency of CYP2D6 poor metabolizers
would be higher (10%) and the frequency of ultra-rapid
metabolizers lower (1–2%) when patients of European
ancestry are prescribed codeine, illustrating ancestry-related
differences in inherited genome variation.
Further reading
Principles of pharmacogenomics
Evans, W.E. and Relling, M.V. (1999). Pharmacogenomics: translating
functional genomics into rational therapeutics. Science 286: 487–
491.
Evans, W.E. and Relling, M.V. (2004). Moving towards individualized
medicine with pharmacogenomics. Nature 429: 464–468.
Gammal, R.S., Crews, K.R., Haidar, C.E. et al. (2016). Pharmacoge-
netics for safe codeine use in sickle cell disease. Pediatrics 138 (1):
e20153479. https://doi.org/10.1542/peds.2015-3479.
He, Y., Hoskins, J.M., and McLeod, H.L. (2011). Copy number
variants in pharmacogenetic genes. Trends Mol. Med. 17: 244–
251.
Hoffman, J.M., Haidar, C.E., Wilkinson, M.R. et al. (2014). PG4KDS:
a model for the pre-emptive implementation of pharmacogenetics.
Am. J. Med. Genet. C Semin. Med. Genet. 166C: 45–55.
Hunt, R.C., Simhadri, V.L., Indoli, M. et al. (2014). Exposing synony-
mous mutations. Trends Genet. 30: 308–321.
Ivanov, M., Barragan, I., and Inglman-Sundberg, M. (2014). Epigenetic
mechanisms of importance for drug treatment. Trends Pharmacol.
Sci. 35: 384–396.
Luzzatto, L. and Seneca, E. (2014). G6PD deficiency: a classic example
of pharmacogenomics with on-going clinical implications. British J.
Haematol. 164: 469–480.
Manalio, T.A. (2013). Bringing genome-wide association findings into
clinical use. Nat. Rev. Genet. 14: 549–558.
Relling, M.V. and Evans, W.E. (2015). Pharmacogenetic in the clinics.
Nature 526: 343–350.
Rukov, J.L., Wilentzik, R., Jaffe, I. et al. (2014). Pharmaco-miR: linking
microRNAs and drug effects. Brief. Bioinform. 15: 648–659.
Pharmacogenomics to improve childhood
all therapy
Annesley, C.E. and Brown, P. (2014). The biology and targeting of FLT3
in pediatric leukemia. Front. Oncol. 4: 263. https://doi.org/10.3389/
fonc.2014.00263.eCollection.
Cheok, M.H. and Evans, W.E. (2006). Acute lymphoblastic leukaemia: a
model for the pharmacogenomics of cancer therapy. Nat. Rev. Cancer
6: 117–129.
Pharmacogenomics 351
Den Boer, M.L., van Slegtenhorst, M., De Menezes, R.X. et al. (2009). A
subtype of childhood acute lymphoblastic leukemia with poor treat-
ment outcome: a genome-wide classification study. Lancet Oncol. 10:
125–134.
Diouf, B., Cheng, Q., Krynetskaia, N. et al. (2011). Somatic deletions of
genes regulating MSH2 protein stability cause DNA mismatch repair
deficiency and drug resistance in human leukemia cells. Nat. Med. 17:
1298–1303.
Diouf, B., Crews, C., Lew, G. et al. (2015). Association of an inher-
ited genetic variant with vincristine-related peripheral neuropathy in
children with acute lymphoblastic leukemia. JAMA 313: 815–823.
Fischer, U., Forster, M., Rinaldi, A. et al. (2015). Genomics and drug
profiling of fatal TCF3-HLF-positive acute lymphoblastic leukemia
identifies recurrent mutation patterns and therapeutic options. Nat.
Genet. 47: 1020–1029.
Karol, S.E., Mattano, L.A., Yang, W. et al. (2016). Genetic risk factors for
the development of osteonecrosis in children under age 10 treated for
acute lymphoblastic leukemia. Blood 127: 558–564.
Karol, S.E., Yang, W., van Driest, S.L. et al. (2015). Genetics of
glucocorticoid-associated osteonecrosis in children with acute lym-
phoblastic leukemia. Blood 126: 1770–1776.
Li, B., Li, H., Bai, Y. et al. (2015). Negative feedback-defective PRPS1
mutants drive thiopurine resistance in relapsed childhood ALL. Nat.
Med. 21: 563–571.
Liu, C., Yang, W., Devidas, M. et al. (2016). Clinical and genetic risk
factors for acute pancreatitis in patients with acute lymphoblastic
leukemia. J. Clin. Oncol. 34: 2133–2140.
Ma, X., Edmonson, M., Yergeau, D. et al. (2015). Rise and fall of sub-
clones from diagnostic to relapse in pediatric B-acute lymphoblastic
leukemia. Nat. Commun. 6: 6604.
Meyer, J.A., Wang, J., Hogan, L.E. et al. (2013). Relapse specific muta-
tions in NT5C2 in childhood acute lymphoblastic leukemia. Nat.
Genet. 45: 290–294.
Moriyama, T., Nishii, R., Perez-Andreu, V. et al. (2016). NUDT15 poly-
morphisms alter thiopurine metabolism and hematopoietic toxicity.
Nat. Genet. 48: 367–373.
Mullighan, C.G., Su, X., Zhang, J. et al. (2009). Deletion of IKZF1 and
prognosis in acute lymphoblastic leukemia. New Engl. J. Med. 360:
470–480.
Paugh, S.W., Bonten, E.J., Savic, D. et al. (2015). NALP3 inflamma-
some upregulation and CASP1 cleavage of the glucocorticoid recep-
tor cause glucocorticoid resistance in leukemia cells. Nat. Genet. 47:
607–614.
Ramsey, L.B., Panetta, J.C., Smith, C. et al. (2013). Genome-wide
study of methotrexate clearance replicates SLCO1B1. Blood 121: 898–
904.
Relling, M.V., Gardner, E.E., Sandborn, W.J. et al. (2014). Clinical Phar-
macogenetics Implementation Consortium guidelines for thiopurine
methyltransferase genotype and thiopurine dosing: 2013 update.
Clin. Pharmacol. Ther. 93: 324–325.
Relling, M.V., McDonagh, E.M., Chang, T. et al. (2013). Clinical Phar-
macogenetics Implementation Consortium guidelines for rasburic-
ase therapy in the context of G6PD deficiency genotype. Clin. Phar-
macol. Ther. 96: 169–174.
Roberts, K.G., Li, Y., Payne-Turner, D. et al. (2014). Targetable kinase-
activating lesions in Ph-like acute lymphoblastic leukemia. N. Engl. J.
Med. 371: 1005–1015.
352 Molecular Hematology
Tzoneva, G., Perez-Garcia, A., Carpenter, Z. et al. (2013). Activating
mutations in the NT5C2 nucleotidase gene drive chemotherapy resis-
tance in relapsed ALL. Nat. Med. 19: 368–371.
Wang, X., Liu, W., Sun, C.L. et al. (2014). Hyaluronan synthase 3 variant
and anthracycline-related cardiomyopathy: a report from the Chil-
dren’s Oncology Group. J. Clin. Oncol. 32: 657–653.
Yang, J.J., Landier, W., Yang, W. et al. (2015). Inherited NUDT15 vari-
ant is a genetic determinant of mercaptopurin intolerance in chil-
dren with acute lymphoblastic leukemia. J. Clin. Oncol. 33: 1235–
1242.
Yang, S.K., Hong, M., Baek, J. et al. (2014). A common missense variant
in NUDT15 confers susceptibility to thiopurine-induced leucopenia.
Nat. Genet. 46: 1017–1020.
Pharmacogenomics to improve oral
antithrombotic therapies
Ismail, S., Lee, M.Y., Duarte, J.D. et al. (2014). Genotype- and
phenotype-directed antiplatelet therapy selection in patients with
acute coronary syndromes. Expert. Rev. Cardiovasc. Ther. 12: 1289–
1303.
Johnson, J.A., Gong, L., Whirl-Carrillo, M. et al. (2011). Clinical Phar-
macogenetics Implementation Consortium guidelines for CYP2C9
and VKORC1 genotypes and warfarin dosing. Clin. Pharmacol. Ther.
90: 625–629.
Mega, J.I., Walker, J.R., Ruff, C.T. et al. (2015). Genetic and the clinical
response to warfarin and edoxaban: findings from the randomized,
double-blind ENGAGE AF-TIMI 48 trial. Lancet 385: 2280–2287.
Nagalla, S. and Bray, P.F. (2016). Personalized medicine in thrombosis:
back to the future. Blood 127: 2665–2671.
Pirmohamed, M., Kamali, F., Daly, A.K. et al. (2015). Oral anticoagula-
tion: a critique of recent advances and controversies. Trends Pharma-
col. Sci. 36: 153–163.
Scott, S.A., Sangkuhl, K., Stein, C.M. et al. (2013). Clinical Pharmaco-
genetics Implementation Consortium guidelines for CYP2C19 geno-
type and clopidogrel therapy: 2013 update. Clin. Pharmacol. Ther. 94:
317–323.
Stergiopoulos, K. and Brown, D.L. (2014). Genotype-guided vs clinical
dosing in warfarin and its analogues. Meta-analysis of randomized
clinical trials. JAMA Intern. Med. 174: 1330–1339.
Chapter 25 History and development of
molecular biology
Paul Moss
School of Cancer Sciences, University of Birmingham, Birmingham, UK
Evolution is the central tenet of biology, 353
The understanding of monogenic and polygenic inheritance, 353
DNA as the conduit of genetic information, 354
The rough guide to the human genome, 357
Evolution is the central tenet of
biology
It is 160 years since the publication of On the Origin of Species
by Means of Natural Selection (Figure 25.1a). Darwin’s far-
reaching insights placed natural variation and adaptation as
the prime determinants of population change. Evolution is
now recognized as the unifying theme of all biology, includ-
ing hematology and medicine, and these landmark observa-
tions can perhaps be recognized as the initiation of the mod-
ern discipline of molecular genetics.
However, despite his unique vision and imagination, Dar-
win was never able to understand the nature of genetic inheri-
tance. It was clear that phenotypic characteristics were passed
from one generation to the next, but usually these charac-
teristics were “blended,” so that a combination of tall and
short parents would produce a child of medium height. Dar-
win proposed his own model for heredity which he termed
pangenesis, in which cells of the body shed gemmules that
collect in the reproductive organs. The concept was that all
the tissues in the body thus had some impact on inheri-
tance, but this concept has clearly been superseded. At this
time there was no concept of the distinction between the
germline and somatic tissue, and it is therefore not surprising
that even brilliant scientists such as Jean-Baptiste de Lamarck
believed that acquired phenotypic features could be passed,
through reproduction, into the subsequent generation. How-
ever, despite his somewhat imperfect conclusions, Lamarck
was actually an extremely important figure in evolutionary
theory, as he developed the concept that a species can change
between different generations, a view entirely in contrast to
Molecular Hematology, 4th Edition. Edited by Drew Provan and John G. Gribben.
© 2020 John Wiley & Sons Ltd. Published 2020 by John Wiley & Sons Ltd.
Experimental techniques and molecular biology, 358
Conclusion, 362
Further reading, 362
the prevailing view that all species had been placed on the
earth by a divine Creator.
Even in Darwin’s time there were fierce arguments about
the nature of biological inheritance. Samuel Butler, later to
become a celebrated novelist, emigrated to New Zealand to
make his fortune in sheep breeding. While there, he read Dar-
win’s work and, through his own highly successful breeding
experiments, came to understand the concept that “informa-
tion” was flowing from one generation to the next. His ideas
were taken further by William Bateson, a Cambridge geneti-
cist who introduced the term genetics in 1906 to describe
the science of inheritance and variation. The units of this
information were subsequently termed genes by Wilhelm
Johannsen.
The understanding of monogenic
and polygenic inheritance
The modern concept of genetic inheritance began with the
work of Gregor Mendel, an Austrian priest who had under-
gone training as a science teacher and is famous for his
pea-breeding experiments published in 1866 (Figure 25.1b).
Mendel used true-breeding varieties of pea to show that the
crossing of two dissimilar strains led to a uniform first gener-
ation (F1 ). However, the crossing of the F1 generation with an
F2 generation produced a population where three-quarters of
the peas had the same (dominant) parental characteristic and
one-quarter had the other (recessive) parental form. In the
course of over 30 000 observations, Mendel went on to breed
subtypes of F2 and laid the rules for dominant and reces-
sive inheritance. He postulated that each characteristic was
determined by two elements from the parents, and that these
were both retained and appeared in the germline as single
forms in equal proportions. The final expression in each plant
353
354 Molecular Hematology
(a) (b)
(d)
(c)
was dependent on the combination of dominant or recessive
elements. These experiments were hugely influential to the
development of molecular biology, as they showed that her-
itable characteristics are particulate and not “blended.” His
work was largely ignored for the next 35 years, until it was
rediscovered in the early twentieth century and widely sup-
ported by Bateson and colleagues. The particulate element
became known as the gene and the modern basis of genetics
was established.
A huge problem in these early days was how to link
the clear biological results of Mendelian experiments with
the discontinuous phenotypes seen in everyday life. For
instance, how can single pieces of information determine
such variables such as height or intelligence? It was left to
the work of R.A. Fisher in 1918 to show that the inher-
itance of multiple different single genes could produce
Fig. 25.1 (a) Charles Darwin. (b) Gregor Mendel.
(c) Francis Crick and James Watson. (d) Fred Sanger.
a phenotype that displayed discontinuous properties, and
the concept of polygenic inheritance of characteristics was
thus born.
DNA as the conduit of genetic
information
From this time, the science of genetics became firmly estab-
lished and rooted in all aspects of biological investigation.
However, the nature of the genetic material was completely
unknown until the pioneering work of Oswald Avery in
1944. Avery trained as a medical doctor and was 67 at
the time of his most famous scientific contribution. He
had been working on the inheritance of S and R strains
of Streptococcus pneumoniae and was able to show that,
 History and development of molecular biology 355

contrary to the current view, the use of proteases could 5ʹ end 3ʹ end
not prevent transmission of genetic information between O O–
strains. In contrast, deoxyribonuclease was highly effective in NH2 O
P
N T OH
preventing transmission, and this work therefore established –O
O N HN N
DNA as the medium for genetic inheritance. Avery died in A
O N N O
1955 and is often regarded as one of the most deserving O
scientists never to have been awarded the Nobel Prize in
Medicine or Physiology. Having established DNA as the O O–
O O
physical basis of heritability, genetic research then proceeded P NH2 O N P
to dominate biological investigation for the next 50 years. –
O O
O O
C N HN G N
O N
Elucidation of the structure of DNA O H2N O

The elucidation of the structure of DNA has become per-

haps the most memorable biological achievement of the last
century. The story is well known, and involves the outstand-
ing X-ray crystallography of Maurice Wilkins and Rosalind
Franklyn, combined with the imagination and intelligence of
James Watson and Francis Crick (Figure 25.1c). Their paper,
published in 1953, took advantage of a wealth of previous
information from other investigators, and this was instru-
mental in allowing a final elucidation.
In 1919, Levene had demonstrated that DNA consisted of
three components: the bases (of which there were four types),
a sugar (which was either ribose or deoxyribose), and a phos-
phate group. He showed that these were linked together in
a unit that he called a nucleotide. The work of Irwin Char-
gaff defined the so-called Chargaff ratio, which relates to the
base composition of DNA. This showed that the amount of
guanine equals that of cytosine, and the amount of adenine
equals that of thymine. Structural analysis revealed that the
hydrogen bonds between guanine and cytosine and between
adenine and thymine were of similar structure, and led to a
model in which the hydrophobic bases were on the inside of
the double helix, leaving the hydrophilic components on the
outside (Figure 25.2).
The structure of DNA has now been well characterized
and comprises three different subunits, a sugar, a base, and a
phosphate group; together these form a nucleotide. DNA is
a polymer of nucleotides and it is the sequence of bases that
determines the genetic information. Certain aspects of the
nucleotide component are worth looking at in more detail.
The backbone of the strand is made from phosphate and
sugar residues, the latter being a 2-deoxyribose five-carbon
(pentose) sugar. Importantly, these sugars are joined either at
their third or fifth carbon atom, and this asymmetry gives a
polarity to the DNA strand, with the 5′ end having a terminal
phosphate group and the 3′ end a terminal hydroxyl residue.
The phosphate groups form phosphodiester bonds, which
contain a phosphorus atom that allows strong covalent
linkage through an ester bond. These groups are negatively
charged and this plays a role in forcing phosphates away
from the center of the DNA helix, as well as combining with
OH O O–
3ʹ end P
–O
O
5ʹ end
Fig. 25.2 The structure of DNA. The deoxyribose pentose sugars are
joined by phosphodiester bonds at C-3 or C-5 to phosphate groups.
The four bases, adenine (A), thymine (T), cytosine (C), and guanine
(G), are attached to this backbone and the double helix is stabilized by
hydrogen bonds (represented by zig-zag lines).
positively charged proteins termed histones that package the
DNA. The core genetic complexity of the DNA molecule
comes from the composition of bases at the center of the
helix. These bases are of two types, purines and pyrim-
idines, the former comprising adenine and guanine, fused
heterocyclic compounds of five- and six-membered rings,
the latter cytosine and thymidine, which are more simple
six-membered rings. The pairing of guanine with cytosine
provides greater strength of interaction due to optimal
hydrogen bonding.
The mechanism by which DNA replicates was a further
question that needed to be resolved. Watson and Crick had
ended their paper with the famous words: “it has not escaped
our notice that the specific pairing we have postulated
immediately suggests a possible copying mechanism for the
genetic material.” This implied that the two helices could
separate and allow copying of each strand. In 1958, Matthew
Meselson and Franklin Stahl showed that the replication of
DNA was semi-conservative. That is, when a copy is made
of a double-stranded DNA template, a complementary
sequence is made against each of the two original strands,
rather than production of a completely new double strand
with retention of the original duplex.
Decoding the sequence of DNA
Later work by Francis Crick predicted that “adapter” inter-
mediates would be found which decoded the DNA sequence
356 Molecular Hematology
into the synthesis of proteins. Transfer RNA was subsequently
isolated and found to comprise a triplet anticodon attached
to a specific amino acid. In this way, the DNA code can be
translated into a protein sequence. There are 20 different sub-
types of amino acid and these are encoded by a range of dif-
ferent triplet sequences on transfer RNA. This is the genetic
code that was elucidated in the early 1960s and was set out by
Crick in the form in which it is now used. Translation usually
commences with an AUG triplet, which codes for methion-
ine. Protein synthesis occurs within the cellular cytoplasm on
ribosomes, which are complex mixtures of RNA and protein
components.
This flow of genetic information from DNA to RNA to pro-
tein became known as the central dogma of molecular biol-
ogy. Although never meant to be as truly closed to critical
evaluation as the term implies, this observation was highly
influential in understanding the centrality of germline DNA
in evolutionary processes, and essentially contradicted the
concept that somatic information could flow back into the
germline, as envisaged in the original Lamarckian theory of
evolution.
RNA is a major component of the cellular nucleus and
three main forms were distinguished in the early days of
molecular biology. Messenger RNA (mRNA) copies the DNA
genetic code into transcripts that travel to the cytoplasm
for protein translation. This process of transcription allows
different cells, all of which carry the same gene content, to
utilize their genome in differential ways, and transcriptional
regulation lies at the heart of cell differentiation programs.
The initial concept that one gene gives rise to a single mRNA
species, and subsequently a single protein molecule, has
been challenged in several ways over the last few decades.
The first observation was that genes are split into discrete
units of expressed code, termed exons, and these DNA
segments are separated by intervening sequences of intronic
DNA. Most of the genes within the human genome have this
intron–exon structure and, although the selective advantage
of this development is unknown, it does allow increased
flexibility for decoding genetic information into a range of
protein transcripts. In order for an initial mRNA species to
encode an accurate representation of the exon structures,
the transcript must be spliced by splicing enzymes within the
nucleus by a process that removes intervening intronic RNA
segments. Many genetic diseases have now been revealed
as due to aberrations in splicing mechanisms, and the tha-
lassemia syndromes, which arguably have told us more about
the principles of molecular regulation than any other genetic
disease, are commonly inactivated by mutations in this
mechanism.
Ribosomal RNA (rRNA) is complexed with ribosomal pro-
teins to produce the large number of ribosomes that are scat-
tered around the cytoplasm and act as the principal sites for
protein translation. There are four different rRNA molecules
within the ribosome, 18S, 5.8S, 28S, and 5S, and together
they constitute over 80% of RNA within the cell. The com-
bination of transcription and translation can be surpris-
ingly slow. It has been estimated that it can take 21 hours
to produce a molecule of dystrophin, the largest gene in
the genome and the one responsible for Duchenne muscular
dystrophy.
Recently, several new classes of RNA species have been
discovered and are likely to play very important roles in the
genetic regulation of eukaryotic cells. A particularly impor-
tant discovery has been the finding that RNA can have cat-
alytic activity, demonstrating that it is more than simply
a passive scaffold. MicroRNA species are approximately 21
nucleotides in length and do not code for proteins but, on
the contrary, carry complementary sequences that are able
to bind to mRNA species and regulate the rate of protein
production. This is done either by binding to mRNA species
and preventing translation on ribosomes, or via recruitment
of ribonucleases that degrade double-stranded RNA species
and thus reduce the half-life of the targeted RNA species.
Current estimates suggest that there are over 1000 micro-
RNA transcripts within the human genome, and it is cur-
rently believed that mutations or deregulation of microRNA
expression can be implicated in a range of malignant dis-
eases; involvement of deletions of microRNAs in the etiology
of chronic lymphocytic leukemia secondary to chromosome
13q deletions is one such example. The net overall effect of
these systems is that the cell thus produces not only mRNA
for protein translation, but also species of RNA with com-
plementary sequence to mRNA transcripts that serve to limit
information derived from each mRNA transcript. This feed-
back control is, of course, a commonly identified mechanism
in many biological systems.
A range of other RNA species are now being identi-
fied. These include small nuclear RNAs (snRNAs), which
are involved in processes such as mRNA splicing, and small
nucleolar RNAs. Many long non-coding RNA species of over
200 nucleotides in length are also produced whose functions
are largely unknown. They are often very highly conserved
in evolution and found in gene-poor regions of the genome.
It is likely that they play important and indispensable roles in
embryonic development.
Regulation of transcriptional activity
The DNA genome encodes the information from which our
complex multicellular structure can be assembled. However,
the code is deciphered in a highly complex process involving
the interaction of many different proteins and nucleic acids.
DNA is copied into RNA by the RNA polymerase II
enzyme, which recognizes a promoter sequence at the 5′ end

of the gene. The organization of promoter sequences is highly
disparate, although many contain the TATAAA sequence
known as the TATA box. Promoter sequences attract a large
number of proteins called transcription factors, which serve
to activate or repress transcription at that gene. These pro-
teins all contain a DNA-binding domain and attract an addi-
tional set of proteins involved in chromatin remodeling that
also regulate transcriptional activity (see Epigenetic Regula-
tion Adds a New Dimension of Complexity).
A further set of DNA regulatory elements are enhancer
sequences that regulate the transcriptional activity of genes
which are often many kilobases away. Their precise mecha-
nism of action is uncertain, but is likely to involve processes
such as “looping out” of DNA and colocalization of transcrip-
tion factors.
The rough guide to the human genome
The human genome contains a sequence of 3000 million
bases of DNA from each parent, equating to some 6 billion
bases in a diploid cell. The Human Genome Project started
in 1990 and 99.3% of the sequence was completed by 2003.
Indeed, the genome has now been fully sequenced for sev-
eral individuals, and this has risen substantially over the last
decade. Large-scale sequencing has revealed a number of sur-
prises about the human genome. The first of these is that
the gene content is much lower than was initially estimated.
Current estimates suggest a gene content of around 25 000,
whereas early proposals placed this nearer 100 000. In addi-
tion, only 2% of the genome actually codes for protein. The
additional 98% has a complex regulatory function and its
importance is still largely unknown. During the early days
of molecular biology, these regions were rather arrogantly
termed “junk DNA,” but they have since turned out to be
extremely important in factors such as determining specia-
tion and regulation of gene expression.
Different individuals show considerable variation within
the genome. This natural polymorphism is typically seen
as a base change of around 1 in every 1300 bases, equating
to over 4 million differences between individuals. These
variants are known as single-nucleotide polymorphisms
(SNPs) and, as well as their intrinsic value in understanding
issues such as natural selection and evolutionary change,
they are also very valuable in determining the genetic linkage
of genes in a variety of human diseases. Indeed, SNP-based
disease association studies are currently the most popular
approach for determining the polygenic basis of disease, and
around 50% of these variants are in non-coding regions of
the genome. Interestingly, most of these variants are quite
common in the human population and there is less variation
between humans than there is between other great apes.
History and development of molecular biology 357
This is believed to reflect the fact that the current human
population of some 6 billion people has grown rapidly over
3000 generations from a founder size “bottleneck” of only
10 000 individuals.
Sequencing has also revealed that copy number variation is
a common finding in the human genome. We seem to have
a dynamic genome in which relatively large areas of deletion
and insertion are seen, and this may underlie a range of disor-
ders that may involve cryptic gene inactivation and translo-
cations. It may be that this high level of genomic instability
has contributed to our rapid evolution, but clearly there also
appears to be a price to pay.
The packaging of genetic material
One fact that almost all of us remember is that each cell in the
human body contains approximately 1 m of double-stranded
DNA, if it were stretched out from end to end. Given that we
have some 1014 cells within our body, the whole content of
our somatic DNA would stretch for some 1014 m. It is no sur-
prise therefore that this length of DNA must be packed very
accurately and densely into a typical cell of around 8 μm. The
first order of packaging is the nucleosome, in which two loops
of DNA are wrapped round a histone protein. As we shall see
later, this structure is critical not only for efficient mechan-
ical packaging, but also for controlling gene regulation. The
second-order structure is coiling of nucleosomes into a rope-
like structure, but further order packaging beyond this is cur-
rently unknown.
Epigenetic regulation adds a new
dimension of complexity
Additional complexities to understanding the transcriptional
regulation of the DNA code are being identified on a regular
basis. It is now clear that protein-induced modifications of
histone structure play an important role in determining the
transcriptional regulation of individual genes. The two major
processes involved are methylation of cytosine bases at CpG
dinucleotides and acetylation of histone tails. A wide range
of methylases, demethylases, acetylases, and deacetylases are
involved in dynamic modification of histone structures, and
this process is now being targeted in clinical practice. Around
75% of all CpG dinucleotides are methylated and unmethy-
lated forms are often grouped at the 5′ end of genes, in areas
known as CpG islands.
In general, methylation serves to downregulate transcrip-
tion from a gene and demethylating agents are used in the
management of diseases such as myelodysplasia or acute
myeloid leukemia. In contrast, acetylation encourages tran-
scriptional activity and deacetylation inhibitors are therefore
also used in a range of hematopoietic disorders. Given the
358 Molecular Hematology
widespread modifications of methylation and acetylation in
the human genome, it is perhaps surprising that the non-
specific activity of these pharmacological agents can prove of
value without engendering significant side effects. However,
current practice suggests that they may be relatively well
tolerated. Epigenetic regulation also represents one example
in which somatic tissues may influence transcription of
the germline. Gender, in particular, can have an important
impact on the transcriptional activity of individual genes.
This process of imprinting can be broadly summarized as
revealing that males tend to encourage fetal growth, whereas
the maternal influence is to restrict fetal developmental.
Aberrations of epigenetic imprinting result in disorders such
as Prader–Willi or Angelman syndrome.
Experimental techniques and
molecular biology
Here I review some of the milestones that have led to the
development of the experimental discipline known as molec-
ular biology. This is a field which is moving perhaps more
rapidly than any other within biology and, while it offers
unparalleled opportunities for clinical practice, it is challeng-
ing for clinicians to maintain their knowledge base within
this area.
The extraction of nucleic acids
A good place to begin is the extraction of nucleic acids from
cells. Nucleic acids are soluble in water, but can be precipi-
tated by the addition of ethanol. In one method the cell or
tissue of interest is lysed to destroy cell membranes and the
protein component degraded by proteinase K or extraction
with phenol and chloroform. The latter approach generates
an aqueous phase containing RNA and an organic phase that
retains protein and lipid, with DNA being trapped at the
interface.
The next step is to purify and concentrate the nucleic acid,
which can be done by ethanol precipitation. Because DNA
is charged, it is highly soluble in water, but the addition of
non-polar ethanol to at least 66% of the final volume draws
it out of solution. The precipitate of nucleic acids can then
be centrifuged, the ethanol drawn off, and the nucleic acid
resuspended in water. Genomic DNA is quite stable in this
form, perhaps not surprising given its natural physiologi-
cal role. However, RNA is highly unstable and susceptible
to RNases, which are ubiquitous enzymes within the envi-
ronment. Inhibitors of RNases and rapid cooling of RNA
solutions are needed for downstream analysis of ribonucleic
acids. Although these chemical steps are effective, and have
been in use for many decades, many laboratories now use
disposable cartridge methodology for extraction of nucleic
acids. These are rapid, often cost-effective, and generally
much easier to use.
Restriction enzymes and expression cloning
Having isolated DNA in an aqueous solution, there is not a
great deal that you can usefully do with it in its native state.
If it is run on an electrophoresis gel, it will be of such high
molecular weight that it will not migrate into the gel to any
extent and is therefore not suitable for downstream analy-
sis. A major advance in molecular biology was the devel-
opment and utilization of bacterial restriction endonucleases.
These are a large family of enzymes that are present in bac-
teria and have evolved to digest double-stranded DNA. They
are important in protecting bacteria from invading bacterio-
phages and thus they “restrict” the vulnerability of the cell
to infection. The critical finding about restriction endonu-
cleases was that they cut DNA at specific DNA sequence
motifs. For instance, the enzyme EcoRI will only cut DNA
at positions that have the sequence 5′ -GAATTC. Note that
this sequence is a palindrome when double stranded and
it is in this form that the DNA is cleaved, after the G on
both strands. If sufficient enzyme is used in a digestion
reaction, every individual site within the target DNA will
be cut.
These enzymes opened up the field of modern molecular
biology, because they led to the development of DNA cloning.
In this procedure, the DNA of interest is digested with a par-
ticular endonuclease and then a vector DNA sequence, usu-
ally in the form of a bacterial plasmid or a virus, is also
digested with the same enzyme. Providing that the vector has
only two of these restriction sites, a portion of the vector will
be excised and then, by combining the digests of the target
DNA and vector, the two populations can be ligated such that
single copies of DNA are cloned into the vector. Typically
these vectors have strong promoters that drive expression of
the cloned gene, as well as genes coding for antibiotics, which
allow selection for vectors containing an insert. In this way a
library of the original DNA is made.
This library can then be introduced into cells for propaga-
tion and functional analysis. Vectors can be transformed into
a bacterial host that can be multiplied using culture medium
to a potentially unlimited population size. Lysis of the bacte-
rial colony, followed by recovery of the plasmid or phage, can
lead to isolation and analysis of large amounts of individual
target DNA sequences. This is a core technique in molecular
biology and has led to most of the subsequent development
within the field. Libraries can also be introduced into eukary-
otic cells through the process of transfection, which typically
involves incubation with cold cations such as calcium or the
use of liposomes.
 History and development of molecular biology 359

