PERS PE C T IV E
From the Veterans Affairs Boston Health 1. Cross SH, Warraich HJ. Changes in the
Care System, Section of General Internal
Medicine (M.W.W.), Brigham and Women’s
Hospital (M.W.W., R.S.S., J.S.W.), and the
Dana–Farber Cancer Institute, Department
of Psychosocial Oncology and Palliative Care
(M.W.W.) — all in Boston; and the Institute
for Health, Health Care Policy, and Aging Re-
search and the Department of Family Medi-
cine and Community Health, Rutgers Uni-
versity, New Brunswick, NJ (E.A.L.).
This article was published on March 12,
2022, at NEJM.org.
The Patient Resident
The Patient Resident
Natasha Z. Rabinowitz Steele, M.D., M.P.H.​​
I slide open the door and slip
into my patient’s room. Mr. C.
is a 64-year-old army veteran
who presented several days ago
with massive hemoptysis. I say
those words aloud each morn-
ing on rounds, but they hardly
describe him.
I learned intimate details about
his body from his laboratory, im-
aging, and pathology reports.
But I only came to know him
through the hours at his bedside.
The woman he loved was work-
ing in the artichoke fields and
struggling to learn English. He
was proud of his three sons,
scattered around the country. Af-
ter a career in combat, he was a
God-fearing man, but he feared
death most of all.
When I first examined him, I
suspected he had cancer. He was
cachectic, and I could feel the
uneven contours of his organs
beneath my fingers. Now, I’ve
come to share the results of the
biopsy.
“There she is,” says Mr. C.
with a smile, pulling himself up
in bed. As his intern, I am a famil-
iar face amidst the ever-changing
personnel.
“I came by to talk about your
1010 n engl j med 386;11  nejm.org  March 17, 2022
The New England Journal of Medicine
Downloaded from nejm.org by Salvador Macias-Díaz on March 23, 2022. For personal use only. No other uses without permission.
Copyright © 2022 Massachusetts Medical Society. All rights reserved.
Los Angeles Times. February 16, 2020
place of death in the United States. N Engl J
Med 2019;​381:​2369-70.
2. Gerrard R, Campbell J, Minton O, et al.
Achieving the preferred place of care for
hospitalized patients at the end of life. Pal-
liat Med 2011;​25:​333-6.
3. Barnato AE, Anthony DL, Skinner J,
Gallagher PM, Fisher ES. Racial and ethnic
differences in preferences for end-of-life
treatment. J Gen Intern Med 2009;​24:​695-
701.
4. Gray N. Think you want to die at home?
You might want to think twice about that.
biopsy,” I say gently. Our eyes
meet, and my body language gives
me away. His eyes begin welling
with tears.
“It’s cancer, isn’t it?” I nod.
His eyes slowly close as the
news sinks in. A monitor alarms
as his heart rate soars. I silence
it — I don’t need technology to
tell me he’s terrified. I gaze down
at his trembling hands — the
hands that once held his chil-
dren, wiped tears from his wife’s
face, that traced the sign of the
cross each Sunday at Mass.
We sit in silence. It’s a loud
silence, full of racing fears of the
unknown. It’s that moment when
everything has changed for him:
he has gone from before to after.
I know this moment all too well.
I look at my own hands —
hands that a year prior first held
my newborn daughter. Hands that,
soon after, clutched my phone as
I read my own radiology report:
“Mediastinal lymphadenopathy
consistent with lymphoma. Bilat-
eral pleural effusions.” I remem-
ber the room started to spin but
I felt paralyzed; then paralysis gave
way to sobs. “I don’t want to
leave her,” I repeated to my hus-
band over and over. Panicked, I
Where Americans Die
(https://www​.­latimes​.­com/​­opinion/​­story/​­2020​
-­02​-­16/​­doctor​-­patients​-­send​-­home​-­to​-­die).
5. Evans WG, Cutson TM, Steinhauser KE,
Tulsky JA. Is there no place like home?
Caregivers recall reasons for and experi-
ence upon transfer from home hospice to
inpatient facilities. J Palliat Med 2006;​ 9:​
100-10.
DOI: 10.1056/NEJMp2112297
Copyright © 2022 Massachusetts Medical Society.
Where Americans Die
called my father, a gentle, kind
family doctor. “Dad . . . do I
have cancer?” I asked, catching
my breath — the words sounded
surreal as I heard them out loud.
My hands held my eyes shut as
he promised he’d be on the next
flight.
Earlier on the day of my diag-
nosis, I’d rounded as a new in-
tern on the cardiology service. I
had brushed off a nagging cough
and weakness: I’d expected to
feel run down after my cesarean
delivery, so I pushed through. I
was doubled over in pain by noon
conference, but it passed. I learned
in medical school to work through
discomfort, prioritize patient care
over my own body, and see pa-
tients and doctors as inherently
different. It didn’t dawn on me
that I could be gravely ill. Then a
rubbery supraclavicular node ap-
peared, and I couldn’t ignore it.
Just 2 weeks into my intern-
ship, the disparate identities of
physician, mother, and cancer
patient had collided inside me. I
didn’t know who I should be at
any given moment. I had expect-
ed to learn medicine by observing
and caring for patients; instead, I
learned about chemotherapeutics,
PE R S PE C T IV E
antiemetics, and granulocyte col-
ony-stimulating factors firsthand.
At a time when I was just begin-
ning a life of healing by day and
parenting by night, I suddenly
became a patient in my own hos-
pital, at the mercy of my mercu-
rial biology.
I was shaking as I removed
my crisp white coat and climbed
into a hospital bed, donning a
patient gown for the first time.
The radiology report had prompt-
ed an echocardiogram that re-
vealed cardiac tamponade — my
heart was straining to pump
blood past an overwhelming tu-
mor burden in my chest. That
evening, I lay awake, remember-
ing the patients I’d rounded on
that morning. It struck me that I
was sicker than any of them. On
the verge of cardiac collapse, I
felt terrifyingly mortal.
After I left the hospital, time
slowed. For years, I had been in a
rush — working to get into
medical school; withstanding the
pressure to survive classes; tran-
sitioning to wards while trying to
maintain my marriage and start a
family; prepping for board exams
and residency interviews. Now,
the cadence of my life followed a
series of 2-week cycles centered
around chemotherapy. My hair fell
out in fistfuls, my postpartum
skin hung off my aching bones,
my breast milk vanished. I feared
holding my daughter too close,
wanting to keep her safe from
the war raging in my body. When
I did hold her, I felt mind-expand-
ing love laced with an over-
whelming fear of loss. She’d
scrunch up her face and give me
a smile identical to mine at her
age, evoking a time when I was
naive to this pain. She was just
starting her life, and I was fight-
ing for mine.
n engl j med 386;11  nejm.org  March 17, 2022
The New England Journal of Medicine
Downloaded from nejm.org by Salvador Macias-Díaz on March 23, 2022. For personal use only. No other uses without permission.
Copyright © 2022 Massachusetts Medical Society. All rights reserved.
Over the next year, I under-
went 12 chemotherapy treatments,
two hospitalizations, two pericar-
diocenteses, four PET scans, and
innumerable blood tests, echo-
cardiograms, and x-rays. Being
sick was a full-time job. Each day
felt like free soloing a cliff: at
any moment, bad news from a lab
or scan could hasten the abyss.
On my last day of treatment,
there was no bell to ring, no
celebratory singing. My hospital
bracelet was clipped off, and I
was discharged with instructions
to return in a few weeks for a fi-
nal scan. The results came swiftly
in a phone call. I was cancer-
free. But being cured was only
the start of healing.
The following day, the most
glaring remnant of my disease,
the long central catheter to my
right atrium, was removed. My
body was returned to me, no lon-
ger merely a vehicle for fighting
a malignant force. But was it even
still me? My oncologist cleared
me to return to residency. No
longer sick but far from well, I
stumbled through my first steps
of survivorship.
I was eager to be on the wards
in this time of Covid — I had
patiently awaited the chance to
match the tempo and bravado of
my coresidents, chasing the diag-
noses and prognoses of others.
But internally, I was struggling
with the now-porous partition
between doctor and patient. Every
patient reminded me of some as-
pect of my own illness. Their pri-
mal fear was familiar, and the
scent of their hospital gowns
evoked memories of my own.
The hospital was full of mean-
ingful landmarks. One day I
paused at the doorway of a pa-
tient with breast cancer, remem-
bering when I’d been hospital-
The Patient Resident
ized in the same room. I caught
a glimpse of myself in the mirror
where I’d once stared at my
sunken face; now a rounder face
looked back at me. I felt my fear
return. Would I get sick again? I
had shed the cloak of cancer, yet
I still carried its weight.
I stopped rushing through my
ward work. When patients spoke,
I slowed to listen. I pledged to be
their ally in a journey through
the unknown, a journey I share.
Slowly, my identities as cancer
patient, doctor, and parent began
to complement one another, their
sum greater than the parts.
The alarm sounds again and
startles me out of my reverie. I
squeeze Mr. C.’s hand, and he
squeezes back. We sit that way
for a long time.
Later that night I’m at the din-
ner table with my family, marking
the first anniversary of my diag-
nosis. My husband puts his arm
around me, and I begin to cry.
Our daughter, just learning to
walk, comes to me. She reaches
for my face, gently touching my
tears in wonderment. Tears that
fall because I’m still here, because
I’m well, because I can witness
my daughter just beginning her
journey. I think about Mr. C., re-
calling the tears and the touch of
someone who is nearing the end
of his own. Though all of our
journeys will have beginnings and
endings, our lives are what we
choose to do with the precious,
unpredictable, terrifying, and
beautiful moments in between.
Identifying details have been changed to
protect the patient’s privacy.
Disclosure forms provided by the author
are available at NEJM.org.
From Stanford Medicine, Stanford, CA.
This article was published on March 12,
2022, at NEJM.org.
DOI: 10.1056/NEJMp2116289
Copyright © 2022 Massachusetts Medical Society.
The Patient Resident
1011
Radiología. 2014;56(3):193---205

www.elsevier.es/rx

RADIOLOGÍA HOY

RECIST y el radiólogo
J. Cervera Deval

Servicio de Radiología, Fundación Instituto Valenciano de Oncología, Valencia, España

Recibido el 25 de septiembre de 2011; aceptado el 12 de marzo de 2012

Disponible en Internet el 16 de agosto de 2012

PALABRAS CLAVE Resumen Mediante las imágenes biomédicas podemos diagnosticar, estadificar, controlar y
RECIST; valorar la efectividad de los procedimientos terapéuticos a los que son sometidos los pacientes
Respuesta del tumor; oncológicos. RECIST (Response Evaluation Criteria In Solid Tumors) es el método para monitori-
Estudios de zar el tratamiento mediante medidas unidimensionales de los tumores, obtenidas con técnicas
seguimiento; de imagen reproducibles como son la TC, la RM y la PET. Los cambios metabólicos inducidos por
Oncología; los nuevos tratamientos modifican la biología y comportamiento del tumor, con discordancia
Imagen diagnóstica; ocasional entre el estado clínico del paciente y la respuesta medida con RECIST, lo que apunta
PET; a la necesidad de incluir pruebas funcionales en la valoración de la respuesta al tratamiento.
Técnicas de El objetivo es revisar los criterios RECIST conociendo la aportación de la imagen funcional
diagnóstico para valorar correctamente la eficacia y la repercusión del tratamiento en los pacientes con
molecular; tumores sólidos.
Tumores sólidos © 2011 SERAM. Publicado por Elsevier España, S.L. Todos los derechos reservados.

KEYWORDS RECIST and the radiologist

RECIST;
Tumor response; Abstract Biomedical imaging makes it possible not only to diagnose and stage cancer, but also
Follow-up studies; to follow up patients and evaluate the response to treatment. RECIST (Response Evaluation
Oncology; Criteria In Solid Tumors) provides a method to monitor the response to treatment based on one
Diagnostic imaging; dimensional measurements of tumors obtained with reproducible imaging techniques like CT,
PET; MRI, and PET. The metabolic changes induced by new treatments modify the biology and beha-
Molecular diagnostic vior of the tumor; occasionally, there is a discrepancy between the patient’s clinical condition
techniques; and the response measured by RECIST, which indicates that functional tests need to be included
Solid tumors in the evaluation of the response to treatment.
The objective is to review the RECIST criteria to include the contribution of functional imaging
to enable the efficacy and effects of the treatment in patients with solid tumors.
© 2011 SERAM. Published by Elsevier España, S.L. All rights reserved.

Correo electrónico: jcervera@Fivo.Org

0033-8338/$ – see front matter © 2011 SERAM. Publicado por Elsevier España, S.L. Todos los derechos reservados.
http://dx.doi.org/10.1016/j.rx.2012.03.010
194 J. Cervera Deval

Introducción
La RAE define «respuesta» como «efecto que se pretende
conseguir con una acción»1 . En el ámbito de la oncología
y la radiología, se trata de medir la respuesta a los tra-
tamientos con biomarcadores de imagen y, así, mantener
un tratamiento o suspenderlo para evitar posibles toxicida-
des y gastos innecesarios. La respuesta al tratamiento de
los tumores sólidos clásicamente se basa en los cambios
de tamaño, elemento base de los criterios RECIST (Res-
ponse Evaluation Criteria In Solid Tumors)2 . Estos cambios
intentan representar el comportamiento biológico de los
tumores, lo que ocasionalmente crea controversia porque,
en los pacientes tratados con nuevos fármacos, no siempre
un aumento del tamaño significa progresión3 .
En ensayos clínicos se han usado diferentes criterios con
el objetivo de normalizar la valoración de la respuesta al tra-
tamiento, elemento fundamental para poder comparar los
efectos de los tratamientos. Más aún cuando se desarrollan
nuevos fármacos y terapias costosas para mejorar la super-
vivencia, y cuyos objetivos son 1) la mejoría clínica, por la
reducción de la carga tumoral, o 2) la curación, por la desa-
parición completa del tumor. El propósito de esta revisión
es repasar los criterios RECIST 1.1 e introducir los métodos
de valoración con imagen funcional en los tratamientos con
los nuevos quimioterápicos.

Criterios RECIST 1.1

A finales de los 90, la OMS4 unifica la valoración de la res-

puesta y de la recidiva, el intervalo libre de enfermedad, y
establece una graduación de la toxicidad del tratamiento.
En el año 2000 aparecen los criterios RECIST2 como modifi-
cación de los criterios de la OMS, actualizados en 2009 a la
versión 1.1 (http://www.recist.com) (fig. 1)5 , que incluye
la resonancia magnética (RM) y la PET con 18 F-FDG como
modalidades de imagen aceptadas para valorar la respuesta6
(tabla 1). Modificaciones importantes que la actualización
1.1 aporta son la reducción del número de lesiones diana
(máximo de 5, solo 2 por órgano); el límite de 5 mm, por
debajo del cual las lesiones no se consideran medibles; la
inclusión de las adenopatías como lesiones diana cuando su
eje corto sea mayor de 15 mm7 ; y la mención especial a las
lesiones óseas, a las lesiones quísticas y a las lesiones previa-
mente irradiadas. El rastreo óseo, la PET o las radiografías no
son adecuados para medir las lesiones óseas, aunque pueden
ser útiles para confirmarlas o determinar su desaparición.
Las metástasis líticas o mixtas líticas-blásticas son medibles
cuando el componente de partes blandas cumpla criterios
de lesión medible. Las blásticas no son medibles (fig. 2).
Las lesiones quísticas con criterios radiológicos de quiste
simple no son medibles porque no se consideran malignas.
Las que correspondan a metástasis quísticas pueden consi-
derarse medibles si cumplen los criterios de lesión medible.
No obstante, cuando existan otras lesiones no quísticas serán
estas las seleccionadas como lesiones diana. RECIST no con-
templa los cambios de densidad (TC) o de intensidad de señal
(RM) durante el seguimiento. Las lesiones localizadas en
áreas previamente irradiadas o sometidas a cualquier tera-
pia locorregional, deben considerarse lesiones no medibles,
a menos que se demuestre su progresión.
Figura 1 Imagen axial de TC sin contraste endovenoso.
Adenocarcinoma de pulmón. (A) Medida de los diámetros
perpendiculares máximos, OMS. (B) Medida unidimensional
máxima, RECIST. La suma de los diámetros máximos de hasta 5
lesiones y solo 2 por órgano (RECIST 1.1) define la carga tumoral
del paciente.
Para aplicar los criterios RECIST, primero hay que defi-
nir en el estudio inicial las lesiones diana representativas y
reproducibles durante el seguimiento, teniendo en cuenta
que su gran variabilidad (confluencia, fragmentación, mala
definición, técnica de imagen, apreciación), y la dificultad
para medirlas, como ocurre en los órganos móviles (ovario,
tracto gastrointestinal), originan discrepancias intraobser-
vador e interobservador8 . La valoración de la respuesta se
realizará con la misma técnica empleada en el estudio ini-
cial, constatando la duración de la respuesta. La suma de las
lesiones diana del estudio basal se usará para seguir y valo-
rar objetivamente la respuesta. Cuando una lesión diana se
fragmente durante el tratamiento se medirán sus partes, se
sumarán, y se considerarán como una única lesión (fig. 3).
Aun con sus limitaciones y excepciones, RECIST es el
método radiológico sencillo, rápido y fiable más usado8 ,
para valorar el tamaño tumoral, aunque en ocasiones no se
correlacione bien con la situación clínica del paciente.
Tipos de respuesta
Se considera respuesta completa (RC) cuando desaparecen
la enfermedad medible y la evaluable, sin que aparezcan

Tabla 1 Resumen comparativo de los parámetros de evaluación según OMS, RECIST 1.0, RECIST 1.1 y PERCIST
OMS
Enfermedad medible Como medir Medida bidimensional:
producto de los diámetros
perpendiculares máximos
Tamaño mínimo No se especifica tamaño
mínimo
Número total de Todas las medibles
lesiones
Ganglios linfáticos No se especifica
Enfermedad no medible Linfangitis, masas abdominales Lesiones óseas, enfermedad
palpables no medibles y
afectación dérmica
RECIST 1.0
Medida unidimensional:
diámetro máximo
≥ 10 mm con TC helicoidal
≥ 20 mm con técnicas
convencionales
Máximo 10 lesiones y hasta 5
por órgano
No se especifica
leptomeníngea, ascitis,
derrame pleural y pericárdico,
carcinoma inflamatorio de la
mama, linfangitis
carcinomatosa de piel o
pulmón y las masas u
organomegalias abdominales
identificadas en la exploración
clínica no medibles por
métodos de imagen
RECIST y el radiólogo
RECIST 1.1 PERCIST
Medida unidimensional: diámetro SUV máximo al menos 1,5
máximo veces el de hígado o > 2
veces un ROI en la aorta
torácica descendente
≥ 10 mm con TC helicoidal o RM Determinada por RECIST 1.1
(no menor que el doble del
espesor de corte)
≥ 20 mm en radiografía de tórax
bien definida y rodeada de aire
Máximo 5 lesiones y hasta 2 por Máximo 5 lesiones y hasta 2
órgano. por órgano, seleccionando
las de máxima captación
Se incluyen metástasis quísticas,
lesiones óseas líticas con
componente de partes blandas.
Normales los ganglios de eje corto
< 10 mm
Ganglios con eje corto ≥ 15 mm
pueden ser lesión diana.
No medibles los de 10 a < 15 mm
Las lesiones inferiores a 10 mm.
Adenopatías con eje corto ≥ 10 y
≤ 15 mm.
Lesiones blásticas.
Quistes simples, indeterminados o
complejos.
Lesiones previamente tratadas o
irradiadas a menos que progresen
195
196 J. Cervera Deval

Tabla 2 Criterios de respuesta OMS, RECIST, EORTC-PET, PERCIST 1.0

Tipos OMS RECIST 1.1 EORTC-PET PERCIST 1.0 Choi
RC Desaparición de Desaparición de Ausencia de captación Desaparición completa Desaparición de todas
18
todas las lesiones todas las lesiones F-FDG, indistinguible de captación de 18 F-FDG las lesiones. Sin nuevas
y gánglios del tejido circundante en lesiones diana lesiones
patológicos medibles con actividad
menor a la hepática e
indistinguible del fondo
vascular
RP >50% disminución ≥ 30% de la suma Disminución del 15-25% Disminución al menos Disminución del tamaño
de la suma de las de los diámetros de SUV después de un del 30% del SUV máximo ≥10% o disminución de
áreas máximos ciclo y más del 25% de las lesiones diana la densidad ≥15% de
después de más ciclos medibles Unidades Hounsfield.
Sin nuevas lesiones.
Ausencia de progresión
de la enfermedad no
mensurable
EE No respuesta No respuesta Incremento de < 25% o Ni respuesta metabólica No cumple criterios de
parcial ni parcial ni disminución de < 15% del completa ni respuesta RC, RP o PE.
progresión progresión SUV, sin incremento de metabólica parcial ni
la extensión de la progresión metabólica
captación.
No hay deterioro
sintomático atribuible a
progresión tumoral
PE ≥ 25% de aumento ≥20% incremento Aumento del SUV > 25%. Aumento del SUV > 30% Aumento de tamaño
de la suma de las de la suma de los en lesiones diana. ≥10% y sin criterios de
áreas o aparición diámetros y RP en la densidad del
de una nueva ≥ 5 mm de tumor.
incremento
absoluto en la
suma de diámetros
Aumento > 20% de la Aumento de la extensión Nuevas lesiones.
extensión de la de la captación del
captación. tumor.
Nuevas captaciones Nuevas captaciones Nódulos nuevos
patológicas no intratumorales o
explicables por efecto aumento de los nódulos
del tratamiento o existentes o un aumento
infección de la parte tisular de
una lesión hipodensa.
Fuente: adaptada de Wahl26 y Choi32 .
EE: enfermedad estable; EORTC: European Organisation for Research and Treatment of Cancer; PE: progresión enfermedad; RC: respuesta
completa; RP: respuesta parcial; SUV: valor estandarizado de captación.

nuevas lesiones, con ausencia de síntomas y normalización
de marcadores, durante al menos 4 semanas. Hablamos de
respuesta parcial (RP) cuando se reduce al menos un 30% la
suma de los diámetros de las lesiones diana. Hay progresión
de la enfermedad (PE) cuando aumenta el 20%, se pro-
duce un incremento absoluto de al menos 5 mm en la suma
de los diámetros de las lesiones diana, o aparecen nuevas
metástasis o progresan las lesiones no diana. La enferme-
dad estable (EE) significa que no hay suficiente reducción
o suficiente incremento para considerarla RP o PE, respec-
tivamente. Se define como respuesta global (RG) a la mejor
respuesta después de iniciado el tratamiento antes de que
recidive o progrese la enfermedad (tablas 2---4).
Es importante tener presente el criterio de aumento
absoluto de 5 mm en la medida de una lesión para conside-
rar progresión. Así, si, por ejemplo, en el estudio basal una
lesión diana mide 30 mm y en el primer control se reduce a
10 mm (reducción del 67%) se considera RP. Si en el siguiente
control mide 13 mm se consideraría EE, ya que además de
aumentar el 20% se requiere el aumento absoluto de 5 mm.
Si en el seguimiento posterior aumenta a 16 mm se conside-
raría PE por incrementar su tamaño más de 5 mm y más del
20% respecto al menor de los tamaños registrados (10 mm)9 .
Volumen tumoral
Se ha propuesto utilizar el volumen tumoral10,11 para evaluar
la respuesta, ya que bastantes neoplasias no son esféri-
cas, y crecen o se reducen de modo irregular12 , y porque
RECIST y el radiólogo 197

Tabla 3 Combinación de las posibles respuestas en pacientes con lesiones medibles y su mejor respuesta global
Lesiones diana Lesiones no diana Nuevas lesiones Respuesta global Confirmación de mejor
respuesta global
RC RC No RC Sí ≥ 4 semanas
RC No RC/No PE No RP Sí ≥ 4 semanas
RC No evaluable No RP
RP No RC/No PE/No evaluable No RP
EE No RC/No PE/No evaluable No EE Sí. Al menos una vez ≥ 4
semanas desde el estudio basal
PE Cualquiera Sí o No PE No previa EE, RP o RC
Cualquiera PE Sí o No PE
Cualquiera Cualquiera Sí PE
Fuente: adaptada de Van Persijn van Meerten8 .
EE: enfermedad estable; PE: progresión enfermedad; RC: respuesta completa; RP: respuesta parcial.

el volumen proporciona una información de la carga tumo-
ral más fiable13 , con una predicción de la respuesta más
exacta, temprana14 y fácilmente reproducible15,16 . Aunque
en RECIST no se ha incluido la valoración volumétrica de los
tumores, sí contempla las medidas coronales de imágenes
isotrópicas5 .
Valoración funcional
Los cambios metabólicos que inducen en los tumores
los nuevos tratamientos (antiangiogénicos, antivasculares,
anti-EGFR [epithelial growth factor receptor], anti Her2,
inhibidores de la vía PI3k/Akt/mTOR, cKIT, terapia hor-
monal) no se reflejan directamente en la forma y, por
tanto, no se pueden valorar adecuadamente con los cri-
terios RECIST17 . Existe una amplia variedad de técnicas
de imágenes funcionales que permiten medir los cambios
fisiopatológicos, metabólicos o bioquímicos18 mediante bio-
marcadores de imagen médica12,19 , los cuales proporcionan
información, tanto para el diagnóstico como para evaluar la
respuesta20 .
Difusión con resonancia magnética

Las técnicas de difusión con RM, basadas en el movi-

miento browniano del agua, han demostrado ser un buen
indicador temprano de la respuesta tumoral21 . General-
mente los cambios en el coeficiente de difusión aparente
(ADC) se correlacionan inversamente con la celularidad. Las
imágenes potenciadas en difusión mejoran la detección y
caracterización de los tumores malignos, ya que estos son
hipercelulares y, por tanto, la difusión está restringida20 .
Un incremento del ADC puede reflejar un incremento en la
movilidad del agua, bien por la pérdida de la integridad de la
membrana, o bien por un incremento en la proporción total
de líquido extracelular, con la correspondiente disminución
del tamaño o número de células, como ocurre en la apoptosis
o necrosis. Por el contrario, una disminución del ADC refleja
una disminución del agua extracelular por un incremento

Tabla 4 Combinación de las posibles respuestas en pacien-

tes con lesiones no medibles y su mejor respuesta global
Lesiones no diana Nuevas lesiones Respuesta global
RC No RC
No RC/No PE No No RC/No PE
Figura 2 Carcinoma de células renales con lesión lítica (flecha
No evaluable No No evaluable
negra) en el hueso ilíaco derecho y con línea de medida inclu-
PE inequívoca Sí o No PE
yendo el componente de partes blandas en la imagen axial de
Cualquiera Sí PE
TC con contraste intravenoso y ventana de partes blandas (A).
Lesiones blásticas en ambos huesos ilíacos y L4 por adenocarci- Fuente: adaptada de Van Persijn van Meerten8 .
noma de próstata en la imagen axial de la pelvis con ventana EE: enfermedad estable; PE: progresión enfermedad; RC: res-
ósea (B). puesta completa; RP: respuesta parcial.
198 J. Cervera Deval

Figura 3 Medida del diámetro máximo (1) en una lesión diana (A) que tras el tratamiento se fragmenta (B). Para su valoración debe
sumarse el diámetro máximo de cada uno de las partes escindidas (en B, 1 + 2 + 3). En el caso contrario, de producirse coalescencia
se mide el máximo de la lesión resultante (1 en A).

Figura 4 Glioblastoma multiforme, estudio basal y tras quimioterapia con fármaco antiangiogénico (bevacizumab) y radioterapia
esteroatáxica fraccionada. En el estudio basal se aprecia la tumoración parafalciana en la imagen axial potenciada en T1 con
contraste intravenoso (A). En la imagen del coeficiente de difusión aparente (B) se aprecia una restricción de la difusión en la parte
sólida, y en el estudio de espectroscopia por RM 1 H muestra un cociente colina/N-acetil aspartato (Cho/NAA) de 3,7 (C), propio de
una tumoración glial de alto grado. El estudio axial T1 con contraste tras el tratamiento muestra una reducción de la tumoración
(D), aumento del ADC respecto al estudio basal (E) ----nótese las áreas de desmielinización debidas a la radioterapia---- y el cociente
Cho/NAA es 0,86 (F). Todo ello sugiere respuesta al tratamiento con disminución del metabolismo y el aumento de producción de
membranas celulares (principal responsable del metabolito Cho en el estudio de espectroscopia).
RECIST y el radiólogo 199

del número o del tamaño de las células, como ocurre en la

fibrosis, en el edema o en la progresión del tumor, lo que
la hace una herramienta muy útil para valorar precozmente
los efectos moleculares y, celulares de los tratamientos17,22
(fig. 4).

Perfusión

La perfusión tumoral19 permite observar cómo el contraste

inyectado se difunde al espacio extravascular extracelular,
lo que aporta información de la permeabilidad endote-
lial y, del volumen del espacio intersticial intercambiable,
y permite valorar la fisiología tisular y microvascular.
Con esta técnica se evalúa la respuesta basada en la
neoangiogénesis (fig. 5), considerando como respuesta la
reducción de los valores relativos de volumen (rCBV)
y flujo tumoral (rCBF). Así, por ejemplo, para dife-
renciar una radionecrosis de la persistencia tumoral en
las metástasis cerebrales, se estima que un rCBV > 2
respecto a la sustancia blanca cerebral sugiere permanencia
tumoral.

Espectroscopia con resonancia magnética

La espectroscopia con resonancia magnética (eRM) de pro-

tón 1 H es una técnica robusta, para uso clínico en el cerebro,
la mama y la próstata17 . Se basa en detectar los metaboli-
tos presentes o aumentados en el tumor y, su disminución
o desaparición durante el seguimiento. Los metabolitos con
importancia biológica son el ATP, la colina (Ch), la creatina
(Cr), el lactato, el citrato (Ci) y el N-acetil-aspartato (NAA).
En el cerebro, la combinación de RM y eRM puede ayu-
dar a evaluar el tipo de tumor cerebral y el grado, así como
a diferenciar el realce tumoral de otras causas de realce.
El criterio metabólico de la eRM de 1 H se basa en que en
las neoplasias malignas cerebrales hay un aumento de Ch
y descenso de NAA, con aumento del cociente Cho/NAA.
Cuando el tumor responde, disminuye este cociente meta-
bólico, tanto en los tumores cerebrales primarios como en
las metástasis.
En la mama, la disminución precoz de Ch es un buen indi-
cador de respuesta. En la próstata, la Ch, Cr y Ci pueden
predecir la existencia de cáncer en la zona periférica, así
como de recidiva tras la prostatectomía radical, radiotera-
pia o crioterapia.

Tomografía por emisión de positrones
La 18 F-FDG es un excelente biomarcador metabólico de la
actividad tumoral23 . El aumento de captación de 18 F-FDG se
relaciona con el aumento de actividad glicolítica cuantifica-
ble de la mayoría de los tumores malignos, lo que posibilita
evaluar la respuesta (fig. 6) e identificar precozmente a los
pacientes que responderán o no24 . Además de los criterios
de respuesta con 18 F-FDG de la European Organisation for
Research and Treatment of Cancer (EORTC)25 , hay que des-
tacar a los criterios de respuesta con la PET (PERCIST 1.0),
para valorar semicuantitativamente la actividad glicolítica
tumoral mediante el valor estandarizado de captación (SUV)
de 18 F-FDG26 .
Figura 5 Estudio de perfusión de un tumor de la base de len-
gua en el plano axial, antes (A) y después (B) de ser tratado con
cetuximab. Existe una disminución del producto del área de per-
meabilidad tumoral de 38 ml/100 ml/min a 33 ml/100 ml/min, y
del flujo sanguíneo de 86,8 ml/100 ml/min a 77 ml/100 ml/min,
considerados como enfermedad estable, por descenso de solo
el 11,29%.
Casos especiales de respuesta
Clásicamente, se ha considerado que una reducción de
tamaño tumoral representaba un signo precoz de mejoría
clínica, aun con los errores subjetivos de medida condi-
cionados por la irregularidad del margen tumoral27 , los
200 J. Cervera Deval

realces heterogéneos y, los diferentes protocolos de imagen.

Algunos autores28 consideran que la reducción del tamaño
tumoral no representa siempre respuesta y, no predice la
supervivencia global de la enfermedad, por lo que propo-
nen emplear la no progresión como predictor de respuesta.
Así, el carcinoma de pulmón no microcítico, responde
ocasionalmente cambiando la densidad del tumor sin modi-
ficar sustancialmente el tamaño, lo que en RECIST no se
consideraría como respuesta29 .
Las nuevos quimioterápicos y los procedimientos inter-
vencionistas de las lesiones focales30 se dirigen a detener
el crecimiento. En ocasiones, estos tratamientos provo-
can que los tumores puedan aumentar de volumen y, en
cambio, tener una excelente respuesta clínica31 . Así, por
ejemplo, las metástasis del tumor del estroma gastrointes-
tinal tratado con imatinib pueden aumentar de tamaño y
disminuir la densidad (fig. 7), lo que es un criterio de res-

Figura 6 (A) Imagen axial potenciada en T1 con contraste y

saturación grasa donde se aprecia una metástasis cerebelosa
izquierda; (B) estudio PET- F18 FDG donde se muestra la acti-
vidad metabólica en esta localización. Tras la radiocirugía se
aprecia un tamaño similar de la lesión en RM, aunque con cam-
bios internos (C) y desaparición de la actividad metabólica en Figura 7 GIST rectal que en el estudio basal presenta 2 metás-
el estudio de PET (D). tasis, la mayor de ellas con 49 UH de densidad (A), que en un
control 6 meses después (B) presenta aumento de tamaño y dis-
minución de densidad a 29 UH, considerándose como criterio de
respuesta propuesto por Choi32 ya que disminuye más del 15%.
RECIST y el radiólogo 201

puesta aceptado, descrito por Choi32 (tabla 2). Criterios
similares se aplican en metástasis hepáticas tratadas con
radioembolización33,34 , a sarcomas de partes blandas tra-
tados con quimioterapia y radioterapia preoperatoria, y al
carcinoma de células renales tratado con inhibidores de la
tirosin-cinasa35,36 .
En el cáncer de esófago, la PET tiene un papel importante
en el diagnóstico y la estadificación, y, además, puede pre-
decir qué tumor responderá al tratamiento. De este modo,
las guías de la National Comprehensive Cancer Network,
aunque recogen que el estudio de TC con contraste es la
primera indicación para valorar la respuesta, este no es
necesario si al paciente se le ha hecho antes un estudio de
PET37 .
Para el glioma cerebral se aplicaban los criterios de res-
puesta de MacDonald et al.38 en 1990 (tabla 5), los cuales
presentaban numerosas limitaciones39,40 en los pacientes
tratados con radioterapia y temozolamida, o con antiangio-
génicos que afectan la permeabilidad vascular del tumor,
especialmente los de mecanismo de acción sobre el factor
de crecimiento del endotelio vascular (VEGF: bevacizu-
mab) y en el receptor VEGF (cediranib). Estos tratamientos

Tabla 5 Criterios de respuesta en gliomas

RECIST 1,1
RC Desaparición de todos los
realces tumorales.
Confirmado a las 4 semanas
RP ≥ 30% disminución de la
suma de los diámetros
máximos. Confirmado a las
4 semanas
EE Ausencia de RC, RP, PE
PE ≥ 20% incremento en la
suma de los diámetros
máximos. Confirmado a las
4 semanas
Fuente: adaptado de Wen39 .
Macdonald (2D)
Desaparición de todos los realces
tumorales. Confirmado a las 4
semanas. No nuevas lesiones. No
corticoides y estabilidad o mejoría
clínica.
≥ 50% disminución en la suma de los
productos de los diámetros máximos
perpendiculares de todas las lesiones
que realzan. Confirmado a las 4
semanas. No nuevas lesiones.
Reducción o estabilidad de dosis de
corticoides y estabilidad o mejoría
clínica.
Ausencia de RC, RP, PE y estabilidad
clínica
≥ 25% incremento en la suma de los
productos de diámetros
perpendiculares de lesiones que
realzan, o nuevas lesiones o
deterioro clínico
RANO
Desaparición completa de todas las lesiones
que captan contraste, sean medibles o no,
mantenidas al menos 4 semanas. No nuevas
lesiones. Estabilidad o reducción de las
lesiones hiperintensas en T2/FLAIR que no
captan contraste. Ausencia de corticoides y
mejoría clínica*
≥ 50% disminución de la suma de productos de
los diámetros transversos de las lesiones
medibles que capten contraste, mantenido al
menos 4 semanas.
No hay nuevas lesiones. Estabilidad o
reducción de las lesiones hiperintensas en
T2/FLAIR que no captan contraste.
Mejoría o estabilidad clínica con menor dosis
de corticoides.
No hay progresión de lesiones no medibles
No cumple los criterios de RC, RP o PE.
Estabilidad o reducción de las lesiones
hiperintensas en T2/FLAIR que no captan
contraste.
Dosis de esteroides igual o inferior a la del
estudio basal.
Aumento ≥ 25% de tamaño de la suma de
productos de los diámetros perpendiculares de
las lesiones que captan contraste. Aumento de
las lesiones no medibles
Dosis de esteroides igual o superior a la que
tomaba al realizar la RM basal.
Aumento significativo de las lesiones no
captantes
Aparición de alguna lesión nueva.
Deterioro clínico no atribuible a otras causas
que no sea directamente al tumor o a cambios
en la dosis de esteroides
Fallecimiento o deterioro clínico severo
*Pacientes sin lesiones medibles la mejor
respuesta es EE
202 J. Cervera Deval

pueden producir una marcada disminución de la capta-

ción de contraste en la fase temprana del tratamiento (1
o 2 días), sin que esto signifique respuesta radiológica-
farmacológica. Los cambios deben permanecer al menos 4
semanas para considerarlos como verdadera respuesta. Con
el fin de soslayar estas limitaciones y estandarizar la valo-
ración de la respuesta en ensayos clínicos, se ha propuesto
una actualización por parte de un amplio grupo de exper-
tos de los criterios de respuesta conocidos como «Response
Assessment in Neuro-Oncology Working Group» (Criterios
RANO)39 . Su aportación es definir las lesiones medibles como
imágenes cuantificables y bien delimitadas, con diámetros
perpendiculares máximos de al menos 10 mm de diámetro en
2 o más imágenes (espesor de corte: 5 mm); y como lesio-
nes no medibles, a las mal definidas o predominantemente
quísticas, visibles solo en un eje, o menor de 10 mm (fig. 8)
y, a las cavidades posquirúrgicas.
En el seguimiento de linfomas se recomienda la PET por
discriminar mejor que la TC a los pacientes que puedan
recaer y a los que tienen posibilidades de remisión41 (fig. 9).

Figura 9 Linfoma de Hodgkin, esclerosis nodular, con adeno-

patías mediastínicas (A) hipermetabólicas en el estudio basal de
PET (B), que tras 6 ciclos de quimioterapia tiene una respuesta
completa (C).

El estudio de PET debe realizarse de 8 a 12 semanas des-

Figura 8 Lesión medible con diámetros transversales (A) aun-
que presenta componente necrótico en su interior. Nódulo (B)
dentro de una cavidad quística mayor de 10 mm (lesión medi-
ble). La lesión quística debe controlarse como no medible.
pués de la radioterapia o quimioterapia, ya que procesos
inflamatorios pueden observarse hasta 2 semanas después
de la quimioterapia.
En el carcinoma hepatocelular los criterios de medida
anatómicos pueden ser erróneos cuando se aplican tera-
pias moleculares, tales como el sorafenib o tratamientos
locorregionales, por lo que se propuso una modificación
de los criterios RECIST (mRECIST)42,43 . En ella se define una
lesión como evaluable si mide al menos 1 cm en una dimen-
sión, demostrable en la fase arterial del estudio dinámico
de TC o RM. Recientemente se ha propuesto un criterio
RECIST y el radiólogo 203

Tabla 6 Criterios mRECIST y RECICL para valoración del hepatocarcinoma

Lesión evaluable
Método de evaluación
Respuesta completa
Respuesta parcial
Enfermedad estable
Progresión enfermedad
Fuente: adaptado de Lencioni42 y Kudo44 .
mRECIST: criterios RECIST modificados; RECICL: criterios de evaluación de respuesta en carcinoma hepático.
mRECIST
Lesiones clasificadas como medibles
con criterio RECIST (1 cm o más en
una dimensión)
Medida unidimensional del tumor
viable (lesión realzada en fase
arterial) y que sea posible su medida
en el seguimiento
Desaparición de cualquier realce
arterial en lesiones diana
Al menos 30% de disminución de la
suma de los diámetros de las lesiones
diana viables (realzadas en fase
arterial), tomando como referencia
la suma basal de los diámetros de las
lesiones diana
No respuesta parcial ni progresión
≥ 20% suma de diámetros de lesiones
diana viables, tomando como
referencia la suma más pequeña de
los diámetros de lesiones viables
recogidas desde el inicio del
tratamiento. Desarrollo de una nueva
lesión
RECICL
Lesiones diana (un máximo de 5)
Suma de todas las lesiones diana. Medida
bidimensional (las lesiones no realzadas en
TC dinámico o con depósito de lipiodol son
medidas como necrosis)
Necrosis del 100% del tumor o 100% de
reducción del tamaño
Efecto necrótico o reducción del tamaño
del 50% a < 100%
No respuesta parcial ni progresión
≥ 25 de aumento del tumor respecto al
efecto necrotizante o aparición de una
nueva lesión (categorizada en 3 grupos:
lesión solitaria intrahepática, lesiones
intrahepáticas múltiples, invasión vascular
o diseminación extrahepática)

de valoración de respuesta (Response Evaluation Criteria
in Cancer of the Liver [RECICL]) (tabla 6), que cuantifica la
combinación de las medidas en 2 dimensiones de las lesio-
nes diana44 . Además, incluyen marcadores tumorales y un
protocolo de valoración en los casos tratados con qui-
mioembolización arterial, así como recomendaciones de
seguimiento según el tipo de tratamiento.
Conclusión
Confidencialidad de los datos. Los autores declaran que en
este artículo no aparecen datos de pacientes.
Derecho a la privacidad y consentimiento informado. Los
autores han obtenido el consentimiento informado de los
pacientes y/o sujetos referidos en el artículo. Este docu-
mento obra en poder del autor de correspondencia.
Conflicto de intereses
El autor declara no tener ningún conflicto de intereses.

Aunque el parámetro más usado para valorar la respuesta

al tratamiento es la reducción tumoral, no siempre una
reducción del volumen tumoral se acompaña de una mejora
de la supervivencia. Las discrepancias entre los criterios
anatómicos y metabólicos de respuesta, los fallos de pre-
dicción, la variabilidad interobservador y la utilidad de
valoraciones volumétricas, hacen que quizás deban revisarse
los criterios RECIST. Probablemente, la combinación de bio-
marcadores de imagen anatómica y funcional refleje mejor
la respuesta tumoral a los tratamientos. La utilización
clínica de biomarcadores funcionales requiere una estan-
darización, una vez hayan podido ser validados en ensayos
multicéntricos.
Responsabilidades éticas
Protección de personas y animales. Los autores declaran
que para esta investigación no se han realizado experimen-
tos en seres humanos ni en animales.
Bibliografía
1. Real Academia Española Diccionario de la Lengua Española. 22.a
ed. Madrid: Espasa-Calpe; 2001.
2. Therasse P, Arbuck SG, Eisenhauer EA, Wanders J, Kaplan RS,
Rubinstein L, et al. New guidelines to evaluate the response to
treatment in solid tumors. European Organization for Research
and Treatment of Cancer, National Cancer Institute of the Uni-
ted States, National Cancer Institute of Canada. J Natl Cancer
Inst. 2000;92:205---16.
3. Benjamin RS, Choi H, Macapinlac HA, Burgess MA, Patel SR,
Chen LL, et al. We should desist using RECIST, at last in GIST. J
Clin Oncol. 2007;25:1760---4.
4. World Health Organization WHO Handbook for Reporting
Results of Cancer Treatment Offset Publication. Geneva.
Switzerland, 1979 [consultado 3 Sep 2011]. Disponible en:
http://whqlibdoc.who.int/publications/9241700483.pdf.
5. Eisenhauer EA, Therase P, Bogaerts J, Schwartz LH,
Sargent D, Ford R, et al. New response evaluation crite-
ria in solid tumours: revised RECIST guideline (version 1.1). Eur
J Cancer. 2009;45:228---47.
204
6. Wahl RL, Jacene H, Kasamos Y, Lodge MA. From RECIST to PER-
CIST: evolving considerations for PET response criteria in solid
tumors. J Nucl Med. 2009;50 Suppl. 1:S122---30.
7. Schwartz LH, Bogaerts J, Ford R, Shanka L, Therasse P,
Gwyther S, et al. Evaluation of lymph nodes with RECIST 1.1.
Eur J Cancer. 2009;45:261---7.
8. van Persijn van Meerten EL, Gelderblom H, Bloem JL. RECIST
revised: implications for the radiologist. A review article on the
modified RECIST guideline. Eur Radiol. 2010;20:1456---67.
9. Nishino M, Jagannathan JP, Ramaiya N, Van den
Abbeele AD. Revised RECIST guideline version 1.1: what
to know and what radiologist need to know. AJR Am J
Roentgenol. 2010;195:281---9.
10. Lee SM, Kim SH, Lee LM, Im SA, Bang YJ, Kim WH, et al. Use-
fulness of CT volumetry for primary gastric lesions in predicting
pathologic response to neoadjuvant chemotherapy in advanced
gastric cancer. Abdom Imaging. 2009;34:430---40.
11. Baghi M, Mack MG, Hambek M, Bisdas S, Muerthel R,
Wagenblast J, et al. Usefulness of MRI volumetric evaluation in
patients with squamous cell cancer of the head and neck treated
with neoadjuvant chemotherapy. Head Neck. 2007;29:104---8.
12. Yaghmai V, Miller FH, Rezai P, Benson AB, Salem R. Response
to treatment series: Part 2, tumor response assessment-
using new and conventional criteria. AJR Am J Roentgenol.
2011;197:18---27.
13. Tran LN, Brown MS, Goldin JG, Yan X, Pals RC, McNitt-Gray MF,
et al. Comparison of treatment response classifications between
unidimensional, bidimensional and volumetric measurements of
metastatic lung lesions on chest computed tomography. Acad
Radiol. 2004;11:1355---60.
14. Curran SD, Muellner AU, Schwartz LH. Imaging response assess-
ment in oncology. Cancer Imaging. 2006;6:S126---30.
15. Buerke B, Puesken M, Müter S, Weckesser M, Gerss J, Heindel
W, et al. Measurement accuracy and reproducibility of semiau-
tomated metric and volumetric lymph node analysis in MDCT.
AJR Am J Roentgenol. 2010;195:979---85.
16. Jaffe CC. Measures of response: RECIST, WHO, and new alter-
natives. J Clin Oncol. 2006;24:3245---51.
17. García Figueiras R, Padhani AR, Vilanova JC, Goh V, Villalba
Martín C. Imagen funcional. Parte 2. Radiología. 2010;52:
208---20.
18. Sullivan DC, Gatsonis C. Response to treatment series: Part I and
introduction, measuring tumor response-challenges in the era
of molecular medicine. AJR Am J Roentgenol. 2011;197:15---7.
19. Martí-Bonmatí L, Sanz Requena R, Alberich Bayarri A. Nue-
vos horizontes en la imagen en oncología. En: Actualizaciones
SERAM: Imagen en Oncología. Madrid: Ed. Médica Panameri-
cana; 2009. p. 13-22.
20. García Figueiras R, Padhani AR, Goh VJ, Vilanova JC, Baleato
González S, Villalba Martín C, et al. Novel oncologic drugs:
what they do and how they affect images. Radiographics.
2011;31:2059---91.
21. Malayeri AA, El Khouli RH, Zaheer A, Jaobs M, Corona-
Villalobos CP, Kamel IR, et al. Principles and applications of
diffusion-weighted imaging in cancer detection, staging, and
treatment follow-up. Radiographics. 2011;31:1773---91.
22. Hamstra DA, Rehemtulla A, Ross BD. Diffusion magnetic reso-
nance imaging: a biomarker for treatment response in oncology.
J Clin Oncol. 2007;25:4104---9.
23. Shankar LK, Hofman JM, Bacharach S, Grahan MM, Karp J,
Lammertsma AA, et al. Consensus recommendations for the
use of 18 F-FDG PET as an indicator of therapeutic response
in patients in National Cancer Institute trials. J Nucl Med.
2006;46:1059---66.
24. Benz MR, Evilevitch V, Allen-Auerbach MS, Eilber FC,
Phelps ME, Czernin J, et al. Treatment monitoring by 18 F-FDG
PET/CT in patients with sarcomas: Interobserver variability
of quantitative parameters in treatment-induced changes in
J. Cervera Deval
histopatholgically responding and non responding tumors. J Nucl
Med. 2008;49:1038---46.
25. Young H, Baum R, Cremerius U, Herholz K, Hoekstra O,
Lammertsma AA, et al. Measurement of clinical and sub-
clinical tumour response using[18 F]-fluorodeoxyglucose and
positron emission tomography: Review and1999 EORTC recom-
mendations reproducibility of quantitative measurements for
evaluating 2-[F-18]-fluoro-2-deoxy-d-glucose uptake at PET. Eur
J Cancer. 1999;35:1773---82.
26. Wahl RL, Jacebe H, Kasamon Y, Lodge MA. From RECIST to PER-
CIST: evolving considerations for PET response criteria in solid
tumors. J Nucl Med. 2009;50:122S---50S.
27. Erasmus JJ, Gladish GW, Broemeling L, Sabloff BS,
Truong MT, Herbst RS, et al. Interobserver and intraob-
server variability in measurement if non-small-cell carcinoma
lung lesions: implications for assessment of tumor response. J
Clin Oncol. 2003;21:2574---82.
28. Tuma RS. Somtimes size doesn’t matter: reevaluating RECIST
and tumor response rate endpoints. J Natl Cancer Inst.
2006;98:1272---4.
29. Pallardó Calatayud Y, Cervantes Ruipérez A, Revert Ventura AJ.
Estadificación del cáncer y valoración de la respuesta al tra-
tamiento. En: Actualizaciones SERAM: Imagen en Oncología.
Madrid: Ed. Médica Panamericana; 2009. p. 1-12.
30. Park MH, Rhim H, Kim YS, Choi D, Lim HK, Lee WJ. Spectrum
of CT findings after radiofrequency ablation of hepatic tumors.
Radiographics. 2008;28:379---92.
31. Riaz A, Miller FH, Kulil LM, Nicolaidis P, Yaghmai V,
Lewandowski RJ, et al. Imaging response in the primary index
lesion and clinical outcomes following transarterial locorre-
gional therapy for hepatocellular carcinoma. JAMA. 2010;303:
1062---9.
32. Choi H, Charnsangavej C, de Castro Faria S, Tamm EP,
Benjamín RS, Johnson MM, et al. CT evaluation of the response
of gastrointestinal stromal tumors after imatinib mesylate
treatment a quantitative analysis correlated with FDG PET fin-
dings. AJR Am J Roentgenol. 2004;183:1619---28.
33. Lewandowski RJ, Geschwind JF, Liapi E, Salem R. Transcathe-
ter intraarterial therapies: rationale and overview. Radiology.
2011;259:641---57.
34. Sato KT, Lewandowski RJ, Mulcahy MF, Atassi B, Ryu RK,
Gates VL, et al. Unresectable chemorefractory liver metasta-
ses: radioembolization with 90Y microspheres- safety, efficacy,
and survival. Radiology. 2008;247:507---15.
35. Katabathina VS, Lassau N, Pedrosa I, Ng CS, Prasad SR. Eva-
luation of treatment response in patients with metastatic renal
cell carcinoma: role of state-of-the-art cross-sectional imaging.
Curr Urol Rep. 2012;13:70---81.
36. Motzer RJ, Michaelson MD, Redman BG, Hudes GR, Wilding G,
Figlin RA, et al. Activity of SU11248, a multitargeted inhibitor of
vascular endotelial growth factor receptor and platelet-derived
growth factor receptor, in patients with metastatic renal cell
carcinoma. J Clin Oncol. 2006;24:16---24.
37. Kwee RM. Prediction of tumor response to neoadjuvant therapy
in patients with esofageal cancer with use of 18 F FDG PET: a
systematic review. Radiology. 2010;254:707---17.
38. MacDonald DR, Cascino TL, Schold SC, Caimcross JG. Response
criteria for phase II studies of supratentorial malignant glioma.
J Clin Oncol. 1990;8:1277---80.
39. Wen PY, MacDonald DR, Reardon DA, Cloughesy TF, Sorensen AG,
Galanis E, et al. Updated response assessment criteria for high-
grade gliomas: response assessment in neuro-oncology working
group. J Clin Oncol. 2010;28:1963---72.
40. Henson JW, Ulmer S, Harris GJ. Brain tumor imaging in clinical
trials. AJNR Am J Neuroradiol. 2008;29:419---24.
41. Juweid ME, Stoobants S, Hoekstra OS, Mottaghy FM,
Dietlein M, Guermazi A, et al. Use of positron emission tomo-
graphy for response assessment of lymphoma: consensus of the
RECIST y el radiólogo
Imaging Subcommittee of International Harmonization Project
in lymphoma. J Clin Oncol. 2007;25:571---8.
42. Lencioni R, Llovet JM. Modified RECIST (mRECIST) assessment
for hepatocellular carcinoma. Semin Liver Dis. 2010;30:52---60.
43. Forner A, Ayuso C, Varela M, Rimola J, Hessheimer AJ, de
Lope CR, et al. Evaluation of tumor response after locorregional
205
therapies in hepatocellular carcinoma: are response evaluation
criteria in solid tumors reliable. Cancer. 2009;115:616---23.
44. Kudo M, Kubo S, Takayasu K, Sakamoto M, Tanaka M, Ikai I, et al.
Response evaluation criteria in cancer of the liver (RECICL) pro-
posed by the liver cancer study group of Japan (2009 revised
version). Hepatol Res. 2010;40:686---92.
European Journal of Cancer xx (2016) 1e6

Available online at www.sciencedirect.com

ScienceDirect

journal homepage: www.ejcancer.com

Current Perspective

RECIST 1.1dUpdate and clarification: From the

RECIST committee

Lawrence H. Schwartz a,b,*, Saskia Litière c, Elisabeth de Vries d,

Robert Ford e, Stephen Gwyther f, Sumithra Mandrekar g,
Lalitha Shankar h, Jan Bogaerts c, Alice Chen i, Janet Dancey r,
Wendy Hayes j, F. Stephen Hodi k, Otto S. Hoekstra l, Erich P. Huang m,
Nancy Lin k, Yan Liu c, Patrick Therasse n, Jedd D. Wolchok o,p,q,
Lesley Seymour r

a
Department of Radiology, Columbia University Medical Center, New York, NY, USA
b
New York Presbyterian Hospital, New York, NY, USA
c
EORTC HQ, Brussels, Belgium
d
Department of Internal Medicine, University Medical Center Groningen, Groningen, The Netherlands
e
Clinical Trials Imaging Consulting, NJ, USA
f
Department of Medical Imaging, East Surrey Hospital, Redhill, Surrey, UK
g
Division of Biomedical Statistics and Informatics, Mayo Clinic, Rochester, MN, USA
h
Clinical Trials Branch, Cancer Imaging Program, Division of Cancer Treatment and Diagnosis, National Cancer Institute,
Bethesda, MD, USA
i
Early Clinical Trials Development Program, Division of Cancer Treatment and Diagnosis, National Cancer Institute,
Bethesda, MD, USA
j
Exploratory Clinical & Translational Research, Bristol-Myers Squibb, Princeton, NJ, USA
k
Department of Medical Oncology, Dana-Farber Cancer Institute, Harvard Medical School, Boston, MA, USA
l
Department of Radiology and Nuclear Medicine, VU University Medical Center, Amsterdam, The Netherlands
m
National Cancer Institute, National Institutes of Health, Rockville, MD, USA
n
Laboratoires Servier, Paris, France
o
Department of Medicine, Memorial Sloan Kettering Cancer Center, New York, NY, USA
p
Weill Cornell Medical and Graduate Colleges, New York, NY, USA
q
Ludwig Institute for Cancer Research, New York, NY, USA
r
Canadian Cancer Trials Group, Queen’s University, Kingston, Canada

Received 25 March 2016; accepted 27 March 2016

* Corresponding author: CUMC, 180 Fort Washington Avenue, New York, NY 10032, USA. Tel.: þ1 212 305 8994; fax: þ1 212 385 4835.
E-mail address: LSchwartz@columbia.edu (L.H. Schwartz).

http://dx.doi.org/10.1016/j.ejca.2016.03.081
0959-8049/ª 2016 Published by Elsevier Ltd.

Please cite this article in press as: Schwartz LH, et al., RECIST 1.1dUpdate and clarification: From the RECIST committee, European Journal
of Cancer (2016), http://dx.doi.org/10.1016/j.ejca.2016.03.081
2 L.H. Schwartz et al. / European Journal of Cancer xx (2016) 1e6

KEYWORDS Abstract The Response Evaluation Criteria in Solid Tumours (RECIST) were developed
RECIST; and published in 2000, based on the original World Health Organisation guidelines first pub-
Clarifications; lished in 1981. In 2009, revisions were made (RECIST 1.1) incorporating major changes,
Tumour response including a reduction in the number of lesions to be assessed, a new measurement method
to classify lymph nodes as pathologic or normal, the clarification of the requirement to
confirm a complete response or partial response and new methodologies for more appropriate
measurement of disease progression. The purpose of this paper was to summarise the ques-
tions posed and the clarifications provided as an update to the 2009 publication.
ª 2016 Published by Elsevier Ltd.

1. Introduction an adjunct to the gold standard end-points of survival

World Health Organisation response guidelines were
first published in 1981 [1,2]. The Response Evaluation
Criteria in Solid Tumors (RECIST) criteria were based
on those criteria and were themselves updated in a 2009
in the European Journal of Cancer (RECIST 1.1) [3]. The
revised guidelines incorporated major changes to the
original RECIST criteria [2], including a reduction in the
number of lesions to be assessed, a new measurement
method to classify lymph nodes as pathologic or
normal, the clarification of the requirement to confirm a
complete response (CR) or partial response (PR) and
new recommendations for the assessment of disease
progression. Supplementary information provided
included imaging guidelines, which better defined image
acquisition and interpretation.
The RECIST criteria have gained widespread adop-
tion and are widely used in oncology clinical trials. The
RECIST 1.1 paper has been cited 3881 times as of
December 2015 Web of Science. End-points categorised
by the RECIST criteria have been used as either primary
or supportive data for regulatory approval of new
therapeutics by both the Food and Drug Administration
and European Medicines Agency (EMA) [4]. RECIST
provides a standardised set of rules for response
assessment using tumour shrinkage, based upon imaging
modalities that are globally available and interpretable
by most clinicians. This standardisation, and the rules
and criteria established, provide a framework for
reproducible analysis and reporting of changes in
tumour size. The reproducibility of these criteria and the
correlations with historical trial results serve an impor-
tant purpose in drug discovery.
Despite the widespread acceptance of RECIST, the
RECIST Working Group continues its work. RECIST
and RECIST 1.1 were developed and tested using data
from clinical trials testing cytotoxic drugs. In the last
decade, there have been substantial changes in the
mechanism of action of cancer therapeutics (targeted
agents, immunotherapies), as well as advances in imag-
ing and clinical trial design and end-points. Although
the majority of clinical trials continue to use RECIST as
and quality of life, the RECIST Working Group con-
tinues to build data warehouses (e.g. data from clinical
trials involving targeted agents and immunotherapies,
FDG-positron emission tomography [PET]/computed
tomography [CT] data) to review the criteria, update
them periodically as required and validate any changes
in a standardised, methodical manner in response to
both therapeutic and imaging technology advances.
Critically, the global oncology community must be able
to implement and adopt any changes proposed to
RECIST in a timely and cost-effective manner.
Since the original publication, users have submitted
questions on the use and interpretation of the guidelines
through the RECIST website (http://www.eortc.org/
recist/). These questions are reviewed by the Steering
Committee and replies provided based on the 2009 pub-
lication. If not addressed in the publication, the Working
Group is consulted to prepare appropriate replies. Rele-
vant and/or frequently asked questions (FAQs) are posted
on a regular basis on the RECIST website.
Here we summarise the questions and clarifications
posed as an update to the 2009 publication.
2. Commonly asked questions regarding RECIST 1.1
1. What is the frequency of tumour evaluation?
The schedule and frequency of tumour response re-
evaluation is protocol specific and is based on the
therapeutic, the disease, the anticipated time to response
and progression, as well as practical considerations,
such as cost and patient convenience.
However, in the context of phase II studies where the
beneficial effect of therapy is not known, follow-up
every 6e8 weeks (timed to coincide with the end of a
cycle) is reasonable. Shorter or longer time intervals
than these could be justified in specific regimens or cir-
cumstances. The protocol should specify which areas
(chest, abdomen) and organs are to be evaluated at
baseline (usually those most likely to be involved with
metastatic disease for the tumour type under study) and
how often evaluations should be repeated [2]. The

Please cite this article in press as: Schwartz LH, et al., RECIST 1.1dUpdate and clarification: From the RECIST committee, European Journal
of Cancer (2016), http://dx.doi.org/10.1016/j.ejca.2016.03.081
 L.H. Schwartz et al. / European Journal of Cancer xx (2016) 1e6
method of assessment should be stated and, ideally,
repeated using the same imaging technique, equipment
and assessor each time. Additional assessments should
be performed if there is suspicion of new site of metas-
tasis or of disease progression based upon clinical
symptoms. The presence of a new lesion(s) should be
documented on an imaging study. All potential sites of
metastases should be evaluated at each time point rather
than following only sites of disease identified at baseline.
2. Should lesions smaller than 5 mm be reported as the actual
size or reported as a default value of 5 mm?
All lesions, both nodal and non-nodal, must be
evaluated, accurately measured (if measurable/target)
and recorded at all time points. If a lesion is no longer
seen, it should be recorded as zero (or absent if non-
measurable/non-target). If a lesion is smaller than 5 mm
and the radiologist believes the lesion can be accurately
measured, then the actual size should be recorded.
It is recognised that lesions become small and ill
defined on a CT scan such that the radiologist cannot
accurately measure them. Some radiologists use the term
‘too small to measure’ to describe this phenomena.
When this occurs, the radiologist may decide that the
lesion is present but he/she does not feel comfortable
providing the oncologist with a precise measurement. If
this occurs, the radiologist may assign the lesion a value
of 5 mm by default. This default value is nominally
derived from the 5-mm CT slice thickness (but should
not be changed with varying CT slice thickness). The
measurement of this type of lesion is potentially non-
reproducible, and providing this default value will pre-
vent a false assessment of response or progressive dis-
ease due to measurement error.
3. If there are three or more measurable lesions in one organ,
and we select two of them as target lesions, how should the
third lesion be considered?
The third lesion should be considered a non-target
lesion and should be recorded and followed as part of
the non-target disease.
4. Is a single-target lesion measurable if a patient has multiple
non-target non-measurable disease?
Yes, a single-target lesion is considered measurable
disease, provided the lesion meets the definition of
measurability as described in RECIST 1.1.
5. How should the limitation of two target lesions per organ be
applied to lymph nodes? Can individual chains/regions be
considered one organ or are lymph nodes (all locations
included) a single organ?
Lymph nodes are considered one organ. Only two
lymph nodes should be measured per patient as target
Please cite this article in press as: Schwartz LH, et al., RECIST 1.1dUpdate and clarification: From the RECIST committee, European Journal
of Cancer (2016), http://dx.doi.org/10.1016/j.ejca.2016.03.081
3
lesions. Other involved lymph nodes should be assessed
and followed as non-target lesions. The analysis of data
from the RECIST warehouse was performed in this
manner.
6. Should double the slice thickness/interval be applied to lymph
nodes as well when a CT slice thickness of 10 mm is used?
It is strongly recommended that CT slice thickness of
5 mm be used. Although RECIST 1.1 (2009) recom-
mends the following: ‘As is described in Appendix II,
when CT scans have slice thickness greater than 5 mm,
the minimum size for a measurable lesion should be
twice the slice thickness’, in 2016 contemporary CT
scanners globally should be able to acquire images with
a slice thickness of 5 mm or less. This is recommended
by the Quantitative Imaging Biomarkers Alliance
guidelines for standardising image acquisition for CT
(http://rsna.org/QIBA_.aspx) [5]. There are many dis-
advantages and no real advantages in obtaining 10-mm
slices. Not only are lesions more difficult to measure, but
new lesion conspicuity is significantly less at 10 mm [6,7].
7. When lymph nodes coalesce forming a conglomerate mass,
which axis should be measured to assess the response: short
or long axis?
The short axis of lymph nodes should always be
measured. As nodal lesions coalesce, a plane between
them may be maintained that would aid in obtaining
maximal short-axis diameter measurements of each in-
dividual lesion. If the nodal lesions have truly coalesced
such that they are no longer separable, the vector of the
longest diameter in this instance should be used to
determine the perpendicular vector for the maximal
short-axis diameter of the coalesced lesion (Fig. 1). Non-
nodal lesions that coalesce should similarly be assessed
by the longest diameter.
Fig. 1. Note optimal manner for measuring the short axis of
coalescing lymph nodes (blue arrow). At baseline, there are two
distinct nodes; therefore, the short axis is measured for each (red
and yellow lines) and at cycle 6 the single-short axis of the coa-
lesced node is now measured as a single line (red) (For interpre-
tation of the references to colour in this figure legend, the reader is
referred to the web version of this article.).
4 L.H. Schwartz et al. / European Journal of Cancer xx (2016) 1e6

8. Clarification of the definition of stable disease.
The definition of stable disease (SD) is clarified as
follows: ‘Neither sufficient shrinkage (compared to
baseline) to qualify for CR or PR nor sufficient increase
(taking as reference the smallest sum of diameters at
baseline or while on study, whichever is smallest) to
qualify for progressive disease (PD)’.
It is important to recognise that the classification of a
response (either CR or PR) occurs in comparison to the
sum of diameters at baseline, while progression is based
on a comparison to the smallest of the sum of diameters
at baseline or the smallest sum of diameters during the
trial (nadir). Most protocols require the criteria for SD
for a specified period (e.g. at least 4 weeks) before SD
can be concluded. Thus, if imaging is conducted at 2 and
4 weeks on-study and at 2 weeks, the criteria for CR,
PR, or PD are not met, but at 4 weeks meets the criteria
for PD, the best overall response is PD, not SD as the
subject was not on-study long enough to qualify for SD.
9. If an abnormal lymph node (or non-nodal disease) ‘disap-
pears’ but then ‘reappears’ should this considered to be PD?
In general, significant lesions that completely regress
and then reappear are indicative of PD. However, it is
important to consider the patient’s entire tumour
burden in order to make certain that the patient is not
falsely classified as having PD based upon a single
measurement or lesion, especially when those lesions are
small or there is a change in optimal imaging assess-
ment. This holds true for all types of metastases (Fig. 2).
If the response was previously considered to be CR,
with resolution of all sites of disease, then the reappear-
ance of any lesion, or the development of a new lesion
considered to be malignant, would generally be considered
PD (see comments below regarding lymph nodes).
In the case of PR or SD, if a previously resolved
lesion reappears, then PD should not be assigned purely
based on the reappearance of the lesion; rather, other
criteria for PD must be met as well, such as the
appearance of new lesions or a sum measurement of
target lesions that has increased more than 20%.
It is important to remember that lymph nodes must
meet the criteria for malignancy (defined pragmatically
as 10 mm) in order to be considered to have reap-
peared. A node which is less than 10 mm is considered
benign and is not PD. In a subject with CR, a new node
that meets the size criteria for a pathologic node is
considered a new site of disease and is, therefore,
consistent with PD.
In subjects with PR or SD:
a. A previously abnormal target node that became
normal and subsequently enlarged in size meeting the
criteria for a pathologic and measurable node (a short
axis of 15 mm) should be added to the sum of the
diameters to determine if the criteria for PD are met
based on target lesions.
b. A previously abnormal non-target node that
became normal and subsequently recurred must meet
the criteria for PD based on NT lesions to call
progression.
c. A normal node at baseline that subsequently be-
comes pathologic is considered a new lesion and results
in PD.
In the circumstances illustrated above, where a single
pathologic node is driving the progression event,
continuation of treatment/follow-up and confirmation
by a subsequent examination should be contemplated. If
it becomes clear that the ‘new node’ has not resolved, or
has significantly increased in size, and truly represents
PD, the date of PD would be the date the new node was
first documented.
10. How should patients be classified who have had surgery/
radiotherapy during trials for which they are being fol-
lowed by RECIST?
For most clinical trials using RECIST, surgery or
radiotherapy after trial inclusion and prior to disease

Fig. 2. Patient with colorectal cancer. Liver metastases at baseline (red circle) appear to be resolved at cycle 3 with ‘reappearance’ at cycle 5 (red
circle). However, notice that the imaging techniques are different and the cycle 3 is of poor image quality to visualise these metastases.
Therefore, these lesions have not truly reappeared and the patient should not be considered to have progressive disease at cycle 5. There is true
progression at cycle 7 (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.).

Please cite this article in press as: Schwartz LH, et al., RECIST 1.1dUpdate and clarification: From the RECIST committee, European Journal
of Cancer (2016), http://dx.doi.org/10.1016/j.ejca.2016.03.081
 L.H. Schwartz et al. / European Journal of Cancer xx (2016) 1e6
progression is a protocol deviation, and if a target lesion
has been surgically removed or treated with radio-
therapy, then the patient’s response is not evaluable. If
surgery or radiotherapy is part of the clinical trial, then
the protocol must define in advance how response and
progression will be handled and ‘censored’ in the ana-
lyses. If the treatment has resulted in inoperable lesions
being suitable for resection, the researcher may wish to
capture that separately as an indicator of ‘activity’.
11. The section on FDG-PET mentions that correlation with
CT is warranted for new lesions and that PD should be
declared if the hot spot on PET corresponds to a pro-
gressing lesion on CT. How should this correlation be
made if the hot spot on PET corresponds to a target lesion
that has increased in size but the sum of the measurements
does not show an increase that is sufficient for PD (other
target lesions have not enlarged or have actually
decreased). Is this PD?
The RECIST 1.1 guidelines state ‘a. Negative FDG-
PET at baseline, with a positive FDG-PET at follow-
up is a sign of PD based on a new lesion. b. No FDG-
PET at baseline and a positive FDG-PET at follow-up:
If the positive FDG-PET at follow-up corresponds to a
new site of disease confirmed by CT, this is PD. If the
positive FDG-PET at follow-up is not confirmed as a
new site of disease on CT, additional follow-up CT
scans are needed to determine if there is truly pro-
gression occurring at that site (if so, the date of PD will
be the date of the initial abnormal FDG-PET scan). If
the positive FDG-PET at follow-up corresponds to a
pre-existing site of disease on CT that is not pro-
gressing on the basis of the anatomic images, this is
not PD’.
Therefore, the scenario is not PD. Currently, PET
scanning is considered a complementary modality pri-
marily to assess for new tumour lesions. Therefore, if a
target lesion is FDG avid and even if the standardised
uptake value on PET has increased, the patient would
only be considered to be PD if the CT metrics for pro-
gression were met, including a greater than 20% increase
in the sum measurement of lesions, non-target un-
equivocal PD or new lesions on CT (regardless of their
presence or SUV on PET).
12. If there is a hot spot on FDG-PET (baseline PET is not
available) that is not associated with a new CT (or
magnetic resonance imaging [MRI]) lesion, the article
states that PD should not be declared. What should the
response be, not evaluable?
FDG-PET is a complementary modality and should
be used in adjunct to the CT or MRI. Therefore, in this
example, the patient’s response should be based solely
upon the measurements and findings on CT and/or
MRI. Since there is no PET at baseline and no corre-
sponding new lesion on CT or MRI, then the patient can
Please cite this article in press as: Schwartz LH, et al., RECIST 1.1dUpdate and clarification: From the RECIST committee, European Journal
of Cancer (2016), http://dx.doi.org/10.1016/j.ejca.2016.03.081
5
only be assessed on the existing lesions and their change
from baseline to follow-up on CT or MRI. Providing
the scan was evaluable and all lesions visible/measur-
able, the response would, therefore, be CR, PR or SD as
no new lesions were seen on CT/MRI.
13. Can the CT information from PET-CT be used as the
basis of CT assessments? Is the technical quality of such
images sufficient for quantification as required in
RECIST?
At present, the low dose or the attenuation correction
CT portion of a combined PET-CT is not of optimal
diagnostic CT quality for use with RECIST measure-
ments. However, if the site has documented that a CT
with appropriate radiation dose for diagnostic quality
and intravenous and oral contrast was used (if not
medically contraindicated), the CT portion of the PET-
CT can be used for RECIST measurements.
14. Can we use coronal or sagittal imaging to measure lesions
in CT if the largest diameter is in a plane other than axial?
It is recommended that the axial imaging plane be
used in all cases on CT scans for consistency and for
ease of measurement especially since reconstructions or
advanced workstations are not always available glob-
ally. It is recognised that the other planes may represent
the true long axis of the tumour but depending on the
CT acquisition parameters across time points, and this
may be difficult to consistently and reproducibly
measure.
3. Conclusions
The RECIST Working Group is currently testing the
Criteria for their applicability with modern therapeutics,
imaging techniques and trial end-points. The current
RECIST 1.1 criteria remain widely used. A number of
users have asked questions regarding the interpretation
of the 2009 Criteria and we have attempted to summa-
rise those questions, and the answers, in this update. For
further updates on the criteria, please visit the RECIST
website at http://www.eortc.org/recist/
Conflict of interest statement
E. de Vries: Research grants from Roche/Genentech,
Amgen, Novartis, Pieris and Servier to the institute,
data monitoring committee Biomarin, advisory board
Synthon. R. Ford: Dr. Ford is not a stockholder in any
pharmaceutical company and holds no options, grants
or patents. He is currently or was previously a consul-
tant for the following companies (either directly through
Clinical Trials Imaging Consulting, LLC., or indirectly
through other pharmaceutical service companies for
which he consults) including ACR Image Metrix,
6 L.H. Schwartz et al. / European Journal of Cancer xx (2016) 1e6
AbbVie, Amgen, Aptiv, Aragon, Bioclinica, Biomedical
Systems, Bristol Myers Squibb, Celldex, Celgene, Cel-
sion, Covance, DNAtrix, Eisai, Exelixis, Genentech/
Roche, ICON Medical Imaging, Image Endpoints,
Janssen, Kyowa, Merck, Mirati Therapeutics, Novartis,
Novocure, Oncothyreon, Ono, Optimer, OrbiMed Ad-
visors, Pfizer, Radiant Sage, Quintiles, Red Hill Pharma,
Sun Advanced Pharma Research Company, Tokai
Pharmaceuticals, Vascular Biogenics and Virtualscopics.
S. Hodi: Consultant for Merck, Novartis, Genentech,
Synta; research support to institution from Bristol-
Myers Squibb. N. Lin: Research funding for clinical
trials from (no direct salary support): Genentech, GSK,
Novartis, Synta, Array Biopharma, Kadmon, and
Puma. J.D. Wolchok: Research grants from Bristol-
Myers Squibb; advisory board member for Brisol-Myers
Squibb, Merck, Medimmune, Genentech. All authors
have no conflicts relevant to the study. LH Schwartz:
Research grants from Novartis, Astellas, Eli Lilly,
Merck, Pfizer, Consultant to GSK, Novartis.
Acknowledgements
Canadian Cancer Trials Group participation was
supported by the Canadian Cancer Society Research
Institute (grant #021039). This publication was sup-
ported by the EORTC Cancer Research Fund and by
the National Cancer Institute (NCI) grant number
5U10-CA11488-45.
Please cite this article in press as: Schwartz LH, et al., RECIST 1.1dUpdate and clarification: From the RECIST committee, European Journal
of Cancer (2016), http://dx.doi.org/10.1016/j.ejca.2016.03.081
References
[1] Miller AB, Hoogstraten B, Staquet M, Winkler A. Reporting re-
sults of cancer treatment. Cancer 1981;47:207e14.
[2] Therasse P, Arbuck SG, Eisenhauer EA, Wanders J, Kaplan RS,
Rubinstein L, Verweij J, Van Glabbeke M, van Oosterom AT,
Christian MC, Gwyther SG. New guidelines to evaluate the
response to treatment in solid tumors. European Organization for
Research and Treatment of Cancer, National Cancer Institute of
the United States, National Cancer Institute of Canada. J Natl
Cancer Inst 2000;92:205e16.
[3] Eisenhauer EA, Therasse P, Bogaerts J, Schwartz LH, Sargent D,
Ford R, Dancey J, Arbuck S, Gwyther S, Mooney M, Rubinstein L,
Shankar L, Dodd L, Kaplan R, Lacombe D, Verweij J. New response
evaluation criteria in solid tumours: revised RECIST guideline
(version 1.1). Eur J Cancer 2009;45:228e47.
[4] Janku F, Berry DA, Gong J, Parsons HA, Stewart DJ,
Kurzrock R. Outcomes of phase II clinical trials with single-agent
therapies in advanced/metastatic nonesmall cell lung cancer pub-
lished between 2000 and 2009. Clin Cancer Res 2012;18:6356e63.
[5] http://www.rsna.org/uploadedFiles/RSNA/Content/Science_and_
Education/QIBA/QIBA_CT%20Vol_
TumorVolumeChangeProfile_v2.2_PubliclyReviewedVersion_
08AUG2012.pdf.
[6] Zhao B, Tan Y, Bell DJ, Marley SE, Guo P, Mann H, Scott ML,
Schwartz LH, Ghiorghiu DC. Exploring intra- and inter-reader
variability in uni-dimensional, bi-dimensional, and volumetric
measurements of solid tumors on CT scans reconstructed at
different slice intervals. Eur J Radiol 2013;82:959e68.
[7] Sullivan DC, Schwartz LH, Zhao B. The imaging viewpoint: how
imaging affects determination of progression-free survival. Clin
Cancer Res 2013;19:2621e8.
Journal Pre-proof

Impact of acetaminophen on the efficacy of immunotherapy in cancer patients.

A. Bessede, A. Marabelle, J.P. Guégan, F.X. Danlos, S. Cousin, F. Peyraud, N.

Chaput, M. Spalato, G. Roubaud, M. Cabart, M. Khettab, A. Chaibi, C. Rey, I. Nafia,
F.X. Mahon, J.C. Soria, A. Italiano
PII: S0923-7534(22)01208-X
DOI: https://doi.org/10.1016/j.annonc.2022.05.010
Reference: ANNONC 941

To appear in: Annals of Oncology

Received Date: 31 March 2022

Revised Date: 9 May 2022
Accepted Date: 24 May 2022

Please cite this article as: Bessede A, Marabelle A, Guégan J, Danlos F, Cousin S, Peyraud F, Chaput
N, Spalato M, Roubaud G, Cabart M, Khettab M, Chaibi A, Rey C, Nafia I, Mahon F, Soria J, Italiano
A, Impact of acetaminophen on the efficacy of immunotherapy in cancer patients., Annals of Oncology
(2022), doi: https://doi.org/10.1016/j.annonc.2022.05.010.

This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition
of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of
record. This version will undergo additional copyediting, typesetting and review before it is published
in its final form, but we are providing this version to give early visibility of the article. Please note that,
during the production process, errors may be discovered which could affect the content, and all legal
disclaimers that apply to the journal pertain.

© 2022 European Society for Medical Oncology. Published by Elsevier Ltd. All rights reserved.
ORIGINAL REPORT

Impact of acetaminophen on the efficacy of immunotherapy in cancer patients.

A Bessede*1; A Marabelle*2; JP Guégan1; FX Danlos2; S Cousin3; F Peyraud3; N Chaput4,5,6 ;

M Spalato3; G Roubaud3; M Cabart3; M Khettab7; A Chaibi3; C Rey3; I Nafia1; FX Mahon3,7; JC

Soria8; A Italiano2,3,7

* These two authors contributed equally to the work

1. Explicyte, 229 cours de l’Argonne, Bordeaux, France

2. Département d’Innovation Précoce et d’Essais Thérapeutiques (DITEP), INSERM

f
oo
U1015 & CIC1428, Université Paris Saclay, Gustave Roussy, Villejuif, France

3. Department of Medicine, Institut Bergonié, Bordeaux, France

r
-p
4. Laboratory of Immunomonitoring in Oncology, Gustave Roussy Cancer Campus,
re
CNRS-UMS 3655 and INSERM-US23, Villejuif, France
lP

5. Faculty of Pharmacy, University Paris-Saclay, Chatenay-Malabry, France

6. Laboratory of Genetic Instability and Oncogenesis, UMR CNRS 8200, Gustave

na

Roussy, Université Paris-Saclay, Villejuif, France

ur

7. Faculty of Medicine, University of Bordeaux, Bordeaux, France

Jo

8. Department of Medicine, Gustave Roussy, Villejuif, France

Correspondence to: Dr. Antoine ITALIANO, MD, PhD

DITEP, Gustave Roussy, Villejuif, France

antoine.italiano@gustaveroussy.fr

1
ABSTRACT

Background: Acetaminophen (APAP) use has been associated with blunted vaccine immune

responses. This study aimed to assess APAP impact on immunotherapy efficacy in patients

with cancer.

Patients and Methods: Exposure to APAP was assessed by plasma analysis and was

correlated with clinical outcome in three independent cohorts of patients with advanced cancer

who were treated with immune checkpoint blockers (ICB). APAP immunomodulatory effects

were evaluated on a pre-clinical tumor model and on human peripheral blood mononuclear

f
cells (PBMCs) from healthy donors.

oo
Results: Detectable plasma APAP levels at treatment onset was associated with a significantly

r
worse clinical outcome in ICB-treated cancer patients, independently of other prognostic
-p
factors. APAP significantly reduced ICB efficacy in the pre-clinical MC38 model, as well as the
re
production of PD1 blockade-related interferon-γ secretion by human PBMCs. Moreover,
lP

reduction of ICB efficacy in vivo was associated with significantly increased tumor infiltration
na

by regulatory T cells (Tregs). Administration of APAP over 24 h induced a significant expansion

of peripheral Tregs in healthy individuals. In addition, interleukin-10, a crucial mediator of Treg-

ur

induced immune suppression, was significantly upregulated upon treatment with ICB in cancer
Jo

patients taking APAP.

Conclusion: This study provides strong pre-clinical and clinical evidence of the role of APAP

as a potential suppressor of antitumor immunity. Hence, APAP should be used with caution in

patients treated with ICB.

KEYWORDS: acetaminophen, immune checkpoint inhibitors, immunotherapy, cancer

2
HIGHLIGHTS

Acetaminophen has been shown to blunt the antibody response to vaccination

Acetaminophen may impact efficacy of immunotherapy in patients with cancer

Further research is needed to decipher the impact of acetaminophen on immunity

f
r oo
-p
re
lP
na
ur
Jo

3
INTRODUCTION

Pain is the most common symptom experienced by patients with advanced cancer.

Acetaminophen (APAP, commonly known as paracetamol) alone or in combination with a

weak opioid, such as codeine or tramadol, is usually considered as the first-line strategy to

manage mild-to-moderate pain in this setting1. Although generally considered to be safe,

evidence suggests that APAP may have negative immunomodulatory effects. Indeed,

preclinical studies demonstrated that APAP can inhibit the proliferation of immune cells and

the T cell-dependent antibody response2,3. Moreover, pioneer clinical studies have suggested

f
that APAP inhibits viral clearance and/or neutralizing antibody response in patients infected

oo
with chicken pox or rhinovirus4. More recently, randomized studies have shown that APAP has

r
a negative impact on vaccination response with decreased antibody levels in subjects
-p
receiving APAP for fever prophylaxis5,6. Given its potential to impair vaccine effectiveness, the
re
World Health Organization stated in 2015 that administration of APAP before or at the time of
lP

vaccination is not recommended7.

na

The development of immune checkpoint blockers (ICBs) was a revolutionary milestone in the

field of immuno-oncology. Anticytotoxic T lymphocyte-associated protein-4 (CTLA-4) and anti-

ur

programmed cell death (PD)-1/PD-ligand 1 (PD-L1) therapies, such as ipilimumab, nivolumab

Jo

or pembrolizumab, are now available for treating various malignancies, including non-small

cell lung cancer, melanoma, and bladder cancer8,9. The objective of our study was to

investigate whether APAP may impair the efficacy of ICB in patients with advanced cancer.

4
METHODS

Patients, treatments, and evaluation

Study design, eligibility criteria, and treatment were previously described for the CheckMate

025 trial10 (NCT01668784, sponsor: Bristol Myers Squibb). The main inclusion criteria for the

BIP (NCT02534649, sponsor: Institut Bergonié) and PREMIS studies (NCT03984318,

sponsor: Gustave Roussy) were age ≥ 18 years, histologically proven malignant tumor,

unresectable and/or metastatic disease, and at least one tumor evaluation by imaging after

immunotherapy onset. All patients included in the BIP and PREMIS studies were treated with

f
anti-PD-(L)1 either in monotherapy or combined with anti-CTLA-4 antibodies, either within

oo
clinical trials, or in the context of approved indications of the European Medicine Agency, or

r
within early access programs. The best response to treatment was evaluated according to
-p
Response Evaluation Criteria in Solid Tumors (RECIST) guidelines11. Routine follow-up and
re
treatment beyond progression therapeutic options were similar within the two studies. To
lP

investigate the pharmacodynamic impact of APAP on peripheral immune cells, four healthy
na

volunteers received 1,000 mg of APAP orally every 6 h, over a period of 24 h (total dose: 4 g).

Peripheral blood mononuclear cells (PBMCs) were collected at baseline and 2h after the last
ur

administration of APAP. This study was approved by the Institutional ethics review board of
Jo

the Institut Bergonié (Bordeaux, France). Written informed consent was obtained from all

subjects involved in the study.

Metabolomic profiling of plasma samples from the BIP study

Detection of APAP and of its metabolite APAP glucuronide was performed for the patients

included in the institutional profiling program BIP (NCT02534649; sponsor: Institut Bergonié)

via liquid chromatography–mass spectrometry. Details are provided in Supplementary

Methods.

Quantitative dosage of APAP

5
Quantitative dosage of APAP and of its metabolite APAP glucuronide was performed for the

patients included in the institutional profiling program PREMIS (NCT03984318; sponsor:

Gustave Roussy) via liquid chromatography–mass spectrometry. Details are provided in

Supplementary Methods.

Human PBMC profiling

PBMCs were isolated from whole blood of healthy donors before and 24 h after APAP dosing

using density gradient centrifugation with Lymphoprep (STEMCELL Technologies, Vancouver,

BC, Canada) as per the manufacturer’s instructions. PBMCs (5 × 105 cells) were first stained

f
oo
with Zombie NIR viability kit (Biolegend, San Diego, CA, USA) for 10 min and then blocked for

5 min with Human TruStain FcX (Biolegend) and True-Stain Monocyte Blocker (Biolegend)

r
-p
solutions. Cells were incubated for 15 min at 4 °C with cell surface marker antibodies, washed
re
in MACS buffer (2 mM EDTA, 0.5% bovine serum albumin, 1x phosphate-buffered saline),
lP

fixed for 30 min at 4 °C with Foxp3 Fixation/Permeabilization solution (eBioscience, Waltham,

MA, USA) and permeabilized with Permeabilization buffer (eBioscience) for 5 min at room
na

temperature. The cells were finally incubated with intracellular marker antibodies for 30 min at
ur

4 °C, washed twice, and resuspended in MACS buffer for analysis on the Novocyte Quanteon
Jo

flow cytometer (Agilent Technologies, Santa Clara, CA, USA). Two different antibody panels

were used for PBMC immunophenotyping: 1)

Zombie/CD45/CD3/CD4/CD8/CD19/CD25/CD56/PD1/CTLA4/TIGIT/TIM3/FoxP3/LAG3, and

2) Zombie/CD45/CD3+CD19/CD11c/CD11b/CD14/CD56/HLA-

R/CD123/PDL1/CD15/IDO1/Arginase1 (Supplementary Table 1). To normalize the signals

obtained before and after APAP dosing, PBMCs from a healthy donor were stained in parallel

of the samples and served as shared control for CytoNorm normalization12. Normalization and

cell phenotyping were both performed on FlowJo v10.8 (BD Biosciences, Franklin Lakes, NJ,

USA). Differences in cell populations and marker expression between pre- and post-treatment

samples were tested using a paired Student’s t-test. Only significant changes in marker

expression were illustrated on the heatmaps (pheatmap R package v1.0.12).

6
Plasma proteomics

Proteomic profiling of plasma samples from cancer patients included in the institutional profiling

program PREMIS (NCT03984318; sponsor: Gustave Roussy) was assessed using the Olink

Target 96 Inflammation panel (Olink Proteomics AB, Uppsala, Sweden) according to the

manufacturer's instructions. Details are provided in Supplementary Methods. Differences in

plasma collected at baseline and week 6 were assessed by paired Student’s t-test using the

limma R package (v3.48.3), and proteins with a fold change > 1.25 and a Benjamini-Hochberg-

adjusted p-value < 0.05 were extracted. Venn diagram was drawn using the ggvenn R package

f
(v0.1.9)

oo
In vitro functional assays

r
-p
PBMCs from three different healthy donors were isolated by Lymphoprep density gradient
re
centrifugation. Cells (1 × 105) were seeded in 96-well plates and treated with anti-CD3 (1
lP

µg/mL, Biolegend) with or without nivolumab (1 µg/mL, Selleckchem) along with increasing

doses of APAP (Sigma-Aldrich). After 72 h, supernatants were collected and interferon-γ

na

release was assessed by homogeneous time resolved fluorescence using the Human IFN
ur

gamma kit (Cisbio, Codolet, France).

Jo

Animal studies

All animal studies were performed under protocols approved by the Institutional Animal Care

and Use Committee of the University of Bordeaux (Bordeaux, France). Details are provided in

Supplementary Methods.

Statistical analysis

The cutoff date for statistical analysis of baseline demographic data and clinical outcome was

by 6/30/2021. Descriptive statistics were used to describe the distribution of variables in the

population. Progression-free survival (PFS) was defined as the time from the start of treatment

until disease progression, death, or last patient contact. Overall survival (OS) was defined as

the time from the start of treatment until death or last patient contact. Survival rates were

7
estimated using the Kaplan–Meier method (survival R package, v3.2-13). Patients from the

Checkmate025 and BIP studies were classified either as “High” or “Low” based on a threshold

value (1x105 mass spectrometry intensity) while patients from the PREMIS study were

classified as “Presence” or “Absence” according to the quantitation of APAP and APAP

glucuronide. Differences between groups were evaluated by chi-square test for categorical

variables and ANOVA with Tuckey tests or Wilcoxon tests for continuous variables. Prognostic

factors were planned to be identified by univariate and multivariate analyses using a Cox

regression model. Variables tested in univariate analysis included age, gender, tumor type,

f
presence of liver metastasis, presence of bone metastasis, number of metastatic sites,

oo
antibiotic use, steroid use, performance status, number of previous lines of treatment, lactate

r
dehydrogenase levels, and presence of detectable levels of APAP in the plasma. Variables
-p
associated with PFS and OS with a p-value < 0.05 in the univariate analysis were planned to
re
be included in the multivariate analysis. Analyses were performed in R using the survival
lP

analysis package (v0.2.0). All statistical tests were two-sided, and p  < 0.05 indicated statistical

significance.
na
ur
Jo

8
RESULTS

Since self-medication with APAP is highly prevalent among the general population, analysis of

medical records is not appropriate to accurately determine APAP exposure. We therefore

analyzed the publicly available serum metabolomics data from 297 patients with advanced

renal cell carcinoma and treated with nivolumab in the context of the randomized phase III trial

CheckMate 025 (NCT01668784)10. We found that patients with detectable levels of APAP or

APAP glucuronide had significantly worse OS than patients without detectable APAP levels at

treatment onset (Figure 1A, B and Supplementary Figure 1A). To note, correlations with

f
objective response rate and PFS were not analyzed as these data were not available. We then

oo
used the same untargeted mass spectrometry-based metabolomics approach to analyze

r
plasma samples of 34 patients included in the institutional profiling program BIP
-p
(NCT02534649) and treated with ICB for advanced disease. Their characteristics are
re
described in Supplementary Table 2. Significant levels of APAP or APAP glucuronide were
lP

detected in 50% of these patients. As shown in Figure 1E, patients exposed to APAP had
na

significantly lower objective response rate (0% vs. 29.4%; p = 0.015), and trended to have

worse PFS (median PFS, 1.87 vs. 4.72 months; 95% confidence interval [CI], 0.30–1.32; p =
ur

0.219; Figure 1C) and OS (median OS, 7.87 vs. 16.56 months; 95% CI, 0.3–1.63; p = 0.412;
Jo

Figure 1D) than patients without detectable APAP levels at treatment onset.

To investigate the robustness of these results, we also evaluated the levels of APAP using a

quantitative mass spectrometry approach in plasma samples collected from 297 patients

enrolled in the PREMIS study (NCT03984318). Their characteristics are described in Table 1.

We found that patients with detectable APAP levels had significantly worse PFS (median PFS:

2.63 vs. 5.03 months; 95% CI, 0.53–0.91; p = 0.009; Figure 1F) and OS (median OS: 8.43 vs.

14.93 months; 95% CI, 0.32–0.69; p < 0.0001; Figure 1G) compared with patients with APAP-

free plasma. Objective response rate was also numerically higher in the APAP-negative group

than in the APAP-positive group (28.9% vs 20.7%; p=0.106); Figure (Figure 1H), although

the difference did not reach statistical significance (except in patients with poor performance

9
status, Supplementary Figure 2). On multivariate analysis, APAP plasma levels remained

independently associated with both PFS and OS (Table 2).

To confirm the impact of APAP on ICB efficacy and provide mechanistic insights, we conducted

preclinical investigations using the MC38 colon tumor model, which has been shown to be

responsive to PD-1/PD-L1 blocking antibodies13. Tumor rejection rates were significantly lower

in mice treated with anti-PD-1/PD-L1 antibodies concomitantly with non-toxic doses of APAP

than in mice treated with anti-PD-1/PD-L1 antibodies alone, with a trend for overall survival

benefit (p=0.17), respectively (Supplementary Figure 3A, B). To investigate the mechanisms

f
oo
underlying APAP activity in vivo, we performed a flow-cytometry-based analysis of tumor-

infiltrating leukocytes 13 days after tumor inoculation (8 days after treatment initiation). An

r
-p
increase in tumor infiltration by regulatory T cells (Tregs) was observed in mice treated with

APAP and, to a higher and significant extent, in mice treated with both APAP and ICB
re
(Supplementary Figure 3D).
lP

To evaluate the direct impact of APAP on immune cells, we exposed human PBMCs from
na

healthy donors to anti-CD3 antibodies in the presence or absence of nivolumab (1 µg/mL) with
ur

increasing concentrations of APAP (0, 100, and 300 µM) for 72 h. As expected14,15, nivolumab

increased the anti-CD3-induced interferon-γ secretion, an event that was drastically limited in
Jo

the presence of APAP (Supplementary Figure 4). We then immunophenotyped PBMCs from

four healthy donors who received 1,000 mg of APAP every 6 h for a period of 24 h. As depicted

in Supplementary Figure 6A and B, APAP induced the expansion of Tregs in all donors, as

well as the expression of the coinhibitory receptors LAG3 and TIM3, which are associated with

a strong immunosuppressive phenotype16. Interestingly, APAP induced a concomitant

expansion of myeloid and plasmacytoid dendritic cells (mDCs and pDCs, respectively), the

latter being characterized by an overexpression of both indoleamine 2,3-dioxygenase 1 (IDO1)

and arginase immunosuppressive enzymes (Supplementary Figure 6C, D). Finally, to further

assess the immunomodulation effect of APAP in cancer patients treated with ICB, the major

surrogate cytokines were profiled in the plasma of patients included in the PREMIS study. In

10
total, 92 cytokines were measured using the Olink Target 96 inflammation panel. We found

that interleukin (IL)-10, a crucial mediator of immune suppression induced by Tregs17, and Flt3-

ligand, an essential growth factor for dendritic cells18–20, were significantly upregulated upon

treatment with ICB, exclusively in patients taking APAP (Supplementary Figure 7).

f
r oo
-p
re
lP
na
ur
Jo

11
DISCUSSION

In this study, we found that patients with advanced cancer taking APAP during immunotherapy

experience worse clinical outcomes, which suggests that APAP decreases T cell-mediated

antitumor immunity. It is unlikely that our data are the result of bias or unmeasured

confounding. First, exposure to acetaminophen was not inferred from medical records which

could have introduced the possibility of recall bias. Second, association with adverse outcome

was observed in three independent cohorts including two academic studies and one-industry

sponsored clinical trial which enrolled patients who were hyperselected (45 inclusion/exclusion

f
oo
criteria). Third, the multivariate analysis considered all the features that may be correlated with

both prognosis and need for taking acetaminophen including antibiotic or steroid use and

r
-p
presence of bone metastases. Fourth, in agreement with our clinical findings, we showed that

APAP reduces the efficacy of ICB in a well-characterized preclinical model of colorectal

re
cancer13. Finally, we were able to demonstrate in vitro, using human PBMCs-based functional
lP

assays, that APAP significantly limits the anti-PD-1 therapy-associated effect on interferon-γ
na

production. This observation is consistent with previous data showing that APAP decreases

the interferon-induced antiviral responses of cultured mammalian cells to influenza virus21.

ur
Jo

The first reports on APAP as an immune response modulator were published in the early 90’s4.

Nonetheless, Prymula et al. were the first to investigate this issue in detail in a randomized

study that comprised 9–16-week-old healthy infants5. They reported reduced immunogenicity

of common pediatric vaccines upon APAP use. The impact of APAP on immunogenicity was

further supported by post-hoc analyses from 10 previous trials with a reduction in antibody

concentrations that was consistent across almost all the antigens studied5. Similarly, a recent

systematic review of clinical studies comprising 2,775 patients reported that prophylactic APAP

administration negatively affected the immune response to pneumococcal conjugate vaccines

in children22.

12
Data related to the potential impact of APAP on antitumor immunity remain almost non-

existent. In a retrospective study, Kostner et al. analyzed the impact of body temperature

achieved during IL-2 infusion in patients with advanced melanoma23. Interestingly, they found

that IL-2-induced fever (peak temperature ≥ 39.5 °C) was associated with an improved survival.

However, when stratifying according to APAP use, the association between fever and

improved survival was not present among patients routinely receiving APAP, thereby

suggesting that APAP intake impairs antitumor immunity.

To date, the mechanisms involved in the immunomodulatory effect of APAP were unknown.

f
As suggested by Kostner et al.23, data from vaccination studies indicated that prevention of

oo
inflammation and fever is unlikely. Indeed, these studies showed that immune responses (and

r
-p
effect of APAP) in children with and without fever were similar5,22. Our preclinical in vivo

experiments indicated that the lower anticancer effects observed when APAP was combined
re
with ICB were associated with a significant increase in tumor infiltrating Tregs. These data are
lP

in line with a recent mice study reporting that APAP has an impact on the maternal immune
na

adaptation to pregnancy, by strongly increasing the proportion FoxP3+ Tregs in uterus-draining

lymph nodes24. FoxP3+ Tregs, which are essential for promoting maternal immune tolerance
ur

toward the fetus, are also involved in antitumor immune response suppression. Similarly, in a
Jo

mouse model of drug-induced liver injury, administration of APAP was shown to promote Treg

induction and IL-10 production via a protective feedback loop that physiologically aims to

alleviate APAP-induced liver injury, thus contributing to an immunosuppressive milieu25.

Strikingly, high-throughput proteomic analysis of the plasma allowed us to identify IL-10, a

crucial mediator of immune suppression induced by Tregs17, as being significantly upregulated

upon treatment with ICB exclusively in patients taking APAP. Moreover, immunophenotyping

of PBMCs from healthy donors before and after APAP consumption showed a significant

upregulation of Tregs, particularly of the subset expressing LAG3 and TIM3 immune

checkpoints, which are critical for Treg-mediated suppression26. In addition, APAP was also

found to promote the expression of Flt3-ligand, an essential growth factor for DCs18–20, and a

13
significant expansion of mDCs and pDCs expressing IDO1 and arginase, two subsets of

immune cell populations known to be associated with induction of immunosuppressive Tregs27.

Altogether, our data suggest that Tregs are key players that may underlie the

immunomodulatory effect of APAP, thus compromising ICB efficacy.

While some controversies still exist concerning the potential predisposing effect of

acetaminophen on cancer development, it is commonly considered as an innocuous drug28.

By combining preclinical experiments and high-throughput profiling of human clinical samples,

our study reports the most comprehensive picture of the immunomodulatory effects of APAP.

f
oo
A recent study raised concerns about the potential deleterious impact of APAP in patients with

coronavirus disease 201929. Our results confirm that more research should be performed to

r
-p
understand the impact of APAP on immunity and present a compelling case for caution in using

this drug in cancer patients treated with ICB. Whether this rule applies at immunotherapy onset
re
or all over the treatment duration, to all antipyretics, all regimens (ICB combined with cytotoxic
lP

chemotherapy or with tyrosine kinase inhibitors), and to other immuno-oncology agents

na

(antitumor vaccines, chimeric antigen receptor T cells) requires further investigations.

ur
Jo

14
ACKNOWLEDGEMENTS
The authors thank the patients who participated in this study and the investigators and
teams who conducted the study. The authors also acknowledge the contributions of
Céline Auzanneau, Maud NGO Camus and Claudio Nicotra for sample logistics.

FUNDING
This work was supported by the Conseil Regional Nouvelle Aquitaine, Fondation
Bergonié, Explicyte and Gustave Roussy. No grant number applicable.

DISCLOSURE
AI reports a consulting or advisory role with AstraZeneca, Bayer, Chugai, Deciphera,
Merck, Parthenon, Roche, Springworks; grants from AstraZeneca, Bayer, BMS, Merck,
MSD, Novartis, Pharmamar, Roche AB, JPG are employees of Explicyte JCS is

f
currently employee of Amgen. The other co-authors declare no conflict of interest.

r oo
REFERENCES
-p
1. Prescott LF. Paracetamol: past, present, and future. Am J Ther. March 2000;7(2):143–
re
147.

2. Ueno K, Yamaura K, Nakamura T, Satoh T, Yano S. Acetaminophen-induced

lP

immunosuppression associated with hepatotoxicity in mice. Res Commun Mol Pathol

Pharmacol. 2000;108(3–4):237–251.
na

3. Yamaura K, Ogawa K, Yonekawa T, Nakamura T, Yano S, Ueno K. Inhibition of the

antibody production by acetaminophen independent of liver injury in mice. Biol Pharm Bull.
February 2002;25(2):201–205.
ur

4. Graham NM, Burrell CJ, Douglas RM, Debelle P, Davies L. Adverse effects of aspirin,
Jo

acetaminophen, and ibuprofen on immune function, viral shedding, and clinical status in
rhinovirus-infected volunteers. J Infect Dis. December 1990;162(6):1277–1282.

5. Prymula R, Siegrist C-A, Chlibek R, Zemlickova H, Vackova M, Smetana J, et al. Effect of

prophylactic paracetamol administration at time of vaccination on febrile reactions and
antibody responses in children: two open-label, randomised controlled trials. Lancet.
October 17, 2009;374(9698):1339–1350.

6. Falup-Pecurariu O, Man SC, Neamtu ML, Chicin G, Baciu G, Pitic C, et al. Effects of
prophylactic ibuprofen and paracetamol administration on the immunogenicity and
reactogenicity of the 10-valent pneumococcal non-typeable Haemophilus influenzae
protein D conjugated vaccine (PHiD-CV) co-administered with DTPa-combined vaccines
in children: An open-label, randomized, controlled, non-inferiority trial. Hum Vaccin
Immunother. March 4, 2017;13(3):649–660.

7. ACIP Vaccine Administration Guidelines for Immunization | CDC [Internet]. 2022 [cited
May 5, 2022]. Available at: https://www.cdc.gov/vaccines/hcp/acip-recs/general-
recs/administration.html

15
8. Hodi FS, O’Day SJ, McDermott DF, Weber RW, Sosman JA, Haanen JB, et al. Improved
survival with ipilimumab in patients with metastatic melanoma. N Engl J Med.
2010;363(8):711–23.

9. Sun L, Zhang L, Yu J, Zhang Y, Pang X, Ma C, et al. Clinical efficacy and safety of anti-
PD-1/PD-L1 inhibitors for the treatment of advanced or metastatic cancer: a systematic
review and meta-analysis. Sci Rep. February 7, 2020;10(1):2083.

10. Motzer RJ, Escudier B, McDermott DF, George S, Hammers HJ, Srinivas S, et al.
Nivolumab versus Everolimus in Advanced Renal-Cell Carcinoma. New England Journal
of Medicine. November 5, 2015;373(19):1803–1813.

11. Eisenhauer EA, Therasse P, Bogaerts J, Schwartz LH, Sargent D, Ford R, et al. New
response evaluation criteria in solid tumours: Revised RECIST guideline (version 1.1).
European Journal of Cancer. January 2009;45(2):228–247.

12. Van Gassen S, Gaudilliere B, Angst MS, Saeys Y, Aghaeepour N. CytoNorm: A

f
oo
Normalization Algorithm for Cytometry Data. Cytometry Part A. 2020;97(3):268–278.

13. Lau J, Cheung J, Navarro A, Lianoglou S, Haley B, Totpal K, et al. Tumour and host cell

r
PD-L1 is required to mediate suppression of anti-tumour immunity in mice. Nat Commun.
February 21, 2017;8:14572. -p
re
14. Peng W, Liu C, Xu C, Lou Y, Chen J, Yang Y, et al. PD-1 blockade enhances T-cell
migration to tumors by elevating IFN-γ inducible chemokines. Cancer Res. October 15,
2012;72(20):5209–5218.
lP

15. Dulos J, Carven GJ, van Boxtel SJ, Evers S, Driessen-Engels LJA, Hobo W, et al. PD-1
blockade augments Th1 and Th17 and suppresses Th2 responses in peripheral blood from
na

patients with prostate and advanced melanoma cancer. J Immunother. March

2012;35(2):169–178.
ur

16. Ma Q, Liu J, Wu G, Teng M, Wang S, Cui M, et al. Co-expression of LAG3 and TIM3
identifies a potent Treg population that suppresses macrophage functions in colorectal
Jo

cancer patients. Clin Exp Pharmacol Physiol. October 2018;45(10):1002–1009.

17. Chaudhry A, Samstein RM, Treuting P, Liang Y, Pils MC, Heinrich J-M, et al. Interleukin-
10 signaling in regulatory T cells is required for suppression of Th17 cell-mediated
inflammation. Immunity. April 22, 2011;34(4):566–578.

18. McKenna HJ, Stocking KL, Miller RE, Brasel K, De Smedt T, Maraskovsky E, et al. Mice
lacking flt3 ligand have deficient hematopoiesis affecting hematopoietic progenitor cells,
dendritic cells, and natural killer cells. Blood. June 1, 2000;95(11):3489–3497.

19. Waskow C, Liu K, Darrasse-Jèze G, Guermonprez P, Ginhoux F, Merad M, et al. The

receptor tyrosine kinase Flt3 is required for dendritic cell development in peripheral
lymphoid tissues. Nat Immunol. June 2008;9(6):676–683.

20. Ginhoux F, Liu K, Helft J, Bogunovic M, Greter M, Hashimoto D, et al. The origin and
development of nonlymphoid tissue CD103+ DCs. J Exp Med. December 21,
2009;206(13):3115–3130.

21. Crocker JF, Digout SC, Lee SH, Rozee KR, Renton K, Field CA, et al. Effects of
antipyretics on mortality due to influenza B virus in a mouse model of Reye’s syndrome.
Clin Invest Med. October 1998;21(4–5):192–202.

16
22. Koufoglou E, Kourlaba G, Michos A. Effect of prophylactic administration of antipyretics on
the immune response to pneumococcal conjugate vaccines in children: a systematic
review. Pneumonia (Nathan). April 25, 2021;13(1):7.

23. Køstner AH, Ellegaard M-BB, Christensen IJ, Bastholt L, Schmidt H. Fever and the use of
paracetamol during IL-2-based immunotherapy in metastatic melanoma. Cancer Immunol
Immunother. March 2015;64(3):349–355.

24. Thiele K, Solano ME, Huber S, Flavell RA, Kessler T, Barikbin R, et al. Prenatal
acetaminophen affects maternal immune and endocrine adaptation to pregnancy, induces
placental damage, and impairs fetal development in mice. Am J Pathol. October
2015;185(10):2805–2818.

25. Wang X, Sun R, Chen Y, Lian Z-X, Wei H, Tian Z. Regulatory T cells ameliorate
acetaminophen-induced immune-mediated liver injury. Int Immunopharmacol. April
2015;25(2):293–301.

f
oo
26. Anderson AC, Joller N, Kuchroo VK. Lag-3, Tim-3, and TIGIT: Co-inhibitory Receptors with
Specialized Functions in Immune Regulation. Immunity. May 17, 2016;44(5):989–1004.

r
27. Kushwah R, Hu J. Role of dendritic cells in the induction of regulatory T cells. Cell Biosci.
May 24, 2011;1(1):20. -p
re
28. Weiss NS. Use of acetaminophen in relation to the occurrence of cancer: a review of
epidemiologic studies. Cancer Causes Control. December 2016;27(12):1411–1418.
lP

29. MANJANI L, DESAI N, KOHLI A, ARYA R, WOODS C, DESALE S. EFFECTS OF

ACETAMINOPHEN ON OUTCOMES IN PATIENTS HOSPITALIZED WITH COVID-19.
Chest. October 2021;160(4):A1072.
na
ur
Jo

17
FIGURES LEGENDS

Figure 1. Acetaminophen (APAP) exposure impairs the efficacy of immune checkpoint

blockers in patients with cancer. (A) Volcano plot representation of the Log Rank p-values of

overall survival (OS) (y axis) and delta median OS (x axis) associated with each plasmatic metabolite in

the CheckMate-025 cohort. Optimal cutoff value for each metabolite marker was used to categorize the

patients as ‘High’ or ‘Low’ status. (B) Kaplan Meier curve of OS according to baseline plasmatic

APAP/APAP glucuronide levels in the CheckMate-025 cohort. (C,D,F,G) Kaplan-Meier curves of the

progression-free survival (C,F) and OS (D,G) according to APAP/APAP glucuronide levels in the BIP

f
(C,D) and PREMIS (F,G) cohorts. (E,H) Proportion of responder (R) and non-responder (NR) patients

oo
in the BIP (E) and PREMIS (H) cohorts according to their baseline plasmatic levels of APAP/APAP

glucuronide classified as ‘Absence’ and ‘Presence’. p-value was calculated by Chi-squared test.

r
-p
re
lP
na
ur
Jo

18
Table 1. Baseline clinical characteristics of patients from the PREMIS Cohort.

Characteristics Patients (N = 297)

Age - yr
Median (range) 63 (26 – 101)
Sex – no. (%)
Female 121 (40.7)
Male 176 (59.3)
ECOG performance-status score* –(%)
0 106 (35.7)
1 142 (47.8)
≥2 48 (16.2)
NA 1 (0.3)
Antibiotic use no. (%)
Yes 6 (2)
No 291 (98)

f
Steroids use no. (%)

oo
Yes 24 (8.1)
No 273 (91.9)
Number of previous treatment regimens –

r
no. (%)
0-1
≥2
-p 94 (31.6)
176 (59.3)
re
NA 27 (9.1)
Number of metastatic sites – no. (%)
0-1 69 (23.2)
lP

≥2 228 (76.8)
Liver metastases no. (%)
Yes 75 (25.2)
na

No 222 (74.8)
Bone metastases no. (%)
Yes 91 (30.6)
ur

No 206 (69.4)
Lactate dehydrogenase levels no(%)
≤ upper limit of normal (ULN) 241 (81.1)
Jo

> upper limit of normal (ULN) 56 (18.9)

Type of immunotherapy – no. (%)
Anti-PD1 164 (55.2)
Anti-PD-L1 102 (34.3)
Combination of immunotherapies 31 (10.4)
Type of cancer
Non-small cell lung cancer 111 (37.4)
Melanoma 24 (8.1)
Soft-tissue sarcoma 22 (7.4)
Renal cell carcinoma 18 (6.1)
Urothelial carcinoma 15 (5)
Others1 107 (36)
RECIST**
PD 186 (62.6)
SD 34 (11.4)
PR 63 (21.2)
CR 12 (4)
NA 2 (0.7)

* Performance-status scores on the Eastern Cooperative Oncology Group (ECOG) scale range from 0
(no disability) to 5 (death).
** Response Evaluation Criteria In Solid Tumors
1Cervix carcinoma, colorectal cancer, gastric cancer, head and neck cancer, renal cancer, soft-tissue
sarcoma, triple negative breast carcinoma

f
r oo
-p
re
lP
na
ur
Jo
Jo
ur
na
lP
re
-p
r oo
f
Table 2. Multivariate analysis of progression-free survival and overall survival in

the PREMIS cohort.

Progression-Free Survival

Independent variables Hazard ratio 95% confidence interval P-value

Age (≥ 63 years, median value) 0.71 0.54–0.94 0.018
APAP exposure (Yes) 1.43 1.07–1.91 0.015
LDH levels (> ULN*) 1.55 1.17–2.06 0.002
ECOG status (≥ 2) 2.05 1.41–2.96 < 0.001
Liver metastasis (Yes) 2.10 1.54–2.87 < 0.001

f
Overall Survival

oo
Independent variables Hazard ratio 95% confidence interval P-value
Sex (Male) 1.45 0.99–2.11 0.056

r
APAP exposure (Yes)
LDH levels (> ULN*)
-p
1.78
1.91
1.18–2.68
1.30–2.81
0.006
0.001
re
Liver metastasis (Yes) 2.60 1.76–3.85 < 0.001
lP

ECOG status (≥ 2) 3.57 2.27–5.60 < 0.001

na

* ULN= upper limit of normal

ur
Jo
Jo
ur
na
lP
re
-p
r oo
f
Jo
ur
na
lP
re
-p
ro
of
 Annals of Oncology 29: 1437–1444, 2018
doi:10.1093/annonc/mdy103
Published online 30 March 2018

ORIGINAL ARTICLE

Negative association of antibiotics on clinical activity

of immune checkpoint inhibitors in patients with
advanced renal cell and non-small-cell lung cancer

L. Derosa1,2,3†, M. D. Hellmann4,5,6†, M. Spaziano7, D. Halpenny8, M. Fidelle1,2,3, H. Rizvi9, N. Long8,

A. J. Plodkowski8, K. C. Arbour4, J. E. Chaft4,5, J. A. Rouche10, L. Zitvogel1,2,3,11, G. Zalcman12,
L. Albiges1,3,13,14, B. Escudier1,13,14 & B. Routy1,2,3,15,16*
1
Gustave Roussy Cancer Campus (GRCC), Villejuif; 2Institut National de la Santé Et de la Recherche Médicale (INSERM) U1015; Equipe Labellisée—Ligue Nationale
Contre le Cancer, Villejuif; 3Université Paris-Sud, Université Paris-Saclay, Gustave Roussy, Villejuif, France; 4Department of Medicine, Thoracic Oncology Service,
Memorial Sloan Kettering Cancer Center, New York; 5Department of Medicine, Weill Cornell Medical College, New York; 6Parker Institute for Cancer
Immunotherapy, New York, USA; 7Cardiology Division, Department of Medicine, McGill University, Montreal, Canada; 8Department of Radiology; 9Druckenmiller
Center for Lung Cancer Research, Memorial Sloan Kettering Cancer Center, New York, USA; 10Department of Imaging, Gustave Roussy, Villejuif; 11Center of Clinical
Investigations in Biotherapies of Cancer (CICBT) 1428, Villejuif; 12Thoracic Oncology Department-CIC1425/CLIP2 Paris-Nord, Hospital Bichat-Claude Bernard, AP-HP,
University Paris-Diderot, Paris; 13Department of Medical Oncology, Gustave Roussy, Villejuif; 14Immunologie Intégrative des Tumeurs et Génétique Oncologique,
GRCC, Villejuif, France; 15Hematology-Oncology Division, Department of Medicine, Centre Hospitalier de l’Université de Montréal (CHUM), Montréal; 16Centre de
Recherche du Centre Hospitalier de l’Université de Montréal (CRCHUM), Montréal, Canada

*Correspondence to: Dr Bertrand Routy, CRCHUM, 900 St-Denis Street R10.474, Montreal, QC, Canada H2X 0A9. Tel: þ1-514-890-8000 (Ext. 31335); E-mail: bertrand.routy@umontreal.ca
†
Both authors contributed equally as senior authors.

Background: The composition of gut microbiota affects antitumor immune responses, preclinical and clinical outcome
following immune checkpoint inhibitors (ICI) in cancer. Antibiotics (ATB) alter gut microbiota diversity and composition leading
to dysbiosis, which may affect effectiveness of ICI.
Patients and methods: We examined patients with advanced renal cell carcinoma (RCC) and non-small-cell lung cancer
(NSCLC) treated with anti-programmed cell death ligand-1 mAb monotherapy or combination at two academic institutions.
Those receiving ATB within 30 days of beginning ICI were compared with those who did not. Objective response, progression-
free survival (PFS) determined by RECIST1.1 and overall survival (OS) were assessed.
Results: Sixteen of 121 (13%) RCC patients and 48 of 239 (20%) NSCLC patients received ATB. The most common ATB were
b-lactam or quinolones for pneumonia or urinary tract infections. In RCC patients, ATB compared with no ATB was associated
with increased risk of primary progressive disease (PD) (75% versus 22%, P < 0.01), shorter PFS [median 1.9 versus 7.4 months,
hazard ratio (HR) 3.1, 95% confidence interval (CI) 1.4–6.9, P < 0.01], and shorter OS (median 17.3 versus 30.6 months, HR 3.5, 95%
CI 1.1–10.8, P ¼ 0.03). In NSCLC patients, ATB was associated with similar rates of primary PD (52% versus 43%, P ¼ 0.26) but
decreased PFS (median 1.9 versus 3.8 months, HR 1.5, 95% CI 1.0–2.2, P ¼ 0.03) and OS (median 7.9 versus 24.6 months, HR 4.4,
95% CI 2.6–7.7, P < 0.01). In multivariate analyses, the impact of ATB remained significant for PFS in RCC and for OS in NSCLC.
Conclusion: ATB were associated with reduced clinical benefit from ICI in RCC and NSCLC. Modulatation of ATB-related
dysbiosis and gut microbiota composition may be a strategy to improve clinical outcomes with ICI.
Key words: antibiotics, immune checkpoint inhibitors, non-small-cell lung cancer, renal cell carcinoma, microbiota

Introduction axis have changed the therapeutic landscape in both renal cell car-
cinoma (RCC) and non-small-cell lung cancer (NSCLC). In RCC,
Immune checkpoint inhibitors (ICIs) targeting the programmed higher objective response rate and longer overall survival (OS)
cell death-1 (PD-1) and programmed cell death ligand-1 (PD-L1) were seen with nivolumab (anti-PD-1 mAb) compared with

C The Author(s) 2018. Published by Oxford University Press on behalf of the European Society for Medical Oncology.
V
All rights reserved. For permissions, please email: journals.permissions@oup.com.
Original article
everolimus [1] and represents a therapeutic option in relapsed
RCC. The robust clinical effect of anti-PD-1 mAb was observed
regarless of PD-L1 expression and standard prognostic factors [2].
In NSCLC, anti-PD-(L)1 mAb improved response rates and sur-
vival in a subset of NSCLC patients [3–5] in both first line or se-
cond line but few clinical features reliably predict benefit.
Primary resistance to ICI is common in patients with both
RCC and NSCLC, ranging from 35% to 44% and remains unpre-
dictable [1, 3]. Much recent effort has focused on biomarkers of
response to immunotherapy (e.g. PD-L1 [6], mutation burden
[5], gene expression signatures of inflammation [7, 8]) but the
identification of more reliable predictors associated with resist-
ance (primary or acquired) is also critical to instruct new strat-
egies to improve precision and broaden responder groups.
Building on recent data [9, 10], we hypothesized that the modu-
lation of gut microbiota by antibiotics (ATB) may be associated
with resistance to anti-PD-1 mAb. The intestinal microbiota repre-
sents a complex ecosystem essential for maintaining gut homeosta-
sis and prevent systemic inflammation [11, 12]. The interactions
between the host and micro-organisms have been identified as a
complex, inter-connected network where certain microbes tailor
local and systemic immune system [13]. ATB are known to affect
gut microbiota, including loss of distinct species (poor diversity),
favoring expansion of others, consequently shifting the metabolic
capacity [14]. ATB-induced dysbiosis has been associated with a
variety of chronic inflammatory disorders [15–17]. In oncology,
converging findings demonstrate the negative impact of ATB-
induced dysbiosis in mice. We built upon this experience to ex-
pand the series of patients examined using patients with advanced
RCC and NSCLC treated with anti-PD-(L)1 therapy at two differ-
ent centers to evaluate the impact of ATB on resistance.
Methods
Patients
Patients with advanced RCC (n ¼ 121) at Gustave Roussy and NSCLC
(n ¼ 239) at Memorial Sloan Kettering Cancer Center treated with ICI
were identified. NSCLC patients and half of RCC patients have been pre-
viously reported [10] but have been fully reanalyzed for multivariate ana-
lysis. All patients with RCC received anti-PD-1 or anti-PD-L1 (PD-(L)1)
mAb alone or in combination with anti-CTLA-4 mAb or bevacizumab
on clinical trials. Patients with NSCLC received anti-PD-(L)1 mAb alone
or in combination with anti-CTLA-4 mAb.
Patient records were reviewed to determine any oral or intravenous
ATB use within the 30 or 60 days before the start of anti-PD-(L)1 therapy.
The class of antibiotic, indication, route of administration and duration
were collected. Clinicopathologic characteristics were collected for all
patients. In RCC patients, additional features included cumulative size of
tumor burden (<10 cm versus  10 cm) [18] and site of metastases. In
NSCLC patients, smoking status and PD-L1 expression (high defined as
50% expression or low/no expression) were also collected if available.
For the evaluation of tumor response, CT scans were reviewed by local
specialized radiologists (JAR at GR; DH, NL, AJP at MSKCC) and re-
sponse was determined by Response Evaluation Criteria in Solid Tumors
(RECIST) version 1.1 [19]. All patients were followed-up until death or
data lock (September 2017 for RCC and March 2017 for NSCLC).
Statistical analysis
Patient characteristics were described according to the status of ATB (ATB
versus no ATB) and compared using Fisher or chi-squared test for
1438 | Derosa et al.
Annals of Oncology
categorical data and t-test for continuous data. Two ATB therapeutic win-
dows were analyzed, either ATB prescribed 30 days or up to 60 days before
the first injection of ICI, termed ATB-30 and ATB-60, respectively.
Survival curves were estimated by the Kaplan–Meier method and com-
pared with the Log-rank test (univariate analysis). Univariate hazard
ratios (HRs) were calculated using log-rank method. Multivariable Cox
regression model was used to determine HRs and 95% confidence inter-
vals (CIs) for progression-free survival (PFS) and OS between ATB and
no ATB, adjusting for other clinicopathologic features. Statistical tests
were two-sided, and a P-value <0.05 was considered statistically signifi-
cant. Statistical analyses were carried out using SPSS.
Results
One hundred and twenty-one patients with RCC and 239 patients
with NSCLC were examined in independent cohorts from two
academic cancer centers. The frequency of ATB use within
30 days of starting PD-(L)1 therapy was relatively similar in
NSCLC (20%) and RCC (13%) patients. b-Lactams 6 inhibitors
were the most commonly administered ATB in both groups
(82% of RCC and 32% of NSCLC) (supplementary Table S1,
available at Annals of Oncology online). Quinolones and sulfona-
mides were also frequently used in NSCLC cohorts.
Among RCC patients, 106 of 121 (88%) received PD-(L)1 mono-
therapy, 10 (8%) PD-(L)1 plus CTLA-4 mAbs and 5 (4%) PD-(L)1
mAbs plus bevacizumab. All RCC patients were treated in clinical
trials. Clinicopathologic characteristics were typical of patients with
advanced RCC and were generally well balanced between those who
received ATB or not, with the exception of tumor burden (larger
tumors in the ATB group) and lines of prior therapy (2þ prior lines
more common in the ATB group) (Table 1A). No patients with
RCC were hospitalized within 30 days of starting PD-(L)1 therapy.
Among NSCLC patients, 48 (20%) patients received ATB with-
in 30 days of PD-(L)1 therapy. Two hundred and five (86%)
received PD-(L)1 monotherapy and 34 (14%) received PD-(L)1
plus CTLA-4 mAbs. Fifty-four (23%) were enrolled in clinical tri-
als. Clinicopathologic features were also typical of patients with
advanced NSCLC, well balanced between the ATB versus no ATB
groups with the exception of recent hospitalization (more com-
mon in the ATB group) and more patients included in the clinical
trials in the no ATB group (Table 1B).
In patients with RCC, recent ATB was associated with
increased rate of primary progressive disease (PD) (75% versus
22%, P < 0.01) (Figure 1A). PFS and OS were also shorter in
those patients with ATB than in those with no ATB (median PFS,
1.9 months versus 7.4 months, HR 3.1, 95% CI 1.4–6.9, P < 0.01;
median OS, 17.3 months versus 30.6 months, HR 3.5, 95% CI
1.1–10.8, P ¼ 0.03) (Figure 1B and C).
In patients with NSCLC, recent ATB was not associated with
an increased rate of primary PD (52% versus 43%, P ¼ 0.26)
(Figure 1D). Similar to RCC, however, PFS and OS were signifi-
cantly shorter in NSCLC patients with ATB than in those with no
ATB (median PFS, 1.9 months versus 3.8 months, HR 1.5, 95%
CI 1.0–2.2, P ¼ 0.03; median OS, 7.9 months versus 24.6 months,
HR 4.4, 95% CI 2.6–7.7, P < 0.01) (Figure 1E and F).
To assess the robustness of the observation that recent ATB was
associated with decreased benefit with ICI, we also examined the
effect of ATB within 60 days of starting therapy [ATB-60 days,
n ¼ 22 for RCC (18%); n ¼ 68 for NSCLC (28%)] (supplementary
Tables S2 and S3, available at Annals of Oncology online). Taking
Volume 29 | Issue 6 | 2018
Annals of Oncology Original article
Table 1A. Baseline characteristics of renal cell carcinoma (RCC) cohort

Characteristics Total ATB 30-0 No ATB P-value

(n 5 121) (n 5 16) (n 5 105)

Age, years Median 61 61 61

Range 28–83 29–83 30–82
Age, years, n (%) <65 79 (65) 8 (50) 69 (66) 0.22
65 42 (35) 8 (50) 36 (34)
Gender, n (%) Male 80 (66) 9 (56) 71 (68) 0.37
Female 41 (34) 7 (44) 34 (32)
Nephrectomy, n (%) Yes 103 (85) 13 (82) 90 (86) 0.64
No 18 (15) 3 (18) 15 (14)
Histology, n (%) Clear cell 115 (95) 14 (88) 101 (96) 0.13
Nonclear cell 6 (5) 2 (12) 4 (4)
IMDC risk group, n (%) Good 25 (21) 3 (19) 22 (21) 0.96
Intermediate 72 (59) 10 (62) 62 (59)
Poor 24 (20) 3 (19) 21 (20)
Tumor burden (mm), n (%) <100 94 (78) 9 (56) 85 (81) 0.03
100 27 (22) 7 (44) 20 (19)
Site of metastasis, n (%) Lung 84 (69) 8 (50) 76 (72) 0.09
Liver 34 (28) 8 (50) 26 (23)
Bone 33 (27) 5 (31) 28 (27)
Brain 14 (12) 0 (0) 14 (13)
Number of prior lines, n (%) 0–1 69 (57) 4 (25) 65 (62) <0.01
2 52 (43) 12 (75) 40 (38)
Treatment, n (%) a-PD-(L)1 mAb 106 (88) 14 (88) 92 (88) 0.86
a-PD-(L)1 mAb þ a-CTLA-4 10 (8) 1 (6) 9 (8)
a-PD-(L)1 mAb þ Bev 5 (4) 1 (6) 4 (4)
ATB, n (%)a Prophylaxis 0 0 0
Therapy 16 (13) 16 (100) 0

a
List of ATB and indication for prophylaxis and therapy are available in supplementary Table S1, available at Annals of Oncology online.
ATB, antibiotics; IMDC, International Metastatic Renal Cell Carcinoma Database Consortium (includes: Karnofsky performance status, time from diagnosis
to treatment, hemoglobin, serum calcium concentration, neutrophil and platelet counts); a, anti; PD-(L)1, PD-1 or PD-L1; mAb, monoclonal antibody; Bev,
bevacizumab.
Bold values reflect statistical significant; *P<0.05, **P<0.01.

into account that gut dysbiosis from ATB can take 1–3 months to
normalize [20], we hypothesized that a trend for similar results
would be seen with the extended timeline. In RCC, ATB-60 days
remained associated with increased risk of primary progression
(P < 0.01) (supplementary Figure S1A, available at Annals of
Oncology online), shorter PFS (median PFS, 3.1 months versus
7.4 months, HR 2.3, 95% CI 1.2–4.4, P < 0.01) (supplementary
Figure S1B, available at Annals of Oncology online) and trended to-
ward worse OS (median OS, 23.4 months versus 30 months, HR
1.9, 95% CI 0.8–4.7, P ¼ 0.15) (supplementary Figure S1C, avail-
able at Annals of Oncology online). In NSCLC, ATB-60 days com-
pared with no ATB group was not different in terms of objective
response or PFS (supplementary Figure S1D and E, available at
Annals of Oncology online), but remained significantly associated
with shorter OS (median OS, 9.8 months versus 21.9 months, HR
2.0, 95% CI 1.3–3.2, P < 0.01) (supplementary Figure S1F, avail-
able at Annals of Oncology online).
We next examined the impact of ATB on PFS and OS within
individual subgroups of patients. As noted, there were a few vari-
ables in both RCC and NSCLC cohorts that were more common
in the respective ATB groups. In both RCC and NSCLC cohorts,
although not statistically significant due to small numbers, ATB
group was generally associated with worse PFS and OS within
nearly every subgroup examined (Figure 2).
Finally, we carried out a multivariate analysis of the effect of
ATB administration, taking into account classical prognostic
factors relevant to RCC and NSCLC, respectively. In RCC, both
ATB and tumor burden were significantly associated with worse
PFS and OS as univariate (Tables 2A and 2B), and were ana-
lyzed in multivariate modeling. ATB and tumor burden
remained independently associated with worse PFS in multi-
variate analysis (HR for ATB 2.0, P < 0.01), while neither vari-
able remained statistically significant for OS (HR for ATB 2.1,
P ¼ 0.11). In NSCLC, ATB and several other clinicopathologic
variables were associated with PFS or OS as univariate and fur-
ther evaluated in multivariate model (Tables 3A and 3B). In the
Cox regression analysis, ATB were not significantly associated
with worse PFS (HR 1.3, 95% CI 0.9–1.8, P ¼ 0.17) but
remained significantly associated with OS (HR 2.5, 95% CI
1.6–3.7, P < 0.01).

Volume 29 | Issue 6 | 2018 doi:10.1093/annonc/mdy103 | 1439

Original article Annals of Oncology
Table 1B. Baseline characteristics of non-small-cell lung cancer (NSCLC) cohort

Characteristics Total ATB 30-0 No ATB P-value

(n 5 239) (n 5 48) (n 5 191)

Age, years Median 66 63 66

Range (22–92) (31–92) (22–88)
Age, years, n (%) <65 110 (46) 27 (56) 83 (43) 0.12
65 129 (54) 21 (44) 108 (57)
Gender, n (%) Male 118 (49) 24 (50) 94 (49) 0.92
Female 121 (51) 24 (50) 97 (51)
ECOG performance status, n (%) 0 53 (22) 5 (11) 48 (25) 0.05
1 183 (77) 43 (89) 140 (74)
2 3 (1) 0 3 (1)
Histology, n (%) Squamous 34 (14) 6 (13) 28 (15) 0.71
Nonsquamous 205 (86) 42 (88) 163 (85)
Smoking status, n (%) Smoker 193 (81) 37 (77) 156 (82) 0.47
Nonsmoker 46 (19) 11 (23) 35 (18)
Number of prior lines, n (%) <3 178 (74) 33 (69) 145 (76) 0.31
3 61 (26) 15 (31) 46 (24)
Hospitalization <30 days, n (%) Yes 20 (8) 16 (33) 4 (2) <0.01
No 219 (93) 32 (67) 187 (98)
Treatment, n (%) a-PD-(L)1 mAb 205 (86) 45 (94) 160 (84) 0.08
a-PD-(L)1 mAb þ a-CTLA-4 34 (14) 3 (6) 31 (16)
Clinical trial, n (%) Yes 54 (23) 5 (11) 49 (25) 0.02
No 185 (77) 43 (89) 142 (74)
PD-L1 expression, n (%) High (50%) 21 (9) 6 (12) 15 (8) 0.12
Low (<50%) 64 (27) 9 (19) 55 (29)
Unknown 154 (64) 33 (69) 121 (63)
ATB, n (%)a Prophylaxis 15 (6) 15 (31) 0
Therapy 33 (14) 33 (69) 0

a
List of ATB and indication for prophylaxis and therapy are available in supplemental Table S1, available at Annals of Oncology online.
ECOG, Eastern Cooperative Oncology Group.
Bold values reflect statistical significant; *P<0.05, **P<0.01.

Discussion taxonomic richness [21]. Another level of complexity exists in the

This study reports the substantial rate of ATB use (13%–20%)
proximal to initiating ICI and the potential clinical impact on
benefit from ICI. In independent cohorts of RCC and NSCLC fol-
lowed-up in two cancer centers, prior ATB was associated with
worse outcomes with ICI.
More precisely, patients in ATB group treated with ICI had a
higher rate of PD in RCC and a significant reduction in PFS and
OS compared with those with no ATB in both cancer types. ATB
appears to be a determinant of poor prognosis in the context of
ICI independently of classical prognostic markers and within in-
dividual patient subgroups. This paper builds upon our prior
publication [10] on the negative impact of ATB, which included
249 patients diagnosed with advanced NSCLC (n ¼ 140), RCC
(n ¼ 67) and urothelial cancer (n ¼ 42) treated with anti-PD-1/
PD-L1 mAb. In that report ATB use was defined as 60 days before
and 30 days after the first injection of ICI, and ATB used was asso-
ciated with decreased PFS from 3.5 to 4.1 months and OS from
11.5 to 20.6 months compared with the no ATB group [10].
In this expanding field, ATB are recognized to be able to shift
the microbiota composition temporally. ATB dysbiosis is associ-
ated with a decrease in microbiota diversity and impacts on
perturbation of microbiome according to the class, the duration,
the route of ATB administration and the presence of resistant
commensals which play a key role in microbiota recolonization
and recomposition. 16S ribosomal RNA sequencing of gene
amplicons of patients’ feces revealed that microbiota returns to
its baseline within 1–3 months after ATB discontinuation.
However, some bacteria may take years to fully recover [14, 20].
To ascertain the difference in ATB timing, we carried out a
sub-group analysis for patients receiving ATB 60 days before
starting ICI. Interestingly, the impact of ATB 60 days before was
not as potent as within the first 30 days before ICI. Partial micro-
biota recovery following ATB might explain the difference in clin-
ical outcomes observed.
Very recent reports highlighted that the composition of gut
microbiota may dictate, at least in part, the anticancer activity of
ICI. In three different cancer types, patients with higher diversity
or richness of the stool sample at baseline correlated with a better
clinical outcome at 6 months. In metastatic melanoma, patients
with ‘favorable’ microbiota composition were enriched in
Ruminococcaceae family members and Bifidobacterium spp. This
favorable microbiota translated into best overall response rate at
6 months [22, 23]. The third study in patients with NSCLC and

1440 | Derosa et al. Volume 29 | Issue 6 | 2018

Annals of Oncology Original article
A P<0.01 B C
100 4% 100 Median PFS 100 Median OS
13% No ATB: 7.4 mo No ATB: 30.6 mo

Progression-free survival (%)

22% ATB: 1.9 mo ATB: 17.3 mo
80 13% 80 80

Overall survival (%)

HR (95%,CI), 3.1 (1.4-6.9) HR (95%,CI), 3.5 (1.1-10.8)
Patients (%) P <0.01 P =0.03
60 60 60
RCC

RCC

RCC
52%
40 75% 40 40

20 20 20
22%
0 0 0
ATB No ATB 0 5 10 15 20 25 30 0 10 20 30 40
n=16 n=105 105 69 35 20 5 4 1 105 81 25 12 6
16 5 3 2 0 0 0 16 10 2 0 0
CR SD
Months Months
PR PD

D E F
100 Median PFS 100 Median OS
100 2% No ATB: 3.8 mo No ATB: 24.6 mo

Progression-free survival (%)

13%
21% ATB: 1.9 mo ATB: 7.9 mo
80 HR (95%,CI), 1.5 (1.0-2.2) 80 HR (95%,CI), 4.4 (2.6-7.6)

Overall survival (%)

80
P =0.03 P<0.01
35%
Patients (%)

NSCLC 60

NSCLC
60
60
NSCLC

34%

40 40
40

52% 20 20
20 43%

0 0
0 0 5 10 15 20 25 30 0 10 20 30 40
ATB No ATB
191 72 33 21 10 5 2
n=48 n=191 191 101 39 20 9
48 17 8 4 2 0 0
Months Months

Figure 1. Best overall response (A), progression-free survival (PFS) (B), and overall survival (OS) (C) in patients with RCC treated with ICI, strati-
fied by use of ATB within 30 days of initiating ICI. Best overall response (D), PFS (E), and OS (F) in patients with NSCLC treated with ICI, strati-
fied by use of ATB within 30 days of initiating ICI. P-values calculated with chi-squared and log-rank tests.

A PFS in RCC B OS in RCC

P-value for P-value for

interaction interaction
All patients (n=121) All patients (n=121)

>65 (n=44) >65 (n=44)

Age 0.11 Age 0.26
<65 (n=77) <65 (n=77)

Male (n=41) Male (n=41)

Sex 0.40 Sex 0.45
Female (n=80) Female (n=80)

Poor (n=24) Poor (n=24)

IMDC Intermediate (n=72) 0.67 IMDC Intermediate (n=72) 0.23

Good (n=25) Good (n=25)

<100 cm (n=94) <100 cm (n=94)

Tumor burden 0.07 Tumor burden 0.53
>100 cm (n=27) >100 cm (n=27)

0-1 (n=69) Number of 0-1 (n=69)

Number of 0.58 0.61
prior lines prior lines
≥2 (n=52) ≥2 (n=52)

0.5 1 2 3 4 5 10 15 0.2 0.5 1 2 3 4 5 10 20

Antibiotics better No antibiotics better Antibiotics better No antibiotics better

C PFS in NSCLC D OS in NSCLC

P-value for P-value for
All patients (n=239) interaction All patients (n=239) interaction

>65 (n=124) >65 (n=124)

0.20 Age 0.10
Age <65 (n=115) <65 (n=115)

Male (n=118) Male (n=118)

Sex 0.15 Sex 0.13
Female (n=121) Female (n=121)

Squamous (n=34) Squamous (n=34)

Histology 0.49 Histology 0.12
Non-squamous (n=205) Non-squamous (n=205)

Number of 3 or more (n=61) Number of 3 or more (n=61)

0.23 0.32
prior lines prior lines 2 or less (n=178)
2 or less (n=178)

1 or more (n=186) 1 or more (n=186)

ECOG 0.55 ECOG 0.48
0 (n=53) 0 (n=53)

Yes (n=54) Yes (n=54)

Clinical trial 0.71 Clinical trial 0.85
No (n=185) No (n=185)

Yes (n=34) Yes (n=34)

Hospitalization 0.16 Hospitalization 0.006
No (n=205) No (n=205)

Treatment Treatment
α-PD-(L)1 mAb (n=205) α-PD-(L)1 mAb (n=205)
0.76 0.42
α-PD-(L)1 mAb + α-CTLA-4 (n=34) α-PD-(L)1 mAb + α-CTLA-4 (n=34)

0.5 1 2 3 4 5 0.5 1 2 3 4 5 10

Antibiotics better No antibiotics better Antibiotics better No antibiotics better

Figure 2. Subgroup analyses of independent prognostic factors for PFS (A and B) and OS (C and D) stratification in RCC and NSCLC, respect-
ively. P-value for interaction calculated with Cox proportional hazards model.

Volume 29 | Issue 6 | 2018 doi:10.1093/annonc/mdy103 | 1441

Original article Annals of Oncology
Table 2A. Uni- and multivariate analyses for progression-free survival (PFS) Table 3A. Uni- and multivariate analyses for PFS in non-small-cell lung can-
in renal cell carcinoma cohort cer cohort

Univariate analysis Multivariate analysis Univariate analysis Multivariate analysis

Prognostic factor PFS P-value PFS P-value Prognostic factor PFS P-value PFS P-value
HR (95% CI) HR (95% CI) HR (95% CI) HR (95% CI)

ATB 30-0/No ATB 2.3 (1.3–4.0) <0.01 2.2 (1.3–3.3) <0.01 ATB 30-0/No ATB 1.4 (1.0– 2.0) 0.04 1.3 (0.9–1.8) 0.17
Age 1.2 (0.8–1.8) 0.43 – – Age 1.2 (0.9–1.6) 0.2 – –
65 years/<65 years  65 years/<65 years
IMDC risk group 1.0 (0.6–1.7) 0.92 – – Histology 1.1 (0.8–1.5) 0.92 – –
Intermediate/good Squamous/nonsquamous
Poor/good 1.7 (0.9–3.2) 0.11 Smoking status 0.7 (0.5–1.0) 0.04 0.7 (0.5–1.0) 0.03
Tumor burden 2.1 (1.3–3.3 ) <0.01 2.0 (1.2–3.2) <0.01 Smoker/nonsmoker
Number of prior regimens 1.4 (1.0–1.9) 0.05 1.2 (0.8–1.6) 0.40
HR, hazard ratio; CI, confidence interval. 3/<3
Bold values reflect statistical significant; *P<0.05, **P<0.01. ECOG performance status 1.7 (1.2–2.4) <0.01 1.5 (1.0–2.2) 0.03
0 versus 1
Clinical trial 0.7 (0.5–1.0) 0.02 0.8 (0.6–1.1) 0.19
Yes versus no
Table 2B. Uni- and multivariate analyses for overall survival (OS) in renal Hospitalization 1.2 (0.8–1.7) 0.45 – –
cell carcinoma cohort Yes versus no

Univariate analysis Multivariate analysis ECOG, Eastern Cooperative Oncology Group.