Electrophoresis and blotting
of nucleic acids
Once DNA has been digested with endonucleases, it is bro-
ken up into smaller pieces, the exact size and pattern of which
will depend on the endonuclease that has been used. Edwin
Southern published a technique whereby this DNA was ini-
tially subjected to electrophoresis on an agarose gel, thus sep-
arating individual pieces on the basis of their size and charge.
If a nitrocellulose filter is then placed on top of the agarose
gel and moderate pressure applied, the DNA will migrate
into the nitrocellulose, where it can be fixed using heat or a
cross-linking agent. If the DNA filter is then hybridized to a
labeled nucleic acid probe, the probe will bind only to that
section of the DNA containing a complementary sequence.
When genomic DNA from different individuals is digested
with the same nuclease, all the resulting DNA fragments will
be of comparable size, notwithstanding the natural variation
found within the human genome. Thus, the hybridization
band should be similar in different individuals. This tech-
nique became known as the Southern blot and was widely
applied to the detection of chromosomal abnormalities such
as deletions, linkage studies, and gene rearrangement iden-
tification. It is somewhat less widely used today due to the
advent of more modern technologies such as SNP mapping
and DNA sequencing.
Southern blot technology has been adopted for use in other
ways and a memorable nomenclature has built up, albeit
based on a mischievous interpretation of the derivation of
“Southern.” The Northern blot refers to a technique in which
RNA is run on an agarose gel and blotted in the same way
as the DNA in a Southern blot. This approach was widely
used to examine differential expression of DNA between tis-
sues, but has largely been replaced by microarray analysis.
Western blotting refers to the technique of separating protein
molecules by electrophoresis prior to blotting onto a mem-
brane. This membrane is then probed with an antibody for
detection of the presence, size, and amount of individual pro-
tein species.
The polymerase chain reaction
The development of the polymerase chain reaction (PCR) in
the 1980s led to the award of a Nobel Prize to Kary Mullis in
1993, and its value in medical practice is perhaps reflected by
the relative swiftness of the Nobel Committee in rewarding
the inventor. The idea is, like most powerful innovations,
fairly straightforward and involves repetitive cycles of DNA
synthesis and dissociation (Figure 25.3). DNA polymerases
are used to copy a specific DNA segment, and this step is
followed by an increase in reaction temperature to dissociate

3ʹ 5ʹ
DNA template
5ʹ 3ʹ

Primer
Newly synthesized
Annealing of 5ʹ
DNA strand
onucleotide Newly synthesized
primers 5ʹ DNA strands

DNA polymerase
copies
from the primer

Double strands are

separated by high
temperature

Temperature is
reduced to allow 5ʹ
annealing of primers

Fig. 25.3 The polymerase chain reaction. Typical

temperatures for annealing, extension, and DNA polymerase
5ʹ
extends the primers
denaturation are 50–60, 72, and 96 °C, respectively.
360 Molecular Hematology

DNA double strands and provide further templates for the biology and has played a large role in driving the discipline
amplification process. forward over the last 20 years.
Mullis performed his initial work using a hot water bath
and had to combine DNA polymerase with free nucleotides
for the initial synthesis, heat the reaction to “melt” the dou- Microarray analysis
ble strands, and allow the mixture to cool for reannealing Although sequence information of germline DNA provides
of the original primer mix, before then adding further poly- highly valuable information, in many situations it is more
merase enzyme, as the original aliquot had been killed by the valuable to know how this information is utilized within indi-
high temperatures used in the melting process. Since then, vidual cells or tissues. One approach to this is to use systems
the technique has been dramatically simplified through the which determine the quantitative profile of mRNA expression
use of thermostable DNA polymerases isolated from bacteria within tissues. Microarrays achieve this by a technique which
that live at high temperatures. Solid-phase technology, which relies on the synthesis of a huge number of oligonucleotides
allows very rapid shifts in temperature, has led to a dramatic that are complementary to each of the approximately 25 000
reduction in incubation and extension times. The technique human genes. These oligonucleotides are placed individually
utilizes short oligonucleotide primers that anneal to the tar- on microarray slides and are then hybridized to sample
get gene of interest. The DNA polymerase extends the DNA tissue. mRNA from the sample is reverse transcribed to
sequence beyond the primary sequence and, as the primers cDNA with the use of fluorescent primers, and then this
are directed toward one another on complementary strands material is hybridized to the oligonucleotides; after washing,
of DNA, the result is a complete double-stranded copy of
the target gene. Typical PCR reactions might allow ampli-
fication of between 500 and 2000 bp of DNA, but innova-
tions do exist that can extend these values. Once a gene has
been amplified in this way it can be sequenced, subjected DNA sample
to enzymatic digestion, or used in a range of downstream
applications.
Fragmentation of
DNA
DNA sequencing
Although the physical structure of DNA was defined in the P B P B Attachment of linkers
1960s, it would be many years before it became possible with biotin (B) group
to sequence the bases in order to reveal the genetic code.
Interestingly, protein sequencing had been reported many Single DNA fragment is
B
attached to a streptavidin-
years earlier, but the technology of DNA sequencing has now S
coated bead
advanced so much that its capability far exceeds current abil-
ities to sequence amino acids.
The two original methodologies were developed by Fred DNA is amplified on
Sanger and Walter Gilbert, who shared the Nobel Prize the bead
for Chemistry in 1980 (Figure 25.1d). The techniques
involved the use of DNA polymerases to sequence DNA
from an oligonucleotide primer. However, within the mix
of nucleotides needed for the building blocks of the new
Beads are dispersed
chains, Sanger’s approach incorporated a small quantity into individual wells
of nucleotides that have dideoxy modifications of their
sequence such that they cannot be extended. Four separate
reactions were used for each DNA sequence, incorporating
The four nucleotides
dideoxy reagents for the A, T, C, or G bases, respectively. 1 T are sequentially passed
If radioactivity or fluorescence is used to label the newly A
over the beads. A
2 primer complementary to
extended chains, these can then be run on a high-resolution
3 C the linker is incorporated
acrylamide gel and from this a ladder of chains is derived,
which shows the stop positions that occur when each base luminescence
C When a nucleotide
detector
terminator is used. These bands can then be read off and can be added it
a DNA sequence can be deduced. This technology was of B CTAG P emits a light signal
tremendous value in initial efforts to understand molecular Fig. 25.4 The principles of 454 next-generation sequencing.

Darwin proposes 1859
theory of natural
selection as mechanism
for evolution
William Bateson first
uses the term “genetics” 1906
Levene elucidates the
1919
three components if DNA and
terms the units “nucleotide”
Oswald Avery demonstrates
that DNA carries genetic 1944
information
Meselson and Stahl show
that DNA replication is 1958
semi-conservative
Arbor, Smith and
1978
Nathans win the Nobel Prize
for discovery of restriction enzymes
1980
Sanger and Gilbert won Nobel
Prize for developing DNA sequencing 1990
The Human Genome Project 2003
reads the DNA code
Fig. 25.5 Some of the key developments in molecular biology.
the amount of hybridized material is assayed using laser
detection. This approach is a highly accurate and repro-
ducible means of measuring the distribution of mRNA
species within cells or tissues. Of course, there is no simple
correlation between mRNA levels and protein expression,
and proteomic analysis allows additional understanding
of cellular differentiation control. Microarray systems are
used widely in diagnostic and investigational approaches to
hematopoietic malignancy.
Next-generation sequencing
Over the last couple of years, new technologies have been
developed that have dramatically altered the capability
of DNA sequencing. Two broad platforms are currently
available and are known as 454 pyrosequencing and the
Illumina/Solexa systems. Both technologies commence with
fragmentation of DNA and then small DNA fragments are
captured individually onto beads prior to an amplification
step. In the 454 system, these beads are then centrifuged
into small wells such that each well contains only one bead,
and these are sequenced individually. The 454 system is
based on luminescence, each base generating a different
color; the cameras used for color detection are based on
History and development of molecular biology 361
Mendel defines the
1866 patterns of dominant
and of recessive inheritence
Fisher develops the
1918 concept of polygenic
basis for phenotype
Watson, Crick, Wilkins, and
1953 Franklyn describe the structure
of the DNA double helix
The genetic code is
1960–1966
elucidated
Ed Southern describes the
1975
“Southern Blot”
The development of the
1983 polymerase chain reaction
1990s Microarray technology
“Next-generation” sequencing
2005+
those developed for the Hubble telescope and have very high
sensitivity (Figure 25.4).
These technologies have led to a step change in the abil-
ity to rapidly sequence DNA and they produce almost fright-
ening levels of information. A single machine run can gen-
erate 500 million base pairs of sequence, and most of the
challenges relate to the bioinformatic analysis of the infor-
mation or simple storage of data. The 454 system sequenced
the entire genome of James Watson in two months, and in
time it is likely that many of our patients, and indeed per-
haps ourselves, will undergo full genome sequencing. Our
current capabilities are probably only limited by our imagi-
nation and they allow experiments such as comparative anal-
ysis of the complete sequence of germline DNA in compari-
son with tumor cells, or study of the evolution of infectious
organisms in a human host almost in real time.
Ultra-deep sequencing refers to the ability to sequence
many different copies of an individual sequence in one run,
also known as in parallel. As such, it can be very powerful for
detecting rare mutations in tissue samples or for following
genetic variation within populations of viruses.
Transcriptome sequencing is the sequence analysis of
the mRNA population in a cell or tissue. There is the
possibility that this approach could replace some of the
362 Molecular Hematology
uses of microarray analysis, as it simultaneously provides
information on the number of mRNA species in the sample
as well as determining their complete nucleotide sequence.
Conclusion
The dramatic developments in molecular biology over the
last 50 years represent one of the most exciting stories of sci-
entific development (Figure 25.5). Refreshingly, they have
also been applied quickly and effectively to human health,
and are now integral to the clinical practice of hematology. A
more detailed analysis of the molecular basis of a wide range
of conditions is discussed more fully in the other chapters of
this book.
Further reading
Ansorge, W.J. (2009). Next-generation DNA sequencing techniques.
Nat. Biotechnol. 25: 195–203.
Crick F, Watson J. (1962). On the genetic code. http://nobelprize.org/
nobel_prizes/medicine/laureates/1962/crick-lecture.html (accessed
May 2019).
Reichard, P. (2002). Osvald T. Avery and the Nobel Prize in Medicine. J.
Biol. Chem. 277: 13355–13362.
Sanger F. (1980). Determination of the nucleotide sequences in
DNA. http://nobelprize.org/nobel_prizes/chemistry/laureates/1980/
sanger-lecture.html (accessed May 2019).
Siomi, H. and Siomi, M.C. (2009). On the road to reading the RNA-
interference code. Nature 457: 396–404.
Chapter 26 Cancer stem cells
Sara Ali & Dominique Bonnet
Haematopoietic Stem Cell Laboratory, The Francis Crick Institute, London, UK

The cancer stem cell concept, 363 CSC-targeted therapies, 365

The cell of origin, 364 Conclusion, 369
Pre-leukemic stem cells, 364 Further reading, 370

a small proportion of leukemic cells (0.2–100/106 leukemic

The cancer stem cell concept
blasts) and to reside in the same compartment as normal
HSCs (lineage negative, CD34+CD38− subset), leading
It was more than a century ago when the term stem cell
to the conclusion that AML likely arises from neoplastic
(Stammzelle) was coined by Haeckel to refer to the unicellular
transformation occurring in primitive stem cells rather than
ancestor of multicellular organisms that stands at the bottom
their committed progeny. These findings sparked interest
of a genealogical tree. Pappenheim later adopted the term
in exploring CSCs in various malignancies, and CSCs have
and applied it in its current meaning to describe the precur-
since been reported in solid tumors as well, providing further
sor cell, which he postulated could give rise to all blood cells.
credence to the CSC concept and extending its applicability
However, it was decades later that experimental proof for the
beyond hematological malignancies.
existence of hematopoietic stem cells (HSCs) was provided.
Although the CSC paradigm appears to hold true for dif-
While studying the effects of radiation on bone marrow func-
ferent cancers, it certainly is not universal. One major caveat
tion, Till and McCulloch made the seminal observation of
that became apparent with the emergence of more permis-
hematopoietic colonies resembling nodules in the spleens of
sive xenotransplantation models is the underestimated fre-
lethally irradiated mice that were transplanted with limiting
quency of tumor-initiating cells (TICs) reported in earlier
numbers of murine bone marrow cells. Even more striking at
studies which employed mouse strains that still had signif-
the time was the observation that upon retransplantation of
icant residual immune function. A typical example illustrat-
these colonies into secondary recipients, some were able to
ing this limitation is melanoma, which was thought, based
generate new multilineage colonies, indicating the ability of
on earlier findings, to follow the CSC model. However, this
these cells to self-renew.
was called into question when it was subsequently reported
Akin to HSCs, the cancer stem cell (CSC) hypothesis
that the frequency of cells with tumorigenic potential in
posits that tumors assume a hierarchical organization in
melanoma could be as high as 25%, arguing against there
which a subpopulation of cells with stem-like properties
being a hierarchical organization. One possible explanation
reside at the apex, and have the capacity to self-renew and
for this observation is provided by the stochastic model,
regenerate tumors that recapitulate the parental tumor from
which proposes that all cells are biologically equivalent with
which they originate. The hierarchical model in cancer was
equal clonogenic potential to facilitate tumor initiation and
first exemplified by the pioneering work of Professor J.E.
growth, provided they receive the appropriate cues, which
Dick’s team, which led to the identification of leukemic stem
may either be intrinsic or extrinsic. It must be noted, how-
cells (LSCs) in acute myeloid leukemia (AML). Leukemia-
ever, that high frequency of TIC does not, in itself, indicate a
initiating cells (LICs), functionally defined by their ability
stochastic pattern of tumor development. What distinguishes
to repopulate non-obese diabetic/severe combined immun-
the CSC paradigm from the stochastic model is that accord-
odeficient (NOD/SCID) mice, were found to represent only
ing to the former framework, only a unique subset of cells
is endowed with long-term tumorigenic potential, and they
can, based on their intrinsic properties, be separated from
Molecular Hematology, 4th Edition. Edited by Drew Provan and John G. Gribben. non-TIC cells. However, this is not possible in cancers that
© 2020 John Wiley & Sons Ltd. Published 2020 by John Wiley & Sons Ltd. follow a stochastic model, as all cells are equally permissive to

363
364 Molecular Hematology
transformation, and therefore TICs can be potentially found
in any cell fraction.
Nevertheless, the two models are by no means mutually
exclusive, and can operate simultaneously within a tumor.
Furthermore, several lines of evidence, based on findings in
normal and malignant tissues, point toward stemness being
a dynamic trait whereby cells can transition between a stem
cell and a differentiated cell state and vice versa. This notion,
known as bidirectional interconversion, provides an alterna-
tive explanation that can reconcile these two models.
The cell of origin
Although often used interchangeably, the CSC is not nec-
essarily the same as the cell of origin. The cell of origin
refers to the cell that sustains the first hit required for tumor
initiation, but does not necessarily have stem cell properties
to allow it to sustain malignant growth. Equally important to
note is that CSCs need not be derived from their normal stem
cell counterparts, but are rather defined as cells with stem
cell properties and tumor-propagating capacity, regardless
of whether they originated from a bona fide stem cell or not.
Indeed, there is significant evidence indicating that CSCs
can arise from downstream progenitor cells as they undergo
oncogenic transformation. In a study by Taussig et al., sorted
leukemic cells from patients with AML with the progenitor
phenotype CD34+ CD38+ were able to induce leukemia in
various immunodeficient mouse strains, indicating that LSC
activity is not restricted to the CD34+ CD38− compartment
as previously hypothesized. Furthermore, lentivirally trans-
duced myeloid progenitors expressing the mixed-lineage
leukemia (MLL) fusion genes MLL-AF9 and MLL-ENL have
been shown to be able to initiate leukemia. The results of
global gene expression profiling of LSC populations in AML
have also challenged the prevailing notion that LSCs in AML
largely arise from stem cells, and suggest instead that LSCs
in AML resemble lymphoid-myeloid multipotent progen-
itors (LMPPs) and granulocyte-macrophage progenitors
(GMP) more closely than HSCs.
Although CSCs can originate from a differentiated pro-
genitor, the cell of origin may still be a stem cell. For example,
while it is widely accepted that chronic myeloid leukemia
(CML) results from neoplastic transformation of an HSC by
the BCR-ABL fusion gene, progression to blast crisis has been
shown to be driven by progenitor cells that have acquired
stem-like properties as a result of a catalogue of events,
including BCR-ABL amplification and B-catenin activation.
Similarly, in AML bearing the translocation t(8;21), resulting
in the AML1/ETO fusion gene, although the translocation
is found in HSC it is not sufficient to induce leukemia,
and the disease is instead driven by cells with a progenitor
phenotype.
Collectively, these data have led to a revised model which
unifies the CSC and stochastic models of tumor heterogene-
ity. According to this hypothesis, CSCs in the early stages
of tumor development are likely to follow the CSC model,
where CSCs account for a small proportion of the tumor
population, with the bulk comprising non-TIC. However,
as the disease progresses, these CSCs acquire advantageous
mutations which enhance their self-renewal ability, leading to
further expansion of the CSC pool, while impairing their dif-
ferentiation capacity. Consequently, as the disease advances,
the hierarchy becomes flatter and the tumor becomes more
homogenous, which would explain the high frequency of
CSCs reported in malignancies such as melanoma.
Pre-leukemic stem cells
The earliest evidence supporting the concept of a pre-
leukemic state came from clonality studies utilizing
X-chromosome inactivation patterns in AML patients het-
erozygous for glucose-6-phosphate dehydrogenase (G6PD)
deficiency. In each of these patients, the leukemic blasts
expressed the same G6PD type, indicating that they arose
from a single cell, and in approximately 25% of the patients,
the abnormal clone at diagnosis persisted even after attain-
ment of complete morphological remission. Similar findings
were reported in AML patients harboring the translocation
t(8;21), who were found to have detectable levels of the
fusion transcript in peripheral blood, despite being in long-
term remission. In recent years, corroborative evidence from
sequencing data has been provided by a number of groups.
Work from the Majeti lab demonstrated that in newly diag-
nosed patients with AML, mutations in epigenetic regulators
such as DNMT3a and ASXL1, were frequently found in
the purified HSC population, whereas mutations in genes
involved in activated signaling and proliferation such as FLT3
and KRAS/NRAS were not detectable, indicating that the
latter mutations were late events. Similarly, Shlush et al. per-
formed targeted sequencing of normal hematopoietic stem,
progenitor, and mature cell fractions in AML patients at diag-
nosis, and found that, in contrast to genetic aberrations such
as NPM1 mutation which was only detected in the leukemic
blasts, mutations in DNMT3a were found in both the
leukemic and non-leukemic compartments, indicating the
presence of a pre-leukemic clone ancestral to the dominant
AML clone.
Indeed, genomic studies of large cohorts of individuals
with no history of hematological malignancy and normal
blood counts have revealed that somatically acquired muta-
tions in hematopoietic cells are a frequent event in the aged
population, occurring in at least 10% of individuals above
the age of 70. Intriguingly, most of the mutations impli-
cated in this phenomenon, termed clonal hematopoiesis of

CHIP
Secondary
mutations
(e.g. DNMT3a (e.g. FLT3-ITD,
TET2, ASXLI)
Leukemia at
diagnosis Remission
Treatment
Normal HSC
Pre-leukemic HSC
LSC
Dominant clone
Minor clone
Fig. 26.1 Model of clonal evolution from clonal hematopoiesis to disease development, and the different patterns of relapse. CHIP, clonal
hematopoiesis of indeterminate potential; HSC, hematopoietic stem cells; LSC, leukemic stem cells.
indeterminate potential (CHIP), occur in epigenetic mod-
ifying genes that are frequently mutated in myeloid malig-
nancies such as DNMT3a, TET2, and ASXL1. In addition to
an increased propensity for developing leukemia, individuals
with CHIP have an increased overall mortality.
As demonstrated in several studies, pre-leukemic clones
can evade chemotherapeutic interventions and persist or
even expand in remission, and it has been shown that this
is more likely to occur among older patients, providing an
additional explanation for the poor outcome observed in this
group of patients. Additionally, persistence of pre-leukemic
mutations in remission in those who did not receive
allogeneic transplant was found to be associated with an
increased risk for relapse. In a study by Ding and colleagues,
whole-genome sequencing of paired samples from diagnosis
and relapse in patients with AML revealed two main patterns
of clonal evolution: in some patients the relapse clone
descended from the dominant clone, which acquired addi-
tional mutations, whereas in others the dominant clone was
eradicated by treatment, but relapse arose from a subclone
that was present at diagnosis. Furthermore, in the study by
Shlush et al., analysis of early and late remission samples from
a patient who was still in remission at three years revealed
a progressive increase of DNMT3a allele frequency in most
of the progenitor and mature cells fractions over time. A
small proportion of the myeloid cells in the late remission
sample was also found to harbor both DNMT3a and NPM1
mutations, which could be explained by reexpansion of the
Cancer stem cells 365
LSC Leukemia at diagnosis
Clonal
expansion
NPMI)
Dominant clone
Relapse
Minor clone
LSC
Preleukemic HSC
diagnostic clone or, alternatively, the NPM1 mutation could
have been acquired independently within a pre-leukemic
clone. Although there has so far been no solid proof of
relapse emerging from a pre-LSC, this route to relapse
remains plausible. From a clinical perspective, the persis-
tence of pre-LSCs in remission raises the question of whether
testing for minimal residual disease should include moni-
toring for pre-leukemic mutations and not only the “driver”
mutations present in leukemic blasts and LSCs, and whether
progressive increase in pre-leukemic burden during remis-
sion would warrant early treatment prior to the emergence
of frank leukemia (see Figure 26.1).
CSC-targeted therapies
Targeting cell surface antigens
Since HSCs and LSCs share the cell surface phenotype
CD34+ /CD38− , several groups have attempted to identify
markers that can accurately distinguish LSCs from HSCs, in
order to enable preferential targeting of LSCs while sparing
normal HSCs. Among the earliest identified LSC antigenic
candidates is interleukin-3 alpha receptor (CD123), which is
strongly expressed in CD34+38− cells of patients with AML,
but virtually absent in the normal HSC counterpart. The LSC
functional capacity of this population has been verified in
vivo, where they were able to engraft and induce leukemia
366 Molecular Hematology
in NOD/SCID mice. Other putative LSC markers include
CD33, CLL-1, Tim-3, and CD47, which are currently being
pursued in clinical trials.
Various therapeutic approaches aimed at targeting surface
markers on LSCs have been explored, and one strategy has
been the use of antibody-drug conjugates (ADCs), consisting
of a monoclonal antibody linked to a therapeutic moiety,
which allows selective delivery of cytotoxic drugs to neoplas-
tic cells bearing the targeted antigen, with minimal impact
on normal tissues. A promising example of this class of drugs
and the first ADC to gain approval for clinical use in cancer
is Gemtuzumab ozogamicin (GO), comprising a humanized
anti-CD33 antibody conjugated to the antitumor antibiotic
calicheamicin. After being initially withdrawn due to lack of
clinical benefit and safety concerns, GO was reapproved by
the US Food and Drug Administration (FDA) when subse-
quent clinical trials using a lower dosing schedule showed
that the addition of GO to standard chemotherapy for
treatment of patients with newly diagnosed AML resulted in
improved outcomes in terms of an event-free survival which
was nearly twice as long (median event-free survival: 17.3
months vs. 9.5 months). Improvement in obtaining complete
remission was also observed when used in the relapse
setting as well as a small, but significant, increase in overall
survival when used as a single agent in older patients who
are ineligible for intensive chemotherapy. It should be noted,
however, that CD33 is not an exclusive marker of LSC, but is
also expressed in leukemic blasts and myeloid progenitors,
and thus far it has not been shown whether the improved
outcome observed with GO relates specifically to its effect on
the LSC compartment or merely the disease bulk.
Other approaches employing the immunophenotypic
properties of LSCs include bispecific antibodies which link
the target pathogenic cell to a cytotoxic effector cell, such as
a T cell or natural killer cell, by simultaneously engaging a
tumor-specific antigen and a receptor expressed on the cyto-
toxic effector cell, thereby inducing cell death. Adoptive cell
therapy using chimeric antigen receptor (CAR)-engineered
T cells has also been pursued as an LSC-targeting avenue.
Albeit promising, targeting LSCs by utilizing their aber-
rant immunophenotype has potential limitations, which may
hinder its clinical utility. One major concern relates to the
specificity of LSC marker, as it is crucial that an antigen used
to identify LSC is not expressed on normal HSC in order
to minimize toxicity. For instance, while CD33 has been
thought not to be expressed in normal HSC, Taussig et al.
reported CD33 expression in normal Lin− CD34+ CD38−
fractions isolated from cord blood and healthy adult bone
marrow, which were able to repopulate NOD/SCID mice and
had self-renewal ability, as determined in serial transplanta-
tion assays. Indeed, CD33 has been recently reported to iden-
tify a subset of cells with serial repopulating capacity within
the most primitive HSC fraction. Furthermore, it is well
recognized that LSCs can be phenotypically heterogeneous
and therefore an LSC-targeted therapy aimed at a single anti-
gen is unlikely to be able to eliminate all LSC clones. The fact
that LSCs undergo clonal evolution as the disease progresses,
and in response to selective pressure exerted by treatment,
adds yet another layer of complexity to the picture.
Targeting CSC self-renewal
Aberrant activation of the Hedgehog (Hh) signaling path-
way has been reported in CSCs of various cancers, including
hematological malignancies. In a mouse model of CML, con-
stitutive activation of Smoothend, a key component of the Hh
pathway, resulted in an increase in CML LSC frequency lead-
ing to disease acceleration, while inhibition of Hh signaling
by cyclopamine caused reduction in CML stem cells, and this
effect was maintained even in imatinib-resistant CML-LSC.
The Wnt pathway effector β-catenin has also been impli-
cated in the regulation of LSC. In murine AML models driven
by MLL-ENL or co-expression of the oncogenes HOXA9
and MEIS1, β-catenin was required for transformation to
AML. Furthermore, functional dependence of LSC on the
canonical Wnt signaling has been demonstrated in T-cell
acute lymphoblastic leukemia (T-ALL). In B-ALL, suppres-
sion of CBP/catenin using a small molecule inhibitor abro-
gated self-renewal of pre-B ALL cells, and promoted their
differentiation, thereby sensitizing them to chemotherapy.
Similarly, LSCs in MLL-driven AML that have acquired resis-
tance to GSK3 inhibitors could be resensitized by inhibiting
β-catenin. Of note is that both Hh and β-catenin appear dis-
pensable for adult HSC function, rendering them attractive
candidates for selective inhibition of LSCs.
The canonical Notch signaling pathway is another pivotal
player in maintaining self-renewal of LSC. There has been
a particular interest in targeting Notch in T-ALL due to the
high prevalence of Notch1 mutations, which are estimated to
occur in more than 50% of patients. Treatment with a gamma
secretase inhibitor, which blocks Notch, has been shown to
abolish the leukemic-initiating stem cell activity in T-ALL
in vivo, regardless of mutational status. Other strategies to
target Notch include the use of monoclonal antibodies and
stapled peptides.
Targeting CSC survival
One of the earliest pieces of evidence implicating the nuclear
factor-kB (NF-kB) in CSC came from AML, where NF-kB
was found to be constitutively activated in the CD34+ frac-
tion of AML cells but not in normal CD34+ HSCs, indicating
that reliance on NF-kB is an exclusive feature of LSCs. Acti-
vation of NF-kB can occur through multiple routes, includ-
ing chromosomal/molecular aberrations, cytokine signaling,
or increased proteasomal activity. Inhibition of NF-kB by a

proteasome inhibitor was shown to induce apoptosis in AML
cells in both the bulk and CD34+ fraction.
Additionally, inhibitors against the PI3/AKT/mtor path-
way have been employed in AML. Constitutive activation of
phosphoinositide 3-kinase (PI3K) is necessary for the sur-
vival of AML, and reduction in AML LSC has been reported
in murine models treated with PI3K inhibitors. Intriguingly,
patients with constitutive PI3K have a favorable prognosis,
which is proposed to be due to PI3K driving LSCs into the S
phase of the cell cycle, thus making them more sensitive to
chemotherapy.
Targeting LSC dormancy
Chemoresistance is often attributed to CSCs, which due
to their largely quiescent nature can evade the effects
of chemotherapy agents that preferentially target actively
cycling cells. Breaking the dormancy of LSCs therefore repre-
sents a rational approach to eliminate the LSC population. In
murine models, AML cells residing in the endosteal region,
where quiescent, chemo-resistant LSCs preferentially home
to, have been shown to enter the cell cycle upon treatment
with the mobilizing agent granulocyte colony-stimulating
factor (GCSF). Furthermore, addition of GCSF to cytara-
bine resulted in enhanced cell death and elimination of LSCs,
as assayed by limiting dilution analysis. These findings are
supported by clinical data from newly diagnosed patients
with AML who received GCSF in conjunction with induc-
tion chemotherapy, showing improved outcomes in terms of
disease-free survival, although this benefit did not extend to
those with poor-risk disease, and no overall survival advan-
tage was noted. From a mechanistic point of view, it is unclear
whether the enhanced chemosensitivity observed with GCSF
relates solely to its effect on cell cycle progression, or whether
there are other mechanisms involved, such as displacement
of LSCs from the niche and potential synergistic interaction
with chemotherapy.
The use of arsenic trioxide, which is licensed for the treat-
ment of PML/RARa acute promyelocytic leukemia, has also
been suggested as an alternative intervention to enhance the
cycling of LSCs and, consequently, their elimination. In a
CML mouse model, loss of PML led to LSC exhaustion and
diminished repopulating ability. Furthermore, PML inhibi-
tion with arsenic trioxide in both murine and human CML-
LSC resulted in cell cycle induction which was more pro-
nounced compared to normal HSC, and was associated with
enhanced chemosensitivity.
Targeting the CSC niche
The bone marrow niche represents a complex and dynamic
microenvironment which provides a host of signaling cues
that are crucial for determining the fate decision of normal
Cancer stem cells 367
HSCs. As with HSCs, niche-derived signals are vital for main-
taining LSCs (see Figure 26.2). Evidence from several murine
models even indicates that alterations in components of the
bone marrow microenvironment may be sufficient to disrupt
normal hematopoiesis and induce neoplastic transformation.
In an elegant study by the Scadden group, selective deletion
of the miRNA processing endonuclease Dicer 1 in osteopro-
genitors, resulting in impaired obsteoblastic differentiation,
induced profound hematopoietic defects mimicking
myelodysplasia, with subsequent transformation to myeloid
sarcoma in some mice. Intriguingly, the same phenotype
was observed when wild-type hematopoietic cells were
transplanted into mice with mutant niche, while the effect
was reversed upon transplantation of hematopoietic cells
from mutant mice into a wild-type environment. This lends
further support to the concept of niche-driven oncogenesis,
whereby microenvironmental aberrations constitute the
primary pathogenic event that drives tumorigenesis, which
consequently facilitates the acquisition of oncogenic hits in
the hematopoietic compartment. Similarly, constitutive acti-
vation of β-catenin in murine osteoblasts has been shown to
alter the differentiation potential of lymphoid and myeloid
progenitors, leading to a differentiation block and the emer-
gence of AML through upregulation of the Notch-ligand
Jagged 1 in osteoblasts and subsequent increase of Notch
signaling in leukemia-initiating HSCs. However, it remains
to be determined whether this mechanism of niche-driven
oncogenesis occurs in human hematological malignancies.
The relationship between the niche and leukemic cells
is clearly not a one-way street and, just as the niche can
dictate tumor behavior, so too can leukemic cells reshape
and manipulate their microenvironment in their favor at the
expense of normal hematopoiesis. For example, in a mouse
model of BCR-ABL CML, reduced expression of CXCL12
was observed in the bone marrow stroma cells secondary
to secreted factors by leukemic cells, leading to reduced
engraftment of normal HSCs, but not of BCR-ABL-positive
HSCs. Similarly, in a murine model of myeloproliferative
neoplasm (MPN), neoplastic myeloid cells were shown to
remodel the endosteal niche by promoting the differentiation
of mesenchymal stroma cells toward the osteoblastic lineage.
Notably, the expanded osteoblasts were functionally altered,
with compromised ability to support normal HSCs but not
LSCs. AML-induced alterations in endothelial cells resulting
in increased vascular permeability and sympathetic neuropa-
thy of the niche in MPN are among other reported examples
of neoplasia-induced remodeling of the bone marrow niche.
Given the important role of the niche in sustaining tumor
development, targeting the niche has been explored as a
potential therapeutic strategy. One approach has been to dis-
rupt homing of LSCs by targeting the CXCL12/CXC4 axis in
leukemia. Pretreatment of AML cells with CXCR4 antibodies
has been shown to impair their homing into the bone marrow
368 Molecular Hematology

mesenchymal
stroma cells
Sympathetic neurons

Osteoclast

?
? Endothelial
cells
Bone

Osteoblast
? Blood vessel

LSC ?
?

Endosteal niche Vascular niche

Adipocvtes
Fig. 26.2 Schematic representation of the interplay between the niche and leukemic cells. The bone marrow (BM) niche consists of
several cellular components including mesenchymal stroma cells, osteoblasts, endothelial cells, adipocytes, and sympathetic neurons. The
leukemia/microenvironment interaction is a dynamic, bidirectional process whereby leukemic cells receive vital cues from the niche that influence
their behavior. Conversely, leukemic cells hijack and remodel the BM niche into a malignant environment that is more permissive to the disease at
the expense of normal hematopoiesis. LSC, leukemic stem cells.

and spleen of transplanted mice. Conversely, mice already
engrafted with AML and treated with CXCR4-neutralizing
antibodies had marked reduction in their AML burden, but
no significant difference in the level of engraftment was
observed in mice engrafted with cord blood mononuclear
cells, indicating that AML cells are more CXCR4 dependent.
Moreover, using a different small molecule CXCL12/CXCR4
inhibitor, CXCR4 inhibition abrogated the protection con-
ferred by the stroma, resulting in increased sensitivity to
chemotherapy. Of note is that higher expression of CXCR4
has been reported in patients with AML, and CXCR4 expres-
sion level appeared to be an independent prognostic indi-
cator across a heterogeneous group of AML patients with
varying mutational backgrounds. Given the role of CXCR4 in
homing, it is not unreasonable to speculate that the negative
impact on outcome may, at least in part, be due to enhanced
CXCR4-mediated retention of LSCs.
Disrupting the LSC–niche interaction can also be achieved
by targeting the integrin very late antigen 4 (VLA4), which
is known for its important role in homing and retention
of HSCs in the niche. VLA-4 binds to the fibronectin
component of the extracellular matrix as well as the vascular
cell adhesion molecular-1 (VCAM-1), expressed by bone
marrow stromal and endothelial cells, and inhibiting
the VLA-4/VCAM-1 interaction induces mobilization of
hematopoietic/stem progenitor cells. In AML, interaction
between VLA-4 on leukemic cells and fibronectin has been
shown to enhance their survival and reduce their chemosen-
sitivity through activation of PI-3k/AKT/BCL2 signaling.
Combination therapy with the anti-VLA4 monoclonal
antibody Natalizumab (NZM) and cytarabine resulted in
increased AML apoptosis, and mice treated with single-agent
NZM had a significant reduction in their AML burden.
Alternative therapeutic approaches include targeting the
adhesion molecule CD44. A monoclonal antibody directed
against CD44 resulted in significant reduction of leukemic
burden in mice transplanted with AML. This effect was medi-
ated not only by impeding LSC homing, but also by altering
the fate of LSCs. A major drawback, however, is that CD44
is also expressed on normal HSCs, and ligation of CD44 has
been shown to affect the migratory behavior of normal HSCs.
Targeting epigenetic modifiers
The use of epigenetic modulating therapies as a means to
target CSCs in hematological malignancies has been mainly

explored in AML, in which mutations in epigenetic modifiers
are a frequent event, particularly in MLL-rearranged AML,
which is primarily driven by epigenetic dysregulation.
The DNA methyltransferase inhibitors, azacytidine and
decitabine, are currently the only two approved epigenetic
targeted therapies in AML. The effect of azacitidine on AML
LSCs has been investigated by Craddock and colleagues, who
found a substantial reduction in the LSC pool size, although
it was never eradicated, even in patients who achieved com-
plete morphological remission.
Histone methyltransferase inhibitors are another class of
epigenetic therapies that have shown demonstrable effect
against LSCs. Key among these is 3-Deazaneplanocin A
(DZNep), which targets several components of the Polycomb
repressor complex 2, including EZH2. In a study by Zhou and
colleagues, DZNep demonstrated significant anti-leukemic
effect, including in LSC-enriched populations, with no
observed impact on normal HSCs in colony-forming assays.
However, more recently, loss of EZH2 has been reported
to correlate with poor prognosis and chemoresistance in
AML, and treatment of AML cells with DZNeP was found
to induce resistance to cytarabine. In contrast to AML,
overexpression of EZH2 has been reported in CML LSCs,
and inactivation of EZH2, in a conditional mouse model and
by clustered regularly interspaced short palindromic repeats
(CRISPR)/Cas9-mediated gene editing, has been shown to
block disease initiation and impair CML LSC function. Sim-
ilar findings were reported by Scott et al., who showed a syn-
ergistic impact on the CML LSC compartment when EZH2
inhibitors were used in conjunction with tyrosine kinase
inhibitors.
In addition to EZH2, the histone 3 lysine 79 (H3K79)
methyltransferase Dot1l has emerged as a critical regulator
of LSC activity. In MLL-AF9 induced leukemia, Dot1l has
been shown to play an important role in promoting LSC
self-renewal, and DOT1l inhibitors are currently being
investigated in a number of clinical trials. Other histone
modifiers with the potential to modulate LSC activity in AML
include the histone demethylases KDM2b/JHDM1b and
LSD1/KMD1A.
Furthermore, anti-leukemic effect against AML LSC has
been shown with the histone deacetylase inhibitor AR-42,
which causes inactivation of NF-κB, as well as inhibitors
against the chromatin reader Bromodomain-containing pro-
tein 4 (BRD4).
Other CSC-targeted therapies
LSCs can be distinguished from normal HSCs by virtue of
their distinct metabolic traits. In a study of chronic-phase
CML, the stem cell–enriched population exhibited an
increase in oxidative metabolism compared to normal HSCs,
and was more susceptible to oxidative phosphorylation
Cancer stem cells 369
inhibition when treated with combination of a tyrosine
kinase inhibitor and tigecycline, an antibiotic that inhibits
mitochondrial protein translation. Reliance on oxida-
tive phosphorylation rather than glycolysis has also been
reported in AML, where LSCs have been shown to have lower
levels of endogenous reactive oxygen species (ROS). Fur-
thermore, BCL-2 was upregulated in ROS-low LSCs, which
when inhibited resulted in elimination of LSC both in vitro
and in vivo. Similarly, in AML with isocitrate dehydrogenase
(IDH) mutations, an RNA interference screen revealed syn-
thetic lethality between IDH1 and BCL-2. Treatment with
BCL-2 inhibitor induced apoptosis in AML cells and affected
the survival of LSCs in xenotransplants through a mecha-
nism involving 2-hydroxybutarate-mediated inhibition of
cytochrome c oxidase activity in the mitochondrial electron
transport chain.
The role of hypoxia in maintaining normal HSC func-
tions, especially self-renewal, is well recognized. Similarly,
hypoxia is important for the maintenance of LSCs. Wang
et al. reported that in patients with AML, HIF1a and its
target GLUT1 were selectively overexpressed in the LSC
fraction. Furthermore, inhibition of HIF-1α by echinomycin
induced apoptosis, which was more pronounced in the LSC
fraction compared to the AML bulk population. The selec-
tive elimination of the LSC population by HIF inhibition was
also demonstrated in vivo by serial transplantation exper-
iments, where mice injected with residual leukemic cells
from echinomycin-treated mice failed to develop leukemia,
indicating an absence of LSCs in the residual leukemic
population. Similar effects of HIF-1α inhibition have been
reported in CML LSCs. It is worthy of note, however, that the
inhibition of HIF by echinomycin is not restricted to HIF-1α,
and that the effect of echinomycin may, at least in part, be
mediated by suppression of HIF-2α, which has been demon-
strated to protect both normal HSC and AML cells from
endoplasmic stress-induced apoptosis.
Conclusion
The mainstay treatment in patients with hematological
malignancies remains traditional therapies that target the
tumor bulk but leave CSCs largely untouched, potentially
contributing to chemoresistance and disease relapse. In
recent years, tremendous advances have been made in our
understanding of CSC biology, which paved the way for the
development of a number of therapeutic agents that tar-
get CSCs. While promising, several considerations should
be taken into account before CSC-targeted therapies can be
incorporated into routine clinical practice.
The timing of administration of a CSC-directed therapy is
likely to influence its efficacy. Ideally, a CSC-targeted therapy
should be given before standard chemotherapy when CSCs
370 Molecular Hematology
are still in a naı̈ve state, as the selective pressure exerted by
chemotherapy can inadvertently lead to the acquisition of
further mutations in CSCs, making them more complex
and potentially drug resistant. However, this may not be
pragmatic in highly proliferative malignancies such as AML,
and the concurrent use of CSC-targeted therapies and “de-
bulking” chemotherapy would therefore be more appropri-
ate. It has also been proposed to use CSC-targeted treatment
in the post-remission setting, either as part of the consolida-
tion regimen or as long-term maintenance strategy in order
to eradicate residual CSCs.
Additionally, caution should be exercised in applying
the same clinical endpoints used to measure the efficacy of
standard cancer therapies, as they could potentially result in
significant underestimates of ultimate clinical benefit, lead-
ing to erroneous conclusions about their efficacy. To make an
accurate assessment of response rates, long follow-up periods
are required; however, in the short term, clinical endpoints
such as progression-free and overall survival may be more
meaningful.
Validation of CSC-targeted therapies in the in vivo setting
is also paramount, as stemness is a functional definition. This
would require having reliable methods to isolate CSCs, which
pose an additional challenge, as CSCs can be phenotypically
unstable. Finally, a rational selection of combination thera-
pies is required in order to maximize efficacy and reduce the
emergence of resistance.
Further reading
The cancer stem cell concept
Ailles, L.E. and Weissman, I.L. (2007). Cancer stem cells in solid tumors.
Curr. Opin. Biotechnol. 18: 460–466.
Al-Hajj, M., Wicha, M.S., Benito-Hernandez, A. et al. (2003). Prospec-
tive identification of tumorigenic breast cancer cells. Proc. Natl. Acad.
Sci. U.S.A. 100: 3983–3988.
Baccelli, I. and Trumpp, A. (2012). The evolving concept of cancer and
metastasis stem cells. J. Cell Biol. 198: 281–293.
Bonnet, D. and Dick, J.E. (1997). Human acute myeloid leukemia is
organized as a hierarchy that originates from a primitive hematopoi-
etic cell. Nat. Med. 3: 730–737.
Maehle, A.H. (2011). Ambiguous cells: the emergence of the stem cell
concept in the nineteenth and twentieth centuries. Notes Rec. R. Soc.
65: 359–378.
Marjanovic, N.D., Weinberg, R.A., and Chaffer, C.L. (2013). Cell plas-
ticity and heterogeneity in cancer. Clin. Chem. 59: 168–179.
Quintana, E., Shackleton, M., Sabel, M.S. et al. (2008). Efficient tumour
formation by single human melanoma cells. Nature 456: 593–598.
Schatton, T., Murphy, G.F., Frank, N.Y. et al. (2008). Identification of
cells initiating human melanomas. Nature 451: 345–349.
Shackleton, M., Quintana, E., Fearon, E.R., and Morrison, S.J. (2009).
Heterogeneity in cancer: cancer stem cells versus clonal evolution.
Cell 138: 822–829.
Singh, S.K., Hawkins, C., Clarke, I.D. et al. (2004). Identification of
human brain tumour initiating cells. Nature 432: 396–401.
Visvader, J.E. and Lindeman, G.J. (2008). Cancer stem cells in solid
tumours: accumulating evidence and unresolved questions. Nat. Rev.
Cancer 8: 755–768.
The cell of origin
Cozzio, A., Passegué, E., Ayton, P.M. et al. (2003). Similar MLL-
associated leukemias arising from self-renewing stem cells and short-
lived myeloid progenitors. Genes Dev. 17: 3029–3035.
Goardon, N., Marchi, E., Atzberger, A. et al. (2011). Coexistence of
LMPP-like and GMP-like leukemia stem cells in acute myeloid
leukemia. Cancer Cell 19: 138–152.
Jamieson, C.H.M., Ailles, L.E., Dylla, S.J. et al. (2004). Granulocyte–
macrophage progenitors as candidate leukemic stem cells in blast-
crisis CML. N. Engl. J. Med. 351: 657–667.
Kreso, A. and Dick, J.E. (2014). Evolution of the cancer stem cell model.
Cell Stem Cell 14: 275–291.
Krivtsov, A.V., Twomey, D., Feng, Z. et al. (2006). Transformation from
committed progenitor to leukaemia stem cell initiated by MLL-AF9.
Nature 442: 818–822.
Miyamoto, T., Weissman, I.L., and Akashi, K. (2000). AML1/ETO-
expressing nonleukemic stem cells in acute myelogenous leukemia
with 8;21 chromosomal translocation. Proc. Natl. Acad. Sci. U.S.A.
97: 7521–7526.
Taussig, D.C., Vargaftig, J., Miraki-Moud, F. et al. (2010). Leukemia-
initiating cells from some acute myeloid leukemia patients with
mutated nucleophosmin reside in the CD34- fraction. Blood 115:
1976–1984.
Pre-leukemic stem cells
Corces-Zimmerman, M.R., Hong, W.-J., Weissman, I.L. et al. (2014).
Preleukemic mutations in human acute myeloid leukemia affect epi-
genetic regulators and persist in remission. Proc. Natl. Acad. Sci.
U.S.A. 111: 2548–2553.
Ding, L., Ley, T.J., Larson, D.E. et al. (2012). Clonal evolution in relapsed
acute myeloid leukaemia revealed by whole-genome sequencing.
Nature 481: 506–510.
Fialkow, P.J., Janssen, J.W., and Bartram, C.R. (1991). Clonal remissions
in acute nonlymphocytic leukemia: evidence for a multistep patho-
genesis of the malignancy. Blood 77: 1415–1417.
Fialkow, P.J., Singer, J.W., Raskind, W.H. et al. (1987). Clonal develop-
ment, stem-cell differentiation, and clinical remissions in acute non-
lymphocytic leukemia. N. Engl. J. Med. 317: 468–473.
Genovese, G., Kähler, A.K., Handsaker, R.E. et al. (2014). Clonal
hematopoiesis and blood-cancer risk inferred from blood DNA
sequence. N. Engl. J. Med. 371: 2477–2487.
Jaiswal, S., Fontanillas, P., Flannick, J. et al. (2014). Age-related clonal
hematopoiesis associated with adverse outcomes. N. Engl. J. Med. 371:
2488–2498.
Jan, M., Snyder, T.M., Corces-Zimmerman, M.R. et al. (2012). Clonal
evolution of preleukemic hematopoietic stem cells precedes human
acute myeloid leukemia. Sci. Transl. Med. 4: 149ra118.
Miyamoto, T., Nagafuji, K., Akashi, K. et al. (1996). Persistence of mul-
tipotent progenitors expressing AML1/ETO transcripts in long-term