Bold values reflect statistical significant; *P<0.05, **P<0.01.
Prognostic factor OS P-value OS P-value
HR (95% CI) HR (95% CI)

ATB 30-0/No ATB 2.4 (1.1–5.7) 0.04 2.1 (0.9–5.0) 0.11

Age 0.8 (0.4–1.5) 0.43 – – Table 3B. Uni- and multivariate analyses for OS in non-small-cell lung
 65 years/< 65 years cancer cohort
IMDC risk group 0.7 (0.3–1.7) 0.45 2.1 (0.8–5.3) 0.12
Univariate analysis Multivariate analysis
Intermediate/good 2.4 (1.0–6.0) 0.06
Poor/good Prognostic factor OS P-value OS P-value
Tumor burden 2.4 (1.2–4.6) <0.01 1.8 (0.9–3.6) 0.09 HR (95% CI) HR (95% CI)

Bold values reflect statistical significant; *P<0.05, **P<0.01. ATB 30-0/No ATB 2.9 (1.9–4.4) <0.01 2.5 (1.6–3.7) <0.01
Age 1.3 (0.9–1.9) 0.23 – –
65 years/<65 years
RCC amenable to anti-PD-1 mAb showed that Akkermansia Histology 1.4 (0.8–1.6) 0.57 – –
Squamous/nonsquamous
muciniphila (accompanied with Ruminococcacae) was signifi-
Smoking status 1.2 (0.7–1.9) 0.55 – –
cantly associated to more favorable outcomes. In this paper, the
Smoker/nonsmoker
changes in the microbiota composition due to ATB uptake were
Number of prior regimens 1.9 (1.3–2.9) <0.01 1.6 (1.1–2.4) 0.02
not addressed. Authors have also characterized other microbiota
3/<3
sites including saliva in cancer patients; however, in MM [23] ECOG performance status 3.6 (1.9–6.5) <0.01 2.6 (1.4–4.9) <0.01
and head and neck squamous-cell carcinoma [24], there was no 0 versus 1
correlation with ICI activity. Clinical trial 0.4 (0.2–0.7) <0.01 0.6 (0.3–1.0) 0.06
In addition, the microbiota composition could account for Yes versus no
uncoupling efficacy from toxicity in metastatic melanoma Hospitalization 1.1 (0.6–1.9) 0.76 – –
patients treated with anti-CTLA-4 [25, 26]. Baseline fecal samples Yes versus no
with high bacterial diversity enriched with Bacteroidetes phylum
and depleted with Firmicutes correlated with the absence of im- ECOG, Eastern Cooperative Oncology Group.
mune colitis. Of note, in the present study, the frequency of im- Bold values reflect statistical significant; *P<0.05, **P<0.01.
mune-related adverse effects was too small (data not shown) to
evaluate the association between ATB and adverse events.
Our study has several limitations. The first one is the retro- patient characteristics, including a higher rate of recent hospital-
spective component of the data entry from two institutions, al- ization in the NSCLC patients treated with ATB. Recent hospital-
though patients’ clinical responses were standardized and ization was not associated with worse PFS or OS as a univariate.
reviewed objectively using the RECIST 1.1 criteria in both cen- In RCC even in the current uncertainty of PD-1 efficacy in naive
ters, the timing of scan acquisition was not prespecified. versus previously treated patients [1, 27–29], we reported a differ-
Furthermore, we acknowledge some notable differences in ence in line of therapy in both groups. To acknowledge the

1442 | Derosa et al. Volume 29 | Issue 6 | 2018

Annals of Oncology
encountered differences in patient characteristics, we examined
the impact of ATB within individual subgroups and found overall
consistency of the effect of ATB. We also conducted multivariable
analysis to adjust for significant prognostic factors identified in
the univariate analysis, which still showed a negative impact of
ATB. Secondly, the analysis did not take into consideration add-
itional factors with a potential impact on the microbiota compos-
ition such as diet, country of origin, or other medications.
Thirdly, we did not characterize the mechanism by which ATB
exert a detrimental effect on clinical outcomes. We speculate that
ATB-related dysbiosis decreases the diversity, shifts the micro-
biota and probably eradicates the most immunogenic bacteria
required to engaged the immune system unleashed by PD-(L)1
blockade. Whether antibiotherapy reflects a general prognostic
association or is causalitively linked with resistance to ICI
remains a matter of debate. It is noteworthy that ATB-based con-
ditionning of tumor bearing mice blunts the efficacy of PD-1 or
PD-1 þ CTLA-4 blockade in otherwise ‘normal’ animals, sug-
gesting a causal link between ATB and primary resistance to ICI
and immunogenic chemotherapy [9, 10, 30, 31]. Finally, the lack
of PD-L1 status characterization in most of the patients (outside
of some NSCLC) leaves open the possibility that tumor-intrinsic
factors could influence these results.
Altogether, these results confirm that ATB-associated dysbiosis
might be deleterious in patients treated with ICI, suggesting that
an intact gut microbiota is needed to mobilize the immune sys-
tem regardless of the tumor site. Efforts to improve antibiotic
stewardship are already ongoing to prevent the emergence of
multidrug resistant organisms [32], which can be particularly
dangerous for cancer patients [33, 34]. The data in this report
may add additional incentive to avoid unnecessary ATB. More
studies are warranted to confirm the deleterious effect of ATB in
large prospective trials and to develop novel diagnostic tools
based on gut microbiota in cancer patients, to predict response/
resistance to ICI and identified the key microbiota signatures for
each tumor site. The discovery of bacteria capable of shifting an
unfavorable ATB-associated dysbiosis to a favorable microbiota
will help build a future therapeutic concept, whereby modulation
of gut microbiota by bacteria could increase ICI clinical activity.
Funding
Gustave Roussy Fondation Philanthropia (no grant number
applies) to LD and BR; RK Smiley Canadian Hematology Society
2017 (National Canadian grant) (no grant number applies).
Disclosure
The authors have declared no conflicts of interest.
References
1. Motzer RJ, Escudier B, McDermott DF et al. Nivolumab versus everoli-
mus in advanced renal-cell carcinoma. N Engl J Med 2015; 373:
1803–1813.
2. Escudier B, Sharma P, McDermott DF et al. CheckMate 025 randomized
phase 3 study: outcomes by key baseline factors and prior therapy for
Volume 29 | Issue 6 | 2018
Original article
nivolumab versus everolimus in advanced renal cell carcinoma. Eur Urol
2017 Mar 3 [Epub ahead of print], doi: 10.1016/j.eururo.2017.02.010.
3. Borghaei H, Paz-Ares L, Horn L et al. Nivolumab versus docetaxel in
advanced nonsquamous non-small-cell lung cancer. N Engl J Med 2015;
373(17): 1627–1639.
4. Reck M, Rodrı́guez-Abreu D, Robinson AG et al. Pembrolizumab versus
chemotherapy for PD-L1-positive non-small-cell lung cancer. N Engl J
Med 2016; 375(19): 1823–1833.
5. Carbone DP, Reck M, Paz-Ares L et al. First-line nivolumab in stage IV or re-
current non-small-cell lung cancer. N Engl J Med 2017; 376(25): 2415–2426.
6. Patel SP, Kurzrock R. PD-L1 Expression as a predictive biomarker in
cancer immunotherapy. Mol Cancer Ther 2015; 14(4): 847–856.
7. Gao J, Shi LZ, Zhao H et al. Loss of IFN-c pathway genes in tumor cells
as a mechanism of resistance to anti-CTLA-4 therapy. Cell 2016; 167(2):
397–404.e9.
8. Ayers M, Lunceford J, Nebozhyn M et al. IFN-c-related mRNA profile predicts
clinical response to PD-1 blockade. J Clin Invest 2017; 127(8): 2930–2940.
9. Vétizou M, Pitt JM, Daillère R et al. Anticancer immunotherapy by
CTLA-4 blockade relies on the gut microbiota. Science 2015; 350(6264):
1079–1084.
10. Routy B, Le Chatelier E, Derosa L et al. Gut microbiome influences efficacy
of PD-1-based immunotherapy against epithelial tumors. Science 2017 Nov
2 [Epub ahead of print], doi: 10.1126/science.aan3706.
11. Lozupone CA, Stombaugh JI, Gordon JI et al. Diversity, stability and re-
silience of the human gut microbiota. Nature 2012; 489(7415): 220–230.
12. Zitvogel L, Ayyoub M, Routy B, Kroemer G. Microbiome and anticancer
immunosurveillance. Cell 2016; 165(2): 276–287.
13. Ubeda C, Pamer EG. Antibiotics, microbiota, and immune defense.
Trends Immunol 2012; 33(9): 459–466.
14. Lange K, Buerger M, Stallmach A, Bruns T. Effects of antibiotics on gut
microbiota. Dig Dis 2016; 34(3): 260–268.
15. Korpela K, de Vos WM. Antibiotic use in childhood alters the gut micro-
biota and predisposes to overweight. Microb Cell 2016; 3(7): 296–298.
16. Leclercq S, Mian FM, Stanisz AM et al. Low-dose penicillin in early life
induces long-term changes in murine gut microbiota, brain cytokines
and behavior. Nat Commun 2017; 8: 15062.
17. Arrieta M-C, Stiemsma LT, Dimitriu PA et al. Early infancy microbial
and metabolic alterations affect risk of childhood asthma. Sci Transl Med
2015; 7: 307ra152.
18. Iacovelli R, Lanoy E, Albiges L, Escudier B. Tumour burden is an inde-
pendent prognostic factor in metastatic renal cell carcinoma. BJU Int
2012; 110(11): 1747–1753.
19. Eisenhauer EA, Therasse P, Bogaerts J et al. New response evaluation cri-
teria in solid tumours: revised RECIST guideline (version 1.1). Eur J
Cancer Oxf Engl 1990 2009; 45: 228–247.
20. Jakobsson HE, Jernberg C, Andersson AF et al. Short-term antibiotic
treatment has differing long-term impacts on the human throat and gut
microbiome. PLoS One 2010; 5(3): e9836.
21. Jernberg C, Löfmark S, Edlund C, Jansson JK. Long-term ecological
impacts of antibiotic administration on the human intestinal microbiota.
ISME J 2007; 1: 56–66.
22. Matson V, Fessler J, Bao R et al. The commensal microbiome is associ-
ated with anti–PD-1 efficacy in metastatic melanoma patients. Science
2018; 359(6371): 104–108.
23. Gopalakrishnan V, Spencer CN, Nezi L et al. Gut microbiome modulates
response to anti-PD-1 immunotherapy in melanoma patients. Science
2017 Nov 2 [Epub ahead of print], doi: 10.1126/science.aan4236.
24. Ferris RL, Blumenschein G, Harrington K et al. Abstract CT022: evalu-
ation of oral microbiome profiling as a response biomarker in squamous
cell carcinoma of the head and neck: analyses from CheckMate 141.
Cancer Res. 2017; 77(13 Suppl): CT022–CT022.
25. Dubin K, Callahan MK, Ren B et al. Intestinal microbiome analyses iden-
tify melanoma patients at risk for checkpoint-blockade-induced colitis.
Nat Commun 2016; 7: 10391.
26. Chaput N, Lepage P, Coutzac C et al. Baseline gut microbiota predicts clinic-
al response and colitis in metastatic melanoma patients treated with ipilimu-
mab. Ann Oncol Off J Eur Soc Med Oncol 2017; 28(suppl_5): 1368–1379.
doi:10.1093/annonc/mdy103 | 1443
Original article
27. Hammers HJ, Plimack ER, Infante JR et al. Safety and efficacy of
nivolumab in combination with ipilimumab in metastatic renal cell
carcinoma: the CheckMate 016 Study. J Clin Oncol 2017; 35(34):
3851–3858.
28. Motzer RJ, Rini BI, McDermott DF et al. Nivolumab for metastatic renal
cell carcinoma: results of a randomized phase II trial. J Clin Oncol 2015;
33(13): 1430–1437.
29. Meeting Library j Safety and efficacy of nivolumab in metastatic renal
cell carcinoma (mRCC): Results from the NIVOREN GETUG-AFU 26
study. https://meetinglibrary.asco.org/record/157690/abstract (3 March
2018, date last accessed).
30. Iida N, Dzutsev A, Stewart CA et al. Commensal bacteria control cancer
response to therapy by modulating the tumor microenvironment.
Science 2013; 342(6161): 967–970.
1444 | Derosa et al.
Annals of Oncology
31. Viaud S, Saccheri F, Mignot G et al. The intestinal microbiota modulates
the anticancer immune effects of cyclophosphamide. Science 2013;
342(6161): 971–976.
32. Rosa RG, Goldani LZ, dos Santos RP. Association between adherence to
an antimicrobial stewardship program and mortality among hospitalised
cancer patients with febrile neutropaenia: a prospective cohort study.
BMC Infect Dis 2014; 14(1): 286.
33. Cometta A, Calandra T, Bille J, Glauser MP. Escherichia coli resistant to
fluoroquinolones in patients with cancer and neutropenia. N Engl J Med
1994; 330(17): 1240–1241.
34. Montassier E, Batard E, Gastinne T et al. Recent changes in bacteremia
in patients with cancer: a systematic review of epidemiology and anti-
biotic resistance. Eur J Clin Microbiol Infect Dis. 2013; 32(7):
841–850.
Volume 29 | Issue 6 | 2018
 ORIGINAL ARTICLE
Do We Really Need to Thank the Beatles for the
Financing of the Development of the Computed
Tomography Scanner?
Zeev V. Maizlin, MD* and Patrick M. Vos, MDÞ
Abstract: It is commonly believed that the revenues from the selling of
the Beatles’ records by Electric and Musical Industries (EMI) allowed the
company to develop the computed tomography (CT) scanner. Some went
to define this as the Beatles’ gift to medicine. However, significant
controversies and discrepancies arise from analysis of this statement,
making its correctness doubtful. The details of financing required for the
CT development and the part of EMI in financial input have never been
publicly announced. This work analyzes the financial contributions to the
CT development and investigates if the revenues received from the sales
of the Beatles’ records were used for the creation of the CT scanner.
Timeline of the development of the EMI CT scanner and the financial
inputs of EMI and British Department of Health and Social Security
(DHSS) were assessed. Without salary expenses to Godfrey Hounsfield
and his team, the development of the CT scanner cost EMI approximately
U100,000. The British DHSS’s expenses were U606,000. Hence, the fi-
nancial contribution of DHSS into the development of the CT scanner
was significantly bigger than that of EMI. Accordingly, British tax payers
and officials of British DHSS are to be thanked for the CT scanner.
The Beatles’ input into the world’s culture is valuable and does not
require decoration by nonexistent connection to the development of CT.
A positive aspect to this misconception is that it keeps in public memory
the name of the company that developed the CT scanner.
Key Words: CT scanner, Hounsfield, history
(J Comput Assist Tomogr 2012;36: 161Y164)
I t is commonly believed that the revenues from the selling of
the Beatles records by Electric and Musical Industries (EMI)
allowed the company to finance its Central Research Laboratory
(CRL) to develop a new medical diagnostic imaging device for
the head scanVthe computed tomography (CT) scanner.1Y4
Some went to define this as the Beatles’ gift to medicine.5
However, there are significant controversies and dis-
crepancies that arise from analysis of this statement, making its
correctness doubtful.
First, the financial costs and sources involved in the CT
development were not clearly stated or publicly announced. To
some extent, it could be due to extensive litigation activity that
followed the introduction of CT scanners in the 1970s.
Second, the claim that flush with cash from selling the
Beatles’ records, EMI went on to finance the development of the
*From the Centre for Medical and Radiology History and Department of
Radiology, McMaster University Medical Centre, Hamilton, Ontario; and
†Department of Radiology, St Paul’s Hospital, University of British Colum-
bia, Vancouver, British Columbia, Canada.
Received for publication November 20, 2011; accepted December 6, 2011.
Reprints: Zeev V. Maizlin, MD, Centre for Medical and Radiology History
and Department of Radiology, McMaster University Medical Centre,
1200 Main St W, Hamilton, Ontario, Canada L8N 3Z5 (e-mail: zeev25@
yahoo.com).
The authors report no conflicts of interest.
Copyright * 2012 by Lippincott Williams & Wilkins
J Comput Assist Tomogr & Volume 36, Number 2, March/April 2012
Copyright © 2012 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
CT scanner, endangering the company’s financial assets (by
stepping into a new unknown foreign, mainly the United States,
market, that was controlled by other major well-established
players), would have been a very risky business strategy that is
hardly ever successful. Even EMI’s history demonstrated that its
control over the company’s own American subsidiary was lim-
ited (at the time when EMI’s Parlophone was making records of
the Beatles in Britain, EMI’s American subdivisionVCapitol
RecordsVdeclined to issue either of the first popular Beatles’
singles (‘‘Please Please Me,’’ ‘‘From Me to You’’), and they were
released through independent US labels.
Third, before launching the development of CT, EMI had
not consulted radiologists, neurologists, or neurosurgeons to get
their opinion and support of the new technology and took no
steps to avoid a substantial risk that the new product would not
be accepted by the doctors and hospital administration; EMI
had no previous significant experience in medical electronics
production and sales. It had also sold off its computer divi-
sion6 in 1962 (not a natural move for a company that would
initiate research and production involving a computer-based
technology).
These considerations question the accuracy of the claim
that EMI did not know what to do with the money it got from
selling the Beatles’ records and hence spent it developing the CT
scanner.
This work assesses the financial contributions to the CT
development and investigates if the revenues received from the
sales of the Beatles records were used for the development of the
CT scanner.
BACKGROUND
Electric and Musical Industries emerged from World War II
with an experience in electronics, which was then largely related
to defense-associated products. The transition to peacetime was
difficult. However, EMI did exciting pioneering work and tried
to become a leading computer company in Britain. With music
business booming, in 1955 EMI acquired Capitol Records in the
United States. This purchase and the subsequent success of the
recording groups, including the Beatles (Parlophone Records, an
EMI subsidiary, signed the recording contract with the Fab Four
in 1962, paying them a minimal royalty rate of 1 farthing per
double-sided disc7), put the company in a very strong financial
position. As a result, in 1970, the company earned U21 million.
In 1967, Godfrey Hounsfield, a research scientist in EMI’s
CRL, conceived the idea of the reversed radar.8 In an attempt to
find the practical application to this device, radiology came to
his attention. This implied that the company would branch out
into the highly competitive field of medical electronics, where it
essentially lacked experience offering only 2 small medical
products (a patient-monitoring device and an infrared ther-
mography device), which together represented less than 0.5% of
the company’s sales.9 It would be difficult to build a medical
imaging operation from scratch without knowledge of the North
www.jcat.org 161
Maizlin and Vos J Comput Assist Tomogr & Volume 36, Number 2, March/April 2012

TABLE 1. Summary of the Funding Allocated to Development of the CT Scanner

Year Research and Engineering Period Financing Financing Source

1967 Hounsfield conceived the idea of the scanner Regular salary from CRL EMI
The salary paid to G. Hounsfield
for his work in CRL
(exact total sum unknown)
1968 ‘‘Lathe bed model’’ created from an old Team worked on a very low EMI
lathe that Hounsfield had been using in a budget of U25,000
previous project working on computer stores.
A lathe bed provided the means for moving
and rotating the gamma-ray source,
Americium 95, around the bottles or Perspex
jars with a photoncounter as the detector
placed on the other side. It took 9 days to
collect sufficient information of 28,000
measurements (and 2.5 h to reconstruct
the image on an ICL 1905 mainframe
computer). This work was done by a very
small team composed of G. Hounsfield,
Stephen Bates (programming), Peter Langstone
(electronics), and Mel King (mechanics)
1968 An x-ray tube was purchased and used, allowing U6000  2 = U12,000 EMI and the DHSS:
more controlled source, as well as having a for the acquisition of an U6000 each
much higher output, thus reducing data x-ray tube and a generator
acquisition time to 9 h
1968Y1969 Completion of the research and construction (1) Funding of the remaining DHSS: U600,000
of the prototype research costs (U69,000)
(2) Order for a prototype scanner EMI: the remaining
and 3 clinical machines research costs:
U150,000 for each: approximately U69,000
U150,000  4 = U600,000
1972 Manufacturing of the first scanners by EMI The DHHS purchased the first 3 DHSSVsee above
EMI scanners and placed them
in Manchester, Glasgow,
and London

American market, where most of the demand for scanners was to
be expected. Moreover, EMI’s experience in electronics work
had been mainly in producing small numbers of highly spe-
cialized defense products on government contracts. If a scanner
was going to be produced, most of the components would be
purchased from subcontractors and had to be integrated into a
functioning system. A diversification move with unclear effect
on the company’s prosperity was going to be dangerous. There
was considerable disagreement among top management at EMI
regarding the CT scanner. Even Dr Len Broadway, head of the
CRL and one of the CT scanner’s earliest supporters, was among
the strongest opponents of EMI’s self-development of the new
business, being convinced that EMI’s potential competitors in
the field had considerably greater technical capabilities and
resources.9
Considerations of the CT scanner supporters were that in
the 1960s the radiology departments in many hospitals were
recognized as important money-making operators. Radiologists
were able to make manufacturers build specially designed and
expensive.
X-ray Systems and Applications
The size of the x-ray market was growing by almost 10%
annually. Initially, EMI estimated that they would be able to sell
5 scanners in the first 12 months. This opinion then changed to
12 and finally to 50 scanners. The company was encouraged by
the opinion that the time might come when neurologists would
feel ethically compelled to order a CT scan before making a
diagnosis.9 Electric and Musical Industries expected that the
scanner cost would be around $400,000, and it was clear that
only the largest and financially strongest hospitals would be able
to afford the scanner.
To save on the development costs and to secure the initial
sale contract, EMI management decided to apply for the assis-
tance of the British government.
Therefore, in 1968, EMI and G. Hounsfield approached the
British Department of Health and Social Security (DHSS). They
asked the Medical Research Council for development funds
in exchange for a share of the profits. The government thought
that it would be commendable foresight and agreed to support
EMI’s development of a head scanner. This development got a
significant and vital support from the DHSS officials: E. Lennon
and G. Higson.
Dr Evan Lennon was a principal medical officer in radi-
ology at the DHSS. He connected between G. Hounsfield and
neuroradiologist Dr James Ambrose, after a few failed attempts
when some reputable radiologists had refused to work with EMI
and Hounsfield on the project (one can only imagine what they
felt later when the CT scanner produced its clinical images!).
With admirable support of Mr Gordon Higson, who worked
at the scientific and technical branch of the DHSS, the backing
of EMI’s project was ensured. The DHSS agreed to fund half the
remaining research costs in exchange for a small royalty on
sales. At the time, it was calculated that it would cost U69,000
to build a complete working system, and the DHSS committed to
pay U150,000 for each of the 4 systems, placing the order for a

162 www.jcat.org * 2012 Lippincott Williams & Wilkins

Copyright © 2012 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
J Comput Assist Tomogr & Volume 36, Number 2, March/April 2012
TABLE 2. Financial Summary for EMI (in U’000)*
1969 1970
EMI sales in medical electronics V V
EMI profit (loss) from sales in medical electronics V V
Profit attributable to ordinary stockholders 7259 8736
*Based on data from Bartlett.9
prototype and 3 clinical machines. It was intended to generate
sufficient income to fund the fifth machine for Hounsfield and
his team to keep and work on.10,11
Dr Ambrose and Dr Louis Kreel supplied the human ca-
daver samples and provided the radiologic consult for the work.
Hounsfield and his small team were stationed in the Radiology
Department at the Atkinson Morley’s Hospital in Wimbledon.
The location was chosen to avoid widespread publicity in the
development phase.1 Thereafter, the development took place
largely in secret.
The first clinical trial was very successful. A prototype EMI
head scanner (Mark I) in October 1971 provided the first clinical
image of a female patient of Atkinson Morley’s Hospital, re-
vealing the presence and location of a brain lesion.
THE TIMELINE AND FINANCING OF THE
DEVELOPMENT OF THE CT SCANNER
Chronological stages and financial contribution of the CT
development are summarized in Table 1.
Based on the available sources, it is difficult to establish if
U69,000, necessary for the funding of the remaining research
costs in 1968Y1969, were shared between EMI and the DHSS or
paid entirely by EMI. To avoid the underestimation of EMI’s
expenses, for calculation purposes it was presumed that U69,000
was paid entirely by EMI. Without salary expenses to G.
Hounsfield and his team, the development of the CT scanner
cost EMI approximately U100,000 (U25,000 + U6000 +
U69,000).
The salary component of EMI expenses cannot be precisely
calculated because of the private character of this information.
However, hiring advertisements in that period12,13 mentioned the
annual salary of engineers, physicists, and technicians to be in
the range of U1500 to U2500. Assuming that G. Hounsfield
could be paid 4 times more and 3 members of his team could be
paid 2 times more than the upper range of salaries at that time, an
approximate estimation of EMI salary expenses for the team is
U25,000 per annum or U100,000 for the years of 1968 to 1971.
Accordingly, the development of the CT scanner cost EMI ap-
proximately U100,000 plus U100,000 of salaries to G. Houns-
field and his team over the years 1967 to 1971. It constituted
total costs for EMI of U200,000Vless than 1% of the company’s
profits (U20,557,000) in this period (Table 2) and could not be
attributed to the revenues from the record sales of one of the
groups. The development of the CT scanner in 1968 to 1970 did
not affect the funds available for distribution to shareholders in
1969 to 1976.9 The drop in 1971 was attributed to extraordinary
losses unrelated to CRL financing and was recovered already in
1972. Later, EMI sales in the United States from the first scanner
in midY1973 until 1975, when 160 machines were installed at an
average cost of U160,000 ($400,000), provided EMI with gain of
U25,600,000 ($64,000,000).
Two important factors negatively affected the financial
input from the Beatles to EMI in the period of the development
* 2012 Lippincott Williams & Wilkins
Copyright © 2012 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
The Beatles and the Development of the CT Scanner
1971 1972 1973 1974 1975 1976
V V 321 5076 20,406 42,104
V V (67) 1242 9230 12,502
4562 7297 10,864 13,327 13,124 24,399
of the CT scanner. Since 1968 (starting from the White Album),
the group’s records were also released by a new Apple Corpo-
ration founded by the Beatles. Moreover, in 1969, the band got
from EMI (temporarily concealing the fact that John Lennon had
already effectively quit the group) the highest royalty rate ever
paid to a recording group up to that time.14
On the other hand, the development process and 4 CT
scanners (including the first) cost the DHSS at least U606,000
(U6000 + U600,000). It is hard to overestimate the value of this
government contract for EMI, which effectively allowed the
company to complete the development and commence the pro-
duction of the CT scanner. What is more, the whole development
was greatly and vitally supported by the British DHSS officials.
CONCLUSIONS
Many factors, significantly more complex and diverse than
the cash inflow from one of the groups, even as successful as the
Beatles, ensured the development of the CT scanner by EMI. It
would be inaccurate to claim that EMI did not have what to do
with the money it got from selling the Beatles’ records. As usual,
these profits were distributed to shareholders and used for the
company’s expansion in the familiar market of the music busi-
ness. Moreover, there is no evidence to presume that the de-
velopment of the CT scanner was the ‘‘gift’’ from the Beatles
because there is no confirmation that the group was ever in-
volved in any way in the decision-making process of the com-
pany’s management. The linkage between 2 successful business
ventures of one of the eminent British corporations was very
beneficial for EMI, helping it to be perceived as a justified (and
there is no doubting this) developer of the first practical CT
scanner.
Comparing the financial contributions into the develop-
ment of the CT scanner from EMI and DHSS, the latter’s part
was significantly bigger. Accordingly, British tax payers and
officials of British DHSS are to be thanked for the CT scanner.
It is clear for all Beatles’ fans (including the authors) that
the group’s input into the world’s cultural heritage is valuable
enough and does not require to be decorated by nonexistent
connection to the development of the new revolutionary imaging
modality. Ironically, there was, however, a positive aspect to this
misconception. It helped to keep in the public’s (and young
radiology residents’) memory the name of the company that
developed the CT scanner.
ACKNOWLEDGMENT
The authors thank Ms Monika Ferrier, BA, whose efforts
and professional expertise helped to enrich this article with
important details. More details about development of the CT
scanner are available in the book ‘‘Wonders of Radiology.’’
REFERENCES
1. Isherwood I. Sir Godfrey Hounsfield. Radiology. 2005;234:975Y976.
2. Rogers LF. ‘‘My word, what is that?’’: Hounsfield and the triumph of
clinical research. Am J Roentgenol. 2003;180:1501.
www.jcat.org 163
Maizlin and Vos
3. Goodman LR. The Beatles, the Nobel Prize, and CT scanning of
the chest. Radiol Clin N Am. 2010;48:1Y7.
4. Berland LL. Commentary on ‘‘Computed tomography of the body:
initial clinical trial with the EMI prototype’’ and ‘‘Computed
tomography of the body: early trends in application
and accuracy of the method’’. Am J Roentgenol. 2008;
191:16Y18.
5. The Beatles gift to medicine. Available at: http://
doctorrw.blogspot.com/2007/03/beatles-gift-to-medicine.html.
Accessed December 10, 2010.
6. The Beatles. 1960Y1963 Business. Available at: http://
www.independent.co.uk/news/obituaries/sir-godfrey-hounsfield-
550312.html. Accessed December 10, 2010.
7. Available at: http://www.beatlemoney.com/beatles6063business.htm.
Accessed September 23, 2011
164 www.jcat.org
Copyright © 2012 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
J Comput Assist Tomogr & Volume 36, Number 2, March/April 2012
8. Strong AB, Hurst RA. EMI patents on computed tomography: history
of legal actions. Br J Radiol. 1994;67:315Y316.
9. Bartlett CA. EMI and the CT Scanner. Boston, MA, Harvard Business
School Press. 1983. Available at: http://www.blackwellpublishing.com/
grant/docs/10EMI.pdf. Accessed December 10, 2010.
10. Beckmann EC. CT scanning the early days. Br J Radiol. 2006;79:5Y8.
11. Süsskind C. The invention of computed tomography. Hist Technol.
1981;6:39Y80.
12. Positions available. New Scientist. April, 30, 1964:323.
13. Positions available. New Scientist. December, 31, 1970:620.
14. Biography for The Beatles. Available at: http://webcache.
googleusercontent.com/search?hl=en&start=40&q=cache:
Nqf FU9cI8MAJ; http://www.imdb.com/name/nm1397313/bio+
beatles+sales+peak+records+emi&ct=clnk. Accessed December 4, 2011.
* 2012 Lippincott Williams & Wilkins
Cell, Vol. 100, 57–70, January 7, 2000, Copyright 2000 by Cell Press
The Hallmarks of Cancer
Douglas Hanahan* and Robert A. Weinberg†
* Department of Biochemistry and Biophysics and
Hormone Research Institute
University of California at San Francisco
San Francisco, California 94143
† Whitehead Institute for Biomedical Research and
Department of Biology
Massachusetts Institute of Technology
Cambridge, Massachusetts 02142
After a quarter century of rapid advances, cancer re-
search has generated a rich and complex body of knowl-
edge, revealing cancer to be a disease involving dy-
namic changes in the genome. The foundation has been
set in the discovery of mutations that produce onco-
genes with dominant gain of function and tumor sup-
pressor genes with recessive loss of function; both
classes of cancer genes have been identified through
their alteration in human and animal cancer cells and
by their elicitation of cancer phenotypes in experimental
models (Bishop and Weinberg, 1996).
Some would argue that the search for the origin and
treatment of this disease will continue over the next
quarter century in much the same manner as it has in
the recent past, by adding further layers of complexity
to a scientific literature that is already complex almost
beyond measure. But we anticipate otherwise: those
researching the cancer problem will be practicing a dra-
matically different type of science than we have experi-
enced over the past 25 years. Surely much of this change
will be apparent at the technical level. But ultimately,
the more fundamental change will be conceptual.
We foresee cancer research developing into a logical
science, where the complexities of the disease, de-
scribed in the laboratory and clinic, will become under-
standable in terms of a small number of underlying prin-
ciples. Some of these principles are even now in the
midst of being codified. We discuss one set of them in
the present essay: rules that govern the transformation
of normal human cells into malignant cancers. We sug-
gest that research over the past decades has revealed
a small number of molecular, biochemical, and cellular
traits—acquired capabilities—shared by most and per-
haps all types of human cancer. Our faith in such simplifi-
cation derives directly from the teachings of cell biology
that virtually all mammalian cells carry a similar molecu-
lar machinery regulating their proliferation, differentia-
tion, and death.
Several lines of evidence indicate that tumorigenesis
in humans is a multistep process and that these steps
reflect genetic alterations that drive the progressive
transformation of normal human cells into highly malig-
nant derivatives. Many types of cancers are diagnosed
in the human population with an age-dependent inci-
dence implicating four to seven rate-limiting, stochastic
events (Renan, 1993). Pathological analyses of a number
of organ sites reveal lesions that appear to represent
the intermediate steps in a process through which cells
Review
evolve progressively from normalcy via a series of pre-
malignant states into invasive cancers (Foulds, 1954).
These observations have been rendered more con-
crete by a large body of work indicating that the ge-
nomes of tumor cells are invariably altered at multiple
sites, having suffered disruption through lesions as sub-
tle as point mutations and as obvious as changes in
chromosome complement (e.g., Kinzler and Vogelstein,
1996). Transformation of cultured cells is itself a
multistep process: rodent cells require at least two intro-
duced genetic changes before they acquire tumorigenic
competence, while their human counterparts are more
difficult to transform (Hahn et al., 1999). Transgenic
models of tumorigenesis have repeatedly supported the
conclusion that tumorigenesis in mice involves multiple
rate-limiting steps (Bergers et al., 1998; see Oncogene,
1999, R. DePinho and T. E. Jacks, volume 18[38], pp.
5248–5362). Taken together, observations of human
cancers and animal models argue that tumor develop-
ment proceeds via a process formally analogous to Dar-
winian evolution, in which a succession of genetic
changes, each conferring one or another type of growth
advantage, leads to the progressive conversion of nor-
mal human cells into cancer cells (Foulds, 1954; Nowell,
1976).
An Enumeration of the Traits
The barriers to development of cancer are embodied
in a teleology: cancer cells have defects in regulatory
circuits that govern normal cell proliferation and homeo-
stasis. There are more than 100 distinct types of cancer,
and subtypes of tumors can be found within specific
organs. This complexity provokes a number of ques-
tions. How many distinct regulatory circuits within each
type of target cell must be disrupted in order for such
a cell to become cancerous? Does the same set of
cellular regulatory circuits suffer disruption in the cells
of the disparate neoplasms arising in the human body?
Which of these circuits operate on a cell-autonomous
basis, and which are coupled to the signals that cells
receive from their surrounding microenvironment within
a tissue? Can the large and diverse collection of cancer-
associated genes be tied to the operations of a small
group of regulatory circuits?
We suggest that the vast catalog of cancer cell geno-
types is a manifestation of six essential alterations in cell
physiology that collectively dictate malignant growth
(Figure 1): self-sufficiency in growth signals, insensitivity
to growth-inhibitory (antigrowth) signals, evasion of pro-
grammed cell death (apoptosis), limitless replicative
potential, sustained angiogenesis, and tissue invasion
and metastasis. Each of these physiologic changes—
novel capabilities acquired during tumor development—
represents the successful breaching of an anticancer
defense mechanism hardwired into cells and tissues.
We propose that these six capabilities are shared in
common by most and perhaps all types of human tu-
mors. This multiplicity of defenses may explain why can-
cer is relatively rare during an average human lifetime.
Cell
58
Figure 1. Acquired Capabilities of Cancer
We suggest that most if not all cancers have acquired the same set
of functional capabilities during their development, albeit through
various mechanistic strategies.
We describe each capability in turn below, illustrate with
a few examples its functional importance, and indicate
strategies by which it is acquired in human cancers.
Acquired Capability: Self-Sufficiency
in Growth Signals
Normal cells require mitogenic growth signals (GS) be-
fore they can move from a quiescent state into an active
proliferative state. These signals are transmitted into the
cell by transmembrane receptors that bind distinctive
classes of signaling molecules: diffusible growth fac-
tors, extracellular matrix components, and cell-to-cell
adhesion/interaction molecules. To our knowledge, no
type of normal cell can proliferate in the absence of
such stimulatory signals. Many of the oncogenes in the
cancer catalog act by mimicking normal growth signal-
ing in one way or another.
Dependence on growth signaling is apparent when
propagating normal cells in culture, which typically pro-
liferate only when supplied with appropriate diffusible
mitogenic factors and a proper substratum for their inte-
grins. Such behavior contrasts strongly with that of tu-
mor cells, which invariably show a greatly reduced
dependence on exogenous growth stimulation. The con-
clusion is that tumor cells generate many of their own
growth signals, thereby reducing their dependence on
stimulation from their normal tissue microenvironment.
This liberation from dependence on exogenously de-
rived signals disrupts a critically important homeostatic
mechanism that normally operates to ensure a proper
behavior of the various cell types within a tissue.
Acquired GS autonomy was the first of the six capabili-
ties to be clearly defined by cancer researchers, in large
part because of the prevalence of dominant oncogenes
that have been found to modulate it. Three common
molecular strategies for achieving autonomy are evi-
dent, involving alteration of extracellular growth signals,
of transcellular transducers of those signals, or of intra-
cellular circuits that translate those signals into action.
While most soluble mitogenic growth factors (GFs) are
made by one cell type in order to stimulate proliferation
of another—the process of heterotypic signaling—many
cancer cells acquire the ability to synthesize GFs to
which they are responsive, creating a positive feedback
signaling loop often termed autocrine stimulation (Fedi
et al., 1997). Clearly, the manufacture of a GF by a cancer
cell obviates dependence on GFs from other cells within
the tissue. The production of PDGF (platelet-derived
growth factor) and TGF␣ (tumor growth factor ␣) by
glioblastomas and sarcomas, respectively, are two illus-
trative examples (Fedi et al., 1997).
The cell surface receptors that transduce growth-
stimulatory signals into the cell interior are themselves
targets of deregulation during tumor pathogenesis. GF
receptors, often carrying tyrosine kinase activities in
their cytoplasmic domains, are overexpressed in many
cancers. Receptor overexpression may enable the can-
cer cell to become hyperresponsive to ambient levels
of GF that normally would not trigger proliferation (Fedi
et al., 1997). For example, the epidermal GF receptor
(EGF-R/erbB) is upregulated in stomach, brain, and
breast tumors, while the HER2/neu receptor is overex-
pressed in stomach and mammary carcinomas (Slamon
et al., 1987; Yarden and Ullrich, 1988). Additionally, gross
overexpression of GF receptors can elicit ligand-inde-
pendent signaling (DiFiore et al., 1987). Ligand-indepen-
dent signaling can also be achieved through structural
alteration of receptors; for example, truncated versions
of the EGF receptor lacking much of its cytoplasmic
domain fire constitutively (Fedi et al., 1997).
Cancer cells can also switch the types of extracellular
matrix receptors (integrins) they express, favoring ones
that transmit progrowth signals (Lukashev and Werb,
1998; Giancotti and Ruoslahti, 1999). These bifunctional,
heterodimeric cell surface receptors physically link cells
to extracellular superstructures known as the extracellu-
lar matrix (ECM). Successful binding to specific moieties
of the ECM enables the integrin receptors to transduce
signals into the cytoplasm that influence cell behavior,
ranging from quiescence in normal tissue to motility,
resistance to apoptosis, and entrance into the active
cell cycle. Conversely, the failure of integrins to forge
these extracellular links can impair cell motility, induce
apoptosis, or cause cell cycle arrest (Giancotti and Ru-
oslahti, 1999). Both ligand-activated GF receptors and
progrowth integrins engaged to extracellular matrix
components can activate the SOS-Ras-Raf-MAP kinase
pathway (Aplin et al., 1998; Giancotti and Ruoslahti,
1999).
The most complex mechanisms of acquired GS auton-
omy derive from alterations in components of the down-
stream cytoplasmic circuitry that receives and pro-
cesses the signals emitted by ligand-activated GF
receptors and integrins. The SOS-Ras-Raf-MAPK cas-
cade plays a central role here. In about 25% of human
Review
59

Figure 2. The Emergent Integrated Circuit of the Cell

Progress in dissecting signaling pathways has begun to lay out a circuitry that will likely mimic electronic integrated circuits in complexity
and finesse, where transistors are replaced by proteins (e.g., kinases and phosphatases) and the electrons by phosphates and lipids, among
others. In addition to the prototypical growth signaling circuit centered around Ras and coupled to a spectrum of extracellular cues, other
component circuits transmit antigrowth and differentiation signals or mediate commands to live or die by apoptosis. As for the genetic
reprogramming of this integrated circuit in cancer cells, some of the genes known to be functionally altered are highlighted in red.

tumors, Ras proteins are present in structurally altered
forms that enable them to release a flux of mitogenic
signals into cells, without ongoing stimulation by their
normal upstream regulators (Medema and Bos, 1993).
We suspect that growth signaling pathways suffer
deregulation in all human tumors. Although this point
is hard to prove rigorously at present, the clues are
abundant (Hunter, 1997). For example, in the best stud-
ied of tumors—human colon carcinomas—about half
of the tumors bear mutant ras oncogenes (Kinzler and
Vogelstein, 1996). We suggest that the remaining colonic
tumors carry defects in other components of the growth
signaling pathways that phenocopy ras oncogene acti-
vation. The nature of these alternative, growth-stimulat-
ing mechanisms remains elusive.
Under intensive study for two decades, the wiring
diagram of the growth signaling circuitry of the mamma-
lian cell is coming into focus (Figure 2). New downstream
effector pathways that radiate from the central SOS-
Ras-Raf-MAP kinase mitogenic cascade are being dis-
covered with some regularity (Hunter, 1997; Rommel
and Hafen, 1998). This cascade is also linked via a variety
of cross-talking connections with other pathways; these
cross connections enable extracellular signals to elicit
multiple cell biological effects. For example, the direct
interaction of the Ras protein with the survival-promot-
ing PI3 kinase enables growth signals to concurrently
evoke survival signals within the cell (Downward, 1998).
While acquisition of growth signaling autonomy by
cancer cells is conceptually satisfying, it is also too
simplistic. We have traditionally explored tumor growth
by focusing our experimental attentions on the geneti-
cally deranged cancer cells (Figure 3, left panel). It is,
however, increasingly apparent that the growth deregu-
lation within a tumor can only be explained once we
understand the contributions of the ancillary cells pres-
ent in a tumor—the apparently normal bystanders such
as fibroblasts and endothelial cells—which must play
key roles in driving tumor cell proliferation (Figure 3,
right panel). Within normal tissue, cells are largely in-
structed to grow by their neighbors (paracrine signals)
or via systemic (endocrine) signals. Cell-to-cell growth
signaling is likely to operate in the vast majority of human
tumors as well; virtually all are composed of several
distinct cell types that appear to communicate via het-
erotypic signaling.
Heterotypic signaling between the diverse cell types
within a tumor may ultimately prove to be as important
Cell
60
in explaining tumor cell proliferation as the cancer cell-
autonomous mechanisms enumerated above. For ex-
ample, we suspect that many of the growth signals driv-
ing the proliferation of carcinoma cells originate from
the stromal cell components of the tumor mass. While
difficult to validate at present, such thinking recasts the
logic of acquired GS autonomy: successful tumor cells
are those that have acquired the ability to co-opt their
normal neighbors by inducing them to release abundant
fluxes of growth-stimulating signals (Skobe and Fu-
senig, 1998). Indeed, in some tumors, these cooperating
cells may eventually depart from normalcy, coevolving
with their malignant neighbors in order to sustain the
growth of the latter (Kinzler and Vogelstein, 1998; Olumi
et al., 1999). Further, inflammatory cells attracted to sites
of neoplasia may promote (rather than eliminate) cancer
cells (Cordon-Cardo and Prives, 1999; Coussens et al.,
1999; Hudson et al., 1999), another example of normal
cells conscripted to enhance tumor growth potential,
another means to acquire necessary capabilities.
Acquired Capability: Insensitivity
to Antigrowth Signals
Within a normal tissue, multiple antiproliferative signals
operate to maintain cellular quiescence and tissue ho-
meostasis; these signals include both soluble growth
inhibitors and immobilized inhibitors embedded in the
extracellular matrix and on the surfaces of nearby cells.
These growth-inhibitory signals, like their positively act-
ing counterparts, are received by transmembrane cell
surface receptors coupled to intracellular signaling cir-
cuits.
Antigrowth signals can block proliferation by two dis-
tinct mechanisms. Cells may be forced out of the active
proliferative cycle into the quiescent (G0) state from
which they may reemerge on some future occasion
when extracellular signals permit. Alternatively, cells
may be induced to permanently relinquish their prolifera-
tive potential by being induced to enter into postmitotic
states, usually associated with acquisition of specific
differentiation-associated traits.
Incipient cancer cells must evade these antiprolifera-
tive signals if they are to prosper. Much of the circuitry
that enables normal cells to respond to antigrowth sig-
nals is associated with the cell cycle clock, specifically
Figure 3. Tumors as Complex Tissues
The field of cancer research has largely been
guided by a reductionist focus on cancer cells
and the genes within them (left panel)—a fo-
cus that has produced an extraordinary body
of knowledge. Looking forward in time, we
believe that important new inroads will come
from regarding tumors as complex tissues in
which mutant cancer cells have conscripted
and subverted normal cell types to serve as
active collaborators in their neoplastic agenda
(right panel). The interactions between the
genetically altered malignant cells and these
supporting coconspirators will prove critical
to understanding cancer pathogenesis and to
the development of novel, effective therapies.
the components governing the transit of the cell through
the G1 phase of its growth cycle. Cells monitor their
external environment during this period and, on the ba-
sis of sensed signals, decide whether to proliferate, to
be quiescent, or to enter into a postmitotic state. At the
molecular level, many and perhaps all antiproliferative
signals are funneled through the retinoblastoma protein
(pRb) and its two relatives, p107 and p130. When in a
hypophosphorylated state, pRb blocks proliferation by
sequestering and altering the function of E2F transcrip-
tion factors that control the expression of banks of genes
essential for progression from G1 into S phase (Wein-
berg, 1995).
Disruption of the pRb pathway liberates E2Fs and
thus allows cell proliferation, rendering cells insensitive
to antigrowth factors that normally operate along this
pathway to block advance through the G1 phase of the
cell cycle. The effects of the soluble signaling molecule
TGF␤ are the best documented, but we envision other
antigrowth factors will be found to signal through this
pathway as well. TGF␤ acts in a number of ways, most
still elusive, to prevent the phosphorylation that inacti-
vates pRb; in this fashion, TGF␤ blocks advance through
G1. In some cell types, TGF␤ suppresses expression
of the c-myc gene, which regulates the G1 cell cycle
machinery in still unknown ways (Moses et al., 1990).
More directly, TGF␤ causes synthesis of the p15INK4B and
p21 proteins, which block the cyclin:CDK complexes
responsible for pRb phosphorylation (Hannon and
Beach, 1994; Datto et al., 1997).
The pRb signaling circuit, as governed by TGF␤ and
other extrinsic factors, can be disrupted in a variety of
ways in different types of human tumors (Fynan and
Reiss, 1993). Some lose TGF␤ responsiveness through
downregulation of their TGF␤ receptors, while others
display mutant, dysfunctional receptors (Fynan and
Reiss, 1993; Markowitz et al., 1995). The cytoplasmic
Smad4 protein, which transduces signals from ligand-
activated TGF␤ receptors to downstream targets, may
be eliminated through mutation of its encoding gene
(Schutte et al., 1996). The locus encoding p15INK4B may be
deleted (Chin et al., 1998). Alternatively, the immediate
downstream target of its actions, CDK4, may become
unresponsive to the inhibitory actions of p15INK4B be-
cause of mutations that create amino acid substitutions
Review
61
in its INK4A/B-interacting domain; the resulting cyclin
D:CDK4 complexes are then given a free hand to inacti-
vate pRb by hyperphosphorylation (Zuo et al., 1996).
Finally, functional pRb, the end target of this pathway,
may be lost through mutation of its gene. Alternatively,
in certain DNA virus-induced tumors, notably cervical
carcinomas, pRb function is eliminated through seques-
tration by viral oncoproteins, such as the E7 oncoprotein
of human papillomavirus (Dyson et al., 1989). In addition,
cancer cells can also turn off expression of integrins and
other cell adhesion molecules that send antigrowth sig-
nals, favoring instead those that convey progrowth sig-
nals; these adherence-based antigrowth signals likely
impinge on the pRb circuit as well. The bottom line is
that the antigrowth circuit converging onto Rb and the
cell division cycle is, one way or another, disrupted in
a majority of human cancers, defining the concept and
a purpose of tumor suppressor loss in cancer.
Cell proliferation depends on more than an avoidance
of cytostatic antigrowth signals. Our tissues also con-
strain cell multiplication by instructing cells to enter irre-
versibly into postmitotic, differentiated states, using di-
verse mechanisms that are incompletely understood; it
is apparent that tumor cells use various strategies to
avoid this terminal differentiation. One strategy for
avoiding differentiation directly involves the c-myc on-
cogene, which encodes a transcription factor. During
normal development, the growth-stimulating action of
Myc, in association with another factor, Max, can be
supplanted by alternative complexes of Max with a
group of Mad transcription factors; the Mad–Max com-
plexes elicit differentiation-inducing signals (Foley and
Eisenman, 1999). However, overexpression of the c-Myc
oncoprotein, as is seen in many tumors, can reverse this
process, shifting the balance back to favor Myc–Max
complexes, thereby impairing differentiation and pro-
moting growth. During human colon carcinogenesis, in-
activation of the APC/␤-catenin pathway serves to block
the egress of enterocytes in the colonic crypts into a
differentiated, postmitotic state (Kinzler and Vogelstein,
1996). Analogously, during the generation of avian eryth-
roblastosis, the erbA oncogene acts to prevent irrevers-
ible erythrocyte differentiation (Kahn et al., 1986).
While the components and interconnections between
the various antigrowth and differentiation-inducing sig-
nals and the core cell cycle machinery are still being
delineated, the existence of an antigrowth signaling cir-
cuitry is clear (Figure 2), as is the necessity for its circum-
vention by developing cancers.
Acquired Capability: Evading Apoptosis
The ability of tumor cell populations to expand in number
is determined not only by the rate of cell proliferation
but also by the rate of cell attrition. Programmed cell
death—apoptosis—represents a major source of this
attrition. The evidence is mounting, principally from
studies in mouse models and cultured cells, as well as
from descriptive analyses of biopsied stages in human
carcinogenesis, that acquired resistance toward apo-
ptosis is a hallmark of most and perhaps all types of
cancer.
Observations accumulated over the past decade indi-
cate that the apoptotic program is present in latent form
in virtually all cell types throughout the body. Once trig-
gered by a variety of physiologic signals, this program
unfolds in a precisely choreographed series of steps.
Cellular membranes are disrupted, the cytoplasmic and
nuclear skeletons are broken down, the cytosol is ex-
truded, the chromosomes are degraded, and the nu-
cleus is fragmented, all in a span of 30–120 min. In the
end, the shriveled cell corpse is engulfed by nearby cells
in a tissue and disappears, typically within 24 hr (Wyllie
et al., 1980).
The apoptotic machinery can be broadly divided into
two classes of components—sensors and effectors. The
sensors are responsible for monitoring the extracellular
and intracellular environment for conditions of normality
or abnormality that influence whether a cell should live
or die. These signals regulate the second class of com-
ponents, which function as effectors of apoptotic death.
The sentinels include cell surface receptors that bind
survival or death factors. Examples of these ligand/
receptor pairs include survival signals conveyed by IGF-
1/IGF-2 through their receptor, IGF-1R, and by IL-3 and
its cognate receptor, IL-3R (Lotem and Sachs, 1996;
Butt et al., 1999). Death signals are conveyed by the
FAS ligand binding the FAS receptor and by TNF␣ bind-
ing TNF-R1 (Ashkenazi and Dixit, 1999). Intracellular
sensors monitor the cell’s well-being and activate the
death pathway in response to detecting abnormalities,
including DNA damage, signaling imbalance provoked
by oncogene action, survival factor insufficiency, or hyp-
oxia (Evan and Littlewood, 1998). Further, the life of most
cells is in part maintained by cell–matrix and cell–cell
adherence-based survival signals whose abrogation
elicits apoptosis (Ishizaki et al., 1995; Giancotti and Ru-
oslahti, 1999). Both soluble and immobilized apoptotic
regulatory signals likely reflect the needs of tissues to
maintain their constituent cells in appropriate architec-
tural configurations.
Many of the signals that elicit apoptosis converge
on the mitochondria, which respond to proapoptotic
signals by releasing cytochrome C, a potent catalyst of
apoptosis (Green and Reed, 1998). Members of the Bcl-2
family of proteins, whose members have either pro-
apoptotic (Bax, Bak, Bid, Bim) or antiapoptotic (Bcl-2,
Bcl-XL, Bcl-W) function, act in part by governing mito-
chondrial death signaling through cytochrome C re-
lease. The p53 tumor suppressor protein can elicit apo-
ptosis by upregulating expression of proapoptotic Bax
in response to sensing DNA damage; Bax in turn stimu-
lates mitochondria to release cytochrome C.
The ultimate effectors of apoptosis include an array
of intracellular proteases termed caspases (Thornberry
and Lazebnik, 1998). Two “gatekeeper” caspases, ⫺8
and ⫺9, are activated by death receptors such as FAS
or by the cytochrome C released from mitochondria,
respectively. These proximal caspases trigger the acti-
vation of a dozen or more effector caspases that execute
the death program, through selective destruction of sub-
cellular structures and organelles, and of the genome.
The possibility that apoptosis serves as a barrier to
cancer was first raised in 1972, when Kerr, Wyllie, and
Currie described massive apoptosis in the cells populat-
ing rapidly growing, hormone-dependent tumors follow-
ing hormone withdrawal (Kerr et al., 1972). The discovery
Cell
62
of the bcl-2 oncogene by its upregulation via chromo-
somal translocation in follicular lymphoma (reviewed in
Korsmeyer, 1992) and its recognition as having anti-
apoptotic activity (Vaux et al., 1988) opened up the in-
vestigation of apoptosis in cancer at the molecular level.
When coexpressed with a myc oncogene in transgenic
mice, the bcl-2 gene was able to promote formation of
B cell lymphomas by enhancing lymphocyte survival, not
by further stimulating their myc-induced proliferation
(Strasser et al., 1990); further, 50% of the infrequent
lymphomas arising in bcl-2 single transgenic transgenic
mice had somatic translocations activating c-myc, con-
firming a selective pressure during lymphomagenesis
to upregulate both Bcl-2 and c-Myc (McDonnell and
Korsmeyer, 1991).
Further insight into the myc-bcl-2 interaction emerged
later from studying the effects of a myc oncogene on
fibroblasts cultured in low serum. Widespread apoptosis
was induced in myc-expressing cells lacking serum; the
consequent apoptosis could be abrogated by exoge-
nous survival factors (e.g., IGF-1), by forced overexpres-
sion of Bcl-2 or the related Bcl-XL protein, or by disrup-
tion of the FAS death signaling circuit (Hueber et al.,
1997). Collectively, the data indicate that a cell’s apo-
ptotic program can be triggered by an overexpressed
oncogene. Indeed, elimination of cells bearing activated
oncogenes by apoptosis may represent the primary
means by which such mutant cells are continually culled
from the body’s tissues.
Other examples strengthen the consensus that apo-
ptosis is a major barrier to cancer that must be circum-
vented. Thus, in transgenic mice where the pRb tumor
suppressor was functionally inactivated in the choroid
plexus, slowly growing microscopic tumors arose, ex-
hibiting high apoptotic rates; the additional inactivation
of the p53 tumor suppressor protein, a component of
the apoptotic signaling circuitry, led to rapidly growing
tumors containing low numbers of apoptotic cells (Sy-
monds et al., 1994). The role of extracellular survival
factors is illustrated by disease progression in trans-
genic mice prone to pancreatic islet tumors. If IGF-2
gene expression, which is activated in this tumorigene-
sis pathway, was abrogated using gene knockout mice,
tumor growth and progression were impaired, as evi-
denced by the appearance of comparatively small, be-
nign tumors showing high rates of apoptosis (Christofori
et al., 1994). In these cells, the absence of IGF-2 did not
affect cell proliferation rates, clearly identifying it as an
antiapoptotic survival factor. Collectively, these obser-
vations argue that altering components of the apoptotic
machinery can dramatically affect the dynamics of tu-
mor progression, providing a rationale for the inactiva-
tion of this machinery during tumor development.
Resistance to apoptosis can be acquired by cancer
cells through a variety of strategies. Surely, the most
commonly occurring loss of a proapoptotic regulator
through mutation involves the p53 tumor suppressor
gene. The resulting functional inactivation of its product,
the p53 protein, is seen in greater than 50% of human
cancers and results in the removal of a key component
of the DNA damage sensor that can induce the apoptotic
effector cascade (Harris, 1996). Signals evoked by other
abnormalities, including hypoxia and oncogene hyper-
expression, are also funneled in part via p53 to the apo-
ptotic machinery; these too are impaired at eliciting
apoptosis when p53 function is lost (Levine, 1997). Addi-
tionally, the PI3 kinase–AKT/PKB pathway, which trans-
mits antiapoptotic survival signals, is likely involved in
mitigating apoptosis in a substantial fraction of human
tumors. This survival signaling circuit can be activated
by extracellular factors such as IGF-1/2 or IL-3 (Evan
and Littlewood, 1998), by intracellular signals emanating
from Ras (Downward, 1998), or by loss of the pTEN
tumor suppressor, a phospholipid phosphatase that
normally attenuates the AKT survival signal (Cantley and
Neel, 1999). Recently, a mechanism for abrogating the
FAS death signal has been revealed in a high fraction
of lung and colon carcinoma cell lines: a nonsignaling
decoy receptor for FAS ligand is upregulated, titrating
the death-inducing signal away from the FAS death re-
ceptor (Pitti et al., 1998). We expect that virtually all
cancer cells harbor alterations that enable evasion of
apoptosis.
It is now possible to lay out a provisional apoptotic
signaling circuitry (Figure 2); while incomplete, it is evi-
dent that most regulatory and effector components are
present in redundant form. This redundancy holds im-
portant implications for the development of novel types
of antitumor therapy, since tumor cells that have lost
proapoptotic components are likely to retain other simi-
lar ones. We anticipate that new technologies will be
able to display the apoptotic pathways still operative in
specific types of cancer cells and that new drugs will
enable cross-talk between the still intact components
of parallel apoptotic signaling pathways in tumor cells,
resulting in restoration of the apoptotic defense mecha-
nism, with substantial therapeutic benefit.
Acquired Capability: Limitless Replicative Potential
Three acquired capabilities—growth signal autonomy,
insensitivity to antigrowth signals, and resistance to
apoptosis—all lead to an uncoupling of a cell’s growth
program from signals in its environment. In principle,
the resulting deregulated proliferation program should
suffice to enable the generation of the vast cell popula-
tions that constitute macroscopic tumors. However, re-
search performed over the past 30 years indicates that
this acquired disruption of cell-to-cell signaling, on its
own, does not ensure expansive tumor growth. Many
and perhaps all types of mammalian cells carry an intrin-
sic, cell-autonomous program that limits their multiplica-
tion. This program appears to operate independently of
the cell-to-cell signaling pathways described above. It
too must be disrupted in order for a clone of cells to
expand to a size that constitutes a macroscopic, life-
threatening tumor.
The early work of Hayflick demonstrated that cells in
culture have a finite replicative potential (reviewed in
Hayflick, 1997). Once such cell populations have pro-
gressed through a certain number of doublings, they
stop growing—a process termed senescence. The se-
nescence of cultured human fibroblasts can be circum-
vented by disabling their pRb and p53 tumor suppressor
proteins, enabling these cells to continue multiplying for
additional generations until they enter into a second
Review
63
state termed crisis. The crisis state is characterized by
massive cell death, karyotypic disarray associated with
end-to-end fusion of chromosomes, and the occasional
emergence of a variant (1 in 107) cell that has acquired
the ability to multiply without limit, the trait termed im-
mortalization (Wright et al., 1989).
Provocatively, most types of tumor cells that are prop-
agated in culture appear to be immortalized, suggesting
that limitless replicative potential is a phenotype that
was acquired in vivo during tumor progression and was
essential for the development of their malignant growth
state (Hayflick, 1997). This result suggests that at some
point during the course of multistep tumor progression,
evolving premalignant cell populations exhaust their en-
dowment of allowed doublings and can only complete
their tumorigenic agenda by breaching the mortality bar-
rier and acquiring unlimited replicative potential.
Observations of cultured cells indicate that various
normal human cell types have the capacity for 60–70
doublings. Taken at face value, these numbers make
little sense when attempting to invoke cell mortality as
an impediment to cancer formation: 60–70 doublings
should enable clones of tumor cells to expand to num-
bers that vastly exceed the number of cells in the human
body. If clues from evaluation of proliferation and apo-
ptotic rates in certain human tumors (Wyllie et al., 1980)
and transgenic mouse models (Symonds et al., 1994;
Shibata et al., 1996; Bergers et al., 1998) prove generaliz-
able, the paradox can be resolved: evolving premalig-
nant and malignant cell populations evidence chronic,
widespread apoptosis and consequently suffer consid-
erable cell attrition concomitant with cell accumulation.
Thus, the number of cells in a tumor greatly underrepre-
sents the cell generations required to produce it, raising
the generational limit of normal somatic cells as a barrier
to cancer.
The counting device for cell generations has been
discovered over the past decade: the ends of chromo-
somes, telomeres, which are composed of several thou-
sand repeats of a short 6 bp sequence element. Replica-
tive generations are counted by the 50–100 bp loss of
telomeric DNA from the ends of every chromosome dur-
ing each cell cycle. This progressive shortening has
been attributed to the inability of DNA polymerases to
completely replicate the 3⬘ ends of chromosomal DNA
during each S phase. The progressive erosion of telo-
meres through successive cycles of replication eventu-
ally causes them to lose their ability to protect the ends
of chromosomal DNA. The unprotected chromosomal
ends participate in end-to-end chromosomal fusions,
yielding the karyotypic disarray associated with crisis
and resulting, almost inevitably, in the death of the af-
fected cell (Counter et al., 1992).
Telomere maintenance is evident in virtually all types of
malignant cells (Shay and Bacchetti, 1997); 85%–90%
of them succeed in doing so by upregulating expression
of the telomerase enzyme, which adds hexanucleotide
repeats onto the ends of telomeric DNA (Bryan and
Cech, 1999), while the remainder have invented a way
of activating a mechanism, termed ALT, which appears
to maintain telomeres through recombination-based in-
terchromosomal exchanges of sequence information
(Bryan et al., 1995). By one or the other mechanism,
telomeres are maintained at a length above a critical
threshold, and this in turn permits unlimited multiplica-
tion of descendant cells. Both mechanisms seem to be
strongly suppressed in most normal human cells in order
to deny them unlimited replicative potential.
The role of telomerase in immortalizing cells can be
demonstrated directly by ectopically expressing the en-
zyme in cells, where it can convey unlimited replicative
potential onto a variety of normal early passage, prese-
nescent cells in vitro (Bodnar et al., 1998; Vaziri and
Benchimol, 1998). Further, late passage cells poised to
enter crisis continue to proliferate without giving any
evidence of crisis when supplied with this enzyme
(Counter et al., 1998; Halvorsen et al., 1999; Zhu et al.,
1999). Additional clues into the importance of telomere
maintenance for cancer comes from analysis of mice
lacking telomerase function. For example, mice carrying
a homozygous knockout of the cell cycle inhibitor
p16INK4A are tumor prone, particularly when exposed to
carcinogens; the tumors that arise show comparatively
elevated telomerase activity. When carcinogens were
applied to p16INK4A-null mice that also lacked telomerase,
tumor incidence was reduced, concomitant with sub-
stantial telomere shortening and karyotypic disarray in
those tumors that did appear (Greenberg et al., 1999).
While telomere maintenance is clearly a key compo-
nent of the capability for unlimited replication, we remain
uncertain about another one, the circumvention of cellu-
lar senescence. The phenomenon of senescence was
originally observed as a delayed response of primary
cells to extended propagation in vitro and has thus been
associated with mechanisms of divisional counting
(Hayflick, 1997). More recently, the senescent state has
been observed to be inducible in certain cultured cells
in response to high level expression of genes such as
the activated ras oncogene (Serrano et al., 1997).
The above-cited observations might argue that senes-
cence, much like apoptosis, reflects a protective mecha-
nism that can be activated by shortened telomeres or
conflicting growth signals that forces aberrant cells irre-
versibly into a G0-like state, thereby rendering them inca-
pable of further proliferation. If so, circumvention of se-
nescence in vivo may indeed represent an essential step
in tumor progression that is required for the subsequent
approach to and breaching of the crisis barrier. But we
consider an alternative model equally plausible: senes-
cence could be an artifact of cell culture that does not
reflect a phenotype of cells within living tissues and
does not represent an impediment to tumor progression
in vivo. Resolution of this quandary will be critical to
completely understand the acquisition of limitless repli-
cative potential.
Acquired Capability: Sustained Angiogenesis
The oxygen and nutrients supplied by the vasculature
are crucial for cell function and survival, obligating virtu-
ally all cells in a tissue to reside within 100 ␮m of a
capillary blood vessel. During organogenesis, this close-
ness is ensured by coordinated growth of vessels and
parenchyma. Once a tissue is formed, the growth of
new blood vessels—the process of angiogenesis—is
transitory and carefully regulated. Because of this de-
pendence on nearby capillaries, it would seem plausible
that proliferating cells within a tissue would have an
Cell
64
intrinsic ability to encourage blood vessel growth. But
the evidence is otherwise. The cells within aberrant pro-
liferative lesions initially lack angiogenic ability, curtail-
ing their capability for expansion. In order to progress
to a larger size, incipient neoplasias must develop angio-
genic ability (Bouck et al., 1996; Hanahan and Folkman,
1996; Folkman, 1997).
Counterbalancing positive and negative signals en-
courage or block angiogenesis. One class of these sig-
nals is conveyed by soluble factors and their receptors,
the latter displayed on the surface of endothelial cells;
integrins and adhesion molecules mediating cell–matrix
and cell–cell association also play critical roles. The
angiogenesis-initiating signals are exemplified by vas-
cular endothelial growth factor (VEGF) and acidic and
basic fibroblast growth factors (FGF1/2). Each binds to
transmembrane tyrosine kinase receptors displayed by
endothelial cells (Fedi et al., 1997; Veikkola and Alitalo,
1999). A prototypical angiogenesis inhibitor is throm-
bospondin-1, which binds to CD36, a transmembrane
receptor on endothelial cells coupled to intracellular Src-
like tyrosine kinases (Bull et al., 1994). There are cur-
rently more than two dozen angiogenic inducer factors
known and a similar number of endogenous inhibitor
proteins.
Integrin signaling also contributes to this regulatory
balance. Quiescent vessels express one class of inte-
grins, whereas sprouting capillaries express another.
Interference with signaling from the latter class of inte-
grins can inhibit angiogenesis (Varner and Cheresh,
1996; Giancotti and Ruoslahti, 1999), underscoring the
important contribution of cell adhesion to the angiogenic
program (Hynes and Wagner, 1996). Extracellular prote-
ases are physically and functionally connected with pro-
angiogenic integrins, and both help dictate the invasive
capability of angiogenic endothelial cells (Stetler-Ste-
venson, 1999).
Experimental evidence for the importance of inducing
and sustaining angiogenesis in tumors is both extensive
and compelling (Bouck et al., 1996; Hanahan and Folk-
man, 1996; Folkman, 1997). The story begins almost 30
years ago with Folkman and colleagues, who used in
vivo bioassays to demonstrate the necessity of angio-
genesis for explosive growth of tumor explants (re-
viewed in Folkman, 1997). Molecular proof of principle
came, for example, when anti-VEGF antibodies proved
able to impair neovascularization and growth of subcu-
taneous tumors in mice (Kim et al., 1993), as did a domi-
nant-interfering version of the VEGF receptor 2 (flk-1)
(Millauer et al., 1994); both results have motivated the
development of specific VEGF/VEGF-R inhibitors now
in late stage clinical trials.
The essential role of angiogenesis is further supported
by the ability of an increasing catalog of antiangiogenic
substances to impair the growth of tumor cells inocu-
lated subcutaneously in mice (Folkman, 1997). Tumors
arising in cancer-prone transgenic mice are similarly
susceptible to angiogenic inhibitors (Bergers et al.,
1999).
The ability to induce and sustain angiogenesis seems
to be acquired in a discrete step (or steps) during tumor
development, via an “angiogenic switch” from vascular
quiescence. When three transgenic mouse models were
analyzed throughout multistep tumorigenesis, in each
case angiogenesis was found to be activated in mid-
stage lesions, prior to the appearance of full-blown tu-
mors. Similarly, angiogenesis can be discerned in pre-
malignant lesions of the human cervix, breast, and skin
(melanocytes) (Hanahan and Folkman, 1996); we expect
that induction of angiogenesis will prove to be an early
to midstage event in many human cancers. These obser-
vations, taken together with the effects of angiogenesis
inhibitors, indicate that neovascularization is a prerequi-
site to the rapid clonal expansion associated with the
formation of macroscopic tumors.
Tumors appear to activate the angiogenic switch by
changing the balance of angiogenesis inducers and
countervailing inhibitors (Hanahan and Folkman, 1996).
One common strategy for shifting the balance involves
altered gene transcription. Many tumors evidence in-
creased expression of VEGF and/or FGFs compared to
their normal tissue counterparts. In others, expression
of endogenous inhibitors such as thrombospondin-1 or
␤-interferon is downregulated. Moreover, both transi-
tions may occur, and indeed be linked, in some tumors
(Singh et al., 1995; Volpert et al., 1997).
The mechanisms underlying shifts in the balances be-
tween angiogenic regulators remain incompletely un-
derstood. In one well-documented example, the inhibi-
tor thrombospondin-1 has been found to positively
regulated by the p53 tumor suppressor protein in some
cell types. Consequently, loss of p53 function, which
occurs in most human tumors, can cause thrombospon-
din-1 levels to fall, liberating endothelial cells from its
inhibitory effects (Dameron et al., 1994). The VEGF gene
is also under complex transcriptional control. For exam-
ple, activation of the ras oncogene or loss of the VHL
tumor suppressor gene in certain cell types causes
upregulation of VEGF expression (Rak et al., 1995; Max-
well et al., 1999).
Another dimension of regulation is emerging in the
form of proteases, which can control the bioavailability
of angiogenic activators and inhibitors. Thus, a variety
of proteases can release bFGF stored in the ECM
(Whitelock et al., 1996), whereas plasmin, a proangio-
genic component of the clotting system, can cleave itself
into an angiogenesis inhibitor form called angiostatin
(Gately et al., 1997). The coordinated expression of pro-
and antiangiogenic signaling molecules, and their mod-
ulation by proteolysis, appear to reflect the complex
homeostatic regulation of normal tissue angiogenesis
and of vascular integrity.
As is already apparent, tumor angiogenesis offers a
uniquely attractive therapeutic target, indeed one that
is shared in common by most and perhaps all types of
human tumors. The next decade will produce a catalog
of the angiogenic regulatory molecules expressed by
different types of tumors, and in many cases, by their
progenitor stages. Use of increasingly sophisticated
mouse models will make it possible to assign specific
roles to each of these regulators and to discern the
molecular mechanisms that govern their production and
activity. Already available evidence indicates that differ-
ent types of tumor cells use distinct molecular strategies
to activate the angiogenic switch. This raises the ques-
tion of whether a single antiangiogenic therapeutic will
suffice to treat all tumor types, or whether an ensemble
of such therapeutics will need to be developed, each
Review
65
responding to a distinct program of angiogenesis that
has been developed by a specific class of human
tumors.
Acquired Capability: Tissue Invasion and Metastasis
Sooner or later during the development of most types
of human cancer, primary tumor masses spawn pioneer
cells that move out, invade adjacent tissues, and thence
travel to distant sites where they may succeed in found-
ing new colonies. These distant settlements of tumor
cells—metastases—are the cause of 90% of human can-
cer deaths (Sporn, 1996). The capability for invasion and
metastasis enables cancer cells to escape the primary
tumor mass and colonize new terrain in the body where,
at least initially, nutrients and space are not limiting. The
newly formed metastases arise as amalgams of cancer
cells and normal supporting cells conscripted from the
host tissue. Like the formation of the primary tumor
mass, successful invasion and metastasis depend upon
all of the other five acquired hallmark capabilities. But
what additional cellular changes enable the acquisition
of these final capabilities during tumorigenesis?
Invasion and metastasis are exceedingly complex
processes, and their genetic and biochemical determi-
nants remain incompletely understood. At the mecha-
nistic level, they are closely allied processes, which justi-
fies their association with one another as one general
capability of cancer cells. Both utilize similar operational
strategies, involving changes in the physical coupling
of cells to their microenvironment and activation of ex-
tracellular proteases.
Several classes of proteins involved in the tethering
of cells to their surroundings in a tissue are altered in
cells possessing invasive or metastatic capabilities. The
affected proteins include cell–cell adhesion molecules
(CAMs)—notably members of the immunoglobulin and
calcium-dependent cadherin families, both of which me-
diate cell-to-cell interactions—and integrins, which link
cells to extracellular matrix substrates. Notably, all of
these “adherence” interactions convey regulatory sig-
nals to the cell (Aplin et al., 1998). The most widely
observed alteration in cell-to-environment interactions
in cancer involves E-cadherin, a homotypic cell-to-cell
interaction molecule ubiquitously expressed on epithe-
lial cells. Coupling between adjacent cells by E-cadherin
bridges results in the transmission of antigrowth and
other signals via cytoplasmic contacts with ␤-catenin
to intracellular signaling circuits that include the Lef/
Tcf transcription factor (Christofori and Semb, 1999).
E-cadherin function is apparently lost in a majority of
epithelial cancers, by mechanisms that include muta-
tional inactivation of the E-cadherin or ␤-catenin genes,
transcriptional repression, or proteolysis of the extracel-
lular cadherin domain (Christofori and Semb, 1999).
Forced expression of E-cadherin in cultured cancer cells
and in a transgenic mouse model of carcinogenesis im-
pairs invasive and metastatic phenotypes, whereas in-
terference with E-cadherin function enhances both
capabilities (Christofori and Semb, 1999). Thus, E-cad-
herin serves as a widely acting suppressor of invasion
and metastasis by epithelial cancers, and its functional
elimination represents a key step in the acquisition of
this capability.
Changes in expression of CAMs in the immunoglobu-
lin superfamily also appear to play critical roles in the
processes of invasion and metastasis (Johnson, 1991).
The clearest case involves N-CAM, which undergoes a
switch in expression from a highly adhesive isoform to
poorly adhesive (or even repulsive) forms in Wilms’ tu-
mor, neuroblastoma, and small cell lung cancer (John-
son, 1991; Kaiser et al., 1996) and reduction in overall
expression level in invasive pancreatic and colorectal
cancers (Fogar et al., 1997). Experiments in transgenic
mice support a functional role for the normal adhesive
form of N-CAM in suppressing metastasis (Perl et al.,
1999).
Changes in integrin expression are also evident in
invasive and metastatic cells. Invading and metastasiz-
ing cancer cells experience changing tissue microenvi-
ronments during their journeys, which can present novel
matrix components. Accordingly, successful coloniza-
tion of these new sites (both local and distant) demands
adaptation, which is achieved through shifts in the spec-
trum of integrin ␣ or ␤ subunits displayed by the migrat-
ing cells. These novel permutations result in different
integrin subtypes (of which there are greater than 22)
having distinct substrate preferences. Thus, carcinoma
cells facilitate invasion by shifting their expression of
integrins from those that favor the ECM present in nor-
mal epithelium to other integrins (e.g., ␣3␤1 and ␣V␤3)
that preferentially bind the degraded stromal compo-
nents produced by extracellular proteases (Varner and
Cheresh, 1996; Lukashev and Werb, 1998). Forced ex-
pression of integrin subunits in cultured cells can induce
or inhibit invasive and metastatic behavior, consistent
with a role of these receptors in acting as central deter-
minants of these processes (Varner and Cheresh, 1996).
Attempts at explaining the cell biological effects of
integrins in terms of a small number of mechanistic rules
have been confounded by the large number of distinct
integrin genes, by the even larger number of heterodi-
meric receptors resulting from combinatorial expression
of various ␣ and ␤ receptor subunits, and by the increas-
ing evidence of complex signals emitted by the cyto-
plasmic domains of these receptors (Aplin et al., 1998;
Giancotti and Ruoslahti, 1999). Still, there is little doubt
that these receptors play central roles in the capability
for tissue invasion and metastasis.
The second general parameter of the invasive and
metastatic capability involves extracellular proteases
(Coussens and Werb, 1996; Chambers and Matrisian,
1997). Protease genes are upregulated, protease inhibi-
tor genes are downregulated, and inactive zymogen
forms of proteases are converted into active enzymes.
Matrix-degrading proteases are characteristically asso-
ciated with the cell surface, by synthesis with a trans-
membrane domain, binding to specific protease re-
ceptors, or association with integrins (Werb, 1997;
Stetler-Stevenson, 1999). One imagines that docking of
active proteases on the cell surface can facilitate inva-
sion by cancer cells into nearby stroma, across blood
vessel walls, and through normal epithelial cell layers.
That notion notwithstanding, it is difficult to unambigu-
ously ascribe the functions of particular proteases solely
to this capability, given their evident roles in other
hallmark capabilities, including angiogenesis (Stetler-
Stevenson, 1999) and growth signaling (Werb, 1997;
Cell
66
Bergers and Coussens, 2000), which in turn contribute
directly or indirectly to the invasive/metastatic capa-
bility.
A further dimension of complexity derives from the
multiple cell types involved in protease expression and
display. In many types of carcinomas, matrix-degrading
proteases are produced not by the epithelial cancer cells
but rather by conscripted stromal and inflammatory cells
(Werb, 1997); once released by these cells, they may be
wielded by the carcinoma cells. For example, certain
cancer cells induce urokinase (uPA) expression in cocul-
tured stromal cells, which then binds to the urokinase
receptor (uPAR) expressed on the cancer cells (Johnsen
et al., 1998).
The activation of extracellular proteases and the al-
tered binding specificities of cadherins, CAMs, and inte-
grins are clearly central to the acquisition of invasive-
ness and metastatic ability. But the regulatory circuits
and molecular mechanisms that govern these shifts re-
main elusive and, at present, seem to differ from one
tissue environment to another. The acquired capability
for invasion and metastasis represents the last great
frontier for exploratory cancer research. We envision
that evolving analytic techniques will soon make it possi-
ble to construct comprehensive profiles of the expres-
sion and functional activities of proteases, integrins, and
CAMs in a wide variety of cancer types, both before and
after they acquire invasive and metastatic abilities. The
challenge will then be to apply the new molecular in-
sights about tissue invasiveness and metastasis to the
development of effective therapeutic strategies.
An Enabling Characteristic: Genome Instability
The acquisition of the enumerated six capabilities during
the course of tumor progression creates a dilemma.
Figure 4. Parallel Pathways of Tumorigen-
esis
While we believe that virtually all cancers
must acquire the same six hallmark capabili-
ties (A), their means of doing so will vary sig-
nificantly, both mechanistically (see text) and
chronologically (B). Thus, the order in which
these capabilities are acquired seems likely
be quite variable across the spectrum of can-
cer types and subtypes. Moreover, in some
tumors, a particular genetic lesion may confer
several capabilities simultaneously, decreas-
ing the number of distinct mutational steps
required to complete tumorigenesis. Thus,
loss of function of the p53 tumor suppressor
can facilitate both angiogenesis and resis-
tance to apoptosis (e.g., in the five-step path-
way shown), as well as enabling the charac-
teristic of genomic instability. In other tumors,
a capability may only be acquired through the
collaboration of two or more distinct genetic
changes, thereby increasing the total number
necessary for completion of tumor progres-
sion. Thus, in the eight-step pathway shown,
invasion/metastasis and resistance to apo-
ptosis are each acquired in two steps.
The available evidence suggests that most are acquired,
directly or indirectly, through changes in the genomes
of cancer cells. But mutation of specific genes is an
inefficient process, reflecting the unceasing, fastidious
maintenance of genomic integrity by a complex array
of DNA monitoring and repair enzymes. These genome
maintenance teams strive to ensure that DNA sequence
information remains pristine. Karyotypic order is guaran-
teed by yet other watchmen, manning so-called check-
points, that operate at critical times in the cell’s life,
notably mitosis. Together, these systems ensure that
mutations are rare events, indeed so rare that the multi-
ple mutations known to be present in tumor cell ge-
nomes are highly unlikely to occur within a human life
span.
Yet cancers do appear at substantial frequency in
the human population, causing some to argue that the
genomes of tumor cells must acquire increased mutabil-
ity in order for the process of tumor progression to reach
completion in several decades time (Loeb, 1991). Mal-
function of specific components of these genomic
“caretaker” systems has been invoked to explain this
increased mutability (Lengauer et al., 1998). The most
prominent member of these systems is the p53 tumor
suppressor protein, which, in response to DNA damage,
elicits either cell cycle arrest to allow DNA repair to take
place or apoptosis if the damage is excessive. Indeed, it
is now clear that the functioning of the p53 DNA damage
signaling pathway is lost in most, if not all, human can-
cers (Levine, 1997). Moreover, a growing number of
other genes involved in sensing and repairing DNA dam-
age, or in assuring correct chromosomal segregation
during mitosis, is found to be lost in different cancers,
labeling these caretakers as tumor suppressors (Len-
gauer et al., 1998). Their loss of function is envisioned
Review
67
to allow genome instability and variability and the gener-
ation of consequently mutant cells with selective advan-
tages. Interestingly, recent evidence suggests that apo-
ptosis may also be a vehicle of genomic instability, in
that DNA within apoptotic cell bodies can be incorpo-
rated into neighboring cells following phagoctytosis
(Holmgren et al., 1999), in principle genetically diversify-
ing any of the constituent cell types of a tumor. We place
this acquired characteristic of genomic instability apart
from the six acquired capabilities associated with tumor
cell phenotype and tumor physiology: it represents the
means that enables evolving populations of premalig-
nant cells to reach these six biological endpoints.
Alternative Pathways to Cancer
The paths that cells take on their way to becoming malig-
nant are highly variable. Within a given cancer type,
mutation of particular target genes such as ras or p53
may be found in only a subset of otherwise histologically
identical tumors. Further, mutations in certain onco-
genes and tumor suppressor genes can occur early in
some tumor progression pathways and late in others. As
a consequence, the acquisition of biological capabilities
such as resistance to apoptosis, sustained angiogen-
esis, and unlimited replicative potential can appear at
different times during these various progressions. Ac-
cordingly, the particular sequence in which capabilities
are acquired can vary widely, both among tumors of the
same type and certainly between tumors of different
types (Figure 4). Furthermore, in certain tumors, a spe-
cific genetic event may, on its own, contribute only par-
tially to the acquisition of a single capability, while in
others, this event may aid in the simultaneous acquisi-
tion of several distinct capabilities. Nonetheless, we be-
lieve that independent of how the steps in these genetic
pathways are arranged, the biological endpoints that
are ultimately reached—the hallmark capabilities of can-
cer—will prove to be shared in common by all types of
tumors.
Synthesis
Cancer cells propagated in culture and dissected into
their molecular components have yielded much of the
wealth of information that we currently possess about
the molecular processes underlying cancer develop-
ment. Yet by simplifying the nature of cancer—por-
traying it as a cell-autonomous process intrinsic to the
cancer cell—these experimental models have turned
their back on a central biological reality of tumor forma-
tion in vivo: cancer development depends upon changes
in the heterotypic interactions between incipient tumor
cells and their normal neighbors. Moreover, once formed,
virtually all types of human tumors, including their meta-
static outgrowths, continue to harbor complex mixtures
of several cell types that collaborate to create malignant
growth (Figure 3). This reconceptualization of cancer
cell biology has begun to drive profound changes in
how we study this disease experimentally. Continuing
elucidation of cancer pathogenesis will depend increas-
ingly upon heterotypic organ culture systems in vitro
and evermore refined mouse models in vivo. Looking
ahead into the future, these systems will help us chart
comprehensive maps of growth signaling networks in
cancer, an endeavor that will depend on defining all of
the signals exchanged between the various cell types
existing symbiotically within a tumor mass and knowing
their effects on the integrated circuits of each of those
cell types.
Our ability to analyze individual human cancers at
the genetic and biochemical levels will also undergo a
dramatic change. At present, description of a recently
diagnosed tumor in terms of its underlying genetic le-
sions remains a distant prospect. Nonetheless, we look
ahead 10 or 20 years to the time when the diagnosis of
all the somatically acquired lesions present in a tumor
cell genome will become a routine procedure. By then,
genome-wide gene expression profiles of tumor cells
will also be routine. With all this information in hand, it
will become possible to test definitively our proposition
that the development of all types of human tumor cells
is governed by a common set of rules such as those
implied by the six acquired capabilities enumerated
here.
We anticipate far deeper insight into the roles played
by inherited alleles in cancer susceptibility and patho-
genesis. At present, our understanding of the interplay
at the cellular level between inherited cancer modifier
genes with oncogenes and tumor suppressor genes that
are altered somatically is rudimentary; modifiers can in
principle act in any of the constituent cell types of a
tumor, or elsewhere in the body, whereas the classical
cancer genes largely act in the cancer cells themselves.
These gaps will be bridged in part by new informatics
technologies, enabling us to process and interpret the
inundation of genetic information that will soon flow from
automated sequencing instruments. New technologies
will also aid us in rationalizing the complex constella-
tions of interacting alleles in terms of a systematics of
cancer formation of the type that we propose here.
The metaphors used to conceptualize cancer cell
function will also shift dramatically. For decades now,
we have been able to predict with precision the behavior
of an electronic integrated circuit in terms of its constit-
uent parts—its interconnecting components, each re-
sponsible for acquiring, processing, and emitting signals
according to a precisely defined set of rules. Two de-
cades from now, having fully charted the wiring dia-
grams of every cellular signaling pathway, it will be pos-
sible to lay out the complete “integrated circuit of the
cell” upon its current outline (Figure 2). We will then be
able to apply the tools of mathematical modeling to
explain how specific genetic lesions serve to reprogram
this integrated circuit in each of the constituent cell
types so as to manifest cancer.
With holistic clarity of mechanism, cancer prognosis
and treatment will become a rational science, unrecog-
nizable by current practitioners. It will be possible to
understand with precision how and why treatment regi-
mens and specific antitumor drugs succeed or fail. We
envision anticancer drugs targeted to each of the hall-
mark capabilities of cancer; some, used in appropriate
combinations and in concert with sophisticated technol-
ogies to detect and identify all stages of disease pro-
gression, will be able to prevent incipient cancers from
developing, while others will cure preexisting cancers,
elusive goals at present. One day, we imagine that can-
cer biology and treatment—at present, a patchwork quilt
of cell biology, genetics, histopathology, biochemistry,
Cell
68
immunology, and pharmacology—will become a sci-
ence with a conceptual structure and logical coherence
that rivals that of chemistry or physics.
Acknowledgments
We wish to thank Terry Schoop of Biomed Arts Associates, San
Francisco, for preparation of the figures, Cori Bargmann and Zena
Werb for insightful comments on the manuscript, and Normita San-
tore for editorial assistance. In addition, we are indebted to Joe
Harford and Richard Klausner, who allowed us to adapt and expand
their depiction of the cell signaling network, and we appreciate
suggestions on signaling pathways from Randy Watnick, Brian Elen-
bas, Bill Lundberg, Dave Morgan, and Henry Bourne. R. A. W. is
a Ludwig Foundation and American Cancer Society Professor of
Biology. His work has been supported by the Department of the
Army and the National Institutes of Health. D. H. acknowledges
the support and encouragement of the National Cancer Institute.
Editorial policy has rendered the citations illustrative but not com-
prehensive.
References
Aplin, A.E., Howe, A., Alahari, S.K., and Juliano, R.L. (1998). Signal
transduction and signal modulation by cell adhesion receptors: the
role of integrins, cadherins, immunoglobulin-cell adhesion mole-
cules, and selectins. Pharmacol. Rev. 50, 197–263.
Ashkenazi, A., and Dixit, V.M. (1999). Apoptosis control by death
and decoy receptors. Curr. Opin. Cell Biol. 11, 255–260.
Bergers, G., and Coussens, L.M. (2000). Extrinsic regulators of epi-
thelial tumor progression: metalloproteinases. Curr. Opin. Genet.
Dev., in press.
Bergers, G., Hanahan, D., and Coussens, L.M. (1998). Angiogenesis
and apoptosis are cellular parameters of neoplastic progression in
transgenic mouse models of tumorigenesis. Int. J. Dev. Biol. 42,
995–1002.
Bergers, G., Javaherian, K., Lo, K.-M., Folkman, J., and Hanahan,
D. (1999). Effects of angiogenesis inhibitors on multistage carcino-
genesis in mice. Science 284, 808–812.
Bishop, J.M., and Weinberg, R.A., eds. (1996). Molecular Oncology
(New York: Scientific American, Inc.).
Bodnar, A.G., Ouellete, M., Frolkis, M., Holt, S.E., Chiu, C., Morin,
G.B., Harley, C.B., Shay, J.W., Lichtsteiner, S., and Wright, W.E.
(1998). Extension of life-span by introduction of telomerase into
normal human cells. Science 279, 349–352.
Bouck, N., Stellmach, V., and Hsu, S.C. (1996). How tumors become
angiogenic. Adv. Cancer Res. 69, 135–174.
Bryan, T.M., and Cech, T.R. (1999). Telomerase and the maintenance
of chromosome ends. Curr. Opin. Cell Biol. 11, 318–324.
Bryan, T.M., Englezou, A., Gupta, J., Bacchetti, S., and Reddel,
R.R. (1995). Telomere elongation in immortal human cells without
detectable telomerase activity. EMBO J. 14, 4240–4248.
Bull, H.A., Brickell, P.M., and Dowd, P.M. (1994). Src-related protein
tyrosine kinases are physically associated with the surface antigen
CD36 in human dermal microvascular endothelial cells. FEBS Lett.
351, 41–44.
Butt, A.J., Firth, S.M., and Baxter, R.C. (1999). The IGF axis and
programmed cell death. Immunol. Cell Biol. 77, 256–262.
Cantley, L.C., and Neel, B.G. (1999). New insights into tumor sup-
pression: PTEN suppresses tumor formation by restraining the
phosphoinositide 3-kinase/AKT pathway. Proc. Natl. Acad. Sci. USA
96, 4240–4245.
Chambers, A.F., and Matrisian, L.M. (1997). Changing views of the
role of matrix metalloproteinases in metastasis. J. Natl. Cancer Inst.
89, 1260–1270.
Chin, L., Pomerantz, J., and DePinho, R.A. (1998). The INK4a/ARF
tumor suppressor: one gene—two products—two pathways. Trends
Biochem. Sci. 23, 291–296.
Christofori, G., and Semb, H. (1999). The role of the cell-adhesion
molecule E-cadherin as a tumour-suppressor gene. Trends Bio-
chem. Sci. 24, 73–76.
Christofori, G., Naik, P., and Hanahan, D. (1994). A second signal
supplied by insulin-like growth factor II in oncogene-induced tumori-
genesis. Nature 369, 414–418.
Cordon-Cardo, C., and Prives, C. (1999). At the crossroads of inflam-
mation and tumorigenesis. J. Exp. Med. 190, 1367–1370.
Counter, C.M., Avilion, A.A., LeFeuvre, C.E., Stewart, N.G., Greider,
C.W., Harley, C.B., and Bacchetti, S. (1992). Telomere shortening
associated with chromosome instability is arrested in immortal cells
which express telomerase activity. EMBO J. 11, 1921–1929.
Counter, C.M., Hahn, W.C., Wei, W., Dickinson Caddle, S., Beijers-
bergen, R.L., Lansdorp, P.M., Sedivy, J.M., and Weinberg, R.A.
(1998). Dissociation between telomerase activity, telomere mainte-
nance and cellular immortalization. Proc. Natl. Acad. Sci. USA 95,
14723–14728.
Coussens, L.M., and Werb, Z. (1996). Matrix metalloproteinases and
the development of cancer. Chem. Biol. 3, 895–904.
Coussens, L.M., Raymond, W.W., Bergers, G., Laig-Webster, M.,
Behrendtsen, O., Werb, Z., Caughey, G.H., and Hanahan, D. (1999).
Inflammatory mast cells up-regulate angiogenesis during squamous
epithelial carcinogenesis. Genes Dev. 13, 1382–1397.
Dameron, K.M., Volpert, O.V., Tainsky, M.A., and Bouck, N. (1994).
Control of angiogenesis in fibroblasts by p53 regulation of throm-
bospondin-1. Science 265, 1582–1584.
Datto, M.B., Hu, P.P., Kowalik, T.F., Yingling, J., and Wang, X.F.
(1997). The viral oncoprotein E1A blocks transforming growth factor
␤-mediated induction of p21/WAF1/Cip1 and p15/INK4B Mol. Cell.
Biol. 17, 2030–2037.
DiFiore, P.P., Pierce, J.H., Kraus, M.H., Segatto, O., King, C.R., and
Aaronson, S.A. (1987). erbB-2 is a potent oncogene when overex-
pressed in NIH/3T3 cells. Science 237, 178–182.
Downward, J. (1998). Mechanisms and consequences of activation
of protein kinase B/Akt. Curr. Opin. Cell Biol. 10, 262–267.
Dyson, N., Howley, P.M., Munger, K., and Harlow, E. (1989). The
human papillomavirus-16 E7 oncoprotein is able to bind to the reti-
noblastoma gene product. Science 243, 934–937.
Evan, G., and Littlewood, T. (1998). A matter of life and cell death.
Science 281, 1317–1322.
Fedi, P., Tronick, S.R., and Aaronson, S.A. (1997). Growth factors.
In Cancer Medicine, J.F. Holland, R.C. Bast, D.L. Morton, E. Frei,
D.W. Kufe, and R.R. Weichselbaum, eds. (Baltimore, MD: Williams
and Wilkins), pp. 41–64.
Fogar, P., Basso, D., Pasquali, C., De Paoli, C., Sperti, C., Roveroni,
G., Pedrazzoli, G., and Plebani, M. (1997). Neural cell adhesion mole-
cule (N-CAM) in gastrointestinal neoplasias. Anticancer Res. 17,
1227–1230.
Foley, K.P., and Eisenman, R.N. (1999). Two MAD tails: what the
recent knockouts of Mad1 and Mx1 tell us about the MYC/MAX/
MAD network. Biochim. Biophys. Acta 1423, M37–47.
Folkman, J. (1997). Tumor angiogenesis. In Cancer Medicine, J.F.
Holland, R.C. Bast, D.L. Morton, E. Frei, D.W. Kufe, and R.R. Weich-
selbaum, eds. (Baltimore, MD: Williams and Wilkins), pp. 181–204.
Foulds, L. (1954). The Experimental Study of Tumor Progression.
Volumes I–III (London: Academic Press).
Fynan, T.M., and Reiss, M. (1993). Resistance to inhibition of cell
growth by transforming growth factor-␤ and its role in oncogenesis.
Crit. Rev. Oncog. 4, 493–540.
Gately, S., Twardowski, P., Stack, M.S., Cundiff, D.L., Grella, D.,
Castellino, F.J., Enghild, J., Kwaan, H.C., Lee, F., Kramer, R.A., et
al. (1997). The mechanism of cancer-mediated conversion of plas-
minogen to the angiogenesis inhibitor angiostatin. Proc. Natl. Acad.
Sci. USA 94, 10868–10872.
Giancotti, F.G., and Ruoslahti, E. (1999). Integrin signaling. Science
285, 1028–1032.
Green, D.R., and Reed, J.C. (1998). Mitochondria and apoptosis.
Science 281, 1309–1312.
Greenberg, R.A., Chin, L., Femino, A., Lee, K.H., Gottlieb, G.J.,
Singer, R.H., Greider, C.W., and DePinho, R.A. (1999). Short dysfunc-
tional telomeres impair tumorigenesis in the INK4a⌬2/3 cancer-prone
mouse. Cell 97, 515–525.
Review
69
Hahn, W.C., Counter, C.M., Lundberg, A.S., Beijersbgern, R.L.,
Brooks, M.W., and Weinberg, R.A. (1999). Creation of human tumor
cells with defined genetic elements. Nature 400, 464–468.
Halvorsen, T.L., Leibowitz, G., and Levine, F. (1999). Telomerase
activity is sufficient to allow transformed cells to escape from crisis.
Mol. Cell. Biol. 19, 1864–1870.
Hanahan, D., and Folkman, J. (1996). Patterns and emerging mecha-
nisms of the angiogenic switch during tumorigenesis. Cell 86,
353–364.
Hannon, G.J., and Beach, D. (1994). P15INK4B is a potential effector
of TGF-beta-induced cell cycle arrest. Nature 371, 257–261.
Harris, C.C. (1996). p53 tumor suppressor gene: from the basic
research laboratory to the clinic—an abridged historical perspec-
tive. Carcinogenesis 17, 1187–1198.
Hayflick, L. (1997). Mortality and immortality at the cellular level. A
review. Biochemistry 62, 1180–1190.
Holmgren, L., Szeles, A., Rajnavolgyi, E., Folkman, J., Klein, G.,
Ernberg, I., and Falk, K.I. (1999). Horizontal transfer of DNA by the
uptake of apoptotic bodies. Blood 93, 3956–3963.
Hudson, J.D., Shoaibi, M.A., Maestro, R., Carnero, A., Hannon, G.J.,
and Beach, D.H. (1999). A proinflammatory cytokine inhibits p53
tumor suppressor activity. J. Exp. Med. 190, 1375–1382.
Hueber, A.O., Zornig, M., Lyon, D., Suda, T., Nagata, S., and Evan,
G.I. (1997). Requirement for the CD95 receptor-ligand pathway in
c-Myc-induced apoptosis. Science 278, 1305–1309.
Hunter, T. (1997). Oncoprotein networks. Cell 88, 333–346.
Hynes, R.O., and Wagner, D.D. (1996). Genetic manipulation of vas-
cular adhesion molecules in mice. J. Clin. Invest. 98, 2193–2195.
Ishizaki, Y., Cheng, L., Mudge, A.W., and Raff, M.C. (1995). Pro-
grammed cell death by default in embryonic cells, fibroblasts, and
cancer cells. Mol. Biol. Cell 6, 1443–1458.
Johnsen, M., Lund, L.R., Romer, J., Almholt, K., and Dano K. (1998).
Cancer invasion and tissue remodeling: common themes in proteo-
lytic matrix degradation. Curr. Opin. Cell Biol. 10, 667–671.
Johnson, J.P. (1991). Cell adhesion molecules of the immunoglobu-
lin supergene family and their role in malignant transformation and
progression to metastatic disease. Cancer Metastasis Rev. 10,
11–22.
Kahn, P., Frykberg, L., Brady, C., Stanley, I., Beug, H., Vennström,
B., and Graf, T. (1986). v-erbA cooperates with sarcoma oncogenes
in leukemic cell transformation. Cell 45, 349–356.
Kaiser, U., Auerbach, B., and Oldenburg, M. (1996). The neural cell
adhesion molecule NCAM in multiple myeloma. Leuk. Lymphoma
20, 389–395.
Kerr, J.F., Wyllie, A.H., and Currie, A.R. (1972). Apoptosis: a basic
biological phenomenon with wide-ranging implications in tissue ki-
netics. Br. J. Cancer 26, 239–257.
Kim, K.J., Li, B., Winer, J., Armanini, M., Gillett, N., Philipps, H.S., and
Ferrara, N. (1993). Inhibition of vascular endothelial growth factor-
induced angiogenesis suppresses tumour growth in vivo. Nature
362, 841–844.
Kinzler, K.W., and Vogelstein, B. (1996). Lessons from hereditary
colorectal cancer. Cell 87, 159–170.
Kinzler, K.W., and Vogelstein, B. (1998). Landscaping the cancer
terrain. Science 280, 1036–1037.
Korsmeyer, S.J. (1992). Chromosomal translocations in lymphoid
malignancies reveal novel proto-oncogenes. Annu. Rev. Immunol.
10, 785–807.
Lengauer, C., Kinzler, K.W., and Vogelstein, B. (1998). Genetic insta-
bilities in human cancers. Nature 396, 643–649.
Levine, A.J. (1997). p53, the cellular gatekeeper for growth and divi-
sion. Cell 88, 323–331.
Loeb, L.A. (1991). Mutator phenotype may be required for multistep
carcinogenesis. Cancer Res. 51, 3073–3079.
Lotem, J., and Sachs, L. (1996). Control of apoptosis in hematopoie-
sis and leukemia by cytokines, tumor suppressor and oncogenes.
Leukemia 10, 925–931.
Lukashev, M.E., and Werb, Z. (1998). ECM signaling: orchestrating
cell behaviour and misbehaviour. Trends Cell Biol. 8, 437–441.
Markowitz, S., Wang, J., Meyeroff, L., Parsons, R., Sun, L., Lutter-
baugh, J., Fan, R., Zborowska, E., Kinzler, K., Vogelstein, B., et al.
(1995). Inactivation of the type II TGF-␤ receptor in colon cancer
cells with microsatellite instability. Science 268, 1336–1338.
Maxwell, P.H., Wiesener, M.S., Chang, G.-W., Clifford, S.C., Vaux,
E.C., Cockman, M.E., Wykoff, C.C., Pugh, C.W., Maher, E.R., and
Ratcliffe, P.J. (1999). The tumour suppressor protein VHL targets
hypoxia-inducible factors for oxygen-dependent proteolysis. Nature
399, 271–275.
McDonnell, T.J., and Korsmeyer, S.J. (1991). Progression from
lymphoid hyperplasia to high-grade malignant lymphoma in mice
transgenic for the t(14;18). Nature 349, 254–256.
Medema, R.H., and Bos, J.L. (1993). The role of p21-ras in receptor
tyrosine kinase signaling. Crit. Rev. Oncog. 4, 615–661.
Millauer, B., Shawver, L.K., Plate, K.H., Risau, W., and Ullrich, A.
(1994). Glioblastoma growth inhibited in vivo by a dominant-negative
Flk-1 mutant. Nature 367, 576–579.
Moses, H.L., Yang, E.Y., and Pietenpol, J.A. (1990). TGF-␤ stimula-
tion and inhibition of cell proliferation: new mechanistic insights.
Cell 63, 245–247.
Nowell, P.C. (1976). The clonal evolution of tumor cell populations.
Science 194, 23–28.
Olumi, A.F., Grossfeld, G.D., Hayward, S.W., Carroll, P.R., Tlsty, T.D.,
and Cunha, G.R. (1999). Carcinoma-associated fibroblasts direct
tumor progression of initiated human prostatic epithelium. Cancer
Res. 59, 5002–5011.
Perl, A.-K., Dahl, U., Wilgenbus, P., Cremer, H., Semb, H., and
Christofori, G. (1999). Reduced expresion of neural cell adhesion
molecule induces metastatic dissemination of pancreatic ␤ tumor
cells. Nat. Med. 5, 286–291.
Pitti, R.M., Marsters, S.A., Lawrence, D.A., Roy, M., Kischkel, F.C.,
Dowd, P., Huang, A., Donahue, C.J., Sherwood, S.W., Baldwin, D.T.,
et al. (1998). Genomic amplification of a decoy receptor for Fas
ligand in lung and colon cancer. Nature 396, 699–703.
Rak, J., Filmus, J., Finkenzeller, G., Grugel, S., Marme, D., and Ker-
bel, R.S. (1995). Oncogenes as inducers of tumor angiogenesis.
Cancer Metastasis Rev. 14, 263–277.
Renan, M.J. (1993). How many mutations are required for tumorigen-
esis? Implications from human cancer data. Mol. Carcinogenesis 7,
139–146.
Rommel, C., and Hafen, E. (1998). Ras—a versatile cellular switch.
Curr. Opin. Genet. Dev. 8, 412–418.
Schutte, M., Hruban, R., Hedrick, L., Cho, K., Nadasdy, G., Weinstein,
C., Bova, G., Isaacs, W., Cairns, P., Nawroz, H., et al. (1996). DPC4
gene in various tumor types. Cancer Res. 56, 2527–2530.
Serrano, M., Lin, A.W., McCurrach, M.E., Beach, D., and Lowe, S.W.
(1997). Oncogeneic ras provokes premature cell senescence associ-
ated with accumulation of p53 and p16INK4A. Cell 88, 593–602.
Shibata, M.A., Maroulakou, I.G., Jorcyk, C.L., Gold, L.G., Ward, J.M.,
and Green, J.E. (1996). p53-independent apoptosis during mammary
tumor progression in C3(1)/SV40 large T antigen transgenic mice:
suppression of apoptosis during the transition from preneoplasia
to carcinoma. Cancer Res. 56, 2998–3003.
Shay, J.W., and Bacchetti, S. (1997). A survey of telomerase activity
in human cancer. Eur. J. Cancer 33, 787–791.
Singh, R.K., Gutman, M., Bucana, C.D., Sanchez, R., Llansa, N.,
and Fidler, I.J. (1995). Interferons alpha and beta down-regulate the
expression of basic fibroblast growth factor in human carcinomas.
Proc. Natl. Acad. Sci. USA 92, 4562–4566.
Skobe, M., and Fusenig, N.E. (1998). Tumorigenic conversion of
immortal human keratinocytes through stromal cell activation. Proc.
Natl. Acad. Sci. USA 95, 1050–1055.
Slamon, D.J., Clark, G.M., Wong, S.G., Levin, W.J., Ullrich, A., and
McGuire, W.L. (1987). Human breast cancer: correlation of relapse
and survival with amplification of the HER-2/neu oncogene. Science
235, 177–182.
Sporn, M.B. (1996). The war on cancer. Lancet 347, 1377–1381.
Stetler-Stevenson, W.G. (1999). Matrix metalloproteinases in angio-
genesis: a moving target for therapeutic intervention. J. Clin. Invest.
103, 1237–1241.
Cell
70

Strasser, A., Harris, A.W., Bath, M.L., and Cory, S. (1990). Novel
primitive lymphoid tumours induced in transgenic mice by coopera-
tion between myc and bcl-2. Nature 348, 331–333.
Symonds, H., Krall, L., Remington, L., Saenz-Robles, M., Lowe, S.,
Jacks, T., and Van Dyke, T. (1994). p53-dependent apoptosis sup-
presses tumor growth and progression in vivo. Cell 78, 703–711.
Thornberry, N.A., and Lazebnik, Y. (1998). Caspases: enemies within.
Science 281, 1312–1316.
Varner, J.A., and Cheresh, D.A. (1996). Integrins and cancer. Curr.
Opin. Cell Biol. 8, 724–730.
Vaux, D.L., Cory, S., and Adams, T.M. (1988). Bcl-2 promotes the
survival of hematopoietic cells and cooperates with c-myc to immor-
talize pre-B cells. Nature 335, 440–442.
Vaziri, H., and Benchimol, S. (1998). Reconstitution of telomerase
activity in normal human cells leads to elongation of telomeres and
extended replicative life span. Curr. Biol. 8, 279–282.
Veikkola, T., and Alitalo, K. (1999). VEGFs, receptors and angiogen-
esis. Semin. Cancer Biol. 9, 211–220.
Volpert, O.V., Dameron, K.M., and Bouck, N. (1997). Sequential de-
velopment of an angiogenic phenotype by human fibroblasts pro-
gressing to tumorigenicity. Oncogene 14, 1495–1502.
Weinberg, R.A. (1995). The retinoblastoma protein and cell cycle
control. Cell 81, 323–330.
Werb, Z. (1997). ECM and cell surface proteolysis: regulating cellular
ecology. Cell 91, 439–442.
Whitelock, J.M., Murdoch, A.D., Iozzo, R.V., and Underwood, P.A.
(1996). The degradation of human endothelial cell-derived perlecan
and release of bound basic fibroblast growth factor by stromelysin,
collagenase, plasmin, and heparanases. J. Biol. Chem. 271, 10079–
10086.
Wright, W.E., Pereira-Smith, O.M., and Shay, J.W. (1989). Reversible
cellular senescence: implications for immortalization of normal hu-
man diploid fibroblasts. Mol. Cell. Biol. 9, 3088–3092.
Wyllie, A.H., Kerr, J.F., and Currie, A.R. (1980). Cell death: the signifi-
cance of apoptosis. Int. Rev. Cytol. 68, 251–306.
Yarden, Y., and Ullrich, A. (1988). EGF and erbB2 receptor overex-
pression in human tumors. Growth factor recepor tyrosine kinases.
Annu. Rev. Biochem. 57, 443–478.
Zhu, J., Wang, H., Bishop, J.M., and Blackburn, E.H. (1999). Telo-
merase extends the lifespan of virus-transformed human cells with-
out net telomere lengthening. Proc. Natl. Acad. Sci. USA 96, 3723–
3728.
Zuo, L., Weger, J., Yang, Q., Goldstein, A.M., Tucker, M.A., Walker,
G.J., Hayward, N., and Dracopoli, N.C. (1996). Germline mutations
in the p16INK4A binding domain of CDK4 in familial melanoma. Nat.
Genet. 12, 97–99.
 Leading Edge
Review

Hallmarks of Cancer: The Next Generation

Douglas Hanahan1,2,* and Robert A. Weinberg3,*
1The Swiss Institute for Experimental Cancer Research (ISREC), School of Life Sciences, EPFL, Lausanne CH-1015, Switzerland
2The Department of Biochemistry & Biophysics, UCSF, San Francisco, CA 94158, USA
3Whitehead Institute for Biomedical Research, Ludwig/MIT Center for Molecular Oncology, and MIT Department of Biology, Cambridge,

MA 02142, USA
*Correspondence: dh@epfl.ch (D.H.), weinberg@wi.mit.edu (R.A.W.)
DOI 10.1016/j.cell.2011.02.013

The hallmarks of cancer comprise six biological capabilities acquired during the multistep develop-
ment of human tumors. The hallmarks constitute an organizing principle for rationalizing the
complexities of neoplastic disease. They include sustaining proliferative signaling, evading growth
suppressors, resisting cell death, enabling replicative immortality, inducing angiogenesis, and acti-
vating invasion and metastasis. Underlying these hallmarks are genome instability, which generates
the genetic diversity that expedites their acquisition, and inflammation, which fosters multiple hall-
mark functions. Conceptual progress in the last decade has added two emerging hallmarks of
potential generality to this list—reprogramming of energy metabolism and evading immune
destruction. In addition to cancer cells, tumors exhibit another dimension of complexity: they
contain a repertoire of recruited, ostensibly normal cells that contribute to the acquisition of hall-
mark traits by creating the ‘‘tumor microenvironment.’’ Recognition of the widespread applicability
of these concepts will increasingly affect the development of new means to treat human cancer.

INTRODUCTION vated by these developments, we now revisit the original hall-

We have proposed that six hallmarks of cancer together consti-
tute an organizing principle that provides a logical framework for
understanding the remarkable diversity of neoplastic diseases
(Hanahan and Weinberg, 2000). Implicit in our discussion was
the notion that as normal cells evolve progressively to
a neoplastic state, they acquire a succession of these hallmark
capabilities, and that the multistep process of human tumor
pathogenesis could be rationalized by the need of incipient
cancer cells to acquire the traits that enable them to become
tumorigenic and ultimately malignant.
We noted as an ancillary proposition that tumors are more than
insular masses of proliferating cancer cells. Instead, they are
complex tissues composed of multiple distinct cell types that
participate in heterotypic interactions with one another. We de-
picted the recruited normal cells, which form tumor-associated
stroma, as active participants in tumorigenesis rather than
passive bystanders; as such, these stromal cells contribute to
the development and expression of certain hallmark capabilities.
During the ensuing decade this notion has been solidified and
extended, revealing that the biology of tumors can no longer
be understood simply by enumerating the traits of the cancer
cells but instead must encompass the contributions of the
‘‘tumor microenvironment’’ to tumorigenesis.
In the course of remarkable progress in cancer research
subsequent to this publication, new observations have served
both to clarify and to modify the original formulation of the hall-
mark capabilities. In addition, yet other observations have raised
questions and highlighted mechanistic concepts that were not
integral to our original elaboration of the hallmark traits. Moti-
marks, consider new ones that might be included in this roster,
and expand upon the functional roles and contributions made
by recruited stromal cells to tumor biology.
HALLMARK CAPABILITIES—CONCEPTUAL PROGRESS
The six hallmarks of cancer—distinctive and complementary
capabilities that enable tumor growth and metastatic dissemina-
tion—continue to provide a solid foundation for understanding
the biology of cancer (Figure 1; see the Supplemental Informa-
tion for downloadable versions of the figures for presentations).
In the first section of this Review, we summarize the essence
of each hallmark as described in the original presentation in
2000, followed by selected illustrations (demarcated by sub-
headings in italics) of the conceptual progress made over the
past decade in understanding their mechanistic underpinnings.
In subsequent sections we address new developments that
broaden the scope of the conceptualization, describing in turn
two enabling characteristics crucial to the acquisition of the six
hallmark capabilities, two new emerging hallmark capabilities,
the constitution and signaling interactions of the tumor microen-
vironment crucial to cancer phenotypes, and we finally discuss
the new frontier of therapeutic application of these concepts.
Sustaining Proliferative Signaling
Arguably the most fundamental trait of cancer cells involves their
ability to sustain chronic proliferation. Normal tissues carefully
control the production and release of growth-promoting signals
that instruct entry into and progression through the cell growth-
and-division cycle, thereby ensuring a homeostasis of cell

646 Cell 144, March 4, 2011 ª2011 Elsevier Inc.


number and thus maintenance of normal tissue architecture and
function. Cancer cells, by deregulating these signals, become
masters of their own destinies. The enabling signals are
conveyed in large part by growth factors that bind cell-surface
receptors, typically containing intracellular tyrosine kinase
domains. The latter proceed to emit signals via branched intra-
cellular signaling pathways that regulate progression through
the cell cycle as well as cell growth (that is, increases in cell
size); often these signals influence yet other cell-biological prop-
erties, such as cell survival and energy metabolism.
Remarkably, the precise identities and sources of the prolifer-
ative signals operating within normal tissues were poorly under-
stood a decade ago and in general remain so. Moreover, we still
know relatively little about the mechanisms controlling the
release of these mitogenic signals. In part, the understanding
of these mechanisms is complicated by the fact that the growth
factor signals controlling cell number and position within tissues
are thought to be transmitted in a temporally and spatially regu-
lated fashion from one cell to its neighbors; such paracrine
signaling is difficult to access experimentally. In addition, the
bioavailability of growth factors is regulated by sequestration in
the pericellular space and extracellular matrix, and by the actions
of a complex network of proteases, sulfatases, and possibly
other enzymes that liberate and activate them, apparently in
a highly specific and localized fashion.
The mitogenic signaling in cancer cells is, in contrast, better
understood (Lemmon and Schlessinger, 2010; Witsch et al.,
2010; Hynes and MacDonald, 2009; Perona, 2006). Cancer cells
can acquire the capability to sustain proliferative signaling in
a number of alternative ways: They may produce growth factor
ligands themselves, to which they can respond via the expres-
sion of cognate receptors, resulting in autocrine proliferative
stimulation. Alternatively, cancer cells may send signals to stim-
ulate normal cells within the supporting tumor-associated
stroma, which reciprocate by supplying the cancer cells with
various growth factors (Cheng et al., 2008; Bhowmick et al.,
2004). Receptor signaling can also be deregulated by elevating
the levels of receptor proteins displayed at the cancer cell
Figure 1. The Hallmarks of Cancer
This illustration encompasses the six hallmark
capabilities originally proposed in our 2000 per-
spective. The past decade has witnessed
remarkable progress toward understanding the
mechanistic underpinnings of each hallmark.
surface, rendering such cells hyperre-
sponsive to otherwise-limiting amounts
of growth factor ligand; the same
outcome can result from structural alter-
ations in the receptor molecules that
facilitate ligand-independent firing.
Growth factor independence may also
derive from the constitutive activation of
components of signaling pathways oper-
ating downstream of these receptors,
obviating the need to stimulate these
pathways by ligand-mediated receptor
activation. Given that a number of distinct downstream signaling
pathways radiate from a ligand-stimulated receptor, the activa-
tion of one or another of these downstream pathways, for
example, the one responding to the Ras signal transducer,
may only recapitulate a subset of the regulatory instructions
transmitted by an activated receptor.
Somatic Mutations Activate Additional Downstream
Pathways
High-throughput DNA sequencing analyses of cancer cell
genomes have revealed somatic mutations in certain human
tumors that predict constitutive activation of signaling circuits
usually triggered by activated growth factor receptors. Thus,
we now know that 40% of human melanomas contain
activating mutations affecting the structure of the B-Raf protein,
resulting in constitutive signaling through the Raf to mitogen-
activated protein (MAP)-kinase pathway (Davies and Samuels
2010). Similarly, mutations in the catalytic subunit of phosphoi-
nositide 3-kinase (PI3-kinase) isoforms are being detected in
an array of tumor types, which serve to hyperactivate the PI3-
kinase signaling circuitry, including its key Akt/PKB signal
transducer (Jiang and Liu, 2009; Yuan and Cantley, 2008). The
advantages to tumor cells of activating upstream (receptor)
versus downstream (transducer) signaling remain obscure, as
does the functional impact of crosstalk between the multiple
pathways radiating from growth factor receptors.
Disruptions of Negative-Feedback Mechanisms that
Attenuate Proliferative Signaling
Recent results have highlighted the importance of negative-
feedback loops that normally operate to dampen various types
of signaling and thereby ensure homeostatic regulation of the
flux of signals coursing through the intracellular circuitry (Wertz
and Dixit, 2010; Cabrita and Christofori, 2008; Amit et al.,
2007; Mosesson et al., 2008). Defects in these feedback mech-
anisms are capable of enhancing proliferative signaling. The
prototype of this type of regulation involves the Ras oncoprotein:
the oncogenic effects of Ras do not result from a hyperactivation
of its signaling powers; instead, the oncogenic mutations
affecting ras genes compromise Ras GTPase activity, which
Cell 144, March 4, 2011 ª2011 Elsevier Inc. 647
operates as an intrinsic negative-feedback mechanism that nor-
mally ensures that active signal transmission is transitory.
Analogous negative-feedback mechanisms operate at
multiple nodes within the proliferative signaling circuitry. A prom-
inent example involves the PTEN phosphatase, which counter-
acts PI3-kinase by degrading its product, phosphatidylinositol
(3,4,5) trisphosphate (PIP3). Loss-of-function mutations in PTEN
amplify PI3K signaling and promote tumorigenesis in a variety
of experimental models of cancer; in human tumors, PTEN
expression is often lost by promoter methylation (Jiang and
Liu, 2009; Yuan and Cantley, 2008).
Yet another example involves the mTOR kinase, a coordinator
of cell growth and metabolism that lies both upstream and down-
stream of the PI3K pathway. In the circuitry of some cancer cells,
mTOR activation results, via negative feedback, in the inhibition
of PI3K signaling. Thus, when mTOR is pharmacologically
inhibited in such cancer cells (such as by the drug rapamycin),
the associated loss of negative feedback results in increased
activity of PI3K and its effector Akt/PKB, thereby blunting the
antiproliferative effects of mTOR inhibition (Sudarsanam and
Johnson, 2010; O’Reilly et al., 2006). It is likely that compromised
negative-feedback loops in this and other signaling pathways
will prove to be widespread among human cancer cells and
serve as an important means by which these cells can achieve
proliferative independence. Moreover, disruption of such self-
attenuating signaling may contribute to the development of
adaptive resistance toward drugs targeting mitogenic signaling.
Excessive Proliferative Signaling Can Trigger Cell
Senescence
Early studies of oncogene action encouraged the notion that
ever-increasing expression of such genes and the signals mani-
fested in their protein products would result in correspondingly
increased cancer cell proliferation and thus tumor growth. More
recent research has undermined this notion, in that excessively
elevated signaling by oncoproteins such as RAS, MYC, and
RAF can provoke counteracting responses from cells, specifi-
cally induction of cell senescence and/or apoptosis (Collado
and Serrano, 2010; Evan and d’Adda di Fagagna, 2009; Lowe
et al., 2004). For example, cultured cells expressing high levels
of the Ras oncoprotein may enter into the nonproliferative but
viable state called senescence; in contrast, cells expressing
lower levels of this protein may avoid senescence and proliferate.
Cells with morphological features of senescence, including
enlarged cytoplasm, the absence of proliferation markers, and
expression of the senescence-induced b-galactosidase
enzyme, are abundant in the tissues of mice engineered to over-
express certain oncogenes (Collado and Serrano, 2010; Evan
and d’Adda di Fagagna, 2009) and are prevalent in some cases
of human melanoma (Mooi and Peeper, 2006). These ostensibly
paradoxical responses seem to reflect intrinsic cellular defense
mechanisms designed to eliminate cells experiencing excessive
levels of certain types of signaling. Accordingly, the relative
intensity of oncogenic signaling in cancer cells may represent
compromises between maximal mitogenic stimulation and
avoidance of these antiproliferative defenses. Alternatively,
some cancer cells may adapt to high levels of oncogenic
signaling by disabling their senescence- or apoptosis-inducing
circuitry.
648 Cell 144, March 4, 2011 ª2011 Elsevier Inc.
Evading Growth Suppressors
In addition to the hallmark capability of inducing and sustaining
positively acting growth-stimulatory signals, cancer cells must
also circumvent powerful programs that negatively regulate
cell proliferation; many of these programs depend on the actions
of tumor suppressor genes. Dozens of tumor suppressors that
operate in various ways to limit cell growth and proliferation
have been discovered through their characteristic inactivation
in one or another form of animal or human cancer; many of these
genes have been validated as bona fide tumor suppressors
through gain- or loss-of-function experiments in mice. The two
prototypical tumor suppressors encode the RB (retinoblas-
toma-associated) and TP53 proteins; they operate as central
control nodes within two key complementary cellular regulatory
circuits that govern the decisions of cells to proliferate or, alter-
natively, activate senescence and apoptotic programs.
The RB protein integrates signals from diverse extracellular
and intracellular sources and, in response, decides whether or
not a cell should proceed through its growth-and-division cycle
(Burkhart and Sage, 2008; Deshpande et al., 2005; Sherr and
McCormick, 2002). Cancer cells with defects in RB pathway
function are thus missing the services of a critical gatekeeper
of cell-cycle progression whose absence permits persistent
cell proliferation. Whereas RB transduces growth-inhibitory
signals that originate largely outside of the cell, TP53 receives
inputs from stress and abnormality sensors that function within
the cell’s intracellular operating systems: if the degree of
damage to the genome is excessive, or if the levels of nucleotide
pools, growth-promoting signals, glucose, or oxygenation are
suboptimal, TP53 can call a halt to further cell-cycle progression
until these conditions have been normalized. Alternatively, in the
face of alarm signals indicating overwhelming or irreparable
damage to such cellular subsystems, TP53 can trigger
apoptosis. Notably, the various effects of activated TP53 are
complex and highly context dependent, varying by cell type as
well as by the severity and persistence of conditions of cell stress
and genomic damage.
Although the two canonical suppressors of proliferation—
TP53 and RB—have preeminent importance in regulating cell
proliferation, various lines of evidence indicate that each oper-
ates as part of a larger network that is wired for functional redun-
dancy. For example, chimeric mice populated throughout their
bodies with individual cells lacking a functional Rb gene are
surprisingly free of proliferative abnormalities, despite the expec-
tation that loss of RB function would allow continuous firing of the
cell division cycle in these cells and their lineal descendants;
some of the resulting clusters of Rb null cells should, by all rights,
progress to neoplasia. Instead, the Rb null cells in such chimeric
mice have been found to participate in relatively normal tissue
morphogenesis throughout the body; the only neoplasia
observed was in the development of pituitary tumors late in life
(Lipinski and Jacks, 1999). Similarly, TP53 null mice develop nor-
mally, show largely proper cell and tissue homeostasis, and
again develop abnormalities later in life, in the form of leukemias
and sarcomas (Ghebranious and Donehower, 1998). Both exam-
ples must reflect the operations of redundantly acting mecha-
nisms that serve to constrain inappropriate replication of cells
lacking these key proliferation suppressors.
Mechanisms of Contact Inhibition and Its Evasion
Four decades of research have demonstrated that the cell-to-
cell contacts formed by dense populations of normal cells prop-
agated in two-dimensional culture operate to suppress further
cell proliferation, yielding confluent cell monolayers. Importantly,
such ‘‘contact inhibition’’ is abolished in various types of cancer
cells in culture, suggesting that contact inhibition is an in vitro
surrogate of a mechanism that operates in vivo to ensure normal
tissue homeostasis, one that is abrogated during the course of
tumorigenesis. Until recently, the mechanistic basis for this
mode of growth control remained obscure. Now, however,
mechanisms of contact inhibition are beginning to emerge.
One mechanism involves the product of the NF2 gene, long
implicated as a tumor suppressor because its loss triggers
a form of human neurofibromatosis. Merlin, the cytoplasmic
NF2 gene product, orchestrates contact inhibition via coupling
cell-surface adhesion molecules (e.g., E-cadherin) to transmem-
brane receptor tyrosine kinases (e.g., the EGF receptor). In so
doing, Merlin strengthens the adhesivity of cadherin-mediated
cell-to-cell attachments. Additionally, by sequestering growth
factor receptors, Merlin limits their ability to efficiently emit mito-
genic signals (Curto et al., 2007; Okada et al., 2005).
A second mechanism of contact inhibition involves the LKB1
epithelial polarity protein, which organizes epithelial structure
and helps maintain tissue integrity. LKB1 can, for example,
overrule the mitogenic effects of the powerful Myc oncogene
when the latter is upregulated in organized, quiescent epithelial
structures; in contrast, when LKB1 expression is suppressed,
epithelial integrity is destabilized, and epithelial cells become
susceptible to Myc-induced transformation (Partanen et al.,
2009; Hezel and Bardeesy, 2008). LKB1 has also been identified
as a tumor suppressor gene that is lost in certain human malig-
nancies (Shaw, 2009), possibly reflecting its normal function as
a suppressor of inappropriate proliferation. It remains to be
seen how frequently these two mechanisms of contact-medi-
ated growth suppression are compromised in human cancers;
no doubt yet other contact-induced proliferative barriers are
yet to be discovered. Clearly mechanisms like these that enable
cells to construct and maintain architecturally complex tissues
represent important means of suppressing and counterbalanc-
ing inappropriate proliferative signals.
Corruption of the TGF-b Pathway Promotes Malignancy
TGF-b is best known for its antiproliferative effects, and evasion
by cancer cells of these effects is now appreciated to be far more
elaborate than simple shutdown of its signaling circuitry (Ikush-
ima and Miyazono, 2010; Massagué, 2008; Bierie and Moses,
2006). In many late-stage tumors, TGF-b signaling is redirected
away from suppressing cell proliferation and is found instead
to activate a cellular program, termed the epithelial-to-mesen-
chymal transition (EMT), that confers on cancer cells traits asso-
ciated with high-grade malignancy, as discussed in further detail
below.
Resisting Cell Death
The concept that programmed cell death by apoptosis serves as
a natural barrier to cancer development has been established by
compelling functional studies conducted over the last two
decades (Adams and Cory, 2007; Lowe et al., 2004: Evan and
Littlewood, 1998). Elucidation of the signaling circuitry governing
the apoptotic program has revealed how apoptosis is triggered
in response to various physiologic stresses that cancer cells
experience during the course of tumorigenesis or as a result of
anticancer therapy. Notable among the apoptosis-inducing
stresses are signaling imbalances resulting from elevated levels
of oncogene signaling, as mentioned earlier, and DNA damage
associated with hyperproliferation. Yet other research has re-
vealed how apoptosis is attenuated in those tumors that
succeed in progressing to states of high-grade malignancy and
resistance to therapy (Adams and Cory, 2007; Lowe et al., 2004).
The apoptotic machinery is composed of both upstream regu-
lators and downstream effector components (Adams and Cory,
2007). The regulators, in turn, are divided into two major circuits,
one receiving and processing extracellular death-inducing
signals (the extrinsic apoptotic program, involving for example
the Fas ligand/Fas receptor), and the other sensing and inte-
grating a variety of signals of intracellular origin (the intrinsic
program). Each culminates in activation of a normally latent
protease (caspases 8 and 9, respectively), which proceeds to
initiate a cascade of proteolysis involving effector caspases
responsible for the execution phase of apoptosis, in which the
cell is progressively disassembled and then consumed, both
by its neighbors and by professional phagocytic cells. Currently,
the intrinsic apoptotic program is more widely implicated as
a barrier to cancer pathogenesis.
The ‘‘apoptotic trigger’’ that conveys signals between the regu-
lators and effectors is controlled by counterbalancing pro- and
antiapoptotic members of the Bcl-2 family of regulatory proteins
(Adams and Cory, 2007). The archetype, Bcl-2, along with its
closest relatives (Bcl-xL, Bcl-w, Mcl-1, A1) are inhibitors of
apoptosis, acting in large part by binding to and thereby suppress-
ing two proapoptotic triggering proteins (Bax and Bak); the latter
are embedded in the mitochondrial outer membrane. When
relieved of inhibition by their antiapoptotic relatives, Bax and
Bak disrupt the integrity of the outer mitochondrial membrane,
causing the release of proapoptotic signaling proteins, the most
important of which is cytochrome c. The released cytochrome c
activates, in turn, a cascade of caspases that act via their proteo-
lytic activities to induce the multiple cellular changes associated
with the apoptotic program. Bax and Bak share protein-protein
interaction domains, termed BH3 motifs, with the antiapoptotic
Bcl-2-like proteins that mediate their various physical interac-
tions. The activities of a subfamily of related proteins, each of
which contains a single such BH3 motif, are coupled to a variety
of sensors of cellular abnormality; these ‘‘BH3-only’’ proteins
act either by interfering with antiapoptotic Bcl-2 proteins or by
directly stimulating the proapoptotic members of this family
(Adams and Cory, 2007; Willis and Adams, 2005).
Although the cellular conditions that trigger apoptosis remain
to be fully enumerated, several abnormality sensors that play
key roles in tumor development have been identified (Adams
and Cory, 2007; Lowe et al., 2004). Most notable is a DNA-
damage sensor that functions via the TP53 tumor suppressor
(Junttila and Evan, 2009); TP53 induces apoptosis by upregulat-
ing expression of the Noxa and Puma BH3-only proteins, doing
so in response to substantial levels of DNA breaks and other
chromosomal abnormalities. Alternatively, insufficient survival
Cell 144, March 4, 2011 ª2011 Elsevier Inc. 649
factor signaling (for instance inadequate levels of interleukin-3 in
lymphocytes or of insulin-like growth factor 1/2 [Igf1/2] in epithe-
lial cells) can elicit apoptosis through a BH3-only protein called
Bim. Yet another condition leading to cell death involves hyper-
active signaling by certain oncoproteins, such as Myc, which
triggers apoptosis (in part via Bim and other BH3-only proteins)
unless counterbalanced by antiapoptotic factors (Junttila and
Evan, 2009; Lowe et al., 2004).
Tumor cells evolve a variety of strategies to limit or circumvent
apoptosis. Most common is the loss of TP53 tumor suppressor
function, which eliminates this critical damage sensor from the
apoptosis-inducing circuitry. Alternatively, tumors may achieve
similar ends by increasing expression of antiapoptotic regulators
(Bcl-2, Bcl-xL) or of survival signals (Igf1/2), by downregulating
proapoptotic factors (Bax, Bim, Puma), or by short-circuiting
the extrinsic ligand-induced death pathway. The multiplicity of
apoptosis-avoiding mechanisms presumably reflects the diver-
sity of apoptosis-inducing signals that cancer cell populations
encounter during their evolution to the malignant state.
The structure of the apoptotic machinery and program, and
the strategies used by cancer cells to evade its actions, were
widely appreciated by the beginning of the last decade. The
most notable conceptual advances since then have involved
other forms of cell death that broaden the scope of ‘‘pro-
grammed cell death’’ as a barrier to cancer.
Autophagy Mediates Both Tumor Cell Survival and Death
Autophagy represents an important cell-physiologic response
that, like apoptosis, normally operates at low, basal levels in cells
but can be strongly induced in certain states of cellular stress,
the most obvious of which is nutrient deficiency (Levine and
Kroemer, 2008; Mizushima, 2007). The autophagic program
enables cells to break down cellular organelles, such as ribo-
somes and mitochondria, allowing the resulting catabolites to
be recycled and thus used for biosynthesis and energy metabo-
lism. As part of this program, intracellular vesicles termed auto-
phagosomes envelope intracellular organelles and then fuse with
lysosomes wherein degradation occurs. In this fashion, low-
molecular-weight metabolites are generated that support
survival in the stressed, nutrient-limited environments experi-
enced by many cancer cells.
Like apoptosis, the autophagy machinery has both regulatory
and effector components (Levine and Kroemer, 2008; Mizush-
ima, 2007). Among the latter are proteins that mediate autopha-
gosome formation and delivery to lysosomes. Of note, recent
research has revealed intersections between the regulatory
circuits governing autophagy, apoptosis, and cellular homeo-
stasis. For example, the signaling pathway involving the PI3-
kinase, AKT, and mTOR kinases, which is stimulated by survival
signals to block apoptosis, similarly inhibits autophagy; when
survival signals are insufficient, the PI3K signaling pathway is
downregulated, with the result that autophagy and/or apoptosis
may be induced (Levine and Kroemer, 2008; Sinha and Levine,
2008; Mathew et al., 2007).
Another interconnection between these two programs resides
in the Beclin-1 protein, which has been shown by genetic studies
to be necessary for induction of autophagy (Levine and Kroemer,
2008; Sinha and Levine, 2008; Mizushima, 2007). Beclin-1 is
a member of the BH3-only subfamily of apoptotic regulatory
650 Cell 144, March 4, 2011 ª2011 Elsevier Inc.
proteins, and its BH3 domain allows it to bind the Bcl-2/Bcl-xL
proteins. Stress-sensor-coupled BH3 proteins can displace Be-
clin-1 from its association with Bcl-2/Bcl-xL, enabling the liber-
ated Beclin-1 to trigger autophagy, much as they can release
proapoptotic Bax and Bak to trigger apoptosis. Hence, stress-
transducing BH3 proteins (e.g., Bid, Bad, Puma, et al.) can
induce apoptosis and/or autophagy depending on the physio-
logic state of the cell.
Mice bearing inactivated alleles of the Beclin-1 gene or of
certain other components of the autophagy machinery exhibit
increased susceptibility to cancer (White and DiPaola, 2009:
Levine and Kroemer, 2008). These results suggest that induction
of autophagy can serve as a barrier to tumorigenesis that may
operate independently of or in concert with apoptosis. Accord-
ingly, autophagy appears to represent yet another barrier that
needs to be circumvented during tumor development (White
and DiPaola, 2009).
Perhaps paradoxically, nutrient starvation, radiotherapy, and
certain cytotoxic drugs can induce elevated levels of autophagy
that are apparently cytoprotective for cancer cells, impairing
rather than accentuating the killing actions of these stress-
inducing situations (White and DiPaola, 2009; Apel et al., 2009;
Amaravadi and Thompson, 2007; Mathew et al., 2007). More-
over, severely stressed cancer cells have been shown to shrink
via autophagy to a state of reversible dormancy (White and
DiPaola, 2009; Lu et al., 2008). This survival response may
enable the persistence and eventual regrowth of some late-
stage tumors following treatment with potent anticancer agents.
Thus, in analogy to TGF-b signaling, which can be tumor sup-
pressing at early stages of tumorigenesis and tumor promoting
later on, autophagy seems to have conflicting effects on tumor
cells and thus tumor progression (Apel et al., 2009; White and
DiPaola, 2009). An important agenda for future research will
involve clarifying the genetic and cell-physiologic conditions
that dictate when and how autophagy enables cancer cells to
survive or causes them to die.
Necrosis Has Proinflammatory and Tumor-Promoting
Potential
In contrast to apoptosis, in which a dying cell contracts into an
almost-invisible corpse that is soon consumed by neighbors,
necrotic cells become bloated and explode, releasing their
contents into the local tissue microenvironment. Although
necrosis has historically been viewed much like organismic
death, as a form of system-wide exhaustion and breakdown,
the conceptual landscape is changing: cell death by necrosis
is clearly under genetic control in some circumstances, rather
than being a random and undirected process (Galluzzi and
Kroemer, 2008; Zong and Thompson, 2006).
Perhaps more important, necrotic cell death releases proin-
flammatory signals into the surrounding tissue microenviron-
ment, in contrast to apoptosis and autophagy, which do not.
As a consequence, necrotic cells can recruit inflammatory cells
of the immune system (Grivennikov et al., 2010; White et al.,
2010; Galluzzi and Kroemer, 2008), whose dedicated function
is to survey the extent of tissue damage and remove associated
necrotic debris. In the context of neoplasia, however, multiple
lines of evidence indicate that immune inflammatory cells can
be actively tumor promoting, given that such cells are capable
of fostering angiogenesis, cancer cell proliferation, and invasive-
ness (see below). Additionally, necrotic cells can release bio-
active regulatory factors, such as IL-1a, which can directly stim-
ulate neighboring viable cells to proliferate, with the potential,
once again, to facilitate neoplastic progression (Grivennikov
et al., 2010). Consequently, necrotic cell death, while seemingly
beneficial in counterbalancing cancer-associated hyperprolifer-
ation, may ultimately do more damage than good. Accordingly,
incipient neoplasias and potentially invasive and metastatic
tumors may gain an advantage by tolerating some degree of
necrotic cell death, doing so in order to recruit tumor-promoting
inflammatory cells that bring growth-stimulating factors to the
surviving cells within these growths.
Enabling Replicative Immortality
By 2000, it was widely accepted that cancer cells require unlim-
ited replicative potential in order to generate macroscopic
tumors. This capability stands in marked contrast to the behavior
of the cells in most normal cell lineages in the body, which are
able to pass through only a limited number of successive cell
growth-and-division cycles. This limitation has been associated
with two distinct barriers to proliferation: senescence, a typically
irreversible entrance into a nonproliferative but viable state, and
crisis, which involves cell death. Accordingly, when cells are
propagated in culture, repeated cycles of cell division lead first
to induction of senescence and then, for those cells that succeed
in circumventing this barrier, to a crisis phase, in which the great
majority of cells in the population die. On rare occasion, cells
emerge from a population in crisis and exhibit unlimited replica-
tive potential. This transition has been termed immortalization,
a trait that most established cell lines possess by virtue of their
ability to proliferate in culture without evidence of either senes-
cence or crisis.
Multiple lines of evidence indicate that telomeres protecting
the ends of chromosomes are centrally involved in the capability
for unlimited proliferation (Blasco, 2005; Shay and Wright, 2000).
The telomeres, composed of multiple tandem hexanucleotide
repeats, shorten progressively in nonimmortalized cells propa-
gated in culture, eventually losing the ability to protect the
ends of chromosomal DNAs from end-to-end fusions; such
fusions generate unstable dicentric chromosomes whose reso-
lution results in a scrambling of karyotype that threatens cell
viability. Accordingly, the length of telomeric DNA in a cell
dictates how many successive cell generations its progeny can
pass through before telomeres are largely eroded and have
consequently lost their protective functions, triggering entrance
into crisis.
Telomerase, the specialized DNA polymerase that adds telo-
mere repeat segments to the ends of telomeric DNA, is almost
absent in nonimmortalized cells but expressed at functionally
significant levels in the vast majority (90%) of spontaneously
immortalized cells, including human cancer cells. By extending
telomeric DNA, telomerase is able to counter the progressive
telomere erosion that would otherwise occur in its absence.
The presence of telomerase activity, either in spontaneously
immortalized cells or in the context of cells engineered to
express the enzyme, is correlated with a resistance to induction
of both senescence and crisis/apoptosis; conversely, suppres-
sion of telomerase activity leads to telomere shortening and to
activation of one or the other of these proliferative barriers.
The two barriers to proliferation—senescence and crisis/
apoptosis—have been rationalized as crucial anticancer
defenses that are hard-wired into our cells, being deployed to
impede the outgrowth of clones of preneoplastic and frankly
neoplastic cells. According to this thinking, most incipient
neoplasias exhaust their endowment of replicative doublings
and are stopped in their tracks by one or the other of these
barriers. The eventual immortalization of rare variant cells that
proceed to form tumors has been attributed to their ability to
maintain telomeric DNA at lengths sufficient to avoid triggering
senescence or apoptosis, achieved most commonly by upre-
gulating expression of telomerase or, less frequently, via an
alternative recombination-based telomere maintenance mech-
anism. Hence, telomere shortening has come to be viewed as
a clocking device that determines the limited replicative poten-
tial of normal cells and thus one that must be overcome by
cancer cells.
Reassessing Replicative Senescence
Whereas telomere maintenance has been increasingly substan-
tiated as a condition critical to the neoplastic state, the concept
of replication-induced senescence as a general barrier requires
refinement and reformulation. (Differences in telomere structure
and function in mouse versus human cells have also complicated
investigation of the roles of telomeres and telomerase in replica-
tive senescence.) Recent experiments have revealed that the
induction of senescence in certain cultured cells can be delayed
and possibly eliminated by the use of improved cell culture
conditions, suggesting that recently explanted primary cells
may be able to proliferate unimpeded in culture up the point of
crisis and the associated induction of apoptosis triggered by crit-
ically shortened telomeres (Ince et al., 2007; Passos et al., 2007;
Zhang et al., 2004; Sherr and DePinho, 2000). In contrast, exper-
iments in mice engineered to lack telomerase indicate that the
consequently shortened telomeres can shunt premalignant cells
into a senescent state that contributes (along with apoptosis) to
attenuated tumorigenesis in mice genetically destined to
develop particular forms of cancer (Artandi and DePinho,
2010). Such telomerase null mice with highly eroded telomeres
exhibit multiorgan dysfunction and abnormalities that include
evidence for both senescence and apoptosis, perhaps analo-
gous to the senescence and apoptosis observed in cell culture
(Artandi and DePinho, 2010; Feldser and Greider, 2007).
Of note, and as discussed earlier, a morphologically similar
form of cell senescence induced by excessive or unbalanced
oncogene signaling is now well documented as a protective
mechanism against neoplasia; the possible interconnections of
this form of senescence with telomerase and telomeres remain
to be ascertained. Thus, cell senescence is emerging conceptu-
ally as a protective barrier to neoplastic expansion that can be
triggered by various proliferation-associated abnormalities,
including high levels of oncogenic signaling and, apparently,
subcritical shortening of telomeres.
Delayed Activation of Telomerase May Both Limit
and Foster Neoplastic Progression
There is now evidence that clones of incipient cancer cells often
experience telomere loss-induced crisis relatively early during
Cell 144, March 4, 2011 ª2011 Elsevier Inc. 651
the course of multistep tumor progression due to their inability to
express significant levels of telomerase. Thus, extensively
eroded telomeres have been documented in premalignant
growths through the use of fluorescence in situ hybridization
(FISH), which has also revealed the end-to-end chromosomal
fusions that signal telomere failure and crisis (Kawai et al.,
2007; Hansel et al., 2006). These results also suggest that such
cells have passed through a substantial number of successive
telomere-shortening cell divisions during their evolution from
fully normal cells-of-origin. Accordingly, the development of
some human neoplasias may be aborted by telomere-induced
crisis long before they succeed in becoming macroscopic,
frankly neoplastic growths.
In contrast, the absence of TP53-mediated surveillance of
genomic integrity may permit other incipient neoplasias to
survive initial telomere erosion and attendant chromosomal
breakage-fusion-bridge (BFB) cycles. The genomic alterations
resulting from these BFB cycles, including deletions and ampli-
fications of chromosomal segments, evidently serve to increase
the mutability of the genome, thereby accelerating the acquisi-
tion of mutant oncogenes and tumor suppressor genes. The real-
ization that impaired telomere function can actually foster tumor
progression has come from the study of mutant mice that lack
both p53 and telomerase function (Artandi and DePinho, 2010,
2000). The proposition that these two defects can cooperatively
enhance human tumorigenesis has not yet been directly docu-
mented.
Circumstantial support for the importance of transient telo-
mere deficiency in facilitating malignant progression has come,
in addition, from comparative analyses of premalignant and
malignant lesions in the human breast (Raynaud et al., 2010;
Chin et al., 2004). The premalignant lesions did not express
significant levels of telomerase and were marked by telomere
shortening and nonclonal chromosomal aberrations. In contrast,
overt carcinomas exhibited telomerase expression concordantly
with the reconstruction of longer telomeres and the fixation (via
clonal outgrowth) of the aberrant karyotypes that would seem
to have been acquired after telomere failure but before the acqui-
sition of telomerase activity. When portrayed in this way, the
delayed acquisition of telomerase function serves to generate
tumor-promoting mutations, whereas its subsequent activation
stabilizes the mutant genome and confers the unlimited replica-
tive capacity that cancer cells require in order to generate clini-
cally apparent tumors.
New Functions of Telomerase
Telomerase was discovered because of its ability to elongate
and maintain telomeric DNA, and almost all telomerase research
has been posited on the notion that its functions are confined to
this crucial function. However, in recent years it has become
apparent that telomerase exerts functions that are relevant to
cell proliferation but unrelated to telomere maintenance. The
noncanonical roles of telomerase, and in particular its protein
subunit TERT, have been revealed by functional studies in
mice and cultured cells; in some cases novel functions have
been demonstrated in conditions where the telomerase enzy-
matic activity has been eliminated (Cong and Shay, 2008).
Among the growing list of telomere-independent functions of
TERT/telomerase is the ability of TERT to amplify signaling by
652 Cell 144, March 4, 2011 ª2011 Elsevier Inc.
the Wnt pathway, by serving as a cofactor of the b-catenin/LEF
transcription factor complex (Park et al., 2009). Other ascribed
telomere-independent effects include demonstrable enhance-
ment of cell proliferation and/or resistance to apoptosis (Kang
et al., 2004), involvement in DNA-damage repair (Masutomi
et al., 2005), and RNA-dependent RNA polymerase function
(Maida et al., 2009). Consistent with these broader roles, TERT
can be found associated with chromatin at multiple sites along
the chromosomes, not just at the telomeres (Park et al., 2009;
Masutomi et al., 2005). Hence, telomere maintenance is proving
to be the most prominent of a diverse series of functions to which
TERT contributes. The contributions of these additional func-
tions of telomerase to tumorigenesis remain to be fully eluci-
dated.
Inducing Angiogenesis
Like normal tissues, tumors require sustenance in the form of
nutrients and oxygen as well as an ability to evacuate metabolic
wastes and carbon dioxide. The tumor-associated neovascula-
ture, generated by the process of angiogenesis, addresses these
needs. During embryogenesis, the development of the vascula-
ture involves the birth of new endothelial cells and their assembly
into tubes (vasculogenesis) in addition to the sprouting (angio-
genesis) of new vessels from existing ones. Following this
morphogenesis, the normal vasculature becomes largely quies-
cent. In the adult, as part of physiologic processes such as
wound healing and female reproductive cycling, angiogenesis
is turned on, but only transiently. In contrast, during tumor
progression, an ‘‘angiogenic switch’’ is almost always activated
and remains on, causing normally quiescent vasculature to
continually sprout new vessels that help sustain expanding
neoplastic growths (Hanahan and Folkman, 1996).
A compelling body of evidence indicates that the angiogenic
switch is governed by countervailing factors that either induce
or oppose angiogenesis (Baeriswyl and Christofori, 2009; Berg-
ers and Benjamin, 2003). Some of these angiogenic regulators
are signaling proteins that bind to stimulatory or inhibitory cell-
surface receptors displayed by vascular endothelial cells. The
well-known prototypes of angiogenesis inducers and inhibitors
are vascular endothelial growth factor-A (VEGF-A) and thrombo-
spondin-1 (TSP-1), respectively.
The VEGF-A gene encodes ligands that are involved in orches-
trating new blood vessel growth during embryonic and postnatal
development, and then in homeostatic survival of endothelial
cells, as well as in physiological and pathological situations in
the adult. VEGF signaling via three receptor tyrosine kinases
(VEGFR-1–3) is regulated at multiple levels, reflecting this
complexity of purpose. Thus, VEGF gene expression can by
upregulated both by hypoxia and by oncogene signaling (Fer-
rara, 2009; Mac Gabhann and Popel, 2008; Carmeliet, 2005).
Additionally, VEGF ligands can be sequestered in the extracel-
lular matrix in latent forms that are subject to release and activa-
tion by extracellular matrix-degrading proteases (e.g., MMP-9;
Kessenbrock et al., 2010). In addition, other proangiogenic
signals, such as members of the fibroblast growth factor (FGF)
family, have been implicated in sustaining tumor angiogenesis
when their expression is chronically upregulated (Baeriswyl
and Christofori, 2009). TSP-1, a key counterbalance in the
angiogenic switch, also binds transmembrane receptors dis-
played by endothelial cells and thereby evokes suppressive
signals that can counteract proangiogenic stimuli (Kazerounian
et al., 2008).
The blood vessels produced within tumors by chronically acti-
vated angiogenesis and an unbalanced mix of proangiogenic
signals are typically aberrant: tumor neovasculature is marked
by precocious capillary sprouting, convoluted and excessive
vessel branching, distorted and enlarged vessels, erratic blood
flow, microhemorrhaging, leakiness, and abnormal levels of
endothelial cell proliferation and apoptosis (Nagy et al., 2010;
Baluk et al., 2005).
Angiogenesis is induced surprisingly early during the multi-
stage development of invasive cancers both in animal models
and in humans. Histological analyses of premalignant, noninva-
sive lesions, including dysplasias and in situ carcinomas arising
in a variety of organs, have revealed the early tripping of the
angiogenic switch (Raica et al., 2009; Hanahan and Folkman,
1996). Historically, angiogenesis was envisioned to be important
only when rapidly growing macroscopic tumors had formed, but
more recent data indicate that angiogenesis also contributes to
the microscopic premalignant phase of neoplastic progression,
further cementing its status as an integral hallmark of cancer.
The past decade has witnessed an astonishing outpouring of
research on angiogenesis. Amid this wealth of new knowledge,
we highlight several advances of particular relevance to tumor
physiology.
Gradations of the Angiogenic Switch
Once angiogenesis has been activated, tumors exhibit diverse
patterns of neovascularization. Some tumors, including such
highly aggressive types as pancreatic ductal adenocarcinomas,
are hypovascularized and replete with stromal ‘‘deserts’’ that are
largely avascular and indeed may even be actively antiangio-
genic (Olive et al., 2009). Many other tumors, including human
renal and pancreatic neuroendocrine carcinomas, are highly
angiogenic and consequently densely vascularized (Zee et al.,
2010; Turner et al., 2003).
Collectively, such observations suggest an initial tripping of
the angiogenic switch during tumor development that is followed
by a variable intensity of ongoing neovascularization, the latter
being controlled by a complex biological rheostat that involves
both the cancer cells and the associated stromal microenviron-
ment (Baeriswyl and Christofori, 2009; Bergers and Benjamin,
2003). Of note, the switching mechanism can vary in its form,
even though the net result is a common inductive signal (e.g.,
VEGF). In some tumors, dominant oncogenes operating within
tumor cells, such as Ras and Myc, can upregulate expression
of angiogenic factors, whereas in others, such inductive signals
are produced indirectly by immune inflammatory cells, as dis-
cussed below. The direct induction of angiogenesis by onco-
genes that also drive proliferative signaling illustrates the impor-
tant principle that distinct hallmark capabilities can be
coregulated by the same transforming agents.
Endogenous Angiogenesis Inhibitors Present Natural
Barriers to Tumor Angiogenesis
Research in the 1990s revealed that TSP-1 as well as fragments
of plasmin (angiostatin) and type 18 collagen (endostatin) can
act as endogenous inhibitors of angiogenesis (Ribatti, 2009;
Kazerounian, et al., 2008; Folkman, 2006, 2002; Nyberg et al.,
2005). The last decade has seen reports of another dozen
such agents (Ribatti, 2009; Folkman, 2006; Nyberg et al.,
2005). Most are proteins, and many are derived by proteolytic
cleavage of structural proteins that are not themselves angio-
genic regulators. A number of these endogenous inhibitors of
angiogenesis can be detected in the circulation of normal
mice and humans. The genes encoding several endogenous
angiogenesis inhibitors have been deleted from the mouse
germline without untoward physiological effects; the growth of
autochthonous and implanted tumors, however, is enhanced
as a consequence (Ribatti, 2009; Nyberg et al., 2005). By
contrast, if the circulating levels of an endogenous inhibitor
are genetically increased (e.g., via overexpression in transgenic
mice or in xenotransplanted tumors), tumor growth is impaired
(Ribatti, 2009; Nyberg et al., 2005); interestingly, wound healing
and fat deposition are impaired or accelerated by elevated or
ablated expression of such genes (Cao, 2010; Seppinen et al.,
2008). The data suggest that such endogenous angiogenesis
inhibitors serve under normal circumstances as physiologic
regulators that modulate transitory angiogenesis during tissue
remodeling and wound healing; they may also act as intrinsic
barriers to induction and/or persistence of angiogenesis by
incipient neoplasias.
Pericytes Are Important Components
of the Tumor Neovasculature
Pericytes have long been known as supporting cells that are
closely apposed to the outer surfaces of the endothelial tubes
in normal tissue vasculature, where they provide important
mechanical and physiologic support to the endothelial cells.
Tumor-associated vasculature, in contrast, was portrayed as
lacking appreciable coverage by these auxiliary cells. However,
careful microscopic studies conducted in recent years have re-
vealed that pericytes are associated, albeit loosely, with the neo-
vasculature of most if not all tumors (Raza et al., 2010; Bergers
and Song, 2005). More importantly, mechanistic studies dis-
cussed below have revealed that pericyte coverage is important
for the maintenance of a functional tumor neovasculature.
A Variety of Bone Marrow-Derived Cells Contribute
to Tumor Angiogenesis
It is now clear that a repertoire of cell types originating in the bone
marrow play crucial roles in pathological angiogenesis (Qian and
Pollard, 2010; Zumsteg and Christofori, 2009; Murdoch et al.,
2008; De Palma et al., 2007). These include cells of the innate
immune system—notably macrophages, neutrophils, mast cells,
and myeloid progenitors—that infiltrate premalignant lesions
and progressed tumors and assemble at the margins of such
lesions; the peri-tumoral inflammatory cells help to trip the angio-
genic switch in previously quiescent tissue and to sustain
ongoing angiogenesis associated with tumor growth, in addition
to facilitating local invasion, as noted below. In addition, they can
help protect the vasculature from the effects of drugs targeting
endothelial cell signaling (Ferrara, 2010). Additionally, several
types of bone marrow-derived ‘‘vascular progenitor cells’’ have
been observed in certain cases to have migrated into neoplastic
lesions and become intercalated into the neovasculature as peri-
cytes or endothelial cells (Patenaude et al., 2010; Kovacic and
Boehm, 2009; Lamagna and Bergers, 2006).
Cell 144, March 4, 2011 ª2011 Elsevier Inc. 653
Activating Invasion and Metastasis
In 2000, the mechanisms underlying invasion and metastasis
were largely an enigma. It was clear that as carcinomas arising
from epithelial tissues progressed to higher pathological grades
of malignancy, reflected in local invasion and distant metastasis,
the associated cancer cells typically developed alterations in
their shape as well as in their attachment to other cells and to
the extracellular matrix (ECM). The best characterized alteration
involved the loss by carcinoma cells of E-cadherin, a key cell-to-
cell adhesion molecule. By forming adherens junctions with
adjacent epithelial cells, E-cadherin helps to assemble epithelial
cell sheets and maintain the quiescence of the cells within these
sheets. Increased expression of E-cadherin was well established
as an antagonist of invasion and metastasis, whereas reduction
of its expression was known to potentiate these phenotypes. The
frequently observed downregulation and occasional mutational
inactivation of E-cadherin in human carcinomas provided strong
support for its role as a key suppressor of this hallmark capability
(Berx and van Roy, 2009; Cavallaro and Christofori, 2004).
Additionally, expression of genes encoding other cell-to-cell
and cell-to-ECM adhesion molecules is demonstrably altered
in some highly aggressive carcinomas, with those favoring cyto-
stasis typically being downregulated. Conversely, adhesion
molecules normally associated with the cell migrations that
occur during embryogenesis and inflammation are often upregu-
lated. For example, N-cadherin, which is normally expressed in
migrating neurons and mesenchymal cells during organogen-
esis, is upregulated in many invasive carcinoma cells. Beyond
the gain and loss of such cell-cell/matrix attachment proteins,
the master regulators of invasion and metastasis were largely
unknown or, when suspected, lacking in functional validation
(Cavallaro and Christofori, 2004).
The multistep process of invasion and metastasis has been
schematized as a sequence of discrete steps, often termed the
invasion-metastasis cascade (Talmadge and Fidler, 2010; Fidler,
2003). This depiction envisions a succession of cell-biologic
changes, beginning with local invasion, then intravasation by
cancer cells into nearby blood and lymphatic vessels, transit of
cancer cells through the lymphatic and hematogenous systems,
followed by escape of cancer cells from the lumina of such
vessels into the parenchyma of distant tissues (extravasation),
the formation of small nodules of cancer cells (micrometasta-
ses), and finally the growth of micrometastatic lesions into
macroscopic tumors, this last step being termed ‘‘colonization.’’
Research into the capability for invasion and metastasis has
accelerated dramatically over the past decade as powerful
new research tools and refined experimental models have
become available, and as critical regulatory genes were identi-
fied. While still an emerging field replete with major unanswered
questions, significant progress has been made in delineating
important features of this complex hallmark capability. An admit-
tedly incomplete representation of these advances is highlighted
below.
The EMT Program Broadly Regulates Invasion
and Metastasis
A developmental regulatory program, referred to as the ‘‘epithe-
lial-mesenchymal transition’’ (EMT), has become prominently
implicated as a means by which transformed epithelial cells
654 Cell 144, March 4, 2011 ª2011 Elsevier Inc.
can acquire the abilities to invade, to resist apoptosis, and to
disseminate (Klymkowsky and Savagner, 2009; Polyak and
Weinberg, 2009; Thiery et al., 2009; Yilmaz and Christofori,
2009; Barrallo-Gimeno and Nieto, 2005). By co-opting a process
involved in various steps of embryonic morphogenesis and
wound healing, carcinoma cells can concomitantly acquire
multiple attributes that enable invasion and metastasis. This
multifaceted EMT program can be activated transiently or stably,
and to differing degrees, by carcinoma cells during the course of
invasion and metastasis.
A set of pleiotropically acting transcriptional factors, including
Snail, Slug, Twist, and Zeb1/2, orchestrate the EMT and related
migratory processes during embryogenesis; most were initially
identified by developmental genetics. These transcriptional
regulators are expressed in various combinations in a number
of malignant tumor types and have been shown in experimental
models of carcinoma formation to be causally important for
programming invasion; some have been found to elicit metas-
tasis when ectopically overexpressed (Micalizzi et al., 2010;
Taube et al., 2010; Schmalhofer et al., 2009; Yang and Weinberg,
2008). Included among the cell-biological traits evoked by such
transcription factors are loss of adherens junctions and associ-
ated conversion from a polygonal/epithelial to a spindly/fibro-
blastic morphology, expression of matrix-degrading enzymes,
increased motility, and heightened resistance to apoptosis—all
traits implicated in the processes of invasion and metastasis.
Several of these transcription factors can directly repress E-cad-
herin gene expression, thereby depriving neoplastic epithelial
cells of this key suppressor of motility and invasiveness (Peinado
et al., 2004).
The available evidence suggests that these transcription
factors regulate one another as well as overlapping sets of target
genes. No rules have yet been established to describe their inter-
actions and the conditions that govern their expression.
Evidence from developmental genetics indicates that contextual
signals received from neighboring cells in the embryo are
involved in triggering expression of these transcription factors
in those cells destined to pass through an EMT (Micalizzi et al.,
2010); in an analogous fashion, increasing evidence suggests
that heterotypic interactions of cancer cells with adjacent
tumor-associated stromal cells can induce expression of the
malignant cell phenotypes that are known to be choreographed
by one or more of these transcriptional regulators (Karnoub and
Weinberg, 2006–2007; Brabletz et al., 2001). Moreover, cancer
cells at the invasive margins of certain carcinomas can be
seen to have undergone an EMT, suggesting that these cancer
cells are subject to microenvironmental stimuli distinct from
those received by cancer cells located in the cores of these
lesions (Hlubek et al., 2007).
Although the evidence is still incomplete, it would appear that
EMT-inducing transcription factors are able to orchestrate most
steps of the invasion-metastasis cascade save the final step of
colonization. We still know rather little about the various manifes-
tations and temporal stability of the mesenchymal state
produced by an EMT. Although expression of EMT-inducing
transcription factors has been observed in certain nonepithelial
tumor types, such as sarcomas and neuroectodermal tumors,
their roles in programming malignant traits in these tumors are
presently poorly documented. Additionally, it remains to be
determined whether invasive carcinoma cells necessarily
acquire their capability through activation of parts of the EMT
program, or whether alternative regulatory programs can also
enable this capability.
Heterotypic Contributions of Stromal Cells to Invasion
and Metastasis
It is increasingly apparent that crosstalk between cancer cells
and cells of the neoplastic stroma is involved in the acquired
capability for invasive growth and metastasis (Egeblad et al.,
2010; Qian and Pollard, 2010; Joyce and Pollard, 2009; Kalluri
and Zeisberg, 2006). Such signaling may impinge on carcinoma
cells and act to alter their hallmark capabilities as suggested
above. For example, mesenchymal stem cells (MSCs) present
in the tumor stroma have been found to secrete CCL5/RANTES
in response to signals released by cancer cells; CCL5 then acts
reciprocally on the cancer cells to stimulate invasive behavior
(Karnoub et al., 2007).
Macrophages at the tumor periphery can foster local invasion
by supplying matrix-degrading enzymes such as metalloprotei-
nases and cysteine cathepsin proteases (Kessenbrock et al.,
2010; Joyce and Pollard, 2009; Palermo and Joyce, 2008; Mo-
hamed and Sloane, 2006); in one model system, the invasion-
promoting macrophages are activated by IL-4 produced by the
cancer cells (Gocheva et al., 2010). And in an experimental
model of metastatic breast cancer, tumor-associated macro-
phages (TAMs) supply epidermal growth factor (EGF) to breast
cancer cells, while the cancer cells reciprocally stimulate the
macrophages with CSF-1; their concerted interactions facilitate
intravasation into the circulatory system and metastatic dissem-
ination of the cancer cells (Qian and Pollard, 2010; Wyckoff et al.,
2007).
Observations like these indicate that the phenotypes of high-
grade malignancy do not arise in a strictly cell-autonomous
manner, and that their manifestation cannot be understood
solely through analyses of tumor cell genomes. One important
implication, still untested, is that the ability to negotiate most of
the steps of the invasion-metastasis cascade may be acquired
in certain tumors without the requirement that the associated
cancer cells undergo additional mutations beyond those that
were needed for primary tumor formation.
Plasticity in the Invasive Growth Program
The role of contextual signals in inducing an invasive growth
capability (often via an EMT) implies the possibility of revers-
ibility, in that cancer cells that have disseminated from a primary
tumor to a more distant tissue site may no longer benefit from the
activated stroma and invasion/EMT-inducing signals that they
experienced while residing in the primary tumor; in the absence
of ongoing exposure to these signals, carcinoma cells may revert
in their new homes to a noninvasive state. Thus, carcinoma cells
that have undergone an EMT during initial invasion and meta-
static dissemination may pass through the reverse process,
termed the mesenchymal-epithelial transition (MET). This plas-
ticity may result in the formation of new tumor colonies of carci-
noma cells exhibiting a histopathology similar to those of carci-
noma cells in the primary tumor that never underwent an EMT
(Hugo et al., 2007). Moreover, the notion that cancer cells
routinely pass through a complete EMT program is likely to be
simplistic; instead, in many cases, cancer cells may enter into
an EMT program only partially, thereby acquiring new mesen-
chymal traits while continuing to express residual epithelial traits.
Distinct Forms of Invasion May Underlie Different
Cancer Types
The EMT program regulates a particular type of invasiveness
that has been termed ‘‘mesenchymal.’’ In addition, two other
distinct modes of invasion have been identified and implicated
in cancer cell invasion (Friedl and Wolf, 2008, 2010). ‘‘Collective
invasion’’ involves nodules of cancer cells advancing en masse
into adjacent tissues and is characteristic of, for example,
squamous cell carcinomas; interestingly, such cancers are
rarely metastatic, suggesting that this form of invasion lacks
certain functional attributes that facilitate metastasis. Less clear
is the prevalence of an ‘‘amoeboid’’ form of invasion (Madsen
and Sahai, 2010; Sabeh et al., 2009), in which individual cancer
cells show morphological plasticity, enabling them to slither
through existing interstices in the extracellular matrix rather
than clearing a path for themselves, as occurs in both the mesen-
chymal and collective forms of invasion. It is presently unre-
solved whether cancer cells participating in the collective and
amoeboid forms of invasion employ components of the EMT
program, or whether entirely different cell-biological programs
are responsible for choreographing these alternative invasion
programs.
Another emerging concept, noted above, involves the facilita-
tion of cancer cell invasion by inflammatory cells that assemble
at the boundaries of tumors, producing the extracellular
matrix-degrading enzymes and other factors that enable inva-
sive growth (Kessenbrock et al., 2010; Qian and Pollard, 2010;
Joyce and Pollard, 2009); these functions may obviate the
need of cancer cells to produce these proteins through activa-
tion of EMT programs. Thus, cancer cells may secrete the
chemoattractants that recruit the proinvasive inflammatory cells
rather than producing the matrix-degrading enzymes them-
selves.
The Daunting Complexity of Metastatic Colonization
Metastasis can be broken down into two major phases: the
physical dissemination of cancer cells from the primary tumor
to distant tissues, and the adaptation of these cells to foreign
tissue microenvironments that results in successful colonization,
i.e., the growth of micrometastases into macroscopic tumors.
The multiple steps of dissemination would seem to be in the
purview of the EMT and similarly acting migratory programs.
Colonization, however, is not strictly coupled with physical
dissemination, as evidenced by the presence in many patients
of myriad micrometastases that have successfully disseminated
but never progress to macroscopic metastatic tumors (Tal-
madge and Fidler, 2010; McGowan et al., 2009; Aguirre-Ghiso,
2007; Townson and Chambers, 2006; Fidler, 2003).
In some types of cancer, the primary tumor may release
systemic suppressor factors that render such micrometastases
dormant, as revealed clinically by explosive metastatic growth
soon after resection of the primary growth (Demicheli et al.,
2008; Folkman, 2002). In others, however, such as breast cancer
and melanoma, macroscopic metastases may erupt decades
after a primary tumor has been surgically removed or pharmaco-
logically destroyed; these metastatic tumor growths evidently
Cell 144, March 4, 2011 ª2011 Elsevier Inc. 655
reflect dormant micrometastases that have solved, after much trial
and error, the complex problem of tissue colonization (Barkan,
et al., 2010; Aguirre-Ghiso, 2007; Townson and Chambers, 2006).
One can infer from such natural histories that micrometasta-
ses may lack other hallmark capabilities necessary for vigorous
growth, such as the ability to activate angiogenesis; indeed the
inability of certain experimentally generated dormant microme-
tastases to form macroscopic tumors has been ascribed to their
failure to activate tumor angiogenesis (Naumov et al., 2008;
Aguirre-Ghiso, 2007). Additionally, recent experiments have
shown that nutrient starvation can induce intense autophagy
that causes cancer cells to shrink and adopt a state of reversible
dormancy; such cells may exit this state and resume active
growth and proliferation when changes in tissue microenviron-
ment, such as access to more nutrients, permit (Kenific et al.,
2010; Lu et al., 2008). Other mechanisms of micrometastatic
dormancy may involve anti-growth signals embedded in normal
tissue extracellular matrix (Barkan et al., 2010) and tumor-sup-
pressing actions of the immune system (Teng et al., 2008;
Aguirre-Ghiso, 2007).
Most disseminated cancer cells are likely to be poorly adap-
ted, at least initially, to the microenvironment of the tissue in
which they have landed. Accordingly, each type of disseminated
cancer cell may need to develop its own set of ad hoc solutions
to the problem of thriving in the microenvironment of one or
another foreign tissue (Gupta et al., 2005). These adaptations
might require hundreds of distinct colonization programs, each
dictated by the type of disseminating cancer cell and the nature
of the tissue microenvironment in which colonization is
proceeding. As further discussed below, however, certain tissue
microenviroments may be preordained to be intrinsically hospi-
table to disseminated cancer cells (Peinado et al., 2011;
Talmadge and Fidler, 2010).
Metastatic dissemination has long been depicted as the last
step in multistep primary tumor progression, and indeed for
many tumors that is likely the case, as illustrated by recent
genome sequencing studies that present genetic evidence for
clonal evolution of pancreatic ductal adenocarcinoma to metas-
tasis (Campbell et al., 2010; Luebeck, 2010; Yachida et al.,
2010). On the other hand, evidence has recently emerged
indicating that cells can disseminate remarkably early,
dispersing from ostensibly noninvasive premalignant lesions in
both mice and humans (Coghlin and Murray, 2010; Klein,
2009). Additionally, micrometastases can be spawned from
primary tumors that are not obviously invasive but possess
a neovasculature lacking in lumenal integrity (Gerhardt and
Semb, 2008). Although cancer cells can clearly disseminate
from such pre-neoplastic lesions and seed the bone marrow
and other tissues, their capability to colonize these sites and
develop into pathologically significant macrometastases
remains unproven. At present, we view this early metastatic
dissemination as a demonstrable phenomenon in mice and hu-
mans whose clinical significance is yet to be established.
Beyond the timing of their dissemination, it also remains
unclear when and where cancer cells develop the ability to colo-
nize foreign tissues as macroscopic tumors. This capability may
arise during primary tumor formation as a result of a tumor’s
particular developmental path prior to any dissemination, such
656 Cell 144, March 4, 2011 ª2011 Elsevier Inc.
that primary tumor cells entering the circulation are fortuitously
endowed with the ability to colonize certain distant tissue sites
(Talmadge and Fidler, 2010). Alternatively, the ability to colonize
specific tissues may only develop in response to the selective
pressure on already disseminated cancer cells to adapt to
growth in foreign tissue microenvironments.
Having developed such tissue-specific colonizing ability, the
cells in metastatic colonies may proceed to disseminate further,
not only to new sites in the body but also back to the primary
tumors in which their ancestors arose. Accordingly, tissue-
specific colonization programs that are evident among cells
within a primary tumor may originate not from classical tumor
progression occurring within the primary lesion but instead
from emigrants that have returned home (Kim et al., 2009).
Such reseeding is consistent with the aforementioned studies
of human pancreatic cancer metastasis (Campbell et al.,
2010; Luebeck, 2010; Yachida et al., 2010). Stated differently,
the phenotypes and underlying gene expression programs of
the populations of cancer cells (and of the cancer stem cells
discussed below) within primary tumors may be significantly
modified by reverse migration of their distant metastatic
progeny.
Implicit in this self-seeding process is another notion: the
supportive stroma that arises in a primary tumor and contributes
to its acquisition of malignant traits may intrinsically provide
a hospitable site for reseeding and colonization by circulating
cancer cells emanating from metastatic lesions.
Clarifying the regulatory programs that enable metastatic
colonization represents an important agenda for future research.
Substantial progress is being made, for example, in defining sets
of genes (‘‘metastatic signatures’’) that correlate with and appear
to facilitate the establishment of macroscopic metastases in
specific tissues (Coghlin and Murray, 2010; Bos et al., 2009;
Olson et al., 2009; Nguyen et al., 2009; Gupta et al., 2005). The
challenge is considerable, given the apparent multitude of
distinct colonization programs cited above. Moreover, coloniza-
tion is unlikely to depend exclusively on cell-autonomous
processes. Instead, it almost certainly requires the establish-
ment of a permissive tumor microenvironment composed of
critical stromal support cells. For these reasons, the process
of colonization is likely to encompass a large number of cell-
biological programs that are, in aggregate, considerably more
complex and diverse than the preceding steps of metastatic
dissemination.
Programming of Hallmark Capabilities
by Intracellular Circuitry
In 2000, we presented a metaphor, in which the numerous
signaling molecules affecting cancer cells operate as nodes
and branches of elaborate integrated circuits that are reprog-
rammed derivatives of the circuits operating in normal cells.
The ensuing decade has both solidified the original depiction
of these circuits and expanded the catalog of signals and the
interconnections of their signaling pathways. It is difficult if not
impossible to graphically portray this circuit comprehensively
and coherently, as was already the case in 2000.
We now suggest a portrayal of this circuitry that is aligned with
individual hallmarks of cancer. Thus, the intracellular integrated
Figure 2. Intracellular Signaling Networks Regulate the Operations of the Cancer Cell
An elaborate integrated circuit operates within normal cells and is reprogrammed to regulate hallmark capabilities within cancer cells. Separate subcircuits,
depicted here in differently colored fields, are specialized to orchestrate the various capabilities. At one level, this depiction is simplistic, as there is considerable
crosstalk between such subcircuits. In addition, because each cancer cell is exposed to a complex mixture of signals from its microenvironment, each of these
subcircuits is connected with signals originating from other cells in the tumor microenvironment, as outlined in Figure 5.

circuit can be segmented into distinct subcircuits, each of which
is specialized to support a discrete cell-biological property in
normal cells and is reprogrammed in order to implement
a hallmark capability in cancer cells (Figure 2). Only a subset of
hallmark capabilities are addressed in this figure, either because
their underlying control circuits remain poorly understood or
because they overlap extensively with those portrayed here.
An additional dimension of complexity involves considerable
interconnections and thus crosstalk between the individual sub-
circuits. For example, certain oncogenic events can affect
multiple capabilities, as illustrated by the diverse effects that
prominent oncogenes, such as mutant RAS and upregulated
MYC, have on multiple hallmark capabilities (e.g., proliferative
signaling, energy metabolism, angiogenesis, invasion, and
survival). We anticipate that future renditions of this integrated
circuit will encompass subcircuits and associated hallmark
capabilities that are still not addressed here.
ENABLING CHARACTERISTICS AND EMERGING
HALLMARKS
We have defined the hallmarks of cancer as acquired functional
capabilities that allow cancer cells to survive, proliferate, and
disseminate; these functions are acquired in different tumor
types via distinct mechanisms and at various times during the
course of multistep tumorigenesis. Their acquisition is made
possible by two enabling characteristics. Most prominent is the
development of genomic instability in cancer cells, which
generates random mutations including chromosomal rearrange-
ments; among these are the rare genetic changes that can
orchestrate hallmark capabilities. A second enabling character-
istic involves the inflammatory state of premalignant and frankly
malignant lesions that is driven by cells of the immune system,
some of which serve to promote tumor progression through
various means.