remission patients with t(8;21) acute myelogenous leukemia. Blood
87: 4789–4796.
Rothenberg-Thurley, M., Amler, S., Goerlich, D. et al. (2018). Persis-
tence of pre-leukemic clones during first remission and risk of relapse
in acute myeloid leukemia. Leukemia 32: 1–11.
Shlush, L.I., Zandi, S., Mitchell, A. et al. (2014). Identification of pre-
leukaemic haematopoietic stem cells in acute leukaemia. Nature 506:
328–333.
Wong, T.N., Miller, C.A., Klco, J.M. et al. (2016). Rapid expansion of pre-
existing nonleukemic hematopoietic clones frequently follows induc-
tion therapy for de novo AML. Blood 127: 893–897.
Xie, M., Lu, C., Wang, J. et al. (2014). Age-related mutations associated
with clonal hematopoietic expansion and malignancies. Nat. Med. 20:
1472–1478.
Targeting cell surface antigens
Hoseini, S.S. and Cheung, N.K. (2017). Acute myeloid leukemia targets
for bispecific antibodies. Blood Cancer J. 7: e522.
Jen, E.Y., Ko, C.-W., Lee, J.E. et al. (2018). FDA approval: gemtuzumab
ozogamicin for the treatment of adults with newly-diagnosed CD33-
positive acute myeloid leukemia. Clin. Cancer Res. 24 (14): 3242–
3246.
Jin, L., Lee, E.M., Ramshaw, H.S. et al. (2009). Monoclonal antibody-
mediated targeting of CD123, IL-3 receptor α chain, eliminates
human acute myeloid leukemic stem cells. Cell Stem Cell 5:
31–42.
Kikushige, Y., Shima, T., Takayanagi, S.I. et al. (2010). TIM-3 is a promis-
ing target to selectively kill acute myeloid leukemia stem cells. Cell
Stem Cell 7: 708–717.
Knapp, D.J.H.F., Hammond, C.A., Hui, T. et al. (2018). Single-cell anal-
ysis identifies a CD33+ subset of human cord blood cells with high
regenerative potential. Nat. Cell Biol. 20: 710–720.
Liu, K., Zhu, M., Huang, Y. et al. (2015). CD123 and its potential clinical
application in leukemias. Life Sci. 122: 59–64.
Majeti, R., Chao, M.P., Alizadeh, A.A. et al. (2009). CD47 is an adverse
prognostic factor and therapeutic antibody target on human acute
myeloid leukemia stem cells. Cell 138: 286–299.
Mardiros, A., Forman, S.J., and Budde, L.E. (2015). T cells expressing
CD123 chimeric antigen receptors for treatment of acute myeloid
leukemia. Curr. Opin. Hematol. 22: 484–488.
Taussig, D.C., Pearce, D.J., Simpson, C. et al. (2005). Hematopoietic stem
cells express multiple myeloid markers: implications for the origin
and targeted therapy of acute myeloid leukemia. Blood 106: 4086–
4092.
Van Rhenen, A., Van Dongen, G.A.M.S., Kelder, A.L. et al. (2007). The
novel AML stem cell-associated antigen CLL-1 aids in discrimination
between normal and leukemic stem cells. Blood 110: 2659–2666.
Targeting CSC self-renewal
Dierks, C., Beigi, R., Guo, G.-R. et al. (2008). Expansion of Bcr-Abl-
positive leukemic stem cells is dependent on hedgehog pathway acti-
vation. Cancer Cell 14: 238–249.
Gang, E.J., Hsieh, Y.T., Pham, J. et al. (2014). Small-molecule inhibition
of CBP/catenin interactions eliminates drug-resistant clones in acute
lymphoblastic leukemia. Oncogene 33: 2169–2178.
Cancer stem cells 371
Giambra, V., Jenkins, C.E., Lam, S.H. et al. (2015). Leukemia stem cells
in T-ALL require active Hif1α and Wnt signaling. Blood 125: 3917–
3927.
Tatarek, J., Cullion, K., Ashworth, T. et al. (2011). Notch1 inhibition
targets the leukemia-initiating cells in a Tal1/Lmo2 mouse model of
T-ALL. Blood 118: 1579–1590.
Wang, Y., Krivtsov, A.V., Sinha, A.U. et al. (2010). The Wnt/beta-catenin
pathway is required for the development of leukemia stem cells in
AML. Science (New York, NY) 327: 1650–1653.
Yeung, J., Esposito, M.T., Gandillet, A. et al. (2010). β-Catenin mediates
the establishment and drug resistance of MLL leukemic stem cells.
Cancer Cell 18: 606–618.
Targeting CSC survival
Guzman, M.L., Neering, S.J., Upchurch, D. et al. (2001). Nuclear factor-
kappaB is constitutively activated in primitive human acute myeloge-
nous leukemia cells. Blood 98: 2301–2307.
Park, S., Chapuis, N., Tamburini, J. et al. (2010). Role of the PI3K/AKT
and mTOR signaling pathways in acute myeloid leukemia. Haemato-
logica 95: 819–828.
Targeting LSC dormancy
Ito, K., Bernardi, R., Morotti, A. et al. (2008). PML targeting eradicates
quiescent leukaemia-initiating cells. Nature 453: 1072–1078.
Löwenberg, B., van Putten, W., Theobald, M. et al. (2003). Effect of
priming with granulocyte colony-stimulating factor on the outcome
of chemotherapy for acute myeloid leukemia. N. Engl. J. Med. 349:
743–752.
Saito, Y., Uchida, N., Tanaka, S. et al. (2010). Induction of cell cycle entry
eliminates human leukemia stem cells in a mouse model of AML. Nat.
Biotechnol. 28: 275–280.
Targeting the CSC niche
Arranz, L., Sánchez-Aguilera, A., Martı́n-Pérez, D. et al. (2014). Neu-
ropathy of haematopoietic stem cell niche is essential for myelopro-
liferative neoplasms. Nature 512: 78–81.
Hsieh, Y., Jiang, E., Pham, J., et al. (2013). VLA4 blockade in acute
myeloid leukemia. In 55th ASH Annual Meeting and Exposition,
122, 7–13 December 2013. New Orleans, LA: American Society of
Hematology.
Jin, L., Hope, K.J., Zhai, Q. et al. (2006). Targeting of CD44 eradi-
cates human acute myeloid leukemic stem cells. Nat. Med. 12: 1167–
1174.
Kode, A., Manavalan, J.S., Mosialou, I. et al. (2014). Leukaemogenesis
induced by an activating β-catenin mutation in osteoblasts. Nature
506: 240–244.
Matsunaga, T., Takemoto, N., Sato, T. et al. (2003). Interaction between
leukemic-cell VLA-4 and stromal fibronectin is a decisive factor for
minimal residual disease of acute myelogenous leukemia. Nat. Med.
9: 1158–1165.
Passaro, D., Di Tullio, A., Abarrategi, A. et al. (2017). Increased vascular
permeability in the bone marrow microenvironment contributes to
disease progression and drug response in acute myeloid leukemia.
Cancer Cell 32: 324–341.e6.
372 Molecular Hematology
Raaijmakers, M.H.G.P., Mukherjee, S., Guo, S. et al. (2010). Bone
progenitor dysfunction induces myelodysplasia and secondary
leukaemia. Nature 464: 852–857.
Schepers, K., Pietras, E.M., Reynaud, D. et al. (2013). Myeloprolifera-
tive neoplasia remodels the endosteal bone marrow niche into a self-
reinforcing leukemic niche. Cell Stem Cell 13: 285–299.
Spoo, A.C., Lübbert, M., Wierda, W.G., and Burger, J.A. (2007). CXCR4
is a prognostic marker in acute myelogenous leukemia. Blood 109:
786–791.
Tavor, S., Petit, I., Porozov, S. et al. (2004). CXCR4 regulates migration
and development of human acute myelogenous leukemia stem cells
in transplanted NOD/SCID mice. Cancer Res. 64: 2817–2824.
Zeng, Z., Shi, Y.X., Samudio, I.J. et al. (2009). Targeting the leukemia
microenvironment by CXCR4 inhibition overcomes resistance to
kinase inhibitors and chemotherapy in AML. Blood 113: 6215–6224.
Zhang, B., Ho, Y.W., Huang, Q. et al. (2012). Altered microenvironmen-
tal regulation of leukemic and normal stem cells in chronic myeloge-
nous leukemia. Cancer Cell 21: 577–592.
Zöller, M. (2015). CD44, hyaluronan, the hematopoietic stem cell, and
leukemia-initiating cells. Front. Immunol. 6: 235.
Targeting epigenetic modifiers
Craddock, C., Quek, L., Goardon, N. et al. (2013). Azacitidine fails to
eradicate leukemic stem/progenitor cell populations in patients with
acute myeloid leukemia and myelodysplasia. Leukemia 27: 1028–
1036.
Göllner, S., Oellerich, T., Agrawal-Singh, S. et al. (2017). Loss of the his-
tone methyltransferase EZH2 induces resistance to multiple drugs in
acute myeloid leukemia. Nat. Med. 23: 69–78.
Guzman, M.L., Yang, N., Sharma, K.K. et al. (2014). Selective activity of
the histone deacetylase inhibitor AR-42 against leukemia stem cells: a
novel potential strategy in acute myelogenous leukemia. Mol. Cancer
Ther. 13: 1979–1990.
Harris, W.J., Huang, X., Lynch, J.T. et al. (2012). The histone demethylase
KDM1A sustains the oncogenic potential of MLL-AF9 leukemia stem
cells. Cancer Cell 21: 473–487.
He, J., Nguyen, A.T., and Zhang, Y. (2011). KDM2b/JHDM1b, an
H3K36me2-specific demethylase, is required for initiation and main-
tenance of acute myeloid leukemia. Blood 117: 3869–3880.
Herrmann, H., Blatt, K., Shi, J. et al. (2013). Small-molecule inhibition of
BRD4 as a new potent approach to eliminate leukemic stem- and pro-
genitor cells in acute myeloid leukemia (AML). Oncotarget 3: 1588–
1599.
Nguyen, A.T., Taranova, O., He, J., and Zhang, Y. (2011). DOT1L,
the H3K79 methyltransferase, is required for MLL-AF9 - mediated
leukemogenesis. Blood 117: 6912–6922.
Scott, M.T., Korfi, K., Saffrey, P. et al. (2016). Epigenetic reprogramming
sensitizes CML stem cells to combined EZH2 and tyrosine kinase
inhibition. Cancer Discov. 6: 1248–1257.
Xie, H., Peng, C., Huang, J. et al. (2016). Chronic myelogenous
leukemia-initiating cells require polycomb group protein EZH2.
Cancer Discov. 6: 1237–1247.
Zhou, J., Bi, C., Cheong, L.L. et al. (2011). The histone methyltransferase
inhibitor, DZNep, up-regulates TXNIP, increases ROS production,
and targets leukemia cells in AML. Blood 118: 2830–2839.
Other CSC-targeted therapies
Chan, S.M., Thomas, D., Corces-Zimmerman, M.R. et al. (2015). Isoci-
trate dehydrogenase 1 and 2 mutations induce BCL-2 dependence in
acute myeloid leukemia. Nat. Med. 21: 178–184.
Kuntz, E.M., Baquero, P., Michie, A.M. et al. (2017). Targeting mito-
chondrial oxidative phosphorylation eradicates therapy-resistant
chronic myeloid leukemia stem cells. Nat. Med. 23: 1234–1240.
Lagadinou, E.D., Sach, A., Callahan, K. et al. (2013). BCL-2 inhibition
targets oxidative phosphorylation and selectively eradicates quiescent
human leukemia stem cells. Cell Stem Cell 12: 329–341.
Rouault-Pierre, K., Lopez-Onieva, L., Foster, K. et al. (2013). HIF-2a
protects human hematopoietic stem/progenitors and acute myeloid
leukemic cells from apoptosis induced by endoplasmic reticulum
stress. Cell Stem Cell: 549–563.
Wang, Y., Liu, Y., Malek, S.N. et al. (2011). Targeting HIF1α eliminates
cancer stem cells in hematological malignancies. Cell Stem Cell 8:
399–411.
Conclusion
Pollyea, D.A. and Jordan, C.T. (2017). Therapeutic targeting of acute
myeloid leukemia stem cells. Blood 129: 1627–1635.
Chapter 27 Molecular basis of transplantation
Francesco Dazzi1,2 & Antonio Galleu1,2
1 School of Cancer & Pharmaceutical Sciences, King’s College London, London, UK
2
King’s Health Partners Cancer Research UK Centre, London, UK
Introduction, 373
Principles and clinical indications of HSCT, 373
Non-immunological factors regulating HSCT, 374
Exploiting HSCT for resetting the immune system, 375
Basic concepts in the immunology of allogeneic HSCT, 376
Introduction
Transplantation is a successful therapeutic modality for a
variety of diseases of different etiology and pathogenesis.
Liver, heart, and kidney failures are common indications
which would be even more widely pursued should the donor
availability not be so limited. Although the basic general
principle underlying transplantation is the replacement of
a malfunctioning tissue with a healthy one, in the case of
hemopoietic stem cell transplantation (HSCT) the proce-
dure is associated with a number of other beneficial effects
which can be exploited not only to restore hemopoietic
failure itself, but also for treating cancer and autoimmune
diseases.
Principles and clinical indications of
HSCT
Hematopoietic stem cells (HSCs) have the capacity to self-
renew and give rise to all formed elements in the blood.
Because of this property, they can be used to rescue the
hemopoietic system from the intensification of anti-cancer
cytotoxic therapies. HSCs can be autologous when donor and
recipient are the same individual, or allogeneic if another
individual is selected as the HSC donor. In autologous
HSCT, patients can be subjected to lethal doses of chemo-
radiotherapy to eradicate the tumor and then receive their
Molecular Hematology, 4th Edition. Edited by Drew Provan and John G. Gribben.
© 2020 John Wiley & Sons Ltd. Published 2020 by John Wiley & Sons Ltd.
Graft-versus-host and graft-versus-Leukemia, 379
Graft-versus-host disease, 381
Conclusions, 386
Further reading, 387
own HSCs, harvested before, to restore the otherwise per-
manently ablated hemopoietic system. Although autologous
HSCT is useful in some solid tumors, its efficacy in the treat-
ment of hemopoietic malignancies is limited by the contam-
ination of the harvested stem cells by the original tumor
and/or by the insufficient activity of the chemotherapy in
eliminating the tumor itself. However, such an approach
maintains some efficacy, because normal HSCs have a tempo-
rary growth advantage at repopulating the recipient as com-
pared to neoplastic stem cells.
A more recent application of autologous HSCT has been
the treatment of severe autoimmune diseases, whereby the
repopulation of the immune system with primitive HSCs
is believed to re-educate the ill immune system. Phase I/II
trials have reported high response rates in systemic lupus
erythematosus, systemic sclerosis, rheumatoid arthritis, and
multiple sclerosis. Randomized studies are ongoing to com-
pare these achievements to conventional immunosuppressive
therapies.
When a compatible donor is available, the use of allo-
geneic HSCT has profoundly modified the outcome of several
hematological malignancies. The conditioning regimen con-
tributes to the eradication of the abnormal cells and ensures
sustained engraftment of the healthy allogeneic stem cells.
However, the efficacy of this approach cannot simply be
ascribed to the chemo-radiotherapy nor to the administra-
tion of healthy HSCs, but is greatly dependent on the immune
recognition of the tumor by the lymphocytes contained in the
donor stem cell preparation.
From these preliminary considerations, it is clear how sev-
eral mechanisms contribute to the outcome of HSCT and
these involve HSC engraftment, expansion, and differentia-
tion as well as, in the case of allogeneic HSCT, the interplay
between donor and recipient immune responses.
373
374 Molecular Hematology
Non-immunological factors regulating
HSCT
HSCs normally face four different outcomes: self-renewal,
differentiation, mobilization, or programmed cell death. All
these processes are tightly regulated and play a fundamen-
tal role during transplantation. Since HSCs are the source
of every blood cell, their self-renewal and lifespan must be
securely controlled. For this reason, a proportion of HSCs
remains quiescent, but available to contribute to blood home-
ostatic renewal. These HSCs reside in specialized areas of the
bone marrow (BM) in which they receive signals required for
their homeostatic quiescent and stationary state. These HSC
niches are constituted by endosteal bone-lining osteoblasts
and other stromal cells which are responsible for generating
the signals. The niche saves stem cells from depletion, while
protecting the host from unnecessary stem cell proliferation.
The hemopoietic stem cell
Probably the best definition of HSC is functional and refers to
a cell capable of providing lifelong reconstitution of all blood
cell lineages after serial transplantation into lethally irradi-
ated recipients.
In mice, the phenotype of HSCs is characterized by the
surface expression of Sca-1 (stem cell antigen-1), c-Kit
(receptor for stem cell factor), with associated low levels
of Thy-1 (CD90) expression, and the absence of other
lineage-specific markers: Lin− Thy1.1+lo c-Kit+ Sca1+ . The
biological activity of the candidate mouse HSC population
has been enumerated by in vitro assays and in vivo competi-
tive repopulation assays. These functional assays have shown
that extensive self-renewal potential and proliferation ability
are gradually lost during commitment and have defined a
hierarchy in the hematopoiesis. Long-term hematopoietic
stem cells (LT-HSCs) give rise to short-term hematopoietic
stem cells (ST-HSCs) that self-renew for six to eight weeks
only, then progress to multipotent progenitors (MPPs),
which have very limited self-renewal ability and are unable
to durably engraft in lethally irradiated mice.
In humans, HSCs have been characterized by the expres-
sion of the cell surface antigens CD34, c-Kit, Thy-1 (CD90),
and the absence of CD38 and other lineage-specific markers,
and are therefore named Lin−c-Kit+Thy-1+CD38−CD34+.
The phenotype is the results of functional studies conducted
not only in vitro, but also using strains of immunodeficient
mice suitable for human HSC engraftment.
Further studies have identified the heterogeneity of the
HSC compartment and identified new markers, like CD133,
which is unique to a rare CD34– subset with LT-HSC func-
tion in vivo, and can similarly identify stem cells of other tis-
sue origin. Very recently a new classification of HSCs and
early progenitors has been proposed based on the selec-
tive expression of signaling lymphocyte activation molecule
(SLAM) family receptors. The highly enriched bone marrow
HSC population is marked by the expression of CD150 and
the absence of CD224 and CD48, whereas multipotent pro-
genitors and restricted B-cell progenitors selectively express
only CD224 and CD48, respectively. A similar expression
pattern of SLAM receptors has been described in fetal
hematopoiesis.
The number of HSCs infused after the conditioning reg-
imen remains a major prognostic factor to predict the
outcome of clinical HSCT. The larger the number of
HSCs administered, the faster the recovery of recipient
hemopoiesis and the lower the transplant-related mortality.
Therefore, much effort is being spent with a view to expand-
ing the number of HSCs ex vivo with a variety of cytokine
cocktails, although very limited success has been achieved
thus far, because expansion is almost invariably associated
with differentiation. However, an aryl hydrocarbon receptor
antagonist (StemRegenin-1, SR-1) that expands CD34+ cells
has been shown to produce a 330-fold increase, with a dra-
matic impact on patient engraftment.
The hemopoietic niche
In the postnatal marrow, stromal cells form a three-
dimensional network investing marrow sinusoids. Four main
cell types of the marrow stromal tissue are known to take
part in supporting hemopoiesis: one is of hemopoietic origin
(macrophages), whereas the others derive from mesenchy-
mal progenitors (reticular cells, adipocytes, and osteoblasts).
Among the various players, osteoblasts expressing N-
cadherin are the most important component of the niche.
Osteoblasts control of HSC proliferation within the niche
is regulated by both cell contact mechanisms and soluble
factors, which mediate their effects via transcription factors,
cell cycle regulators, adhesion molecules, and chromoso-
mal modifiers. These include a series of molecules which
regulate the quiescent state of HSCs and their population
size, including Notch1/Jagged1, the Wnt/β-catenin signaling
pathways, Tie-2/Ang-1, osteopontin, and tenascin C. Ani-
mals deficient in these molecules have a reduced number
and function of HSCs. A wide variety of cytokines produced
by stromal cells has also been described, they maintain HSCs
in quiescence or promote their self-renewal rather than
differentiation. Among these are stem cell factor, leukemia
inhibitory factor, the SDF-1/CXCL12 chemokine axis, bone
morphogenetic protein-4 (BMP-4), the cytokine fms-like
tyrosine kinase-3 (Flt-3), and transforming growth factor-β
(TGF-β). Stroma also produces a variety of interleukins
and, under certain circumstances, also cytokines, which act
on more mature hemopoietic progenitors, such as gran-
ulocyte macrophage colony-stimulating factor (GM-CSF)

and granulocyte colony-stimulating factor (G-CSF). Lastly,
adhesion molecules also play an important role in the
control of HSC proliferation, because the engagement of β-1
integrins prevents the progression of CD34+ cells from the
G1 to S phase of the cell cycle, thus inhibiting progenitor
proliferation.
Although osteoblasts are ultimately the stromal cell mak-
ing contact with and regulating the HSC compartment, bone
is continuously subject to bone remodeling, a process gen-
erated by the cooperation of osteoblasts, the bone-forming
cells, and osteoclasts, the bone-resorbing cells. As part of
this tightly regulated process, osteoblasts regulate osteoclast
maturation and proliferation, and there is emerging evidence
that the hemopoietic niche is the result of this process. In
fact, pharmacological agents with the ability to modulate the
number and/or functions of these cells have an impact on the
number of HSCs and their ability to engraft in myeloablated
recipients.
Mobilization and homing of HSCs
The function of the niche is not confined to the control of
HSC size, but also to their mobilization. The ability to migrate
is a fundamental property of HSCs throughout their devel-
opment. During ontogeny, blood formation occurs in dis-
tinct extra-embryonic and embryonic sites which involve in
sequence the yolk sac, aorto-gonadomesonephros (AGM),
placenta, and fetal liver, before hemopoiesis finally takes
place only in the bone marrow. Migration from one site to
another is critical, as demonstrated by the fact that mice made
genetically deficient in CXCL12, a chemokine responsible
for the homing of HSCs to the bone marrow, or its receptor
CXCR4, have normal production of fetal liver HSCs, but fail
to transition to hematopoiesis in the marrow space.
The migratory nature of HSCs is maintained in adulthood,
during which HSCs are found in the blood even under home-
ostatic conditions. Following the administration of cytokines,
including G-CSF, GM-CSF, and interleukin-8 (IL-8), the
number of HSCs in the peripheral blood changes dramat-
ically, with different kinetics depending on the mechanism
involved. For example, G-CSF can modify niche cells by
reducing the production of CXCL12 by osteoblasts, despite
the absence of G-CSF receptors on these cells. Of particular
interest is the observation that mice with altered sympathetic
nervous system function lack the ability to mobilize stem cells
from the bone marrow in response to G-CSF, thus indicating
the possibility that stem cells might be able to “sense” changes
such as those occurring distantly at the site of an injury.
Further developments in the area of HSC mobilization
have recently identified the use of chemokine antagonists
as an alternative and probably more effective strategy than
G-CSG. AMD3100, a specific antagonist of the chemokine
receptor CXCR4, is currently in clinical development, and
Molecular basis of transplantation 375
clinical trials have demonstrated its efficacy even in patients
or donors refractory to G-CSF.
The concept of BM space and
cell competition
A large bulk of studies in both animals and humans have
shown that recipients of HSCs must receive some form
of myeloablation for the HSCs to engraft, irrespective of
whether there is an underlying malignancy to eradicate or
an allogeneic immune system to suppress. This has intro-
duced the notion that the quantitative reduction of host cel-
lular components provides “space” for the incoming HSCs
to expand and compete more favorably with recipient cells,
reduced in numbers by the conditioning regimen. Further-
more, there is also evidence that such homeostatic expan-
sion might also affect some of the immunological mecha-
nisms involved in transplantation tolerance that we will dis-
cuss later.
However, “opening space” is not the only mechanism at
the basis of engraftment. High levels of long-term marrow
engraftment are obtained with the infusion of high levels of
marrow cells in untreated mice. This indicates that syngeneic
engraftment is also determined by stem cell competition in
the hemopoietic niche.
Exploiting HSCT for resetting the
immune system
While the long-term efficacy of autologous HSCT in malig-
nancies is confined to a minority of tumors, the fact that
HSCs are also the precursors of immune cells has prompted
investigators to test the ability of autologous HSCT to treat
conditions characterized by abnormal immune responses. In
support of this initiative is the evidence that the maturation
of new T cells in the thymus continues, albeit at a decreased
rate, also in adult life.
Autologous HSCT is currently being explored with
remarkable success in severe forms of autoimmune diseases,
including multiple sclerosis, systemic lupus erythematosus,
systemic sclerosis, rheumatoid arthritis, and Crohn’s disease.
The main rationale for applying HSCT to autoimmune
diseases has been the idea that intensive immune deple-
tion could eliminate the pathogenic repertoire, and that
reconstitution of a new immune system from hematopoietic
precursors could restore immune tolerance, halting ongoing
inflammatory activity and preventing relapses.
Recent studies have confirmed the notion that HSCT
induces alterations of the immune system which are
beyond the effects of a dose-escalating immunosuppressive
approach. HSCT has been shown not only to affect the B-cell
populations associated with the production of autoantibod-
ies, but also to profoundly perturb the T-cell compartment,
376 Molecular Hematology
as illustrated by the normalization of the deregulated T-cell
receptor (TCR) repertoire in multiple sclerosis. Furthermore,
it appears that following myeloablation, a subset of T cells
(regulatory T cells) with the specific function of controlling
immunity to self antigen selectively expands and could con-
tribute to the control of the underlying autoimmune disease.
On the contrary, the immune reconstitution following
allogeneic HSCT remains incomplete for several months
or years, depending on the histocompatibility differences
between donor and recipient. This is one of the several prob-
lems associated with the various immune responses gener-
ated in an allogeneic setting.
Basic concepts in the immunology of
allogeneic HSCT
Allogeneic HSCT triggers a network of immune responses
which fundamentally affect the outcome of the procedure,
in terms of both complications and therapeutic success.
Whereas recipient anti-donor immune responses (host ver-
sus graft, HvG) are important in solid organ transplan-
tation, in allogeneic stem cell transplantation (SCT) they
are profoundly inhibited by the conditioning regimen. The
immunologically competent cells, present in the HSC prepa-
ration, play a more important role because they mediate a
reaction against the host which targets recipient normal tis-
sues (graft-versus-host, GvH), but also mediate the effect at
the basis of the eradication of residual neoplastic cells (graft-
versus-leukemia, GvL).
The major histocompatibility complex
The major histocompatibility complex (MHC) defines a
genetic region which includes genes encoding class I and
class II membrane-bound cell surface glycoproteins. The
function of MHC proteins is to present peptide antigens to
T cells, a vital part of initiating an antigen-specific immune
response. MHC proteins are also involved in the recognition
of virally infected cells, or those cells in which genetic anoma-
lies arise, by natural killer (NK) cells.
There are two major classes of genes within the MHC
region, namely class I and class II MHC genes. In addition
to these, the MHC class III region encodes other proteins of
the immune system, like certain complement and cytokine
genes. In humans, the MHC region is found on the short arm
of chromosome 6 and encodes for the human leukocyte anti-
gens (HLAs). Different loci are designated by a letter, thus the
major class I loci are HLA-A, HLA-B, and HLA-C. HLA class
II genes are collectively designated HLA-D, and individual
loci identified by a second letter, HLA-DR, HLA-DP, HLA-
DQ. In addition to the “classical” class I and class II MHC
genes, there exists a number of non-classical MHC genes, like
HLA-E, HLA-F, and HLA-G, with functions which have not
yet been fully elucidated.
The role of MHC molecules is to present peptide antigens
to T cells. Class I molecules present endogenous peptides,
like those that are virus or tumor encoded, which are
generated in the cytosol, transported to the endoplasmic
reticulum (ER), and finally presented to the cell surface. HLA
class II molecules are assembled in the ER, then transported
through the Golgi to endosomal compartments, where they
load peptides that have entered the cell via endocytosis or
receptor-mediated internalization. The selection of peptides
by MHC molecules is dictated by the sequence of the MHC
antigen-binding groove. Although MHC molecules have
limited polymorphism as compared to TCRs, they exhibit
different avidity for different peptides, thus accounting for
individual variability to respond to the same antigen and
against different moieties.
Mechanisms of allorecognition
Allorecognition is a particular form of antigen presenta-
tion, occurring only after transplantation of tissues between
genetically disparate individuals of the same species (or after
in vitro simulation of such). HvG reactions refer to the
immune response of the host to disparate antigens expressed
on donor cells, which results in graft rejection. The current
understanding of HvG responses derives mainly from stud-
ies of solid organ transplantation, in which the host does
not receive any conditioning prior to the transplant and thus
maintains the ability to reject it.
Three pathways of allorecognition have been described:
direct, indirect, and the most recently described semi-direct
allorecognition (Figure 27.1). The term direct allorecogni-
tion was initially used to describe the recognition by host
T cells of intact donor MHC-peptide complex directly on
the surface of donor antigen-presenting cells (APCs), but
it has also later been extended to include the recognition
of other donor-derived transplantation antigens (minor his-
tocompatibility antigens) presented on donor APCs by an
MHC molecule that is shared between donor and recipient.
The frequency of alloreactive T cells using this pathway has
been estimated at between 0.1% and 10%, as compared to
approximately 10−5 for nominal peptide antigens. An expla-
nation for such a high frequency is that direct allorecognition
arises as a consequence of the cross-reactivity of self MHC-
restricted T cells.
Indirect allorecognition is the recognition of donor-
derived antigens which have been processed and are
presented on the cell surface of host APCs in the context
of self MHC. Therefore, the distinction between direct and
indirect presentation is thus the source of the APCs on
which alloantigens are presented. The mechanism of indirect
allorecognition is indistinguishable from the physiological

Direct presentation
MHC
Fig. 27.1 Mechanisms of allo-recognition. MHC,
major histocompatibility complex; TCR, T-cell Donor
receptor. Recipient
processing and presentation of pathogen-derived peptide
antigens. Donor MHC molecules can be processed, and
peptide fragments presented to T cells, by host APCs.
Although only approximately 5–10% of alloreactive T cells
are specific for indirectly presented donor antigens, they
are still thought to play a major role in chronic allograft
rejection, which occurs at a time when donor APCs are no
longer thought to be present. The mechanisms by which
indirectly primed T cells mediate graft rejection are unclear.
Following organ transplantation, the donor endothelial
layer is repopulated with recipient cells. One theory is that
donor antigen is presented to direct-pathway T cells by recip-
ient MHC I by the recipient endothelium. An alternative
hypothesis is that graft destruction is the result of bystander
killing following re-encounter of antigen on the surface of
graft-infiltrating APCs.
More recently, a predominant role of CD4+ T cells
has been advocated, whereby they orchestrate both the
production of donor-specific antibodies and the initiation
of inflammatory signals through myeloid cells. Following
the presentation of donor-derived antigens, host B cells are
induced to produce specific antibodies by CD4+ T cells. Allo-
graft injury is ultimately mediated through opsonization and
activation of host NK and macrophages. Host macrophages
can also interact with CD4+ T cells as host APCs.
A further mechanism of allorecognition has been
described as the semi-direct presentation of donor antigens
and the underlying mechanisms identified very recently.
The original hypothesis assumed that intact donor-derived
MHC complexes were transferred to and presented by
host-derived APCs to activate alloantigen-specific CD4+
T cells through the indirect pathway and specific CD8+ T
cells through the direct pathway. It was recently found that
high numbers of recipient APCs cross-dressed with donor
MHC are present in the lymph nodes draining the graft.
Donor MHCs are derived from allogeneic exosomes, which
can induce pro-inflammatory allo-responses even without
Molecular basis of transplantation 377
Indirect presentation Semi-direct presentation
TCR
T cells
transplantation. The contribution of this pathway to the
rejection of the transplanted organ is yet to be elucidated.
Transplantation tolerance: clonal deletion
The establishment of tolerance to antigens expressed by
donor tissues is a major goal of allogeneic transplantation.
There are three main mechanisms which contribute to trans-
plantation tolerance: clonal deletion, anergy, and peripheral
regulation (Figure 27.2).
During T-cell ontogeny, intrathymic clonal deletion of self-
reactive T cells is the primary process for the selection of the
mature T-cell repertoire. The transplantation of donor HSCs
into sublethally or lethally myeloablated recipients gives rise
to an immune system which is immunologically tolerant to
donor antigens. The chimerism is established not only at the
level of the BM, but also in the thymus, where recipient T
cells learn to recognize the donor as “self” after interaction
with donor-derived professional APCs, which orchestrate the
continuous depletion of alloreactive T and B cells. Since it is
mainly based on clonal deletion, such tolerance is long term
and does not require immunosuppression except for the first
few months after the transplant.
There is now plenty of anecdotal evidence that patients
undergoing allogeneic HSCT can subsequently receive an
organ from the original HSCT donor without the need for
immunosuppression. The induction of allograft tolerance
through HSCT in both HLA-matched and mismatched
kidney transplants has been actively pursued in three major
North American centers (Stanford, Northwestern University,
and Massachusetts General Hospital). By 2017, more than
80 patients had undergone this approach, and the results
of these clinical trials seem very encouraging, with the vast
majority of patients (50–76%) being off immunosuppression,
with improved quality of life and very low transplant-related
mortality rates (0–5%). Also negligible is the occur-
rence of graft-versus-host disease (GvHD) and infectious
378 Molecular Hematology
Recipient thymus
Natural Treg
Clonal deletion
Donor stem cells
Selection of
mature T cell
repertoire
Recipient
bone marrow
Naïve and
memory T cells
Peripheral anergy
complications. Graft rejection is less than 25% at long-term
follow-ups. Notably, although in most cases hematopoietic
chimerism was achieved during the induction phase of the
treatment, it was often not durable, thus questioning its
requirement in the maintenance of the allograft tolerance
in the long term. A better understanding of the underlying
mechanisms of tolerance induction after HSCT and the
improvement of the safety of the protocols used will cer-
tainly extend the success of this approach to allografts from
non-living donors.
Clonal deletion can also occur at extra-thymic sites
and can account for the removal of mature alloreactive T
cells, thus leading to the establishment and maintenance of
donor-specific tolerance in experimental models of mixed
chimerism induction following MHC-mismatched HSCT. T
cells are deleted in the periphery by either activation-induced
cell death (AICD) or passive cell death, both leading to
apoptosis.
T-cell anergy describes a persistent state of unrespon-
siveness of T cells to their cognate antigen. This functional
inactivation occurs as a consequence of TCR engagement
in the absence of full costimulatory signals. T cells receive
costimulatory signals via ligation of surface CD28 with B7
molecules expressed on APCs. Other costimulatory path-
ways include CD40/CD40 ligand (CD154). Furthermore,
signaling through CTLA-4 (CD152) has also been impli-
cated in the generation of hyporesponsive T cells. Anergic
T cells fail to produce sufficient interleukin (IL)-2 and to
proliferate in response to antigen, but the unresponsive
state can be overcome by the exogenous addition of IL-2
Induced Treg
Fig. 27.2 Mechanisms of transplant tolerance.
Treg, T-regulatory cells.
or if anergic T cells are cultured in the absence of their
cognate antigen. The blockade of costimulatory pathways
has been widely exploited as a mechanism to induce tol-
erance in solid organ transplantation and HSCT. In the
context of experimental models of allogeneic HSCT, the
administration of anti-CD154 mAb was shown to induce
mixed lympho-hematopoietic chimerism and subsequent
permanent skin graft acceptance if used as a single agent, in
combination with reduced-intensity HSCT, or in combina-
tion with the antagonists of CTLA-4. However, there is little
evidence in support of a role for anergy in the induction of
transplantation tolerance.
Transplantation tolerance: peripheral
regulation
Since the demonstration that tolerance to allogeneic graft can
be transferred to naı̈ve recipients by T cells derived from a
tolerant host, induction and maintenance of transplantation
tolerance have largely been ascribed to immunoregulation.
The initial problems in the identification of a suppressor
T cell in the 1970s have been partially overcome more
recently with the discovery of a distinct CD4+ T cell subset
constitutively expressing CD25. Besides CD25, regulatory
T cells (Treg) are characterized by the expression of the
transcription factor Foxp3, crucially involved in suppressing
overactive immune responses, as demonstrated by the
high incidence of self-reactive lymphocytes in immune
dysregulation, polyendocrinopathy, enteropathy, X-linked
(IPEX) syndrome in humans and scurfy in mice. Further
 Molecular basis of transplantation 379

markers to identify Treg cells include the expression of The immunological targets of GvHD
CTLA-4, glucocorticoid-induced tumor necrosis factor and GvL
receptor (GITR), folate receptor-4 (FR4), and the absence
We have previously discussed how differences between
of CD127. CD4+CD25+ naturally occurring Treg cells are
donor and recipient at the MHC level produce vigorous
thymic derived and have been shown to be important for
immune responses. Since genetic differences are the major
maintaining self-tolerance, regulating the homeostasis of the
factors influencing the outcome of allogeneic HSCT, the
peripheral T-cell pool, and contributing to tolerance induc-
selection of the donor is crucial, and where possible it
tion in various models of solid organ transplantation, as well
should be an MHC-identical sibling or an MHC-matched
as in allogeneic HSCT. In vivo depletion of Treg cells in ani-
unrelated donor. However, MHC matching is not sufficient
mal models increases the incidence of autoimmune diseases
for long-term graft survival and/or to prevent GvH reactions
and immune responses to tumors. It has also been observed
without the use of potent immunosuppressive regimens. The
that in a condition of full or mixed donor chimerism after
existence of additional histocompatibility loci, first indicated
HSCT, there is a selective advantage in the homeostatic
in inbred mice, was then clearly demonstrated in humans
expansion of Treg cells as compared to effector T cells. The
in HSCT, whereby they were associated with severe GvH. In
presence of donor antigens during this phase skews the Treg
this genetic situation, immune responses are directed against
repertoire toward the preferential expansion of those recog-
alloantigens encoded by histocompatibility (H) loci outside
nizing the donor antigens, thus contributing to the induction
the MHC, the so-called minor H loci. Minor H antigens
of transplantation tolerance. The use of conditioning regi-
are polymorphic self-derived peptides expressed on the cell
mens aimed at expanding Treg cells in vivo or the infusion of
surface in association with MHC class I and II molecules.
Treg cells expanded in vitro and co-transplanted with HSCs
The fact that they are recognized, as are viruses, by MHC-
have been proven efficacious in prolonging the survival of the
restricted T cells with specificity for peptides brought to
allograft.
the cell surface during biosynthesis of MHC class I and II
The mechanisms underlying naturally occurring Treg cell
molecules makes their identification more difficult. Unlike
immunosuppression require cell-to-cell contact. However,
in vitro T-cell responses to MHC antigens, which can be
adaptive CD4+CD25+ regulatory cells with a similar phe-
measured readily by proliferation in mixed lymphocyte reac-
notype have been described, they derive from mature effec-
tions (MLRs) without previous exposure to antigen, those
tor T cells and exert their suppressive activity via inhibitory
against minor H antigens need prior in vivo immunization.
cytokines such as TGF-β or IL-10. It is likely that a combi-
Although there is a large number of polymorphic pro-
nation of naturally occurring and adaptive regulatory cells
teins, not all give rise to peptides recognized as minor H
is involved in transplantation tolerance, depending on the
antigens. Furthermore, there are several factors that appear
time and the conditioning regimens. Although a proportion
in practice to severely limit the number of minor H antigens
of Treg cells specifically recognizes alloantigens, their over-
eliciting an in vivo response in a particular donor/recipient
all suppressive activity is not limited to cells expressing their
combination. In fact, it is clear that there is a hierarchy of
cognate antigens, thus implicating the possibility of unde-
responsiveness. In a genetic situation where there are many
sirable effects on virus- or tumor-specific immunity. There-
minor H disparities, one or a few are immunodominant,
fore, alloantigen-specific Treg cells are the most promising
with clones of CD8+ T cells responding to such an antigen
approach for their therapeutic application in the context of
expanding selectively, while T cells against others appear
transplantation.
transiently early in the response, or not at all. In most
models, immunodominance results from competition for
APC resources among responding CD8+ T cells, because it
Graft-versus-host and
disappears when competing epitopes are presented on dif-
graft-versus-Leukemia
ferent APCs or when APCs are present in large excess. There
is strong statistical and genetic evidence that this also occurs
A unique and prominent feature of allogeneic HSCT as com-
in human immune responses to multiple minor H antigens,
pared to solid organ transplantation is the presence, in the
thus in principle making it possible to predict, measure, and
graft, of immunologically competent cells which have the
manipulate the immune response following allografting.
ability to recognize normal and malignant recipient tissues.
As a consequence, donor lymphocytes mediate a reaction
Graft-versus-leukemia
against normal tissues which is defined as GvHD, but also
mediate the fundamental therapeutic effect at the basis of the Clinical studies and experimental animal models have
eradication of residual neoplastic cells, the GvL. It is not sur- shown that the efficacy of allogeneic HSCT in hematological
prising, therefore, that the molecular targets of GvH and GvL malignancies is related not only to the intensive chemo-
largely overlap. radiotherapy, but also to an immunological anti-tumor effect
380 Molecular Hematology
exerted by the graft itself. This effect, referred to as GvL,
has been initially recognized because patients who received
T-cell-depleted HSC preparations with the intention of
reducing GvHD had a much higher incidence of leukemia
recurrence after the transplant. Further lines of evidence
support this concept. For example, the risk of relapse is
higher if donor and recipient are identical twins, and some
reports have indicated that remission can be re-established
by the withdrawal of post-transplant immunosuppressive
treatment and/or by the recurrence of GvHD. The proof
of principle of the GvL effect came from the evidence that
the infusion of lymphocytes from the original stem cell
donor could restore complete remission in patients with
chronic myeloid leukemia (CML) relapsed after allogeneic
SCT. These data support the requirement of two main
components for GvL to occur: the presence of T cells in the
donor preparation, and the existence of antigenic differences
between donor and recipients.
Although the GvL effect of donor lymphocyte infusions
(DLIs) is extremely successful in CML (the response rate
is more than 90%), the experience in other malignancies
is not as good and, in some cases, rather disappointing.
However, the introduction of reduced-intensity condition-
ing has extended the indications for allogeneic HSCT, and
thus promoted the use of DLI in several diseases. The rate
of response after reduced-intensity allograft in patients with
high-risk acute myeloid leukemia (AML) and myelodysplas-
tic syndrome (MDS) has been reported to be comparable
with the results obtained using myeloablative regimens. In
these diseases, DLI can produce better outcomes if admin-
istered prophylactically or pre-emptively, guided by reduc-
tion of donor chimerism. Durable responses, which exceed
the response durations seen both after conventional trans-
plantation and following previous chemotherapy cycles, were
reported in a small cohort of patients with Hodgkin lym-
phoma and in low-grade non-Hodgkin lymphomas. Initial
enthusiasm with multiple myeloma has been subsequently
disappointing, since it has become apparent that responses
can be achieved only in association with severe GvHD. It
is possible that higher doses of donor lymphocyte are nec-
essary, but then they would be incompatible with the side
effects.
Exploitation of the beneficial effect of allogeneic immune
responses in non-hematological malignancies has indicated
that renal cell carcinoma (RCC) may be susceptible to the
GvH/tumor effect. However, no further evidence was pro-
vided, thus suggesting it had a minor impact.
The reason for the different sensitivity to DLI among var-
ious malignancies remains unclear. A possible explanation
for low susceptibility to DLI of acute lymphoblastic leukemia
(ALL) has been attributed to the limited duration and mag-
nitude of leukemia-specific T-cell response generated in vivo,
which has been attributed to clonal exhaustion. As widely
documented by studies on solid cancers, several mechanisms
implicated in the tumor evasion from immune surveillance
can explain the resistance to graft-versus-tumor responses.
An important component for the induction of GvL
reactions is a fully functional antigen presentation. There
is emerging evidence that host dendritic cells (DCs) of
leukemic origin may be harnessed in this process by induc-
ing the generation of anti-leukemic cytotoxic T cells in in
vitro and in vivo preclinical models. Early efforts to trans-
late this approach into the clinical setting seem encourag-
ing. The principle has been adopted by exploiting the abil-
ity of cytarabine and azacytidine to induce the maturation of
leukemic blasts into DCs, and the consequent provision of
antigenic and accessory stimulation of donor naı̈ve T cells.
There are reports of complete hematological remissions in
AML patients relapsed after HSCT by using a combination
of low-dose cytarabine, G-CSF-mobilized donor blood cells
and GM-CSF post-transplant, or a combination of azacyti-
dine and DLI. The efficacy of such a strategy has been associ-
ated with the differentiation of central memory T cells and
the induction of a selective, GvHD-sparing, anti-leukemia
effect.
Another cell subset involved in the elimination of residual
leukemic cells is NK cells. NK are fundamental players
of innate immunity. Their killing activity is tightly tuned
by the balance of activating and inhibitory signals trans-
mitted by the interaction of different receptors on the
surface of NK cells, such as killer-cell immunoglobulin-like
receptors (KIRs), NK Group 2 member D (NKG2D), or
NKG2A/CD94, and the corresponding ligands on the target
cell membrane. NK cells lyse target cells which lack MHC
class I molecules specific to the inhibitory receptors KIRs
(“missing-self hypothesis”). Therefore, NK cytotoxicity can
be particularly important in HLA-mismatched or haploiden-
tical donor–recipient pairs. Importantly, since NK cells are
the first immune population to recover after HSCT, they are
also the first cell subset to recognize normal and malignant
allogeneic targets. The involvement of “alloreactive” NK
cells in GvL, suggested in preclinical models, has been
corroborated in clinical studies, whereby the transplantation
of haploidentical NK-“alloreactive” donors in high-risk
AML patients has been shown to prevent disease relapse
and improve survival. Accordingly, a slow NK recovery after
HSCT has been correlated to a higher risk of relapse.
Are tumor-specific T cells generated
after allografting?
There is evidence that the immune system can mount a
response against a tumor by recognizing antigens which
are either specific or associated to the tumor. Tumor-
specific antigens (TSAs) include antigenic peptides gener-
ated by genetic mutations specific to an individual cancer.

Tumor-associated antigens (TAAs) come from normal non-
polymorphic proteins overexpressed or aberrantly expressed
in the tumor.
Almost all patients with CML harbor a specific molec-
ular abnormality, the reciprocal chromosome translocation
t(9,22)(q34;q11), which juxtaposes BCR and ABL genes.
The fusion gene encodes for a chimeric protein, namely
p210BCR-ABL . Peptides derived from p210BCR-ABL can be pre-
sented by both MHC class I and class II molecules on CML
cells in vitro and p210BCR-ABL -specific CD8+ T cells have
been detected in the peripheral blood of CML patients. How-
ever, it is unclear whether these T cells have any purpose
in the control of the leukemia. Early-phase clinical trials
using p210BCR-ABL vaccination have showed some encourag-
ing results. In patients with CML receiving a peptide-based
vaccination, the development of bcr-abl peptide-specific
immune responses was associated with clinical responses
and/or further reduction of residual disease.
Other non-polymorphic proteins, such as proteinase-3
(Pr3) and Wilm’s tumor 1 (WT1), have also been investigated
as targets for T cells in CML because of their high expression
levels in this leukemia. Pr3- or WT1-specific T cells at very
low frequency have been identified in patients with CML in
remission after allogeneic HSCT, but the findings have never
been confirmed and their significance is not supported by
proper clinical trials.
Several other tumor-associated and TSAs, such as
preferentially expressed antigen in melanoma (PRAME),
melanoma family antigen (MAGE), receptor for hyaluron-
mediated motility (RHAMM), New York esophageal
squamous cell carcinoma-1 cancer-testis antigen (NY3ESO),
or nucleophosmin (NPM1), are often expressed in leukemic
blasts. Cytotoxic T cells specific for the cognate antigen
have been isolated from the bone marrow of healthy donors
as well as from non-transplanted AML, ALL, and CML
patients, indicating that the alloreactivity post-transplant is
more prominent than leukemia-reactive immune responses.
The presence of higher titers of PRAME-specific CD8+
T cells after HSCT has been correlated with a higher
probability of longer remission. However, most immune
reactivities against any of these antigens traced after DLI
are weak, and the results so far obtained with the use of
TAA-specific T cells as immunotherapeutic agents have been
disappointing.
Graft-versus-host disease
GvHD is the main drawback of allogeneic SCT and is largely
responsible for non-relapse mortality. In relation to the
time of onset, GvHD can be classified as acute or chronic,
depending on whether it develops within 100 days from the
transplant. However, although there are clearly two types
Molecular basis of transplantation 381
of diseases in terms of symptoms, organ involvement, and
pathological changes, there is not necessarily a temporal
correlation, with several cases presenting as overlapping
syndromes. The uncertainty in the clinical classification is
the result of the poor understanding of the pathogenesis of
the disease and the very limited access to its monitoring.
Acute GvHD
Acute GvHD is a potentially lethal disease which usually
targets skin, liver, and the gastrointestinal tract. The typi-
cal pathological finding in the skin is epithelial apoptotic
damage, with characteristic nuclear alterations and a small
inflammatory infiltrate. In the liver, the first injury involves
vascular endothelial cells, and later the biliary epithelium is
destroyed as a result of portal lymphocytic infiltration. In the
gastrointestinal tract there is destruction of intestinal crypts
with acute or chronic inflammatory infiltrates and flattening
of the villous architecture. The clinical score is graded at four
levels, with grade 4 being very often lethal.
The cellular effectors of alloreactivity after HSCT are
cytotoxic T cells and NK cells, depending on the histo-
incompatibility of the donor–recipient pair. Cytotoxic T cells
are activated CD8+ T cells that recognize their specific anti-
gen bound to MHC class I molecules on the target cells. They
exert their functions through two main contact-dependent
cytolytic pathways: the Fas/Fas-ligand (Fas-FasL, CD95-
CD178)-mediated apoptosis and the release of cytotoxic
granules containing perforin/granzyme. Studies performed
in mice genetically deficient for these molecules showed that
a prominent but not exclusive role for the perforin/granzyme
pathway in GvHD-induced organ damage. In contrast, the
same pathway appears to be critical for the GvL effect, thus
suggesting the possibility of selectively impairing the alter-
native pathways to inhibit GvHD without compromising
GvL. In fact, during experimental and clinical GvHD, FasL
expression on donor T cells is increased and elevated serum
levels of soluble FasL and Fas are correlated with severity of
GvHD, and the inhibition of the Fas-FasL pathway markedly
reduces GvHD of the liver in animal models.
NK cells mediate direct cytotoxicity with the same
pathways of cytotoxic lymphocytes, albeit recognizing
different targets. Experimental studies have shown that NK
cells are neither necessary nor sufficient to induce GvHD,
but in the setting wherein GvHD is already underway,
NK cells contribute to the overall outcome. The effect of
donor NK-mediated cytotoxicity in vivo, in the absence of
alloreactive T cells, is shown to be confined to recipient
lympho-hematopoietic cells. Although the mechanism
responsible for such selectivity is unclear, it has been hypoth-
esized that donor “alloreactive” NK cells could prevent
GvHD manifestations by killing off host DCs, thus avoiding
alloantigen presentation to donor T cells.
382 Molecular Hematology
Chronic GvHD
Chronic GvHD is a syndrome characterized by multiorgan
involvement and is clinically similar to an autoimmune dis-
order. It usually involves skin, liver, gastrointestinal, and res-
piratory mucosa, but several immunological functions are
impaired. The incidence of chronic GvHD ranges from 6%
to 80%, mainly depending on the degree of disparity in the
major histocompatibility antigen and the previous appear-
ance of acute GvHD. Pathological findings are not typical
and are characterized by epithelial atrophy, increased hyaline
deposits, fibrosis, and chronic inflammatory infiltrates. His-
torically, chronic GvHD was classified as limited or extensive
on the basis of the results of a small retrospective study, but
that has been shown neither to be reproducible nor predic-
tive of the clinical outcome, and a new scoring system is being
proposed.
The pathogenesis of GvHD
Acute GvHD and the “Cytokine Storm”
Donor T cells recognizing major and/or minor H antigen dis-
parities on recipient tissues are certainly critical in the induc-
tion of acute GvHD. However, other factors need to be taken
into account. In particular, the incidence of acute GvHD has
been reported to be much higher if allogeneic donor lym-
phocytes are administered in concomitance with the condi-
tioning regimen than if they are infused a few months later
in the form of DLI. The reason for this is that the injuries
produced by the cytotoxic agents and the infections at time
of transplant activate the innate immune system and induce
pro-inflammatory changes in endothelial and epithelial cells.
This process, described as the “cytokine storm,” consists of
three main phases: (i) the destruction of natural barriers by
the conditioning regimens (first phase); (ii) the activation of
DC and presentation of alloantigens (second phase); and (iii)
the recruitment and proliferation of activated effector cells
(third phase).
In the first phase, the destruction of natural barriers by
the conditioning regimens and the release of inflammatory
cytokines, like tumor necrosis factor (TNF)-α and IL-1,
allows bacterial products to permeate in the tissues. The T
cells exposed to bacterial products like lipopolysaccharide
exhibit enhanced migration and survival. In accord with this
notion, the use of low-intensity preparatory regimens sub-
stantially reduces the incidence of acute GvHD. There is
increasing evidence that lower diversity of the microbiota in
the gut at the time of the HSCT is associated with higher
incidence of GvHD rates and mortality. Studies in humans
and mice indicate that active GvHD is associated with an
imbalance in favor of specific bacterium phyla, with a pre-
dominance of Enterococci. Furthermore, the use of specific
antibiotic associations active against anaerobic bacteria, such
as imipenem-cilastatin and piperacillin-tazobactam, seems
to correlate with higher GvHD grades and mortality. Con-
versely, the use of antibiotics with less perturbative effects on
the gut microbiota (such as rifaximin) was associated with
lower incidence of gut GvHD and GvHD-related mortality.
The studies of microbiota composition point out that its
role in GvHD development is not merely related to the
release of bacterial particles. In fact, microbiota seems to
actively participate in the generation of a specific environ-
ment which can be pro- or anti-GvHD through the pro-
duction of metabolites. Short-chain fatty acids, for exam-
ple, stimulate the extrathymic maturation of Treg cells and
one of these acids (butyric acid) can reduce the production
of pro-inflammatory cytokines while inducing the produc-
tion of indoleamine 2,3-dioxygenase (IDO), which can ame-
liorate GvHD. Based on these findings, several approaches
have been tested in clinical trials with the aim of restor-
ing the diversity of the microbiota by the use of probiotics
or specific combinations of antibiotics for decontamination
of the gut at the HSCT. Recently, the use of fecal micro-
biota transplantation has also been investigated for the man-
agement of steroid-resistant gut GvHD. Although the ratio-
nale for these approaches is very intriguing, clinical results
are still controversial and they have not yet entered clinical
practice.
The second step in the cytokine storm involves the
activation of DCs and the presentation of alloantigens. The
inflammatory changes promoted in recipient tissues by con-
ditioning regimens can initiate DC maturation and license
them to potently induce T-cell responses. While host APCs
are necessary and sufficient for GvHD development, the role
of donor APCs appears confined to the intensification of
ongoing disease. The differential role of donor and recipient
APCs can be accounted for by the fact that endogenously
synthesized minor H antigen-derived peptides are presented
to CD8+ T cells on host MHC class I (induction phase).
Subsequently, the same antigens can be processed by donor
APCs through the exogenous route of antigen processing and
presented to the T cells, thus accounting for augmenting the
GvHD initiated by the host APCs (maintenance phase). This
two-step process is similar to that described for the rejection
of solid organ transplants (direct and indirect alloresponses).
For its full development, GvHD finally requires cellular
effectors to be recruited. Although activated donor T and
NK cells effect the killing of host cells by contact-dependent
cytotoxicity, the release of inflammatory cytokines also plays
an important role. In accordance, target organ injury can be
partially prevented by the neutralization of TNF-α and IL-1
in animal models, and the blockade of TNF-α is of some
efficacy in pilot clinical studies. The importance of multiple
inflammatory effectors in GvHD suggests that more than
one pro-inflammatory cytokine should be simultaneously

inhibited to effectively antagonize systemic GvHD. The
case of IL-6 blockade can well explain the complexity
of the matter. IL-6 is a cytokine usually associated with
pro-inflammatory functions, but it also plays an important
role in controlling local and systemic inflammation. This
dual activity depends on the fact that IL-6 can transmit
signals through alternative pathways that can target different
subsets of CD4+ cells. While the small proportion of CD4+
cells expressing IL-6R are subject to a suppressive signal,
the vast majority respond to IL-6 linked to a soluble form of
IL-6R present in the serum, and that drives an inflammatory
response. Blockade of IL-6 signaling has been successfully
used in murine models of GvHD, but its efficacy in patients
seems to be confined to its prophylactic use.
An obvious candidate to inhibit the expansion of allore-
active GvHD T cells is IL-2, but the efficacy of anti-CD25
antibodies has not been sufficiently supported by the clini-
cal experience. In contrast, the use of low-dose IL-2 has been
demonstrated to effectively increase the number of Treg cells
and to be effective at treating the chronic form of GvHD. The
therapeutic efficacy of IL-2 could also be enhanced by com-
bination with rapamycin, which synergizes in increasing the
proliferation and expansion of Treg cells.
Acute GvHD and T-Cell Trafficking
Although essentially all tissues express transplantation anti-
gens, clinical manifestations of acute GvHD display remark-
able tissue tropism involving primarily gut, liver, and skin.
Recent studies have suggested that the skewed organ involve-
ment is related to the homing properties of activated allo-
geneic T cells and the presence of local tissue inflammation.
Naı̈ve T cells injected in lethally irradiated allogeneic
recipients are initially retained within secondary lymphoid
tissues where, activated by recipient-derived APCs, they
undergo a rapid burst of proliferation, enter the peripheral
circulation, and subsequently accumulate in the gut, liver,
and skin. However, local inflammation also appears crucial
in permitting the entry of activated T cells to target tissues,
because the infusion of alloreactive T cells produces GvHD
in irradiated recipients, but not in recipients in which mixed
hemopoietic chimerism has already been established, despite
similar levels of alloreactive T cells in the periphery. To con-
firm the role of local inflammation, GvHD can be induced
also in mixed chimeric mice if inflammatory stimuli are
administered together with the donor T-cell transfer.
Chronic GvHD
The pathogenesis of chronic GvHD largely remains to be elu-
cidated. Mature donor alloreactive T cells infused with the
graft are supposed to play a key role in chronic GvHD. Sim-
ilarly, APCs have been shown to be important although, in
Molecular basis of transplantation 383
contrast to acute GvHD, the development of the chronic form
is dependent on the APCs of donor rather than recipient ori-
gin, thus justifying the possibility of a broader repertoire of
antigens and more widespread organ involvement.
However, one of the most remarkable features of chronic
GvHD is the clinical and pathological similarities with
systemic autoimmune disease. Elevated levels of serum
autoantibodies (i.e. antinuclear, anti-dsDNA, anti-smooth-
muscle antibodies) in up to 70% of chronic GvHD patients
support the hypothesis that functionally relevant autoreac-
tive T and B cells can be generated during chronic GvHD. It
has been suggested that antibodies to ubiquitously expressed
minor H antigens may be analogous to pathogenic autoan-
tibodies in systemic autoimmune diseases, thus justifying
the similarities of tissue pathology. The importance of B-cell
responses in chronic GvHD was documented by clinical
improvement following B-cell depletion with anti-CD20
monoclonal antibody.
Studies in animal models have been carried out with
the aim of investigating the autoimmune manifestations of
chronic GvHD. Using an HSCT model in which donor and
recipient differed for multiple minor H antigens, it was pos-
sible to isolate from mice developing chronic GvHD autore-
active CD4+ T cells with a specificity for an MHC class
II-encoded determinant common to donor and recipient.
Moreover, donor CD4+ T cells isolated from recipients with
chronic GvHD cause autoimmune disease if transferred to
new syngeneic recipients. The development of donor T cells
that recognize antigens shared by donor and host has also
been attributed to impaired mechanisms of deletion or reg-
ulation during chronic GvHD. It has been hypothesized that
autoreactive T cells may be the consequence of a damaged
recipient thymus, unable to select the new T-cell repertoire
following the pre-transplant therapy or acute GvHD. How-
ever, it seems that also thymectomized animals can develop
GvHD.
Not dissimilarly from the acute disease, cytokine dysreg-
ulation has a causative role in chronic GvHD. High levels
of IL-1β, IL-6, interferon (IFN)-γ, TNF-α, and low levels
of the inhibitory cytokine IL-10, are associated with more
severe forms. Multiple cytokines produced by activated T
cells, such as INF-γ and TGF-β, were shown to promote
increased collagen deposition in preclinical models. Other
soluble mediators, such as the CCL2 and CCL3 chemokines,
seem to modulate collagen turnover and deposition by
sending signals to fibroblast via macrophages or indirectly
through stimulation of TGF-β.
Can GvHD and GvL be dissected?
Currently, the major aim in allogeneic HSCT is to identify
a way to the separate the beneficial GvL from GvHD (sum-
marized in Figure 27.3). Several attempts have been made
384 Molecular Hematology