Cell 144, March 4, 2011 ª2011 Elsevier Inc. 657


Yet other distinct attributes of cancer cells have been
proposed to be functionally important for the development of
cancer and might therefore be added to the list of core hallmarks
(Negrini et al., 2010; Luo et al., 2009; Colotta et al., 2009). Two
such attributes are particularly compelling. The first involves
major reprogramming of cellular energy metabolism in order to
support continuous cell growth and proliferation, replacing the
metabolic program that operates in most normal tissues and
fuels the physiological operations of the associated cells. The
second involves active evasion by cancer cells from attack and
elimination by immune cells; this capability highlights the dichot-
omous roles of an immune system that both antagonizes and
enhances tumor development and progression. Both of these
capabilities may well prove to facilitate the development and
progression of many forms of human cancer and therefore can
be considered to be emerging hallmarks of cancer. These
enabling characteristics and emerging hallmarks, depicted in
Figure 3, are discussed individually below.
An Enabling Characteristic: Genome Instability
and Mutation
Acquisition of the multiple hallmarks enumerated above depends
in large part on a succession of alterations in the genomes of
neoplastic cells. Simply depicted, certain mutant genotypes
confer selective advantage on subclones of cells, enabling their
outgrowth and eventual dominance in a local tissue environment.
Accordingly, multistep tumor progression can be portrayed as
a succession of clonal expansions, each of which is triggered
by the chance acquisition of an enabling mutant genotype.
Because heritable phenotypes, e.g., inactivation of tumor
suppressor genes, can also be acquired through epigenetic
mechanisms such as DNA methylation and histone modifications
(Berdasco and Esteller, 2010; Esteller, 2007; Jones and Baylin,
2007), some clonal expansions may well be triggered by nonmu-
tational changes affecting the regulation of gene expression.
658 Cell 144, March 4, 2011 ª2011 Elsevier Inc.
Figure 3. Emerging Hallmarks and Enabling
Characteristics
An increasing body of research suggests that two
additional hallmarks of cancer are involved in the
pathogenesis of some and perhaps all cancers.
One involves the capability to modify, or repro-
gram, cellular metabolism in order to most effec-
tively support neoplastic proliferation. The second
allows cancer cells to evade immunological
destruction, in particular by T and B lymphocytes,
macrophages, and natural killer cells. Because
neither capability is yet generalized and fully vali-
dated, they are labeled as emerging hallmarks.
Additionally, two consequential characteristics of
neoplasia facilitate acquisition of both core and
emerging hallmarks. Genomic instability and thus
mutability endow cancer cells with genetic alter-
ations that drive tumor progression. Inflammation
by innate immune cells designed to fight infections
and heal wounds can instead result in their inad-
vertent support of multiple hallmark capabilities,
thereby manifesting the now widely appreciated
tumor-promoting consequences of inflammatory
responses.
The extraordinary ability of genome maintenance systems to
detect and resolve defects in the DNA ensures that rates of
spontaneous mutation are usually very low during each cell
generation. In the course of acquiring the roster of mutant genes
needed to orchestrate tumorigenesis, cancer cells often
increase the rates of mutation (Negrini et al., 2010; Salk et al.,
2010). This mutability is achieved through increased sensitivity
to mutagenic agents, through a breakdown in one or several
components of the genomic maintenance machinery, or both.
In addition, the accumulation of mutations can be accelerated
by compromising the surveillance systems that normally monitor
genomic integrity and force genetically damaged cells into either
senescence or apoptosis (Jackson and Bartek, 2009; Kastan,
2008; Sigal and Rotter, 2000). The role of TP53 is central here,
leading to its being called the ‘‘guardian of the genome’’ (Lane,
1992).
A diverse array of defects affecting various components of the
DNA-maintenance machinery—often referred to as the ‘‘care-
takers’’ of the genome (Kinzler and Vogelstein, 1997)—have
been documented. The catalog of defects in these caretaker
genes includes those whose products are involved in (1) detect-
ing DNA damage and activating the repair machinery, (2) directly
repairing damaged DNA, and (3) inactivating or intercepting
mutagenic molecules before they have damaged the DNA
(Negrini et al., 2010; Ciccia and Elledge, 2010; Jackson and
Bartek, 2009; Kastan, 2008; Harper and Elledge, 2007; Friedberg
et al., 2006). From a genetic perspective, these caretaker genes
behave much like tumor suppressor genes, in that their functions
can be lost during the course of tumor progression, with such
losses being achieved either through inactivating mutations or
via epigenetic repression. Mutant copies of many of these care-
taker genes have been introduced into the mouse germline and
result, predictably, in increased cancer incidence, supporting
their potential involvement in human cancer development
(Barnes and Lindahl, 2004).
 In the decade since we first enumerated the cancer hallmarks,
another major source of tumor-associated genomic instability
has been uncovered: as described earlier, the loss of telomeric
DNA in many tumors generates karyotypic instability and associ-
ated amplification and deletion of chromosomal segments
(Artandi and DePinho, 2010). When viewed in this light, telome-
rase is more than an enabler of the hallmark capability for
unlimited replicative potential and must also be added to the
list of critical caretakers responsible for maintaining genome
integrity.
Advances in the molecular-genetic analysis of cancer cell
genomes have provided the most compelling demonstrations
of function-altering mutations and of ongoing genomic instability
during tumor progression. One type of analysis—comparative
genomic hybridization (CGH)—documents the gains and losses
of gene copy number across the cell genome; in many tumors,
the pervasive genomic aberrations revealed by CGH provide
clear evidence for loss of control of genome integrity. Impor-
tantly, the recurrence of specific aberrations (both amplifications
and deletions) at particular sites in the genome indicates that
such sites are likely to harbor genes whose alteration favors
neoplastic progression (Korkola and Gray, 2010).
More recently, with the advent of efficient and economical
DNA-sequencing technologies, higher-resolution analyses
have become possible. Early studies are revealing distinctive
patterns of DNA mutations in different tumor types (see http://
cancergenome.nih.gov/). In the not-too-distant future, the
sequencing of entire cancer cell genomes promises to clarify
the prevalence of ostensibly random mutations scattered across
cancer cell genomes. Thus, recurring genetic alterations may
point to a causal role of particular mutations in tumor pathogen-
esis.
Although the specifics of genome alteration vary dramatically
between different tumor types, the large number of genome
maintenance and repair defects that have already been docu-
mented in human tumors, together with abundant evidence of
widespread destabilization of gene copy number and nucleotide
sequence, persuade us that instability of the genome is inherent
to the great majority of human cancer cells. This leads, in turn, to
the conclusion that the defects in genome maintenance and
repair are selectively advantageous and therefore instrumental
for tumor progression, if only because they accelerate the rate
at which evolving premalignant cells can accumulate favorable
genotypes. As such, genome instability is clearly an enabling
characteristic that is causally associated with the acquisition of
hallmark capabilities.
An Enabling Characteristic: Tumor-Promoting
Inflammation
Pathologists have long recognized that some tumors are densely
infiltrated by cells of both the innate and adaptive arms of the
immune system and thereby mirror inflammatory conditions
arising in non-neoplastic tissues (Dvorak, 1986). With the advent
of better markers for accurately identifying the distinct cell types
of the immune system, it is now clear that virtually every
neoplastic lesion contains immune cells present at densities
ranging from subtle infiltrations detectable only with cell type-
specific antibodies to gross inflammations that are apparent
even by standard histochemical staining techniques (Pagès
et al., 2010). Historically, such immune responses were largely
thought to reflect an attempt by the immune system to eradicate
tumors, and indeed, there is increasing evidence for antitumoral
responses to many tumor types with an attendant pressure on
the tumor to evade immune destruction, as discussed below.
By 2000, there were already clues that the tumor-associated
inflammatory response had the unanticipated, paradoxical effect
of enhancing tumorigenesis and progression, in effect helping
incipient neoplasias to acquire hallmark capabilities. In the
ensuing decade, research on the intersections between inflam-
mation and cancer pathogenesis has blossomed, producing
abundant and compelling demonstrations of the functionally
important tumor-promoting effects that immune cells—largely
of the innate immune system—have on neoplastic progression
(DeNardo et al., 2010; Grivennikov et al., 2010; Qian and Pollard,
2010; Colotta et al., 2009). Inflammation can contribute to
multiple hallmark capabilities by supplying bioactive molecules
to the tumor microenvironment, including growth factors that
sustain proliferative signaling, survival factors that limit cell
death, proangiogenic factors, extracellular matrix-modifying
enzymes that facilitate angiogenesis, invasion, and metastasis,
and inductive signals that lead to activation of EMT and other
hallmark-facilitating programs (DeNardo et al., 2010;
Grivennikov et al., 2010; Qian and Pollard, 2010; Karnoub and
Weinberg, 2006–2007).
Importantly, inflammation is in some cases evident at the
earliest stages of neoplastic progression and is demonstrably
capable of fostering the development of incipient neoplasias
into full-blown cancers (Qian and Pollard, 2010; de Visser et al.,
2006). Additionally, inflammatory cells can release chemicals,
notably reactive oxygen species, that are actively mutagenic for
nearby cancer cells, accelerating their genetic evolution toward
states of heightened malignancy (Grivennikov et al., 2010). As
such, inflammation can be considered an enabling characteristic
for its contributions to the acquisition of core hallmark capabil-
ities. The cells responsible for this enabling characteristic are
described in the section below on the tumor microenvironment.
An Emerging Hallmark: Reprogramming Energy
Metabolism
The chronic and often uncontrolled cell proliferation that repre-
sents the essence of neoplastic disease involves not only
deregulated control of cell proliferation but also corresponding
adjustments of energy metabolism in order to fuel cell growth
and division. Under aerobic conditions, normal cells process
glucose, first to pyruvate via glycolysis in the cytosol and there-
after to carbon dioxide in the mitochondria; under anaerobic
conditions, glycolysis is favored and relatively little pyruvate is
dispatched to the oxygen-consuming mitochondria. Otto
Warburg first observed an anomalous characteristic of cancer
cell energy metabolism (Warburg, 1930, 1956a, 1956b): even in
the presence of oxygen, cancer cells can reprogram their
glucose metabolism, and thus their energy production, by
limiting their energy metabolism largely to glycolysis, leading to
a state that has been termed ‘‘aerobic glycolysis.’’
The existence of this metabolic switch in cancer cells has been
substantiated in the ensuing decades. Such reprogramming of
Cell 144, March 4, 2011 ª2011 Elsevier Inc. 659
energy metabolism is seemingly counterintuitive, in that cancer
cells must compensate for the 18-fold lower efficiency of
ATP production afforded by glycolysis relative to mitochondrial
oxidative phosphorylation. They do so in part by upregulating
glucose transporters, notably GLUT1, which substantially
increases glucose import into the cytoplasm (Jones and Thomp-
son, 2009; DeBerardinis et al., 2008; Hsu and Sabatini, 2008).
Indeed, markedly increased uptake and utilization of glucose
have been documented in many human tumor types, most
readily by noninvasively visualizing glucose uptake using posi-
tron emission tomography (PET) with a radiolabeled analog of
glucose (18F-fluorodeoxyglucose, FDG) as a reporter.
Glycolytic fueling has been shown to be associated with
activated oncogenes (e.g., RAS, MYC) and mutant tumor
suppressors (e.g., TP53) (DeBerardinis et al., 2008; Jones and
Thompson, 2009), whose alterations in tumor cells have been
selected primarily for their benefits in conferring the hallmark
capabilities of cell proliferation, avoidance of cytostatic controls,
and attenuation of apoptosis. This reliance on glycolysis can be
further accentuated under the hypoxic conditions that operate
within many tumors: the hypoxia response system acts pleio-
tropically to upregulate glucose transporters and multiple
enzymes of the glycolytic pathway (Semenza, 2010a; Jones
and Thompson, 2009; DeBerardinis et al., 2008). Thus, both
the Ras oncoprotein and hypoxia can independently increase
the levels of the HIF1a and HIF2a transcription factors, which
in turn upregulate glycolysis (Semenza, 2010a, 2010b; Kroemer
and Pouyssegur, 2008).
A functional rationale for the glycolytic switch in cancer cells
has been elusive, given the relatively poor efficiency of gener-
ating ATP by glycolysis relative to mitochondrial oxidative phos-
phorylation. According to one long-forgotten (Potter, 1958) and
recently revived and refined hypothesis (Vander Heiden et al.,
2009), increased glycolysis allows the diversion of glycolytic
intermediates into various biosynthetic pathways, including
those generating nucleosides and amino acids; this facilitates,
in turn, the biosynthesis of the macromolecules and organelles
required for assembling new cells. Moreover, Warburg-like
metabolism seems to be present in many rapidly dividing em-
bryonic tissues, once again suggesting a role in supporting the
large-scale biosynthetic programs that are required for active
cell proliferation.
Interestingly, some tumors have been found to contain two
subpopulations of cancer cells that differ in their energy-gener-
ating pathways. One subpopulation consists of glucose-depen-
dent (‘‘Warburg-effect’’) cells that secrete lactate, whereas cells
of the second subpopulation preferentially import and utilize the
lactate produced by their neighbors as their main energy source,
employing part of the citric acid cycle to do so (Kennedy and Dew-
hirst, 2010; Feron, 2009; Semenza, 2008). These two populations
evidently function symbiotically: the hypoxic cancer cells depend
on glucose for fuel and secrete lactate as waste, which is im-
ported and preferentially used as fuel by their better-oxygenated
brethren. Although this provocative mode of intratumoral symbi-
osis has yet to be generalized, the cooperation between lactate-
secreting and lactate-utilizing cells to fuel tumor growth is in fact
not an invention of tumors but rather again reflects cooption of
a normal physiological mechanism, in this case one operating in
660 Cell 144, March 4, 2011 ª2011 Elsevier Inc.
muscle (Kennedy and Dewhirst, 2010; Feron, 2009; Semenza,
2008). Additionally, it is becoming apparent that oxygenation,
ranging from normoxia to hypoxia, is not necessarily static in
tumors but instead fluctuates temporally and regionally (Hardee
et al., 2009), likely as a result of the instability and chaotic organi-
zation of the tumor-associated neovasculature.
Altered energy metabolism is proving to be as widespread in
cancer cells as many of the other cancer-associated traits that
have been accepted as hallmarks of cancer. This realization
raises the question of whether deregulating cellular energy
metabolism is therefore a core hallmark capability of cancer cells
that is as fundamental as the six well-established core hallmarks.
In fact, the redirection of energy metabolism is largely orches-
trated by proteins that are involved in one way or another in
programming the core hallmarks of cancer. When viewed in
this way, aerobic glycolysis is simply another phenotype that is
programmed by proliferation-inducing oncogenes.
Interestingly, activating (gain-of-function) mutations in the iso-
citrate dehydrogenase 1/2 (IDH) enzymes have been reported in
glioma and other human tumors (Yen et al., 2010). Although
these mutations may prove to have been clonally selected for
their ability to alter energy metabolism, there is confounding
data associating their activity with elevated oxidation and
stability of the HIF-1 transcription factors (Reitman and Yan,
2010), which could in turn affect genome stability and angiogen-
esis/invasion, respectively, thus blurring the lines of phenotypic
demarcation. Currently, therefore, the designation of reprog-
rammed energy metabolism as an emerging hallmark seems
most appropriate, to highlight both its evident importance as
well as the unresolved issues surrounding its functional indepen-
dence from the core hallmarks.
An Emerging Hallmark: Evading Immune Destruction
A second, still-unresolved issue surrounding tumor formation
involves the role that the immune system plays in resisting or
eradicating formation and progression of incipient neoplasias,
late-stage tumors, and micrometastases. The long-standing
theory of immune surveillance proposes that cells and tissues
are constantly monitored by an ever-alert immune system, and
that such immune surveillance is responsible for recognizing
and eliminating the vast majority of incipient cancer cells
and thus nascent tumors. According to this logic, solid tumors
that do appear have somehow managed to avoid detection
by the various arms of the immune system or have been able
to limit the extent of immunological killing, thereby evading
eradication.
The role of defective immunological monitoring of tumors
would seem to be validated by the striking increases of certain
cancers in immunocompromised individuals (Vajdic and van
Leeuwen, 2009). However, the great majority of these are
virus-induced cancers, suggesting that much of the control of
this class of cancers normally depends on reducing viral burden
in infected individuals, in part through eliminating virus-infected
cells. These observations, therefore, seem to shed little light
on the possible role of the immune system in limiting formation
of the >80% of tumors of nonviral etiology. In recent years,
however, an increasing body of evidence, both from genetically
engineered mice and from clinical epidemiology, suggests that
the immune system operates as a significant barrier to tumor
formation and progression, at least in some forms of non-virus-
induced cancer.
When mice genetically engineered to be deficient for various
components of the immune system were assessed for the devel-
opment of carcinogen-induced tumors, it was observed that
tumors arose more frequently and/or grew more rapidly in the
immunodeficient mice relative to immunocompetent controls.
In particular, deficiencies in the development or function of
CD8+ cytotoxic T lymphocytes (CTLs), CD4+ Th1 helper T cells,
or natural killer (NK) cells each led to demonstrable increases
in tumor incidence; moreover, mice with combined immunodefi-
ciencies in both T cells and NK cells were even more susceptible
to cancer development. The results indicated that, at least in
certain experimental models, both the innate and adaptive
cellular arms of the immune system are able to contribute signif-
icantly to immune surveillance and thus tumor eradication (Teng
et al., 2008; Kim et al., 2007).
In addition, transplantation experiments have shown that
cancer cells that originally arose in immunodeficient mice are
often inefficient at initiating secondary tumors in syngeneic
immunocompetent hosts, whereas cancer cells from tumors
arising in immunocompetent mice are equally efficient at initi-
ating transplanted tumors in both types of hosts (Teng et al.,
2008; Kim et al., 2007). Such behavior has been interpreted as
follows: Highly immunogenic cancer cell clones are routinely
eliminated in immunocompetent hosts—a process that has
been referred to as ‘‘immunoediting’’—leaving behind only
weakly immunogenic variants to grow and generate solid
tumors; such weakly immunogenic cells can thereafter colonize
both immunodeficient and immunocompetent hosts. Con-
versely, when arising in immunodeficient hosts, the immuno-
genic cancer cells are not selectively depleted and can, instead,
prosper along with their weakly immunogenic counterparts.
When cells from such nonedited tumors are serially transplanted
into syngeneic recipients, the immunogenic cancer cells are
rejected when they confront, for the first time, the competent
immune systems of their secondary hosts (Smyth et al., 2006).
(Unanswered in these particular experiments is the question of
whether the chemical carcinogens used to induce such tumors
are prone to generate cancer cells that are especially immuno-
genic.)
Clinical epidemiology also increasingly supports the existence
of antitumoral immune responses in some forms of human
cancer (Bindea et al., 2010; Ferrone and Dranoff, 2010; Nelson,
2008). For example, patients with colon and ovarian tumors
that are heavily infiltrated with CTLs and NK cells have a better
prognosis than those that lack such abundant killer lymphocytes
(Pagès et al., 2010; Nelson, 2008); the case for other cancers is
suggestive but less compelling and is the subject of ongoing
investigation. Additionally, some immunosuppressed organ
transplant recipients have been observed to develop donor-
derived cancers, suggesting that in the ostensibly tumor-free
donors, the cancer cells were held in check, in a dormant state,
by a fully functional immune system (Strauss and Thomas, 2010).
Still, the epidemiology of chronically immunosuppressed
patients does not indicate significantly increased incidences of
the major forms of nonviral human cancer, as noted above.
This might be taken as an argument against the importance of
immune surveillance as an effective barrier to tumorigenesis
and tumor progression. We note, however, that HIV and pharma-
cologically immunosuppressed patients are predominantly
immunodeficient in the T and B cell compartments and thus do
not present with the multicomponent immunological deficiencies
that have been produced in the genetically engineered mutant
mice lacking both NK cells and CTLs; this leaves open the possi-
bility that such patients still have residual capability for an immu-
nological defense against cancer that is mounted by NK and
other innate immune cells.
In truth, the above discussions of cancer immunology simplify
tumor-host immunological interactions, as highly immunogenic
cancer cells may well evade immune destruction by disabling
components of the immune system that have been dispatched
to eliminate them. For example, cancer cells may paralyze infil-
trating CTLs and NK cells, by secreting TGF-b or other immuno-
suppressive factors (Yang et al., 2010; Shields et al., 2010). More
subtle mechanisms operate through the recruitment of inflam-
matory cells that are actively immunosuppressive, including
regulatory T cells (Tregs) and myeloid-derived suppressor cells
(MDSCs). Both can suppress the actions of cytotoxic lympho-
cytes (Mougiakakos et al., 2010; Ostrand-Rosenberg and Sinha,
2009).
In light of these considerations and the still-rudimentary
demonstrations of antitumor immunity as a significant barrier
to tumor formation and progression in humans, we present
immunoevasion as another emerging hallmark, whose gener-
ality as a core hallmark capability remains to be firmly estab-
lished.
THE TUMOR MICROENVIRONMENT
Over the past decade, tumors have increasingly been recog-
nized as organs whose complexity approaches and may even
exceed that of normal healthy tissues. When viewed from this
perspective, the biology of a tumor can only be understood
by studying the individual specialized cell types within it
(Figure 4, upper) as well as the ‘‘tumor microenvironment’’
that they construct during the course of multistep tumorigenesis
(Figure 4, lower). This depiction contrasts starkly with the
earlier, reductionist view of a tumor as nothing more than
a collection of relatively homogeneous cancer cells, whose
entire biology could be understood by elucidating the cell-
autonomous properties of these cells. We enumerate here
a set of cell types known to contribute in important ways to
the biology of many tumors and discuss the regulatory signaling
that controls their individual and collective functions. Most of
these observations stem from the study of carcinomas, in which
the neoplastic epithelial cells constitute a compartment (the
parenchyma) that is clearly distinct from the mesenchymal cells
forming the tumor-associated stroma.
Cancer Cells and Cancer Stem Cells
Cancer cells are the foundation of the disease; they initiate
tumors and drive tumor progression forward, carrying the
oncogenic and tumor suppressor mutations that define cancer
as a genetic disease. Traditionally, the cancer cells within tumors
Cell 144, March 4, 2011 ª2011 Elsevier Inc. 661

have been portrayed as reasonably homogeneous cell popula-
tions until relatively late in the course of tumor progression,
when hyperproliferation combined with increased genetic
instability spawn distinct clonal subpopulations. Reflecting
such clonal heterogeneity, many human tumors are histopatho-
logically diverse, containing regions demarcated by various
degrees of differentiation, proliferation, vascularity, inflamma-
tion, and/or invasiveness. In recent years, however, evidence
has accumulated pointing to the existence of a new dimension
of intratumor heterogeneity and a hitherto-unappreciated
subclass of neoplastic cells within tumors, termed cancer stem
cells (CSCs).
Although the evidence is still fragmentary, CSCs may prove to
be a common constituent of many if not most tumors, albeit
being present with widely varying abundance. CSCs are defined
operationally through their ability to efficiently seed new tumors
upon inoculation into recipient host mice (Cho and Clarke, 2008;
Lobo et al., 2007). This functional definition is often comple-
mented by including the expression in CSCs of markers that
are also expressed by the normal stem cells in the tissue-of-
origin (Al-Hajj et al., 2003).
CSCs were initially implicated in the pathogenesis of hemato-
poietic malignancies (Reya et al., 2001; Bonnet and Dick, 1997)
and then years later were identified in solid tumors, in particular
breast carcinomas and neuroectodermal tumors (Gilbertson and
Rich, 2007; Al-Hajj et al., 2003). Fractionation of cancer cells on
the basis of displayed cell-surface markers has yielded subpop-
ulations of neoplastic cells with a greatly enhanced ability, rela-
tive to the corresponding majority populations, to seed new
tumors upon implantation in immunodeficient mice. These
662 Cell 144, March 4, 2011 ª2011 Elsevier Inc.
Figure 4. The Cells of the Tumor Microenviron-
ment
(Upper) An assemblage of distinct cell types constitutes
most solid tumors. Both the parenchyma and stroma of
tumors contain distinct cell types and subtypes that
collectively enable tumor growth and progression.
Notably, the immune inflammatory cells present in tumors
can include both tumor-promoting as well as tumor-killing
subclasses.
(Lower) The distinctive microenvironments of tumors. The
multiple stromal cell types create a succession of tumor
microenvironments that change as tumors invade normal
tissue and thereafter seed and colonize distant tissues.
The abundance, histologic organization, and phenotypic
characteristics of the stromal cell types, as well as of the
extracellular matrix (hatched background), evolve during
progression, thereby enabling primary, invasive, and then
metastatic growth. The surrounding normal cells of the
primary and metastatic sites, shown only schematically,
likely also affect the character of the various neoplastic
microenvironments. (Not shown are the premalignant
stages in tumorigenesis, which also have distinctive
microenvironments that are created by the abundance
and characteristics of the assembled cells.)
often-rare tumor-initiating cells proved to share
transcriptional profiles with certain normal
tissue stem cell populations, motivating their
designation as stem-like.
The origins of CSCs within a solid tumor have
not been clarified and indeed may well vary from
one tumor type to another. In some tumors, normal tissue stem
cells may serve as the cells-of-origin that undergo oncogenic
transformation to yield CSCs; in others, partially differentiated
transit-amplifying cells, also termed progenitor cells, may suffer
the initial oncogenic transformation thereafter assuming more
stem-like character. Once primary tumors have formed, the
CSCs, like their normal counterparts, may self-renew as well
as spawn more differentiated derivatives; in the case of
neoplastic CSCs, these descendant cells form the great bulk of
many tumors. It remains to be established whether multiple
distinct classes of increasingly neoplastic stem cells form during
inception and subsequent multistep progression of tumors, ulti-
mately yielding the CSCs that have been described in fully devel-
oped cancers.
Recent research has interrelated the acquisition of CSC traits
with the EMT transdifferentiation program discussed above
(Singh and Settleman, 2010; Mani et al., 2008; Morel et al.,
2008). Induction of this program in certain model systems can
induce many of the defining features of stem cells, including
self-renewal ability and the antigenic phenotypes associated
with both normal and cancer stem cells. This concordance
suggests that the EMT program not only may enable cancer cells
to physically disseminate from primary tumors but also can
confer on such cells the self-renewal capability that is crucial
to their subsequent clonal expansion at sites of dissemination
(Brabletz et al., 2005). If generalized, this connection raises an
important corollary hypothesis: the heterotypic signals that
trigger an EMT, such as those released by an activated, inflam-
matory stroma, may also be important in creating and maintain-
ing CSCs.
 An increasing number of human tumors are reported to
contain subpopulations with the properties of CSCs, as defined
operationally through their efficient tumor-initiating capabilities
upon xenotransplantation into mice. Nevertheless, the im-
portance of CSCs as a distinct phenotypic subclass of
neoplastic cells remains a matter of debate, as does their oft-
cited rarity within tumors (Boiko et al., 2010; Gupta et al., 2009;
Quintana et al., 2008). Indeed, it is plausible that the phenotypic
plasticity operating within tumors may produce bidirectional
interconversion between CSCs and non-CSCs, resulting in
dynamic variation in the relative abundance of CSCs. Such
plasticity could complicate definitive measurement of their prev-
alence. Analogous plasticity is already implicated in the EMT
program, which can be engaged reversibly (Thiery and Sleeman,
2006).
These complexities notwithstanding, it is evident that this
new dimension of tumor heterogeneity holds important implica-
tions for successful cancer therapies. Increasing evidence in
a variety of tumor types suggests that cells with properties of
CSCs are more resistant to various commonly used chemother-
apeutic treatments (Singh and Settleman, 2010; Creighton
et al., 2009; Buck et al., 2007). Their persistence may help to
explain the almost-inevitable disease recurrence following
apparently successful debulking of human solid tumors by radi-
ation and various forms of chemotherapy. Indeed, CSCs may
well prove to underlie certain forms of tumor dormancy,
whereby latent cancer cells persist for years or even decades
after surgical resection or radio/chemotherapy, only to
suddenly erupt and generate life-threatening disease. Hence,
CSCs may represent a double-threat, in that they are more
resistant to therapeutic killing and, at the same time, endowed
with the ability to regenerate a tumor once therapy has been
halted.
This phenotypic plasticity implicit in CSC state may also
enable the formation of functionally distinct subpopulations
within a tumor that support overall tumor growth in various
ways. For example, an EMT can convert epithelial carcinoma
cells into mesenchymal, fibroblast-like cancer cells that may
well assume the duties of cancer-associated fibroblasts (CAFs)
in some tumors. Remarkably, several recent reports have
documented the ability of glioblastoma cells (or possibly their
associated CSC subpopulations) to transdifferentiate into endo-
thelial-like cells that can substitute for bona fide host-derived
endothelial cells in forming a tumor-associated neovasculature
(Soda et al., 2011; El Hallani et al., 2010; Ricci-Vitiani et al.,
2010; Wang et al., 2010). Observations like these indicate that
certain tumors may acquire stromal support by inducing some
of their own cancer cells to undergo various types of metamor-
phosis to produce stromal cell types rather than relying on
recruited host cells to provide their functions.
The discovery of CSCs and biological plasticity in tumors
indicates that a single, genetically homogeneous population of
cells within a tumor may nevertheless be phenotypically hetero-
geneous due to the presence of cells in distinct states of differ-
entiation. However, an equally important source of phenotypic
variability may derive from the genetic heterogeneity within
a tumor that accumulates as cancer progression proceeds.
Thus, elevated genetic instability operating in later stages of
tumor progression may drive rampant genetic diversification
that outpaces the process of Darwinian selection, generating
genetically distinct subpopulations far more rapidly than they
can be eliminated.
Such thinking is increasingly supported by in-depth sequence
analysis of tumor cell genomes, which has become practical due
to recent major advances in DNA (and RNA) sequencing tech-
nology. Thus the sequencing of the genomes of cancer cells
microdissected from different sectors of the same tumor
(Yachida et al., 2010) has revealed striking intratumoral genetic
heterogeneity. Some of this genetic diversity may be reflected
in the long-recognized histological heterogeneity within indi-
vidual human tumors. Alternatively, this genetic diversification
may enable functional specialization, producing subpopulations
of cancer cells that contribute distinct, complementary capabil-
ities, which then accrue to the common benefit of overall tumor
growth as described above.
Endothelial Cells
Much of the cellular heterogeneity within tumors is found in
their stromal compartments. Prominent among the stromal
constituents are the cells forming the tumor-associated vascu-
lature. Mechanisms of development, differentiation, and
homeostasis of endothelial cells composing the arteries, veins,
and capillaries were already well understood in 2000. So too
was the concept of the ‘‘angiogenic switch,’’ which activates
quiescent endothelial cells, causing them to enter into a cell-
biological program that allows them to construct new blood
vessels (see above). Over the last decade, a network of inter-
connected signaling pathways involving ligands of signal-trans-
ducing receptors displayed by endothelial cells (e.g., Notch,
Neuropilin, Robo, and Eph-A/B) has been added to the
already-prominent VEGF, angiopoietin, and FGF signals. These
newly characterized pathways have been functionally impli-
cated in developmental and tumor-associated angiogenesis
and illustrate the complex regulation of endothelial cell pheno-
types (Pasquale, 2010; Ahmed and Bicknell, 2009; Dejana
et al., 2009; Carmeliet and Jain, 2000).
Other avenues of research are revealing distinctive gene
expression profiles of tumor-associated endothelial cells and
identifying cell-surface markers displayed on the lumenal
surfaces of normal versus tumor endothelial cells (Nagy et al.,
2010; Ruoslahti et al., 2010; Ruoslahti, 2002). Differences in
signaling, in transcriptome profiles, and in vascular ‘‘ZIP codes’’
will likely prove to be important for understanding the conversion
of normal endothelial cells into tumor-associated endothelial
cells. Such knowledge may lead, in turn, to opportunities to
develop novel therapies that exploit these differences in order
to selectively target tumor-associated endothelial cells.
Closely related to the endothelial cells of the general circula-
tion are those forming lymphatic vessels (Tammela and Alitalo,
2010). Their role in the tumor-associated stroma, specifically in
supporting tumor growth, is poorly understood. Indeed, because
of high interstitial pressure within solid tumors, intratumoral
lymphatic vessels are typically collapsed and nonfunctional; in
contrast, however, there are often functional, actively growing
(‘‘lymphangiogenic’’) lymphatic vessels at the peripheries of
tumors and in the adjacent normal tissues that cancer cells
Cell 144, March 4, 2011 ª2011 Elsevier Inc. 663
invade. These associated lymphatics likely serve as channels for
the seeding of metastases in the draining lymph nodes that are
commonly observed in a number of cancer types.
Pericytes
As noted earlier, pericytes represent a specialized mesenchymal
cell type (related to smooth muscle cells) with finger-like projec-
tions that wrap around the endothelial tubing of blood vessels. In
normal tissues, pericytes are known to provide paracrine
support signals to the normally quiescent endothelium. For
example, Ang-1 secreted by pericytes conveys antiproliferative
stabilizing signals that are received by the Tie2 receptors
expressed on the surface of endothelial cells; some pericytes
also produce low levels of VEGF that serve a trophic function
in endothelial homeostasis (Gaengel et al., 2009; Bergers and
Song, 2005). Pericytes also collaborate with the endothelial cells
to synthesize the vascular basement membrane that anchors
both pericytes and endothelial cells and helps vessel walls to
withstand the hydrostatic pressure of blood flow.
Genetic and pharmacological perturbation of the recruitment
and association of pericytes has demonstrated the functional
importance of these cells in supporting the tumor endothelium
(Pietras and Ostman, 2010; Gaengel et al., 2009; Bergers and
Song, 2005). For example, pharmacological inhibition of
signaling through the PDGF receptor expressed by tumor peri-
cytes and bone marrow-derived pericyte progenitors results in
reduced pericyte coverage of tumor vessels, which in turn desta-
bilizes vascular integrity and function (Pietras and Ostman, 2010;
Raza et al., 2010; Gaengel et al., 2009); interestingly, and in
contrast, the pericytes of normal vessels are not prone to such
pharmacological disruption, providing another example of the
differences in regulation of normal quiescent and tumor vascula-
ture. An intriguing hypothesis, still to be fully substantiated, is
that tumors with poor pericyte coverage of their vasculature
may be more prone to permit cancer cell intravasation into the
circulatory system, enabling subsequent hematogenous
dissemination (Raza et al., 2010; Gerhardt and Semb, 2008).
Immune Inflammatory Cells
As also discussed above, infiltrating cells of the immune system
are increasingly accepted to be generic constituents of tumors.
These inflammatory cells operate in conflicting ways: both
tumor-antagonizing and tumor-promoting leukocytes can be
found, in various proportions, in most if not all neoplastic lesions.
Although the presence of tumor-antagonizing CTLs and NK cells
is not surprising, the prevalence of immune cells that functionally
enhance hallmark capabilities was largely unanticipated.
Evidence began to accumulate in the late 1990s that the infiltra-
tion of neoplastic tissues by cells of the immune system serves,
perhaps counterintuitively, to promote tumor progression. Such
work traced its conceptual roots back to the association of sites
of chronic inflammation with tumor formation, and to the obser-
vation that tumors could be portrayed as wounds that never heal
(Schäfer and Werner, 2008: Dvorak, 1986). In the course of
normal wound healing and fighting infections, immune inflamma-
tory cells appear transiently and then disappear, in contrast to
their persistence in sites of chronic inflammation, where their
presence has been associated with various tissue pathologies,
664 Cell 144, March 4, 2011 ª2011 Elsevier Inc.
including fibrosis, aberrant angiogenesis, and neoplasia (Griven-
nikov et al., 2010; Karin et al., 2006).
Over the past decade, the manipulation of genes involved in
the determination or effector functions of various immune cell
types, together with pharmacological inhibitors of such cells or
their functions, has shown them to play diverse and critical roles
in fostering tumorigenesis. The roster of tumor-promoting
inflammatory cells now includes macrophage subtypes, mast
cells, and neutrophils, as well as T and B lymphocytes (Coffelt
et al., 2010; DeNardo et al., 2010; Egeblad et al., 2010; Johans-
son et al., 2008; Murdoch et al., 2008; DePalma et al., 2007).
Such studies are yielding a growing list of signaling molecules
released by inflammatory cells that serve as effectors of their
tumor-promoting actions. These include the tumor growth factor
EGF, the angiogenic growth factor VEGF, other proangiogenic
factors such as FGF2, chemokines, and cytokines that amplify
the inflammatory state; in addition, these cells may produce
proangiogenic and/or proinvasive matrix-degrading enzymes,
including MMP-9 and other matrix metalloproteinases, cysteine
cathepsin proteases, and heparanase (Qian and Pollard, 2010;
Murdoch et al., 2008). Consistent with their expression of these
diverse effectors, tumor-infiltrating inflammatory cells have been
shown to induce and help sustain tumor angiogenesis, to stimu-
late cancer cell proliferation, to facilitate, via their presence at the
margins of tumors, tissue invasion, and to support the metastatic
dissemination and seeding of cancer cells (Coffelt et al., 2010;
Egeblad et al., 2010; Qian and Pollard, 2010; Mantovani, 2010;
Joyce and Pollard, 2009; Mantovani et al., 2008; Murdoch
et al., 2008; DePalma et al., 2007).
In addition to fully differentiated immune cells present in tumor
stroma, a variety of partially differentiated myeloid progenitors
have been identified in tumors (Murdoch et al., 2008). Such cells
represent intermediaries between circulating cells of bone
marrow origin and the differentiated immune cells typically found
in normal and inflamed tissues. Importantly, these progenitors,
like their more differentiated derivatives, have demonstrable
tumor-promoting activity. Of particular interest, a class of
tumor-infiltrating myeloid cells (defined as coexpressing the
macrophage marker CD11b and the neutrophil marker Gr1)
has been shown to suppress CTL and NK cell activity, having
been independently identified as MDSCs (Qian and Pollard,
2010; Ostrand-Rosenberg and Sinha, 2009). This attribute raises
the possibility that recruitment of certain myeloid cells may be
doubly beneficial for the developing tumor, by directly promoting
angiogenesis and tumor progression while at the same time
affording a means to evade immune destruction.
The counterintuitive existence of both tumor-promoting and
tumor-antagonizing immune cells can be rationalized by
invoking the diverse roles of the immune system: On the one
hand, the immune system specifically detects and targets infec-
tious agents with the adaptive immune response, which is sup-
ported by cells of the innate immune system. On the other, the
innate immune system is involved in wound healing and clearing
dead cells and cellular debris. These specialized tasks are
accomplished by distinct subclasses of inflammatory cells,
namely a class of conventional macrophages and neutrophils
(engaged in supporting adaptive immunity), and subclasses of
‘‘alternatively activated’’ macrophages, neutrophils, and
myeloid progenitors that are engaged in wound healing and
tissue housecleaning (Egeblad et al., 2010; Mantovani, 2010;
Qian and Pollard, 2010; Johansson et al., 2008). The latter
subtypes of immune cells are one of the major sources of the
angiogenic, epithelial, and stromal growth factors and matrix-re-
modeling enzymes that are needed for wound healing, and it is
these cells that are recruited and subverted to support
neoplastic progression. Similarly, subclasses of B and T
lymphocytes may facilitate the recruitment, activation, and
persistence of such wound-healing and tumor-promoting
macrophages and neutrophils (DeNardo et al., 2010; Egeblad
et al., 2010; Biswas and Mantovani, 2010). Of course, other
subclasses of B and T lymphocytes and innate immune cell
types can mount demonstrable tumor-killing responses. The
balance between the conflicting inflammatory responses in
tumors is likely to prove instrumental in prognosis and, quite
possibly, in therapies designed to redirect these cells toward
tumor destruction.
Cancer-Associated Fibroblasts
Fibroblasts are found in various proportions across the spectrum
of carcinomas, constituting in many cases the preponderant cell
population of the tumor stroma. The term ‘‘cancer-associated
fibroblast’’ subsumes at least two distinct cell types: (1) cells
with similarities to the fibroblasts that create the structural foun-
dation supporting most normal epithelial tissues and (2) myofi-
broblasts, whose biological roles and properties differ markedly
from those of tissue-derived fibroblasts. Myofibroblasts are
identifiable by their expression of a-smooth muscle actin
(SMA). They are rare in most healthy epithelial tissues, although
certain tissues, such as the liver and pancreas, contain appre-
ciable numbers of a-SMA-expressing cells. Myofibroblasts tran-
siently increase in abundance in wounds and are also found in
sites of chronic inflammation. Although beneficial to tissue
repair, myofibroblasts are problematic in chronic inflammation,
contributing to the pathological fibrosis observed in tissues
such as lung, kidney, and liver.
Recruited myofibroblasts and reprogrammed variants of
normal tissue-derived fibroblastic cells have been demonstrated
to enhance tumor phenotypes, notably cancer cell proliferation,
angiogenesis, and invasion and metastasis; their tumor-
promoting activities have largely been defined by transplantation
of cancer-associated fibroblasts admixed with cancer cells into
mice, and more recently by genetic and pharmacologic pertur-
bation of their functions in tumor-prone mice (Dirat et al., 2010;
Pietras and Ostman, 2010; Räsänen and Vaheri, 2010; Shimoda
et al., 2010; Kalluri and Zeisberg, 2006; Bhowmick et al., 2004).
Because they secrete a variety of extracellular matrix compo-
nents, cancer-associated fibroblasts are implicated in the forma-
tion of the desmoplastic stroma that characterizes many
advanced carcinomas. The full spectrum of functions contrib-
uted by both subtypes of cancer-associated fibroblasts to tumor
pathogenesis remains to be elucidated.
Stem and Progenitor Cells of the Tumor Stroma
The various stromal cell types that constitute the tumor microen-
vironment may be recruited from adjacent normal tissue—the
most obvious reservoir of such cell types. However, in recent
years, the bone marrow has increasingly been implicated as
a key source of tumor-associated stromal cells (Bergfeld and
DeClerck, 2010; Fang and Salven, 2011; Giaccia and Schipani,
2010; Patenaude et al., 2010; Lamagna and Bergers, 2006).
Mesenchymal stem and progenitor cells have been found to
transit into tumors from the marrow, where they may differentiate
into the various well-characterized stromal cell types. Some of
these recent arrivals may also persist in an undifferentiated or
partially differentiated state, exhibiting functions that their more
differentiated progeny lack.
The bone marrow origins of stromal cell types have
been demonstrated using tumor-bearing mice in which the
bone marrow cells and thus their disseminated progeny have
been selectively labeled with reporters such as green fluorescent
protein (GFP). While immune inflammatory cells have been long
known to derive from the bone marrow, more recently the
progenitors of pericytes and of various subtypes of cancer-asso-
ciated fibroblasts originating from the bone marrow have been
described in various mouse models of cancer (Bergfeld and
DeClerck, 2010; Fang and Salven, 2011; Giaccia and Schipani,
2010; Lamagna and Bergers, 2006); the prevalence and func-
tional importance of endothelial progenitors for tumor angiogen-
esis is currently unresolved (Fang and Salven, 2011; Patenaude
et al., 2010). Taken together, these various lines of evidence indi-
cate that tumor-associated stromal cells may be supplied to
growing tumors by proliferation of preexisting stromal cells, by
differentiation in situ of local stem/progenitor cells originating
in the neighboring normal tissue, or via recruitment of bone
marrow-derived stem/progenitor cells.
Heterotypic Signaling Orchestrates
the Cells of the Tumor Microenvironment
Depictions of the intracellular circuitry governing cancer cell
biology (e.g., Figure 2) will need to be complemented by similar
diagrams charting the complex interactions between the
neoplastic and stromal cells within a tumor and the dynamic
extracellular matrix that they collectively erect and remodel (Ege-
blad et al., 2010; Kessenbrock et al., 2010; Pietras and Ostman,
2010; Polyak et al., 2009). A reasonably complete, graphic
depiction of the network of microenvironmental signaling inter-
actions is still far beyond our reach, as the great majority of
signaling molecules and pathways remain to be identified. We
provide instead a hint of such interactions in Figure 5, upper.
These few well-established examples are intended to exemplify
a signaling network of remarkable complexity that is of critical
importance to tumor pathogenesis.
Another dimension of complexity is not represented in this
simple schematic: both neoplastic cells and the stromal cells
around them change progressively during the multistep transfor-
mation of normal tissues into high-grade malignancies. This
histopathological progression must reflect underlying changes
in heterotypic signaling between tumor parenchyma and stroma.
Such stepwise progression is likely to depend on back-and-
forth reciprocal interactions between the neoplastic cells and
the supporting stromal cells, as depicted in Figure 5, lower.
Thus, incipient neoplasias begin the interplay by recruiting and
activating stromal cell types that assemble into an initial preneo-
plastic stroma, which in turn responds reciprocally by enhancing
Cell 144, March 4, 2011 ª2011 Elsevier Inc. 665
Figure 5. Signaling Interactions in the Tumor Microenvironment during Malignant Progression
(Upper) The assembly and collective contributions of the assorted cell types constituting the tumor microenvironment are orchestrated and maintained by
reciprocal heterotypic signaling interactions, of which only a few are illustrated.
(Lower) The intracellular signaling depicted in the upper panel within the tumor microenvironment is not static but instead changes during tumor progression as
a result of reciprocal signaling interactions between cancer cells of the parenchyma and stromal cells that convey the increasingly aggressive phenotypes that
underlie growth, invasion, and metastatic dissemination. Importantly, the predisposition to spawn metastatic lesions can begin early, being influenced by the
differentiation program of the normal cell-of-origin or by initiating oncogenic lesions. Certain organ sites (sometimes referred to as ‘‘fertile soil’’ or ‘‘metastatic
niches’’) can be especially permissive for metastatic seeding and colonization by certain types of cancer cells, as a consequence of local properties that are either
intrinsic to the normal tissue or induced at a distance by systemic actions of primary tumors. Cancer stem cells may be variably involved in some or all of the
different stages of primary tumorigenesis and metastasis.

666 Cell 144, March 4, 2011 ª2011 Elsevier Inc.

the neoplastic phenotypes of the nearby cancer cells. The
cancer cells, which may further evolve genetically, again feed
signals back to the stroma, continuing the reprogramming of
normal stromal cells to serve the budding neoplasm; ultimately
signals originating in the tumor stroma enable cancer cells to
invade normal adjacent tissues and disseminate.
This model of reciprocal heterotypic signaling must be
extended to encompass the final stage of multistep tumor
progression—metastasis (Figure 5, lower right). The circulating
cancer cells that are released from primary tumors leave a micro-
environment created by the supportive stroma of such tumors.
However, upon landing in a distant organ, these cancer cells
encounter a naive, fully normal, tissue microenvironment.
Consequently, many of the heterotypic signals that shaped their
phenotype while they resided within primary tumors may be
absent in sites of dissemination, constituting a barrier to growth
of the seeded cancer cells. Thus, the succession of reciprocal
cancer cell to stromal cell interactions that defined multistep
progression in the primary tumor now must be repeated anew
in distant tissues as disseminated cancer cells proceed to colo-
nize their newfound organ sites.
Although this logic applies in some cases of metastasis, in
others, as mentioned earlier, certain tissue microenvironments
may, for various reasons, already be supportive of freshly
seeded cancer cells; such permissive sites have been referred
to as ‘‘metastatic niches’’ (Peinado et al., 2011; Coghlin and
Murray, 2010). Implicit in this term is the notion that cancer cells
seeded in such sites may not need to begin by inducing
a supportive stroma because it already preexists, at least in
part. Such permissivity may be intrinsic to the tissue site
(Talmadge and Fidler, 2010) or preinduced by circulating factors
released by the primary tumor (Peinado et al., 2011). The most
well-documented components of induced premetastatic niches
are tumor-promoting inflammatory cells, although other cell
types and the ECM may well prove to play important roles in
different metastatic contexts.
The likelihood that signaling interactions between cancer cells
and their supporting stroma evolve during the course of multi-
stage tumor development clearly complicates the goal of fully
elucidating the mechanisms of cancer pathogenesis. For
example, this reality poses challenges to systems biologists
seeking to chart the crucial regulatory networks than orchestrate
malignant progression. Moreover, it seems likely that under-
standing these dynamic variations will become crucial to the
development of novel therapies designed to successfully target
both primary and metastatic tumors.
THERAPEUTIC TARGETING
The introduction of mechanism-based targeted therapies to
treat human cancers has been heralded as one of the fruits of
three decades of remarkable progress of research into the
mechanisms of cancer pathogenesis. We do not attempt here
to enumerate the myriad therapies that are under development
or have been introduced of late into the clinic. Instead, we
consider how the description of hallmark principles is beginning
to inform therapeutic development at present and may increas-
ingly do so in the future.
The rapidly growing armamentarium of targeted therapeutics
can be categorized according to their respective effects on
one or more hallmark capabilities, as illustrated in the examples
presented in Figure 6. Indeed, the observed efficacy of these
drugs represents, in each case, a validation of a particular capa-
bility: if a capability is truly important for the biology of tumors,
then its inhibition should impair tumor growth and progression.
We note that most of the hallmark-targeting cancer drugs
developed to date have been deliberately directed toward
specific molecular targets that are involved in one way or another
in enabling particular capabilities. Such specificity of action has
been considered a virtue, as it presents inhibitory activity against
a target while having, in principle, relatively fewer off-target
effects and thus less nonspecific toxicity. In fact, resulting clin-
ical responses have generally been transitory, being followed
by almost-inevitable relapses.
One interpretation of this history, supported by growing exper-
imental evidence, is that each of the core hallmark capabilities is
regulated by partially redundant signaling pathways. Conse-
quently, a targeted therapeutic agent inhibiting one key pathway
in a tumor may not completely shut off a hallmark capability, al-
lowing some cancer cells to survive with residual function until
they or their progeny eventually adapt to the selective pressure
imposed by the therapy being applied. Such adaptation, which
can be accomplished by mutation, epigenetic reprogramming,
or remodeling of the stromal microenvironment, can reestablish
the functional capability, permitting renewed tumor growth and
clinical relapse. Given that the number of parallel signaling path-
ways supporting a given hallmark must be limited, it may
become possible to target all of these supporting pathways ther-
apeutically, thereby preventing the development of adaptive
resistance.
In response to therapy, cancer cells may also reduce their
dependence on a particular hallmark capability, becoming
more dependent on another; this represents a quite different
form of acquired drug resistance. This concept is exemplified
by recent discoveries of unexpected responses to antiangio-
genic therapies. Some have anticipated that effective inhibition
of angiogenesis would render tumors dormant and might even
lead to their dissolution (Folkman and Kalluri, 2004). Instead,
the clinical responses to antiangiogenic therapies have been
found to be transitory (Azam et al., 2010; Ebos et al., 2009; Berg-
ers and Hanahan, 2008).
In certain preclinical models, where potent angiogenesis inhib-
itors succeed in suppressing this hallmark capability, tumors
adapt and shift from a dependence upon continuing angiogen-
esis to heightening the activity of another instead—invasiveness
and metastasis (Azam et al., 2010: Ebos et al., 2009; Bergers and
Hanahan, 2008). By invading nearby tissues, initially hypoxic
cancer cells evidently gain access to normal, preexisting tissue
vasculature. Initial clinical validation of this adaptive/evasive
resistance is apparent in the increased invasion and local metas-
tasis seen when human glioblastomas are treated with antian-
giogenic therapies (Ellis and Reardon, 2009; Norden et al.,
2009; Verhoeff et al., 2009). The applicability of this lesson to
other human cancers has yet to be established.
Analogous adaptive shifts in dependence on other hallmark
traits may also limit efficacy of analogous hallmark-targeting
Cell 144, March 4, 2011 ª2011 Elsevier Inc. 667
Figure 6. Therapeutic Targeting of the Hallmarks of Cancer
Drugs that interfere with each of the acquired capabilities necessary for tumor growth and progression have been developed and are in clinical trials or in some
cases approved for clinical use in treating certain forms of human cancer. Additionally, the investigational drugs are being developed to target each of the
enabling characteristics and emerging hallmarks depicted in Figure 3, which also hold promise as cancer therapeutics. The drugs listed are but illustrative
examples; there is a deep pipeline of candidate drugs with different molecular targets and modes of action in development for most of these hallmarks.

therapies. For example, the deployment of apoptosis-inducing
drugs may induce cancer cells to hyperactivate mitogenic
signaling, enabling them to compensate for the initial attrition
triggered by such treatments. Such considerations suggest
that drug development and the design of treatment protocols
will benefit from incorporating the concepts of functionally
discrete hallmark capabilities and of the multiple biochemical
pathways involved in supporting each of them. Thus, in partic-
ular, we can envisage that selective cotargeting of multiple
core and emerging hallmark capabilities and enabling character-
istics (Figure 6) in mechanism-guided combinations will result in
more effective and durable therapies for human cancer.
CONCLUSION AND FUTURE VISION
We have sought here to revisit, refine, and extend the concept of
cancer hallmarks, which has provided a useful conceptual
framework for understanding the complex biology of cancer.
The six acquired capabilities—the hallmarks of cancer—have
stood the test of time as being integral components of most
forms of cancer. Further refinement of these organizing princi-
ples will surely come in the foreseeable future, continuing the
remarkable conceptual progress of the last decade.
Looking ahead, we envision significant advances during the
coming decade in our understanding of invasion and metastasis.
Similarly, the role of aerobic glycolysis in malignant growth will
be elucidated, including a resolution of whether this metabolic
reprogramming is a discrete capability separable from the core
hallmark of chronically sustained proliferation. We remain
perplexed as to whether immune surveillance is a barrier that
virtually all tumors must circumvent, or only an idiosyncrasy of
an especially immunogenic subset of them; this issue too will
be resolved in one way or another.
Yet other areas are currently in rapid flux. In recent years, elab-
orate molecular mechanisms controlling transcription through
chromatin modifications have been uncovered, and there are

668 Cell 144, March 4, 2011 ª2011 Elsevier Inc.

clues that specific shifts in chromatin configuration occur during
the acquisition of certain hallmark capabilities (Berdasco and Es-
teller, 2010). Functionally significant epigenetic alterations seem
likely to be factors not only in the cancer cells but also in the
altered cells of the tumor-associated stroma. It is unclear at
present whether an elucidation of these epigenetic mechanisms
will materially change our overall understanding of the means by
which hallmark capabilities are acquired or simply add additional
detail to the regulatory circuitry that is already known to govern
them.
Similarly, the discovery of hundreds of distinct regulatory mi-
croRNAs has already led to profound changes in our under-
standing of the genetic control mechanisms that operate in
health and disease. By now dozens of microRNAs have been
implicated in various tumor phenotypes (Garzon et al., 2010),
and yet these only scratch the surface of the real complexity,
as the functions of hundreds of microRNAs known to be present
in our cells and altered in expression in different forms of cancer
remain total mysteries. Here again, we are unclear as to whether
future progress will cause fundamental shifts in our under-
standing of the pathogenetic mechanisms of cancer or only
add detail to the elaborate regulatory circuits that have already
been mapped out.
Finally, the circuit diagrams of heterotypic interactions
between the multiple distinct cell types that assemble and
collaborate to produce different forms and progressively malig-
nant stages of cancer are currently rudimentary. In another
decade, we anticipate that the signaling circuitry describing
the intercommunication between these various cells within
tumors will be charted in far greater detail and clarity, eclipsing
our current knowledge. And, as before (Hanahan and Weinberg,
2000), we continue to foresee cancer research as an increasingly
logical science, in which myriad phenotypic complexities are
manifestations of a small set of underlying organizing principles.
SUPPLEMENTAL INFORMATION
Supplemental Information includes six figures that are downloadable for
presentations and can be found with this article online at doi:10.1016/j.cell.
2011.02.013.
ACKNOWLEDGMENTS
We thank Terry Schoop (OFC Graphics, Kensington, CA, USA) for exceptional
efforts in preparing the figures. And we thank Gerard Evan (Cambridge, UK),
Erwin Wagner (Madrid, ESP), and Zena Werb (San Francisco, USA) for valu-
able comments and suggestions on the manuscript. D.H. and R.A.W. are
American Cancer Society Research Professors. Research in the authors’ labo-
ratories has been largely supported by the U.S. National Cancer Institute. Due
to space limitations, many primary and historical publications have not been
cited, particularly in cases where topical reviews are available.
REFERENCES
Adams, J.M., and Cory, S. (2007). The Bcl-2 apoptotic switch in cancer devel-
opment and therapy. Oncogene 26, 1324–1337.
Aguirre-Ghiso, J.A. (2007). Models, mechanisms and clinical evidence for
cancer dormancy. Nat. Rev. Cancer 7, 834–846.
Ahmed, Z., and Bicknell, R. (2009). Angiogenic signalling pathways. Methods
Mol. Biol. 467, 3–24.
Al-Hajj, M., Wicha, M.S., Benito-Hernandez, A., Morrison, S.J., and Clarke,
M.F. (2003). Prospective identification of tumorigenic breast cancer cells.
Proc. Natl. Acad. Sci. USA 100, 3983–3988.
Amaravadi, R.K., and Thompson, C.B. (2007). The roles of therapy-induced
autophagy and necrosis in cancer treatment. Clin. Cancer Res. 13, 7271–7279.
Amit, I., Citri, A., Shay, T., Lu, Y., Katz, M., Zhang, F., Tarcic, G., Siwak, D., La-
had, J., Jacob-Hirsch, J., et al. (2007). A module of negative feedback regula-
tors defines growth factor signaling. Nat. Genet. 39, 503–512.
Apel, A., Zentgraf, H., Büchler, M.W., and Herr, I. (2009). Autophagy-A double-
edged sword in oncology. Int. J. Cancer 125, 991–995.
Artandi, S.E., and DePinho, R.A. (2000). Mice without telomerase: what can
they teach us about human cancer? Nat. Med. 6, 852–855.
Artandi, S.E., and DePinho, R.A. (2010). Telomeres and telomerase in cancer.
Carcinogenesis 31, 9–18.
Azam, F., Mehta, S., and Harris, A.L. (2010). Mechanisms of resistance to anti-
angiogenesis therapy. Eur. J. Cancer 46, 1323–1332.
Baeriswyl, V., and Christofori, G. (2009). The angiogenic switch in carcinogen-
esis. Semin. Cancer Biol. 19, 329–337.
Baluk, P., Hashizume, H., and McDonald, D.M. (2005). Cellular abnormalities
of blood vessels as targets in cancer. Curr. Opin. Genet. Dev. 15, 102–111.
Barkan, D., Green, J.E., and Chambers, A.F. (2010). Extracellular matrix:
a gatekeeper in the transition from dormancy to metastatic growth. Eur. J.
Cancer 46, 1181–1188.
Barnes, D.E., and Lindahl, T. (2004). Repair and genetic consequences of endog-
enous DNA base damage in mammalian cells. Annu. Rev. Genet. 38, 445–476.
Barrallo-Gimeno, A., and Nieto, M.A. (2005). The Snail genes as inducers of cell
movement and survival: implications in development and cancer. Develop-
ment 132, 3151–3161.
Berdasco, M., and Esteller, M. (2010). Aberrant epigenetic landscape in
cancer: How cellular identity goes awry. Dev. Cell 19, 698–711.
Bergers, G., and Benjamin, L.E. (2003). Tumorigenesis and the angiogenic
switch. Nat. Rev. Cancer 3, 401–410.
Bergers, G., and Hanahan, D. (2008). Modes of resistance to anti-angiogenic
therapy. Nat. Rev. Cancer 8, 592–603.
Bergers, G., and Song, S. (2005). The role of pericytes in blood-vessel forma-
tion and maintenance. Neuro-oncol. 7, 452–464.
Bergfeld, S.A., and DeClerck, Y.A. (2010). Bone marrow-derived mesen-
chymal stem cells and the tumor microenvironment. Cancer Metastasis Rev.
29, 249–261.
Berx, G., and van Roy, F. (2009). Involvement of members of the cadherin
superfamily in cancer. Cold Spring Harb. Perspect. Biol. 1, a003129.
Bhowmick, N.A., Neilson, E.G., and Moses, H.L. (2004). Stromal fibroblasts in
cancer initiation and progression. Nature 432, 332–337.
Bierie, B., and Moses, H.L. (2006). Tumour microenvironment: TGFbeta: the
molecular Jekyll and Hyde of cancer. Nat. Rev. Cancer 6, 506–520.
Bindea, G., Mlecnik, B., Fridman, W.H., Pagès, F., and Galon, J. (2010). Natural
immunity to cancer in humans. Curr. Opin. Immunol. 22, 215–222.
Biswas, S.K., and Mantovani, A. (2010). Macrophage plasticity and interaction
with lymphocyte subsets: cancer as a paradigm. Nat. Immunol. 11, 889–896.
Blasco, M.A. (2005). Telomeres and human disease: ageing, cancer and
beyond. Nat. Rev. Genet. 6, 611–622.
Boiko, A.D., Razorenova, O.V., van de Rijn, M., Swetter, S.M., Johnson, D.L.,
Ly, D.P., Butler, P.D., Yang, G.P., Joshua, B., Kaplan, M.J., et al. (2010).
Human melanoma-initiating cells express neural crest nerve growth factor
receptor CD271. Nature 466, 133–137.
Bonnet, D., and Dick, J.E. (1997). Human acute myeloid leukemia is organized
as a hierarchy that originates from a primitive hematopoietic cell. Nat. Med. 3,
730–737.
Bos, P.D., Zhang, X.H., Nadal, C., Shu, W., Gomis, R.R., Nguyen, D.X., Minn, A.J.,
van de Vijver, M.J., Gerald, W.L., Foekens, J.A., and Massagué, J. (2009). Genes
that mediate breast cancer metastasis to the brain. Nature 459, 1005–1009.
Cell 144, March 4, 2011 ª2011 Elsevier Inc. 669
Brabletz, T., Jung, A., Reu, S., Porzner, M., Hlubek, F., Kunz-Schughart, L.A.,
Knuechel, R., and Kirchner, T. (2001). Variable beta-catenin expression in
colorectal cancers indicates tumor progression driven by the tumor environ-
ment. Proc. Natl. Acad. Sci. USA 98, 10356–10361.
Brabletz, T., Jung, A., Spaderna, S., Hlubek, F., and Kirchner, T. (2005).
Opinion: migrating cancer stem cells - an integrated concept of malignant
tumour progression. Nat. Rev. Cancer 5, 744–749.
Buck, E., Eyzaguirre, A., Barr, S., Thompson, S., Sennello, R., Young, D.,
Iwata, K.K., Gibson, N.W., Cagnoni, P., and Haley, J.D. (2007). Loss of homo-
typic cell adhesion by epithelial-mesenchymal transition or mutation limits
sensitivity to epidermal growth factor receptor inhibition. Mol. Cancer Ther.
6, 532–541.
Burkhart, D.L., and Sage, J. (2008). Cellular mechanisms of tumour suppres-
sion by the retinoblastoma gene. Nat. Rev. Cancer 8, 671–682.
Cabrita, M.A., and Christofori, G. (2008). Sprouty proteins, masterminds of
receptor tyrosine kinase signaling. Angiogenesis 11, 53–62.
Campbell, P.J., Yachida, S., Mudie, L.J., Stephens, P.J., Pleasance, E.D.,
Stebbings, L.A., Morsberger, L.A., Latimer, C., McLaren, S., Lin, M.L., et al.
(2010). The patterns and dynamics of genomic instability in metastatic pancre-
atic cancer. Nature 467, 1109–1113.
Cao, Y. (2010). Adipose tissue angiogenesis as a therapeutic target for obesity
and metabolic diseases. Nat. Rev. Drug Discov. 9, 107–115.
Carmeliet, P. (2005). VEGF as a key mediator of angiogenesis in cancer.
Oncology 69 (Suppl 3), 4–10.
Carmeliet, P., and Jain, R.K. (2000). Angiogenesis in cancer and other
diseases. Nature 407, 249–257.
Cavallaro, U., and Christofori, G. (2004). Cell adhesion and signalling by cad-
herins and Ig-CAMs in cancer. Nat. Rev. Cancer 4, 118–132.
Cheng, N., Chytil, A., Shyr, Y., Joly, A., and Moses, H.L. (2008). Transforming
growth factor-beta signaling-deficient fibroblasts enhance hepatocyte growth
factor signaling in mammary carcinoma cells to promote scattering and inva-
sion. Mol. Cancer Res. 6, 1521–1533.
Chin, K., de Solorzano, C.O., Knowles, D., Jones, A., Chou, W., Rodriguez,
E.G., Kuo, W.L., Ljung, B.M., Chew, K., Myambo, K., et al. (2004). In situ anal-
yses of genome instability in breast cancer. Nat. Genet. 36, 984–988.
Cho, R.W., and Clarke, M.F. (2008). Recent advances in cancer stem cells.
Curr. Opin. Genet. Dev. 18, 1–6.
Ciccia, A., and Elledge, S.J. (2010). The DNA damage response: making it safe
to play with knives. Mol. Cell 40, 179–204.
Coffelt, S.B., Lewis, C.E., Naldini, L., Brown, J.M., Ferrara, N., and De Palma,
M. (2010). Elusive identities and overlapping phenotypes of proangiogenic
myeloid cells in tumors. Am. J. Pathol. 176, 1564–1576.
Coghlin, C., and Murray, G.I. (2010). Current and emerging concepts in tumour
metastasis. J. Pathol. 222, 1–15.
Collado, M., and Serrano, M. (2010). Senescence in tumours: evidence from
mice and humans. Nat. Rev. Cancer 10, 51–57.
Colotta, F., Allavena, P., Sica, A., Garlanda, C., and Mantovani, A. (2009).
Cancer-related inflammation, the seventh hallmark of cancer: links to genetic
instability. Carcinogenesis 30, 1073–1081.
Cong, Y., and Shay, J.W. (2008). Actions of human telomerase beyond telo-
meres. Cell Res. 18, 725–732.
Creighton, C.J., Li, X., Landis, M., Dixon, J.M., Neumeister, V.M., Sjolund, A.,
Rimm, D.L., Wong, H., Rodriguez, A., Herschkowitz, J.I., et al. (2009). Residual
breast cancers after conventional therapy display mesenchymal as well as
tumor-initiating features. Proc. Natl. Acad. Sci. USA 106, 13820–13825.
Curto, M., Cole, B.K., Lallemand, D., Liu, C.H., and McClatchey, A.I. (2007).
Contact-dependent inhibition of EGFR signaling by Nf2/Merlin. J. Cell Biol.
177, 893–903.
Davies, M.A., and Samuels, Y. (2010). Analysis of the genome to personalize
therapy for melanoma. Oncogene 29, 5545–5555.
670 Cell 144, March 4, 2011 ª2011 Elsevier Inc.
DeBerardinis, R.J., Lum, J.J., Hatzivassiliou, G., and Thompson, C.B. (2008).
The biology of cancer: Metabolic reprogramming fuels cell growth and prolif-
eration. Cell Metab. 7, 11–20.
Dejana, E., Orsenigo, F., Molendini, C., Baluk, P., and McDonald, D.M. (2009).
Organization and signaling of endothelial cell-to-cell junctions in various
regions of the blood and lymphatic vascular trees. Cell Tissue Res. 335, 17–25.
Demicheli, R., Retsky, M.W., Hrushesky, W.J., Baum, M., and Gukas, I.D.
(2008). The effects of surgery on tumor growth: a century of investigations.
Ann. Oncol. 19, 1821–1828.
DeNardo, D.G., Andreu, P., and Coussens, L.M. (2010). Interactions between
lymphocytes and myeloid cells regulate pro- versus anti-tumor immunity.
Cancer Metastasis Rev. 29, 309–316.
De Palma, M., Murdoch, C., Venneri, M.A., Naldini, L., and Lewis, C.E. (2007).
Tie2-expressing monocytes: regulation of tumor angiogenesis and therapeutic
implications. Trends Immunol. 28, 519–524.
Deshpande, A., Sicinski, P., and Hinds, P.W. (2005). Cyclins and cdks in devel-
opment and cancer: a perspective. Oncogene 24, 2909–2915.
de Visser, K.E., Eichten, A., and Coussens, L.M. (2006). Paradoxical roles of
the immune system during cancer development. Nat. Rev. Cancer 6, 24–37.
Dirat, B., Bochet, L., Escourrou, G., Valet, P., and Muller, C. (2010). Unraveling
the obesity and breast cancer links: a role for cancer-associated adipocytes?
Endocr. Dev. 19, 45–52.
Dvorak, H.F. (1986). Tumors: wounds that do not heal. Similarities between
tumor stroma generation and wound healing. N. Engl. J. Med. 315, 1650–1659.
Ebos, J.M., Lee, C.R., and Kerbel, R.S. (2009). Tumor and host-mediated path-
ways of resistance and disease progression in response to antiangiogenic
therapy. Clin. Cancer Res. 15, 5020–5025.
Egeblad, M., Nakasone, E.S., and Werb, Z. (2010). Tumors as organs: complex
tissues that interface with the entire organism. Dev. Cell 18, 884–901.
El Hallani, S., Boisselier, B., Peglion, F., Rousseau, A., Colin, C., Idbaih, A.,
Marie, Y., Mokhtari, K., Thomas, J.L., Eichmann, A., et al. (2010). A new alter-
native mechanism in glioblastoma vascularization: tubular vasculogenic
mimicry. Brain 133, 973–982.
Ellis, L.M., and Reardon, D.A. (2009). Cancer: The nuances of therapy. Nature
458, 290–292.
Esteller, M. (2007). Cancer epigenomics: DNA methylomes and histone-modi-
fication maps. Nat. Rev. Genet. 8, 286–298.
Evan, G.I., and d’Adda di Fagagna, F. (2009). Cellular senescence: hot or
what? Curr. Opin. Genet. Dev. 19, 25–31.
Evan, G., and Littlewood, T. (1998). A matter of life and cell death. Science 281,
1317–1322.
Fang, S., and Salven, P. (2011). Stem cells in tumor angiogenesis. J. Mol. Cell.
Cardiol. 50, 290–295.
Feron, O. (2009). Pyruvate into lactate and back: from the Warburg effect to
symbiotic energy fuel exchange in cancer cells. Radiother. Oncol. 92, 329–
333.
Feldser, D.M., and Greider, C.W. (2007). Short telomeres limit tumor progres-
sion in vivo by inducing senescence. Cancer Cell 11, 461–469.
Ferrara, N. (2009). Vascular endothelial growth factor. Arterioscler. Thromb.
Vasc. Biol. 29, 789–791.
Ferrara, N. (2010). Pathways mediating VEGF-independent tumor angiogen-
esis. Cytokine Growth Factor Rev. 21, 21–26.
Ferrone, C., and Dranoff, G. (2010). Dual roles for immunity in gastrointestinal
cancers. J. Clin. Oncol. 28, 4045–4051.
Fidler, I.J. (2003). The pathogenesis of cancer metastasis: the ‘seed and soil’
hypothesis revisited. Nat. Rev. Cancer 3, 453–458.
Folkman, J. (2002). Role of angiogenesis in tumor growth and metastasis.
Semin. Oncol. 29(6, Suppl 16), 15–18.
Folkman, J. (2006). Angiogenesis. Annu. Rev. Med. 57, 1–18.
Folkman, J., and Kalluri, R. (2004). Cancer without disease. Nature 427, 787.
Friedberg, E.C., Aguilera, A., Gellert, M., Hanawalt, P.C., Hays, J.B., Lehmann,
A.R., Lindahl, T., Lowndes, N., Sarasin, A., and Wood, R.D. (2006). DNA repair:
from molecular mechanism to human disease. DNA Repair (Amst.) 5, 986–996.
Friedl, P., and Wolf, K. (2008). Tube travel: the role of proteases in individual
and collective cancer cell invasion. Cancer Res. 68, 7247–7249.
Friedl, P., and Wolf, K. (2010). Plasticity of cell migration: a multiscale tuning
model. J. Cell Biol. 188, 11–19.
Gaengel, K., Genové, G., Armulik, A., and Betsholtz, C. (2009). Endothelial-
mural cell signaling in vascular development and angiogenesis. Arterioscler.
Thromb. Vasc. Biol. 29, 630–638.
Galluzzi, L., and Kroemer, G. (2008). Necroptosis: a specialized pathway of
programmed necrosis. Cell 135, 1161–1163.
Garzon, R., Marcucci, G., and Croce, C.M. (2010). Targeting microRNAs in
cancer: rationale, strategies and challenges. Nat. Rev. Drug Discov. 9, 775–
789.
Gerhardt, H., and Semb, H. (2008). Pericytes: gatekeepers in tumour cell
metastasis? J. Mol. Med. 86, 135–144.
Ghebranious, N., and Donehower, L.A. (1998). Mouse models in tumor
suppression. Oncogene 17, 3385–3400.
Giaccia, A.J., and Schipani, E. (2010). Role of carcinoma-associated fibro-
blasts and hypoxia in tumor progression. Curr. Top. Microbiol. Immunol.
345, 31–45.
Gilbertson, R.J., and Rich, J.N. (2007). Making a tumour’s bed: glioblastoma
stem cells and the vascular niche. Nat. Rev. Cancer 7, 733–736.
Gocheva, V., Wang, H.W., Gadea, B.B., Shree, T., Hunter, K.E., Garfall, A.L.,
Berman, T., and Joyce, J.A. (2010). IL-4 induces cathepsin protease activity
in tumor-associated macrophages to promote cancer growth and invasion.
Genes Dev. 24, 241–255.
Grivennikov, S.I., Greten, F.R., and Karin, M. (2010). Immunity, inflammation,
and cancer. Cell 140, 883–899.
Gupta, G.P., Minn, A.J., Kang, Y., Siegel, P.M., Serganova, I., Cordón-Cardo,
C., Olshen, A.B., Gerald, W.L., and Massagué, J. (2005). Identifying site-
specific metastasis genes and functions. Cold Spring Harb. Symp. Quant.
Biol. 70, 149–158.
Gupta, P.B., Chaffer, C.L., and Weinberg, R.A. (2009). Cancer stem cells:
mirage or reality? Nat. Med. 15, 1010–1012.
Hanahan, D., and Folkman, J. (1996). Patterns and emerging mechanisms of
the angiogenic switch during tumorigenesis. Cell 86, 353–364.
Hanahan, D., and Weinberg, R.A. (2000). The hallmarks of cancer. Cell 100,
57–70.
Hansel, D.E., Meeker, A.K., Hicks, J., De Marzo, A.M., Lillemoe, K.D., Schulick,
R., Hruban, R.H., Maitra, A., and Argani, P. (2006). Telomere length variation in
biliary tract metaplasia, dysplasia, and carcinoma. Mod. Pathol. 19, 772–779.
Hardee, M.E., Dewhirst, M.W., Agarwal, N., and Sorg, B.S. (2009). Novel
imaging provides new insights into mechanisms of oxygen transport in tumors.
Curr. Mol. Med. 9, 435–441.
Harper, J.W., and Elledge, S.J. (2007). The DNA damage response: Ten years
after. Mol. Cell 28, 739–745.
Hezel, A.F., and Bardeesy, N. (2008). LKB1; linking cell structure and tumor
suppression. Oncogene 27, 6908–6919.
Hlubek, F., Brabletz, T., Budczies, J., Pfeiffer, S., Jung, A., and Kirchner, T.
(2007). Heterogeneous expression of Wnt/beta-catenin target genes within
colorectal cancer. Int. J. Cancer 121, 1941–1948.
Hugo, H., Ackland, M.L., Blick, T., Lawrence, M.G., Clements, J.A., Williams,
E.D., and Thompson, E.W. (2007). Epithelial—mesenchymal and mesen-
chymal—epithelial transitions in carcinoma progression. J. Cell. Physiol.
213, 374–383.
Hsu, P.P., and Sabatini, D.M. (2008). Cancer cell metabolism: Warburg and
beyond. Cell 134, 703–707.
Hynes, N.E., and MacDonald, G. (2009). ErbB receptors and signaling path-
ways in cancer. Curr. Opin. Cell Biol. 21, 177–184.
Ikushima, H., and Miyazono, K. (2010). TGFbeta signalling: a complex web in
cancer progression. Nat. Rev. Cancer 10, 415–424.
Ince, T.A., Richardson, A.L., Bell, G.W., Saitoh, M., Godar, S., Karnoub, A.E.,
Iglehart, J.D., and Weinberg, R.A. (2007). Transformation of different human
breast epithelial cell types leads to distinct tumor phenotypes. Cancer Cell
12, 160–170.
Jackson, S.P., and Bartek, J. (2009). The DNA-damage response in human
biology and disease. Nature 461, 1071–1078.
Jiang, B.H., and Liu, L.Z. (2009). PI3K/PTEN signaling in angiogenesis and
tumorigenesis. Adv. Cancer Res. 102, 19–65.
Johansson, M., Denardo, D.G., and Coussens, L.M. (2008). Polarized immune
responses differentially regulate cancer development. Immunol. Rev. 222,
145–154.
Jones, P.A., and Baylin, S.B. (2007). The epigenomics of cancer. Cell 128, 683–
692.
Jones, R.G., and Thompson, C.B. (2009). Tumor suppressors and cell metab-
olism: a recipe for cancer growth. Genes Dev. 23, 537–548.
Joyce, J.A., and Pollard, J.W. (2009). Microenvironmental regulation of metas-
tasis. Nat. Rev. Cancer 9, 239–252.
Junttila, M.R., and Evan, G.I. (2009). p53—a Jack of all trades but master of
none. Nat. Rev. Cancer 9, 821–829.
Kalluri, R., and Zeisberg, M. (2006). Fibroblasts in cancer. Nat. Rev. Cancer 6,
392–401.
Kang, H.J., Choi, Y.S., Hong, S.B., Kim, K.W., Woo, R.S., Won, S.J., Kim, E.J.,
Jeon, H.K., Jo, S.Y., Kim, T.K., et al. (2004). Ectopic expression of the catalytic
subunit of telomerase protects against brain injury resulting from ischemia and
NMDA-induced neurotoxicity. J. Neurosci. 24, 1280–1287.
Karin, M., Lawrence, T., and Nizet, V. (2006). Innate immunity gone awry: link-
ing microbial infections to chronic inflammation and cancer. Cell 124, 823–835.
Karnoub, A.E., and Weinberg, R.A. (2006–2007). Chemokine networks and
breast cancer metastasis. Breast Dis. 26, 75–85.
Karnoub, A.E., Dash, A.B., Vo, A.P., Sullivan, A., Brooks, M.W., Bell, G.W.,
Richardson, A.L., Polyak, K., Tubo, R., and Weinberg, R.A. (2007). Mesen-
chymal stem cells within tumour stroma promote breast cancer metastasis.
Nature 449, 557–563.
Kastan, M.B. (2008). DNA damage responses: mechanisms and roles in
human disease: 2007 G.H.A. Clowes Memorial Award Lecture. Mol. Cancer
Res. 6, 517–524.
Kawai, T., Hiroi, S., Nakanishi, K., and Meeker, A.K. (2007). Telomere length
and telomerase expression in atypical adenomatous hyperplasia and small
bronchioloalveolar carcinoma of the lung. Am. J. Clin. Pathol. 127, 254–262.
Kazerounian, S., Yee, K.O., and Lawler, J. (2008). Thrombospondins in cancer.
Cell. Mol. Life Sci. 65, 700–712.
Kenific, C.M., Thorburn, A., and Debnath, J. (2010). Autophagy and metas-
tasis: another double-edged sword. Curr. Opin. Cell Biol. 22, 241–245.
Kennedy, K.M., and Dewhirst, M.W. (2010). Tumor metabolism of lactate: the
influence and therapeutic potential for MCT and CD147 regulation. Future
Oncol. 6, 127–148.
Kessenbrock, K., Plaks, V., and Werb, Z. (2010). Matrix metalloproteinases:
Regulators of the tumor microenvironment. Cell 141, 52–67.
Kim, M.Y., Oskarsson, T., Acharyya, S., Nguyen, D.X., Zhang, X.H., Norton, L.,
and Massagué, J. (2009). Tumor self-seeding by circulating cancer cells. Cell
139, 1315–1326.
Kim, R., Emi, M., and Tanabe, K. (2007). Cancer immunoediting from immune
surveillance to immune escape. Immunology 121, 1–14.
Kinzler, K.W., and Vogelstein, B. (1997). Cancer-susceptibility genes. Gate-
keepers and caretakers. Nature 386, 761–763.
Klein, C.A. (2009). Parallel progression of primary tumours and metastases.
Nat. Rev. Cancer 9, 302–312.
Klymkowsky, M.W., and Savagner, P. (2009). Epithelial-mesenchymal transi-
tion: a cancer researcher’s conceptual friend and foe. Am. J. Pathol. 174,
1588–1593.
Cell 144, March 4, 2011 ª2011 Elsevier Inc. 671
Korkola, J., and Gray, J.W. (2010). Breast cancer genomes—form and func-
tion. Curr. Opin. Genet. Dev. 20, 4–14.
Kovacic, J.C., and Boehm, M. (2009). Resident vascular progenitor cells: an
emerging role for non-terminally differentiated vessel-resident cells in vascular
biology. Stem Cell Res. (Amst.) 2, 2–15.
Kroemer, G., and Pouyssegur, J. (2008). Tumor cell metabolism: Cancer’s
Achilles’ heel. Cancer Cell 13, 472–482.
Lamagna, C., and Bergers, G. (2006). The bone marrow constitutes a reservoir
of pericyte progenitors. J. Leukoc. Biol. 80, 677–681.
Lane, D.P. (1992). Cancer. p53, guardian of the genome. Nature 358, 15–16.
Lemmon, M.A., and Schlessinger, J. (2010). Cell signaling by receptor tyrosine
kinases. Cell 141, 1117–1134.
Levine, B., and Kroemer, G. (2008). Autophagy in the pathogenesis of disease.
Cell 132, 27–42.
Lipinski, M.M., and Jacks, T. (1999). The retinoblastoma gene family in differ-
entiation and development. Oncogene 18, 7873–7882.
Lobo, N.A., Shimono, Y., Qian, D., and Clarke, M.F. (2007). The biology of
cancer stem cells. Annu. Rev. Cell Dev. Biol. 23, 675–699.
Lowe, S.W., Cepero, E., and Evan, G. (2004). Intrinsic tumour suppression.
Nature 432, 307–315.
Luebeck, E.G. (2010). Cancer: Genomic evolution of metastasis. Nature 467,
1053–1055.
Lu, Z., Luo, R.Z., Lu, Y., Zhang, X., Yu, Q., Khare, S., Kondo, S., Kondo, Y., Yu,
Y., Mills, G.B., et al. (2008). The tumor suppressor gene ARHI regulates au-
tophagy and tumor dormancy in human ovarian cancer cells. J. Clin. Invest.
118, 3917–3929.
Luo, J., Solimini, N.L., and Elledge, S.J. (2009). Principles of cancer therapy:
Oncogene and non-oncogene addiction. Cell 136, 823–837.
Mac Gabhann, F., and Popel, A.S. (2008). Systems biology of vascular endo-
thelial growth factors. Microcirculation 15, 715–738.
Madsen, C.D., and Sahai, E. (2010). Cancer dissemination—Lessons from
leukocytes. Dev. Cell 19, 13–26.
Maida, Y., Yasukawa, M., Furuuchi, M., Lassmann, T., Possemato, R., Oka-
moto, N., Kasim, V., Hayashizaki, Y., Hahn, W.C., and Masutomi, K. (2009).
An RNA-dependent RNA polymerase formed by TERT and the RMRP RNA.
Nature 461, 230–235.
Mani, S.A., Guo, W., Liao, M.J., Eaton, E.N., Ayyanan, A., Zhou, A.Y., Brooks, M.,
Reinhard, F., Zhang, C.C., Shipitsin, M., et al. (2008). The epithelial-mesen-
chymal transition generates cells with properties of stem cells. Cell 133, 704–715.
Mantovani, A. (2010). Molecular pathways linking inflammation and cancer.
Curr. Mol. Med. 10, 369–373.
Mantovani, A., Allavena, P., Sica, A., and Balkwill, F. (2008). Cancer-related
inflammation. Nature 454, 436–444.
Massagué, J. (2008). TGFb in cancer. Cell 134, 215–230.
Masutomi, K., Possemato, R., Wong, J.M., Currier, J.L., Tothova, Z., Manola,
J.B., Ganesan, S., Lansdorp, P.M., Collins, K., and Hahn, W.C. (2005). The
telomerase reverse transcriptase regulates chromatin state and DNA damage
responses. Proc. Natl. Acad. Sci. USA 102, 8222–8227.
Mathew, R., Karantza-Wadsworth, V., and White, E. (2007). Role of autophagy
in cancer. Nat. Rev. Cancer 7, 961–967.
McGowan, P.M., Kirstein, J.M., and Chambers, A.F. (2009). Micrometastatic
disease and metastatic outgrowth: clinical issues and experimental
approaches. Future Oncol. 5, 1083–1098.
Micalizzi, D.S., Farabaugh, S.M., and Ford, H.L. (2010). Epithelial-mesen-
chymal transition in cancer: parallels between normal development and tumor
progression. J. Mammary Gland Biol. Neoplasia 15, 117–134.
Mizushima, N. (2007). Autophagy: process and function. Genes Dev. 21, 2861–
2873.
Mohamed, M.M., and Sloane, B.F. (2006). Cysteine cathepsins: multifunc-
tional enzymes in cancer. Nat. Rev. Cancer 6, 764–775.
672 Cell 144, March 4, 2011 ª2011 Elsevier Inc.
Mooi, W.J., and Peeper, D.S. (2006). Oncogene-induced cell senescence—
halting on the road to cancer. N. Engl. J. Med. 355, 1037–1046.
Morel, A.-P., Lièvre, M., Thomas, C., Hinkal, G., Ansieau, S., and Puisieux, A.
(2008). Generation of breast cancer stem cells through epithelial-mesen-
chymal transition. PLoS ONE 3, e2888.
Mosesson, Y., Mills, G.B., and Yarden, Y. (2008). Derailed endocytosis: an
emerging feature of cancer. Nat. Rev. Cancer 8, 835–850.
Mougiakakos, D., Choudhury, A., Lladser, A., Kiessling, R., and Johansson,
C.C. (2010). Regulatory T cells in cancer. Adv. Cancer Res. 107, 57–117.
Murdoch, C., Muthana, M., Coffelt, S.B., and Lewis, C.E. (2008). The role of myeloid
cells in the promotion of tumour angiogenesis. Nat. Rev. Cancer 8, 618–631.
Nagy, J.A., Chang, S.H., Shih, S.C., Dvorak, A.M., and Dvorak, H.F. (2010).
Heterogeneity of the tumor vasculature. Semin. Thromb. Hemost. 36,
321–331.
Naumov, G.N., Folkman, J., Straume, O., and Akslen, L.A. (2008). Tumor-
vascular interactions and tumor dormancy. APMIS 116, 569–585.
Negrini, S., Gorgoulis, V.G., and Halazonetis, T.D. (2010). Genomic insta-
bility—an evolving hallmark of cancer. Nat. Rev. Mol. Cell Biol. 11, 220–228.
Nelson, B.H. (2008). The impact of T-cell immunity on ovarian cancer
outcomes. Immunol. Rev. 222, 101–116.
Nguyen, D.X., Bos, P.D., and Massagué, J. (2009). Metastasis: from dissemi-
nation to organ-specific colonization. Nat. Rev. Cancer 9, 274–284.
Norden, A.D., Drappatz, J., and Wen, P.Y. (2009). Antiangiogenic therapies for
high-grade glioma. Nat. Rev. Neurol. 5, 610–620.
Nyberg, P., Xie, L., and Kalluri, R. (2005). Endogenous inhibitors of angiogen-
esis. Cancer Res. 65, 3967–3979.
Okada, T., Lopez-Lago, M., and Giancotti, F.G. (2005). Merlin/NF-2 mediates
contact inhibition of growth by suppressing recruitment of Rac to the plasma
membrane. J. Cell Biol. 171, 361–371.
Olive, K.P., Jacobetz, M.A., Davidson, C.J., Gopinathan, A., McIntyre, D., Hon-
ess, D., Madhu, B., Goldgraben, M.A., Caldwell, M.E., Allard, D., et al. (2009).
Inhibition of Hedgehog signaling enhances delivery of chemotherapy in
a mouse model of pancreatic cancer. Science 324, 1457–1461.
Olson, P., Lu, J., Zhang, H., Shai, A., Chun, M.G., Wang, Y., Libutti, S.K., Na-
kakura, E.K., Golub, T.R., and Hanahan, D. (2009). MicroRNA dynamics in the
stages of tumorigenesis correlate with hallmark capabilities of cancer. Genes
Dev. 23, 2152–2165.
O’Reilly, K.E., Rojo, F., She, Q.B., Solit, D., Mills, G.B., Smith, D., Lane, H., Hof-
mann, F., Hicklin, D.J., Ludwig, D.L., et al. (2006). mTOR inhibition induces
upstream receptor tyrosine kinase signaling and activates Akt. Cancer Res.
66, 1500–1508.
Ostrand-Rosenberg, S., and Sinha, P. (2009). Myeloid-derived suppressor
cells: linking inflammation and cancer. J. Immunol. 182, 4499–4506.
Pagès, F., Galon, J., Dieu-Nosjean, M.C., Tartour, E., Sautès-Fridman, C., and
Fridman, W.H. (2010). Immune infiltration in human tumors: a prognostic factor
that should not be ignored. Oncogene 29, 1093–1102.
Palermo, C., and Joyce, J.A. (2008). Cysteine cathepsin proteases as pharma-
cological targets in cancer. Trends Pharmacol. Sci. 29, 22–28.
Park, J.I., Venteicher, A.S., Hong, J.Y., Choi, J., Jun, S., Shkreli, M., Chang, W.,
Meng, Z., Cheung, P., Ji, H., et al. (2009). Telomerase modulates Wnt signalling
by association with target gene chromatin. Nature 460, 66–72.
Partanen, J.I., Nieminen, A.I., and Klefstrom, J. (2009). 3D view to tumor
suppression: Lkb1, polarity and the arrest of oncogenic c-Myc. Cell Cycle 8,
716–724.
Pasquale, E.B. (2010). Eph receptors and ephrins in cancer: bidirectional sig-
nalling and beyond. Nat. Rev. Cancer 10, 165–180.
Passos, J.F., Saretzki, G., and von Zglinicki, T. (2007). DNA damage in telo-
meres and mitochondria during cellular senescence: is there a connection?
Nucleic Acids Res. 35, 7505–7513.
Patenaude, A., Parker, J., and Karsan, A. (2010). Involvement of endothelial
progenitor cells in tumor vascularization. Microvasc. Res. 79, 217–223.
Peinado, H., Lavothskin, S., and Lyden, D. (2011). The secreted factors
responsible for pre-metastatic niche formation: Old sayings and new thoughts.
Semin. Cancer Biol. Published online January 18, 2011. 10.1016/j.semcancer.
2011.01.002.
Peinado, H., Marin, F., Cubillo, E., Stark, H.J., Fusenig, N., Nieto, M.A., and
Cano, A. (2004). Snail and E47 repressors of E-cadherin induce distinct inva-
sive and angiogenic properties in vivo. J. Cell Sci. 117, 2827–2839.
Perona, R. (2006). Cell signalling: growth factors and tyrosine kinase recep-
tors. Clin. Transl. Oncol. 8, 77–82.
Pietras, K., and Ostman, A. (2010). Hallmarks of cancer: interactions with the
tumor stroma. Exp. Cell Res. 316, 1324–1331.
Polyak, K., and Weinberg, R.A. (2009). Transitions between epithelial and
mesenchymal states: acquisition of malignant and stem cell traits. Nat. Rev.
Cancer 9, 265–273.
Polyak, K., Haviv, I., and Campbell, I.G. (2009). Co-evolution of tumor cells and
their microenvironment. Trends Genet. 25, 30–38.
Potter, V.R. (1958). The biochemical approach to the cancer problem. Fed.
Proc. 17, 691–697.
Qian, B.Z., and Pollard, J.W. (2010). Macrophage diversity enhances tumor
progression and metastasis. Cell 141, 39–51.
Quintana, E., Shackleton, M., Sabel, M.S., Fullen, D.R., Johnson, T.M., and
Morrison, S.J. (2008). Efficient tumour formation by single human melanoma
cells. Nature 456, 593–598.
Raica, M., Cimpean, A.M., and Ribatti, D. (2009). Angiogenesis in pre-malig-
nant conditions. Eur. J. Cancer 45, 1924–1934.
Räsänen, K., and Vaheri, A. (2010). Activation of fibroblasts in cancer stroma.
Exp. Cell Res. 316, 2713–2722.
Raynaud, C.M., Hernandez, J., Llorca, F.P., Nuciforo, P., Mathieu, M.C.,
Commo, F., Delaloge, S., Sabatier, L., André, F., and Soria, J.C. (2010). DNA
damage repair and telomere length in normal breast, preneoplastic lesions,
and invasive cancer. Am. J. Clin. Oncol. 33, 341–345.
Raza, A., Franklin, M.J., and Dudek, A.Z. (2010). Pericytes and vessel matura-
tion during tumor angiogenesis and metastasis. Am. J. Hematol. 85, 593–598.
Reitman, Z.J., and Yan, H. (2010). Isocitrate dehydrogenase 1 and 2 mutations
in cancer: alterations at a crossroads of cellular metabolism. J. Natl. Cancer
Inst. 102, 932–941.
Reya, T., Morrison, S.J., Clarke, M.F., and Weissman, I.L. (2001). Stem cells,
cancer, and cancer stem cells. Nature 414, 105–111.
Ribatti, D. (2009). Endogenous inhibitors of angiogenesis: a historical review.
Leuk. Res. 33, 638–644.
Ricci-Vitiani, L., Pallini, R., Biffoni, M., Todaro, M., Invernici, G., Cenci, T.,
Maira, G., Parati, E.A., Stassi, G., Larocca, L.M., and De Maria, R. (2010).
Tumour vascularization via endothelial differentiation of glioblastoma stem-
like cells. Nature 468, 824–828.
Ruoslahti, E. (2002). Specialization of tumour vasculature. Nat. Rev. Cancer 2,
83–90.
Ruoslahti, E., Bhatia, S.N., and Sailor, M.J. (2010). Targeting of drugs and
nanoparticles to tumors. J. Cell Biol. 188, 759–768.
Sabeh, F., Shimizu-Hirota, R., and Weiss, S.J. (2009). Protease-dependent
versus -independent cancer cell invasion programs: three-dimensional amoe-
boid movement revisited. J. Cell Biol. 185, 11–19.
Salk, J.J., Fox, E.J., and Loeb, L.A. (2010). Mutational heterogeneity in human
cancers: origin and consequences. Ann. Rev. Pathol. 5, 51–75.
Schäfer, M., and Werner, S. (2008). Cancer as an overhealing wound: an old
hypothesis revisited. Nat. Rev. Mol. Cell Biol. 9, 628–638.
Schmalhofer, O., Brabletz, S., and Brabletz, T. (2009). E-cadherin, beta-cate-
nin, and ZEB1 in malignant progression of cancer. Cancer Metastasis Rev. 28,
151–166.
Semenza, G.L. (2008). Tumor metabolism: cancer cells give and take lactate.
J. Clin. Invest. 118, 3835–3837.
Semenza, G.L. (2010a). HIF-1: upstream and downstream of cancer metabo-
lism. Curr. Opin. Genet. Dev. 20, 51–56.
Semenza, G.L. (2010b). Defining the role of hypoxia-inducible factor 1 in
cancer biology and therapeutics. Oncogene 29, 625–634.
Seppinen, L., Sormunen, R., Soini, Y., Elamaa, H., Heljasvaara, R., and Pihla-
janiemi, T. (2008). Lack of collagen XVIII accelerates cutaneous wound healing,
while overexpression of its endostatin domain leads to delayed healing. Matrix
Biol. 27, 535–546.
Shaw, R.J. (2009). Tumor suppression by LKB1: SIK-ness prevents metas-
tasis. Sci. Signal. 2, pe55.
Shay, J.W., and Wright, W.E. (2000). Hayflick, his limit, and cellular ageing. Nat.
Rev. Mol. Cell Biol. 1, 72–76.
Sherr, C.J., and DePinho, R.A. (2000). Cellular senescence: Mitotic clock or
culture shock? Cell 102, 407–410.
Sherr, C.J., and McCormick, F. (2002). The RB and p53 pathways in cancer.
Cancer Cell 2, 103–112.
Shields, J.D., Kourtis, I.C., Tomei, A.A., Roberts, J.M., and Swartz, M.A.
(2010). Induction of lymphoidlike stroma and immune escape by tumors that
express the chemokine CCL21. Science 328, 749–752.
Shimoda, M., Mellody, K.T., and Orimo, A. (2010). Carcinoma-associated
fibroblasts are a rate-limiting determinant for tumour progression. Semin.
Cell Dev. Biol. 21, 19–25.
Sigal, A., and Rotter, V. (2000). Oncogenic mutations of the p53 tumor
suppressor: the demons of the guardian of the genome. Cancer Res. 60,
6788–6793.
Singh, A., and Settleman, J. (2010). EMT, cancer stem cells and drug resis-
tance: an emerging axis of evil in the war on cancer. Oncogene 29, 4741–4751.
Sinha, S., and Levine, B. (2008). The autophagy effector Beclin 1: a novel BH3-
only protein. Oncogene 27 (Suppl 1), S137–S148.
Smyth, M.J., Dunn, G.P., and Schreiber, R.D. (2006). Cancer immunosurveil-
lance and immunoediting: the roles of immunity in suppressing tumor develop-
ment and shaping tumor immunogenicity. Adv. Immunol. 90, 1–50.
Soda, Y., Marumoto, T., Friedmann-Morvinski, D., Soda, M., Liu, F., Michiue,
H., Pastorino, S., Yang, M., Hoffman, R.M., Kesari, S., and Verma, I.M. (2011).
Feature Article: Transdifferentiation of glioblastoma cells into vascular endo-
thelial cells. Proc. Natl. Acad. Sci. USA. Published online January 24, 2011.
Sudarsanam, S., and Johnson, D.E. (2010). Functional consequences of
mTOR inhibition. Curr. Opin. Drug Discov. Devel. 13, 31–40.
Strauss, D.C., and Thomas, J.M. (2010). Transmission of donor melanoma by
organ transplantation. Lancet Oncol. 11, 790–796.
Talmadge, J.E., and Fidler, I.J. (2010). AACR centennial series: the biology of
cancer metastasis: historical perspective. Cancer Res. 70, 5649–5669.
Tammela, T., and Alitalo, K. (2010). Lymphangiogenesis: Molecular mecha-
nisms and future promise. Cell 140, 460–476.
Taube, J.H., Herschkowitz, J.I., Komurov, K., Zhou, A.Y., Gupta, S., Yang, J.,
Hartwell, K., Onder, T.T., Gupta, P.B., Evans, K.W., et al. (2010). Core epithe-
lial-to-mesenchymal transition interactome gene-expression signature is
associated with claudin-low and metaplastic breast cancer subtypes. Proc.
Natl. Acad. Sci. USA 107, 15449–15454.
Teng, M.W.L., Swann, J.B., Koebel, C.M., Schreiber, R.D., and Smyth, M.J.
(2008). Immune-mediated dormancy: an equilibrium with cancer. J. Leukoc.
Biol. 84, 988–993.
Thiery, J.P., and Sleeman, J.P. (2006). Complex networks orchestrate epithe-
lial-mesenchymal transitions. Nat. Rev. Mol. Cell Biol. 7, 131–142.
Thiery, J.P., Acloque, H., Huang, R.Y., and Nieto, M.A. (2009). Epithelial-
mesenchymal transitions in development and disease. Cell 139, 871–890.
Townson, J.L., and Chambers, A.F. (2006). Dormancy of solitary metastatic
cells. Cell Cycle 5, 1744–1750.
Turner, H.E., Harris, A.L., Melmed, S., and Wass, J.A. (2003). Angiogenesis in
endocrine tumors. Endocr. Rev. 24, 600–632.
Vajdic, C.M., and van Leeuwen, M.T. (2009). Cancer incidence and risk factors
after solid organ transplantation. Int. J. Cancer 125, 1747–1754.
Cell 144, March 4, 2011 ª2011 Elsevier Inc. 673
Vander Heiden, M.G., Cantley, L.C., and Thompson, C.B. (2009). Under-
standing the Warburg effect: the metabolic requirements of cell proliferation.
Science 324, 1029–1033.
Verhoeff, J.J., van Tellingen, O., Claes, A., Stalpers, L.J., van Linde, M.E., Richel,
D.J., Leenders, W.P., and van Furth, W.R. (2009). Concerns about anti-angio-
genic treatment in patients with glioblastoma multiforme. BMC Cancer 9, 444.
Wang, R., Chadalavada, K., Wilshire, J., Kowalik, U., Hovinga, K.E., Geber, A.,
Fligelman, B., Leversha, M., Brennan, C., and Tabar, V. (2010). Glioblastoma
stem-like cells give rise to tumour endothelium. Nature 468, 829–833.
Warburg, O. (1956a). On the origin of cancer cells. Science 123, 309–314.
Warburg, O. (1956b). On respiratory impairment in cancer cells. Science 124,
269–270.
Warburg, O.H. (1930). The Metabolism of Tumours: Investigations from the
Kaiser Wilhelm Institute for Biology, Berlin-Dahlem (London, UK: Arnold
Constable).
Wertz, I.E., and Dixit, V.M. (2010). Regulation of death receptor signaling by the
ubiquitin system. Cell Death Differ. 17, 14–24.
White, E., Karp, C., Strohecker, A.M., Guo, Y., and Mathew, R. (2010). Role of
autophagy in suppression of inflammation and cancer. Curr. Opin. Cell Biol.
22, 212–217.
White, E., and DiPaola, R.S. (2009). The double-edged sword of autophagy
modulation in cancer. Clin. Cancer Res. 15, 5308–5316.
Willis, S.N., and Adams, J.M. (2005). Life in the balance: how BH3-only
proteins induce apoptosis. Curr. Opin. Cell Biol. 17, 617–625.
Witsch, E., Sela, M., and Yarden, Y. (2010). Roles for growth factors in cancer
progression. Physiology (Bethesda) 25, 85–101.
Wyckoff, J.B., Wang, Y., Lin, E.Y., Li, J.F., Goswami, S., Stanley, E.R., Segall,
J.E., Pollard, J.W., and Condeelis, J. (2007). Direct visualization of macro-
674 Cell 144, March 4, 2011 ª2011 Elsevier Inc.
phage-assisted tumor cell intravasation in mammary tumors. Cancer Res.
67, 2649–2656.
Yachida, S., Jones, S., Bozic, I., Antal, T., Leary, R., Fu, B., Kamiyama, M., Hru-
ban, R.H., Eshleman, J.R., Nowak, M.A., et al. (2010). Distant metastasis
occurs late during the genetic evolution of pancreatic cancer. Nature 467,
1114–1117.
Yang, J., and Weinberg, R.A. (2008). Epithelial-mesenchymal transition: At the
crossroads of development and tumor metastasis. Dev. Cell 14, 818–829.
Yang, L., Pang, Y., and Moses, H.L. (2010). TGF-beta and immune cells: an
important regulatory axis in the tumor microenvironment and progression.
Trends Immunol. 31, 220–227.
Yen, K.E., Bittinger, M.A., Su, S.M., and Fantin, V.R. (2010). Cancer-associ-
ated IDH mutations: biomarker and therapeutic opportunities. Oncogene 29,
6409–6417.
Yilmaz, M., and Christofori, G. (2009). EMT, the cytoskeleton, and cancer cell
invasion. Cancer Metastasis Rev. 28, 15–33.
Yuan, T.L., and Cantley, L.C. (2008). PI3K pathway alterations in cancer: vari-
ations on a theme. Oncogene 27, 5497–5510.
Zee, Y.K., O’Connor, J.P., Parker, G.J., Jackson, A., Clamp, A.R., Taylor, M.B.,
Clarke, N.W., and Jayson, G.C. (2010). Imaging angiogenesis of genitourinary
tumors. Nat. Rev. Urol. 7, 69–82.
Zhang, H., Herbert, B.S., Pan, K.H., Shay, J.W., and Cohen, S.N. (2004).
Disparate effects of telomere attrition on gene expression during replicative
senescence of human mammary epithelial cells cultured under different condi-
tions. Oncogene 23, 6193–6198.
Zong, W.X., and Thompson, C.B. (2006). Necrotic death as a cell fate. Genes
Dev. 20, 1–15.
Zumsteg, A., and Christofori, G. (2009). Corrupt policemen: inflammatory cells
promote tumor angiogenesis. Curr. Opin. Oncol. 21, 60–70.
REVIEW

Hallmarks of Cancer: New Dimensions

Douglas Hanahan

ABSTRACT The hallmarks of cancer conceptualization is a heuristic tool for distilling the vast
complexity of cancer phenotypes and genotypes into a provisional set of underly-
ing principles. As knowledge of cancer mechanisms has progressed, other facets of the disease have
emerged as potential refinements. Herein, the prospect is raised that phenotypic plasticity and dis-
rupted differentiation is a discrete hallmark capability, and that nonmutational epigenetic reprogram-
ming and polymorphic microbiomes both constitute distinctive enabling characteristics that facilitate
the acquisition of hallmark capabilities. Additionally, senescent cells, of varying origins, may be added
to the roster of functionally important cell types in the tumor microenvironment.

Significance: Cancer is daunting in the breadth and scope of its diversity, spanning genetics, cell and
tissue biology, pathology, and response to therapy. Ever more powerful experimental and computa-
tional tools and technologies are providing an avalanche of “big data” about the myriad manifestations
of the diseases that cancer encompasses. The integrative concept embodied in the hallmarks of cancer
is helping to distill this complexity into an increasingly logical science, and the provisional new dimen-
sions presented in this perspective may add value to that endeavor, to more fully understand mecha-
nisms of cancer development and malignant progression, and apply that knowledge to cancer medicine.

INTRODUCTION steps of tumor pathogenesis. Certainly, the diversity of malig-

The Hallmarks of Cancer were proposed as a set of func-
tional capabilities acquired by human cells as they make
their way from normalcy to neoplastic growth states, more
specifically capabilities that are crucial for their ability to
form malignant tumors. In these articles (1, 2), Bob Weinberg
and I enumerated what we imagined were shared commonali-
ties that unite all types of cancer cells at the level of cellular
phenotype. The intent was to provide a conceptual scaffold
that would make it possible to rationalize the complex phe-
notypes of diverse human tumor types and variants in terms
of a common set of underlying cellular parameters. Initially
we envisaged the complementary involvement of six distinct
hallmark capabilities and later expanded this number to
eight. This formulation was influenced by the recognition
that human cancers develop as products of multistep pro-
cesses, and that the acquisition of these functional capabili-
ties might be mapped in some fashion to the distinguishable
Ludwig Institute for Cancer Research – Lausanne Branch, Lausanne, Swit-
zerland. The Swiss Institute for Experimental Cancer Research (ISREC)
within the School of Life Sciences at the Swiss Federal Institute of Tech-
nology Lausanne (EPFL), Lausanne, Switzerland. The Swiss Cancer Center
Leman (SCCL), Lausanne, Switzerland.
Corresponding Author: Douglas Hanahan, Agora Translational Cancer
Research Center, Rue du Bugnon 25A, Lausanne CH-1011, Switzerland.
Phone: 41-21-545-1119; E-mail: douglas.hanahan@epfl.ch
Cancer Discov 2022;12:31–46
doi: 10.1158/2159-8290.CD-21-1059
©2021 American Association for Cancer Research
nant pathogenesis spanning multiple tumor types and an
increasing plethora of subtypes includes various aberrations
(and hence acquired capabilities and characteristics) that are
the result of tissue-specific barriers necessarily circumvented
during particular tumorigenesis pathways. While appreciat-
ing that such specialized mechanisms can be instrumental,
we limited the hallmarks designation to parameters having
broad engagement across the spectrum of human cancers.
The eight hallmarks currently comprise (Fig.  1, left) the
acquired capabilities for sustaining proliferative signaling,
evading growth suppressors, resisting cell death, enabling
replicative immortality, inducing/accessing vasculature,
activating invasion and metastasis, reprogramming cellular
metabolism, and avoiding immune destruction. In the most
recent elaboration of this concept (2), deregulating cellular
metabolism and avoiding immune destruction were segre-
gated as “emerging hallmarks,” but now, eleven years later, it
is evident that they, much like the original six, can be consid-
ered core hallmarks of cancer, and are included as such in the
current depiction (Fig. 1, left).
As we noted at the time, these hallmark traits, on their
own, fail to address the complexities of cancer pathogenesis,
that is, the precise molecular and cellular mechanisms that
allow evolving preneoplastic cells to develop and acquire
these aberrant phenotypic capabilities in the course of tumor
development and malignant progression. Accordingly, we
added another concept to the discussion, portrayed as “ena-
bling characteristics,” consequences of the aberrant condi-
tion of neoplasia that provide means by which cancer cells
and tumors can adopt these functional traits. As such, the

JANUARY 2022 CANCER DISCOVERY | 31

Downloaded from cancerdiscovery.aacrjournals.org on January 12, 2022. © 2022 American Association for Cancer
Research.
REVIEW Hanahan

Sustaining Evading Emerging hallmarks &

proliferative signaling growth suppressors enabling characteristics
Unlocking Nonmutational
phenotypic epigenetic
Deregulating Avoiding plasticity reprogramming
cellular immune
metabolism destruction

Resisting Enabling
cell death replicative
immortality

Genome
instability & Tumor-promoting
mutation inflammation
Senescent Polymorphic
cells microbiomes
Inducing or accessing Activating invasion
vasculature & metastasis

Figure 1.  In essence: the Hallmarks of Cancer, circa 2022. Left, the Hallmarks of Cancer currently embody eight hallmark capabilities and two enabling
characteristics. In addition to the six acquired capabilities—Hallmarks of Cancer—proposed in 2000 (1), the two provisional “emerging hallmarks” intro-
duced in 2011 (2)—cellular energetics (now described more broadly as “reprogramming cellular metabolism”) and “avoiding immune destruction”—have
been sufficiently validated to be considered part of the core set. Given the growing appreciation that tumors can become sufficiently vascularized either
by switching on angiogenesis or by co-opting normal tissue vessels (128), this hallmark is also more broadly defined as the capability to induce or oth-
erwise access, principally by invasion and metastasis, vasculature that supports tumor growth. The 2011 sequel further incorporated “tumor-promoting
inflammation” as a second enabling characteristic, complementing overarching “genome instability and mutation,” which together were fundamentally
involved in activating the eight hallmark (functional) capabilities necessary for tumor growth and progression. Right, this review incorporates additional
proposed emerging hallmarks and enabling characteristics involving “unlocking phenotypic plasticity,” “nonmutational epigenetic reprogramming,” “poly-
morphic microbiomes,” and “senescent cells.”

enabling characteristics reflected upon molecular and cel-
lular mechanisms by which hallmarks are acquired rather
than the aforementioned eight capabilities themselves.
These two enabling processes were genome instability and
tumor-promoting inflammation.
We further recognized that the tumor microenvironment
(TME), herein defined to be composed of heterogeneous and
interactive populations of cancer cells and cancer stem cells
along with a multiplicity of recruited stromal cell types—the
transformed parenchyma and the associated stroma—is now
widely appreciated to play an integral role in tumorigenesis
and malignant progression.
Given the continued interest in these formulations and our
enduring intent to encourage ongoing discussion and refine-
ment of the Hallmarks scheme, it is appropriate to consider
a frequently posed question: are there additional features of
this conceptual model that might be incorporated, respecting
the need to ensure that they are broadly applicable across the
spectrum of human cancers? Accordingly, I present several
prospective new hallmarks and enabling characteristics, ones
that might in due course become incorporated as core com-
ponents of the hallmarks of cancer conceptualization. These
parameters are “unlocking phenotypic plasticity,” “nonmu-
tational epigenetic reprogramming,” “polymorphic microbi-
omes,” and “senescent cells” (Fig. 1, right). Importantly, the
examples presented in support of these propositions are illus-
trative but by no means comprehensive, as there is a growing
and increasingly persuasive body of published evidence in
support of each vignette.
UNLOCKING PHENOTYPIC PLASTICITY
During organogenesis, the development, determination,
and organization of cells into tissues in order to assume
homeostatic functions is accompanied by terminal differen-
tiation, whereby progenitor cells—sometimes irrevocably—
stop growing upon culmination of these processes. As such,
the end result of cellular differentiation is in most cases
antiproliferative and constitutes a clear barrier to the con-
tinuing proliferation that is necessary for neoplasia. There
is increasing evidence that unlocking the normally restricted
capability for phenotypic plasticity in order to evade or
escape from the state of terminal differentiation is a criti-
cal component of cancer pathogenesis (3). This plasticity
can operate in several manifestations (Fig. 2). Thus, nascent
cancer cells originating from a normal cell that had advanced
down a pathway approaching or assuming a fully differenti-
ated state may reverse their course by dedifferentiating back
to progenitor-like cell states. Conversely, neoplastic cells
arising from a progenitor cell that is destined to follow a
pathway leading to end-stage differentiation may short-
circuit the process, maintaining the expanding cancer cells in
a partially differentiated, progenitor-like state. Alternatively,
transdifferentiation may operate, in which cells that were
initially committed into one differentiation pathway switch
to an entirely different developmental program, thereby
acquiring tissue-specific traits that were not preordained by
their normal cells-of-origin. The following examples sup-
port the argument that differing forms of cellular plasticity,

32 | CANCER DISCOVERY JANUARY 2022 AACRJournals.org

Downloaded from cancerdiscovery.aacrjournals.org on January 12, 2022. © 2022 American Association for Cancer
Research.
Hallmarks of Cancer: New Dimensions
Unlocking
phenotypic
plasticity
Figure 2.  Unlocking phenotypic plasticity. Left, phenotypic plasticity is arguably an acquired hallmark capability that enables various disruptions of
cellular differentiation, including (i) dedifferentiation from mature to progenitor states, (ii) blocked (terminal) differentiation from progenitor cell states,
and (iii) transdifferentiation into different cell lineages. Right, depicted are three prominent modes of disrupted differentiation integral to cancer patho-
genesis. By variously corrupting the normal differentiation of progenitor cells into mature cells in developmental lineages, tumorigenesis and malignant
progression arising from cells of origin in such pathways is facilitated.
when taken together, constitute a functionally distinct
hallmark capability.
Dedifferentiation
Colon carcinogenesis exemplifies disrupted differentiation,
in that there is a teleological necessity for incipient cancer
cells to escape from the conveyer belt of terminal differen-
tiation and exfoliation, which could in principle occur via
dedifferentiation of not yet irrevocably terminally differenti-
ated colonic epithelial cells, or via blocked differentiation
of progenitor/stem cells in the crypts that spawn these dif-
ferentiating cells. Both differentiated cells and stem cells
have been implicated as cell-of-origin for colon cancer (4–6).
Two developmental transcription factors (TF), the homeobox
protein HOXA5 and SMAD4, the latter involved in BMP
signal transmission, are highly expressed in differentiating
colonic epithelial cells, and typically lost in advanced colon
carcinomas, which characteristically express markers of stem
and progenitor cells. Functional perturbations in mouse
models have shown that forced expression of HOXA5 in
colon cancer cells restores differentiation markers, suppresses
stem cell phenotypes, and impairs invasion and metastasis,
providing a rationale for its characteristic downregulation
(7, 8). SMAD4, by contrast, both enforces differentiation and
thereby suppresses proliferation driven by oncogenic WNT
signaling, revealed by the engineered loss of SMAD4 expres-
sion, providing an explanation for its loss of expression so as
to enable dedifferentiation and, subsequently, WNT-driven
hyperproliferation (5). Notably, the loss of both of these “dif-
ferentiation suppressors” with consequent dedifferentiation
is associated with acquisition of other hallmark capabilities,
as are other hallmark-inducing regulators, which complicates
Downloaded from cancerdiscovery.aacrjournals.org on January 12, 2022. © 2022 American Association for Cancer
Research.
REVIEW
Progenitor cell Differentiated cell
Normal differentiation
Dedifferentiation
Blocked
differentiation
Transdifferentiation
the strict definition of this provisional hallmark as separable
and independent.
Another line of evidence involves suppressed expression
of the MITF master regulator of melanocyte differentiation,
which is evidently involved in the genesis of aggressive forms
of malignant melanoma. Loss of this developmental TF is
associated with the reactivation of neural crest progenitor
genes and the downregulation of genes that characterize
fully differentiated melanocytes. The reappearance of the
neural crest genes indicates that these cells revert to the pro-
genitor state from which melanocytes arise developmentally.
Moreover, a lineage tracing study of BRAF-induced melano-
mas established mature pigmented melanocytes as the cells
of origin, which undergo dedifferentiation during the course
of tumorigenesis (9). Of note, the mutant BRAF oncogene,
which is found in more than half of cutaneous melanomas,
induces hyperproliferation that precedes and hence is mecha-
nistically separable from the subsequent dedifferentiation
arising from downregulation of MITF. Another study func-
tionally implicated upregulation of the developmental TF
ATF2, whose characteristic expression in mouse and human
melanomas indirectly suppresses MITF1, concomitant with
malignant progression of the consequently dedifferentiated
melanoma cells (10). Conversely, expression in melanomas
of mutant forms of ATF2 that fail to repress MITF results in
well-differentiated melanomas (11).
Additionally, a recent study (12) has associated lineage
dedifferentiation with malignant progression from pancre-
atic islet cell neoplasias into metastasis-prone carcinomas;
these neuroendocrine cells and derivative tumors arise from
a developmental lineage that is distinct from the one gen-
erating the far larger number of adjacent cells that form
the exocrine and pancreas and the ductal adenocarcinomas
JANUARY 2022 CANCER DISCOVERY | 33
REVIEW
that arise therefrom. Notably, the multistep differentiation
pathway of islet progenitor cells into mature  β cells has been
thoroughly characterized (13). Comparative transcriptome
profiling reveals that adenoma-like islet tumors are most sim-
ilar to immature but differentiated insulin-producing β cells,
whereas the invasive carcinomas are most similar to embry-
onic islet cell precursors. The progression toward poorly
differentiated carcinomas involves a first step of dedifferen-
tiation that does not initially involve increased proliferation
or reduced apoptosis when compared with the well-differen-
tiated adenomas, both of which rather occur later. Thus, the
discrete step of dedifferentiation is not driven by observable
alterations in the hallmark traits of sustained proliferation
and resistance to apoptosis. Rather, upregulation of a miRNA
previously implicated in specifying the islet progenitor state,
one that is downregulated during terminal differentiation
of  β cells, has been shown to orchestrate the observed dedif-
ferentiation occurring during malignant progression (12).
Blocked Differentiation
While the above examples illustrate how suppression of
differentiation factor expression can facilitate tumorigenesis
by enabling more well-differentiated cells to dedifferentiate
into progenitors, in other cases incompletely differentiated
progenitor cells can suffer regulatory changes that actively
block their continued advance into fully differentiated, typi-
cally nonproliferative states.
Acute promyelocytic leukemia (APL) has long been docu-
mented to result from a chromosomal translocation that
fuses the PML locus with the gene encoding the retinoic
acid  α  nuclear receptor (RARα). Myeloid progenitor cells
bearing such translocations are evidently unable to continue
their usual terminal differentiation into granulocytes, result-
ing in cells trapped in a proliferative, promyelocytic progeni-
tor stage (14). Proof-of-concept of this scheme comes from
treating cultured APL cells, mouse models of this disease,
as well as afflicted patients, with retinoic acid, the ligand of
RARα; this therapeutic treatment causes the neoplastic APL
cells to differentiate into ostensibly mature nonproliferating
granulocytes, short-circuiting their continuing proliferative
expansion (14–16).
A variation on this theme involves another form of acute
myeloid leukemia, this one carrying the t(8;21) translocation,
which produces the AML1–ETO fusion protein. This protein
can, on its own, transform myeloid progenitors, at least in
part by blocking their differentiation. Therapeutic interven-
tion in mouse models and in patients with a pharmaco-
logic inhibitor of a chromatin-modifying histone deacetylase
(HDAC) causes the myeloid leukemia cells to recommence
their differentiation into cells with a more mature myeloid
cell morphology. Concomitant with this response is a reduc-
tion in proliferative capacity, thereby impairing the progres-
sion of this leukemia (17, 18).
A third example, in melanoma, involves a developmental
TF, SOX10, which is normally downregulated during mel-
anocyte differentiation. Gain- and loss-of-function studies
in a zebrafish model of BRAF-induced melanoma have dem-
onstrated that aberrantly maintained expression of SOX10
blocks differentiation of neural progenitor cells into melano-
cytes, enabling BRAF-driven melanomas to form (19).
34 | CANCER DISCOVERY JANUARY 2022 AACRJournals.org
Downloaded from cancerdiscovery.aacrjournals.org on January 12, 2022. © 2022 American Association for Cancer
Research.
Hanahan
Other examples of differentiation modulators involve the
metabolite alpha-ketoglutarate (αKG), a necessary cofactor
for a number of chromatin-modifying enzymes, which is
demonstrably involved in stimulating certain differentiated
cell states. In pancreas cancer, the tumor suppressor p53
stimulates the production of αKG and maintenance of a more
well-differentiated cell state, whereas prototypical loss of p53
function results in reductions in  αKG levels and consequent
dedifferentiation associated with malignant progression (20).
In one form of liver cancer, mutation of an isocitrate dehy-
drogenase gene (IDH1/2) results in the production not of
differentiation-inducing  αKG but rather a related “onco-
metabolite,” D-2-hydroxygluterate (D2HG), which has been
shown to block hepatocyte differentiation from liver progeni-
tor cells by D2HG-mediated repression of a master regulator
of hepatocyte differentiation and quiescence, HNF4a. The
D2HG-mediated suppression of HNF4a function elicits a
proliferative expansion of the hepatocyte progenitor cells in
the liver, which become susceptible to oncogenic transforma-
tion upon subsequent mutational activation of the KRAS
oncogene that drives malignant progression to liver cholan-
giocarcinoma (21). Mutant IDH1/2 and their oncometabolite
D2HG are also operative in a variety of myeloid and other
solid tumor types, where D2HG inhibits  αKG-dependent
dioxygenases necessary for histone and DNA methylation
events that mediate alterations in chromatin structure dur-
ing developmental lineage differentiation, thereby freezing
incipient cancer cells in a progenitor state (22, 23).
An additional, related concept is “circumvented differen-
tiation,” wherein partially or undifferentiated progenitor/
stem cells exit the cell cycle and become dormant, residing in
protective niches, with the potential to reinitiate proliferative
expansion (24), albeit still with the selective pressure to dis-
rupt their programmed differentiation in one way or another.
Transdifferentiation
The concept of transdifferentiation has long been rec-
ognized by pathologists in the form of tissue metaplasia,
wherein cells of a particular differentiated phenotype mark-
edly change their morphology to become clearly recognizable
as elements of another tissue, of which one prominent exam-
ple is Barrett’s esophagus, where chronic inflammation of
the stratified squamous epithelium of the esophagus induces
transdifferentiation into a simple columnar epithelium that
is characteristic of the intestine, thereby facilitating the sub-
sequent development of adenocarcinomas, and not the squa-
mous cell carcinomas that would be anticipated to arise from
this squamous epithelium (3). Now, molecular determinants
are revealing mechanisms of transdifferentiation in various
cancers, both for cases where gross tissue metaplasia is evi-
dent and for others where it is rather more subtle, as the
following examples illustrate.
One illuminating case for transdifferentiation as a dis-
crete event in tumorigenesis involves pancreatic ductal ade-
nocarcinoma (PDAC), wherein one of the implicated cells
of origin, the pancreatic acinar cell, can become transdiffer-
entiated into a ductal cell phenotype during the initiation
of neoplastic development. Two TFs—PTF1a and MIST1—
govern, via their expression in the context of self-sustaining,
“feed-forward” regulatory loops, the specification and
Hallmarks of Cancer: New Dimensions
maintenance of the differentiated pancreatic acinar cell
state (25). Both of these TFs are frequently downregulated
during neoplastic development and malignant progression
of human and mouse PDAC. Functional genetic studies in
mice and cultured human PDAC cells have demonstrated
that experimentally forced expression of PTF1a impairs
KRAS-induced transdifferentiation and proliferation, and
can also force the redifferentiation of already neoplastic
cells into a quiescent acinar cell phenotype (26). Conversely,
suppression of PTF1a expression elicits acinar-to-ductal
metaplasia, namely transdifferentiation, and thereby sensi-
tizes the duct-like cells to oncogenic KRAS transformation,
accelerating subsequent development of invasive PDAC
(27). Similarly, forced expression of MIST1 in KRAS-express-
ing pancreas also blocks transdifferentiation and impairs
the initiation of pancreatic tumorigenesis otherwise facili-
tated by the formation of premalignant duct-like (PanIN)
lesions, whereas genetic deletion of MIST1 enhances their
formation and the initiation of KRAS-driven neoplastic
progression (28). Loss of either PTF1 or MIST1 expression
during tumorigenesis is associated with elevated expression
of another developmental regulatory TF, SOX9, which is
normally operative in the specification of ductal cells (27,
28). Forced upregulation of SOX9, obviating the need to
downregulate PTF1a and MIST1, has also been shown to
stimulate transdifferentiation of acinar cells into a ductal
cell phenotype that is sensitive to KRAS-induced neoplasia
(29), implicating SOX9 as a key functional effector of their
downregulation in the genesis of human PDAC. Thus,
three TFs that regulate pancreatic differentiation can be
variously altered to induce a transdifferentiated state that
facilitates—in the context of mutational activation of KRAS—
oncogenic transformation and the initiation of tumorigen-
esis and malignant progression.
Additional members of the SOX family of chromatin-
associated regulatory factors are on the one hand broadly
associated both with cell fate specification and lineage switch-
ing in development (30), and on the other with multiple
tumor-associated phenotypes (31). Another salient example
of SOX-mediated transdifferentiation involves a mechanism
of therapeutic resistance in prostate carcinomas. In this case,
loss of the RB and p53 tumor suppressors—whose absence
is characteristic of neuroendocrine tumors—in response to
antiandrogen therapy is necessary but not sufficient for the
frequently observed conversion of well-differentiated pros-
tate cancer cells into carcinoma cells that have entered a dif-
ferentiation lineage with molecular and histologic features
of neuroendocrine cells, which notably do not express the
androgen receptor. In addition to loss of RB and p53, the
acquired resistance to antiandrogen therapy requires upregu-
lated expression of the SOX2 developmental regulatory gene,
which is demonstrably instrumental in inducing transdif-
ferentiation of the therapy-responsive adenocarcinoma cells
into derivatives that reside in a neuroendocrine cell state that
is refractory to the therapy (32).
A third example also reveals transdifferentiation as a strat-
egy employed by carcinoma cells to avoid elimination by
a lineage-specific therapy, in this case involving basal cell
carcinomas (BCC) of the skin treated with a pharmaco-
logic inhibitor of the Hedgehog-Smoothened (HH/SMO)
Downloaded from cancerdiscovery.aacrjournals.org on January 12, 2022. © 2022 American Association for Cancer
Research.
REVIEW
oncogenic signaling pathway known to drive the neoplastic
growth of these cells (33). Drug-resistant cancer cells switch,
via broad epigenetic shifts in specific chromatin domains and
the altered accessibility of two superenhancers, to a devel-
opmentally related but distinct cell type. The newly gained
phenotypic state of the BCC cells enables them to sustain
expression of the WNT oncogenic signaling pathway, which
in turn imparts independence from the drug-suppressed HH/
SMO signaling pathway (34). As might be anticipated from
this transdifferentiation, the transcriptome of the cancer cells
shifts from a gene signature reflecting the implicated cell-of-
origin of BCCs, namely the stem cells of hair follicle bulge,
to one indicative of the basal stem cells that populate the
interfollicular epidermis. Such transdifferentiation to enable
drug resistance is being increasingly documented in different
forms of cancer (35).
Developmental lineage plasticity also appears to be
prevalent among the major subtypes of lung carcinomas,
that is, neuroendocrine carcinomas [small-cell lung cancer
(SCLC)] and adenocarcinomas + squamous cell carcinomas
[collectively non–small cell lung cancer (NSCLC)]. Single-
cell RNA sequencing has revealed remarkably dynamic
and heterogeneous interconversion among these subtypes
as well as distinct variations thereof during the stages in
lung tumorigenesis, subsequent malignant progression,
and responses to therapy (36–38). Thus, rather than the
simple conceptualization of a pure clonal switch from
one lineage into another, these studies paint a much more
complex picture, of dynamically interconverting subpopu-
lations of cancer cells exhibiting characteristics of multiple
developmental lineages and stages of differentiation, a
sobering realization in regard to lineage-based therapeutic
targeting of human lung cancer. Regulatory determinants
of this dynamic phenotypic plasticity are beginning to be
identified (37, 39, 40).
Synopsis
The three classes of mechanism described above highlight
selective regulators of cellular plasticity that are separable—at
least in part—from core oncogenic drivers and other hallmark
capabilities. Beyond these examples lies a considerable body
of evidence associating many forms of cancer with disrupted
differentiation concomitant with the acquisition of tran-
scriptome signatures and other phenotypes—for example,
histologic morphology—associated with progenitor or stem
cell stages observed in the corresponding normal tissue-of-
origin or in other more distantly related cell types and line-
ages (41–43). As such, these three subclasses of phenotypic
plasticity—dedifferentiation of mature cells back to progeni-
tor states, blocked differentiation to freeze developing cells
in progenitor/stem cell states, and transdifferentiation to
alternative cell lineages—appear to be operative in multiple
cancer types during primary tumor formation, malignant
progression, and/or response to therapy. There are, however,
two conceptual considerations. First, dedifferentiation and
blocked differentiation are likely intertwined, being indis-
tinguishable in many tumor types where the cell-of-origin—
differentiated cell or progenitor/stem cell—is either unknown
or alternatively involved. Second, the acquisition or mainte-
nance of progenitor cell phenotypes and loss of differentiated
JANUARY 2022 CANCER DISCOVERY | 35
REVIEW
Figure 3.  Nonmutational epigenetic reprogramming. Much as during embryogenesis and tissue differentiation and homeostasis, growing evidence
makes the case that instrumental gene-regulatory circuits and networks in tumors can be governed by a plethora of corrupted and co-opted mechanisms
that are independent from genome instability and gene mutation.
features is in most cases an imprecise reflection of the normal
developmental stage, being immersed in a milieu of other
hallmark-enabling changes in the cancer cell that are not
present in naturally developing cells. In addition, yet another
form of phenotypic plasticity involves cell senescence, dis-
cussed more generally below, wherein cancer cells induced
to undergo ostensibly irreversible senescence are instead able
to escape and resume proliferative expansion (44). Finally,
as with other hallmark capabilities, cellular plasticity is not
a novel invention or aberration of cancer cells, but rather
the corruption of latent but activatable capabilities that
various normal cells use to support homeostasis, repair, and
regeneration (45).
Collectively, these illustrative examples encourage consid-
eration of the proposition that unlocking cellular plasticity
to enable various forms of disrupted differentiation consti-
tutes a discrete hallmark capability, distinguishable in regu-
lation and cellular phenotype from the well-validated core
hallmarks of cancer (Fig. 2).
NONMUTATIONAL EPIGENETIC
REPROGRAMMING
The enabling characteristic of genome (DNA) instabil-
ity and mutation is a fundamental component of cancer
formation and pathogenesis. At present, multiple interna-
tional consortia are cataloging mutations across the genome
of human cancer cells, doing so in virtually every type of
human cancer, at different stages of malignant progression,
36 | CANCER DISCOVERY JANUARY 2022 AACRJournals.org
Downloaded from cancerdiscovery.aacrjournals.org on January 12, 2022. © 2022 American Association for Cancer
Research.
Hanahan
Nonmutational
epigenetic reprogramming
including metastatic lesions, and during the development
of adaptive resistance to therapy. One result is the now
widespread appreciation that mutations in genes that organ-
ize, modulate, and maintain chromatin architecture, and
thereby globally regulate gene expression, are increasingly
detected and functionally associated with cancer hallmarks
(46–48).
There is, in addition, a case to be made for another appar-
ently independent mode of genome reprogramming that
involves purely epigenetically regulated changes in gene
expression, one that might be termed “nonmutational epi-
genetic reprogramming” (Fig.  3). Indeed, the proposition
of mutation-less cancer evolution and purely epigenetic
programming of hallmark cancer phenotypes was raised
almost a decade ago (49) and is increasingly discussed
(46, 50–52).
The concept of nonmutational epigenetic regulation of
gene expression is of course well established as the central
mechanism mediating embryonic development, differentia-
tion, and organogenesis (53–55). In the adult, for example,
long-term memory involves changes in gene and histone
modification, in chromatin structure, and in the triggering of
gene expression switches that are stably maintained over time
by positive and negative feedback loops (56, 57). Growing
evidence supports the proposition that analogous epigenetic
alterations can contribute to the acquisition of hallmark
capabilities during tumor development and malignant pro-
gression. A few examples are presented below in support of
this hypothesis.
Hallmarks of Cancer: New Dimensions
Microenvironmental Mechanisms
of Epigenetic Reprogramming
If not solely by consequence of oncogenic mutations, how
then is the cancer cell genome reprogrammed? A growing
body of evidence indicates that the aberrant physical proper-
ties of the tumor microenvironment can cause broad changes
in the epigenome, from which changes beneficial to the phe-
notypic selection of hallmark capabilities can result in clonal
outgrowth of cancer cells with enhanced fitness for prolif-
erative expansion. One common characteristic of tumors (or
regions within tumors) is hypoxia, consequent to insufficient
vascularization. Hypoxia, for example, reduces the activity of
the TET demethylases, resulting in substantive changes in
the methylome, in particular hypermethylation (58). Insuf-
ficient vascularization likely also limits the bioavailability of
critical blood-borne nutrients, and nutrient deprivation has
been shown for example to alter translational control and
consequently enhance the malignant phenotype of breast
cancer cells (59).
A persuasive example of hypoxia-mediated epigenetic regu-
lation involves a form of invariably lethal pediatric epend-
ymoma. Like many embryonic and pediatric tumors, this
form lacks recurrent mutations, in particular a dearth of
driver mutations in oncogenes and tumor suppressors.
Rather, the aberrant growth of these cancer cells is demon-
strably governed by a gene regulatory program induced by
hypoxia (60, 61). Notably, the putative cell-of-origin of this
cancer resides in a hypoxic compartment, likely sensitizing
cells resident therein to the initiation of tumorigenesis by as
yet unknown cofactors.
Another persuasive line of evidence for microenvironmen-
tally mediated epigenetic regulation involves the invasive
growth capability of cancer cells. A classic example involves
the reversible induction of invasiveness of cancer cells at the
margins of many solid tumors, orchestrated by the devel-
opmental regulatory program known as the epithelial-to-
mesenchymal transition (EMT; refs. 62–64). Notably, a mas-
ter regulator of the EMT, ZEB1, has been recently shown to
induce expression of a histone methyltransferase, SETD1B,
that in turn sustains ZEB1 expression in a positive feedback
loop that maintains the (invasive) EMT regulatory state
(65). A previous study similarly documented that induction
of EMT by upregulated expression of a related TF, SNAIL1,
caused marked alterations in the chromatin landscape con-
sequent to induction of a number of chromatin modifiers,
whose activity was demonstrably necessary for the mainte-
nance of the phenotypic state (66). Furthermore, a roster of
conditions and factors to which cancer cells at the margins of
tumors are exposed, including hypoxia and cytokines secreted
by stromal cells, can evidently induce the EMT and in turn
invasiveness (67, 68).
A distinctive example of microenvironmental program-
ming of invasiveness, ostensibly unrelated to the EMT pro-
gram, involves autocrine activation, in pancreas cancer cells
and others, via interstitial pressure–driven fluid flow, of a
neuronal signaling circuit involving secreted glutamate and
its receptor NMDAR (69, 70). Notably, the prototypical stiff-
ness of many solid tumors, embodied in extensive alterations
to the extracellular matrix (ECM) that envelop the cells
Downloaded from cancerdiscovery.aacrjournals.org on January 12, 2022. © 2022 American Association for Cancer
Research.
REVIEW
within, has broad effects on the invasive and other phe-
notypic characteristics of cancer cells. Compared with the
normal tissue ECM from which tumors originate, the tumor
ECM is typically characterized by increased cross-linking and
density, enzymatic modifications, and altered molecular com-
position, which collectively orchestrate—in part via integrin
receptors for ECM motifs—stiffness-induced signaling and
gene-expression networks that elicit invasiveness and other
hallmark characteristics (71).
In addition to such regulatory mechanisms endowed by
the physical tumor microenvironment, paracrine signaling
involving soluble factors released into the extracellular milieu
by the various cell types populating solid tumors can also con-
tribute to the induction of several morphologically distinct
invasive growth programs (72), only one of which—dubbed
“mesenchymal”—seems to involve the aforementioned EMT
epigenetic regulatory mechanism.
Epigenetic Regulatory Heterogeneity
A growing knowledge base is heightening appreciation of
the importance of intratumoral heterogeneity in generating
the phenotypic diversity where the fittest cells for prolif-
erative expansion and invasion outgrow their brethren and
hence are selected for malignant progression. Certainly, one
facet of this phenotypic heterogeneity is founded in chronic
or episodic genomic instability and consequent genetic het-
erogeneity in the cells populating a tumor. In addition, it
is increasingly evident that there can be non–mutationally
based epigenetic heterogeneity. A salient example involves
the linker histone H1.0, which is dynamically expressed and
repressed in subpopulations of cancer cells within a number
of tumor types, with consequent sequestration or accessibil-
ity, respectively, of megabase-sized domains, including ones
conveying hallmark capabilities (73). Notably, the population
of cancer cells with repressed H1.0 were found to have stem-
like characteristics, enhanced tumor-initiating capability, and
an association with poor prognosis in patients.
Another example of epigenetically regulated plasticity has
been described in human oral squamous cell carcinomas
(SCC), wherein cancer cells at the invasive margins adopt a
partial EMT (p-EMT) state lacking the aforementioned mes-
enchymal TFs but expressing other EMT-defining genes that
are not expressed in the central core of the tumors (74). The
p-EMT cells evidently do not represent a clonal compartmen-
talization of mutationally altered cells: cultures of primary
tumor-derived cancer cells contain dynamic mixtures of both
p-EMThi and p-EMTlo cells, and when p-EMThi/lo cells were
FACS-purified and cultured, both reverted to mixed popula-
tions of p-EMThi and p-EMTlo cells within 4 days. Moreover,
although paracrine signals from the adjacent stroma could
be envisaged as deterministic for the p-EMThi state, the sta-
ble presence and regeneration of the two epigenetic states in
culture argues for a cancer cell–intrinsic mechanism. Notably,
this conclusion is supported by analysis of 198 cell lines rep-
resenting 22 cancer types, including SCC, wherein 12 stably
heterogeneous epigenetic states (including the p-EMT in
SCC) were variously detected in the cell line models as well
as their cognate primary tumors (75). Again, the heteroge-
neous phenotypic states could not be linked to detectable
genetic differences, and in several cases FACS-sorted cells of
JANUARY 2022 CANCER DISCOVERY | 37
REVIEW
a particular state were shown to dynamically reequilibrate
upon culture, recapitulating a stable balance among the het-
erogeneous states seen in the original cell lines.
Additionally, technologies for genome-wide profiling of
diverse attributes—beyond DNA sequence and its mutational
variation—are illuminating influential elements of the cancer
cell genome’s annotation and organization that correlate
with patient prognosis, and increasingly with hallmark capa-
bilities (76–78). Epigenomic heterogeneity is being revealed
by increasingly powerful technologies for profiling genome-
wide DNA methylation (79, 80), histone modification (81),
chromatin accessibility (82), and posttranscriptional modifi-
cation and translation of RNA (83, 84). A challenge in regard
to the postulate being considered herein will be to ascertain
which epigenomic modifications in particular cancer types
(i) have regulatory significance and (ii) are representative of
purely nonmutational reprogramming, as opposed to being
mutation-driven and thus explainable by genome instability.
Epigenetic Regulation of the Stromal Cell Types
Populating the Tumor Microenvironment
In general, the accessory cells in the tumor microenvi-
ronment that functionally contribute to the acquisition
of hallmark capabilities are not thought to suffer genetic
instability and mutational reprogramming to enhance their
tumor-promoting activities; rather it is inferred that these
cells—cancer-associated fibroblasts, innate immune cells, and
endothelial cells and pericytes of the tumor vasculature—
are epigenetically reprogrammed upon their recruitment by
soluble and physical factors that define the solid tumor micro-
environment (2, 85). It can be anticipated the multi-omic pro-
filing technologies currently being applied to cancer cells will
increasingly be used to interrogate the accessory (stromal)
cells in tumors to elucidate how normal cells are corrupted
to functionally support tumor development and progression.
For example, a recent study (86) suggests that such reprogram-
ming can involve modifications of the epigenome in addition
to the inductive interchange of cytokines, chemokines, and
growth factors that alter intracellular signaling networks in
all of these cell types: when mouse models of metastasis to
lung were treated with a combination of a DNA methyltrans-
ferase inhibitor (5-azacytidine) and an inhibitor of histone
modification (an HDAC), the infiltrating myeloid cells were
found to have switched from an immature (tumor-promot-
ing) progenitor state into cells resembling mature interstitial
(tumor-antagonizing) macrophages, which, in contrast to
their counterparts in untreated tumors, were incapable of
supporting the hallmark capabilities necessary for efficient
metastatic colonization (86). It can be envisaged that multi-
omic profiling and pharmacologic perturbation will serve to
elucidate the reprogrammed epigenetic state in such myeloid
cells as well as other hallmark-enabling accessory cell types
populating tumor microenvironments.
Synopsis
Collectively, these illustrative snapshots support the prop-
osition that nonmutational epigenetic reprograming will
come to be accepted as a bona fide enabling characteristic
that serves to facilitate the acquisition of hallmark capa-
bilities (Fig. 3), distinct from that of genomic DNA instability
38 | CANCER DISCOVERY JANUARY 2022 AACRJournals.org
Downloaded from cancerdiscovery.aacrjournals.org on January 12, 2022. © 2022 American Association for Cancer
Research.
Hanahan
and mutation. Notably, it can be anticipated that nonmuta-
tional epigenetic reprogramming will prove to be integrally
involved in enabling the provisional new hallmark capability
of phenotypic plasticity discussed above, in particular being
a driving force in the dynamic transcriptomic heterogeneity
that is increasingly well documented in cancer cells populat-
ing malignant TMEs. The advance of single cell multi-omic
profiling technologies is envisaged to illuminate the respec-
tive contributions of and interplay between mutation-driven
versus nonmutational epigenetic regulation to the evolution
of tumors during malignant progression and metastasis.
POLYMORPHIC MICROBIOMES
An expansive frontier in biomedicine is unfolding via
illumination of the diversity and variability of the plethora
of microorganisms, collectively termed the microbiota, that
symbiotically associate with the barrier tissues of the body
exposed to the external environment—the epidermis and the
internal mucosa, in particular the gastrointestinal tract, as
well as the lung, the breast, and the urogenital system. There
is growing appreciation that the ecosystems created by resi-
dent bacteria and fungi—the microbiomes—have profound
impact on health and disease (87), a realization fueled by
the capability to audit the populations of microbial species
using next-generation sequencing and bioinformatic tech-
nologies. For cancer, the evidence is increasingly compelling
that polymorphic variability in the microbiomes between
individuals in a population can have a profound impact on
cancer phenotypes (88, 89). Association studies in human
and experimental manipulation in mouse models of can-
cer are revealing particular microorganisms, principally but
not exclusively bacteria, which can have either protective
or deleterious effects on cancer development, malignant
progression, and response to therapy. So too can the global
complexity and constitution of a tissue microbiome at large.
Indeed, while the gut microbiome has been the pioneer of
this new frontier, multiple tissues and organs have associ-
ated microbiomes, which have distinctive characteristics in
regard to population dynamics and diversity of microbial
species and subspecies. This growing appreciation of the
importance of polymorphically variable microbiomes in
health and disease posits the question: is the microbiome
a discrete enabling characteristic that broadly affects, both
positively and negatively, the acquisition of hallmark capa-
bilities for cancer? I reflect on this possibility below, illustrat-
ing evidence for some of the prominent tissue microbiomes
implicated in cancer hallmarks (Fig.  4), beginning with the
most prominent and evidently impactful microbiome, that
of the intestinal tract.
Diverse Modulatory Effects of the Gut Microbiome
It has long been recognized that the gut microbiome is fun-
damentally important for the function of the large intestine
(colon) in degrading and importing nutrients into the body
as part of metabolic homeostasis, and that distortions in the
microbial populations—dysbiosis—in the colon can cause a
spectrum of physiologic maladies (87). Among these has
been the suspicion that the susceptibility, development, and
Hallmarks of Cancer: New Dimensions REVIEW

Gut

Skin Lung
Modulating
tumor
Growth
Inflammation
Immune
evasions
Genome
instability
Vaginal/ Therapy
Oral resistance
cervical

Polymorphic
microbiomes Tumor

Figure 4.  Polymorphic microbiomes. Left, while intersecting with the enabling characteristics of tumor-promoting inflammation and genomic instabil-
ity and mutation, there is growing reason to conclude that polymorphic microbiomes in one individual versus another, being resident in the colon, other
mucosa and connected organs, or in tumors themselves, can diversely influence—by either inducing or inhibiting—many of the hallmark capabilities,
and thus are potentially an instrumental and quasi-independent variable in the puzzle of how cancers develop, progress, and respond to therapy. Right,
multiple tissue microbiomes are implicated in modulating tumor phenotypes. In addition to the widely studied gut microbiome, other distinctive tissue
microbiomes, as well as the tumor microbiome, are implicated in modulating the acquisition—both positively and negatively—of the illustrated hallmark
capabilities in certain tumor types.

pathogenesis of colon cancer is influenced by the gut micro-
biome. In recent years, persuasive functional studies, involv-
ing fecal transplants from colon tumor–bearing patients and
mice into recipient mice predisposed to develop colon cancer
has established a principle: there are both cancer-protective
and tumor-promoting microbiomes, involving particular
bacterial species, which can modulate the incidence and
pathogenesis of colon tumors (90).
The mechanisms by which microbiota impart these modu-
latory roles are still being elucidated, but two general effects
are increasingly well established for tumor-promoting micro-
biomes and in some cases for specific tumor-promoting
bacterial species. The first effect is mutagenesis of the colonic
epithelium, consequent to the production of bacterial tox-
ins and other molecules that either damage DNA directly,
or disrupt the systems that maintain genomic integrity, or
stress cells in other ways that indirectly impair the fidel-
ity of DNA replication and repair. A case in point is E. coli
carrying the PKS locus, which demonstrably mutagenizes
the human genome and is implicated in conveying hallmark-
enabling mutations (91).
Additionally, bacteria have been reported to bind to the
surface of colonic epithelial cells and produce ligand mimet-
ics that stimulate epithelial proliferation, contributing in
neoplastic cells to the hallmark capability for proliferative
signaling (88). Another mechanism by which specific bacterial
species promote tumorigenesis involves butyrate-producing
bacteria, whose abundance is elevated in patients with colo-
rectal cancer (92). The production of the metabolite butyrate
has complex physiologic effects, including the induction of
senescent epithelial and fibroblastic cells. A mouse model
of colon carcinogenesis populated with butyrate-producing
bacteria developed more tumors than mice lacking such
bacteria; the connection between butyrate-induced senes-
cence and enhanced colon tumorigenesis was demonstrated
by the use of a senolytic drug that kills senescent cells,
which impaired tumor growth (92). In addition, bacterial-
produced butyrate has pleiotropic and paradoxical effects on
differentiated cells versus undifferentiated (stem) cells in the
colonic epithelium in conditions where the intestinal barrier
is disrupted (dysbiosis) and the bacteria are invasive, affect-
ing, for example, cellular energetics and metabolism, histone
modification, cell-cycle progression, and (tumor-promoting)
innate immune inflammation that is immunosuppressive of
adaptive immune responses (93).
Indeed, a broad effect of polymorphic microbiomes
involves the modulation of the adaptive and innate immune
systems via multifarious routes, including the production by
bacteria of “immunomodulatory” factors that activate dam-
age sensors on epithelial or resident immune cells, resulting
in the expression of a diverse repertoire of chemokines and
cytokines that can sculpt the abundance and characteristics
of immune cells populating the colonic epithelia and its
underlying stroma and draining lymph nodes. In addition,
certain bacteria can breach both the protective biofilm and
the mucus lining the colonic epithelia and proceed to dis-
rupt the epithelial cell–cell tight junctions that collectively
maintain the integrity of the physical barrier that normally
compartmentalizes the intestinal microbiome. Upon invad-
ing the stroma, bacteria can trigger both innate and adap-
tive immune responses, eliciting secretion of a repertoire of
cytokines and chemokines. One manifestation can be the
creation of tumor-promoting or tumor-antagonizing immune
microenvironments, consequently protecting against or
facilitating tumorigenesis and malignant progression. Con-
cordantly, the modulation by distinctive microbiomes in

JANUARY 2022 CANCER DISCOVERY | 39

Downloaded from cancerdiscovery.aacrjournals.org on January 12, 2022. © 2022 American Association for Cancer
Research.
REVIEW
individual patients of the intertwined parameters of (i) elicit-
ing (innate) tumor promoting inflammation and (ii) escap-
ing (adaptive) immune destruction can be associated not
only with prognosis, but also with responsiveness or resist-
ance to immunotherapies involving immune checkpoint
inhibitors and other therapeutic modalities (89, 94–96).
Provisional proof-of-concept has come from recent stud-
ies demonstrating restored efficacy to immunotherapy
following transplants of fecal microbiota from therapy-
responsive patients into patients with melanoma who had
progressed during prior treatment with immune checkpoint
blockade (97, 98).
An ongoing mystery has involved the molecular mecha-
nisms by which particular and variable constituents of the
gut microbiome systemically modulate the activity of the
adaptive immune system, either enhancing antitumoral
immune responses evoked by immune checkpoint blockade,
or rather eliciting systemic or local (intratumoral) immu-
nosuppression. A recent study has shed some light: certain
strains of Enterococcus (and other bacteria) express a pepti-
doglycan hydrolyase called SagA that releases mucopeptides
from the bacterial wall, which can then circulate systemi-
cally and activate the NOD2 pattern receptor, which in turn
can enhance T-cell responses and the efficacy of checkpoint
immunotherapy (99). Other immunoregulatory molecules
produced by specific bacterial subspecies are being identi-
fied and functionally evaluated, including bacteria-produced
inosine, a rate-limiting metabolite for T-cell activity (100).
These examples and others are beginning to chart the molec-
ular mechanisms by which polymorphic microbiomes are
indirectly and systemically modulating tumor immunobiol-
ogy, above and beyond immune responses consequent to
direct physical interactions of bacteria with the immune
system (101, 102).
Beyond the causal links to colon cancer and melanoma, the
gut microbiome’s demonstrable ability to elicit the expres-
sion of immunomodulatory chemokines and cytokines that
enter the systemic circulation is evidently also capable of
affecting cancer pathogenesis and response to therapy in
other organs of the body (94, 95). An illuminating example
involves the development of cholangiocarcinomas in the
liver: gut dysbiosis allows the entry and transport of bacteria
and bacterial products through the portal vein to the liver,
where TLR4 expressed on hepatocytes is triggered to induce
expression of the chemokine CXCL1, which recruits CXCR2-
expressing granulocytic myeloid cells (gMDSC) that serve to
suppress natural killer cells so as to evade immune destruc-
tion (103), and likely convey other hallmark capabilities (85).
As such, the gut microbiome is unambiguously implicated as
an enabling characteristic that can alternatively facilitate or
protect against multiple forms of cancer.
Beyond the Gut: Implicating Distinctive
Microbiomes in Other Barrier Tissues
Virtually all tissues and organs exposed, directly or indi-
rectly, to the outside environment are also repositories for com-
mensal microorganisms (104). Unlike the intestine, where the
symbiotic role of the microbiome in metabolism is well recog-
nized, the normal and pathogenic roles of resident microbiota
in these diverse locations is still emerging. There are evidently
40 | CANCER DISCOVERY JANUARY 2022 AACRJournals.org
Downloaded from cancerdiscovery.aacrjournals.org on January 12, 2022. © 2022 American Association for Cancer
Research.
Hanahan
organ/tissue-specific differences in the constitution of the
associated microbiomes in homeostasis, aging, and cancer,
with both overlapping and distinctive species and abundan-
cies to that of the colon (104, 105). Moreover, association
studies are providing increasing evidence that local tumor-
antagonizing/protective versus tumor-promoting tissue
microbiomes, similarly to the gut microbiome, can modulate
susceptibility and pathogenesis to human cancers arising in
their associated organs (106–109).
Impact of Intratumoral Microbiota?
Finally, pathologists have long recognized that bacteria can
be detected within solid tumors, an observation that has now
been substantiated with sophisticated profiling technologies.
For example, in a survey of 1,526 tumors encompassing seven
human cancer types (bone, brain, breast, lung, melanoma,
ovary, and pancreas), each type was characterized by a distinc-
tive microbiome that was largely localized inside cancer cells
and immune cells, and within each tumor type, variations in
the tumor microbiome could be detected and inferred to be
associated with clinicopathologic features (110). Microbiota
have been similarly detected in genetically engineered de novo
mouse models of lung and pancreas cancer, and their absence
in germ-free mice and/or their abrogation with antibiotics can
demonstrably impair tumorigenesis, functionally implicating
the tumor microbiome as an enabler of tumor-promoting
inflammation and malignant progression (111, 112). Associa-
tion studies in human pancreatic ductal adenocarcinoma and
functional tests via fecal transplants into tumor-bearing mice
have established that variations in the tumor microbiome—
and the associated gut microbiome—modulate immune phe-
notypes and survival (113). An important challenge for the
future will be to extend these implications to other tumor
types, and to delineate the potentially separable contribu-
tions of constitution and variation in the tumor microbiome
to that of the gut (and local tissue of origin) microbiome,
potentially by identifying specific microbial species that are
functionally influential in one location or the other.
Synopsis
Among the fascinating questions for the future is whether
microbiota resident in different tissues or populating incipi-
ent neoplasias have the capability to contribute to or interfere
with the acquisition of other hallmark capabilities beyond
immunomodulation and genome mutation, thereby influ-
encing tumor development and progression. There are clues
that particular bacterial species can directly stimulate the
hallmark of proliferative signaling, for example, in colonic
epithelium (88), and modulate growth suppression by alter-
ing tumor suppressor activity in different compartments of
the intestine (114), whereas direct effects on other hallmark
capabilities, such as avoiding cell death, inducing angiogene-
sis, and stimulating invasion and metastasis, remain obscure,
as does the generalizability of these observations to multiple
forms of human cancer. Irrespective, there is an increasingly
compelling case to be made that polymorphic variation in
microbiomes of the intestine and other organs constitutes
a distinctive enabling characteristic for the acquisition of
hallmark capabilities (Fig.  4), albeit intersecting with and
Hallmarks of Cancer: New Dimensions
complementing those of genome instability and mutation,
and tumor-promoting inflammation.
SENESCENT CELLS
Cellular senescence is a typically irreversible form of pro-
liferative arrest, likely evolved as a protective mechanism for
maintaining tissue homeostasis, ostensibly as a complemen-
tary mechanism to programmed cell death that serves to
inactivate and in due course remove diseased, dysfunctional,
or otherwise unnecessary cells. In addition to shutting down
the cell division cycle, the senescence program evokes changes
in cell morphology and metabolism and, most profoundly,
the activation of a senescence-associated secretory phenotype
(SASP) involving the release of a plethora of bioactive pro-
teins, including chemokines, cytokines, and proteases whose
identity is dependent on the cell and tissue type from which a
senescent cell arises (115–117). Senescence can be induced in
cells by a variety of conditions, including microenvironmen-
tal stresses such as nutrient deprivation and DNA damage, as
well as damage to organelles and cellular infrastructure, and
imbalances in cellular signaling networks (115, 117), all of
which have been associated with the observed increase in the
abundance of senescent cells in various organs during aging
(118, 119).
Cellular senescence has long been viewed as a protective
mechanism against neoplasia, whereby cancerous cells are
induced to undergo senescence (120). Most of the afore-
mentioned instigators of the senescent program are asso-
ciated with malignancy, in particular DNA damage as a
consequence of aberrant hyperproliferation, so-called onco-
gene-induced senescence due to hyperactivated signaling,
and therapy-induced senescence consequent to cellular and
genomic damage caused by chemotherapy and radiotherapy.
Indeed, there are well-established examples of the protective
benefits of senescence in limiting malignant progression (118,
119). To the contrary, however, an increasing body of evidence
reveals quite the opposite: in certain contexts, senescent cells
variously stimulate tumor development and malignant pro-
gression (119, 121). In one illuminating case study, senescent
cells were pharmacologically ablated in aging mice, in par-
ticular depleting senescent cells characteristically expressing
the cell-cycle inhibitor p16−INK4a: in addition to delaying mul-
tiple age-related symptoms, the depletion of senescent cells
in aging mice resulted in reduced incidences of spontaneous
tumorigenesis and cancer-associated death (122).
The principal mechanism by which senescent cells pro-
mote tumor phenotypes is thought to be the SASP, which is
demonstrably capable of conveying, in paracrine fashion to
viable cancer cells in proximity, as well as to other cells in the
TME, signaling molecules (and proteases that activate and/
or desequester them) so as to convey hallmark capabilities.
Thus, in different experimental systems, senescent cancer
cells have been shown to variously contribute to prolifera-
tive signaling, avoiding apoptosis, inducing angiogenesis,
stimulating invasion and metastasis, and suppressing tumor
immunity (116, 118, 120, 121).
Yet another facet to the effects of senescent cancer cells
on cancer phenotypes involves transitory, reversible senes-
cent cell states, whereby senescent cancer cells can escape
Downloaded from cancerdiscovery.aacrjournals.org on January 12, 2022. © 2022 American Association for Cancer
Research.
REVIEW
from their SASP-expressing, nonproliferative condition, and
resume cell proliferation and manifestation of the associ-
ated capabilities of fully viable oncogenic cells (44). Such
transitory senescence is most well documented in cases of
therapy resistance (44), representing a form of dormancy
that circumvents therapeutic targeting of proliferating can-
cer cells, but may well prove to be more broadly operative in
other stages of tumor development, malignant progression,
and metastasis.
Moreover, the hallmark-promoting capabilities of senes-
cent cells are not limited to senescent cancer cells. Cancer-
associated fibroblasts (CAF) in tumors have been shown to
undergo senescence, creating senescent CAFs that are demon-
strably tumor-promoting by virtue of conveying hallmark
capabilities to cancer cells in the TME (115, 116, 121). More-
over, senescent fibroblasts in normal tissues produced in
part by natural aging or environmental insults have similarly
been implicated in remodeling tissue microenvironments via
their SASP so as to provide paracrine support for local inva-
sion (so-called “field effects”) and distant metastasis (116) of
neoplasias developing in proximity. Additionally, senescent
fibroblasts in aging skin have been shown to recruit—via their
SASP—innate immune cells that are both immunosuppres-
sive of adaptive antitumoral immune responses anchored
by CD8 T cells, and stimulatory of skin tumor growth (123),
with the latter effect potentially reflecting paracrine contribu-
tions of such innate immune cells (myeloid cells, neutrophils,
and macrophages) to other hallmark capabilities.
While less well established, it seems likely that other abun-
dant stromal cells populating particular tumor microen-
vironments will prove to undergo senescence, and thereby
modulate cancer hallmarks and consequent tumor phe-
notypes. For example, therapy-induced senescent tumor
endothelial cells can enhance proliferation, invasion, and
metastasis in breast cancer models (124, 125).
Certainly, such clues warrant investigation in other tumor
types to assess generality of fibroblastic, endothelial, and
other stromal cell senescence as a driving force in tumor
evolution. Also currently unresolved are the regulatory
mechanisms and functional determinants through which
a particular senescent cell type in a given TME evokes a
tumor-promoting versus a tumor-antagonizing SASP, which
can seeming be alternatively induced in the same senescing
cell type, perhaps by different instigators when immersed in
distinctive physiologic and neoplastic microenvironments.
Synopsis
The concept that tumors are composed of genetically
transformed cancer cells interacting with and benefiting from
recruited and epigenetically/phenotypically corrupted acces-
sory (stromal) cells is well established as instrumental to the
pathogenesis of cancer. The considerations discussed above
and described in the reviews and reports cited herein (and
elsewhere) make a persuasive case for the proposition that
senescent cells (of whatever cellular origin) should be consid-
ered for addition to the roster of functionally significant cells
in the tumor microenvironment (Fig.  5). As such, senescent
cells warrant being factored into the quest for deep knowl-
edge of cancer mechanisms. Furthermore, the realization of
JANUARY 2022 CANCER DISCOVERY | 41
REVIEW Hanahan

Senescent cells
(multiple origins)

Senescent cells

Figure 5.  Senescent cells. Heterogeneous cancer cell subtypes as well as stromal cell types and subtypes are functionally integrated into the manifes-
tations of tumors as outlaw organs. Clues are increasingly implicating senescent cell derivatives of many of these cellular constituents of the TME, and
their variable SASPs, in modulating hallmark capabilities and consequent tumor phenotypes.

their importance motivates the ancillary goal to therapeuti-
cally target tumor-promoting senescent cells of all constitu-
tions, be it by pharmacologic or immunologic ablation, or by
reprogramming the SASP into tumor-antagonizing variants
(115, 121, 126).
CONCLUDING REMARKS
While the eight hallmarks of cancer and their two ena-
bling characteristics have proved of enduring heuristic value in
the conceptualization of cancer, the considerations presented
above suggest that there may be new facets of some general-
ity and hence of relevance to more fully understanding the
complexities, mechanisms, and manifestations of the disease.
By applying the metric of discernable if not complete inde-
pendence from the 10 core attributes, it is arguable that these
four parameters may well—pursuant to further validation and
generalization beyond the case studies presented—become inte-
grated into the hallmarks of cancer schematic (Fig. 6). Thus, cel-
lular plasticity may come to be added to the roster of hallmark
capabilities. Notably, while the eight core and this nouveau
capability are each, by their definition as a hallmark, concep-
tually distinguishable, aspects of their regulation are at least
partially interconnected in some and perhaps many cancers.
For example, multiple hallmarks are coordinately modulated
in some tumor types by canonical oncogenic drivers, including
(i) KRAS (https://cancer.sanger.ac.uk/cosmic/census-page/
KRAS),
(ii) MYC (https://cancer.sanger.ac.uk/cosmic/census-page/
MYC),
(iii) NOTCH (https://cancer.sanger.ac.uk/cosmic/census-
page/NOTCH1; ref. 127), and
(iv) TP53 (https://cancer.sanger.ac.uk/cosmic/census-page/
TP53),
highlighting the important challenge to more fully elu-
cidate the regulatory networks governing these acquired
capabilities.
In addition to adding cellular plasticity to the roster,
nonmutational epigenetic reprogramming and polymorphic
variations in organ/tissue microbiomes may come to be incor-
porated as mechanistic determinants—enabling characteris-
tics—by which hallmark capabilities are acquired, along with
tumor-promoting inflammation (itself partially intercon-
nected to the microbiome), above and beyond the mutations
and other aberrations that manifest the afore-mentioned
oncogenic drivers.
Finally, senescent cells of different origins—including can-
cer cells and various stromal cells—that functionally contrib-
ute to the development and malignant progression of cancer,
albeit in markedly distinctive ways to those of their nonsenes-
cent brethren, may become incorporated as generic compo-
nents of the TME. In conclusion, it is envisaged that raising
these provisional “trial balloons” will stimulate debate, dis-
cussion, and continuing experimental investigation in the
cancer research community about the defining conceptual
parameters of cancer biology, genetics, and pathogenesis.

42 | CANCER DISCOVERY JANUARY 2022 AACRJournals.org

Downloaded from cancerdiscovery.aacrjournals.org on January 12, 2022. © 2022 American Association for Cancer
Research.
Hallmarks of Cancer: New Dimensions REVIEW

Sustaining Evading
proliferative signaling growth suppressors

Unlocking Nonmutational
phenotypic plasticity epigenetic reprogramming

Deregulating Avoiding immune

cellular destruction
metabolism

Resisting cell Enabling

death replicative
immortality

Genome
Tumor-promoting
instability &
inflammation
mutation

Polymorphic
Senescent cells
microbiomes

Inducing or accessing Activating invasion &

vasculature metastasis

Figure 6.  Hallmarks of Cancer—new additions. Depicted are the canonical and prospective new additions to the “Hallmarks of Cancer.” This treatise
raises the possibility, aiming to stimulate debate, discussion, and experimental elaboration, that some or all of the four new parameters will come to be
appreciated as generic to multiple forms of human cancer and hence appropriate to incorporate into the core conceptualization of the hallmarks of cancer.

Author’s Disclosures 5. Perekatt AO, Shah PP, Cheung S, Jariwala N, Wu A, Gandhi V,

No disclosures were reported.
Acknowledgments
First and foremost, I profoundly thank Bob Weinberg for an
exceptional tradition of insightful and formative discussions, and
for excellent comments and suggestions to the first vignette of this
manuscript. Additionally, I wish to thank: Ben Stanger; Bradley
Bernstein, Giovanni Ciriello, and William Flavahan; Jennifer Wargo;
and Sheila Stewart for their valuable comments and suggestions
on the four vignettes, respectively, and SayoStudio for assistance in
crafting the figures.
Received August 4, 2021; revised November 17, 2021; accepted
November 18, 2021; published first January 12, 2022.
REFERENCES
1. Hanahan D, Weinberg RA. The hallmarks of cancer. Cell 2000;100:
57–70.
2. Hanahan D, Weinberg RA. Hallmarks of cancer: the next generation.
Cell 2011;144:646–74.
3. Yuan S, Norgard RJ, Stanger BZ. Cellular plasticity in cancer. Cancer
Discov 2019;9:837–51.
4. Barker N, Ridgway RA, van Es JH, van de Wetering M, Begthel H,
van den Born M, et al. Crypt stem cells as the cells-of-origin of intes-
tinal cancer. Nature 2009;457:608–11.
et  al. SMAD4 suppresses WNT-driven dedifferentiation and onco-
genesis in the differentiated gut epithelium. Cancer Res 2018;78:
4878–90.
6. Shih IM, Wang TL, Traverso G, Romans K, Hamilton SR, Ben-Sasson S,
et  al. Top-down morphogenesis of colorectal tumors. Proc Natl
Acad Sci U S A 2001;98:2640–5.
7. Ordóñez-Morán P, Dafflon C, Imajo M, Nishida E, Huelsken J.
HOXA5 counteracts stem cell traits by inhibiting Wnt signaling in
colorectal cancer. Cancer Cell 2015;28:815–29.
8. Tan SH, Barker N. Stemming colorectal cancer growth and metas-
tasis: HOXA5 forces cancer stem cells to differentiate. Cancer Cell
2015;28:683–5.
9. Köhler C, Nittner D, Rambow F, Radaelli E, Stanchi F, Vandamme N,
et  al. Mouse cutaneous melanoma induced by mutant BRaf arises
from expansion and dedifferentiation of mature pigmented mel-
anocytes. Cell Stem Cell 2017;21:679–93.
10. Shah M, Bhoumik A, Goel V, Dewing A, Breitwieser W, Kluger H,
et al. A role for ATF2 in regulating MITF and melanoma develop-
ment. PLoS Genet 2010;6:e1001258.
11. Claps G, Cheli Y, Zhang T, Scortegagna M, Lau E, Kim H, et al. A
transcriptionally inactive ATF2 variant drives melanomagenesis.
Cell Rep 2016;15:1884–92.
12. Saghafinia S, Homicsko K, Di Domenico A, Wullschleger S,
Perren A, Marinoni I, et  al. Cancer cells retrace a stepwise differ-
entiation program during malignant progression. Cancer Discov
2021;11:2638–57.
13. Yu X-X, Qiu W-L, Yang L, Zhang Y, He M-Y, Li L-C, et al. Defining
multistep cell fate decision pathways during pancreatic develop-
ment at single-cell resolution. EMBO J 2019;38:e100164.