1. Reduce 2. Deplete CD8*/

the dose Enrich CD4*

3. Remove alloreactive T cells

(suicide-inducible gene)
4. Induce anergy to alloantigens
(costimulation, Treg cells

5. Th1–Th2 shift Hemopoietic-restricted

minor Ags

6. Selection of
Leukemia-Associated
7. Block cytokines leukemia-specific
Ags
clones

TNFα, IFNγ
IL-2, IL-6 Cytotoxic pathways
JAK Inhibitors (FAS vs. perforin)

8. NK cells infusion

+ IL-15/IL-2
+ BiKEs
Leukemic APCs

9. DLI in association
with differentiating
factors
+ Cytarabine
+ Azacytidine
Leukemic blasts

DLI cells
Fig. 27.3 Strategies to separate GvHD from GvL. Ags, antigens; APCs, antigen-presenting cells; BiKEs, bispecific killer engagers; DLI, donor
lymphocyte infusion; IFN, interferon; IL, interleukin; JAK, Janus kinase; NK, natural killer; Th1, T helper type 1; Th2, T helper type 2; TNF, tumor
necrosis factor; Treg, T-regulatory cells.

during the last 15 years, either by selectively controlling
GvHD or selectively inducing GvL.
Control of GvHD
The simpler and more pragmatic approach is to escalate
the dose of donor lymphocytes until a clinical response is
achieved. This permits the incidence of GvHD to be reduced
to less than 20% of cases and has proved more useful than
depleting T-cell subsets like CD8+. Alternatively, the route of
selectively depleting in vitro recipient-reactive donor T cells
before their infusion is valuable and its application has been
successfully tested in haploidentical transplantation.
Alternative approaches to diminishing GvHD have been
based on the use of donor T cells genetically engineered
with suicide genes that can be activated in case of GvHD.
Such a strategy allows donor T cells to be safely exploited for
immune reconstitution and the GvL effect, with the option
of switching them off in case of GvHD. The positive experi-
ence with the gene encoding herpes simplex virus thymidine
kinase (HSV-TK) is being confirmed by the use of caspase 9
transgene. While the first relies on the administration of the
antiviral drug ganciclovir to induce necrosis in transduced
donor T cells, the second is based on the drug-induced acti-
vation of the intrinsic apoptotic pathway. The results of new
clinical trials are keenly awaited.
The possibility of interfering with the non-specific effects
of the cytokine storm with neutralizing monoclonal antibod-
ies is currently under scrutiny and the topic has been dis-
cussed before. Based on the observation that some of the
cytokines involved in GvHD development act through the
STA3/JAK2, JAK1, or JAK3 pathways, it is not surprising that

Treg cells
MSCs
MDSCs
Fig. 27.4 Cell-based immunosuppression in
hemopoietic stem cell transplantation. GvHD,
graft-versus-host disease; GvL, graft-versus-
leukemia; MDSCs, myeloid-derived suppressor
cells; MSCs, Mesenchymal stromal cells; Treg,
T-regulatory cells.
the availability of Janus kinase (JAK) inhibitors has paved the
way to testing these reagents in the management of GvHD.
The results are encouraging, but not yet sufficient to support
their use on a vast scale.
Active immune regulation of donor/recipient reactive cells
is emerging as a key mechanism for inducing and maintain-
ing tolerance to alloantigens (summarized in Figure 27.4).
Treg cells are probably the main player in this process and
a few studies in animal models have documented that their
adoptive transfer can also prevent and partially treat GvHD.
There are also suggestions that Treg cells may preserve the
GvL activity, but data obtained in patients after HSCT have
failed to show a significant correlation with GvHD and have
found that leukemia relapses are associated with an incre-
ment in the number of Treg cells in the peripheral blood of
patients. Similarly, the amount of Treg infiltration seems to
correlate with a bad prognosis in ovarian cancer.
Employing immunosuppressive networks represents a fur-
ther approach to controlling GvHD. Treg cells isolated from
umbilical cord blood and expanded in vitro or freshly isolated
from adult peripheral blood have been used for GvHD pro-
phylaxis in two different studies. In both cases, there was a
reduction in the development of GvHD in comparison with
historical controls and the treatment was well tolerated. How-
ever, Treg-treated patients exhibited a higher frequency of
opportunistic infections than controls.
Mesenchymal stromal cells (MSCs) have received center-
stage attention as potent modulators of immune responses,
especially because of the virtual absence of any side effects.
These cells exhibit a potent immunosuppressive activity in
vitro and in vivo, which targets virtually any cells of the
immune system. Their effect is not cognate dependent, thus
not requiring MHC identity between MSCs and immune
cells. There is extensive experience with the use of third-
party MSCs infused into patients with steroid-resistant acute
GvHD. All studies have shown a significant improvement
of overall survival in responding patients, although no ran-
domized trial has yet formally proven their efficacy. A very
Molecular basis of transplantation 385
↓ Tissue damage
↓ GvHD severity
↓ GvHD mortality
No effect on GvL
recent commercial study has claimed the final evidence in
a large phase III clinical trial, but the results have not yet
been made publicly available. The heterogeneous clinical
results are likely to result from the poor knowledge of the
mechanisms underlying immunosuppression and the lack
of criteria to stratify patients for treatment with MSCs.
Recent studies have shown that MSCs must undergo apop-
tosis to be effective, and that this apoptosis is induced
by the recipient cytotoxic cells responsible for GvHD.
This observation can provide a new perspective to stratify
patients for treatment and refine the use of this therapeutic
tool.
Finally, preclinical studies have tested the efficacy of
myeloid-derived suppressor cells (MDSCs), which are a
heterogenous myeloid population with immunosuppressive
activity. The adoptive transfer of MDSCs in animal models
of GvHD profoundly inhibits the development of GvHD.
Induction of Selective GvL
Cell dose and the time of administration are probably the
most important factors affecting the efficacy and toxicity
of DLI. Clinical responses are largely dependent on disease
stage, but the cell dose plays an important role, because the
number of responders increases with increasing the cell dose
and the effective cell dose correlates with disease stage.
Surprisingly, the incidence of GvHD is not affected by the
cell dose, but rather influenced by the time at which DLI is
administered. The longer the interval between the transplant
and DLI, the lower the incidence of acute and chronic GvHD.
Although these observations would strongly advise delaying
the administration of DLI, the fast kinetics of disease relapse
in diseases like AML and ALL make DLI completely ineffec-
tive if given at the time of full-blown relapse. Furthermore,
leukemia with high proliferative capacity are more prone to
escape T-cell donor immune control. Leukemic clones which
had become resistant to donor lymphocytes after losing mis-
matched HLA haplotypes by acquired uniparental disomy
386 Molecular Hematology
have been described in relapsed AML patients treated with
repeated DLI infusions.
Developing a DLI with a more selective antileukemia activ-
ity has been the holy grail of hematological immunotherapy.
It is generally true that minor H antigens are the targets of
both GvHD and GvL, thus in principle making the separa-
tion of the two processes impossible. However, there are two
strategies to pursue selectivity. One is based on the fact that
not all minor H antigens are equally expressed on the various
tissues and some of them are preferentially, if not exclusively,
expressed on hemopoietic tissues. In the case of leukemia, for
example, the generation of T cells with a specificity against
hemopoietic polymorphisms may limit the collateral dam-
age by confining the GvH activity to that tissue and sparing
the others. Further selectivity could be achieved by concen-
trating on lineage-specific polymorphisms, as demonstrated
for B-cell malignancies (Figure 27.5). A promising approach
could be the production of effector CD4+ T cells specific
for minor H antigens presented in the context of HLA class
II molecules, which are mainly expressed on cells of the
hematopoietic system. However, limitations to this approach
are represented by the observation that non-hematopoietic
precursors are induced to express HLA class II peptides in the
presence of inflammation. Furthermore, quiescent leukemic
stem cells do not express HLA class II epitopes.
The selective infusion of “CD4-enriched DLI” would also
induce the reversal of host antileukemic CD8+ cells with
exhausted phenotype by “awakening” dormant endogenous
GvL effector cells. The use of CD4+-selected DLI has been
associated with a high incidence of durable responses
Hemopoietic
GVHD GVL
failure
Normal Normal
Leukemia
tissues hemopoietic tissues
Alloreactive donor T cell
Polymorphisms expressed on all normal and malignant tissues
Hemopoietic-restricted polymorphisms
Leukemia-restricted polymorphisms
Fig. 27.5 Minor histocompatibility antigens in graft-versus-host
disease (GvHD) and graft-versus-leukemia (GvL).
and a low rate of GvHD. Responding patients exhibited a
significant downregulation of exhaustion-related genes in
their bone marrow CD8+ cells. These approaches could also
benefit from the association with drugs able to block the
immunological checkpoints, thus further acting against the
immunosuppressive environment of the tumor.
An alternative is to exploit the phenomenon of immun-
odominance that we described before. T cells directed against
a broadly expressed minor H antigen, but specific for the
immunodominant epitope, appear to be able to selectively
exert a powerful GvL without GvHD. The mechanisms for
this tumor-specific effect largely remain unknown.
Long-term responses can be achieved with more aggres-
sive strategies based on the use of pre-emptive or prophy-
lactic DLI in association with differentiating factors able to
induce the maturation of DCs from leukemic blasts and the
stimulation of naı̈ve donor lymphocytes. Promising also is
the combination of DLI with other therapeutic agents like
lenalidomide and sorafenib, although the mechanistic basis
of their interaction is still poorly understood.
The evidence that NK cells can produce a potent
antileukemic effect with minimal risk of GvHD makes such
an approach ideal. However, significant limitations are rep-
resented by the relatively small number of NK cells obtained
with leukapheresis and the poor knowledge of how to main-
tain NK therapeutic activity after their ex vivo expansion. The
administration of recombinant IL-2 and IL-15 has been used
in patients after NK infusions with a view to augmenting their
number and activation in vivo.
The use of bispecific killer engagers (BiKEs) has been
recently tested in clinic. BiKEs are small molecules made by
two antibody regions of different specificity, the first specific
for a TAA/TSA and the second CD16a specific, thus cross-
linking NK cells to a specific target cell. Such approaches have
been tested for the treatment of MDS and AML using a com-
bination of CD16a and CD33.
Much research is still needed to improve the therapeutic
efficacy of these procedures and to further extend their use
in clinical practice. It is conceivable that in the near future
a combination of strategies will become the most successful
approach to maximizing the GvL effect after HSCT.
Conclusions
HSCT is a potent clinical methodology which, by leveraging
on both recipient and donor immune systems, gener-
ates long-term immunological tolerance and antitumor
immunity. Despite the limitations and toxicities of the
procedure, the many successes of HSCT in malignant and
non-malignant conditions have provided a unique platform
for the development of new cellular therapies that are
currently being tested in other ailments.

Further reading
Clinical use of HSCT
Jethava, Y.S., Sica, S., Savani, B. et al. (2017). Conditioning regimens
for allogeneic hematopoietic stem cell transplants in acute myeloid
leukemia. Bone Marrow Transplant. 52 (11): 1504–1511.
Lindsay, J.O., Allez, M., Clark, M. et al., ASTIC trial group; European
Society for Blood and Marrow Transplantation Autoimmune Disease
Working Party; European Crohn’s and Colitis Organisation (2017).
Autologous stem-cell transplantation in treatment-refractory Crohn’s
disease: an analysis of pooled data from the ASTIC trial. Lancet Gas-
troenterol. Hepatol. 2 (6): 399–406.
Muraro, P.A., Pasquini, M., Atkins, H.L. et al. (2017). Multiple Sclerosis–
Autologous Hematopoietic Stem Cell Transplantation (MS-AHSCT)
long-term outcomes study group. JAMA Neurol. 74 (4): 459–469.
Takami A. Hematopoietic stem cell transplantation for acute
myeloid leukemia. Int. J. Hematol. 2018; 107(5): 513–518. doi:
https://doi.org/10.1007/s12185-018-2412-8
van Laar, J.M., Farge, D., Sont, J.K. et al., EBMT/EULAR Scleroderma
Study Group (2014). Autologous hematopoietic stem cell transplan-
tation vs intravenous pulse cyclophosphamide in diffuse cutaneous
systemic sclerosis: a randomized clinical trial. JAMA 311 (24): 2490–
2498.
The hemopoietic niche
Crane, G.M., Jeffery, E., and Morrison, S.J. (2017). Adult haematopoietic
stem cell niches. Nat. Rev. Immunol. 17 (9): 573–590. https://doi.org/
10.1038/nri.2017.53.
Korn, C. and Méndez-Ferrer, S. (2017). Myeloid malignancies and
the microenvironment. Blood 129 (7): 811–822. https://doi.org/
10.1182/blood-2016-09-670224.
Trento, C., Marigo, I., Pievani, A. et al. (2017). Bone marrow mesenchy-
mal stromal cells induce nitric oxide synthase-dependent differenti-
ation of CD11b+ cells that expedite hematopoietic recovery. Haema-
tologica 102 (5): 818–825. https://doi.org/10.3324/haematol.2016.
Transplantation tolerance
Chhabra, A.Y., Leventhal, J., Merchak, A.R., and Ildstad, S. (2017).
HSCT-based approaches for tolerance induction in renal transplant.
Transplantation 101 (11): 2682–2690.
Molecular basis of transplantation 387
Fozza, C. and Dazzi, F. (2012). Regulatory T cells in stem cell transplan-
tation: main characters or walk-on actors? Crit. Rev. Oncol. Hematol.
84 (1): 18–25.
Kitagawa, Y. and Sakaguchi, S. (2017). Molecular control of regulatory
T cell development and function. Curr. Opin. Immunol. 49: 64–70.
Marino, J., Babiker-Mohamed, M.H., Crosby-Bertorini, P. et al. (2016).
Donor exosomes rather than passenger leukocytes initiate allore-
active T cell responses after transplantation. Sci. Immunol. 1
(1):aaf8759.
Waldmann, H., Chen, T.C., Graca, L. et al. (2006). Regulatory T cells in
transplantation. Semin. Immunol. 18 (2): 111–119.
GvHD and GvL
Brunstein, C.G., Miller, J.S., McKenna, D.H. et al. (2016). Umbilical cord
blood-derived T regulatory cells to prevent GVHD: kinetics, toxicity
profile, and clinical effect. Blood 127 (8): 1044–1051.
Galleu, A., Riffo-Vasquez, Y., Trento, C. et al. (2017). Apoptosis
in mesenchymal stromal cells induces in vivo recipient-mediated
immunomodulation. Sci. Transl. Med. 9 (416). pii: eaam7828.
Martin, P.J., Rizzo, J.D., Wingard, J.R. et al. (2012). First- and second-line
systemic treatment of acute graft-versus-host disease: recommenda-
tions of the American Society of Blood and Marrow Transplantation.
Biol. Blood Marrow Transplant. 18 (8): 1150–1163.
McLornan, D., Potter, V., and Dazzi, F. (2000). Donor lymphocyte infu-
sion: rationale, benefits, and limitations. In: Hematopoietic Cell Trans-
plants: Concepts, Controversies and Future Directions (eds. H. Lazarus,
R. Gale, A. Keating, et al.), 223–231. Cambridge: Cambridge Univer-
sity Press https://doi.org/10.1017/9781316335727.026.
Schroeder MA, Choi J, Satser K, Di Persio JF 2018. The role of
janus kinase signaling in graft-versus-host disease and graft versus
leukemia. Biol. Blood Marrow Transplant. 24(6): 1125–1134.
Shlomchik, W.D. (2007). Graft-versus-host disease. Nat. Rev. Immunol.
7 (5): 340–352.
Simonetta, F., Alvarez, M., and Negrin, R.S. (2017). Natural killer cells
in graft-versus-host-disease after allogeneic hematopoietic cell trans-
plantation. Front. Immunol. 8: 465.
Staffas, A., Burgos da Silva, M., and van den Brink, M.R. (2017). The
intestinal microbiota in allogeneic hematopoietic cell transplant and
graft-versus-host disease. Blood 129 (8): 927–933.
von Dalowski, F., Kramer, M., Wermke, M. et al. (2016). Mesenchy-
mal stromal cells for treatment of acute steroid-refractory graft versus
host disease: clinical responses and long-term outcome. Stem Cells 34
(2): 357–366.
Index
5q− syndrome 53
11q23, acute myeloid leukemia 39
AAT see α1-antitrypsin
AAV see adeno-associated virus
Abelson gene (ABL) 63
ABL see Abelson gene
accelerated phase chronic myeloid leukemia
(CML-AP) 83
aceruloplasminemia 170
acquired bone marrow failure syndromes
131–137
idiopathic aplastic anemia 131–134
inherited glycosylphosphatidylinositol
deficiency 136–137
paroxysmal nocturnal hemoglobinurua
134–136
activated factor V (FVa) 209–212
activated factor VII (FVIIa) 207–209
activated factor VIII (FVIIIa) 210
activated factor IX (FIXa) 207
activated factor X (FXa) 207–210, 212
activated protein C (APC) 210–213
activated signaling
acute myeloid leukemia 41–42
myelodysplastic syndromes 56
ACTs see artemisinin-based combination
therapies
acute graft-versus-host disease 381–383
acute lymphoblastic leukemia (ALL) 59–70
biological consequences 60
B-precursor 62–65
chromosomal translocations 63–66
cooperating genetic events 60–62
cytogenic subgroups 60–61
development 62
donor lymphocyte infusions 380
hyperdiploidy 64–65
hypodiploidy 65
minimal residual disease evaluation
69–70
Molecular Hematology, 4th Edition. Edited by Drew Provan and John G. Gribben.
© 2020 John Wiley & Sons Ltd. Published 2020 by John Wiley & Sons Ltd.
mixed-lineage leukemia gene 64
molecular methodologies 59–60
pharmacogenomics 342–348
Philadelphia chromosome 63–64
Philadelphia-like 68
relapsing 68
T cell 65–68
acute myeloid leukemia (AML)
activated signaling 41–42
chromatin modification 45
chromosomal abnormalities 39–40
cohesin complex 44
core-binding factor 38
core concepts 50
cytogenic abnormalities 53–54
donor lymphocyte infusions 380
driver gene mutations 40–41
genetics 37–47
cytogenic abnormalities 37–40
monitoring and therapeutics 45–46
next generation sequencing 40–45
recurrent mutations 56–57
International Prognostic Scoring System
Revised 54
KMT2A rearrangements 39
methylation 44–45
recurrent mutations 56–57
retinoic acid receptor 38–39
secondary 54
spliceosome 43–44
therapy-related chromosomal aberrations
54
transcription factors 42–43
tumor suppressors 43
acute promyelocytic leukemia 38–39
adaptive immunity 300–303
additional sex combs like-1 (ASXL1)
45
adeno-associated virus (AAV) vectors 322,
330–331
adenoviral vectors 322, 329–330
adhesion
blood group antigens 291–293
malaria and host cells 200–202
adhesion-related platelet disorders 253–256
ADR see adverse drug reactions
adult hemoglobin, structure 2
adverse drug reactions (ADR)
single nucleotide polymorphisms 344
tyrosine kinase inhibitors 81
AE1 see anion-exchanger protein 1
African siderosis 170
AGM see aorta-gonad-mesonephrous region
agonist receptor defects, platelets 257
AKT 88–89
ALCL see anaplastic large cell lymphoma
ALG see anti-lymphocyte globulin
ALK see anaplastic lymphoma kinases
alkylating agent-related chromosomal
aberrations 54
allogeneic stem cell transplantation 373–387
chronic myeloid leukemia 82–83
clonal deletion 377–378
Diamond–Blackfan anemia 144
dyskeratosis congenita 142
Fanconi anemia 140
graft-versus-host disease 377–386
graft-versus-leukemia 379, 383–384
homing and mobilization 375
idiopathic aplastic anemia 133
immune system reset 375–376
immunology 376–379
indications 373
microenvironment 374–375
non-immunological factors 374–375
principles 373
sickling disorders 183–184
space and competition 375
tumor-specific antigens 380–381
allorecognition mechanisms 376–377
allostery, hemoglobin 173–175
all-trans-retinoic acid (ATRA) 38–39
389
390 Index

α1-antitrypsin (AAT) deficiency 330–331 human studies 306–307 glycophorin C/D 290–291
α globins 2, 9–12 idiopathic aplastic anemia 131–133 Rh antigens 289–290
α-granule defects 258 infections 306, 313–314 BMF see bone marrow failure
α2β1 integrins 255 mouse models 305–306 BMPR see bone morphogenetic protein
alpha-ketoglutarate 55–56 non-infectious triggers 306–307 receptor
α0 thalassemia 9–10 pathophysiology 308 BMT see bone marrow transplantation
α+ thalassemia 10–11 T cell selection 301–303 bone marrow, multiple myeloma 127–128
α thalassemias 9–12, 15–16 thrombocytopenia 297, 305–306, 308–311 bone marrow failure (BMF) syndromes
AML see acute myeloid leukemia tissue damage 308 131–153
anaplastic large cell lymphoma (ALCL) 107 treatment 311–315 acquired 131–137
anaplastic lymphoma kinases (ALK) 107 autosomal dominant siderosis 170 anemia of chronic disease 156
anemia Diamond–Blackfan anemia 142–144
of chronic disease 155–159 BAFF see B cell activating factor dyskeratosis congenita 140–142
diagnosis 155–156 B cell activating factor (BAFF) 127 Fanconi anemia 138–140
pathogenesis 156–158 B cell precursor acute lymphoblastic leukemia glycosylphosphatidylinositol deficiency
treatment 158–159 (B-precursor ALL) 62–65 136–137
fetal 276 B-cell receptor (BCR) signaling pathway idiopathic aplastic anemia 131–134
malarial 202–204 115–116 inherited 138–144
sickle cell 175–187 BCL-2 103–104, 106, 108, 116 paroxysmal nocturnal hemoglobinurua
anemia of inflammation see anemia, of BCL11A 184 134–136
chronic disease Bcl-xL 89 red cell enzymopathies 144–151
anion-exchanger protein 1 (AE1) 285–289 BCR see B-cell receptor, breakpoint cluster bone marrow space 375
ankyrin 286 region bone marrow transplantation (BMT)
antenatal diagnosis of hemophilia 229–230 BCR-ABL1 51, 73–74 sickling disorders 183–184
anti-D/anti-K prophylaxis 276–277 Bernard–Soulier syndrome (BSS) 253–254 see also allogeneic stem cell transplantation
antigen-presenting cells (APCs) 297–298, β-catenin 366, 367 bone morphogenetic protein receptor (BMPR)
302–303 β globins 2, 7–9 166
anti-lymphocyte globulin (ALG) 131, 133 β thalassemias 7–9, 13–15 bosutinib 74–75, 81
antiplatelet antibodies 310 with hemoglobin E 187 2,3-BPG see 2,3-diphosphoglycerate
antithrombin (AT) 209 with sickle cell anemia 184 B-precursor ALL see B cell precursor acute
deficiencies 213–214 Beth Israel hemoglobin 189 lymphoblastic leukemia
pharmacogenomics 348–350 BL see Burkitt lymphoma BRC-ABL1 51
aorta-gonad-mesonephrous (AGM) region blastic phase chronic myeloid leukemia breakpoint cluster region (BCR) signaling
29 (CML-BP) 83 pathway 108
APC see activated protein C bleeding disorders see hemophilia, platelet brush border ferrireductases 162
APCs see antigen-presenting cells disorders, thrombophilia, von Bruton’s tyrosine kinase (BTK) 107, 108,
APECED see autoimmune polyglandular Willebrand disease 115–116
endocrinopathy with candidiasis and blood cell alloantigens 267–283 BSS see Bernard–Soulier syndrome
ectodermal dysplasia class I 271–272 BTK see Bruton’s tyrosine kinase
apical membrane antigen 1 (AMA-1) 199 class II genes 272 Burkitt lymphoma (BL) 103–105
apoptosis suppression 103–104 genetics 268–271 Bushwick hemoglobin 188
artemisinin-based combination therapies hemolytic disease of the fetus and newborn busulfan (1,4-dimethane-sulfonyl-oxybutane)
(ACTs) 193 273–277 74, 82
AT see antithrombin molecular typing 268–271
ATA boxes 4 neonatal alloimmune neutropenia C4b-binding protein (C4BP) 211, 215
ATR-16 syndrome 11 280–282 CAFCs see cobblestone area-forming cells
ATRA see all-trans-retinoic acid neonatal alloimmune thrombocytopenia CALM-AF10 see PICALM-MLLT 10
atransferrinemia 170 277–280 CALR 93–94
ATR-X syndrome 11–12 see also blood group antigens Campath-1H 311–312
autoimmune polyglandular endocrinopathy blood coagulation 207–211 cancer stem cells (CSCs) 363–372
with candidiasis and ectodermal protein C system 208–211 cell of origin 364
dysplasia (APECED) 304 regulation 209 dormancy 367
autoimmunity 297–317 see also thrombophilia epigenetics 368–369
benefits of 303–304 blood group antigens 285–296 microenvironment 367–368
epitope spread 307 adhesion molecules 291–293 pre-leukemic 364–365
genetics 304–305 anion-exchanger protein 1 285–289 surface antigens 365–366
hormonal influences 307 diversity of functions 292 targeted therapies 365–369