JANUARY 2022 CANCER DISCOVERY | 43

Downloaded from cancerdiscovery.aacrjournals.org on January 12, 2022. © 2022 American Association for Cancer
Research.
REVIEW
14. de Thé H. Differentiation therapy revisited. Nat Rev Cancer 2018;
18:117–27.
15. He LZ, Merghoub T, Pandolfi PP. In vivo analysis of the molecular
pathogenesis of acute promyelocytic leukemia in the mouse and its
therapeutic implications. Oncogene 1999;18:5278–92.
16. Warrell RP, de Thé H, Wang ZY, Degos L. Acute promyelocytic leu-
kemia. N Engl J Med 1993;329:177–89.
17. Bots M, Verbrugge I, Martin BP, Salmon JM, Ghisi M, Baker A, et al.
Differentiation therapy for the treatment of t(8;21) acute myeloid leu-
kemia using histone deacetylase inhibitors. Blood 2014;123:1341–52.
18. Ferrara FF, Fazi F, Bianchini A, Padula F, Gelmetti V, Minucci S,
et  al. Histone deacetylase-targeted treatment restores retinoic acid
signaling and differentiation in acute myeloid leukemia. Cancer Res
2001;61:2–7.
19. Kaufman CK, Mosimann C, Fan ZP, Yang S, Thomas AJ, Ablain J,
et al. A zebrafish melanoma model reveals emergence of neural crest
identity during melanoma initiation. Science 2016;351:aad2197.
20. Morris JP, Yashinskie JJ, Koche R, Chandwani R, Tian S, Chen C-C,
et al. α-Ketoglutarate links p53 to cell fate during tumour suppres-
sion. Nature 2019;573:595–9.
21. Saha SK, Parachoniak CA, Ghanta KS, Fitamant J, Ross KN,
Najem MS, et al. Mutant IDH inhibits HNF-4α to block hepatocyte
differentiation and promote biliary cancer. Nature 2014;513:110–4.
22. Dang L, Su S-SM. Isocitrate dehydrogenase mutation and (R)-2-hy-
droxyglutarate: from basic discovery to therapeutics development.
Annu Rev Biochem 2017;86:305–31.
23. Waitkus MS, Diplas BH, Yan H. Biological role and therapeutic
potential of IDH mutations in cancer. Cancer Cell 2018;34:186–95.
24. Phan TG, Croucher PI. The dormant cancer cell life cycle. Nat Rev
Cancer 2020;20:398–411.
25. Jiang M, Azevedo-Pouly AC, Deering TG, Hoang CQ, DiRenzo D,
Hess DA, et al. MIST1 and PTF1 collaborate in feed-forward regula-
tory loops that maintain the pancreatic acinar phenotype in adult
mice. Mol Cell Biol 2016;36:2945–55.
26. Krah NM, Narayanan SM, Yugawa DE, Straley JA, Wright CVE,
MacDonald RJ, et  al. Prevention and reversion of pancreatic tum-
origenesis through a differentiation-based mechanism. Dev Cell
2019;50:744–54.
27. Krah NM, De La O J-P, Swift GH, Hoang CQ, Willet SG, Chen Pan F,
et al. The acinar differentiation determinant PTF1A inhibits initia-
tion of pancreatic ductal adenocarcinoma. eLife 2015;4:e07125.
28. Shi G, DiRenzo D, Qu C, Barney D, Miley D, Konieczny SF. Main-
tenance of acinar cell organization is critical to preventing Kras-
induced acinar-ductal metaplasia. Oncogene 2013;32:1950–8.
29. Kopp JL, von Figura G, Mayes E, Liu F-F, Dubois CL, Morris JP,
et al. Identification of Sox9-dependent acinar-to-ductal reprogram-
ming as the principal mechanism for initiation of pancreatic ductal
adenocarcinoma. Cancer Cell 2012;22:737–50.
30. Julian LM, McDonald AC, Stanford WL. Direct reprogramming with
SOX factors: masters of cell fate. Curr Opin Genet Dev 2017;46:24–36.
31. Grimm D, Bauer J, Wise P, Krüger M, Simonsen U, Wehland M, et al.
The role of SOX family members in solid tumours and metastasis.
Semin Cancer Biol 2020;67:122–53.
32. Mu P, Zhang Z, Benelli M, Karthaus WR, Hoover E, Chen C-C, et al.
SOX2 promotes lineage plasticity and antiandrogen resistance in
TP53- and RB1-deficient prostate cancer. Science 2017;355:84–8.
33. Von Hoff DD, LoRusso PM, Rudin CM, Reddy JC, Yauch RL,
Tibes  R, et  al. Inhibition of the hedgehog pathway in advanced
basal-cell carcinoma. N Engl J Med 2009;361:1164–72.
34. Biehs B, Dijkgraaf GJP, Piskol R, Alicke B, Boumahdi S, Peale F,
et al. A cell identity switch allows residual BCC to survive Hedgehog
pathway inhibition. Nature 2018;562:429–33.
35. Boumahdi S, de Sauvage FJ. The great escape: tumour cell plasticity in
resistance to targeted therapy. Nat Rev Drug Discov 2020;19:39–56.
36. Groves SM, Ireland A, Liu Q, Simmons AJ, Lau K, Iams WT,
et al. Cancer Hallmarks Define a Continuum of Plastic Cell States
between Small Cell Lung Cancer Archetypes [Internet]. Systems
Biology; 2021 Jan. Available from: http://biorxiv.org/lookup/doi/10.
1101/2021.01.22.427865.
44 | CANCER DISCOVERY JANUARY 2022 AACRJournals.org
Downloaded from cancerdiscovery.aacrjournals.org on January 12, 2022. © 2022 American Association for Cancer
Research.
Hanahan
37. LaFave LM, Kartha VK, Ma S, Meli K, Del Priore I, Lareau C, et al.
Epigenomic state transitions characterize tumor progression in
mouse lung adenocarcinoma. Cancer Cell 2020;38:212–28.
38. Marjanovic ND, Hofree M, Chan JE, Canner D, Wu K, Trakala M,
et  al. Emergence of a high-plasticity cell state during lung cancer
evolution. Cancer Cell 2020;38:229–46.
39. Drapkin BJ, Minna JD. Studying lineage plasticity one cell at a time.
Cancer Cell 2020;38:150–2.
40. Inoue Y, Nikolic A, Farnsworth D, Liu A, Ladanyi M, Somwar  R,
et  al. Extracellular signal-regulated kinase mediates chromatin
rewiring and lineage transformation in lung cancer [Internet]. Can-
cer Biology; 2020 Nov. Available from: http://biorxiv.org/lookup/
doi/10.1101/2020.11.12.368522.
41. Dravis C, Chung C-Y, Lytle NK, Herrera-Valdez J, Luna G, Trejo CL,
et  al. Epigenetic and transcriptomic profiling of mammary gland
development and tumor models disclose regulators of cell state
plasticity. Cancer Cell 2018;34:466–82.
42. Malta TM, Sokolov A, Gentles AJ, Burzykowski T, Poisson L,
Weinstein JN, et  al. Machine learning identifies stemness features
associated with oncogenic dedifferentiation. Cell 2018;173:338–54.
43. Miao Z-F, Lewis MA, Cho CJ, Adkins-Threats M, Park D, Brown JW,
et  al. A dedicated evolutionarily conserved molecular network
licenses differentiated cells to return to the cell cycle. Dev Cell
2020;55:178–94.
44. De Blander H, Morel A-P, Senaratne AP, Ouzounova M, Puisieux A.
Cellular plasticity: a route to senescence exit and tumorigenesis.
Cancers 2021;13:4561.
45. Merrell AJ, Stanger BZ. Adult cell plasticity in vivo: de-differentia-
tion and transdifferentiation are back in style. Nat Rev Mol Cell Biol
2016;17:413–25.
46. Baylin SB, Jones PA. Epigenetic determinants of cancer. Cold Spring
Harb Perspect Biol 2016;8:a019505.
47. Flavahan WA, Gaskell E, Bernstein BE. Epigenetic plasticity and the
hallmarks of cancer. Science 2017;357:eaal2380.
48. Jones PA, Issa J-PJ, Baylin S. Targeting the cancer epigenome for
therapy. Nat Rev Genet 2016;17:630–41.
49. Huang S. Tumor progression: Chance and necessity in Darwinian
and Lamarckian somatic (mutationless) evolution. Prog Biophys
Mol Biol 2012;110:69–86.
50. Darwiche N. Epigenetic mechanisms and the hallmarks of cancer:
an intimate affair. Am J Cancer Res 2020;10:1954–78.
51. Feng Y, Liu X, Pauklin S. 3D chromatin architecture and epigenetic
regulation in cancer stem cells. Protein Cell 2021;12:440–54.
52. Nam AS, Chaligne R, Landau DA. Integrating genetic and non-
genetic determinants of cancer evolution by single-cell multi-omics.
Nat Rev Genet 2021;22:3–18.
53. Bitman-Lotan E, Orian A. Nuclear organization and regulation of
the differentiated state. Cell Mol Life Sci CMLS 2021;78:3141–58.
54. Goldberg AD, Allis CD, Bernstein E. Epigenetics: a landscape takes
shape. Cell 2007;128:635–8.
55. Zeng Y, Chen T. DNA methylation reprogramming during mam-
malian development. Genes 2019;10:257.
56. Hegde AN, Smith SG. Recent developments in transcriptional and
translational regulation underlying long-term synaptic plasticity
and memory. Learn Mem 2019;26:307–17.
57. Kim S, Kaang B-K. Epigenetic regulation and chromatin remodeling
in learning and memory. Exp Mol Med 2017;49:e281.
58. Thienpont B, Van Dyck L, Lambrechts D. Tumors smother their
epigenome. Mol Cell Oncol 2016;3:e1240549.
59. Gameiro PA, Struhl K. Nutrient deprivation elicits a transcriptional
and translational inflammatory response coupled to decreased pro-
tein synthesis. Cell Rep 2018;24:1415–24.
60. Lin GL, Monje M. Understanding the deadly silence of posterior
fossa A ependymoma. Mol Cell 2020;78:999–1001.
61. Michealraj KA, Kumar SA, Kim LJY, Cavalli FMG, Przelicki D,
Wojcik JB, et al. Metabolic regulation of the epigenome drives lethal
infantile ependymoma. Cell 2020;181:1329–45.
62. Bakir B, Chiarella AM, Pitarresi JR, Rustgi AK. EMT, MET, plasticity,
and tumor metastasis. Trends Cell Biol 2020;30:764–76.
Hallmarks of Cancer: New Dimensions
63. Gupta PB, Pastushenko I, Skibinski A, Blanpain C, Kuperwasser C.
Phenotypic plasticity: driver of cancer initiation, progression, and
therapy resistance. Cell Stem Cell 2019;24:65–78.
64. Lambert AW, Weinberg RA. Linking EMT programmes to normal
and neoplastic epithelial stem cells. Nat Rev Cancer 2021;21:325–38.
65. Lindner P, Paul S, Eckstein M, Hampel C, Muenzner JK, Erlenbach-
Wuensch K, et  al. EMT transcription factor ZEB1 alters the epige-
netic landscape of colorectal cancer cells. Cell Death Dis 2020;11:147.
66. Javaid S, Zhang J, Anderssen E, Black JC, Wittner BS, Tajima K,
et  al. Dynamic chromatin modification sustains epithelial-mesen-
chymal transition following inducible expression of Snail-1. Cell
Rep 2013;5:1679–89.
67. Serrano-Gomez SJ, Maziveyi M, Alahari SK. Regulation of epithelial-
mesenchymal transition through epigenetic and post-translational
modifications. Mol Cancer 2016;15:18.
68. Skrypek N, Goossens S, De Smedt E, Vandamme N, Berx G.
Epithelial-to-mesenchymal transition: epigenetic reprogramming
driving cellular plasticity. Trends Genet TIG 2017;33:943–59.
69. Li L, Hanahan D. Hijacking the neuronal NMDAR signaling circuit
to promote tumor growth and invasion. Cell 2013;153:86–100.
70. Li L, Zeng Q, Bhutkar A, Galván JA, Karamitopoulou E, Noordermeer
D, et al. GKAP acts as a genetic modulator of NMDAR signaling to
govern invasive tumor growth. Cancer Cell 2018;33:736–51.
71. Mohammadi H, Sahai E. Mechanisms and impact of altered tumour
mechanics. Nat Cell Biol 2018;20:766–74.
72. Odenthal J, Takes R, Friedl P. Plasticity of tumor cell invasion:
governance by growth factors and cytokines. Carcinogenesis 2016;
37:1117–28.
73. Torres CM, Biran A, Burney MJ, Patel H, Henser-Brownhill T,
Cohen A-HS, et al. The linker histone H1.0 generates epigenetic and
functional intratumor heterogeneity. Science 2016;353:aaf1644.
74. Puram SV, Tirosh I, Parikh AS, Patel AP, Yizhak K, Gillespie S, et al.
Single-cell transcriptomic analysis of primary and metastatic tumor
ecosystems in head and neck cancer. Cell 2017;171:1611–24.
75. Kinker GS, Greenwald AC, Tal R, Orlova Z, Cuoco MS, McFarland
JM, et  al. Pan-cancer single-cell RNA-seq identifies recurring pro-
grams of cellular heterogeneity. Nat Genet 2020;52:1208–18.
76. Murtha M, Esteller M. Extraordinary cancer epigenomics: think-
ing outside the classical coding and promoter box. Trends Cancer
2016;2:572–84.
77. Nebbioso A, Tambaro FP, Dell’Aversana C, Altucci L. Cancer epige-
netics: moving forward. PLoS Genet 2018;14:e1007362.
78. Tavernari D, Battistello E, Dheilly E, Petruzzella AS, Mina M,
Sordet-Dessimoz J, et al. Non-genetic evolution drives lung adeno-
carcinoma spatial heterogeneity and progression. Cancer Discov
2021;11:1490–507.
79. Heyn H, Vidal E, Ferreira HJ, Vizoso M, Sayols S, Gomez A, et  al.
Epigenomic analysis detects aberrant super-enhancer DNA methyla-
tion in human cancer. Genome Biol 2016;17:11.
80. Saghafinia S, Mina M, Riggi N, Hanahan D, Ciriello G. Pan-cancer
landscape of aberrant DNA methylation across human tumors. Cell
Rep 2018;25:1066–80.
81. Audia JE, Campbell RM. Histone modifications and cancer. Cold
Spring Harb Perspect Biol 2016;8:a019521.
82. Corces MR, Granja JM, Shams S, Louie BH, Seoane JA, Zhou W,
et al. The chromatin accessibility landscape of primary human can-
cers. Science 2018;362:eaav1898.
83. Esteve-Puig R, Bueno-Costa A, Esteller M. Writers, readers and
erasers of RNA modifications in cancer. Cancer Lett 2020;474:
127–37.
84. Janin M, Coll-SanMartin L, Esteller M. Disruption of the RNA
modifications that target the ribosome translation machinery in
human cancer. Mol Cancer 2020;19:70.
85. Hanahan D, Coussens LM. Accessories to the crime: functions
of cells recruited to the tumor microenvironment. Cancer Cell
2012;21:309–22.
86. Lu Z, Zou J, Li S, Topper MJ, Tao Y, Zhang H, et al. Epigenetic ther-
apy inhibits metastases by disrupting premetastatic niches. Nature
2020;579:284–90.
Downloaded from cancerdiscovery.aacrjournals.org on January 12, 2022. © 2022 American Association for Cancer
Research.
REVIEW
87. Thomas S, Izard J, Walsh E, Batich K, Chongsathidkiet P, Clarke G,
et al. The host microbiome regulates and maintains human health:
a primer and perspective for non-microbiologists. Cancer Res
2017;77:1783–812.
88. Dzutsev A, Badger JH, Perez-Chanona E, Roy S, Salcedo R, Smith CK,
et al. Microbes and cancer. Annu Rev Immunol 2017;35:199–228.
89. Helmink BA, Khan MAW, Hermann A, Gopalakrishnan V, Wargo JA.
The microbiome, cancer, and cancer therapy. Nat Med 2019;25:
377–88.
90. Sears CL, Garrett WS. Microbes, microbiota, and colon cancer. Cell
Host Microbe 2014;15:317–28.
91. Pleguezuelos-Manzano C, Puschhof J, Rosendahl Huber A, van
Hoeck A, Wood HM, Nomburg J, et  al. Mutational signature in
colorectal cancer caused by genotoxic pks+ E. coli. Nature 2020;580:
269–73.
92. Okumura S, Konishi Y, Narukawa M, Sugiura Y, Yoshimoto S,
Arai  Y, et  al. Gut bacteria identified in colorectal cancer patients
promote tumourigenesis via butyrate secretion. Nat Commun 2021;
12:5674.
93. Salvi PS, Cowles RA. Butyrate and the intestinal epithelium: modu-
lation of proliferation and inflammation in homeostasis and dis-
ease. Cells 2021;10:1775.
94. Fessler J, Matson V, Gajewski TF. Exploring the emerging role of
the microbiome in cancer immunotherapy. J Immunother Cancer
2019;7:108.
95. Gopalakrishnan V, Helmink BA, Spencer CN, Reuben A, Wargo JA.
The influence of the gut microbiome on cancer, immunity, and
cancer immunotherapy. Cancer Cell 2018;33:570–80.
96. Zitvogel L, Ma Y, Raoult D, Kroemer G, Gajewski TF. The micro-
biome in cancer immunotherapy: diagnostic tools and therapeutic
strategies. Science 2018;359:1366–70.
97. Baruch EN, Youngster I, Ben-Betzalel G, Ortenberg R, Lahat A, Katz L,
et  al. Fecal microbiota transplant promotes response in immuno-
therapy-refractory melanoma patients. Science 2021;371:602–9.
98. Davar D, Dzutsev AK, McCulloch JA, Rodrigues RR, Chauvin J-M,
Morrison RM, et  al. Fecal microbiota transplant overcomes resist-
ance to anti–PD-1 therapy in melanoma patients. Science 2021;
371:595–602.
99. Griffin ME, Espinosa J, Becker JL, Luo J-D, Carroll TS, Jha JK,
et al. Enterococcus peptidoglycan remodeling promotes checkpoint
inhibitor cancer immunotherapy. Science 2021;373:1040–6.
100. Mager LF, Burkhard R, Pett N, Cooke NCA, Brown K, Ramay H,
et  al. Microbiome-derived inosine modulates response to check-
point inhibitor immunotherapy. Science 2020;369:1481–9.
101. Ansaldo E, Belkaid Y. How microbiota improve immunotherapy.
Science 2021;373:966–7.
102. Ansaldo E, Farley TK, Belkaid Y. Control of immunity by the micro-
biota. Annu Rev Immunol 2021;39:449–79.
103. Zhang Q, Ma C, Duan Y, Heinrich B, Rosato U, Diggs LP, et al. Gut
microbiome directs hepatocytes to recruit MDSCs and promote
cholangiocarcinoma. Cancer Discov 2021;11:1248–67.
104. Ding T, Schloss PD. Dynamics and associations of microbial com-
munity types across the human body. Nature 2014;509:357–60.
105. Byrd AL, Liu M, Fujimura KE, Lyalina S, Nagarkar DR, Charbit B,
et  al. Gut microbiome stability and dynamics in healthy donors
and patients with non-gastrointestinal cancers. J Exp Med 2021;
218:e20200606.
106. Healy CM, Moran GP. The microbiome and oral cancer: more ques-
tions than answers. Oral Oncol 2019;89:30–3.
107. Swaney MH, Kalan LR. Living in your skin: microbes, molecules and
mechanisms. Infect Immun 2021;89:e00695–20.
108. Willis JR, Gabaldón T. The human oral microbiome in health
and disease: from sequences to ecosystems. Microorganisms 2020;
8:308.
109. Xu J, Peng J-J, Yang W, Fu K, Zhang Y. Vaginal microbiomes and
ovarian cancer: a review. Am J Cancer Res 2020;10:743–56.
110. Nejman D, Livyatan I, Fuks G, Gavert N, Zwang Y, Geller LT, et al.
The human tumor microbiome is composed of tumor type-specific
intracellular bacteria. Science 2020;368:973–80.
JANUARY 2022 CANCER DISCOVERY | 45
REVIEW
111. Jin C, Lagoudas GK, Zhao C, Bullman S, Bhutkar A, Hu B, et  al.
Commensal microbiota promote lung cancer development via γδ T
cells. Cell 2019;176:998–1013.
112. Pushalkar S, Hundeyin M, Daley D, Zambirinis CP, Kurz E,
Mishra A, et al. The pancreatic cancer microbiome promotes onco-
genesis by induction of innate and adaptive immune suppression.
Cancer Discov 2018;8:403–16.
113. McAllister F, Khan MAW, Helmink B, Wargo JA. The tumor micro-
biome in pancreatic cancer: bacteria and beyond. Cancer Cell
2019;36:577–9.
114. Kadosh E, Snir-Alkalay I, Venkatachalam A, May S, Lasry A, Elyada
E, et  al. The gut microbiome switches mutant p53 from tumour-
suppressive to oncogenic. Nature 2020;586:133–8.
115. Birch J, Gil J. Senescence and the SASP: many therapeutic avenues.
Genes Dev 2020;34:1565–76.
116. Faget DV, Ren Q, Stewart SA. Unmasking senescence: context-
dependent effects of SASP in cancer. Nat Rev Cancer 2019;19:
439–53.
117. Gorgoulis V, Adams PD, Alimonti A, Bennett DC, Bischof O,
Bishop C, et  al. Cellular senescence: defining a path forward. Cell
2019;179:813–27.
118. He S, Sharpless NE. Senescence in health and disease. Cell 2017;
169:1000–11.
119. Kowald A, Passos JF, Kirkwood TBL. On the evolution of cellular
senescence. Aging Cell 2020;19:e13270.
46 | CANCER DISCOVERY JANUARY 2022 AACRJournals.org
Downloaded from cancerdiscovery.aacrjournals.org on January 12, 2022. © 2022 American Association for Cancer
Research.
Hanahan
120. Lee S, Schmitt CA. The dynamic nature of senescence in cancer. Nat
Cell Biol 2019;21:94–101.
121. Wang B, Kohli J, Demaria M. Senescent cells in cancer therapy:
friends or foes? Trends Cancer 2020;6:838–57.
122. Baker DJ, Childs BG, Durik M, Wijers ME, Sieben CJ, Zhong J, et al.
Naturally occurring p16(Ink4a)-positive cells shorten healthy lifes-
pan. Nature 2016;530:184–9.
123. Ruhland MK, Loza AJ, Capietto A-H, Luo X, Knolhoff BL,
Flanagan KC, et al. Stromal senescence establishes an immunosup-
pressive microenvironment that drives tumorigenesis. Nat Com-
mun 2016;7:11762.
124. Hwang HJ, Lee Y-R, Kang D, Lee HC, Seo HR, Ryu J-K, et  al.
Endothelial cells under therapy-induced senescence secrete CXCL11,
which increases aggressiveness of breast cancer cells. Cancer Lett
2020;490:100–10.
125. Wang D, Xiao F, Feng Z, Li M, Kong L, Huang L, et  al. Sunitinib
facilitates metastatic breast cancer spreading by inducing endothe-
lial cell senescence. Breast Cancer Res 2020;22:103.
126. Amor C, Feucht J, Leibold J, Ho Y-J, Zhu C, Alonso-Curbelo D, et al.
Senolytic CAR T cells reverse senescence-associated pathologies.
Nature 2020;583:127–32.
127. Aster JC, Pear WS, Blacklow SC. The varied roles of notch in cancer.
Annu Rev Pathol 2017;12:245–75.
128. Kuczynski EA, Vermeulen PB, Pezzella F, Kerbel RS, Reynolds AR.
Vessel co-option in cancer. Nat Rev Clin Oncol 2019;16:469–93.
Hallmarks of Cancer: New Dimensions
Douglas Hanahan

Cancer Discov 2022;12:31-46.

Updated version Access the most recent version of this article at:
http://cancerdiscovery.aacrjournals.org/content/12/1/31

Cited articles This article cites 126 articles, 28 of which you can access for free at:
http://cancerdiscovery.aacrjournals.org/content/12/1/31.full#ref-list-1

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications
Subscriptions Department at pubs@aacr.org.

Permissions To request permission to re-use all or part of this article, use this link
http://cancerdiscovery.aacrjournals.org/content/12/1/31.
Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC)
Rightslink site.

Downloaded from cancerdiscovery.aacrjournals.org on January 12, 2022. © 2022 American Association for Cancer
Research.
 VOLUME 34 • NUMBER
JOURNAL OF CLINICAL ONCOLOGY
All authors: Massachusetts General
Hospital, Harvard Medical School, Boston,
MA.
Published online ahead of print at
www.jco.org on December 7, 2015.
Supported by Award No. R01CA139118
(A.G.T.) and Award No. P50CA089393
(A.G.T.) from the National Cancer Institute,
and the Adele McKinnon Research Fund
for Breast Cancer-Related Lymphedema.
Terms in blue are defined in the glossary,
found at the end of this article and online
at www.jco.org
Presented at the San Antonio Breast
Cancer Symposium, San Antonio, TX,
December 9-12, 2014.
The content is solely the responsibility of
the authors and does not necessarily
represent the official views of the National
Cancer Institute or the National Institutes
of Health.
Authors’ disclosures of potential conflicts
of interest are found in the article online at
www.jco.org. Author contributions are
found at the end of this article.
Corresponding author: Alphonse G.
Taghian, MD, PhD, Department of
Radiation Oncology, Massachusetts
General Hospital, 100 Blossom St,
Boston, MA 02114; e-mail: ataghian@
partners.org.
© 2015 by American Society of Clinical
Oncology
0732-183X/16/3407w-691w/$20.00
DOI: 10.1200/JCO.2015.61.5948
7 • MARCH 1, 2016
O R I G I N A L R E P O R T
Impact of Ipsilateral Blood Draws, Injections, Blood Pressure
Measurements, and Air Travel on the Risk of Lymphedema for
Patients Treated for Breast Cancer
Chantal M. Ferguson, Meyha N. Swaroop, Nora Horick, Melissa N. Skolny, Cynthia L. Miller, Lauren S. Jammallo,
Cheryl Brunelle, Jean A. O’Toole, Laura Salama, Michelle C. Specht, and Alphonse G. Taghian
See accompanying article on page 655
A B S T R A C T
Purpose
The goal of this study was to investigate the association between blood draws, injections, blood
pressure readings, trauma, cellulitis in the at-risk arm, and air travel and increases in arm volume in a
cohort of patients treated for breast cancer and screened for lymphedema.
Patients and Methods
Between 2005 and 2014, patients undergoing treatment of breast cancer at our institution were
screened prospectively for lymphedema. Bilateral arm volume measurements were performed
preoperatively and postoperatively using a Perometer. At each measurement, patients reported the
number of blood draws, injections, blood pressure measurements, trauma to the at-risk arm(s),
and number of flights taken since their last measurement. Arm volume was quantified using the
relative volume change and weight-adjusted change formulas. Linear random effects models were
used to assess the association between relative arm volume (as a continuous variable) and non-
treatment risk factors, as well as clinical characteristics.
Results
In 3,041 measurements, there was no significant association between relative volume change
or weight-adjusted change increase and undergoing one or more blood draws (P = .62), injections
(P = .77), number of flights (one or two [P = .77] and three or more [P = .91] v none), or duration of
flights (1 to 12 hours [P = .43] and 12 hours or more [P = .54] v none). By multivariate analysis, factors
significantly associated with increases in arm volume included body mass index $ 25 (P = .0236), axillary
lymph node dissection (P , .001), regional lymph node irradiation (P = .0364), and cellulitis (P , .001).
Conclusion
This study suggests that although cellulitis increases risk of lymphedema, ipsilateral blood draws, injections,
blood pressure readings, and air travel may not be associated with arm volume increases. The results may
help to educate clinicians and patients on posttreatment risk, prevention, and management of lymphedema.
J Clin Oncol 34:691-698. © 2015 by American Society of Clinical Oncology
the efficacy of such precautionary behaviors do
INTRODUCTION
not exist, highlighted in a recent statement by the
National Lymphedema Network.5 The guidelines
Clinicians and national guidelines strongly advise place a large amount of burden on patients and
patients with breast cancer to avoid blood draws, clinicians, who go to great lengths to exercise
injections, blood pressure readings, and trauma to precautionary behaviors and face high levels of
the at-risk arm during and after treatment to anxiety when they accidentally do not abide by the
reduce the risk of developing cellulitis and breast guidelines. Therefore, we sought to investigate the
cancer–related lymphedema (BCRL).1-6 Patients association between blood draws, injections,
are also advised to exercise caution when flying by blood pressure readings, trauma, and cellulitis in
wearing prophylactic compression sleeves.1,3,5 the at-risk arm and flying on increases in arm
These guidelines are based on anecdotal infor- volume in a large prospective cohort of patients
mation, and comprehensive data demonstrating undergoing treatment of breast cancer.
© 2015 by American Society of Clinical Oncology 691
 Ferguson et al
Lymphedema is a potential adverse effect of breast cancer
treatment characterized by arm swelling, discomfort, and impaired
upper extremity function in its later stages. The development of
lymphedema is the result of a compromised lymphatic drainage
system after the removal of lymph nodes or radiation to the axillary
lymph nodes. Patients with breast cancer carry a lifelong risk of
developing lymphedema; the average time to onset is 14.4 months
after treatment completion.7,8 Consequently, if patients comply
with cautionary guidelines, they must exercise risk-reducing
practices for the rest of their lives.
Well-defined risk factors for developing lymphedema include
axillary lymph node dissection (ALND),9-15 regional lymph node
irradiation (RLNR),8,16-19 higher body mass index (BMI), and
older age at diagnosis.7,8,11,14,20-30 It remains unclear why, among
patients with similar demographic and treatment-related charac-
teristics, some go on to develop lymphedema. This variation has
prompted speculation that events including blood draws, injec-
tions, blood pressure readings, trauma to the arm, air travel, and
cellulitis might incite lymphedema.6,31 Despite the prevalence and
persistence of recommendations to pursue precautionary behavior
after breast cancer treatment, few data exist to support these
practices, and investigation into the topic is warranted.5
PATIENTS AND METHODS
Study Design
Patients with newly diagnosed breast cancer between 2009 and 2014
who were prospectively screened for lymphedema at our institution were
included in this analysis. Bilateral arm volume measurements were obtained
using an optoelectric Perometer (Pero-system, Wuppertal, Germany), which
uses infrared light to measure limb circumference and calculate arm volume.
All patients had a preoperative baseline measurement, a postoperative
measurement, a measurement after chemotherapy and/or radio therapy, and
measurements at regular follow-up intervals corresponding with oncology
visits.32-35 Regular follow-up intervals correspond to time periods of between
3 and 7 months. Occasionally patients are measured more frequently at their
request. The protocol for lymphedema screening has been previously
published and approved by the institutional review board33 (ClinicalTrials.
gov identification number NCT01521741).
Arm volume changes were quantified using the relative volume change
(RVC)33 formula after unilateral breast surgery or the weight-adjusted volume
change (WAC) formula after bilateral breast surgery.34 The RVC equation
accounts for preoperative differences in arm volume by using a baseline
measurement and for nontreatment-related changes, including weight gain or
loss, by comparing the surgical arm to the nonsurgical arm.33,34 Briefly, RVC =
{[A(2)U(1)/U(2)A(1)} 2 1), where A(1) and A(2) are arm volumes on the
surgical or ipsilateral side at preoperative and postoperative measurements, and
U(1) and U(2) are arm volumes on the contralateral side at corresponding time
points.33 The WAC calculates changes in arm volume compared with a
preoperative arm measurement for each arm independently and accounts for
weight changes that could influence arm size.34 RVC and WAC have been
shown to classify lymphedema similarly.34 Lymphedema was defined as RVC
or WAC $ 10%.7,15,36
At each Perometer measurement, patients were asked to complete a
survey on which they reported the number of blood draws, injections,
blood pressure readings, and trauma to the at-risk arm(s); number of
flights and the length of the flight since their last measurement; and use of
compression sleeve while flying. Self-reported trauma ranged from
bruising to arm fractures.
This analysis includes information from each patient visit at which
both an arm volume measurement and a survey were obtained.
692 © 2015 by American Society of Clinical Oncology
Patient Population
All 632 patients included in this study had a new diagnosis of invasive
breast cancer at their baseline measurement, more than 6 months of
postsurgical follow-up, and at least one follow-up arm measurement.
Patient and clinocopathologic characteristics were collected via medical
record review. Patients who undergo sentinel lymph node biopsy are still at
risk for developing lymphedema and are often advised to exercise risk
reduction practices just like those who undergo ALND. Therefore, both
groups were included in the study. Episodes of cellulitis were defined as
clinical signs of infection in the ipsilateral arm and/or breast that required
antibiotics. Each side was evaluated independently for patients who
underwent bilateral breast surgery, resulting in a total of 760 at-risk arms
available for analysis. In our hospital, the majority of patients who chose to
undergo contralateral prophylactic mastectomy undergo sentinel lymph
node biopsy on their unaffected breast to stage an occult breast cancer if
discovered on final pathology. Therefore, the unaffected breasts for patients
undergoing bilateral mastectomies were included in the analysis because
they are also at risk for developing lymphedema. Patients were censored
from the analysis after a diagnosis of distant metastases and recurrence and
if they wore a compression sleeve while flying.
Statistical Analysis
The Kaplan-Meier method was used to estimate the 2-year cumulative
incidence of lymphedema, defined by RVC/WAC $ 10%. Linear random
effects models were used to assess the association between relative arm volume
(as a continuous variable), clinical risk factors, and episodes of non-
precautionary behavior. These models account for the correlation between
arm volume measurements obtained from the same patient and, for patients
who underwent bilateral breast surgery, on the same side of the body. At each
measurement, the number of blood draws, injections, blood pressure
readings, trauma, and episodes of cellulitis were analyzed both as continuous
variables and as dichotomous variables, categorized according to whether
patients reported having had one or more events versus none. Number of and
hours spent on flights in total since last follow-up were analyzed as both
continuous and trichotomous variables. Model estimates give the mean
change in RVC or WAC associated with a one-unit increase in a continuous
risk factor and the mean difference in RVC or WAC between subgroups for
categorical variables. Univariate model results were used to estimate and plot
the mean RVC or WAC within each subgroup for categorical clinical and
nontreatment-related variables, along with the 95% CI for the mean and the P
value associated with the comparison of means. The multivariate model was
chosen by starting with a model that included all variables that were sig-
nificant at the .10 significance level in the univariate analysis, as well as all
nonprecautionary behaviors, and removing one variable at a time until only
significant (P , .05) variables remained. Because cellulitis is in the causal
pathway between the risk events and arm volume increase, it was not included
in the model selection process to avoid overadjustment bias.
Cox proportional hazards models were used to assess the association
between risk of cellulitis and risk events. Patients who developed cellulitis in the
postoperative period were excluded from this analysis because of lack of
information on risk events. In addition, in analyzing the association between risk
events and subsequent cellulitis, we evaluated the relationship between the risk
of infection and the number of risk events reported before the cellulitis episode.
RESULTS
Patient Population
Six hundred thirty-two patients with a total of 3,041 post-
operative measurements were included. Median age at diagnosis
was 52 years (range, 28 to 81 years), and median BMI was 26 lb/in2
(range, 16 to 59 lb/in2; Table 1); 92.1% of the cohort was white, 0.6%
Hispanic, 2% African American, 2.7% Asian, and 2.6% unknown.
Patients were followed for a median of 24 months (6 to 60 months)
JOURNAL OF CLINICAL ONCOLOGY
 Risk Reduction Practices for Lymphedema

Table 1. Summary of Clinicopathologic Factors Stratified by Lymphedema Table 1. Summary of Clinicopathologic Factors Stratified by Lymphedema
(RVC/WAC $ 10%) (RVC/WAC $ 10%) (continued)
Lymphedema Lymphedema
No (n = 697) Yes (n = 63) Overall (N = 760) No (n = 697) Yes (n = 63) Overall (N = 760)
Age at diagnosis, years* 52 51 52 Chest wall only
BMI, lb/in2* 25.4 29.1 25.6 No. 316 16 332
Follow-up, months* 23.59 34.31 24.03 % 45.3 25.4 43.7
No. of postoperative visits* 4 5 4 RLNR
Months between visits* 7.14 6.45 7.09 No. 130 30 160
Tumor type % 18.7 47.6 21.1
None Not available
No. 112 13 125 No. 3 0 3
% 16.1 20.6 16.4 % 0.4 0 0.4
Invasive Cellulitis
No. 583 50 633 No
% 83.6 79.4 83.3 No. 659 56 715
Not available % 94.5 88.9 94.1
No. 2 0 2 Yes
% 0.3 0 0.3 No. 38 7 45
Laterality % 5.5 11.1 5.9
Unilateral
No. 473 31 504 Abbreviations: ALND, axillary lymph node dissection; BMI, body mass index;
RNLR, regional lymph node irradiation; RVC, relative volume change; SLNB,
% 67.9 49.2 66.3
sentinel lymph node biopsy; WAC, weight-adjusted volume change.
Bilateral *Values shown, median.
No. 224 32 256
% 32.1 50.8 33.7
Breast surgery
Lumpectomy and had a median of four postoperative visits (one to 16). The
No. 359 18 377
% 51.5 28.6 49.6
median time between measurements was 7 months (interquartile
Mastectomy range, 6.2 months). Thirteen percent of patients had one mea-
No. 338 45 383 surement, 34% had two to three measurements, 40% had four to six
% 48.5 71.4 50.4 measurements, and 13% had seven to 16 measurements. By medical
Axillary surgery
None
record review, 5.9% (45 of 760) of treated breasts had at least one
No. 53 7 60 episode of cellulitis any time after surgery requiring antibiotics
% 7.6 11.1 7.9 (Table 1); 47% (21 of 45) of cellulitis episodes occurred in the
SLNB
immediate postoperative period and therefore before the first
No. 519 22 541
% 74.5 34.9 71.2 postoperative arm measurement. At 24 months, the cumulative
ALND incidence of BCRL as defined by an RVC or WAC $ 10% was 7.72%
No. 125 34 159 (95% CI, 5.82% to 10.22%).
% 17.9 54 20.9
Neoadjuvant chemotherapy
No Nonprecautionary Behaviors
No. 606 50 656
% 86.9 79.4 86.3
In 8.5% (251 of 2,965) of responses, patients reported having
Yes one or more blood draw in their affected arm(s) since their last
No. 91 13 104 measurement, 2.1% (63 of 2,961) of responses reported having one
% 13.1 20.6 13.7 or more injection, 16.3% (482 of 2,961) of responses reported
Adjuvant chemotherapy
No
having one or more blood pressure measurements taken, and 1.0%
No. 416 25 441 (37 of 2,999) of responses reported trauma to the at-risk arm
% 59.7 39.7 58 (Table 2). Thirty percent (878 of 2,960) of patients had flown since
Yes their last measurement (Table 2).
No. 281 38 319
% 40.3 60.3 42
Hormonal therapy Univariate and Multivariate Analyses
No
No. 121 12 133 By univariate analysis, there was no significant association
% 17.4 19 17.5 between an increased RVC or WAC and undergoing one or more
Yes blood draws (P = .62), injections (P = .77), trauma to the at-risk
No. 576 51 627
arm (P = .08), number of flights (one or two [P = .77] and three or
% 82.6 81 82.5
Radiation therapy more [P = .91] v none), or duration of flights (1 to 12 hours [P =
None .43] and 12 hours or more [P = .54] v none) (Fig 1). Having a
No. 248 17 265 BMI $ 25 lb/in2 at time of diagnosis (P = .0064), undergoing
% 35.6 27 34.9
ALND (P = .0003), having blood pressure readings (P = .034),
(continued in next column)
RLNR (P , .001), and cellulitis (P , .001) were significantly
associated with arm volume increases (Table 3; Fig 1). When

www.jco.org © 2015 by American Society of Clinical Oncology 693

 Ferguson et al

analyzed as continuous variables, age at diagnosis (P = .25), the

Table 2. Summary of Reported Risk Events
number of ipsilateral blood draws (P = .92), injections (P = .85),
No. of Measurements With Event blood pressure readings (P = .15), and number (P = .34) or duration
Lymphedema of flights (P = .98) were not associated with an increase in RVC or
Risk Factor No (n = 192) Yes (n = 2,849) Overall (N = 3,041) WAC by univariate analysis. In a subset of patients who underwent
Blood draws ALND, BMI $ 25 lb/in2 (P = .0051) and cellulitis (P , .001)
None remained significant, but blood draws (P = .26), injections (P = .35),
No. 2,538 176 2,714
% 89.1 91.7 89.2
blood pressures (P = .39), trauma (P = .23), and air travel (one or
1+ two [P = .96] and three or more [P = .88] v no flights) or duration of
No. 241 10 251 flights (1 to 12 hours [P = .85] and 12 hours or more [P = .69] v
% 8.5 5.2 8.3
Not available
none) were not associated with increased arm swelling.
No. 70 6 76 By multivariate analysis, factors significantly associated with
% 2.5 3.1 2.5 increases in arm volume included BMI $ 25 lb/in2 (P = .0236),
Injections
None
ALND (P ,.001), RLNR (P = .0364), and cellulitis (P ,.001) (Fig 2).
No. 2,726 183 2,909 None of the nonprecautionary behaviors were significantly
% 95.7 95.3 95.7 associated with arm volume change in the multivariate analysis.
1+
No. 60 3 63
Further investigation of the association between risk events
% 2.1 1.6 2.1 and subsequent cellulitis revealed that blood draws (hazard ratio
Not available [HR], 0.977; P = .91), injections (HR, 1.101; P = .5), and blood
No. 63 6 69
% 2.2 3.1 2.3
pressure readings (HR, 0.943; P = .1) on the ipsilateral arm were
Blood pressure not significantly associated with cellulitis.
None
No. 2,305 174 2,479
% 80.9 90.6 81.5
1+ DISCUSSION
No. 470 12 482
% 16.5 6.3 15.9
Not available
A large, prospective cohort of patients with breast cancer was
No. 74 6 80 analyzed to investigate the association of nonprecautionary
% 2.6 3.1 2.6 behaviors and the subsequent risk of lymphedema. No significant
Trauma
No association was found between undergoing ipsilateral blood draws,
No. 2,780 182 2,962 injections, and blood pressure readings, and trauma to the at-risk
% 97.6 94.8 97.4 arm, air travel, and arm swelling. Increase in arm swelling was
Yes
No. 29 8 37 associated with ALND, RLNR, BMI $ 25 lb/in2, and cellulitis.
% 1 4.2 1.2 Previous reports postulate that blood draws, injections, blood
Not available pressure readings, and trauma can cause infection and injury to the
No. 40 2 42
% 1.4 1 1.4 at-risk arm.3,6,37-39 The theory is that patients with surgically
Flights, No. removed lymph nodes have compromised lymphatic systems that
None cannot clear fluid resulting from an immune response, leading to
No. 1,949 133 2,082
% 68.4 69.3 68.5 build-up of lymphatic fluid in the arm and subsequent lymph-
1-2 edema.37 One of the first papers to recommend precautionary
No. 453 28 481 behaviors was published in 1992 as a case report of a woman who
% 15.9 14.6 15.8
3+ developed swelling 30 years after treatment in response to a finger
No. 389 8 397 pinprick.40 Smith37 advised avoidance of traumatic events to the at-
% 13.7 4.2 13.1
risk arm after examining 10 patients with BCRL, four of whom
Not available
No. 58 23 81 reported blood draws as triggering their arm swelling. These
% 2 12 2.7 studies are anecdotal and based on information from a small
Flights, hours
number of patients. Despite minimal data, the authors provided
None
No. 1,946 132 2,078 persistent recommendations to avoid blood draws, blood pressure
% 68.3 68.8 68.3 readings, and injections in the at-risk arm to reduce the risk of
1-12
lymphedema.6,31,37,41 The results of our prospective study provide
No. 429 23 452
% 15.1 12 14.9 evidence-based data that call these guidelines into question.
12+ Clark et al41 evaluated 188 patients with breast cancer who
No. 408 13 421
% 14.3 6.8 13.8
underwent ALND and correlated skin punctures with risk of
Not available developing lymphedema. With a median follow-up of 3 years,
No. 66 24 90 20.7% (39 of 188) of patients developed arm lymphedema, 9.6%
% 2.3 12.5 3
(18 of 188) of all patients had a skin puncture, and therefore 44%
(8 of 18) of patients with skin puncture developed lymphedema
during the follow-up period.41 This analysis performed a mea-
surement for lymphedema at 6 months and 3 years after surgery

694 © 2015 by American Society of Clinical Oncology JOURNAL OF CLINICAL ONCOLOGY

 Risk Reduction Practices for Lymphedema

Risk Factor P 95% CI

No blood draws –0.22 to 0.49

1+ Blood draws .617 –0.58 to 0.59
No injections –0.22 to 0.48
1+ Injections .7697 –1.03 to1.01
No blood pressures –0.15 to 0.57
1+ Blood pressures .0338 –0.72 to 0.25
No trauma –0.24 to 0.47
Fig 1. Univariate analysis of association
Trauma .0811 –0.06 to 2.58 between risk events and arm volume
No cellulitis –0.36 to 0.34 increases. RVC, relative volume change;
WAC, weight-adjusted volume change.
Cellulitis < .001 1.99 to 4.13
No flights –0.28 to 0.45
1–2 Flights .7667 –0.33 to 0.61
3+ Flights .9133 –0.45 to 0.57
No flying hours –0.26 to 0.46
1–12 Flying hours .4261 –0.22 to 0.74
12+ Flying hours .5377 –0.54 to 0.46

–2.00% –1.00% 0.00% 1.00% 2.00% 3.00% 4.00%

Mean Difference in Arm Volume (RVC/WAC) in Subgroup

and did not specify the timing of skin puncture; therefore, it did In contrast, McLaughlin et al6 found no association in risk-
not take into account other events that could have contributed to reducing behaviors of patients with and without lymphedema
lymphedema.42 In our subset analysis of 159 patients undergoing when evaluated using tape measured circumferential arm mea-
ALND, nonprecautionary risk behaviors and specifically skin surements. Circumferential tape measurement is limited because
puncture were not correlated with increased arm swelling. By the measurements can be subject to both inter- and intrapersonal
performing arm measurements at regularly timed intervals for as variation.43 Unlike tape measurements, measurements by the
long as 5 years and recording the number of nonprecautionary Perometer and use of the RVC and WAC formulas account for
events since the previous measurement, we were able to reduce the preexisting arm volume differences and detect small volume
recall bias and more specifically evaluate effects of nonprecau- changes. In addition, obtaining reproducible, prospective arm
tionary behavior on arm swelling. measurements using the Perometer at the same time as collecting
individual incidence of risk-related events could have biased our
data to attribute small volume changes that may have not rep-
Table 3. Univariate Analysis of Categorical Treatment Factors resented swelling detectable via circumferential tape measurement
Patient- or Treatment- Mean RVC/WAC to risk events. Similarly, we found no association with non-
Related Risk Factor in Subgroup (%) 95% CI P precautionary behavior and increase in arm volume.
Surgical characteristics The Physical Activity and Lymphedema (PAL) trial assessed
Lumpectomy 0.04 20.42 to 0.51 —* the safety of progressive strength training in breast cancer sur-
Mastectomy v 0.26 20.28 to 0.79 .5542
lumpectomy vivors and investigated the association of 30 risk factors and arm
No nodal surgery 20.22 21.22 to 0.78 — swelling.44,45 The authors reported no significant association
SLNB v none 20.35 20.73 to 0.04 .8113 between ipsilateral blood draws, blood pressure readings, and air
ALND v none 1.89 1.22 to 2.55 ,.001
travel and incident arm swelling.44 The study included clinical
Radiation therapy
None 20.00 20.43 to 0.43 — evaluation of arm swelling using water displacement and patient-
Breast/chest wall 20.20 20.65 to 0.25 .42341 reported risk-reducing behaviors at 3, 6, and 12 months after
irradiation v none enrollment.44,45 Patients were given compression sleeves to wear
Breast/chest wall + 1.29 0.65 to 1.88 ,.001
RLNR v none during exercise and participated in a biweekly, year-long exercise
BMI, lb/in2 program. The incidence of lymphedema as defined as interlimb
, 25 20.43 20.98 to 0.11 — difference increase of $ 5% was low, which may be due to wearing
$ 25 0.57 0.10 to 1.03 .0064
a compression sleeve or the exercise program.44,45 In addition, the
Abbreviations: ALND, axillary lymph node dissection; BMI, body mass index; incidence of nonprecautionary behaviors was also low, which may be
RLNR, regional lymph node irradiation; RVC, relative volume change; WAC,
weight-adjusted volume change. due to the education about lymphedema provided to participants.
*Indicates the specified variable or comparison was not analyzed. Patients in our cohort do not routinely receive recommendations to
avoid blood draws, injections, or blood pressure readings or to wear a

www.jco.org © 2015 by American Society of Clinical Oncology 695

 Ferguson et al
BMI ≥ 25
ALND
RLNR
Cellulitis
–1.00% 0.00% 1.00% 2.00% 3.00% 4.00%
Mean Difference in Arm Volume (RVC/WAC) for Risk Factor
compressive sleeve while flying. However, our incidence of risk events
was low, and we enhanced our patient cohort to include patients with
both unilateral and bilateral breast surgery, increasing both the
number of risk events and the overall incidence of lymphedema.
It has been reported that flying incites the development
of lymphedema because of changes in cabin air pressure.6,46 Many
clinicians recommend using risk-reducing compression sleeves when
flying based on few, mostly retrospective, studies.47 In these studies it
is unclear if the reported lymphedema was preexisting or influenced by
other factors. Kilbreath et al48 evaluated changes in arm volume after
short- and long-distance flights and found no change in arm volume
as measured by bioimpedance. In our analysis, we prospectively col-
lected information regarding air travel and quantified arm swelling
with short intervals of follow-up. This design enabled the investigation
of the immediate effects of flying on arm volume and minimized the
risk of recall bias due to the shorter interval of follow up.
In our study, BMI $ 25 lb/in2 at diagnosis, ALND, RLNR, and
cellulitis were significantly associated with arm swelling. High
BMI, RLNR, and ALND are consistently cited in the literature as
risk factors for arm swelling and lymphedema, and our data further
support these claims.6,8,10,11,13-16,19,20,24,49,50 The increased risk for
developing lymphedema after infection including cellulitis is also
well documented in the literature,2,6,22,51 and our results support
these findings. Interestingly in our cohort, episodes of cellulitis
were not associated with a risk event that occurred before the
cellulitis. This may be due to a lower incidence of risk events or to
an overall lower risk of infection after risk events in a modern era of
attention to sterility.
Limitations of the study include a relatively short median
follow-up time of 24 months, low incidence of risk events,
potential for recall bias, and lack of information about patients’
receipt of physical therapy. As lymphedema has been shown to
develop an average of 14.4 months after surgery,8 longer follow-up
time would strengthen the study. Although efforts were made to
ensure regular follow-up visits, future studies using repeat ques-
tionnaires and sensitivity analyses might elucidate the effects of
recall bias. In addition, our analysis did not adjust for the number
of positive lymph nodes because we have not found it to be sig-
nificant in a previous analysis; however, this is an important area
for future investigation.52 Finally, the study did not capture and
696 © 2015 by American Society of Clinical Oncology
P 95% CI
.0236 0.06% to 1.38%
< .001 0.94% to 2.55% Fig 2. Multivariate analysis of risk factors
associated with arm volume increases. ALND,
.0364 0.05% to 1.31% axillary lymph node dissection; BMI, body mass
index (in lb/in2); RNLR, regional lymph node
irradiation; RVC, relative volume change; WAC,
< .001 1.72% to 3.80%
weight-adjusted volume change.
therefore was not able to account for women who had physical
therapy, which might have placed them at lower risk for developing
swelling, even after risk events.
This study represents the largest cohort of patients with breast
cancer prospectively screened for lymphedema using preoperative
arm measurements to determine the impact of risk-reducing
practices on arm swelling. We evaluated the association between
risk events and changes in arm volume and found no significant
associations between air travel, blood draws, injections, and blood
pressure readings in the at-risk arm with arm volume increase.
Patients who have a BMI $ 25 lb/in2, underwent ALND or RLNR,
and have had an episode of cellulitis should be more closely
monitored for changes in arm volume because of the significant
association with arm volume increases. The recommendations set
forth in the studies discussed still persist today and unfortunately are
supported with few evidence-based data. Although we cannot
affirmatively state that risk-reduction practices have no effect on arm
swelling, we hope to generate evidence that brings reasonable doubt
to burdensome guidelines and encourage further investigation into
nonprecautionary behaviors and the risk of lymphedema.
AUTHORS’ DISCLOSURES OF POTENTIAL CONFLICTS
OF INTEREST
Disclosures provided by the authors are available with this article at
www.jco.org.
AUTHOR CONTRIBUTIONS
Conception and design: Chantal M. Ferguson, Meyha N. Swaroop, Melissa
N. Skolny, Cynthia L. Miller, Lauren S. Jammallo, Cheryl Brunelle, Jean A.
O’Toole, Michelle C. Specht, Alphonse G. Taghian
Collection and assembly of data: Chantal M. Ferguson, Meyha N.
Swaroop, Melissa N. Skolny, Cynthia L. Miller, Lauren S. Jammallo, Jean A.
O’Toole, Laura Salama, Alphonse G. Taghian
Data analysis and interpretation: Chantal M. Ferguson, Meyha N.
Swaroop, Nora Horick, Melissa N. Skolny, Cynthia L. Miller, Lauren S.
Jammallo, Cheryl Brunelle, Michelle C. Specht, Alphonse G. Taghian
Manuscript writing: All authors
Final approval of manuscript: All authors
JOURNAL OF CLINICAL ONCOLOGY

REFERENCES
1. American Cancer Society: Lymphedema: What
every woman with breast cancer should know—For
women at risk for lymphedema. http://www.cancer.
org/acs/groups/cid/documents/webcontent/002876-
pdf.pdf
2. Indelicato DJ, Grobmyer SR, Newlin H, et al:
Delayed breast cellulitis: An evolving complication of
breast conservation. Int J Radiat Oncol Biol Phys 66:
1339-1346, 2006
3. Loudon L, Petrek J: Lymphedema in women
treated for breast cancer. Cancer Pract 8:65-71,
2000
4. National Lymphedema Network: Supplement
to NLN position breast cancer screening: Screening
and early detection of breast cancer-related lym-
phedema—The imperative. http://www.lymphnet.
org/pdfDocs/PP_Lymphedema_BC_Supplement.pdf
5. National Lymphedema Network: Position
statement of the National Lymphedema Network:
Lymphedema risk reduction practices. http://www.
lymphnet.org/pdfDocs/nlnriskreduction.pdf
6. McLaughlin SA, Bagaria S, Gibson T, et al:
Trends in risk reduction practices for the prevention
of lymphedema in the first 12 months after breast
cancer surgery. J Am Coll Surg 216:380-389, 2013;
quiz 511-513
7. DiSipio T, Rye S, Newman B, et al: Incidence
of unilateral arm lymphoedema after breast cancer: A
systematic review and meta-analysis. Lancet Oncol
14:500-515, 2013
8. Specht MC, Miller CL, Russell TA, et al:
Defining a threshold for intervention in breast cancer-
related lymphedema: What level of arm volume
increase predicts progression? Breast Cancer Res
Treat 140:485-494, 2013
9. Shih YC, Xu Y, Cormier JN, et al: Incidence,
treatment costs, and complications of lymphedema
after breast cancer among women of working age: A
2-year follow-up study. J Clin Oncol 27:2007-2014,
2009
10. Mansel RE, Fallowfield L, Kissin M, et al:
Randomized multicenter trial of sentinel node biopsy
versus standard axillary treatment in operable breast
cancer: The ALMANAC Trial. J Natl Cancer Inst 98:
599-609, 2006
11. Norman SA, Localio AR, Kallan MJ, et al: Risk
factors for lymphedema after breast cancer treat-
ment. Cancer Epidemiol Biomarkers Prev 19:
2734-2746, 2010
12. Yang EJ, Park WB, Seo KS, et al: Longitudinal
change of treatment-related upper limb dysfunction
and its impact on late dysfunction in breast cancer
survivors: A prospective cohort study. J Surg Oncol
101:84-91, 2010
13. Lucci A, McCall LM, Beitsch PD, et al:
American College of Surgeons Oncology Group:
Surgical complications associated with sentinel
lymph node dissection (SLND) plus axillary lymph
node dissection compared with SLND alone in the
American College of Surgeons Oncology Group Trial
Z0011. J Clin Oncol 25:3657-3663, 2007
14. Park JH, Lee WH, Chung HS: Incidence and
risk factors of breast cancer lymphoedema. J Clin
Nurs 17:1450-1459, 2008
15. Tsai RJ, Dennis LK, Lynch CF, et al: The risk of
developing arm lymphedema among breast cancer
survivors: A meta-analysis of treatment factors. Ann
Surg Oncol 16:1959-1972, 2009
www.jco.org
Risk Reduction Practices for Lymphedema
16. Warren LE, Miller CL, Horick N, et al: The
impact of radiation therapy on the risk of lymph-
edema after treatment for breast cancer: A pro-
spective cohort study. Int J Radiat Oncol Biol Phys
88:565-571, 2014
17. Coriddi M, Khansa I, Stephens J, et al: Analysis
of factors contributing to severity of breast cancer-
related lymphedema. Ann Plast Surg 74:22-25, 2015
18. Tiwari P, Coriddi M, Lamp S: Lymphedema
strategies for investigation and treatment: A review.
Plast Surg Nurs 32:173-177, 2012; quiz 178-179
19. Hayes SB, Freedman GM, Li T, et al: Does
axillary boost increase lymphedema compared with
supraclavicular radiation alone after breast con-
servation? Int J Radiat Oncol Biol Phys 72:1449-1455,
2008
20. Jammallo LS, Miller CL, Singer M, et al: Impact
of body mass index and weight fluctuation on lym-
phedema risk in patients treated for breast cancer.
Breast Cancer Res Treat 142:59-67, 2013
21. Ahmed RL, Schmitz KH, Prizment AE, Folsom
AR: Risk factors for lymphedema in breast cancer
survivors, the Iowa Women’s Health Study. Breast
Cancer Res Treat 130:981-991, 2011
22. McLaughlin SA, Wright MJ, Morris KT, et al:
Prevalence of lymphedema in women with breast
cancer 5 years after sentinel lymph node biopsy or
axillary dissection: Objective measurements. J Clin
Oncol 26:5213-5219, 2008
23. Wilke LG, McCall LM, Posther KE, et al: Sur-
gical complications associated with sentinel lymph
node biopsy: Results from a prospective international
cooperative group trial. Ann Surg Oncol 13:491-500,
2006
24. Soran A, Wu WC, Dirican A, et al: Estimating
the probability of lymphedema after breast cancer
surgery. Am J Clin Oncol 34:506-510, 2011
25. Swenson KK, Nissen MJ, et al: Case-control
study to evaluate predictors of lymphedema after
breast cancer surgery. Oncol Nurs Forum 36:
185-193, 2009
26. Nesvold IL, Dahl AA, Løkkevik E, et al: Arm and
shoulder morbidity in breast cancer patients after
breast-conserving therapy versus mastectomy. Acta
Oncol 47:835-842, 2008
27. Hayes SC, Janda M, Cornish B, et al: Lym-
phedema after breast cancer: Incidence, risk factors,
and effect on upper body function. J Clin Oncol 26:
3536-3542, 2008
28. Paskett ED, Naughton MJ, McCoy TP, et al:
The epidemiology of arm and hand swelling in pre-
menopausal breast cancer survivors. Cancer Epi-
demiol Biomarkers Prev 16:775-782, 2007
29. Crosby MA, Card A, Liu J, et al: Immediate
breast reconstruction and lymphedema incidence.
Plast Reconstr Surg 129:789e-795e, 2012
30. Ozaslan C, Kuru B: Lymphedema after treat-
ment of breast cancer. Am J Surg 187:69-72, 2004
31. Cole T: Risks and benefits of needle use in
patients after axillary node surgery. Br J Nurs 15: 969-
974, 976-979, 2006
32. Ancukiewicz M, Miller CL, Skolny MN, et al:
Comparison of relative versus absolute arm size
change as criteria for quantifying breast cancer-
related lymphedema: The flaws in current studies
and need for universal methodology. Breast Cancer
Res Treat 135:145-152, 2012
33. Ancukiewicz M, Russell TA, Otoole J, et al:
Standardized method for quantification of developing
lymphedema in patients treated for breast cancer. Int
J Radiat Oncol Biol Phys 79:1436-1443, 2011
n n n
34. Miller CL, Specht MC, Horick N, et al: A novel,
validated method to quantify breast cancer-related
lymphedema (BCRL) following bilateral breast sur-
gery. Lymphology 46:64-74, 2013
35. O’Toole J, Russell T, Singer M, et al: A general
hospital’s multi-disciplinary approach to BCRL.
J Lymphoedema 5:81-85, 2010
36. Armer JM, Heckathorn PW: Post-breast can-
cer lymphedema in aging women: Self-management
and implications for nursing. J Gerontol Nurs 31:
29-39, 2005
37. Smith J: The practice of venepuncture in
lymphoedema. Eur J Cancer Care (Engl) 7:97-98,
1998
38. Davis S: Lymphedema following breast cancer
treatment. Radiol Technol 70:42-56, 1998; quiz 57-60
39. Drury AN, Jones NW: Observations upon the
rate at which edema forms when the veins of the
human limb are congested. Heart 2:580-581, 1927
40. Brennan MJ, Weitz J: Lymphedema 30 years
after radical mastectomy. Am J Phys Med Rehabil
71:12-14, 1992
41. Clark B, Sitzia J, Harlow W: Incidence and risk
of arm oedema following treatment for breast cancer:
A three-year follow-up study. QJM 98:343-348, 2005
42. Cemal Y, Pusic A, Mehrara BJ: Preventative
measures for lymphedema: Separating fact from
fiction. J Am Coll Surg 213:543-551, 2011
43. Deltombe T, Jamart J, Recloux S, et al: Reli-
ability and limits of agreement of circumferential,
water displacement, and optoelectronic volumetry in
the measurement of upper limb lymphedema.
Lymphology 40:26-34, 2007
44. Showalter SL, Brown JC, Cheville AL, et al:
Lifestyle risk factors associated with arm swelling
among women with breast cancer. Ann Surg Oncol
20:842-849, 2013
45. Schmitz KH, Troxel AB, Cheville A, et al:
Physical Activity and Lymphedema (the PAL trial):
Assessing the safety of progressive strength training
in breast cancer survivors. Contemp Clin Trials 30:
233-245, 2009
46. Casley-Smith JR, Casley-Smith JR: Lymph-
edema initiated by aircraft flights. Aviat Space Environ
Med 67:52-56, 1996
47. Graham PH: Compression prophylaxis may
increase the potential for flight-associated lym-
phoedema after breast cancer treatment. Breast 11:
66-71, 2002
48. Kilbreath SL, Ward LC, Lane K, et al: Effect of
air travel on lymphedema risk in women with history
of breast cancer. Breast Cancer Res Treat 120:
649-654, 2010
49. Shahpar H, Atieh A, Maryam A, et al: Risk
factors of lymph edema in breast cancer patients. Int
J Breast Cancer 2013:641818, 2013
50. Graham P, Jagavkar R, Browne L, et al:
Supraclavicular radiotherapy must be limited laterally
by the coracoid to avoid significant adjuvant breast
nodal radiotherapy lymphoedema risk. Australas
Radiol 50:578-582, 2006
51. Mak SS, Yeo W, Lee YM, et al: Predictors of
lymphedema in patients with breast cancer under-
going axillary lymph node dissection in Hong Kong.
Nurs Res 57:416-425, 2008
52. Miller CL, Skolny M, Ferguson C, et al: Does
number of sentinel lymph nodes removed impact the
risk of breast cancer- related lymphedema (BCRL)?
The National Lymphedema Network Conference,
11th International Conference, Washington DC , Sep-
tember 3-7, 2014
© 2015 by American Society of Clinical Oncology 697
 Ferguson et al

GLOSSARY TERM

sentinel lymph node: the lymph node that is anatomically
located such that it is the first site of lymph drainage from the
location of the primary tumor. It is suspected and assumed that if a
malignancy is going to disseminate via the lymphatic system,
metastases will first be evident in the sentinel lymph node. In this
manner, this lymph node is said to stand guard or sentinel over the
metastatic state of the tumor. For many cancers, the sentinel lymph
node is biopsied as part of the staging process and presence of macro-
or micrometastases in the sentinel lymph node is a negative prog-
nostic factor.

698 © 2015 by American Society of Clinical Oncology JOURNAL OF CLINICAL ONCOLOGY

 Risk Reduction Practices for Lymphedema

AUTHORS’ DISCLOSURES OF POTENTIAL CONFLICTS OF INTEREST

Impact of Ipsilateral Blood Draws, Injections, Blood Pressure Measurements, and Air Travel on the Risk of Lymphedema for Patients Treated for
Breast Cancer
The following represents disclosure information provided by authors of this manuscript. All relationships are considered compensated. Relationships are
self-held unless noted. I = Immediate Family Member, Inst = My Institution. Relationships may not relate to the subject matter of this manuscript. For more
information about ASCO’s conflict of interest policy, please refer to www.asco.org/rwc or jco.ascopubs.org/site/ifc.
Chantal M. Ferguson Cheryl Brunelle
No relationship to disclose No relationship to disclose
Meyha N. Swaroop Jean A. O’Toole
No relationship to disclose No relationship to disclose
Nora Horick Laura Salama
No relationship to disclose No relationship to disclose
Melissa N. Skolny Michelle C. Specht
No relationship to disclose No relationship to disclose
Cynthia L. Miller Alphonse G. Taghian
No relationship to disclose Research Funding: ImpediMed L-Dex
Lauren S. Jammallo
No relationship to disclose

www.jco.org © 2015 by American Society of Clinical Oncology

 Research

JAMA | Original Investigation

Effect of Acupuncture vs Sham Acupuncture or Waitlist Control

on Joint Pain Related to Aromatase Inhibitors Among Women
With Early-Stage Breast Cancer
A Randomized Clinical Trial
Dawn L. Hershman, MD, MS; Joseph M. Unger, PhD, MS; Heather Greenlee, ND, PhD; Jillian L. Capodice, MS, LAc;
Danika L. Lew, MA; Amy K. Darke, MS; Alice T. Kengla, MD; Marianne K. Melnik, MD; Carla W. Jorgensen, MD;
William H. Kreisle, MD; Lori M. Minasian, MD; Michael J. Fisch, MD; N. Lynn Henry, MD; Katherine D. Crew, MD, MS

Supplemental content
IMPORTANCE Musculoskeletal symptoms are the most common adverse effects of aromatase Related article at
inhibitors and often result in therapy discontinuation. Small studies suggest that acupuncture jamaoncology.com
may decrease aromatase inhibitor–related joint symptoms.

OBJECTIVE To determine the effect of acupuncture in reducing aromatase inhibitor–related

joint pain.

DESIGN, SETTING, AND PATIENTS Randomized clinical trial conducted at 11 academic centers
and clinical sites in the United States from March 2012 to February 2017 (final date of
follow-up, September 5, 2017). Eligible patients were postmenopausal women with
early-stage breast cancer who were taking an aromatase inhibitor and scored at least 3 on the
Brief Pain Inventory Worst Pain (BPI-WP) item (score range, 0-10; higher scores indicate
greater pain).

INTERVENTIONS Patients were randomized 2:1:1 to the true acupuncture (n = 110), sham
acupuncture (n = 59), or waitlist control (n = 57) group. True acupuncture and sham
acupuncture protocols consisted of 12 acupuncture sessions over 6 weeks (2 sessions per
week), followed by 1 session per week for 6 weeks. The waitlist control group did not receive
any intervention. All participants were offered 10 acupuncture sessions to be used between
weeks 24 and 52.

MAIN OUTCOMES AND MEASURES The primary end point was the 6-week BPI-WP score. Mean
6-week BPI-WP scores were compared by study group using linear regression, adjusted for
baseline pain and stratification factors (clinically meaningful difference specified as 2 points).

RESULTS Among 226 randomized patients (mean [SD] age, 60.7 [8.6] years; 88% white;
mean [SD] baseline BPI-WP score, 6.6 [1.5]), 206 (91.1%) completed the trial. From baseline
to 6 weeks, the mean observed BPI-WP score decreased by 2.05 points (reduced pain)
in the true acupuncture group, by 1.07 points in the sham acupuncture group, and by 0.99
points in the waitlist control group. The adjusted difference for true acupuncture vs sham
acupuncture was 0.92 points (95% CI, 0.20-1.65; P = .01) and for true acupuncture vs waitlist
control was 0.96 points (95% CI, 0.24-1.67; P = .01). Patients in the true acupuncture group
experienced more grade 1 bruising compared with patients in the sham acupuncture group
(47% vs 25%; P = .01).

CONCLUSIONS AND RELEVANCE Among postmenopausal women with early-stage breast

cancer and aromatase inhibitor–related arthralgias, true acupuncture compared with sham
acupuncture or with waitlist control resulted in a statistically significant reduction in joint pain Author Affiliations: Author
affiliations are listed at the end of this
at 6 weeks, although the observed improvement was of uncertain clinical importance.
article.
Corresponding Author: Dawn L.
TRIAL REGISTRATION ClinicalTrials.gov Identifier: NCT01535066 Hershman, MD, MS, Columbia
University, 161 Ft Washington Ave,
10-1068, New York, NY 10032
JAMA. 2018;320(2):167-176. doi:10.1001/jama.2018.8907 (dlh23@columbia.edu).

(Reprinted) 167
© 2018 American Medical Association. All rights reserved.

Downloaded From: by a Tufts Univ. Hirsh Health Sciences Library User on 07/10/2018
 Research Original Investigation Acupuncture vs Control for Joint Pain Related to Aromatase Inhibitor Treatment for Breast Cancer
A
romatase inhibitors have proven efficacy for the treat-
ment of hormone-sensitive breast cancer.1 However,
many patients experience adverse effects, including
aromatase inhibitor–related arthralgias (pain and stiffness),
which occur in approximately 50% of patients2-4 and contrib-
ute to nonadherence with therapy.5,6 Data from clinical trials
suggest that nonadherence to endocrine therapy is associ-
ated with reduced disease-free survival.7
Acupuncture is a traditional Chinese therapy that in-
volves insertion of fine, single-use, sterile needles in acu-
points according to a system of channels and meridians that
were developed by early practitioners of Traditional Chinese
Medicine.8 The needles are stimulated by manual manipula-
tion, electrical stimulation, or heat.8 Acupuncture is a popu-
lar nonpharmacological modality used for treating a variety of
conditions and may result in reduced opioid use.9 Several stud-
ies have evaluated acupuncture as a treatment for aromatase
inhibitor–related arthralgias.10-13 However, these studies pro-
vided uncertain interpretations because of small sample size,
ineffective blinding, and implementation at single centers. Con-
cerns were also raised that the sham control intervention may
have had positive physiological effects and therefore could be
an active control.
This multicenter randomized blinded sham- and waitlist-
controlled clinical trial was conducted to evaluate the effect
of acupuncture on joint pain related to aromatase inhibitors
among women with early-stage breast cancer.
Methods
Study Oversight
The study was conducted at 11 academic and community sites
and was approved by appropriate individual institutional re-
view boards. Sites were required to have 2 trained acupunc-
turists for the duration of the trial. Study participants were in-
formed of the investigational nature of the study and provided
written informed consent. Race and ethnicity, which were self-
reported by patients and recorded by site investigators, were
reported to aid in the determination of the generalizability of
the findings.14 The trial protocol is reported in Supplement 1.
Patient Characteristics
Eligible study participants were women with histologically con-
firmed (stages I-III) primary invasive estrogen receptor–
positive carcinoma of the breast, progesterone receptor–
positive carcinoma of the breast, or both. Patients were required
to be postmenopausal or premenopausal with use of a
gonadotropin-releasing hormone agonist, to be taking a third-
generation aromatase inhibitor for more than 30 days prior to
registration with plans to continue for at least 1 additional year,
to have recovered from the effects of surgery or chemo-
therapy, and to have a Zubrod performance status of 0 to 1.15
Inclusion criteria included a score of greater than 3 (range, 0-10;
higher scores indicate greater pain) on the worst pain item of
the Brief Pain Inventory-Short Form (BPI-SF) that (by patient
report) started or increased since starting aromatase inhibi-
tor therapy.16
168 JAMA July 10, 2018 Volume 320, Number 2 (Reprinted)
© 2018 American Medical Association. All rights reserved.
Downloaded From: by a Tufts Univ. Hirsh Health Sciences Library User on 07/10/2018
Key Points
Question Does acupuncture reduce joint pain related to
aromatase inhibitors among postmenopausal women with
early-stage breast cancer?
Findings In this multicenter randomized clinical trial of 226
women with early-stage breast cancer, patients in the acupuncture
group, compared with those in the sham acupuncture group or the
waitlist control group, had significant reductions in changes in joint
pain scores from baseline to 6 weeks (between-group difference
vs sham acupuncture, 0.92 points and 0.96 points vs the waitlist
control group [0- to 10-point scale]).
Meaning Acupuncture was associated with statistically significant
reductions in aromatase inhibitor–related joint pain at 6 weeks,
although the magnitude of the improvement was of uncertain
clinical importance.
Exclusion criteria included prior acupuncture treatment
for joint symptoms at any time, although patients were al-
lowed to have received 2 or fewer acupuncture treatments
within 12 months prior to registration for other reasons. Study
participants must not have received opioid analgesics, topi-
cal analgesics, oral corticosteroids, intramuscular corticoste-
roids, or intra-articular steroids, or any other medical therapy,
alternative therapy, or physical therapy for the treatment of
joint pain or joint stiffness within 28 days prior to registra-
tion. Patients must not have had a history of bone fracture or
surgery of the affected knees, hands, or both within 6 months
prior to enrollment. Patients with a severe bleeding disorder
or a latex allergy were excluded.
Study Intervention
Study participants were randomized 2:1:1 to the true acupunc-
ture group, the sham acupuncture group, or the waitlist con-
trol group, with randomization dynamically balanced by study
site to account for potential differences in acupuncture
administration.12 The acupuncture study interventions were
developed by a consensus of acupuncture experts based on pre-
vious studies of acupuncture for aromatase inhibitor–related
arthralgias with adherence to the Standards for Reporting of
Controlled Trials in Acupuncture (STRICTA) recommendations.13
The details of the acupuncture point protocol (ie, body site of
needle placement) and extensive training and standardization
methods (in-person training, study manuals, monthly phone
calls, and remote quality assurance monitoring) have been pre-
viously described.17
Briefly, both true acupuncture and sham acupuncture con-
sisted of twelve 30- to 45- minute sessions administered over
a period of 6 weeks (2 per week) followed by 1 session per week
for 6 weeks. For true acupuncture, stainless steel, single-use,
sterile and disposable needles were used and inserted at tra-
ditional depths and angles. The joint-specific protocol was tai-
lored to as many as 3 of the patient’s most painful joint areas.18
Needles were restimulated manually once during each ses-
sion. The sham acupuncture consisted of a core standardized
prescription of minimally invasive, shallow needle insertion
using thin and short needles at nonacupuncture points. The
jama.com
 Acupuncture vs Control for Joint Pain Related to Aromatase Inhibitor Treatment for Breast Cancer
sham acupuncture protocol also included joint-specific treat-
ments and an auricular sham based on the application of ad-
hesives to nonacupuncture points on the ear. The waitlist con-
trol group received no acupuncture treatments and received
no other intervention for 24 weeks after randomization. At 24
weeks, all patients received vouchers for 10 true acupuncture
bonus sessions to be used prior to the 52-week visit.
Patient-Reported Outcome Measures
Primary End Point
There is no consensus on the best measure for aromatase
inhibitor–induced arthralgias. The BPI-SF was used in several
prior interventional trials of aromatase inhibitor arthralgias.10,19,20
This 14-item questionnaire asks individuals to rate joint pain over
the prior week and the degree to which the pain interferes with
activities using a 0- to 10-point scale (0 [no pain] to 10 [pain as
bad as you can imagine]).16 For this study, the primary end point
was the BPI Worst Pain Item (BPI-WP) score at 6 weeks of treat-
ment. A reduction of 2 points on the BPI-WP has been identified
as a clinically meaningful change for a patient.21
Prespecified Secondary End Points
Prespecified secondary end points were measured and evalu-
ated using several instruments. The BPI-SF was used to mea-
sure worst pain, worst stiffness, pain severity, and pain-
related interference scores (range from no symptoms to
worst, 0-10); administered at 6, 12, 16, 20, 24, and 52 weeks
to assess the duration of effect following completion of the
intervention.
The Western Ontario and McMaster Universities Osteoar-
thritis Index (WOMAC) version 3.1, a validated measure for
assessing osteoarthritis of the knees or hips, evaluated
patients using 24 questions related to 3 subscales (pain
[0-500], stiffness [0-200], and physical function [0-1700])22;
administered at 6, 12, 24, and 52 weeks. The Modified Score
for the Assessment and Quantification of Chronic Rheuma-
toid Affections of the Hands (M-SACRAH) was used to assess
hand pain, stiffness, and functional status (range, 0-100)23;
administered at 6, 12, 24, and 52 weeks. For both instru-
ments, higher scores indicate greater symptom severity.
The Functional Assessment of Cancer Therapy-Endocrine
Symptoms (FACT-ES) measured physical and functional well-
being and endocrine symptoms (score range, 0-128); admin-
istered at 6, 12, 24, and 52 weeks. The PROMIS Pain Impact-
Short Form (PROMIS PI-SF; score range, 6-30) 24,25 was
administered at 6, 12, 24, and 52 weeks. Both insturments have
5 response levels (low scores [not at all] to high scores [very
much]), and higher scores reflect better symptoms. Addi-
tional prespecified secondary end points were analgesic and
opioid use and safety and tolerability of acupuncture. Other
outcomes included an assessment of masking among pa-
tients enrolled in the true acupuncture and sham acupunc-
ture study groups.
Statistical Analyses
The primary hypothesis was that true acupuncture would
decrease joint pain associated with the use of aromatase
inhibitors (according to the BPI-SF worst pain item [BPI-WP])
jama.com
© 2018 American Medical Association. All rights reserved.
Downloaded From: by a Tufts Univ. Hirsh Health Sciences Library User on 07/10/2018
Original Investigation Research
compared with both sham acupuncture or waitlist control at
6 weeks. For a 2-point difference between groups and an
assumed 3.0-point standard deviation at 6 weeks, 208 eli-
gible patients were required for 82% power (using 2-sided
tests; α = .025) for true acupuncture vs sham acupuncture
and true acupuncture vs waitlist control. The design further
specified an estimated 5% nonadherence and 10% dropout
rate at the primary end point evaluation time of 6 weeks.
In addition, the design incorporated a 10% contamination
rate. The primary outcome was analyzed using the intention-
to-treat principle (ie, as randomized), using a complete-case
approach given limited missing data (<10%), without using
imputation.
The primary outcome at week 6 and the secondary out-
comes at weeks 6 and 12 were examined using multivariable
linear regression and adjusting for the baseline score, indica-
tor variables for study sites, and 2 indicator variables for in-
tervention group. The baseline score was not included as a com-
ponent of the outcome because it was obtained prior to
intervention administration.
Longitudinal measures through week 24 (12 weeks fol-
lowing completion of the intervention) were analyzed as sec-
ondary outcomes using linear mixed models with a random
effect for patient and an a priori unstructured covariance
matrix.26 Fixed-effect factors associated with the outcomes
of interest included intervention assignment using separate
indicator variables (with true acupuncture as the reference
category), continuous time (both as a linear variable and qua-
dratic variable), an intervention × time interaction (both lin-
ear and quadratic time), the baseline score, and indicator
variables for study sites.
In a post hoc analysis, we reported the proportion of pa-
tients reporting an individual-level clinically meaningful re-
duction (2 points from baseline on a scale of 0-10 or [calcu-
lated separately] a relative reduction of 30% ), based on current
guidelines for clinical trials evaluating interventions to re-
duce pain.27,28 Risk differences (RDs) and relative risks (RRs)
were estimated using Poisson regression with robust stan-
dard errors.29
Statistical analyses were conducted using SAS version 9.44
(SAS Institute), including SAS proc mixed for linear mixed
models.30 For the primary end point, Bonferroni adjustment
was used for multiple comparisons (setting α = .025) for each
independent comparison of true acupuncture vs sham acu-
puncture and true acupuncture vs waitlist control. No adjust-
ments for multiple comparisons were made for all secondary
or post hoc analyses, which are considered exploratory.
Results
Patient Characteristics
From March 2012 to February 2017, a total of 226 patients were
randomly assigned to the true acupuncture (n = 110), sham acu-
puncture (n = 59), or waitlist control (n = 57) group (Figure 1).
Patient baseline characteristics are shown in Table 1. Median
age was lower in the sham acupuncture group (57.0 years) than
in the waitlist control (60.6 years) or true acupuncture (60.8
(Reprinted) JAMA July 10, 2018 Volume 320, Number 2 169
 Research Original Investigation Acupuncture vs Control for Joint Pain Related to Aromatase Inhibitor Treatment for Breast Cancer

Figure 1. Flow of Randomized Patients for the Effect of True Acupuncture vs Sham Acupuncture and True Acupuncture vs Waitlist Control on Joint
Pain Using the 6-Week Brief Pain Inventory-Short Form Worst Pain Score

226 Patients randomizeda

110 Randomized to receive true acupuncture
108 Received true acupuncture
as randomized
2 Did not receive true acupuncture
as randomized
59 Randomized to receive sham acupuncture
57 Received sham acupuncture
as randomized
2 Did not receive sham acupuncture
as randomized
57 Randomized to the waitlist control group
56 Were in the waitlist control group
as randomized
1 Was not in the waitlist control group
as randomized

10 Not evaluable 5 Not evaluable (did not have a 6 Not evaluable

9 Did not have a 6-wk BPI measure baseline BPI measure) 5 Did not have a 6-wk BPI measure
1 Did not have a baseline BPI measure 1 Did not have a baseline BPI measure

100 Included in the 6-wk primary analysis
2 Received no treatmentb
101 Included in the 12-wk analysisc
54 Included in the 6-wk primary analysis
2 Received no treatmentb
1 Received true acupuncture for 2 sessionsb
54 Included in the 12-wk analysis
51 Included in the 6-wk primary analysis
1 Did not wait for treatmentb
51 Included in the 12-wk analysis

b
BPI indicates Brief Pain Inventory. Patients’ status as major deviations began immediately in the course of
a
Data were not collected to report the number of patients screened for follow-up and was applied to both the 6-week and 12-week analyses.
c
eligibility prior to randomization. One patient who was not available for 6-week evaluation was evaluated
at 12 weeks.

years) groups. Fewer Hispanic patients were randomized to the
waitlist control group, and more Asian patients were random-
ized to the true acupuncture group. The overall median time
for receiving aromatase inhibitor therapy prior to enrollment
was 1.1 years. The patient characteristics were well balanced
by study site and clinical factors.
Major protocol deviations were reported for 6 patients: 1
patient randomized to the sham acupuncture group received
true acupuncture for 2 sessions, 4 patients randomized to
acupuncture groups (true, 2; sham, 2) received no treatment,
and 1 patient randomized to the waitlist control group was in
too much pain and declined to participate (but did complete
the patient-reported questionnaires). Twenty-one patients
were not evaluable (Figure 1). No patients reported receiving
acupuncture outside of the trial during the 12-week interven-
tion period.
Worst Pain Scores
In total, 91% (100/110) of patients in the true acupuncture
group, 92% (54/59) in the sham acupuncture group, and 89%
(51/57) in the waitlist control group had both baseline and
6-week BPI-WP scores for analysis (Figure 1). Reporting rates
at weeks 6 and 12 were similar. There was no statistically sig-
nificant difference by study group in the amount of available
data at follow-up (P = .93). Mean baseline BPI-WP score was
6.84 in the true acupuncture group, 6.55 in the sham acupunc-
ture group, and 6.48 in the waitlist control group.
Primary Outcome
Compared with baseline, the mean observed BPI-WP score was
2.05 points lower (reduced pain) at 6 weeks in the true acu-
puncture group, 1.07 points lower in the sham acupuncture
group, and 0.99 points lower for the waitlist control group, with
differences in adjusted 6-week mean BPI-WP scores between
true acupuncture vs sham acupuncture of 0.92 points (95% CI,
0.20-1.65; P = .01) and between true acupuncture vs waitlist
control of 0.96 points (95% CI, 0.24-1.67; P = .01) (Table 2).
Thus, the study results rejected the null hypothesis that true
acupuncture generated the same outcomes as sham acupunc-
ture and waitlist control, although the magnitude of the ef-
fect did not achieve the prespecified difference of 2 points.
Secondary Outcomes
Patients randomized to the true acupuncture group had sta-
tistically significant improved symptom scores compared with
those randomized to the sham acupuncture and waitlist con-
trol groups at 6 weeks according to BPI average pain, pain se-
verity, and worst stiffness (Table 2). Patients randomized to
true acupuncture had improved symptoms at 6 weeks com-
pared with those in the sham acupuncture group, but not
compared with the waitlist control group according to pain in-
terference. At 12 weeks, patients randomized to the true acu-
puncture group compared with the sham acupuncture group
had statistically significant improvements in average pain, but
no significant improvement in worst pain, pain interference,
pain severity, or worst stiffness (Table 2). Compared with the
waitlist control group, patients randomized to the true acu-
puncture group had improved pain by all BPI measures at
week 12 (P≤.003).
Patients randomized to the true acupuncture group had
improved symptoms compared with those in the sham acu-
puncture group at 6 weeks according to the M-SACRAH,
WOMAC, and PROMIS PI-SF measures (eTable 3 in Supple-
ment 2). Patients randomized to the true acupuncture
group had improved symptoms compared with those in the
waitlist control group at 6 weeks according to the M-SACRAH
and WOMAC measures. At week 12, patients randomized
to the true acupuncture group had improved symptoms

170 JAMA July 10, 2018 Volume 320, Number 2 (Reprinted) jama.com

© 2018 American Medical Association. All rights reserved.

Downloaded From: by a Tufts Univ. Hirsh Health Sciences Library User on 07/10/2018
 Acupuncture vs Control for Joint Pain Related to Aromatase Inhibitor Treatment for Breast Cancer Original Investigation Research

Table 1. Baseline Characteristics for Patients in the True Acupuncture, Sham Acupuncture,
or Waitlist Control Group

No. (%)a
True Acupuncture Sham Acupuncture Waitlist Control
(n = 110) (n = 59) (n = 57)
Age, median (range), yb 60.8 (34.1-80.6) 57.0 (40.6-77.5) 60.6 (27.1-76.0)
Hispanic
Yes 11 (10) 7 (12) 3 (5)
Unknown 0 1 0
Race
White 88 (83) 54 (93) 51 (91)
Black 6 (6) 2 (3) 2 (4)
Asian 11 (10) 2 (3) 2 (4)
Pacific Islander 0 0 1 (2)
American Indian 1 (1) 0 0
Unknown 4 1 1
Study sitec
Columbia University 14 (13) 8 (14) 7 (12)
Minority Underserved NCORP
Kaiser Permanente Medical Center 27 (25) 14 (24) 15 (26)
Spectrum Health Medical Group 34 (31) 17 (29) 17 (30)
Other (8) 35 (32) 20 (34) 18 (32)
Breast cancer stage
I 41 (39) 28 (47) 28 (49)
II 53 (50) 23 (39) 23 (40)
III 12 (11) 8 (14) 6 (11)
Prior chemotherapy 56 (51) 31 (53) 24 (42)
Prior tamoxifen 18 (16) 15 (25) 10 (18)
Current or prior aromatase inhibitor therapyd
Anastrozole 80 (73) 44 (75) 40 (70)
Currently using 64 (58) 37 (63) 37 (65)
Not currently using 15 (14) 7 (12) 3 (5)
Not reported 1 (1) 0 0
Letrozole 36 (33) 17 (29) 17 (30)
Currently using 29 (26) 13 (22) 11 (19)
Not currently using 7 (6) 4 (7) 6 (11) Abbreviations: NCORP, National
Cancer Institute Community
Exemestane 21 (19) 10 (17) 10 (18)
Oncology Research Program;
Currently using 15 (14) 9 (15) 8 (14) NSAID, nonsteroidal
Not currently using 6 (5) 1 (2) 2 (4) anti-inflammatory drug.
a
Time receiving aromatase inhibitors, 1.0 (0.1-8.0) 1.1 (0.1-9.0) 1.1 (0.1-3.1) Values are reported as No. (%)
median (range), y unless otherwise indicated.
b
Prior acupuncture 19 (17) 13 (22) 12 (21) Younger women were either
Pain medications (any use) d surgically or chemically
menopausal.
Acetaminophen/paracetamol 26 (24) 12 (20) 8 (14) c
See eTable 5 in Supplement 2 for
Ibuprofen 31 (28) 20 (34) 22 (39) specific site names and locations.
Other NSAID 15 (14) 8 (14) 8 (14) d
Indicates the number of patients
Other pain medication 6 (5) 3 (5) 4 (7) who answered yes. More than 1
subcategory type is allowed.
≥1 medication 66 (60) 36 (61) 33 (58)
Therefore, items may not sum.

compared with those in the sham acupuncutre group ac-
cording to the WOMAC and PROMIS PI-SF measures, and
they had improved symptoms compared with those in the
waitlist control group according to the M-SACRAH, WOMAC,
FACT-ES, and PROMIS PI-SF measures.
In linear mixed-model analysis, there was no significant
interaction between intervention assignment and time, and
there was no significant quadratic relationship of BPI-WP
over time (eTable 4 in Supplement 2). Through 24 weeks,
adjusted mean BPI-WP scores were 0.59 points lower (95%
CI, 0.34-1.14; P = .04) in the true acupuncture group com-
pared with the sham acupuncture group and were 1.23 points
lower (95% CI, 0.66-1.80; P<.001) in the true acupuncture
group compared with the waitlist control group (Figure 2).
Similar findings for trends are shown for the secondary mea-
sures (eFigure 1 in Supplement 2).

jama.com (Reprinted) JAMA July 10, 2018 Volume 320, Number 2 171

© 2018 American Medical Association. All rights reserved.

Downloaded From: by a Tufts Univ. Hirsh Health Sciences Library User on 07/10/2018
 172
Table 2. Observed and Fitted Group Mean Results for Brief Pain Inventory-Short Form Scores at Weeks 6 and 12 in Each Group

Group Mean Differencea Proportion With >30% Improvement

Mean (95% CI) Fitted Difference (95% CI) Risk Difference, % (95% CI)
Analysis No. of True vs Sham True vs Sham
c
by Group Patients Baseline Follow-up True vs Waitlist P Value % True vs Waitlist P Value Relative Risk (95% CI)b P Value
Worst Paind
Week 6e
True 100 6.84 (6.55 to 7.13) 4.79 (4.36 to 5.22) 49.0 1 [Reference]
Research Original Investigation

Sham 54 6.55 (6.14 to 6.98) 5.48 (4.78 to 6.18) 0.92 (0.20 to 1.65) .01 24.1 24.9 (9.9 to 40.0) .001 1.95 (1.16 to 3.28) .01
Waitlist 51 6.48 (6.07 to 6.89) 5.49 (4.81 to 6.17) 0.96 (0.24 to 1.67) .01 23.5 25.5 (10.3 to 40.7) .001 2.04 (1.19 to 3.48) .009
Week 12
True 101 6.84 (6.55 to 7.13) 4.53 (4.05 to 5.01) 51.5 1 [Reference]
Sham 54 6.55 (6.14 to 6.98) 5.04 (4.33 to 5.74) 0.73 (−0.07 to 1.53) .08 46.3 5.2 (−11.3 to 21.7) .54 1.09 (0.77 to 1.54) .62
Waitlist 51 6.48 (6.07 to 6.89) 6.29 (5.73 to 6.86) 1.95 (1.19 to 2.70) <.001 15.7 35.8 (21.9 to 49.8) <.001 3.24 (1.66 to 6.34) <.001
Average Pain

JAMA July 10, 2018 Volume 320, Number 2 (Reprinted)

Week 6
True 100 4.95 (4.64 to 5.27) 3.50 (3.21 to 3.88) 43.0 1 [Reference]
Sham 54 4.69 (4.22 to 5.17) 3.93 (3.40 to 4.45) 0.60 (0.03 to 1.17) .04 25.9 17.1 (1.9 to 32.3) .03 1.60 (0.97 to 2.63) .07
Waitlist 51 4.95 (4.52 to 5.37) 4.14 (3.65 to 4.63) 0.71 (0.15 to 1.28) .01 17.7 25.4 (11.1 to 39.6) .0005 2.44 (1.28 to 4.62) .006
Week 12
True 101 4.95 (4.64 to 5.27) 3.00 (2.62 to 3.38) 60.4 1 [Reference]
Sham 53 4.69 (4.22 to 5.17) 3.62 (3.06 to 4.18) 0.79 (0.16 to 1.42) .02 45.3 15.1 (−1.3 to 31.6) .07 1.31 (0.94 to 1.83) .11

Downloaded From: by a Tufts Univ. Hirsh Health Sciences Library User on 07/10/2018
Waitlist 51 4.95 (4.52 to 5.37) 4.33 (3.80 to 4.87) 1.38 (0.76 to 2.00) <.001 29.4 31.0 (15.3 to 46.7) .0001 2.04 (1.30 to 3.21) .002
Pain Interference
Week 6
True 100 4.24 (3.86 to 4.62) 2.55 (2.11 to 2.98) 59.0 1 [Reference]
Sham 54 3.85 (3.34 to 4.37) 3.03 (2.44 to 3.62) 0.73 (0.11 to 1.35) .02 44.4 14.6 (−1.8 to 30.9) .08 1.36 (0.98 to 1.89) .06
Waitlist 51 4.01 (3.54 to 4.48) 3.07 (2.47 to 3.67) 0.63 (−0.03 to 1.29) .06 35.3 23.7 (7.4 to 40.0) .004 1.68 (1.12 to 2.52) .01
Week 12
True 101 4.24 (3.86 to 4.62) 2.44 (2.00 to 2.88) 68.3 1 [Reference]
Sham 54 3.85 (3.34 to 4.37) 2.40 (1.84 to 2.96) 0.19 (−0.40 to 0.77) .53 64.8 3.5 (−12.1 to 19.1) .66 1.06 (0.84 to 1.33) .64

© 2018 American Medical Association. All rights reserved.

Waitlist 51 4.01 (3.54 to 4.48) 3.31 (2.69 to 3.94) 0.99 (0.36 to 1.63) .003 37.3 31.1 (15.0 to 47.1) <.001 1.85 (1.26 to 2.70) .002
Pain Severity
Week 6
True 100 4.71 (4.41 to 5.01) 3.21 (2.85 to 3.57) 50.0 1 [Reference]
Sham 54 4.75 (5.32 to 5.19) 3.75 (3.21 to 5.28) 0.56 (0.00 to 1.11) .05 33.3 16.7 (0.7 to 32.6) .04 1.52 (1.00 to 2.31) .05
Waitlist 51 4.64 (4.25 to 5.03) 3.82 (3.26 to 4.37) 0.71 (0.15 to 1.28) .01 33.3 16.7 (0.4 to 32.9) .04 1.51 (0.98 to 2.34) .06
Week 12
True 101 4.71 (4.41 to 5.01) 2.89 (2.52 to 3.26) 57.4 1 [Reference]
Sham 54 4.75 (5.32 to 5.19) 3.41 (2.86 to 3.95) 0.53 (−0.06 to 1.11) .08 44.4 13.0 (−3.4 to 29.4) .12 1.29 (0.92 to 1.81) .15
Waitlist 51 4.64 (4.25 to 5.03) 4.25 (3.72 to 4.78) 1.43 (0.84 to 2.02) <.001 21.6 35.9 (21.0 to 50.7) <.001 2.64 (1.53 to 4.55) <.001

(continued)

jama.com
Acupuncture vs Control for Joint Pain Related to Aromatase Inhibitor Treatment for Breast Cancer
 Acupuncture vs Control for Joint Pain Related to Aromatase Inhibitor Treatment for Breast Cancer Original Investigation Research

Patients randomized to the true acupuncture group were

weeks (70% vs 43%, P = .009). However, there was no evidence that the intervention effect with respect to the
acupuncture group to believe they were receiving true acupuncture at 6 weeks (68% vs 36%, P<.001) and at 12
P Value
more likely to believe they were receiving true acupuncture

.002

primary outcome of worst pain differed between those believing vs not believing they were receiving true
.04
.02

.62
than those receiving sham acupuncture (68% vs 36%; P<.001).
However, the intervention effect for BPI-WP did not differ be-
tween those believing vs not believing they were receiving true
acupuncture when interaction tests were used (P for interac-

acupuncture at either 6 weeks (P for interaction = .16) or 12 weeks (P for interaction = .75).
Relative Risk (95% CI)b
tion = .16).

1.56 (1.03 to 2.37)

1.76 (1.11 to 2.78)

1.09 (0.78 to 1.53)

2.54 (1.42 to 4.54)
1 [Reference] Post Hoc Analysis

1 [Reference]
The proportion of patients experiencing at least a 2-point re-
duction in BPI-WP at 6 weeks was 58.0% (n = 58) for the true
acupuncture group, 33.3% (n = 18) for the sham acupuncture
group (RD, 24.7% [95% CI, 8.8%-40.5%]; RR, 1.64 [95% CI, 1.10-
2.44]; P = .02), and 31.4% (n = 16) for the waitlist control group
P Value

(difference for true acupuncture vs waitlist control, 26.6% [95%

.005

<.001
.02

.66

CI, [10.6%-42.6%]; RR, 1.75 [95% CI, 1.13-2.69]; P = .01; eFig-

Risk Difference, % (95% CI)

ure 2 and eTable 2 in Supplement 2) Similar results by treat-

Proportion With >30% Improvement

ment group were evident for the proportion of patients expe-

3.7 (−12.8 to 20.2)
30.4 (15.7 to 45.1)

Category indicates the primary end point.

18.7 (2.7 to 34.6)
22.6 (6.7 to 38.5)

riencing an improvement of greater than 30% (Figure 2; eTable

True vs Waitlist

2 and eTable 3 in Supplement 2).

True vs Sham
Table 2. Observed and Fitted Group Mean Results for Brief Pain Inventory-Short Form Scores at Weeks 6 and 12 in Each Group (continued)

Adverse Events
Bruising was the most common adverse event reported for
those receiving true acupuncture or sham acupuncture. More
52.0
33.3
29.4

50.0
46.3
19.6

patients in the true acupuncture group experienced grade 1

%

bruising (47%) than in the sham acupuncture group (25%;

P = .01) (eTable 1 in Supplement 2). There was 1 episode of grade
P Value

<.001

2 presyncope in the true acupuncture group and 1 episode in

.02
.01

.08

the sham acupuncture group. No other differences by study

e

group for selected adverse events were observed, and no grade

Fitted Difference (95% CI)

regression (adjusting for the baseline score and the stratification factor). Positive values favor true acupuncture.
Data for examination of relative risks were calculated using Poisson regression (adjusting for the baseline score

3 or higher adverse events were reported.

0.72 (−0.08 to 1.53)

Patients randomized to the true acupuncture group were more likely than patients randomized to the sham
1.00 (0.19 to 1.81)
1.09 (0.26 to 1.92)

1.80 (1.03 to 2.57)

Data for examination of group mean differences by study group were calculated using multivariable linear
True vs Waitlist
True vs Sham

Discussion
In this multicenter, sham- and waitlist-controlled clinical trial
of patients with early-stage breast cancer and aromatase
inhibitor–related joint pain, there were statistically signifi-
4.46 (3.94 to 4.98)
5.50 (4.86 to 6.14)
5.53 (4.91 to 6.14)

4.35 (3.82 to 4.88)

5.07 (4.47 to 5.68)
6.12 (5.59 to 6.64)

cant but modest improvements in pain scores with true acu-

puncture administered twice a week for 6 weeks compared
with both sham acupuncture and waitlist control. However,
Follow-up

and the stratification factor). Positive values favor true acupuncture.

the magnitude of the effect did not achieve the prespecified

between-group difference of 2 points in the primary end point
Data were calculated among patients with follow-up scores.

of BPI-WP, which had been considered to represent a clini-

Group Mean Differencea

cally meaningful difference by the design.

6.65 (6.26 to 7.04)
6.59 (6.03 to 7.16)
6.68 (6.26 to 7.10)

6.65 (6.26 to 7.04)

6.59 (6.03 to 7.16)
6.68 (6.26 to 7.10)

The adjusted mean between-group difference at 6 weeks

was 0.92 points for true acupuncture vs sham acupuncture
Mean (95% CI)

and 0.96 points for true acupuncture vs waitlist control.

Baselinec

Although there is some uncertainty about the clinical mean-

ing of this between-group difference,27,28 the observed differ-
ence in this study is consistent with other randomized stud-
ies of pain control interventions, which have reported
Patients
No. of

between-group mean differences ranging from 0.7 to

100
54
51

100
54
51

1.0 points.31,32 The pain-reporting guidelines describe the

Worst Stiffness

clinical meaning of an individual’s response, with studies

Waitlist

Waitlist

suggesting that a reduction of 2 points on an 11-point scale (or

by Group

Week 12
Analysis

Sham

Sham
Week 6
True

True

30%) represents a clinically important reduction (ie, much

improved).21,28 A post hoc hypothesis-generating, responder
b

d
a

jama.com (Reprinted) JAMA July 10, 2018 Volume 320, Number 2 173

© 2018 American Medical Association. All rights reserved.

Downloaded From: by a Tufts Univ. Hirsh Health Sciences Library User on 07/10/2018
 Research Original Investigation Acupuncture vs Control for Joint Pain Related to Aromatase Inhibitor Treatment for Breast Cancer

Figure 2. Linear Mixed-Model Results Through 24 Weeks for Selected Brief Pain Inventory-Short Form End Points

A Worst pain B Pain interference

8 8

6 Waitlist control 6
Mean BPI-SF Score

Mean BPI-SF Score

Sham acupuncture
True acupuncture
4 4
Waitlist control
Sham acupuncture
True acupuncture
2 2

0 0
Baseline 6 12 16 20 24 Baseline 6 12 16 20 24
Assessment Week Assessment Week
No. of patients No. of patients
Waitlist control 56 51 51 48 45 50 Waitlist control 56 51 51 48 45 50
Sham acupuncture 59 54 54 53 52 54 Sham acupuncture 59 54 54 53 52 54
True acupuncture 109 100 101 94 97 97 True acupuncture 109 100 101 94 97 97

C Pain severity D Worst stiffness

8 8

6 6 Waitlist control
Mean BPI-SF Score

Mean BPI-SF Score

Sham acupuncture
True acupuncture
4 Waitlist control 4
Sham acupuncture
True acupuncture

2 2

0 0
Baseline 6 12 16 20 24 Baseline 6 12 16 20 24
Assessment Week Assessment Week
No. of patients No. of patients
Waitlist control 56 51 51 48 45 50 Waitlist control 56 51 51 48 45 50
Sham acupuncture 59 54 54 53 52 54 Sham acupuncture 59 54 54 53 52 54
True acupuncture 109 100 101 94 97 97 True acupuncture 109 100 101 94 97 97

Horizontal bars at baseline indicate the observed baseline means. Vertical bars
indicate the interquartile range (IQR) for each study group at each assessment
time and are offset by a small margin to show the IQR for each group. Fitted
regression lines for each group are also shown, with P values for the effect of
intervention on Brief Pain Inventory-Short Form outcomes derived from
multivariable linear mixed-model analyses with a random effect for patient,
adjusting for continuous time, the baseline score, and indicator variables for
study sites, and report values for comparisons between true acupuncture
vs sham acupuncture and true acupuncture vs waitlist control for week 6, 12, 16,
20, and 24 assessment times.
A, P = .04 (vs sham acupuncture); P<.001 (vs waitlist control).
B, P = .18 (vs sham acupuncture); P = .003 (vs waitlist control).
C, P = .08 (vs sham acupuncture); P = .003 (vs waitlist control).
D, P = .02 (vs sham acupuncture); P<.001 (vs waitlist control).

analysis suggested that the proportion of patients with a
2-point improvement at 6 weeks in the true acupuncture
group (58%) was greater than that in the sham acupuncture
group (33%) and in the waitlist control group (31%).
In this study, the addition of maintenance true acu-
puncture once a week for an additional 6 weeks, compared
with sham acupuncture, was associated with statistically
significant improvements in average pain, but no significant
improvement in worst pain, pain interference, pain severity,
or worst stiffness at 12 weeks. Compared with waitlist con-
trol, additional maintenance true acupuncture was associ-
ated with significant improvements in worst pain, average
pain, pain interference, pain severity and worst stiffness at
12 weeks.
Both pharmacological and nonpharmacological inter-
ventions have been studied to treat aromatase inhibitor–
associated arthralgias.19,20,33,34 Although some interventions
have shown a benefit from the intervention, most have been
difficult to interpret due to methodological limitations.
Placebo-controlled trials of pharmacological interventions,
such as vitamin D, omega-3-fatty acids, and duloxetine, have
consistently shown placebo effects of as much as 50%;
however, only duloxetine was found to show a statistically
significant between-group difference. 1 9,3 3 ,3 4 Trials of
unblinded, modality-based therapies, such as exercise, have
shown benefit; however, no placebo effect was observed in
the control group,20 raising questions about response bias.35
The lack of a placebo effect in prior acupuncture trials raised

174 JAMA July 10, 2018 Volume 320, Number 2 (Reprinted) jama.com

© 2018 American Medical Association. All rights reserved.

Downloaded From: by a Tufts Univ. Hirsh Health Sciences Library User on 07/10/2018
 Acupuncture vs Control for Joint Pain Related to Aromatase Inhibitor Treatment for Breast Cancer Original Investigation Research

concerns about the integrity of the sham blinding, uninten- Limitations

tional crossover in the sham treatment groups, or a possibil-
ity of a physiological effect of the sham. The current trial
was designed with a waitlist control group to address prior
concerns about the sham effects.
The 4 prior studies of acupuncture for aromatase inhibi-
tor arthralgias did not evaluate maintenance or durability of
effect.10,11,36,37 Most studies of acupuncture in patients with
cancer have evaluated the benefits of true acupuncture
among women with hot flashes, and several evaluated persis-
tent effects following completion of the intervention, sug-
gesting that the effects may be long lasting. For example, in a
randomized unblinded trial of weekly acupuncture for 12
weeks, the intervention was associated with a significantly
lower hot flash score at 3 months and also at 6 months fol-
lowing completion of the acupuncture intervention.38 Acu-
puncture is appealing to some patients because the adverse
effects are generally limited compared with medications
used to control pain, such as opiates or duloxetine. However,
despite the benefits and limited toxicities, acupuncture is not
covered by many insurance plans.
This study has several strengths. The intervention was
conducted at multiple sites by acupuncturists who under-
went rigorous standardized training and administered the
repeat assessments throughout the time frame of the study.
Acupuncturists had diverse training backgrounds, and
patients were treated at both academic and community sites
throughout the United States, both of which increase the
generalizability of the findings. The retention and comple-
tion rates were similar across groups, reducing the concern
about differential dropout.
This study also has several limitations. First, blinding was not
possible for patients randomized to the waitlist control group.
Second, a significantly higher proportion of patients in the true
acupuncture group compared with the sham acupuncture
group believed they were receiving true acupuncture, sug-
gesting the possibility of lack of successful masking of this
group. Third, primary outcomes were based on relatively short-
term measurements, and long-term follow-up beyond 12
months was not available. Fourth, the study design did not in-
corporate the pain literature regarding a comparison be-
tween groups of a clinically meaningful reduction for an in-
dividual, and therefore, these results are post hoc. Fifth, there
was no correction for multiple comparisons for the second-
ary analyses, and therefore, these findings should be consid-
ered exploratory. Sixth, accrual was slower than expected due
to acupuncture staff turnover and retraining, expansion of
study sites, scheduling difficulties, and the widespread avail-
ability of acupuncture, limiting patients’ enthusiasm for en-
tering a randomized trial. Seventh, uptake in the community
may be limited by the expense of the intervention, logistical
challenges, and uncertain experience of the acupuncturist with
the study methodology.
Conclusion
Among postmenopausal women with early-stage breast can-
cer and aromatase inhibitor–related arthralgias, true acu-
puncture compared with sham acupuncture or with waitlist
control, resulted in a statistically significant reduction in
joint pain at 6 weeks, although the observed improvement
was of uncertain clinical importance.

ARTICLE INFORMATION
Accepted for Publication: June 6, 2018.
Author Affiliations: Columbia University Medical
Center, New York, New York (Hershman); Fred
Hutchinson Cancer Research Center, Seattle,
Washington (Unger, Greenlee, Lew, Darke); SWOG
Statistics and Data Management Center, Seattle,
Washington (Unger, Lew, Darke); Mount Sinai
Hospital, New York, New York (Capodice); Kaiser
Permanente Medical Center, Walnut Creek,
California (Kengla); Spectrum Health Medical
Group, Grand Rapids, Michigan (Melnik); NCORP of
the Carolinas (Greenville Health System),
Greenville, South Carolina (Jorgensen); St Luke’s
Mountain States Tumor Institute (PCRC NCORP),
Boise, Idaho (Kreisle); Division of Cancer
Prevention, National Cancer Institute, Bethesda,
Maryland (Minasian); AIM Specialty Health,
Chicago, Illinois (Fisch); University of Utah
Huntsman Cancer Institute, Salt Lake City (Henry).
Author Contributions: Drs Hershman and Unger
had full access to all of the data in the study and
takes responsibility for the integrity of the data
and the accuracy of the data analysis.
Concept and design: Hershman, Unger, Greenlee,
Capodice, Minasian, Crew.
Acquisition, analysis, or interpretation of data:
Hershman, Unger, Greenlee, Lew, Darke, Kengla,
Melnik, Jorgensen, Kreisle, Fisch, Henry, Crew.
Drafting of the manuscript: Hershman, Unger,
Capodice, Kreisle, Crew.
Critical revision of the manuscript for important
intellectual content: Hershman, Unger, Greenlee,
Lew, Darke, Kengla, Melnik, Jorgensen, Minasian,
Fisch, Henry, Crew.
Statistical analysis: Hershman, Unger, Lew, Darke.
Obtained funding: Hershman, Crew.
Administrative, technical, or material support:
Hershman, Greenlee, Capodice, Jorgensen, Crew.
Supervision: Hershman, Unger, Greenlee, Capodice,
Kengla, Crew.
Conflict of Interest Disclosures: All authors have
completed and submitted the ICMJE Form for
Disclosure of Potential Conflicts of Interest. Ms
Darke reports receipt of grants from the National
Cancer Institute (NCI) during the conduct of the
study. No other disclosures were reported.
Funding/Support: Funding was provided by the
National Institutes of Health (NIH) National Center
for Complementary and Integrative Health and
the Office of Research on Women’s Health
R01AT006376; NIH/NCI/DCP (Division of Cancer
Prevention) grant UG1CA189974 and legacy grant
U10CA37429.
Role of the Funder: The funder had no role in the
design and conduct of the study; collection,
management, analysis, and interpretation of the
data; preparation, review or approval of the
manuscript; and decision to submit the manuscript
for publication.
Disclaimer: The content is solely the responsibility
of the authors and does not necessarily represent
the official views of the NIH.
Meeting Presentation: This research was
presented at the San Antonio Breast Cancer
Symposium, December 2017.
REFERENCES
1. Burstein HJ, Lacchetti C, Anderson H, et al.
Adjuvant endocrine therapy for women with
hormone receptor-positive breast cancer: American
Society of Clinical Oncology Clinical Practice
Guideline Update on Ovarian Suppression. J Clin
Oncol. 2016;34(14):1689-1701.
2. Crew KD, Greenlee H, Capodice J, et al.
Prevalence of joint symptoms in postmenopausal
women taking aromatase inhibitors for early-stage
breast cancer. J Clin Oncol. 2007;25(25):3877-3883.
3. Castel LD, Hartmann KE, Mayer IA, et al. Time
course of arthralgia among women initiating
aromatase inhibitor therapy and a postmenopausal
comparison group in a prospective cohort. Cancer.
2013;119(13):2375-2382.
4. Henry NL, Azzouz F, Desta Z, et al. Predictors of
aromatase inhibitor discontinuation as a result of
treatment-emergent symptoms in early-stage
breast cancer. J Clin Oncol. 2012;30(9):936-942.
5. Hershman DL, Kushi LH, Shao T, et al. Early
discontinuation and nonadherence to adjuvant

jama.com (Reprinted) JAMA July 10, 2018 Volume 320, Number 2 175

© 2018 American Medical Association. All rights reserved.

Downloaded From: by a Tufts Univ. Hirsh Health Sciences Library User on 07/10/2018
 Research Original Investigation
hormonal therapy in a cohort of 8,769 early-stage
breast cancer patients. J Clin Oncol. 2010;28(27):
4120-4128.
6. Goss PE, Ingle JN, Pritchard KI, et al. Extending
Aromatase-Inhibitor Adjuvant Therapy to 10 Years.
N Engl J Med. 2016;375(3):209-219.
7. Chirgwin JH, Giobbie-Hurder A, Coates AS, et al.
Treatment adherence and its impact on
disease-free survival in the breast international
group 1-98 trial of tamoxifen and letrozole, alone
and in sequence. J Clin Oncol. 2016;34(21):2452-
2459.
8. Helms J. Acupuncture Energetics: A Clinical
Approach for Physicians. Berkeley, CA: Medical
Acupuncture Publishers; 1995.
9. Tedesco D, Gori D, Desai KR, et al. Drug-free
interventions to reduce pain or opioid consumption
after total knee arthroplasty. JAMA Surg. 2017;152
(10):e172872.
10. Crew KD, Capodice JL, Greenlee H, et al.
Randomized, blinded, sham-controlled trial of
acupuncture for the management of aromatase
inhibitor-associated joint symptoms in women with
early-stage breast cancer. J Clin Oncol. 2010;28(7):
1154-1160.
11. Mao JJ, Xie SX, Farrar JT, et al. A randomised
trial of electro-acupuncture for arthralgia related to
aromatase inhibitor use. Eur J Cancer. 2014;50(2):
267-276.
12. Pocock SJ, Simon R. Sequential treatment
assignment with balancing for prognostic factors in
the controlled clinical trial. Biometrics. 1975;31(1):
103-115.
13. MacPherson H, Altman DG, Hammerschlag R,
et al; STRICTA Revision Group. Revised standards
for reporting interventions in clinical trials of
acupuncture (STRICTA): extending the CONSORT
statement. J Evid Based Med. 2010;3(3):140-155.
14. National Institutes of Health. Racial and ethnic
categories and definitions for NIH diversity
programs and for other reporting purposes. Notice
number NOT-OD-15-089. https://grants.nih.gov
/grants/guide/notice-files/not-od-15-089.html.
April 8, 2015. Accessed June 21, 2018.
15. Oken MM, Creech RH, Tormey DC, et al. Toxicity
and response criteria of the Eastern Cooperative
Oncology Group. Am J Clin Oncol. 1982;5(6):649-655.
16. Daut RL, Cleeland CS, Flanery RC. Development
of the Wisconsin Brief Pain Questionnaire to assess
pain in cancer and other diseases. Pain. 1983;17(2):
197-210.
176 JAMA July 10, 2018 Volume 320, Number 2 (Reprinted)
© 2018 American Medical Association. All rights reserved.
Downloaded From: by a Tufts Univ. Hirsh Health Sciences Library User on 07/10/2018
Acupuncture vs Control for Joint Pain Related to Aromatase Inhibitor Treatment for Breast Cancer
17. Greenlee H, Crew KD, Capodice J, et al.
Methods to standardize a multicenter acupuncture
trial protocol to reduce aromatase inhibitor-related
joint symptoms in breast cancer patients.
J Acupunct Meridian Stud. 2015;8(3):152-158.
18. Hui KK, Nixon EE, Vangel MG, et al.
Characterization of the “deqi” response in
acupuncture. BMC Complement Altern Med.
2007;7:33.
19. Hershman DL, Unger JM, Crew KD, et al.
Randomized multicenter placebo-controlled trial of
omega-3 fatty acids for the control of aromatase
inhibitor-induced musculoskeletal pain: SWOG
S0927. J Clin Oncol. 2015;33(17):1910-1917.
20. Irwin ML, Cartmel B, Gross CP, et al.
Randomized exercise trial of aromatase
inhibitor-induced arthralgia in breast cancer
survivors. J Clin Oncol. 2015;33(10):1104-1111.
21. Farrar JT, Young JP Jr, LaMoreaux L, Werth JL,
Poole RM. Clinical importance of changes in chronic
pain intensity measured on an 11-point numerical
pain rating scale. Pain. 2001;94(2):149-158.
22. Bellamy N, Buchanan WW, Goldsmith CH,
Campbell J, Stitt LW. Validation study of WOMAC:
a health status instrument for measuring clinically
important patient relevant outcomes to
antirheumatic drug therapy in patients with
osteoarthritis of the hip or knee. J Rheumatol. 1988;
15(12):1833-1840.
23. Sautner J, Andel I, Rintelen B, Leeb BF.
Development of the M-SACRAH, a modified,
shortened version of SACRAH (Score for the
Assessment and Quantification of Chronic
Rheumatoid Affections of the Hands).
Rheumatology (Oxford). 2004;43(11):1409-1413.
24. Fallowfield LJ, Leaity SK, Howell A, Benson S,
Cella D. Assessment of quality of life in women
undergoing hormonal therapy for breast cancer.
Breast Cancer Res Treat. 1999;55(2):189-199.
25. Garcia SF, Cella D, Clauser SB, et al.
Standardizing patient-reported outcomes
assessment in cancer clinical trials. J Clin Oncol.
2007;25(32):5106-5112.
26. Laird NM, Ware JH. Random-effects models for
longitudinal data. Biometrics. 1982;38(4):963-974.
27. Dworkin RH, Turk DC, McDermott MP, et al.
Interpreting the clinical importance of group
differences in chronic pain clinical trials: IMMPACT
recommendations. Pain. 2009;146(3):238-244.
28. Dworkin RH, Turk DC, Wyrwich KW, et al
Interpreting the clinical importance of treatment
outcomes in chronic pain clinical trials: IMMPACT
recommendations. J Pain. 2008;9(2):105-121.
29. Zou G. A modified poisson regression approach
to prospective studies with binary data. Am J
Epidemiol. 2004;159(7):702-706.
30. SAS Institute Inc. SAS 9.4 System Options. 5th
ed. Cary, NC: SAS Institute Inc; 2016.
31. Smith EM, Pang H, Cirrincione C, et al; Alliance
for Clinical Trials in Oncology. Effect of duloxetine
on pain, function, and quality of life among patients
with chemotherapy-induced painful peripheral
neuropathy. JAMA. 2013;309(13):1359-1367.
32. Kroenke K, Theobald D, Wu J, et al. Effect of
telecare management on pain and depression in
patients with cancer. JAMA. 2010;304(2):163-171.
33. Shapiro AC, Adlis SA, Robien K, et al.
Randomized, blinded trial of vitamin D3 for treating
aromatase inhibitor-associated musculoskeletal
symptoms (AIMSS). Breast Cancer Res Treat. 2016;
155(3):501-512.
34. Henry NL, Unger JM, Schott AF, et al.
Randomized, multicenter, placebo-controlled
clinical trial of duloxetine versus placebo for
aromatase inhibitor-associated arthralgias in
early-stage breast cancer: SWOG S1202. J Clin Oncol.
2018;36(4):326-332.
35. Henry NL, Griggs JJ. The power of the placebo
in symptom management. J Clin Oncol. 2015;33(17):
1870-1872.
36. Bao T, Cai L, Giles JT, et al. A dual-center
randomized controlled double blind trial assessing
the effect of acupuncture in reducing
musculoskeletal symptoms in breast cancer
patients taking aromatase inhibitors. Breast Cancer
Res Treat. 2013;138(1):167-174.
37. Oh B, Kimble B, Costa DS, et al. Acupuncture
for treatment of arthralgia secondary to aromatase
inhibitor therapy in women with early breast
cancer. Acupunct Med. 2013;31(3):264-271.
38. Lesi G, Razzini G, Musti MA, et al. Acupuncture
as an integrative approach for the treatment of hot
flashes in women with breast cancer. J Clin Oncol.
2016;34(15):1795-1802.
jama.com
 original reports
Dietary Supplement Use During Chemotherapy
and Survival Outcomes of Patients With Breast
Cancer Enrolled in a Cooperative Group Clinical
Trial (SWOG S0221)
Christine B. Ambrosone, PhD1; Gary R. Zirpoli, PhD2; Alan D. Hutson, PhD1; William E. McCann1; Susan E. McCann, PhD, RD1;
William E. Barlow, PhD3; Kara M. Kelly, MD1; Rikki Cannioto, PhD, EdD1; Lara E. Sucheston-Campbell, PhD4; Dawn L. Hershman, MD5;
Joseph M. Unger, PhD3; Halle C.F. Moore, MD6; James A. Stewart, MD7; Claudine Isaacs, MD8; Timothy J. Hobday, MD9;
Muhammad Salim, MD10; Gabriel N. Hortobagyi, MD11; Julie R. Gralow, MD12; George T. Budd, MD6; and Kathy S. Albain, MD13
abstract

PURPOSE Despite reported widespread use of dietary supplements during cancer treatment, few empirical data
with regard to their safety or efficacy exist. Because of concerns that some supplements, particularly anti-
oxidants, could reduce the cytotoxicity of chemotherapy, we conducted a prospective study ancillary to
a therapeutic trial to evaluate associations between supplement use and breast cancer outcomes.
METHODS Patients with breast cancer randomly assigned to an intergroup metronomic trial of cyclophos-
phamide, doxorubicin, and paclitaxel were queried on their use of supplements at registration and during
treatment (n =1,134). Cox proportional hazards regression adjusting for clinical and lifestyle variables was used.
Recurrence and survival were indexed at 6 months after enrollment using a landmark approach.
RESULTS There were indications that use of any antioxidant supplement (vitamins A, C, and E; carotenoids;
coenzyme Q10) both before and during treatment was associated with an increased hazard of recurrence
(adjusted hazard ratio [adjHR], 1.41; 95% CI, 0.98 to 2.04; P = .06) and, to a lesser extent, death (adjHR, 1.40;
95% CI, 0.90 to 2.18; P = .14). Relationships with individual antioxidants were weaker perhaps because of small
numbers. For nonantioxidants, vitamin B12 use both before and during chemotherapy was significantly as-
sociated with poorer disease-free survival (adjHR, 1.83; 95% CI, 1.15 to 2.92; P , .01) and overall survival
(adjHR, 2.04; 95% CI, 1.22 to 3.40; P , .01). Use of iron during chemotherapy was significantly associated with
recurrence (adjHR, 1.79; 95% CI, 1.20 to 2.67; P , .01) as was use both before and during treatment (adjHR,
1.91; 95% CI, 0.98 to 3.70; P = .06). Results were similar for overall survival. Multivitamin use was not as-
sociated with survival outcomes.
CONCLUSION Associations between survival outcomes and use of antioxidant and other dietary supplements
both before and during chemotherapy are consistent with recommendations for caution among patients when
considering the use of supplements, other than a multivitamin, during chemotherapy.
J Clin Oncol 38:804-814. © 2019 by American Society of Clinical Oncology
ASSOCIATED
CONTENT
Appendix INTRODUCTION affecting treatment efficacy. To address this question
Data Supplement Use of dietary supplements after a cancer diagnosis is prospectively, we initiated the Diet, Exercise, Lifestyle
Author affiliations common1,2; however, few empirical data provide a basis and Cancer Prognosis (DELCaP) study,8 a correlative
and support
for clinical recommendations for use during chemo- to a phase III trial led by SWOG (S0221) that randomly
information (if assigned patients with breast cancer to different
applicable) appear therapy.3 Because one cytotoxic mechanism of cancer
therapeutics is through the generation of reactive oxygen treatment schedules with doxorubicin, cyclophos-
at the end of this
article. species (ROS), there has been concern that use of di- phamide, and paclitaxel9 and queried participants on
Accepted on October etary supplements during treatment, particularly anti- their use of supplements at randomization and at
9, 2019 and oxidants, could reduce treatment efficacy.4 In fact, there completion of chemotherapy. We previously reported
published at
have been clinical recommendations that patients not on the prevalence of supplement use in S0221, par-
ascopubs.org/journal/ ticularly in relation to recommendations from treating
jco on December 19, take antioxidant supplements during chemotherapy.2,5-7
oncologists8 and in relation to chemotherapy-induced
2019: DOI https://doi.
org/10.1200/JCO.19. The use of supplements, however, could possibly peripheral neuropathy (CIPN).10 No associations were
01203 reduce treatment-related adverse effects while not found between the use of antioxidants and CIPN, but

804 Volume 38, Issue 8

Downloaded from ascopubs.org by 200.92.202.106 on August 19, 2022 from 200.092.202.106
Copyright © 2022 American Society of Clinical Oncology. All rights reserved.
 Supplement Use During Chemotherapy and Breast Cancer Survival

there was reduced risk of CIPN with use of multivitamins
before diagnosis and, to a lesser extent, during chemo-
therapy. These associations, however, must be balanced
against potential adverse effects of supplement use on
recurrence and mortality. Here, we address the primary
goal of the DELCaP study: To determine whether use of
supplements during chemotherapy, particularly antioxi-
dants, has any effects on survival outcomes.
METHODS
Study Population
DELCaP began approximately 20 months after S0221
opened, with information embedded in the informed con-
sent for the trial as previously described.8 Of the 2,014
patients who were eligible, 1,607 (80%) agreed to partic-
ipate in DELCaP (Fig 1). After initial contact, a baseline
questionnaire (Q1) was sent for completion before begin-
ning chemotherapy, which queried about regular use
(at least once per week) before diagnosis and between
diagnosis and enrollment in the trial. The validated sur-
vey was adapted from the VITAL study11 and piloted
among patients with cancer enrolled in the DataBank and
BioRepository at Roswell Park Comprehensive Cancer
Center.12 If surveys were not returned within 2 weeks, each
patient was called with a reminder and an offer to complete
the questionnaire over the telephone, followed by monthly
calls if no response. After 4 months of nonresponse, the
patient was withdrawn as a passive refuser. For Q1 late
returns, 17% of surveys were received between 3 and
6 months after having been sent, and 2% were received
after 6 months. Approximately 6 months after randomi-
zation, when chemotherapy was scheduled to be com-
pleted, a second questionnaire (Q2) was sent to patients
who had completed Q1 (n = 1,340; 83%) that queried
about supplement use during chemotherapy; 1,134 (85%)
of these patients completed Q2. An example of questions
for Q1 and Q2 is shown in Data Supplement. Question-
naires were reviewed for missing data, and patients with
missing responses were contacted to maximize complete
data. Reasons for not returning questionnaires are shown in
Figure 1, with passive refusal referring to patients who did
not return questionnaires but did not formally withdraw
from the study. Patients who did not return Q2 were sig-
nificantly more likely than those who did to be nonwhite, to
be younger (48.35 v 51.65 years), and to have less grade 3
and 4 hematologic toxicity (5.16% v 11.73%). For a limited
number of institutions that did not allow contact with their

Randomly assigned to SWOG S0221 (N = 2,716) Enrolled in S0221 before DELCaP

Eligible for DELCaP (n = 2,014) (n = 702)

Did not consent to DELCaP

(n = 407)

FIG 1. Study schema for participants

included in the Diet, Exercise, Life-
Consented to DELCaP ancillary style and Cancer Prognosis (DELCaP)
(questionnaire component; n = 1,607) analysis, an observational study an-
Did not complete Q1 cillary to SWOG S0221, a randomized
(n = 267) treatment trial for high-risk breast
CRA Mail Total cancer. CRA, clinical research as-
Deceased 1 2 3 sociate; Q1, questionnaire 1 completed
Not eligible 5 25 30 at registration in the trial before initi-
Passive refusal 5 205 210 ation of chemotherapy; Q2, ques-
Active refusal 1 19 20 tionnaire 2 completed at completion
of active treatment (approximately
Completed Q1 (n = 1,340)
CRA administered (n = 119) 6 months after registration in S0221;
Did not complete Q2 ‘other’ patients were collapsed into
(n = 206) other categories). *Other: mental
CRA Mail Total disability, loss of vision, physician re-
Deceased 0 4 4 moved patient from study, dropped
Not eligible 0 1 1 on suspicion of metastasis.
Passive refusal 6 178 184
Active refusal 3 14 17
Total 9 197 206
Completed Q1 and Q2 (n = 1,134)
CRA administered (n = 43)

Journal of Clinical Oncology 805

Downloaded from ascopubs.org by 200.92.202.106 on August 19, 2022 from 200.092.202.106

Copyright © 2022 American Society of Clinical Oncology. All rights reserved.
 Ambrosone et al
patients (14 of 320), nurses or clinical research associates
administered the questionnaires; this applied to 8% of
participants. This study was approved by the institutional
review boards at Roswell Park Comprehensive Cancer
Center and at institutions participating in S0221.
Statistical Analyses
A landmark survival analysis was performed that desig-
nated a time point that occurred during the follow-up period
(landmark time) and analyzed only those patients who
survived to that point13; here, the 6-month survey was the
landmark time point. Among patients alive without re-
currence at 6 months, disease-free survival (DFS) was
defined as the time from then until the first documentation
of breast cancer recurrence, new breast cancer, or death,
whichever came first. Patients who had not experienced
recurrence were censored on the date of last clinical
contact. Overall survival (OS) was defined as the time from
6 months after enrollment until death as a result of any
cause, with patients who were still alive censored on their
last follow-up date. The median follow-up time for those
who did not have an event was 8.1 years. Baseline char-
acteristics in relation to DFS and OS were compared, and
Cox proportional hazards regressions were used to evaluate
associations. Because use of supplements before enroll-
ment may influence relationships between use during
treatment and outcomes, we categorized use (yes [Y]/no
[N]) at baseline and during chemotherapy combined as
NN, NY, YN, and YY, with the NN group (did not use at
either time) set as the referent. Tests for proportionality
using time-varying covariates were carried out at a = .05 for
YY, NY, and YN versus NN in the univariable setting;
no violations were observed. Minimally adjusted models
included dosing arm as a stratification factor that corre-
sponded to the original randomized treatment assign-
ment and age (model 1). Fully adjusted models also
included tumor characteristics (estrogen receptor [ER],
progesterone receptor, human epidermal growth factor
receptor 2, lymph node status, and tumor size [stage was
not available]), grades 3 or 4 hematologic toxicity, body
mass index, physical activity, smoking, and alcohol con-
sumption (model 2). To account for additional sources
of antioxidants and the effects of use of multiple supple-
ments, we further adjusted for multivitamin use (model 3).
In sensitivity analyses, patients who took five or more
supplements (n = 189) were removed from analysis. In
exploratory analyses, we created a variable for number of
supplements, which was considered in models, and per-
formed further analyses stratified by ER status.
The landmark analysis at 6 months was based on the Q2
survey response among patients who also completed Q1.
The small fraction of OS events in the first 6 months (n = 4)
precluded those patients from completing the 6-month
survey; thus, among patients who returned Q2, there
were no DFS events within the first 6 months. Patients who
did not complete Q2 and died after the landmark time point
806 © 2019 by American Society of Clinical Oncology
Downloaded from ascopubs.org by 200.92.202.106 on August 19, 2022 from 200.092.202.106
Copyright © 2022 American Society of Clinical Oncology. All rights reserved.
were included in the passive refusers category. All statistical
tests were two-sided, and P , .05 was considered statis-
tically significant. All analyses were performed using SAS
9.4 statistical software (SAS Institute, Cary NC).
RESULTS
Characteristics of the 1,340 patients who completed Q1 are
listed in Table 1. Patients who experienced recurrence
(n = 310) or died (n = 222) were more likely to be older, to
be postmenopausal, and to have a higher body mass index
(all P , .05); to have completed grade school or some high
school (P = .03); to self-report as black (P = .10); and to
have poor prognosis factors ($ 4 positive lymph nodes, ER
negativity, progesterone receptor negativity, human epi-
dermal growth factor receptor 2 negativity; P , .01).
Among the 1,134 patients who completed both Q1 and Q2,
there were 251 recurrences and 181 deaths. As previously
reported8 and listed in Appendix Table A1 (online only), the
prevalence of supplement use, particularly antioxidants,
was low compared with reports in the literature of use by
patients with cancer1,2 and tended to decrease during
treatment. For example, vitamin C was used by 20.5% of
patients before treatment but only 12.2% during therapy.
Vitamins E and A were taken during treatment by , 10% of
patients. Use of any antioxidant during treatment (vitamins
C, A, and E; carotenoids; or coenzyme Q10) was observed
among 17.5% of patients, whereas 44% of patients took
multivitamins during chemotherapy. Appendix Table A2
(online only) lists the prevalence of use at both time points
according to DFS and OS.
Table 2 lists the relationships between DFS and OS and
antioxidant supplement use before and during chemo-
therapy. In fully adjusted models, patients who used any
antioxidant both before and during treatment were at an
increased hazard of recurrence (adjusted hazard ratio
[adjHR], 1.41; 95% CI, 0.98 to 2.04; P = .06; Fig 2A) and,
to a lesser degree, death (HR, 1.40; 95% CI, 0.90 to 2.18;
Fig 2B). There were no relationships with outcomes for use
of antioxidants only before treatment initiation or only
during chemotherapy.
Associations for the majority of single antioxidant supple-
ments were not statistically significant. There was a sug-
gestion that use of vitamins C, E, and coenzyme Q10 before
and during treatment was associated with an increased
hazard of recurrence (adjHR, 1.36 [95% CI, 0.87 to 2.13],
1.47 [95% CI, 0.81 to 2.67], and 1.75 [95% CI, 0.77 to
4.01], respectively), with similar HRs for OS. Although the
number of users was quite low (n = 5), use of vitamin A both
before and during chemotherapy was associated with
a significantly increased hazard of recurrence and death,
with similar relationships for use of carotenoid supple-
ments. Inclusion of multivitamin use in fully adjusted
models resulted in slightly attenuated estimates for some
variables (Table 2).
Volume 38, Issue 8
 Supplement Use During Chemotherapy and Breast Cancer Survival

TABLE 1. Baseline Demographic and Clinical Characteristics of the Overall DELCaP Study Population in S0221 by DFS and OS Status
(n = 1,340)
DFS, No. (%) OS, No. (%)

Characteristic Progression Free Progression P Alive Deceased P

No. of patients 1,030 (76.87) 310 (23.13) 1,118 (83.44) 222 (16.56)
Demographic
Mean age, years (SD) 50.87 (9.64) 52.75 (10.56) , .01* 50.89 (9.72) 53.39 (10.47) , .01
Mean BMI, kg/m2 (SD) 29.14 (6.75) 30.16 (7.05) .02 29.15 (6.77) 30.49 (7.07) , .01
Menopausal status , .01 , .01
Pre 506 (80.32) 124 (19.68) 545 (86.51) 85 (13.49)
Post 515 (73.99) 181 (26.01) 563 (80.89) 133 (19.11)
Self-reported race .38 .10
White 864 (77.28) 254 (22.72) 937 (83.81) 181 (16.19)
Black 64 (68.09) 30 (31.91) 69 (73.40) 25 (26.60)
Multiracial 35 (77.78) 10 (22.22) 38 (84.44) 7 (15.56)
American Indian 10 (76.92) 3 (23.08) 11 (84.62) 2 (15.38)
Asian/Pacific Islander 40 (83.33) 8 (16.67) 44 (91.17) 4 (0.08)
Other 14 (77.78) 4 (22.22) 15 (83.33) 3 (16.67)
Self-reported ethnicity .52 .73
Non-Hispanic 978 (76.95) 293 (23.05) 1,059 (83.32) 212 (16.68)
Hispanic 44 (73.33) 16 (26.67) 51 (85.00) 9 (15.00)
Education .06 .03
Grade school or some high school 61 (65.59) 32 (34.41) 69 (74.19) 24 (25.81)
High school graduate or GED 219 (76.31) 68 (23.69) 234 (81.53) 53 (18.47)
Some college or technical school 380 (78.51) 104 (21.49) 412 (85.12) 72 (14.88)
College graduate 227 (79.37) 59 (20.63) 248 (86.71) 38 (13.29)
Advanced degree 137 (70.46) 47 (25.54) 149 (80.98) 35 (19.02)
Clinical
Nodal status , .01 , .01
Negative 290 (83.09) 59 (16.91) 309 (88.54) 40 (11.46)
1-3 positive nodes 406 (80.88) 96 (19.12) 433 (86.25) 69 (13.75)
$ 4 positive nodes 332 (68.17) 155 (31.83) 374 (76.80) 113 (23.20)
ER/PR status .01 , .01
Positive (either or both positive) 690 (79.04) 183 (20.96) 751 (86.03) 122 (13.97)
Negative (both negative) 338 (72.84) 126 (27.16) 365 (78.66) 99 (21.34)
HER2 status , .01 .02
Negative 793 (75.17) 262 (24.83) 867 (82.18) 188 (17.82)
Positive 231 (82.80) 48 (17.20) 245 (87.81) 34 (12.19)

NOTE. P values reflect pooled t test for age and BMI.

Abbreviations: BMI, body mass index; DELCaP, Diet, Exercise, Lifestyle and Cancer Prognosis; DFS, disease-free survival; ER, estrogen
receptor; GED, general educational development; HER2, human epidermal growth factor receptor 2; OS, overall survival; PR, progesterone
receptor; SD, standard deviation.
*Satterthwaite approximation for age and progression-free survival (rejecting equal variance at P # .05); otherwise, x2 test.

Table 3 lists associations with nonantioxidant supplements. treatment outcomes and use of vitamin B12 and iron
Multivitamin use before, during, or at both time points was supplements (Figs 3A and 3B). Use of vitamin B12 both
not associated with survival outcomes (Figs 2C and 2D), nor before and during treatment was associated with poorer DFS
was vitamin D use. Striking associations were found between (HR, 1.83; 95% CI, 1.15 to 2.92) and OS (HR, 2.04; 95% CI,

Journal of Clinical Oncology 807

Downloaded from ascopubs.org by 200.92.202.106 on August 19, 2022 from 200.092.202.106

Copyright © 2022 American Society of Clinical Oncology. All rights reserved.
 Ambrosone et al

TABLE 2. Adjusted Models for Associations Between Use of Antioxidant Supplements Before and During Chemotherapy and DFS and OS in S0221: DELCaP
Study Participants Who Completed Both Questionnaires 1 and 2 (n = 1,134)
Reported DFS, HR (95% CI) OS, HR (95% CI)
Antioxidant Use
Before and
During Fully Adjusted With Fully Adjusted With
Chemotherapy Minimally Adjusted* Fully Adjusted† MV Use‡ Minimally Adjusted* Fully Adjusted† MV Use‡
Any
NN 1.0 1.0 1.0 1.0 1.0 1.0
NY 0.79 (0.43 to 1.43) 0.93 (0.52 to 1.64) 0.92 (0.52 to 1.64) 0.97 (0.50 to 1.87) 1.03 (0.53 to 1.98) 1.01 (0.52 to 1.97)
YN 1.02 (0.74 to 1.43) 1.10 (0.79 to 1.53) 1.04 (0.74 to 1.47) 1.08 (0.27 to 1.58) 1.19 (0.81 to 1.76) 1.11 (0.75 to 1.66)
YY 1.42 (1.00 to 2.03) 1.41 (0.98 to 2.04) 1.38 (0.94 to 2.02) 1.37 (0.90 to 2.09) 1.40 (0.90 to 2.18) 1.34 (0.85 to 2.13)
Vitamin C
NN 1.0 1.0 1.0 1.0 1.0 1.0
NY 1.01 (0.56 to 1.83) 1.14 (0.65 to 2.01) 1.14 (0.64 to 2.03) 1.11 (0.56 to 2.20) 1.14 (0.58 to 2.27) 1.15 (0.58 to 2.31)
YN 1.05 (0.73 to 1.50) 1.10 (0.76 to 1.59) 1.04 (0.72 to 1.52) 1.19 (0.79 to 1.79) 1.35 (0.89 to 2.04) 1.27 (0.83 to 1.93)
YY 1.37 (0.89 to 2.11) 1.36 (0.87 to 2.13) 1.31 (0.83 to 2.08) 1.42 (0.86 to 2.34) 1.42 (0.83 to 2.40) 1.37 (0.80 to 2.34)
Vitamin A
NN 1.0 1.0 1.0 1.0 1.0 1.0
NY 1.61 (0.78 to 3.32) 1.51 (0.71 to 3.23) 1.51 (0.70 to 3.29) 1.27 (0.46 to 3.49) 1.20 (0.44 to 3.28) 1.25 (0.45 to 3.49)
YN 0.85 (0.39 to 1.87) 0.75 (0.33 to 1.71) 0.71 (0.31 to 1.63) 0.77 (0.29 to 2.05) 0.71 (0.26 to 1.94) 0.66 (0.24 to 1.83)
YY 4.05 (1.46 to 11.25) 4.23 (1.32 to 13.57) 4.06 (1.26 to 13.16) 4.14 (1.81 to 9.51) 3.53 (1.05 to 11.93) 3.20 (0.93 to 10.99)
Vitamin E
NN 1.0 1.0 1.0 1.0 1.0 1.0
NY 1.16 (0.60 to 2.24) 1.15 (0.60 to 2.18) 1.13 (0.59 to 2.16) 1.13 (0.52 to 2.49) 1.22 (0.56 to 2.62) 1.19 (0.55 to 2.58)
YN 0.99 (0.68 to 1.43) 1.03 (0.71 to 1.50) 0.98 (0.67 to 1.44) 1.0 (0.66 to 1.53) 1.10 (0.71 to 1.71) 1.04 (0.66 to 1.62)
YY 1.62 (0.92 to 2.88) 1.47 (0.81 to 2.67) 1.38 (0.75 to 2.54) 1.65 (0.86 to 3.17) 1.53 (0.77 to 3.04) 1.39 (0.68 to 2.82)
Coenzyme Q10
NN 1.0 1.0 1.0 1.0 1.0 1.0
NY 1.16 (0.53 to 2.56) 1.34 (0.59 to 3.02) 1.35 (0.59 to 3.06) 1.11 (0.42 to 2.92) 1.28 (0.47 to 3.48) 1.34 (0.49 to 3.67)
YN 1.33 (0.70 to 2.52) 1.30 (0.66 to 2.56) 1.28 (0.65 to 2.51) 1.16 (0.54 to 2.47) 1.13 (0.50 to 2.58) 1.08 (0.47 to 2.48)
YY 2.02 (0.92 to 4.47) 1.75 (0.77 to 4.01) 1.68 (0.73 to 3.89) 2.20 (0.92 to 5.26) 2.04 (0.82 to 5.09) 1.88 (0.75 to 4.76)
Carotenoid
NN 1.0 1.0 1.0 1.0 1.0 1.0
NY 2.59 (0.78 to 8.57) 3.15 (1.14 to 8.65) 3.20 (1.16 to 8.87) 3.59 (0.93 to 13.84) 2.98 (0.91 to 9.73) 3.21 (0.97 to 10.61)
YN 1.46 (0.62 to 3.44) 1.04 (0.38 to 2.83) 0.99 (0.36 to 2.70) 0.83 (0.20 to 3.44) 0.80 (0.20 to 3.28) 0.74 (0.18 to 3.04)
YY 4.03 (1.12 to 14.50) 2.46 (0.75 to 8.03) 2.24 (0.68 to 7.37) 2.68 (0.87 to 8.27) 1.74 (0.41 to 7.42) 1.50 (0.35 to 6.55)

NOTE. Use of vitamin C, vitamin A, vitamin E, carotenoids, or coenzyme Q10.

Abbreviations: DELCaP, Diet, Exercise, Lifestyle and Cancer Prognosis; DFS, disease-free survival; HR, hazard ratio; MV, multivitamin; NN, neither before
nor during; NY, not before but during; OS, overall survival; YN, before but not during; YY, both before and during.
*Stratified on treatment assignment and adjusted for age.
†Stratified on treatment assignment and adjusted for age, tumor characteristics (estrogen receptor, progesterone receptor, human epidermal growth factor
receptor 2, lymph node status, and tumor size), grades 3 or 4 hematologic toxicity, body mass index, physical activity, smoking, and alcohol consumption.
‡Stratified on treatment assignment and adjusted for age, tumor characteristics (estrogen receptor, progesterone receptor, human epidermal growth factor
receptor 2, lymph node status, and tumor size), grades 3 or 4 hematologic toxicity, body mass index, physical activity, smoking, alcohol consumption, and
MV use.

1.22 to 3.40), as was iron supplementation, with greater
recurrence with use during treatment (HR, 1.79; 95% CI,
1.20 to 2.67) and with use both before and during che-
motherapy (HR, 1.91; 95% CI, 0.98 to 3.70). There were
similar relationships with death (Fig 3D). Use of omega-3
fatty acids both before and during treatment was associated
with DFS (HR, 1.67; 95% CI, 1.12 to 2.49) but not OS. Other
supplements commonly taken were not associated with
poorer outcomes in adjusted models. The addition of pol-
ysupplement use to models was nonsignificant and not

808 © 2019 by American Society of Clinical Oncology Volume 38, Issue 8

Downloaded from ascopubs.org by 200.92.202.106 on August 19, 2022 from 200.092.202.106

Copyright © 2022 American Society of Clinical Oncology. All rights reserved.
 Supplement Use During Chemotherapy and Breast Cancer Survival

A B
1.0 1.0
Log-rank P = .1579 Log-rank P = .3323
DFS (probability)

OS (probability)
0.8 0.8

0.6 0.6

0.4 NN 0.4 NN
NY NY
0.2 0.2
YN YN
0.2 YY 0.2 YY

0 24 48 72 96 120 144 0 24 48 72 96 120 144

Time (months) Time (months)
No. at risk: No. at risk:
NN 739 721 693 657 626 583 451 388 316 232 107 22 0 NN 739 736 717 695 669 624 487 421 347 254 117 25 0
NY 75 72 70 68 63 62 45 40 32 25 10 1 0 NY 75 75 73 72 68 63 45 41 33 26 11 1 0
YN 207 200 190 183 178 162 133 118 93 58 30 4 0 YN 207 206 200 194 190 177 147 129 102 65 33 6 0
YY 126 125 113 108 103 89 75 57 43 30 14 4 0 YY 126 126 122 116 111 101 85 65 48 34 16 4 0

C D
1.0 1.0
Log-rank P = .6834 Log-rank P = .4326
DFS (probability)

OS (probablity)
0.8 0.8

0.6 0.6

0.4 0.4
NN NN
0.2 NY 0.2 NY
YN YN
0.2 YY
0.2 YY

0 24 48 72 96 120 144 0 24 48 72 96 120 144

Time (months) Time (months)
No. at risk: No. at risk:
NN 460 451 431 410 396 365 277 233 184 129 63 12 0 NN 460 458 447 434 421 388 297 250 200 142 67 13 0
NY 128 124 122 118 110 105 87 79 67 53 28 8 0 NY 128 128 123 122 119 113 93 85 73 58 32 8 0
YN 178 172 163 153 146 134 114 99 81 57 19 5 0 YN 178 176 172 166 162 147 126 108 88 63 22 7 0
YY 373 363 342 327 310 284 224 191 151 106 51 6 0 YY 373 373 362 347 328 308 246 212 168 116 56 8 0

FIG 2. Product limit estimates for use of antioxidant supplements and (A) disease-free survival (DFS) and (B) overall survival (B) and for use of
multivitamins and (C) DFS and (D) OS. NN, neither before nor during; NY, not before but during; YN, before but not during; YY, both before and during.

retained in any of the stepwise models (data not shown).
Analyses stratified by ER status are listed in Appendix Table
A3 (online only). The resultant small sample sizes for the
majority of supplements precluded us from drawing con-
clusions. However, for use of any antioxidant, while re-
lationships for DFS were similar by ER status, the association
with OS was driven largely by ER-positive breast cancer (HR,
1.70; 95% CI, 0.99 to 2.92), with no association for ER-
negative disease (HR, 0.93; 95% CI, 0.46 to 1.88).
DISCUSSION
In this observational study ancillary to an intergroup clinical
trial for high-risk breast cancer, we found some support for
the notion that use of dietary supplements during chemo-
therapy could have a negative impact on recurrence and OS.
When we considered use of any antioxidant (vitamins C, A,
and E; carotenoids; and coenzyme Q10), there was a 41%
increase in hazard of recurrence with use both before and
during treatment of borderline significance, with a similar but
weaker association with mortality. Although there were
suggestions of increased risk with individual antioxidants,
including vitamins A, E, and C and carotenoids and co-
enzyme Q, the numbers of users and events were small, and
risk estimates were unstable. Nonetheless, these findings of
increased risk of poor outcomes with use of antioxidant
supplements are congruent with concerns that use of an-
tioxidants during chemotherapy could reduce the cytotoxic
effects of ROS generated by numerous chemotherapy
agents4,8,14 and seem to support the recommendations by
some that antioxidant supplements not be consumed during
cancer therapy.2,5,15 However, a review by Ladas and Kelly16
in 2010 concluded that insufficient evidence existed with
regard to safety of dietary supplements to make recom-
mendations, and that may still be the case.
No associations were found with regard to use of multivi-
tamins either before or during chemotherapy, consistent
with the only other study of lifestyle factors embedded in
a clinical trial for colon cancer CALGB89803.17 However,

Journal of Clinical Oncology 809

Downloaded from ascopubs.org by 200.92.202.106 on August 19, 2022 from 200.092.202.106

Copyright © 2022 American Society of Clinical Oncology. All rights reserved.
 Ambrosone et al

TABLE 3. Adjusted Models for Associations Between Use of Dietary Supplements Other Than Antioxidants Before and During Chemotherapy and DFS and
OS in S0221: DELCaP Study Participants Who Completed Both Questionnaires 1 and 2 (n = 1,134)
DFS, HR (95% CI) OS, HR (95% CI)
Reported Supplement Use
Before and During Fully Adjusted With Fully Adjusted With
Chemotherapy Minimally Adjusted* Fully Adjusted† MV Use‡ Minimally Adjusted* Fully Adjusted† MV Use‡
MVs
NN 1.0 1.0 — 1.0 1.0 —
NY 1.02 (0.67 to 1.55) 1.02 (0.67 to 1.56) — 0.92 (0.55 to 1.56) 0.91 (0.54 to 1.55) —
YN 1.20 (0.98 to 1.71) 1.27 (0.88 to 1.84) — 1.20 (0.78 to 1.83) 1.35 (0.87 to 2.09) —
YY 1.11 (0.83 to 1.49) 1.21 (0.90 to 1.64) — 1.20 (0.85 to 1.68) 1.31 (0.92 to 1.88) —
Vitamin D
NN 1.0 1.0 1.0 1.0 1.0 1.0
NY 1.25 (0.86 to 1.81) 1.20 (0.83 to 1.74) 1.19 (0.81 to 1.74) 1.13 (0.73 to 1.75) 1.06 (0.68 to 1.66) 1.05 (0.66 to 1.65)
YN 1.01 (0.67 to 1.51) 1.02 (0.66 to 1.56) 0.96 (0.62 to 1.48) 1.07 (0.67 to 1.71) 1.16 (0.71 to 1.90) 1.07 (0.65 to 1.77)
YY 1.17 (0.81 to 1.70) 1.25 (0.84 to 1.86) 1.22 (0.81 to 1.84) 1.10 (0.71 to 1.70) 1.18 (0.73 to 1.91) 1.11 (0.67 to 1.82)
Vitamin B6
NN 1.0 1.0 1.0 1.0 1.0 1.0
NY 0.86 (0.61 to 1.23) 0.90 (0.64 to 1.28) 0.89 (0.63 to 1.27) 0.92 (0.61 to 1.39) 0.98 (0.64 to 1.48) 0.97 (0.64 to 1.47)
YN 0.79 (0.43 to 1.47) 0.70 (0.38 to 1.29) 0.65 (0.35 to 1.22) 0.96 (0.48 to 1.90) 0.96 (0.43 to 1.73) 0.79 (0.39 to 1.60)
YY 1.09 (0.61 to 1.97) 1.11 (0.61 to 2.03) 1.07 (0.58 to 1.96) 1.13 (0.57 to 2.24) 1.20 (0.60 to 2.41) 1.13 (0.56 to 2.29)
Vitamin B12
NN 1.0 1.0 1.0 1.0 1.0 1.0
NY 1.08 (0.66 to 1.75) 1.08 (0.66 to 1.76) 1.08 (0.66 to 1.77) 0.81 (0.42 to 1.58) 0.85 (0.44 to 1.63) 0.85 (0.44 to 1.64)
YN 0.91 (0.55 to 1.51) 0.86 (0.51 to 1.44) 0.80 (0.47 to 1.36) 0.80 (0.42 to 1.52) 0.76 (0.39 to 1.47) 0.70 (0.36 to 1.36)
YY 1.87 (1.20 to 2.98) 1.83 (1.15 to 2.92) 1.77 (1.10 to 2.84) 2.07 (1.37 to 3.38) 2.04 (1.22 to 3.40) 1.91 (1.13 to 3.22)
Iron
NN 1.0 1.0 1.0 1.0 1.0 1.0
NY 1.63 (1.09 to 2.44) 1.79 (1.20 to 2.67) 1.79 (1.18 to 2.70) 1.62 (1.02 to 2.58) 1.71 (1.06 to 2.76) 1.67 (1.02 to 2.72)
YN 0.66 (0.34 to 1.31) 0.60 (0.29 to 1.24) 0.58 (0.28 to 1.19) 0.59 (0.25 to 1.41) 0.54 (0.22 to 1.35) 0.50 (0.20 to 1.26)
YY 2.54 (1.36 to 4.78) 1.91 (0.98 to 3.70) 1.88 (0.96 to 3.67) 2.67 (1.36 to 5.27) 1.90 (0.90 to 4.02) 1.80 (0.85 to 3.84)
Folic acid
NN 1.0 1.0 1.0 1.0 1.0 1.0
NY 1.21 (0.74 to 1.99) 1.22 (0.73 to 2.04) 1.21 (0.72 to 2.04) 1.05 (0.55 to 2.02) 1.11 (0.58 to 2.14) 1.11 (0.58 to 2.16)
YN 0.82 (0.49 to 1.36) 0.76 to 0.45 to 1.28) 0.72 (0.42 to 1.21) 0.75 (0.40 to 1.41) 0.69 (0.40 to 1.33) 0.63 (0.32 to 1.22)
YY 1.32 (0.69 to 2.52) 1.35 (0.71 to 2.60) 1.32 (0.68 to 2.54) 1.54 (0.78 to 3.06) 1.79 (0.89 to 3.59) 1.70 (0.84 to 3.43)
Calcium
NN 1.0 1.0 1.0 1.0 1.0 1.0
NY 1.23 (0.81 to 1.87) 1.18 (0.77 to 1.81) 1.17 (0.76 to 1.80) 1.05 (0.62 to 1.77) 0.99 (0.58 to 1.70) 0.96 (0.55 to 1.66)
YN 1.28 (0.91 to 1.79) 1.31 (0.93 to 1.85) 1.24 (0.87 to 1.78) 1.44 (0.98 to 2.11) 1.60 (1.08 to 2.36) 1.49 (0.99 to 2.24)
YY 1.22 (0.88 to 1.70) 1.26 (0.89 to 1.76) 1.20 (0.84 to 1.74) 1.19 (0.81 to 1.75) 1.31 (0.87 to 1.96) 1.19 (0.77 to 1.84)
Omega-3 fatty acids
NN 1.0 1.0 1.0 1.0 1.0 1.0
NY 1.23 (0.66 to 2.26) 1.15 (0.62 to 2.13) 1.14 (0.61 to 2.12) 1.07 (0.50 to 2.29) 1.03 (0.48 to 2.23) 1.00 (0.46 to 2.18)
YN 0.77 (0.51 to 1.16) 0.82 (0.54 to 1.25) 0.77 (0.50 to 1.18) 0.74 (0.45 to 1.20) 0.88 (0.54 to 1.45) 0.81 (0.49 to 1.34)
YY 1.54 (1.04 to 2.29) 1.67 (1.12 to 2.49) 1.63 (1.07 to 2.48) 1.23 (0.75 to 2.04) 1.35 (0.81 to 2.25) 1.23 (0.72 to 2.10)
Glucosamine
NN 1.0 1.0 1.0 1.0 1.0 1.0
NY 0.21 (0.03 to 1.49) 0.20 (0.03 to 1.40) 0.20 (0.03 to 1.44) — — —
YN 0.83 (0.51 to 1.35) 0.86 (0.52 to 1.40) 0.82 (0.50 to 1.35) 0.80 (0.45 to 1.43) 0.84 (0.47 to 1.52) 0.79 (0.44 to 1.44)
YY 1.40 (0.84 to 2.32) 1.39 (0.84 to 2.31) 1.32 (0.78 to 2.22) 1.61 (0.96 to 2.70) 1.52 (0.87 to 2.65) 1.41 (0.80 to 2.50)
(continued on following page)

810 © 2019 by American Society of Clinical Oncology Volume 38, Issue 8

Downloaded from ascopubs.org by 200.92.202.106 on August 19, 2022 from 200.092.202.106

Copyright © 2022 American Society of Clinical Oncology. All rights reserved.
 Supplement Use During Chemotherapy and Breast Cancer Survival

TABLE 3. Adjusted Models for Associations Between Use of Dietary Supplements Other Than Antioxidants Before and During Chemotherapy and DFS and
OS in S0221: DELCaP Study Participants Who Completed Both Questionnaires 1 and 2 (n = 1,134) (continued)
DFS, HR (95% CI) OS, HR (95% CI)
Reported Supplement Use
Before and During Fully Adjusted With Fully Adjusted With
Chemotherapy Minimally Adjusted* Fully Adjusted† MV Use‡ Minimally Adjusted* Fully Adjusted† MV Use‡
Melatonin
NN 1.0 1.0 1.0 1.0 1.0 1.0
NY 0.69 (0.16 to 2.92) 0.68 (0.17 to 2.76) 0.70 (0.17 to 2.83) 1.05 (0.25 to 4.50) 1.12 (0.27 to 4.57) 1.16 (0.28 to 4.75)
YN 0.69 (0.21 to 2.24) 0.79 (0.25 to 2.50) 0.78 (0.25 to 2.47) 0.97 (0.30 to 3.08) 1.01 (0.31 to 3.24) 1.00 (0.31 to 3.21)
YY 1.51 (0.47 to 4.92) 0.98 (0.31 to 3.18) 0.98 (0.31 to 3.16) 1.38 (0.36 to 5.36) 0.87 (0.21 to 3.65) 0.85 (0.20 to 3.54)
Acidophilus
NN 1.0 1.0 1.0 1.0 1.0 1.0
NY 1.36 (0.63 to 2.96) 1.29 (0.60 to 2.78) 1.27 (0.59 to 2.74) 1.73 (0.74 to 4.05) 1.70 (0.74 to 3.91) 1.67 (0.73 to 3.84)
YN 1.34 (0.57 to 3.10) 1.34 (0.59 to 3.06) 1.33 (0.58 to 3.02) 1.19 (0.42 to 3.37) 1.15 (0.42 to 3.17) 1.11 (0.40 to 3.04)
YY 1.90 (0.76 to 4.71) 1.43 (0.58 to 3.51) 1.44 (0.58 to 3.55) 1.93 (0.74 to 5.02) 1.52 (0.55 to 4.20) 1.54 (0.56 to 4.24)

NOTE. Use of vitamin C, vitamin A, vitamin E, carotenoids, and coenzyme Q10.

Abbreviations: DELCaP, Diet, Exercise, Lifestyle and Cancer Prognosis; DFS, disease-free survival; HR, hazard ratio; MV, multivitamin; NN, neither before
nor during; NY, not before but during; OS, overall survival; YN, before but not during; YY, both before and during.
*Stratified on treatment assignment and adjusted for age.
†Stratified on treatment assignment and adjusted for age, tumor characteristics (estrogen receptor, progesterone receptor, human epidermal growth factor
receptor 2, lymph node status, and tumor size), grades 3 or 4 hematologic toxicity, body mass index, physical activity, smoking, and alcohol consumption.
‡Stratified on treatment assignment and adjusted for age, tumor characteristics (estrogen receptor, progesterone receptor, human epidermal growth factor
receptor 2, lymph node status and tumor size), grades 3 or 4 hematologic toxicity, body mass index, physical activity, smoking, alcohol consumption, and
multivitamin use.

use of omega-3 fatty acid supplements, iron, and vitamin
B12 both before and during chemotherapy was associated
with DFS and OS. In DELCaP, omega-3 fatty acid intake was
derived from the question, “Did you take fish oil, EPA
[eicosapentaenoic acid], omega-3, flaxseed, or cod liver
oil?” Thus, it is unclear what supplement was most com-
monly used and how that may drive associations with in-
creased risk. In CALGB89803, omega-3 polyunsaturated
fatty acid was associated with a decreased risk of re-
currence, but levels were derived from fish intake18 and not
from supplements.
Increased risk with vitamin B12 and iron use both before
and during chemotherapy, as well as any antioxidant, infers
that habitual use over time might be associated with greater
recurrence and mortality. Habitual use could be due to
indication, with patients using these supplements to treat
an existing condition such as anemia, which could, of itself,
be related to breast cancer recurrence and death.19
However, the addition of hematologic toxicities to models
did not diminish observed associations. On the other hand,
relationships between iron and poorer outcomes were also
observed for those who used iron supplements only during
treatment as well as during both time periods. Iron sup-
plementation may independently play a role in poorer DFS
and OS because iron plays unique roles in tumor initiation
and progression directly and through effects on the tumor
microenvironment.20-22 Iron enables the production of ROS,
which can contribute to malignant transformation, and
once established, tumors require high amounts of iron for
proliferation. Iron may also impair antitumor immunity.
Thus, iron supplementation could be associated with DFS
through these mechanisms.
How the use of vitamin B12 both before and during che-
motherapy could be associated with poorer outcomes remains
to be understood. Studies that examined circulating levels of
vitamin B12 (cobalamin) in relation to cancer and cancer
outcomes had mixed results.23 In a cohort study of . 25,000
patients with measured cobalamin levels before diagnosis,
elevated levels were associated with higher mortality,23 but it is
unknown whether the higher levels were indicators of un-
derlying pathologic conditions. Earlier randomized controlled
trials of B vitamins or placebo in patients with ischemic heart
disease found higher cancer incidence and all-cause mortality
in patients who were randomly assigned to folic acid plus
vitamin B12 but not with vitamin B6.24
Unlike the majority of observational studies that evaluated
supplements and cancer outcomes,8 the strength of
DELCaP is that it was conducted in the context of a ther-
apeutic clinical trial, with surveys before beginning che-
motherapy and at completion of treatment. This allowed us
to directly assess the potential interaction of dietary sup-
plements with chemotherapy. Furthermore, all patients in
the trial received the same drugs, a common chemotherapy
regimen, but with varying dosing schedules. This relative
homogeneity in treatments allowed for more direct in-
ferences about supplement use in relation to treatment
outcomes, unlike population-based studies where treat-
ments may be extremely heterogeneous.

Journal of Clinical Oncology 811

Downloaded from ascopubs.org by 200.92.202.106 on August 19, 2022 from 200.092.202.106

Copyright © 2022 American Society of Clinical Oncology. All rights reserved.
 Ambrosone et al

A B
1.0 1.0 Log-rank P = .0023
Log-rank P = .0196
DFS (probability)

OS (probability)
0.8 0.8

0.6 0.6

0.4 0.4
NN NN
0.2 NY 0.2 NY
YN YN
0.2 YY 0.2 YY

0 24 48 72 96 120 144 0 24 48 72 96 120 144

Time (months) Time (months)
No. at risk: No. at risk:
NN 918 897 856 815 778 724 571 497 403 296 141 29 0 NN 918 914 891 866 835 776 618 535 438 323 153 33 0
NY 78 74 72 70 68 63 54 41 33 21 8 1 0 NY 78 78 75 72 70 66 58 46 38 24 11 2 0
YN 86 83 83 79 75 65 48 40 28 16 7 0 YN 86 86 85 81 80 72 54 46 31 18 7 0
YY 60 59 50 47 44 39 30 24 20 12 5 1 0 YY 60 60 56 53 48 46 33 28 23 14 6 1 0

C D
1.0 1.0 Log-rank P = .0044
Log-rank P = .0012
DFS (probability)

OS (probability)
0.8 0.8

0.6 0.6

0.4 NN
0.4 NN
0.2 NY 0.2 NY
YN YN
0.2 YY 0.2 YY

0 24 48 72 96 120 144 0 24 48 72 96 120 144

Time (months) Time (months)
No. at risk: No. at risk:
NN 979 957 912 867 831 766 603 520 418 305 138 25 0 NN 979 976 950 917 886 822 650 561 454 331 150 29 0
NY 88 84 80 76 69 64 49 39 31 20 9 2 0 NY 88 88 86 84 79 71 57 47 38 25 11 2 0
YN 53 52 52 52 51 49 41 34 27 17 12 3 0 YN 53 53 52 52 51 50 43 37 29 19 13 4 0
YY 22 21 18 17 15 14 11 10 8 3 2 1 0 YY 22 21 19 19 17 17 14 11 9 4 3 1 0

FIG 3. Product limit estimates for use of vitamin B12 supplements and (A) disease-free survival (DFS) and (B) overall survival (OS) and for use of iron
supplements and (C) DFS and (D) OS. NN, neither before nor during; NY, not before but during; YN, before but not during; YY, both before and during.

Despite the strengths of the study design, we were still
hampered by the limits of the data, and as in many ob-
servational studies, there was the potential for a number of
biases, including selection, recall, and confounding biases.
Although there are reports in the literature of up to 60% of
patients taking antioxidant supplements during adjuvant
treatment,1,8,10 this was not the case in S0221. The fairly
small numbers of patients taking antioxidants and other
supplements thus reduced our statistical power to be able
to identify moderate or weak associations or potentially
resulted in spurious associations. Although it is possible
that patients under-reported use, we took every effort to
obtain accurate information by asking patients to check
supplement bottles for details. Nonetheless, under-
reporting may have been a possibility, particularly for the
small number of patients for whom the questionnaire was
administered by the clinical research associate. Use may
also have been low because patients were adhering to
recommendations from their clinicians not to take sup-
plements during treatment, as previously reported,8 and
were perhaps more compliant because of their enrollment
in a clinical trial for high-risk breast cancer. Furthermore,
although we collected detailed data about supplement use,
the relatively low numbers of users precluded our evalu-
ation of higher dosages of antioxidants. However, the
majority of users also took a multivitamin, so it might be
assumed that use of additional single supplements could
represent somewhat higher doses than that usually found
in a multivitamin, and we further controlled for multivitamin
use in fully adjusted models.
For the small proportion of patients who did not return
questionnaires within the desired time frame, responses
could be affected by recall bias. This may be relevant for Q1
with regard to supplement use in the past, but we expect
that late responses about use during treatment may be less
subject to recall bias because of the relative recency of
treatment.
Patients who took supplements during treatment could
have differed from those who did not, which would result in

812 © 2019 by American Society of Clinical Oncology Volume 38, Issue 8

Downloaded from ascopubs.org by 200.92.202.106 on August 19, 2022 from 200.092.202.106

Copyright © 2022 American Society of Clinical Oncology. All rights reserved.
 Supplement Use During Chemotherapy and Breast Cancer Survival

uncontrolled confounding. Indeed, patients who took any
antioxidant, as well as vitamin B12 and iron, tended to be
older than those who did not. However, adjustment for age,
additional lifestyle factors, and tumor characteristics only
somewhat attenuated risk estimates and widened the CIs.
An additional concern could be that patients may be taking
a number of supplements, which could be the true drivers
of associations with antioxidants. Adjustment for use of
multivitamins did not greatly affect associations, however,
and in sensitivity analyses where the small number of
patients taking more than five supplements were removed
(n = 129), associations remained the same.
The constraints of the intergroup trial design also limited
availability of data. A possibility exists that patients who
chose to take supplements were less likely to adhere to
adjuvant hormonal therapy, which could be associated
with poorer survival outcomes, but information on ad-
herence was not available for this trial. We also lack data
on possible adjuvant radiation therapy. In a randomized
trial of a-tocopherol and b-carotene during radiation
treatment of patients with head and neck cancer, the
intervention was associated with higher mortality.25 It is
possible that relationships with the use of antioxidants
during chemotherapy could be primarily among patients
who also received adjuvant radiation subsequent to this
trial. Finally, although a landmark analysis approach
appropriately separates predictor (supplement use) from
survival outcomes, it does not rule out the possibility of
selection bias between groups.
In summary, our findings from this prospective study in
the context of a cooperative group clinical trial indicate
that use of antioxidant supplements during chemother-
apy, as well as iron and vitamin B12, may increase the
risk of breast cancer recurrence and mortality. Short of
a randomized trial of supplements in patients with cancer,
the findings provide some empirical data for consideration
when discussing with patients the use of dietary sup-
plements during chemotherapy.

AFFILIATIONS CA180863, CA180858, CA180828, CA180801, CA68183, CA04919,

1
Roswell Park Comprehensive Cancer Center, Buffalo, NY CA13612, CA46282; and, in part, by Amgen.
2
Boston University, Boston, MA
3
Fred Hutchinson Cancer Research Center, Seattle, WA AUTHORS’ DISCLOSURES OF POTENTIAL CONFLICTS OF
4
The Ohio State University, Columbus, OH INTEREST AND DATA AVAILABILITY STATEMENT
5
Columbia University, New York, NY
6
Disclosures provided by the authors and data availability statement (if
Cleveland Clinic, Cleveland, OH
7
applicable) are available with this article at DOI https://doi.org/10.1200/
Baystate Medical Center, Springfield, MA
8
JCO.19.01203.
Georgetown University, Washington, DC
9
Mayo Clinic, Rochester, MN
10
Allan Blair Cancer Centre, Regina, Saskatchewan, Canada AUTHOR CONTRIBUTIONS
11
The University of Texas MD Anderson Cancer Center, Houston, TX Conception and design: Christine B. Ambrosone, Gary R. Zirpoli, Susan E.
12
Seattle Cancer Care Alliance, Seattle, WA McCann, Dawn L. Hershman, Joseph M. Unger, James A. Stewart, George
13
Loyola University Chicago Stritch School of Medicine, Chicago, IL T. Budd, Kathy S. Albain
Financial support: Christine B. Ambrosone
Administrative support: Christine B. Ambrosone, Gabriel N. Hortobagyi
CORRESPONDING AUTHOR
Provision of study material or patients: Christine B. Ambrosone, Halle C.F.
Christine B. Ambrosone, PhD, Roswell Park Comprehensive Cancer
Moore, James A. Stewart, Gabriel N. Hortobagyi, George T. Budd, Kathy
Center, Elm and Carlton Sts, Buffalo, NY 14263; e-mail:
S. Albain
christine.ambrosone@roswellpark.org.
Collection and assembly of data: Christine B. Ambrosone, Gary R. Zirpoli,
William E. McCann, William E. Barlow, Timothy J. Hobday, Muhammad
EQUAL CONTRIBUTION Salim, Gabriel N. Hortobagyi, George T. Budd
G.T.B. and K.S.A. are joint senior authors. Data analysis and interpretation: Christine B. Ambrosone, Gary R. Zirpoli,
Alan D. Hutson, William E. McCann, Susan E. McCann, William E.
Barlow, Kara M. Kelly, Rikki Cannioto, Lara E. Sucheston-Campbell,
SUPPORT
Dawn L. Hershman, Joseph M. Unger, Halle C.F. Moore, Claudine Isaacs,
Supported by grants R01 CA116395 (C.B.A.) and R01 CA139426
Timothy J. Hobday, Gabriel N. Hortobagyi, Julie R. Gralow, George T.
(C.B.A.), the Breast Cancer Research Foundation (C.B.A.), and Roswell
Budd, Kathy S. Albain
Park Comprehensive Cancer Center Cancer Center Support Grant P30
Manuscript writing: All authors
CA016056. S0221 was supported, in part, by National Cancer Institute
Final approval of manuscript: All authors
(NCI), Division of Cancer Prevention, SWOG NCI Community Oncology
Accountable for all aspects of the work: All authors
Research Program Research Base Grant No. 5UG1 CA189974-02; by
NCI, National Clinical Trials Network, Grant No. CA180888, CA180819,

REFERENCES
1. Velicer CM, Ulrich CM: Vitamin and mineral supplement use among US adults after cancer diagnosis: A systematic review. J Clin Oncol 26:665-673, 2008
2. Cassileth BR, Vickers AJ: High prevalence of complementary and alternative medicine use among cancer patients: Implications for research and clinical care.
J Clin Oncol 23:2590-2592, 2005
3. Harvie M: Nutritional supplements and cancer: Potential benefits and proven harms. Am Soc Clin Oncol Educ Book 34:e478-e486, 2014

Journal of Clinical Oncology 813

Downloaded from ascopubs.org by 200.92.202.106 on August 19, 2022 from 200.092.202.106

Copyright © 2022 American Society of Clinical Oncology. All rights reserved.
 Ambrosone et al

4. Lawenda BD, Kelly KM, Ladas EJ, et al: Should supplemental antioxidant administration be avoided during chemotherapy and radiation therapy? J Natl Cancer
Inst 100:773-783, 2008
5. Norman HA, Butrum RR, Feldman E, et al: The role of dietary supplements during cancer therapy. J Nutr 133:3794S-3799S, 2003
6. Rock CL, Doyle C, Demark-Wahnefried W, et al: Nutrition and physical activity guidelines for cancer survivors. CA Cancer J Clin 62:242-274, 2012
7. World Cancer Research Fund and American Institute for Cancer Research: Cancer survivors: Evidence on survivors of breast and other cancers. https://
www.wcrf.org/dietandcancer/cancer-survivors
8. Zirpoli GR, Brennan PM, Hong CC, et al: Supplement use during an intergroup clinical trial for breast cancer (S0221). Breast Cancer Res Treat 137:903-913,
2013
9. Budd GT, Barlow WE, Moore HC, et al: SWOG S0221: A phase III trial comparing chemotherapy schedules in high-risk early-stage breast cancer. J Clin Oncol
33:58-64, 2015
10. Zirpoli GR, McCann SE, Sucheston-Campbell LE, et al: Supplement use and chemotherapy-induced peripheral neuropathy in a cooperative group trial (S0221):
The DELCaP study. J Natl Cancer Inst 109:djx098, 2017
11. Satia-Abouta J, Patterson RE, King IB, et al: Reliability and validity of self-report of vitamin and mineral supplement use in the vitamins and lifestyle study. Am
J Epidemiol 157:944-954, 2003
12. Luc L, Baumgart C, Weiss E, et al: Dietary supplement use among participants of a databank and biorepository at a comprehensive cancer centre. Public Health
Nutr 18:916-926, 2015
13. Anderson JR, Cain KC, Gelber RD: Analysis of survival by tumor response. J Clin Oncol 1:710-719, 1983
14. Ambrosone C, Ahn J, Schoenenberger V: Antioxidant supplements, genetics and chemotherapy outcomes. Curr Cancer Ther Rev 1:251-258, 2005
15. American Institute for Cancer Research: Nutrition of the Cancer Patient. Washington, DC, American Institute for Cancer Research, 2000. https://www.aicr.org/
patients-survivors/healthy-or-harmful/supplements.html
16. Ladas E, Kelly KM: The antioxidant debate. Explore (NY) 6:75-85, 2010
17. Ng K, Meyerhardt JA, Chan JA, et al: Multivitamin use is not associated with cancer recurrence or survival in patients with stage III colon cancer: Findings from
CALGB 89803. J Clin Oncol 28:4354-4363, 2010
18. Van Blarigan EL, Fuchs CS, Niedzwiecki D, et al: Marine v-3 Polyunsaturated fatty acid and fish intake after colon cancer diagnosis and survival: CALGB 89803
(Alliance). Cancer Epidemiol Biomarkers Prev 27:438-445, 2018
19. Caro JJ, Salas M, Ward A, et al: Anemia as an independent prognostic factor for survival in patients with cancer: A systemic, quantitative review. Cancer 91:
2214-2221, 2001
20. Pfeifhofer-Obermair C, Tymoszuk P, Petzer V, et al: Iron in the tumor microenvironment-connecting the dots. Front Oncol 8:549, 2018
21. Ludwig H, Evstatiev R, Kornek G, et al: Iron metabolism and iron supplementation in cancer patients. Wien Klin Wochenschr 127:907-919, 2015 [Erratum:
Wien Klin Wochenschr 127:920-921, 2015]
22. Torti SV, Torti FM: Iron and cancer: More ore to be mined. Nat Rev Cancer 13:342-355, 2013
23. Arendt JF, Farkas DK, Pedersen L, et al: Elevated plasma vitamin B12 levels and cancer prognosis: A population-based cohort study. Cancer Epidemiol 40:
158-165, 2016
24. Ebbing M, Bønaa KH, Nygård O, et al: Cancer incidence and mortality after treatment with folic acid and vitamin B12. JAMA 302:2119-2126, 2009
25. Bairati I, Meyer F, Jobin E, et al: Antioxidant vitamins supplementation and mortality: A randomized trial in head and neck cancer patients. Int J Cancer 119:
2221-2224, 2006

n n n

814 © 2019 by American Society of Clinical Oncology Volume 38, Issue 8

Downloaded from ascopubs.org by 200.92.202.106 on August 19, 2022 from 200.092.202.106

Copyright © 2022 American Society of Clinical Oncology. All rights reserved.
 Supplement Use During Chemotherapy and Breast Cancer Survival

AUTHORS’ DISCLOSURES OF POTENTIAL CONFLICTS OF INTEREST

Dietary Supplement Use During Chemotherapy and Survival Outcomes of Patients With Breast Cancer Enrolled in a Cooperative Group Clinical Trial (SWOG
S0221)
The following represents disclosure information provided by authors of this manuscript. All relationships are considered compensated unless otherwise noted.
Relationships are self-held unless noted. I = Immediate Family Member, Inst = My Institution. Relationships may not relate to the subject matter of this manuscript.
For more information about ASCO’s conflict of interest policy, please refer to www.asco.org/rwc or ascopubs.org/journal/jco/site/ifc.
Open Payments is a public database containing information reported by companies about payments made to US-licensed physicians (Open Payments).

William E. Barlow Gabriel N. Hortobagyi

Research Funding: AbbVie (Inst), Merck (Inst), AstraZeneca (Inst) Consulting or Advisory Role: Novartis, Peregrine Pharmaceuticals, Agendia
Research Funding: Novartis (Inst)
Kara M. Kelly
Travel, Accommodations, Expenses: Novartis
Research Funding: Merck (Inst)
Travel, Accommodations, Expenses: Bristol-Myers Squibb Julie R. Gralow
Consulting or Advisory Role: Novartis, Genentech, Pfizer, Merck, Puma
Dawn L. Hershman
Biotechnology, Sandoz, AstraZeneca, Immunomedics, Genomic Health
Consulting or Advisory Role: AIM Specialty Health
George T. Budd
Halle C.F. Moore
Research Funding: Genentech (Inst), Roche (Inst), TRACON Pharma (Inst),
Research Funding: Puma Biotechnology (Inst), AbbVie (Inst)
Daiichi Sankyo (Inst), Eli Lilly (Inst), Macrogenics (Inst), Ambrx (Inst), Deciphera
Claudine Isaacs (Inst)
Honoraria: Genentech, Roche, AstraZeneca, Pfizer
Kathy S. Albain
Consulting or Advisory Role: Pfizer, Genentech, Roche, Novartis, AstraZeneca,
Consulting or Advisory Role: Novartis, Pfizer, Myriad Genetics, Genomic Health,
Medivation, NanoString Technologies, Syndax, Puma Biotechnology, Context
Agendia, Genentech, Roche
Therapeutics
Research Funding: Seattle Genetics (Inst)
Speakers’ Bureau: Genentech, Pfizer, AstraZeneca
Other Relationship: Puma Biotechnology
Research Funding: Novartis (Inst), Pfizer (Inst), Genentech (Inst), Tesaro (Inst)
Patents, Royalties, Other Intellectual Property: McGraw Hill Publishing, No other potential conflicts of interest were reported.
UpToDate, Wolters Kluwer, Elsevier
Timothy J. Hobday
Consulting or Advisory Role: AbbVie
Research Funding: Novartis (Inst)

Journal of Clinical Oncology

Downloaded from ascopubs.org by 200.92.202.106 on August 19, 2022 from 200.092.202.106

Copyright © 2022 American Society of Clinical Oncology. All rights reserved.
 Ambrosone et al

APPENDIX

TABLE A1. Supplement Use Before and During Breast Cancer TABLE A1. Supplement Use Before and During Breast Cancer
Treatment Among Patients Who Completed Questionnaires 1 and 2 in Treatment Among Patients Who Completed Questionnaires 1 and 2 in
the DELCaP Study (n = 1,134) the DELCaP Study (n = 1,134) (continued)
Supplement Use During Supplement Use During
Treatment, No. (%) Treatment, No. (%)
Supplement Use Supplement Use
Before Diagnosis No Yes Total, No. (%) Before Diagnosis No Yes Total, No. (%)
Vitamin C Yes 86 (7.6) 60 (5.3) 146 (12.9)
No 844 (74.4) 57 (5.0) 901 (79.5) Total 997 (87.9) 137 (12.1) 1,134 (100)
Yes 152 (13.4) 81 (7.1) 233 (20.5) Iron
Total 996 (87.8) 138 (12.2) 1,134 (100) No 973 (85.8) 86 (7.6) 1,059 (93.4)
Vitamin A Yes 53 (4.7) 22 (1.9) 75 (6.6)
No 1,077 (94.9) 21 (1.8) 1,098 (96.8) Total 1,028 (90.1) 109 (9.6) 1,134 (100)
Yes 31 (2.7) 5 (0.1) 36 (3.2) Folic acid
Total 1,108 (97.7) 26 (2.3) 1,134 (100) No 962 (84.8) 58 (5.1) 1,020 (90.0)
Vitamin E Yes 80 (7.1) 34 (2.9) 114 (10.0)
No 911 (80.3) 41 (3.6) 952 (84.0) Total 1,042 (91.9) 92 (8.1) 1,134 (100)
Yes 147 (12.9) 35 (3.1) 182 (16.0) Calcium
Total 1,058 (93.3) 76 (6.7) 1,134 (100) No 634 (55.9) 110 (9.7) 744 (65.6)
Coenzyme Q10 Yes 181 (15.9) 209 (18.4) 390 (34.4)
No 1,057 (93.4) 26 (2.3) 1,083 (95.7) Total 815 (71.9) 319 (28.1) 1,134 (100)
Yes 33 (2.9) 15 (1.3) 48 (4.3) Omega-3 fatty acids
Total 1,090 (96.4) 41 (3.6) 1,131 (100) No 831 (73.3) 46 (4.1) 877 (77.3)
Carotenoid Yes 160 (14.1) 97 (8.6) 257 (22.7)
No 1,106 (97.5) 6 (0.6) 1,112 (98.1) Total 991 (87.4) 143 (12.6) 1,134 (100)
Yes 17 (1.5) 5 (0.5) 22 (1.9) Glucosamine
Total 1,123 (99.0) 11 (1.0) 1,134 (100) No 967 (85.3) 18 (1.6) 985 (87.0)
Any antioxidant* Yes 88 (7.8) 60 (5.3) 148 (13.0)
No 736 (65.0) 74 (6.5) 810 (71.6) Total 1,055 (93.1) 78 (6.9) 1,133 (100)
Yes 206 (18.2) 126 (11.1) 332 (29.3) Melatonin
Total 932 (82.3) 200 (17.6) 1,132 (100) No 1,089 (96.3) 14 (1.2) 1,108 (97.4)
Multivitamins Yes 19 (1.7) 10 (0.9) 29 (2.6)
No 459 (40.4) 127 (11.2) 588 (51.6) Total 1,108 (97.9) 24 (2.1) 1,132 (100)
Yes 178 (15.7) 370 (32.6) 551 (48.4) Acidophilus
Total 637 (56.1) 497 (43.8) 1,134 (100) No 1,067 (94.4) 27 (2.4) 1,094 (96.8)
Vitamin D Yes 23 (2.0) 13 (1.2) 36 (3.2)
No 736 (64.9) 140 (12.3) 876 (77.2) Total 1,090 (96.4) 40 (3.5) 1,130 (100)
Yes 119 (10.5) 139 (12.3) 258 (22.7)
NOTE. Total numbers for each supplement may vary because of
Total 855 (75.4) 279 (24.6) 1,134 (100) missing data.
Vitamin B6 Abbreviation: DELCaP, Diet, Exercise, Lifestyle and Cancer
Prognosis.
No 846 (74.5) 178 (15.8) 1,022 (90.3)
*Use of vitamin C, vitamin A, vitamin E, carotenoids, and coenzyme
Yes 62 (5.5) 48 (4.2) 110 (9.7) Q10.
Total 908 (80.0) 226 (20.0) 1,134 (100)
Vitamin B12
No 911 (80.3) 77 (6.8) 988 (87.1)
(continued in next column)

© 2019 by American Society of Clinical Oncology Volume 38, Issue 8

Downloaded from ascopubs.org by 200.92.202.106 on August 19, 2022 from 200.092.202.106

Copyright © 2022 American Society of Clinical Oncology. All rights reserved.
 Supplement Use During Chemotherapy and Breast Cancer Survival

TABLE A2. Prevalence of Use of Dietary Supplements Before and During Chemotherapy for High-Risk Breast Cancer in S0221: DELCaP Study
Participants Who Completed Both Questionnaires 1 and 2 (n = 1,134)
Progression, No. (%) Death, No. (%)
Reported Supplement Use Before
and During Chemotherapy No Yes No Yes

Any antioxidant*
NN 580 (79) 156 (21) 629 (85) 110 (15)
NY 62 (84) 12 (16) 65 (87) 10 (13)
YN 160 (78) 46 (22) 172 (83) 35 (17)
YY 89 (71) 37 (29) 100 (79) 26 (21)
Vitamin C
NN 663 (79) 181 (21) 721 (85) 127 (15)
NY 45 (79) 12 (21) 49 (84) 9 (16)
YN 117 (77) 35 (23) 124 (82) 28 (18)
YY 58 (72) 23 (28) 64 (79) 17 (21)
Vitamin A
NN 844 (78) 234 (22) 914 (84) 169 (16)
NY 14 (67) 7 (33) 17 (81) 4 (19)
YN 25 (81) 6 (19) 27 (87) 4 (13)
YY 2 (40) 3 (60) 2 (40) 3 (60)
Vitamin E
NN 715 (78) 196 (22) 776 (85) 140 (15)
NY 31 (76) 10 (24) 34 (83) 7 (17)
YN 114 (78) 33 (22) 122 (83) 25 (17)
YY 23 (66) 12 (34) 26 (74) 9 (26)
Coenzyme Q10
NN 829 (78) 228 (21) 897 (85) 164 (15)
NY 20 (77) 6 (23) 22 (85) 4 (15)
YN 24 (73) 9 (27) 28 (82) 6 (18)
YY 9 (60) 6 (40) 10 (67) 5 (33)
Carotenoid
NN 868 (78) 240 (22) 938 (84) 174 (16)
NY 3 (50) 3 (50) 4 (57) 3 (43)
YN 12 (70) 5 (29) 15 (88) 2 (12)
YY 2 (40) 3 (60) 3 (60) 2 (40)
Multivitamins
NN 365 (80) 94 (20) 394 (86) 66 (14)
NY 99 (78) 28 (22) 110 (86) 18 (14)
YN 135 (76) 43 (24) 147 (83) 31 (17)
YY 284 (77) 86 (23) 306 (82) 67 (18)
Vitamin D
NN 582 (79) 155 (21) 628 (85) 113 (15)
NY 105 (75) 35 (25) 116 (83) 24 (17)
YN 93 (78) 26 (22) 99 (83) 20 (17)
YY 105 (76) 34 (24) 116 (83) 24 (17)
Calcium
NN 513 (58) 123 (19) 554 (87) 86 (13)
NY 84 (76) 27 (24) 94 (85) 17 (15)
YN 133 (73) 48 (27) 141 (78) 40 (22)
YY 157 (75) 52 (25) 172 (82) 38 (18)
Vitamin B6
(continued on following page)

Journal of Clinical Oncology

Downloaded from ascopubs.org by 200.92.202.106 on August 19, 2022 from 200.092.202.106

Copyright © 2022 American Society of Clinical Oncology. All rights reserved.
 Ambrosone et al

TABLE A2. Prevalence of Use of Dietary Supplements Before and During Chemotherapy for High-Risk Breast Cancer in S0221: DELCaP Study
Participants Who Completed Both Questionnaires 1 and 2 (n = 1,134) (continued)
Progression, No. (%) Death, No. (%)
Reported Supplement Use Before
and During Chemotherapy No Yes No Yes

NN 656 (78) 190 (22) 715 (84) 134 (16)

NY 141 (79) 37 (21) 152 (84) 28 (16)
YN 51 (82) 11 (18) 53 (85) 9 (15)
YY 36 (75) 12 (25) 39 (81) 9 (19)
Vitamin B12
NN 718 (79) 195 (21) 776 (85) 142 (15)
NY 60 (77) 18 (23) 68 (87) 10 (13)
YN 70 (81) 16 (19) 76 (88) 10 (12)
YY 38 (63) 22 (37) 41 (68) 19 (32)
Iron
NN 771 (79) 204 (21) 831 (85) 148 (15)
NY 60 (70) 27 (31) 68 (77) 20 (23)
YN 45 (85) 8 (15) 48 (91) 5 (9)
YY 12 (55) 10 (45) 14 (63) 8 (36)
Folic acid
NN 754 (78) 210 (22) 816 (84) 152 (16)
NY 43 (73) 16 (27) 49 (83) 10 (17)
YN 65 (81) 15 (19) 71 (88) 10 (12)
YY 24 (71) 10 (29) 25 (74) 9 (26)
Omega-3 fatty acids
NN 647 (78) 182 (22) 698 (84) 135 (16)
NY 34 (76) 11 (24) 38 (84) 7 (16)
YN 133 (83) 27 (17) 14 (88) 19 (12)
YY 67 (69) 30 (31) 79 (81) 19 (19)
Glucosamine
NN 753 (78) 214 (22) 820 (84) 152 (16)
NY 17 (94) 1 (6) 18 (100) 0
YN 70 (80) 18 (20) 75 (85) 13 (15)
YY 42 (70) 18 (30) 44 (73) 16 (27)
Melatonin
NN 845 (78) 244 (22) 919 (84) 175 (16)
NY 12 (86) 2 (14) 12 (86) 2 (14)
YN 16 (84) 3 (16) 16 (84) 3 (16)
YY 7 (70) 3 (30) 8 (80) 2 (20)
Acidophilus
NN 837 (78) 230 (22) 906 (78) 166 (15)
NY 20 (74) 7 (26) 21 (74) 6 (22)
YN 17 (74) 6 (26) 19 (83) 4 (17)
YY 8 (62) 5 (38) 9 (69) 4 (31)

NOTE. Total numbers for each supplement may vary because of missing data.
Abbreviations: DELCaP, Diet, Exercise, Lifestyle and Cancer Prognosis; NN, neither before nor during; NY, not before but during; YN, before
but not during; YY, both before and during.
*Use of vitamin C, vitamin A, vitamin E, carotenoids, or coenzyme Q10.

© 2019 by American Society of Clinical Oncology Volume 38, Issue 8

Downloaded from ascopubs.org by 200.92.202.106 on August 19, 2022 from 200.092.202.106

Copyright © 2022 American Society of Clinical Oncology. All rights reserved.
 Supplement Use During Chemotherapy and Breast Cancer Survival

TABLE A3. Associations Between Supplements Before and During Chemotherapy for High-Risk Breast Cancer and DFS and OS in S0221
Stratified by ER Status: DELCaP Study Participants Who Completed Both Questionnaires 1 and 2 (n = 1,134)
Progression,
No. (%) Death, No. (%)

Reported Supplement Use No Yes DFS, HR (95% CI) No Yes OS, HR (95% CI)
Antioxidants
Vitamin C
YY 58 (72) 23 (28) 1.37 (0.89 to 2.11) 64 (79) 17 (21) 1.41 (0.86 to 2.34)
ER positive 38 (76) 12 (24) 1.28 (0.70 to 2.34) 41 (82) 9 (18) 1.55 (0.78 to 3.10)
ER negative 20 (67) 10 (33) 1.25 (0.66 to 2.37) 23 (77) 7 (23) 1.04 (0.48 to 2.24)
Vitamin A
YY 2 (40) 3 (60) 4.05 (1.46 to 11.25) 2 (40) 3 (60) 4.14 (1.81 to 9.50)
ER positive 2 (40) 3 (60) 5.34 (1.76 to 16.17) 2 (67) 3 (33) 5.21 (2.20 to 12.34)
ER negative 0 (0) 0 (0) — 0 (0) 0 (0) —
Vitamin E
YY 23 (66) 12 (34) 1.62 (0.92 to 2.88) 26 (74) 9 (26) 1.65 (0.86 to 3.17)
ER positive 13 (62) 8 (38) 1.98 (0.96 to 4.08) 16 (76) 5 (24) 1.76 (0.74 to 4.15)
ER negative 10 (71) 4 (29) 1.05 (0.39 to 2.84) 10 (71) 4 (29) 1.36 (0.50 to 3.74)
Coenzyme Q10
YY 9 (60) 6 (40) 2.02 (0.92 to 4.47) 10 (67) 5 (33) 2.20 (0.92 to 5.26)
ER positive 6 (67) 3 (33) 1.76 (0.59 to 5.25) 7 (78) 2 (22) 1.60 (0.41 to 6.27)
ER negative 3 (50) 3 (50) 2.22 (0.71 to 6.97) 3 (50) 3 (50) 2.79 (0.90 to 8.64)
Carotenoid
YY 2 (40) 3 (60) 4.03 (1.12 to 14.50) 3 (60) 2 (40) 2.68 (0.87 to 8.27)
ER positive 0 (0) 0 (0) — 0 (0) 0 (0) —
ER negative 2 (40) 3 (60) 3.15 (1.07 to 9.32) 3 (60 2 (40) 1.88 (1.64 to 5.55)
Any antioxidant*
YY 89 (71) 37 (29) 1.42 (1.00 to 2.03) 100 (79) 26 (21) 1.37 (0.90 to 2.08)
ER positive 61 (73) 22 (27) 1.42 (0.89 to 2.24) 67 (81) 16 (19) 1.70 (0.99 to 2.92)
ER negative 28 (67) 14 (33) 1.32 (0.74 to 2.36) 33 (79) 9 (21) 0.93 (0.46 to 1.88)
Nonantioxidants
Multivitamins
YY 284 (77) 86 (23) 1.11 (0.83 to 1.48) 306 (82) 67 (18) 1.20 (0.85 to 1.68)
ER positive 204 (79) 55 (21) 1.17 (0.81 to 1.69) 218 (84) 41 (16) 1.40 (0.89 to 2.19)
ER negative 79 (71) 33 (29) 1.06 (0.65 to 1.73) 87 (78) 25 (22) 1.01 (0.59 to 1.73)
Vitamin D
YY 105 (76) 34 (24) 1.17 (0.81 to 1.70) 116 (83) 24 (17) 1.09 (0.71 to 1.70)
ER positive 71 (76) 22 (24) 1.38 (0.87 to 2.20) 78 (84) 15 (16) 1.50 (0.85 to 2.63)
ER negative 34 (74) 12 (26) 0.83 (0.44 to 1.57) 38 (83) 8 (17) 0.65 (0.31 to 1.36)
Vitamin B6
YY 36 (75) 12 (25) 1.09 (0.61 to 1.97) 39 (81) 9 (19) 1.13 (0.56 to 2.24)
ER positive 21 (70) 9 (30) 1.62 (0.81 to 3.24) 24 (80) 6 (20) 1.51 (0.64 to 3.54)
ER negative 15 (83) 3 (17) 0.54 (0.17 to 1.72) 15 (83) 3 (17) 0.75 (0.23 to 2.44)
Vitamin B12
YY 38 (63) 22 (37) 1.87 (1.20 to 2.98) 41 (68) 19 (32) 2.07 (1.37 to 3.38)
ER positive 25 (64) 14 (36) 2.15 (1.21 to 3.82) 27 (69) 12 (31) 2.53 (1.35 to 4.73)
ER negative 13 (65) 7 (35) 1.32 (0.58 to 2.99) 14 (70) 6 (30) 1.31 (0.60 to 3.04)
(continued on following page)

Journal of Clinical Oncology

Downloaded from ascopubs.org by 200.92.202.106 on August 19, 2022 from 200.092.202.106

Copyright © 2022 American Society of Clinical Oncology. All rights reserved.
 Ambrosone et al

TABLE A3. Associations Between Supplements Before and During Chemotherapy for High-Risk Breast Cancer and DFS and OS in S0221
Stratified by ER Status: DELCaP Study Participants Who Completed Both Questionnaires 1 and 2 (n = 1,134) (continued)
Progression,
No. (%) Death, No. (%)

Reported Supplement Use No Yes DFS, HR (95% CI) No Yes OS, HR (95% CI)
Iron
YY 12 (55) 10 (45) 2.54 (1.36 to 4.78) 14 (63) 8 (36) 2.67 (1.36 to 5.27)
ER positive 8 (50) 8 (50) 3.28 (1.60 to 6.70) 10 (63) 6 (37) 3.44 (1.62 to 7.32)
ER negative 4 (67) 2 (33) 1.52 (0.31 to 7.42) 4 (67) 2 (33) 1.90 (0.39 to 9.36)
Folic acid
YY 24 (71) 10 (29) 1.32 (0.69 to 2.52) 25 (74) 9 (26) 1.54 (0.78 to 3.086)
ER positive 17 (68) 8 (32) 1.77 (0.82 to 3.79) 18 (72) 7 (28) 2.13 (0.95 to 4.79)
ER negative 7 (78) 2 (22) 0.70 (0.18 to 2.70) 7 (78) 2 (22) 0.92 (0.23 to 3.79)
Calcium
YY 157 (75) 52 (25) 1.22 (0.88 to 1.70) 172 (82) 38 (18) 1.54 (0.78 to 3.06)
ER positive 108 (77) 32 (23) 1.12 (0.74 to 1.70) 117 (84) 23 (16) 1.30 (0.79 to 2.14)
ER negative 48 (71) 20 (29) 1.37 (0.78 to 2.41) 54 (79) 14 (21) 1.03 (0.55 to 1.93)
Omega-3 fatty acids
YY 67 (69) 30 (31) 1.54 (1.03 to 2.29) 79 (81) 19 (19) 1.23 (0.75 to 2.04)
ER positive 43 (73) 16 (27) 1.36 (0.82 to 2.35) 51 (86) 8 (14) 0.92 (0.44 to 1.96)
ER negative 24 (60) 14 (40) 1.60 (0.86 to 2.98) 28 (74) 10 (26) 1.37 (0.69 to 2.73)
Glucosamine
YY 42 (70) 18 (30) 1.40 (0.84 to 2.32) 44 (73) 16 (27) 1.61 (0.96 to 2.70)
ER positive 30 (70) 13 (30) 1.69 (0.91 to 3.14) 31 (72) 12 (28) 2.19 (1.16 to 4.10)
ER negative 12 (75) 4 (25) 0.79 (0.30 to 2.11) 13 (81) 3 (19) 0.70 (0.23 to 2.12)
Melatonin
YY 7 (70) 3 (30) 1.51 (0.47 to 4.92) 8 (80) 2 (20) 1.38 (0.36 to 5.36)
ER positive 4 (67) 2 (33) 1.79 (0.42 to 7.56) 4 (67) 2 (33) 2.77 (0.66 to 11.70)
ER negative 3 (75) 1 (25) 1.10 (0.12 to 9.95) 4 (100) 0 (0) 0.0
Acidophilus
YY 8 (62) 5 (38) 1.90 (0.76 to 4.71) 8 (62) 4 (31) 1.93 (0.74 to 5.02)
ER positive 8 (80) 2 (20) 0.93 (0.22 to 4.03) 8 (80) 2 (20) 1.40 (0.35 to 5.64)
ER negative 0 (0) 3 (100) 5.81 (3.91 to 8.63) 1 (33) 2 (77) 3.66 (1.12 to 11.97)

NOTE. Total numbers for each supplement may vary because of missing data.
Abbreviations: DELCaP, Diet, Exercise, Lifestyle and Cancer Prognosis; ER, estrogen receptor; HR, hazard ratio; YY, both before and during.
*Use of vitamin C, vitamin A, vitamin E, carotenoids, or coenzyme Q10.

© 2019 by American Society of Clinical Oncology Volume 38, Issue 8

Downloaded from ascopubs.org by 200.92.202.106 on August 19, 2022 from 200.092.202.106

Copyright © 2022 American Society of Clinical Oncology. All rights reserved.
Printed by Salvador Macias on 8/24/2022 5:23:20 AM. For personal use only. Not approved for distribution. Copyright © 2022 National Comprehensive Cancer Network, Inc., All Rights Reserved.

NCCN Guidelines Version 3.2022 NCCN Guidelines Index

Table of Contents
Non-Small Cell Lung Cancer Discussion

FINDINGS FOLLOW-UPc,d,g,h
<6 mm No routine follow-up
Consider CT
Low riske 6–8 mm CT at 6–12 mo Stable
at 18–24 mo
Consider CT at 3 mo,
>8 mm
PET/CT,i or biopsyj
Incidental
finding: solid
nodule(s) on
chest CT CT at 12 mo
<6 mm Stable No routine follow-up
(optional)
Repeat CT at
High riskf 6–8 mm CT at 6–12 mo Stable
18–24 mo

Consider CT at 3 mo,
>8 mm
PET/CT,i or biopsyj

c Principles of Diagnostic Evaluation (DIAG-A 1 of 3).
d The most important radiologic factor is change or stability
imaging study.
e Low risk = minimal or absent history of smoking or other known risk factors.
f High risk = history of smoking or other known risk factors. Known risk factors include
history of lung cancer in a first-degree relative; exposure to asbestos, radon, or
uranium.
g Non-solid (ground-glass) nodules may require longer follow-up to exclude indolent
adenocarcinoma.
h Adapted from Fleischner Society Guidelines: MacMahon H, Naidich DP, Goo
JM, et al. Guidelines for management of incidental pulmonary nodules detected
on CT images: From the Fleischner Society 2017. Radiology 2017;284:228-243.
©
Radiological Society of North America. Fleischner Society Guidelines do not direct
whether or not contrast is necessary or if an LDCT is appropriate. LDCT is preferred
unless there is a reason for contrast enhancement for better diagnostic resolution.
i PET/CT performed skull base to knees or whole body. A positive PET result is defined
compared with a previous as a standardized uptake value (SUV) in the lung nodule greater than the baseline
mediastinal blood pool. A positive PET scan finding can be caused by infection or
inflammation, including absence of lung cancer with localized infection, presence
of lung cancer with associated (eg, postobstructive) infection, and presence of lung
cancer with related inflammation (eg, nodal, parenchymal, pleural). A false-negative
PET scan can be caused by a small nodule, low cellular density (nonsolid nodule or
ground-glass opacity [GGO]), or low tumor avidity for FDG (eg, adenocarcinoma in
situ [previously known as bronchoalveolar carcinoma], carcinoid tumor).
j If empiric therapy is contemplated without tissue confirmation, multidisciplinary
evaluation that at least includes interventional radiology, thoracic surgery, and
interventional pulmonology is required to determine the safest and most efficient
approach for biopsy, or to provide consensus that a biopsy is too risky or difficult and
that the patient can proceed with therapy without tissue confirmation. (IJsseldijk MA,
et al. J Thorac Oncol 2019;14:583-595.)

Note: All recommendations are category 2A unless otherwise indicated.

Clinical Trials: NCCN believes that the best management of any patient with cancer is in a clinical trial. Participation in clinical trials is especially encouraged.

Version 3.2022, 03/16/22 © 2022 National Comprehensive Cancer Network® (NCCN®), All rights reserved. NCCN Guidelines® and this illustration may not be reproduced in any form without the express written permission of NCCN.
DIAG-2
Printed by Salvador Macias on 8/24/2022 5:23:20 AM. For personal use only. Not approved for distribution. Copyright © 2022 National Comprehensive Cancer Network, Inc., All Rights Reserved.

NCCN Guidelines Version 3.2022 NCCN Guidelines Index

Table of Contents
Non-Small Cell Lung Cancer Discussion

FINDINGS FOLLOW-UPc,d,g,h
Solitary pure <6 mm No routine follow-up
ground-glass CT at 6–12 mo to confirm no growth or
nodules ≥6 mm development of a solid component, then
CT every 2 y until 5 y
Incidental
finding: Solitary <6 mm No routine follow-up
subsolid part-solid • CT at 3–6 mo to confirm no growth or change in solid
nodule(s) nodules ≥6 mm component, then annual CT for 5 y
on chest CT • If solid component ≥6 mm, consider PET/CTi or biopsyj

• CT at 3–6 mo
<6 mm
If stable, consider CT at 2 and 4 y
Multiple
subsolid
• CT at 3–6 mo
nodules
• Subsequent management
≥6 mm
based on most
suspicious nodule(s)

c Principles of Diagnostic Evaluation (DIAG-A 1 of 3).
d The most important radiologic factor is change or stability
compared with a previous
imaging study.
g Non-solid (ground-glass) nodules may require longer follow-up to exclude indolent
adenocarcinoma.
h Adapted from Fleischner Society Guidelines: MacMahon H, Naidich DP, Goo
JM, et al. Guidelines for management of incidental pulmonary nodules detected
on CT images: From the Fleischner Society 2017. Radiology 2017;284:228-243.
©
Radiological Society of North America. Fleischner Society Guidelines do not direct
whether or not contrast is necessary or if an LDCT is appropriate. LDCT is preferred
unless there is a reason for contrast enhancement for better diagnostic resolution.
i PET/CT performed skull base to knees or whole body. A positive PET result is defined
as a SUV in the lung nodule greater than the baseline mediastinal blood pool. A
positive PET scan finding can be caused by infection or inflammation, including
absence of lung cancer with localized infection, presence of lung cancer with
associated (eg, postobstructive) infection, and presence of lung cancer with related
inflammation (eg, nodal, parenchymal, pleural). A false-negative PET scan can be
caused by a small nodule, low cellular density (nonsolid nodule or GGO), or low tumor
avidity for FDG (eg, adenocarcinoma in situ [previously known as bronchoalveolar
carcinoma], carcinoid tumor).
j If empiric therapy is contemplated without tissue confirmation, multidisciplinary
evaluation that at least includes interventional radiology, thoracic surgery, and
interventional pulmonology is required to determine the safest and most efficient
approach for biopsy, or to provide consensus that a biopsy is too risky or difficult and
that the patient can proceed with therapy without tissue confirmation. (IJsseldijk MA,
et al. J Thorac Oncol 2019;14:583-595.)

Note: All recommendations are category 2A unless otherwise indicated.

Clinical Trials: NCCN believes that the best management of any patient with cancer is in a clinical trial. Participation in clinical trials is especially encouraged.

Version 3.2022, 03/16/22 © 2022 National Comprehensive Cancer Network® (NCCN®), All rights reserved. NCCN Guidelines® and this illustration may not be reproduced in any form without the express written permission of NCCN.
DIAG-3
 17
CAPÍTULO 
Nódulo pulmonar solitario
E. Guzmán de Alba
El cáncer de pulmón se ha convertido en nuestro país, y
en todo el mundo, en uno de los más frecuentes y más
mortales, motivos por los que, al igual que en cualquier otra
neoplasia, el diagnóstico temprano y el tratamiento ade-
cuado y bien dirigido permiten alcanzar los mejores resul-
tados y el mejor pronóstico para el paciente. El nódulo pul-
monar solitario se define como un nódulo de 3 cm de
diámetro como máximo, completamente rodeado de pa-
rénquima pulmonar aparentemente sano y que no se aso-
cia a ganglios linfáticos aumentados de tamaño ni a derra-
me pleural1. La detección de un nódulo pulmonar solitario
en general se produce en pacientes asintomáticos a quie-
nes se les realizó un estudio radiológico de tórax por encon-
trarse dentro de un programa de detección temprana de
cáncer de pulmón o por alguna otra razón, como el segui-
miento por alguna neoplasia previa, el estudio preoperato-
rio previo a alguna cirugía o en el contexto de alguna otra
enfermedad pulmonar o cardíaca que requiere de estudios
radiológicos de tórax. La radiografía de tórax simple en ge-
neral nos da poca información sobre las características pro-
pias del nódulo, y el estudio de elección para determinar si
el nódulo tiene características de un posible cáncer pulmo-
nar es la tomografía computarizada (TC) de tórax2.
FACTORES DE RIESGO PARA CÁNCER
DE PULMÓN EN PACIENTE CON UN
NÓDULO PULMONAR
Parte fundamental del estudio de un nódulo pulmonar
solitario es determinar si el paciente cuenta con factores de
riesgo para ser portador de un nódulo pulmonar maligno.
En múltiples estudios se ha demostrado que los factores de
riesgo para que un nódulo pulmonar encontrado en estu-
dios radiológicos sea maligno son la edad del paciente, la
historia de malignidad previa en el paciente, la historia de
tabaquismo del paciente y el tiempo posterior al cese del
tabaquismo3-5. En la tabla 1 se pude observar cómo catego-
rizar a los pacientes con un nódulo pulmonar en alto, me-
dio o bajo riesgo por sus características clínicas6.
106
En estudios grandes se ha encontrado que la probabili-
dad de malignidad de un nódulo es muy baja, del 0 al 1.5%
en nódulos < 5 mm, riesgo que aumenta desde el 33 hasta
el 64% en nódulos de 11 a 20 mm de diámetro y que pue-
de ser tan elevada como del 82% en nódulos > 20 mm7. La
historia de tabaquismo es preponderante en el riesgo de
desarrollar un cáncer de pulmón, por lo que es uno de los
factores que se debe evaluar de manera más profunda en
el momento de determinar la posibilidad de que un nódu-
lo pulmonar sea de origen neoplásico8. La intensidad en el
consumo del tabaco (índice tabáquico) se puede medir
mediante la determinación del número de paquetes de
cigarrillos fumados al año por una persona utilizando la
siguiente fórmula:
n.º de cigarrillos
fumados al día
x n.º de años de fumador
= n.º de paquetes/año
20
De esta manera se puede determinar que una persona
que ha fumado una cajetilla de cigarrillos al día por 20 años
tiene un índice tabáquico de 20, lo que la coloca en un área
de alto riesgo para padecer cáncer de pulmón9.
CARACTERIZACIÓN RADIOLÓGICA
Como ya se había mencionado anteriormente, la radio-
grafía simple de tórax no es un estudio suficiente para eva-
luar las características radiológicas de un nódulo pulmonar;
sin embargo, es el estudio en el que generalmente se rea-
liza el hallazgo del nódulo (Fig. 1).
En el momento en el que se encuentra un nódulo pul-
monar, se debe hacer todo lo posible para obtener la his-
toria radiológica del paciente. En el caso de que éste cuen-
te con estudios radiológicos previos, se deberá realizar una
búsqueda intencionada del nódulo que estamos viendo en
el estudio actual: si la lesión se encuentra presente en es-
tudios de más de dos años previos al actual y sus
 Nódulo pulmonar solitario

Tabla 1. Características de riesgo del paciente para tener un nódulo pulmonar maligno
Característica Riesgo
Bajo Medio Alto

Tamaño del nódulo < 8 mm 8-20 mm > 20 mm

Edad < 45 años 45-60 años > 60 años
Antecedentes neoplásicos Sin antecedentes Con antecedentes
Tabaquismo Sin tabaquismo < 20 paquetes/año > 20 paquetes/año
Cese del tabaquismo > 7 años < 7 años Activo
EPOC No Sí
Características del nódulo Aspecto benigno Indeterminado Aspecto maligno
en la TC

relaciona directamente con las posibilidades de ser malig-

Figura 1. Radiografía simple de tórax con presencia de nódulo en región
parahiliar izquierda.
características se mantienen sin cambios, el paciente no
requerirá mayor estudio de ese nódulo; si, por el contrario,
la lesión actual no está presente en estudios previos o sus
características han cambiado, como el aumento de tamaño
o el que se asocie a retracción o engrosamiento pleural, se
deberá continuar con el estudio del paciente mediante la
obtención de una TC de tórax.
La TC de tórax es el estudio de elección para poder
determinar las características radiológicas del tumor y con
ello proponer alguna otra medida de diagnóstico radioló-
gico o histológico. Las características que normalmente se
estudian en un nódulo pulmonar en la TC son el tamaño,
los bordes, las características internas (densidad, atenua-
ción, calcificaciones, perfusión) y las características del te-
jido que lo circunda, como la presencia de retracción pleu-
ral, broncograma aéreo y alteración de la estructura normal
del parénquima pulmonar que lo rodea. Como se había
mencionado anteriormente, el tamaño del nódulo se
no: los nódulos < 5 mm son muy poco probables de ser
malignos, mientras que los nódulos > 11 mm aumentan
sus posibilidades de ser malignos hasta un 33% y los nódu-
los > 20 mm llegan hasta un 80% de malignidad10. En rela-
ción con los bordes del nódulo, generalmente los nódulos
con un contorno bien definido, de bordes lisos y bien re-
dondeados se asocian a mayor posibilidad de ser benignos,
mientras que los bordes irregulares, espiculados y mal de-
finidos se asocian con mayor frecuencia a malignidad11. La
presencia de las espiculaciones en los bordes de los nódu-
los se encuentra en relación con el crecimiento de células
neoplásicas a lo largo del intersticio pulmonar, mientras
que los bordes irregulares y mal definidos están asociados
al crecimiento irregular de diferentes zonas dentro de una
neoplasia maligna12, por esto son de gran ayuda para de-
terminar el riesgo de que un nódulo sea de origen maligno.
La presencia de bordes espiculados, llamados en ocasiones
imágenes de corona radiada, tiene un alto valor predictivo
de malignidad, que puede llegar a ser del 90%, motivo por
el que los nódulos que presentan esta característica en la
mayoría de las ocasiones deben ser estudiados a través de
métodos de citología o histología13 (Fig. 2).
La presencia de calcificaciones dentro del nódulo gene-
ralmente aumenta la posibilidad de benignidad. Las calci-
ficaciones que normalmente se asocian a benignidad son
las difusas, centrales, laminares y lobuladas o en «palomita
de maíz». Las calcificaciones difusas, centrales o laminares
en general se asocian con la presencia de un granuloma,
mientras que las calcificaciones lobuladas se asocian ma-
yormente a tumores benignos de tipo del hamartoma14
(Fig. 3).
Es importante resaltar que se puede observar algún
tipo de calcificación en carcinomas de pulmón; sin embar-
go, estas calcificaciones en general son puntiformes, irre-
gulares o espiculadas dentro del nódulo y no siguen el
mismo patrón de las calcificaciones benignas mencionadas
anteriormente15.

107
Oncología general para profesionales de la salud de primer contacto
A B
Figura 2. Característica de los bordes de un nódulo en la TC. A: nódulo pulmonar con características benignas, de aspecto redondeado, bordes lisos
y regulares que no alteran la anatomía normal del parénquima pulmonar. B: nódulo pulmonar con características de malignidad, con bordes irregulares,
espiculados y áreas de retracción pleural.
Figura 3. Nódulo pulmonar en el que se observan calcificaciones periféricas
sugestivas de granuloma.
La evolución de la tecnología en la TC ha permitido que
cada vez seamos más capaces de ver características finas
de los nódulos, lo que ha facilitado localizar nódulos que
no están completamente sólidos dentro del parénquima
pulmonar; a estos nódulos se les llama subsólidos y se ca-
racterizan ser nódulos con áreas denominadas de «vidrio
despulido». Estos nódulos en su mayoría se deben a proce-
sos infecciosos o inflamatorios; sin embargo, hasta el 50%
de los adenocarcinomas pulmonares inician con patrones
de nódulo subsólido, por lo que es importante mantener-
los bajo una vigilancia estrecha con el fin de detectar cam-
bios en sus características que puedan sugerir una neopla-
sia maligna, como aumento del tamaño, confluencia de las
áreas subsólidas que forman un nódulo sólido, retracción
de las estructuras normales del pulmón produciendo espi-
culaciones o retracción de la pleura, lo que en su caso ame-
ritaría el estudio histológico de ese pulmón con el fin de
descartar la presencia de un tumor maligno que se esté
desarrollando en ese sitio16 (Fig. 4).
108
Durante la realización de la TC para el estudio de un
nódulo pulmonar se puede solicitar la realización de un
estudio de perfusión del nódulo, durante el cual se realizan
mediciones de las UH del nódulo antes de la administra-
ción intravenosa de medio de contraste iónico y a los 2, 3
y 4 min. El aumento de la densidad medida en el nódulo
está relacionada con la vascularidad y riego sanguíneo del
mismo, el cual es mayor en lesiones malignas, por lo que
un aumento de 20 unidades Houdsfield (HD) o más en el
nódulo es altamente sugestivo de una neoplasia maligna17.
En caso de que hayamos decidido mantener en obser-
vación y dar seguimiento a un paciente con un nódulo
pulmonar solitario, uno de los factores determinantes para
decidir realizar alguna intervención en dicho paciente es el
crecimiento del nódulo. En la TC el crecimiento del nódulo
se mide calculando el tiempo que tarda en duplicar su
volumen. Recordemos que, a pesar de que en estudios ra-
diológicos vemos un nódulo como una circunferencia, en
realidad es una esfera; por tanto, duplicar el volumen de un
nódulo implica el aumento de alrededor del 30% de su
circunferencia, es decir, que para que un nódulo de 10 mm
duplique su volumen, en el siguiente estudio deberá medir
13 mm aproximadamente. La velocidad de duplicación se
mide en días: un nódulo maligno tiene un promedio de
velocidad de duplicación de 20 a 400 días, con una media
de 100 días, por lo que se recomienda realizar los estudios
de seguimiento con una diferencia de cuatro meses entre
cada uno para poder definir si el nódulo está creciendo. En
caso de que un nódulo sólido se mantenga sin cambios
durante un periodo de dos años, podemos afirmar que es
un nódulo benigno; mientras que un nódulo que aumenta
su tamaño en un 30% en cualquiera de nuestros estudios
de control deberá ser considerado un nódulo altamente
sugestivo de malignidad y se deberán tomar medidas para
su tratamiento18. Los nódulos subsólidos presentan un cre-
cimiento mucho más lento que los nódulos sólidos, por lo
 Nódulo pulmonar solitario

A B

Figura 4. Tomografía computarizada con nódulos subsólidos. A: se observa claramente la presencia de la imagen nodular de aspecto en «vidrio despulido».
B: nódulo subsólido de aspecto maligno que ya presenta un área sólida central (flecha).

A B

Figura 5. A: TC con nódulo pulmonar de aspecto maligno. B: mismo paciente en estudio de PET-CT con gran captación del radiotrazador, lo que sugiere
malignidad.

que el seguimiento se deberá mantener por al menos cinco
años, y en algunas publicaciones se refieren hasta siete
años19.
El uso de la tomografía por emisión de positrones-to-
mografía computarizada (PET-CT) en pacientes con nódulo
pulmonar solitario está reservado para aquellos pacientes
con una muy alta sospecha de malignidad y en los que se
desee descartar la presencia de enfermedad metastásica.
Como la PET-CT es un estudio caro, se deberá utilizar de
manera juiciosa y con indicaciones claras. El uso de la PET-
CT en nódulos < 10 mm tiene baja sensibilidad, por lo que
no se requiere su uso en estos casos. En casos de pacientes
con nódulos > 10 mm de diámetro en su parte sólida, se
dará como positivos a la posibilidad de malignidad cuando
la captación en el nódulo sea ≥ 2.4 Standard Uptake Value
(SUV). En el caso de que se cuente con un estudio de PET-
CT positivo en el nódulo pulmonar, se deberá efectuar una
corroboración citológica o histológica de malignidad del
mismo, ya que existen otras etiologías que pueden emular
la presencia de un nódulo maligno en una PET-CT, como
las enfermedades granulomatosas infecciosas o inflamato-
rias, como la tuberculosis pulmonar o las micosis pulmona-
res20,21 (Fig. 5).
DIAGNÓSTICO HISTOLÓGICO DEL
NÓDULO PULMONAR SOLITARIO
Una vez determinado que el nódulo pulmonar que pre-
senta el paciente es de alto riesgo para ser una neoplasia,
ya sea porque es un nódulo sugestivo de malignidad en
estudios radiológicos o porque el paciente es un paciente
de alto riesgo (fumador mayor de 60 años de edad) con un
nódulo indeterminado, el siguiente paso es obtener un
diagnóstico citológico o histológico de certeza de benigni-
dad o malignidad. Los métodos diagnósticos utilizados
para este fin son: punción transtorácica con guía radiológi-
ca o sin ella, broncoscopia o procedimientos quirúrgicos
como resección por toracoscopia o toracotomía. Lo ideal

109
Oncología general para profesionales de la salud de primer contacto
Figura 6.
siempre será que el estudio histológico de un nódulo pul-
monar se realice en un centro especializado en este tipo de
padecimientos.
Actualmente las recomendaciones para el uso de méto-
dos diagnósticos menos invasivos a la cirugía son para
aquellos nódulos con probabilidad baja-intermedia de ma-
lignidad, ya que en pacientes con nódulos con alta proba-
bilidad de ser malignos la recomendación es la resección
quirúrgica con estudio histológico transoperatorio22.
El uso de la punción transtorácica guiada por TC ha
demostrado una buena sensibilidad; sin embargo, siempre
se deberán tomar en cuenta el tamaño del nódulo, la proxi-
midad a la pared torácica, el calibre de la aguja utilizada y
la posibilidad de contar con un citopatólogo en el sitio
para que determine que la muestra es adecuada, pues
todo ello aumenta las posibilidades de tener un buen diag-
nóstico. Por ello, nódulos ≥ 10 mm y que se encuentran a
una distancia promedio de 30 mm de la pleura parietal al-
canzan un diagnóstico en un 66% de los casos. Sin embar-
go, este procedimiento no está libre de complicaciones, y
se llega a presentar un neumotórax en el 52% de los pa-
cientes, con necesidad de colocación de un tubo de drena-
je pleural en el 9% de éstos23.
El uso de la broncoscopia para nódulos pulmonares so-
litarios es cada vez más aceptado con el uso de técnicas
como el ultrasonido radial o la electronavegación; sin em-
bargo, aun así, el diagnóstico certero es muy variable, ya
que por definición un nódulo pulmonar solitario es un
110
nódulo < 30 mm, periférico y que en general se encuentra
lejos del alcance del broncoscopio tradicional. Mediante el
uso de técnicas convencionales de broncoscopia se ha en-
contrado una certeza diagnóstica tan baja como del 30%
en pacientes con nódulos < 2 cm, que aumenta al 60%
cuando el nódulo es > 2 cm, y más central aún en las ma-
nos más expertas en el uso de esta técnica, mientras que
el uso de fluoroscopia como guía para la toma de biopsias
no ha demostrado aumentar la tasa de éxito24.
TRATAMIENTO QUIRÚRGICO
DEL NÓDULO PULMONAR
El tratamiento quirúrgico del nódulo pulmonar es el
método de elección para manejar este tipo de problemas.
La resección quirúrgica de un nódulo pulmonar se encuen-
tra indicada en tres instancias25,26:
• El nódulo se diagnosticó como maligno, en cuyo caso
la resección quirúrgica es el tratamiento de elección.
• El nódulo es sugestivo de malignidad a pesar de tener
biopsias u otros estudios que no han demostrado ma-
lignidad en sus resultados.
• El nódulo es altamente sugestivo de malignidad en
un paciente de alto riesgo, en cuyo caso no se reco-
miendan otros estudios y es preferible realizar la re-
sección quirúrgica como primer paso de diagnóstico
y terapéutico.

La resección quirúrgica del nódulo permite tener un
diagnóstico preciso de la etiología del mismo y, al mismo
tiempo, realizar un procedimiento de tratamiento definiti-
vo en caso de que en el estudio transoperatorio el resulta-
do sea de un nódulo maligno. En caso de que el nódulo sea
secundario a un cáncer pulmonar, el procedimiento quirúr-
gico estándar para su tratamiento es la lobectomía con la
resección de los nodos ganglios linfáticos regionales, inclu-
yendo los ganglios mediastinales. Existen otros procedi-
mientos aceptados como la segmentectomía anatómica y
la resección en cuña acompañados de la resección de los
ganglios regionales y mediastinales; sin embargo, la deci-
sión de alguno de estos procedimientos depende de las
características del nódulo y del paciente, y deberá ser rea-
lizada exclusivamente por cirujanos altamente capacitados
en tratar pacientes con cáncer de pulmón27.
Actualmente existen dos modalidades de tratamiento
quirúrgico para un nódulo pulmonar: la cirugía de mínima
invasión o videotoracoscopia, la cual actualmente es el mé-
todo de elección, ya que representa una técnica con menor
molestia para el paciente, menor tiempo de recuperación
e inicio de sus actividades cotidianas, con baja morbimor-
talidad y con los mismos resultados desde el punto de
vista oncológicos, y la resección del nódulo mediante tora-
cotomía28 (Fig. 5).
En la figura 6 se muestra en flujograma para el buen
diagnóstico de un nódulo pulmonar solitario.
BIBLIOGRAFÍA
1. Ost DE, Gould MK. Decision making in patients with pulmonary nodules. Am J
Respir Crit Care Med. 2012;185(4):363-72.
2. Ost D, Fein AM, Feinsilver SH. The Solitary Pulmonary Nodule. N Engl J Med.
2003;348(25):2535-42.
3. Al-Ameri A, Malhotra P, Thygesen H, et al. Risk of malignancy in pulmonary
nodules: A validation study of four prediction models. Lung Cancer.
2015;89(1):27-30.
4. Swensen SJ, Silverstein MD, Ilstrup DM, Schleck CD, Edell ES. The probability of
malignancy in solitary pulmonary nodules. Application to small radiologically
indeterminate nodules. Arch Intern Med. 1997;157(8):849-55.
5. Gould MK, Ananth L, Barnett PG, Veterans Affairs SNAP Cooperative Study
Group. A clinical model to estimate the pretest probability of lung cancer in
patients with solitary pulmonary nodules. Chest. 2007;131(2):383-8.
6. Zhan P, Xie H, Xu C, Hao K, Hou Z, Song Y. Management strategy of solitary
pulmonary nodules. J Thorac Dis. 2013;5(6):824-9.
Nódulo pulmonar solitario
7. Wahidi MM, Govert JA, Goudar RK, Gould MK, McCrory DC, American College
of Chest Physicians. Evidence for the Treatment of Patients With Pulmonary
Nodules: When Is It Lung Cancer? Chest. 2007;132(3):94S-107S.
8. Henschke CI, McCauley DI, Yankelevitz DF, et al. Early Lung Cancer Action
Project: overall design and findings from baseline screening. Lancet.
1999;354(9173):99-105.
9. Torre Bouscoulet L, De la Garza Salazar J, García Sancho Figueroa C, Palomar
Lever A. Generalidades del cáncer de pulmón de células no pequeñas. En:
Arriete Rodríguez O, Guzmán de Alba E, eds. Cáncer de pulmón de células no
pequeñas. México; 2016. p. 1-14.
10. Naidich DP, Bankier AA, MacMahon H, et al. Recommendations for the Manage-
ment of Subsolid Pulmonary Nodules Detected at CT: A Statement from the
Fleischner Society. Radiology. 2013;266(1):304-17.
11. Gurney JW, Lyddon DM, McKay JA. Determining the likelihood of malignancy
in solitary pulmonary nodules with Bayesian analysis. Part II. Application. Ra-
diology. 1993;186(2):415-22.
12. Heitzman ER, Markarian B, Raasch BN, Carsky EW, Lane EJ, Berlow ME. Pathways
of tumor spread through the lung: radiologic correlations with anatomy and
pathology. Radiology. 1982;144(1):3-14.
13. Winer-Muram HT. The Solitary Pulmonary Nodule. Radiology. 2006;239(1):34-49.
14. Truong MT, Ko JP, Rossi SE, et al. Update in the Evaluation of the Solitary Pul-
monary Nodule. RadioGraphics. 2014;34(6):1658-79.
15. Grewal RG, Austin JH. CT demonstration of calcification in carcinoma of the
lung. J Comput Assist Tomogr. 1994;18(6):867-71.
16. Lee HY, Choi YL, Lee KS, et al. Pure Ground-Glass Opacity Neoplastic Lung
Nodules: Histopathology, Imaging, and Management. Am J Roentgenol.
2014;202(3):W224-33.
17. Swensen SJ, Viggiano RW, Midthun DE, et al. Lung Nodule Enhancement at CT:
Multicenter Study. Radiology. 2000;214(1):73-80.
18. Nathan MH, Collins VP, Adams RA. Differentiation of Benign and Malignant
Pulmonary Nodules by Growth Rate. Radiology. 1962;79(2):221-32.
19. Aoki T, Nakata H, Watanabe H, et al. Evolution of Peripheral Lung Adenocarci-
nomas. Am J Roentgenol. 2000;174(3):763-8.
20. Cronin P, Dwamena BA, Kelly AM, Bernstein SJ, Carlos RC. Solitary pulmonary
nodules and masses: a meta-analysis of the diagnostic utility of alternative
imaging tests. Eur Radiol. 2008;18(9):1840-56.
21. Yilmaz F, Tastekin G. F-FDG PET in evaluation of solitary pulmonary nodules. Int
J Clin Exp Med. 2015;8(1):45-51.
22. Gould MK, Donington J, Lynch WR, et al. Evaluation of Individuals With Pulmo-
nary Nodules: When Is It Lung Cancer? Chest. 2013;143(5):e93S-e120S.
23. Ng YL, Patsios D, Roberts H, et al. CT-guided percutaneous fine-needle aspira-
tion biopsy of pulmonary nodules measuring 10 mm or less. Clin Radiol.
2008;63(3):272-7.
24. Rivera MP, Mehta AC, American College of Chest Physicians. Initial Diagnosis of
Lung Cancer. Chest. 2007;132(3):131S-48.
25. Baldwin DR. Management of pulmonary nodules according to the 2015 British
Thoracic Society guidelines. Key messages for clinical practice. Polish Arch In-
tern Med. 2016;126(4):262-74.
26. Sihoe ADL, Hiranandani R, Wong H, Yeung ESL. Operating on a suspicious lung
mass without a preoperative tissue diagnosis: pros and cons. Eur J Cardiothorac
Surg. 2013;44(2):231-7; discussion 237.
27. Liu Y, Huang C, Liu H, Chen Y, Li S. Sublobectomy versus Lobectomy for stage
IA (T1a) non-small-cell lung cancer: a meta-analysis study. World J Surg Oncol.
2014;12(1):138.
28. Cai Y, Fu X, Xu Q, Sun W, Zhang N. Thoracoscopic Lobectomy versus Open
Lobectomy in Stage I Non-Small Cell Lung Cancer: A Meta-Analysis. PLoS One.
2013;8(12):e82366.
111
 Clinical Review & Education

JAMA | US Preventive Services Task Force | RECOMMENDATION STATEMENT

Screening for Lung Cancer

US Preventive Services Task Force Recommendation Statement
US Preventive Services Task Force

Viewpoint page 933 and

IMPORTANCE Lung cancer is the second most common cancer and the leading cause of Editorial page 939
cancer death in the US. In 2020, an estimated 228 820 persons were diagnosed with lung Multimedia
cancer, and 135 720 persons died of the disease. The most important risk factor for lung
Related articles pages 971 and
cancer is smoking. Increasing age is also a risk factor for lung cancer. Lung cancer has a
988 and JAMA Patient Page
generally poor prognosis, with an overall 5-year survival rate of 20.5%. However, early-stage page 1016
lung cancer has a better prognosis and is more amenable to treatment.
Supplemental content
OBJECTIVE To update its 2013 recommendation, the US Preventive Services Task Force
CME Quiz at
(USPSTF) commissioned a systematic review on the accuracy of screening for lung cancer
jamacmelookup.com and CME
with low-dose computed tomography (LDCT) and on the benefits and harms of screening for Questions page 1000
lung cancer and commissioned a collaborative modeling study to provide information about
the optimum age at which to begin and end screening, the optimal screening interval, and the Related articles at
jamaoncology.com
relative benefits and harms of different screening strategies compared with modified
jamanetworkopen.com
versions of multivariate risk prediction models. jamasurgery.com
POPULATION This recommendation statement applies to adults aged 50 to 80 years who have
a 20 pack-year smoking history and currently smoke or have quit within the past 15 years.

EVIDENCE ASSESSMENT The USPSTF concludes with moderate certainty that annual screening
for lung cancer with LDCT has a moderate net benefit in persons at high risk of lung cancer
based on age, total cumulative exposure to tobacco smoke, and years since quitting smoking.

RECOMMENDATION The USPSTF recommends annual screening for lung cancer with LDCT
in adults aged 50 to 80 years who have a 20 pack-year smoking history and currently smoke
or have quit within the past 15 years. Screening should be discontinued once a person
has not smoked for 15 years or develops a health problem that substantially limits life Author/Group Information: The US
Preventive Services Task Force
expectancy or the ability or willingness to have curative lung surgery. (B recommendation) (USPSTF) members are listed at the
This recommendation replaces the 2013 USPSTF statement that recommended annual end of this article.
screening for lung cancer with LDCT in adults aged 55 to 80 years who have a 30 pack-year Corresponding Author: Alex H.
smoking history and currently smoke or have quit within the past 15 years. Krist, MD, MPH, Virginia
Commonwealth University, 830 E
Main St, One Capitol Square,
JAMA. 2021;325(10):962-970. doi:10.1001/jama.2021.1117 Sixth Floor, Richmond, VA 23219
(chair@uspstf.net).

Summary of Recommendation

Adults aged 50 to 80 years who have a 20 pack-year
smoking history and currently smoke or have quit
within the past 15 years
The USPSTF recommends annual screening for lung cancer with low-dose computed
tomography (LDCT) in adults aged 50 to 80 years who have a 20 pack-year smoking
history and currently smoke or have quit within the past 15 years. Screening should
B
be discontinued once a person has not smoked for 15 years or develops a health
problem that substantially limits life expectancy or the ability or willingness to have
curative lung surgery.

See the Figure for a more detailed summary of the recommendation for clinicians. USPSTF indicates US Preventive Services Task Force.

L
ung cancer is the second most common cancer and the lead-
ing cause of cancer death in the US. In 2020, an estimated
228 820 persons were diagnosed with lung cancer, and
135 720 persons died of the disease.1
Themostimportantriskfactorforlungcancerissmoking.2,3 Smok-
ingisestimatedtoaccountforabout90%ofalllungcancercases,2 with
a relative risk of lung cancer approximately 20-fold higher in smokers
than in nonsmokers.3 Increasing age is also a risk factor for lung cancer.
The median age of diagnosis of lung cancer is 70 years.4,5
Lung cancer has a generally poor prognosis, with an overall
5-year survival rate of 20.5%.1 However, early-stage lung cancer has
a better prognosis and is more amenable to treatment.

962 JAMA March 9, 2021 Volume 325, Number 10 (Reprinted) jama.com

© 2021 American Medical Association. All rights reserved.

Downloaded From: https://jamanetwork.com/ Mexico | Access Provided by JAMA by Salvador Macias Diaz on 10/21/2021
 USPSTF Evidence Report: Screening for Lung Cancer US Preventive Services Task Force Clinical Review & Education

Figure. Clinician Summary: Screening for Lung Cancer

What does the USPSTF Adults aged 50 to 80 years who have a 20 pack-year smoking history and currently smoke or have quit within the past 15 years:
recommend? • Screen for lung cancer with low-dose computed tomography (CT) every year.
• Stop screening once a person has not smoked for 15 years or has a health problem that limits life expectancy or the ability
to have lung surgery.
Grade: B

To whom does this Adults aged 50 to 80 years who have a 20 pack-year smoking history and currently smoke or have quit within the past 15 years.
recommendation apply? (See below for definition of pack-year.)

What’s new? The USPSTF has revised the recommended ages and pack-years for lung cancer screening. It expanded the age range to 50 to
80 years (previously 55 to 80 years) and reduced the pack-year history to 20 pack-years of smoking (previously 30 pack-years).

How to implement this 1. Assess risk based on age and pack-year smoking history: Is the person aged 50 to 80 years and have they accumulated 20
recommendation? pack-years or more of smoking?
a. A pack-year is a way of calculating how much a person has smoked in their lifetime. One pack-year is the equivalent of
smoking an average of 20 cigarettes—1 pack—per day for a year.
2. Screen: If the person is aged 50 to 80 years and has a 20 pack-year or more smoking history, engage in shared decision-making
about screening.
a. The decision to undertake screening should involve a discussion of its potential benefits, limitations, and harms.
b. If a person decides to be screened, refer them for lung cancer screening with low-dose CT, ideally to a center with experience
and expertise in lung cancer screening.
c. If the person currently smokes, they should receive smoking cessation interventions.

How often? • Screen every year with low-dose CT.

• Stop screening once a person has not smoked for 15 years or has a health problem that limits life expectancy or the
ability to have lung surgery.

What are other The USPSTF has made recommendations on interventions to prevent the initiation of tobacco use in children and adolescents,
relevant USPSTF and on behavioral and pharmacotherapy interventions for tobacco smoking cessation in adults, including pregnant women.
recommendations? These recommendations are available at https://www.uspreventiveservicestaskforce.org

Where to read the full Visit the USPSTF website (https://www.uspreventiveservicestaskforce.org) to read the full recommendation statement.
recommendation This includes more details on the rationale of the recommendation, including benefits and harms; supporting evidence;
statement? and recommendations of others.

The USPSTF recognizes that clinical decisions involve more considerations than evidence alone. Clinicians should understand the evidence but individualize
decision-making to the specific patient or situation.

USPSTF indicates US Preventive Services Task Force.

Table. Summary of USPSTF Rationale

USPSTF Assessment of Magnitude of Net Benefit
Rationale Assessment
Detection The USPSTF found adequate evidence that LDCT has
The US Preventive Services Task Force (USPSTF) concludes with sufficient sensitivity and specificity to detect early–stage
moderate certainty that annual screening for lung cancer with LDCT lung cancer
has a moderate net benefit in persons at high risk of lung cancer Benefits of early The USPSTF found adequate evidence that annual screening
detection and for lung cancer with LDCT in a defined population of
based on age, total cumulative exposure to tobacco smoke, and years intervention and high-risk persons can prevent a substantial number of lung
since quitting smoking. The moderate net benefit of screening de- treatment cancer–related deaths
pends on limiting screening to persons at high risk, the accuracy of Harms of early • The harms associated with LDCT screening include
detection and false-positive results leading to unnecessary tests and
image interpretation being similar to or better than that found in clini- intervention and invasive procedures, incidental findings, short-term
cal trials, and the resolution of most false-positive results with se- treatment increases in distress due to indeterminate results,
overdiagnosis, and radiation exposure
rial imaging rather than invasive procedures. • The USPSTF found adequate evidence that the harms of
screening for lung cancer with LDCT are moderate in
See the Figure, Table, and eFigure in the Supplement for more magnitude
information on the USPSTF recommendation rationale and assess- USPSTF The USPSTF concludes with moderate certainty that annual
ment. For more details on the methods the USPSTF uses to deter- assessment screening for lung cancer with LDCT is of moderate net
benefit for persons at high risk of lung cancer based on age,
mine the net benefit, see the USPSTF Procedure Manual.6 total cumulative exposure to tobacco smoke, and years
since quitting smoking
Abbreviations: LDCT, low-dose computed tomography; USPSTF, US Preventive
Services Task Force.
Practice Considerations
Patient Population Under Consideration Assessment of Risk
This recommendation applies to adults aged 50 to 80 years who Smoking and older age are the 2 most important risk factors for
have a 20 pack-year smoking history and currently smoke or have lung cancer. 3-5 The risk of lung cancer in persons who smoke
quit within the past 15 years. increases with cumulative quantity and duration of smoking and

jama.com (Reprinted) JAMA March 9, 2021 Volume 325, Number 10 963

© 2021 American Medical Association. All rights reserved.

Downloaded From: https://jamanetwork.com/ Mexico | Access Provided by JAMA by Salvador Macias Diaz on 10/21/2021
 Clinical Review & Education US Preventive Services Task Force
with age but decreases with increasing time since quitting for
persons who formerly smoked.3 The USPSTF considers adults aged
50 to 80 years who have a 20 pack-year smoking history and cur-
rently smoke or have quit within the past 15 years to be at high risk
and recommends screening for lung cancer with annual LDCT in
this population.
African American/Black (Black) men have a higher incidence of
lung cancer than White men, and Black women have a lower inci-
dence than White women.1 These differences are likely related to
differences in smoking exposure (ie, prevalence of smoking) and re-
lated exposure to carcinogens in cigarettes.7,8 The differences may
also be related to other social risk factors.
Other risk factors for lung cancer include environmental
exposures, prior radiation therapy, other (noncancer) lung dis-
eases, and family history. Lower level of education is also associ-
ated with a higher risk of lung cancer.7 The task force recommends
using age and smoking history to determine screening eligibility
rather than more elaborate risk prediction models because there is
insufficient evidence to assess whether risk prediction model–
based screening would improve outcomes relative to using the
risk factors of age and smoking history for broad implementation
in primary care.
Screening Tests
Low-dose computed tomography has high sensitivity and reason-
able specificity for the detection of lung cancer, with demon-
strated benefit in screening persons at high risk.9-11 Other potential
screening modalities that are not recommended because they have
not been found to be beneficial include sputum cytology, chest ra-
diography, and measurement of biomarker levels.12,13
Screening Intervals
The 2 lung cancer screening trials that showed a benefit of lung
cancer screening used different screening intervals. The National
Lung Screening Trial (NLST) screened annually for 3 years.9 The
Nederlands-Leuvens Longkanker Screenings Onderzoek (NELSON)
trial screened at intervals of 1 year, then 2 years, then 2.5 years.11
Modeling studies from the Cancer Intervention and Surveillance
Modeling Network (CISNET)14,15 suggest that annual screening for
lung cancer leads to greater benefit than does biennial screening.
Based on the available evidence and these models, the USPSTF
recommends annual screening.
Treatment and Interventions
Lung cancer can be treated with surgery, chemotherapy, radiation
therapy, targeted therapies, immunotherapy, or combinations of
these treatments.16 Surgical resection is generally considered the
current treatment of choice for patients with stage I or II non–small
cell lung cancer (NSCLC).17
Implementation of Lung Cancer Screening
Available data indicate that uptake of lung cancer screening is low.
One recent study using data for 10 states found that 14.4% of per-
sons eligible for lung cancer screening (based on 2013 USPSTF cri-
teria) had been screened in the prior 12 months.18 Increasing lung
cancer screening discussions and offering screening to eligible per-
sons who express a preference for it is a key step to realizing the po-
tential benefit of lung cancer screening.
964 JAMA March 9, 2021 Volume 325, Number 10 (Reprinted)
© 2021 American Medical Association. All rights reserved.
Downloaded From: https://jamanetwork.com/ Mexico | Access Provided by JAMA by Salvador Macias Diaz on 10/21/2021
USPSTF Evidence Report: Screening for Lung Cancer
Screening Eligibility, Screening Intervals, and Starting
and Stopping Ages
As noted above, the USPSTF recommends annual screening for lung
cancer with LDCT in adults aged 50 to 80 years who have at least a
20 pack-year smoking history. Screening should be discontinued
once a person has not smoked for 15 years.
The NLST9 and the NELSON trial11 enrolled generally healthy per-
sons, so those study findings may not accurately reflect the bal-
ance of benefits and harms in persons with comorbid conditions. The
USPSTF recommends discontinuing screening if a person develops
a health problem that substantially limits life expectancy or the abil-
ity or willingness to have curative lung surgery.
Smoking Cessation Counseling
All persons enrolled in a screening program who are current smok-
ers should receive smoking cessation interventions. To be consis-
tent with the USPSTF recommendation on counseling and inter-
ventions to prevent tobacco use and tobacco-caused disease,19
persons referred for lung cancer screening through primary
care should receive these interventions concurrent with referral.
Because many persons may enter screening through pathways
besides referral from primary care, the USPSTF encourages incor-
porating such interventions into all screening programs.
Shared Decision-making
Shared decision-making is important when clinicians and patients
discuss screening for lung cancer. The benefit of screening varies
with risk because persons at higher risk are more likely to benefit.
Screening does not prevent most lung cancer deaths; thus,
smoking cessation remains essential. Lung cancer screening
has the potential to cause harm, including false-positive results
and incidental findings that can lead to subsequent testing and
treatment, including the anxiety of living with a lung lesion that
may be cancer. Overdiagnosis of lung cancer and the risks of
radiation exposure are harms, although their exact magnitude is
uncertain. The decision to undertake screening should involve a
thorough discussion of the potential benefits, limitations, and
harms of screening.
Standardization of LDCT Screening and Follow-up
of Abnormal Findings
The randomized clinical trials (RCTs) that provide evidence for
the benefit of screening for lung cancer with LDCT were pri-
marily conducted in academic centers with expertise in the per-
formance and interpretation of LDCT and the management of
lung lesions seen on LDCT. Clinical settings that have similar expe-
rience and expertise are more likely to duplicate the beneficial
results found in trials.
In an effort to minimize the uncertainty and variation about
the evaluation and management of lung nodules and to standard-
ize the reporting of LDCT screening results, the American College
of Radiology developed the Lung Imaging Reporting and Data Sys-
tem (Lung-RADS) classification system and endorses its use in lung
cancer screening.20 Lung-RADS provides guidance to clinicians on
which findings are suspicious for cancer and the suggested man-
agement of lung nodules detected on LDCT. Data suggest that the
use of Lung-RADS may decrease the rate of false-positive results in
lung cancer screening.21
jama.com
 USPSTF Evidence Report: Screening for Lung Cancer
Additional Tools and Resources
The Centers for Disease Control and Prevention has several web-
sites with many resources to help patients stop smoking:
• “How to Quit” resources (https://www.cdc.gov/quit)
• Smoking Cessation: Fast Facts (https://www.cdc.gov/tobacco/
data_statistics/fact_sheets/cessation/smoking-cessation-fast-
facts/)
• Tips From Former Smokers (https://www.cdc.gov/tobacco/
campaign/tips/)
The National Cancer Institute has developed resources to help
patients stop smoking (https://www.smokefree.gov). It has also
developed patient and clinician guides on screening for lung cancer:
• Lung Cancer Screening (PDQ)–Patient Version (https://www.cancer.
gov/types/lung/patient/lung-screening-pdq)
• Lung Cancer Screening (PDQ)–Health Professional Version (https://
www.cancer.gov/types/lung/hp/lung-screening-pdq)
Other Related USPSTF Recommendations
Prevention of initiation of smoking and smoking cessation for
those who smoke are the most important interventions to prevent
lung cancer. The USPSTF has made recommendations on interven-
tions to prevent the initiation of tobacco use in children and
adolescents22 and on the use of pharmacotherapy and counseling
for tobacco cessation.19
Update of Previous USPSTF Recommendation
This recommendation replaces the 2013 USPSTF recommendation
on screening for lung cancer. In 2013 the USPSTF recommended
annual screening for lung cancer with LDCT in adults aged 55 to
80 years who have a 30 pack-year smoking history and currently
smoke or have quit within the past 15 years (abbreviated as
A-55-80-30-15).23 For this updated recommendation, the USPSTF
has changed the age range and pack-year eligibility criteria and
recommends annual screening for lung cancer with LDCT in adults
aged 50 to 80 years who have a 20 pack-year smoking history and
currently smoke or have quit within the past 15 years (A-50-80-
20-15). Abbreviations for screening recommendations are
expanded in the Box.
As in the 2013 recommendation, the USPSTF recommends
that screening should be discontinued once a person has not
smoked for 15 years or develops a health problem that substantially
limits life expectancy or the ability or willingness to have curative
lung surgery.
Supporting Evidence
Scope of Review
To update its 2013 recommendation, the USPSTF commissioned a
systematic review24,25 on the accuracy of screening for lung cancer
with LDCT and the benefits and harms of screening for lung cancer.
The review also assessed whether the benefits of screening vary by
subgroup (eg, by race or sex) or by the number or frequency of LDCT
scans and whether the harms associated with screening and the
evaluation of lung nodules differ with the use of Lung-RADS, Inter-
national Early Lung Cancer Action Program (I-ELCAP), or similar ap-
jama.com
© 2021 American Medical Association. All rights reserved.
Downloaded From: https://jamanetwork.com/ Mexico | Access Provided by JAMA by Salvador Macias Diaz on 10/21/2021
US Preventive Services Task Force Clinical Review & Education
Box. US Preventive Services Task Force Low-Dose Computed
Tomographic Screening Recommendations for Lung Cancer
A-55-80-30-15
In 2013, The US Preventive Services Task Force (USPSTF)
recommended annual screening for lung cancer with low-dose
computed tomography (LDCT) for adults aged 55 to 80 years who
have a 30 pack-year smoking history and currently smoke or have
quit within the past 15 years (abbreviated as A-55-80-30-15).23
A-50-80-20-15
For this updated recommendation, the USPSTF has changed the
age range and pack-year eligibility criteria and recommends annual
screening for lung cancer with LDCT for adults aged 50 to 80 years
who have a 20 pack-year smoking history and currently smoke or
have quit within the past 15 years (abbreviated as A-50-80-20-15).
proaches (eg, to reduce false-positive results). In addition, the re-
view assessed whether the use of risk prediction models for
identifying adults at higher risk of lung cancer mortality improves
the balance of benefits and harms of screening compared with the
use of trial eligibility criteria or variants of the prior USPSTF recom-
mendation criteria.
In addition to the systematic evidence review, the USPSTF com-
missioned collaborative modeling studies from CISNET14,15 to pro-
vide information about the optimal age at which to begin and end
screening, the optimal screening interval, and the relative benefits
and harms of different screening strategies, including risk factor–
based strategies using age, pack-year smoking history, and years since
quitting smoking for former smokers, compared with modified ver-
sions of multivariate risk prediction models. The modeling studies
complement the evidence that the systematic review provides.
Accuracy of Screening Tests
The USPSTF reviewed several RCTs and cohort studies that re-
ported on the sensitivity, specificity, or predictive value of LDCT, using
eventual diagnosis of lung cancer as the reference standard.24,25 Not
all of the reviewed studies reported all test accuracy data. In the stud-
ies that reported it, sensitivity ranged from 59% to 100%, specific-
ity ranged from 26.4% to 99.7%, positive predictive value ranged
from 3.3% to 43.5%, and negative predictive value ranged from
97.7% to 100%.
In the NLST26 and NELSON trial,11 the reported sensitivities were
93.1% and 59%, respectively, and reported specificities were 76.5%
and 95.8%, respectively. Although the negative predictive values
were similar for the NLST and NELSON trial (99.9% and 97.7%, re-
spectively), the positive predictive values were very different (3.3%
and 43.5%, respectively). This discrepancy is largely accounted for
by the trials’ differing definitions of a positive finding and screening
protocols—the NELSON trial used a volumetric approach and added
an indeterminate nodule result category (ie, an indeterminate find-
ing was not considered a positive result even if it led to additional
testing). The NLST used an approach of maximum diameter with-
out an indeterminate category (ie, any nodule meeting the diam-
eter criteria was considered a positive result).
Three retrospective studies compared how various ap-
proaches for nodule classification would alter the accuracy of
LDCT.21,27,28 The first study demonstrated that using Lung-RADS in
(Reprinted) JAMA March 9, 2021 Volume 325, Number 10 965
 Clinical Review & Education US Preventive Services Task Force
the NLST would have increased specificity while decreasing
sensitivity.21 The other 2 studies found that use of I-ELCAP criteria
(increase in nodule size threshold to an average diameter of 5 mm,
6 mm, or larger) would increase positive predictive value.27,28
Benefits of Early Detection and Treatment
The USPSTF reviewed 7 RCTs that evaluated lung cancer screening
with LDCT.24,25 The NLST9 and the NELSON trial11 were the only trials
adequately powered to detect a lung cancer mortality benefit.
The NLST, the largest RCT to date (n = 53 454), enrolled par-
ticipants aged 55 to 74 years at the time of randomization who had
a tobacco use history of at least 30 pack-years and were current
smokers or had quit within the past 15 years. The mean pack-year
smoking history in NLST participants was 56 pack-years.9 The
NELSON trial (n = 15 792) enrolled participants aged 50 to 74 years
who had a tobacco use history of at least 15 cigarettes a day
(three-fourths of a pack per day) for more than 25 years or 10 ciga-
rettes a day (one-half of a pack per day) for more than 30 years and
were current smokers or had quit within the past 10 years. The
median pack-year smoking history in NELSON trial participants was
38 pack-years.11
The NLST reported a relative risk reduction in lung cancer mor-
tality of 20% (95% CI, 6.8%-26.7%)9; a subsequent analysis of
NLST data with additional follow-up and end point verification
reported a relative risk reduction of 16% (95% CI, 5%-25%).10 At
10 years of follow-up, the NELSON trial reported 181 lung cancer
deaths among participants in the screening group and 242 in
the control group (incidence rate ratio [IRR], 0.75 [95% CI,
0.61-0.90]).11,24 The NLST also found a reduction in all-cause mor-
tality with LDCT screening compared with chest radiography (IRR,
0.93 [95% CI, 0.88-0.99]). Results of the other trials were impre-
cise, without any statistically significant differences between
screening with LDCT and chest radiography or no screening.24
Evidence on screening interval comes from the NLST and the
NELSON trial and CISNET modeling studies. The NLST screened an-
nually for 3 years.9 The NELSON trial screened at intervals of 1 year,
then 2 years, then 2.5 years.11 The CISNET modeling studies sug-
gest that annual screening with LDCT provides greater benefit in de-
creasing lung cancer mortality and in life-years gained compared with
biennial screening.14
Several lines of evidence suggest that screening for lung can-
cer in persons with fewer pack-years of smoking (ie, fewer than the
30 pack-year eligibility criterion of the 2013 USPSTF recommenda-
tion) and at an earlier age can increase the benefits of screening. As
noted, the NELSON trial enrolled persons aged 50 to 74 years (about
one-fourth of participants were younger than 55 years) who had ac-
cumulated fewer pack-years of smoking (half of a pack per day for
more than 30 years or three-fourths of a pack per day for more than
25 years).11 This trial provides empirical evidence for the benefit of
screening for lung cancer with LDCT in persons aged 50 to 55 years
and with lighter pack-year smoking histories.
The CISNET modeling studies also provided data that helped in-
form the pack-year eligibility criterion for lung cancer screening and
the ages at which to start and stop screening. The USPSTF focused
on screening programs in the 1960 birth cohort (more representa-
tive of current smoking patterns compared with earlier cohorts) that
yielded lung cancer mortality reductions at least as great as the 2013
USPSTF screening program (A-55-80-30-15). For screening pro-
966 JAMA March 9, 2021 Volume 325, Number 10 (Reprinted)
© 2021 American Medical Association. All rights reserved.
Downloaded From: https://jamanetwork.com/ Mexico | Access Provided by JAMA by Salvador Macias Diaz on 10/21/2021
USPSTF Evidence Report: Screening for Lung Cancer
grams that provide this level of mortality benefit and also maxi-
mize, or come close to maximizing, both lung cancer deaths averted
and life-years gained for any given level of LDCT screening, in at least
3 of the 4 CISNET models (ie, “consensus-efficient” programs), the
majority (52%) have a minimum pack-year eligibility criterion of 20
pack-years. Almost all have a starting age of 50 or 55 years, and all
have a stopping age of 80 years.14,15
Relative to the 2013 USPSTF screening program (A-55-80-30-
15), CISNET modeling analyses suggest that annually screening per-
sons aged 50 to 80 years who have at least a 20 pack-year smoking
history and currently smoke or have quit within the past 15 years
(A-50-80-20-15) would be associated with lung cancer mortality
reduced by 13.0% vs 9.8%, with avoiding 503 vs 381 lung cancer
deaths, and with 6918 life-years gained vs 4882 life-years gained
per 100 000 persons in the population aged 45 to 90 years over a
lifetime of screening.14 Thus, this screening program would be
associated with important reductions in lung cancer deaths and
increases in life-years gained compared with the previous recom-
mendation and is supported by new trial data and the CISNET mod-
eling studies.
Screening for lung cancer in persons at an earlier age and with
fewer pack-years of smoking (ie, 20 pack-years) may also help par-
tially ameliorate racial disparities in screening eligibility. Data sug-
gest that Black persons who smoke have a higher risk of lung can-
cer than do White persons, and this risk difference is more
apparent at lower levels of smoking intensity.7 One recent analysis
of Southern Community Cohort Study participants found that 17%
of Black persons who smoke were eligible for lung cancer screen-
ing based on the 2013 USPSTF eligibility criteria compared with
31% of White persons who smoke. In the same study, among per-
sons diagnosed with lung cancer, a significantly lower percentage
of Black persons who smoke (32%) were eligible for screening
than were White persons (56%).29 Data also suggest that Latinx/
Hispanic persons who smoke accumulate fewer pack-years than
White persons who smoke.30,31 A strategy of annually screening
persons aged 50 to 80 years who have at least a 20 pack-year
smoking history and currently smoke or have quit within the past
15 years (A-50-80-20-15) would increase the relative percentage
of persons eligible for screening by 87% overall—78% in non-
Hispanic White adults, 107% in non-Hispanic Black adults, and
112% in Hispanic adults compared with 2013 USPSTF criteria
(A-55-80-30-15).14 Similarly, a strategy of screening persons aged
50 to 80 years who have at least a 20 pack-year smoking history
and currently smoke or have quit within the past 15 years (A-50-
80-20-15) would increase the relative percentage of persons eli-
gible for screening by 80% in men and by 96% in women, 14
because they accumulate fewer pack-years than men.32
Simulation studies suggest that risk prediction models to
determine eligibility for lung cancer screening could be associated
with reduced lung cancer deaths and the number of participants
needed to screen to prevent 1 lung cancer death. The CISNET
modeling studies commissioned by the USPSTF thus compared
the benefits and harms of screening programs based on risk pre-
diction models vs risk factor–based screening (ie, using age and
smoking history). The risk prediction models used were modified
versions of the PLCOm2012 model,33 the Lung Cancer Death Risk
Assessment Tool (LCDRAT) model,34 and the Bach model,35 lim-
ited to age, sex (for those models that include sex as a variable,
jama.com
 USPSTF Evidence Report: Screening for Lung Cancer
such as the LCDRAT and Bach models), smoking intensity, and
smoking duration (and setting other potential variables such as
race, education, body mass index, personal history of cancer, or
family history of lung cancer to their reference value). Because age
is an important risk factor for lung cancer, these risk prediction
models shifted screening to persons of older age and increased
the number of lung cancer deaths averted, but screening occurs at
older ages when there are fewer years to be gained. Thus, some
risk prediction model–based screening programs were associated
with slightly increased life-years gained, while some were not or
were associated with even slightly decreased life-years gained.
Risk prediction models also were associated with increased the
number of over diagnosed lung cancers, which are more common
in older individuals.14
It is possible that the use of a more complex risk prediction model
to determine eligibility might impose a barrier to wider implemen-
tation and uptake of lung cancer screening, a service that currently
has low uptake. Currently, there are no studies that have prospec-
tively compared the use of USPSTF criteria considering age, pack-
year smoking history, and number of years since quitting vs risk pre-
diction models as criteria for lung cancer screening, so it is uncertain
whether using a risk prediction model would improve lung cancer
detection and clinical outcomes. The International Lung Screening
Trial (ILST), a prospective cohort study that is comparing the accu-
racy of the PLCOm2012 model against the 2013 USPSTF criteria for
detecting lung cancer, may provide some evidence regarding this
issue.36 In summary, determining eligibility for lung cancer screen-
ing using more complex risk prediction models may represent an
implementation barrier, and there is currently insufficient evi-
dence to assess whether risk prediction model–based screening
would improve outcomes relative to simply using the risk factors of
age and smoking history.
Harms of Screening and Treatment
Harms of screening can include false-positive results leading to unnec-
essarytestsandinvasiveprocedures,overdiagnosis,radiation-induced
cancer, incidental findings, and increases in distress or anxiety.
The NLST reported false-positive rates of 26.3% for baseline,
27.2% for year 1, and 15.9% for year 2.9 The NELSON trial reported
false-positive rates of 19.8% at baseline, 7.1% at year 1, 9.0% for males
at year 3, and 3.9% for males at year 5.5 of screening.11,37 An imple-
mentation study through the Veterans Health Administration re-
vealed a false-positive rate of 28.9% of veterans eligible for screen-
ing (58% of those who were actually screened) at baseline.38 Both
of these studies were conducted prior to the use of the Lung-RADS
protocol for nodule classification, the use of which may reduce false-
positives, albeit at the cost of some false-negatives. One retrospec-
tive study assessed how use of Lung-RADS would have changed the
false-positive result rate in the NLST and found a false-positive rate
among baseline results for Lung-RADS of 12.8% (95% CI, 12.4%-
13.2%) vs 26.6% (95% CI, 26.1%-27.1%) for the NLST approach.21
The further workup of false-positive results can result in signifi-
cant harms such as additional imaging, biopsy, or surgical proce-
dures. Fourteen studies reported on the evaluation of false-
positive results. Among all patients screened, the percentage who
had a needle biopsy for false-positive results ranged from 0.09%
to 0.56%. Complication rates from needle biopsy for false-positive
results ranged from 0.03% to 0.07% of all patients screened. Sur-
jama.com
© 2021 American Medical Association. All rights reserved.
Downloaded From: https://jamanetwork.com/ Mexico | Access Provided by JAMA by Salvador Macias Diaz on 10/21/2021
US Preventive Services Task Force Clinical Review & Education
gical procedures for false-positive results were reported in 0.5% to
1.3% of all screened participants.24
In the NLST, false-positive results led to invasive procedures
(needle biopsy, thoracotomy, thoracoscopy, mediastinoscopy, and
bronchoscopy) in 1.7% of patients screened. Complications oc-
curred in 0.1% of patients screened, and death in the 60 days fol-
lowing the most invasive procedure performed to evaluate a false-
positive result occurred in 0.007% of those screened.9 One study
estimated that the use of Lung-RADS criteria would have pre-
vented 23.4% of invasive procedures due to false-positive results.21
In the CISNET modeling studies, the false-positive rate varied
based on screening eligibility criteria. Relative to the 2013 USPSTF
criteria (A-55-80-30-15), the 2021 USPSTF criteria (A-50-80-20-
15) would result in 2.2 vs 1.9 false-positive results per person over a
lifetime of screening.14 Note that screening programs that start at
younger ages or use a lower pack-year eligibility screen a larger total
number of persons.
Determining the rate of overdiagnosis in screening trials is
challenging because the duration of follow-up affects the calcula-
tion of excess, potentially over diagnosed, cancers in the screening
vs control groups. Initially, the NLST reported 119 additional lung
cancers (1060 total cancers with LDCT vs 941 with chest radiogra-
phy) after 3 screening rounds and 6.5 years of follow-up (IRR, 1.12
[95% CI, 1.02-1.22]).9,24 With extended follow-up, the NLST found
no statistically significant difference between groups for overall
lung cancer incidence; however, this study had some methodologi-
cal limitations, including use of a different ascertainment method
during posttrial follow-up, lack of information on any posttrial
screening that may have occurred in either the LDCT or chest radi-
ography group, and missing data.39 In the NELSON trial, 40 excess
lung cancers (344 cancers in the LDCT group vs 304 in the control
group) were reported in the LDCT group after the a priori planned
10 years of follow-up; after 11 years of follow-up, there was an
excess of 14 cancers with LDCT.11
In the CISNET modeling studies, which account for lifetime fol-
low-up, the 2013 USPSTF screening program (A-55-80-30-15) would
result in 6.3% of screen-detected cases of lung cancer being over
diagnosed lung cancers vs 6.0% lung cancers being overdiagnoses
with the 2021 screening program (A-50-80-20-15).14
In the 9 publications reporting on radiation exposure associ-
ated with LDCT,24 the radiation exposure associated with 1 LDCT scan
ranged from 0.65 to 2.36 mSv. For context, average annual back-
ground radiation exposure in the US is 2.4 mSv. Two of the studies
estimated the cumulative radiation exposure for participants un-
dergoing screening with LDCT. Using estimated radiation exposure
from screening and follow-up evaluations and estimates of the risk
of radiation-induced cancer deaths, the Italian Lung Cancer Screen-
ing Trial (ITALUNG) estimated a lifetime risk of fatal cancer of 0.11
cases per 1000 persons for LDCT after the 4 screening rounds,40
and the Continuing Observation of Smoking Subjects study esti-
mated a lifetime risk of 2.6 to 8.1 major cancers per 10 000 persons
screened after 10 rounds of annual screening.41
The CISNET modeling studies found that lifetime estimates of
radiation-related lung cancer deaths varied by eligibility criteria for
screening. Relative to the 2013 USPSTF recommendation (A-55-80-
30-15), the 2021 USPSTF recommendation (A-50-80-20-15) would
be associated with an estimated 38.6 vs 20.6 radiation-related lung
cancer deaths per 100 000 persons in the total population aged 45
(Reprinted) JAMA March 9, 2021 Volume 325, Number 10 967
 Clinical Review & Education US Preventive Services Task Force
to 90 years, or 1 death caused for every 13.0 vs 18.5 lung cancer
deaths avoided by screening.14
When comparing LDCT groups vs control groups for smoking
cessation or abstinence outcomes, evidence does not indicate that
screening leads to lower rates of smoking cessation or continued ab-
stinence or to higher rates of relapse. Several studies suggest that,
compared with no screening, individuals who receive LDCT screen-
ing do not have worse health-related quality of life, anxiety, or dis-
tress over 2 years of follow-up. However, screening participants who
receive true-positive or indeterminate results may experience worse
health-related quality of life, anxiety, or distress in the short-term.24
Studies reported a wide range of screening-related incidental
findings that were deemed significant or required further evalua-
tion (4.4% to 40.7%), in part because of inconsistent definitions of
what constitutes an incidental finding and which findings were clini-
cally significant.24 Older age was associated with a greater likeli-
hood of incidental findings. Common incidental findings included
coronary artery calcification, aortic aneurysms, emphysema, and in-
fectious and inflammatory processes. Other common findings were
masses, nodules, or cysts of the kidney, breast, adrenal gland, liver,
thyroid, pancreas, spine, and lymph nodes. Cancers involving the kid-
ney, thyroid, or liver were ultimately diagnosed in 0.39% of NLST
participants in the LDCT group during screening.42
Incidental findings led to downstream evaluation, including con-
sultations, additional imaging, and invasive procedures with asso-
ciated costs and burdens. The benefits of incidental detection of non-
lung cancer conditions and the balance of benefits and harms of
incidental findings on LDCT screening remain uncertain.
Response to Public Comment
A draft version of this recommendation statement was posted for
public comment on the USPSTF website from July 7, 2020, to August
3, 2020. Most comments generally agreed with the draft recom-
mendation, although some requested broadening the eligibility cri-
teria for lung cancer screening and others mentioned that addi-
tional risk factors for lung cancer other than smoking exist or that
lung cancer can occur in persons who never smoked. In response,
the USPSTF acknowledges that there are risk factors for lung can-
cer other than smoking; however, current evidence does not sup-
port the incorporation of these risk factors as determinants of eli-
gibility for lung cancer screening. The USPSTF also acknowledges
that lung cancer can occur in persons who never smoked or among
persons who currently smoke or formerly smoked who do not meet
screening eligibility criteria. Nevertheless, smoking is the major risk
factor for lung cancer, all trials of screening for lung cancer have been
conducted among persons who smoke or were former smokers, and
trial and modeling data support the current USPSTF recommenda-
tion as offering a reasonable balance of benefits and harms.
Some comments suggested the use of more complex risk pre-
diction models to determine eligibility for lung cancer screening. In
response, the USPSTF clarified language that use of these risk pre-
diction models might make implementation more difficult, and that
there are currently no lung cancer screening trials prospectively com-
paring USPSTF eligibility criteria with risk prediction models. The
USPSTF also added a reference to the ILST, a prospective cohort
study addressing this issue.
In response to comments, the USPSTF also added information
about the currently low uptake of lung cancer screening and data
968 JAMA March 9, 2021 Volume 325, Number 10 (Reprinted)
© 2021 American Medical Association. All rights reserved.
Downloaded From: https://jamanetwork.com/ Mexico | Access Provided by JAMA by Salvador Macias Diaz on 10/21/2021
USPSTF Evidence Report: Screening for Lung Cancer
on the effect of the current recommendation on eligibility for screen-
ing in Latinx/Hispanic persons. Last, the USPSTF added and up-
dated resources and website links in the Additional Tools and
Resources section.
How Does the Evidence Fit With Biological Understanding?
Lung cancer is a proliferation of malignant cells that originate in lung
tissue. Smoking is the strongest risk factor for lung cancer. Older age
is also associated with increasing incidence of lung cancer. Lung can-
cer is classified into 2 major categories based on cell type and im-
munohistochemical and molecular characteristics: NSCLC, which col-
lectively comprises adenocarcinoma, squamous cell carcinoma, and
large cell carcinoma, and small cell lung cancer. Screening is aimed
at early detection of NSCLC rather than small cell lung cancer be-
cause the latter is much less common and typically spreads too
quickly to be reliably detected at an early, potentially curable stage
by screening.
Currently, 79% of patients present with lung cancer that has
spread to regional lymph nodes or metastasized to distant sites. Only
17% of patients present with localized disease. Patients with local-
ized disease have a 59% 5-year survival rate, compared with 32%
for those with regional spread and 6% for those with distant
metastases.1 By leading to earlier detection and treatment, screen-
ing for lung cancer can give patients a greater chance for cure.
Research Needs and Gaps
• Implementation research addressing how best to increase the up-
take of lung cancer screening discussions in clinical practice is
needed, particularly among the populations at higher risk of death
from lung cancer or populations that are socially and economi-
cally disadvantaged.
• Research is needed to evaluate whether, as lung cancer screening
is implemented in more diverse community settings, including
among racial/ethnic minorities, among populations socially and eco-
nomically disadvantaged (for whom smoking prevalence and lung
cancer incidence is higher), and in settings that screen greater num-
bers of women, the balance of benefits and harms differs from
those observed in RCTs.
• Research to identify biomarkers that can accurately identify per-
sons at high risk is needed to improve detection and minimize false-
positive results.
• Research to identify technologies that can help more accurately
discriminate between benign and malignant lung nodules
is needed.
• Research is needed on the benefits and harms of using risk pre-
diction models to select patients for lung cancer screening, includ-
ing whether use of risk prediction models represents a barrier to
wider implementation of lung cancer screening in primary care.
Recommendations of Others
The American Association for Thoracic Surgery recommends an-
nual lung cancer screening with LDCT for North Americans aged 55
to 79 years with a 30 pack-year history of smoking. It also recom-
mends offering annual lung cancer screening with LDCT starting at
jama.com
 USPSTF Evidence Report: Screening for Lung Cancer US Preventive Services Task Force Clinical Review & Education

age 50 years to persons with a 20 pack-year smoking history if there
is an additional cumulative risk of developing lung cancer of 5% or
greater over the following 5 years.43
The American Cancer Society recommends annual lung cancer
screening with LDCT for persons aged 55 to 74 years who are in fairly
good health, have at least a 30 pack-year smoking history, and cur-
rently smoke or have quit within the past 15 years. It also recom-
mends smoking cessation counseling for current smokers, shared
decision-making about lung cancer screening, and that screening be
conducted in a high-volume, high-quality lung cancer screening and
treatment center.44
The American College of Chest Physicians suggests that an-
nual screening with LDCT should be offered to asymptomatic smok-
ers and former smokers aged 55 to 77 years who have smoked 30
pack-years or more and either continue to smoke or have quit within
the past 15 years. It also recommends that screening not be per-
formed for individuals with comorbidities that adversely influence
their ability to tolerate the evaluation of screen-detected findings
or tolerate treatment of an early-stage screen-detected lung can-
cer or that substantially limit their life expectancy.45
The National Comprehensive Cancer Network recommends an-
nual screening for lung cancer with LDCT in persons aged 55 to 77
years who have at least a 30 pack-year smoking history and cur-
rently smoke or have quit within the past 15 years or in persons 50
years or older who have at least a 20 pack-year smoking history and
have at least 1 additional risk factor for lung cancer.46
The American Academy of Family Physicians has concluded that
the evidence is insufficient to recommend for or against screening
for lung cancer with LDCT in persons at high risk of lung cancer based
on age and smoking history.47

ARTICLE INFORMATION conflict-of-interest-disclosures. All members of the 2021. https://seer.cancer.gov/statfacts/html/lungb.

Accepted for Publication: January 27, 2021.
The US Preventive Services Task Force (USPSTF)
members: Alex H. Krist, MD, MPH; Karina W.
Davidson, PhD, MASc; Carol M. Mangione, MD,
MSPH; Michael J. Barry, MD; Michael Cabana, MD,
MA, MPH; Aaron B. Caughey, MD, PhD; Esa M.
Davis, MD, MPH; Katrina E. Donahue, MD, MPH;
Chyke A. Doubeni, MD, MPH; Martha Kubik, PhD,
RN; C. Seth Landefeld, MD; Li Li, MD, PhD, MPH;
Gbenga Ogedegbe, MD, MPH; Douglas K. Owens,
MD, MS; Lori Pbert, PhD; Michael Silverstein, MD,
MPH; James Stevermer, MD, MSPH; Chien-Wen
Tseng, MD, MPH, MSEE; John B. Wong, MD.
Affiliations of The US Preventive Services Task
Force (USPSTF) members: Fairfax Family Practice
Residency, Fairfax, Virginia (Krist); Virginia
Commonwealth University, Richmond (Krist);
Feinstein Institute for Medical Research at
Northwell Health, Manhasset, New York
(Davidson); University of California, Los Angeles
(Mangione); Harvard Medical School, Boston,
Massachusetts (Barry); University of California,
San Francisco (Cabana); Oregon Health & Science
University, Portland (Caughey); University of
Pittsburgh, Pittsburgh (Davis); University of North
Carolina at Chapel Hill (Donahue); Mayo Clinic,
Rochester, Minnesota (Doubeni); George Mason
University, Fairfax, Virginia (Kubik); University of
Alabama at Birmingham (Landefeld); University of
Virginia, Charlottesville (Li); New York University,
New York (Ogedegbe); Stanford University,
Stanford, California (Owens); University of
Massachusetts Medical School, Worcester (Pbert);
Boston University, Boston, Massachusetts
(Silverstein); University of Missouri, Columbia
(Stevermer); University of Hawaii, Honolulu
(Tseng); Pacific Health Research and Education
Institute, Honolulu, Hawaii (Tseng); Tufts University
School of Medicine, Boston, Massachusetts (Wong).
Author Contributions: Dr Krist had full access to all
of the data in the study and takes responsibility for
the integrity of the data and the accuracy of the
data analysis. The USPSTF members contributed
equally to the recommendation statement.
Conflict of Interest Disclosures: All authors
have completed and submitted the ICMJE
Form for Disclosure of Potential Conflicts of
Interest. Authors followed the policy regarding
conflicts of interest described at https://www.
uspreventiveservicestaskforce.org/Page/Name/
USPSTF receive travel reimbursement and an
honorarium for participating in USPSTF meetings.
Dr Barry reported receiving grants and personal
fees from Healthwise.
Funding/Support: The USPSTF is an independent,
voluntary body. The US Congress mandates that
the Agency for Healthcare Research and Quality
(AHRQ) support the operations of the USPSTF.
Role of the Funder/Sponsor: AHRQ staff assisted
in the following: development and review of the
research plan, commission of the systematic
evidence review from an evidence-based practice
center, coordination of expert review and public
comment of the draft evidence report and draft
recommendation statement, and the writing and
preparation of the final recommendation statement
and its submission for publication. AHRQ staff had
no role in the approval of the final recommendation
statement or the decision to submit for publication.
Disclaimer: Recommendations made by the
USPSTF are independent of the US government.
They should not be construed as an official position
of AHRQ or the US Department of Health and
Human Services.
Additional Contributions: We thank Howard
Tracer, MD (AHRQ), who contributed to the writing
of the manuscript, and Lisa Nicolella, MA (AHRQ),
who assisted with coordination and editing.
Additional Information: The US Preventive
Services Task Force (USPSTF) makes
recommendations about the effectiveness of
specific preventive care services for patients
without obvious related signs or symptoms. It
bases its recommendations on the evidence of both
the benefits and harms of the service and an
assessment of the balance. The USPSTF does not
consider the costs of providing a service in this
assessment. The USPSTF recognizes that clinical
decisions involve more considerations than
evidence alone. Clinicians should understand the
evidence but individualize decision-making to the
specific patient or situation. Similarly, the USPSTF
notes that policy and coverage decisions involve
considerations in addition to the evidence of clinical
benefits and harms.
REFERENCES
1. Cancer Stat Facts: lung and bronchus cancer.
National Cancer Institute. Accessed January 15,
html
2. Alberg AJ, Brock MV, Ford JG, et al.
Epidemiology of lung cancer: diagnosis and
management of lung cancer, 3rd ed: American
College of Chest Physicians evidence-based clinical
practice guidelines. Chest. 2013;143(5 suppl):e1S-
e29S.
3. Samet JM. Health benefits of smoking cessation.
Clin Chest Med. 1991;12(4):669-679.
4. Key statistics for lung cancer. American Cancer
Society. Accessed January 15, 2021. http://www.
cancer.org/cancer/lung-cancer/about/key-statistics.
html
5. Siegel RL, Miller KD, Jemal A. Cancer statistics,
2015. CA Cancer J Clin. 2015;65(1):5-29. doi:10.
3322/caac.21254
6. Procedure Manual. US Preventive Services Task
Force. Published 2015. Accessed January 21, 2021.
https://uspreventiveservicestaskforce.org/
uspstf/procedure-manual
7. Haiman CA, Stram DO, Wilkens LR, et al. Ethnic
and racial differences in the smoking-related risk of
lung cancer. N Engl J Med. 2006;354(4):333-342.
doi:10.1056/NEJMoa033250
8. Stram DO, Park SL, Haiman CA, et al.
Racial/ethnic differences in lung cancer incidence in
the multiethnic cohort study: an update. J Natl
Cancer Inst. 2019;111(8):811-819. doi:10.1093/jnci/
djy206
9. Aberle DR, Adams AM, Berg CD, et al; National
Lung Screening Trial Research Team. Reduced
lung-cancer mortality with low-dose computed
tomographic screening. N Engl J Med. 2011;365(5):
395-409. doi:10.1056/NEJMoa1102873
10. Pinsky PF, Church TR, Izmirlian G, Kramer BS.
The National Lung Screening Trial: results stratified
by demographics, smoking history, and lung cancer
histology. Cancer. 2013;119(22):3976-3983. doi:10.
1002/cncr.28326
11. de Koning HJ, van der Aalst CM, de Jong PA,
et al. Reduced lung-cancer mortality with volume
CT screening in a randomized trial. N Engl J Med.
2020;382(6):503-513. doi:10.1056/NEJMoa1911793
12. Hirales Casillas CE, Flores Fernández JM, Padilla
Camberos E, Herrera López EJ, Leal Pacheco G,
Martínez Velázquez M. Current status of circulating
protein biomarkers to aid the early detection of

jama.com (Reprinted) JAMA March 9, 2021 Volume 325, Number 10 969

© 2021 American Medical Association. All rights reserved.

Downloaded From: https://jamanetwork.com/ Mexico | Access Provided by JAMA by Salvador Macias Diaz on 10/21/2021
 Clinical Review & Education US Preventive Services Task Force
lung cancer. Future Oncol. 2014;10(8):1501-1513.
doi:10.2217/fon.14.21
13. Oken MM, Hocking WG, Kvale PA, et al; PLCO
Project Team. Screening by chest radiograph and
lung cancer mortality: the Prostate, Lung,
Colorectal, and Ovarian (PLCO) randomized trial.
JAMA. 2011;306(17):1865-1873. doi:10.1001/jama.
2011.1591
14. Meza R, Jeon J, Toumazis I, et al. Evaluation of
the Benefits and Harms of Lung Cancer Screening
With Low-Dose Computed Tomography:
A Collaborative Modeling Study for the U.S.
Preventive Services Task Force. Agency for Healthcare
Research and Quality; 2021. AHRQ publication 20-
05266-EF-2.
15. Meza R, Jeon J, Toumazis I, et al. Evaluation of
the benefits and harms of lung cancer screening
with low-dose computed tomography: modeling
study for the US Preventive Services Task Force.
JAMA. Published March 9, 2021. doi:10.1001/jama.
2021.1077
16. Non-Small Cell Lung Cancer Treatment
(PDQ®)–Health Professional Version. National
Cancer Institute. Updated November 18, 2020.
Accessed January 15, 2021. https://www.cancer.
gov/types/lung/hp/non-small-cell-lung-treatment-
pdq
17. Yan TD, Black D, Bannon PG, McCaughan BC.
Systematic review and meta-analysis of
randomized and nonrandomized trials on safety
and efficacy of video-assisted thoracic surgery
lobectomy for early-stage non-small-cell lung
cancer. J Clin Oncol. 2009;27(15):2553-2562. doi:
10.1200/JCO.2008.18.2733
18. Zahnd WE, Eberth JM. Lung cancer screening
utilization: a behavioral risk factor surveillance
system analysis. Am J Prev Med. 2019;57(2):250-255.
doi:10.1016/j.amepre.2019.03.015
19. Siu AL; US Preventive Services Task Force.
Behavioral and pharmacotherapy interventions for
tobacco smoking cessation in adults, including
pregnant women: U.S. Preventive Services Task
Force recommendation statement. Ann Intern Med.
2015;163(8):622-634. doi:10.7326/M15-2023
20. Lung CT Screening Reporting & Data System
(Lung-RADS). American College of Radiology.
Accessed January 15, 2021. https://www.acr.org/
Clinical-Resources/Reporting-and-Data-Systems/
Lung-Rads
21. Pinsky PF, Gierada DS, Black W, et al.
Performance of Lung-RADS in the National Lung
Screening Trial: a retrospective assessment. Ann
Intern Med. 2015;162(7):485-491. doi:10.7326/M14-
2086
22. Owens DK, Davidson KW, Krist AH, et al; US
Preventive Services Task Force. Primary care
interventions for prevention and cessation of
tobacco use in children and adolescents: US
Preventive Services Task Force recommendation
statement. JAMA. 2020;323(16):1590-1598. doi:10.
1001/jama.2020.4679
23. Moyer VA; US Preventive Services Task Force.
Screening for lung cancer: U.S. Preventive Services
970 JAMA March 9, 2021 Volume 325, Number 10 (Reprinted)
© 2021 American Medical Association. All rights reserved.
Downloaded From: https://jamanetwork.com/ Mexico | Access Provided by JAMA by Salvador Macias Diaz on 10/21/2021
Task Force recommendation statement. Ann Intern
Med. 2014;160(5):330-338. doi:10.7326/M13-2771
24. Jonas D, Reuland DS, Reddy SM, et al.
Screening for Lung Cancer With Low-Dose
Computed Tomography: An Evidence Review for the
U.S. Preventive Services Task Force. Evidence
Synthesis No. 198. Agency for Healthcare
Research and Quality; 2021. AHRQ publication
20-05266-EF-1.
25. Jonas DE, Reuland DS, Shivani SM, et al.
Screening for lung cancer with low-dose computed
tomography: updated evidence report and
systematic review for the US Preventive Services
Task Force. JAMA. Published March 9, 2021. doi:10.
1001/jama.2021.0377
26. Pinsky PF, Gierada DS, Nath H, Kazerooni EA,
Amorosa J. ROC curves for low-dose CT in the
National Lung Screening Trial. J Med Screen. 2013;
20(3):165-168. doi:10.1177/0969141313500666
27. Yip R, Henschke CI, Yankelevitz DF, Smith JP.
CT screening for lung cancer: alternative definitions
of positive test result based on the National Lung
Screening Trial and International Early Lung Cancer
Action Program databases. Radiology. 2014;273(2):
591-596. doi:10.1148/radiol.14132950
28. Henschke CI, Yip R, Yankelevitz DF, Smith JP;
International Early Lung Cancer Action Program
Investigators. Definition of a positive test result
in computed tomography screening for lung cancer:
a cohort study. Ann Intern Med. 2013;158(4):246-252.
doi:10.7326/0003-4819-158-4-201302190-00004
29. Aldrich MC, Mercaldo SF, Sandler KL, Blot WJ,
Grogan EL, Blume JD. Evaluation of USPSTF lung
cancer screening guidelines among African
American adult smokers. JAMA Oncol. 2019;5(9):
1318-1324. doi:10.1001/jamaoncol.2019.1402
30. Kaplan RC, Bangdiwala SI, Barnhart JM, et al.
Smoking among U.S. Hispanic/Latino adults: the
Hispanic community health study/study of Latinos.
Am J Prev Med. 2014;46(5):496-506. doi:10.1016/j.
amepre.2014.01.014
31. Bandiera FC, Assari S, Livaudais-Toman J,
Pérez-Stable EJ. Latino and Black smokers in the
Health and Retirement Study are more likely to
quit: the role of light smoking. Tob Induc Dis. 2016;
14:23. doi:10.1186/s12971-016-0090-y
32. Pinsky PF. Racial and ethnic differences in lung
cancer incidence: how much is explained by
differences in smoking patterns? (United States).
Cancer Causes Control. 2006;17(8):1017-1024. doi:
10.1007/s10552-006-0038-2
33. Tammemägi MC, Katki HA, Hocking WG, et al.
Selection criteria for lung-cancer screening. N Engl J
Med. 2013;368(8):728-736. doi:10.1056/
NEJMoa1211776
34. Katki HA, Kovalchik SA, Petito LC, et al.
Implications of nine risk prediction models for
selecting ever-smokers for computed tomography
lung cancer screening. Ann Intern Med. 2018;169(1):
10-19. doi:10.7326/M17-2701
35. Bach PB, Kattan MW, Thornquist MD, et al.
Variations in lung cancer risk among smokers. J Natl
Cancer Inst. 2003;95(6):470-478. doi:10.1093/
jnci/95.6.470
USPSTF Evidence Report: Screening for Lung Cancer
36. Lim KP, Marshall H, Tammemägi M, et al; ILST
(International Lung Screening Trial) Investigator
Consortium. Protocol and rationale for the
International Lung Screening Trial. Ann Am Thorac
Soc. 2020;17(4):503-512. doi:10.1513/AnnalsATS.
201902-102OC
37. van Klaveren RJ, Oudkerk M, Prokop M, et al.
Management of lung nodules detected by volume
CT scanning. N Engl J Med. 2009;361(23):2221-2229.
doi:10.1056/NEJMoa0906085
38. Kinsinger LS, Anderson C, Kim J, et al.
Implementation of lung cancer screening in the
Veterans Health Administration. JAMA Intern Med.
2017;177(3):399-406. doi:10.1001/jamainternmed.
2016.9022
39. National Lung Screening Trial Research Team.
Lung cancer incidence and mortality with extended
follow-up in the National Lung Screening Trial.
J Thorac Oncol. 2019;14(10):1732-1742. doi:10.1016/
j.jtho.2019.05.044
40. Mascalchi M, Belli G, Zappa M, et al.
Risk-benefit analysis of X-ray exposure associated
with lung cancer screening in the Italung-CT trial.
AJR Am J Roentgenol. 2006;187(2):421-429. doi:10.
2214/AJR.05.0088
41. Rampinelli C, De Marco P, Origgi D, et al.
Exposure to low dose computed tomography for
lung cancer screening and risk of cancer: secondary
analysis of trial data and risk-benefit analysis. BMJ.
2017;356:j347. doi:10.1136/bmj.j347
42. Nguyen XV, Davies L, Eastwood JD, Hoang JK.
Extrapulmonary findings and malignancies in
participants screened with chest CT in the National
Lung Screening Trial. J Am Coll Radiol. 2017;14(3):
324-330. doi:10.1016/j.jacr.2016.09.044
43. Jaklitsch MT, Jacobson FL, Austin JH, et al. The
American Association for Thoracic Surgery
guidelines for lung cancer screening using low-dose
computed tomography scans for lung cancer
survivors and other high-risk groups. J Thorac
Cardiovasc Surg. 2012;144(1):33-38. doi:10.1016/j.
jtcvs.2012.05.060
44. Lung cancer screening guidelines. American
Cancer Society. Accessed January 15, 2021. https://
www.cancer.org/health-care-professionals/
american-cancer-society-prevention-early-
detection-guidelines/lung-cancer-screening-
guidelines.html
45. Mazzone PJ, Silvestri GA, Patel S, et al.
Screening for lung cancer: CHEST guideline and
expert panel report. Chest. 2018;153(4):954-985.
doi:10.1016/j.chest.2018.01.016
46. NCCN Guidelines for Patients: lung cancer
screening. National Comprehensive Cancer
Network. Published 2020. Accessed January 15,
2021. https://www.nccn.org/patients/guidelines/
content/PDF/lung_screening-patient.pdf
47. Clinical Preventive Service Recommendation:
lung cancer. American Academy of Family
Physicians. Published 2013. Accessed January 15,
2021. https://www.aafp.org/patient-care/clinical-
recommendations/all/lung-cancer.html
jama.com
 VOLUME 36 • NUMBER 11 • APRIL 10, 2018

JOURNAL OF CLINICAL ONCOLOGY O R I G I N A L R E P O R T

Nut Consumption and Survival in Patients With Stage III

Colon Cancer: Results From CALGB 89803 (Alliance)
Temidayo Fadelu, Sui Zhang, Donna Niedzwiecki, Xing Ye, Leonard B. Saltz, Robert J. Mayer, Rex B. Mowat,
Renaud Whittom, Alexander Hantel, Al B. Benson, Daniel M. Atienza, Michael Messino, Hedy L. Kindler, Alan
Venook, Shuji Ogino, Kimmie Ng, Kana Wu, Walter Willett, Edward Giovannucci, Jeffrey Meyerhardt, Ying Bao,
and Charles S. Fuchs

Author affiliations and support information

(if applicable) appear at the end of this
article.
Published at jco.org on February 28, 2018.
J.M., Y.B., and C.S.F. contributed equally
to this work.
Nonfederal sponsors did not participate in
the design or conduct of the study;
collection, management, analysis, or
interpretation of the data; or preparation,
review, or approval of the article. The
views expressed in this article are those of
the authors and not an official position of
their affiliated institutions or funders.
Clinical trial information: NCT00003835.
Corresponding author: Charles S. Fuchs,
MD, Yale Cancer Center, 333 Cedar St,
New Haven, CT 06510; e-mail: charles.
fuchs@yale.edu.
© 2018 by American Society of Clinical
Oncology. Creative Commons Attribution
Non-Commercial No Derivatives 4.0
License.
$ =
A B S T R A C T
Purpose
Observational studies have reported increased colon cancer recurrence and mortality in patients
with states of hyperinsulinemia, including type 2 diabetes, obesity, sedentary lifestyle, and high
glycemic load diet. Nut intake has been associated with a lower risk of type 2 diabetes, metabolic
syndrome, and insulin resistance. However, the effect of nut intake on colon cancer recurrence and
survival is not known.
Patients and Methods
We conducted a prospective, observational study of 826 eligible patients with stage III colon cancer
who reported dietary intake on food frequency questionnaires while enrolled onto a randomized
adjuvant chemotherapy trial. Using Cox proportional hazards regression, we assessed associations
of nut intake with cancer recurrence and mortality.
Results
After a median follow-up of 6.5 years, compared with patients who abstained from nuts, individuals
who consumed two or more servings of nuts per week experienced an adjusted hazard ratio (HR) for
disease-free survival of 0.58 (95% CI, 0.37 to 0.92; Ptrend = .03) and an HR for overall survival of 0.43
(95% CI, 0.25 to 0.74; Ptrend = .01). In subgroup analysis, the apparent benefit was confined to tree
nut intake (HR for disease-free survival, 0.54; 95% CI, 0.34 to 0.85; Ptrend = .04; and HR for overall survival,
0.47; 95% CI, 0.27 to 0.82; Ptrend = .04). The association of total nut intake with improved outcomes was
maintained across other known or suspected risk factors for cancer recurrence and mortality.

0732-183X/18/3611w-1112w/$20.00
Conclusion
Diets with a higher consumption of nuts may be associated with a significantly reduced incidence of
cancer recurrence and death in patients with stage III colon cancer.

J Clin Oncol 36:1112-1120. © 2018 by American Society of Clinical Oncology. Creative Commons
Attribution Non-Commercial No Derivatives 4.0 License: https://creativecommons.org/licenses/by-nc-nd/4.0/

plasma C-peptide or low insulin-like growth

INTRODUCTION
Recent prospective observational studies among
patients with colon cancer suggest that diet and
lifestyle factors may significantly influence the
risk of colon cancer recurrence and death.1-6 In
aggregate, these studies indicate that states of
energy excess, including type 2 diabetes (T2D),7
obesity,6 sedentary lifestyle,2,5 Western-pattern
diet,3 increased dietary glycemic load,1 and
ASSOCIATED CONTENT
high intake of sugar-sweetened beverages,8 are
Appendix
DOI: https://doi.org/10.1200/JCO.
associated with an increased risk of colon
2017.75.5413 cancer recurrence and mortality. Moreover,
DOI: https://doi.org/10.1200/JCO.2017. increased cancer mortality was observed among
75.5413 patients with colorectal cancer with elevated
factor–binding protein-1 levels,9 suggesting
that the association between energy excess and
increased risk of colon cancer recurrence may
be mediated, in part, by long-term hyper-
insulinemia. In fact, increased coffee intake,
which has been associated with decreased risk
of T2D,10-14 lower plasma C-peptide levels,15,16
and increased insulin sensitization,17,18 con-
ferred improved disease-free survival (DFS) in
patients with stage III colon cancer.19
In prospective cohort studies, increased nut
intake has been associated with a reduced risk of
T2D and metabolic syndrome20-23 and a re-
duction in insulin resistance.20,24-27 Nuts are
nutrient-dense foods that are rich in unsaturated

1112 © 2018 by American Society of Clinical Oncology

 Nut Consumption and Stage III Colon Cancer Outcome

fatty acids, fiber, vitamins, minerals, and other bioactive substances

such as phenolic antioxidants and phytosterols.28-30 CALGB 89803
total enrollment
In light of evidence supporting a link between excess energy (N = 1,264)
balance, hyperinsulinemia, and increased recurrence in patients
with colon cancer, we prospectively examined the association of Did not complete Q1 (n = 169)
nut intake with cancer recurrence and mortality in a cohort of Enrolled prior to diet (n = 87)
and lifestyle incorporation
patients with stage III colon cancer enrolled onto a National into CALGB 89803
Cancer Institute–sponsored randomized clinical trial of adjuvant Refused to complete (n = 23)
chemotherapy. In the trial, detailed data on pathologic stage, or did not return Q1
performance status, postoperative treatment, and follow-up were Experienced cancer (n = 59)
recurrence, died, or were
prospectively captured. In addition, comprehensive data on diet removed from CALGB
and lifestyle were collected before any documentation of cancer 89803 before Q1
recurrence.
Completed Q1
(n = 1,095)

PATIENTS AND METHODS

Did not complete Q2
(n = 117)
Study Population
Patients in this prospective cohort participated in the National Cancer
Institute–sponsored Cancer and Leukemia Group B (now Alliance for
Clinical Trials in Oncology) 89803 adjuvant therapy trial for stage III
colon cancer, comparing therapy with weekly fluorouracil and leucovorin to Completed Q1 and Q2
weekly irinotecan, fluorouracil, and leucovorin (ClinicalTrials.gov identifier: (n = 978)
NCT000038350).31 Between April 1999 and May 2001, 1,264 patients were
enrolled. An amendment was introduced after enrolling 87 patients that
required enrollees to complete a self-administered semiquantitative food Excluded: Q2 invalid (n = 111)
frequency questionnaire (FFQ) that captured diet and lifestyle habits. The Caloric intake (Q2; n = 23)
questionnaires were administered midway through adjuvant therapy exclusion* and/or
(4 months after surgery; questionnaire 1 [Q1]) and again 6 months after left > 70 food items blank
completion of treatment (14 months after surgery; questionnaire 2 [Q2]). Experienced cancer (n = 88)
Patients were eligible if they underwent a complete surgical resection recurrence before
completing Q2
of the primary tumor within 56 days before trial entry, had no prior
chemotherapy or radiation therapy for treatment of the tumor, had re-
gional lymph node metastases without evidence of distant metastases, had Valid Q2
(n = 867)
a baseline Eastern Cooperative Oncology Group performance status of 0 to
2,32 and had adequate bone marrow, renal, and hepatic function.31 Because
of possible dietary modifications immediately after colectomy,33 we Excluded: ineligible (n = 41)
for analysis
a priori limited the primary analysis to nut intake reported on Q2. Patients
Did not answer (n = 9)
were excluded if they reported significantly abnormal caloric intake (, 600
both nut-related
or . 4,200 calories per day for men; , 500 or . 3,500 calories per day questions on Q2
for women), left . 70 food items blank, or left blank the nut Experienced cancer (n = 32)
consumption–related questions on Q2. Finally, patients were excluded if recurrence or died within
they had cancer recurrence or death before completion of Q2 or within 60 days after completing Q2
60 days of completing Q2 to avoid misattribution bias. The resulting
cohort included 826 eligible patients (Fig 1). The protocol was reviewed by Final sample
the institutional review board of each participating center, and all patients (N = 826)
provided study-specific informed consent.
Fig 1. Derivation of the study cohort. (*) Calorie exclusion: , 600 or . 4,200
calories per day for men and , 500 calories or . 3,500 calories per day for women.
Dietary Assessment Q1, questionnaire 1 (midway through adjuvant therapy); Q2, questionnaire 2
Patients in this analysis completed semiquantitative FFQs that in- (6 months after completion of adjuvant therapy). CALGB, Cancer and Leukemia
cluded 131 food items, vitamin and mineral supplements, and open-ended Group B.
sections for other supplements and foods not specifically listed.34,35 Par-
ticipants were asked how often, on average, over the previous 3 months they
Total nut intake was calculated as the weighted proportional summation of
consumed a specific food portion size, with up to nine possible responses,
tree nuts and peanuts. We similarly assessed self-reported peanut butter
which ranged from never to to six or more times per day. We computed
intake. In a previous cohort study evaluating the validity of our FFQ in
nutrient intake by multiplying the frequency of consumption of each food by
measuring intake of nuts and peanut butter, the correlation coefficients
the nutrient content of the specified portions.36 Nutrient values were energy
between dietary records and the FFQs were 0.75 and 0.75, respectively.38
adjusted using the residuals methods.37
On the questionnaire, we separately assessed self-reported intake of
1-oz servings of tree nuts and peanuts with the following eight ordered Tumor Assessments for KRAS, BRAF, PIK3CA, and TP53
categories for intake: none; less than once per month; one to three servings Mutations, Microsatellite Instability, and PTGS2 Expression
per month; one serving per week; two to four servings per week; five to six Polymerase chain reaction (PCR) and pyrosequencing targeted for
servings per week; one serving per day; and two or more servings per day. mutation hotspots in PIK3CA exons 9 and 20,39 BRAF codon 600,40 and

jco.org © 2018 by American Society of Clinical Oncology 1113

 Fadelu et al
KRAS codons 12 and 13 were performed, as previously described.41,42 Mu-
tations in TP53 exons 5 to 8 were determined by Sanger sequencing and
sequencing by hybridization, as previously described.43 Microsatellite in-
stability (MSI) was assessed by PCR for 10 microsatellite markers; tumors with
instability in $ 50% of the loci were classified as MSI-high; for 28 patients
without PCR MSI results, those with loss of MLH1 or MSH2 expression were
classified as MSI-high, as previously described.44 PTGS2 (COX-2) expression
was assessed by immunohistochemistry, as previously described.45
Study End Points
The primary end point was DFS, defined as time from completion of
Q2 to tumor recurrence, occurrence of a new primary colon tumor, or
death from any cause. We also assessed recurrence-free survival (RFS),
defined as time from completion of Q2 to tumor recurrence or occurrence
of a new primary colon tumor. For RFS, patients who died without known
tumor recurrence were censored at last documented physician evaluation.
Finally, we assessed overall survival (OS), defined as the time from
completion of Q2 to death from any cause.
Statistical Analysis
There was no statistically significant difference in OS or DFS in both
arms of the randomized trial.31 Therefore, patient data from both arms
were combined and analyzed according to frequency categories of dietary
intake. The primary analysis was done with total nut intake, combining
intake of tree nuts and peanuts. For consistency with prior published
studies on nut intake and to conserve statistical power for the analysis,
categories of nut intake were consolidated into the following five ordered
categories: never, less than one serving per month, one to three servings per
month, one serving per week, and two or more servings per week.46-49
Differences in distribution of baseline characteristics by nut consumption
categories were evaluated using the x2 test, except for continuous variables,
which were evaluated using the Kruskal-Wallis test.
Cox proportional hazards regression analysis was used to determine
the simultaneous affect of other variables potentially associated with each
outcome.50 The covariates in the model were fixed as measured on Q2. To
account for the caloric content of nuts, our initial model adjusted for total
calorie intake.37 The final model further adjusted for potential con-
founders including age, sex, depth of invasion through bowel wall, number
of positive lymph nodes, baseline performance status, treatment group,
body mass index (calculated as weight in kilograms divided by height in
meters squared), physical activity (measured in metabolic equivalent task
hours per week), aspirin use, and glycemic load (as previously described).1
Selection of covariates for the final model was based on clinical signifi-
cance, previous studies, and degree of correlation with the exposure.
Covariates with missing variables were coded with indicator variables.
The statistical analysis was performed using nut consumption as
a continuous measure to minimize bias created by selected categorization.
We tested for linear trends across frequency categories of intake by
assigning each participant the median value for each frequency category
and modeling this value as a continuous variable, consistent with prior
studies.1,2,8,19 P values for trend were calculated using the Wald test of the
score variable. Secondary exploratory analysis was performed individually
for peanuts and tree nuts; we further collapsed intake into four categories
to conserve power (never, , one serving per month, one to three servings
per month, and $ one serving per week). Stratified exploratory analyses
were also performed for other risk factors; the likelihood ratio test was used
to test for interaction. In addition, we performed several sensitivity an-
alyses to confirm the robustness of the results. The Cox regression models
were tested for and met the assumption of proportionality by both time-
dependent covariate and Schoenfeld residuals methods. All analyses were
performed using SAS version 9.4 (SAS Institute, Cary, NC). A threshold
level of significance of P , .05 was considered statistically significant. All
P values are two-sided and were not adjusted for multiple comparisons.
Data collection and management were conducted by the Alliance Statistics
and Data Center. Data quality was ensured by review of data by the Alliance
1114 © 2018 by American Society of Clinical Oncology
Statistics and Data Center and by the study chairperson following Alliance
policies. All analyses were based on the study database frozen on November
9, 2009. See the Appendix (online only) for additional analyses.
RESULTS
Baseline Characteristics
Baseline characteristics by frequency of total nut consumption
are listed in Table 1. Patients who consumed more nuts were more
frequently male, consumed more alcohol, and had a lower glycemic
load diet. Although nut users consumed greater total calories, nut
users did not present with a higher body mass index. There were
statistical differences in age, Western and prudent dietary patterns,
and dietary glycemic load by categories of nut intake, although
clear trends in these factors by nut intake were less apparent.
Impact of Nut Intake on Cancer Recurrence and Death
Median follow-up time from completion of Q2 was 6.5 years.
During follow-up, 199 of the 826 patients experienced cancer
recurrence or developed new primary tumors. Of the 826 patients,
177 patients died; of those, 39 patients died without documented
cancer recurrence.
Increasing total nut intake was associated with significant
reduction in recurrence and mortality after adjusting for other
predictors of cancer recurrence (Table 2). Compared with patients
who abstained, individuals who consumed diets with two or more
servings of nuts per week experienced an adjusted hazard ratio
(HR) for DFS of 0.58 (95% CI, 0.37 to 0.92; Ptrend = .03). Higher
nut intake was also associated with a significant improvement in
OS (HR, 0.43; 95% CI, 0.25 to 0.74; Ptrend = .01) and a trend toward
improved RFS, which did not reach statistical significance (HR,
0.70; 95% CI, 0.42 to 1.16; Ptrend = .15). Unadjusted survival curves
are presented in Appendix Figures A1, A2, and A3 (online only).
We further examined the associations with nut intake models
after controlling for other possible confounders (Western and
prudent dietary patterns, race, smoking, and alcohol use) and
found that our findings were largely unchanged (Appendix Table
A1, online only).
Sensitivity Analyses
Because of possible dietary modifications immediately after
colectomy,33 we a priori limited the primary analysis to nut intake
reported on Q2. Nonetheless, we repeated our analyses using the
cumulative average of nut intake on Q1 and Q2 and found that the
association with DFS was unchanged (Appendix Tables A2 and A3,
online only).
We considered the possibility that changes in dietary habits
could reflect occult cancer or impending death; as such, our
primary analyses excluded patients who developed cancer re-
currence or died within 60 days of completing Q2. To further
address this issue, we repeated the models after excluding re-
currences or deaths within 180 days of Q2 completion (n = 783),
and our results remained largely unchanged (HR, 0.54; 95% CI,
0.32 to 0.89; Ptrend = .02).
Furthermore, we considered that nut intake may be a marker
of dentition and oral health, potential confounders that were not
JOURNAL OF CLINICAL ONCOLOGY
 Nut Consumption and Stage III Colon Cancer Outcome

Table 1. Baseline Characteristics: Total Nut Consumption (N = 826)

Servings of Nuts
, 1 per Month 1-3 per Month 1 per Week $ 2 per Week
Characteristic Never (n = 145) (n = 98) (n = 211) (n = 214) (n = 158) P*
Median nut intake, servings per week (IQR) 0 (0-0) 0.2 (0.2-0.2) 0.4 (0.4-0.4) 1.0 (0.7-1.0) 3.5 (3.0-6.0)
Median age, years (IQR) 64 (54-70) 60 (52-69) 58 (50-67) 59 (51-69) 62 (53-69) .01†
Male 74 (51.0) 46 (46.9) 114 (54.0) 129 (60.3) 103 (65.2) .02
Race .08
White 125 (86.2) 83 (84.7) 188 (89.1) 193 (90.2) 151 (95.6)
Black 11 (7.6) 9 (9.2) 17 (8.1) 10 (4.7) 5 (3.2)
Other 9 (6.2) 6 (6.1) 6 (2.8) 11 (5.1) 2 (1.3)
Performance status‡ .32
0 103 (71.0) 69 (70.4) 155 (73.5) 163 (76.2) 125 (79.1)
1-2 39 (26.9) 29 (29.6) 49 (23.2) 48 (22.4) 31 (19.6)
Unknown 3 (2.1) 0 (0.0) 7 (3.3) 3 (1.4) 2 (1.3)
No. of positive nodes .48
1-3 94 (64.8) 67 (68.4) 139 (65.9) 151 (70.6) 97 (61.4)
$4 48 (33.1) 31 (31.6) 66 (31.3) 60 (28.0) 59 (37.3)
Unknown 3 (2.1) 0 (0.0) 6 (2.8) 3 (1.4) 2 (1.3)
Tumor stage§ .61
T1, 2 29 (20.0) 15 (15.3) 33 (15.6) 32 (15.0) 29 (18.4)
T3, 4 116 (80.0) 83 (84.7) 177 (83.9) 182 (85.0) 129 (81.6)
Unknown 0 (0.0) 0 (0.0) 1 (0.5) 0 (0.0) 0 (0.0)
Differentiation .51
Well/moderate 108 (74.5) 75 (76.5) 154 (73.0) 171 (79.9) 122 (77.2)
Poor/undifferentiated 34 (23.4) 23 (23.5) 50 (23.7) 40 (18.7) 34 (21.5)
Unknown 3 (2.1) 0 (0.0) 7 (3.3) 3 (1.4) 2 (1.3)
Obstruction 32 (22.1) 18 (18.4) 47 (22.3) 50 (23.4) 34 (21.5) .91
Perforation 4 (2.8) 2 (2.0) 8 (3.8) 9 (4.2) 6 (3.8) .87
Treatment .18
FU+LV 82 (56.6) 44 (44.9) 97 (46.0) 115 (53.7) 76 (48.1)
CPT-11+FU+LV 63 (43.4) 54 (55.1) 114 (54.0) 99 (46.3) 82 (51.9)
Median calorie consumption, kcal/d (IQR) 1,624 (1,298-2,034) 1,547 (1,248-2,019) 1,692 (1,305-2,067) 1,755 (1,398-2,205) 2,162 (1,770-2,630) , .001†
Median BMI, kg/m2 (IQR) 28.7 (24.8-32.5) 27.7 (24.4-32.8) 28.7 (25.8-32.7) 28.8 (25.3-33.1) 27.9 (24.5-31.9) .28†
Median physical activity, MET-h/wk (IQR) 6.7 (0.6-17.2) 9.9 (2.3-20.2) 7.5 (1.9-21.5) 7.7 (2.1-21.3) 10.8 (2.7-26.4) .03†
Median caffeine, mg/d (IQR) 140 (30-318) 164 (46-352) 163 (73-356) 160 (46-327) 146 (51-301) .3†
Median alcohol, g/wk (IQR) 0.2 (0.1-14.2) 3.3 (0.1-56.2) 9.7 (0.1-48.9) 9.7 (0.1-71.7) 21.9 (0.1-96.0) , .001†
Median SSB, servings per week (IQR) 0.7 (0.0-3.4) 1.3 (0.4-5.5) 1.2 (0.2-4.5) 1 (0.2-4.0) 1.2 (0.2-4.0) .13†
Never-smokers 57 (39.3) 46 (46.9) 92 (43.6) 104 (48.6) 72 (45.6) .5
Aspirin use 27 (18.6) 8 (8.2) 25 (11.8) 33 (15.4) 28 (17.7) .1
Western diet, No. , median (%) 82 (56.6) 54 (55.1) 112 (53.1) 113 (52.8) 52 (32.9) , .001†
Prudent diet, No. , median (%) 81 (55.9) 63 (64.3) 132 (62.6) 87 (40.7) 51 (32.3) , .001†
Glycemic load, No. , median (%) 52 (35.9) 37 (37.8) 101 (47.9) 112 (52.3) 111 (70.3) , .001†

NOTE. Data presented as No. (%) unless otherwise indicated.

Abbreviations: BMI, body mass index; CPT-11, irinotecan; FU, fluorouracil; IQR, interquartile range; LV, leucovorin; MET, metabolic equivalent task; SSB, sugar-
sweetened beverage.
*By x2 test unless otherwise noted.
†By Kruskal-Wallis nonparametric test.
‡Baseline performance status: 0 indicates fully active; 1 indicates restricted in physically strenuous activity but ambulatory and able to carry out light work; 2 indicates
ambulatory and capable of all self-care but unable to carry out any work activities, up and about . 50% of waking hours.
§T1, 2 indicates level of invasion through the bowel wall not beyond the muscle layer; T3, 4 indicates level of invasion through the bowel wall beyond the muscle layer.

assessed in this cohort. To address this issue, we ran the models
using a new reference group that combined the lowest intake
categories of never and less than one serving per month; the results
remained largely unchanged. Individuals reporting two or more
servings of nuts per week demonstrated an improvement in DFS
compared with the new reference group (HR, 0.57; 95% CI, 0.38 to
0.87; Ptrend = .03).
Stratified Analyses
We evaluated the influence of total nut intake on DFS across
strata of other potential predictors and confounders of patient
outcome, comparing patients with an intake of two of more servings
per week to nonconsumers (Fig 2). In this exploratory analysis, the
association between total nut intake and DFS was consistent across
strata of patient, disease, and treatment characteristics; these in-
cluded strata of clinically relevant genomic alterations (MSI and
KRAS, BRAF, and PIK3CA mutations). However, in these stratified
analyses, statistical power to adequately detect differences was
limited by the sample size, and such analyses should be considered
exploratory.
Subgroup Analyses
In separate analyses of the subtypes of nuts, we observed
a significant improvement in both DFS and OS with increasing tree

jco.org © 2018 by American Society of Clinical Oncology 1115

 Fadelu et al

Table 2. Associations Between Colon Cancer Recurrence and Mortality and Total Nut Consumption
Total Nut Consumption Categories of Intake (servings)
Outcome Never , 1 per Month 1-3 per Month 1 per Week $ 2 per Week Ptrend*
DFS
No. of events/total No. of patients 52/145 36/98 56/211 58/214 36/158
HR (95% CI), adjusted 1† Ref 1.03 (0.68 to 1.58) 0.72 (0.49 to 1.04) 0.71 (0.48 to 1.03) 0.58 (0.37 to 0.90) .02
HR (95% CI), adjusted 2‡ Ref 1.02 (0.66 to 1.57) 0.71 (0.48 to 1.04) 0.69 (0.47 to 1.02) 0.58 (0.37 to 0.92) .03
RFS
No. of events/total No. of patients 39/145 31/98 51/211 47/214 31/158
HR (95% CI), adjusted 1† Ref 1.18 (0.74 to 1.90) 0.88 (0.58 to 1.33) 0.77 (0.50 to 1.18) 0.68 (0.42 to 1.11) .08
HR (95% CI), adjusted 2‡ Ref 1.13 (0.70 to 1.82) 0.83 (0.54 to 1.27) 0.75 (0.48 to 1.15) 0.70 (0.42 to 1.16) .15
OS
No. of events/total No. of patients 44/145 27/98 43/211 39/214 24/158
HR (95% CI), adjusted 1† Ref 0.89 (0.55 to 1.43) 0.64 (0.42 to 0.98) 0.56 (0.36 to 0.86) 0.45 (0.27 to 0.75) .01
HR (95% CI), adjusted 2‡ Ref 0.91 (0.56 to 1.49) 0.64 (0.42 to 0.98) 0.54 (0.34 to 0.84) 0.43 (0.25 to 0.74) .01

Abbreviations: DFS, disease-free survival; HR, hazard ratio; OS, overall survival; Ref, reference; RFS, recurrence-free survival.
*Two-sided P value. Trend across consumption levels with the categorical median.
†Model adjusted 1: Adjusted using Cox proportional hazards regression for calorie intake.
‡Model adjusted 2: Adjusted using Cox proportional hazards regression further for age, sex, depth of invasion through bowel wall, number of positive lymph nodes,
baseline performance status, treatment group, body mass index, physical activity, aspirin use, and glycemic load.

nut consumption, whereas the associations with peanut intake did
not reach statistical significance (Table 3). We similarly examined
the influence of peanut butter consumption on patient outcome
and found no association between peanut butter intake and DFS
(Ptrend = .15), OS (Ptrend = .12), or RFS (Ptrend = .09).
DISCUSSION
In this prospective cohort of patients with stage III colon cancer
who received adjuvant chemotherapy, a diet with increased total
nut intake was associated with a significant improvement in cancer
recurrence or mortality (DFS) and all-cause mortality (OS).
Moreover, these associations seemed to be independent of other
predictors of patient outcome, diet, and lifestyle factors, and the
effect of total nut consumption was maintained across other
known or suspected risk factors of cancer recurrence. In contrast to
tree nuts, we did not observe a significant association with peanuts,
which are legumes, although the statistical power to assess peanut
intake was limited.
To our knowledge, this is the first study to examine the as-
sociation between nut intake and colon cancer recurrence and
survival. Prior observational cohort studies have explored the
relationship between nut intake and risk of developing colorectal
cancer, finding inconsistent results.48,51,52 Other studies have re-
ported reduced cancer-related mortality in association with in-
creased nut intake49,53-55; however, the attributable mortality
reductions in colorectal cancer or other individual cancer types
were not described in these studies.
Our data are consistent with a wealth of existing observational
and clinical trial data reporting health benefits of nut consumption
on many chronic diseases, including reductions in the risk of T2D,
insulin resistance, metabolic syndrome, cardiovascular disease, and
all-cause mortality.49,56-65 Nutrients in nuts, such as unsaturated
fatty acids, high-quality protein, fiber, vitamins, minerals, phy-
tochemicals, and other bioactive substances, may confer potential
anticarcinogenic, anti-inflammatory, and antioxidant properties.28-30
Clinical trials have demonstrated beneficial effects of nuts on
relevant intermediate markers including oxidation,25,66 endothelial
dysfunction,67 hyperglycemia,25 and insulin resistance.20,68 In
addition, preclinical studies suggest that the contents of nuts may
inhibit the growth of a colon cancer cell line69 and may decrease the
rate of colorectal cancer growth and angiogenesis in mice.70
Multiple prior observational studies in patients with colorectal
cancer, including analyses from our study population, suggest that
states of excess energy balance (eg, obesity, T2D, sedentary lifestyle,
Western-pattern diet, greater glycemic load diet, and high sugar-
sweetened beverage intake) are associated with greater risk of
cancer recurrence and mortality.1-3,8 Therefore, we hypothesize
that the effect of nuts on hyperinsulinemia and energy balance
may, in part, explain the association between nut intake and
improved patient outcome in our study. Nonetheless, there may be
other possible mechanisms through which nut intake may influ-
ence survival in patients with colon cancer.
There are several strengths of our analysis. Embedding the
cohort within a randomized trial ensures equal distribution of po-
tential confounders. All patient had stage III colon cancer, minimizing
the effect of heterogeneity of the disease stage on the outcomes.
Treatment and patient follow-up were rigorous because they were
prescribed and monitored by the clinical trial; the outcomes of death
and cancer recurrence were prospectively collected through regular
detailed medical examinations during the course of the follow-up
period. Furthermore, detailed information about potential con-
founders and effect modifiers was collected prospectively.
However, our study has a few limitations. Nut intake was self-
reported, potentially subject to measurement errors, and some-
what limited in range of intake in this patient population. However,
in previous validation studies, nut intake as recorded on our
questionnaire showed good concordance with independent dietary
records.38 Moreover, because dietary data were collected pro-
spectively, any misclassification of nut intake would underestimate
a true association. Given the observational nature of this study, we
cannot exclude the possibility that the associations between nut
intake and improved outcome are a result of confounding vari-
ables or that error in the measurement of adjusted confounding
variables could result in residual confounding. However, these

1116 © 2018 by American Society of Clinical Oncology JOURNAL OF CLINICAL ONCOLOGY

 Nut Consumption and Stage III Colon Cancer Outcome

Subgroup No. of Patients Ptrend Pinter

Age, years .07

< 60 399 .86
≥ 60 427 < .01
Sex .96
Male 466 .07
Female 360 .23
Treatment .35
FU+LV 414 .04
CPT-11+FU+LV 412 .31
Performance status .44
0 630 .14
1-2 196 .09
No. of positive lymph nodes .65
1-3 (N1) 562 .06
≥ 4 (N2) 264 .27
Body mass index .72
2 197 .32
< 25 kg/m
≥ 25 kg/m2 629 .04
Physcial activity .34
< 9 MET-h/wk 449 .03
≥ 9 MET-h/wk 377 .37
Glycemic load .64
< Median 413 .07
≥ Median 413 .38
Aspirin .20
No 705 .13
Yes 121 .05
Microsatellite .17
Stable 504 .19
Unstable 93 .08
KRAS .71
Wild type 289 .13
Mutant 152 .49
BRAF .29
Wild type 378 .23
Mutant 59 .15
PIK3CA .21
Wild type 370 .66
Mutant 50 .14
COX2 expression .80
Low 287 .38
High 135 .41

0 0.5 1 1.5 2 2.5

HR
Fig 2. Multivariable hazard ratios (HRs) and 95% CIs for cancer recurrence or mortality across strata of various factors. The analyses used five categories of total nut
intake (never, less than one serving per month, one to three servings per month, one serving per week, and two or more servings per week). The forest plot represents the
HRs of the comparison of never nut consumers versus consumers of two or more servings of nuts per week, adjusting for calorie intake, age, sex, depth of invasion through
bowel wall, number of positive lymph nodes, baseline performance status, treatment group, body mass index, physical activity, aspirin use, and glycemic load. P values are
two-sided; Pinter indicates P for interaction between strata and nut intake; Ptrend indicates P for trend across levels of nut intake. COX2, cyclooxygenase-2; CPT-11,
irinotecan; FU, fluorouracil; LV, leucovorin; MET-h/wk, metabolic equivalent task hours per week.

associations persisted even after controlling for known or sus-
pected predictors of patient outcome. Furthermore, the associa-
tions remained largely consistent on multiple sensitivity analyses.
Our cohort did not collect data on dentition. Poor dentition
may influence the intake of foods such as nuts, which are hard and
require significant mastication. Observational studies have shown
an association between dental disease and increased incidence and
mortality of cardiovascular disease.71-77 Periodontal disease has
also been linked with a higher incidence of some solid cancers,78-85
although studies on the association with colon cancer incidence
have yielded conflicting results79,82,86 and the effect on colon
cancer outcomes is largely unknown. Our additional sensitivity
analysis to address this potential confounder, collapsing the two
lowest categories into a new control group, yielded consistent

jco.org © 2018 by American Society of Clinical Oncology 1117

 Fadelu et al

Table 3. Associations Between Colon Cancer Recurrence and Mortality by Nut Subtype: Tree Nuts and Peanuts
Categories of Intake (servings)
Outcome Never , 1 per Month 1-3 per Month $ 1 per Week Ptrend*
Tree nuts
DFS
No. of events/total No. of patients 96/267 77/287 40/152 25/120
HR (95% CI), adjusted 1† Ref 0.72 (0.53 to 0.97) 0.66 (0.46 to 0.96) 0.53 (0.34 to 0.84) .02
HR (95% CI), adjusted 2‡ Ref 0.68 (0.51 to 0.93) 0.66 (0.45 to 0.96) 0.54 (0.34 to 0.85) .04
RFS
No. of events/total No. of patients 76/267 71/287 32/152 20/120
HR (95% CI), adjusted 1† Ref 0.84 (0.61 to 1.16) 0.67 (0.44 to 1.02) 0.55 (0.33 to 0.91) .03
HR (95% CI), adjusted 2‡ Ref 0.79 (0.57 to 1.10) 0.66 (0.43 to 1.01) 0.56 (0.33 to 0.94) .06
OS
No. of events/total No. of patients 77/267 55/287 28/152 17/120
HR (95% CI), adjusted 1† Ref 0.65 (0.46 to 0.92) 0.61 (0.39 to 0.94) 0.47 (0.27 to 0.80) .02
HR (95% CI), adjusted 2‡ Ref 0.62 (0.44 to 0.88) 0.58 (0.37 to 0.91) 0.47 (0.27 to 0.82) .04
Peanuts
DFS
No. of events/total No. of patients 70/215 78/280 51/183 39/148
HR (95% CI), adjusted 1† Ref 0.84 (0.61 to 1.16) 0.83 (0.58 to 1.19) 0.78 (0.52 to 1.17) .41
HR (95% CI), adjusted 2‡ Ref 0.87 (0.63 to 1.21) 0.85 (0.59 to 1.23) 0.81 (0.53 to 1.23) .46
RFS
No. of events/total No. of patients 55/215 66/280 44/183 34/148
HR (95% CI), adjusted 1† Ref 0.91 (0.64 to 1.30) 0.92 (0.62 to 1.37) 0.90 (0.58 to 1.40) .76
HR (95% CI), adjusted 2‡ Ref 0.94 (0.66 to 1.35) 0.93 (0.62 to 1.39) 0.97 (0.61 to 1.53) .98
OS
No. of events/total No. of patients 57/215 57/280 37/183 26/148
HR (95% CI), adjusted 1† Ref 0.75 (0.52 to 1.08) 0.73 (0.48 to 1.10) 0.62 (0.39 to 1.01) .15
HR (95% CI), adjusted 2‡ Ref 0.77 (0.53 to 1.12) 0.73 (0.48 to 1.11) 0.60 (0.37 to 0.98) .11

Abbreviations: DFS, disease-free survival; HR, hazard ratio; OS, overall survival; Ref, reference; RFS, recurrence-free survival.
*Two-sided P value. Trend across consumption levels with the categorical median.
†Model adjusted 1: Adjusted using Cox proportional hazards regression for calorie intake.
‡Model adjusted 2: Adjusted using Cox proportional hazards regression further for age, sex, depth of invasion through bowel wall, number of positive lymph nodes,
baseline performance status, treatment group, body mass index, physical activity, aspirin use, and glycemic load.

results. Nonetheless, as with many nutritional epidemiology

studies, the potential for residual confounding remains.
Finally, patients in randomized trials may differ from the
population at large. However, the distribution of dietary and
lifestyle habits reported by our cohort was similar to that of other
US cohorts,87 supporting generalizability of our results. Still, it is
imperative to replicate our findings in other cohorts.
In conclusion, this prospective study of patients with stage III
colon cancer suggests that a diet with increased nut consumption is
associated with a significant reduction in cancer recurrence and
mortality. Although the findings of our observational study do not
establish causality, the results offer further support of the role of
diet and lifestyle as modifiable risk factors of outcomes in patients
with colon cancer.
AUTHORS’ DISCLOSURES OF POTENTIAL CONFLICTS
OF INTEREST
AUTHOR CONTRIBUTIONS
Conception and design: Leonard B. Saltz, Robert J. Mayer, Rex B. Mowat,
Alexander Hantel, Michael Messino, Hedy L. Kindler, Alan Venook, Walter
Willett, Edward Giovannucci, Charles S. Fuchs
Financial support: Shuji Ogino
Administrative support: Robert J. Mayer, Charles S. Fuchs
Provision of study materials or patients: Renaud Whittom, Michael
Messino, Hedy L. Kindler
Collection and assembly of data: Renaud Whittom, Daniel M. Atienza,
Shuji Ogino, Charles S. Fuchs
Data analysis and interpretation: Temidayo Fadelu, Sui Zhang, Donna
Niedzwiecki, Xing Ye, Al B. Benson, Alan Venook, Shuji Ogino, Kimmie
Ng, Kana Wu, Walter Willett, Edward Giovannucci, Jeffrey Meyerhardt,
Ying Bao, Charles S. Fuchs
Manuscript writing: All authors
Final approval of manuscript: All authors
Accountable for all aspects of the work: All authors

Disclosures provided by the authors are available with this article at

jco.org.

REFERENCES
1. Meyerhardt JA, Sato K, Niedzwiecki D, et al:
Dietary glycemic load and cancer recurrence and
survival in patients with stage III colon cancer:
Findings from CALGB 89803. J Natl Cancer Inst 104:
1702-1711, 2012
2. Meyerhardt JA, Heseltine D, Niedzwiecki D, et al:
Impact of physical activity on cancer recurrence and
survival in patients with stage III colon cancer: Findings
from CALGB 89803. J Clin Oncol 24:3535-3541, 2006
3. Meyerhardt JA, Niedzwiecki D, Hollis D, et al:
Association of dietary patterns with cancer re-
currence and survival in patients with stage III colon
cancer. JAMA 298:754-764, 2007
4. Meyerhardt JA, Niedzwiecki D, Hollis D, et al:
Impact of body mass index and weight change after

1118 © 2018 by American Society of Clinical Oncology JOURNAL OF CLINICAL ONCOLOGY

 Nut Consumption and Stage III Colon Cancer Outcome
treatment on cancer recurrence and survival in pa-
tients with stage III colon cancer: Findings from
Cancer and Leukemia Group B 89803. J Clin Oncol
26:4109-4115, 2008
5. Meyerhardt JA, Giovannucci EL, Holmes MD,
et al: Physical activity and survival after colorectal
cancer diagnosis. J Clin Oncol 24:3527-3534, 2006
6. Dignam JJ, Polite BN, Yothers G, et al: Body
mass index and outcomes in patients who receive
adjuvant chemotherapy for colon cancer. J Natl
Cancer Inst 98:1647-1654, 2006
7. Meyerhardt JA, Catalano PJ, Haller DG, et al:
Impact of diabetes mellitus on outcomes in pa-
tients with colon cancer. J Clin Oncol 21:433-440,
2003
8. Fuchs MA, Sato K, Niedzwiecki D, et al: Sugar-
sweetened beverage intake and cancer recurrence
and survival in CALGB 89803 (Alliance). PLoS One 9:
e99816, 2014
9. Wolpin BM, Meyerhardt JA, Chan AT, et al:
Insulin, the insulin-like growth factor axis, and mor-
tality in patients with nonmetastatic colorectal can-
cer. J Clin Oncol 27:176-185, 2009
10. Bhupathiraju SN, Pan A, Manson JE, et al:
Changes in coffee intake and subsequent risk of type
2 diabetes: Three large cohorts of US men and
women. Diabetologia 57:1346-1354, 2014
11. Ding M, Bhupathiraju SN, Chen M, et al:
Caffeinated and decaffeinated coffee consumption
and risk of type 2 diabetes: A systematic review and
a dose-response meta-analysis. Diabetes Care 37:
569-586, 2014
12. Bhupathiraju SN, Pan A, Malik VS, et al: Caf-
feinated and caffeine-free beverages and risk of type
2 diabetes. Am J Clin Nutr 97:155-166, 2013
13. van Dam RM, Hu FB: Coffee consumption and
risk of type 2 diabetes: A systematic review. JAMA
294:97-104, 2005
14. Salazar-Martinez E, Willett WC, Ascherio A,
et al: Coffee consumption and risk for type 2 diabetes
mellitus. Ann Intern Med 140:1-8, 2004
15. Wu T, Willett WC, Hankinson SE, et al: Caf-
feinated coffee, decaffeinated coffee, and caffeine in
relation to plasma C-peptide levels, a marker of in-
sulin secretion, in U.S. women. Diabetes Care 28:
1390-1396, 2005
16. Fung TT, Hu FB, Schulze M, et al: A dietary
pattern that is associated with C-peptide and risk of
colorectal cancer in women. Cancer Causes Control
23:959-965, 2012
17. Wedick NM, Brennan AM, Sun Q, et al: Effects
of caffeinated and decaffeinated coffee on biological
risk factors for type 2 diabetes: A randomized con-
trolled trial. Nutr J 10:93, 2011
18. Williams CJ, Fargnoli JL, Hwang JJ, et al:
Coffee consumption is associated with higher
plasma adiponectin concentrations in women with or
without type 2 diabetes: A prospective cohort study.
Diabetes Care 31:504-507, 2008
19. Guercio BJ, Sato K, Niedzwiecki D, et al:
Coffee intake, recurrence, and mortality in stage III
colon cancer: Results from CALGB 89803 (Alliance).
J Clin Oncol 33:3598-3607, 2015
20. Casas-Agustench P, López-Uriarte P, Bulló M,
et al: Effects of one serving of mixed nuts on serum
lipids, insulin resistance and inflammatory markers in
patients with the metabolic syndrome. Nutr Metab
Cardiovasc Dis 21:126-135, 2011
21. Holligan SD, West SG, Gebauer SK, et al: A
moderate-fat diet containing pistachios improves
emerging markers of cardiometabolic syndrome in
healthy adults with elevated LDL levels. Br J Nutr
112:744-752, 2014
jco.org
22. O’Neil CE, Fulgoni VL III, Nicklas TA: Tree nut
consumption is associated with better adiposity
measures and cardiovascular and metabolic syn-
drome health risk factors in U.S. adults: NHANES
2005-2010. Nutr J 14:64, 2015
23. Ros E: Health benefits of nut consumption.
Nutrients 2:652-682, 2010
24. Hernández-Alonso P, Salas-Salvadó J, Baldrich-
Mora M, et al: Beneficial effect of pistachio con-
sumption on glucose metabolism, insulin resistance,
inflammation, and related metabolic risk markers: A
randomized clinical trial. Diabetes Care 37:3098-3105,
2014
25. Jenkins DJ, Kendall CW, Josse AR, et al: Al-
monds decrease postprandial glycemia, insulinemia,
and oxidative damage in healthy individuals. J Nutr
136:2987-2992, 2006
26. Jiang R, Jacobs DR Jr, Mayer-Davis E, et al:
Nut and seed consumption and inflammatory
markers in the multi-ethnic study of atherosclerosis.
Am J Epidemiol 163:222-231, 2006
27. Sweazea KL, Johnston CS, Ricklefs KD, et al:
Almond supplementation in the absence of dietary
advice significantly reduces C-reactive protein in
subjects with type 2 diabetes. J Funct Foods 10:
252-259, 2014
28. Kris-Etherton PM, Hu FB, Ros E, et al: The role
of tree nuts and peanuts in the prevention of coronary
heart disease: Multiple potential mechanisms. J Nutr
138:1746S-1751S, 2008
29. González CA, Salas-Salvadó J: The potential of
nuts in the prevention of cancer. Br J Nutr 96:
S87-S94, 2006 (suppl 2)
30. Pfundstein B, Haubner R, Würtele G, et al:
Pilot walnut intervention study of urolithin bio-
availability in human volunteers. J Agric Food Chem
62:10264-10273, 2014
31. Saltz LB, Niedzwiecki D, Hollis D, et al: Irino-
tecan fluorouracil plus leucovorin is not superior to
fluorouracil plus leucovorin alone as adjuvant treat-
ment for stage III colon cancer: Results of CALGB
89803. J Clin Oncol 25:3456-3461, 2007
32. Zubrod CG, Schneiderman M, Frei E, et al:
Appraisal of methods for the study of chemotherapy
of cancer in man: Comparative therapeutic trial of
nitrogen mustard and triethylene thiophosphor-
amide. J Chronic Dis 11:7-33, 1960
33. Sturrock PR, Maykel JA: Postoperative
management, in colorectal surgery. https://www-
clinicalkey-com.ezp-prod1.hul.harvard.edu/#!/content/
book/3-s2.0-B9781437717242000040?scrollTo=%
23hl0000956
34. Willett WC, Reynolds RD, Cottrell-Hoehner S,
et al: Validation of a semi-quantitative food frequency
questionnaire: Comparison with a 1-year diet record.
J Am Diet Assoc 87:43-47, 1987
35. Willett WC, Sampson L, Stampfer MJ, et al:
Reproducibility and validity of a semiquantitative food
frequency questionnaire. Am J Epidemiol 122:51-65,
1985
36. US Department of Agriculture: Composition of
Foods: Raw, Processed, and Prepared, 1963-1988.
Washington, DC, Department of Agriculture, US
Government Printing Office, 1989
37. Willett W, Stampfer MJ: Total energy intake:
Implications for epidemiologic analyses. Am J Epi-
demiol 124:17-27, 1986
38. Salvini S, Hunter DJ, Sampson L, et al: Food-
based validation of a dietary questionnaire: The ef-
fects of week-to-week variation in food consumption.
Int J Epidemiol 18:858-867, 1989
39. Ogino S, Liao X, Imamura Y, et al: Predictive
and prognostic analysis of PIK3CA mutation in stage
III colon cancer intergroup trial. J Natl Cancer Inst
105:1789-1798, 2013
40. Ogino S, Shima K, Meyerhardt JA, et al:
Predictive and prognostic roles of BRAF mutation
in stage III colon cancer: Results from intergroup
trial CALGB 89803. Clin Cancer Res 18:890-900,
2012
41. Ogino S, Kawasaki T, Brahmandam M, et al:
Sensitive sequencing method for KRAS mutation
detection by pyrosequencing. J Mol Diagn 7:
413-421, 2005
42. Ogino S, Meyerhardt JA, Irahara N, et al: KRAS
mutation in stage III colon cancer and clinical out-
come following intergroup trial CALGB 89803. Clin
Cancer Res 15:7322-7329, 2009
43. Warren RS, Atreya CE, Niedzwiecki D, et al:
Association of TP53 mutational status and gender
with survival after adjuvant treatment for stage III
colon cancer: Results of CALGB 89803. Clin Cancer
Res 19:5777-5787, 2013
44. Bertagnolli MM, Niedzwiecki D, Compton CC,
et al: Microsatellite instability predicts improved re-
sponse to adjuvant therapy with irinotecan, fluoro-
uracil, and leucovorin in stage III colon cancer: Cancer
and Leukemia Group B protocol 89803. J Clin Oncol
27:1814-1821, 2009
45. Chan AT, Ogino S, Fuchs CS: Aspirin and the
risk of colorectal cancer in relation to the ex-
pression of COX-2. N Engl J Med 356:2131-2142,
2007
46. Bao Y, Hu FB, Giovannucci EL, et al: Nut
consumption and risk of pancreatic cancer in women.
Br J Cancer 109:2911-2916, 2013
47. Wang W, Yang M, Kenfield SA, et al: Nut
consumption and prostate cancer risk and mortality.
Br J Cancer 115:371-374, 2016
48. Yang M, Hu FB, Giovannucci EL, et al: Nut
consumption and risk of colorectal cancer in women.
Eur J Clin Nutr 70:333-337, 2016
49. Bao Y, Han J, Hu FB, et al: Association of nut
consumption with total and cause-specific mortality.
N Engl J Med 369:2001-2011, 2013
50. Cox DR: Regression models and life-tables.
J R Stat Soc B 34:187-220, 1972
51. Yeh C-C, You S-L, Chen C-J, et al: Peanut
consumption and reduced risk of colorectal cancer in
women: A prospective study in Taiwan. World J
Gastroenterol 12:222-227, 2006
52. Jenab M, Ferrari P, Slimani N, et al: Associa-
tion of nut and seed intake with colorectal cancer risk
in the European Prospective Investigation into Can-
cer and Nutrition. Cancer Epidemiol Biomarkers Prev
13:1595-1603, 2004
53. Guasch-Ferré M, Bulló M, Martı́nez-González
MÁ, et al: Frequency of nut consumption and mor-
tality risk in the PREDIMED nutrition intervention trial.
BMC Med 11:164, 2013
54. Fernández-Montero A, Bes-Rastrollo M, Barrio-
López MT, et al: Nut consumption and 5-y all-cause
mortality in a Mediterranean cohort: The SUN project.
Nutrition 30:1022-1027, 2014
55. Bonaccio M, Di Castelnuovo A, De Curtis A,
et al: Nut consumption is inversely associated with
both cancer and total mortality in a Mediterranean
population: Prospective results from the Moli-sani
study. Br J Nutr 114:804-811, 2015
56. Jiang R, Manson JE, Stampfer MJ, et al: Nut
and peanut butter consumption and risk of type 2
diabetes in women. JAMA 288:2554-2560, 2002
57. Villegas R, Gao Y-T, Yang G, et al: Legume and
soy food intake and the incidence of type 2 diabetes
in the Shanghai Women’s Health Study. Am J Clin
Nutr 87:162-167, 2008
© 2018 by American Society of Clinical Oncology 1119
 Fadelu et al

58. Pan A, Sun Q, Manson JE, et al: Walnut
consumption is associated with lower risk of type 2
diabetes in women. J Nutr 143:512-518, 2013
59. Fernández-Montero A, Bes-Rastrollo M,
Beunza JJ, et al: Nut consumption and incidence of
metabolic syndrome after 6-year follow-up: The SUN
(Seguimiento Universidad de Navarra, University of
Navarra Follow-up) cohort. Public Health Nutr 16:
2064-2072, 2013
60. Singh PN, Fraser GE: Dietary risk factors for
colon cancer in a low-risk population. Am J Epidemiol
148:761-774, 1998
61. Djoussé L, Rudich T, Gaziano JM: Nut con-
sumption and risk of hypertension in US male phy-
sicians. Clin Nutr 28:10-14, 2009
62. Tsai C-J, Leitzmann MF, Hu FB, et al: A pro-
spective cohort study of nut consumption and the
risk of gallstone disease in men. Am J Epidemiol 160:
961-968, 2004
63. Tsai C-J, Leitzmann MF, Hu FB, et al: Frequent
nut consumption and decreased risk of cholecys-
tectomy in women. Am J Clin Nutr 80:76-81, 2004
64. Strate LL, Liu YL, Syngal S, et al: Nut, corn, and
popcorn consumption and the incidence of di-
verticular disease. JAMA 300:907-914, 2008
65. Gopinath B, Buyken AE, Flood VM, et al:
Consumption of polyunsaturated fatty acids, fish, and
nuts and risk of inflammatory disease mortality. Am J
Clin Nutr 93:1073-1079, 2011
66. Torabian S, Haddad E, Rajaram S, et al: Acute
effect of nut consumption on plasma total poly-
phenols, antioxidant capacity and lipid peroxidation.
J Hum Nutr Diet 22:64-71, 2009
67. Ma Y, Njike VY, Millet J, et al: Effects of walnut
consumption on endothelial function in type 2 di-
abetic subjects: A randomized controlled crossover
trial. Diabetes Care 33:227-232, 2010
68. Tapsell LC, Batterham MJ, Teuss G, et al:
Long-term effects of increased dietary poly-
unsaturated fat from walnuts on metabolic parame-
ters in type II diabetes. Eur J Clin Nutr 63:1008-1015,
2009
69. Lux S, Scharlau D, Schlörmann W, et al: In vitro
fermented nuts exhibit chemopreventive effects in
HT29 colon cancer cells. Br J Nutr 108:1177-1186,
2012
70. Nagel JM, Brinkoetter M, Magkos F, et al:
Dietary walnuts inhibit colorectal cancer growth in
mice by suppressing angiogenesis. Nutrition 28:
67-75, 2012
71. DeStefano F, Anda RF, Kahn HS, et al: Dental
disease and risk of coronary heart disease and
mortality. BMJ 306:688-691, 1993
72. Joshipura KJ, Hung H-C, Rimm EB, et al:
Periodontal disease, tooth loss, and incidence of is-
chemic stroke. Stroke 34:47-52, 2003
73. Jimenez M, Krall EA, Garcia RI, et al: Peri-
odontitis and incidence of cerebrovascular disease in
men. Ann Neurol 66:505-512, 2009
74. Watt RG, Tsakos G, de Oliveira C, et al: Tooth
loss and cardiovascular disease mortality risk: Re-
sults from the Scottish Health Survey. PLoS One 7:
e30797, 2012
75. Desvarieux M, Demmer RT, Rundek T, et al:
Relationship between periodontal disease, tooth
loss, and carotid artery plaque: The Oral Infections
and Vascular Disease Epidemiology Study (INVEST).
Stroke 34:2120-2125, 2003
76. Dietrich T, Jimenez M, Krall Kaye EA, et al:
Age-dependent associations between chronic
periodontitis/edentulism and risk of coronary heart
disease. Circulation 117:1668-1674, 2008
77. Tu Y-K, Galobardes B, Smith GD, et al: As-
sociations between tooth loss and mortality patterns
in the Glasgow Alumni Cohort. Heart 93:1098-1103,
2007
78. Migliorati CA: Periodontal diseases and can-
cer. Lancet Oncol 9:510-512, 2008
79. Michaud DS, Liu Y, Meyer M, et al: Periodontal
disease, tooth loss, and cancer risk in male health
professionals: A prospective cohort study. Lancet
Oncol 9:550-558, 2008
80. Abnet CC, Qiao Y-L, Mark SD, et al: Pro-
spective study of tooth loss and incident esophageal
and gastric cancers in China. Cancer Causes Control
12:847-854, 2001
81. Garrote LF, Herrero R, Reyes RMO, et al: Risk
factors for cancer of the oral cavity and oro-pharynx in
Cuba. Br J Cancer 85:46-54, 2001
82. Hiraki A, Matsuo K, Suzuki T, et al: Teeth loss and
risk of cancer at 14 common sites in Japanese. Cancer
Epidemiol Biomarkers Prev 17:1222-1227, 2008
83. Abnet CC, Kamangar F, Dawsey SM, et al:
Tooth loss is associated with increased risk of gastric
non-cardia adenocarcinoma in a cohort of Finnish
smokers. Scand J Gastroenterol 40:681-687, 2005
84. Boylan MR, Khalili H, Huang ES, et al: A pro-
spective study of periodontal disease and risk of
gastric and duodenal ulcer in male health pro-
fessionals. Clin Transl Gastroenterol 5:e49, 2014
85. Hujoel PP, Drangsholt M, Spiekerman C, et al:
An exploration of the periodontitis-cancer associa-
tion. Ann Epidemiol 13:312-316, 2003
86. Momen-Heravi F, Babic A, Tworoger SS, et al:
Periodontal disease, tooth loss and colorectal cancer
risk: Results from the Nurses’ Health Study. Int J
Cancer 140:646-652, 2017
87. Michaud DS, Fuchs CS, Liu S, et al: Dietary
glycemic load, carbohydrate, sugar, and colorectal
cancer risk in men and women. Cancer Epidemiol
Biomarkers Prev 14:138-147, 2005

Affiliations
Temidayo Fadelu, Sui Zhang, Robert J. Mayer, Shuji Ogino, Kimmie Ng, Jeffrey Meyerhardt, and Charles S. Fuchs, Dana-Farber/
Partners CancerCare; Shuji Ogino, Kana Wu, Walter Willett, and Edward Giovannucci, Harvard TH Chan School of Public Health;
Shuji Ogino, Edward Giovannucci, and Ying Bao, Brigham and Women’s Hospital, and Harvard Medical School, Boston, MA; Donna
Niedzwiecki and Xing Ye, Alliance Statistics and Data Center, Duke University, Durham; Michael Messino, Southeast Clinical Oncology
Research Consortium, Mission Hospitals, Asheville, NC; Leonard B. Saltz, Memorial Sloan Kettering Cancer Center, New York, NY; Rex B.
Mowat, Toledo Community Hospital Oncology Program, Toledo, OH; Renaud Whittom, Hôpital du Sacré-Coeur de Montréal, Montreal,
Quebec, Canada; Alexander Hantel, Loyola University Stritch School of Medicine, Naperville; Al B. Benson, Robert H Lurie Comprehensive
Cancer Center, Northwestern University; Hedy L. Kindler, University of Chicago Comprehensive Cancer, Chicago, IL; Daniel M. Atienza,
Virginia Oncology Associates, Norfolk, VA; Alan Venook, University of California at San Francisco Comprehensive Cancer Center, San
Francisco, CA; and Charles S. Fuchs, Yale Cancer Center, Yale School of Medicine, New Haven, CT.
Support
Supported by the National Cancer Institute of the National Institutes of Health under Awards No. U10CA180821 and U10CA180882
(to Alliance for Clinical Trials in Oncology), U10CA180820 (to Eastern Cooperative Oncology Group–American College of Radiology
Imaging Network), and U10CA180888 (to Southwest Oncology Group). Also supported by Grants No. U10CA60138, U10CA138561,
U10CA180791, U10CA180836, U10CA180867, and UG1CA189858, and the Pharmacia & Upjohn Company, now Pfizer Oncology. C.S.F.,
K.N., J.M., D.N., and X.Y. are supported in part by grants from the National Cancer Institute (Grants No. K07 CA148894, R01 CA118553,
R01 CA149222, R01 CA169141, and P50 CA127003). Y.B. is supported in part by Grant No. P30 DK046200 from the National Institutes of
Health and the International Tree Nut Council Nutrition Research and Education Foundation. S.O. is supported in part by Grant No. R35
CA197735 from the National Institutes of Health.

nnn

1120 © 2018 by American Society of Clinical Oncology JOURNAL OF CLINICAL ONCOLOGY

 Nut Consumption and Stage III Colon Cancer Outcome

AUTHORS’ DISCLOSURES OF POTENTIAL CONFLICTS OF INTEREST

Nut Consumption and Survival in Patients With Stage III Colon Cancer: Results From CALGB 89803 (Alliance)
The following represents disclosure information provided by authors of this manuscript. All relationships are considered compensated. Relationships are
self-held unless noted. I = Immediate Family Member, Inst = My Institution. Relationships may not relate to the subject matter of this manuscript. For more
information about ASCO’s conflict of interest policy, please refer to www.asco.org/rwc or ascopubs.org/jco/site/ifc.
Temidayo Fadelu Daniel M. Atienza
No relationship to disclose Employment: Virginia Oncology Associates
Sui Zhang Michael Messino
No relationship to disclose No relationship to disclose
Donna Niedzwiecki Hedy L. Kindler
No relationship to disclose Consulting or Advisory Role: Aduro Biotech, MedImmune, Bayer,
Genentech, Verastem, Celgene, GlaxoSmithKline, AstraZeneca, Merck
Xing Ye Research Funding: Aduro Biotech, AstraZeneca, Bayer, GlaxoSmithKline,
No relationship to disclose Merck, MedImmune, Verastem, Bristol-Myers Squibb, Eli Lilly
Leonard B. Saltz Expert Testimony: Aduro Biotech
Consulting or Advisory Role: Eli Lilly, McNeil PPC (I), Abbvie Alan Venook
Research Funding: Taiho Pharmaceutical Consulting or Advisory Role: Genentech
Robert J. Mayer Research Funding: Genentech (Inst), Bristol-Myers Squibb (Inst)
Honoraria: Taiho Pharmaceutical Patents, Royalties, Other Intellectual Property: Royalties from Now-
Consulting or Advisory Role: CASI Pharmaceuticals UptoDate for authoring and maintaining two chapters
Travel, Accommodations, Expenses: Genentech, Taiho Pharmaceutical,
Rex B. Mowat Roche
No relationship to disclose
Shuji Ogino
Renaud Whittom No relationship to disclose
Honoraria: AstraZeneca
Consulting or Advisory Role: Boehringer Ingelheim, AstraZeneca, Pfizer, Kimmie Ng
Bristol-Myers Squibb Consulting or Advisory Role: Bayer, Tarrex, Genentech, Eli Lilly
Research Funding: Astellas (Inst), Genentech (Inst) Research Funding: Gilead Sciences, Tarrex, Genentech (Inst)
Travel, Accommodations, Expenses: Seattle Genetics, Celgene Kana Wu
Alexander Hantel No relationship to disclose
Travel, Accommodations, Expenses: Genentech Walter Willett
Al B. Benson No relationship to disclose
Consulting or Advisory Role: Genentech, Bristol-Myers Squibb, Merck Edward Giovannucci
Serono, Celgene, Vicus Therapeutics, Pharmacyclics, Precision No relationship to disclose
Therapeutics, Taiho Pharmaceutical, Bayer, Alchemia, Infinity
Pharmaceuticals, Boehringer Ingelheim, Astellas Pharma, EMD Serono, Jeffrey Meyerhardt
IntegraGen, Guardant Health, Opsona Therapeutics, Lexicon, Novartis, Honoraria: Chugai Pharma
Boston Biomedical, Helsinn, Guerbet, Eli Lilly, Spectrum Pharmaceuticals, Consulting or Advisory Role: Genentech (Inst)
Advanced Accelerator Applications, Bayer/Onyx, TRM Oncology, DAVA
Pharmaceuticals, Sanofi, Eli Lilly/ImClone Ying Bao
Research Funding: Gilead Sciences (Inst), Amgen (Inst), Astellas Pharma No relationship to disclose
(Inst), Advanced Accelerator Applications (Inst), Bayer/Onyx (Inst), Charles S. Fuchs
Novartis (Inst), Alchemia (Inst), AVEO (Inst), Infinity Pharmaceuticals Leadership: CytomX Therapeutics
(Inst), Genentech (Inst), Merck Sharp & Dohme (Inst), Taiho Consulting or Advisory Role: Eli Lilly, Sanofi, Merck, Entrinsic Health,
Pharmaceutical (Inst) Five Prime Therapeutics, Agios, Gilead Sciences, Merrimack, Taiho
Travel, Accommodations, Expenses: Genentech, Eli Lilly/ImClone, Bayer, Pharmaceutical, KEW Group, Genentech
Sanofi, Spectrum Pharmaceuticals, AVEO, Gilead Sciences, Astellas
Pharma, Boehringer Ingelheim, DAVA Oncology, TRM Oncology,
Guardant Health, Helsinn, Guerbet, Boston Biomedical

jco.org © 2018 by American Society of Clinical Oncology

 Fadelu et al

Appendix

Methods
In supplemental analyses, we offer additional multivariable models using Cox proportional hazards regression, whereby
additional covariates were added as part of sensitivity analyses. In successive models, Western and prudent dietary patterns (model 3)
and race (as a categorical variable: white, black, or other), smoking as binary (ever or never), and alcohol (as a continuous variable in
grams per day; model 4) were added to the final model that included age, sex, depth of invasion through bowel wall, number of positive
lymph nodes, baseline performance status, treatment group, body mass index (calculated as weight in kilograms divided by height in
meters squared), physical activity (measured in metabolic equivalent task hours per week), aspirin use, and glycemic load. Glycemic
load and Western and prudent dietary patterns were determined by methods previously described by our study cohort (Wu K, et al:
Cancer Causes Control 15:853-862, 2004; Fung T, et al: Arch Intern Med 163:309-314, 2003).1,3
Because of possible dietary modifications immediately after colectomy, we a priori limited the primary analysis to nut intake
reported on questionnaire 2 (Q2). In secondary analyses, we repeated the models using cumulative nut intake from questionnaire 1
(Q1) and Q2. Cumulative nut exposure average was calculated based on weighting proportional to the time between Q1 and Q2,
and then the time between Q2 and the outcome survival period, using previously published methods (Hu FB, et al: Am J Epidemiol
149:531-540, 1999).1 In this analysis, the follow-up period began from the completion of Q1. Consistent with previous analyses in
our study cohort, patients were excluded if they experienced cancer recurrence or death within 90 days of completing Q1 to prevent
misattribution bias.1,8,18 Survival estimate curves for disease-free survival, recurrence-free survival, and overall survival by nut
intake categories were calculated using the Kaplan-Meier methods, and the log-rank testing was used to compare survival between
groups (Kaplan EL, Meier P: J Am Stat Assoc 53:457-481, 1958).