CAP structures 5
carrier testing, hemophilia 228–229
CARs see chimeric antigen receptors
CBF see core-binding factor
CBFB 38
CBFβ–MYH11 fusion protein 38
CCAAT boxes 4
CCAAT/enhancer-binding protein α
(CEBPA) 42–43, 57
CD34+ cells 30
CD44 291–292
CD123 see interleukin-3 alpha receptor
CD177 281
CDAE1 see N-terminal cytoplasmic domain
of AE 1
CDKIs see cyclin-dependent kinase
inhibitors
CDKN2A 66
CEBPA see CCAAT/enhancer-binding
protein α
cell-cycle-active drugs 31
cell cycle control
hematopoiesis 24
lymphomas 106
multiple myeloma 125
cell-extrinsic regulators, hematopoiesis
24–29
cell-intrinsic regulators, hematopoiesis
24
cell of origin (COO)
cancer stem cells 364
chronic lymphocytic leukemia 112–113
cellular compartment model 21–23
central dogma 356
central tolerance 302–303, 377–378
ceruloplasmin 170
CFCs see colony-forming cells
CFU-S see colony-forming units, spleen
CGH see comparative genomic
hybridization
Chédiak–Higashi syndrome 259
chemokine receptors, hematopoiesis 26–29,
30
childhood acute lymphoblastic leukemia
342–348
Chile hemoglobin 190
chimeric antigen receptors (CARs)
331–334
CHIP see clonal hematopoiesis of
indeterminate potential
ChIP assays see chromatin
immunoprecipitation assays
cHL see classical Hodgkin lymphoma
chorionic villus sampling 229
Christmas disease 226
chromatin immunoprecipitation (ChIP)
assays 103
chromatin modification
acute myeloid leukemia 45
see also epigenetics
chromosomal abnormalities
acute myeloid leukemia 39–40
chronic lymphocytic leukemia 113
multiple myeloma 122–123
chromosomal translocations
acute lymphoblastic leukemia 60, 63–66
chronic myeloid leukemia 73–74
myeloid malignancies 51–52
Philadelphia chromosome 63–64
chromosome 1, multiple myeloma 126
chromosome 7, acute myeloid leukemia
39–40
chromosome 9, Philadelphia chromosome
63–64
chromosome 13, multiple myeloma 125
chromosome 17, multiple myeloma 126
chromosome 22
B-precursor ALL 63–64
chronic myeloid leukemia 51
see also Philadelphia chromosome
chronic eosinophilic leukemia 95
chronic graft-versus-host disease 382, 383
chronic lymphocytic leukemia (CLL)
B-cell receptor signaling 115–116
cell of origin 112–113
chromosomal alterations 113
environmental factors 112
epidemiology 111–112
epigenetics 115
genetics 107–108
impaired immunity 116–117
microenvironment 116
micro RNA alterations 114–115
molecular biology 111–120
prognostic factors 117–118
Richter transformation 118–119
somatic mutations 113–114
stereotype receptors 113
chronic myeloid leukemia (CML) 71–85
BCR-ABL1 73–74
chromosome 22 51
clinical presentation 71–72
donor lymphocyte infusions 380
epidemiology 71
prognostic models 73
SRC family of kinases 74
therapy 74–83
accelerated/blastic phases 83
allogeneic stem cell transplantation
82–83
busulfan 74, 82
discontinuation 77
drug dose schedules 79–81
endpoints 77–78
Index 391
frontline 75–78
hydroxyurea 74, 82
interferon alpha 81–82
monitoring 78–79
Philadelphia chromosome-negative 83
secondary 79
side effects 81
tyrosine kinase inhibitors 74–81
treatment failure 78–79
chronic neutrophilic leukemia (CNL) 95
chronic refractory immune thrombocytopenia
311, 313–314
CIDRs see cysteine-rich interdomain regions
circumsporozoite protein (CSP) 195
c-kit 25, 38, 42
classical Hodgkin lymphoma (cHL) 108
classifications
multiple myeloma 124
von Willebrand disease 236–238
clinical factors
chronic myeloid leukemia 71–72
dyskeratosis congenita 140
Fanconi anemia 138, 140
glucose-6-phosphate dehydrogenase
deficiency 148
hemophilia 221–223
idiopathic aplastic anemia 133
immune thrombocytopenia 310
JAK2V617F myeloproliferative neoplasms
89–91
neonatal alloimmune thrombocytopenia
279–280
paroxysmal nocturnal hemoglobinurua
136
red cell enzymopathies 144
sickle cell anemia 177–180
von Willebrand disease 236–238
see also diagnosis
clinical features
chronic myeloid leukemia 71–72
JAK2V617F myeloproliferative neoplasms
89–91
CLL see chronic lymphocytic leukemia
clonal deletion 302–303, 377–378
clonal hematopoiesis 49, 133–134
clonal hematopoiesis of indeterminate
potential (CHIP) 49
clopidogrel, pharmacogenomics 350
CLP see common lymphoid progenitor
clustered regularly interspersed short
palindromic repeats (CRISPR)-Cas9
system 103, 184, 320
CML-AP see accelerated phase chronic
myeloid leukemia
CML-BP see blastic phase chronic myeloid
leukemia
c-MYC 66–68, 103–105, 125–127
392 Index
CNL see chronic neutrophilic leukemia
CNV see copy number variants
coagulation 207–211
protein C system 208–211
regulation 209
see also hemophilia, platelet disorders,
thrombophilia, von Willebrand disease
cobblestone area-forming cells (CAFCs)
22
codeine, pharmacogenomics 341
codons 5, 356
cohesin complex
acute myeloid leukemia 44
myelodysplastic syndromes 56
collagen receptors, platelet disorders
255–256
colony-forming cells (CFCs) 22–23, 30
colony-forming units, spleen (CFU-S) 22
colony-stimulating factor 3 receptor (CSF3R)
mutations 95
commitment, hematopoiesis 24
common lymphoid progenitor (CLP)
populations 23
comparative genomic hybridization (CGH)
102
competition, stem cell transplantation
375
complement system 135, 298–299
congenital cyanosis 17
conventional metaphase cytogenic analysis
50–51
COO see cell of origin
copy number variants (CNV) and
pharmacogenomics 341
core-binding factor (CBF) 38
costimulatory blockade 313
coumarins, pharmacogenomics 348–350
counseling
hemophilia 228–229
neonatal alloimmune thrombocytopenia
280
CRISPR see clustered regularly interspersed
short palindromic repeats
crizanlizumab 183
CRLF2 see cytokine receptor-like factor 2
cryptic splice sites 8
CSCs see cancer stem cells
CSF3R see colony-stimulating factor 3
receptor
CSP see circumsporozoite protein
CXCR4, hematopoiesis 30
cyanosis, congenital 17
cyclin D1 106, 123–124
cyclin-dependent kinase inhibitors (CDKIs)
24
cyclin-dependent kinases (CDKs) 106
cyclophosphamide, hematopoiesis 31
CYP2C9
clopidogrel efficacy 350
warfarin efficacy 349
CYP2D6 and codeine 341
cysteine-rich interdomain regions (CIDRs)
201
cytarabine 367, 369
cytogenic abnormalities
acute lymphoblastic leukemia 60–61
acute myeloid leukemias 37–40
myelodysplastic syndromes 52–53
cytokine receptor-like factor 2 (CRLF2)
deregulation 65
cytokines
anemia of chronic disease 158
autoimmunity 305
graft-versus-host disease 382–385
hematopoiesis regulation 24–25
immunity 300–301
JAK2 signaling 87–89
MPL mutations 92
cytoskeletal platelet disorders 260
dasatinib 74–77, 79–81
side effects 81
DBA see Diamond–Blackfan anemia
DBP see Duffy binding proteins
DC see dyskeratosis congenita
DCYTB see duodenal cytochrome b
3-deazaneplanocin A (DZNep) 369
deficiencies
α1-antitrypsin 330–331
antithrombin 213–214
glucose-6-phosphate dehydrogenase 345,
364
protein C 214–215
protein S 215
DEK-NUP214 39
deletions, α thalassemias 10–11, 12
δ-granule defects 258
δβ thalassemias 12
dense δ-granule defects 258
deoxy(T) hemoglobin 2
desmopressin 244–245
deterministic theory of hematopoiesis 23
development
acute lymphoblastic leukemia 62
globin expression 3, 6
hematopoietic stem cells 22, 29
multiple myeloma 125–126
stem cells 22
diagnosis
anemia of chronic disease 155–156
autoimmunity 310
Diamond–Blackfan anemia 143
dyskeratosis congenita 142
Fanconi anemia 140
genetic analysis techniques 102–103
glucose-6-phosphate dehydrogenase
deficiency 148, 149
hemophilia 229–230
immune thrombocytopenia 310
multiple myeloma 124–125
myelodysplastic syndromes 52–53
myeloid malignancies 50–51
myeloproliferative neoplasms 95–96
neonatal alloimmune neutropenia 282
neonatal alloimmune thrombocytopenia
279–280
pharmacogenomics 339–352
platelet disorders 252–253
pyruvate kinase deficiency 147
red cell enzymopathies 144
von Willebrand disease 238
see also clinical factors
Diamond–Blackfan anemia (DBA)
142–144
Dicer 1 367
differentiation 21
diffuse large B-cell lymphoma (DLBCL) 104,
105, 118–119
1,4-dimethane-sulfonyl-oxybutane see
busulfan
2,3-diphosphoglycerate (2,3-BPG) 3
discontinuation, chronic myeloid leukemia
treatment 77
divalent metal transporter 1 (DMT1) 162
DKC1 140–142
DLBCL see diffuse large B-cell lymphoma
DLIs see donor lymphocyte infusions
DMT1 see divalent metal transporter 1
DNA
discovery of 354–357
see also epigenetics, genetics, transcription
DNA methyltransferase 3A (DNMT3A) 44,
55, 57, 94–95
DNA sequencing 103, 360–362. See also
next-generation sequencing
DNMT3A see DNA methyltransferase 3A
donor lymphocyte infusions (DLIs) 380,
385–386
dormancy, leukemic stem cells 367
Dot1l 369
double-hit lymphomas 104–106
driver gene mutations 40–41
drug dose schedules
chronic myeloid leukemia 79–81
see also management, treatment
Duffy antigen 196–197
Duffy binding proteins (DBP) 197–198
duodenal cytochrome b (DCYTB) 162
dyskeratosis congenita (DC) 140–142
dyskerin 140
DZNep see 3-deazaneplanocin A
 Index 393

eADA see erythrocyte adenosine deaminase erythropoiesis FGFR3 125

early T-cell precursor acute lymphoblastic iron restriction 156–157 fibronectin 30–31
leukemia (ETP ALL) 67 malaria 204–205 FISH see fluorescent in situ hybridization
EBA ligands 197–198 erythropoietin (EPOR) FLT3-ITD 39, 41–42, 57
EBP see erythrocyte-binding proteins JAK2 signaling 87–89 Flt3 receptor
EBV see Epstein–Barr virus resistance and inflammation 158 acute myeloid leukemia 41–42, 57
eculizumab 136 E-selectins, hematopoiesis 30 hematopoiesis 25
egress of hematopoietic stem cells 30–31 essential thrombocythemia (ET) 87–96 fluorescence-activated cell sorting (FACS) 32
electrophoresis 359 CALR mutations 93–94 fluorescent in-situ hybridization (FISH) 102
eltrombopag 133, 314 JAK2 mutations, other 91–92 follicular lymphoma, genetics 106
Embden–Meyerhof pathway 147 JAK2V617F mutation 89–91 frequencies of hemoglobin variants 17–18
endoplasmic reticulum (ER), CALR mutations MPL mutations 92 frontline therapy
93–94 ESXL1 94–95 chronic myeloid leukemia 75–78
endosomes, iron transport 163 ethanol precipitation 358 malaria 193
endothelial protein C receptor (EPCR) 209 ethnic groups, pharmacogenomics 342 functional analysis, hematopoietic stem cells
engraftment 29, 29–30 ETP ALL see early T-cell precursor acute 33–34
environmental factors lymphoblastic leukemia functions of hemoglobin 2–3
autoimmunity 306 ETV6 65 FVa see activated factor V
chronic lymphocytic leukemia 112 expression FVII see factor VII
EPCR see endothelial protein C receptor globins FVIIa see activated factor VII
epidemiology developmental changes 6 FVIII see factor VIII
autoimmunity 306–307 gene action 4–5 FVIIIa see activated factor VIII
chronic lymphocytic leukemia 111–112 genetic control 3–4 FX see factor X
chronic myeloid leukemia 71 regulation 5–6 FXa see activated factor X
FV Leiden 212 expression cloning 358
hemoglobin variants 17–18, 187 extracellular-signal-regulated kinase (ERK) G6PD see glucose-6-phosphate
malaria 193–195 pathway 88–89, 108 dehydrogenase
neonatal alloimmune thrombocytopenia ex-vivo expansion 32–33 gain of function mutations, prothrombin
279 EZH2 55, 94–95, 369 215–217
sickling disorders 176, 184–186 γ chains, hemoglobin 2–3
epigenetics 2, 357–358 F1P1L1-PDGFRA fusion gene 95 ganciclovir 384
acute myeloid leukemias 44–45 FACS see fluorescence-activated cell sorting GATA2 39
cancer stem cells 368–369 factor V (FV) 209–212 GBT440 183
chronic lymphocytic leukemia 115 Leiden 211–215, 217 G-CSF see granulocyte colony-stimulating
Fanconi anemia 139 factor VII (FVII) 207–209 factor
myelodysplastic syndromes 55–56 factor VIII (FVIII) 210 gemtuzumab ozogamicin (GO) 366
myeloproliferative neoplasms 94–95 hemophilia 223–225 gene action, hemoglobin 4–5
pharmacogenomics 342 inhibitory antibodies 226–228 gene expression profiling (GEP) 102
epitope spread, autoimmunity 307 von Willebrand disease 236, 238, 243–245 gene therapy 319–337
EPOR see erythropoietin factor IX (FIX) 207, 209, 225–226 cancer 331–335
Epstein–Barr virus (EBV) 105 gene therapy 330 factor IX 330
ERK see extracellular-signal-regulated inhibitory antibodies 226–228 hemophilia 231–232
kinase factor X (FX) 207–210, 212 lysosomal storage diseases 319, 327
erythrocyte adenosine deaminase (eADA) factor XI (FXI) 207–209 severe combined immunodeficiency
142 factor FXII (FXII) 207 326–327
erythrocyte-binding proteins (EBP) Fanconi anemia 138–140 suicide induction 334–335
197–198, 200–201 FBXW7 66 vectors 321–331
erythrocyte lifespan, inflammation effects FCGR3B 281 adeno-associated 322, 330–331
158 Fcγ receptors 305 adenoviral 322, 329–330
erythrocytes ferritin 155–156, 164 lentiviral 322–323, 327–329
adhesion molecules 291–293 ferroportin 162–165 oncoretroviral 321–327
anion-exchanger protein 1 285–289 hemochromatosis 169 genetic counseling, hemophilia 228–229
antigens 267–296 siderosis 170 genetic immunotherapy 331–335
glycophorin C/D 290–291 fetal anemia 276 genetics
malarial invasion 196–200 fetal hemoglobin 2, 3, 6–7 acute lymphoblastic leukemia 60–62
Rh antigens 289–290 fetal tests for hemophilia 229–230 acute myeloid leukemias 37–47
see also blood, hemoglobin, red blood cells FGFR1 95 analytical techniques 102–103
394 Index

genetics (Continued) GM-CSF receptor see granulocyte homing 29–30, 375

anaplastic large cell lymphoma 107
autoimmunity 304–305
Burkitt lymphoma 103–105
chronic lymphocytic leukemia 107–108,
111–112
Diamond–Blackfan anemia 142–143
diffuse large B-cell lymphoma 104, 105
double-hit lymphomas 104–106
dyskeratosis congenita 140–142
Fanconi anemia 139
follicular lymphoma 106
glucose-6-phosphate dehydrogenase
deficiency 148
hemophilia 223
HFE-associated hemochromatosis
167–169
HLA genes 268–271
Hodgkin lymphoma 108
lymphomas 101–110
lymphoplasmacytic lymphoma 104,
106–107
mantle cell lymphoma 106
mucosa-associated lymphoid tissue
lymphomas 107
neonatal alloimmune neutropenia
281–282
neonatal alloimmune thrombocytopenia
280
pyruvate kinase deficiency 146
Rh antigens 273–276
sickle cell anemia 175–177
thrombophilia 211–217
type 1 diabetes 304
von Willebrand disease 238–244
von Willebrand factor 236
see also pharmacogenomics
genome annotation 1
genome-wide association studies (GWAS) 6
JAK2V617F myeloproliferative neoplasms
90–91
genomic microarrays 51
genotype-phenotype relationships,
thalassemias 12–16
geography of malaria 194
GEP see gene expression profiling
giant platelet disorders 260
Glanzmann thrombasthenia (GT) 256–257
globins
genetic control 3–4
structural variants 6–7, 16–18
synthesis 4–6
glucose-6-phosphate dehydrogenase (G6PD)
deficiency 148–150, 345, 364
glycophorin A 197
glycophorin C/D (GPC/GPD) 290–291
glycosylphosphatidylinositol (GPI) 134–137
macrophage colony-stimulating factor
receptor
GO see gemtuzumab ozogamicin
Gower hemoglobin 2
GPC/GPD see glycophorin C/D
GPCRs see G-protein-coupled receptors
GPI see glycosylphosphatidylinositol
G-protein-coupled receptors (GPCRs)
26–29
GPVI 255–256
graft-versus-host disease (GvHD) 377–386
acute 381–383
chronic 382, 383
control 384–385
pathogenesis 382–383
separation from graft-versus-leukemia
383–386
graft-versus-leukemia (GvL) 379, 383–386
selective induction 385–386
granulocyte colony-stimulating factor
(G-CSF) 23–24, 31, 367, 374–375
granulocyte macrophage colony-stimulating
factor (GM-CSF) receptor 23, 25,
374–375
Gray platelet syndrome 258–259
Griscelli syndrome 259
GT see Glanzmann thrombasthenia
GvHD see graft-versus-host disease
GvL see graft-versus-leukemia
GWAS see genome-wide association studies
Hammersmith hemoglobin 188
haplotypes and pharmacogenomics 341
HApMAp consortium 341
Hb see hemoglobin
HDFN see hemolytic disease of the fetus and
newborn
Heinz bodies 188
Helicobacter pylori eradication 313–314
hematopoiesis 21–31
cellular compartment model 21–22
lineage commitment 23–24
microenvironmental niche 29
regulation 24–29
trafficking 29–31
hematopoietic-specific factors 6
hematopoietic stem cells (HSCs) 21–35
adeno-associated virus vectors 330–331
adenoviral vectors 329–330
cell-extrinsic regulators 24–29
cell-intrinsic regulators 24
cellular compartment model 21–23
egress 30–31
engraftment 29–30
functional analysis 33–34
gene therapy 319–334
isolation 31–34
lentiviral vectors 322–323, 327–329
lineage commitment 23–24
microenvironment 29
mobilization 31, 375
oncoretroviral vectors 321–327
primitive cell trafficking 29–31
regulation 24–29
sources 22
hematopoietic stem cell transplantation
(HSCT) 373–387
clonal deletion 377–378
Diamond–Blackfan anemia 144
dyskeratosis congenita 142
Fanconi anemia 140
graft-versus-host disease 377–386
graft-versus-leukemia 379, 383–386
homing and mobilization 375
idiopathic aplastic anemia 133
immune system reset 375–376
immunology 376–379
indications 373
microenvironment 374–375
non-immunological factors 374–375
peripheral tolerance 378–379
principles 373
space and competition 375
tumor-specific antigens 380–381
hematopoietic stem/progenitor cells (HSPCs)
29
hemochromatosis 167–170
ferroportin-associated 169
HFE-associated 167–169
juvenile 169
TFR2-associated 169
hemoglobin (Hb) 1–19
allostery 173–175
α thalassemias 9–12, 15–16
β thalassemias 7–9, 13–15
developmental changes 6
δβ thalassemias 12
gene action 4–5, 12–16
Gower 2
high oxygen affinity 188–189
Lepores 12
nitric oxide 278–279
oxygen dissociation curves 2–3
pathologies 6–18
Portland 2
prevention & treatments 18
regulation 3–4, 5–6
structural variants 6–7, 16–18, 173–191
structure and function 2–3, 173–175
synthesis 4–6
tetramers 2–3
unstable 9, 17, 188

hemoglobinopathies 173–191
with β thalassemias 184, 187
hemoglobin C 185
hemoglobin E 187
hemoglobin SC 185–186
high oxygen affinity 188–189
low oxygen affinity 189
M hemoglobins 189–190
sickle cell disease 16–17, 175–187
sickle/β thalassemia 184
unstable 9, 17, 188
hemojuvelin 169
hemolytic disease of the fetus and newborn
(HDFN) 273–277
frequency 276
Kell system 276
Rh antigens 273–276
treatment and prevention 277
hemophilia 221–234
antenatal diagnosis 229–230
A-type 223–225
B-type 225–226
carrier testing 228–229
clinical features 221–223
genetics 223
inhibitory antibodies 226–228
treatment 230–232
heparin 209, 218
hepcidin 162, 164–166
hephaestin 162
hereditary factors see genetics
hereditary persistence of fetal hemoglobin
(HPFH) 6–7, 12
Hermansky–Pudlak syndrome 259
herpes simplex virus thymidine kinase
(HSV-TK) 384
herpes virus vectors 322
HFE 166–169
HFE-associated hemochromatosis
167–169
2-HG see 2-hydroxyglutarate
high-grade B-cell lymphomas with MYC and
BCL2 or BCL6 translocations
104–106
high hyperdiploidy, acute lymphoblastic
leukemia 64–65
high oxygen affinity hemoglobins
188–189
high-throughput RNA sequencing (RNA-Seq)
103
histone arginine methyltransferases 89
histone modifications
cancer stem cells 368–369
myelodysplastic syndromes 55
HL see Hodgkin lymphoma
HLA see human leukocyte antigen
HNAs see human neutrophil antigens
Hodgkin lymphoma (HL)
genetics 108
Richter transformation 118–119
Hoechst 33342 31–32
homing, hematopoietic stem cells 29–30,
375
host cell adhesion, malaria 200–202
host ligands, malarial invasion 196–197
Hoyerall–Hreidarsson syndrome 140
HPA see human platelet antigen
HPFH see hereditary persistence of fetal
hemoglobin
HRD tumors see hyperdiploidy
HSCs see hematopoietic stem cells
HSCT see hematopoietic stem cell
transplantation
HSPCs see hematopoietic stem/progenitor
cells
HSV-TK see herpes simplex virus thymidine
kinase
Human Genome Project 1, 357
human immunodeficiency virus (HIV) 323,
327–329
human leukocyte antigen (HLA) genes
268–273
class I 271–272
class II 272
identification 268–271
stem cell transplantation 376–378
human neutrophil antigens (HNAs)
280–283
human platelet antigen (HPA) 267, 277–280
human studies, autoimmunity 306–307
hydroxycarbamide (hydroxyurea) 74, 82,
183
2-hydroxyglutarate (2-HG) 55–56
5-hydroxymethylfurfural (5-HMF) 183
hyperdiploidy
acute lymphoblastic leukemia 64–65
multiple myeloma 122–123
hypnozoites 196
hypodiploidy, acute lymphoblastic leukemia
65
hypoferremia, anemia of chronic disease 156
hypomethylation, acute myeloid leukemia
44–45
hypoxia
cancer stem cells 369
iron homeostasis 165
IAA see idiopathic aplastic anemia
ibrutinib 108, 116
ICAM see intercellular adhesion molecules
ICLs see intrastrand DNA cross-links
idelalisib 108
IDH1/2 44
acute myeloid leukemia 57
Index 395
myelodysplastic syndromes 55–56
myeloproliferative neoplasms 94–95
idiopathic aplastic anemia (IAA) 131–134
IFN-γ see interferon-γ
IGD see inherited
glycosylphosphatidylinositol
deficiency
IgH see immunoglobulin heavy chain
IL see interleukin
imatinib 63–64, 74–77, 79–81
side effects 81
immune thrombocytopenia (ITP) 297,
305–306, 308–311
immunity
adaptive 300–303
clonal deletion 377–378
complement system 298–299
innate 297–299
peripheral tolerance 302, 378–379
resetting 375–376
see also autoimmunity
immunoglobulin heavy chain (IgH)
transcription elements 104
immunoglobulin heavy chain (IgH)
translocations, multiple myeloma
122, 123
immunoglobulins
hemophilia 226–228
neonatal alloimmune neutropenia 282
immunoglobulin translocations, multiple
myeloma 122
immunology
graft-versus-host disease 379
stem cell transplantation 376–379
impaired immunity, chronic lymphocytic
leukemia 116–117
Indian blood group antigens 291–292
induced pluripotent stem cells (iPSCs) 21,
142
induction, graft-versus-leukemia 385–386
infections and autoimmunity 306, 313–314
inflammation
anemia of chronic disease 155–159
erythrocyte lifespan 158
erythropoietin resistance 158
iron absorption and release 158, 164–166
inherited bone marrow failure syndromes
138–144
Diamond–Blackfan anemia 142–144
dyskeratosis congenita 140–142
Fanconi anemia 138–140
inherited glycosylphosphatidylinositol
deficiency (IGD) 136–137
inhibitory antibodies, hemophilia 226–228
initiation codons 5
innate immunity 297–299
instructive model of hematopoiesis 23
396 Index
integrins
hematopoiesis 30–31
platelet disorders 255
intercellular adhesion molecules (ICAM)
201
interferon alpha, chronic myeloid leukemia
81–82
interferon-β (IFN-β), anemia of chronic
diease 158
interferon-γ (IFN-γ), anemia of chronic
diease 158
interferon-γ (IFN-γ) receptors, hematopoiesis
25
interleukin-3 alpha receptor (CD123)
365–366
interleukin (IL) receptors 25, 158, 365–366
internal tandem duplications (ITDs) 39,
41–42, 57
International Prognostic Scoring System
Revised (IPSS-R) 54
intervening sequences (IVSs) 3–4
intestinal iron transport 162
intracellular signaling, platelet disorders
257–258
intrastrand DNA cross-links (ICLs) 138–139
introns see intervening sequences
in vitro assays, hematopoietic stem cells 33
in vitro colony-forming cells (CFCs) 22–23.
See also colony-forming cells
in vivo assays, hematopoietic stem cells
33–34
iPCS see induced pluripotent stem cells
IPSS-R see International Prognostic Scoring
System Revised
iron chelation 143–144
iron metabolism 161–171
disorders 166–170
homeostatic regulation 164–166
storage 164
transport mechanisms 161–164
iron parameters, anemia of chronic disease
155
iron restriction, erythropoiesis 156–157
iron storage 164
iron transport 161–164
general principles 161–162
intestinal 162
transferrin 162–164
isocitrate dehydrogenases 44, 55–56, 57,
94–95, 369
isolation of hematopoietic stem cells
31–34
ITGAL 282
ITGAM 282
ITP see immune thrombocytopenia
IVSs see intervening sequences
Iwate hemoglobin 189
JAK2 mutations
cytokine signaling 87–89
myeloproliferative neoplasms 87–92
other 91–92
V617F 87–91
JAK homology domains (JH) 88–89
Janus kinase (JAK) inhibitors 385
JH see JAK homology domains
juvenile hemochromatosis 169
Kansas hemoglobin 189
karyotypes
acute lymphoblastic leukemia 60–61
hemophilia 223
International Prognostic Scoring System
Revised 54
multiple myeloma 122–123
myeloid malignancies 52–54
Kell system 276
KMT2A 39, 45, 64
Köln hemoglobin 188
K-RAS 125
laminins 292
LCA see Leber congenital amaurosis
LCR see locus control regions
Leber congenital amaurosis (LCA)
330
Leiden Factor V allele 211–215, 217
lentiglobin vectors, sickle cell disease
184
lentiviral vectors 322–323, 327–329
Lepores hemoglobin 12
leukemia inhibitory factor (LIF) 25
leukemic stem cells (LSC) 364–369
dormancy 367
surface antigens 365–366
life cycle, malaria 195–196
lineage commitment 23–24
lineage determination, cell-intrinsic 24
linkage disequilibrium 341
locus control regions (LCR) 4, 5–6
long-term culture-initiating cells (LTC-ICs)
22, 30
low oxygen affinity hemoglobins 189
LPL see lymphoplasmacytic lymphoma
LSC see leukemic stem cells
L-selectins, hematopoiesis 30
LTC-ICs see long-term culture-initiating
cells
Lutheran blood group antigens 292
lymphocytes
iron transport 162–163
see also B cell, T cell
lymphomas
common translocations 103–104
genetic immunotherapy 331–335
genetics 101–110
common 103–104
techniques 102–103
see also individual lymphoma types
lymphoplasmacytic lymphoma (LPL) 104,
106–107
lysosomal storage diseases (LSDs) 319,
327
macrophage colony-stimulating factor
(M-CSF) 23, 25
macrophages, iron transport 163–164
major histocompatibility complex (MHC)
300, 304–305, 376
malaria 193–206
anemia 202–204
epidemiology 193–195
erythropoiesis 204–205
host cell adhesion 200–202
invasion 196–200
life cycle 195–196
malaria resistance 17–18
MALT lymphomas see mucosa-associated
lymphoid tissue lymphomas
management
anemia of chronic disease 158–159
glucose-6-phosphate dehydrogenase
deficiency 149–150
graft-versus-host disease 384–385
hemoglobin disorders 18
hemophilia 230–231
pyruvate kinase deficiency 147–148
red cell enzymopathies 146
thrombophilia 217–218
see also treatment
mantle cell lymphoma 106
MAPK see mitogen-activated protein kinase
markers, primitive hematopoietic stem cells
31
mature cells 23
MAX 103
M-CSF see macrophage colony-stimulating
factor
MDS see myelodysplastic syndromes
MDSCs see myeloid-derived suppressor cells
MECOM activation 39
mental retardation, α thalassemias 11–12
merozoites 195–196
merozoite surface protein 1 (MSP-1)
199–200
mesenchymal stromal cells (MSCs) 385
messenger RNA (mRNA) 4–6, 356
β thalassemias 7–9
methylation 44–45, 55–56, 94–95
MHC see major histocompatibility complex
M hemoglobins 189–190
microarray analysis 360–361