1.0 +
Censor
0.9 Log-rank P = .0377

0.8
0.7
0.6
DFS

0.5
0.4 Nuts
Never
0.3 < 1 per month
0.2 1-3 per month
1 per week
0.1
≥ 2 per week

0 1 2 3 4 5 6 7 8
No. at risk Time (years)
Never 145 124 112 100 91 80 66 19 1
< 1 per month 98 87 70 68 62 55 46 4 0
1-3 per month 211 184 165 152 144 138 115 28 0
1 per week 214 196 174 166 150 139 117 22 0
≥ 2 per week 158 137 128 126 121 108 97 25 1

Fig A1. Kaplan-Meier curves for disease-free survival (DFS) by total nut con-
sumption. Curves depict survival for five categories of total nut intake (never, less
than one serving per month, one to three servings per month, one serving per
week, and two or more servings per week).

© 2018 by American Society of Clinical Oncology JOURNAL OF CLINICAL ONCOLOGY

 Nut Consumption and Stage III Colon Cancer Outcome

1.0 +
Censor
0.9 Log-rank P = .1782

0.8
0.7
0.6

RFS
0.5
0.4 Nuts
Never
0.3
< 1 per month
0.2 1-3 per month
1 per week
0.1
≥ 2 per week

0 1 2 3 4 5 6 7 8

No. at risk Time (years)

Never 145 124 112 100 91 80 66 19 1
< 1 per month 98 87 70 68 62 55 46 4 0
1-3 per month 211 184 165 152 144 138 115 28 0
1 per week 214 196 174 165 149 139 117 22 0
≥ 2 per week 158 137 128 126 121 108 97 25 1

Fig A2. Kaplan-Meier curves for recurrence-free survival (RFS) by total nut
consumption. Curves depict survival for five categories of total nut intake (never,
less than one serving per month, one to three servings per month, one serving per
week, and two or more servings per week).

+
Censor
1.0 Log-rank P = .0052
0.9
0.8
0.7
0.6
OS

0.5
0.4 Nuts
Never
0.3
< 1 per month
0.2 1-3 per month
0.1 1 per week
≥ 2 per week

0 1 2 3 4 5 6 7 8

No. at risk Time (years)

Never 145 140 127 118 106 89 71 21 1
< 1 per month 98 96 87 81 76 67 53 6 0
1-3 per month 211 203 186 173 162 151 124 31 1
1 per week 214 208 198 185 169 155 125 27 1
≥ 2 per week 158 151 147 141 137 122 105 25 1

Fig A3. Kaplan-Meier curves for overall survival (OS) by total nut consumption. Curves
depict survival for five categories of total nut intake (never, less than one serving per
month, one to three servings per month, one serving per week, and two or more
servings per week).

jco.org © 2018 by American Society of Clinical Oncology

 Fadelu et al

Table A1. Sensitivity Analyses Further Controlling for Additional Confounders: Dietary Pattern, Race, Smoking, and Alcohol Use
Total Nut Consumption Categories of Intake (servings)
Outcome Never , 1 per Month 1-3 per Month 1 per Week $ 2 per Week Ptrend*
DFS
No. of events/total No. of patients 52/145 36/98 56/211 58/214 36/158
HR (95% CI), model 3† Ref 1.03 (0.67 to 1.59) 0.71 (0.48 to 1.05) 0.71 (0.48 to 1.04) 0.60 (0.38 to 0.95) .05
HR (95% CI), model 4‡ Ref 1.02 (0.66 to 1.57) 0.69 (0.47 to 1.02) 0.69 (0.47 to 1.02) 0.59 (0.37 to 0.93) .05
RFS
No. of events/total No. of patients 39/145 31/98 51/211 47/214 31/158
HR (95% CI), model 3† Ref 1.15 (0.71 to 1.86) 0.84 (0.55 to 1.29) 0.76 (0.49 to 1.18) 0.72 (0.43 to 1.20) .18
HR (95% CI), model 4‡ Ref 1.13 (0.70 to 1.82) 0.82 (0.53 to 1.25) 0.75 (0.48 to 1.16) 0.70 (0.42 to 1.17) .16
OS
No. of events/total No. of patients 44/145 27/98 43/211 39/214 24/158
HR (95% CI), model 3† Ref 0.94 (0.58 to 1.54) 0.65 (0.42 to 0.99) 0.55 (0.35 to 0.85) 0.45 (0.26 to 0.78) .01
HR (95% CI), model 4‡ Ref 0.92 (0.56 to 1.50) 0.64 (0.41 to 0.98) 0.54 (0.34 to 0.85) 0.46 (0.26 to 0.78) .01

Abbreviations: BMI, body mass index; DFS, disease-free survival; HR, hazard ratio; OS, overall survival; Q2, questionnaire 2; Ref, reference; RFS, recurrence-free
survival.
*Two-sided P value. Trend across consumption levels with the categorical median.
†Model 3: Cox proportional hazards regression adjusting for age, sex, depth of invasion through bowel wall, number of positive lymph nodes, baseline performance
status, treatment group, and aspirin use (Q2), BMI (Q2), physical activity (Q2), dietary glycemic load (Q2), calorie intake (Q2), Western dietary pattern (Q2), and prudent
dietary pattern.
‡Model 4: Cox proportional hazards regression adjusting for age, sex, depth of invasion through bowel wall, number of positive lymph nodes, baseline performance
status, treatment group, and aspirin use (Q2), BMI (Q2), physical activity (Q2), glycemic load (Q2), and calorie intake (Q2), plus race (white, black, or other), smoking (never
or ever), and alcohol (grams per day, Q2).

Table A2. Consumption of Nuts on Q1 and Q2 in Servings per Day

Nut Type Q1 Mean (95% CI) Q1 Median Q2 Mean (95% CI) Q2 Median P
Peanuts 0.09 (0.08 to 0.11) 0.03 0.10 (0.09 to 0.11) 0.03 , .001
Tree nuts 0.07 (0.05 to 0.08) 0.03 0.08 (0.07 to 0.09) 0.03 , .001
Peanut butter 0.28 (0.25 to 0.32) 0.07 0.23 (0.21 to 0.26) 0.07 , .001

NOTE. P value generated from the paired Wilcoxon test for non-normality.
Abbreviations: Q1, questionnaire 1 (midway through adjuvant therapy); Q2, questionnaire 2 (6 months after completion of adjuvant therapy).

Table A3. Association Between Colon Cancer Recurrence and Mortality by Cumulative Average Total Nut Consumption of Q1 and Q2
Total Nut Consumption Categories of Intake (servings)
Outcome Never , 1 per Month 1-3 per Month 1 per Week $ 2 per Week Ptrend*
DFS
No. of events/total No. of patients 89/152 43/117 99/260 81/251 81/236
HR (95% CI), model 1† Ref 0.49 (0.34 to 0.70) 0.51 (0.39 to 0.69) 0.41 (0.30 to 0.56) 0.44 (0.32 to 0.60) .008
HR (95% CI), model 2‡ Ref 0.53 (0.36 to 0.76) 0.51 (0.38 to 0.68) 0.41 (0.30 to 0.56) 0.45 (0.33 to 0.62) .01
RFS
No. of events/total No. of patients 78/152 39/117 90/260 71/251 71/236
HR (95% CI), model 1† Ref 0.52 (0.35 to 0.76) 0.55 (0.40 to 0.74) 0.42 (0.30 to 0.58) 0.45 (0.33 to 0.63) .01
HR (95% CI), model 2‡ Ref 0.54 (0.36 to 0.79) 0.53 (0.39 to 0.73) 0.42 (0.30 to 0.58) 0.46 (0.32 to 0.64) .02
OS
No. of events/total No. of patients 74/152 32/117 78/260 62/251 64/236
HR (95% CI), model 1† Ref 0.44 (0.29 to 0.66) 0.49 (0.36 to 0.68) 0.38 (0.27 to 0.54) 0.43 (0.30 to 0.60) .02
HR (95% CI), model 2‡ Ref 0.48 (0.32 to 0.73) 0.48 (0.34 to 0.66) 0.38 (0.26 to 0.53) 0.43 (0.30 to 0.61) .03

NOTE. Survival time starts at Q1, and patients with events within 90 days of Q1 were excluded.
Abbreviations: DFS, disease-free survival; HR, hazard ratio; OS, overall survival; Q1, questionnaire 1 (midway through adjuvant therapy); Q2, questionnaire 2 (6 months
after completion of adjuvant therapy); Ref, reference; RFS, recurrence-free survival.
*Two-sided P value. Trend across consumption levels with the categorical median.
†Model 1: Cox proportional hazards regression adjusting for time-varying calorie consumption.
‡Model 2: Cox proportional hazards regression adjusting for age, sex, depth of invasion through bowel wall, number of positive lymph nodes, baseline performance
status, treatment group, regular aspirin use (both Q1 and Q2), and time-varying calorie consumption, body mass index, physical activity, and glycemic load.

© 2018 by American Society of Clinical Oncology JOURNAL OF CLINICAL ONCOLOGY

 EUROPEAN UROLOGY 70 (2016) 974–982

available at www.sciencedirect.com
journal homepage: www.europeanurology.com

Platinum Priority – Prostate Cancer

Editorial by Sarah K. Holt, George R. Schade and John L. Gore on pp. 983–984 of this issue

Ejaculation Frequency and Risk of Prostate Cancer:

Updated Results with an Additional Decade of Follow-up

Jennifer R. Ridera,b,1,*, Kathryn M. Wilson a,c,1, Jennifer A. Sinnott a,d, Rachel S. Kelly c,
Lorelei A. Mucci a,c, Edward L. Giovannucci a,c,e
a
Department of Epidemiology, Harvard T.H. Chan School of Public Health, Boston, MA, USA; b Department of Epidemiology, Boston University School of Public
Health, Boston, MA, USA; c Channing Division of Network Medicine, Department of Medicine, Brigham and Women’s Hospital and Harvard Medical School,
Boston, MA, USA; d Department of Biostatistics, Harvard T.H. Chan School of Public Health, Boston, MA, USA; e Department of Nutrition, Harvard T.H. Chan
School of Public Health, Boston, MA, USA

Article info Abstract

Article history: Background: Evidence suggests that ejaculation frequency may be inversely related to the risk
Accepted March 16, 2016 of prostate cancer (PCa), a disease for which few modifiable risk factors have been identified.
Objective: To incorporate an additional 10 yr of follow-up into an original analysis and to
comprehensively evaluate the association between ejaculation frequency and PCa, accounting for
Associate Editor:
screening, clinically relevant disease subgroups, and the impact of mortality from other causes.
Matthew Cooperberg Design, setting, and participants: A prospective cohort study of participants in the Health
Professionals Follow-up Study utilizing self-reported data on average monthly ejaculation
frequency. The study includes 31 925 men who answered questions on ejaculation frequency
Keywords:
on a 1992 questionnaire and followed through to 2010. The average monthly ejaculation
Epidemiology frequency was assessed at three time points: age 20–29 yr, age 40–49 yr, and the year before
Ejaculation questionnaire distribution.
Behavioral risk factors Outcome measurements and statistical analysis: Incidence of total PCa and clinically relevant
disease subgroups. Cox models were used to estimate hazard ratios (HRs) and 95% confidence
intervals (CIs).
Results and limitations: During 480 831 person-years, 3839 men were diagnosed with PCa.
Ejaculation frequency at age 40–49 yr was positively associated with age-standardized body
mass index, physical activity, divorce, history of sexually transmitted infections, and consump-
tion of total calories and alcohol. Prostate-specific antigen (PSA) test utilization by 2008, number
of PSA tests, and frequency of prostate biopsy were similar across frequency categories. In
multivariable analyses, the hazard ratio for PCa incidence for 21 compared to 4–7 ejaculations
per month was 0.81 (95% confidence interval [CI] 0.72–0.92; p < 0.0001 for trend) for frequency
at age 20–29 yr and 0.78 (95% CI 0.69–0.89; p < 0.0001 for trend) for frequency at age 40–49 yr.
Associations were driven by low-risk disease, were similar when restricted to a PSA-screened
cohort, and were unlikely to be explained by competing causes of death.
Conclusions: These findings provide additional evidence of a beneficial role of more frequent
ejaculation throughout adult life in the etiology of PCa, particularly for low-risk disease.
Patient summary: We evaluated whether ejaculation frequency throughout adulthood is
related to prostate cancer risk in a large US-based study. We found that men reporting higher
compared to lower ejaculatory frequency in adulthood were less likely to be subsequently
diagnosed with prostate cancer.
# 2016 European Association of Urology. Published by Elsevier B.V. All rights reserved.

1
These authors contributed equally to this work.
* Corresponding author. Department of Epidemiology, Boston University School of Public Health,
715 Albany Street, Boston, MA 02118, USA. Tel. +1 617 6385035; Fax: +1 617 5667805.
E-mail address: rider@bu.edu (J.R. Rider).
http://dx.doi.org/10.1016/j.eururo.2016.03.027
0302-2838/# 2016 European Association of Urology. Published by Elsevier B.V. All rights reserved.
 EUROPEAN UROLOGY 70 (2016) 974–982
1. Introduction
Prostate cancer (PCa) accounts for approximately 15% of all
new cancer diagnoses among men worldwide, and the
burden of disease continues to increase globally [1]. While
diet and physical activity may provide some promise for
secondary prevention [2–5], there are no evidence-based
recommendations to offer healthy adult men to reduce PCa
risk. The few established disease risk factors—age, race,
family history, and germline polymorphisms—are not
modifiable [6].
Sexual behaviors represent potential modifiable risk
factors and may influence PCa development through a
variety of specific mechanisms. One biological mechanism
involves prostatic accumulation of potentially carcinogenic
secretions, which may create more opportunity for PCa
development, sometimes referred to as the prostate
stagnation hypothesis [7,8]. On the basis of this premise, a
prospective report from the Health Professionals Follow-up
Study (HPFS) cohort published in 2004 found a statistically
significant inverse association between monthly ejacula-
tion frequency and PCa risk based on 8 yr of follow-up [8].
Compared to men reporting an average of 4–7 ejaculations
per month (EPM), the risk of PCa among men reporting 21
EPM in middle age was 50% lower. Although these initial
findings were intriguing, the strongest reduction in risk was
noted for ejaculation frequency in the time period
immediately before questionnaire administration, raising
concerns about the potential influence of undiagnosed PCa
on the results.
To confirm and build on these results [8], we conducted
an updated study within the HPFS cohort with an additional
10 yr of follow-up and 3839 PCa cases, more than double the
number included in the original report. This updated
analysis allows us to address possible reverse causation,
investigate the potential impact of PSA screening, and
determine whether the association between ejaculation
frequency and PCa differed according to the clinical disease
characteristics, as has been observed for other PCa risk
factors [9]. Finally, because ejaculation frequency may be an
indicator of health status and could be related to mortality
from multiple causes, the current analysis considers the
impact of competing causes of death on our findings. Thus,
this updated analysis represents a comprehensive evalua-
tion of the association between ejaculation frequency and
PCa in a large US-based prospective cohort.
2. Patients and methods
2.1. Study population
The HPFS is an ongoing prospective cohort study among 51 529 US male
health professionals [8]. In brief, cancer-free, predominantly Caucasian
(>91%) health professionals aged 40–75 yr were recruited in 1986 and
have been followed with biennial questionnaires on medical history and
lifestyle, including known or suspected cancer and chronic disease risk
factors, diet, use of supplements, and preventive behaviors. Ejaculation
frequency was assessed in the 1992 questionnaire, which was completed
by 46 213 men. Men with a diagnosis of cancer before 1992 (excluding
non-melanoma skin cancer) were excluded from the analysis,
975
leaving 41 201 men. Of these, 9276 did not complete all three questions
on ejaculation frequency, leaving 31 925 men in the study population for
the current analysis. Nonresponders who provided information on
weight, physical activity, and diet appeared to be similar to the
responders. Among participants who were alive in 2010, follow-up was
96% complete. All participants provided informed consent and the study
was approved by the human subjects committee of the Harvard T.H.
Chan School of Public Health, Boston.
2.2. Exposure and covariate assessment
In 1992, participants were asked the following question: ‘‘On average,
how many ejaculations did you have per month during these ages?: ages
20–29; ages 40–49; past year.’’ The frequency at each time point was
reported in the categories none, 1–3, 4–7, 8–12, 13–20, and >20 EPM. To
limit the burden for participants and because the question was designed
specifically to address the prostate stagnation hypothesis, no informa-
tion on the specific type of activity leading to ejaculation was requested.
Information on potential confounders was ascertained in the 1992 ques-
tionnaire and most were updated on the biennial questionnaires
throughout follow-up. PSA testing was first assessed in the 1994 ques-
tionnaire; starting in 1994, men were also asked if they had an elevated
PSA level and whether they had undergone a prostate biopsy or rectal
ultrasound.
2.3. Outcome assessment
For men reporting a diagnosis of PCa, we retrieved medical records and
pathology reports to confirm the diagnosis and to obtain information on
age at diagnosis, PSA level, and tumor stage and grade. Cases were
followed through biennial questionnaires to collect information on the
clinical course, including the development of metastases and treatments.
Deaths were ascertained through repeated mailings and telephone calls
to participants, as well as periodic searches of the National Death Index.
Cause of death was assigned following a review of death certificates,
information from the family, and medical records.
Total PCa incidence was the primary endpoint of interest. Men
diagnosed with stage T1a cancers were excluded from analyses. To
determine whether the association between ejaculation frequency and
PCa differed according to the clinical disease characteristics, we also used
clinical information to group PCa diagnoses into four risk categories
according to National Comprehensive Cancer Network (NCCN) guidelines
[10]. Locally advanced and metastatic disease categories were combined
owing to limited numbers. Men were assigned to the highest category for
which they were eligible: low risk = T1/T2 tumor, PSA < 10 ng/ml,
Gleason score 6; intermediate risk = T1/T2 tumor, PSA 10–<20 ng/ml,
Gleason score 7; high risk = T3 tumor, PSA 20–<50 ng/ml, Gleason score 8;
and regional or distant metastases = T4/N1/M1 tumor, PSA 50 ng/ml. To
more carefully explore differences in risk for indolent and aggressive
disease, we also considered the following subgroups as secondary
analyses: lethal disease (defined as PCa death or metastases to bone or
other organs before the end of follow-up), advanced disease (stage T3b, T4
or N1 or M1 at diagnosis or lethal disease during follow-up), organ-
confined disease (low-grade stage T1 or T2 and N0, M0 at diagnosis and no
progression to metastasis or death during follow-up); and categories of
Gleason score based on prostatectomy or biopsy pathology reports
(Gleason  3 + 4 and Gleason  4 + 3).
2.4. Statistical analyses
Person-time was calculated from the return date for the 1992 question-
naire to the date of PCa diagnosis, death, or the end of the follow-up
period (January 31, 2010). Actuarial curves for PCa-free survival
were generated according to the ejaculation frequency category for
976 EUROPEAN UROLOGY 70 (2016) 974–982
age 40–49 yr using the Kaplan-Meier method. Cox proportional hazards
models were used to estimate the hazard ratio (HR) and 95% confidence
intervals (CI) for total PCa and for each of the clinical subgroups for each
ejaculation frequency category. As in the 2004 report [8], 4–7 EPM was
selected as the reference category as relatively few men reported an
average of 0–3 EPM. The top two categories were combined for some
analyses owing to small numbers of men in the 21 EPM group. Age-
adjusted and multivariable models were evaluated. Age-adjusted
models are adjusted for age in months (as the time scale) and for
calendar time. Multivariable models were additionally adjusted for: race
(Caucasian, African-America, Asian, other ancestry, missing); family
history of PCa (yes/no); vigorous physical activity (quintiles); body mass
index (BMI; <21, 21–<23, 23–<25, 25–<27.5, 27.5–<30, 30 kg/m2,
missing); height (quintiles); diabetes (yes/no); marital status (married,
divorced, other); intake of energy, processed meat, tomato sauce,
calcium, alcohol, and a-linolenic acid (all quintiles); multivitamin use
(yes, no, missing); smoking (never, quit >10 yr ago, quit 10 yr ago,
current, missing); history of vasectomy (yes/no); and history of PSA
testing (yes/no in the previous 2-yr questionnaire cycle). Statistical
significance was evaluated based on a p trend estimated by assigning the
minimum frequency for each category. The missing indicator method
was used for missing data on most covariates [11]. All participants had
baseline questionnaire data for food frequency, and missing data on
nutrients were carried forward from previous reported values. For
activity, missing data were assumed to be in the lowest reference
category. For height, missing data were assumed to be in the middle
category.
2.5. Sensitivity analyses and effect modification
If men with erectile dysfunction have lower ejaculatory frequency and
serious comorbidities associated with a higher risk of premature death
from other causes, a spurious association between less frequent
ejaculation and reduced risk of PCa could result. To address this issue,
we performed a sensitivity analysis that excluded men who reported a
history of erectile dysfunction, defined as poor or very poor ability to
maintain an erection without treatment during the period 1990–1994,
as assessed in the 2000 questionnaire. To eliminate a possible effect of
undiagnosed disease on ejaculation frequency reported, we also
performed analyses that excluded cases diagnosed within the first 4
yr of follow-up. Finally, to address the fact that diagnostic intensity may
vary according to ejaculation frequency, we conducted a sensitivity
analysis restricted to a PSA-screened subset of men who reported a PSA
test before 1994 with follow-up from 1994 to 2010. Stratified analyses
were performed to evaluate potential effect modification by age, BMI,
and vasectomy status. The statistical significance of effect modification
was tested using likelihood ratio tests to compare models with and
without interaction terms between the potential effect modifier and
ejaculation frequency. All aforementioned analyses were performed
using SAS statistical software (release 9.3; SAS Institute, Cary, NC, USA).
2.6. Competing risks analysis
PCa has a long natural history [12], is very sensitive to diagnostic
intensity [13], and is often indolent [14]. Thus, to better understand the
interplay between PCa and deaths due to other causes, we modeled these
events jointly using a multistate model in a semi-competing risks
framework [15], as shown in Supplementary Figure 1. Specifically, we
modeled PCa diagnoses as intermediate states and deaths due to PCa or
other causes as final states. We grouped men into the four NCCN risk
categories described above. Each transition hazard was modeled
assuming that hazards are proportional across ejaculation frequency
levels. We present results for a simple model using age as the time scale
and considering event occurrence by levels of ejaculation frequency
reported at age 40–49 yr. We also investigated causes of death across
ejaculation frequency categories to better understand the different
mortality rates. Analyses were run in R using the mstate package [16]. For
all analyses, p < 0.05 was considered statistically significant.
3. Results
3.1. Baseline characteristics
Ejaculation frequency declined with age. The proportion of
men reporting average frequency of 13 EPM was 57% at age
20–29 yr but dropped to 32% at age 40–49 yr. The Spearman
correlation between ejaculation frequency as an ordinal
variable and ages 20–29 and 40–49 yr was 0.66. Some 40% of
men were in the same frequency category for ages 20–29 and
40–49 yr, and 47% of men moved down a single category from
age 20–29 yr to age 40–49 yr.
The baseline age-standardized characteristics of the study
population (n = 31 925) according to average monthly
ejaculation frequency at age 40–49 yr are presented in
Table 1. Having had a PSA test by 1994 or by 2008 was not
monotonically associated with ejaculation frequency at age
40–49 yr, and the total number of PSA tests was similar across
frequency categories. Among 17 093 men who reported an
initial elevated PSA, the percentage who subsequently
reported prostate biopsy was similar across ejaculation
categories. Men reporting 21 EPM who were subsequently
diagnosed with PCa were somewhat less likely to undergo
radical prostatectomy and more likely to report radiation
compared to men reporting lower frequencies.
Overall, associations between the covariates investigat-
ed and ejaculation frequency were similar for frequency at
age 20–29 yr and in the year before the questionnaire
(Supplementary Tables 1 and 2). Although the associations
were not monotonic, there was some evidence that men in
the highest ejaculation frequency category in the year
before the questionnaire were less likely to have had a PSA
screening test by 1994 (45.4% vs 52.6%) and by 2008 (87.5%
vs 91.8%). However, the associations between covariates
and ejaculatory frequency remained similar to those for the
overall cohort when analysis was restricted to the subset of
13 405 screened men who reported having had a PSA test in
1994 (Supplementary Table 3).
3.2. Ejaculation frequency and PCa risk
During 480 831 person-years of follow-up, a total of
3839 incident PCa cases were diagnosed. As shown in
Figure 1, PCa was less frequently diagnosed among men in
the higher ejaculation frequency categories. The age-
adjusted and multivariable-adjusted HRs according to
average monthly ejaculation frequency are presented in
Table 2. We also present results excluding 10 103 men who
reported erectile dysfunction, leaving 21 822 men and
2704 total cases.
Results for total PCa were similar for the age-adjusted
and multivariable analyses for all three time points at which
ejaculation frequency was assessed, and for the sensitivity
analysis excluding men with erectile dysfunction (Table 2).
 EUROPEAN UROLOGY 70 (2016) 974–982 977

Table 1 – Age-standardized characteristics at baseline in 1992 for the 31 925 men from the Health Professionals Follow-up Study according
to reported ejaculation frequency at age 40–49 yra

Ejaculation frequency

0–3 EPM 4–7 EPM 8–12 EPM 13–20 EPM 21 EPM

Cases (n) 1713 7812 12147 7440 2813

Age (yr) 59.9 (10.4) 59.1 (9.6) 58.9 (9.1) 58.5 (8.9) 58.0 (8.9)
Height (cm) 178 (7) 178 (7) 178 (7) 178 (7) 178 (7)
Body mass index in 1992 (kg/m2) 25.6 (3.6) 25.5 (3.3) 25.8 (3.4) 26.0 (3.4) 26.3 (3.7)
Total activity (MET h/wk) 31 (35) 33 (38) 37 (39) 41 (47) 42 (46)
Total calorie intake (kcal/d) 1847 (583) 1890 (561) 1923 (576) 1968 (590) 2013 (619)
Alcohol intake (g/d) 8.4 (12.9) 9.4 (13.3) 10.4 (14.2) 11.3 (15.3) 12.2 (17.0)
Processed meat intake (servings/d) 0.3 (0.4) 0.3 (0.4) 0.3 (0.4) 0.3 (0.4) 0.3 (0.4)
Tomato sauce intake (servings/wk) 0.9 (0.9) 0.9 (0.9) 0.9 (0.9) 1.0 (1.0) 1.0 (1.0)
Calcium intake (mg/d) 908 (396) 916 (391) 909 (389) 908 (404) 927 (417)
a-Linolenic acid intake (g/d) 1.0 (0.4) 1.1 (0.4) 1.0 (0.4) 1.0 (0.4) 1.0 (0.4)
Divorced 4.9 4.1 5.1 7.8 11.8
Smoked in the past 10 yr 33.5 37.5 38.8 39.3 39.6
Self-reported history of syphilis or gonorrhea 1.2 2.6 3 3.3 4.6
Vasectomy history 18.2 25 27.6 27.2 27.9
Race
White 93.4 96.3 96.7 96.9 96.2
African-American 0.6 0.6 0.8 0.6 0.8
Asian 4 1.7 1.3 1.2 1.1
Other 2.1 1.3 1.2 1.4 2.0
Screening behavior
Had physical examination in past 2 yr 67.8 71.3 71.1 70.7 68.9
Had rectal examination in past 2 yr 65.4 67.4 68.1 67.1 66.3
Had PSA test by 1994 48.2 54.4 55.6 53.8 50.4
Had PSA test by 2008 90.9 92.7 93.1 92.2 90.6
Total periods with PSA test by 2008 (n)b 4.6 (2.6) 5.1 (2.5) 5.1 (2.5) 4.9 (2.6) 4.8 (2.6)
Had prostate biopsy after first report of elevated PSAc 47.2 47.4 48 46.7 47.2
Primary treatment among 3839 cases
Cases (n) 192 1041 1509 807 290
Radical prostatectomy 38.4 41.8 40.8 41.8 36.2
Radiation therapy 30.2 31.9 34.2 34.3 39.3
Hormone therapy 8.3 5.9 5.7 6.6 8.4
Active surveillance/none 8.4 8.2 8.3 6.7 5.1
Other 2.2 1.6 1.8 2.9 0.7
Missing treatment information 12.6 10.6 9.3 7.7 10.2

EPM = ejaculations per month; MET = metabolic equivalent of task; PSA = prostate-specific antigen.
a
Data are presented as the age-standardized mean (standard deviation) for continuous variables and the age-standardized percentage for categorical variables.
All characteristics except age and primary treatment are age-standardized to the distribution for the full cohort in 1992. Primary treatment is standardized to the
age distribution for all prostate cancer cases.
b
Number of 2-yr questionnaire cycles in which a PSA test was reported in the previous 2 yr (maximum = 8).
c
Among 17 093 men who reported ‘‘elevated PSA’’; based only on first report of elevated PSA.

Compared to men with an average monthly frequency of
4–7 ejaculations, men reporting 21 EPM at ages 20–29 and
40–49 yr and in 1991 had a significantly lower risk of total
PCa, with a multivariable-adjusted HR of 0.81 (95% CI 0.72–
0.92), 0.78 (95% CI 0.69–0.89), and 0.76 (95% CI 0.61–0.94),
respectively. Trend tests at each time point when excluding
men in the lowest ejaculation frequency category, who may
be more likely to have serious comorbidities, were similar to
results when considering all five categories (p < 0.0001 for
ages 20–29 and 40–49 yr, p = 0.06 for 1991).
The absolute PCa incidence rate for frequency at age
20–29 yr was 6.56 cases/1000 person-years for 21 EPM
and 8.95 cases/1000 person-years for 4–7 EPM (incidence
rate difference [IRD] 2.39 cases/1000 person-years). For
frequency at age 40–49 yr, the absolute rate was 6.74 cases/
1000 person-years for 21 EPM and 8.94 cases/1000
person-years for 4–7 EPM (IRD 2.20 cases/1000 person-
years). For frequency in the year before the questionnaire,
the rate was 4.49 cases/1000 person-years for 21 EPM and
8.35 cases/1000 person-years for 4–7 EPM (IRD 3.89 cases/
1000 person-years).
Men who reported an average frequency of 21 EPM at
both age 20–29 yr and age 40–49 yr experienced the same
risk reduction for total PCa as men in the highest EPM
category at age 40–49 yr (HR 0.78, 95% CI 0.68–0.90). The
association appeared to be driven by frequency at age
40–49 yr when frequency at both time points was included
in the same model. Compared to men with 4–7 EPM, the HR
for men with 13 EPM at age 40–49 yr was 0.85 (95% CI
0.76–0.94; p = 0.005 for trend) after adjusting for frequency
at age 20–29 yr. The HR for 13 EPM compared to 4–7 EPM
at age 20–29 yr was 0.95 (95% CI 0.83–1.08; p = 0.30 for
trend) after adjusting for frequency at age 40–49 yrs.
However, the correlation between frequencies at different
time points makes it challenging to completely disentangle
the associations.
978 EUROPEAN UROLOGY 70 (2016) 974–982
[(Fig._1)TD$IG]
A 1.0 Ejaculation frequency at any time point was not signifi-
cantly associated with diagnosis of high-risk PCa or
regional/distant metastases. However, for age 20–29 yr
0.8
there was a suggestion of an inverse association between
ejaculation frequency and local/distant metastases (HR
0.6 0.89, 95% CI 0.68–1.15; p = 0.07 for 13 vs 4–7 EPM).
Survival

The risk of both organ-confined and low-grade PCa was

significantly lower for 13 compared to 4–7 EPM for all
0.4 0 3 EPM
three time periods (Table 4). For high-grade PCa, there was a
4 7 EPM
8 12 EPM suggestion of higher risk for men in the lowest frequency
0.2 13 20 EPM category at age 20–29 yr (HR 1.32, 95% CI 0.91–1.92) and
21 EPM age 40–49 yr (HR 1.38, 95% CI 1.03–1.86). However, there
was some evidence that higher ejaculation frequency in the
0.0
year before questionnaire distribution is associated with
higher risk of advanced PCa (HR 1.37, 95% CI 1.00–1.86) or
B 1.00 lethal PCa (HR 1.48, 95% CI 1.02–2.15), but the trend tests
were only marginally significant (p values 0.11 for advanced
and 0.05 for lethal PCa). When we excluded men diagnosed
within the first 4 yr of follow-up to address the impact of
0.95 undiagnosed disease or early symptoms on the results, the
associations for 13 versus 4–7 EPM were attenuated for
Survival

both advanced PCa (HR 1.15, 95% CI 0.79–1.69; p = 0.36 for

trend) and lethal PCa (HR 1.19, 95% CI 0.73–1.94; p = 0.33
0.90 for trend; data not shown).
Sensitivity analyses, including analyses excluding men
diagnosed with PCa in the first 4 yr of follow-up and
analyses restricted to a screened cohort, produced results
similar to the overall findings (Supplementary Material and
0.85
Supplementary Tables 4 and 5). Several stratified analyses
0 4 8 12 16 were conducted to explore potential effect modifiers.
Time (yr) Results stratified by age at baseline, age at diagnosis, BMI
Number at risk: at diagnosis, and history of vasectomy provided no evidence
0 3 EPM: 1713 1609 1464 1283 1118 that any of these factors modified the association between
4 7 EPM: 7812 7381 6761 6117 5430 ejaculation frequency and PCa risk (data not shown).
8 12 EPM: 12147 11519 10695 9719 8573
13 20 EPM: 7440 7080 6589 6005 5365 3.3. Competing causes of death
21 EPM: 2813 2686 2519 2287 2046

Because ejaculation frequency may be an indicator of health

Fig. 1 – Kaplan-Meier curve for prostate cancer–free survival according
to ejaculation frequency category for age 40–49 yr (1992–2010). (A) Plot
over the full prostate cancer–free survival range. (B) Magnified plot for
a restricted survival range. EPM = ejaculations per month.
According to the four NCCN risk groups, 1585 cases had
localized low-risk PCa, 1493 had localized intermediate-risk
disease, 604 had localized high-risk PCa, and 157 patients
had evidence of regional or distant metastases at diagnosis.
Information on clinical disease characteristics was missing
for 434 (11%) men, who were classified in the two lowest
risk categories depending on whether their PCa diagnosis
occurred after a PSA test (n = 336; 21.2% of the localized
low-risk group) or in the absence of a PSA test (n = 98; 6.6%
of the localized intermediate-risk group). For all three time
periods, 13 EPM was associated with a significantly lower
risk (25–28%) of low-risk PCa in comparison to 4–7 EPM
(Table 3). Ejaculation frequency at age 20–29 yr was
also significantly associated with intermediate-risk PCa
(p = 0.0003), with a 27% reduction for 13 versus 4–7 EPM.
status, we fitted the multistate model in Supplementary
Figure 1 to examine PCa incidence over time across the four
NCCN risk groups and the cumulative incidence of lethal
PCa in light of other causes of death. None of the ejaculation
categories was significantly associated with changes in PCa-
specific survival after diagnosis, but categories for the
lowest (0–3 EPM) and highest (13 EPM) frequency had a
higher risk of other-cause mortality (Supplementary
Tables 6 and 7). However, the predicted probability of
events over time for men who were cancer-free at age 50 yr
(Supplementary Fig.) shows that the reduction in PCa risk in
the highest category cannot be fully explained by death
from other causes. Additional details on the results from
this analysis are included in the Supplementary Material.
4. Discussion
The results of this prospective cohort study involving
31 925 men, 18 yr of follow-up, and 3839 PCa cases offer
additional evidence of a role for ejaculation frequency in
the etiology of PCa, particularly for low-risk disease.
 EUROPEAN UROLOGY 70 (2016) 974–982 979

Table 2 – Hazard ratio for incidence of total prostate cancer among 31 925 men in the Health Professionals Follow-up Study according to
reported ejaculation frequency in different time periodsa

Ejaculation frequency

0–3 EPM 4–7 EPM 8–12 EPM 13–20 EPM 21 EPM p trend 13 EPM p trend

Age 20–29 yr
Cases (n) 137 438 1208 1303 753 2056
Age-adjusted HR (95% CI) 0.97 (0.80–1.17) 1.00 1.03 (0.92–1.15) 0.92 (0.83–1.03) 0.80 (0.71–0.90) <0.0001 0.87 (0.79–0.97) 0.0001
Age-adjusted no-ED HR (95% CI) 0.90 (0.70–1.15) 1.00 0.98 (0.86–1.12) 0.89 (0.78–1.01) 0.78 (0.67–0.89) <0.0001 0.84 (0.75–0.96) 0.0009
MV HR (95% CI) 0.99 (0.81–1.19) 1.00 1.03 (0.92–1.14) 0.92 (0.82–1.02) 0.81 (0.72–0.92) <0.0001 0.88 (0.79–0.97) 0.0002
MV no-ED HR (95% CI) 0.91 (0.71–1.17) 1.00 0.99 (0.86–1.13) 0.90 (0.78–1.02) 0.80 (0.69–0.92) <0.0001 0.86 (0.76–0.97) 0.002
Age 40–49 yr
Cases (n) 192 1041 1509 807 290 1097
Age-adjusted HR (95% CI) 0.87 (0.74–1.01) 1.00 0.91 (0.84–0.98) 0.80 (0.72–0.87) 0.77 (0.67–0.87) <0.0001 0.79 (0.72–0.86) <0.0001
Age-adjusted no-ED HR (95% CI) 0.91 (0.75–1.11) 1.00 0.93 (0.85–1.02) 0.80 (0.72–0.89) 0.80 (0.68–0.93) <0.0001 0.80 (0.72–0.88) <0.0001
MV HR (95% CI) 0.88 (0.76–1.03) 1.00 0.90 (0.83–0.98) 0.80 (0.73–0.88) 0.78 (0.69–0.89) <0.0001 0.80 (0.73–0.87) <0.0001
MV no-ED HR (95% CI) 0.91 (0.75–1.11) 1.00 0.93 (0.84–1.02) 0.81 (0.72–0.90) 0.82 (0.70–0.96) 0.0006 0.81 (0.73–0.90) 0.0002
Year before questionnaire (1991)
Cases (n) 1293 1299 831 322 94 416
Age-adjusted HR (95% CI) 1.03 (0.95–1.11) 1.00 0.95 (0.87–1.04) 0.90 (0.80–1.02) 0.73 (0.59–0.90) 0.0004 0.86 (0.76–0.96) 0.002
Age-adjusted no-ED HR (95% CI) 1.06 (0.96–1.17) 1.00 0.96 (0.87–1.06) 0.91 (0.80–1.05) 0.71 (0.56–0.90) 0.0005 0.86 (0.76–0.97) 0.002
MV HR (95% CI) 1.05 (0.97–1.13) 1.00 0.96 (0.87–1.05) 0.93 (0.82–1.05) 0.76 (0.61–0.94) 0.001 0.89 (0.79–0.99) 0.004
MV no-ED HR (95% CI) 1.07 (0.97–1.19) 1.00 0.96 (0.87–1.06) 0.94 (0.82–1.08) 0.74 (0.58–0.94) 0.002 0.89 (0.78–1.01) 0.007

EPM = ejaculations per month; HR = hazard ratio; CI = confidence interval; ED = erectile dysfunction; MV = multivariate.
a
Age-adjusted models are adjusted for age in months and calendar time. Multivariate models are additionally adjusted for: race; family history of prostate
cancer; vigorous physical activity (quintiles); body mass index (six categories); height (quintiles); diabetes; marital status; intake of energy, processed meat,
tomato sauce, calcium, alcohol, and a-linolenic acid (all quintiles); multivitamin use; smoking (never, quit >10 yr ago, quit 10 yr ago, current); history of
vasectomy; and history of PSA testing. No-ED models exclude men who reported poor or very poor ability to have and maintain an erection in the time period
1990–1994, for which analyses include 21 822 participants and 2704 prostate cancer events.

The absolute difference in PCa rate between 21 and
4–7 EPM was 2.39 cases/1000 person-years for frequency at
age 20–29 yr, 2.20 cases/1000 person-years for frequency
at age 40–49 yr, and 3.89 cases/1000 person-years for
frequency in the year before questionnaire distribution.
An initial report published in 2004 for this cohort found
that more frequent ejaculation was related to a lower risk of
total PCa, with strongest associations for higher frequency
in the year before questionnaire distribution [8]. With an
additional decade of follow-up, we demonstrate that

Table 3 – Hazard ratio for prostate cancer by disease severity among 31 925 men in the Health Professionals Follow-up Study according to
reported ejaculation frequency at different timesa,b

0–3 EPM 4–7 EPM 8–12 EPM 13 EPM p value

Cases MV HR Cases MV HR Cases MV HR Cases MV HR for trend

(n) (95% CI) (n) (95% CI) (n) (95% CI) (n) (95% CI)

Age 20–29 yr
Low-risk disease 40 0.84 (0.59–1.18) 160 1.00 (Ref) 407 0.91 (0.75–1.09) 687 0.75 (0.63–0.89) 0.0006
Intermediate-risk disease 39 0.85 (0.60–1.20) 156 1.00 (Ref) 390 0.88 (0.73–1.06) 664 0.73 (0.61–0.88) 0.0003
High-risk disease 43 0.99 (0.70–1.39) 142 1.00 (Ref) 435 1.13 (0.93–1.36) 775 1.00 (0.84–1.20) 0.46
Regional/distant metastases 28 1.20 (0.77–1.87) 70 1.00 (Ref) 186 1.01 (0.77–1.33) 320 0.89 (0.68–1.15) 0.07
Age 40–49 yr
Low-risk disease 65 0.95 (0.72–1.23) 347 1.00 (Ref) 502 0.88 (0.77–1.01) 335 0.72 (0.61–0.83) <0.0001
Intermediate-risk disease 52 0.70 (0.52–0.94) 363 1.00 (Ref) 579 0.99 (0.87–1.13) 401 0.83 (0.72–0.96) 0.11
High-risk disease 39 1.13 (0.80–1.61) 160 1.00 (Ref) 217 0.85 (0.69–1.04) 188 0.89 (0.72–1.11) 0.17
Regional/distant metastases 7 0.61 (0.27–1.35) 44 1.00 (Ref) 55 0.85 (0.57–1.27) 51 0.96 (0.64–1.45) 0.65
Year before
questionnaire (1991)
Low-risk disease 365 1.05 (0.91–1.21) 446 1.00 (Ref) 303 0.94 (0.81–1.09) 135 0.75 (0.62–0.92) 0.002
Intermediate-risk disease 432 1.01 (0.89–1.16) 481 1.00 (Ref) 325 1.00 (0.87–1.15) 157 0.89 (0.74–1.07) 0.27
High-risk disease 232 1.18 (0.96–1.45) 180 1.00 (Ref) 129 1.20 (0.95–1.50) 63 1.12 (0.84–1.51) 0.94
Regional/distant metastases 72 0.94 (0.64–1.38) 48 1.00 (Ref) 15 0.59 (0.33–1.07) 22 1.82 (1.07–3.10) 0.19

EPM = ejaculations per month; MV = multivariate; HR = hazard ratio; CI = confidence interval; PSA = prostate-specific antigen.
a
Risk groups are based on National Comprehensive Cancer Network guidelines. Men were assigned to the highest category for which they were eligible:
low risk = T1/T2 tumor, PSA <10 ng/ml, Gleason score 6; intermediate risk = T1/T2 tumor, PSA 10–<20 ng/ml, Gleason score 7; High risk = T3 tumor, PSA level
20–<50 ng/ml, Gleason score 8; regional or distant metastases = T4/N1/M1 tumor, PSA >50 ng/ml.
b
Age-adjusted models are adjusted for age in months and calendar time. Multivariate models are additionally adjusted for: race; family history of prostate
cancer; vigorous physical activity (quintiles); body mass index (six categories); height (quintiles); diabetes; marital status; intake of energy, processed meat,
tomato sauce, calcium, alcohol, and a-linolenic acid (all quintiles); multivitamin use; smoking (never, quit >10 yrs ago, quit 10 yrs ago, current); history of
vasectomy; and history of PSA testing.
980 EUROPEAN UROLOGY 70 (2016) 974–982

Table 4 – Multivariable prostate cancer incidence by disease severity among 31 925 men in the Health Professionals Follow-up Study
according to reported ejaculation frequency at different timesa

Ejaculation frequency

0–3 EPM 4–7 EPM 8–12 EPM 13 EPM p value

Cases MV HR Cases MV HR Cases MV HR Cases MV HR for trend

(n) (95% CI) (n) (95% CI) (n) (95% CI) (n) (95% CI)

Age 20–29 yr
Low grade (GS  3 + 4) 76 0.88 (0.68–1.13) 283 1.00 (Ref) 785 1.00 (0.87–1.15) 1364 0.86 (0.76–0.98) 0.006
High grade (GS  4 + 3) 39 1.32 (0.91–1.92) 93 1.00 (Ref) 244 1.00 (0.78–1.27) 451 0.93 (0.74–1.16) 0.08
Organ-confined 98 1.04 (0.82–1.30) 309 1.00 (Ref) 877 1.04 (0.91–1.18) 1508 0.89 (0.78–1.00) 0.0008
Advanced 21 0.79 (0.48–1.29) 72 1.00 (Ref) 156 0.87 (0.65–1.15) 268 0.80 (0.62–1.05) 0.23
Lethal 14 0.63 (0.35–1.14) 57 1.00 (Ref) 113 0.83 (0.60–1.14) 200 0.82 (0.60–1.10) 0.71
Age 40–49 yr
Low grade (GS  3 + 4) 106 0.76 (0.62–0.93) 689 1.00 (Ref) 1010 0.90 (0.82–0.99) 703 0.76 (0.68–0.85) <0.0001
High grade (GS  4 + 3) 57 1.38 (1.03–1.86) 197 1.00 (Re) 318 1.01 (0.85–1.21) 255 0.99 (0.82–1.19) 0.23
Organ-confined 132 0.86 (0.71–1.03) 749 1.00 (Ref) 1126 0.93 (0.84–1.02) 785 0.78 (0.71–0.86) <0.0001
Advanced 33 0.98 (0.67–1.44) 146 1.00 (Ref) 185 0.83 (0.66–1.03) 153 0.85 (0.67–1.07) 0.16
Lethal 24 0.95 (0.61–1.49) 107 1.00 (Ref) 133 0.83 (0.64–1.08) 120 0.95 (0.73–1.24) 0.78
Year before
questionnaire (1991)
Low grade (GS  3 + 4) 778 1.04 (0.94–1.15) 869 1.00 (Ref) 587 0.97 (0.87–1.08) 274 0.82 (0.72–0.95) 0.003
High grade (GS  4 + 3) 284 1.08 (0.91–1.29) 264 1.00 (Ref) 180 1.07 (0.88–1.30) 99 1.13 (0.89–1.43) 0.68
Organ-confined 901 1.08 (0.98–1.18) 952 1.00 (Ref) 648 1.00 (0.90–1.11) 291 0.82 (0.72–0.94) 0.0008
Advanced 211 0.99 (0.79–1.22) 157 1.00 (Ref) 88 0.98 (0.75–1.28) 61 1.37 (1.00–1.86) 0.11
Lethal 169 0.95(0.74–1.22) 114 1.00 (Ref) 60 1.01 (0.74–1.39) 41 1.48(1.02–2.15) 0.05

EPM = ejaculations per month; MV = multivariable; HR = hazard ratio; CI = confidence interval; GS = Gleason score.
a
Organ-confined disease: stage T1/2N0M0 at diagnosis that did not progress to metastasis or death during the follow-up period. Advanced disease: stage
T3b/4 or N1 or M1 at diagnosis or prostate cancer death or distant metastasis during follow-up. Lethal disease: prostate cancer death or metastases to bone
or other organs before the end of the follow-up period.

ejaculation frequency at three different time points during
adulthood is associated with statistically significant modest
reductions in risk of total PCa. The association with
frequency at age 20–29 yr became more pronounced with
additional follow-up, while the associations with frequency
at age 40–49 yr and in the year before the questionnaire
remained statistically significant but were somewhat
attenuated. Taken together with the fact that strong inverse
associations remained after excluding men diagnosed in the
first 4 yr of follow-up, the updated results are unlikely to
be strongly influenced by the effects of undiagnosed disease
on ejaculation frequency.
Importantly, our findings were robust to adjustment for
time-varying factors such as BMI, physical activity, and diet
that differed with ejaculation frequency and that have also
been associated with PCa and its progression [5], as well as
other factors associated with PCa risk in this cohort. Because
men in the higher ejaculation frequency categories had
some exposure patterns that might put them at higher risk
of morbidity and mortality due to other causes—higher BMI,
greater alcohol consumption, and more frequent history of
smoking and sexually transmitted infections—we were
concerned that the reduction in PCa risk we observed in this
group might be attributable to premature death from other
causes among men who may have had undiagnosed PCa.
Thus, a strength of our study is the consideration of a model
for semi-competing risks. From this model, the increase in
risk of death by age 80 yr among men with the lowest
ejaculation frequency is 3.8%, while the reduction in PCa
risk is 2.2%. By comparison, men reporting 13 EPM have an
increase of only 1.8% in the risk of dying from other causes
by age 80 yr, while their decrease in PCa risk is 3.8%. Thus, in
both cases the reduction in PCa risk may be partly explained
by premature death due to other causes, but the reduction
among men reporting high ejaculation frequency seemingly
cannot be explained by this effect alone.
Several important limitations of our study should be
noted. Ascertainment of the exposure relied on reporting of
sexual activity in the past. This may introduce measure-
ment error, particularly in the reporting of frequency at age
20–29 yr. However, the use of an anonymized questionnaire
may have resulted in more accurate reporting of sexual
behaviors than a one-on-one interview [17]. Previous
studies suggest that the validity of data on sensitive
information is further improved by: (1) an understanding
among participants that their data will be kept confidential,
most likely true in this large ongoing cohort in which men
had already responded to at least one and potentially two
previous questionnaires; and (2) the avoidance of implying
a ‘‘normal’’ response category [18]. Most importantly,
however, the data were collected prospectively, preventing
differential misclassification (ie, recall bias). Thus, our
results can be considered conservative estimates of the true
association.
While the utilization of a large prospective study has
numerous advantages, the clinical information available for
men at the time of their PCa diagnosis is more limited than
in clinical settings. Thus, we are not able to distinguish very
low-risk disease from low-risk disease in subgroup
analyses. The abundance of data on potential confounders
 EUROPEAN UROLOGY 70 (2016) 974–982
is an advantage of working within the well-annotated HPFS
cohort, but we still cannot rule out residual confounding by
other lifestyle factors. Furthermore, our cohort consisted
primarily of Caucasian men and the frequency of ejaculation
may vary across populations. However, the results may still
be generalizable to other men, as we would not expect a
true biological association between ejaculation frequency
and PCa to differ by race or ethnicity.
The literature exploring the role of sexual activity in the
etiology of PCa is inconsistent [7,19–30]. Previous studies
are primarily retrospective case-control studies, raising
concerns about recall bias, especially given that erectile
dysfunction, ejaculatory dysfunction, and decreased libido
are common consequences of both PCa and its treatment
[31,32]. Moreover, few previous studies have considered
ejaculation frequency per se, with most utilizing proxies of
sexual activity, such as age at first marriage, marital status,
number of sexual partners, and number of children. Few
previous studies have examined associations according to
tumor grade or stage despite their particular importance for
PCa; spurious associations with more favorable disease may
result from confounding by early detection. We do, in fact,
find that the inverse association with overall PCa is driven
by low-risk disease, which could indicate that more
sexually active men might undergo less screening and
follow-up testing. This alternative explanation for our
findings is especially plausible given the potential resulting
side effects of PCa and its treatment on sexual function.
However, PSA screening history and biopsy utilization after
elevated PSA were quite similar across the ejaculation
frequency categories. Moreover, the results were consistent
even when we restricted the analysis to a screened cohort
and PSA history was taken into account. Nonetheless, we
cannot rule out residual confounding by screening or post-
screening biopsy behaviors.
Our results identified suggestive but not statistically
significant associations between higher ejaculation fre-
quency in the year before the questionnaire and both
advanced and lethal PCa. However, the findings appear to be
driven by men diagnosed in the period immediately
following the questionnaire. The attenuated association
in sensitivity analyses excluding men diagnosed in the first
4 yr of follow-up, together with the fact that these
suggestive positive associations were only found for
ejaculation frequency in the year before the questionnaire
distribution and not at younger ages, is consistent with men
with undiagnosed aggressive PCa experiencing symptoms
that promoted more frequent ejaculation. While we are not
aware of any literature supporting ejaculation for relief of
PCa symptoms, it nonetheless seems unlikely that these
suggestive associations with advanced and lethal disease
reflect causality.
In addition to the prostate stagnation hypothesis [7], a
number of mechanisms have been proposed to explain an
inverse association between ejaculation frequency and PCa.
More frequent ejaculation may influence the function of
peripheral-zone epithelial cells, hindering the metabolic
switch from citrate secretion to citrate oxidation known to
occur early in prostate tumorigenesis [33]. Alternatively,
981
more frequent ejaculation may reduce the development of
prostatic intraluminal crystalloids, which have been asso-
ciated with higher risk of PCa [34,35]. Higher ejaculatory
frequency may be linked to lowering of psychological
tension and central sympathetic nervous system suppres-
sion, which could dampen the stimulation of prostate
epithelial cell division [36]. Given the lack of modifiable risk
factors identified for PCa to date, the specific biological
mechanisms underlying these associations are worthy of
further investigation.
5. Conclusions
This large prospective study provides the strongest
evidence to date of a beneficial role of ejaculation in
prevention of PCa, a disease for which relatively little is
understood about etiology generally and knowledge of
modifiable risk factors is particularly scant. The results are
robust to adjustment for many dietary, lifestyle, and
screening behaviors, but additional work on the underlying
biological mechanisms should be undertaken to corrobo-
rate these findings given the potential for residual
confounding. More frequent ejaculation in the absence of
risky sexual behaviors could represent an important means
of reducing the profound medical costs and physical and
psychological side effects of unnecessary diagnosis and
treatment of low-risk tumors, even though it appears to be
less strongly associated with aggressive disease.
Author contributions: Jennifer R. Rider had full access to all the data in
the study and takes responsibility for the integrity of the data and the
accuracy of the data analysis.
Study concept and design: Rider, Mucci, Giovannucci.
Acquisition of data: Rider, Wilson, Mucci, Giovannucci.
Analysis and interpretation of data: Rider, Wilson, Sinnott, Mucci,
Giovannucci.
Drafting of the manuscript: Rider, Wilson.
Critical revision of the manuscript for important intellectual content: Rider,
Wilson, Sinnott, Kelly, Mucci, Giovannucci.
Statistical analysis: Rider, Wilson, Sinnott, Kelly.
Obtaining funding: Giovannucci.
Administrative, technical, or material support: None.
Supervision: Mucci, Giovannucci.
Other: None.
Financial disclosures: Jennifer R. Rider certifies that all conflicts of
interest, including specific financial interests and relationships and
affiliations relevant to the subject matter or materials discussed in the
manuscript (eg, employment/affiliation, grants or funding, consultan-
cies, honoraria, stock ownership or options, expert testimony, royalties,
or patents filed, received, or pending), are the following: None.
Funding/Support and role of the sponsor: The Health Professionals
Follow-Up Study is supported in part by grants UM1 CA167552, P01
CA055075, P01 CA133891, and P01 CA141298. Jennifer R. Rider, Kathryn
M. Wilson, and Lorelei A. Mucci are supported by Prostate Cancer
Foundation Young Investigator Awards. The sponsors played no role in
the study.
Acknowledgments: We are grateful to the participants and staff of the
Health Professionals Follow-up Study for their valuable contributions. In
addition, we would like to thank the following state cancer registries for
982 EUROPEAN UROLOGY 70 (2016) 974–982
their cooperation and assistance: AL, AZ, AR, CA, CO, CT, DE, FL, GA, ID, IL,
IN, IA, KY, LA, ME, MD, MA, MI, NE, NH, NJ, NY, NC, ND, OH, OK, OR, PA, RI,
SC, TN, TX, VA, WA, and WY. JRR and KMW assume full responsibility for
analyses and interpretation of these data. Finally, we are indebted to
David Havelick for his enthusiastic support of this project.
Appendix A. Supplementary data
Supplementary data associated with this article can be
found, in the online version, at http://dx.doi.org/10.1016/j.
eururo.2016.03.027.
References
[1] Soerjomataram I, Lortet-Tieulent J, Parkin DM, et al. Global burden
of cancer in 2008: a systematic analysis of disability-adjusted life-
years in 12 world regions. Lancet 2012;380:1840–50.
[2] Gardner JR, Livingston PM, Fraser SF. Effects of exercise on treat-
ment-related adverse effects for patients with prostate cancer
receiving androgen-deprivation therapy: a systematic review. J Clin
Oncol 2014;32:335–46.
[3] Kenfield SA, Stampfer MJ, Giovannucci E, Chan JM. Physical activity
and survival after prostate cancer diagnosis in the Health Profes-
sionals Follow-up Study. J Clin Oncol 2011;29:726–32.
[4] Richman EL, Kenfield SA, Stampfer MJ, Paciorek A, Carroll PR, Chan
JM. Physical activity after diagnosis and risk of prostate cancer
progression: data from the cancer of the prostate strategic urologic
research endeavor. Cancer Res 2011;71:3889–95.
[5] Wilson KM, Giovannucci EL, Mucci LA. Lifestyle and dietary factors
in the prevention of lethal prostate cancer. Asian J Androl
2012;14:365–74.
[6] Discacciati A, Wolk A. Lifestyle and dietary factors in prostate
cancer prevention. Recent Results Cancer Res 2014;202:27–37.
[7] Isaacs JT. Prostatic structure and function in relation to the etiology
of prostatic cancer. Prostate 1983;4:351–66.
[8] Leitzmann MF, Platz EA, Stampfer MJ, Willett WC, Giovannucci E.
Ejaculation frequency and subsequent risk of prostate cancer. JAMA
2004;291:1578–86.
[9] Giovannucci E, Liu Y, Platz EA, Stampfer MJ, Willett WC. Risk factors
for prostate cancer incidence and progression in the Health Profes-
sionals Follow-up Study. Int J Cancer 2007;121:1571–8.
[10] National Comprehensive Cancer Network. Clinical practice guide-
lines in oncology: prostate cancer. Fort Washington, PA: NCCN; 2016.
[11] Greenland S, Finkle WD. A critical look at methods for handling
missing covariates in epidemiologic regression analyses. Am J
Epidemiol 1995;142:1255–64.
[12] Popiolek M, Rider JR, Andren O, et al. Natural history of early,
localized prostate cancer: a final report from three decades of
follow-up. Eur Urol 2013;63:428–35.
[13] Thompson IM, Pauler DK, Goodman PJ, et al. Prevalence of prostate
cancer among men with a prostate-specific antigen level 4.0 ng
per milliliter. N Engl J Med 2004;350:2239–46.
[14] Jahn JL, Giovannucci EL, Stampfer MJ. The high prevalence of
undiagnosed prostate cancer at autopsy: implications for epidemi-
ology and treatment of prostate cancer in the prostate-specific
antigen era. Int J Cancer 2015;137:2795–802.
[15] Putter H, Fiocco M, Geskus RB. Tutorial in biostatistics: competing
risks and multi-state models. Stat Med 2007;26:2389–430.
[16] R Core Team. R: a language and environment for statistical com-
puting. Vienna, Austria: R Foundation for Statistical Computing;
2014.
[17] Fenton KA, Johnson AM, McManus S, Erens B. Measuring sexual
behaviour: methodological challenges in survey research. Sex
Transm Infect 2001;77:84–92.
[18] Tourangeau R, Yan T. Sensitive questions in surveys. Psychol Bull
2007;133:859–83.
[19] Sarma AV, McLaughlin JC, Wallner LP, et al. Sexual behavior,
sexually transmitted diseases and prostatitis: the risk of prostate
cancer in black men. J Urol 2006;176:1108–13.
[20] Fernandez L, Galan Y, Jimenez R, et al. Sexual behaviour, history of
sexually transmitted diseases, and the risk of prostate cancer: a
case-control study in Cuba. Int J Epidemiol 2005;34:193–7.
[21] Hayes RB, Pottern LM, Strickler H, et al. Sexual behaviour, STDs and
risks for prostate cancer. Br J Cancer 2000;82:718–25.
[22] Dennis LK, Dawson DV. Meta-analysis of measures of sexual activi-
ty and prostate cancer. Epidemiology 2002;13:72–9.
[23] Honda GD, Bernstein L, Ross RK, Greenland S, Gerkins V, Henderson
BE. Vasectomy, cigarette smoking, and age at first sexual inter-
course as risk factors for prostate cancer in middle-aged men. Br J
Cancer 1988;57:326–31.
[24] Spence AR, Rousseau MC, Parent ME. Sexual partners, sexually
transmitted infections, and prostate cancer risk. Cancer Epidemiol
2014;38:700–7.
[25] Rotkin ID. Studies in the epidemiology of prostatic cancer: expand-
ed sampling. Cancer Treat Rep 1977;61:173–80.
[26] Ahmadi H, Allameh F, Baradaran N, et al. Circulating sex hormones
play no role in the association between sexual activity and the risk
of prostate cancer. J Sex Med 2011;8:905–13.
[27] Giles GG, Severi G, English DR, et al. Sexual factors and prostate
cancer. BJU Int 2003;92:211–6.
[28] Hayes RB, de Jong FH, Raatgever J, et al. Physical characteristics and
factors related to sexual development and behaviour and the risk
for prostatic cancer. Eur J Cancer Prev 1992;1:239–45.
[29] Rosenblatt KA, Wicklund KG, Stanford JL. Sexual factors and the risk
of prostate cancer. Am J Epidemiol 2001;153:1152–8.
[30] Oishi K, Okada K, Yoshida O, et al. A case-control study of prostatic
cancer in Kyoto, Japan: sexual risk factors. Prostate 1990;17:269–79.
[31] Saitz TR, Serefoglu EC, Trost LW, Thomas R, Hellstrom WJ. The pre-
treatment prevalence and types of sexual dysfunction among
patients diagnosed with prostate cancer. Andrology 2013;1:859–63.
[32] Choo R, Long J, Gray R, Morton G, Gardner S, Danjoux C. Prospective
survey of sexual function among patients with clinically localized
prostate cancer referred for definitive radiotherapy and the impact
of radiotherapy on sexual function. Support Care Cancer 2010;18:
715–22.
[33] Costello LC, Franklin RB. The clinical relevance of the metabolism of
prostate cancer; zinc and tumor suppression: connecting the dots.
Mol Cancer 2006;5:17.
[34] Svatek RS, Karam JA, Rogers TE, Shulman MJ, Margulis V, Benaim EA.
Intraluminal crystalloids are highly associated with prostatic ade-
nocarcinoma on concurrent biopsy specimens. Prostate Cancer
Prostatic Dis 2007;10:279–82.
[35] Del Rosario AD, Bui HX, Abdulla M, Ross JS. Sulfur-rich prostatic
intraluminal crystalloids: a surgical pathologic and electron probe
X-ray microanalytic study. Hum Pathol 1993;24:1159–67.
[36] Newman HF, Reiss H, Northup JD. Physical basis of emission,
ejaculation, and orgasm in the male. Urology 1982;19:341–50.
 original article Annals of Oncology 20: 542–549, 2009
doi:10.1093/annonc/mdn644
Published online 6 January 2009

A phase III randomized, double-blind,

placebo-controlled trial of gabapentin in the
management of hot flashes in men (N00CB)
C. L. Loprinzi1*, A. C. Dueck1, B. S. Khoyratty2, D. L. Barton1, S. Jafar3, K. M. Rowland Jr4,
P. J. Atherton1, G. W. Marsa5, W. H. Knutson6, J. D. Bearden III7, L. Kottschade1 & T. R. Fitch8
1
Division of Medical Oncology, Mayo Clinic and Mayo Foundation, Rochester; 2Metro-Minnesota Community Clinical Oncology Program, St Louis Park; 3Michigan
Cancer Research Consortium, Ann Arbor; 4Carle Cancer Center CCOP, Urbana; 5Toledo Community Hospital Oncology Program CCOP, Toledo; 6Duluth CCOP,
Duluth, MN; 7Upstate Carolina CCOP, Spartanburg; 8Mayo Clinic Arizona, Scottsdale, USA

Received 2 July 2008; revised 26 August 2008; accepted 27 August 2008

Introduction: Hot flashes represent a significant problem in men undergoing androgen deprivation therapy.
Materials and methods: Via a prospective, double-blind, placebo-controlled clinical trial, men with hot flashes, on
a stable androgen deprivation therapy program for prostate cancer, received a placebo or gabapentin at target doses
of 300, 600, or 900 mg/day. Hot flash frequencies and severities were recorded daily during a baseline week and for
4 weeks while the patients took the study medication.
Results: In the 214 eligible patients who began the study drug on this trial, comparing the fourth treatment week to
the baseline week, mean hot flash scores decreased in the placebo group by 4.1 units and in the three increasing
dose gabapentin groups by, 3.2, 4.6, and 7.0 units. Comparing the three combined gabapentin arms to the placebo
original
article

arm did not result in significant hot flash differences. Wilcoxon rank-sum P values for change in hot flash scores and
frequencies after 4 weeks of treatment were 0.10 and 0.02, comparing the highest dose gabapentin arm to the
placebo arm, respectively. The gabapentin was well tolerated in this trial.
Conclusion: These results support that gabapentin decreases hot flashes, to a moderate degree, in men with
androgen ablation-related vasomotor dysfunction.
Key words: hot flashes, men, prostate cancer

introduction Anecdotal experience reported a reduction of hot flashes in five

Hot flashes can be a prominent problem in men with prostate
cancer undergoing androgen deprivation therapy. Whether
androgen deprivation therapy is carried out by medical or
surgical means, hot flashes affect up to 75% of patients [1, 2].
There are limited therapeutic options for treating men with
hot flashes. While clonidine moderately helps hot flashes in
women, a randomized, placebo-controlled, double-blind trial
demonstrated no significant benefit in men [3]. Placebo-
controlled trials evaluating estrogen or progestational agents
have demonstrated efficacy in treating male hot flashes, with
about a 75% reduction seen with hormones versus a 25%
reduction with a placebo [4, 5]. Nonetheless, these hormonal
treatments can have side-effects. Estrogen can cause breast
tenderness and enlargement. Progestational agents have been
associated with rising prostate-specific antigen levels in some
men with prostate cancer [6].
Gabapentin is an antiseizure agent which has been
utilized for the treatment of a number of pain syndromes.
*Correspondence to: Dr C. L. Loprinzi, Division of Medical Oncology, Mayo Clinic, 200 First
Street SW, Rochester, MN 55905, USA. Tel: +1-507-284-8964; Fax: +1-507-284-1803;
E-mail: cloprinzi@mayo.edu
women and one man [7]. Subsequent pilot trials also suggested
efficacy [8, 9].
Given this promising evidence and the similar treatment
benefits of some drugs for hot flashes in men and women, it
was hypothesized that gabapentin would be effective against hot
flashes in men receiving androgen ablation therapy. This
current clinical trial was developed to study the efficacy and
side-effects of three relatively low gabapentin doses in this
patient population.
materials and methods
patient eligibility characteristics
Men with a history of prostate cancer, who were on a stable program of
androgen ablation hormone therapy for the prior 4 weeks and who were
not planning on discontinuing hormone therapy for 5 weeks, were eligible
for this trial if they had bothersome hot flashes. Bothersome hot flashes
were defined by their occurrence of at least 14 times per week and of
sufficient severity to make the patient desire therapeutic intervention, with
the hot flashes being present for at least 1 month before study entry.
Patients were required to have an estimated life expectancy of at least
6 months and a good performance status.

ª The Author 2009. Published by Oxford University Press on behalf of the European Society for Medical Oncology.
All rights reserved. For permissions, please email: journals.permissions@oxfordjournals.org
Annals of Oncology
Eligible patients could not have had a history of any significant renal
insufficiency (defined as a serum creatinine of 1.5 times normal in the prior
2 years) nor could they have any of the following concurrent (£4 weeks) or
planned therapies: antineoplastic chemotherapy, androgens, estrogens, or
progestational agents. They could not have used gabapentin previously.
Antidepressant use was allowed if the patient had been on a stable dose for
at least 1 month and did not plan to modify this treatment during the
ensuing 5 weeks.
All patients provided written informed consent for participating in this
trial, monitored by local Internal Review Boards as mandated by United
States Federal regulations. Patients were stratified before randomization on
the basis of the patient’s reported number of hot flashes per day (2–3 versus
4–9 versus ‡10) and the reported duration of hot flash symptoms (<9
versus ‡9 months). Patients were then randomized, using a dynamic
allocation method that balances the marginal distributions of the
stratification factors [10], to receive one of four treatment schedules:
gabapentin 300 mg at bedtime for 28 days, versus gabapentin 300 mg at
bedtime for 7 days and then 300 mg twice daily for 21 days, versus
gabapentin 300 mg at bedtime for 7 days then 300 mg twice daily for 7 days
and then 300 mg three times daily for 14 days, versus a placebo for
28 days. Patients receiving placebo were divided into three groups so that
one group received one tablet at bedtime for 28 days, a second group
received one tablet at bedtime for 7 days then one tablet b.i.d. for 21 days,
and a third group received one tablet at bedtime for 7 days then one
tablet b.i.d. for 7 days then one tablet t.i.d. for 14 days. The assigned
treatment was administered following the completion of a 1-week period
to collect baseline hot flash data. All treatment assignments were blinded
to the patients and clinical investigators, with only the North Central
Cancer Treatment Group (NCCTG) randomization office, the pharmacists,
and the study statisticians having access to the drug assignments for
individual patients. Gabapentin and matching placebo tablets were
provided by Pfizer Corporation, New York.
At study entry, the patient was given a booklet that contained patient
instructions, hot flash definitions, hot flash diary questionnaires to be
original article
completed daily for 5 weeks, symptom experience diaries to be completed
weekly for 5 weeks, and the 30-item Profile of Mood States-Brief
(POMS-B) [11] form to be completed at the end of the baseline week and
then at the end of 4 weeks of treatment. Hot flash definitions were
provided to each patient, derived from other men who had participated in
a previous hot flash trial [12]. The hot flash diary was similar to
a validated diary that has been utilized in a series of clinical trials which
have now involved >1500 subjects [13]. This diary questionnaire inquired
regarding the numbers of mild, moderate, severe, and very severe hot
flashes per 24-hour period. The weekly symptom experience diaries asked
about the severity of potential side-effects of gabapentin on a 0–10 scale
including appetite loss, undesirable appetite increase, nausea, mouth
dryness, fatigue, dizziness, trouble walking/balance problems, muscle
pain, trouble concentrating, constipation, sleepiness, sleeping troubles,
blurry vision, nervousness, and mood changes. It also inquired about
hot flash distress, satisfaction with hot flash control, quality of life (QoL),
and how much hot flashes affected QoL, each on a 0–10 scale. The
POMS-B asked about the severity of weekly mood changes. Throughout
the presented results, all patient-reported outcomes have been
transformed onto a 0–100 scale with 0 indicating the most negative
outcome and 100 indicating the most positive outcome, for ease of
comparability. On the last day of each treatment week, patients were
asked about their compliance with taking the study medication by asking
them how many tablets they were taking per day.
Each patient was to be contacted by telephone weekly for the first 5 weeks
to document compliance, encourage completion of the questionnaires, and
address questions.
On the last scheduled day of the randomized double-blind treatment,
each patient, after assuring that he had completed his questionnaires, was
allowed to be told which dose of gabapentin he received and was
subsequently given the choice of whether he would like to continue with
gabapentin or to start active drug if he was on the placebo.

Table 1. Baseline eligible patient characteristics

Placebo (n = 53) 300 mg/day (n = 54) 600 mg/day (n = 53) 900 mg/day (n = 54) P valuea
Age 0.89
Mean (standard deviation) 70 (8.89) 69 (8.53) 70 (7.02) 69 (7.69)
Median 71 69 69 72
Range (47-87) (50-89) (58-86) (52-85)
Race/ethnicity 0.66
White 49 (93%) 49 (91%) 49 (93%) 50 (93%)
Black or African-American 3 (6%) 5 (9%) 3 (6%) 4 (7%)
Asian 1 (2%) 0 (0%) 0 (0%) 0 (0%)
Hispanic 0 (0%) 0 (0%) 1 (2%) 0 (0%)
ECOG performance score 0.39
0 41 (77%) 47 (87%) 42 (79%) 40 (74%)
1 12 (23%) 7 (13%) 11 (21%) 14 (26%)
No. of hot flashes per day 0.62
1: 2–3 7 (13%) 5 (9%) 4 (7%) 10 (19%)
2: 4–9 22 (42%) 26 (48%) 28 (53%) 23 (43%)
3: ‡10 24 (45%) 23 (43%) 21 (40%) 21 (39%)
Duration of hot flashes (months) 0.99
1: <9 31 (58%) 33 (61%) 31 (58%) 32 (59%)
2: ‡9 22 (42%) 21 (39%) 22 (42%) 22 (41%)
a
P values for Kruskal–Wallis test for age and chi-squared test for all other variables.
ECOG, Eastern Cooperative Oncology Group.

Volume 20 | No. 3 | March 2009 doi:10.1093/annonc/mdn644 | 543

original article Annals of Oncology

statistical methods
Methods used to analyze the data were similar to those used for a previous
dose-finding hot flash study of venlafaxine [14]. The primary end point
in this current study was change from baseline in hot flash score after
4 weeks of treatment. The hot flash score was computed for each patient
using data provided on the hot flash diary by assigning points (1 = mild, 2 =
moderate, 3 = severe, and 4 = very severe) to each hot flash based on
reported severity, summing the points for each day, and averaging the daily
scores across each week of the study.
A sample size of 50 patients per arm was required to provide at least 80%
power to detect a clinically meaningful difference in the primary end point
between the collective gabapentin arm (150 patients) and the placebo
arm (50 patients) with a two-sided 5% type I error rate using the Wilcoxon
rank-sum test. For comparisons between any pair of arms, the power
was 66% for detecting the same clinically meaningful difference in the
primary end point using a Wilcoxon rank-sum test with a two-sided 5%
type I error rate. For this study, a clinically meaningful difference in the
primary end point was defined as 0.5 standard deviations, which, based on
previous data, equates to 2.5 units of hot flash score. Cohen identifies such
an effect as being of moderate size [15]. For each Wilcoxon rank-sum
comparison, the Hodges–Lehmann (HL) shift estimator (median of all
pairwise differences) with 95% confidence interval (CI) is reported herein.
Additional analyses included Kruskal–Wallis testing of the primary end
point across the three gabapentin arms (i.e. dose levels) followed by
pairwise comparisons using Wilcoxon rank-sum tests among the three
gabapentin arms. Additionally, the primary end point for each gabapentin
arm was compared with the placebo arm, using Wilcoxon rank-sum tests.
Secondary analyses included analysis of hot flash frequency similar to
that carried out using hot flash score, where a clinically meaningful
difference equates to 1 hot flash per day. CIs were constructed for the
median reductions in hot flash score and frequency by arm, and plots
were created of average hot flash scores and frequencies (as percent of
baseline) during this double-blind study by treatment arm. Additionally,
changes at week 4, from baseline values, in side-effects as measured by the
symptom experience diary, mood (POMS-B), hot flash distress, satisfaction
with hot flash control, QoL, and hot flash effect on QoL tools were

223
randomized
patients

8 withdrew
prior to
receiving study
medication

215 started
study
medication

54 assigned to 53 assigned to 54 assigned to

54 assigned to
gabapentin gabapentin gabapentin
Placebo
300 mg/d 600 mg/d 900 mg/d

1 ineligible 46 with 42 with 43 with

due to use of evaluable hot evaluable hot evaluable hot
megestrol flash data for flash data for flash data for
acetate the 5 weeks the 5 weeks the 5 weeks

46 with
evaluable hot
flash data for
the 5 weeks

Figure 1. Consort diagram of patient flow in this trial.

544 | Loprinzi et al. Volume 20 | No. 3 | March 2009

Annals of Oncology
compared between the gabapentin arms and the placebo arm using
Wilcoxon rank-sum tests. Two-sided P values £0.05 were considered
statistically significant throughout.
The impact of the missing data on conclusions was investigated using
imputation. Analyses were first carried out using all available data and then
carried out on datasets with missing values imputed using the last value
carried forward and the average value carried forward. Conclusions were
consistent throughout. Results of the analyses using all available data are
presented in the following section.
As confirmatory analyses to the prespecified primary and secondary
analyses, hot flash scores (and frequencies) at week 4 were compared using
analysis of covariance with baseline scores (frequencies) as the covariate.
Results of these analyses were very similar to those of our primary and
secondary analyses (results not shown).
results
From 12 February 2002 to 9 November 2006, 223 patients were
entered on this clinical trial from 45 institutions in the United
States. Eight patients canceled before receiving study treatment
and one patient, on the placebo arm, was deemed ineligible due
to concomitant treatment with megestrol acetate, leaving 214
eligible patients. Baseline characteristics (Table 1) were well
balanced across the treatment arms. The flow of patients on this
trial is illustrated in Figure 1.
Changes from baseline hot flash scores and frequencies
original article
0 (0, 0), Wilcoxon P = 0.002; constipation HL (95% CI) =
0 (0, 0), Wilcoxon P = 0.02], with the significant differences
favoring the gabapentin arms (i.e. patients on the placebo arm
reported more trouble with both appetite loss and
constipation). In comparing the highest gabapentin dose arm
to the placebo arm, the significant differences included
appetite loss [HL (95% CI) = 0 (0, 0), Wilcoxon P = 0.02),
fatigue [HL (95% CI) = 10 (0, 20), Wilcoxon P = 0.05), and
indigestion/belching [HL (95% CI) = 0 (0, 10), Wilcoxon P =
0.02), each, again favoring the gabapentin arm. The
proportion of patients ending active treatment early due to
side-effects was not significantly different in any of the four
study arms or in the combined gabapentin arms as compared
with the placebo arm (Fisher’s exact test; 5.9% versus 4.3%,
P = 0.69).
Hot flash distress changes are illustrated in Figure 3. The
difference between the combined gabapentin arms and the
placebo arm in change from baseline at week 4 was significant
for satisfaction with hot flash control [HL (95% CI) = 10 (0,
20), Wilcoxon P = 0.03) and the effect of hot flashes on QoL
[HL (95% CI) = 10 (0, 20), Wilcoxon P = 0.01), with the
120
A
100
(percent of baseline)
Hot flash frequency

during the fourth treatment week were available for 177 of 214 Placebo
80 300
(83%) patients—46 of 54 (85%), 46 of 54 (85%), 42 of 53 600
(79%), and 43 of 54 (80%) patients for the placebo arm and the 60 900
300, 600, and 900 mg/day gabapentin arms, respectively.
40
Changes in hot flash score from the baseline week over the four
treatment weeks are illustrated in Figure 2 and Table 2. The 20
primary comparison of change from baseline in hot flash score
0
after 4 weeks of treatment between the collective gabapentin
Baseline 1 2 3 4
arm and the placebo arm was not statistically significant [HL
Week
(95% CI) = 20.6 (22.3, 1.0); Wilcoxon P = 0.48]. However,
the comparisons of changes from baseline hot flash score and B 120
frequency at week 4 between the 900 mg/day gabapentin arm 100
(percent of baseline)

(the highest gabapentin dose arm) and the placebo were

Hot flash score

Placebo
associated with Wilcoxon rank-sum P values of 0.10 and 0.02 80
300
[HL (95% CI) of 21.9 (24.1, 0.3) and 21.4 (22.7, 20.3)], 600
60
respectively. P values for all pairwise Wilcoxon rank-sum 900
comparisons among the four treatment arms appear in Table 3. 40
There was no apparent difference in efficacy between sets of
20
patients grouped by whether they had 10 or more baseline hot
flashes per day, versus less. 0
With regard to side-effects, the only statistically significant Baseline 1 2 3 4
differences between the combined gabapentin arms and the Week
placebo arm were appetite loss and constipation [change
Figure 2. Mean changes from baseline for hot flash scores (A) and
during week 4, from baseline; appetite loss HL (95% CI) =
frequencies (B) for all study arms.

Table 2. Median percent decreases in hot flash score and frequency during treatment week 4, from baseline

Measure Median % change from baseline to week 4 (95% confidence interval)

Placebo 300 mg/day 600 mg/day 900 mg/day
Hot flash frequency 21.5 (11.3–30.9) 22.8 (12.1–33.0) 31.8 (16.5–40.5) 45.5 (31.1–50.6)
Hot flash score 27.0 (12.1–36.1) 29.7 (13.1–36.9) 33.8 (22.2–47.1) 44.4 (35.2–56.3)

Volume 20 | No. 3 | March 2009 doi:10.1093/annonc/mdn644 | 545

original article Annals of Oncology

Table 3. Wilcoxon rank-sum comparisons of change from baseline at week 4 for hot flash score and frequency

Placebo 300 mg/day 600 mg/day 900 mg/day Collective gabapentin Hodges–Lehmann Wilcoxon
(n = 46) (n = 46) (n = 42) (n = 43) (n = 131) shift estimator P value
Table values are median changes from baseline at week 4 (95% confidence
interval)
Hot flash score Collective gabapentin 22.8 23.3 20.6 (22.3, 1.0) 0.48
versus placeboa
300 mg/day 22.8 22.4 0.1 (21.7, 2.2) 0.80
versus placebo
600 mg/day 22.8 23.1 20.3 (22.6, 1.6) 0.72
versus placebo
900 mg/day 22.8 24.3 21.9 (24.1, 0.3) 0.10
versus placebo
300 mg/day 22.4 23.1 0.7 (21.3, 2.9) 0.44
versus 600 mg/day
300 mg/day 22.4 24.3 2.1 (0.0, 4.1) 0.05
versus 900 mg/day
600 mg/day 23.1 24.3 1.4 (20.9, 3.4) 0.19
versus 900 mg/day
Hot flash frequency Collective gabapentin 21.6 22.1 20.6 (21.6, 0.3) 0.19
versus placebo
300 mg/day 21.6 21.8 20.1 (21.3, 1.0) 0.75
versus placebo
600 mg/day 21.6 22.0 20.3 (21.6, 1.0) 0.60
versus placebo
900 mg/day 21.6 22.6 21.4 (22.7, 20.3) 0.02
versus placebo
300 mg/day 21.8 22.0 0.3 (21.0, 1.6) 0.73
versus 600 mg/day
300 mg/day 21.8 22.6 1.3 (0.1, 2.7) 0.03
versus 900 mg/day
600 mg/day 22.0 22.6 1.1 (20.3, 2.5) 0.10
versus 900 mg/day
a
Primary comparison.

25 Wilcoxon P = 0.003], with the significant differences again

Change in hot flash distress

20
15
10
5 Placebo
0 week, were insignificant (all Wilcoxon P > 0.05) for the
Baseline 1 2
Week
Figure 3. Median changes from baseline in hot flash distress, for each
study arm (higher scores are better, on a total scale of 100 points).
significant differences favoring the gabapentin arms. The
differences between the highest gabapentin dose arm and the
placebo arm in changes in the week 4 values, from the baseline
week, were significant for hot flash distress [HL (95% CI) = 10
(0, 20), Wilcoxon P = 0.04), satisfaction with hot flash
control [HL (95% CI) = 10 (0, 30), Wilcoxon P = 0.03), and
the effect of hot flashes on QoL [(95% CI) = 10 (0, 20),
900 favoring the gabapentin arm. Overall QoL (as changes during
week 4, from baseline) was not significantly different between
600
the gabapentin arm and the placebo arm [HL (95% CI) =
0 (0, 10), Wilcoxon P = 0.49] or between the highest dose of
gabapentin and the placebo arm [HL (95% CI) = 0 (0, 10),
300
Wilcoxon P = 0.42).
The differences between the combined gabapentin arms and
the placebo arm in changes during week 4, from the baseline
3 4
POMS-B Total Mood Disturbance and all subscale scores
(tension/anxiety, depression/dejection, confusion/
bewilderment, fatigue/inertia, and anger/hostility) except for
vigor/activity [HL (95% CI) = 5 (0, 10), Wilcoxon P = 0.03],
with the significant difference in favor of the combined
gabapentin arms. The difference between the highest
gabapentin dose arm and the placebo arm in change during
week 4, from the baseline week, was insignificant (all Wilcoxon
P > 0.05) for the POMS-B Total Mood Disturbance and all
subscale scores except for depression/dejection [HL (95% CI) =
5 (0, 10), Wilcoxon P = 0.02), with the significant difference
again favoring the gabapentin arm.

546 | Loprinzi et al. Volume 20 | No. 3 | March 2009

Annals of Oncology
Compliance information, obtained from patient self-
reports, appears in Table 4. During each week of treatment,
compliance rates for patients who did not drop out of the
study were high (‡89% for each of the individual treatment
regimens and ‡95% for the combination of all study
arms) during this double-blind study. There were no
significant differences in compliance rates across the
gabapentin arms and matching placebo regimens at each
treatment week.
The proportion of patients choosing to continue on with
gabapentin (or start taking gabapentin if initially on the placebo
arm) at the end of the double-blind study was 78%, 77%, 56%,
and 62% in the placebo and three increasing gabapentin dose
arms.
The placebo arm was broken up into three subsets, based on
whether the target dose was 1, 2, or 3 tablets per day. When
comparing these three arms to each other, there were no
substantial differences between them in terms of hot flash
scores (which, during the fourth treatment week were 78%,
80%, and 72% of the baseline week in the three subsets,
respectively), toxicity, or compliance (Table 4).
discussion
The efficacy of gabapentin seen in this current trial parallels the
efficacy seen with similar doses in women with hot flashes, with
each of these revealing an 50% reduction in hot flashes at
a gabapentin dose of 300 mg given three times daily [16, 17].
A cross study comparison of the results from this current trial
and the results of the two trials in women with 900 mg of
gabapentin per day is illustrated in Figure 4, demonstrating
similarities among the three different trials. Recent data from
another trial in women also revealed very similar results, with a
original article
51% reduction of hot flashes with gabapentin, compared with a
26% reduction with a placebo (P < 0.0001) [18].
The gabapentin was extremely well tolerated in this trial,
without any evidence of side-effects being observed more than
what was seen in the placebo group. Similar results were seen in
a placebo-controlled trial looking at gabapentin for treatment
of chemotherapy neuropathy, with this trial utilizing up to 2700
mg/day in divided doses [19].
The mechanism of action of gabapentin’s effect against hot
flashes is not clear. It has been hypothesized that modulation of
calcium currents may be involved [17]. Nonetheless, there is
not definitive information regarding the mechanism for its
efficacy.
It is worth providing a few words with regards to the
methodology of this trial. The utilized methodology has been
validated [13] and has been used in a series of Mayo Clinic/
NCCTG clinical trials involving both women and men with
hot flashes [14, 20–25]. A similar methodology has also been
used by investigators in other hot flash trials [16, 26, 27].
With regards to trial length, it might be argued that the
duration of treatment in this trial (4 weeks) is shorter than
what has been utilized in other trials evaluating hot flash
treatment regimens. Nonetheless, trials that investigated
nonhormonal agents for hot flashes over 8–12 weeks nicely
demonstrate that there is a plateau effect on the curves from
4 weeks on to 6–12 weeks [16, 17, 25–27].
Another methodology question deals with the placebo
program that was utilized in this trial. The placebo arm could
be broken up into three subsets, based on whether the target
dose was 1, 2, or 3 tablets per day. One might argue that the
patients getting more tablets per day might demonstrate greater
placebo or nocebo effects or be less compliant. Nonetheless,
the results of this trial did not illustrate a difference in efficacy,

Table 4. Patient-reported compliance: the placebo arms are designated as those that had target doses of 1, 2, or 3 tablets per day to correspond to
gabapentin doses of 300, 600, and 900 mg/day, respectively

Placebo (1/day) Placebo (2/day) Placebo (3/day) 300 mg/day 600 mg/day 900 mg/day Total P valuea
Compliance week 1 0.21
Missing 0 0 0 2 1 3 6
Yes 17 (100%) 15 (94%) 16 (94%) 49 (100%) 42 (89%) 43 (96%) 182 (95%)
No 0 (0%) 1 (6%) 1 (6%) 0 (0%) 5 (11%) 2 (4%) 9 (5%)
Compliance week 2 0.31
Missing 0 0 0 1 1 2 4
Yes 16 (100%) 15 (94%) 17 (100%) 49 (100%) 44 (98%) 44 (100%) 185 (99%)
No 0 (0%) 1 (6%) 0 (0%) 0 (0%) 1 (2%) 0 (0%) 2 (1%)
Compliance week 3 0.33
Missing 0 0 0 2 0 3 5
Yes 16 (100%) 14 (93%) 17 (100%) 48 (98%) 42 (93%) 43 (100%) 180 (97%)
No 0 (0%) 1 (7%) 0 (0%) 1 (2%) 3 (7%) 0 (0%) 5 (3%)
Compliance week 4 0.16
Missing 0 1 0 6 2 3 12
Yes 15 (100%) 13 (93%) 17 (100%) 44 (100%) 39 (91%) 41 (98%) 169 (97%)
No 0 (0%) 1 (7%) 0 (0%) 0 (0%) 4 (9%) 1 (2%) 6 (3%)

Compliant patient status was defined by the patient stating that he took the correct number of daily pills, with the given percentage applying to all patients
that answered that question.
a
Each week is a 2 · 6 two-way table [compliance (yes, no) versus arm (placebo 1/day, placebo 2/day, placebo 3/day, 300 mg/day, 600 mg/day, 900 mg/day)]
and each P value is for a chi-squared test with 5 degrees of freedom (for the week).

Volume 20 | No. 3 | March 2009 doi:10.1093/annonc/mdn644 | 547

original article
Men current
900 mg/d P=0.10
Placebo
Guttuso et al
900 mg/d
P=0.02
Placebo
Pandya et al
900 mg/d
P=0.0001
Placebo
0 20 40 60 80 100 120
Residual hot flash score at 4 weeks (%)
Figure 4. Cross study comparisons of data from two studies of gabapentin
in women [16, 17] and the current one in men. Error bars represent 95%
confidence intervals.
presumed toxicity, or compliance among the three subsets of
the placebo arm (Figure 4).
It is reasonable to ask whether this current project should be
considered as a positive clinical study. On the one hand, it
could be argued that the primary analysis written in the
statistical section of this protocol, that being a comparison of
the combined gabapentin arms to the placebo arm, did not
reveal a statistically significant difference, and thus this is
a negative clinical study. On the other hand, it can easily be
argued that this is not the best analytical means to look at
a study such as this. It was expected that there was likely to be
a dose–response effect for the chosen low doses of gabapentin
and that only the higher doses might have shown a positive
result, as was seen in another similarly designed clinical trial
conducted by the NCCTG [14] and in a placebo-controlled
double-blind, randomized trial in women examining
gabapentin at 300 and 900 mg/day [16]. This was, in fact, what
was observed.
Facts that support that gabapentin moderately decreases
hot flashes in men include a statistically significant comparison
of hot flash frequencies and a trend for improvement in hot
flash scores between the 900 mg/day and placebo arms;
statistically significant improvements of hot flash frequencies
and scores between the 900 mg/day and 300 mg/day arms; that
dose–response relationships are apparent in Figures 1 and 2;
that the results in men look so similar to what was seen in three
trials in women [16–18]; that there were significant differences
for satisfaction with hot flash control and the effect of hot
flashes on QoL for the combined gabapentin arms compared
with the placebo arm, and that there were significant differences
between the highest gabapentin dose arm and the placebo
arm in changes in the week 4 values, from the baseline week, for
hot flash distress, satisfaction with hot flash control, and the
effect of hot flashes on QoL.
Thus, it appears reasonable to utilize gabapentin in clinical
practice for hot flashes in men, noting that it would be ideal
to have further data to substantiate its benefit. Nonetheless,
the magnitude of benefit with this therapy appears to be
relatively moderate. More information about gabapentin’s
ability to relieve hot flashes in men is desirable. Better therapies
for alleviating hot flashes in men on androgen deprivation
therapy are still needed.
548 | Loprinzi et al.
Annals of Oncology
funding
Public Health Service (CA-25224, CA-37404, CA-35103, CA-
63849, CA-63848, CA-35195, CA-35272, CA-35269, CA-35101,
CA-60276, CA-52352, CA-37417, CA-35448); National Cancer
Institute (U10CA037404).
acknowledgements
The content is solely the responsibility of the authors and does
not necessarily represent the official views of the National
Cancer Institute or the National Institutes of Health. This study
was conducted as a collaborative trial of the North Central
Cancer Treatment Group and Mayo Clinic. Additional
participating institutions include Iowa Oncology Research
Association CCOP, Des Moines, IA 50314 (Roscoe F. Morton,
MD); Rapid City Regional Oncology Group, Rapid City, SD
59709 (Richard C. Tenglin, MD); CentraCare Clinic, St Cloud,
MN 56301 (Harold E. Windschitl, MD); Quain and Ramstad
Clinic, Bismarck, ND 58506 (Edward J. Wos, DO); Sioux
Community Cancer Consortium, Sioux Falls, SD 57105 (Loren
K. Tschetter, MD); Missouri Valley Cancer Consortium,
Omaha, NE 68106 (Gamini S. Soori, MD); Siouxland
Hematology-Oncology Associates, Sioux City, IA 51105
(Donald B. Wender, MD); Wichita Community Clinical
Oncology Program, Wichita, KS 67214-3882 (Shaker R. Dakhil,
MD); Illinois Oncology Research Assn. CCOP, Peoria, IL
61615-7828 (John W. Kugler, MD); Montana Cancer
Consortium, Billings, MT 59101 (Benjamin T. Marchello, MD);
Cedar Rapids Oncology Project CCOP, Cedar Rapids, IA 52403
(Martin Wiesenfeld, MD); Allegheny General Hospital,
Pittsburgh, PA 15212 (Jane M. Raymond, MD); Columbus
CCOP, Columbus, OH 53215 (J. Philip Kuebler, MD, PhD);
Hawaii Minority-Based CCOP (William S. Loui, MD); and
Howard University Cancer Center, Washington, DC 20060
(Lucile Adams-Campbell, PhD).
references
1. Schow DA, Renfer LG, Rozanski TA, Thompson IM. Prevalence of hot flushes
during and after neoadjuvant hormonal therapy for localized prostate cancer.
South Med J 1998; 91: 855–857.
2. Charig CR, Rundle JS. Flushing. Long-term side effect of orchiectomy in
treatment of prostatic carcinoma. Urology 1989; 33: 175–178.
3. Loprinzi C, Goldberg RM, O’Fallon JR et al. Transdermal clonidine for
ameliorating postorchiectomy hot flashes. J Urol 1993; 151: 634–636.
4. Loprinzi CL, Michalak JC, Quella SK et al. Megestrol acetate for the prevention of
hot flashes. N Engl J Med 1994; 331: 347–352.
5. Atala A, Amin M, Harty JI. Diethylstilbestrol in treatment of postorchiectomy
vasomotor symptoms and its relationship with serum follicle-stimulating
hormone, luteinizing hormone, and testosterone. Urology 1992; 39: 108–110.
6. Burch PA, Loprinzi CL. Prostate-specific antigen decline after withdrawal of low-
dose megestrol acetate. J Clin Oncol 1999; 17: 1087–1088.
7. Guttuso TJ Jr. Gabapentin’s effects on hot flashes and hypothermia. Neurology
2000; 54: 2161–2163.
8. Pandya KJ, Thummala AR, Griggs JJ et al. Pilot study using gabapentin for
tamoxifen-induced hot flashes in women with breast cancer. Breast Cancer Res
Treat 2004; 83: 87–89.
9. Loprinzi C, Barton D, Sloan J et al. Pilot evaluation of gabapentin for treating hot
flashes. Mayo Clin Proc 2002; 77: 1155–1163.
Volume 20 | No. 3 | March 2009
Annals of Oncology
10. Pocock SJ, Simon R. Sequential treatment assignment with balancing for
prognostic factors in the controlled clinical trial. Biometrics 1975; 31: 103–115.
11. McNair DM, Loor M, Droppleman LF. Profile of Mood States Manual. New York:
Multi-Health Systems 2003.
12. Quella S, Loprinzi CL, Dose AM. A qualitative approach to defining ‘‘hot flashes’’
in men. Urol Nurs 1994; 14: 155–158.
13. Sloan JA, Loprinzi CL, Novotny PJ et al. Methodologic lessons learned from hot
flash studies. J Clin Oncol 2001; 19: 4280–4290.
14. Loprinzi CL, Kugler JW, Sloan JA et al. Venlafaxine in management of hot flashes
in survivors of breast cancer: a randomised controlled trial. Lancet 2000; 356:
2059–2063.
15. Cohen J. Statistical Power Analysis for the Behavioral Sciences. Hillsdale, NJ:
Lawrence Erlbaum Associates, Inc 1988.
16. Pandya KJ, Morrow GR, Roscoe JA et al. Gabapentin for hot flashes in 420
women with breast cancer: a randomised double-blind placebo-controlled trial.
Lancet 2005; 366: 818–824.
17. Guttuso T Jr., Kurlan R, McDermott MP, Kieburtz K. Gabapentin’s effects on hot
flashes in postmenopausal women: a randomized controlled trial. Obstet Gynecol
2003; 101: 337–345.
18. Butt DA, Lock M, Lewis JE et al. Gabapentin for the treatment of menopausal hot
flashes: a randomized controlled trial. Menopause 2008; 15: 310–318.
19. Rao RD, Michalak JC, Sloan JA et al. Efficacy of gabapentin in the management of
chemotherapy-induced peripheral neuropathy: a phase 3 randomized, double-blind,
placebo-controlled, crossover trial (N00C3). Cancer 2007; 110: 2110–2118.
Volume 20 | No. 3 | March 2009
original article
20. Goldberg RM, Loprinzi CL, O’Fallon JR et al. Transdermal clonidine for
ameliorating tamoxifen-induced hot flashes. J Clin Oncol 1994; 12: 155–158.
21. Barton DL, Loprinzi CL, Quella SK et al. Prospective evaluation of vitamin E for
hot flashes in breast cancer survivors. J Clin Oncol 1998; 16: 495–500.
22. Quella S, Loprinzi C, Barton D. Evaluation of soy phytoestrogens for the
treatment of hot flashes in breast cancer survivors. J Clin Oncol 1999; 16:
495–500.
23. Loprinzi CL, Sloan JA, Perez EA et al. Phase III evaluation of fluoxetine for
treatment of hot flashes. J Clin Oncol 2002; 20: 1578–1583.
24. Pockaji B, Gallagher J, Loprinzi C et al. Phase III double-blinded, randomized trial
to evaluate the use of black cohosh in the treatment of hot flashes: a North
Central Cancer Treatment Group Study. Proc Am Soc Clin Oncol 2005; 23:
732S.
25. Loprinzi C, Levitt R, Sloan J et al. Medroxyprogesterone acetate (MPA) versus
venlafaxine for hot flashes: a North Central Cancer Treatment Group Trial.
2005 Annual Meeting of the American Society of Clinical Oncol. J Clin Oncol
2005 vol 23, No 16s:732s (Abstr 8014).
26. Stearns V, Beebe KL, Iyengar M, Dube E. Paroxetine controlled release in the
treatment of menopausal hot flashes: a randomized controlled trial. JAMA 2003;
289: 2827–2834.
27. Stearns V, Slack R, Greep N et al. Paroxetine is an effective treatment for hot
flashes: results from a prospective randomized clinical trial. J Clin Oncol 2005;
23: 6919–6930.
doi:10.1093/annonc/mdn644 | 549
Author's Accepted Manuscript

Prophylactic Sildenafil Citrate Improves Selected Aspects of Sexual Function in

Men Treated by Radiotherapy for Prostate Cancer

Michael J. Zelefsky, Daniel Shasha, Rebekah Dunn Branco, Marisa Kollmeier,

Raymond E. Baser, Xin Pei, Ronald Ennis, Richard Stock, Natan Bar-Chama, John
P. Mulhall

PII: S0022-5347(14)00355-3
DOI: 10.1016/j.juro.2014.02.097
Reference: JURO 11150

To appear in: The Journal of Urology

Accepted Date: 21 February 2014

Please cite this article as: Zelefsky MJ, Shasha D, Branco RD, Kollmeier M, Baser RE, Pei X, Ennis R,
Stock R, Bar-Chama N, Mulhall JP, Prophylactic Sildenafil Citrate Improves Selected Aspects of Sexual
Function in Men Treated by Radiotherapy for Prostate Cancer, The Journal of Urology® (2014), doi:
10.1016/j.juro.2014.02.097.

DISCLAIMER: This is a PDF file of an unedited manuscript that has been accepted for publication. As a
service to our subscribers we are providing this early version of the article. The paper will be copy edited
and typeset, and proof will be reviewed before it is published in its final form. Please note that during the
production process errors may be discovered which could affect the content, and all legal disclaimers
that apply to The Journal pertain.

All press releases and the articles they feature are under strict embargo until uncorrected proof of
the article becomes available online. We will provide journalists and editors with full-text copies of the
articles in question prior to the embargo date so that stories can be adequately researched and written.
The standard embargo time is 12:01 AM ET on that date.
 ACCEPTED MANUSCRIPT

Original Article

Prophylactic Sildenafil Citrate Improves Selected Aspects of Sexual Function in Men Treated

by Radiotherapy for Prostate Cancer

PT
Michael J. Zelefsky, MD,* Daniel Shasha, MD, Rebekah Dunn Branco, MA, MBA,

Marisa Kollmeier, MD, Raymond E. Baser, MS, Xin Pei, PhD, Ronald Ennis, MD, Richard

RI
Stock, MD, Natan Bar-Chama, MD, and John P Mulhall, MD, MSc, FECSM, FACS

SC
From The Department of Radiation Oncology and Urology (MJZ, RBD, MK, XP, JPM),

U
Memorial Sloan-Kettering Cancer Center, the Department of Radiation Oncology (DS, RE),
AN
Continuum Cancer Center of New York, the Department of Radiation Oncology and Urology

(RS, NB-C), Mt Sinai Medical Center, and the Department of Epidemiology and Biostatistics
M

(REB), Memorial Sloan-Kettering Cancer Center, New York, New York

D
TE

* Correspondence: Michael J. Zelefsky, MD, Memorial Sloan-Kettering Cancer Center, 1275

York Avenue, New York, NY 10065. Phone: 212-639-6802. Fax: 212-639-8876. E-mail:
EP

zelefskm@mskcc.org.
C

Number of words: 2594

AC

Running Title: Prophylactic Sildenafil for ED after RT

Key Words: Prostate Cancer, Radiotherapy, Sildenafil Citrate, Erectile Dysfunction

1
 ACCEPTED MANUSCRIPT

Funding/Support: Pfizer provided an educational grant that was used for research

assistance support and supplied drug for the study.

PT
RI
U SC
AN
M
D
TE
C EP
AC

2
 ACCEPTED MANUSCRIPT

ABSTRACT

Purpose: Adjuvant daily sildenafil citrate (SC) during and after radiotherapy for prostate cancer

for erectile function preservation was studied.

Methods: A randomized, prospective trial of 279 patients with localized prostate cancer treated

PT
with radiotherapy compared daily SC (50 mg) to placebo (2:1 randomization).

RI
Medication/placebo was initiated 3 days pretreatment and continued daily for 6 months. Patients

completed the International Index of Erectile Function (IIEF) including the erectile function

SC
domain (EF) and International Prostate Symptom Score questionnaires pretherapy and the Rand

36-Item Health Survey at 3, 6, 9, 12, 18, and 24 months post-radiotherapy. All IIEF domains

were scored.
U
AN
Results: At 12 months, EF scores were better for SC (P=0.018) than placebo, 73% of SC vs 50%

of placebo patients had mild/no erectile dysfunction (ED) (P=0.024), and the SC arm had
M

superior overall satisfaction (P=0.028) and IIEF total scores (P=0.044). At 24 months, EF and
D

IIEF scores were no longer significantly better for SC (P=0.172 and 0.09, respectively), yet
TE

overall satisfaction scores were higher (P=0.032). SC sexual desire scores were higher (P=0.049)

at 24 months despite completing drug therapy 18 months prior. At 24 months, 81.6% of SC and
EP

56.0% of placebo patients (P=0.045) had functional erections (with or without ED medication).

Conclusions: Daily SC during and after radiotherapy for prostate cancer was associated with
C

improved overall sexual function compared with placebo for various sexual function domains.
AC

This is the largest randomized, prospective, controlled trial to demonstrate utility of a PDE5

inhibitor as a rehabilitation strategy in prostate cancer radiation therapy patients.

3
 ACCEPTED MANUSCRIPT

ERECTILE dysfunction (ED) is recognized as a significant treatment-related complication after

definitive treatment for prostate cancer such as surgery and radiotherapy (RT). ED is a significant

concern for patients as they select therapy, and can negatively impact overall quality of life. After

RT, the overall reported ED incidence is 24-59%, and sexual function often begins to decline only

PT
12-18 months following completion of therapy.[1]

RI
The importance of penile rehabilitation after prostate cancer therapy has been evaluated

by several investigators [2-6], based on recognition that preservation of corpus cavernosal

SC
endothelial and smooth muscle integrity might improve long-term sexual function. Several

animal studies have demonstrated a positive effect for regular use of a PDE5i such as sildenafil

U
citrate (SC) through neuro- and vascular-protective effects on tissues involved in erectile
AN
function.[7] In animals chronically treated with PDE5i, reduced deposition of collagen within the

corpora cavernosa and endothelial cell preservation has been observed in penile tissues.[8]
M

Clinical studies suggest a benefit to rehabilitation in radical prostatectomy (RP) patients. In

D

one retrospective case-matched study,[4] RP patients were given early SC within 6 months of

surgery or after 6 months from surgery. Improved erectile function was noted post-therapy
TE

among early-SC patients compared with historical controls (58% vs 30%). In a randomized trial
EP

evaluating penile rehabilitation among RP patients, 628 men were randomized to receive

placebo, nightly vardenafil, or on-demand vardenafil. The study had a 9-month double-blind
C

treatment period, and the primary outcome measure was the percentage of subjects with IIEF-EF
AC

score ≥22. No statistically significant differences were observed between groups.[5]

For RP patients, a significant component of post-treatment ED is related to neural injury,

but the etiology of post-RT function appears to be more related to vascular (endothelial)

dysfunction.[9] We hypothesized that, in RT patients, adjuvant SC may demonstrate more

significant improvements given its recognized vascular-protective effects. In one study that was

4
 ACCEPTED MANUSCRIPT

grossly underpowered, including 27 patients randomized, no benefits were observed for the use

of prophylactic SC.[10] To prospectively study the impact of prophylactic SC among prostate RT

patients with pretherapy functional erections, we undertook the current study. The primary

objective was to determine if there is a role for penile rehabilitation among potent patients

PT
receiving definitive RT for prostate cancer, specifically if prophylactic daily SC during and after

RI
RT can preserve sexual function 24 months from the start of RT.

SC
MATERIALS AND METHODS

This is a double-blind, placebo-controlled trial to assess whether concurrent and adjuvant SC in

U
conjunction with RT for clinically localized prostate cancer will improve sexual function
AN
following RT at 24 months post-treatment.
M

Patient Population
D

Two hundred seventy-nine patients with organ-confined prostate cancer were randomized.
TE

Eligible patients included those who were to receive definitive treatment with external beam

radiation therapy (EBRT), brachytherapy, or brachytherapy combined with EBRT, who also had
EP

a baseline International Index of Erectile Function (IIEF) score ≥17. Following enrollment of

≈50 patients, eligibility criteria were expanded to broaden accrual, allowing patients receiving
C

neo-adjuvant androgen-deprivation therapy (ADT), starting within 1 month prior to enrollment.

AC

These patients were included to determine if a benefit would be observed despite the sexual

depressive effects associated with ADT.

Patients using erectogenic agents >4 times per month or those taking 0.8 mg of

tamsulosin daily pretherapy were excluded. Also excluded were patients with a clinically

5
 ACCEPTED MANUSCRIPT

significant penile deformity, a history of penile implant, or prior history of prostate cancer

therapy, or if SC was contraindicated. The patients were enrolled across three institutions. There

were 279 randomized patients and 202 initiated treatment with the study drug or placebo,

including 218 and 162, respectively, at Memorial Sloan-Kettering Cancer Center (MSKCC); 48

PT
and 30 at Continuum Cancer Center; and 13 and 10 at Mt Sinai School of Medicine.

RI
This research protocol was approved by the institutional review board from each

participating institution and all participants provided written informed consent prior to

SC
enrollment.

Those not taking the study drug either decided against RT following randomization or

U
otherwise withdrew without taking the first dose. Table 1 provides baseline demographic,
AN
disease, and comorbidity information for participants who initiated therapy. The median EBRT

dose was 86.4 Gy when given as monotherapy and 50.4 Gy when given with brachytherapy.
M

There were no significant differences between the sildenafil and placebo groups as far as the
D

prescription dose levels delivered to the prostate (i.e. percentage who underwent EBRT alone
TE

versus BRT or combined modality treatment). When ADT was used, luteinizing hormone-

releasing hormone agonist in a neoadjuvant and concurrent setting was often given in
EP

combination with anti-testosterone therapy, and in general ADT was discontinued after

completion of RT.
C
AC

Treatment

Patients were randomized 2:1 to receive daily orally administered study therapy with SC 50 mg

or placebo. For ADT patients, study drug was initiated simultaneously with or within 1 month

after starting of ADT. Participants not receiving ADT began study therapy between 3 days prior

6
 ACCEPTED MANUSCRIPT

to starting RT through 2 weeks following the first fraction of radiation or implantation of

brachytherapy. All participants were instructed to continue daily administration of SC or placebo

until 6 months following the start of RT. Patients were asked to keep a daily pill diary to

document compliance.

PT
RI
Assessments and Follow-up

Participants were evaluated for changes in sexual function, urinary symptoms, quality-of-life

SC
measures, and toxicities for the duration of study therapy and for 2 years after initiation of RT.

Urinary symptoms and quality of life were measured through patient-reported questionnaires

U
(International Prostate Symptom Score (IPSS), and the Rand 36-Item Health Survey (QOL),
AN
respectively). Sexual function was assessed by the IIEF-6 and by patient self-reports at follow-up

visits. IIEF domains were evaluated and scored for each treatment, including Erectile Function
M

(EF), Orgasmic Function (OF), Sexual Desire (SD), Intercourse Satisfaction (IS), and Overall
D

Satisfaction (OS). Patient characteristics including age, use of brachytherapy, use of ADT, and
TE

baseline IIEF scores were balanced between treatment groups.

Participants were assessed for these outcomes and quality-of-life measures at baseline
EP

and 3, 6, 9, 12, 18, and 24 months following start of RT. Treatment-related toxicities were

assessed throughout the course of therapy. Baseline assessments occurred in the 4-week interval
C

prior to starting study drug.

AC

Response Criteria

The primary objective was to determine if prophylactic SC used during and after a course of RT

(EBRT and/or brachytherapy with or without ADT) can prevent loss of spontaneous erectile

7
 ACCEPTED MANUSCRIPT

function 2 years from the start of RT, evaluated based on responses to the IIEF at 2 years post-

RT. We also assessed the response among only patients not treated with ADT at both 1 and 2

years. A four-point change in IIEF score vs baseline was considered significant in erectile

function.[11] An additional analysis was performed on patients treated at MSKCC, where the

PT
information was more readily available to assess erectile function based on participant self-report

RI
at regular clinical follow-ups, and graded according to the Common Terminology Criteria for

Adverse Events (CTCAE) v 3.0. Participants with grade 0-1 ED were considered fully

SC
functional, grade 2 ED partially potent, and grade 3 ED non-functional.

Acute and late urinary side effects were assessed by IPSS scores and participant self-

U
report in the clinic. Acute toxicity was defined as any toxicity occurring within 4 months from
AN
the completion of therapy. Late toxicities were defined as those observed after 4 months
M

following completion of therapy.

D

Statistical Methods

Outcome analysis was based on the modified intention-to-treat principle in which data from all
TE

randomized patients who received RT were analyzed. Baseline demographic and medical
EP

characteristics of the treatment groups were compared using χ2 tests for categorical variables and

t-tests for continuous variables.

C

Average IIEF total and subscale scores and IPSS total scores were summarized for each
AC

treatment arm at baseline and each follow-up assessment time (e.g., 12-months, 24-months, etc.)

using median values. Score variability was summarized by the first and third quartiles of the

score distributions. Wilcoxon rank sums tests assessed statistical differences between study arms

in their score distributions at each assessment time. Differences were evaluated with two-sided

tests at baseline and one-sided tests at each follow-up, with the one-sided alternative hypothesis

8
 ACCEPTED MANUSCRIPT

that SC patients had more functional scores than placebo patients. Proportions of MSKCC

patients potent at clinical follow-ups corresponding to the study assessment times, and

proportions of patients experiencing acute and late GU toxicities were compared using Fisher

exact tests. For each scale, effect of treatment arm on scores assessed at 12 months and later,

PT
controlling for baseline score and age, was evaluated using linear regression models estimated

RI
with generalized estimating equations accounting for multiple scores per patient (i.e., at 12, 18,

and 24 months). For all tests, P≤0.05 was considered statistically significance, while P≤0.10 was

SC
considered marginally statistically significant.

U
RESULTS
AN
Among patients treated with or without ADT, SC patients tended to have higher EF and IIEF total

scores at 24 months than placebo patients, but these differences were not significant (Table 2). SC
M

patients had significantly higher OS subscale scores at 24 months (P=0.048).

As shown in Table 3, ADT patients tended to experience worse erectile function outcomes
D

than non-ADT patients (Table 4), regardless of treatment arm. Additionally, the IIEF total and
TE

subscale scores among the ADT patients were considerably more variable at each time-point

than among non-ADT patients. These data suggest that sexually depressive effects of ADT
EP

mitigated potential benefits of prophylactic SC.

C

For the rest of the analysis we focused on the overwhelming majority of patients not
AC

receiving ADT (90%). In these there were no significant erectile function differences between

treatment arms at baseline. In general, SC patients had significantly better scores for all domains

at 3 and 6 months post-therapy, with these differences diminishing with time (Table 5, Figs 1 and

2). However, significant differences persisted at 12 and 24 months. At 12 months, EF scores of

SC patients (median, 26.0; IQR, 19.3-29.0) were significantly better (P=0.018) than those of

9
 ACCEPTED MANUSCRIPT

placebo patients (21.5, 14.0-28.0), and 73% of SC patients had only mild or no ED by EF score

vs 50% of placebo patients (P=0.024). The SC arm reported superior overall satisfaction (OS)

subscale (P=0.028) and IIEF total scores (P=0.044) at 12 months. EF scores were no longer

significantly better at 24 months for SC patients (P=0.172), but OS scores were significantly

PT
higher (P=0.032) and IIEF scores marginally but not significantly higher (P=0.097). SC patients

RI
also experienced significantly higher sexual desire (SD) scores at 24 months (P=0.049)

compared with placebo despite having completed drug therapy 18 months prior.

SC
At 3 and 6 months, 91.4% and 78.4% of SC patients reported functional erections and were

not taking any medications for ED at their physician visits vs 63.6% and 47.6% for placebo (p =

U
0.015 and p = 0.022, respectively). At 24 months, 81.6% of SC and 56.0% of placebo patients (p =
AN
0.045) reported functional erections (with or without ED medication). We observed no differences

in erectile function outcomes among EBRT patients compared with brachytherapy or a combined-
M

modality treatment.
D

There were no significant differences in acute or late treatment-related toxicities between

the groups. Among SC patients, 20.7% had no acute GU toxicity, 52.6% had grade 1 and 26.7%
TE

had grade 2-3 acute GU toxicities vs 11.7% of control patients with no toxicity, 54.5% with
EP

grade 1, and 33.8% with grade 2-3 acute GU toxicity (P=0.194). 39.7%, 41.3%, and 19.0% of SC

patients experienced no, grade 1, and grade 2-3 late toxicities, respectively, compared with
C

11.7%, 54.5%, and 33.8% of control patients, respectively (P=0.576). The two study arms also
AC

had similar IPSS scores at baseline and at each follow-up study assessment. Drug-related

symptoms were more often reported in the SC group compared with the placebo cohort.

Specifically, SC patients more often complained of dizziness (11% vs 0%), headaches (18% vs

5%), blurred vision (6% vs 5%), and flushing (17% vs 8%).

10
 ACCEPTED MANUSCRIPT

DISCUSSION

Our findings are the first indications in a randomized trial of evidence for a benefit using

prophylactic SC in prostate cancer patients treated with RT. We observed significant

improvements in various sexual function domains at 6 and 12 months post-therapy, and

PT
specifically for OS and SD. In addition, these differences persisted as far as 24 months after

RI
therapy. We view these important findings as proof of the principle that prophylactic SC can play

a critical role in erectile function preservation for prostate cancer patients treated with RT. The

SC
greatest improvement was at 6 and 12 months after treatment, suggesting that our 6-month usage

of prophylactic SC may not be of adequate duration. At 24 months however the differences were

U
less pronounced yet with persistently improved desire and satisfaction in the SC arm noted, and
AN
at 24 months, 82% of SC and 56.0% of placebo patients (p = 0.045) reported functional erections

(with or without ED medication)..A longer course of SC (> 6 months) may be required to provide
M

even better functional outcomes beyond 12 months from therapy. Only further prospective
D

studies will clarify this issue.

TE

Our results support the concept that PDE5 inhibitors might limit the damage radiation

inflicts on vascular tissue in cavernosal tissue. Desouza et al published data that unequivocally
EP

demonstrated the positive effect of sildenafil on endothelial function.[12] In patients with ED

and type 2 diabetes, sildenafil improved flow-mediated dilation. This has also been demonstrated
C

using tadalafil.[13] The duration of effect of these agents on endothelial function was shown to
AC

be longer than the presence of either drug within the blood, suggesting that the impact of PDE5

inhibitors on endothelium at the cellular level may be self-sustaining. Furthermore, all

commercially available PDE5i have been shown to generate endothelial progenitor cells from the

bone marrow,[14] suggesting another mechanism by which PDE5i may protect endothelium. As

patients in this study were not prospectively followed with penile duplex Doppler studies, we

11
 ACCEPTED MANUSCRIPT

could not validate our hypothesis that the erectile function improvements we observed with SC

was related to improved vascular function. We recognize a limitation of this study is that not all

patients underwent serial testosterone measurements which could have detected in some

testosterone deficiency syndrome. Nevertheless, we would not expect, in the non-hormonal

PT
cohort of patients (where the greatest benefit of prophylactic SC was observed) any differences

RI
in testosterone reductions as a result of localized radiotherapy to the prostate.

PDE5i have been used as the backbone to penile rehabilitation interventions following

SC
RP.[4] While definitive data does not yet exist proving their utility in this population, two of

three randomized, placebo-controlled trials have demonstrated a signal that regular PDE5i use

U
results in improved rates of men returning to their baseline preoperative erectile function
AN
(sildenafil)[15] and earlier return to PDE5i response (tadalafil).[16] Therefore, it seems

reasonable to explore such agents in RT patients to limit the negative effects of RT on

M

endothelium and improve erectile function preservation. Schiff et al analyzed patients who
D

underwent prostate brachytherapy who subsequently used PDE5 inhibitors on a regular

TE

basis.[17] Patients were stratified into early-use (<1 year post-brachytherapy) and late-use groups

(>1 year post-brachytherapy). Higher rates of erectile function preservation, based on sexual
EP

health inventory for men (SHIM) scores, were seen in the early-use group.

Our study also suggests that prophylactic SC was not as beneficial among patients treated
C

with ADT in conjunction with RT. Due to the limited number of these patients in our study, it is
AC

difficult to draw conclusions about the value of penile rehabilitation among patients with a

depressed libido as a result of ADT. Further studies in this population are warranted and longer

duration of adjuvant usage well beyond testosterone recovery may be necessary to achieve more

optimal outcomes.

12
 ACCEPTED MANUSCRIPT

CONCLUSIONS

Our findings indicate that daily use of SC during and for 6 months after initiation of

prostate cancer RT is associated with improved sexual function outcomes. Patients reported

PT
significantly improved sexual functions for SC compared to placebo. Radiotherapy-induced

RI
trauma to the vascular supply of the cavernosal tissue may undergo rehabilitation and repair

during administration of SC. Longer SC usage post-RT may further improve sexual function and

SC
will require further prospective studies.

U
AN
M
D
TE
C EP
AC

13
 ACCEPTED MANUSCRIPT

Abbreviations and Acronyms

ADT = androgen-deprivation therapy

EBRT = external beam radiation therapy

ED = erectile dysfunction

PT
EF = erectile function

RI
GU = genitourinary

IIEF = International Index of Erectile Function

SC
IPSS = International Prostate Symptom Score

IQR = interquartile range

IS = intercourse satisfaction
U
AN
MSKCC = Memorial Sloan-Kettering Cancer Center

OF = orgasmic function
M

OS = overall satisfaction
D

PSA = prostate-specific antigen

TE

Q1 = first quartile

Q3 = third quartile
EP

QOL = quality of life

RP = radical prostatectomy
C

RT = radiotherapy
AC

SC = sildenafil citrate

SD = sexual desire

14
 ACCEPTED MANUSCRIPT

REFERENCES

1. van der Wielen GJ, Mulhall JP, Incrocci L (2007) Erectile dysfunction after radiotherapy

for prostate cancer and radiation dose to the penile structures: a critical review. Radiother Oncol

84:107-113

PT
2. Montorsi F, Guazzoni G, Strambi LF, Da Pozzo LF, Nava L, Barbieri L, Rigatti P,

RI
Pizzini G, Miani A (1997) Recovery of spontaneous erectile function after nerve-sparing radical

retropubic prostatectomy with and without early intracavernous injections of alprostadil: results

SC
of a prospective, randomized trial. J Urol 158:1408-1410

3. Mulhall J, Land S, Parker M, Waters WB, Flanigan RC (2005) The use of an erectogenic

U
pharmacotherapy regimen following radical prostatectomy improves recovery of spontaneous
AN
erectile function. J Sex Med 2:532-540; discussion 540-532

4. Mulhall JP, Parker M, Waters BW, Flanigan R (2010) The timing of penile rehabilitation
M

after bilateral nerve-sparing radical prostatectomy affects the recovery of erectile function. BJU
D

Int 105:37-41
TE

5. Montorsi F, Brock G, Lee J, Shapiro J, Van Poppel H, Graefen M, Stief C (2008) Effect

of nightly versus on-demand vardenafil on recovery of erectile function in men following

EP

bilateral nerve-sparing radical prostatectomy. Eur Urol 54:924-931

6. McCullough AR, Hellstrom WG, Wang R, Lepor H, Wagner KR, Engel JD (2010)
C

Recovery of erectile function after nerve sparing radical prostatectomy and penile rehabilitation
AC

with nightly intraurethral alprostadil versus sildenafil citrate. J Urol 183:2451-2456

7. Mulhall JP, Slovick R, Hotaling J, Aviv N, Valenzuela R, Waters WB, Flanigan RC

(2002) Erectile dysfunction after radical prostatectomy: hemodynamic profiles and their

correlation with the recovery of erectile function. J Urol 167:1371-1375

15
 ACCEPTED MANUSCRIPT

8. Vignozzi L, Filippi S, Morelli A, Ambrosini S, Luconi M, Vannelli GB, Donati S,

Crescioli C, Zhang XH, Mirone V, Forti G, Maggi M (2006) Effect of chronic tadalafil

administration on penile hypoxia induced by cavernous neurotomy in the rat. J Sex Med 3:419-

431

PT
9. Zelefsky MJ, Eid JF (1998) Elucidating the etiology of erectile dysfunction after

RI
definitive therapy for prostatic cancer. Int J Radiat Oncol Biol Phys 40:129-133

10. Ilic D, Hindson B, Duchesne G, Millar JL (2013) A randomised, double-blind, placebo-

SC
controlled trial of nightly sildenafil citrate to preserve erectile function after radiation treatment

for prostate cancer. J Med Imaging Radiat Oncol 57:81-88

11.
U
Mulhall J (2003) Deciphering erectile dysfunction drug trials. J Urol 170:353-358
AN
12. Desouza C, Parulkar A, Lumpkin D, Akers D, Fonseca VA (2002) Acute and prolonged

effects of sildenafil on brachial artery flow-mediated dilatation in type 2 diabetes. Diabetes Care
M

25:1336-1339
D

13. Rosano GM, Aversa A, Vitale C, Fabbri A, Fini M, Spera G (2005) Chronic treatment
TE

with tadalafil improves endothelial function in men with increased cardiovascular risk. Eur Urol

47:214-220; discussion 220-212

EP

14. Foresta C, De Toni L, Magagna S, Galan A, Garolla A (2010) Phosphodiesterase-5

inhibitor tadalafil acts on endothelial progenitor cells by CXCR4 signalling. Curr Drug Deliv
C

7:274-282
AC

15. Padma-Nathan H, McCullough AR, Levine LA, Lipshultz LI, Siegel R, Montorsi F,

Giuliano F, Brock G (2008) Randomized, double-blind, placebo-controlled study of

postoperative nightly sildenafil citrate for the prevention of erectile dysfunction after bilateral

nerve-sparing radical prostatectomy. Int J Impot Res 20:479-486

16
 ACCEPTED MANUSCRIPT

16. Montorsi F, Aversa A, Moncada I, Perimenis P, Porst H, Barker C, Shane M, Sorsaburu S

(2011) A randomized, double-blind, placebo-controlled, parallel study to assess the efficacy and

safety of once-a-day tadalafil in men with erectile dysfunction who are naïve to PDE5 inhibitors.

J Sex Med 8:2617-2624

PT
17. Schiff JD, Bar-Chama N, Cesaretti J, Stock R (2006) Early use of a phosphodiesterase

RI
inhibitor after brachytherapy restores and preserves erectile function. BJU Int 98:1255-1258

U SC
AN
M
D
TE
C EP
AC

17
 ACCEPTED MANUSCRIPT

Table 1. Demographic characteristics and baseline comorbidities data for patients who initiated

study therapy

Placebo Sildenafil Citrate

PT
N = 77 N = 125

RI
No. % No. %

SC
Demographics

Age (years)

≤65 50
U 65% 83 66%
AN
>65 27 35% 42 34%

Pretreatment PSA
M

<10 67 87% 113 90%

D

10-20 9 12% 9 7%
TE

>20 1 1% 1 1%

Gleason score
EP

6 44 57% 75 60%

7 29 38% 47 38%
C

>7 4 5% 3 2%
AC

RT type

EBRT 17 22% 25 20%

Brachytherapy 33 43% 57 46%

Combined EBRT + 27 35% 43 34%

18
 ACCEPTED MANUSCRIPT

brachytherapy

Hormone therapy

No ADT 69 90% 112 90%

ADT 8 10% 13 10%

PT
Baseline comorbidities

RI
Diabetes 5 6% 4 3%

Hypertension 33 43% 44 35%

SC
Heart disease 9 12% 12 10%

Smoking history 32 42% 57 46%

U
ADT, androgen-deprivation therapy; EBRT, external beam radiotherapy; PSA prostate-
AN
specific antigen; and RT, radiotherapy.
M
D
TE
C EP
AC

19
 ACCEPTED MANUSCRIPT

Table 2. Erectile Function (EF), International Index of Erective Function (IIEF) Total, and

Overall Satisfaction (OS) scores at baseline, 6, 12, and 24 months after therapy according to

treatment arm among patients treated with or without androgen-deprivation therapy

PT
Placebo Sildenafil Citrate

Time Points Median [Q1; Q3] Median [Q1; Q3] P Value

RI
Baseline EF 28.00 [25.00;29.50] 28.00 [25.00;30.00] 0.493

IIEF total 64.00 [56.75;69.00] 64.00 [57.00;71.00] 0.577

SC
OS 8.00 [7.00;10.00] 9.00 [7.25;10.00] 0.232

U
Month 6 EF 25.00 [11.00;28.00]
AN 27.00 [19.75;30.00] 0.036

IIEF total 58.00 [35.00;64.00] 59.00 [49.00;67.75] 0.091

OS 7.50 [4.25;8.00] 8.00 [7.00;9.00] 0.013

M

Month 12 EF 20.70 [13.25;27.75] 25.00 [18.50;29.00] 0.024

IIEF total 51.00 [34.50;63.50] 58.00 [41.50;66.75] 0.070

D

OS 6.00 [4.00;8.00] 8.00 [4.00;9.00] 0.069

TE

Month 24 EF 24.00 [8.75;29.00] 24.50 [14.00;29.00] 0.262

IIEF total 54.50 [29.75;64.75] 58.00 [39.00;65.00] 0.186

EP

OS 6.00 [4.00;8.00] 8.00 [5.00;9.00] 0.048

C

Q1, first quartile; Q3, third quartile. Baseline P values are from two-sided Wilcoxon rank sums
AC

tests. Post-baseline P values are from one-sided Wilcoxon rank sums tests with one-sided

alternative hypothesis that sildenafil citrate patients had higher scores than placebo patients.

20
 ACCEPTED MANUSCRIPT

Table 3. Erectile Function (EF), International Index of Erective Function (IIEF) Total, and

Overall Satisfaction (OS) scores at baseline, 6, 12 and 24 months after therapy according to

treatment arm in the androgen-deprivation therapy patients

PT
Placebo Sildenafil Citrate

Time Points Median [Q1; Q3] Median [Q1; Q3] P Value

RI
Baseline EF 26.00 [20.50;27.50] 27.00 [25.50;29.00] 0.173

IIEF total 56.00 [49.00;63.00] 61.00 [44.00;67.00] 0.965

SC
OS 7.00 [5.00;9.00] 6.00 [4.00;10.00] 0.624

U
Month 6 EF 22.50 [8.75;25.75]
AN 18.00 [6.50;24.50] 0.628

IIEF total 56.00 [56.00;57.00] 43.00 [17.00;51.00] 0.959

OS 7.00 [6.00;9.00] 5.50 [4.50;6.50] 0.908

M

Month 12 EF 9.00 [2.50;15.75] 17.00 [1.00;23.00] 0.274

IIEF total 22.00 [9.75;36.75] 19.00 [5.50;44.50] 0.666

D

OS 4.00 [3.50;5.00] 3.00 [2.25;5.25] 0.707

TE

Month 24 EF 21.00 [10.50;26.50] 15.00 [10.80;24.00] 0.711

IIEF total 52.00 [32.50;61.75] 38.50 [19.85;59.50] 0.759

EP

OS 6.50 [4.25;8.00] 7.00 [2.50;8.00] 0.566

C

Q1, first quartile; Q3, third quartile. Baseline P values are from two-sided Wilcoxon rank sums
AC

tests. Post-baseline P values are from one-sided Wilcoxon rank sums tests with one-sided

alternative hypothesis that sildenafil citrate patients had higher scores than placebo patients.

21
 ACCEPTED MANUSCRIPT

Table 4. Erectile Function (EF), IIEF total, and Overall Satisfaction (OS) scores at baseline, 6,

12, and 24 months after therapy according to treatment arm in the non-androgen-deprivation

therapy patients

Time Placebo Sildenafil Citrate

PT
points Median [Q1; Q3] Median [Q1; Q3] P Value

RI
Baseline EF 28.00 [25.00; 30.00] 28.00 [25.00; 30.00] 0.727

IIEF total 66.00 [58.00; 69.50] 65.00 [58.00; 71.00] 0.741

SC
OS 8.00 [7.25; 10.00] 9.00 [8.00; 10.00] 0.259

Month 6 EF 25.00 [11.50; 28.00] 27.00 [20.00; 30.00] 0.021

IIEF total
U
58.50 [34.50; 64.00] 60.50 [51.75; 69.00] 0.030
AN
OS 8.00 [4.00; 8.00] 8.00 [7.00; 9.50] 0.003

Month 12 EF 21.50 [14.00; 28.00] 26.00 [19.25; 29.00] 0.018

M

IIEF total 53.00 [37.70; 64.50] 60.00 [44.50; 67.50] 0.043

D

OS 7.00 [5.00; 8.00] 8.00 [5.25; 9.75] 0.027

TE

Month 24 EF 25.00 [9.00; 29.00] 25.00 [17.00; 29.00] 0.172

IIEF total 54.50 [33.00; 65.00] 58.50 [45.75; 66.00] 0.097

EP

OS 6.00 [4.50; 8.00] 8.00 [5.50; 9.50] 0.033

Q1, first quartile; Q3, third quartile. Baseline P values are from two-sided Wilcoxon rank sums
C

tests. Post-baseline P values are from one-sided Wilcoxon rank sums tests with one-sided
AC

alternative hypothesis that sildenafil citrate patients had higher scores than placebo patients.

22
 ACCEPTED MANUSCRIPT

Table 5. Linear regressions fit using generalized estimating equations testing effect of sildenafil

citrate (SC) arm on all scores 12 months or longer after radiotherapy for Erectile Function (EF),

IIEF Total, and Overall Satisfaction (OS) among non-androgen-deprivation therapy patients,

controlling for baseline scores and age

PT
Outcome Predictor Regression Standard

RI
Coefficient Error P Value

EF Intercept 19.879 0.996 <0.001

SC
Baseline EF1 0.739 0.163 <0.001

Age1 -0.055 0.077 0.473

SC Arm2 1.63
U 1.241 0.189
AN
IIEF Total Intercept 48.963 2.34 <0.001

Baseline IIEF
M

0.816 0.133 <0.001

Total1

Age1 -0.091 0.173 0.599

D

SC Arm2 4.145 2.773 0.135

TE

OS Intercept 6.55 0.289 <0.001

Baseline OS1 0.583 0.109 <0.001

EP

Age1 -0.005 0.022 0.827

SC Arm2 0.667 0.359 0.063

C

1
Baseline scores and age were centered around their means.
AC

2
The regression coefficients for the predictor “SC Arm” represent the average differences

between the SC and placebo arms in the scale scores assessed at 12 months or later from the start

of radiotherapy, controlling for baseline score and age. Positive values indicate better scores in

the SC arm.

23
 ACCEPTED MANUSCRIPT

PT
RI
U SC
AN
M
D
TE
C EP
AC

Fig 1. International Index of Erectile Function (IIEF) median scores and inter-quartile ranges

according to the assigned arms during the 2-year study period among the non-androgen-

deprivation patients.

24
 ACCEPTED MANUSCRIPT

PT
RI
U SC
AN
M
D
TE
C EP
AC

Fig 2. Erectile function (EF) median scores and inter-quartile ranges according to the assigned

arms during the 2-year study period among the non-androgen-deprivation patients.

25
 Prostate Cancer and Prostatic Disease (2014) 17, 180–186 OPEN
& 2014 Macmillan Publishers Limited All rights reserved 1365-7852/14
www.nature.com/pcan

ORIGINAL ARTICLE
A double-blind, placebo-controlled randomised trial evaluating the
effect of a polyphenol-rich whole food supplement on PSA
progression in men with prostate cancer—the UK NCRN Pomi-T
study
R Thomas1,2,3, M Williams1, H Sharma1, A Chaudry2 and P Bellamy3

BACKGROUND: Polyphenol-rich foods such as pomegranate, green tea, broccoli and turmeric have demonstrated anti-neoplastic
effects in laboratory models involving angiogenesis, apoptosis and proliferation. Although some have been investigated in small,
phase II studies, this combination has never been evaluated within an adequately powered randomised controlled trial.
METHODS: In total, 199 men, average age 74 years, with localised prostate cancer, 60% managed with primary active surveillance
(AS) or 40% with watchful waiting (WW) following previous interventions, were randomised (2:1) to receive an oral capsule
containing a blend of pomegranate, green tea, broccoli and turmeric, or an identical placebo for 6 months.
RESULTS: The median rise in PSA in the food supplement group (FSG) was 14.7% (95% confidence intervals (CIs) 3.4–36.7%), as
opposed to 78.5% in the placebo group (PG) (95% CI 48.1–115.5%), difference 63.8% (P ¼ 0.0008). In all, 8.2% of men in the FSG and
27.7% in the PG opted to leave surveillance at the end of the intervention (w2 P ¼ 0.014). There were no significant differences
within the predetermined subgroups of age, Gleason grade, treatment category or body mass index. There were no differences in
cholesterol, blood pressure, blood sugar, C-reactive protein or adverse events.
CONCLUSIONS: This study found a significant short-term, favourable effect on the percentage rise in PSA in men managed with AS and
WW following ingestion of this well-tolerated, specific blend of concentrated foods. Its influence on decision-making suggests that this
intervention is clinically meaningful, but further trials will evaluate longer term clinical effects, and other makers of disease progression.

Prostate Cancer and Prostatic Disease (2014) 17, 180–186; doi:10.1038/pcan.2014.6; published online 11 March 2014
Keywords: nutrition; active surveillance; watchful waiting

INTRODUCTION Australian study linked a similar supplement intake with more

Diets deficient in polyphenols and other natural plant-based
phytochemicals found in herbs, spices, fruit, teas, colourful
vegetables and other healthy plant-based foods, have been linked
with higher risks of cancer particularly breast,1,2 pancreas,3 ovary,4
skin,5 prostate,6,7 bowel8 and oesophagus.9
The protective benefits of polyphenols, however, do not to stop
after a diagnosis of cancer. Breast cancer survivors taking higher
levels of fruit and vegetables had a lower recurrence10 and those
with a higher dietary intake of lignans, isoflavones, flavanones
within soy-rich foods or green tea had a lower risk of breast cancer
death.11–13 Individuals with skin cancer who had higher leafy,
green vegetable intake had a lower rate of new cancer formation.5
Men adopting healthy diets after prostate cancer were shown to
have slower PSA progression.14,15
The potential benefits of concentrating foods into a pill has
been the subject of extentive evaluation. Up to now, research has
focused on supplements containing specific, extracted chemicals
believed to be the anti-cancer candidates and although some
studies have shown benefits,16–18 most have not or were actually
linked to an increased risk of cancer. For example, the two vitamin
A and E studies increased the risk of lung cancer19,20 and an
subsequent skin cancers.5 Long-term folate supplementation after
myocardial infarct resulted in a higher cancer risk.21 A supplement
containing vitamin C, copper and manganese did not slow PSA
progression;15 the selenium and vitamin E cancer prevention trial
(SELECT) study showed an increased prostate cancer incidence
following long-term intake of vitamin E and selenium22 as did men
in the Health Professionals Study, who took zinc.23 Despite some
initial encouragement from cohort and small prospective studies,
lycopene, saw palmetto or genistein extracts evaluated within
more scientifically robust analyses did not demonstrate a benefit
for either prostate cancer, benign prostatic hypertrophy or other
malignancies.1,6,18,24–26
As a consequence of these data, scientific attention has been
turning towards the evaluation of concentrated polyphenol-rich
whole food supplements, rather than extracted chemicals, as
convenient ways to boost poor diets or further enhance already
adequate diets. There are some laboratory and phase II studies to
support their further evaluation and hence the rationale for this
study. Men with prostate cancer, managed with active surveillance
(AS) or watchful waiting (WW) for a PSA relapse after radical
treatments, were selected as an ideal cohort to evaluate a lifestyle

1
The Primrose Research Unit, Bedford Hospital, Bedford, UK; 2Department of Oncology, Addenbrooke’s Cambridge University NHS Trust, Cambridge, UK and 3Department of
Postgraduate Medicine, Cranfield University, Cranfield, UK. Correspondence: Professor R Thomas, Bedford and Addenbrooke’s Cambridge University NHS Hospitals, c/o The
Research Unit, Primrose Oncology, Bedford Hospital, Bedford MK42 9DJ, UK.
E-mail: robert.thomas@bedfordhospital.nhs.uk
Received 13 October 2013; revised 14 January 2014; accepted 26 January 2014; published online 11 March 2014

intervention as they have a useful serum marker of their
disease, PSA, and medical interventions are often not indicated
initially.15
Rationale for the ingredients of this interventional food
supplement
Pomegranate, rich in ellagic acid, has been shown in in vitro
studies to inhibit proliferation, markers of migration, induce
apoptosis and cell adhesion in breast and prostate cancer cell
lines.27 In humans, a phase II study reported a prolongation of PSA
doubling following pomegranate juice consumption and markers
of oxidative stress improved.28 A further phase II study, gave men
pomegranate seed extract with similar effects.29 Its influence was
not felt to be via hormonal route as it affected both androgen-
sensitive and -resistant human prostate cancer cells, and one of
the clinical studies showed no change in testosterone levels
following regular intake.25,30–34
Green tea, rich in epigallocatechin gallate, has been shown to
block ornithine decarboxylase, an enzyme which signals cells to
proliferate faster and bypass apoptosis.35–37 It has also been
reported to reduce several growth factors which promote breast
and prostate cancer cell line growth, block de-differentiation and
angiogenesis.38 Men given an extract of tea illustrated a significant
reduction in the levels of several growth factors that promote
cancer, as well as a beneficial effect on PSA.35
Broccoli, rich in isothiocyanate and its metabolite sulphora-
phane, has been shown to inhibit growth and promote apoptosis
in cancer cells.39 In humans, a study found that regular broccoli
intake downregulated cancer genes linked to cancer promotion
and up-regulated genes link to cancer suppression.5,7,39,40
Curcumin, which gives turmeric its yellow colour, has been
shown to slow prostate cancer cell growth, increase apoptosis,
reduce markers of invasion and migration of cells.41–44 It has been
shown to inhibit tyrosine kinase activity of the epidermal growth
factor receptor,45 have cyco-oxidase-I-mediated anti-inflammatory
properties46 and halt the growth of stem cells that give rise to
breast cancer without influencing normal breast cells.46
Finally, all four ingredients also have some anti-oxidant proper-
ties that are thought to protect the DNA against oxidative damage
from ingested or environmental carcinogens,36,37,47 although this
precise mechanism has not been confirmed clinically.48
The rationale and hypothesis for selecting the ingredients of
this supplement were that as they originate different food sources
(fruit, herb, vegetable and leaf), each with their unique profile and
concentration of polyphenols, their separate anti-cancer mechan-
isms, summarised above, could be synergistic49,50 yet at the same
time their variable composition would avoid over-consumption of
one particular phytochemical.25,30,32–34,41–44 The trial committee
determined the concentration of each ingredient based on the
amounts safely used within previous clinical studies.25,30,32–34,41–44
MATERIALS AND METHODS
This was a placebo-controlled, double-blind, randomised trial designed
with the aim of establishing whether supplementing the diet with a
polyphenol-rich, whole food supplement containing a blend of green tea,
pomegranate, broccoli and curcumin (turmeric) influenced the rate of PSA
progression, compared with placebo among men with prostate cancer
either managed with primary AS or WW, following a PSA relapse post-
radical treatments. The 203 participants originated from across the UK and
were all consented and randomised at The Primrose Oncology Unit,
Bedford Hospital, from a total of 208 reviewed for eligibility, between
November 2011 and July 2012. The intervention consisted of a tablet taken
three times a day containing:
 broccoli powder (Brassica oleracea) 100 mg,
 turmeric powder (Curcuma longa) 100 mg,
 pomegranate whole fruit powder (Punica granatum) 100 mg,
& 2014 Macmillan Publishers Limited
Polyphenol rich food supplementation
R Thomas et al
181
 green tea 5:1 extract (Camellia sinensis) 20 mg equivalent to 100 mg of
green tea and
 bulking agent (di-calcium phosphate), anti-caking agents (modified maize-
based starch, maltodextrin and magnesium stearate) removed post trial.
The control group took a placebo containing identical bulking and anti-
caking agents with 10 mg of watercress extract to provide an identical
colour and substance.
All participants were men with an average age of 74 years (range 53–89
years), with histologically confirmed prostate cancer, 121 (60%) were being
managed with AS, and 78 (40%) managed with WW following previous
radical interventions and radical local salvage therapies had been excluded
(primary radiotherapy, 65; surgery followed by radiotherapy, 8; and
brachytherapy, 9). The baseline characteristics are summarised in Table 1.
Following written informed consent, men were randomised by
externally generated, numerically sequenced, opaque, tamper-proof
envelopes to the food supplement group (FSG) or placebo group (PG).
There was a 2:1 randomisation which resulted in 136 in the FSG and 67 in
the PG. Four men withdrew consent after initial randomisation and
proceeded to intervention before the 3-month consultation (two from
each group), had no further relevant PSA and as such could not be
included in an intention to treat analysis (Figure 1). At baseline, 3 months
and 6 months post intervention, PSA, full blood count, urea and
electrolytes, liver function profile, blood glucose, fasting cholesterol,
C-reactive protein, body weight, height and blood pressure were
measured. Adverse and favourable events were recorded in the Case
Report Form according to the National Cancer Institute (NCI) common
toxicity grading scale.
Certification and quality assurance
This trial was approved by the National Ethics Committee, was peer
reviewed by the National Cancer Research Institute (NCRI) Complementary
Therapies Research Committee and formally adopted by the National
Cancer Research Network (NCRN). The Medicines and Health Regulatory
Agency (MHRA) confirmed that no MHRA licence was required as the
intervention was not classified as a medicinal product. The randomisation
process was outsourced and the trial methodology, collection and storage
of data were verified and independently audited by an external agency to
ensure adherence to European Good Clinical Practice. At the end of the
trial, data were externally audited for a second time to ensure that there
were no data inconsistencies or deviation from the trial design, before the
database was sealed and sent for blinded analysis by the statistician at
Cranfield University. The UK manufacturers of the food supplement (Power
Health Products, York, UK) adhered to good manufacturing practice
guidelines and performed in-house analysis for authenticity and purity (the
manufacturing analysis certificates were presented to the reviewers of the
publication). The food supplement and placebo tablets were supplied to
the trials unit in tamper-proof, sealed containers. A batch of the
supplement has been securely stored by the trust secretary and can be
sent to any regulatory body at request.
Table 1. Summary of baseline characteristic in the randomly assigned
groups
Baseline characteristic FSG (134) PG (65)
Age (mean years) 71.8 76.4a
PSA (mean mg l  1) 6.5 6.5
Gleason gradep7 127 (95%) 57 (88%)
Gleason grade 47 7 (5%) 8 (12%)
Gleason grade mean (mg l  1) 6.5 6.2
BMI (mean kg m  2) 28.1 28.3
Cholesterol (mean mmol l  1) 4.87 4.72
BP (mean systolic/diastolic mm Hg) 146/83 150/82
Serum glucose (mean mmol l  1) 5.15 5.30
C-reactive protein (mean mg l  1) 1.51 1.74
Abbreviations: BMI, body mass index; BP, blood pressure; FSG, food
supplement group; PG, placebo group.
The mean age in the PG was older by 4.4 years (t-test P ¼ 0.013) so age was
included in the analysis of the percentage change in PSA as a covariate.
a
Randomisation produced no statistical difference in the group character-
istics except for age.
Prostate Cancer and Prostatic Disease (2014), 180 – 186
 Polyphenol rich food supplementation
R Thomas et al
182

Figure 1. Consort diagram highlighting the flow of patients through the National Cancer Research Network Pomi-T study.

Statistical considerations Median percentage PSA change - all men

The patient representatives on the NCRI Development Committee felt that Difference 63.8% ANCOVA p=0.0008
Percentage rise plus 95% CI

a 2:1 randomisation would be more acceptable to participants. This 120

appeared to be a correct assertion because all but five eligible men invited
to enter this study agreed to be randomised. The PSA value used for the
final analysis was pre-determined to be the 6-month value or the 3-month
value in the men who withdraw at this stage.
Statistical methods
The percentage change in PSA from baseline to final measurement was
evaluated using an analysis of covariance (ANCOVA) which assessed the
effect of the FSG versus PG, as well as the predetermined subgroups of
Gleason grade, body mass index and treatment category. Age was
included as a covariate to adjust for any differences in age between
groups. The interactions between the effect of the FS and each of the other
categorical effects were also included in the analysis. To satisfy the
assumptions of this analysis, the change in PSA had to be transformed to
logarithms, P-values given in this section followed by ANCOVA are from
the F-value in this ANCOVA. All median values were back transformed from
this ANCOVA and therefore allow for the effect of any differences in other
subgroups or age.
The analysis of the number of men at the end of the trial with the same
or lower PSA was analysed using a w2 test with 1 degree of freedom. The
differences in toxicity measures between the two groups were tested using
an appropriate t-test (that is, with equal or unequal variances), and
transforming the measure if required to satisfy the assumptions of a t-test.
RESULTS
Primary end point
In the FSG, the mean PSA rose from 6.50 to 6.81 ug l  6 from
baseline to the end of the intervention, a median PSA percentage
rise of 14.7% (95% confidence interval (CI)  3.4% to 36.7%). In
the PG, the mean PSA increased from 6.50 to 10.98 ug l  1, a
median percentage rise of 78.5% (95% CI 48.1–115.5%). The
median percentage PSA increased at a significantly slower rate in
the FSG group compared with the PG (difference 63.8% ANCOVA,
P ¼ 0.0008).
Secondary end point
The number of men with a PSA lower or the same value at trial
completion was 61 (46%) in the FSG as opposed to 9 (14%) in the
100
All men n=199
80
60
40
20
0
-20
FS group Placebo group
Key: FS = Food supplement, ANCOVA = Analysis of covariance,
= 95% confidence intervals
Figure 2. Median percentage rise in PSA between men taking the
food supplement versus placebo.
PG. This difference was statistically significant (w2 value with 1
degree of freedom ¼ 19.58, P ¼ 0.000010).
Decisions to remain on AS or WW
Twenty-five men opted to leave AS or WW at 3 months (11 in the
FSG and 14 PG), and 4 after 6 months, 11 of 134 in the FSG (8.2%)
and 18 of 65 (27.7%) in the PG (this difference of 19.5% was
significant w2 value with 1 degree of freedom, P ¼ 0.014). The
reasons for opting out were multifactorial and at the discretion of
the physician and patient who were both blind to the intervention
arm but all had a rising PSA.
Predetermined subgroup analysis
A separate analysis of the cohort of men managed with AS
(n ¼ 121) revealed that in the FSG the mean PSA dropped by
0.14% (95% CI  7.57 to 7.95), whereas in the PG it rose by 46.98%
(95% CI 28.51–68.31); difference 47.12% (ANVOCA, P ¼ 0.001), see
Figure 2. A separate analysis of the cohort of men managed with
WW (n ¼ 78) revealed that in the FSG the mean PSA rose by 8.78%
(95% CI  6.32 to 26.62), whereas in the PG it rose by 80.34%

Prostate Cancer and Prostatic Disease (2014), 180 – 186 & 2014 Macmillan Publishers Limited

Median percentage change in PSA (AS)
Difference 47.12% ANCOVA p=0.001
100
80
Men on active surveillance n=121
60
40
20
0
-20
FS group Placebo group
Key: FS = Food supplement, ANCOVA = Analysis of covariance,
= 95% confidence intervals
Figure 3. Subgroup analysis: median percentage change in PSA for
the 121 men managed with active surveillance (AS).
Median percentage change in PSA (WW)
Difference 71.56% ANCOVA p=0.001
110
90
Men on watchful waiting n=78
70
50
30
10
-10
FS group Placebo group
Key: FS = Food supplement, ANCOVA = Analysis of covariance,
= 95% confidence intervals
Figure 4. Subgroup analysis: median percentage change in PSA for
the 78 men managed with watchful watching (WW; PSA relapse
following previous radiotherapy).
(95% CI 50.54–116.55); difference 71.56% (ANVOCA, P ¼ 0.001), see
Figure 3. This difference between the median percentage change
in PSA in either the AS or WW cohorts was not statistically
significant (P ¼ 0.805 ANCOVA) (Figure 4).
There was no statistical difference between the median
percentage PSA rise between men in the FSG or the PG as to
whether they were overweight (X25 kg m  2) or not (P ¼ 0.564
ANCOVA). Although men with higher Gleason grade for the entire
cohort tended to progress at a faster rate, there was no difference
between the median percentage change in PSA between men in
the FSG or the PG whichever Gleason grade category they were
(P ¼ 0.089 ANCOVA). There was a significant effect of age on the
median percentage change in PSA for the entire cohort with older
men tending to progress at a slower rate (P ¼ 0.0272 ANCOVA;
slope of log(MPC-PSA) ¼ 0.0098, P ¼ 0.0272). As there were more
older men randomised to the PG, age was included as a covariate
in the analysis so that the difference between the median
percentage change in PSA for FSG and PG was adjusted to the
mean age of the whole cohort, so that any effect on PSA between
FSG and PG would have been removed.
& 2014 Macmillan Publishers Limited
Polyphenol rich food supplementation
R Thomas et al
183
Other measures
There were no significant differences at the beginning or the end
of the study between groups for cholesterol, blood pressure,
serum glucose or C-reactive protein. Compliance, measured by
counting remaining tablets in returned pots, was excellent and
similar at 98.4% in the PG and 96.5% in the FSG. Sex hormones
were not a predetermined analysis, as the ingredients were
specifically chosen not to have phytoestrogenic properties but
were measured in 64 men who had been taking the supplement
for at least 3 months or more. Three of these men (5.5%) had
testosterone levels below our laboratory normal range. The
average testosterone (13.4 nmol l  1) was within the normal range
as were the other sex hormones: follicle stimulating hormone,
9.2 iu l  1; luteinizing hormone, 7.4 iu l  1; sex hormone binding
globulin, 41.4 nmol l  1; and free androgen index, 24.3%. Magnetic
resonance images (MRI) of the prostate were also not included at
specific time points in the trial protocol but a total of 74 of the 121
men on AS, were taken as part of their routine management
clinical protocol. All these scans, in addition to their original report,
were scrutinised within our multidisciplinary team meetings.
Twelve (16%) men had radiologically progressive disease and in
these men the average PSA rose from 7.65 to 8.67 ug l  1. Eight
(11%) had radiological regression and in these the average PSA
dropped from 7.2 to 4.1 ug l  1. No man in the FSG had
radiological progression with a stable PSA but one man (1.3%)
not taking the FS had radiological progression with a falling PSA.
Although these figures were not subjected to statistical analysis as
MRI was not a predetermined end point but they do give some
reassurance that PSA change is linked to underlying disease
status.
Adverse events
There were 34 (24%) men who recorded adverse events (any NCI
score) in the FSG group and 23 (34%) in the PG. These differences
were not statistically significant (w2 value with 1 degree of
freedom ¼ 2.2, P ¼ 0.14). There were no grade X3 toxicities, but
one man in the FSG group had grade 2 diarrhoea. Gastrointestinal
events, considered separately, occurred in 21 (15.5%) in the FSG
group as opposed to 5 (7.5%) in the PG, but this difference
was not statistically significant (w2 value with 1 degree of
freedom ¼ 2.24, P ¼ 0.11). No man in either group reported central
nervous system symptoms, such as agitation, insomnia or tremors,
none of the 30 men on warfarin reported any unexpected change
in the international normalized ratio, nor did the 43 men taking
ramipril report an unexpected change in their blood pressure.
Positive events
There were 16 (12%) men who recorded positive events (mainly
improved bowel and urinary function) in the FSG, and 3 (4.6%) in
the PG. These differences were not statistically significant (w2 value
with 1 degree of freedom ¼ 2.67, P ¼ 0.10) (Table 2).
DISCUSSION
This study demonstrated a significant effect on the rate of PSA
progression among men with prostate cancer, randomised to take
this nutritional supplement compared with placebo. The differ-
ence was large, the patient characteristics were well-balanced and
the trial had sufficient numbers to ensure adequate statistical
power. However, there are some caveats with the trial design to
consider and discuss.
The first caveat was the relatively short, 6 months, duration of
the intervention which although enough time to detect an early
difference, as men are often managed with surveillance for many
years, a longer design would exclude the possibility that this effect
was short lived. On the other hand, a longer study would have
been beyond the funding constraints of this non-commercial
Prostate Cancer and Prostatic Disease (2014), 180 – 186
 Polyphenol rich food supplementation
R Thomas et al
184
Another issue with the trial design was that the cohort included
Table 2. Summary of the adverse and positive events
men both on primary AS and those experiencing a PSA relapse
FSG PG Difference % after radical treatments, deemed not eligible for salvage local
(% of 134) (% of 65) (significance) therapies. The design committee believe that this was not a
negative caveat as there were appropriate reasons for this choice.
Adverse gastro-intestinal events First, in our experience, this is precisely the group of men who are
Loose bowels 6 (4.5%) 0 (0%) 4.5 (ns) most interested in lifestyle strategies15 and a trial cohort should,
Diarrhoea (grade 1) 2 (1.5%) 0 (0%) 1.5 (ns) where possible, reflect the wider population considering the study
Diarrhoea (grade 2) 2 (1.5%) 1 (1.5%) 0 (ns) intervention. Both groups of men are similar as they have viable
Diarrhoea (grade 0 (0%) 0 (0%) 0 (ns)
prostate cancer, and are not receiving other systemic therapies.
X3)
Constipation 2 (1.5%) 0 (0%) 1.5 (ns) Both groups have serum PSA being monitored as part of their
Flatulence 5 (3.6%) 0 (0%) 3.6 (ns) routine clinical management. Previous nutritional studies have
Rectal bleeding 0 (0%) 1 (1.5%) 1.5 (ns) also included both groups.15,17,26 To support this rationale, in
Nausea 0 (0%) 1 (1.5%) 1.5 (ns) this study, there was no statistical difference in the effect on
Bloating 4 (3%) 2 (3%) 0 (ns) PSA between both cohorts, suggesting a similar effect of
All GI adverse 21 (15.5%) 5 (7.5%) 8 (ns) polyphenol-rich foods, regardless of whether patients had
events received radiotherapy or not.
Despite these caveats, although PSA has its short falls, men
Other adverse events
managed with AS or WW are greatly concerned with their serum
Gout exacerbation 2 (1.5%) 1 (1.5%) 0 (ns)
Worsening urinary 2 (1.5%) 2 (3%) 1.5 (ns) levels and a rise is often a trigger for a change in their
flow management.31 This was borne out in the trial, as 29 men opted
Worsening renal 2 (1.5%) 0 (0%) 1.5 (ns) for to leave AS or WW during or at the end of the study when their
function PSA rose, twice the percentage in the PG than the FSG. This
Weight loss 0 (0%) 2 (3%) 3 (ns) difference, as well as achieving strong statistical significance,
Non-specific 2 (1.5%) 4 (6%) 4.5 (ns) suggests a clinically meaningful effect. Notwithstanding avoidance
‘feeling unwell’ of the inevitable toxicities of androgen deprivation therapy, with
Miscellaneous 5 (3.6%) 9 (13.4%) 9.8 (ns) over a 10-fold difference in price between this food supplement
unrelated
and androgen deprivation therapy, a future trial should include a
All adverse events 34 (24%) 23 (34%) 10 (ns)
cost-effectiveness analysis in order to determine the magnitude of
Positive events potential savings for healthcare providers.15
Improved erectile 1 (0.75%) 0 (0%) 0.75 (ns) As well as price, other attractive features of this supplement
function were its tolerance and safety. There were no overall statistically
Improve urinary 4 (3%) 1 (1.5%) 1.5 (ns) significant difference in symptoms compared with placebo, although
flow more men in the supplement group experienced non-significant
Reduced prostatic 1 (0.75%) 0 (0%) 0.75 (ns) bloating or diarrhoea, but almost as many reported beneficial
discomfort effects including improved digestion and urinary symptoms. The
All positive 6 (4.5%) 1 (1.5%) 3 (ns)
trend towards better urinary symptoms is worthy of future scrutiny
prostatic symptoms
Improved bowel 8 (6%) 0 (0%) 6 (ns) in the next study, especially as previous studies have linked
function turmeric with improved symptoms of prostatitis presumably via its
Improved well 2 (1.5%) 2 (3%) 0 (ns) anti-inflammatory properties.41,46 One of the ingredients of the
being supplement, pomegranate, has been cited to be a weak inhibitor
All positive event 16 (12%) 3 (4.5%) 7.5 (ns) of cytochrome P450 but, reassuringly, the quantities used in this
study resulted in no unexpected changes in blood pressure or
Abbreviation: FSG, food supplement group; GI, gastrointestinal; ns, non-
significant difference; PG, placebo group.
international normalized ratio levels, which could have related to
interference of the metabolism of ramipril or warfarin.
Although, these ingredients have previously demonstrated
study. Furthermore, a trial designed to take a potential placebo for fundamental effects on cancer progression in laboratory studies
longer would have been a less attractive option for men, so the including markers of angiogenesis, metastasis, adhesion and
rapid recruitment achieved in this study would have been less apoptotis,25,30,32–34,41–44 our next trial will explore these mech-
likely. anisms in more detail building upon the translational designs
The second caveat was that the main end point relied on PSA instigated by previous researchers.14 Furthermore, although these
measurement, without other formal indicators of disease progres- ingredients were specifically selected not to have phytoestrogenic
sion including MRI or prostate biopsies which are now becoming properties and the average sex hormone levels were reassuringly
routine in the management of men on AS. Although biopsy would normal in the subgroup who had them measured, the next trial
have enhanced scientific competence, not all men currently will also explore the physiological effects on participants in more
consent a repeat intervention and hence this inclusion would have detail, including serum testosterone and markers of oxidative
reduced the rate of recruitment, increased complexity. The cost of stress.2,34
including MRI before and after the trial period would also have In conclusion, this statistically valid double-blind randomised
been prohibitive for the level of sponsorship available but controlled trial has demonstrated a significant short-term effect on
fortunately, images were taken as part of their routine manage- PSA and is food for thought for men living with prostate cancer,
ment in over half of the men in the AS group. It is of some 50–70% of whom are reported to have taken ‘over-the-counter’
reassurance, that PSA reflected underlying disease status as the supplements.47 The favourable effect on PSA progression was
percentage change in PSA was 10-fold lower in the men with significant both in men on primary AS and those experiencing a
disease shrinking on MRI compared with disease progression and PSA relapse after radiotherapy. This low cost food supplement was
no man taking the food supplement had disease progression with well-tolerated and also influenced clinically relevant decisions, as
a stable PSA. All men choosing to continue to supplement after to whether to switch to interventions with more toxicity. Although
the trial period will now receive annual MRI’s this data will subject these results do not prove a long-term effect, they have provided
to a future publication. significant encouragement to design a major study with more

Prostate Cancer and Prostatic Disease (2014), 180 – 186 & 2014 Macmillan Publishers Limited

comprehensive physiological, radiological and translational
laboratory end points, in order to get a deeper understanding
of the evidence of benefit and role for these complex and readily
available foods.
CONFLICT OF INTEREST
The authors declare no conflict of interest.
ACKNOWLEDGEMENTS
The authors thank the charities Prostate Action and The Primrose Oncology Fund for
sponsoring this study and the enthusiasm and co-operation of the participants. The
full trial protocol is available on cancernet.co.uk/pomi-t.htm. Peer reviewed
sponsorship was received from the charity ‘Prostate Action’, and the study was also
supported by ‘The Primrose Oncology Fund’. No sponsorship was received from the
UK manufacturer of the investigational supplement used in this study.
REFERENCES
1 Hu F, Wang YB, Liang J, Yi B, Zhang W, Liang J et al. Carotenoids and breast cancer
risk: a meta-analysis and meta-regression. Breast Cancer Res Treat 2012; 131:
239–253.
2 Wu AH, Yu MC. Tea, hormone-related cancers and endogenous hormone levels.
Mol Nutr Food Res 2006; 50: 160–169.
3 Banim PJ, Luben R, McTaggart A, Welch A, Wareham N, Khaw KT et al.
Dietary antioxidants and the aetiology of pancreatic cancer. Gut 2013; 62:
1489–1496.
4 Tung K, Wilkens LR, Wu AH, McDuffie K, Hankin JH, Nomura AM et al. Association
of dietary vitamin A, carotenoids and other antioxidants with the risk of ovarian
cancer. Cancer Epidemiol Biomarkers Prev 2005; 14: 669–676.
5 Heinen MM, Hughes MC, Ibiebele TI, Marks GC. Intake of antioxidant nutrients and
the risk of skin cancer. EJC 2007; 43: 2707–2716.
6 Giovannucci E, Rimm EB, Liu Y, Meir J, Stampfer MJ, Walter C et al. A prospective
study of tomato products, lycopene, and prostate cancer risk. J Natl Cancer Inst
2002; 94: 391–398.
7 Joseph MA, Moysich KB, Freudenheim JL, Shields PG, Bowman ED, Zhang Y et al.
Cruciferous vegetables, genetic polymorphisms and prostate cancer risk. Nutr
Cancer 2004; 50: 206–213.
8 Steward A, Gescher AJ. Curcumin in cancer management: recent results of
analogue design and clinical studies and desirable future research. Mol Nutr Food
Res 2008; 52: 1005–2009.
9 Sun CL, Yuan JM, Koh WP, Lee HP, Yu MC. Green tea and black tea consumption
and colorectal cancer risk: The Singapore Chinese Health Study. Carcinogenesis
2007; 28: 2143–2148.
10 Pierce JP, Natarajan L, Caan BJ, Parker BA, Greenberg ER, Flatt SW et al. Influence
of a diet very high in vegetables, fruit, and fiber and low in fat on prognosis
following treatment for breast cancer: The Women’s Healthy Eating and Living
(WHEL) randomized trial. JAMA 2007; 298: 289–298.
11 Buck K, Vrieling A, Zaineddin AK, Becker S, Hüsing A, Kaaks R et al. Serum
enterolactone and prognosis of postmenopausal breast cancer. J Clin Oncol 2011;
29: 3730–3738.
12 Boyapati SM, Shu XO, Ruan ZX, Dai Q, Cai Q, Gao YT et al. Soy food intake and
breast cancer survival: a follow up of the Shanghai Breast Cancer Study. Breast
Cancer Res Treat 2005; 92: 11–17.
13 Ogunleye AA, Xue F, Michels KB. Green tea and breast cancer risk of recurrence: a
meta-analysis. Breast Cancer Res Treat 2010; 119: 477.
14 Ornish D, Magbanua MJM, Weidner G, Weinberg V, Kemp C, Green C et al.
Intensive lifestyle changes may affect the progression of prostate cancer. J Urol
2005; 174: 1065–1070.
15 Thomas R, Oakes R, Gordon J, Russell S, Blades M, Williams M. A randomised,
double-blind phase II study of lifestyle counselling and salicylate compounds in
patients with progressive prostate cancer. Nutr Food Sci 2009; 39: 295–305.
16 Meyer F, Galan P, Douville P, Bairati I, Kegle P, Bertrais S et al. Antioxidant vitamin
and mineral supplementation and prostate cancer prevention in the SU.VI.MAX
trial. Int J Cancer 2005; 116: 182–186.
17 Schröder FH, Roobol MJ, Boevé ER, de Mutsert R, Zuijdgeest-van Leeuwen SD,
Kersten I et al. Randomized, double-blind, placebo-controlled crossover study in
men with prostate cancer and rising PSA: effectiveness of a dietary supplement.
Eur Urol 2005; 48: 922–930.
18 Brasky TM, Kristal AR, Navarro SL, Lampe JW, Peters U, Patterson RE et al. Specialty
supplements and prostate cancer risk in the VITamins And Lifestyle (VITAL)
cohort. Nutr Cancer 2011; 63: 573–582.
& 2014 Macmillan Publishers Limited
Polyphenol rich food supplementation
R Thomas et al
185
19 Omenn GS, Goodman GE, Thornquist MD, Balmes J, Cullen MR, Glass A et al. Risk
factors for lung cancer and for intervention effects in CARET, the beta-carotene in
retinol efficacy trial. J Natl Cancer Inst 1996; 88: 1550–1559.
20 Heinonen O, Albanes D, Virtamo J, Taylor PR, Huttunen JK, Hartman AM et al.
Prostate cancer and supplementation with alpha-tocopherol and beta-
carotene: incidence and mortality in a controlled trial. J Natl Cancer Inst 1998; 90:
440–446.
21 Figueiredo JC, Grau MV, Haile RW, Sandler RS, Summers RS, Bresalier RS et al. Folic
acid and risk of prostate cancer: results from a randomized clinical trial. J Natl
Cancer Inst 2009; 101: 432–435.
22 Klein EA, Thompson IM, Tangen CM, Crowley JJ, Lucia MS, Goodman PJ et al.
Vitamin E and the risk of prostate cancer: the selenium and vitamin E cancer
prevention Trial (SELECT). JAMA 2011; 306: 1549–1556.
23 Leitzmann MF, Stampfer MJ, Wu K, Colditz GA, Willet WC, Giovannucci EL. Zinc
supplementation and the risks of prostate cancer. J Natl Cancer Inst 2003; 95:
1004–1007.
24 Bent S, Kane C, Shinohara K, Neuhaus J, Hudes ES, Goldberg H et al. Saw Palmetto
for benign prostatic hyperplasia. N Engl J Med 2006; 354: 557–566.
25 Barber NJ, Zhang X, Zhu G, Pramanik R, Barber JA, Martin FL et al.
Lycopene inhibits DNA synthesis in primary prostate epithelial cells in vitro
and its administration is associated with a reduced prostate-specific
antigen velocity in a phase II clinical study. Prostate Cancer Prostatic Dis 2006; 9:
407–413.
26 Clark PE, Hall MC, Borden LS, Clark PE, Miller AA, Hu JJ et al. Phase I-II prospective
dose-escalating trial of lycopene in patients with biochemical relapse of prostate
cancer after definitive local therapy. Urology 2006; 67: 1257–1261.
27 Wang L, Alcon A, Yuan H, Ho J, Li QJ, Martins-Green M. Cellular and molecular
mechanisms of pomegranate juice-induced anti-metastatic effect on prostate
cancer cells. Integr Biol (Camb) 2011; 3: 742–754.
28 Pantuck AJ, Leppert JT, Zomorodian N, Seeram N, Seiler D, Liker H et al. Phase II
study of pomegranate juice for men with rising PSA following surgery or radiation
for prostate cancer. J Urol 2005; 173: 225–226.
29 Paller CJ, Ye X, Wozniak PJ, Gillespie BK, Sieber PR, Greengold HR et al.
A randomised phase II study of pomegranate extract for men with rising PSA
following initial therapy for localized prostate cancer. Prostate Cancer Prostatic Dis
2013; 16: 50–55.
30 Rettig MB, Heber D, An J, Seeram NP, Rao JY, Liu H et al. Pomegranate extract
inhibits androgen-independent prostate cancer growth through a nuclear factor-
kappaB-dependent mechanism. Mol Cancer Ther 2008; 7: 2662–2671.
31 Lansky EP, Harrison G, Froom P, Jiang WG. Pomegranate (Punica granatum) pure
chemicals show possible synergistic inhibition of human PC-3 prostate cancer cell
invasion across Matrigel. Invest New Drugs 2005; 23: 121–122.
32 Malik A, Mukhtar H. Prostate cancer prevention through pomegranate fruit. Cell
Cycle 2006; 5: 371–373.
33 Khan N, Hadi N, Afaq F, Syed DN, Kweon MH, Mukhtar H. Pomegranate fruit
extract inhibits prosurvival pathways in human A549 lung carcinoma cells and
tumor growth in athymic nude mice. Carcinogenesis 2007; 28: 163–173.
34 Choi C, Kim JK, Choi SH, Jung HS, Kim HJ, Cho SY et al. Identification of steroid
hormones in pomegranate (Punica granatum) using HPLC and GC–mass spec-
trometry. Food Chemistry 2006; 96: 562–571.
35 Liao S, Kao YH, Hiipalla RA. Green tea: biochemical and biological basis for health
benefits. Vitam Horm 2001; 62: 1–94.
36 Porrini M, Riso P. Factors influencing the bioavailability of antioxidants in foods: a
critical appraisal. Nutr Metab Cardiovas 2008; 80: 353–361.
37 Liao J, Yang GY, Park ES, Meng X, Sun Y, Jia D et al. Inhibition of lung carcino-
genesis and effects on angiogenesis and apoptosis in mice given green tea. Nutr
Cancer 2004; 48: 44–53.
38 Yang CS, Maliakal P, Meng X. Inhibition of carcinogenesis by tea. Annu Rev
Pharmacol Toxicol 2002; 42: 25–54.
39 Traka M, Gasper AV, Melchini A, Bacon JR, Needs PW, Frost V et al. The dietary
isothiocyanate sulforaphane modulates gene expression and alternative gene
splicing in a PTEN null preclinical murine model of prostate cancer. Mol Cancer
2010; 9: doi:10.1186/1476-4598-9-189.
40 Gasper AV, Al-Janobi A, Smith JA, Bacon RJ, Fortun P, Atherton C et al.
Glutathione S-transferase M1 polymorphism and metabolism of sulforaphane
from standard and high-glucosinolate broccoli. Am J Clin Nutr 2005; 82:
1283–1291.
41 Shah BH, Nawaz Z, Pertani SA, Roomi A, Mahmood H, Saeed SA et al. Inhibitory
effect of curcumin, a food spice from turmeric, on platelet-activating factor-
and arachidonic acidmediated platelet aggregation through inhibition of
thromboxane formation and Ca2 þ signaling. Biochem Pharmacol 1999; 58:
1167–1172.
42 Zhang HN, Yu CX, Chen WW, Jiang AL, Kong F, Deng DJ et al. Curcumin down
regulates gene NKX3.1 in prostate cancer cell lines (LNcaP). Acta Pharmacol Sin
2007; 28: 423–430.
Prostate Cancer and Prostatic Disease (2014), 180 – 186
 Polyphenol rich food supplementation
R Thomas et al
186
43 Iqbal M, Sharma SD, Okazaki Y, Fujisawa M, Okada S. Dietary supplementation of
curcurmin enhances antioxidant phase II metabolosing enzymes in mice. Phar-
macol Toxicol 2003; 92: 33–38.
44 Somasundaram S, Edmund NA, Moore DT, Small GW, Shhi YY, Orlowski RZ.
Curcumin inhibits chemotherapy-induced apoptosis in models of cancer. Cancer
Res 2002; 62: 3868–3875.
45 Dorai T, Gehani N, Buttyan R, Katz AE. Therapeutic potential of curcumin in human
prostate cancer. Curcumin inhibits tyrosine kinase activity of the Epidermal
growth factor receptor. Mol Urol 2000; 4: 1–6.
46 Handler N, Jaeger W, Puschacher H, Leisser K, Erker T. Synthesis of noval curcumin
analogues and their evaluation as selective cyclooxygenase-1 inhibitors. Chem
Pharm Bull 2007; 55: 64–71.
47 Bauer CM, Johnson EK, Beebe-Dimmer JL, Cooney KA. Prevalence and correlates
of vitamin and supplement usage amongst men with prostate cancer. Integr
Cancer Ther 2012; 11: 83–89.
Prostate Cancer and Prostatic Disease (2014), 180 – 186
48 Freedland SJ, Carducci M, Kroeger N, Partin A, Rao JY, Jin Y et al. A double blind
randomised neoadjuvant study of the tissue effects of Pomx pills in men with
prostate cancer before radical prostatectomy. Cancer Prev Res (Phila) 2013; 6:
1120–1127.
49 Parada J, Aguilera JM. Food combinations and microstructure affects the bioa-
vailability of several nutrients. J Food Sci 2007; 72: 21–23.
50 Kakarla M, Brenner DE, Khorkya H, Cheng C, Tazi K, Ginestier C et al. Targeting
breast stem cells with the preventive compounds curcumin and piperine. Breast
Cancer Res Treat 2010; 122: 777–785.
This work is licensed under a Creative Commons Attribution 3.0
Unported License. To view a copy of this license, visit http://
creativecommons.org/licenses/by/3.0/
& 2014 Macmillan Publishers Limited
 PSA Testing for Prostate Cancer Screening
Invited Commentary
Prostate-Specific Antigen Testing for Prostate Cancer Screening—
Is the Message Getting Through?
Freddie C. Hamdy, MD
Since the discovery of prostate-specific antigen (PSA) more than
40 years ago and its subsequent wide adoption for the detec-
tion of prostate cancer, screening for the disease has been
plagued with continuing controversies that remain largely
unresolved. There is a pro-
Related article
foundly engrained public per-
ception embraced by many
professionals that cancer detection as early as possible saves
lives and therefore should be promoted at any cost, be it finan-
cial or in kind, such as the inevitable adverse effects caused by
radical treatments. Although this may be true for some can-
cers, more than 3 decades of PSA testing among asymptomatic
men across the globe, particularly in the Western world, has
proved otherwise. Leapman et al1 illustrate this point in their
article on the changes in the rates of PSA testing associated
with recent updates in the US Preventive Services Task Force
(USPSTF) recommendations on prostate cancer screening.
Currently, the biggest challenges to making recommen-
dations on PSA testing as a public health screening policy for
prostate cancer are 3-fold: (1) the avoidance of overdetection
of disease that would not otherwise manifest itself as clini-
cally important during an individual’s lifetime, (2) the subse-
quent inevitable overtreatment with potential unwanted ad-
verse effects on quality of life, and (3) our inability to determine
accurately the lethal potential of prostate cancer at diagno-
sis, resulting in undertreatment. It has been clearly demon-
strated that screening by PSA testing saves lives, but at the un-
acceptable cost of overdetection and overtreatment, which has
prevented many policymakers and organizations from recom-
mending PSA testing as a public health policy for asymptom-
atic men in the general community.
Guidelines usually rely on evidence to make recommen-
dations about any health intervention, but for prostate can-
cer screening, evidence has been scarce over decades. Two
large randomized clinical trials of screening by PSA testing in
the US and Europe have been major sources of high-level data
on interval screening by PSA, complemented by the more re-
cent Cluster Randomized Trial of PSA Testing for prostate can-
cer (CAP)2 in the UK investigating the benefits of screening with
a single PSA test. The Prostate, Lung, Colon and Ovary (PLCO)
study in the US has ruled itself out as providing evidence be-
cause of heavy contamination and other unresolvable issues.
The European Randomized Trial of Screening for Prostate
Cancer (ERSPC) in Europe showed a benefit in favor of screen-
ing but at a high cost of overdetection,3 and the UK CAP study
showed no survival benefit with a single PSA test but in-
creased detection of low-risk, low-volume prostate cancer.2 At
the same time, trials of the effectiveness in screening for dis-
ease, such as the ProtecT (Prostate Testing for Cancer and Treat-
ment) study in the UK4 and the Prostate Cancer Intervention
vs Observation Trial (PIVOT)5 in the US, showed no survival
jamaoncology.com
© 2021 American Medical Association. All rights reserved.
Downloaded From: https://jamanetwork.com/ by a Mexico | Access Provided by JAMA User on 11/21/2021
Invited Commentary
benefit in favor of early radical treatment, although ProtecT
showed that early intervention reduced the rate of disease pro-
gression and metastasis in favor of radical interventions com-
pared with active monitoring4 and recent reports from PIVOT
suggest a benefit of early treatment for intermediate-risk
disease.5 The most mature treatment effectiveness random-
ized clinical trial is the Scandinavian Prostate Cancer Group-4
(SPCG-4) trial, which showed consistent benefits of radical
treatments compared with watchful waiting in disease-
specific, overall survival and progression, but approximately
half of the SPCG-4 cohort already had nonorgan-confined dis-
ease, and most were not detected via PSA test, which largely
explains the benefit patients received from early treatment.6
Based on emerging evidence from these few global ran-
domized clinical trials, the USPSTF made recommendations
in 2012 that PSA testing should not be offered to any man at
any age with the purpose of prostate cancer screening (grade
D recommendation). This recommendation has been associ-
ated with a notable reduction in indiscriminate PSA testing in
the US but was deplored by many clinicians and institutions
as in turn being associated with an increase in the number of
men presenting with metastatic and advanced disease. As
the trials matured, these recommendations were amended in
2018, suggesting that men between 55 and 69 years of age can
be offered a PSA test but that men aged 70 years or older should
not be tested.
Leapman et al1 report the variations in PSA testing in a US
cohort of men aged 40 to 89 years during 2 distinct periods
associated with changes in the USPSTF recommendations in
2012 and 2018. The analysis was triggered because of the per-
ceived increase in the presentation of late-stage disease,
claimed by some as being caused by the reduction in PSA test-
ing. The findings of the analysis are that there was a progres-
sive and significant increase in PSA testing among men of all
ages after the revision of the USPSTF grade C recommenda-
tion in 2018 for men aged 55 to 59 years, with particular con-
cerns about the increase in testing for men aged 40 to 55 years
and 70 years or older falling outside the age category in
the USPSTF recommendations and causing potential harm.
The findings are valuable, and the authors should be com-
mended for undertaking this informative analysis, which re-
flects a flawed attitude toward PSA testing in the community.
Clinicians often advocate that the answer to the overde-
tection of prostate cancer is the adoption of safe active sur-
veillance protocols. This is the incorrect approach. Whenever
an asymptomatic man receives a diagnosis of prostate can-
cer, irrespective of its indolent nature, he is given a “new” iden-
tity of becoming a patient with cancer. A “cancer passport” is
delivered, and its consequences may be far reaching. Men
should therefore be counseled properly regarding the ben-
efits and harms of screening before a decision is taken to mea-
(Reprinted) JAMA Oncology Published online November 11, 2021 E1
 Invited Commentary PSA Testing for Prostate Cancer Screening

sure their PSA level. Quality of life in the SPCG-4 randomized
clinical trial of watchful waiting vs radical prostatectomy
among a cohort of patients with prostate cancer was ana-
lyzed and compared with a matched cohort of men without
cancer. It was revealing that the harm caused to quality of life
was likely due to the diagnosis of cancer rather than the treat-
ment administered.7 Similar findings of anxiety were re-
ported by the ERSPC, affecting men’s mental health after PSA
testing. In the ProtecT study, patient-reported outcomes
showed no evidence of a difference in anxiety and depres-
sion levels between men who received active monitoring and
men who underwent radical treatments.4
The prostate cancer screening landscape continues to
evolve with the emergence of prebiopsy imaging using mul-
tiparametric magnetic resonance imaging of the prostate,
which has transformed the diagnostic pathway in many coun-
tries but with slower uptake in the US. It is now evident that
the long-term practice of a PSA test followed by systematic
biopsies of the prostate is antiquated. In the CAP trial, for in-
stance, a single PSA test missed more than one-third of lethal
prostate cancers in men who underwent screening. 2 Al-
though genomic testing and the use of other emerging risk
calculators, such as polygenic hazard scores, have limited evi-
dence to be adopted in a risk-stratification approach to screen-
ing, it is likely that with imaging and targeted biopsies, the field
will progress further, minimizing the risk of overdetection
and overtreatment and focusing on identifying early disease
that warrants tailored treatments to improve outcomes and
save lives. But when will the message get through to the pub-
lic, clinicians, and health care professionals that inappropri-
ate PSA testing outside evidence-based recommendations
should cease?

ARTICLE INFORMATION
Author Affiliation: Nuffield Department
of Surgical Sciences, University of Oxford,
Oxford, United Kingdom.
Corresponding Author: Freddie C. Hamdy, MD,
Nuffield Department of Surgical Sciences,
University of Oxford, Old Road Campus
Research Bldg, Headington, Oxford OX3 7DQ,
United Kingdom (freddie.hamdy@nds.ox.ac.uk).
Published Online: November 11, 2021.
doi:10.1001/jamaoncol.2021.5129
Conflict of Interest Disclosures: Dr Hamdy
reported receiving grants from the National
Institute for Health Research HTA Programme
and Cancer Research UK during the conduct of
the study; and personal fees from Intuitive
Surgical outside the submitted work; serving as
editor-in-chief, British Journal of Urology
International; and being chief investigator of the
National Institute for Health Research HTA ProtecT
trial (96/20/99) and coinvestigator of the CAP
Trial (Cancer Research UK C11043/A4286, C18281/
A8145, C18281/A11326, C18281/A15064).
REFERENCES
1. Leapman MS, Wang R, Park H, et al. Changes in
prostate-specific antigen testing relative to the
revised US Preventive Services Task Force
recommendation on prostate cancer screening.
JAMA Oncol. Published online November 11, 2021.
doi:10.1001/jamaoncol.2021.5143
2. Martin RM, Donovan JL, Turner EL, et al;
CAP Trial Group. Effect of a low-intensity PSA-based
screening intervention on prostate cancer
mortality: the CAP randomized clinical trial. JAMA.
2018;319(9):883-895. doi:10.1001/jama.2018.0154
3. Hugosson J, Roobol MJ, Månsson M, et al;
ERSPC investigators. A 16-yr follow-up of the
European Randomized Study of Screening for
Prostate Cancer. Eur Urol. 2019;76(1):43-51.
doi:10.1016/j.eururo.2019.02.009
4. Hamdy FC, Donovan JL, Lane JA, et al; ProtecT
Study Group. 10-Year outcomes after monitoring,
surgery, or radiotherapy for localized prostate
cancer. N Engl J Med. 2016;375(15):1415-1424.
doi:10.1056/NEJMoa1606220
5. Wilt TJ, Vo TN, Langsetmo L, et al. Radical
prostatectomy or observation for clinically localized
prostate cancer: extended follow-up of the Prostate
Cancer Intervention Versus Observation Trial
(PIVOT). Eur Urol. 2020;77(6):713-724. doi:10.1016/
j.eururo.2020.02.009
6. Bill-Axelson A, Holmberg L, Garmo H, et al.
Radical prostatectomy or watchful waiting in
prostate cancer—29-year follow-up. N Engl J Med.
2018;379(24):2319-2329. doi:10.1056/
NEJMoa1807801
7. Johansson E, Steineck G, Holmberg L, et al;
SPCG-4 Investigators. Long-term quality-of-life
outcomes after radical prostatectomy or watchful
waiting: the Scandinavian Prostate Cancer Group-4
randomised trial. Lancet Oncol. 2011;12(9):891-899.
doi:10.1016/S1470-2045(11)70162-0

E2 JAMA Oncology Published online November 11, 2021 (Reprinted) jamaoncology.com

© 2021 American Medical Association. All rights reserved.

Downloaded From: https://jamanetwork.com/ by a Mexico | Access Provided by JAMA User on 11/21/2021

 Gynecologic Oncology 117 (2010) 270–275

Contents lists available at ScienceDirect

Gynecologic Oncology
j o u r n a l h o m e p a g e : w w w. e l s e v i e r. c o m / l o c a t e / y g y n o

Impact of hydronephrosis on outcome of stage IIIB cervical cancer patients with

disease limited to the pelvis, treated with radiation and concurrent chemotherapy:
A Gynecologic Oncology Group study
Peter G. Rose a,⁎, Shamshad Ali b, Charles W. Whitney c, Rachelle Lanciano d, Frederick B. Stehman e
a
Cleveland Clinic Foundation and Case Western Reserve University, Cleveland, OH, USA
b
Gynecologic Oncology Group Statistical and Data Center, Roswell Park Cancer Institute, Buffalo, NY, USA
c
Christiana Gynecologic Oncology, LLC, Apex Medical Center, Newark, DE, USA
d
Delaware County Memorial Hospital, Drexel Hill, PA, USA
e
Indiana University School of Medicine, Indianapolis, IN, USA

a r t i c l e i n f o a b s t r a c t

Article history: Objectives. To estimate the significance of hydronephrosis and impact of ureteral obstruction relief on
Received 7 December 2009 outcome in patients with stage IIIB cervical cancer treated with radiation and concurrent chemotherapy.
Available online 24 February 2010 Methods. We retrospectively studied stage IIIB cervical cancer patients treated on GOG trials 56, 85, 120
and 165 evaluating radiation and concurrent chemotherapy. Eligible patient records were reviewed to assess
Keywords: the presence of hydronephrosis and treatment of ureteral obstruction. Patients were classified into three
Stage IIIB cervical cancer
groups; no hydronephrosis, hydronephrosis relieved from ureteral obstruction via stent or percutaneous
Hydronephrosis
Performance status
nephrostomy and hydronephrosis without treatment of ureteral obstruction.
Survival Results. 539 stage IIIB patients were studied. Hydronephrosis was present in 238 (44.2%). Patient age, race,
and tumor characteristics (size, histology and grade) were not significantly different between patients with or
without hydronephrosis. Patients with hydronephrosis received similar doses of radiation and cisplatin-based
chemotherapy. Both overall and progression-free survival were worse with hydronephrosis (log-rank test
p value = 0.0189 and 0.0186, respectively). Univariable analysis identified five prognostic factors; pelvic nodal
metastasis (p = 0.0001), tumor diameter (p = 0.0007), cisplatin-based concurrent chemoradiation
(p = 0.0031), hydronephrosis (p = 0.0189), and performance status (p = 0.0359). Hydronephrosis was
associated with worse performance status (p b 0.001). On multivariable analysis hydronephrosis was not a
significant prognostic factor. Ureteral obstruction relief occurred for 88% of patients and was associated with
improved survival.
Conclusion. In patients with stage IIIB cervical cancer restricted to the pelvis, hydronephrosis at
presentation is a significant but not independent prognostic factor associated with poor performance status
and poorer survival. Relief of ureteral obstruction is correlated with improved outcome.
© 2010 Published by Elsevier Inc.

Introduction without concurrent cisplatin-based chemotherapy. Stage IIIB is

The prognosis for patients with advanced stage cervical cancer is
based on both the extent of local pelvic disease and the presence of
metastases. The Federation International Obstetrics and Gynecology
(FIGO) staging system for cervical cancer describes various stages and
substages of disease which, with the exception of the most advanced
stage (stage IVB), describe the extent of local/regional pelvic disease.
Despite the use of relatively standard radiation protocols, survival
worsens with increasing disease stage for patients treated with or
⁎ Corresponding author. Division of Gynecologic Oncology, Department of Obstetrics
and Gynecology, Cleveland Clinic Foundation, A81, 9500 Euclid Avenue, Cleveland,
OH 44195, USA. Fax: +1 216 444 8551.
E-mail address: rosep@ccf.org (P.G. Rose).
defined as local extension of the cervical tumor to the pelvic sidewall
or the presence of hydronephrosis. Hydronephrosis, particularly if
bilateral, may result in varying degrees of renal insufficiency which
may limit the ability to deliver concurrent cisplatin-based chemo-
therapy which has been recommended in a National Cancer Institute
Clinical Alert [1]. Renal function may be improved or restored if the
renal obstruction is relieved by ureteral stents or percutaneous
nephrostomies. The impact of hydronephrosis and interventions to
correct it has not been extensively reported in advanced stage cervical
cancer patients treated with curative intent with radiation and
concurrent chemotherapy [2–6].
The disease extent in patients with stage IIIB disease can vary
depending on whether there is unilateral or bilateral parametrial
disease, whether there is unilateral or bilateral sidewall involvement,

0090-8258/$ – see front matter © 2010 Published by Elsevier Inc.

doi:10.1016/j.ygyno.2010.01.045
 P.G. Rose et al. / Gynecologic Oncology 117 (2010) 270–275 271

whether hydronephrosis is present and if present is it unilateral or Table 1

bilateral, and overall tumor size. In addition to these local factors Patient factors by hydronephrosis.

patients with stage IIIB are at increased risk for extrapelvic metastasis. Characteristic Hydronephrosis statusa
Although basic radiologic imaging with plain radiographs is allowed by
No Hyn Hyn WR Hyn WOR
the FIGO staging system, more advanced techniques including
n (%) n (%) n (%)
lymphangiography, computerized tomography (CT), magnetic reso-
nance imaging (MRI) and more recently positive emission tomogra- Age group ≤25 2 (0.7) 2 (1.0) 0 (0.0)
26–35 31 (10.3) 25 (12.0) 3 (10.3)
phy (PET) which have been utilized to evaluate patients for extrapelvic
36–45 80 (26.6) 44 (21.1) 9 (31)
nodal or distant metastases are not accepted [7]. Furthermore, the 46–55 83 (27.6) 75 (35.9) 11 (37.9)
prognosis of patients with stage IIIB disease depends on how metas- 56–65 62 (20.6) 36 (17.2) 6 (20.7)
tases to the paraaortic nodes have been excluded, i.e. radiologically or ≥66 43 (14.3) 27 (12.9) 0 (0.0)
by surgical lymphadenectomy [8]. Surgical findings are not utilized in Race Black 60 (19.9) 37 (17.7) 8 (27.6)
White 154 (51.2) 111 (53.1) 14 (48.3)
the FIGO staging system and therefore disease identified by retroper-
Other 9 (3.0) 8 (3.8) 0 (0.0)
itoneal lymphadenectomy does not change the patient's disease stage. Unknown 78 (25.9) 53 (25.4) 7 (24.1)
The purpose of the current study was to estimate the significance GOG performance 0 179 (59.5) 115 (55.0) 19 (65.5)
of ureteral obstruction as a prognostic variable in stage IIIB cervical 1 107 (35.5) 78 (37.3) 9 (31)
cancer with disease limited to the pelvis. Additionally, we sought to 2 13 (4.3) 14 (6.7) 0 (0.0)
3 2 (0.7) 2 (1.0) 1 (3.4)
estimate the impact of relieving ureteral obstruction by ureteral stent Tumor size (cm) ≤2.0 2 (0.7) 0 (0.0) 0 (0.0)
or percutaneous nephrostomy in patients with stage IIIB cervical 2.1–6.0 127 (42.2) 84 (40.2) 10 (34.5)
cancer treated with chemoradiation. 6.1–10.0 152 (50.5) 112 (53.6) 17 (58.6)
≥10.1 17 (5.6) 11 (5.3) 2 (6.9)
Not reported 3 (1.0) 2 (1.0) 0 (0.0)
Materials and methods
Tumor grade 1 13 (4.3) 14 (6.7) 2 (6.9)
2 205 (68.1) 141 (67.5) 19 (65.5)
We retrospectively reviewed patients entered on four randomized 3 79 (26.2) 51 (24.4) 8 (27.6)
clinical trials performed by the Gynecologic Oncology Group (GOG) in Not graded 4 (1.3) 3 (1.4) 0 (0.0)
locally advanced stage disease (stages IIB–IVA) conducted from June Cell type Squamous cell carcinoma 277 (92.0) 195 (93.3) 28 (96.6)
Adeno-squamous 12 (4.0) 8 (3.8) 1 (3.4)
1981 to August 2000 [9–12]. These studies have all been reported carcinoma
previously and all utilized external and intracavitary radiation and Adenocarcinoma, NS 10 (3.3) 5 (2.4) 0 (0.0)
concurrent chemotherapy. It is important to note that eligibility for Other 2 (0.7) 1 (0.5) 0 (0.0)
these trials included a serum creatinine less than 2.0 mg/dl prior to Parametrial None 3 (1.0) 2 (1.0) 1 (3.4)
involvement Unilateral 148 (49.2) 109 (52.2) 14 (48.3)
enrollment. Information regarding the presence of unilateral or bilat-
Bilateral 142 (47.2) 95 (45.5) 12 (41.4)
eral parametrial disease, overall tumor size, whether hydronephrosis Not done 8 (2.7) 3 (1.4) 2 (6.9)
is present and if present is it unilateral or bilateral, and what inter- Pelvic node status Negative 163 (54.2) 94 (45) 16 (55.2)
ventions were utilized to correct it prior to initiation of therapy Positive 43 (14.3) 28 (13.4) 6 (20.7)
was abstracted from individual patient records. Additionally, data Not done 95 (31.6) 87 (41.6) 7 (24.1)
a
regarding the presence of pelvic nodal metastasis if sampled, maxi- Hyn = hydronephrosis, WR = with relief, WOR = without relief.
mum tumor diameter, use of cisplatin-containing chemotherapy
during radiation, the presence of hydronephrosis, and performance cumulative doses of radiation and cisplatin-based chemotherapy com-
status were abstracted and analyzed. The maximum tumor diameter pared to patients without hydronephrosis (median radiation dose to
(cm) was assessed as a continuous variable. Performance status was point A was 81 Gy for both groups and median number of cycles of
evaluated on an ordinal scale (0, 1, 2, and 3). Due to differences in cisplatin-based therapy was 5 for both groups). There was no signifi-
progression-free survival (PFS) and overall survival (OS) for cisplatin- cant difference in observed toxicity with cisplatin-containing therapy
containing chemotherapy regimens we examined adverse effects in the patients with or without hydronephrosis (Supplementary
of cisplatin-based chemotherapy in patients with relative to those Table 1). Both PFS and OS were significantly worse for patients with
without baseline hydronephrosis. hydronephrosis (p = 0.0189 and 0.0186, respectively) (Figs. 1 and 2).
Product-limit estimates were calculated according to Kaplan– Approximately 12% of patients presenting with hydronephrosis
Meier method and differences in PFS and OS were assessed utilizing had bilateral hydronephrosis. The median survivals were 18.6 and
the log-rank test [13,14]. The Cox model was used to adjust for prog- 31.7 months for patients with bilateral or unilateral hydronephrosis,
nostic factors (reported by Stehman et al.) and to estimate hazard respectively (Fig. 3). However, this difference was not statistically
ratios (and 95% confidence interval) of PFS and OS [15]. In this analysis, different (p value = 0.2537).
performance status of (2 or 3) was compared to (0 or 1). The incidence Univariable analysis identified five unfavorable prognostic factors;
of maximum adverse event grade among patients treated with plati- pelvic nodal metastasis (p = 0.0001), increasing maximum tumor
num based chemotherapy was summarized and tabulated by hydro- diameter (p = 0.0007), the lack of cisplatin-based concurrent chemor-
nephrosis status at baseline. Fisher's exact test was implemented to adiation (p = 0.0031), the presence of hydronephrosis (p = 0.0189),
determine associations between hydronephrosis status and baseline and performance status N1 (p = 0.0359) (Table 2A). Hydronephrosis
traits [16]. was associated with poorer GOG performance status (p b 0.001).
Adjusting for tumor size, pelvic node status, performance status and
Results treatment, hydronephrosis was not a significant prognostic factor
(Table 2B).
Five hundred and thirty-nine patients with stage IIIB cervical Ureteral obstruction relief by either stent or percutaneous
cancer from prior GOG protocols 56 (N = 100), 85 (N = 118), 120 nephrostomy occurred for 88% of patients presenting with hydrone-
(N = 220) and 165 (N = 101) were studied. The demographics of these phrosis. In particular, patients presenting without hydronephrosis
patients are presented in Table 1. Hydronephrosis was present in 238 showed the most favorable outcome (median PFS = 48 months,
stage IIIB patients (44.2%). Patient age, race, and tumor characteristics OS = 70 months) followed by those who got treatment for ureteral
(size, histology and grade) were not significantly different between obstruction (median PFS = 18 months and OS = 34 months). The
the groups. Patients presenting with hydronephrosis received similar group of patients presenting with hydronephrosis in whom ureteral
272 P.G. Rose et al. / Gynecologic Oncology 117 (2010) 270–275

Fig. 1. PFS by hydronephrosis among stage IIIB patients (log-rank test, p value = 0.0189).

obstruction was not invasively relieved before treatment had the
worst outcome (median PFS = 10 months and OS = 17 months). Both
PFS and OS were significantly different across groups (log-rank test p
value = 0.0190 and 0.0078 respectively) (Figs. 4 and 5).
Discussion
The strengths of this study include the large number (539) of stage
IIIB cervical cancer patients treated in four prospective randomized
trials utilizing standardized radiation doses and schedules. Prognostic
factors identified included pelvic nodal metastasis, tumor diameter,
cisplatin-based concurrent chemoradiation, hydronephrosis, and
performance status. The FIGO staging system is a clinical staging
system which is widely applicable but lacks recognition of any of the
above identified prognostic factors. While tumor dimensions are the
basis of stage I substages (IA1–IB2) they are not utilized to assess stage
IIB–IVA disease. The presence of hydronephrosis among stage IIIB
patients was associated with poorer PFS and OS in this study. This was
seen despite the fact that patients with hydronephrosis received similar
doses of radiation and cisplatin-based chemotherapy. The presence of
hydronephrosis is not taken into consideration in the current clinical
staging system. Although not an independent prognostic variable on
multivariable analysis, hydronephrosis is an easily determined factor
associated with poorer performance status and poorer OS in univariable
analysis.
Logsdon and Eifel reported the MD Anderson experience with stage
IIIB cervical cancer from 1960 to 1993 [2]. Of 983 patients treated with
curative intent fewer than 5% received concurrent chemotherapy.
Thirty percent (291) had hydronephrosis and these patients had a
poorer disease specific survival than patients without hydronephrosis

Fig. 2. OS by hydronephrosis among IIIB patients (log-rank test, p value = 0.0186).

 P.G. Rose et al. / Gynecologic Oncology 117 (2010) 270–275 273

Fig. 3. OS by hydronephrosis type (log-rank test, p value = 0.2537).

(28% vs. 40%). Improved survival was noted in recent years with the
routine use of intracavitary radiation [2]. The outcome for patients
with bilateral vs. unilateral hydronephrosis was not different. The
management of patients with hydronephrosis was not addressed in this
study. Chao et al. reported the Millinckrodt Institutes experience with
hydronephrosis in 297 patients with stage IIIB cervical cancer [3]. These
authors noted a poorer outcome and a higher distant failure rate for
stage IIIB patients that did not reach the sidewall in whom ureteral
obstruction was presumably due to nodal metastasis. In contrast for
patients with sidewall extension no difference in survival was noted for
patients with or without hydronephrosis.
The poor survival of patients with primary cervical cancer and
obstructive uropathy (0–50%) has been documented in previous
reports [2–6]. Many of these patients have been treated with palliative
intent [4]. Since the number of patients with primary cervical cancer
and obstructive uropathy treated with curative intent with radiation
and concurrent chemotherapy in previous reports is small, optimal
management is not established. However, the patient with ureteral
obstruction is often a candidate for pelvic radiation therapy with
curative intent. The current series is unique in that all patients in-
cluded in this study were enrolled in National Cancer Institute trials
and treated with curative intent.
Ureteral obstruction from cervical cancer occurs most often in the
cardinal ligament 3–6 cm proximal to the ureterovesical junction
[12]. A number of surgical procedures can be employed to relieve
obstructive uropathy with endoscopic retrograde ureteral catheteri-
zation, surgical nephrostomy, surgical ureterostomy, and percutane-
ous nephrostomy most commonly employed. However, endoscopic
retrograde ureteral catheterization is not technically feasible in over
70% of patients with malignant ureteral obstruction [17]. Percutaneous
nephrostomy is currently the preferred supravesical diversion because
of its minimal morbidity and mortality [18,19].
Taylor and Anderson studied 18 patients from two institutions
with untreated cervical cancer and obstructive uropathy treated with
either medical (hemodialysis and peritoneal dialysis) or surgical
management and found no method clearly superior [5]. Horan et al.
reported their retrospective experience with 13 patients; five with
unilateral and eight with bilateral ureteral obstruction [6]. Eight
required ureteral stenting which was usually performed prior to
radiation therapy. Ureteral stenting during radiation resulted in
significant delays in treatment. They also prospectively studied three
patients with unilateral ureteral obstruction whose renal function and
hydronephrosis during treatment remained stable or improved with
Table 2B
Multivariate analysis of prognostics factors by Cox proportional hazard regression.

Table 2A Factor OS PFS

Univariate analysis of prognostics factors by Cox proportional hazard regression. HR (95% CI) p HR (95% CI) p
Factor OS PFS Tumor size (cm) 1.079 (1.018–1.143) 0.0099 1.059 (1.000–1.122) 0.0459
Performance 2.380 (1.276–4.438) 0.0064 1.790 (0.965–3.320) 0.0647
HR (95% CI) p HR (95% CI) p
status
Tumor size (cm) 1.077 (1.032–1.125) 0.0007 1.069 (1.024–1.116) 0.0025 (2, 3 vs. 0, 1)
Performance status 1.560 (1.030–2.364) 0.0359 1.501 (0.999–2.254) 0.0506 Pelvic node status 1.857 (1.378–2.504) b 0.0001 1.726 (1.285–2.318) 0.0003
(2, 3 vs. 0, 1) (positive vs.
Pelvic node status 1.780 (1.326–2.390) 0.0001 1.671 (1.248–2.237) 0.0006 negative)
(positive vs. Hydronephrosis 1.244 (0.945–1.639) 0.1199 1.248 (0.955–1.629) 0.1045
negative) (yes vs. no)a
Hydronephrosis 1.291 (1.043–1.597) 0.0189 1.282 (1.041–1.579) 0.0193 Chemotherapy 0.747 (0.568–0.981) 0.0362 0.745 (0.570–0.973) 0.0305
(yes vs. no) (cisplatin vs.
Chemotherapy 0.722 (0.582–0.896) 0.0031 0.718 (0.582–0.887) 0.0021 other)
(cisplatin vs. a
In the multivariate analysis, when adjusting for tumor size, pelvic node status,
other)
performance status and treatment, hydronephrosis was no longer significant.
274 P.G. Rose et al. / Gynecologic Oncology 117 (2010) 270–275

Fig. 4. PFS by hydronephrosis relief status (log-rank test, p value = 0.0190).

time. Two of these three received concurrent cisplatin chemotherapy.
The authors recommended elective urinary diversion for patients
prior to radiation therapy for patients with bilateral obstruction and a
creatinine clearance b50 ml/min.
Ureteral obstruction may compromise the ability to deliver poten-
tially nephrotoxic cisplatin chemotherapy concurrently with radiation
therapy which has been demonstrated in randomized trials to be supe-
rior to radiation alone or radiation and non-cisplatin chemotherapy.
However, in the current study the presence of hydronephrosis did not
limit the doses of cisplatin-containing chemotherapy and similar radia-
tion doses were able to be given. Eligibility for the trials in this report
included a serum creatinine less than 2.0 mg/dl prior to enrollment. Our
analysis of the impact of hydronephrosis in this trial is limited to
patients whose renal function was either normal or improved with a
nephrostomy or ureteral stent prior to enrollment. Therefore, patients
whose renal function was more abnormal or failed to improve are not
represented in this study. Additional weaknesses of this study are
variations in radiation therapy techniques in the most recent study
(Protocol #165) including a slightly higher (5%) radiation dose, more
compacted treatment time and the use of high dose-rate brachytherapy.

Fig. 5. OS by hydronephrosis relief status (log-rank test, p value = 0.0078).

 P.G. Rose et al. / Gynecologic Oncology 117 (2010) 270–275
Patients with FIGO stage IIIB cervical cancer and hydronephrosis
had a shorter PFS and OS than those whose stage was based only on
palpation of cancer extension to the pelvic sidewall. Thus, the pres-
ence of hydronephrosis is a clinical surrogate of poorer performance
status and poorer OS.
Conflict of interest statement
The authors have no conflicts of interest to declare.
Acknowledgments
This study was supported by National Cancer Institute grants to
the Gynecologic Oncology Group Administrative Office (CA 27469)
and the Gynecologic Oncology Group Statistical Office (CA 37517).
The following Gynecologic Oncology Group member institutions
participated in the primary treatment studies: University of Alabama
at Birmingham, Oregon Health Sciences University, Duke University
Medical Center, Abington Memorial Hospital, University of Rochester
Medical Center, Walter Reed Army Medical Center, University of
Minnesota Medical School, University of Southern California at Los
Angeles, University of Mississippi Medical Center, Colorado Gyneco-
logic Oncology Group P.C., University of California at Los Angeles,
University of Miami School of Medicine, Milton S. Hershey Medical
Center, Georgetown University Hospital, University of Cincinnati,
University of North Carolina School of Medicine, University of Iowa
Hospitals and Clinics, Indiana University School of Medicine, Wake
Forest University School of Medicine, Albany Medical College,
University of California Medical Center at Irvine, Rush-Presbyterian-
St. Luke's Medical Center and Eastern Virginia Medical School.
Appendix A. Supplementary data
Supplementary data associated with this article can be found, in
the online version, at doi:10.1016/j.ygyno.2010.01.045.
References
[1] NCI Clinical Announcement, U.S. Dept. of Health and Human Services, Public
Health Service, National Institutes of Health, February 1999.
275
[2] Logsdon MD, Eifel PJ. FIGO IIIB squamous cell carcinoma of the cervix: an analysis
of prognostic factors emphasizing the balance between external beam and
intracavitary radiation therapy. Int J Radiat Oncol Biol Phys 1999;43:763–75.
[3] Chao KS, Leung WM, Grigsby PW, Mutch DG, Herzog T, Perez CA. The clinical
implications of hydronephrosis and the level of ureteral obstruction in stage IIIB
cervical cancer. Int J Radiat Oncol Biol Phys 1998;40:1095–100.
[4] Krynicki R, Bidzinski M, Wierzba W, Panek G, Jonska J, Lindner B. Hydronephrosis
as a prognostic factor for patients with FIGO IIIB cervical cancer. Ginekol Pol 2003;74:
32–9.
[5] Taylor Jr PT, Anderson WA. Untreated cervical cancer complicated by obstructive
uropathy and olioguric renal failure. Gynecol Oncol 1981;11:162–74.
[6] Horan G, McArdle O, Martin J, Collins CD, Faul C. Pelvic radiotherapy in patients
with hydronephrosis in stage IIIB cancer of the cervix: renal effects and the
optimal timing of urinary diversion. Gynecol Oncol 2006;101:441–4.
[7] Grigsby PW. The prognostic value of PET and PET/CT in cervical cancer. Cancer
Imaging 2008;8:146–55.
[8] Gold MA, Tian C, Whitney CW, Rose PG, Lanciano R. Surgical versus radiographic
determination of para-aortic lymph node metastases before chemoradiation for
locally advanced cervical carcinoma: a Gynecologic Oncology Group study. Cancer
2008;112:1954–63.
[9] Stehman FB, Bundy BN, Thomas G, Keys HM, d'Ablaing III G, Fowler Jr WC, et al.
Hydroxyurea versus misonidazole with radiation in cervical carcinoma: long-term
follow-up of a Gynecologic Oncology Group trial. J Clin Oncol 1993;11:1523–8.
[10] Whitney CW, Sause W, Bundy BN, Malfetano JH, Hannigan EV, Fowler Jr WC, et al.
A randomized comparison of fluorouracil plus cisplatin versus hydroxyurea as an
adjunct to radiation therapy in stages IIB–IVA carcinoma of the cervix with nega-
tive para-aortic lymph nodes. A Gynecologic Oncology Group and Southwest
Oncology Group study. J Clin Oncol 1999;17:1339–48.
[11] Rose PG, Bundy BN, Watkins EB, Thigpen JT, Deppe G, Maiman MA, et al. Concurrent
cisplatin-based chemoradiation improves progression free and overall csurvival in
advanced cervical cancer: results of a randomized Gynecologic Oncology Group
study. N Engl J Med 1999;340:1144–53.
[12] Lanciano R, Calkins A, Bundy B, Parham G, Lucci III JA, Moore DH, et al. Randomized
comparison of weekly cisplatin or protracted venous infusion of fluorouracil in
combination with pelvic radiation in advanced cervix cancer: a Gynecologic
Oncology Group study. J Clin Oncol 2005;23:8289–95.
[13] Bland JM, Altman DG. The logrank test. BMJ 2004;328:1073.
[14] Kaplan EL, Meier P. Nonparametric estimation from incomplete observations. J Am
Stat Assoc 1958;53:457–81.
[15] Stehman FB, Bundy BN, DiSaia PJ, Keys HM, Larson JE, Fowler WC. Carcinoma of the
cervix treated with radiation therapy I: a multi-variate analysis of prognostic
variables in the Gynecologic Oncology Group. Cancer 1991;67:2776–85.
[16] Agresti A. A survey of exact inference for contingency tables. Stat Science 1992;7:
131–77.
[17] Kearney GP, Tumeh SS. Urinary tract involvement in gynecologic cancer. In: Knapp
RC, Berkowitz RS, editors. Gynecologic Oncology. 2nd ed. New York: McGraw-Hill;
1993. p. 341.
[18] Zadra JA, Jewett MA, Keresteci AG, Rankin JT, St. Louis E, Grey RR, et al. Nonoperative
urinary diversion for malignant ureteral obstruction. Cancer 1987;60:1353–7.
[19] Soper JT, Blaszczyk TM, Oke E, Clarke-Pearson D, Creasman WT. Percutaneous
nephrostomy in gynecologic oncology patients. Am J Obstet Gynecol 1988;158:
1126–31.

JAMA Clinical Guidelines Synopsis
Cervical Cancer Screening Guideline for Individuals at Average Risk
Julie Chor, MD, MPH; Andrew M. Davis, MD, MPH; Jennifer M. Rusiecki, MD
GUIDELINE TITLE Cervical Cancer Screening for Individuals
at Average Risk: 2020 Guideline Update From the
American Cancer Society
RELEASE DATE July 30, 2020
PRIOR VERSION 2012
DEVELOPER AND FUNDING SOURCE American Cancer
Society (ACS)
TARGET POPULATION Individuals with cervix in situ
MAJOR RECOMMENDATIONS
• Begin screening at age 25 years regardless of sexual history
or HPV vaccination status (strong recommendation)
• Primary HPV testing every 5 years through age 65 years
(strong recommendation)
• If primary HPV testing is not available use cotesting
(HPV+ cytology) every 5 years or every 3 years if cytology
only (strong recommendation)
• Discontinue screening at age 65 years if no history of
cervical intraepithelial neoplasia grade 2 or more severe
diagnosis in last 25 years and adequate negative prior
screening in last 10 years (qualified recommendation)
Summary of the Clinical Problem
Even though screening has significantly reduced cervical cancer–
related deaths in the US, an estimated 14 480 individuals will be di-
agnosed with cervical cancer and an estimated nearly 4300 will die
from this disease during 2021.1 Worldwide, cervical cancer is the
fourth most common cancer in women and is the leading cause of
cancer-related deaths among women in Africa.2 In the US, Black
women experience significantly higher cervical cancer mortality rates
than White women,3 and disruptions of care during the COVID pan-
demic have affected cancer screening in general. Testing for cervi-
cal infection with high-risk human papillomavirus (HPV) has much
higher sensitivity than traditional cytology for the dysplastic cervi-
cal changes that precede cancer, making this approach an increas-
ingly important prediction tool for invasive cervical cancer. The 2020
American Cancer Society (ACS) Cervical Cancer Screening Guide-
line provides the latest recommendations, including when to initi-
ate screening, frequency of testing, and which modality to use.4
Characteristics of the Guideline Source
A volunteer ACS guideline development group (GDG) of 18 individu-
als with wide expertise including gynecology, oncology, and pri-
mary care reviewed the evidence, supported modeling analyses, and
received input from various groups, including many professional so-
cieties and patient-based organizations.4 Required disclosures in-
cluded financial and nonfinancial relationships (Table).
jama.com
© 2021 American Medical Association. All rights reserved.
Downloaded From: https://jamanetwork.com/ by a Mexico | Access Provided by JAMA User on 11/21/2021
Clinical Review & Education
Evidence Base
The 2020 ACS guideline updates the 2012 ACS guideline for individu-
als at average risk, defined as those with a cervix in situ who are ini-
tiating cervical cancer screening, have normal cervical cancer screen-
ing results in the past, or who have been returned to routine cervical
cancer screening based on standard follow-up recommendations.4
The recommendations do not apply to individuals at increased risk for
cervical cancer due to solid organ or stem cell transplant, HIV infec-
tion, immunosuppression from other causes, or in utero exposure
to diethylstilbestrol.4
The 2020 ACS guideline includes 3 important changes: (1) the
age of initial screening is increased to 25 years (previously 21 years);
(2) the interval for testing is extended to every 5 years (previously
3 years); and (3) an updated recommendation for initial screening
approach deemphasizes the 2012 guideline’s use of cervical
cytology.4 The 2020 guidelines recommend primary HPV testing
alone every 5 years, with the option of cotesting with cytology and
HPV testing every 5 years or cytology alone every 3 years when pri-
mary HPV testing is not available.4 The reason that cotesting may
be available when primary HPV testing is not available is that the spe-
cific HPV testing performed in a given hospital laboratory may not
be approved for primary HPV testing. There is no change to the prior
recommendation to stop screening at age 65 years.
Importantly, these recommendations do not differ by HPV vacci-
nation status despite data from Sweden that demonstrated a decrease
in cervical cancer rates from 94 cases/100 000 persons among unvac-
cinated people to 47/100 000 persons among a widely vaccinated
population.5 While specific risk reduction to US populations cannot be
directlyinferred,theguidelinestates,“futurerecommendationsforcer-
vicalcancerscreeningwillneedtoincorporateHPVvaccinationstatus.”4
The ACS 2020 guideline was informed primarily by 2 reports
commissioned by the US Preventive Services Task Force (USPSTF)
2018 cervical cancer screening update: a systematic review and a
mathematical disease-simulation model,6,7 which accompanied the
USPSTF recommendation statement.8 The decision to delay initia-
tion of screening reflects the low incidence and mortality of cervi-
cal cancer among individuals younger than 25 years, the high inci-
dence of transient HPV infections among those younger than 25
years, and the potential for treatment of cervical abnormalities
to result in adverse obstetric events (eg, premature delivery).4
Table. Guideline Rating
Standard Rating
Establishing transparency Good
Management of conflict of interest in the guideline development group Good
Guideline development group composition Good
Clinical practice guideline-systematic review intersection Good
Establishing evidence foundations and rating strength for each of the Good
guideline recommendations
Articulation of recommendations Good
External review Good
Updating Good
Implementation issues Fair
(Reprinted) JAMA Published online November 12, 2021 E1
 Clinical Review & Education JAMA Clinical Guidelines Synopsis

The recommendation to use primary HPV testing for screening was
based on data that showed increased sensitivity and long-term nega-
tive predictive value of HPV-based screening compared with cytol-
ogy alone, as well as the approval of primary HPV screening tests
by the US Food and Drug Administration.4
The ACS GDG commissioned a decision analysis from the model-
ing group that supported the 2018 USPSTF cervical cancer screening
update.Modelinganalysescomparedseveralscreeningstrategies,vary-
ing age of initiation (21 or 25 years), interval (3, 4, or 5 years), and mo-
dality (cytology alone, cotesting, or primary HPV). The models com-
pared outcomes based on effectiveness (ability to identify treatable
cervicalabnormalities)andefficiency(minimizingunnecessaryscreen-
ing and colposcopies per life-year gained and cancer averted).
All strategies conclude screening at age 65 for individuals with-
out a history of cervical intraepithelial neoplasia (CIN) 2 or worse in
the past 25 years, and with documented adequate screening in the
past 10 years, defined as 2 consecutive tests within the past 10 years,
with the most recent within the past 3 to 5 years, depending on the
test used. This recommendation holds even if a new sexual partner
of any age is reported after age 65. Individuals of any age who have
had surgical removal of the cervix (and no history of CIN 2 or more
severe diagnosis) should not be screened with cytology or HPV test-
ing for lower genital tract malignancies, such as vaginal cancer.
Benefits and Harms
The GDG prioritized reduction in cervical cancer cases while maxi-
mizing life-years gained and minimizing unnecessary screening and
colposcopies. Delaying the starting age for cervical cytology test-
ing (ie, Papanicolaou testing) may lessen emotional harms associ-
ated with low-risk screening, including anxiety and distress from ex-
aminations, and may reduce the economic and emotional harms that
may occur after interventions in response to less serious and tran-
sient cytopathologic findings.
A potential concern is that the first Pap test brings many indi-
viduals in for their first preventive reproductive health visit. Wait-
ing until age 25 years for initial screening may delay counseling and
screening for effective contraception and intimate partner vio-
lence, as well as service delivery around healthy sexuality. Guide-
lines suggesting a later age of screening have been linked to re-
duced screening for sexually transmitted infections (STIs).9
Discussion
The2020ACSguidelines4 andthe2018USPSTFcervicalcancerscreen-
ingguidelines8 differonageofinitiation,frequency,andmodality.Based
on additional analyses obtained through systematic review and deci-
sion analysis, the 2020 ACS guidelines aim to incorporate technologi-
cal advances, increase effectiveness, and decrease potential risks and
burdens of cervical cancer screening (discomfort, anxiety, and addi-
tional procedures) compared with the 2018 USPSTF guidelines.
Clinicians who wish to optimize implementation of the 2020 ACS
guidelinesandreconciletheserecommendationswiththe2018USPSTF
guidelinesareencouragedtoengagepatientsinshareddecision-making
regarding medical considerations, risk of HPV exposure, HPV vacci-
nation status, and prior abnormal cervical cancer screening results.
Ideally,thesediscussionswillexplorepatients’values,experiences,and
preferences regarding cervical cancer screening and their comfort with
pelvic examinations. Fostering strong patient-clinician relationships in
earlyadulthoodmaykeeppatientsengagedinpreconceptualreproduc-
tive health care and also promote earlier prenatal care.
Theseconsiderationsargueforplanningandpoliciestoengageado-
lescents and adults younger than 25 years in preventive reproductive
health care. This need is even more salient in vulnerable communities,
givenlong-standingracialandethnicreproductivehealthdisparities,in-
cludinglowerratesofcontraceptiveuseandHPVvaccination,andhigher
rates of STIs and reproductive cancers. Such inequities and barriers to
care result in a greater risk of preterm and low-birth-weight deliveries
and markedly higher maternal morbidity and mortality.10
Areas in Need of Future Study or Ongoing Research
Additional research is needed to evaluate how guideline implemen-
tation affects key outcomes, including initiation and adherence with
Pap testing, access to effective contraception and timely and trusted
prenatal care, and STI rates in persons younger than 25 years. Evalu-
ation should include assessment for potentially widening racial and
ethnic and socioeconomic inequities in reproductive health. The psy-
chosocial value of delayed screening initiation and less frequent pel-
vic examinations, particularly in those with a history of painful ex-
aminations or trauma, deserves study. Finally, because the new ACS
guideline does not incorporate vaccination status in its recommen-
dations, monitoring of cervical cancer detection in populations with
varying degrees of HPV vaccine coverage will be important.

ARTICLE INFORMATION
Author Affiliations: Department of Obstetrics and
Gynecology, University of Chicago, Chicago, Illinois
(Chor); Section of General Internal Medicine,
Department of Medicine, University of Chicago,
Chicago, Illinois (Davis, Rusiecki).
Corresponding Author: Andrew M. Davis, MD,
MPH, Section of General Internal Medicine,
Department of Medicine, University of Chicago,
5841 S Maryland Ave, Chicago, IL 60637
(amd@uchicago.edu).
Published Online: November 12, 2021.
doi:10.1001/jama.2021.13448
Conflict of Interest Disclosures: None reported.
REFERENCES
1. Siegel RL, Miller KD, Fuchs HE, Jemal A. Cancer
statistics, 2021. CA Cancer J Clin. 2021;71(1):7-33.
doi:10.3322/caac.21654
2. Arbyn M, Weiderpass E, Bruni L, et al. Estimates
of incidence and mortality of cervical cancer in
2018. Lancet Glob Health. 2020;8(2):e191-e203.
doi:10.1016/S2214-109X(19)30482-6
3. Beavis AL, Gravitt PE, Rositch AF.
Hysterectomy-corrected cervical cancer mortality
rates reveal a larger racial disparity in the United
States. Cancer. 2017;123(6):1044-1050. doi:10.
1002/cncr.30507
4. Fontham ETH, Wolf AMD, Church TR, et al.
Cervical Cancer Screening for Individuals at Average
Risk: 2020 guideline update from the American
Cancer Society. CA Cancer J Clin. 2020;70(5):321-346.
doi:10.3322/caac.21628
5. Lei J, Ploner A, Elfström KM, et al. HPV
vaccination and the risk of invasive cervical cancer.
N Engl J Med. 2020;383(14):1340-1348.
6. Melnikow J, Henderson JT, Burda BU, et al.
Screening for cervical cancer with high-risk human
papillomavirus testing: updated evidence report
and systematic review for the US Preventive
Services Task Force. JAMA. 2018;320(7):687-705.
doi:10.1001/jama.2018.10400
7. Kim JJ, Burger EA, Regan C, Sy S. Screening for
cervical cancer in primary care: a decision analysis
for the US Preventive Services Task Force. JAMA.
2018;320(7):706-714. doi:10.1001/jama.2017.19872
8. US Preventive Services Task Force. Screening for
cervical cancer: US Preventive Services Task Force
Recommendation Statement. JAMA. 2018;320(7):
674-686. doi:10.1001/jama.2018.10897
9. Ursu A, Sen A, Ruffin M. Impact of cervical
cancer screening guidelines on screening for
chlamydia. Ann Fam Med. 2015;13(4):361-363.
10. Sutton MY, Anachebe NF, Lee R, Skanes H.
Racial and ethnic disparities in reproductive health
services and outcomes, 2020. Obstet Gynecol.
2021;137(2):225-233.

E2 JAMA Published online November 12, 2021 (Reprinted) jama.com

© 2021 American Medical Association. All rights reserved.

Downloaded From: https://jamanetwork.com/ by a Mexico | Access Provided by JAMA User on 11/21/2021

European Journal of Cancer (2014) 50, 1628– 1637

Available at www.sciencedirect.com

ScienceDirect
journal homepage: www.ejcancer.com

Effects of hormone replacement therapy on the rate

of recurrence in endometrial cancer survivors: A
meta-analysis

Seung-Hyuk Shim, Sun Joo Lee, Soo-Nyung Kim ⇑

Department of Obstetrics and Gynecology, Konkuk University School of Medicine, Seoul, Republic of Korea

Received 12 December 2013; received in revised form 3 March 2014; accepted 6 March 2014
Available online 28 March 2014

KEYWORDS Abstract Background: To quantify the effect of hormone replacement therapy (HRT) on the
Endometrial cancer recurrence in endometrial cancer (EC) survivors through a meta-analysis.
Hormone replacement Methods: A systematic literature review was conducted through October 2013 and included
therapy studies reporting estimates of effect size for the relationship between HRT use and the risk
Recurrence of EC recurrence. Study design features that may affect the selection of participants, the detec-
Meta-Analysis tion of EC recurrence and manuscript publication were assessed. If there was no significant
statistical heterogeneity across studies, then a fixed effects model was used to obtain pooled
estimates for the effect of HRT use on EC recurrence by combining study-specific estimates
of the odds ratio (OR).
Results: One randomised trial and five observational studies included 896 EC survivors who
used HRT and 1079 non-users. Over the combined study period, 19 of the 896 HRT users
experienced recurrence, whereas 64 of the 1079 controls did. The meta-analysis based on
the fixed effects model indicates no significant increase in the risk of recurrence in EC survi-
vors using HRT relative to the control group (OR: 0.53; 95% confidence interval: 0.30–0.96,
I2 = 49.0). This pattern was also observed in the subgroup analysis for the stage and type of
HRT. There was no evidence of any publication bias.
Conclusions: Although based mainly on observational studies, the literature does not provide
support for a positive relationship between HRT use and the risk of EC recurrence. Future
research should verify this relationship through randomised controlled trials over a longer term.
Ó 2014 Elsevier Ltd. All rights reserved.

⇑ Corresponding author: Address: Department of Obstetrics and Gynecology, Konkuk University School of Medicine, 1 Hwayang-dong,
Gwangjin-gu, Seoul 143-701, Korea. Tel.: +82 2 2030 7641; fax: +82 2 2030 7748.
E-mail address: snkim@kuh.ac.kr (S.-N. Kim).

http://dx.doi.org/10.1016/j.ejca.2014.03.006
0959-8049/Ó 2014 Elsevier Ltd. All rights reserved.
 S.-H. Shim et al. / European Journal of Cancer 50 (2014) 1628–1637
1. Introduction
Endometrial cancer (EC) is the most common gyne-
cologic malignancy in Western countries. In the United
States (US), 47,130 cases were newly diagnosed in
2012, and 8010 women died of EC in the same year
[1]. The number of EC patients has increased rapidly
in Korea, where it was third most common gyneco-
logic malignancy with 1616 new estimated cases in
2010 [2].
Surgical staging procedures such as hysterectomy and
bilateral salpingooophorectomy represent the corner-
stone treatment option for EC. Although EC is gener-
ally a postmenopausal disease, about 25% of EC
patients are premenopausal [3]. Therefore, menopausal
symptoms, particularly in reproductive women undergo-
ing surgical treatment for EC, are common. In addition,
menopausal symptoms caused by surgery tend to be
more severe than those resulting from normal meno-
pause [4]. Accordingly, relieving these symptoms is an
important issue in terms of the quality of life after treat-
ment because a majority of EC patients show an early-
stage disease and a favourable prognosis.
Although no scientific data demonstrate the detri-
mental effect of hormonal replacement therapy (HRT)
in EC survivors, many clinicians remain reluctant to pre-
scribe HRT for these patients because of the concern
over oestrogen use as an important aetiologic factor
for EC [5]. Without estimates of the effect of HRT on
EC recurrence, it is difficult for EC survivors and their
physicians to weigh the risk and benefits of HRT. In this
regard, the present study provides a meta-analysis to
estimate the effect of HRT on EC recurrence in patients
previously treated for EC by considering observational
and randomised studies.
2. Materials and methods
2.1. Literature search
A systematic review was conducted using developed
guidelines for reporting a systematic review [6,7]. MED-
LINE and EMBASE databases and the Cochrane Cen-
tral Register for Controlled Trials were searched with no
language limitations up to 15th October 2013. Papers
ahead of their publication were also included. The fol-
lowing keywords were used for this search: ‘endometrial
neoplasm or cancer or carcinoma or malignancy,’ ‘uter-
ine neoplasm or cancer or carcinoma or malignancy,’
‘oestrogen,’ ‘oestrogen replacement therapy,’ ‘hormone
replacement therapy,’ and ‘estradiol.’ The titles and
abstracts were checked to exclude any clearly unrelated
articles. The full text of remaining papers was evaluated
to determine their relevancy. In addition, the references
cited in selected papers and published reviews were con-
sidered to include any additional studies of relevance.
1629
All searches were performed independently by the two
authors (S.S. and S.L.).
2.2. Eligibility criteria
Studies were included if they met the following crite-
ria: (1) randomised controlled trials, a prospective or
retrospective cohort study, a nested case-control study,
or a population-based case-control study; (2) partici-
pants of interest being patients receiving surgical treat-
ment for EC; (3) the intervention of interest being
HRT exposure and (4) the outcome measure being EC
recurrence measured by the relative risk (RRs), the odds
ratio (ORs) or the hazard ratio (HRs) estimated with the
95% confidence interval (CI) (or sufficient data for cal-
culating them). Single-arm cohort studies were not
included in the meta-analysis.
2.3. Data extraction
Three authors (S.S., S.L. and S.K.) independently
extracted data, including the name of the first author,
the year of publication, the study design, the study loca-
tion, the study period, age, sample size (cases and con-
trols or cohort size), tumour stage, grade, exposure to
HRT use, disease-free interval (DFI) between the initial
EC diagnosis and HRT administration, type of HRT,
duration of HRT, follow-up after HRT administration,
EC recurrence and covariates controlled for in the anal-
ysis. When not reported, the follow-up after the initia-
tion of HRT was assumed to equal the duration of
HRT use. Each study was systematically reviewed for
features that could introduce some bias, including pro-
cedures for identifying participants, whether risk factors
for recurrence were similar at diagnosis, and whether the
follow-up duration was similar between HRT users and
non-users. Discrepancies in data extraction were jointly
reviewed until a consensus was reached.
2.4. Quality assessment
Study quality was evaluated independently by two
authors (S.S. and S.L.) by using the nine-star Newcas-
tle–Ottawa scale (NOS) [8]. The NOS assesses study
quality in three categories: selection, comparability and
exposure (case-control studies) or outcomes (cohort
studies). Nine points on the NOS reflects the highest
study quality. Based on quality assessment standards in
previous meta-analyses [9], a study awarded five or more
stars was defined to be of high quality in the present
meta-analysis. Any disagreement was resolved after a
discussion and reevaluation with the third author (SK).
2.5. Data synthesis and analysis
From the original study data, the OR and the 95% CI
were calculated for each study for the recurrence rate for
1630 S.-H. Shim et al. / European Journal of Cancer 50 (2014) 1628–1637
HRT users and non-users. Heterogeneity across studies
was examined using I2, which measures the percentage
of the total variation across studies [10]. Here substan-
tial heterogeneity was defined as an I2 value greater than
50% [11]. In the absence of significant heterogeneity, a
fixed effects model was used. When there was statistical
heterogeneity, a random effects model was used to esti-
mate the combined OR for randomised and observa-
tional studies. Then a subgroup analysis was
conducted for the type of study design a [randomised
control trial (RCT) or a non-randomised observational
study (NRS)], HRT (oestrogen only or oestrogen plus
progestin) and stage (stage I, stages I and II or all
stages). The subgroup analysis was planned a priori
before data were collected and analysed.
A sensitivity analysis was conducted to investigate the
robustness of conclusions by eliminating each study in
the meta-analysis one at a time to determine its effect
on the pooled OR. Publication bias was evaluated using
the fail-safe N test [12] and the Begg–Mazumdar rank
correlation test [13]. A funnel plot was built to assess
this bias by using the standard error and the diagnostic
OR [14,15]. Comprehensive Meta-Analysis version 2.0
(Biostat, Englewood, NJ, US) was used for all statistical
tests. P < 0.05 was considered significant for this meta-
analysis. Data from this meta-analysis are presented
according to the checklist based on the Meta-Analysis
of Observational Studies in Epidemiology [6].
3. Results
3.1. Literature search
Fig. 1 presents a flow diagram showing how relevant
studies were identified. A total of 11 potentially relevant
papers were found by focusing on HRT use and EC
recurrence [16–26]. Among those, one was an incom-
plete study providing no exact recurrence data [22],
Fig. 1. Flow diagram of the procedure for the literature search.
one was a letter to the editor [23], two were case series
without control groups [24,25] and one was a study from
duplication GOG data used in the study by Barakat
et al. [26]. These were excluded from further analysis.
The remaining six reported on HRT exposure and the
EC recurrence and thus were included in the meta-
analysis [16–21].
3.2. Study characteristics
A total of six papers were published between 1986
and 2006, including a total of 896 EC survivors with
HRT and 1079 without. One was a double-blind RCT
[20] and five were NRSs [16–19,21]. The studies were
from the US [16–20] and Turkey [21] and assessed to
range from six to eight on the nine-star NOS (Supple-
mental Table, online only). In the selection category,
all the included studies were awarded three stars. How-
ever, only half of the included studies [19–21] attempted
to adjust confounding factors by restrictive inclusion cri-
teria. For the comparability category, five studies [16–
19,21] were awarded one star because age and/or stage
were controlled. One study [20] was awarded two stars
as they satisfied both comparability criteria. For the
exposure assessment category, two studies [16,18] lost
stars for not adequately presenting loss to follow-up.
Table 1 shows the study characteristics in detail.
In two studies, the type of HRT was oestrogen only
[16,20]. That of HRT was combined oestrogen plus pro-
gestin in another study [21]. In the other studies, that of
HRT was a mixture of oestrogen only and combined
oestrogen plus progestin [17–19]. Five studies reported
similar stages and grade distributions for HRT users
and non-users at the beginning of the observation period
[16,18–21], although one reported a higher number of
high-risk (stage IC, grade 3) patients in the control
group (37% of non-users) [17]. Not all studies reported
the DFI before initiating HRT in the control group,
Table 1
Characteristics of studies included in the meta-analysis.
Study/design Location/ Treatment Mean Stage Grade HRT type Oestrogen HRT route Mean Mean Mean Mean Recurrence Adjusted variables
study age alone duration follow- DFI follow-
period (yr) of HRT up before up
(Mo) period HRT period
(Mo) (Mo) after
HRT
Creasman et al. [16]/ US/ HRT 57* IA-30 1–29 Conjugated oestrogen 47 (100%) Oral-7 26* 60* 15* 32* 1 Stage, LN, grade, depth of
Retrospective 1975– (n = 47) IB-17 2–13 only (0.625 or Vaginal-34 myometrial invasion, cytology
case-control study 1980 3–5 1.25 mg/d) Both-6
No HRT 62* IA- 1–87 42* 26
(n = 174) 115 2–60

S.-H. Shim et al. / European Journal of Cancer 50 (2014) 1628–1637

IB-57 3–27
Lee et al. [17]/ US/ HRT 55.7 IA-24 1, 2–44 Conjugated oestrogen 29 (66%) Oral-44 (100%) 64* 87* <12 Mo: 64*,  0 Stage
Retrospective 1975– (n = 44) IB-20 3–0 (0.625 or 1.25 mg/d) 57%
case-control study 1985 with or without 13–
progestin 24 Mo:
9%
25–
60 Mo:
21%
>60 Mo:
13%
No HRT 60.1 IA-24 1, 2–87 63* 8
(n = 99) IB-43 3–12
IC-31
Chapman et al. [18]/ US/ HRT 57.5 IA, B- 1–39 Conjugated oestrogen 29 (47%) Oral-57 39.5§ 57.1 8* 39.5 2 Preoperative HRT, radiation
Retrospective 1982– (n = 62) 54 2–17 (0.625 or 1.25 mg/d or Vagina-1 (<12 Mo:
case-control study 1994 IC-6 3–6 transdermal patch) Oral + Vaginal- 60%
II-2 with or without 4 13–
progesterone (oral, 24 Mo:
2.5 mg/d) 19.4%
>25 Mo:
21%)
No HRT 69.3 IA, B- 1–25 39.1 6
(n = 61) 43 2–26
IC-9 3–10
II-9
Suriano et al. [19]/ US/ HRT 59.7 IA-14 1–31 Conjugated equine 38 (51%) Oral-75 (100%) NR 83 <6 Mo: NR 2 Age, stage
Retrospective 1984– (n = 75) IB-44 2–29 oestrogen (oral, 43 (57%)
matched cohort 1998 IC-6 3–15 0.625 mg/d) with or 7–12 Mo:
study IIA-4 without 12 (16%)
IIB-3 medroxyprogesterone >12 Mo:
IIIA- acetate (oral, 2.5 mg/ 20 (27%)
3IIIC- d)
1
(continued on next page)

1631
 1632
Table 1 (continued)
Study/design Location/ Treatment Mean Stage Grade HRT type Oestrogen HRT route Mean Mean Mean Mean Recurrence Adjusted variables
study age alone duration follow- DFI follow-
period (yr) of HRT up before up
(Mo) period HRT period
(Mo) (Mo) after
HRT
No HRT 61.1 IA-14 1–24 69 11
(n = 75) IB-43 2–32
IC-7 3–19
IIA-5
IIB-2
IIIA-

S.-H. Shim et al. / European Journal of Cancer 50 (2014) 1628–1637

3
IIIC-1

Ayhan et al. [21]/ Turkey/ HRT 51.2 I-44 1–40 Conjugated equine 0 Oral-50 (100%) 49.1 49.1 <1–2 Mo: 49.1  0 Age, parity, systemic disease,
Prospective 1992– (n = 50) II-6 2–7 oestrogen (oral, 50 (100%) Preoperative HRT, stage,
matched cohort 2001 3–3 0.625 mg/d) plus grade, depth of myometiral
study medroxyprogesterone invasion, LVSI, tumour size,
acetate (oral, 2.5 mg/ resected LN number, mean
d) follow-up period
No HRT 54.3 I-46 1–38 53.2 1
(n = 52) II-6 2–11
3–3
Barakat et al. [20]/ US/ HRT 57* IA- 1–366 Oestrogen (oral, dose 618 Oral-618 36* 35.7* <5 Mo: 36*,  14 Age, stage, grade, histology,
Randomised 1997– (n = 618) 232 2–203 not otherwise (100%) (100%) 618 adjuvant therapy, type of
controlled trial 2003 IB- 3–45 specified) (100%) surgery
305
IC-48
IIA-
22
IIB-8
No HRT 57* IA- 1–367 35.7* 12
(n = 618) 238 2–198
IB- 3–48
302
IC-42
IIA-
17
IIB-
16
Non-randomised HRT 56.6à IA, 1,2– 143 Oral-198 44.0à 67.9à NA 45.4à 5
observational (n = 278) IB- 249
studies combined 156 (89.6%)
IC-12 3–29
II-18 (10.4%)
III-4
 S.-H. Shim et al. / European Journal of Cancer 50 (2014) 1628–1637 1633

Abbreviations: HRT, hormone replacement therapy; DFI, disease-free interval; LN, lymph node; LVSI, lymphovascular space invasion; NR, not reported; NA, not able to calculate; PET/CT, positron
but several reported matching control individuals
according to the DFI before HRT [20,21]. On average,
HRT users were more than two years younger than
non-users. The mean duration of HRT use was
38.0 months, with the mean duration of the follow-up
after the initiation of HRT at 38.3 months. The mean
duration of the follow-up was 45.7 months for HRT
users and 42.6 months for non-users.

3.3. Meta-analysis

Six studies included a total of 1975 patients with a

52

19

64

combined total of 83 recurrences (19 among 896 HRT

38.3à

users). Fig. 2A shows the ORs for the risk of EC recur-

rence for each study and all studies combined comparing
the use of HRT with no use of HRT. The pooled OR for
HRT use and the risk of EC recurrence was 0.534 (95%
NA

NA

NA

CI: 0.297–0.960). Heterogeneity was low across the stud-

ies (P = 0.081 and I2 = 49.0). Fig. 2B–D show the
51.8à

45.7à

42.6à

results of the subgroup analysis for EC survivors based

on the study design, type of HRT and stage. In the sen-
38.0à

sitivity analysis, the results based on the omission of one

study at a time and the calculation of the pooled OR for
the remaining studies show that no study had a signifi-
cant effect on the pooled OR (Supplemental Fig. 1,
Oral-816

online only).

3.4. Subgroup analysis by the type of study design

Fig. 2B shows the OR for each study and the pooled

761

OR for categories of NRS and RCT. Only one RCT was

conducted with a total of 1236 patients. Here the median
age was 57 years, the DFI was less than 20 weeks before
HRT initiation, the median duration of HRT use was
36 months and the median follow-up period after
HRT initiation was 36 months (Table 1). A total of 26
emission tomography/computed tomography; CI, confidence interval.

Assumed equal to the mean follow-up period after HRT initiation.

recurrences (14 among 618 HRT users) were observed

(84.6%)

(91.3%)

(88.5%)

(11.0%)

in this trial, and the OR for the RCT was 1.17 (95%
(8.3%)

3–119
(15.4)
3–71

3–74
1,2–

1,2–

1,2–

CI: 0.54–2.56). Five NRSs involved a total of 739

390

818

955

patients with the mean age of 59.7 years, the mean dura-
IC-47

IC-60

IC-89
II-22

II-48

II-55
III-4

III-4

III-4

tion of HRT use of 44 months, and the mean follow-up

385

784

925
IA,

IA,

IA,
IB-

IB-

IB-

period after HRT initiation of 45.4 months (Table 1). A

No HRT 61.5à

56.9à

No HRT 58.9à

combined total of 57 recurrences (five among 278 HRT

Assumed equal to the duration of HRT use.

users) were observed in these five studies. The pooled

(n = 1079)
(n = 461)

(n = 896)

OR for observational studies was 0.19 (95% CI: 0.08–

0.46; P = 0.943 and I2 = 0).
HRT

3.5. Subgroup analysis by the type of HRT

Fig. 2C shows the OR for each study and the

Weighted average.

pooled OR for categories of oestrogen only and com-

bined oestrogen plus progestin. Oestrogen only was
All combined

not associated with EC recurrence (OR: 0.35; 95%

Median.

CI: 0.06–2.10; P = 0.054 and I2 = 65.7). However, a

significant negative association (i.e. protective effect)
was observed in studies with combined oestrogen plus
*
 
à
§
1634 S.-H. Shim et al. / European Journal of Cancer 50 (2014) 1628–1637
progestin (OR: 0.23; 95% CI: 0.08–0.66; P = 0.824 and
I2 = 0).
3.6. Subgroup analysis by stage
Fig. 2D shows the OR for each study and the pooled
OR for categories of stage I, stages I and II and stages
I–III. HRT use in stage I EC survivors did not lead to
a significant increase in the risk of EC recurrence (OR:
0.12; 95% CI: 0.02–0.64; P = 0.99 and I2 = 0). HRT in
stages I and II was also not associated with EC recur-
rence (OR: 0.87; 95% CI: 0.44–1.74; P = 0.295 and
I2 = 18.1). For stages I–III, only one study was reported
(OR: 0.16; 95% CI: 0.03–0.75) [19].
3.7. Publication bias
The Begg-Mazumdar rank correlation test shows no
evidence of publication bias (P = 0.85). The fail-safe N
test indicates that for the combined two-tailed P-value
to be non-significant (P > 0.05), nine more studies with
significant findings would be required. A funnel plot
for publication bias was symmetric (Supplemental
Fig. 2, online only).
4. Discussion
According to published data, HRT use had no nega-
tive effect on recurrence and overall survival after EC
treatment. However, the individual sample size in most
studies was not large enough to sufficiently explain the
effect of HRT on EC recurrence, and no quantitative
analysis has estimated the effect of HRT on EC recur-
rence. The results of the present meta-analysis indicate
that HRT use by women with a history of EC did not
lead to a significant increase in the risk of EC recur-
rence. This pattern was also observed in the subgroup
analysis for the stage and type of HRT.
The pooled OR for HRT use and the risk of EC
recurrence were calculated using a fixed effects model
because of low heterogeneity across studies. Nonethe-
less, the results of the subgroup analysis based on the
study design show different patterns. More specifically,
estimates from the NRS indicate that HRT prevented
EC recurrence, whereas those from the RCT do not.
This result may be related to design features of those
NRSs that could have introduced selection bias, despite
the analytic technique used to remove their potential
effect. Given the caution against prescribing HRT to
women with a history of EC, those women with a more
favourable prognosis might have been selected to
receive HRT. Actually, in NRSs, HRT users were
approximately five years younger than non-users. In
addition, in NRSs, the average follow-up period for
HRT users was longer than that for non-users (67.9
versus 51.8 months). This difference implies that the
administration of HRT to women started approxi-
mately 15 months after initial treatment. Because most
EC recurrences occur within two years after the initial
diagnosis, patient candidates for recurrence in NRSs
might have been eliminated in this period [27]. We also
found that the control groups in five NRSs have a
higher-than-expected recurrence rate (52 out of 461
patients = 11%) compared to the RCT (12 out of 618
patients = 2%). As seen in Table 1, baseline risk factors
of recurrence were different between the control
groups; the control groups of NRSs have more diseases
with grade 3 (15.4% versus 7.8%) and deep myometrial
invasion (10.2% versus 6.8%) compared to the RCT.
This feature from selection bias may lead to the high
rate of recurrence in the control group of the observa-
tional studies.
Several studies added progesterone to HRT based on
findings that progesterone inhibits the stimulatory effect
of oestrogen on normal and hyperplastic endometrium
[18,19,21]. In the subgroup analysis by the type of
HRT, there was a protective effect of combined HRT
(OR: 0.23; 95% CI: 0.08–0.66) on EC recurrence, whereas
there was no such effect for oestrogen only (OR: 0.35;
95% CI: 0.06–2.10). However, this should be interpreted
with caution because there was significant heterogeneity
(I2 = 65.7) across oestrogen-only group studies. In addi-
tion, in studies with combined HRT, HRT users were
approximately five years younger than non-users (56.7
versus 61.9 years). Here it should be noted that only half
of the HRT users received progestin in all studies with
combined HRT [17–19] except for Ayhan et al. [21]. It
remains unclear whether progestin truly suppresses the
oestrogen-induced stimulation of incipient tumour cells.
Nonetheless, adding progestin to the HRT regimen
deserves a more informed discussion.
Most studies targeted patients with stage I or II EC
for HRT use [16–18,20,21]. The analysis of EC survivors
with stage I or II EC reveals that HRT did not increase
the risk of recurrence in these patients. Although one
study included stage III EC patients [19], only four
(5.3%) patients received HRT. Therefore, the evidence
provides no support for the argument that HRT can
be safely given to patients with advanced EC. Residual
malignant cells after a suboptimal surgical effort for
advanced EC may be restimulated by HRT, provoking
a recurrence.
Although most studies used the standard oral dose of
oestrogen (0.625 mg of conjugated oestrogen daily), the
dosage was up to 1.25 mg of conjugated oestrogen daily
in three studies [16–18]. However, because none of the
studies described the exact number of patients who
received this daily dose, it was not possible to conduct
a subgroup analysis for the oestrogen dose. The admin-
istration route was generally oral except for two studies
in which 34 (72.3%) [16] and 1 (1.6%) [18] received HRT
through only vagina. This suggests that, because topical
 S.-H. Shim et al. / European Journal of Cancer 50 (2014) 1628–1637 1635

oestrogen can significantly increase the level of systemic

estradiol [28], topical vaginal oestrogen was considered
to represent systemic use, as in oral administration
[18,19].
Among the included studies, the duration of HRT
ranged from 2 months [16,17] to 132 months [17]. How-
ever, the mean duration of HRT in all studies was longer
than 36 months except for Creasman et al. (26 months),
which was well beyond the interval for a majority of
recurrences. Therefore, no subgroup analysis was con-
ducted to determine whether this difference in the period
of hormone exposure would have differential effects on
EC recurrence.
The mean DFI before HRT administration varied
across studies. HRT started within 6 months after sur-
gery in only two studies [20,21], and the timing of
HRT initiation ranged from immediate postoperation
to a delay of up to 108 months [16–19]. Approximately
57–73% of the patients in three studies initiated HRT
within a year, whereas the remaining 27–43% had
HRT later [17–19]. These latter figures may represent
patients who were less likely to face EC recurrence for
the reason discussed earlier.
The RCT reported by Barakat et al. [20] overcomes
many of the shortcomings of the NRS and provides
the best available data on the effect of HRT in EC sur-
vivors. However, the study was closed prematurely after
the publication of the findings of the Women’s Health
Initiative [29]. In addition, patients primarily with early,
good prognosis tumours were enroled (60% < 60 years
old; 38% and 50% with stage IA and IB EC, respec-
tively). On the other hand, 8% had stage IC EC, and
only 7% had grade 3 EC. The insufficient enrolment of
high-risk EC patients likely led to an insufficient number
of recurrences and thus suboptimal power for the clini-
cal trial.
Fig. 2. (A) The odds ratio for the risk of EC recurrence for each study
According to preclinical data, in vitro treatment of
and all studies combined comparing hormone replacement therapy
Ishikawa endometrial cancer cells with oestrogen upreg- (HRT) use with no HRT use in a meta-analysis based on the fixed
ulates oestrogen receptor expression and augments the effects model. Heterogeneity was low across studies (P = 0.081,
growth of these cells [30,31]. However, no study has ver- I2 = 49.0). HRT use and the risk of EC recurrence in the subgroup
ified in vivo growth of residual microscopic endometrial meta-analysis by (B) study design, (C) the type of HRT and (D) the
stage of HRT. The size of each square is proportional to sample size
cancer in patients. Several laboratory studies have sug-
for each study, and the horizontal line through the square indicates the
gested that oestrogen exposure is associated with the 95% confidence interval for that study. For the pooled analysis, the
increased mitosis of normal endometrial cells, thereby diamond indicates the pooled value, and the right and left ends of the
placing them in a particular molecular configuration diamond indicate the 95% confidence interval for the analysis.
susceptible to DNA damage [32,33]. This suggests that
oestrogen can act as a tumour promoter for already and EC recurrence has yet to be clearly illustrated,
genetically altered cells. Therefore, oestrogen may not and therefore further molecular biology research is
have any further deleterious effect once the neoplastic necessary.
transformation takes place [19]. In other words, when The results of this study should be interpreted with
the source of endometrial tissue is removed through hys- caution because of several limitations. First, a majority
terectomy as part of primary cancer treatment, recur- of the studies were retrospective and a meta-analysis can-
rence then becomes a function of occult residual not address problems related to confounding factors
tumour cells, not a result of some de novo hormonal inherent in studies considered. All of the included studies
stimulation or neoplastic transformation [19]. The were awarded three stars on the NOS selection category.
mechanism underlying the relationship between HRT However, only three studies tried to control for selection
1636 S.-H. Shim et al. / European Journal of Cancer 50 (2014) 1628–1637
bias by indication or other confounding factors. The abil-
ity of this study to control for potential confounders was
limited by a small number of studies considered and a
lack of comprehensive reporting by those studies.
Although the included studies adjusted for potential con-
founding factors, residual or unknown confounding fac-
tors could not be excluded as a potential explanation for
the observed findings. Second, the overall use of HRT
varied across studies: 21.3% (47/221) in Ref. [16], 30.8%
(44/143) in Ref. [17], 50.5% (62/123) in Ref. [18] and
52.2% (130/249) in Ref. [19]. Third, it was not possible
to consider a further subgroup analysis for patient age
(less or greater than 50), dosage of HRT and tumour
grade because of a lack of data. In addition, it was not
possible to address the question of whether HRT is con-
traindicated for patients with advanced-stage EC because
only one study included stage III EC patients [19].
Finally, although no major publication bias was found,
its presence cannot be completely excluded because small
studies showing a negative effect of HRT on recurrence
are not likely to be published (the so-called ‘small-study
effect’) [34]. This potential bias can underestimate the
true risk of recurrence in HRT users. The Egger regres-
sion asymmetry test performs better than the rank corre-
lation test in evaluating the small-study effect, but it lacks
power unless the number of included trials is greater than
or equal to 10 [34]. In this regard, the results of the pres-
ent meta-analysis should be interpreted accordingly.
In sum, the results suggest that HRT use does not
increase the risk of recurrence in women with stage I
and/or II EC, although they are based mainly on retro-
spective, case, or cohort-controlled studies subject to a
variety of biases. The selection of healthier and younger
women to begin HRT and publication bias may explain
the protective effect of HRT on recurrence in EC survi-
vors. Although the results do not completely exclude
the possibility of HRT increasing the risk of recurrence,
they suggest that the magnitude of such a risk is not large.
The positive effect of HRT on the quality of life seems to
outweigh the unfounded suspicion about an increased
risk of recurrence. In this regard, additional well-designed
RCTs considering all potential confounding factors,
including cancer characteristics and treatment, different
types of HRT, the DFI before HRT initiation and the
duration of HRT use, are needed to verify the results of
this meta-analysis. In the meantime, EC survivors inter-
ested in HRT should be informed of available data and
not be exposed only to unsubstantiated bias.
Source of funding
This study received no financial support.
Conflict of interest statement
None declared.
Acknowledgement
This work was supported by Konkuk University.
Appendix A. Supplementary data
Supplementary data associated with this article can
be found, in the online version, at http://dx.doi.org/
10.1016/j.ejca.2014.03.006.
References
[1] Siegel R, Naishadham D, Jemal A. Cancer statistics, 2012. CA
Cancer J Clin 2012;62:10–29.
[2] Lim MC, Moon E-K, Shin A, et al. Incidence of cervical,
endometrial, and ovarian cancer in Korea, 1999–2010. J Gynecol
Oncol 2013;24:298–302.
[3] Darai E, Gompel A, Deval B, et al. Hormone replacement
therapy after endometrial or ovarian cancer. Gynecol Obstet
Fertil 2000;28:198–204.
[4] Michaelson-Cohen R, Beller U. Managing menopausal symptoms
after gynecological cancer. Curr Opin Oncol 2009;21:407–11.
[5] Hancke K, Foeldi M, Zahradnik HP, Gitsch G, Gilbert L,
Denschlag D. Estrogen replacement therapy after endometrial
cancer: a survey of physicians’ prescribing practice. Climacteric
2010;13:271–7.
[6] Stroup DF, Berlin JA, Morton SC, et al. Meta-analysis of
observational studies in epidemiology: a proposal for reporting.
Meta-analysis Of Observational Studies in Epidemiology
(MOOSE) group. JAMA 2000;283:2008–12.
[7] Moher D, Liberati A, Tetzlaff J, Altman DG, Group P. Preferred
reporting items for systematic reviews and meta-analyses: the
PRISMA statement. PLoS Med 2009;6:e1000097.
[8] Wells G, Shea B, O’connell D, et al. The Newcastle-Ottawa Scale
(NOS) for assessing the quality of nonrandomised studies in meta-
analyses. In: 3rd Symposium on systematic reviews: beyond the
basics; 2000. p. 3–5.
[9] Aziz O, Constantinides V, Tekkis PP, et al. Laparoscopic versus
open surgery for rectal cancer: a meta-analysis. Ann Surg Oncol
2006;13:413–24.
[10] Higgins JP, Thompson SG. Quantifying heterogeneity in a meta-
analysis. Stat Med 2002;21:1539–58.
[11] Higgins JP, Thompson SG, Deeks JJ, Altman DG. Measuring
inconsistency in meta-analyses. BMJ 2003;327:557–60.
[12] Orwin RG. A fail-safe N for effect size in meta-analysis. J Educ.
Stat. 1983;8:157.
[13] Begg CB, Mazumdar M. Operating characteristics of a rank
correlation test for publication bias. Biometrics
1994;50:1088–101.
[14] Sterne JA, Egger M. Funnel plots for detecting bias in meta-
analysis: guidelines on choice of axis. J Clin Epidemiol
2001;54:1046–55.
[15] Sterne JA, Egger M, Smith GD. Systematic reviews in health care:
investigating and dealing with publication and other biases in
meta-analysis. BMJ 2001;323:101–5.
[16] Creasman WT, Henderson D, Hinshaw W, Clarke-Pearson DL.
Estrogen replacement therapy in the patient treated for endome-
trial cancer. Obstet Gynecol 1986;67:326–30.
[17] Lee RB, Burke TW, Park RC. Estrogen replacement therapy
following treatment for stage I endometrial carcinoma. Gynecol
Oncol 1990;36:189–91.
[18] Chapman JA, DiSaia PJ, Osann K, Roth PD, Gillotte DL,
Berman ML. Estrogen replacement in surgical stage I and II
endometrial cancer survivors. Am J Obstet Gynecol
1996;175:1195–200.
 S.-H. Shim et al. / European Journal of Cancer 50 (2014) 1628–1637
[19] Suriano KA, McHale M, McLaren CE, Li KT, Re A, DiSaia PJ.
Estrogen replacement therapy in endometrial cancer patients: a
matched control study. Obstet Gynecol 2001;97:555–60.
[20] Barakat RR, Bundy BN, Spirtos NM, Bell J, Mannel RS,
Gynecologic Oncology Group S. Randomized double-blind trial
of estrogen replacement therapy versus placebo in stage I or II
endometrial cancer: a Gynecologic Oncology Group Study. J Clin
Oncol 2006;24:587–92.
[21] Ayhan A, Taskiran C, Simsek S, Sever A. Does immediate
hormone replacement therapy affect the oncologic outcome in
endometrial cancer survivors? Int J Gynecol Cancer
2006;16:805–8.
[22] Lauritzen C. Östrogensubstitution in der Postmenopause vor und
nach behandeltem Genital- und Mammakarzinom: Menopause:
Hormonsubstitution heute. Basel: Aesopus Verlag; 1993, p. 76–
80.
[23] Bryant GW. Administration of estrogens to patients with a
previous diagnosis of endometrial adenocarcinoma. South Med J
1990;83:725–6.
[24] Baker DP. Estrogen-replacement therapy in patients with previ-
ous endometrial carcinoma. Compr Ther 1990;16:28–35.
[25] Gitsch G, Hanzal E, Jensen D, Hacker NF. Endometrial cancer in
premenopausal women 45 years and younger. Obstet Gynecol
1995;85:504–8.
[26] Maxwell GL, Tian C, Risinger JI, Hamilton CA, Barakat RR,
Gynecologic Oncology Group S. Racial disparities in recurrence
among patients with early-stage endometrial cancer: is recurrence
increased in black patients who receive estrogen replacement
therapy? Cancer 2008;113:1431–7.
1637
[27] Levgur M. Estrogen and combined hormone therapy for women
after genital malignancies: a review. J Reprod Med
2004;49:837–48.
[28] Kendall A, Dowsett M, Folkerd E, Smith I. Caution: vaginal
estradiol appears to be contraindicated in postmenopausal
women on adjuvant aromatase inhibitors. Ann Oncol
2006;17:584–7.
[29] Rossouw JE, Anderson GL, Prentice RL, et al. Risks and benefits
of estrogen plus progestin in healthy postmenopausal women:
principal results from the Women’s Health Initiative randomized
controlled trial. JAMA 2002;288:321–33.
[30] Holinka CF, Hata H, Kuramoto H, Gurpide E. Responses to
estradiol in a human endometrial adenocarcinoma cell line
(Ishikawa). J Steroid Biochem 1986;24:85–9.
[31] Farnell YZ, Ing NH. The effects of estradiol and selective estrogen
receptor modulators on gene expression and messenger RNA
stability in immortalized sheep endometrial stromal cells and
human endometrial adenocarcinoma cells. J Steroid Biochem Mol
Biol 2003;84:453–61.
[32] Samsioe G. The endometrium: effects of estrogen and estrogen-
progestogen replacement therapy. Int J Fertil Menopausal Stud
1994;39(Suppl. 2):84–92.
[33] Li SF, Shiozawa T, Nakayama K, Nikaido T, Fujii S. Stepwise
abnormality of sex steroid hormone receptors, tumor suppressor
gene products (p53 and Rb), and cyclin E in uterine endometrioid
carcinoma. Cancer 1996;77:321–9.
[34] Sterne JA, Gavaghan D, Egger M. Publication and related bias in
meta-analysis: power of statistical tests and prevalence in the
literature. J Clin Epidemiol 2000;53:1119–29.
 The n e w e ng l a n d j o u r na l of m e dic i n e

Review Article

Dan L. Longo, M.D., Editor

Hematuria in Adults
Julie R. Ingelfinger, M.D.​​

V
isible (macroscopic) hematuria, documented since ancient This article was updated on July 13, 2021,
times,1,2 is striking, particularly when there is no prior event, such as trau- at NEJM.org.

ma, dysuria due to cystitis, or flank pain with passage of a kidney stone, to N Engl J Med 2021;385:153-63.
provide a clear explanation. In contrast, nonvisible (microscopic) hematuria, or DOI: 10.1056/NEJMra1604481
Copyright © 2021 Massachusetts Medical Society.
microhematuria, may go undetected for years.2 Many patients are found to have
microhematuria when a urinalysis is performed for other reasons.
The reported prevalence of microhematuria varies greatly, from a small percent-
age of patients at screening to more than 40% in some urology clinics, probably
owing to referral bias regarding patients and their signs and symptoms.3-5 The
American Urological Association (AUA) guidelines note a prevalence ranging from
2.1 to 31.4%.5 Thus, a discussion of epidemiology per se is less important than a
consideration of individual patient circumstances and geographic region. In certain
parts of the world — for example, in northern Africa, where Schistosoma haematobium
is endemic — microhematuria is common, reflecting prevalent bladder infestation.
In addition, the causes of hematuria differ between men and women, and the
evaluation, accordingly, should reflect that difference.6-8 Furthermore, the implica-
tions of hematuria vary greatly, depending on the underlying cause. Given the
association of both visible hematuria and microhematuria with bladder and kidney
cancer, the focus of evaluation has long been to rule out cancer. Cancer is more
likely to be identified in men than in women when they are evaluated for micro-
hematuria. Worldwide, estimates of the incidence of kidney cancer are 6.0 cases
per 100,000 person-years for men and 3.1 cases per 100,000 person-years for
women, and estimates of the incidence of bladder cancer are 9.0 cases per 100,000
person-years for men and 2.2 cases per 100,000 person-years for women, accord-
ing to GLOBOCAN, a registry of data on the global incidence of cancer,9 a result
also found in a 2020 meta-analysis.10 However, evaluation in women is often de-
layed, which may contribute to worse outcomes of bladder cancer among women.
Beyond cancer, hematuria is associated with protean symptoms and causes.
This review focuses on major causes, the evaluation, and the implications of he-
maturia. Since therapy depends on the cause, treatment is not addressed here.

De tec t ion of Hem at ur i a

Urine hue normally ranges from nearly colorless, when dilute, to dark amber,
when concentrated. Red, pink, “rusty,” or brown urine may suggest hematuria but
may be due to substances besides blood. Available dipstick tests detect red cells,
as well as hemoglobin and myoglobin, necessitating microscopic examination of
the urine if a dipstick test is heme-positive, followed by other tests to confirm
hematuria and, potentially, to determine its cause.11,12
Commonly used dipstick tests incorporate a benzidine compound reduced
with a buffered organic peroxide (3,3′,5,5′-tetramethylbenzidine and diisopropyl-
benzene dihydroperoxide). When the test strip is dipped in urine that includes an
oxidizing substrate, a color reaction ensues; for example, the pseudoperoxidase

n engl j med 385;2  nejm.org  July 8, 2021 153

The New England Journal of Medicine
Downloaded from nejm.org by Salvador Macias-Díaz on August 6, 2021. For personal use only. No other uses without permission.
Copyright © 2021 Massachusetts Medical Society. All rights reserved.
 The n e w e ng l a n d j o u r na l of m e dic i n e

Table 1. Selected Nonheme Causes of Pigmenturia, as Compared with Red Cells or Free Hemoglobin.*

Variable Red Cells Hemoglobin Myoglobin Porphyria Bile Pigments Alkaptonuria

Urine color Red to rusty Pink to red Rusty Turns black, Brown Turns dark in
brown, or red in sunlight
sunlight
Heme test Positive Positive Positive Negative Negative Negative
Usual microscopic Red cells, No cells No red cells; Normal Normal Normal
findings casts casts may be
present
Plasma Normal Pink Normal Normal Icteric Normal

* Shown are selected endogenous causes of pigmenturia. A heme test for exogenous causes, including beets, rhubarb,
azo dye, sulfonamides, and phenolphthalein, is negative.

activity of hemoglobin will oxidize the benzi-
dine compound, turning the dipstick blue. Free
hemoglobin and myoglobin also react with the
test strip, as would be expected because of their
molecular structure.13
Certain substances may lead to false positive
dipstick reactions: sodium hypochlorite, peroxi-
dases from vegetable or bacterial sources, and se-
men. In contrast, very high ascorbic acid levels in
urine may produce false negative results. The list of
substances that can be confused with hematuria
is lengthy (see the partial listing in Table 1). The
absence of red cells on microscopic examination of
a “heme-positive” urine specimen suggests ei-
ther that the red cells have all lysed or that iso-
lated hemoglobinuria or myoglobinuria may be
present. Hemoglobinuria is dipstick-positive, but
if it is the sole cause of the positive test, red cells
will not be present in the urinary sediment. In
some instances, both hematuria and hemoglobin-
uria are present and red cells will be identified.
Distinguishing blood from myoglobin in the
urine (which may signify rhabdomyolysis and
may or may not be suspected, given a patient’s
history) is clinically important.14 Frequently, cen-
trifugation is helpful, since red cells will sedi-
ment, leaving a clear supernatant, whereas myo-
globin will not. However, free hemoglobin also
will not precipitate in a routine laboratory cen-
trifuge. In all, the absence of red cells in the
sediment of a “heme-positive” specimen suggests
that either isolated hemoglobinuria or myoglo-
binuria may be present. Myoglobinuria can then
be identified biochemically through ammonium
sulfate precipitation, as well as through electro-
phoretic and immunologic tests.
Thus, after a positive dipstick test, micro-
scopic examination of the urinary sediment is
essential.15 True microhematuria is most often
defined as more than 2 or 3 red cells per high-
power field, and this finding should be con-
firmed on two or three separate urinalyses.16,17
Some experts suggest that if even a single urinaly-
sis is positive, a patient should have follow-up
urinalyses for at least a year, so as to avoid miss-
ing an intermittent source of microhematuria
that could signify a clinically important problem.
Careful examination of the urinary sediment
is central to differentiating glomerular from
other forms of hematuria.15 Although micro-
graphs of red cells crossing the glomerular base-
ment membranes are as rare as hen’s teeth, red
cells that have traversed the glomerular base-
ment membrane into the glomerular filtrate are
much the worse for wear and appear with blebs
and other irregularities.18 In contrast, lower uri-
nary tract hematuria is characterized by normal
red cells (Fig. 1). If dysmorphic red cells are
found, the diagnostic evaluation should first
focus on the possibility of a glomerulopathy (see
the pragmatic algorithm in Fig. 2), particularly
if casts, especially red-cell casts, are present.
Isolated hematuria is less likely to be associ-
ated with a glomerulopathy than is hematuria
with albuminuria or decreased kidney function,
though IgA nephropathy and familial nephropa-
thies may not always be characterized by albu-
minuria.15 Bacteria in the unspun urine and in
the sediment suggests urinary tract infection.
Crystals may suggest nephrolithiasis.
If examination of the urinary sediment shows
normal-appearing red cells, then imaging and
urologic referral should be considered. Most
urologists will focus on imaging results and cys-
toscopy. In patients with risk factors for cancer,
many urologists and guidelines would suggest

154 n engl j med 385;2  nejm.org  July 8, 2021

The New England Journal of Medicine

Downloaded from nejm.org by Salvador Macias-Díaz on August 6, 2021. For personal use only. No other uses without permission.
Copyright © 2021 Massachusetts Medical Society. All rights reserved.
 Hematuria in Adults

A Nondysmorphic Red Cells B Dysmorphic Red Cells

Figure 1. Nondysmorphic and Dysmorphic Red Cells in Freshly Voided Urine.

Panel A shows nondysmorphic red cells, and Panel B shows dysmorphic red cells, which have characteristic blebs.
Micrographs are courtesy of the estate of the late Michael Linshaw, M.D.

either urethrocystoscopy and computed tomo-
graphic (CT) urography19 or ultrasonography and
cystoscopy.5
Glomerular hematuria in the absence of pro-
teinuria or kidney dysfunction has largely been
considered to be an innocuous finding or per-
haps a marker of changes in the glomerular
filtration apparatus or of inflammation, yet some
data suggest that isolated glomerular hematuria
may not always be innocuous.20 Over time, even
in conditions considered to have a good progno-
sis, microhematuria has been associated with
progressive kidney failure, and macrohematuria
has been associated with acute tubular dysfunc-
tion and acute kidney injury.20 For instance, epi-
sodes of gross hematuria in persons with IgA
nephropathy may be accompanied by diminution
of the glomerular filtration rate or frank acute
kidney injury, but the patient generally recov-
ers.21 Vivante et al.22 reported that microhematu-
ria among army recruits was associated with a
risk of end-stage kidney disease many years later.
Furthermore, kidney donors who have had micro-
hematuria have been reported to have an increased
risk of kidney dysfunction after donation.23
The mechanism of kidney injury may be due
to the effects of tubular inflammation and oxi-
dative stress induced by breakdown products of
red cells.24,25 Whether glomerular hematuria it-
self induces progression or is a surrogate for
inflammation is unclear. Moreno et al.24,25 postu-
late that red-cell casts induce damage to tubular
cells through obstruction, that hemoglobin and
heme are directly toxic to the tubules, and that
renal tubular cells subsequently engage in erythro-
phagocytosis, which has been linked to further
nephrotoxicity induced by iron and hemoglobin.26
C onfir med V isibl e Hem at ur i a
Visible hematuria often has an obvious explana-
tion, which is congruent with the patient’s his-
tory, signs, and symptoms (e.g., the passage of
a kidney stone, acute hemorrhagic cystitis, or a
sickle-cell crisis). However, many other condi-
tions can cause visible hematuria (Fig. 3). Thus,
the evaluation of gross hematuria should be
focused on the basis of the patient’s symptoms
and concomitant clinical and laboratory find-
ings.27 In adults over the age of 40 years (some
sources suggest a cutoff age of 35 years), an
unexplained episode of visible hematuria may
signify bladder or upper urinary tract cancer,
and most experts suggest urologic referral
along with imaging and cystoscopy to rule out
cancer.28-31
Medical History and Physical Examination
A thorough history is critical in focusing the
diagnostic evaluation of a patient with gross
hematuria.27-31 Has the patient had trauma, and
might there be a renal contusion, vascular com-
promise, or infarction? A history suggestive of
urinary tract infection or renal colic calls for an
initial evaluation to confirm or rule out an infec-
tion or a kidney stone. If the patient has just
undergone surgery, is there a complication, even
if the surgical procedure did not directly involve

n engl j med 385;2 nejm.org July 8, 2021 155

The New England Journal of Medicine
Downloaded from nejm.org by Salvador Macias-Díaz on August 6, 2021. For personal use only. No other uses without permission.
Copyright © 2021 Massachusetts Medical Society. All rights reserved.
 The n e w e ng l a n d j o u r na l of m e dic i n e

Hematuria

If negative, pseudohematuria
Dipstick test administered
diagnosis

If positive, proceed with urinalysis and

basic metabolic panel

Dysmorphic red cells, casts Nondysmorphic red cells

Glomerular bleeding Nonglomerular bleeding

Evaluate blood and urine for Urine culture and sensitivity

UTI symptoms Yes
possible glomerulopathies test; treatment for UTI
Consider kidney biopsy
No

Nephrolithiasis symptoms Yes CT and stone evaluation

No

Diagnosis
Evaluate cancer risk No
confirmed

Yes

Moderate-to-high risk Low risk

Cystoscopy and CT (if high risk)

Findings still positive Repeat urinalysis in 4–6 wk Treatment
Ultrasonography and cystoscopy
(if moderate risk)

No diagnosis; consider evaluation Resolved; perform follow-up

Diagnosis and treatment
for glomerular bleeding urinalysis for 1–2 yr

Figure 2. Algorithm Incorporating Risk Assessment in the Evaluation of Hematuria.

The algorithm is based on the current American Urological Association guidelines. UTI denotes urinary tract infection.

the kidneys? Is the patient receiving anticoagu-
lants? Has the patient engaged in strenuous ac-
tivity, such as a long-distance run, which may
occasionally be associated with visible hematu-
ria? Has the patient’s voiding pattern changed,
which could be suggestive of prostatic hypertro-
phy or obstruction? Does the patient have flank
pain or suprapubic pain?
The physical examination further focuses the
workup.27-31 Are vital signs stable, and is there
any sign of vascular instability that would sug-
gest bleeding or some other acute, potentially
life-threatening problem? Costovertebral angle
or suprapubic tenderness may suggest a kidney
stone or a urinary tract infection. In male pa-
tients, rectal examination to palpate the prostate
may help assess the possibility of prostatitis or
prostate cancer. In female patients, it is impor-

156 n engl j med 385;2  nejm.org  July 8, 2021

The New England Journal of Medicine

Downloaded from nejm.org by Salvador Macias-Díaz on August 6, 2021. For personal use only. No other uses without permission.
Copyright © 2021 Massachusetts Medical Society. All rights reserved.
 Hematuria in Adults

Glomerular Hematuria Nonglomerular Hematuria

IgA nephropathy Polycystic kidney disease
Alport’s syndrome Sickle-cell disease and papillary necrosis
Thin basement membrane disease Nephrolithiasis
C3 nephropathy Tumors
Postinfectious glomerulonephritis Leukemia
Trauma
Infection
Strenuous exercise
Kidney

Ureteral Hematuria Ureter

Stones Vesicular Hematuria
Infection
UTI
Trauma
Stones
Tumor
Infections
Bladder Irritation
Tumor
Urethra Foreign body

Sexual Dimorphism Urethral Hematuria

Prostatic bleeding Trauma
Vaginal bleeding Infection
Endometriosis of the urinary tract

Figure 3. Hematuria According to Location.

Categories of hematuria are shown according to the location in the urinary tract.

tant to be sure that the bleeding is from the a mean age of 65.7 years) with visible or micro-
urinary tract, not the reproductive tract. Signs of scopic hematuria, 10% of the cohort had can-
systemic disease should also be sought. cers, mainly bladder cancer (in 8.0% of all pa-
tients evaluated); cancer was more common
Imaging among those with visible hematuria.33 Renal
Imaging is generally considered necessary in parenchymal cancer was seen in 1.0% of the
adults with gross hematuria, given the possibil- patients, upper tract transitional-cell cancer in
ity of cancer, unless there is an obvious cause 0.7%, and prostate cancer in 0.3%. In this study,
(e.g., hemorrhagic cystitis). For many patients, cancer was most likely to be diagnosed in the
ultrasonography plus cystoscopy suffices. CT male patients, especially older men and men
urography is recommended by some sources with a history of smoking. The results of most
but is more costly than ultrasonography and trials, taken together, have suggested that any
usually includes the administration of contrast patient with visible blood in the urine should
material.27-31 The use of magnetic resonance im- undergo urologic investigation.5,33
aging is not generally recommended. It is said
that dysuria is present in approximately 80% of C onfir med Microhem at ur i a
patients with bladder cancer and that dysuria
doubles the likelihood of finding a bladder The most common causes of microhematuria
cancer.32 are nonmalignant5: glomerulopathies (e.g., IgA
In DETECT 1, a prospective, observational nephropathy or thin glomerular basement mem-
study involving 3556 adults (59% were men, with brane disease) and inflammatory conditions of

n engl j med 385;2 nejm.org July 8, 2021 157

The New England Journal of Medicine
Downloaded from nejm.org by Salvador Macias-Díaz on August 6, 2021. For personal use only. No other uses without permission.
Copyright © 2021 Massachusetts Medical Society. All rights reserved.
 The n e w e ng l a n d j o u r na l
the urethra, prostate, and bladder, as well as
renal calculi and benign prostatic hypertrophy.
The causes of microhematuria are many and
disparate, as summarized in Figure 3.
Most sources state that confirmed micro­
hematuria requires evaluation, since overlooking
a bladder cancer or kidney cancer would be po-
tentially fatal, but that viewpoint has been chal-
lenged in recent years.34,35 Urinary tract cancer is
rare in women, and recent data suggest that the
likelihood of cancer is less than 5% in a woman
under 50 years of age who has never smoked and
who has asymptomatic microhematuria with
fewer than 25 red cells per high-power field.8
Even for men, some guidelines suggest individ­
ualized evaluation. Such recommendations note
that evaluation of microhematuria is time-con-
suming and costly and may cause more harm
than good. In one study, up to 57% of patients
referred to a urology service for microhematuria
turned out to have pseudohematuria.12 Given all
these considerations, the decision to evaluate a
patient may best be based on a mutual decision
arrived at through discussion between patient
and clinician.
There is debate about whether evaluation of
microhematuria should include invasive testing
such as cystoscopy, even though the annual inci-
dence of urinary tract cancers is substantial. For
example, in 2009, an estimated 70,980 bladder
cancers and 57,760 renal cancers were diagnosed
in the United States. In 2020, the estimated
numbers of new bladder and renal cancer cases
were 81,400 and 73,750, respectively, according
to the American Cancer Society.36
Although, as noted above, most patients with
urinary tract cancer present with hematuria,
asymptomatic microhematuria has a consistently
poor performance as a trigger for diagnostic
investigations. For example, Jung et al.37 exam-
ined the medical records of 156,692 patients
with microhematuria; the incidence of urinary
tract cancer was 0.7% over a period of 3 years.37
An age of more than 40 years, the presence of
more than 25 red cells per high-power field, and
male sex increase the risk of cancer. Additional
risk factors include occupational exposure to
carcinogens, particularly aromatic amines and
hydrocarbons (involved in chimney maintenance
and dye work); analgesic abuse; pelvic irradia-
tion; chemotherapy courses involving alkylating
158 n engl j med 385;2  nejm.org  July 8, 2021
The New England Journal of Medicine
Downloaded from nejm.org by Salvador Macias-Díaz on August 6, 2021. For personal use only. No other uses without permission.
Copyright © 2021 Massachusetts Medical Society. All rights reserved.
of m e dic i n e
agents (e.g., cyclophosphamide); and prolonged
exposure to foreign bodies (e.g., catheters).
In the above-mentioned DETECT 1 trial,33
10% of patients had malignant tumors, the ma-
jority of which were bladder cancers. In another
study, 234 male patients who had had a negative
evaluation for microhematuria were followed for
14 years; urinary tract cancer developed in 2 of
these patients (0.9%).38
Despite the low frequency of urinary tract
cancer, expert opinion favors evaluation.5,39-42
Nielsen and Qaseem, for example, recommend
that if a patient has persistent microhematuria,
evaluation should proceed, even if that patient is
receiving antiplatelet or anticoagulant medica-
tion.39 In contrast, the 2012 AUA guidelines,
based on a review of 191 evidence-based reports
and on expert opinion for areas and concepts
lacking such data, recommended evaluation for
all patients 35 years of age or older in whom
microhematuria does not have an obvious, identi-
fied benign cause, according to the history, physi-
cal examination, and urinalysis.42 That workup,
according to the AUA, should include cystoscopy
and upper urinary tract imaging with CT.
Medical History
As with visible hematuria, it is crucial to obtain
a good history in order to focus the evaluation of
microhematuria.5,15 Is the patient symptomatic,
and if so, what are those symptoms?
The evaluation of microhematuria is focused
on ruling out diagnoses that are very important
not to miss — mainly, urinary tract cancers.
Microhematuria is detected by dipstick testing
or examination of the urinary sediment. It is
important to ascertain that the patient does not
have a urinary tract infection, is not menstruat-
ing, and has not just exercised, engaged in sex-
ual activity, or had exposure to instrumentation
(e.g., through cystoscopy).
A careful examination of the urinary sediment
is key. It is best to obtain a clean-catch mid-
stream urine sample — ideally, the first void of
the day — and examine it within 1 to 2 hours.
The first void of the day has higher osmolality
and lower pH (both of which are helpful in pre-
serving the morphologic features of formed
sediment elements) than a sample obtained later
in the day.5 There are standard ways to perform
a urinalysis, generally spinning the sample at
 Hematuria in Adults
4000 rpm for 5 minutes after obtaining a dip-
stick reading.11,15 The normal number of red
cells in the urinary sediment is generally agreed
to be fewer than 2 or 3 per high-power field.5,11,15
The presence of red cells in one urinalysis
should be confirmed in one or two more uri-
nalyses. If microhematuria is not confirmed in a
second or third urinalysis, many sources suggest
repeated testing over a period of a year or so. If
microhematuria is confirmed, then it is impor-
tant next to consider whether the findings sug-
gest an upper or a lower urinary tract origin.
The morphologic characteristics of red cells,43-47
as well as a review of dipstick results, the
formed elements in the urinary sediment, and
any other urinary abnormalities detected, may
suggest whether upper or lower urinary tract
bleeding is more likely, with dysmorphic red
cells in the sediment suggesting an upper uri-
nary tract (glomerular) source of bleeding.5 Ac-
cording to some sources, a finding that more
than 25% of urinary red cells are dysmorphic
points to glomerular disease.47 Urinary red-cell
mean corpuscular volume has also been used to
help differentiate between upper and lower uri-
nary tract disease.48 The sensitivity of both dys-
morphic and small red-cell findings has been
challenged. The presence of proteinuria should
impel the clinician to launch an investigation for
a major nephropathy.15 In sum, isolated hema-
turia with dysmorphic red cells or red cells with
a decreased mean corpuscular volume, with or
without casts, should prompt an evaluation for
glomerulopathy as well. The most commonly
associated glomerulopathies are IgA nephropa-
thy and Alport’s syndrome (familial nephritis).
But the list of potential glomerulopathies is
long, and an evaluation of kidney function,
complement status, and autoimmune markers is
indicated, depending on the individual patient.
Even if the blood in the urine appears to be due
to lower urinary tract disease, kidney function
and the estimated glomerular filtration rate
should be checked. The specific studies will de-
pend on the possibilities that seem likely (Fig. 2).
Cytologic Analysis
Urinary cytologic analysis49-52 has long been rec-
ommended as a possible adjunct to cystoscopy
and can serve as a way to find evidence of small
carcinomas that are overlooked on cystoscopy.
n engl j med 385;2  nejm.org  July 8, 2021
The New England Journal of Medicine
Downloaded from nejm.org by Salvador Macias-Díaz on August 6, 2021. For personal use only. No other uses without permission.
Copyright © 2021 Massachusetts Medical Society. All rights reserved.
In a study by Mishriki et al.,49 in which 2278
patients underwent urinary cytologic analysis,
only 2 were determined to have cancer on the
basis of the cytologic findings. Cytologic analy-
sis in a study by Hofland et al.50 showed cancer
in 4 of 1000 urine samples, and in 2 of those
4 samples, cancer was detected only on the basis
of cytologic findings. Current practice is to
limit the use of cytologic analysis to cases of
gross hematuria or symptomatic hematuria.5,52
Imaging
Adults with established microhematuria should
undergo imaging, and as with visible hematuria,
the most cost-effective combination of studies
has been found to be ultrasonography plus cys-
toscopy.5 Using decision analysis, Halpern et al.34
looked at simulated cancer detection rates in
persons with asymptomatic microhematuria and
associated costs in terms of the incremental
costs of cancer detection. They examined four
imaging approaches: CT urography alone, cys-
toscopy alone, CT urography plus cystoscopy,
and ultrasonography plus cystoscopy. Ultraso-
nography plus cystoscopy detected the most
cancers per incremental cost of cancer detection,
and exchanging CT urography for ultrasonogra-
phy would have detected only one additional
cancer.
Cystoscopy
For patients who have gross hematuria without
an obvious cause, cystoscopy is generally recom-
mended in order to rule out cancer, most often
bladder cancer. The 2012 AUA guidelines42 recom-
mended cystoscopy for all patients with micro-
hematuria who are older than 35 years of age,
but the 2020 AUA guidelines5 added risk levels
and recommended immediate cystoscopy only
for patients at increased risk.
Molecular Biomarkers
The possibility that molecular markers might
focus the workup of hematuria is attractive.53
Biomarkers for various urinary tract cancers,54,55
for glomerular disease,56,57 and for genetic dis-
eases58 are of great interest. Bladder cancer is
heterogeneous, and if there were prospectively
validated markers, they might serve as important
adjuncts in clinical decision making.53 Whereas
90 to 95% of bladder cancers are urothelial
159
 The n e w e ng l a n d j o u r na l
(transitional-cell) carcinomas, only 3 to 7% are
squamous-cell carcinomas (usually in the renal
pelvis and ureters), and less than 3% are adeno-
carcinomas (most often in the trigonal region).
The relevant markers for distinct types of can-
cer appear to differ. Sathianathen et al.54 car-
ried out a meta-analysis in which 17 studies
met the specified criteria. The authors found
substantial heterogeneity among the studies.
Evaluating several biomarkers in combination
(NMP22, UroVysion, and uCyt+) or alone (blad-
der tumor–associated antigen [BTA], NMP22,
uCyt+, Cxbladder, and AssureMDx), they calcu-
lated that the biomarkers had high sensitivity
but insufficient specificity to eliminate the need
for cystoscopy. Wilson et al.55 noted that BTA
and NMP22, which are approved by the Food
and Drug Administration for use together with
cystoscopy for the diagnosis of bladder cancer,
do not work well if used alone, since hematuria
in the absence of cancer, as well as infections
and stones, is associated with false positive test
results. To date, despite the interest in bio-
markers, the types of prospective studies need-
ed to render biomarkers clinically useful are
lacking.
Nucleic acids detected in plasma, including
DNA (genomic, as well as mitochondrial and
viral), RNA, and microRNA, have been recog-
nized for some time as potential biomarkers.
Circulating tumor DNAs (ctDNAs) are used as
biomarkers in multiple areas of oncology. The
detection of ctDNA in muscle-invasive bladder
cancer is being piloted for use in diagnosis and
for monitoring during treatment and follow-up
care.59 Cell-free DNA has been used similarly in
renal-cell carcinoma.60
Microhematuria with No Identified Cause
The specific cause of hematuria may be elusive.
In some series, more than half of the patients
have microhematuria with no definable cause.
Bolenz et al.32 suggest a “risk-adapted investiga-
tion,” connoting a personalized and risk–benefit
approach to evaluation that is based on the view
that every patient with hematuria should be
evaluated, but not every possible test should be
used. Furthermore, if no cause is identified, at a
certain point, it makes sense to live with uncer-
tainty. Follow-up of the patient with microhema-
turia depends on the history and evaluation. In
160 n engl j med 385;2  nejm.org  July 8, 2021
The New England Journal of Medicine
Downloaded from nejm.org by Salvador Macias-Díaz on August 6, 2021. For personal use only. No other uses without permission.
Copyright © 2021 Massachusetts Medical Society. All rights reserved.
of m e dic i n e
the absence of a specific diagnosis, monitoring
at intervals is considered to be important — ini-
tially after an interval of a few months and then
once or twice yearly.
Guidance for Screening and Evaluation
Given the wide differential diagnosis and the
disparate reports about the import of micro­
hematuria, current guidance varies. The Ameri-
can College of Physicians has stated (in 2016)39
that asymptomatic adults as a group should not
be screened by means of urinalysis for cancer
detection and that for those who are nonetheless
screened and found to have a positive dipstick
test, the presence of blood should be confirmed
by microscopic urinalysis. The U.S. Preventive
Services Task Force does not recommend screen-
ing for microhematuria, citing inadequate evi-
dence of benefit versus harm.61 Others have also
considered the issue of harms and benefits.62,63
Table 2. Risk Stratification for Bladder Cancer.
Low risk (all criteria must be met)
<40 Yr of age for men, <50 yr of age for women
Never smoked or <10 pack-yr of smoking
3–10 Red cells per high-power field on one urinalysis
No risk factors for urothelial cancer
Intermediate risk (one criterion raises the risk to inter-
mediate)
40–59 Yr of age for men and women
10–30 Pack-yr of smoking
11–25 Red cells per high-power field on repeat urinalysis
Additional risk factors for urothelial cancer
High risk (one criterion raises the risk to high)
≥60 Yr of age for men and women
>30 Pack-yr of smoking
>25 Red cells per high-power field on single urinalysis
History of gross hematuria
Additional risk factors for urothelial cancer
Irritative lower urinary tract symptoms
Previous pelvic radiation therapy
Previous chemotherapy with cyclophosphamide
or ifosfamide
Family history of urothelial cancer or Lynch syndrome
Occupational exposure to benzene or aromatic amines
Chronic indwelling foreign body in the urinary tract
 Hematuria in Adults

In a 2018 review of guidelines and guidance19
on microhematuria, the lack of consensus on
several points was clear. However, most of the
guidelines state that the presence of 3 or more
red cells in one, two, or three urinalyses justifies
the diagnosis of microhematuria, whereas the
Japanese Urological Association requires 5 or
more red cells. Some other guidelines require a
positive dipstick test.
Tan et al. recently developed a Haematuria
Cancer Risk Score (HCRS)64 to aid clinicians in
considering evaluation. Using the DETECT 1
cohort, the investigators created a test cohort of
3539 patients in 40 hospitals in the United King-
dom (ClinicalTrials.gov number, NCT02676180)
and a Swiss validation cohort of 656 persons.
Use of the HCRS score, which is based on the
patient’s age and sex, type of hematuria, and
smoking history, appeared to detect more can-
cers than were found by following the United
Kingdom’s National Institute for Health and
Care Excellence (NICE) guidelines65 or the AUA
guidelines.5 There is also another risk score,
which is not in general use.66
The AUA guidelines were updated in 2020,5
through a panel created 2 years earlier by AUA
Education and Research (AUAER) and the Soci-
ety of Urodynamics, Female Pelvic Medicine and
Urogenital Reconstruction. The assessment now
favored is based on risk categories (Table 2). The
current guidelines comprise recommendations
that are intended to control costs and limit risks
associated with “overevaluation” of persons with
a low likelihood of having urinary tract cancer
while avoiding “underevaluation” of persons at
high risk. Not surprisingly, the guidelines sug-
gest a shared decision-making plan between
patient and clinician. Many other guidelines and
guidance documents67-72 are available, and they
agree about the importance of not missing a
urinary tract cancer.
Sum m a r y
Hematuria is an important sign that may con-
note serious disease, yet on many occasions no
specific cause is identified. A high index of sus-
picion and good communication between clini-
cian and patient can result in a reasoned and
reasonable approach to evaluation and therapy.
No potential conflict of interest relevant to this article was
reported.
Disclosure forms provided by the author are available with the
full text of this article at NEJM.org.

References
1. Shokeir AA, Hussein MI. The urology
of Pharaonic Egypt. BJU Int 1999;​84:​755-
61.
2. Armstrong JA. Urinalysis in Western
culture: a brief history. Kidney Int 2007;​
71:​384-7.
3. Mohr DN, Offord KP, Owen RA,
Melton LJ III. Asymptomatic microhema-
turia and urologic disease. A population-
based study. JAMA 1986;​256:​224-9.
4. Woolhandler S, Pels RJ, Bor DH, Him-
melstein DU, Lawrence RS. Dipstick uri-
nalysis screening of asymptomatic adults
for urinary tract disorders. I. Hematuria
and proteinuria. JAMA 1989;​262:​1214-9.
5. Barocas DA, Boorjian SA, Alvarez RD,
et al. Microhematuria: AUA/SUFU Guide-
line. J Urol 2020;​204:​778-86.
6. Ark JT, Alvarez JR, Koyama T, et al.
Variation in the diagnostic evaluation
among persons with hematuria: influ-
ence of gender, race and risk factors for
bladder cancer. J Urol 2017;​198:​1033-8.
7. Cohn JA, Vekhter B, Lyttle C, Stein-
berg GD, Large MC. Sex disparities in di-
agnosis of bladder cancer after initial
presentation with hematuria: a nation-
wide claims-based investigation. Cancer
2014;​120:​555-61.
8. Rabinowitz DE, Buford K, Wood AM,
et al. Incidence of significant findings of
microhematuria workup in women-what
guidelines work best? Urology 2021;​151:​
19-23.
9. Ferlay J, Soerjomataram I, Dikshit R,
et al. Cancer incidence and mortality
worldwide: sources, methods and major
patterns in GLOBOCAN 2012. Int J Can-
cer 2015;​136(5):​E359-E386.
10. Jubber I, Shariat SF, Conroy S, et al.
Non-visible haematuria for the detection
of bladder, upper tract, and kidney can-
cer: an updated systematic review and
meta-analysis. Eur Urol 2020;​77:​583-98.
11. Simerville JA, Maxted WC, Pahira JJ.
Urinalysis: a comprehensive review. Am
Fam Physician 2005;​71:​1153-62.
12. Rao PK, Gao T, Pohl M, Jones JS. Dip-
stick pseudohematuria: unnecessary con-
sultation and evaluation. J Urol 2010;​183:​
560-4.
13. Litwin MS, Graham SD Jr. False-posi-
tive hematuria. JAMA 1985;​254:​1724.
14. Adams EC. Differentiation of myoglo-
bin and hemoglobin in biological fluids.
Ann Clin Lab Sci 1980;​10:​493-9.
15. Cohen RA, Brown RS. Microscopic
hematuria. N Engl J Med 2003;​ 348:​
2330-8.
16. Grossfeld GD, Litwin MS, Wolf JS,
et al. Evaluation of asymptomatic micro-
scopic hematuria in adults: the American
Urological Association best practice pol-
icy — part I: definition, detection, preva-
lence, and etiology. Urology 2001;​57:​599-
603.
17. Sutton JM. Evaluation of hematuria in
adults. JAMA 1990;​263:​2475-80.
18. Makino H, Nishimura S, Soda K,
Takaoka M, Kaneshige T, Ota Z. Mecha-
nism of hematuria. I. Electron microscop-
ic demonstration of the passage of a red
blood cell through a glomerular capillary
wall in rat masugi nephritis. Virchows
Arch B Cell Pathol Incl Mol Pathol 1986;​
50:​199-208.
19. Linder BJ, Bass EJ, Mostafid H, Boor-
jian SA. Guideline of guidelines: asymp-
tomatic microscopic haematuria. BJU Int
2018;​121:​176-83.
20. Yuste C, Gutierrez E, Sevillano AM,

n engl j med 385;2  nejm.org  July 8, 2021 161

The New England Journal of Medicine
Downloaded from nejm.org by Salvador Macias-Díaz on August 6, 2021. For personal use only. No other uses without permission.
Copyright © 2021 Massachusetts Medical Society. All rights reserved.
 The n e w e ng l a n d j o u r na l
et al. Pathogenesis of glomerular haema-
turia. World J Nephrol 2015;​4:​185-95.
21. Praga M, Gutierrez-Millet V, Navas JJ,
et al. Acute worsening of renal function
during episodes of macroscopic hematu-
ria in IgA nephropathy. Kidney Int 1985;​
28:​69-74.
22. Vivante A, Afek A, Frenkel-Nir Y, et al.
Persistent asymptomatic isolated micro-
scopic hematuria in Israeli adolescents
and young adults and risk for end-stage
renal disease. JAMA 2011;​306:​729-36.
23. Kido R, Shibagaki Y, Iwadoh K, et al.
Persistent glomerular hematuria in living
kidney donors confers a risk of progres-
sive kidney disease in donors after hemi-
nephrectomy. Am J Transplant 2010;​ 10:​
1597-604.
24. Moreno JA, Sevillano Á, Gutiérrez E,
et al. Glomerular hematuria: cause or
consequence of renal inflammation? Int
J Mol Sci 2019;​20:​2205.
25. Moreno JA, Martín-Cleary C, Gutiér-
rez E, et al. AKI associated with macro-
scopic glomerular hematuria: clinical and
pathophysiologic consequences. Clin J Am
Soc Nephrol 2012;​7:​175-84.
26. Sheerin NS, Sacks SH, Fogazzi GB. In
vitro erythrophagocytosis by renal tubu-
lar cells and tubular toxicity by haemoglo-
bin and iron. Nephrol Dial Transplant
1999;​14:​1391-7.
27. Willis GC, Tewelde SZ. The approach
to the patient with hematuria. Emerg Med
Clin North Am 2019;​37:​755-69.
28. Mishriki SF, Vint R, Somani BK. Half
of visible and half of recurrent visible he-
maturia cases have underlying pathology:
prospective large cohort study with long-
term followup. J Urol 2012;​187:​1561-5.
29. Mishriki SF, Grimsley SJS, Nabi G. In-
cidence of recurrent frank hematuria and
urological cancers: prospective 6.9 years
of followup. J Urol 2009;​182:​1294-8.
30. Expert Panel on Urological Imaging,
Wolfman DJ, Marko J, et al. ACR Appro-
priateness Criteria® hematuria. J Am Coll
Radiol 2020;​17(5S):​S138-S147.
31. Wood DP Jr. Is it time to revise our
evaluation of hematuria? J Urol 2009;​182:​
1253-4.
32. Bolenz C, Schröppel B, Eisenhardt A,
Schmitz-Dräger BJ, Grimm M-O. The in-
vestigation of hematuria. Dtsch Arztebl
Int 2018;​115:​801-7.
33. Tan WS, Feber A, Sarpong R, et al.
Who should be investigated for haematu-
ria? Results of a contemporary prospec-
tive observational study of 3556 patients.
Eur Urol 2018;​74:​10-4.
34. Halpern JA, Chughtai B, Ghomrawi
H. Cost-effectiveness of common diagnos-
tic approaches for evaluation of asymp-
tomatic microscopic hematuria. JAMA
Intern Med 2017;​177:​800-7.
35. Subak LL, Grady D. Asymptomatic
microscopic hematuria — rethinking the
162 n engl j med 385;2  nejm.org  July 8, 2021
The New England Journal of Medicine
Downloaded from nejm.org by Salvador Macias-Díaz on August 6, 2021. For personal use only. No other uses without permission.
Copyright © 2021 Massachusetts Medical Society. All rights reserved.
of m e dic i n e
diagnostic algorithm. JAMA Intern Med
2017;​177:​808-9.
36. American Cancer Society. Cancer sta-
tistics center (https://cancerstatisticscenter​
.­cancer​.­org/​­?_ga=​­2​.­82720342​.­1930708337​
.­1595527626​-­1692772379​.­1595527626#!/​­).
37. Jung H, Gleason JM, Loo RK, Patel
HS, Slezak JM, Jacobsen SJ. Association of
hematuria on microscopic urinalysis and
risk of urinary tract cancer. J Urol 2011;​
185:​1698-703.
38. Madeb R, Golijanin D, Knopf J, et al.
Long-term outcome of patients with a neg-
ative work-up for asymptomatic micro-
hematuria. Urology 2010;​75:​20-5.
39. Nielsen M, Qaseem A. Hematuria as
a marker of occult urinary tract cancer:
advice for high-value care from the Amer-
ican College of Physicians. Ann Intern
Med 2016;​164:​488-97.
40. Murakami S, Igarashi T, Hara S,
Shimazaki J. Strategies for asymptomatic
microscopic hematuria: a prospective
study of 1,034 patients. J Urol 1990;​144:​
99-101.
41. Khadra MH, Pickard RS, Charlton M,
Powell PH, Neal DE. A prospective analy-
sis of 1,930 patients with hematuria to
evaluate current diagnostic practice. J Urol
2000;​163:​524-7.
42. Davis R, Jones JS, Barocas DA, et al.
Diagnosis, evaluation and follow-up of
asymptomatic microhematuria (AMH) in
adults: AUA guideline. J Urol 2012;​188:​
2473-81.
43. Birch DF, Fairley KF. Haematuria: glo-
merular or non-glomerular? Lancet 1979;​
2:​845-6.
44. Pollock C, Liu PL, Györy AZ, et al.
Dysmorphism of urinary red blood cells
— value in diagnosis. Kidney Int 1989;​36:​
1045-9.
45. Köhler H, Wandel E, Brunck B. Acan-
thocyturia — a characteristic marker for
glomerular bleeding. Kidney Int 1991;​40:​
115-20.
46. Kitamoto Y, Tomita M, Akamine M,
et al. Differentiation of hematuria using
a uniquely shaped red cell. Nephron 1993;​
64:​32-6.
47. Hamadah AM, Gharaibeh K, Mara
KC, et al. Urinalysis for the diagnosis of
glomerulonephritis: role of dysmorphic
red blood cells. Nephrol Dial Transplant
2018;​33:​1397-403.
48. Shichiri M, Hosoda K, Nishio Y, et al.
Red-cell-volume distribution curves in di-
agnosis of glomerular and non-glomeru-
lar haematuria. Lancet 1988;​1:​908-11.
49. Mishriki SF, Aboumarzouk O, Vint R,
Grimsley SJ, Lam T, Somani B. Routine
urine cytology has no role in hematuria
investigations. J Urol 2013;​189:​1255-8.
50. Hof land CA, Mariani AJ. Is cytology
required for a hematuria evaluation?
J Urol 2004;​171:​324-6.
51. Yafi FA, Brimo F, Auger M, Aprikian
A, Tanguay S, Kassouf W. Is the perfor-
mance of urinary cytology as high as re-
ported historically? A contemporary analy-
sis in the detection and surveillance of
bladder cancer. Urol Oncol 2014;​ 32(1):​
27.e1-6.
52. Kassouf W. The value of urine cytol-
ogy in the workup of hematuria. Cancer
Cytopathol 2016;​124:​303-4.
53. Soria F, Krabbe L-M, Todenhöfer T,
et al. Molecular markers in bladder can-
cer. World J Urol 2019;​37:​31-40.
54. Sathianathen NJ, Butaney M, Weight
CJ, Kumar R, Konety BR. Urinary bio-
markers in the evaluation of primary
hematuria: a systematic review and meta-
analysis. Bladder Cancer 2018;​4:​353-63.
55. Wilson JL Jr, Antoniassi MP, Lopes PI,
Azevedo H. Proteomic research and diag-
nosis in bladder cancer: state of the art
review. Int Braz J Urol 2021;​47:​503-14.
56. Suzuki H. Biomarkers for IgA ne-
phropathy on the basis of multi-hit patho-
genesis. Clin Exp Nephrol 2019;​23:​26-31.
57. Coppo R. Biomarkers and targeted
new therapies for IgA nephropathy. Pedi-
atr Nephrol 2017;​32:​725-31.
58. Benonisdottir S, Kristjansson RP,
Oddsson A, et al. Sequence variants as-
sociating with urinary biomarkers. Hum
Mol Genet 2019;​28:​1199-211.
59. Tan MP, Attard G, Huddart RA. Circu-
lating tumour DNA in muscle-invasive
bladder cancer. Int J Mol Sci 2018;​ 19:​
2568.
60. Nuzzo PV, Berchuck JE, Korthauer K,
et al. Detection of renal cell carcinoma
using plasma and urine cell-free DNA
methylomes. Nat Med 2020;​26:​1041-3.
61. Moyer VA. Screening for bladder can-
cer: U.S. Preventive Services Task Force
recommendation statement. Ann Intern
Med 2011;​155:​246-51.
62. Bauer SR, Carroll PR, Grady D. Hema-
turia practice guidelines that explicitly
consider harms and costs. JAMA Intern
Med 2019;​179:​1362-4.
63. Georgieva MV, Wheeler SB, Erim D,
et al. Comparison of the harms, advan-
tages, and costs associated with alterna-
tive guidelines for the evaluation of hema-
turia. JAMA Intern Med 2019;​179:​1352-62.
64. Tan WS, Ahmad A, Feber A, et al. De-
velopment and validation of a haematuria
cancer risk score to identify patients at
risk of harbouring cancer. J Intern Med
2019;​285:​436-45.
65. Suspected cancer:​ recognition and
referral. NICE guideline. London:​Na-
tional Institute for Health and Care Ex-
cellence, 2015 (https://www​.­n ice​.­org​.­u k/​
­g uidance/​­ng12/​­resources/​­suspected​
-­c ancer​-­recognition​-­a nd​-­referral​-­pdf​
-­1837268071621).
66. Wu X, Lin J, Grossman HB, et al. Pro-
jecting individualized probabilities of
developing bladder cancer in white indi-
 Hematuria in Adults

viduals. J Clin Oncol 2007;​25:​4974-81.
67. van der Molen AJ, Hovius MC. Hema-
turia: a problem-based imaging algo-
rithm illustrating the recent Dutch guide-
lines on hematuria. AJR Am J Roentgenol
2012;​198:​1256-65.
68. Assmus MA, Beyer DB, Hanks J, et al.
Quality and cost assessment of Canadian
Urological Association microscopic hema-
turia guidelines in clinical practice: turn-
ing urine into gold. Can Urol Assoc J
2019;​13:​406-11.
69. Wollin T, Laroche B, Psooy K. Cana-
dian guidelines for the management of
asymptomatic microscopic hematuria in
adults. Can Urol Assoc J 2009;​3:​77-80.
70. Anderson J, Fawcett D, Feehally J,
Goldberg L, Kelly J, MacTier R. Joint con-
sensus statement on the initial assess-
ment of haematuria. July 2008 (https://
www​.­baus​.­org​.­u k/​­_userfiles/​­pages/​­f iles/​
­News/​­haematuria_consensus_guidelines
_July_2008​.­pdf).
71. Woldu SL, Ng CK, Loo RK, et al. Eval-
uation of the new American Urological
Association guidelines risk classification
for hematuria. J Urol 2021;​205:​1387-93.
72. Skaggs AW, Loehfelm TW, Fanana-
pazir G, Dall’Era M, Corwin MT. Utiliza-
tion and yield of CT urography: are the
American Urological Association guide-
lines for imaging of patients with asymp-
tomatic microscopic hematuria being fol-
lowed? AJR Am J Roentgenol 2021;​216:​
106-10.
Copyright © 2021 Massachusetts Medical Society.

images in clinical medicine

The Journal welcomes consideration of new submissions for Images in Clinical
Medicine. Instructions for authors and procedures for submissions can be found
on the Journal’s website at NEJM.org. At the discretion of the editor, images that
are accepted for publication may appear in the print version of the Journal,
the electronic version, or both.

n engl j med 385;2  nejm.org  July 8, 2021 163

The New England Journal of Medicine
Downloaded from nejm.org by Salvador Macias-Díaz on August 6, 2021. For personal use only. No other uses without permission.
Copyright © 2021 Massachusetts Medical Society. All rights reserved.
 The New England Journal of Medicine
Downloaded from nejm.org at The University Of Illinois on November 26, 2015. For personal use only. No other uses without permission.
From the NEJM Archive. Copyright © 2009 Massachusetts Medical Society. All rights reserved.
 The New England Journal of Medicine
Downloaded from nejm.org at The University Of Illinois on November 26, 2015. For personal use only. No other uses without permission.
From the NEJM Archive. Copyright © 2009 Massachusetts Medical Society. All rights reserved.
 MARIJUANA USE DOES NOT SPURIOUSLY ELEVATE
SERUM HUMAN CHORIONIC GONADOTROPIN LEVELS

GLENN D. BRAUNSTEIN, M.D.

RONALD THOMPSON, M.D.
STANLEY GROSS, M.D.
JAMES R. SOARES, PH.D.

From the Departments of Medicine and Radiation Therapy,

Cedars-Sinai Medical Center-UCLA School of Medicine,
Los Angeles, California

ABSTRACT-Marijuana use has been reported to spuriously elevate immunoreactive human cho-
rionic gonadotropin (HCG) in the serum of patients with testicular germ cell tumors. To reinvesti-
gate this finding, we measured serum HCG and AQ-tetrahydrocannabinol (A”-THC) levels in 16
men known or suspected to be marijuana users. Eight of the serum samples had measurable levels of
AQ-THC, while eight did not. None contained immunoreactive HCG. The addition of Ae-THC
directly to pooled male serum had no effect on the HCG radioimmunoassay. These results indicate
that marijuana does not artificially elevate serum HCG concentrations.

It has been well established that human cho-
rionic gonadotropin (HCG) measurements are
useful for monitoring disease activity in patients
with germ cell tumors of the testes.’ Spuriously
elevated HCG concentrations have been noted
in some assay systems because of cross-reaction
with human pituitary luteinizing hormone
(HLH) ,2-5 lipid interference,s presence of
heterophilic antibodies,7,8 or HCG binding pro-
teins.Q-12 In addition, Garnick13 suggested that
marijuana abuse may result in an elevation of
HCG in the serum of patients with germ cell
tumors of the testes. However, we were unable
to demonstrate an alteration by marijuana on
HCG concentrations during pregnancy. l4
Therefore, we examined directly the effects of
marijuana on HCG measurements to determine
whether or not cannabinoids interfere with
HCG determinations.
Material and Methods
Blood was obtained from 16 men suspected
or known to be marijuana users. The serum was
separated and stored at - 20°C until assayed.
Serum AQ-tetrahydrocannabinol (Ag-THC) was
measured by a specific radioimmunoassay. l5
HCG measurements were performed by our
previously described double antibody beta sub-
unit radioimmunoassay method using reagents
supplied by the National Hormone and Pi-
tuitary Program, Baltimore, MD.le
To investigate the possibility of cross-reaction
of marijuana in the HCG radioimmunoassay,
we added the active ingredient of marijuana,
A9-tetrahydrocannabinol, directly to normal
pooled male serum and measured the pool in
HCG radioimmunoassay (RIA).
Results
Eight of the 16 patients had measurable
quantities of A9-THC (range 5.6-34.2 ng/ml) in
their sera, while 8 had no detectable AQ-THC.
None of the 16 samples had measurable levels of
immunoreactive HCG.
No apparent HCG activity was present in the
pooled male serum samples after the addition

UROLOGY i JUNE 1985 i VOLUMEXXV,NUMBER6 605

from 5 to 500 pg of Ag-THC to the samples prior
to the assay.
Comment
Our results indicate that neither Ag-THC nor
other cannabinoid metabolites present in the
serum of individuals who smoke the drug inter-
feres in the HCG radioimmunoassay or leads to
the apparent, but spurious, presence of HCG in
blood. This finding is supported by the recent
report by Hogan and co-workers” who found
no effect of the synthetic cannabinoid nabilone
on serum HCG concentrations in men receiving
combination chemotherapy for testicular can-
cer. In addition, we previously had shown no
difference in the serum concentrations of three
trophoblastic proteins (HCG, pregnancy-spe-
cific-betal-glycoprotein [SPr], and placenta
lactogen) at any time during pregnancy in 13
women who used marijuana regularly through-
out pregnancy in comparison with 13 pregnant
control women who did not use marijuana.14
Thus, it is unlikely that even if a patient with a
testicular tumor containing trophoblastic ele-
ments which produced HCG smoked mari-
juana heavily, A9-THC or one of its derivatives
would stimulate the tumor to secrete increased
quantities of the hormone.
It should be stressed that our beta HCG ra-
dioimmunoassay utilizes reagents supplied by
the National Hormone and Pituitary Program,
Baltimore, MD. Since the antisera in the com-
mercial beta HCG RIA kits differ in their reac-
tivity to HCG and HLH,18 our results may not
be found with other HCG assays. Several
authors have noted that high concentrations of
HLH may result in false positive elevations of
HCG with some HCG RIA kits in patients with
testicular tumors.2-5 This problem can be cir-
cumvented by using more specific reagents or
by measuring the HCG concentration following
suppression of HLH by testosterone administra-
tion.2.3
A rise in serum HCG concentration measured
by a reliable, specific assay in a patient with a
testicular tumor should be considered evidence
of relapse and not be attributed to the use of
cannabis derivatives.
606
Los Angeles, California 90048
(DR. BRAUNSTEIN)
ACKNOWLEDGMENTS. To Ms. Helene Zauderer for cleri-
cal help, and the support of the Linda Simon Cancer Re-
search Fund.
References
1. Braunstein GD: HCG expression in trophoblastic and non-
trophoblastic tumors, in Fishman WH (Ed): Oncodevelopmental
Markers, Biologic, Diagnostic and Monitoring Aspects, New York,
Academic Press, 1983, p 351.
2. Catalona WJ, Vaitukaitis JL, and Fair WR: Falsely positive
specific human chorionic gonadotropin assays in patients with tes-
titular tumors: conversion to negative with testosterone adminis-
tration, J Urol 122: 126 (1981).
3. Ketchum C, et al: False increases of beta-subunit cho-
riogonadotropin in commercial kit results because of cross reactiv-
ity with lutropin, Clin Chem 28: 1998 (1982).
4. Javadpour N, and Soares T: False-positive and false-nega-
tive alphafetoprotein and human chorionic gonadotropin assays
in testicular cancer: a double blind studv. Cancer 48: 2279 (1981).
5. Fowler JE, et al: Commercial radioimmunoassay for beta
subunit of human chorionic gonadotropin: false positive deter-
minations due to elevated serum luteinizing hormone, ibid 49:
136 (1982).
6. Fritz TJ, Bunker DM, and Lawlor J: Hyperlipidemia inter-
ference in radioimmunoassavs. Clin Chem 28: 2325 (1982).
7. Baulieu JL, et al: Falsely elevated results of radidimmunoas-
says using double antibody method: arguments for a third anti-
rabbit IgG antibody present in certain human sera, Eur J Nucl
Med 7: 121 (1982).
8. Vladutiu AO, et al: Heterophilic antibodies interfering with
radioimmunoassay. A false-positive pregnancy test, JAMA 248:
2489 (1982).
9. Chan DW, Bill MJ, and Rosenshein N: Specific binding pro-
teins for human chorionic gonadotrophin in a patient with
trophoblastic disease, Am J Clin Path01 76: 214 (1981).
10. Braunstein GD, et al: Characterization of antihuman cho-
rionic gonadotropin serum antibody appearing after ovulation in-
duction, J Clin Endocrinol Metab 57: 1164 (1983).
11. Musch K, Wolf AS, and Lauritzen C: Antibodies to human
chorionic gonadotropin in humans, Clin Chim Acta 113: 95
(1981).
12. Nieschlag E, Bernitz S, and Topert M: Antigenicity of hu-
man chorionic gonadotropin preparations in men, Clin Endo-
crinol 16: 483 (1982).
13. Garnick MB: Spurious rise in human chorionic gonadotro-
pin induced by marihuana in patients with testicular cancer, N
Engl J Med 303: 1177 (1980).
14. Braunstein GD, et al: Pregnancy hormone concentrations
in marijuana users, Life Sci 33: 195 (1983).
15. Yeager ER et al: A8-tetrahydrocannabinol by GLC-MS val-
idated radioimmunoassays of hemolyzed blood or serum, J Anal
Toxic01 5: 81 (1981).
16. Braunstein GD, et al: First trimester chorionic gonadotro-
pin measurements as an aid in the diagnosis of early pregnancy
disorders, Am J Obstet Gynecol 131: 25 (1978).
17. Hogan P, et al: Cannabinoids and hCG levels in patients
with testicular cancer, Lancet 2: 1144 (1983).
18. Rasor JL, Farber S, and Braunstein GD: An evaluation of
10 kits for determination of human choriogonadotropin in serum,
Clin Chem 29: 1828 (1983).
UROLOGY / JUNE 1985 / VOLUME XXV, NUMBER 6
 VOLUME 33 䡠 NUMBER 8 䡠 MARCH 10 2015

JOURNAL OF CLINICAL ONCOLOGY O R I G I N A L R E P O R T

Randomized Controlled Trial of the Prophylactic Effect of

ZhengGang Ren and Sheng-Long Ye,
Liver Cancer Institute, Zhongshan
Hospital, Fudan University, and Key
Laboratory of Carcinogenesis and
Cancer Invasion, Ministry of Education;
ZengQiang Qu, WeiPing Zhou, and
LeHua Shi, Eastern Hepatobiliary
Surgery Hospital of the Second Military
Medical University, Shanghai;
KangShun Zhu and MinQiang Lu, The
Third Affiliated Hospital of Sun Yat-sen
University; LiGong Lu, Guangdong
Provincial People’s Hospital, Guang-
dong; HaiYan Kang, 301 Military Hospi-
tal; YongPing Yang, 302 Military
Hospital; XiaoFeng Guo, Chinese Anti-
Cancer Association, Beijing; TianQiang
Song, Tianjin Cancer Hospital, Tianjin;
Hui Wang, Jilin Provincial Tumor Hospi-
tal, Jilin; WeiZhu Yang, Union Hospital
of Fujian Medical University, Fujian;
Xuan Wang, The 81 Hospital of the
Chinese People’s Liberation Army,
Nanjing; and YuXian Bai, Heilongjiang
Provincial Cancer Hospital, Heilongjiang,
People’s Republic of China.
Published online ahead of print at
www.jco.org on February 9, 2015.
Sponsored by the Chinese Anti-Cancer
Association and Bayer HealthCare.
Authors’ disclosures of potential
conflicts of interest and author
contributions are found at the end of
this article.
Clinical trial information: NCT01934829.
Urea-Based Cream on Sorafenib-Associated Hand-Foot
Skin Reactions in Patients With Advanced
Hepatocellular Carcinoma
ZhengGang Ren, KangShun Zhu, HaiYan Kang, MinQiang Lu, ZengQiang Qu, LiGong Lu, TianQiang Song,
WeiPing Zhou, Hui Wang, WeiZhu Yang, Xuan Wang, YongPing Yang, LeHua Shi, YuXian Bai,
XiaoFeng Guo, and Sheng-Long Ye
A B S T R A C T
Purpose
To assess whether urea-based cream (UBC) has prophylactic benefits on sorafenib-induced
hand-foot skin reaction (HFSR) in patients with advanced hepatocellular carcinoma (HCC).
Patients and Methods
In this randomized, open-label trial, 871 patients with advanced HCC throughout China were
treated with 10% UBC three times per day plus best supportive care (BSC; n ⫽ 439) or BSC alone
excluding all creams (n ⫽ 432), starting on day 1 of sorafenib treatment, for up to 12 weeks. HFSR
was assessed every 2 weeks and at 14 weeks for patients completing the study. Once HFSR
occurred, patients were allowed any cream, including a UBC.
Results
The 12-week incidence of any grade HFSR was significantly lower in the UBC group versus the
BSC-alone group (56.0% v 73.6%, respectively; odds ratio [OR], 0.457; 95% CI, 0.344 to 0.608;
P ⬍ .001), as was the incidence of grade ⱖ 2 HFSR (20.7% v 29.2%, respectively; OR, 0.635; 95%
CI, 0.466 to 0.866; P ⫽ .004). Median time to first occurrence of HFSR was significantly longer in
the UBC group than the BSC-alone group (84 v 34 days, respectively; hazard ratio, 0.658; 95% CI,
0.541 to 0.799; P ⬍ .001). Elevated AST was associated with increased risk of HFSR but did not
alter the treatment effect of UBC. UBC plus BSC, compared with BSC alone, did not affect the
sorafenib dose reduction or interruption rate (9.1% v 11.8%, respectively; P ⫽ .1937), response
rate (11.1% v 10.1%, respectively; P ⫽ .6674), or disease control rate (98.8% v 98.2%,
respectively; P ⫽ .5350) at week 12.
Conclusion
UBC prophylaxis in patients with advanced HCC starting sorafenib reduced HFSR rates, extended
the time to first occurrence of HFSR, and improved patient quality of life compared with BSC.
Blinded, randomized, placebo-controlled trials to determine the role of UBC on the incidence and
severity of HFSR are warranted.
J Clin Oncol 33:894-900. © 2015 by American Society of Clinical Oncology

Corresponding author: Sheng-Long Ye,

MD, Liver Cancer Institute, Zhongshan
Hospital, Fudan University, 180 Fenglin
Rd, Shanghai 200032, People’s Repub-
lic of China; e-mail: slye@shmu.edu.cn.
© 2015 by American Society of Clinical
Oncology
0732-183X/15/3308w-894w/$20.00
DOI: 10.1200/JCO.2013.52.9651
INTRODUCTION
The multikinase inhibitor sorafenib1-3 was shown to
be effective in two phase III clinical trials in patients
with advanced hepatocellular carcinoma (HCC)4,5
and has become standard systemic therapy for HCC
worldwide, including in China.6-11 Dermatologic
adverse events (AEs), however, are frequent, includ-
ing hand-foot skin reaction (HFSR; palmar and/or
plantar erythrodysesthesia), rash, and alopecia.
HFSR associated with tyrosine kinase inhibitors
(TKIs) is characterized by keratinocytic necrosis, der-
mal edema, parakeratosis, and hyperkeratosis.12-15 A
meta-analysis of 24 clinical trials including more
than 6,000 patients with various malignancies found
that the rates of any grade and grade 3 HFSR in
sorafenib-treated patients were 39% and 9%, re-
spectively.16 The two phase III trials of sorafenib in
patients with HCC reported any grade HFSR rates of
21% and 45% and grade 3 HFSR rates of 8% and 11%,
withHFSRbeingoneofthemostcommonAEsleading
to dose reduction.4,5 High rates of any grade HFSR
(72% to 82%) have been reported during sorafenib
treatment of Asian patients with HCC.17,18

894 © 2015 by American Society of Clinical Oncology

Downloaded from ascopubs.org by 200.92.202.106 on August 19, 2022 from 200.092.202.106

Copyright © 2022 American Society of Clinical Oncology. All rights reserved.
 Prophylactic Urea-Based Cream for Sorafenib HFSR in Advanced HCC

Screened
(N = 901)

Excluded (n = 30)

Randomly assigned
(n = 871)

Assigned prophylactic Assigned BSC only

urea-based cream + BSC excluding all creams
(n = 439) (n = 432)

Evaluable for incidence of HFSR Evaluable for incidence of HFSR

Fig 1. Study design. BSC, best support-
(n = 439) (n = 432)
ive care; HFSR, hand-foot skin reaction.