microenvironmental niches
cancer stem cells 367–368
chronic lymphocytic leukemia 116
hematopoiesis 29
hematopoietic stem cells 374–375
multiple myeloma 127–128
micro RNA (miRNA) 103, 114–115, 356
minimal residual disease (MRD) monitoring
acute lymphoblastic leukemia 69–70
acute myeloid leukemias 45–46
miRNA see micro RNA
mitogen-activated protein kinase (MAPK)
88–89
mixed-lineage leukemia (MLL) 64, 66. See
also KMT2A
mixed lymphocyte reactions (MLRs) 379
MLL see mixed-lineage leukemia
MLLT3 39
MLRs see mixed lymphocyte reactions
MM see multiple myeloma
MMLV see Moloney murine leukemia virus
mobilization, hematopoietic stem cells 31,
375
molecular biology
chronic lymphocytic leukemia 111–120
history and development 353–362
multiple myeloma 121–130
molecular genetics see genetics
molecular medicine, definition 1
Moloney murine leukemia virus (MMLV)
321–327
monitoring
acute myeloid leukemias 45–46
chronic myeloid leukemia therapy 78–79
monoclonal gammopathy of undetermined
significance (MGUS) 121–130
transformations 125–126
see also multiple myeloma
monogenic inheritance 353–354
mouse models, autoimmunity 305–306
MPL mutations, myeloproliferative neoplasms
92
MPN see myeloproliferative neoplasms
MRD see minimal residual disease
mRNA see messenger RNA
MSCs see mesenchymal stromal cells
MSP-1 see merozoite surface protein 1
mucosa-associated lymphoid tissue (MALT)
lymphomas 107
multiple myeloma (MM)
chromosome alterations 122–123
classification 124–125
immunoglobulin translocations 122
karyotypic complexity 122
microenvironmental niche 127–128
molecular biology 121–130
neoplasm stages 121–122
pathogenensis 126–127
post-germinal center B cells 121
prognostic factors 124–125
transformative events 125–126
treatment 127–128
multiple sclerosis 315
multipotent stem cells 21. See also
hematopoietic stem cells
MYC oncogene 66–68, 103–105, 125–127
myelodysplastic syndromes (MDS) 12
cell signaling genes 56
cohesin complex 56
core concepts 49–50
cytogenic abnormalities 52–53
donor lymphocyte infusions 380
epigenetics 55–56
histone modifications 55
idiopathic aplastic anemia 133
International Prognostic Scoring System
Revised 54
methylation 55–56
recurrent mutations 54–56
splicing factor mutations 55
transcription factors 56
treatment selection 52–53
myeloid-derived suppressor cells (MDSCs)
385
myeloid malignancies
chromosomal translocations 51–52
diagnosis 50–51
International Prognostic Scoring System
Revised 54
karyotypes 52–54
recurrent mutations 54–57
risk assessment 51–52
therapy-related chromosomal aberrations
54
treatment selection 52–54
see also acute lymphoblastic leukemia, acute
myeloid leukemia, chronic myeloid
leukemia, myelodysplastic syndromes
myeloproliferative neoplasms (MPN) 87–99
CALR mutations 93–94
chronic neutrophilic/eosinophilic leukemia
95
diagnosis 95–96
JAK2 mutations
other 91–92
V617F 87–91
methylation disorders 94–95
microenvironmental niche 367–368
MPL mutations 92
splicesome disorders 95
splicing factor mutations 55
NAITP see neonatal alloimmune
thrombocytopenia
Index 397
NCOA4 see nuclear receptor coactivator 4
neomycin phosphotransferase 324
neonatal alloimmune neutropenia 280–283
neonatal alloimmune thrombocytopenia
(NAITP) 277–280
neovascular age-related macular degeneration
(NVAMD) 329
next-generation sequencing (NGS) 40–45,
51, 103, 271, 361–362
NFκB pathway
multiple myeloma 126
targeting 366–367
NGS see next-generation sequencing
NHRD tumors see non-hyperdiploid tumors
nilotinib 74–77, 79–81
side effects 76
nitric oxide (NO) 287–289
non-deletion α thalassemias 11
non-heme iron absorption 162
non-Hodgkin lymphoma (NHL) 105
non-hyperdiploid (NHRD) tumors 122–123
non-infectious triggers of autoimmunity
306–307
nonsense-mediated decay 9
non-transferrin-bound iron (NTBI) 163
Northern blotting 359
NOTCH1 66–68, 366
NPM 1 see nucleophosmin 1
N-RAS 125
NT5C2 347
NT5C3 150
NTBI see non-transferrin-bound iron
N-terminal cytoplasmic domain of AE1
(CDAE1) 286–289
nuclear receptor coactivator 4 (NCOA4) 164
nucleic acid extraction 358
nucleophosmin 1 (NPM 1) 40–41, 57, 107
nucleosomes 357
NUDT15 346–347
numerical chromosomal abnormalities
acute myeloid leukemia 39–40
multiple myeloma 122–123
NVAMD see neovascular age-related
macular degeneration
omacetaxine 74–75, 81
oncoretroviral vectors 321–327
oxygen dissociation curves 2–3
oxy(R) hemoglobin 2
P2RY8-CRLF2 65
P2Y receptor 257
p53 43, 56
pancytopenia 131–133
paroxysmal nocturnal hemoglobinurua
(PNH) 134–136
partial tandem duplications (PTDs) 45
398 Index
pathogenesis
anemia of chronic disease 156–158
graft-versus-host disease 382–383
malarial anemia 203–204
pathologies, hemoglobin 6–18
pathophysiology
α thalassemias 15–16
autoimmunity 308
β thalassemias 13–15
Diamond–Blackfan anemia 142–143
dyskeratosis congenita 140–141
glucose-6-phosphate dehydrogenase
deficiency 149
hemolytic disease of the fetus and newborn
277
idiopathic aplastic anemia 131–133
malaria 193–206
multiple myeloma 126–127
paroxysmal nocturnal hemoglobinurua
134–136
platelet disorders 257–261
pyruvate kinase deficiency 146–147
red cell enzymopathies 144–146
sickle cell anemia 180–183
PCM1-JAK2 95
PCR see polymerase chain reaction
PCRB see poly(rC)-binding proteins
PCR-SSOP see polymerase chain
reaction-sequence-specific
oligonucleotide probes
PCR-SSP see polymerase chain
reaction-sequence-specific primers
PDGFRA 95
PDGFRB 95
PECAM see platelet endothelial cell
adhesion molecule
perforin-like protein 1 (PfPLP1) 195
peripheral tolerance 302, 378–379
permissive model of hematopoiesis
23–24
PfEMP1 see Plasmodium falciparum
erythrocyte membrane protein 1
PfPLP1 see perforin-like protein 1
PfSPECT see sporozoite microneme protein
essential for cell traversal
pharmacogenomics 339–352
childhood acute lymphoblastic leukemia
342–348
clopidogrel 350
codeine 341
coumarins 348–350
oral antithrombin therapy 348–350
principles 340–342
rasburicase 345
thiopurines 345–347
phenotypes
α thalassemias 9–12, 15–16
β thalassemias 9, 13–15 polymerase chain reaction-sequence-specific
JAK2V617F myeloproliferative neoplasms primers (PCR-SSP) 270–271
89–91 poly(rC)-binding proteins (PCRB) 163
phenotypic diversity, β thalassemias 14–15 ponatinib
Philadelphia chromosome 63–64 chronic myeloid leukemia 74–75, 81,
chronic myeloid leukemia 73–74 83
Philadelphia chromosome-negative chronic side effects 81
myeloid leukemia 83 Poole hemoglobin 188
Philadelphia-like acute lymphoblastic Portland hemoglobin 2
leukemia 68 positional cloning 1
phosphatidylinositol 3-kinase (PI3K) 88–89 positive selection, T cells 302
Ph-pos ALL see Philadelphia chromosome post-germinal center B cells, multiple
physical isolation, hematopoietic stem cells myeloma 121
31 pox virus vectors 322
PICALM-MLLT10 (CALM-AF10) 66 precursor cells 23
PIGA 133–136 pregnancy, tyrosine kinase inhibitors 82
PIGM 137 prekallikrein 207
PK see pyruvate kinase pre-leukemic stem cells 364–365
plasma cell neoplasm stages 121–122 prevention
plasmacytoma (PCT) 121–122 hemoglobin disorders 18
Plasmodium falciparum erythrocyte hemolytic disease of the fetus and newborn
membrane protein 1 (PfEMP1) 277
200–202 malaria 200
Plasmodium falciparum erythrocyte Protein 1 primary myelofibrosis (PMF) 87–96
(PfEMP1) 17–18 CALR mutations 93–94
Plasmodium Spp. see malaria JAK2 mutations, other 91–92
platelet disorders 251–265 JAK2V617F mutation 89–91
adhesion-related 253–256 MPL mutations 92
agonist receptor defects 257 PRMT5 89
collagen receptor defects 255–256 progenitor cells 23
cytoskeletal 260 prognostic factors
diagnosis 252–253 chronic lymphocytic leukemia 117–118
Glanzmann thrombasthenia 256–257 multiple myeloma 124–125
intracellular signaling 257–258 see also risk assessment
pathophysiology 257–261 prognostic models, chronic myeloid leukemia
polymorphisms 260–261 73
procoagulant regulation 259–260 programmed cell death ligands, chronic
risk assessment 260–261 lymphocytic leukemia 117
storage pool defects 258–259 promoter elements, hemoglobin 4
transcription factors 260 promyelocytic leukemia (PML), retinoic acid
von Willebrand disease 254–255 receptor fusion 38–39, 52
platelet endothelial cell adhesion molecule prophylactic platelet transfusions 273
(PECAM) 201 protein C 208–211
platelets, normal function 251–252 activated 210–213
pluripotent stem cells 21 anticoagulation system 209–211
PNH see paroxysmal nocturnal deficiencies 214–215
hemoglobinurua protein S 211, 215
polyA 5 protein serine-threonine kinase receptors
polycythemia vera (PV) 87, 94–96 25–26
polygenic inheritance 353–354 proteomics 1–2
polymerase chain reaction (PCR) 102, prothrombin 207, 215–217
359–360 PRPS1 347–348
HLA genes 268–271 P-selectins 30
myeloid malignancies 51 psoriasis 315
polymerase chain reaction-sequence-specific PTDs see partial tandem duplications
oligonucleotide probes (PCR-SSOP) PTEN, acute lymphoblastic leukemia
268–270 66–67

pyrimidine 5ʹ-nuclotidase deficiency
150–151
pyruvate kinase (PK) deficiencies 146–148
Quebec platelet disorder 259
RARS see ring sideroblasts
rasburicase pharmacogenomics 345
RAS pathway, multiple myeloma 125
RB see retinoblastoma, ribosome proteins
RBCs see red blood cells
RBM15-MKL1 39
RBP see reticulocyte-binding proteins
receptor tyrosine kinases 25
recombinant blood products, hemophilia
230–231
recurrent mutations
acute myeloid leukemia 56–57
myelodysplastic syndromes 54–56
red blood cells (RBCs)
adhesion molecules 291–293
anion-exchanger protein 1 285–290
glycophorin C/D 290–291
malarial invasion 196–200
Rh antigens 289–290
red cell enzymopathies 144–151
glucose-6-phosphate dehydrogenase
148–150
pyrimidine 5ʹ-nuclotidase 150–151
pyruvate kinase 146–148
regulation
blood coagulation 209
hematopoiesis 24–29
hemoglobin synthesis 3–4, 5–6
transcription 356–357
relapses
acute lymphoblastic leukemia 68
post-transplantation chronic myeloid
leukemia 82–83
reporter genes 324
resistance
malaria 17–18
thrombophilia, APC 212–213
restriction enzymes 358
retention 29
reticulocyte-binding proteins (RBP) 199
retinoblastoma (RB) 106, 125
retinoic acid receptor (RARα) 38–39, 52
reverse genetics 1
Rh-123 see rhodamine- 123
Rh see rhesus antigens
R hemoglobin see oxy hemoglobin
rhesus (Rh) antigens 273–277, 289–290
rheumatoid arthritis 315
rhodamine-123 (Rh-123) 32
rhoptry proteins 199
ribosomal RNA (rRNA) 356
ribosome proteins (RB), Diamond–Blackfan
anemia 142–143
Richter transformation (RT) 118–119
ring sideroblasts (RARS) 55
RIPA assay see ristocetin-induced platelet
agglutination assay
risk assessment
chronic lymphocytic leukemia 117–118
hemolytic disease of the fetus and newborn
277
myeloid malignancies 51–52
neonatal alloimmune thrombocytopenia
279–280
platelet disorders 260–261
ristocetin-induced platelet agglutination
(RIPA) assay 238
rituximab 105, 312–313
rivipansel 183
RNA polymerase 4–5
RNA-Seq see high-throughput RNA
sequencing
romiplostim 314
RPS14 39
RPS19 142–143
rRNA see ribosomal RNA
RT see Richter transformation
RUNX1 38, 42, 52, 57, 65
St Mande hemoglobin 189
SCD see sickle cell disease
SCID see severe combined
immunodeficiency
SCL44A2 282
SCL 24, 66
Scott syndrome 259–260
SCT see stem cell transplantation
secondary acute myeloid leukemia 54
selectins, hematopoiesis 30
selective induction, graft-versus leukemia
385–386
self-renewal 21, 24, 366
sequencing-based typing 271
sequencing technology development
359–362
serum ferritin, anemia of chronic disease
155–156
severe combined immunodeficiency (SCID)
326–327
SF3B1 55, 95
SFKs see SRC family of kinases
sialic acid ligands, malaria 197
sickle cell disease (SCD) 16–17, 175–187
clinical presentation 177–180
epidemiology 176, 184–186
genetics 175–177, 186–187
hemoglobin C 185
hemoglobin SC 185–186
Index 399
pathophysiology 180–183
treatment 183–184
side effects, tyrosine kinase inhibitors 76, 81
siderosis 170
SIL-TAL1 chimeric gene 66
single-nucleotide polymorphisms (SNPs) 2
adverse drug reactions 344
arrays 102–103
JAK2V617F myeloproliferative neoplasms
90–91
myeloid malignancies 51
pharmacogenomics 340–341, 344
small nuclear RNAs (snRNAs) 356
SMM see smoldering multiple myeloma
smoldering multiple myeloma (SMM)
121–122
S-nitrothiols (SNOs) 287–289
SNPs see single-nucleotide polymorphisms
somatic mutations
chronic lymphocytic leukemia 113–114
pharmacogenomics 341–342
sources, stem cells 22
Southern blotting 1, 359
spliceosome 43–44, 95
splice site consensus sequences 8
splicing factors 55
sporozoite microneme protein essential for
cell traversal (PfSPECT) 195
SRC family of kinases (SFKs) 74
SRSF2 43–44, 45, 55, 95
STAT5 signaling 88–89
stem cells 21–35
cancerous 363–372
cell of origin 364
dormancy 367
epigenetics 368–369
microenvironment 367–368
pre-leukemic 364–365
surface antigens 365–366
targeted therapies 365–369
cell of origin 364
cellular compartment model 21–23
clinical usage 31–34
definitions and distinctions 21
lineage commitment 23–24
pre-leukemic 364–365
sources 22
see also cancer stem cells, hematopoietic
stem cells
stem cell transplantation (SCT)
chronic myeloid leukemia 82–83
see also hematopoietic stem cell
transplantation
stereotype receptors 113
stochastic theory of hematopoiesis 23
storage of iron 164
storage of platelets 258–259
400 Index

Stormorken syndrome 260 TGF-β see transforming growth factor-β
structurally abnormal globin chains 6–7, thalassemias 2
16–18, 173–191 α-type 9–12, 15–16
β thalassemias 184, 187 β-type 7–9, 13–15, 184, 187
hemoglobin C disease 185 with hemoglobin E 187
hemoglobin SC disease 185–186 with sickle cell anemia 184
high oxygen affinity 188–189 T hemoglobin see deoxy hemoglobin
low oxygen affinity 189 therapy see treatment
M hemoglobins 189–190 therapy-related chromosomal aberrations 54
sickling 175–187 thiopurine pharmacogenomics 345–347
unstable 9, 17, 188 thrombin 210
structure thrombocytopenia
hemoglobin 2–3 immune 297, 305–306, 308–311, 313–314
von Willebrand factor 236 neonatal alloimmune 277–280
suicide gene therapy 334–335 thrombomodulin 209
supravital stains 31–32 thrombophilia 207, 211–218
SUSTAIN study 183 antithrombin deficiencies 213–214
symptomatic myeloma 121–122 genetics 211–217
synthesis, hemoglobin 4–6 management 217–218
multifactorial nature 217
t(5;14)(q35;q32) acute lymphoblastic leukemia protein C deficiencies 214–215
66 protein S deficiencies 215
t(12;21)(p13;q22) acute lymphoblastic prothrombin gain of function 215–217
leukemia 65 thrombopoietin receptor agonists 133, 314
T see thymine thrombopoietin receptors, myeloproliferative
TAL1 66 neoplasms 92
TALENs see transcription activator-like thrombopoietin (TPOR), JAK2 signaling
effector nucleases 88–89
T ALL see T cell acute lymphoblastic thrombospondin-related adhesive protein
leukemia (TRAP) 195
targeted therapies, cancer stem cells 365–369 thymine (T) 5
T cell acute lymphoblastic leukemia (T ALL) thymus 301–303, 377–378
65–68 TICs see tumor-initiating cells
T cell receptors (TCR) 300–303, 331–333 tigecycline 369
T cell selection 301–303, 377–378 tissue factor pathway inhibitor (TFPI) 209
T cell trafficking, graft-versus-host disease TKD see tyrosine kinase domains
383 TLX3 dysregulation 66
TCF3 105 TNF-α see tumor necrosis factor-α
TCF3-HLF 348 TNFR see tumor necrosis factor receptor
TCL5 66 tolerance, transplantations 377–378
TEL 65 toxicities, tyrosine kinase inhibitors 81
telomerase, dyskeratosis congenita 140–142 TP53 see p53, tumor protein 53
telomerase reverse transcriptase (TERT) TPMT 342, 345–346
140–142 TpoR 133
telomerase RNA component (TERC) TPOR see thrombopoietin
140–142 trafficking
TERC see telomerase RNA component graft-versus-host disease 383
(TERC) hematopoiesis 29–31
termination codons 5 transcription 356
TERT see telomerase reverse transcriptase β thalassemias 7–9
TET2 see Tet methylcytosine dioxygenase 2 globin genes 4–5
Tet methylcytosine dioxygenase 2 (TET2) regulation 356–357
45, 55–56, 94–95 transcription activator-like effector nucleases
tetramers, hemoglobin 2–3 (TALENs) 320
TFPI see tissue factor pathway inhibitor transcription factors 6, 357
TFR2-associated hemochromatosis 169 acute myeloid leukemia 42–43
TFRs see transferrin receptors hematopoiesis 24
myelodysplastic syndromes 56
platelet disorders 260
transcriptome sequencing 361–362
transfection 359
transferrin 155, 162–164, 170
transferrin receptors (TFRs) 163–164, 199
transfer RNA (tRNA) 5, 356
transforming growth factor-β (TGF-β)
25–26
transfusions
Diamond–Blackfan anemia 143–144
hemophilia 230–231
platelets 273
sickle cell anemia 183
von Willebrand disease 245–246
translation 9, 356
TRAP see thrombospondin-related adhesive
protein
treatment
acute myeloid leukemia 45–46
anemia of chronic disease 158–159
autoimmunity 311–315
cancer stem cells 365–369
chronic myeloid leukemia 74–83
Diamond–Blackfan anemia 143–144
dyskeratosis congenita 142
Fanconi anemia 140
graft-versus-host disease 384–385
hemoglobin disorders 18
hemolytic disease of the fetus and newborn
277
hemophilia 230–232
idiopathic aplastic anemia 133
immune thrombocytopenia 311
multiple myeloma 127–128
myeloid malignancies 52–54
neonatal alloimmune thrombocytopenia
280
paroxysmal nocturnal hemoglobinurua
136
sickle cell anemia 183–184
thrombophilia 217–218
see also management
treatment endpoints, chronic myeloid
leukemia 77–78
treatment failure, chronic myeloid leukemia
78–79
treatment value, chronic myeloid leukemia
77–78
trisomies, multiple myeloma 122–123
trophozoites 196
TSAs see tumor-specific antigens
tumor-initiating cells (TICs) 363–364
tumor necrosis factor receptor (TNFR) 29
tumor necrosis factor-α (TNF-α) 158, 315
tumor protein 53 (TP53) 43, 56, 126
tumor-specific antigens (TSAs) 380–381

tumor suppressors, acute myeloid leukemia
43
type 1 diabetes, genetics 304
type I cytokine receptors, hematopoiesis
24–25
type II cytokine receptors, hematopoiesis 25
tyrosine kinase domains (TKD), FLT3-ITD
39, 41–42, 57
tyrosine kinase inhibitors (TKIs)
acute lymphoblastic leukemia 63–64
anaplastic large cell lymphoma 107
cancer stem cells 369
chronic myeloid leukemia 74–81
choice 80–81
drug dose schedules 79–81
endpoints 77–78
monitoring 78–79
diffuse large B-cell lymphoma 105
pregnancy 82
side effects 76, 81
tyrosine kinases, JAK2 mutations 87–92
U2AF1 55, 95
U see uracil
ultra-deep sequencing 361
unstable variants of hemoglobin 9, 17, 188
uracil (U) 5
vascular cell-adhesion molecule-1 (VCAM-1)
30–31, 368
vectors 321–331
adeno-associated 322, 330–331
adenoviral 322, 329–330
lentiviral 322–323, 327–329
oncoretroviral 321–327
venetoclax 116
venous thromboembolism
antithrombin deficiencies 213–214
genetics 211–217
management 217–218
multifactorial nature 217
protein C deficiencies 214–215
protein S deficiencies 215
prothrombin gain of function 215–217
very late antigen 4 (VLA4) 368
vitamin K 209–211, 349–350
VKORC1 and coumarins 349–350
VLA4 see very late antigen 4
von Willebrand disease (VWD) 235–249
classifications 236–238
diagnosis 238
factor VIII 236, 238, 243–245
genetics 238–244
platelet disorders 254–255
treatment 244–246
Index 401
type 1 240–241
type 2a 241–242
type 2b 242–243
type 2M 243
type 2N 243
type 3 243–244
von Willebrand factor (VWF) 210,
235–249
gene organization 236
structure and function 235–236
VRX496 328
Waldenstrom macroglobulinemia (WM)
107
warfarin metabolism 349
WAS see Wiskott–Aldrich syndrome
WES see whole-exome sequencing
whole-exome sequencing (WES) 103
Wilms tumor 1 (WT1) 43
Wiskott–Aldrich syndrome (WAS) 259
WM see Waldenstrom macroglobulinemia
WT1 see Wilms tumor 1
ZBTB16 38
zinc finger nucleases (ZFN) 320
zinc protoporphyrin 156
Zurich hemoglobin 188