Nonevaluable (n = 85) Nonevaluable patients (n = 87)

patients Missing informed (n = 5)
Missing informed (n = 4) consent date
consent date Calculated time (n = 6)
Calculated time (n = 14) to first event
to first event outside of limit
outside of limit Missing date or (n = 76)
Missing date or (n = 67) grade of HFSR
grade of HFSR

Evaluable for time to first HFSR Evaluable for time to first HFSR
(n = 354) (n = 345)

Despite prolonging survival in patients with HCC, the advan-
tages of sorafenib are reduced by its toxicities, including
HFSR.12,13,19,20 Although not life threatening, HFSR can severely af-
fect health-related quality of life (HRQoL), including physical, psy-
chological, and social well-being,12,19,21 resulting in sorafenib dose
reductions and/or discontinuations and potentially altering survival
outcomes. Appropriate HFSR prophylaxis and management are nec-
essary to ensure proper sorafenib dosing and enhance HRQoL.
Mild hyperkeratosis, an early sign of HFSR, may be the only
manifestation of sorafenib-associated HFSR. Creams containing urea
are widely used to treat hyperkeratotic conditions, including psoria-
sis,22 and have been recommended for TKI-related HFSR.23 These
creams are generally inexpensive, readily available, and well tolerated.
At present, however, there is no consensus on best supportive care
(BSC) for management of TKI-associated HFSR, and to our knowl-
edge, no randomized controlled trials have evaluated methods to
prevent or palliate sorafenib-associated HFSR.23,24 This large random-
ized controlled trial evaluated the prophylactic effects of a urea-based
cream (UBC) on the incidence of sorafenib-associated HFSR in pa-
tients with advanced HCC.
PATIENTS AND METHODS
Patients
This investigator-initiated, randomized, open-label trial assessed
whether a topical UBC could reduce the incidence and severity of HFSR in
patients with advanced HCC starting sorafenib at 64 centers throughout China
(Appendix Table A1, online only). Other inclusion criteria included willing-
ness and ability to report for clinic visits every 2 weeks for the first 3 months of
treatment, willingness and ability (with assistance if necessary) to fill out
patient-reported questionnaires, discontinuation of cancer therapy at least 3
weeks before study entry, and life expectancy of ⱖ 3 months. Exclusion criteria
included previous treatment with sorafenib, ongoing concomitant anticancer
therapy, or concurrent enrollment in another clinical trial. Each patient pro-
vided written informed consent, and the study protocol was approved by the
investigative review and/or ethics board of each participating institution.
Study Design
All patients receiving sorafenib were randomly assigned 1:1 to either
UBC (10% urea; Eucerin, Beiersdorf, Hamburg, Germany) three times per day
plus BSC (n ⫽ 439) for AEs other than HFSR or BSC alone (n ⫽ 432)
excluding use of all creams (Fig 1). HFSR was evaluated using a modified
grading system specific to HFSR symptoms (Appendix Table A2, online only).
All other AEs were evaluated by the treating physician using National Cancer
Institute Common Terminology Criteria for Adverse Events (version 3). Once
HFSR occurred, patients were allowed any cream, per the treating physician.
Patients were examined every 2 weeks for 12 weeks and followed up at week 14.
During each visit, concomitant medications, laboratory values (hematology
and chemistry), and 26 preselected/precoded AEs were recorded, including
AEs commonly associated with sorafenib treatment. Patients returned any
unused urea cream at each visit. Patients underwent abdominal computed
tomography or magnetic resonance imaging scans to measure and evaluate
tumors at screening, at week 6 or 8, and at the final (week 12) or follow-up
(week 14) visit.
Sorafenib dose modifications (interruptions and reductions) for derma-
tologic toxicities, including HFSR, were allowed as predefined in the study
protocol. Dosing was uninterrupted for grade 1 toxicity. Doses were reduced
for grade 2 toxicities until resolution to grade 0 to 1, with resumption at the
next cycle of full-dose sorafenib (400 mg twice per day) after the first incident,
reduction of sorafenib dose to 400 mg once every day after the second incident,
and reduction of sorafenib dose to 400 mg every 2 days after subsequent

www.jco.org © 2015 by American Society of Clinical Oncology 895

Downloaded from ascopubs.org by 200.92.202.106 on August 19, 2022 from 200.092.202.106

Copyright © 2022 American Society of Clinical Oncology. All rights reserved.
 Ren et al

incidents. Sorafenib doses were interrupted for grade 3 toxicities until resolu- Response rates, rates of common AEs, and rates of dose reductions and
tion to grade 0 to 1, with reduction at the next cycle to sorafenib 400 mg once discontinuations as a result of HFSR were compared using ␹2 or Fisher’s exact
every day after the first incident and to sorafenib 400 mg every 2 days after test. Mean HFSR symptom score and HF-QoL daily activity score were com-
subsequent incidents. pared using the Wilcoxon rank sum test.

Study End Points

The primary end point was the incidence of any grade HFSR within 12 RESULTS
weeks of starting sorafenib. Secondary end points included incidence of grade
ⱖ 2 HFSR within 12 weeks, time to first occurrence of HFSR, duration of
Of the 901 screened patients with HCC, 871 were randomly as-
HFSR, percentages of patients with sorafenib dose reductions/interruptions
and discontinuations, HFSR-associated HRQoL using the Hand-Foot Reac- signed to UBC plus BSC (n ⫽ 439) or BSC alone (n ⫽ 432) when
tion Quality of Life (HF-QoL) questionnaire, and tumor response according starting sorafenib 400 mg twice per day (Fig 1). At baseline, the two
to RECIST criteria. The population evaluable for HFSR consisted of all patients groups were generally well matched demographically and clini-
who received at least one dose of sorafenib. The efficacy population consisted cally, including in the etiology of HCC, ECOG PS, previous anti-
of all patients who received at least one dose of sorafenib and for whom cancer therapy, and history of disease (Table 1), except that
computed tomography scans were available for tumor assessment. transarterial embolizartion was significantly more frequent in the
The HF-QoL questionnaire (Anderson et al, manuscript in preparation)
UBC plus BSC group (10.1% v 6.3% for BSC alone; P ⫽ .0409).
was adapted from the validated Dermatology-Specific QoL questionnaire, a
multidimensional questionnaire assessing HRQoL across a variety of skin Analysis of history of dermatosis showed that one patient in the
diseases.25 The HF-QoL includes 48 items covering HFSR symptoms on the UBC plus BSC group had a skin eruption, and one patient each in
feet, hands and fingers, and other body areas; the impact of these symptoms on the BSC group had psoriasis and an unidentified dermatosis.
a patient’s ability to sleep and perform daily activities; and the effects of HFSR
on a patient’s social life and feelings or mood. The HF-QoL questionnaire
includes the following three global items: a visual scale measuring overall
health, a question on overall rating of HFSR severity, and a global rating of
change in HFSR. Table 1. Baseline Demographics and Clinical Characteristics of the
For HF-QoL parameters, the minimal clinically important difference, or Evaluable Population
the smallest change indicative of a clinically important improvement or de- UBC ⫹ BSC BSC Alone
cline, was estimated from change in scores corresponding to a unit change in (n ⫽ 439) (n ⫽ 432)
toxicity from grade 0 to grade 1, change in self-rated severity of 1 (better), and Demographic or Clinical No. of No. of
change from no to mild symptoms. Thus, differences of ⱖ 4 and ⱖ 6 units in Characteristic Patients % Patients %
HF-QoL Daily Activities and Symptoms, respectively, were considered clini-
Age, years
cally meaningful.
Median 51.8 52.0
Range 20.4-86.4 20.2-84.3
Statistical Analysis Male 377 85.9 368 85.2
We estimated that 45% of patients with HCC would develop HFSR Weight, kg
during the first 12 weeks of sorafenib treatment and that the UBC would Median 67.0 66.5
reduce the relative risk by 25%, resulting in a nominal HFSR incidence of Range 40.5-88.0 40.5-101.0
33.75%. With a one-sided ␣ ⫽ .025 and a power of 90%, we calculated a HCC etiology
requirement for 412 patients per group, or a total of 824 patients. To allow for Hepatitis B 339 77.2 332 76.9
a drop-out rate of 10%, 900 patients were screened. Hepatitis C 14 3.2 11 2.5
The primary end point of any grade HFSR within 12 weeks of starting ECOG performance score
sorafenib and the incidence of HFSR at each study visit were compared in the 0 209 47.7 201 46.5
two groups using the ␹2 or Fisher’s exact test. Correlations between treatment 1 211 48.2 215 49.8
effects and severity of HFSR were assessed using the Mantel-Haenszel test, 2 17 3.9 16 3.7
comparing grade 0 versus grade 1 versus grade 2 or 3. An exploratory logistic Previous therapyⴱ
regression analysis was used to identify demographic and clinical factors pre- Any 301 68.7 293 67.8
dictive of HFSR within 12 weeks of starting sorafenib. To identify potential Hepatectomy 107 24.4 100 23.1
predictors of HFSR, a multivariable logistic regression model was used with a TACE 201 45.8 221 51.3
stepwise variable selection process.26 Covariates and factors investigated in- TAE 44 10.1† 27 6.3
cluded age (⬍ v ⱖ 65 years), sex, Eastern Cooperative Oncology Group Disease history
(ECOG) performance score (PS; 0 v ⱖ 1), body mass index (BMI; ⬍ v ⱖ 23.1 Alcohol/drug addiction 3 0.7 4 0.9
kg/m2), and ALT and AST concentrations (ⱖ 4 and ⱕ 40 v ⬎ 40 U/L for both). Allergy 4 0.9 2 0.5
Inclusion and exclusion criteria of P ⬍ .1 were used for stepwise selection. If the Sleep disorder 19 4.4 12 2.8
identified model contained predictive covariates other than treatment, inter- Arthritis 18 4.1 16 3.7
action terms were included to test for differential treatment effects. Hypertension 23 5.3 19 4.4
Time to first HFSR was defined as the time from date of random assign- Diabetes 11 2.5 14 3.3
ment to date of last visit or first report of HFSR; patients not experiencing Dermatosis‡ 1 0.2 2 0.5
HFSR were censored at the date of last visit. Patients who missed or with
Abbreviations: BSC, best supportive care; ECOG, Eastern Cooperative On-
invalid (eg, before random assignment) visits or who did not provide valid
cology Group; HCC, hepatocellular carcinoma; TACE, transarterial chemoem-
signed informed consent were excluded from analysis. Time to first HFSR was bolization; TAE, transarterial embolization; UBC, urea-based cream.
ⴱ
determined using the Kaplan-Meier method and compared using log-rank Patients may have received more than one therapy.
tests. Hazard ratio (HR) and 95% CIs were estimated by proportional hazards †P ⫽ .0409 compared with the BSC-alone group.
‡Including one patient with skin eruption in the UBC plus BSC group and
regression. Using a Cox regression model and prespecified covariates (age, sex,
one patient each with psoriasis and unidentified dermatosis in the
ECOG PS, BMI, and liver enzyme concentrations), an exploratory analysis was BSC-alone group.
performed to identify significant predictors of time to HFSR.

896 © 2015 by American Society of Clinical Oncology JOURNAL OF CLINICAL ONCOLOGY

Downloaded from ascopubs.org by 200.92.202.106 on August 19, 2022 from 200.092.202.106

Copyright © 2022 American Society of Clinical Oncology. All rights reserved.
 Prophylactic Urea-Based Cream for Sorafenib HFSR in Advanced HCC

Table 2. Incidence of Worst-Grade Treatment-Emergent HFSR in the Evaluable Population Within 12 Weeks
UBC ⫹ BSC (n ⫽ 439) BSC Alone (n ⫽ 432)

Grade of HFSR No. of Patients % No. of Patients % Odds Ratio 95% CI Pⴱ

Model 1
None 193 44.0† 114 26.4† Reference group ⬍ .001
Grade 1 155 35.3 192 44.4 0.477 0.349 to 0.653
Grade 2 72 16.4 98 22.7 0.434 0.296 to 0.636
Grade 3 19 4.3 28 6.5 0.401 0.214 to 0.750
Model 2
None 193 44.0† 114 26.4† Reference group ⬍ .001
Grade 1 155 35.3 192 44.4 0.477 0.349 to 0.653
Grade 2/3 91 20.7 126 29.2 0.427 0.299 to 0.609
Model 3
None 193 44.0† 114 26.4† Reference group ⬍ .001
Grade 1/2/3 246 56.0 318 73.6 0.457 0.344 to 0.608
Model 4
None/grade 1 348 79.3 306 70.8 Reference group ⬍ .004
Grade 2/3 91 20.7 126 29.2 0.635 0.466 to 0.866

Abbreviations: BSC, best supportive care; HFSR, hand-foot skin reaction; UBC, urea-based cream.
ⴱ
Mantel-Haenszel test.
†Includes all patients without recorded incidence of HFSR.

The incidence of any grade treatment-emergent HFSR within 12
weeks of starting sorafenib was significantly lower in the UBC plus
BSC group than in the BSC-alone group (56.0% v 73.6%, respectively;
odds ratio [OR], 0.457; 95% CI, 0.344 to 0.608; P ⬍ .001; Table 2).
Rates of moderate (grade 2) and severe (grade 3) HFSR were also
significantly lower in the UBC plus BSC group than the BSC-alone
group (20.7% v 29.2%, respectively; OR, 0.635; 95% CI, 0.466 to 0.866;
P ⫽ .004). The similarity of the ORs suggested that treatment effect
was not associated with HFSR severity and was similar across all grades
of HFSR. The incidence of any grade HFSR at each study visit was
lower in the UBC plus BSC group than in the BSC-alone group (Fig 2),
as were the rates of any grade HFSR for the 12-week observation
period (56.0% v 73.6%, respectively; P ⬍ .001) and the entire study
period (53.8% v 70.4%, respectively; P ⬍ .001; Table 3).
A stepwise exploratory logistic regression analysis evaluated vari-
ables associated with increased risk of HFSR. Although age, sex, ECOG
PS, BMI, and ALT were not predictive of increased risk, abnormally
high AST levels were associated with an OR of 2.123 (95% CI, 1.521 to
2.962). A logistic regression model testing for the interaction between
treatment and AST found a similar treatment effect in patients with
normal and abnormal AST (P ⫽ .2814). A univariable logistic regres-
sion model, using only treatment, yielded results consistent with the
multivariable model using treatment and AST (OR, 0.457; 95% CI,
0.344 to 0.608). Application of the same univariable model to the
subsets of patients with normal (OR, 0.424; 95% CI, 0.295 to 0.609)
and abnormal (OR, 0.594; 95% CI, 0.362 to 0.974) AST yielded results
similar to the multivariable model.
The median time to first occurrence of HFSR was significantly
longer in the UBC plus BSC group than in the BSC-alone group (84 v
Table 3. Any-Grade Drug-Related Adverse Events Associated With
Sorafenib Occurring in ⬎ 1.0% of the Evaluable Populations of Both Groups

UBC ⫹ BSC BSC Alone

(n ⫽ 439) (n ⫽ 432)
60
Urea cream + BSC
Adverse No. of No. of
BSC
50 Event Patients % Patients % P
HFSR 236 53.8 304 70.4 ⬍ .001ⴱ
40 Diarrhea 46 10.5 44 10.2 .8870ⴱ
Percent

Rash 27 6.2 22 5.1 .4982ⴱ

30 Fatigue 13 3.0 19 4.4 .2597ⴱ
Alopecia 13 3.0 12 2.8 .8712ⴱ
20 Anorexia 8 1.8 10 2.3 .6095ⴱ
Hypertension 7 1.6 7 1.6 .9758ⴱ
10 Headache 6 1.4 12 2.8 .1433ⴱ
Dry skin 5 1.1 13 3.0 .0524ⴱ
0 Erythema 4 0.9 6 1.4 .5435†
0 2 4 6 8 10 12 14
NOTE. Adverse events recorded according to Common Terminology Criteria
Time (weeks) for Adverse Events (version 3).
P .1445 .0081 < .001 < .001 < .001 < .001 .0445 Abbreviations: BSC, best supportive care; HFSR, hand-foot skin reaction;
UBC, urea-based cream.
ⴱ 2
␹ test.
Fig 2. Prevalence of any-grade hand-foot skin reaction at each study visit. †Fisher’s exact test.
Statistical analysis was conducted using the ␹2 test. BSC, best supportive care.

www.jco.org © 2015 by American Society of Clinical Oncology 897

Downloaded from ascopubs.org by 200.92.202.106 on August 19, 2022 from 200.092.202.106

Copyright © 2022 American Society of Clinical Oncology. All rights reserved.
 Ren et al

1.0 A 15
Urea cream + BSC
BSC
HFSR-Free Probability

12.4
0.8 11.5 11.1
10 8.9
9.3 9

Score
8.3
0.6 7.5 7.1
6.4
5.2 5.1
0.4 Urea cream + BSC 5 4.8 4.9
Median 84 days
95% CI: 45 to 93 days
HR: 0.658
0.2 BSC 95% CI: 0.541, 0.799 P .2681 .0171 < .001 < .001 .0036 .0041 .0255
Median 34 days P < .001
95% CI: 29 to 43 days 0 2 4 6 8 10 12 14

0 10 20 30 40 50 60 70 80 90 100 Time (weeks)

Time (days)
No. at risk
Urea cream 354 297 237 217 203 186 173 156 139 129
B 15 Urea cream + BSC
BSC
BSC 345 290 209 186 160 140 116 99 77 4

Fig 3. Kaplan-Meier analysis of time to first hand-foot skin reaction (HFSR) 10 9.6

Score
event in patients who received urea-based cream plus best supportive care (BSC) 8.3 7.6
and patients who received BSC alone. HR, hazard ratio. 7.2 6.9 6.4 6.2
6.0 5.3
5 4.9
4.0 4.9
3.9
3.8

34 days, respectively; HR, 0.658; 95% CI, 0.541 to 0.799; P ⬍ .001; Fig P .8630 .2406 .0015 .0365 .3876 .4937 .7796
3), with a significant reduction in HFSR occurrence by week 12 of 0 2 4 6 8 10 12 14
treatment. Exploratory Cox regression analysis showed that treatment
(HR, 0.621; 95% CI, 0.503 to 0.766; P ⬍ .001) and abnormal AST
Time (weeks)
No. at risk
concentration (HR, 1.949; 95% CI, 1.581 to 2.403; P ⫽ .001) were Urea 439 424 418 413 373 345 336 256
significant predictors of increased risk of HFSR over time. A Cox cream
BSC 432 415 414 408 372 341 332 249
model testing the interaction between treatment and AST on risk of
HFSR over time found similar treatment effects in patients with nor-
Fig 4. Hand-Foot Reaction Quality of Life outcomes. (A) Hand-foot skin
mal and abnormal AST (P ⫽ .0732). Of the 871 patients, 172 patients reaction (HFSR) symptoms and (B) HFSR daily activity over time in patients who
(19.7%), 85 in the UBC plus BSC group and 87 in the BSC-alone received urea-based cream plus best supportive care (BSC) and in patients who
group, could not be assessed for time to HFSR as a result of missing or received BSC alone. Scores expressed as mean ⫾ SE. P values were calculated
using the Wilcoxon rank sum test.
incorrect visits or lack of valid informed consent.
Treatment with UBC was associated with significant reductions
in risk of HFSR during the first 12 weeks of treatment (OR, 0.457; 95% Analysis of responses showed that the HFSR symptom score was
CI, 0.344 to 0.608; P ⬍ .001) and during the entire study period (OR, lower at each study visit from week 4 to week 14 in patients treated
0.490; 95% CI, 0.370 to 0.647; P ⬍ .001). These findings are consistent with UBC plus BSC versus BSC alone, indicating a lower symptom
with the significant reduction in risk of HFSR per unit time (HR, burden in the former group. In addition, the HFSR daily activity score
0.658; 95% CI, 0.541 to 0.799; P ⬍ .001). was lower at weeks 6 and 8 in patients receiving UBC plus BSC,
Twenty-six preselected/precoded AEs commonly associated indicating that HFSR had a reduced impact on these patients’ daily
with sorafenib were recorded at each visit. Except for HFSR, there activities (Fig 4). Analysis by RECIST criteria of the 670 patients
were no significant between-group differences in the incidence of evaluable for tumor assessment (333 patients in the UBC plus BSC
drug-related AEs (Table 3). Only HFSR and diarrhea occurred in group and 337 patients in the BSC-alone group) showed similar over-
more than 10% of patients. all response rates (11.1% v 10.1%, respectively; P ⫽ .6674) and disease
Assessment of the effects of UBC on treatment with sorafenib control rates (98.8% v 98.2%, respectively; P ⫽ .5350) at week 12.
showed that, in the UBC plus BSC and BSC-alone groups, the rates of
sorafenib dose reductions/interruptions as a result of HFSR (9.1% v
11.8%, respectively; P ⫽ .1937) and sorafenib discontinuations (1.1% DISCUSSION
v 0.7%, respectively; P ⫽ .4893) were similar.
At the initial visit, 424 (96.6%) of 439 patients in the UBC HFSR usually develops 2 to 4 weeks after initiation of treatment with a
plus BSC group and 415 (96.1%) of 432 patients in the BSC- TKI, including sorafenib, with the risk being highest at 5 weeks fol-
alone group completed HF-QoL questionnaires, making the rates lowed by a gradual decline.23,27 HFSR may be severe enough to require
of missing questionnaires at this visit 3.4% and 3.9%, respectively. TKI dose reduction and to seriously impair activities of daily liv-
At visits 2, 3, 4, 5, and 6, 418 (95.2%), 413 (94.1%), 374 (85.2%), ing.12,28,29 Despite relatively high HFSR rates, the management of
345 (78.6%), and 336 patients (76.5%), respectively, in the UBC HFSR has been reactive rather than proactive. Recent consensus rec-
plus BSC group and 413 (95.6%), 408 (94.4%), 372 (86.1%), 341 ommendations by an international, interdisciplinary expert panel30
(78.9%), and 332 patients (76.9%), respectively, in the BSC-alone included the use of over-the-counter and prescription-strength
group returned questionnaires. creams and moisturizers during treatment with TKIs.23 Anecdotal

898 © 2015 by American Society of Clinical Oncology JOURNAL OF CLINICAL ONCOLOGY

Downloaded from ascopubs.org by 200.92.202.106 on August 19, 2022 from 200.092.202.106

Copyright © 2022 American Society of Clinical Oncology. All rights reserved.
 Prophylactic Urea-Based Cream for Sorafenib HFSR in Advanced HCC

evidence has suggested that several prescription-strength topical
agents, including those containing 40% urea, 0.1% tazarotene, and 5%
fluorouracil, benefit patients with HFSR and other dermatologic con-
ditions.22,31,32 To our knowledge, however, no previous trial has sys-
tematically tested the ability of a topical agent, started at the initiation
of TKI treatment, to prevent and/or treat TKI-associated HFSR. Thus,
this randomized, open-label trial tested whether treatment with UBC
for 12 weeks had a prophylactic effect in a study of 871 patients with
advanced HCC starting sorafenib at 64 centers throughout China.
The incidence of sorafenib-related HFSR has been found to vary
in clinical trials of patients with advanced HCC. These differences may
be a result of differences in tumor type and starting dose33 and sex and
PS.27 The incidence of any grade HFSR in this study was higher than in
most other studies of sorafenib4,5,34 but was not highest overall.18 The
disparities between this trial and others may also have been a result of
the use of different HFSR grading systems and of increased awareness
of HFSR by clinicians and patients during treatment with sorafenib.
Importantly, however, the rates of any grade and grade ⱖ 2 HFSR were
significantly lower in the UBC plus BSC group than in the BSC-alone
group, with similar reductions in each grade of HFSR severity. More-
over, UBC significantly extended the median time to first occurrence
of HFSR, decreasing its risk through week 12 by 34%.
Although the mechanism underlying sorafenib-induced HFSR is
unknown, factors such as age, sex, ECOG PS, and laboratory values at
baseline have been assessed.27 Multivariable regression analyses indi-
cated that prophylactic use of UBC reduced HFSR incidence and
extended the time to first occurrence. In addition, these analyses
identified patients with abnormal AST as being at higher risk for
HFSR. Abnormal AST may reflect underlying liver disease in these
patients, which may interfere with liver metabolism of sorafenib and
enhance its plasma concentration. Although plasma concentrations of
sorafenib were not measured in this study, the benefits of prophylactic
UBC were similar in patients with higher (elevated AST) and lower
(normal AST) risk of HFSR.
However, UBC had little effect on sorafenib dose reductions,
interruptions, and discontinuations; overall response rate; and disease
control rate. This may have been a result of the short duration of the
trial or of the ability of patients in the BSC-alone group to switch to
UBC after the development of HFSR. The latter may have masked the
effects of UBC on sorafenib dose reductions, interruptions, and dis-
continuations, but not on the rates of any grade and grade ⱖ 2 HFSR
or time to first occurrence of HFSR. Moreover, UBC improved patient
HFSR-associated quality of life during sorafenib treatment, reducing
HFSR symptoms and their impact on daily activities.
Although prophylactic UBC reduced the incidence and severity
of HFSR, patients in the control arm were not allowed use of creams
before development of HFSR, preventing a determination of whether
the benefits of UBC were a result of the urea or of the cream itself.
Other limitations included the nonblinded nature of the study and the
evaluation of HFSR by treating physicians and not by central review of
photographs of the affected areas. Concurrent central review was not
possible because the symptoms were acute and required immediate
attention.20 Longer follow-up may determine whether prophylactic
use of UBC reduces HFSR rates and intensity, increasing drug
exposure and improving therapeutic outcomes. Further studies in
a blinded, placebo-controlled, randomized setting are warranted.
In conclusion, prophylactic UBC in patients starting sorafenib for
advanced HCC reduced the incidence and severity of HFSR, de-
layed the time to its first occurrence, and improved HFSR-
associated quality of life.
AUTHORS’ DISCLOSURES OF POTENTIAL CONFLICTS
OF INTEREST
Although all authors completed the disclosure declaration, the following
author(s) and/or an author’s immediate family member(s) indicated a
financial or other interest that is relevant to the subject matter under
consideration in this article. Certain relationships marked with a “U” are
those for which no compensation was received; those relationships marked
with a “C” were compensated. For a detailed description of the disclosure
categories, or for more information about ASCO’s conflict of interest policy,
please refer to the Author Disclosure Declaration and the Disclosures of
Potential Conflicts of Interest section in Information for Contributors.
Employment or Leadership Position: None Consultant or Advisory
Role: None Stock Ownership: None Honoraria: None Research
Funding: Sheng-Long Ye, Bayer Expert Testimony: None Patents,
Royalties, and Licenses: None Other Remuneration: None
AUTHOR CONTRIBUTIONS
Conception and design: ZhengGang Ren, WeiPing Zhou,
Sheng-Long Ye
Provision of study materials or patients: WeiZhu Yang, Sheng-Long Ye
Collection and assembly of data: KangShun Zhu, HaiYan Kang,
MinQiang Lu, ZengQiang Qu, LiGong Lu, TianQiang Song, WeiPing
Zhou, Hui Wang, WeiZhu Yang, Xuan Wang
Data analysis and interpretation: YongPing Yang, LeHua Shi, YuXian
Bai, XiaoFeng Guo
Manuscript writing: All authors
Final approval of manuscript: All authors

REFERENCES
1. Wilhelm SM, Carter C, Tang L, et al: BAY 43-9006
exhibits broad spectrum oral antitumour activity and
targets the RAF/MEK/ERK pathway and receptor ty-
rosine kinases involved in tumor progression and angio-
genesis. Cancer Res 64:7099-7109, 2004
2. Wilhelm SM, Adnane L, Newell P, et al:
Preclinical overview of sorafenib, a multikinase in-
hibitor that targets both Raf and VEGF and PDGF
receptor tyrosine kinase signaling. Mol Cancer Ther
7:3129-3140, 2008
3. Liu L, Cao Y, Chen C, et al: Sorafenib blocks
the RAF/MEK/ERK pathway, inhibits tumor angio-
genesis, and induces tumor cell apoptosis in hepa-
tocellular carcinoma model PLC/PRF/5. Cancer Res
66:11851-11858, 2006
4. Llovet JM, Ricci S, Mazzaferro V, et al:
Sorafenib in advanced hepatocellular carcinoma.
N Engl J Med 359:378-390, 2008
5. Cheng AL, Kang YK, Chen Z, et al: Efficacy and
safety of sorafenib in patients in the Asia-Pacific region
with advanced hepatocellular carcinoma: A phase III
randomised, double-blind, placebo-controlled trial. Lancet
Oncol 10:25-34, 2009
6. Bruix J, Sherman M, American Association
for the Study of Liver Diseases: Management of
hepatocellular carcinoma: An update. Hepatology
53:1020-1022, 2011
7. European Association for the Study of the
Liver, European Organisation for Research and
Treatment of Cancer: EASL-EORTC clinical practice
guidelines: Management of hepatocellular carci-
noma. J Hepatol 56:908-943, 2012
8. Verslype C, Rosmorduc O, Rougier P, et al:
Hepatocellular carcinoma: ESMO-ESDO clinical
practice guidelines for diagnosis, treatment and
follow-up. Ann Oncol 56:vii41-vii48, 2012 (suppl 7)
9. Chinese Anti-Cancer Association Society of Liver
Cancer, Chinese Society of Clinical Oncology, Chinese
Society of Hepatology Liver Cancer Study Group: The
expert consensus on the treatment standards for hepa-
tocellular carcinoma. Dig Dis Endosc 3:40-51, 2009
10. Ministry of Health of the People’s Republic of
China: The diagnosis and treatment guideline of
primary liver cancer (version 2011). Chin Clin Oncol
6:929-946, 2011

www.jco.org © 2015 by American Society of Clinical Oncology 899

Downloaded from ascopubs.org by 200.92.202.106 on August 19, 2022 from 200.092.202.106

Copyright © 2022 American Society of Clinical Oncology. All rights reserved.

11. Song P, Tobe RG, Inagaki Y, et al: The man-
agement of hepatocellular carcinoma around the
world: A comparison of guidelines from 2001 to
2011. Liver Int 32:1053-1063, 2012
12. Robert C, Soria JC, Spatz A, et al: Cutaneous
side-effects of kinase inhibitors and blocking anti-
bodies. Lancet Oncol 6:491-500, 2005
13. Beldner M, Jacobson M, Burges GE, et al:
Localized palmar-plantar epidermal hyperplasia: A
previously undefined dermatologic toxicity to
sorafenib. Oncologist 12:1178-1182, 2007
14. Lacouture ME, Reilly LM, Gerami P, et al:
Hand foot skin reaction in cancer patients treated
with the multikinase inhibitors sorafenib and
sunitinib. Ann Oncol 19:1955-1961, 2008
15. Yang CH, Lin WC, Chuang CK, et al: Hand-foot
skin reaction in patients treated with sorafenib: A
clinicopathological study of cutaneous manifesta-
tions due to multitargeted kinase inhibitor therapy.
Br J Dermatol 158:592-596, 2008
16. Zhang L, Zhou Q, Ma L, et al: Meta-analysis of
dermatological toxicities associated with sorafenib.
Clin Exp Dermatol 36:344-350, 2011
17. Ogasawara S, Kanai F, Obi S, et al: Safety and
tolerance of sorafenib in Japanese patients with
advanced hepatocellular carcinoma. Hepatol Int
5:850-856, 2011
18. Kudo M, Imanaka K, Chida N, et al: Phase III
study of sorafenib after transarterial chemoemboli-
sation in Japanese and Korean patients with unre-
sectable hepatocellular carcinoma. Eur J Cancer
47:2117-2127, 2011
900 © 2015 by American Society of Clinical Oncology
Downloaded from ascopubs.org by 200.92.202.106 on August 19, 2022 from 200.092.202.106
Copyright © 2022 American Society of Clinical Oncology. All rights reserved.
Ren et al
19. Porta C, Paglino C, Imarisio I, et al: Uncovering
Pandora’s vase: The growing problem of new toxic-
ities from novel anticancer agents—The case of
sorafenib and sunitinib. Clin Exp Med 7:127-134,
2007
20. Autier J, Escudier B, Wechsler J, et al: Pro-
spective study of the cutaneous adverse effects of
sorafenib, a novel multikinase inhibitor. Arch Derma-
tol 144:886-892, 2008
21. Nardone B, Hensley JR, Kulik L, et al: The
effect of hand-foot skin reaction associated with the
multikinase inhibitors sorafenib and sunitinib on
health-related quality of life. J Drugs Dermatol 11:
e61-e65, 2012
22. Hagemann I, Proksch E: Topical treatment by
urea reduces epidermal hyperproliferation and in-
duces differentiation in psoriasis. Acta Derm Vene-
reol 76:353-356, 1996
23. Lacouture ME, Wu S, Robert C, et al: Evolving
strategies for the management of hand-foot skin
reaction associated with the multitargeted kinase
inhibitors sorafenib and sunitinib. Oncologist 13:
1001-1011, 2008
24. Anderson R, Jatoi A, Robert C, et al: Search
for evidence-based approaches for the prevention
and palliation of hand-foot skin reaction (HFSR)
caused by the multikinase inhibitors (MKIs). Oncol-
ogist 14:291-302, 2009
25. Anderson RT, Rajagopalan R: Development and
validation of a quality of life instrument for cutaneous
diseases. J Am Acad Dermatol 37:41-50, 1997
26. SAS Institute: SAS/STAT(R) User’s Guide,
Version 8. Cary, NC, SAS Institute, 1999
■ ■ ■
27. Dranitsaris G, Vincent MD, Yu J, et al: Devel-
opment and validation of a prediction index for
hand-foot skin reaction in cancer patients receiving
sorafenib. Ann Oncol 23:2103-2108, 2012
28. Manchen E, Robert C, Porta C: Management
of tyrosine kinase inhibitor-induced hand-foot skin
reaction: Viewpoints from the medical oncologist,
dermatologist, and oncology nurse. J Support Oncol
9:13-23, 20113
29. Robert C, Mateus C, Spatz A, et al: Dermatologic
symptoms associated with the multikinase inhibitor
sorafenib. J Am Acad Dermatol 50:299-305, 2009
30. Wood LS, Lemont H, Jatoi A, et al: Practical
considerations in the management of hand-foot skin
reaction caused by multikinase inhibitors. Commu-
nity Oncol 7:23-29, 2010
31. Duvic M, Asano AT, Hager C, et al: The
pathogenesis of psoriasis and the mechanism of
action of tazarotene. J Am Acad Dermatol 39:S129-
S133, 1998
32. Scheithauer W, McKendrick J, Begbie S, et al:
Oral capecitabine as an alternative to i.v. 5-fluorouracil-
based adjuvant therapy for colon cancer: Safety results of
a randomized, phase III trial. Ann Oncol 14:1735-1743,
2003
33. Chu D, Lacouture ME, Fillos T, et al: Risk of
hand-foot skin reaction with sorafenib: A systematic
review and meta-analysis. Acta Oncol 47:176-186, 2008
34. Abou-Alfa GK, Schwartz L, Ricci S, et al:
Phase II study of sorafenib in patients with ad-
vanced hepatocellular carcinoma. J Clin Oncol 24:
4293-4300, 2006
JOURNAL OF CLINICAL ONCOLOGY
 Prophylactic Urea-Based Cream for Sorafenib HFSR in Advanced HCC

Acknowledgment
We acknowledge the contributions of all the other investigators (Appendix Table A1) and the patients in this trial and of BelMed Professional
Resources (New Rochelle, NY) for editorial support.

Appendix

Table A1. Investigators and Institutions Enrolling Patients Onto This Trial

Name of Investigator Institution

YuXian Bai Heilongjiang Provincial Cancer Hospital
JianQiang Cai Cancer Institute and Hospital Chinese Academy of Medical Sciences
LiPing Cao The Second Affiliated Hospital of Zhejiang University School of Medical
Hong Chen The Hospital of Armed Police Force
HuanWei Chen The First People’s Hospital of Fushan
Hui Chen Beijing Cancer Hospital
Jun Chen Jiangsu Cancer Hospital
PeiFeng Chen Zhejiang Provincial Hospital of Traditional Chinese Medicine
YanLing Chen Fujian Medical University Union Hospital
YuTang Chen Zhejiang Cancer Hospital
DianYuan Cui First Affiliated Hospital of Dalian Medical University
Sheng Dong ShangHai Changzheng Hospital
YongChong Dou Shengzhen People’s Hospital
Wen Fang Zhongda Hospital Southeast University
GuoSheng Feng GuangXi People’s Hospital
YuJia Gao Liaoning Cancer Hospital and Institute
Sheng Guan The First Affiliated Hospital of Zhengzhou University
WuHua Guo The Second Affiliated Hospital to Nanchang University
Zhi Guo Tianjin Medical University Cancer Institute and Hospital
ChunYi Hao Beijing Cancer Hospital
XiangDong Hua Liaoning Cancer Hospital and Institute
XingYu Huang Shanghai Sixth People’s Hospital
Feng Huo General Hospital of Guanzhou Military Command of People’s Liberation Army
ZhiXiang Jian Guangdong General Hospital
JiaJi Jiang The First Affiliated Hospital of Fujian Medical University
TianJun Jiang People’s Liberation Army 302 Hospital
HaiYan Kang People’s Liberation Army 301 Hospital
Bing Li The Second Affiliated Hospital of Zhejiang University School of Medical
Bing Li The First Affiliated Hospital Xiamen University
Bo Li West China Hospital Sichuan University
HaiLiang Li Henan Provincial Cancer Hospital
Huai Li Cancer Institute and Hospital Chinese Academy of Medical Sciences
WenTao Li Fudan University Shanghai Cancer Center
XiaoWu Li Southwest Hospital
Jun Liang The Affiliated Hospital of Qingdao University
LiJian Liang The First Affiliated Hospital Sun Yat-sen University
ZhengYin Liao West China Hospital Sichuan University
HaiLan Lin Fujian Provincial Cancer Hospital
FengYong Liu People’s Liberation Army 301 Hospital
JingFeng Liu The First Affiliated Hospital of Fujian Medical University
ZhaoYu Liu Shengjing Hospital of China Medical University
Zhe Liu People’s Liberation Army 301 Hospital
ZhenWen Liu People’s Liberation Army 302 Hospital
GuoLiang Lou Shanghai Changhai Hospital
LiGong Lu Guangdong General Hospital
MinQiang Lu The Third Affiliated Hospital Sun Yat-sen University
AiYing Ma People’s Liberation Army 455 Hospital
Rui Ma Liaoning Cancer Hospital and Institute
YiLong Ma Guangxi Medical School Affiliated Cancer Hospital
XiongYing Miao The Second Xiangya Hospital of Central South University
Wei Mou Southwest Hospital
(continued on following page)

www.jco.org © 2015 by American Society of Clinical Oncology

Downloaded from ascopubs.org by 200.92.202.106 on August 19, 2022 from 200.092.202.106

Copyright © 2022 American Society of Clinical Oncology. All rights reserved.
 Ren et al

Table A1. Investigators and Institutions Enrolling Patients Onto This Trial (continued)

Name of Investigator Institution

Lei Nie Hubei Cancer Hospital
YuDong Qiu Nanjing Drum Tower Hospital
ZenQiang Qu Eastern Hepatobiliary Surgery Hospital
JiaLing Shen Renji Hospital Affiliated to ShangHai Jiaotong University School of Medicine
LiJun Sheng Shangdong Cancer Hospital
LeHua Shi Eastern Hepatobiliary Surgery Hospital
Zheng Shi The First Affiliated Hospital of Fujian Medical University
GuangMing Shu Peking University People’s Hospital
YiJin Shu Zhejiang Provincial Hospital of Traditional Chinese Medicine
HongLi Song Tianjing First Central Hospital
TianQiang Song Tianjin Medical University Cancer Institute and Hospital
LiYing Sun Tianjing First Central Hospital
Tao Sun Liaoning Cancer Hospital and Institute
ZhenHua Tu The First Affiliated Hospital of Zhejiang University School of Medical
Hui Wang Jilin Provincial Cancer Hospital
TongShan Wang Jiangsu Province Hospital
Xuan Wang People’s Liberation Army 81 Hospital
YaoDong Wang Fujiang Provincial Hospital
JinShu Wu Hunan Provincial People’s Hospital
BaoCai Xing Beijing Cancer Hospital
AiBing Xu Nantong Tumor Hospital
GuoHui Xu The Second People’s Hospital of Sichuan
JianMing Xu People’s Liberation Army 307 Hospital
LinFeng Xu The Second Affiliated Hospital Sun Yat-sen University
WeiZhu Yang Fujian Medical University Union Hospital
WuWei Yang People’s Liberation Army 307 Hospital
YongPing Yang People’s Liberation Army 302 Hospital
Hua Ye The Second Affiliated Hospital Sun Yat-sen University
ShengLong Ye Zhongshan Hospital Fudan University
Tao Yin Hubei Cancer Hospital
JiaSheng Zheng Beijing YouAn Hospital Capital Medical University
SuiLiang Zhang Eastern Hepatobiliary Surgery Hospital
XiPing Zhang Zhejiang Cancer Hospital
GuangWen Zhou Ruijin Hospital Affiliated to ShangHai Jiaotong University School of Medicine
Shi Zhou Affiliated Hospital of Guiyang Medical College
WeiPing Zhou Eastern Hepatobiliary Surgery Hospital
XinWen Zhou Jiangxi Cancer Hospital
Yun Zhou Henan Provincial People’s Hospital
KangShun Zhu The Third Affiliated Hospital Sun Yat-sen University
Li Zhuang The First Affiliated Hospital of Zhejiang University School of Medical

Table A2. Grading of Hand-Foot Skin Reaction

Grade Description
1 Numbness, dysesthesia/paresthesia, tingling, painless swelling or erythema of the hands and/or feet and/or discomfort, which does not
disrupt normal activities
2 Painful erythema and swelling of the hands and/or feet and/or discomfort affecting the patient’s activities
3 Moist desquamation, ulceration, blistering, or severe pain of the hands and/or feet and/or severe discomfort that causes the patient to
be unable to work or perform activities of daily living

© 2015 by American Society of Clinical Oncology JOURNAL OF CLINICAL ONCOLOGY

Downloaded from ascopubs.org by 200.92.202.106 on August 19, 2022 from 200.092.202.106

Copyright © 2022 American Society of Clinical Oncology. All rights reserved.
 ORIGINAL CONTRIBUTION
Effect of Duloxetine on Pain, Function,
and Quality of Life Among Patients
With Chemotherapy-Induced
Painful Peripheral Neuropathy
A Randomized Clinical Trial
Ellen M. Lavoie Smith, PhD
Herbert Pang, PhD
Constance Cirrincione, MS
Stewart Fleishman, MD
Electra D. Paskett, PhD
Tim Ahles, PhD
Linda R. Bressler, PharmD
Camilo E. Fadul, MD
Chetaye Knox, BS
Nguyet Le-Lindqwister, MD
Paul B. Gilman, MD
Charles L. Shapiro, MD
for the Alliance for Clinical Trials in
Oncology
A
PPROXIMATELY 20% TO 40% OF
patients with cancer who re-
ceiveneurotoxicchemotherapy
(eg, taxanes, platinums, vinca
alkaloids, bortezomib) will develop pain-
ful chemotherapy-induced peripheral
neuropathy.1-3 Painful chemotherapy-
induced neuropathy can persist from
months to years beyond chemotherapy
completion,causingsignificantchallenges
for cancer survivors due to its negative in-
fluence on function and quality of life
(QOL).4-8 Chemotherapy–induced pe-
ripheral neuropathy is difficult to man-
age,andmostrandomizedcontrolledtrials
CME available online at
www.jamanetworkcme.com
and questions on p 1412.
Author Video Interview available at
www.jama.com.
©2013 American Medical Association. All rights reserved.
Downloaded From: https://jamanetwork.com/ Mexico | Access Provided by JAMA by Salvador Macias Diaz on 08/19/2022
Importance There are no known effective treatments for painful chemotherapy-
induced peripheral neuropathy.
Objective To determine the effect of duloxetine, 60 mg daily, on average pain severity.
Design, Setting, and Patients Randomized, double-blind, placebo-controlled cross-
over trial at 8 National Cancer Institute (NCI)–funded cooperative research networks
that enrolled 231 patients who were 25 years or older being treated at community
and academic settings between April 2008 and March 2011. Study follow-up was com-
pleted July 2012. Stratified by chemotherapeutic drug and comorbid pain risk, pa-
tients were randomized to receive either duloxetine followed by placebo or placebo
followed by duloxetine. Eligibility required that patients have grade 1 or higher sen-
sory neuropathy according to the NCI Common Terminology Criteria for Adverse Events
and at least 4 on a scale of 0 to 10, representing average chemotherapy-induced pain,
after paclitaxel, other taxane, or oxaliplatin treatment.
Interventions The initial treatment consisted of taking 1 capsule daily of either 30
mg of duloxetine or placebo for the first week and 2 capsules of either 30 mg of du-
loxetine or placebo daily for 4 additional weeks.
Main Outcome Measures The primary hypothesis was that duloxetine would be more
effective than placebo in decreasing chemotherapy-induced peripheral neuropathic pain.
Pain severity was assessed using the Brief Pain Inventory-Short Form “average pain” item
with 0 representing no pain and 10 representing as bad as can be imagined.
Results Individuals receiving duloxetine as their initial 5-week treatment reported a mean
decrease in average pain of 1.06 (95% CI, 0.72-1.40) vs 0.34 (95% CI, 0.01-0.66) among
those who received placebo (P=.003; effect size, 0.513). The observed mean difference
in the average pain score between duloxetine and placebo was 0.73 (95% CI, 0.26-
1.20). Fifty-nine percent of those initially receiving duloxetine vs 38% of those initially
receiving placebo reported decreased pain of any amount.
Conclusion and Relevance Among patients with painful chemotherapy-induced
peripheral neuropathy, the use of duloxetine compared with placebo for 5 weeks re-
sulted in a greater reduction in pain.
Trial Registration clinicaltrials.gov Identifier: NCT00489411
JAMA. 2013;309(13):1359-1367 www.jama.com
testing a variety of drugs with diverse serotonin and norepinephrine are key
mechanisms of action revealed no effec- neurotransmitters that suppress trans-
tive treatment.9 mission of painful peripheral stimuli by
There is mounting evidence that
serotonin and norepinephrine dual Author Affiliations are listed at the end of this article.
reuptake inhibitors are effective in treat- Corresponding Author: Ellen M. Lavoie Smith, PhD,
University of Michigan School of Nursing, Room 2151,
ing neuropathy-related pain.10 Both Ann Arbor, MI 48109 (ellenls@umich.edu).
JAMA, April 3, 2013—Vol 309, No. 13 1359
 DULOXETINE ON CHEMOTHERAPY-INDUCED PERIPHERAL NEUROPATHY PAIN
inhibiting input to the spinal dorsal
horn neurons.11 Several phase 3 stud-
ies show that duloxetine is an effec-
tive treatment for painful diabetic neu-
ropathy.12-15 Based on these trials, our
hypothesis was that duloxetine would
ameliorate chemotherapy-induced pe-
ripheral neuropathic pain as well. A ran-
domized phase 3 trial was conducted
to test this hypothesis.
METHODS
STUDY DESIGN
The Cancer and Leukemia Group B
(CALGB/Alliance) conducted a ran-
domized phase 3 double-blind, placebo-
controlled crossover trial to assess
whether 60 mg of duloxetine taken
orally once daily decreases the sever-
ity of chemotherapy-induced periph-
eral neuropathy (CALGB-170601,
NCT00489411). The primary hypoth-
esis was that duloxetine would be more
effective than placebo in decreasing the
average pain score after a 5-week treat-
ment period. Secondary aims were to
assess duloxetine’s effect on QOL and
function and on adverse events. The
study was approved by each site’s in-
stitutional review board, and partici-
pants provided signed informed con-
sent. Enrollment occurred between
April 2008 and March 2011. Study fol-
low-up was completed July 2012.
Patients
Using the National Cancer Institute’s
(NCI’s) Clinical Trials Support Unit
mechanism to facilitate accrual, par-
ticipants were recruited from 8 multi-
site NCI-funded cooperative research
networks, resulting in a geographi-
cally diverse population of patients dis-
tributed throughout the United States.
The diagnosis of chemotherapy-
induced peripheral neuropathy was de-
termined based on symptom history,
loss of deep tendon reflexes, or the pres-
ence of symmetrical stocking-glove
numbness or paresthesias beginning af-
ter neurotoxic chemotherapy. Eligible
patients were 25 years or older, had at
least grade 1 sensory pain based on the
NCI Common Terminology Criteria for
Adverse Events version 3.0 grading
1360 JAMA, April 3, 2013—Vol 309, No. 13
Downloaded From: https://jamanetwork.com/ Mexico | Access Provided by JAMA by Salvador Macias Diaz on 08/19/2022
scale, and reported 4 on a 10-point
scale, average neuropathic pain, for 3
or more months after completing che-
motherapy. Patients with any cancer di-
agnosis or stage were potentially eli-
gible. To diminish the likelihood that
symptoms would spontaneously re-
solve over the course of the study,
efficacy data were obtained over 5
weeks.
Initially, patients who had received
paclitaxel or oxaliplatin could partici-
pate, but eligibility was later ex-
panded to allow prior treatment with
single-agent docetaxel, nanoparticle al-
bumin–bound paclitaxel, or cisplatin.
Prior or ongoing treatment with other
neurotoxic chemotherapeutic agents
was not allowed. Participants with a
documented medical history of neu-
ropathy from any type of nerve com-
pression (eg, carpal or tarsal tunnel syn-
drome, radiculopathy, spinal stenosis,
brachial plexopathy), leptomeningeal
carcinomatosis, severe depression, sui-
cidal ideation, bipolar disease, alcohol
abuse, a major eating disorder, and
markedly abnormal renal or liver func-
tion tests were ineligible.
Despite scant evidence supporting
the association between certain comor-
bid illnesses and the risk of develop-
ing severe pain,9,16 patients with diabe-
tes mellitus and peripheral vascular
disease, whose pain was thought to be
from chemotherapy-induced periph-
eral neuropathy, were eligible but were
defined as high risk. We controlled for
comorbid illness as a potential con-
founder by assigning equal numbers of
high-risk patients to each treatment
group. Concurrent use of other drugs
known to influence serotonin levels was
not allowed. Concomitant use of se-
lected analgesics was allowed (eg, opi-
oids, acetaminophen, aspirin, nonste-
roidal anti-inflammatory drugs), but
only patients receiving stable doses in
the 2 weeks before randomization could
participate: (1) no new analgesics were
added, (2) no analgesics were discon-
tinued, and (3) the weekly 24-hour total
analgesic dose did not fluctuate up or
down by more than 10% in the 2 weeks
before study registration.
©2013 American Medical Association. All rights reserved.
Intervention
Eligible patients were randomized
using a 1:1 allocation ratio to either
group A or group B. In this crossover
design, group A received 60 mg of
duloxetine daily during the initial
treatment period and placebo at
crossover period. Group B received
placebo as initial treatment and
duloxetine as the crossover treat-
ment. Randomization, provided by
the CALGB/Alliance Statistical Cen-
ter, was stratified by neurotoxic drug
class (taxanes vs platinums) and by
pain risk (high risk vs no risk). A
computer-generated kit number was
used to order the blinded study
drug from a distribution center. Drug
labels were applied to the capsule
bottles at the distribution center
before being mailed to study sites;
thus, all patients and personnel were
blinded to the treatment assignment.
The initial (weeks 1-5) and cross-
over (weeks 8-12) treatment periods
each consisted of receiving 1 capsule of
either placebo or 30 mg of duloxetine
for the first week and 2 capsules of
either placebo or 30 mg of duloxetine
for 4 weeks. The initial 5-week period
was followed by a 2-week washout pe-
riod for a total study duration of 14
weeks (FIGURE 1).
Data Collection and Instruments
Patient-reported pain severity and func-
tional interference was assessed weekly
using the well-validated Brief Pain In-
ventory Short Form (BPI-SF).17-19 The
BPI-SF contains 4 items assessing av-
erage, worst, least, and immediate pain
severity in the last 24 hours. Pain se-
verity items are scored using an 11-
point numeric rating scale (0, no pain;
10, pain as bad as you can imagine). The
BPI-SF worse pain severity item has
been shown to be reliable and valid for
use as a single item.20 However, we
chose average pain severity as our
primary outcome measure based on
recommendations from the Initiative
on Methods, Measurement, and
Pain Assessment in Clinical Trials
(IMMPACT).21 In addition, average pain
has been defined as the primary out-

come measure in numerous dulox-
etine phase 3 studies involving pa-
tients with peripheral and central
neuropathic pain conditions due to dia-
betic- and oxaliplatin-induced neu-
ropathy, fibromyalgia, and osteoarthri-
tis13-15,22-30 and thereby was chosen to
facilitate comparison of our findings
across similar studies. The minimal clini-
cally important difference in pain sever-
ity was defined for the current study as
a 0.98 difference in mean average pain
severity between the duloxetine and
placebo groups. Using an accepted
method for assessing the influence of
pain on function,13-15,26-29 7 BPI-SF items
were used to quantify the degree to
which pain interfered with daily activi-
ties or function (0, does not interfere;
10, completely interferes). The 7 items
were summed to obtain a total inter-
ference score. Patient-reported QOL
was assessed using the Functional As-
sessment of Cancer Treatment, Gyne-
cologic Oncology Group Neurotoxic-
ity (FACT/GOG-Ntx) subscale on day
1 of weeks 1, 6, 8, and 13. The assess-
ment’s strong psychometric proper-
ties have been previously demon-
strated.31-33 The instrument contains 11
questions assessing numbness, tin-
gling, and discomfort in the hands or
feet; difficulty hearing; tinnitus; joint
pain or muscle cramps; weakness; or
trouble walking, buttoning buttons, or
feeling small shapes when placed in the
hand. Items are scored from 0 to 4 (0,
not at all; 4, very much) and summed
(total score range, 0-44).33 Because there
are no published data defining a cut
point for determining a clinically im-
portant change in the score, we de-
fined a 2- to 3-point change as a clini-
cally meaningful improvement in QOL
per published recommendations spe-
cific to similar measures.34-37
Using the NCI CTCAE version 3.0,
adverse events were reported weekly
and graded on a 0 to 4 scale (0, nor-
mal; 4, life-threatening) 38 as was base-
line sensory chemotherapy-induced pe-
ripheral neuropathy (0, normal; 1,
asymptomatic, weakness on physical
examination, loss of reflexes, or pares-
thesias not interfering with function; 2,
©2013 American Medical Association. All rights reserved.
Downloaded From: https://jamanetwork.com/ Mexico | Access Provided by JAMA by Salvador Macias Diaz on 08/19/2022
DULOXETINE ON CHEMOTHERAPY-INDUCED PERIPHERAL NEUROPATHY PAIN
Figure 1. Patient Flow Chart of Initial and Crossover Periods a
231 Patients enrolled
231 Randomized
Group A
115 Randomized to receive duloxetine followed by
crossover to placebo
109 Received duloxetine as randomized
6 Withdrew consent prior to receiving the
intervention
88 Completed initial intervention
21 Discontinued duloxetine early
12 Toxic effects
5 Withdrew consent
1 Other
3 Unknown
87 Included in primary analysis
1 Excluded (incomplete data)
85 Crossed over and received placebo
74 Completed crossover intervention
11 Discontinued intervention early
2 Toxic effects
4 Withdrew consent
1 Received alternate treatment
4 Other
67 Included in crossover analysis
7 Excluded (incomplete data)
a The number screened and the number offered participation but declined was not captured, nor were the
reasons captured for why patients did not participate in the crossover treatment period.
weakness and sensory alterations in-
terfering with function; 3, weakness and
sensory alterations interfering with ac-
tivities of daily living or requiring brac-
ing or assistive devices; and 4, life
threatening, paralysis, or disabling).
Statistical Analyses
Statistical analyses were performed by
CALGB/Alliance statisticians. The pri-
mary study end point was the change
from start to end of the initial treat-
ment period (week 1 to week 5, mea-
sured on day 1 of weeks 1 and 6) in av-
erage pain based on the BPI-SF average
pain severity item. The comparison of
interest was the difference between the
2 treatment groups in pain change.
With 232 patients (assuming 20% at-
trition), the study had 90% power (2-
sided ␣ of .05) to detect a 0.98-point
change between the 2 groups assum-
ing a standard error of 0.31. The tar-
get difference of 0.98 is consistent with
Group B
116 Randomized to receive placebo followed by
crossover to duloxetine
111 Received placebo as randomized
5 Withdrew consent prior to receiving the
intervention
99 Completed initial intervention
12 Discontinued placebo early
1 Toxic effects
6 Withdrew consent
2 Disease progression
3 Unknown
94 Included in primary analysis
5 Excluded (incomplete data)
93 Crossed over and received duloxetine
81 Completed crossover intervention
12 Discontinued intervention early
5 Toxic effects
2 Withdrew consent
2 Disease progression
3 Other
74 Included in crossover analysis
7 Excluded (incomplete data)
diabetic research.30 We calculated the
proportion of patients experiencing any
decrease in pain. Also, using another ac-
cepted approach for assessing clinical
significance, 13-15,22-24,26,39-41 we con-
ducted an exploratory responder analy-
sis based on the proportion of patients
in both groups who experienced a 30%
(or 50%) decrease in pain severity. Rela-
tive risk-benefit was the proportion of
patients with 30% (or 50%) pain re-
duction in the duloxetine group rela-
tive to that in the placebo group. It was
calculated from contingency tables. We
also conducted an exploratory sub-
group analyses based on chemo-
therapy class.
Secondary end points were change in
chemotherapy-induced peripheral neu-
ropathy–related QOL, measured by the
total score of the FACT/GOG-Ntx, and
degree of pain-related functional inter-
ference based on the BPI-SF interfer-
ence score. Changes attributable to ini-
JAMA, April 3, 2013—Vol 309, No. 13 1361
 DULOXETINE ON CHEMOTHERAPY-INDUCED PERIPHERAL NEUROPATHY PAIN

sion 9.2 (SAS Institute Inc). Data qual-

Table. Patient Demographics and Baseline Features
ity was ensured by review of data by the
No. (%)
of Participants CALGB/Alliance Statistical Center and
by the study chairperson. The study un-
Group A Group B Total P
Characteristics (n = 109) (n = 111) (n = 220) Value derwent standard biannual monitor-
Demographics ing by the CALGB/Alliance data and
Age, y safety monitoring board with 1 formal
30-39 2 (2) 1 (1) 3 (1)
40-49 16 (15) 23 (21) 39 (18)
interim efficacy analysis resulting in
50-59 43 (39) 39 (35) 82 (37) study continuation.
.70 a
60-69 30 (28) 29 (26) 59 (27)
RESULTS
ⱖ70 18 (17) 19 (17) 37 (17)
Mean (SD) 60 (10.4) 59 (10.6) 59 (10.5) Patients
Sex Patient disposition for the initial treat-
Men 38 (35) 44 (40) 82 (37)
.46 ment period is illustrated in Figure 1.
Women 71 (65) 67 (60) 138 (63)
Of the 231 patients recruited to the
Race
White 91 (83) 87 (78) 178 (81) study, 115 were allocated to group A
Black 14 (13) 17 (15) 31 (14) .50 (duloxetine first, placebo second), and
Other 4 (4) 5 (5) 9 (4) 116 to group B (placebo first, dulox-
Not reported 0 (0) 2 (2) 2 (1) etine second). Eleven patients never re-
Stratification factors ceived treatment, leaving 220 treated
Neurotoxic agent
Paclitaxel 44 (40) 43 (39) 87 (40)
patients. The dropout rate due to ad-
Oxaliplatin 63 (58) 66 (59) 129 (59) N/A
verse events in the duloxetine-first
Other taxane 2 (2) 2 (2) 4 (2) group was 11% vs 1% in the placebo-
Other platinum 0 (0) 0 (0) 0 (0) first group (P⬍.001). Despite using an
High risk of CIPN intent-to-treat analysis approach, 6 pa-
No 46 (42) 51 (46) 97 (44) tients (1 in group A; 5 in group B) dur-
N/A
Yes 63 (58) 60 (54) 123 (56)
ing the initial treatment period were ex-
Disease-related features
Primary disease cluded from the primary analysis
Breast 41 (38) 42 (38) 83 (38) because they provided no data at all.
GI 62 (57) 62 (56) 124 (56) This resulted in a 19% dropout rate.
Both breast & GI tract 0 (0) 1 (1) 1 (0) .99 b Patient characteristics are de-
Genitourinary 4 (4) 4 (4) 8 (3) scribed in the TABLE. Both groups were
Other 2 (2) 1 (1) 3 (1) similar at baseline, except for the mean
Missing 0 (0) 1 (1) 1 (0)
(SD) pain score of 6.1 (1.7) in group
(continued)
A, duloxetine first, and 5.6 (1.6) in
group B, placebo first (P=.02).
tial treatment were defined as the equations were used to determine
difference between the week 1 and week whether there was a treatment effect Pain (Primary Outcome)
5 scores (measured on day 1 of weeks when combining data from both the ini- At the end of the initial treatment pe-
1 and 6). Changes for crossover treat- tial and crossover periods. Multiple riod, patients in the duloxetine-first
ment used week 8 and week 12 scores imputation and pattern-mixture model group reported a larger decrease in av-
(measured on day 1 of weeks 8 and 13). were used to evaluate the pattern and erage pain (mean change score, 1.06;
To test for a group effect during the potential influence of missing values for 95% CI, 0.72-1.40) than those in the
initial treatment period on the pri- the primary end point (eAppendix placebo-first group (mean change score,
mary and secondary end points, we used available at www.http://www.jama 0.34; 95% CI, 0.01-0.66; P = .003;
3 separate models of analysis of cova- .com). To univariately compare treat- FIGURE 2). The effect size attributed to
riance, each stratified by neurotoxic ment groups, the Wilcoxon rank test duloxetine was moderately large at
agent and risk of painful chemotherapy- was used for continuous variables and 0.513.42 The observed mean differ-
induced peripheral neuropathy, and the ␹2 test for proportions. The 95% ence in the average pain score between
including the baseline measure of the confidence intervals for proportions the duloxetine-first and placebo-first
corresponding end point. Least square used exact binomial methods. Analy- groups was 0.73 (95% CI, 0.26-1.20).
means and their 95% confidence inter- ses included only patients who began Results of the sensitivity analysis tak-
vals were taken from analysis of cova- protocol therapy. Statistical analyses ing missing data into consideration are
riance models. Additionally, for the pri- with 2-sided significance threshold of consistent with the primary findings of
mary end point, generalized estimating P⬍.05 were performed using SAS ver- the trial (eTables 1 and 2 available at
1362 JAMA, April 3, 2013—Vol 309, No. 13 ©2013 American Medical Association. All rights reserved.

Downloaded From: https://jamanetwork.com/ Mexico | Access Provided by JAMA by Salvador Macias Diaz on 08/19/2022
 DULOXETINE ON CHEMOTHERAPY-INDUCED PERIPHERAL NEUROPATHY PAIN

http://www.jama.com): multiple im-

Table. Patient Demographics and Baseline Features (continued)
putation (P = .002) and pattern-
No. (%)
mixture model control–based imputa- of Participants
tion (P=.004). The primary analysis P Group A Group B Total P
value lies between the slightly more lib- Characteristics (n = 109) (n = 111) (n = 220) Value
eral multiple imputation approach and Disease-related features
Disease stage
the more conservative pattern- Early, I-II 38 (44) 38 (41) 76 (35)
mixture model. A post hoc power cal- III 34 (39) 42 (45) 76 (35) .69
culation is provided in the eAppendix. Metastatic 14 (16) 12 (13) 26 (12)
Of the patients treated with dulox- Missing 1 (1) 2 (2) 3 (1)
etine first, 59% reported any decrease Concurrent medications use
in pain vs 38% of patients treated with No 68 (62) 58 (52) 126 (57)
placebo first. Thirty percent of dulox- Yes 31 (28) 43 (39) 74 (34) .10
etine-treated patients reported no Missing 10 (9) 10 (9) 20 (9)
change in pain and 10% reported in- Performance status
0 60 (55) 60 (54) 120 (55)
creased pain. Based on the explor-
1 44 (40) 47 (42) 91 (41)
atory responder analysis, the propor- .78
2 4 (4) 2 (2) 6 (3)
tion of patients achieving various levels
3 1 (1) 1 (1) 2 (1)
of pain reduction is illustrated in
Missing 0 (0) 1 (1) 1 (0)
FIGURE 3. Compared with placebo, the
Pretreatment sensory neuropathy grade
relative risk of experiencing a 30% pain 1, Asymptomatic 1 (1) 2 (2) 3 (1)
reduction with duloxetine was 1.96 2, Moderate 77 (71) 84 (76) 161 (73)
(95% CI, 1.15-3.35) and experiencing .47
3, Severe 31 (28) 24 (22) 55 (25)
a 50% pain reduction was 2.43 (95% CI, 4, Life-threatening 0 (0) 0 (0) 0 (0)
1.11-5.30; eTable 3). Missing 0 (0) 1 (1) 1 (0)
Although the study was powered to Pretreatment pain score
⬍4 2 (2) 2 (2) 4 (2)
detect differences between treatments
4-5 44 (40) 58 (52) 102 (46)
as main effects only, in exploratory
analyses, we examined the potential in- 6-7 39 (36) 35 (32) 74 (34) .02 a
8-10 24 (22) 15 (14) 39 (18)
teraction between treatment group and
Missing 0 (0) 1 (1) 1 (0)
chemotherapy class. Results sug-
Mean (SD) 6.1 (1.7) 5.6 (1.6) 5.8 (1.7)
gested that patients who received plati-
Abbreviations: CIPN, chemotherapy-induced peripheral neuropathy; GI, gastrointestinal; N/A, not applicable (comparative
nums (oxaliplatin) experienced more testing is not applicable because these are stratification variables).
a Tested as a continuous variable.
benefit from duloxetine than those who b When testing for differences between groups based on primary disease, the groups were collapsed into 3 categories
received taxanes (P = .13). The ob- (breast, gastrointestinal, or other).

Figure 2. Duloxetine and Placebo Effects on Average Pain Severity During the Initial and Crossover Treatment Periods

Initial treatment period Crossover treatment period

6.5 6.5
Duloxetine first (group A) Placebo second (group A)
6.0 Placebo first (group B) 6.0 Duloxetine second (group B)
Mean Pain Score

Mean Pain Score

5.5 5.5

5.0 5.0

4.5 4.5

4.0 4.0

3.5 3.5

3.0 3.0
1 2 3 4 5 6 1 2 3 4 5 6
Week (Washout) Week
No. patients No. patients
Duloxetine first (group A) 87 86 86 87 85 87 Placebo second (group A) 67 65 65 66 63 67
Placebo first (group B) 94 92 91 91 91 94 Duloxetine second (group B) 74 68 71 69 66 74

The mean average pain score was measured on the first day of each week in the initial and crossover treatments periods. Day 1 of the first week begins the initial
treatment period. Day 1 of week 6 begins the washout period when patients received 1 capsule of duloxetine or placebo. Patients took no drug during week 7. Error
bars represent 95% CIs.

©2013 American Medical Association. All rights reserved. JAMA, April 3, 2013—Vol 309, No. 13 1363

Downloaded From: https://jamanetwork.com/ Mexico | Access Provided by JAMA by Salvador Macias Diaz on 08/19/2022
 DULOXETINE ON CHEMOTHERAPY-INDUCED PERIPHERAL NEUROPATHY PAIN
Figure 3. Percent Decrease in Pain Score Due to Duloxetine vs Placebo
100
90
80
Percentage of Patients
70
60
50
40
30
20
10
0
≥0
The plot shows the proportion of patients achieving various levels of pain reduction at the completion of the
initial treatment period.
served mean difference in platinum-
related average pain score between the
duloxetine and placebo treatment was
1.06 (95% CI, 0.48 to 1.63) vs 0.19
(95% CI, ⫺0.61 to 0.98) for patients
treated with taxane. Results of the ex-
ploratory responder analysis revealed
that compared with placebo, the rela-
tive risk of experiencing a reduction in
pain by 30% with duloxetine among pa-
tients treated with platinum was 3.05
(95% CI, 1.49-6.27) and a reduction by
50% was 3.78 (95% CI, 1.32-10.84;
eTables 4-6 available at http://www
.jama.com). However, compared with
placebo, the relative risk of experienc-
ing a 30% pain reduction with dulox-
etine among patients treated with tax-
ane was 0.97 (95% CI, 0.41-2.32) and
a 50% pain reduction was 1.22 (95% CI,
0.35-4.18), neither of which was sta-
tistically significant. There was no dif-
ference in duloxetine efficacy based on
risk of developing painful chemo-
therapy-induced peripheral neuropathy.
We also concurrently evaluated
changes in pain severity during the
crossover treatment period. After ad-
justing for the study stratifiers (che-
motherapy class and pain risk), there
was a statistically significant treat-
ment effect on change in pain score
(P⬍ .001), but an order effect was not
significant (P = .43). An additional
generalized estimating equation model
with carryover effect was performed.
1364 JAMA, April 3, 2013—Vol 309, No. 13
Downloaded From: https://jamanetwork.com/ Mexico | Access Provided by JAMA by Salvador Macias Diaz on 08/19/2022
Duloxetine (n = 87)
Placebo (n = 94)
≥10 ≥20 ≥30 ≥40 ≥50 ≥60 ≥70 ≥80 ≥90 100
Pain Reduction From Baseline to End Point
of Initial Treatment Period, %
The treatment remained significant
(P = .002), whereas the period (P =.55)
and carryover (P=.95) effects were not.
The change in mean pain score during
the crossover treatment period for
group A (placebo second) was 0.41
(95% CI, 0.06-0.89) and for group B
(duloxetine second) was 1.42 (95% CI,
0.97-1.87). The mean difference be-
tween the 2 groups in mean change
score during the crossover period was
1.01 (95% CI, 0.36-1.65).
Secondary Outcome
Pain Interference With Daily Function.
At the end of the initial treatment pe-
riod, when compared with placebo, pa-
tients treated with duloxetine first re-
ported a greater decrease in the amount
that pain had interfered with daily func-
tioning (P=.01; eFigure 1). The change
in mean interference score for patients
treated with duloxetine first was 7.9
(95% CI, 5.4-10.5) vs 3.5 (95% CI, 1.1-
5.9) for patients treated with placebo
first. The mean difference between the
2 groups in mean change score was 4.40
(95% CI, 0.93-7.88).
Quality of Life. Pain-related QOL im-
proved to a greater degree for those
treated with duloxetine during the ini-
tial treatment than for those treated with
placebo. The mean change in the FACT/
GOG-Ntx total score was 2.44 (95% CI,
0.43-4.45) for patients treated with du-
loxetine first vs 0.87 (95% CI, 1.09-
©2013 American Medical Association. All rights reserved.
2.82) for patients treated with placebo
first (P=.03). The mean difference be-
tween the 2 groups in mean change
score was 1.58 (95% CI, 0.15-3.00;
P=.03).
Adverse Effects. No hematologic or
grade 4 (moderately severe) or 5 (se-
vere) adverse events were reported. In
the initial treatment period, 16% treated
with duloxetine and 27% treated with
placebo reported grade 2 (mild) and 7%
treated with duloxetine and 3% treated
with placebo reported grade 3 (mod-
erate) nonhematologic adverse events.
Fatigue (7%), insomnia (5%), and nau-
sea (5%) were the most common ad-
verse effects reported by patients treated
by duloxetine, whereas somnolence
(8%), insomnia (7%), and fatigue (5%)
were the most common adverse ef-
fects reported by patients treated with
placebo (eTable 7 available at http:
//www.jama.com).
Other Findings
Nonpainful Symptoms. After the ini-
tial treatment period, 41% of patients
treated with duloxetine reported a de-
crease in numbness and tingling in the
feet (95% CI, 31%- 52%) vs 23% of pa-
tients treated with placebo (95% CI, 15%-
33%). The trend held through the cross-
over period with 41% of patients treated
with duloxetine (95% CI, 31%-53%) vs
21% of patients treated with placebo
(95% CI, 13%-32%). The proportion of
patients with improved hand numb-
ness and tingling at the end of the ini-
tial treatment period was similar among
those treated with duloxetine (36%) and
those treated with placebo (34%).
Ancillary Analgesics. Compared
with group A (duloxetine first), a higher
proportion of those in group B (pla-
cebo first) were taking concomitant
medications at the start (43% of pa-
tients in group B vs 31% in group A)
and at the end of the initial treatment
period (36% in group B vs 29% in group
A).Twenty-seven percent of patients
who started out taking concomitant
medications in the group A discontin-
ued all medications by the end of the
initial treatment period compared with
19% of patients in group B.

COMMENT
Treatment of painful chemotherapy-
induced peripheral neuropathy contin-
ues to be a challenge because most drugs
tested to date have fallen short of pro-
viding adequate pain relief.43-47 To our
knowledge, the current study is the first
large phase 3 trial to elucidate an effec-
tive intervention for painful chemo-
therapy-induce peripheral neuropathy
caused by platinum and taxane agents
(mainly paclitaxel or oxaliplatin). Dur-
ing initial treatment, the mean differ-
ence between the 2 groups of the change
in average pain score was 0.73 (P=.003),
which compares favorably to mean dif-
ferences in average pain scores (range,
0.60-0.98) observed in patients receiv-
ing duloxetine for US Food and Drug
Administration–approved indications
for painful diabetic neuropathy, fibro-
myalgia, and osteoarthritis (eTable 8
available at www.http://www.jama
.com).13,26,30 The observed mean differ-
ence in the average pain score between
the duloxetine and placebo groups in pa-
tients treated with platinum was larger
than results reported by Goldstein and
colleagues13 (FIGURE 4). However, du-
loxetine-related clinically meaningful
improvement in other painful condi-
tions may not be directly comparable
with painful chemotherapy-induced pe-
ripheral neuropathy.
In addition to the magnitude of the im-
provement, several other factors should
be considered when judging clinical sig-
nificance, such as the treatment effect
size.39 Our results revealed a moderately
large treatment effect size (0.513). Based
on the IMMPACT recommendations,
clinical meaningfulness is also based on
the results of a responder analysis, spe-
cifically the proportion of patients expe-
riencing a 30% or a 50% improvement in
painseverity.21,39,48 A10%to20%decrease
in pain severity is considered to represent
a minimal clinically important change, a
30% change represents a moderately im-
portant improvement and a 50% change
represents a substantially important im-
provement.21,39,48 Duringinitialtreatment,
the mean change in average pain score re-
portedbypatientstreatedwithduloxetine
inthecurrentstudywas1.06,animprove-
©2013 American Medical Association. All rights reserved.
Downloaded From: https://jamanetwork.com/ Mexico | Access Provided by JAMA by Salvador Macias Diaz on 08/19/2022
DULOXETINE ON CHEMOTHERAPY-INDUCED PERIPHERAL NEUROPATHY PAIN
Figure 4. Comparison of the Mean Differences in Average Pain Scores Across Duloxetine
Chronic Pain Studies
No. of
Source Patients
Goldstein et al,13 2005 176
Russell et al,30 2008 294
Chappell et al,26 2011 225
Current study 181
Neurotoxic agents, current study
Platinums 106
Taxanes 75
Duloxetine was compared against placebo in studies involving patients with diabetic neuropathy, Goldstein et
al; fibromyalgia, Russell et al; osteoarthritis, Chappell et al; and chemotherapy-induced peripheral neuropa-
thy, the current study, including a breakdown by neurotoxic agent that patients in the current study took.
ment of approximately 10% that is con-
sistent with the IMMPACT definition of
a minimal clinically important difference.
In addition, results of the exploratory re-
sponder analysis suggest that the relative
risk of experiencing a 30% and 50% im-
provement in pain severity statistically fa-
vored duloxetine.
Other factors to consider when judg-
ing clinical significance include how
quickly the drug takes effect, tolerabil-
ity, and the drug’s influence on other ef-
ficacy end points such as function and
QOL.39 In the current study, pain scores
decreased in patients treated with du-
loxetine relatively quickly, within the first
week of therapy (Figure 2). Consistent
with our results (eTable 7), several stud-
ies show that duloxetine is safe and well-
tolerated.13,14,26,30,49-51 Furthermore, du-
loxetine is associated with improved
function and QOL. Therefore, after con-
sidering the many factors in addition to
the magnitude of improvement in pain
scores, study results strongly suggest that
duloxetine treatment is associated with
a clinically meaningful improvement in
chemotherapy-induced peripheral neu-
ropathic pain.
Resultsfromourexploratorysubgroup
analysis lend support to the premise that
differences in pathophysiologic mecha-
nisms may help to explain duloxetine re-
sponse rate variations across neuropathic
painconditions.Justasresponseratesmay
vary when duloxetine is used to treat dia-
betic vs chemotherapy-induced periph-
eral neuropathy due to differences in
Mean Difference in Mean Favors Favors
Pain Scores (95% CI) Placebo Duloxetine
0.98 (0.37 to 1.60)
0.60 (0.05 to 1.16)
0.79 (0.26 to 1.32)
0.73 (0.26 to 1.20)
1.06 (0.48 to 1.63)
0.19 (–0.61 to 0.98)
–2 –1 0 1 2
Mean Difference (95% CI)
nerve injury mechanisms,16,52-55 the
mechanisms of taxane- vs platinum-
induced peripheral nerve injury are quite
different,16 possibly explaining why pa-
tients treated with platinum reported less
pain in the current study.
The current trial has several strengths
and limitations. The strengths include the
prospective, randomized, placebo-
controlled trial design and the geographi-
cally diverse sample. Regarding limita-
tions, first, there was an imbalance in the
dropout rate due to adverse effects in pa-
tients treated with duloxetine vs pla-
cebo (11% vs 1%, respectively), despite
similar adverse effect rates in both
groups. One reason for this differential
in the dropout rate may be the higher
proportion of grade 3 adverse events re-
ported by patients treated with dulox-
etine. These patients may have been able
to guess which drug they were taking,
and those experiencing no or minimal
pain relief may have dropped out.
Another potential study limitation has
to do with how baseline pain was deter-
mined. Oncology providers commonly
use the NCI CTCAE, or other similar
grading scales, to guide a history and
physical examination focused on
chemotherapy-induced peripheral neu-
ropathy and to grade its severity.9,16
Therefore, we relied on standard
CTCAE-guided practices for determin-
ing severity at baseline. We did not spe-
cifically train the examiners regarding
CTCAE use because its grading is deeply
embedded into oncology practice. Of
JAMA, April 3, 2013—Vol 309, No. 13 1365
 DULOXETINE ON CHEMOTHERAPY-INDUCED PERIPHERAL NEUROPATHY PAIN
note, despite its everyday use in oncol-
ogy clinical settings, the CTCAE is
known for its suboptimal interrater reli-
ability and poor sensitivity to detect
subtle changes.32,56-60 As such, use of the
CTCAE could have resulted in misdi-
agnosis. Despite its limitations, CTCAE
use is consistent with community and
academic clinical practice,9 and thus the
generalizability of our findings is
enhanced. Other limitations are that
changes in concurrent ancillary analge-
sic dosage were not assessed, study find-
ings may not be applicable to patients
with painful chemotherapy-induced
peripheral neuropathy caused by other
neurotoxic agents, and the study did not
address long-term duloxetine treat-
ment (beyond 5 weeks).
Despiteduloxetine’sacceptableadverse
effect profile, the risk of duloxetine-drug
interactions should not be overlooked.
Duloxetine should not be used with other
drugs that inhibit serotonin reuptake due
to the associated increased risk of sero-
tonin syndrome.61 Also, because dulox-
etine is a moderate cytochrome (CYP)
P450 2D6 enzyme-inhibiter, coadminis-
tration with CYP P450 2D6 substrates
can lead to increased substrate drug
serum concentrations and associated
toxicities.62-65 Concurrent use of dulox-
etine with warfarin, nonsteroidal anti-
inflammatory drugs, or both may also
increase bleeding risk.61,66 Lastly, if du-
loxetineandtamoxifenaretakentogether,
duloxetine-induced CYP P450 2D6 en-
zyme inhibition could inhibit tamoxifen
conversion to its active metabolite, en-
doxifen.61,63,67,68
In conclusion, 5 weeks of duloxetine
treatment was associated with a statisti-
cally and clinically significant improve-
ment in pain compared with placebo. Ex-
ploratory analyses raise the possibility
that duloxetine may work better for
oxaliplatin-induced rather than taxane-
induced painful chemotherapy-induced
peripheral neuropathy.
Author Affiliations: University of Michigan School of
Nursing, Ann Arbor, Michigan (Dr Smith); Alliance Sta-
tistics and Data Center (Dr Pang and Ms Cirrincione)
and Department of Biostatistics and Bioinformatics (Dr
Pang), Duke University, Durham, North Carolina; Can-
cer Supportive Services, Continuum Cancer Centers
of New York, Beth Israel Hospital, and St Luke’s–
Roosevelt, New York, New York (Dr Fleishman); The
1366 JAMA, April 3, 2013—Vol 309, No. 13
Downloaded From: https://jamanetwork.com/ Mexico | Access Provided by JAMA by Salvador Macias Diaz on 08/19/2022
Ohio State University Comprehensive Cancer Center,
College of Medicine, Department of Internal Medi-
cine, Columbus (Dr Paskett); Department of Psychia-
try and Behavioral Sciences, Memorial Sloan-
Kettering Cancer Center, New York, New York (Dr
Ahles); Cancer and Leukemia Group B/Alliance Central
Office, Chicago, Illinois (Dr Bressler); Department of
Medicine (Hematology/Oncology) and Neurology,
Norris Cotton Cancer Center, Dartmouth-Hitchcock
Medical Center, Lebanon, New Hampshire (Dr Fadul);
Illinois Oncology Research Association (Ms Knox and
Dr Le-Lindqwister); Main Line Hematology and On-
cology Associates, Wynnewood, Pennsylvania (Dr Gil-
man); and Division of Oncology, Department of In-
ternal Medicine, The Ohio State University Wexner
Medical Center and Comprehensive Cancer Center,
and James Cancer Hospital, College of Medicine, De-
partment of Internal Medicine Columbus (Dr
Shapiro).
Author Contributions: Drs Smith and Pang had full
access to all of the data in the study and take respon-
sibility for the integrity of the data and the accuracy
of the data analysis.
Study concept and design: Smith, Pang, Fleishman,
Paskett, Ahles, Bressler, Fadul, Shapiro.
Acquisition of data: Cirrincione, Fleishman, Knox,
Le-Lindqwister, Gilman.
Analysis and interpretation of data: Smith, Pang,
Cirrincione, Fleishman, Ahles, Fadul.
Drafting of the manuscript: Smith, Pang, Cirrincione,
Fleishman, Bressler, Knox.
Critical revision of the manuscript for important in-
tellectual content: Smith, Pang, Cirrincione, Fleish-
man, Paskett, Ahles, Fadul, Le-Lindqwister, Gilman,
Shapiro.
Statistical analysis: Smith, Pang, Cirrincione.
Obtained funding: Smith.
Administrative, technical, or material support: Pang,
Fleishman, Paskett, Ahles, Bressler, Knox,
Le-Lindqwister, Shapiro.
Study supervision: Smith, Pang, Fleishman, Paskett,
Fadul, Gilman.
Conflict of Interest Disclosures: All authors have com-
pleted and submitted the ICMJE Form for Disclosure
of Potential Conflicts of Interest. Dr Smith reported
receiving support from CALGB/Alliance for travel to
meetings. Dr Paskett reported institutional support from
CALGB/Alliance for travel to meetings. Dr Ahles re-
ported receiving support from CALBG/Alliance for
travel to meetings. Dr Fadul reported pending insti-
tutional grants from Genentech. Dr Gilman reported
institutional and direct grants pending from the NCI.
No other financial disclosures were made.
Funding/Support: This study was supported by grant
CA31946 from the NCI Division of Cancer Preven-
tion, the Alliance Statistics and Data Center, and the
Alliance Chairman. Drug and placebo were supplied
by Eli Lilly. The NCI provided funding for data man-
agement and statistical analysis.
Role of the Sponsor: The NCI and Eli Lilly each re-
viewed and approved the study concept via the usual
peer-review process. Minor suggestions were made
by each group regarding aspects of the study design.
Neither the NCI Division of Cancer Prevention nor Eli
Lilly had a role in data collection or management, analy-
sis, interpretation of the data, or with manuscript prepa-
ration, review, or approval.
Online-Only Material: The eAppendix, 8 eTables, eFig-
ure, and Author Video Interview are available at http:
//www.jama.com.
Additional Contributions: We thank the study partici-
pants and the research staff at all participating sites. Spe-
cial thanks are extended to the CALGB Oncology Nurs-
ing Committee, Richard Schilsky, MD (American Society
of Clinical Oncology), Xiaofei Wang, PhD (Duke Uni-
versity), John Taylor, MA (Alliance), Brandelyn Pitcher,
MS (Duke University), Sara Jasinski, BA (Duke Univer-
sity), Celia Bridges, BSN, RN (University of Michigan),
©2013 American Medical Association. All rights reserved.
and Asa B. Smith, Student (University of Michigan). Dr
Wang, Mr Taylor, Ms Pitcher, and Ms Jasinski received
salary support from the Alliance Cooperative Group Grant
for their contributions to this work. All others provided
assistance without receiving financial compensation.
REFERENCES
1. Smith EM, Cohen JA, Pett MA, Beck SL. The reli-
ability and validity of a modified total neuropathy score-
reduced and neuropathic pain severity items when used
to measure chemotherapy-induced peripheral neu-
ropathy in patients receiving taxanes and platinums.
Cancer Nurs. 2010;33(3):173-183.
2. Kautio AL, Haanpää M, Kautiainen H, Kalso E, Saarto
T. Burden of chemotherapy-induced neuropathy—a
cross-sectional study. Support Care Cancer. 2011;
19(12):1991-1996.
3. Loprinzi CL, Reeves BN, Dakhil SR, et al. Natural
history of paclitaxel-associated acute pain syndrome:
prospective cohort study NCCTG N08C1. J Clin Oncol.
2011;29(11):1472-1478.
4. Smith EML, Bakitas MA, Homel P, et al. Prelimi-
nary assessment of a neuropathic pain treatment and
referral algorithm for patients with cancer. J Pain Symp-
tom Manage. 2011;42(6):822-838.
5. Bakitas MA. Background noise: the experience of
chemotherapy-induced peripheral neuropathy. Nurs
Res. 2007;56(5):323-331.
6. Shimozuma K, Ohashi Y, Takeuchi A, et al. Taxane-
induced peripheral neuropathy and health-related qual-
ity of life in postoperative breast cancer patients un-
dergoing adjuvant chemotherapy: N-SAS BC 02, a
randomized clinical trial. Support Care Cancer. 2012;
20(12):3355-3364.
7. Dodd MJ, Cho MH, Cooper BA, Miaskowski C. The
effect of symptom clusters on functional status and
quality of life in women with breast cancer. Eur J On-
col Nurs. 2010;14(2):101-110.
8. Tofthagen C. Surviving chemotherapy for colon can-
cer and living with the consequences. J Palliat Med.
2010;13(11):1389-1391.
9. Stubblefield MD, Burstein HJ, Burton AW, et al.
NCCN task force report: management of neuropa-
thy in cancer. J Natl Compr Canc Netw. 2009;
7(suppl 5):S1-S26.
10. Saarto T, Wiffen PJ. Antidepressants for neuro-
pathic pain. Cochrane Database Syst Rev. 2007;
(4):CD005454.
11. Willis WD, Westlund KN. Neuroanatomy of the
pain system and of the pathways that modulate pain.
J Clin Neurophysiol. 1997;14(1):2-31.
12. Bril V, England JD, Franklin GM, et al; American
Academy of Neurology; American Association of Neu-
romuscular and Electrodiagnostic Medicine; American
Academy of Physical Medicine and Rehabilitation. Evi-
dence-based guideline: treatment of painful diabetic neu-
ropathy—report of the American Association of Neu-
romuscular and Electrodiagnostic Medicine, the American
Academy of Neurology, and the American Academy of
Physical Medicine & Rehabilitation. Muscle Nerve. 2011;
43(6):910-917.
13. Goldstein DJ, Lu Y, Detke MJ, Lee TC, Iyengar S.
Duloxetine vs placebo in patients with painful dia-
betic neuropathy. Pain. 2005;116(1-2):109-118.
14. Wernicke JF, Pritchett YL, D’Souza DN, et al. A
randomized controlled trial of duloxetine in diabetic
peripheral neuropathic pain. Neurology. 2006;
67(8):1411-1420.
15. Raskin J, Pritchett YL, Wang F, et al. A double-
blind, randomized multicenter trial comparing dulox-
etine with placebo in the management of diabetic pe-
ripheral neuropathic pain. Pain Med. 2005;6(5):
346-356.
16. Argyriou AA, Bruna J, Marmiroli P, Cavaletti G.
Chemotherapy-induced peripheral neurotoxicity
(CIPN): an update. Crit Rev Oncol Hematol. 2012;
82(1):51-77.

17. Tan G, Jensen MP, Thornby JI, Shanti BF. Vali-
dation of the Brief Pain Inventory for chronic nonma-
lignant pain. J Pain. 2004;5(2):133-137.
18. Cleeland CS. The brief pain inventory user guide.
http://www.mdanderson.org/education-and-research
/departments-programs-and-labs/departments-and
-divisions/symptom-research/symptom-assessment
-tools/BPI_UserGuide.pdf. Updated 2009. Accessed
July 28, 2009.
19. Cleeland CS, Ryan KM. Pain assessment: global
use of the Brief Pain Inventory. Ann Acad Med
Singapore. 1994;23(2):129-138.
20. Atkinson TM, Mendoza TR, Sit L, et al. The Brief
Pain Inventory and its “pain at its worst in the last 24
hours” item: clinical trial end point considerations. Pain
Med. 2010;11(3):337-346.
21. Dworkin RH, Turk DC, Peirce-Sandner S, et al. Re-
search design considerations for confirmatory chronic
pain clinical trials: IMMPACT recommendations. Pain.
2010;149(2):177-193.
22. Kaur H, Hota D, Bhansali A, Dutta P, Bansal D,
Chakrabarti A. A comparative evaluation of amitrip-
tyline and duloxetine in painful diabetic neuropathy:
a randomized, double-blind, cross-over clinical trial.
Diabetes Care. 2011;34(4):818-822.
23. Yang YH, Lin JK, Chen WS, et al. Duloxetine im-
proves oxaliplatin-induced neuropathy in patients with
colorectal cancer: an open-label pilot study. Support
Care Cancer. 2011;20(7):1491-1497.
24. Pritchett YL, McCarberg BH, Watkin JG, Robinson
MJ. Duloxetine for the management of diabetic pe-
ripheral neuropathic pain: response profile. Pain Med.
2007;8(5):397-409.
25. Mease PJ, Spaeth M, Clauw DJ, et al. Estimation
of minimum clinically important difference for pain in
fibromyalgia. Arthritis Care Res (Hoboken). 2011;
63(6):821-826.
26. Chappell AS, Desaiah D, Liu-Seifert H, et al. A
double-blind, randomized, placebo-controlled study
of the efficacy and safety of duloxetine for the treat-
ment of chronic pain due to osteoarthritis of the knee.
Pain Pract. 2011;11(1):33-41.
27. Chappell AS, Ossanna MJ, Liu-Seifert H, et al. Du-
loxetine, a centrally acting analgesic, in the treat-
ment of patients with osteoarthritis knee pain: a 13-
week, randomized, placebo-controlled trial. Pain. 2009;
146(3):253-260.
28. Marangell LB, Clauw DJ, Choy E, et al. Compara-
tive pain and mood effects in patients with comorbid
fibromyalgia and major depressive disorder: second-
ary analyses of four pooled randomized controlled trials
of duloxetine. Pain. 2011;152(1):31-37.
29. Frakes EP, Risser RC, Ball TD, Hochberg MC,
Wohlreich MM. Duloxetine added to oral nonsteroi-
dal anti-inflammatory drugs for treatment of knee pain
due to osteoarthritis: results of a randomized, double-
blind, placebo-controlled trial. Curr Med Res Opin.
2011;27(12):2361-2372.
30. Russell IJ, Mease PJ, Smith TR, et al. Efficacy and safety
of duloxetine for treatment of fibromyalgia in patients
with or without major depressive disorder: results from
a6-month,randomized,double-blind,placebo-controlled,
fixed-dose trial. Pain. 2008;136(3):432-444.
31. Calhoun EA, Welshman EE, Chang CH, et al. Psy-
chometric evaluation of the Functional Assessment of
Cancer Therapy/Gynecologic Oncology Group-
Neurotoxicity (FACT/GOG-Ntx) questionnaire for pa-
tients receiving systemic chemotherapy. Int J Gyne-
col Cancer. 2003;13(6):741-748.
32. Griffith KA, Merkies ISJ, Hill EE, Cornblath DR. Mea-
sures of chemotherapy-induced peripheral neuropa-
thy: a systematic review of psychometric properties.
J Peripher Nerv Syst. 2010;15(4):314-325.
33. Huang HQ, Brady MF, Cella D, Fleming G. Vali-
dation and reduction of FACT/GOG-Ntx subscale for
platinum/paclitaxel-induced neurologic symptoms: a
gynecologic oncology group study. Int J Gynecol
Cancer. 2007;17(2):387-393.
©2013 American Medical Association. All rights reserved.
Downloaded From: https://jamanetwork.com/ Mexico | Access Provided by JAMA by Salvador Macias Diaz on 08/19/2022
DULOXETINE ON CHEMOTHERAPY-INDUCED PERIPHERAL NEUROPATHY PAIN
34. Cella D, Eton DT, Fairclough DL, et al. What is a
clinically meaningful change on the Functional As-
sessment of Cancer Therapy-Lung (FACT-L) ques-
tionnaire? results from Eastern Cooperative Oncol-
ogy Group (ECOG) study 5592. J Clin Epidemiol. 2002;
55(3):285-295.
35. Norman GR, Sloan JA, Wyrwich KW. Interpreta-
tion of changes in health-related quality of life: the
remarkable universality of half a standard deviation.
Med Care. 2003;41(5):582-592.
36. Dubois D, Dhawan R, van de Velde H, et al. De-
scriptive and prognostic value of patient-reported out-
comes: the bortezomib experience in relapsed and re-
fractory multiple myeloma. J Clin Oncol. 2006;
24(6):976-982.
37. Cella D, Eton DT, Lai JS, Peterman AH, Merkel
DE. Combining anchor and distribution-based meth-
ods to derive minimal clinically important differences
on the Functional Assessment of Cancer Therapy
(FACT) anemia and fatigue scales. J Pain Symptom
Manage. 2002;24(6):547-561.
38. Common terminology criteria for adverse events
version 4.02 (CTCAE). http://evs.nci.nih.gov/ftp1
/CTCAE/About.html. Updated 2009. Accessed Janu-
ary 7, 2010.
39. Dworkin RH, Turk DC, McDermott MP, et al. In-
terpreting the clinical importance of group differ-
ences in chronic pain clinical trials: IMMPACT
recommendations. Pain. 2009;146(3):238-244.
40. Dworkin RH, Turk DC, Wyrwich KW, et al. In-
terpreting the clinical importance of treatment out-
comes in chronic pain clinical trials: IMMPACT
recommendations. J Pain. 2008;9(2):105-121.
41. Kajdasz DK, Iyengar S, Desaiah D, et al. Dulox-
etine for the management of diabetic peripheral neu-
ropathic pain: evidence-based findings from post hoc
analysis of three multicenter, randomized, double-
blind, placebo-controlled, parallel-group studies. Clin
Ther. 2007;29(suppl):2536-2546.
42. Cohen J. Statistical Power Analysis for the Be-
havioral Sciences. 2nd ed. Hillsdale, NJ: Lawrence Earl-
baum; 1988.
43. Hammack JE, Michalak JC, Loprinzi CL, et al. Phase
III evaluation of nortriptyline for alleviation of symp-
toms of cis-platinum-induced peripheral neuropathy.
Pain. 2002;98(1-2):195-203.
44. Kautio AL, Haanpää M, Leminen A, Kalso E,
Kautiainen H, Saarto T. Amitriptyline in the preven-
tion of chemotherapy-induced neuropathic symptoms.
Anticancer Res. 2009;29(7):2601-2606.
45. Barton DL, Wos EJ, Qin R, et al. A double-blind,
placebo-controlled trial of a topical treatment for
chemotherapy-induced peripheral neuropathy: NC-
CTG trial N06CA. Support Care Cancer. 2011;
19(6):833-841.
46. Rao RD, Flynn PJ, Sloan JA, et al. Efficacy of la-
motrigine in the management of chemotherapy-
induced peripheral neuropathy: a phase 3 random-
ized, double-blind, placebo-controlled trial, N01C3.
Cancer. 2008;112(12):2802-2808.
47. Rao RD, Michalak JC, Sloan JA, et al; North Cen-
tral Cancer Treatment Group. Efficacy of gabapentin
in the management of chemotherapy-induced pe-
ripheral neuropathy: a phase 3 randomized, double-
blind, placebo-controlled, crossover trial (N00C3).
Cancer. 2007;110(9):2110-2118.
48. Farrar JT, Young JP Jr, LaMoreaux L, Werth JL,
Poole RM. Clinical importance of changes in chronic
pain intensity measured on an 11-point numerical pain
rating scale. Pain. 2001;94(2):149-158.
49. Hall JA, Wang F, Oakes TM, Utterback BG, Crucitti
A, Acharya N. Safety and tolerability of duloxetine in
the acute management of diabetic peripheral neuro-
pathic pain: analysis of pooled data from three placebo-
controlled clinical trials. Expert Opin Drug Saf. 2010;
9(4):525-537.
50. Skljarevski V, Desaiah D, Liu-Seifert H, et al. Ef-
ficacy and safety of duloxetine in patients with chronic
low back pain. Spine (Phila Pa 1976). 2010;35
(13):E578-E585.
51. Raskin J, Wang F, Pritchett YL, Goldstein DJ. Du-
loxetine for patients with diabetic peripheral neuro-
pathic pain: a 6-month open-label safety study. Pain
Med. 2006;7(5):373-385.
52. Spallone V, Lacerenza M, Rossi A, Sicuteri R,
Marchettini P. Painful diabetic polyneuropathy: ap-
proach to diagnosis and management. Clin J Pain.
2012;28(8):726-743.
53. Morales-Vidal S, Morgan C, McCoyd M, Hornik
A. Diabetic peripheral neuropathy and the manage-
ment of diabetic peripheral neuropathic pain. Post-
grad Med. 2012;124(4):145-153.
54. Smith AG, Singleton JR. Diabetic neuropathy. Con-
tinuum (Minneap Minn). 2012;18(1):60-84.
55. Callaghan BC, Cheng HT, Stables CL, Smith AL,
Feldman EL. Diabetic neuropathy: clinical manifesta-
tions and current treatments. Lancet Neurol. 2012;
11(6):521-534.
56. Postma TJ, Heimans JJ, Muller MJ, Ossenkoppele
GJ, Vermorken JB, Aaronson NK. Pitfalls in grading se-
verity of chemotherapy-induced peripheral neuropathy.
Ann Oncol. 1998;9(7):739-744.
57. Kuroi K, Shimozuma K, Ohashi Y, et al. Prospec-
tive assessment of chemotherapy-induced peripheral
neuropathy due to weekly paclitaxel in patients with
advanced or metastatic breast cancer (CSP-HOR 02
study). Support Care Cancer. 2009;17(8):1071-
1080.
58. Cavaletti G, Frigeni B, Lanzani F, et al. Chemo-
therapy-induced peripheral neurotoxicity assess-
ment: a critical revision of the currently available tools.
Eur J Cancer. 2010;46(3):479-494.
59. Lavoie Smith EM, Cohen JA, Pett MA, Beck SL.
The validity of neuropathy and neuropathic pain mea-
sures in patients with cancer receiving taxanes and
platinums. Oncol Nurs Forum. 2011;38(2):133-
142.
60. Frigeni B, Piatti M, Lanzani F, et al. Chemotherapy-
induced peripheral neurotoxicity can be misdiag-
nosed by the National Cancer Institute Common
Toxicity scale. J Peripher Nerv Syst. 2011;16(3):
228-236.
61. Spina E, Trifirò G, Caraci F. Clinically significant
drug interactions with newer antidepressants. CNS
Drugs. 2012;26(1):39-67.
62. Patroneva A, Connolly SM, Fatato P, et al. An as-
sessment of drug-drug interactions: the effect of des-
venlafaxine and duloxetine on the pharmacokinetics
of the CYP2D6 probe desipramine in healthy subjects.
Drug Metab Dispos. 2008;36(12):2484-2491.
63. Skinner MH, Kuan HY, Pan A, et al. Duloxetine
is both an inhibitor and a substrate of cytochrome
P4502D6 in healthy volunteers. Clin Pharmacol Ther.
2003;73(3):170-177.
64. Hua TC, Pan A, Chan C, et al. Effect of dulox-
etine on tolterodine pharmacokinetics in healthy
volunteers. Br J Clin Pharmacol. 2004;57(5):652-
656.
65. Preskorn SH, Greenblatt DJ, Flockhart D, et al.
Comparison of duloxetine, escitalopram, and sertra-
line effects on cytochrome P450 2D6 function in
healthy volunteers. J Clin Psychopharmacol. 2007;
27(1):28-34.
66. Glueck CJ, Khalil Q, Winiarska M, Wang P. In-
teraction of duloxetine and warfarin causing severe
elevation of international normalized ratio. JAMA.
2006;295(13):1517-1518.
67. Jin Y, Desta Z, Stearns V, et al. CYP2D6 geno-
type, antidepressant use, and tamoxifen metabolism
during adjuvant breast cancer treatment. J Natl Can-
cer Inst. 2005;97(1):30-39.
68. Caraci F, Crupi R, Drago F, Spina E. Metabolic drug
interactions between antidepressants and anticancer
drugs: focus on selective serotonin reuptake inhibi-
tors and hypericum extract. Curr Drug Metab. 2011;
12(6):570-577.
JAMA, April 3, 2013—Vol 309, No. 13 1367
 EUROPEAN JOURNAL OF CANCER 4 2 ( 2 0 0 6 ) 2 8 2 –2 8 9

available at www.sciencedirect.com

journal homepage: www.ejconline.com

Review

Efficacy of homeopathic therapy in cancer treatment

Stefania Milazzoa, Nancy Russellb, Edzard Ernst a,*

a
Department of Complementary Medicine, Peninsula Medical School, Universities of Exeter and Plymouth,
Institute of Health and Social Care, 25 Victoria Park Road, Exeter EX2 4NT, United Kingdom
b
Integrative Medicine Program Education, Unit 147, The University of Texas M. D. Anderson Cancer Center, 1515 Holcombe Blvd, Houston,
TX 77030-0049, United States

A R T I C L E I N F O A B S T R A C T

Article history: Many cancer patients use homeopathic approaches to increase their bodyÕs ability to fight
Received 23 September 2005 cancer, improve their physical and emotional well-being, and alleviate their pain resulting
Accepted 28 September 2005 from the disease or conventional treatments. Homeopathy is highly controversial as there
Available online 11 January 2006 is no plausible mode of action for these highly diluted remedies. The aim of this systematic
review is to summarize and critically evaluate the efficacy of homeopathic remedies used
Keywords: as a sole or additional therapy in cancer care. We have searched the literature using the
Homeopathy databases: Amed (from 1985); CINHAL (from 1982); EMBASE (from 1974); Medline (from
Homeopathic remedy 1951); and CAMbase (from 1998). Randomised and non-randomised controlled clinical trials
Cancer including patients with cancer or past experience of cancer receiving single or combined
homeopathic interventions were included. The methodological quality of the trials was
assessed by Jadad score. Six studies met our inclusion criteria (five were randomised clin-
ical trials and one was a non-randomised study); but the methodological quality was var-
iable including some high standard studies. Our analysis of published literature on
homeopathy found insufficient evidence to support clinical efficacy of homeopathic ther-
apy in cancer care.
Ó 2005 Elsevier Ltd. All rights reserved.

1. Introduction As a palliative or supportive treatment, homeopathy is

Cancer is the second most frequent cause of death in devel-
oped countries according to a World Health Organization
(WHO) report from 2003 [1]. Even though non-surgical
orthodox treatments can control or even cure cancer, many
adverse effects limit their use [2]. Cancer patients therefore
often turn towards complementary therapies, including
homeopathy [2]. A recent European survey has shown that
homeopathy is amongst the most commonly used comple-
mentary therapies for cancer in 7 out of 14 European
countries [3].
used mainly to strengthen the body in its fight against can-
cer, to improve general well-being, and to alleviate pain
resulting from disease or conventional treatments [2,3].
Homeopathy is controversial as no plausible mode of action
has been identified for substances that are so highly diluted
that they can not be measured [4]. Homeopathic remedies
are believed to be most effective when they are selected to
address a total set of symptoms and characteristics [5] and
in classical or individualized homeopathy, choice of remedies
are based on the match of a patientÕs particular symptoms
with a remedy picture rather than conventional diagnosis

* Corresponding author: Tel.: +44 1392 424839; fax: +44 1392 427562.
E-mail addresses: stesincro@yahoo.com (S. Milazzo), nrussell@mdanderson.org (N. Russell), edzard.ernst@pms.ac.uk (E. Ernst).
0959-8049/$ - see front matter Ó 2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.ejca.2005.09.025
 EUROPEAN JOURNAL OF CANCER 4 2 ( 2 0 0 6 ) 2 8 2 –2 8 9 283

[6]. Prescribing homeopathic substances is based on its pro-
posed law of similars that suggests that ‘‘like cures like’’ [7].
Although Hahnemann initially diluted these substances in
order to reduce toxicity, he came to believe that the actual
process of diluting and shaking imparted additional potency
to each solution [5]. His process of testing natural substances
in healthy individuals became known as ‘‘drug proving’’ and
results continue to be collected into an encyclopaedia of
homeopathic drug effects known as the Materia Medica
[8,9]. In ‘‘classical homeopathy’’ single remedies are given to
patients, whereas in ‘‘complex homeopathy’’ several homeo-
pathic medicines are combined into one formula, where con-
centration tends to be below 24X and usually below 12X [10]
(the numbers indicate the dilution of the homeopathic rem-
edy; that is, remedies are obtained by ‘‘decimal dilution’’, one
part substance to nine parts alcohol, and then labelled by the
letter X or D).
In the 1950s, Hans H. Reckeweg developed a new form of
homeopathy known as homotoxicology [11], which generally
uses formulations that contains measurable amounts of
homeopathically prepared active ingredients, designed to
work with the bodyÕs defence mechanisms and facilitate the
bodyÕs elimination of toxic substances (homotoxins). Homo-
toxicological remedies are prepared according to the rules of
homeopathy and are used in combinations as complex reme-
dies. Some experts fail to differentiate between homeopathic
and homotoxicological medicines. However, there are impor-
tant differences. Homeopathy follows the ‘‘like cures like’’
principle, while homotoxicology does not [12]. Homotoxicol-
ogy often makes use of biological material that would be
atypical in homeopathy, such as material from pigs. Some re-
ports suggest the efficacy of homotoxicology for defined con-
ditions, but many caveats exist [12].
Isopathy is another subset of homeopathy that was devel-
oped by Johann Lux in the 1830s. It differs from homeopathy
in that remedies are prepared from those substances that
cause the illness (e.g., allergens or bacteria) [13] and several
trials have suggested its clinical efficacy [14].
The aim of this systematic review is to summarize and crit-
ically evaluate the clinical trial evidence for the effectiveness
of any type of homeopathic remedy in cancer care.
2. Methods
Electronic literature searches were conducted using the
following databases: Amed (from 1985); CINHAL (from 1982);

Table 1 – Search terms to identify the studies for this systematic review
Database Intervention(s) Cancer patients Types of study design Location of studies

PubMed 1. Homeopath*.ab, de. 9. Cancer*.ti, ab, rw, sh. 21. Clinical trial.pt. 33. 8 AND 20 AND 32
2. Homeopath* remed*.ab, de. 10. Neoplasm*.ti, ab, rw, sh. 22. Controlled clinical trial.pt.
3. Homeopath* drug*.ab, de. 11. Terminal adj.illness.ti,ab 23. Randomised controlled
trial.pt.
4. Simil*.ab, de. 12. Carcin*.ti, ab, rw, sh. 24. Random*.tw.
5. Homotoxicology.ab, de. 13. Oncol*.ti, ab, rw, sh. 25. Double blind.tw.
6. Single remedy.ab, de. 14. Sarcoma.ti, ab, rw, sh. 26. Placebo*.tw.
7. Combination remedies 15. Tumour.ti, ab, rw, sh. 27. Control*.tw.
8. OR/1–7 16. Leuk*.ti, ab, rw, sh. 28. Symptom*.tw.
17. Aden*.ti, ab, rw, sh. 29. Quality of life.tw.
18. Malignant.ti, ab, rw, sh. 30. Palliative treatment.tw.
19. Lymphoma.ti, ab, rw, sh. 31. Palliative care.tw.
20. OR/9–19 32. OR/21–31

NHS dialog 1. Homeopath$2.AB.DE. 9. Cancer$6.AB. 21. Clinical trial.AB. 33. 8 AND 20 AND 32
(AMED, CINHAL, 2. Homeopath$2 10. Neoplasm$4.AB. 22. Control$3.AB.
EMBASE, MEDLINE) adj.remed$3.AB.DE
3. Homeopath$2 11. Terminal 23. Study.AB.
adj.drug$1.AB.DE adj.illness.AB.DE
4. Simil$4.DE. 12. Carcin$6.AB. 24. Random$7.AB.
5. Homotoxicology.DE. 13. Oncol$4.AB. 25. Double adj.Blind.AB.
6. Single adj.remedy.DE. 14. Sarcoma.AB. 26. Placebo.AB.
7. Combination 15. Tumour$3.AB. 27. Control$4.AB.DE
adj.remedies.DE.
8. OR/1–7 16. Leuk$5.AB. 28. Sympto$2.AB.
17. Aden$6.AB. 29. Quality adj.life.AB.
18. Malignant.AB. 30. Palliative adj.treatment.DE.
19. Lymphoma.AB. 31. Palliative adj.care.DE.
20. OR/9–19 32. OR/21–31

CAMbase N/A N/A N/A Homeopathy for cancer;

homeopathic complex
remedies for cancer;
trial of homeopathy
for cancer

sh, subheading; ab, abstract; tw, textword; *, truncation; ti, title word; $, truncation; de, drug effect; pt, publication type.
284 EUROPEAN JOURNAL OF CANCER
EMBASE (from 1974); Medline (from 1951); and CAMbase (from
1998). AMED, CINHAL, and EMBASE were searched through
the NHS Dialog portal, whereas MEDLINE was searched
through NHS Dialog as well as through PubMed. CAMbase
was searched by simply typing a naturally spoken sentence,
as the system requires.
Randomised and non-randomised controlled clinical trials
(RCTs and CCTs, respectively) were included if they investi-
gated patients with cancer or past experience of cancer who
received single or combined homeopathic interventions as
sole treatment or as adjuvant to conventional treatments,
and their outcomes were compared to any other intervention
or no intervention. Details of the literature searches are sum-
marised in Table 1.
The main outcome measures we considered were efficacy
of homeopathic remedies for treating symptoms in cancer
patients and cancer survivors. Secondary outcome measures
included tumour response and quality of life. All articles were
read by two independent reviewers and data from the articles
were validated and extracted according to pre-defined crite-
ria. The methodological quality of all studies was indepen-
dently assessed by the two reviewers using Jadad score [26].
Discrepancies between reviewers were resolved by a third
independent reviewer. Statistically significant results of each
trial were documented.
3. Results
The searches identified 55 potentially relevant studies, of
which six met our inclusion criteria (Fig. 1). Three were dou-
ble-blinded RCTs, one was a triple-blinded study, and another
one was a CCT. Key data are summarised in Tables 2 and 3.
Kulkarni [15] conducted an RCT in cancer patients to as-
sess the effectiveness of homeopathy on the severity of radio-
therapy-related side effects. Patients with different types of
cancer were randomized into three parallel arms: placebo;
cobaltum 30; and causticum 30 (types of dilution were not
specified). These homeopathic remedies were selected be-
cause they mimic various symptoms of radiation reaction.
All the patients were evaluated once a week according to
an 18-point radiation reaction profile, and the average grading
was calculated at the end of the study: 0–5 for minimal reac-
tion; 6–10 for moderate but tolerable reaction; and >11 for
Total articles retrieved for homeopathic and
homotoxicology treatments in cancer care: 55
Animal and in vitro
studies: 12
Included studies: 6*
*Included studies:
Randomized Clinical trials: 5
Controlled Clinical trials: 1
Fig. 1 – Flowchart of inclusion process.
4 2 ( 2 0 0 6 ) 2 8 2 –2 8 9
severe degree of reaction. Reaction index was lower in both
intervention groups compared to placebo (5 for homeopathic
groups vs. 8.5 for placebo group). No significant differences in
tumour reduction were observed in the study.
A study by Oberbaum [16] tested he effects of TraumeelSÒ
on chemotherapy-induced stomatitis (mouth sores) in a non-
randomized CCT. This treatment for 20 children and teenag-
ers was compared with seven randomly chosen controls from
the same age group with similar stages of cancer, who re-
ceived no treatments for stomatitis.
TraumeelSÒ is an homeopathic preparation containing: ar-
nica 2X, calendula 2X, millefolium 3X, chamomilla 3X, sym-
phytum 6X, belladonna 2X ana 0.1 ml, aconitum 2X 0.06 ml,
bellis perennis 2X 0.05 ml, hypericum 2X 0.03 ml, echinacea
angustifolia 2X, echinacea purpurea 2X ana 0.025 ml, hama-
melis 1X 0.01, mercurius sol. 6X 0.05 g, and hepar sulfuris
6X 0.1 g. It was administered orally using liquid doses pro-
vided in vials. The primary outcome measure was the level
of pain measured according to opiate requirements. Ulcer
severity was assessed according to the WHO staging of dis-
ease: 0 for no ulcer; 1 for oral pain with no ulcers; 2 for oral
pain with ulcers but the ability to eat is retained; 3 for liquid
diet only; and 4 for inability to eat or to drink. In all treated
patients, the treatment was followed by an immediate de-
crease in pain. There was a non-significant trend suggesting
less patients in the intervention group required opiates com-
pared to the control group (P = 0.09) and some patients in this
group also reported a mood improvement. Symptom duration
was numerically different between the two groups favouring
the treated group (6 vs. 13 days). Oberbaum [17] subsequently
carried out a larger RCT to assess the effectiveness of Traume-
elSÒ for chemotherapy-induced stomatitis in cancer patients
after allogeneic or autologous stem-cell transplantation. Pa-
tients (ages 3–25 years) were randomised to two groups
receiving, in addition to standard mouth-washes, TraumeelSÒ
oral rinse or a placebo rinse. TraumeelSÒ preparation con-
tained high dilutions (10 1–10 9) of different extracts (see
above). The main outcomes were occurrence of stomatitis
and time to worsening of symptoms. Subjective symptom
scores were recorded during the first 7 days of the trial. Signif-
icant differences occurred in the reduction of severity and/or
duration of stomatitis in the intervention group (P < 0.01)
compared to the placebo group. Mean ‘‘area under the curve’’
Observational studies:
37
Table 2 – List of included studies
First Jadad Study designa Sample Patient Intervention Control Condition Outcome Main results P value Adverse
author score and follow up size condition (n) intervention investigated measures events
(year) [26] (n) (nP:nI)

Kulkarni 1 RCT with three 82 Cancer Cobaltum 30 Placebo (28) Radiation Primary outcome: Lower radiation NS None
1988 [15] parallel arms patients (28) reaction Degree of reaction in reported
Follow up: NS undergoing Causticum 30 reaction (0–5; intervention groups
radiation (26) 6–10; >11)
therapy

Oberbaum 0 CCT 27 Leukemia TraumeelÒ No treatment Chemotherapy Primary outcome: Lower opiate 0.09 None
1998 [16] Induced- Opiate requirements in reported

EUROPEAN JOURNAL OF CANCER

Stomatitis requirements treated group
for pain
Follow up: NS Secondary: Decrease in duration NS
Duration of of symptoms in
symptoms; intervention group
Quality of life
Mood improvement NS
in intervention
group

Balzarini 4 Randomised 61 Breast Belladonna Placebo (32) Radiodermatitis Primary outcome: Skin temperature 0.008, 0.016, Hot flushes,
2000 [18] double-blind cancer and 7cH and X-ray Skin heat; decrease at week 3, 0.023, 0.011, perspiration
placebo- undergoing 15cH (29) Hyperpig- 4, 6, 8 and 10 0.250 and
controlled radio- mentation; migraine

4 2 ( 2 0 0 6 ) 2 8 2 –2 8 9
clinical trial therapy Erythema; (0:1)
Oedema
Follow up: 10 Secondary: Total Less frequent 0.050,
weeks severity of hyperpigmentation 0.060
symptoms at week 5 and 10
Less intense colour 0.280
of the skin at week
10
High frequency of 0.025
oedema at week 5, 6
and 10
Decrease of total 0.025
severity of
symptoms during
recovery
0.890
0.05

Report of main results in the intervention groups.

nP:nI number of patients in the placebo group:number of patients in the intervention group; NS, not specified.
a If not stated otherwise trials have two parallel arms.

285
 286
Table 3 – List of included studies. Report of main results in the intervention group
First author Jadad Study design* Sample Patient Intervention Control Condition Outcome Main results P value Adverse
(year) score and follow up size condition (n) intervention investigated measure events
[26] (n) (nP:nI)

Oberbaum 4 Random. 30 Patients TraumeelSÒ Placebo (15) Chemotherapy- Primary outcome: Reduction of <0.01 Graft versus
2001[17] double-blind affected by (15) induced Occurrence of severity and host disease
placebo blood malignant stomatitis stomatitis; Time duration of (6:3)
controlled cancer, who to worsening of stomatitis in
clinical trial underwent BMT symptoms intervention group
Follow up: Secondary: Oral Decrease of time to <0.001 Sepsis (8:3)
44 weeks pain worsening of

EUROPEAN JOURNAL OF CANCER

symptoms in
intervention group
Reduction in oral NS
pain and discomfort;
dryness of mouth
and tongue;
difficulty to swallow;
and dysphagia in the
intervention group
Gastrointe-
stinal
complication
(5:0)

4 2 ( 2 0 0 6 ) 2 8 2 –2 8 9
Venous
occlusive
disease (0:4)
Pneumonitis
(0:4)

Jacobs 2005 5 Triple-blind RCT 83 Breast cancer HylandÕs Identical Menopausal Primary outcome: Increase of hot 0.01; Headaches in
[19] with three survivors menopause placebo (27) symptoms Hot flushes flushes severity <0.001 combination
parallel arms (30) severity; Hot score in remedy group
flushes combination (NS)
frequency remedy group
compared to placebo
and single remedy
Follow up: 35 different Secondary: Increase of hot 0.006;
1 year single Quality of life flushes frequency in 0.002
remedies (26) comb. group
compared to placebo
and single remedy
Lower KMI score in Not
single remedy group signific.
Improvement of <0.05
quality of life in both
intervention groups
 EUROPEAN JOURNAL OF CANCER 4 2 ( 2 0 0 6 ) 2 8 2 –2 8 9 287

score was 10.4 compared to 24.3. There was also a statistically

significant difference between the two groups in the mean
time to worsening of symptoms with an average of 6.9 days
Adverse
events
(nP:nI)

in the intervention group compared to 4.3 days in the placebo

group (P < 0.001). However the median times were similar (4.7
NS

compared to 4.0, respectively). In a subgroup of patients, aged

P value

less than 15 years, the lower severity score in the intervention

group (11 compared to 25.9 of placebo) was also statistically
/

significant (P < 0.01). Patients in the TraumeelSÒ group

Neither positive nor

showed a reduction in oral pain and discomfort; dryness of

significant results

mouth and tongue; difficulty to swallow; and dysphagia (no

were observed
Main results

P values of statistical significance for these symptoms were

provided). Nausea was reported in two patients in each group
at the beginning of the trial and they were not included in the
analysis. Serious complications occurred in both treatment
groups without significant differences between the two
severity; Quality
MYMOP Activity
Primary outcome:

groups. Graft vs. host disease, sepsis, and gastrointestinal

score; MYMOP

frequency and
Secondary: Hot
Profile score

complications occurred mainly in the placebo group. On the

Outcome
measure

other hand, more patients in the intervention group experi-

flushes

of life

enced venous occlusive disease and pneumonitis.

Balzarini [18] investigated the effectiveness of homeopathic
treatment for skin reactions during radiotherapy for breast
investigated

cancer treatment. Patients were randomised into two groups

withdrawal
Condition

symptoms

receiving homeopathic remedies or placebo. Both groups also

Estrogen

received a topical medication containing fluocortolone.

Homeopathic treatment consisted of three granules of Bella-
nP:nI number of patients in the placebo group:number of patients in the intervention group; NS, not specified.

donna 7cH (twice a day) and X-ray 15cH (once a day). The main
able placebo (25)

outcomes measured in this study were: erythema; skin heat;

intervention

Indistinguish-

cutaneous and subcutaneous oedema; and hyperpigmenta-

tion. Patients treated with homeopathy appeared to experience
Control

transient benefits. Less hyperpigmentation (P value at 5th

(n)

week = 0.050) and decrease of skin heat (P value at 8th

week = 0.011) were observed although these differences were
Intervention

Tablets liquid

no longer significant by the end of the 10-week follow up. Total

71 different

or granules
remedies.

severity score was positive in favour of homeopathic treatment

during radiotherapy and recovery, but statistical significance
(28)

for the difference was noticed only during recovery (P = 0.05).

(n)

High frequency of oedema was observed in the interven-

tion group, and statistical significance was reached at the
Breast cancer

5th and 6th week (P = 0.025). Otherwise, adverse events were

condition

equally distributed between the two groups.

survivor
Sample Patient

Jacobs and colleagues [19] evaluated the homeopathic ef-

fects on menopausal symptoms in breast cancer survivors. Pa-
tients who suffered from menopausal symptoms; had a
* If not stated otherwise trials have 2 parallel arms.

history of carcinoma in situ or stage I–III breast cancer; had

53
size

completed all surgical, chemo- or radiation therapy; and had

an average of three hot flushes per day for a month before
score and follow up
Jadad Study design*

the trial were included in the trial. Following a ‘‘double-dum-

double-blind

clinical trial

my’’ design, each patient was prescribed an individualized

Follow up:
controlled

16 weeks
Random.

placebo-

homeopathic medication and randomized into three treat-

ment groups: a placebo combination and a verum single rem-
edy; a verum combination medicine and a verum single
remedy; and two placebo combinations. Single remedies con-
[26]

sisted of 35 different homeopathic medications, mainly: sepia,

calcarea carbonica, sulfur, lachesis, and kali carbonicum. The
First author

combination remedy was ‘‘HylandÕs menopause’’, which con-

Thompson
2005 [20]

tained: amyl nitrate, sanguinaria canadensis, and lachesis.

Main outcome measures were hot flush severity and num-
(year)

ber of hot flushes, measured according to the Kupperman

Menopausal Index (KMI) score. Quality of life score was
288 EUROPEAN JOURNAL OF CANCER
measured using the Short Form 36 (SF-36), which evaluated
mental and physical health status. Although the overall single
remedy homeopathic group tended to experience lower sever-
ity score and fewer hot flushes, no significant differences
were found among the three groups in the univariate model
of all patients adjusted for baseline, time, and tamoxifen
use over 1 year period. A statistically significant improvement
in general health score was observed in both homeopathy
groups compared to placebo (combination: P < 0.03; single:
P = 0.02). A subgroup analysis was carried out for the patients
taking (60%) or not taking (40%) tamoxifen. In the subgroup
not receiving tamoxifen, there was a statistical significant in-
crease in the hot flush severity score in the combination
homeopathy group compared to placebo (P = 0.01, 95% CI,
6.2–47.1) and a highly significant difference when compared
to single remedy (P < 0.001, 95% CI, 51.9–15.0). In the group
not receiving tamoxifen, the total number of hot flushes in-
creased statistically significantly in the combination homeo-
pathic remedy group compared to the placebo and to the
single remedy (combination P = 0.006; single remedy
P = 0.002; 95% CI not provided). Statistically significant in-
crease of headaches was also observed in the combination
homeopathic group (P = 0.03).
Thompson and co-workers [20] conducted a double-blind
RCT to evaluate the effect of homeopathy in 53 breast cancer
survivors affected by oestrogen withdrawal symptoms. Selec-
tion criteria of patients included: more than three hot flushes
per day; not having metastatic disease; no concurrent treat-
ment for hot flushes; no severe concurrent illness; and not
undergoing chemotherapy or about to receive any adjuvant
chemotherapy. Patients were randomized to receive homeo-
pathic remedies or placebo. Patients randomized to homeop-
athy were individually prescribed 71 different remedies, most
commonly: sulfur, sepia, carcinosin, natrum muriaticum, bel-
ladonna, and arnica. Primary outcome measures were activity
score and overall profile score, measured according to the
Measure Yourself Medical Outcome Profile (MYMOP). Second-
ary outcome measures included: hot flush frequency and hot
flush severity (measured with Menopausal symptom ques-
tionnaire); and quality of life (measured according to EORTC
QLQ-C30, plus Breast module). No significant differences be-
tween intervention and placebo group for both MYMOP activ-
ity and overall profile scores were noted (P = 0.17, 95% CI
1.0–0.2; P=0.13, 95% CI 0.9–0.1, respectively). There were
no differences between the groups for any other secondary
outcomes measured at final follow up. Adverse events were
experienced by approximately one quarter of women in both
groups and no significant difference was observed between
active remedy and placebo group.
4. Discussion
Five out of six trials included in this systematic review yielded
positive results, which suggest the effectiveness of homeo-
pathic remedies for cancer care. Cancer patients appear to
have benefited from homeopathic interventions specifically
for chemotherapy-induced stomatitis, radiodermatitis and
general adverse events from radiotherapy. Breast cancer sur-
vivors, suffering from menopausal symptoms, experienced
a general improvement on their quality of life.
4 2 ( 2 0 0 6 ) 2 8 2 –2 8 9
Among the six studies we included, five were randomised
[15,17–20] and only one was not randomised [16]. Statistical
analysis for significance was performed in all the studies,
but only four provided statistical features in their result sec-
tions. Oberbaum [16] discussed highly significant differences
between the groups regarding duration of symptoms, but no
statistical features for those differences were given. This
was a pilot study conducted in order to test the rationale for
performing a more rigorously designed trial [17]. The study
by Kulkarni [15], investigating the effectiveness of some
homeopathic remedies for radiation protection, concluded
that the remedies reduced the degree of radiation reaction
significantly, but results of their statistical analysis were not
shown. This study also lacked complete information regard-
ing patients and remedies, as well as essential methodologi-
cal details, such as randomisation method.
Even trials with a Jadad score of 5 were not devoid of flaws.
The small sample size in Jacobs [19] precluded definitive con-
clusions, and a major flaw in the study was the use of combi-
nation remedies in an ongoing daily regimen, without
following over-the-counter instructions that suggested to dis-
continue use if adverse effects occurred. In the trial con-
ducted by Thompson [20], the high placebo response can
probably be related to a type II error in the study, since their
sample size was only adequate for detecting large differences
in response.
Of the 6 trials included in this systematic review, only two
reported statistically significant positive results of their pri-
mary outcomes. One of these was a RCT conducted by Ober-
baum [17] that showed encouraging results confirming the
need to perform the trial on a larger scale. The other study
was the RCT carried out by Balzarini [18], although this trial
showed statistically significant differences in favour of the
intervention group, these differences were not consistently
observed at all time points. The rest of the studies indicated
a positive trend towards homeopathic interventions for
improvements in quality of life and symptom management,
which seems to justify further investigations.
Homeopathic remedies are thought to trigger the bodyÕs
own defence and self-regulatory response. However, their
mode of action is unclear. As homeopathic remedies are often
diluted beyond AvogadroÕs number, no pharmacological ac-
tion can be expected. Sceptics therefore insist that homeopa-
thyÕs clinical success is solely due to a placebo response [21].
Homeopaths counter this criticism by postulating that
homeopathic remedies work through mechanisms other than
pharmacological ones [22,23]. The evidence for homeopathic
remedies in cancer care may not be fully conclusive but it
does seem to warrant further study. Clinical trials of homeop-
athy should be rigorously designed to minimize bias. The
existing trials have a number of limitations, e.g. sample size,
which should be addressed in future research. Such research
can be expensive and it is therefore a precondition that ade-
quate research funds for homeopathy are made available.
Considering that positive results have been obtained in some
cases [24] using the controversial remedy ‘‘Carcinosin’’ (a car-
cinogenic substance) and in some animals studies [25] inves-
tigating the possible anticancer effect of homeopathic
interventions, further studies testing homeopathy for tumour
response should also be undertaken.
 EUROPEAN JOURNAL OF CANCER
The main limitation of our systematic review is the lack
and sometimes poor quality of the primary data. The studies
we evaluated were highly heterogeneous in virtually every re-
spect. In some studies, individualized remedies were applied.
Although individualization of therapy allows homeopathy to
be practiced in its traditional fashion, this increases the com-
plexity of comparing outcomes. In conclusion, the evidence
emerging from this systematic review is encouraging but
not convincing. Further research should attempt to answer
the many open questions related to homeopathy.
Conflict of interest statement
None declared.
R E F E R E N C E S
1. http://www.who.int/cancer/media/en/788.pdf [accessed
01.06.05].
2. Montfort H. A new homeopathic approach to neoplastic
diseases: from cell destruction to carcinogen-induced
apoptosis. Brit Homeopath J 2000;89(2):78–83.
3. Molassiotis et al. Use of complementary and alternative
medicine in cancer patients: a European survey. Ann Oncol
2005;16(4):655–63.
4. Shang A, Huwiler-Müntener K, Nartey L, et al. Are the clinical
effects of homoeopathy placebo effects? Comparative study
of placebo-controlled trials of homoeopathy and allopathy.
The Lancet 2005;366(9487):726–32.
5. Jonas WB, Kaptchuk TJ, Linde K. A critical overview of
homeopathy. Ann Intern Med 2003;138:393–9.
6. Linde K, Melchart D. Randomized controlled trials of
individualized homeopathy: a state-of -the art review. J Altern
Complem Med 1998;4:371–88.
7. Robinson K. Homeopathy: questions and answers. New
Mexico: Elegance publishing; 1994.
8. Bellavite P, Signorini A. Homeopathy: a frontier in medical
science. North Atlantic Books.
9. Ernst E, Kaptchuk TJ. Homeopathy revisited. Arch Int Med
1996;156(19):2162–4.
10. https://www.cedrugstorenews.com/userapp/lessons/
page_view_ui.cfm?lessonuid=&pageid=
E44F4F4C5554FDD448D601472A02FFA0
[accessed 01.06.05].
11. http://www.heel.com.au/homotoxicology/index.shtml
[accessed 12.06.05].
4 2 ( 2 0 0 6 ) 2 8 2 –2 8 9 289
12. Ernst E, Schmidt K. Homotoxicology – a review of randomised
clinical trials. Eur J Clin Pharmacol 2004;60(5):299–306. Epub
2004 Jun 9.
13. http://www.bioresourceinc.com/articles/vol1no2.html
[accessed 12.06.05].
14. Reilly DT, Taylor MA, McSharry C, et al. Is homoeopathy a
placebo response? Controlled trial of homoeopathic potency,
with pollen in hayfever as model. Lancet 1986;2:881–6.
15. Kulkarni A, Nagarkar BM, Burde GS. Radiation protection by
use of homoeopathic medicines. Hahnemann Homoeopath Sand
1988;12:20–3.
16. Oberbaum M. Experimental treatment of
chemotherapy-induced stomatitis using a homeopathic
complex preparation: a preliminary study. Biol Med
1998;3(6):104–8.
17. Oberbaum M, Yaniv I, Ben-Gal Y, et al. A randomized,
controlled clinical trial of the homeopathic medication
TRAUMEEL S in the treatment of chemotherapy-induced
stomatitis in children undergoing stem cell transplantation.
Cancer 2001;92(3):684–90.
18. Balzarini A, Felisi E, Martini A, et al. Efficacy of homeopathic
treatment of skin reactions during radiotherapy for breast
cancer: a randomised, double-blind clinical trial. Brit
Homeopath J 2000;89(1):8–12.
19. Jacobs J, Herman P, Heron K, et al. Homeopathy for
menopausal symptoms in breast cancer survivors: a
preliminary randomized controlled trial. J Altern Complem Med
2005;11(1):21–7.
20. Thompson EA, Montgomery A, Douglas D, et al. A pilot,
randomized, double-blinded, placebo-controlled trial of
individualized homeopathy for symptoms of estrogen
withdrawal in breast-cancer survivors. J Altern Complem Med
2005;11(1):13–20.
21. Vandenbroucke JP. Homoeopathy and ‘‘the growth of truth’’.
The Lancet 2005;366(9487):691–2.
22. Fisher P. The information medicine hypothesis. In: Shulte J,
Endler PC, editors. Fundamental research in high
dilutions. Dordrecht: Kluwer; 1998. p. xi–xiv.
23. Witt C, Ludtke R, Weisshuhn TE, et al. High homeopathic
potencies are different from potentized solvent when
investigated with the REDEM technology. Forsch Komp Klass
Nat 2005;12(1):6–13.
24. Masiello DJ. Melanoma. Am J Holistic Med 2003–2004;96(4):
318–20.
25. Sato DY, Wal R, de Oliveira CC, et al. Histopathological and
immunophenotyping studies on normal and sarcoma
180-bearing mice treated with a complex homeopathic
medication. Homeopathy 2005;94(1):26–32.
26. Jadad A, Moore A, Carroll D. Assessing the quality of
randomized trials: is blinding necessary?. Control Clin Trial
1996;17:1–12.
Tumores sólidos 2022 (actualizado).

49 . Cáncer en el embarazo:
consideraciones generales y
agentes antineoplásicos
Rachel Cossetti, Ana Carolina Salles de Mendonça y Antonio Carlos Buzaid

DEFINICIÓN
El cáncer asociado al embarazo se define como el cáncer diagnosticado
desde el primer día de la concepción hasta 1 año después del parto [ Best
Pract Res Clin Obstet Gynaecol 33:2, 2016 ].

EPIDEMIOLOGIA
El diagnóstico de cáncer asociado con el embarazo ocurre en
aproximadamente uno de cada 1000 a 1500 embarazos [ Oncologist 7:279,
2002 ; Am J Obstet Gynecol 189:1128, 2003 ; Int J Gynecol Cancer 27:613,
2017 ]. Esta incidencia ha ido en aumento durante las últimas décadas y
parece reflejar una mayor incidencia de cáncer en general y el retraso en
la edad del primer embarazo hasta la tercera y cuarta décadas de vida para
un número cada vez mayor de mujeres [ Surg Oncol 20:e175 , 2011 ]. Los
tipos de cáncer más comunes durante el embarazo son el cáncer de
mama, el melanoma, el cáncer de cuello uterino, el cáncer de ovario, las
neoplasias malignas hematológicas, el cáncer de tiroides y el cáncer
colorrectal [ J Clin Oncol 27:45, 2009 ].

CONSIDERACIONES GENERALES
La aparición de cáncer durante el embarazo agrega más complejidad al
manejo oncológico de la paciente, ya que la madre y el feto pueden verse
afectados por el tratamiento. El melanoma, las neoplasias malignas
hematológicas y el cáncer de pulmón son las únicas neoplasias malignas
con diseminación a la placenta y al feto descritas en la bibliografía [ J Clin
Oncol 21:2179, 2003 ]. Aunque no se comprende por completo, los cambios
hormonales, la supresión inmunitaria y el aumento de la vascularización y
la permeabilidad capilar están asociados con la fisiopatología del cáncer
en el embarazo [ World J Oncol 10:28, 2019 ]. El manejo ideal de estos casos
debe buscar las recomendaciones terapéuticas estándar para la
enfermedad oncológica de base, de acuerdo al pronóstico, guías
terapéuticas y supervivencia estimada, pero se debe considerar la
seguridad fetal al tomar decisiones, además de factores éticos,
psicológicos, religiosos y legales [ Best Pract Res Clin Obstet Gynaecol 33:2,
2016 ]. El embarazo debe manejarse como un embarazo de alto riesgo, con
monitoreo fetal regular mediante ecografía morfológica (US) y Doppler de
la arteria umbilical [ ESMO Open 1:e000016, 2016 ]. El objetivo debe ser
lograr un parto a término (≥ 37 semanas), ya que la prematuridad es uno
de los principales factores pronósticos que afectan el desarrollo pediátrico
hasta los 3 años de edad [ Lancet Oncol 19:337, 2018]. La modalidad de
parto debe elegirse con base en la recomendación obstétrica, excepto en
casos de neoplasias del tracto ginecológico. Sin embargo, en muchos
casos se prefiere la cesárea por la posibilidad de determinar la fecha exacta
del parto, lo que permite una mejor planificación de la fecha del último
tratamiento sistémico, entre otras intervenciones. Siempre se debe
evaluar cuidadosamente la placenta en busca de signos de
micrometástasis, particularmente en casos de melanoma [ ESMO Open
1:e000016, 2016 ]. En la mayoría de los casos es posible ofrecer un
tratamiento oncológico adecuado a la paciente con mantenimiento del
bienestar fetal, pero la falta de experiencia y conocimiento de los
profesionales sanitarios en el manejo de estos casos puede dar lugar a
retrasos en el diagnóstico, tratamiento incorrecto y mayor riesgo materno
-fetal [Mundo J Oncol 10:28, 2019 ]. Cabe señalar que, en muchos casos, los
retrasos en el diagnóstico y tratamiento pueden ser más perjudiciales que
los posibles riesgos fetales. Se debe considerar la edad gestacional al
determinar los posibles efectos nocivos de las intervenciones diagnósticas
y terapéuticas sobre el bienestar fetal [ Best Pract Res Clin Obstet
Gynaecol 33:2, 2016 ].

— Hasta 10 días después de la concepción: fase de “todo o nada”. Si el

embrión sobrevive a la exposición a agresiones externas, es probable que
no ocurra ningún efecto adverso.

— 10 días a 8 semanas: período de organogénesis. La exposición a

agentes teratogénicos puede causar malformaciones fetales importantes.

— 2° y 3° trimestres: período de crecimiento fetal. Algunos sistemas

fetales en crecimiento siguen siendo vulnerables, como el sistema
nervioso central (SNC) y el sistema hematológico. Pueden ocurrir
complicaciones como el parto prematuro y la restricción del crecimiento
intrauterino con la exposición a agresiones externas.
El embarazo y el impacto en la evolución del
cáncer
En general, las características patológicas y el pronóstico del cáncer
diagnosticado durante el embarazo son similares a las de las pacientes no
embarazadas, con HR = 1,03 para el grupo de embarazadas (IC 95 %: 0,86-
1,22) y HR = 1,02 (IC 95 %: 0,86). -1,22) para el grupo lactante. Estudios previos
sugieren que las mujeres diagnosticadas de mama (HR=1,95; IC 95%: 1,36-
2,78) y de ovario (HR=2,23; IC 95%: 1,05-4,73) durante la lactancia tenían un
mayor riesgo de muerte por causa específica. El riesgo de mortalidad por
causas específicas se redujo en mujeres que experimentaron un
embarazo después del diagnóstico de cáncer (HR=0,49; IC del 95 %: 0,41-
0,9) [ J Clin Oncol 27:45, 2009]. Sin embargo, un estudio de cohortes que
incluyó mujeres diagnosticadas con cáncer de mama en el embarazo
después de la década de 2000 evaluó el pronóstico materno de pacientes
que recibieron quimioterapia (QT) durante el embarazo y mostró una
supervivencia libre de progresión (PFS) y una supervivencia general (SG)
similares para mujeres embarazadas y embarazadas. mujeres no
embarazadas (SLP: HR=1,0; IC 95 %: 0,82-1,27; p=0,83; OS: HR=1,08; IC 95 %:
0,81 -1,45; p=0,59). En un análisis de subgrupos de pacientes que recibieron
más del 60 % de QT durante el embarazo, la supervivencia fue comparable
a la de mujeres no embarazadas (SLP: HR=0,81; IC 95 %: 0,62-1,06; p =0,13;
SG: HR= 0,85; IC del 95 %: 0,58-1,23; p=0,39) [ J Clin Oncol 39:abstr 515, 2021 ].

Interrupción electiva del embarazo

El mantenimiento del embarazo no parece estar asociado con un peor
pronóstico, siempre que el tratamiento del cáncer se realice
correctamente. La interrupción puede ser discutida en pacientes con
enfermedad metastásica, especialmente en casos de crisis visceral en el
primer trimestre, en los que existe riesgo inminente de muerte para la
madre o en casos de enfermedad avanzada e imposibilidad de
tratamiento adecuado, en vista de los riesgos al paciente feto. Se debe
realizar un abordaje multidisciplinar del paciente, con apoyo y
seguimiento del equipo de psicología [ ESMO Open 1:e000016, 2016 ].

SÍNTOMAS Y PRESENTACIÓN
Los cambios fisiológicos del embarazo suponen un reto para el
diagnóstico oncológico, con frecuentes retrasos, especialmente en los
casos de cáncer de mama. El diagnóstico temprano es esencial para lograr
mejores resultados, pero lamentablemente se retrasa por factores que
confunden su alcance [ Best Pract Res Clin Obstet Gynaecol 33:19,
2016 ]. Por otro lado, el embarazo provoca cambios tisulares que afectan a
condiciones benignas y pueden conducir a cambios sugestivos de
malignidad. El patólogo siempre debe estar informado del estado de
embarazo de la paciente para minimizar los errores de diagnóstico [ Eur J
Cancer 46:3158, 2010 ].

COMO ESTADIAR
Es necesario realizar pruebas de estadificación, según el sitio primario,
respetando las particularidades relacionadas con el embarazo, lo que
minimiza los riesgos maternos y fetales [ Cancer Biol Med 15:6, 2018]. Se
deben preferir las técnicas de imagen sin exposición a la radiación
ionizante, como la ecografía y la resonancia magnética (RM) sin contraste,
durante el embarazo. La ecografía suele estar ampliamente disponible y
representa el método de elección para la evaluación inicial de una masa
abdominal o lesiones en la mama, la cabeza y el cuello y los tejidos blandos,
y se puede realizar en cualquier trimestre del embarazo. La RM sin
contraste debe realizarse preferentemente después del primer trimestre
del embarazo, lo que permite una valoración más detallada de órganos y
tejidos, con valoración funcional de las lesiones mediante la técnica de
difusión, que se considera segura durante el embarazo. Existen reportes
en la literatura de la asociación causal del uso de gadolinio durante el
embarazo con malformaciones fetales, aborto y muerte neonatal, [JAMA
316:952, 2016 ; Best Pract Res Clin Obstet Gynaecol 33:19, 2016 ]. La
resonancia magnética de cuerpo entero sin contraste es una prueba de
estadificación importante, que contribuye a la evaluación del tórax, el
abdomen y los huesos, y debe realizarse preferiblemente después del
período de organogénesis [ Radiographics 32:897, 2012 ; Cáncer Biol Med
15:6, 2018]. Se debe considerar el uso de técnicas de imagen con radiación
ionizante cuando sea esencial para el manejo del caso, pero se debe
monitorear la exposición a la radiación, con una dosis fetal acumulada
máxima recomendada de 100 mGy. Con este umbral de dosis, existe un
riesgo aproximadamente un 1 % mayor de malformación fetal y cáncer
infantil. Varios métodos de imagen con radiación ionizante se consideran
seguros, con exposición fetal por debajo de la dosis máxima recomendada
si se usa con protección abdominal adecuada, y no deben evitarse cuando
sea necesario, como la radiografía simple no abdominal, la mamografía y
la tomografía computarizada (TC) de cabeza y cuello. La tomografía
computarizada por emisión de positrones (PET-CT), la gammagrafía ósea,
las TC de abdomen y pelvis y los exámenes guiados por fluoroscopia
cruzan el umbral de seguridad fetal, deberían ser evitados. También se
debe evitar el uso de contraste yodado en mujeres embarazadas
[Tratamiento del cáncer Rev 27:1, 2001 ; Lancet Oncol 6:328, 2005 ; J Vasc
Interv Radiol 23:19, 2012 ; Mundo J Oncol 10:28, 2019 ].

Marcadores tumorales
Los niveles séricos de marcadores tumorales pueden tener menor
sensibilidad y especificidad como resultado de cambios fisiológicos
durante el embarazo. Los niveles de CA 15-3, SCC (antígeno de carcinoma
de células escamosas), CA 125 y alfa-fetoproteína están aumentados en el
embarazo, y no se recomienda su uso. Otros marcadores, como ACE, CA
19-9, DHL y la hormona antimulleriana, no se ven alterados por el estado
del embarazo y pueden ser útiles. Como excepciones, la inhibina B está
elevada en el último trimestre del embarazo y la DHL es un marcador de
cambios hipertensivos en el embarazo [ BMC Med 10:86, 2012 ; Mundo J
Oncol 10:28, 2019 ].

TRATAMIENTO
Cirugía
La cirugía materna durante el embarazo puede estar asociada con un
mayor riesgo de parto prematuro y perfusión uteroplacentaria alterada,
con un mayor riesgo de hipoxia, daño neurológico y muerte fetal. El
segundo trimestre del embarazo es el período más propicio para realizar
procedimientos quirúrgicos, debido al menor riesgo de aborto
espontáneo y parto prematuro [ Asian Pac J Clin Oncol 15:296,
2019 ]. Cuando esté indicado, el procedimiento quirúrgico no debe
posponerse debido al riesgo de progresión de la enfermedad materna y
compromiso fetal por malignidad. Una revisión sistemática de 12 452
mujeres que se sometieron a procedimientos quirúrgicos no obstétricos
durante el embarazo no identificó un mayor riesgo de aborto espontáneo
o defectos fetales [ Am J Surg 190:467, 2005]. Se recomienda reducir el
tiempo bajo anestesia general, minimizar la dosis y la concentración de los
agentes utilizados y evitar la anestesia por inhalación [ Am J Obstet
Gynecol 218:98, 2018 ].

Radioterapia
La exposición a la radiación se asocia con teratogénesis, aborto
espontáneo, retraso mental y mayor riesgo de malignidad en niños y
adultos después de la exposición intrauterina. Por lo tanto, se recomienda
esperar al final del embarazo para el tratamiento con radioterapia. Hay, sin
embargo, reportes de radioterapia durante el embarazo sin asociación con
efectos adversos fetales en casos de cáncer de mama, linfoma, cáncer de
cabeza y cuello y SNC, probablemente relacionados con una protección
uterina eficiente y baja dispersión de la radiación con el uso de tecnologías
más modernas. Los eventos adversos relacionados con la exposición a la
radiación durante el embarazo dependen del período de desarrollo fetal y
de la dosis de radiación. Las dosis inferiores a 0,1 Gy no parecen afectar el
desarrollo fetal. En el primer trimestre (organogénesis), los riesgos de
malformaciones, la restricción del crecimiento de la cabeza y el retraso
mental son mayores. En el segundo trimestre, los principales efectos
adversos son similares a los del primer trimestre. En el tercer trimestre, el
riesgo de retraso mental y restricción del crecimiento es menor, pero aún
puede ocurrir [Crit Rev Oncol Hematol 136:13, 2019 ].

TRATAMIENTO SISTÉMICO
Consideraciones generales
Las indicaciones de tratamiento sistémico durante el embarazo son
similares a las de las pacientes diagnosticadas de cáncer fuera de este
período, es decir, se consideran los mismos factores de riesgo. Sin
embargo, al proponer el tratamiento a la mujer embarazada, se debe
evaluar cuidadosamente la seguridad para el feto. El gran reto es no
infratratar a la paciente oncológica y elegir el tratamiento que mejor se
adapte al trimestre gestacional para preservar la salud fetal. Es imperativo
que el oncólogo tenga un conocimiento profundo de los medicamentos
contra el cáncer seguros durante el embarazo. No se recomienda el
tratamiento sistémico en el primer trimestre del embarazo, considerando
el riesgo de teratogenicidad fetal, prematuridad y aborto espontáneo. En
los casos de pacientes diagnosticados de enfermedad avanzada o
metastásica en el primer trimestre, se debe considerar la gravedad de la
enfermedad de base, riesgo para la salud materna, necesidad de
tratamiento específico y mantenimiento o no del embarazo. A partir del
segundo trimestre del embarazo, existen datos de estudios retrospectivos
así como series de casos que sugieren la seguridad del uso de ciertos
agentes antineoplásicos. [Ann Intern Med 94:547, 1981 ; J Clin Oncol
23:4192, 2005 ; Tratamiento del cáncer Rev 41:301, 2015]. En su mayoría, los
agentes antineoplásicos, la terapia hormonal (HT) y los agentes biológicos
utilizados en el tratamiento del cáncer se consideran categoría D para el
tratamiento de mujeres embarazadas con cáncer por la FDA y deben
evitarse. Un estudio observacional multicéntrico que evaluó a niños que
estuvieron expuestos a QT en el período prenatal mostró que la exposición
fetal a QT no se asoció con un aumento de la morbilidad cardíaca o
auditiva, ni con daño a la salud general, incluidos los hitos en el
crecimiento y desarrollo infantil, en comparación con la población en
general. Es importante destacar que, en este estudio, la prematuridad fue
común entre los fetos que fueron expuestos a quimioterapia, y este fue un
factor asociado con el deterioro cognitivo y del desarrollo. Por lo tanto, el
parto prematuro iatrogénico debe evitarse tanto como sea posible.
[Lancet Oncol 19:337, 2018 ]. Otro estudio de casos y controles corrobora
estos hallazgos [ N Engl J Med 373:1824, 2015 ]. Los cambios fisiológicos del
embarazo, tales como los cambios en la función renal y hepática, el
volumen plasmático y el efecto del tercer espacio relacionado con el
líquido amniótico, pueden alterar la farmacocinética de los agentes
antineoplásicos [ Semin Oncol 16:337, 1989 ; Best Pract Res Clin Obstet
Gynaecol 33:86, 2016 ]. El cálculo de la dosis de quimioterapia en pacientes
embarazadas también debe basarse en el área de superficie corporal, con
ajustes basados en el aumento de peso continuo. Deben evitarse las
reducciones de dosis debido a la probable mayor eliminación de fármacos
durante el embarazo [ Lancet Oncol 5:283, 2004 ].

Amamantamiento
Las mujeres que se someten a un tratamiento oncológico activo no deben
amamantar. Hay informes de excreción de fármacos citotóxicos en la
leche materna con riesgo de toxicidad neonatal, como leucopenia y
trombocitopenia. Todavía son escasos los datos sobre la seguridad del uso
de la terapia sistémica con nuevas clases de medicamentos, y se debe
evitar la lactancia [ Cancer Treat Rev 39:207, 2013 ; Medicina de la Lactancia
Materna 12:91, 2017 ].

Quimioterapia
La mayoría de los agentes quimioterapéuticos pueden atravesar la barrera
placentaria (peso molecular < 500 kDa) y llegar al feto [ Clin
Pharmacokinet 43:487, 2004 ]. Los riesgos para el feto asociados con el uso
de quimioterapia durante el embarazo dependen de los fármacos
utilizados y de la edad gestacional [ Best Pract Res Clin Obstet Gynaecol
33:86, 2016 ]. En el primer trimestre del embarazo (período de
organogénesis), el uso de agentes quimioterapéuticos está
contraindicado debido al alto riesgo de teratogenicidad, con mayor riesgo
de anomalías congénitas, anomalías cromosómicas, prematuridad y
aborto espontáneo. El riesgo de malformaciones es mayor con los
fármacos combinados que con el uso de un solo agente [ Ann Oncol 15:146,
2004 ; Defectos de nacimiento Res A Clin Mol Teratol 94:626, 2012; Best
Pract Res Clin Obstet Gynaecol 33:86, 2016 ]. Algunos medicamentos se
consideran seguros durante el segundo y tercer trimestre [ J Clin Oncol
23:4192, 2005 ]. El riesgo de malformación fetal es del 3-4% en la población
general, y el tratamiento con fármacos considerados seguros en los
últimos trimestres se asoció con un riesgo bajo (alrededor del 3%), pero
existe el riesgo de restricción del crecimiento intrauterino (RCIU),
prematuridad y bajo peso al nacer hasta en un 50% de los casos. Aunque
el uso de quimioterapia durante el segundo y tercer trimestre del
embarazo se considera seguro, debe evitarse durante las 3 semanas
previas a la fecha prevista del parto para reducir el riesgo de
mielosupresión fetal y complicaciones durante el parto como hemorragia,
sepsis y muerte neonatal. [Defectos de nacimiento Res A Clin Mol Teratol
94:626, 2012 ; Best Pract Res Clin Obstet Gynaecol 33:86, 2016 ].

Antraciclinas: El uso de doxorrubicina y epirrubicina se considera

seguro en el segundo y tercer trimestre del embarazo. Debe evitarse el uso
de idarubicina debido al riesgo de teratogenicidad. Un estudio
prospectivo del MD Anderson Cancer Center (MDACC) que incluyó a 87
pacientes con cáncer de mama en el embarazo tratadas, en el segundo o
tercer trimestre, con el régimen FAC la terapia adyuvante o neoadyuvante
ha demostrado seguridad en el uso de antraciclinas en el embarazo. La
gran mayoría de los recién nacidos no presentó complicaciones
perinatales, ocurriendo complicaciones en solo el 3% de ellos, tasa similar
a la población general. Hubo un caso de síndrome de Down, un caso de
reflujo urotelial y un caso de pie zambo congénito. No hubo aborto
espontáneo ni muerte perinatal [ Breast Cancer Res 16:500, 2014 ; BMC
Cancer 17:777, 2017 ].

Taxanos: El uso de taxanos en el segundo y tercer trimestre del embarazo

parece ser seguro. Una revisión sistemática de la literatura que incluye 16
estudios (50 embarazos) que usan taxanos durante el embarazo informó
un 76,7 % de los resultados con el nacimiento de niños completamente
sanos [ Clin Breast Cancer 13:16, 2013 ]. Además, otros estudios describen su
uso en cáncer de ovario, cervical y de pulmón de células no pequeñas con
resultados prometedores [ Arch Gynecol Obstet 283:97, 2011 ; Arco Gynecol
Obstet 284:779, 2011 ; J Thorac Oncol 4:559, 2009 ].

Platino: los datos sugieren seguridad con el uso de derivados de platino

en el segundo y tercer trimestre del embarazo. En un metaanálisis que
incluyó 39 estudios, con 88 mujeres con cáncer de cuello uterino, la terapia
neoadyuvante basada en platino se consideró una opción favorable para
el tratamiento de pacientes, con recién nacidos completamente normales
en el 80,7 % de los casos [ Drug Des Devel Ther 13:79, 2018 ]. La revisión de
la literatura que usa platino combinado con taxanos para el tratamiento
del cáncer de ovario también informó datos de seguridad de la
combinación [ Int J Clin Pharmacol Ther 55:753, 2017 ]. En una revisión
sistemática de 43 mujeres tratadas con platino durante el embarazo (36
recibieron cisplatino , 6 recibieron carboplatino y 1 recibió ambos
fármacos), 8 pacientes expuestas a cisplatino tuvieron algún impacto en
el embarazo o deterioro de la salud fetal, 3 tenían RCIU, 1 polihidramnios, 1
ventriculomegalia y 3 casos de prematuridad. Ninguno de los pacientes
tratados con carboplatino experimentó complicaciones [ Cancer 113:3069,
2008 ].

Ciclofosfamida: Se considera seguro para el tratamiento de mujeres

embarazadas en el segundo o tercer trimestre del embarazo. La mayoría
de los estudios evaluaron el uso de ciclofosfamida en regímenes
asociados con antraciclinas [ Rev Assoc Med Bras 59:174, 2013 ; Cáncer de
mama Res 16:500, 2014 ].

Fluoruracilo: Hay informes de seguridad del tratamiento del cáncer en

mujeres embarazadas con este agente de quimioterapia después del
primer trimestre del embarazo. Los datos provienen de estudios que
evaluaron la seguridad y eficacia de la combinación de antraciclinas
más ciclofosfamida con o sin fluorouracilo (régimen FAC ) en el
tratamiento del cáncer de mama en el embarazo [ Breast Cancer Res
16:500, 2014 ; Cáncer 107:1219, 2006 ].

Metotrexato: el uso de metotrexato conlleva un alto riesgo de

teratogenicidad y debe evitarse durante el embarazo [ Birth Defects Res A
Clin Mol Teratol 94:626, 2012 ].

Inhibidores de la tirosina quinasa

Los inhibidores de la tirosina cinasa ( tyrosine kinase inhibitors - TKI)
actúan de forma selectiva sobre los receptores de tirosina cinasa (TKR) que
se activan en algunas neoplasias. Aunque se considera selectivo para el
TKR de interés, los sitios de unión de ATP están muy conservados, lo que
permite la unión a TKR no específicos [ Sci Rep 9:2535, 2019 ]. La
transferencia placentaria de imatinib al feto está documentada en
humanos. La tasa de malformación después de la exposición en el primer
trimestre es de alrededor del 11%, sin reportes en el segundo y tercer
trimestre del embarazo. Por lo tanto, se debe evitar el uso de imatinib en
el primer trimestre del embarazo. Datos del uso de imatinib después del
primer trimestre están restringidas a un pequeño número de pacientes, y
su uso debe evaluarse cuidadosamente en cada caso [ Blood 111:5505,
2008 ; Cancer Treat Rev 41:301, 2015 ]. Los TKI de segunda
generación, dasatinib y nilotinib , parecen tener un perfil de toxicidad
similar al de imatinib durante el embarazo [ Anticancer Drugs 23:754,
2012 ; JHematol Oncol 2:42, 2009 ]. Sin embargo, se debe evitar su uso
hasta que haya más datos disponibles [ Cancer Treat Rev 41:301,
2015 ]. La vía del receptor del factor de crecimiento epidérmico– EGFR)
participa en la proliferación y diferenciación celular y juega un papel
importante en varias etapas del desarrollo embrionario. Los informes de
uso involuntario de erlotinib (2 casos), gefitinib (1 caso), erlotinib seguido
de gefitinib (1 caso) y crizotinib (2 casos) durante el embarazo no
mostraron anomalías fetales. Sin embargo, el uso de estos medicamentos,
así como los inhibidores de EGFR de segunda y tercera generación, no se
recomienda durante el embarazo debido a la escasez de datos [ J Adv Res
7:571, 2016 ]. Ácido transretinoico ( ATRA) es esencial para el desarrollo
embrionario, regulando la morfogénesis, la proliferación celular y la
diferenciación [ Mol Med Rep 16:5915, 2017 ]. Su transferencia placentaria
está bien documentada en humanos y está contraindicada en el primer
trimestre del embarazo [ Br J Haematol 94:699, 1996 ; Acta Haematol
114:167, 2005 ]. Su uso parece ser seguro después del primer trimestre del
embarazo [ Cancer Treat Rev 41:301, 2015 ]. El interferón -alfa no atraviesa
la barrera placentaria en cantidades significativas y se considera una
opción terapéutica plausible para su uso durante el embarazo [ Reprod
Toxicol 33:265, 2012 ].

Agentes biológicos
Los agentes biológicos, como los anticuerpos monoclonales, son
moléculas de alto peso molecular que requieren un transporte placentario
activo para llegar al feto, lo que no suele ocurrir hasta la semana 14 de
gestación, lo que sugiere que la exposición a estos agentes en el primer
trimestre presenta un riesgo menor. efectos adversos, mientras que los
riesgos son mayores en el segundo y tercer trimestre del embarazo, pero
hay escasez de datos sobre el uso de estos agentes en el primer trimestre,
por lo que se debe tener precaución y evitar su uso [ Cancer Treat Rev
41:301, 2015 ]. La tasa de eventos adversos con el uso de anticuerpos
monoclonales es de hasta un 40 % [ Birth Defects Res A Clin Mol Teratol
94:626, 2012 ].

Terapia anti-HER-2: El uso de trastuzumab en el tratamiento del cáncer

de mama HER-2 positivo en el embarazo está contraindicado en cualquier
trimestre. Su uso se ha asociado con casos de oligohidramnios, que
pueden estar relacionados con casos de hipoplasia pulmonar, anomalías
esqueléticas e incluso muerte neonatal [ Cancer Treat Rev 41:301, 2015 ]. En
un meta-análisis de 17 estudios, la aparición de oligohidramnios fue del 61,1
%, con una mayor incidencia de prematuridad [ J Clin Oncol 36:abstr 10065,
2018 ]. De manera similar, el uso de lapatinib [ Clin Breast Cancer 7:339,
2006 ], pertuzumab [ Anticancer Drugs 29:810, 2018 ] y ado-trastuzumab
emtansine[ Ther Adv Med Oncol 6:202, 2014 ] también está contraindicado
en el embarazo.

Rituximab: existe un riesgo potencial de citopenia neonatal y posneonatal

con el uso de rituximab durante el embarazo, pero se deben sopesar los
beneficios de usar este fármaco en el control de la enfermedad y la SG en
portadoras de linfomas de células B durante el embarazo [ Cancer Treat
Rev 41:301, 2015 ; Front Pharmacol 12:621247, 2021 ].

Inhibidores de EGFR: Cetuximab y panitumumab son anticuerpos

monoclonales anti-EGFR. A pesar de ser moléculas grandes, ambas
atraviesan la barrera placentaria. La vía EGFR es importante para el
crecimiento fetal y el mantenimiento del embarazo, con teratogenicidad
reportada en modelos animales. No existen datos sobre la seguridad de su
uso en humanos durante el embarazo, y debe evitarse [ Oncologist 21:563,
2016 ; Cancer Treat Rev 41:301, 2015 ].

Agentes antiangiogénicos: debido a la escasez de datos en la literatura,

se debe evitar el uso de antiangiogénicos durante el embarazo [ Cancer
Treat Rev 41:301, 2015 ]. Bevacizumab es un anticuerpo monoclonal que
inhibe la vía del receptor del factor de crecimiento endotelial vascular
(VEGF). En un modelo animal, hay transferencia placentaria dependiente
de la dosis en todas las etapas del embarazo [ Birth Defects Res B Dev
Reprod Toxicol 95:363, 2012 ], con hallazgos de eventos adversos fetales [ J
Matern Fetal Neonatal Med 27:1744, 2014 ] . También hay datos de
teratogenicidad fetal en modelos animales que
usan sunitinib y sorafenib anti-VEGFR multi-TKI [Defectos de nacimiento
Res B Dev Reprod Toxicol 86:204, 2009 ]. No hay datos sobre el uso de estos
medicamentos durante el embarazo en humanos.

Inmunoterapia
Durante el embarazo, las vías de señalización de puntos de control, en
particular el eje PD-1/PDL-1, son importantes para la tolerancia inmunitaria
fetal. La evidencia sugiere que los inhibidores de los puntos
de control inmunitarios pueden atravesar la barrera placentaria y causar
daño al feto y un mayor riesgo de aborto espontáneo [ Cancer Treat Rev
41:301, 2015 ; Best Pract Res Clin Obstet Gynaecol 33:86, 2016 ; Semín Oncol
45:164, 2018 ].

Terapia hormonal
Estudios con el uso de tamoxifeno durante el embarazo en humanos
reportan 16 casos de nacidos vivos con malformaciones congénitas y 122
de nacidos vivos sin malformaciones. De estos últimos, se informaron 85
casos en el estudio de prevención, sin que se informaran hallazgos de
anomalías fetales [ Lancet 342:168, 1993 ]. La base de datos de seguridad
de AstraZeneca informó 11 casos de malformaciones congénitas en 44
nacidos vivos, 12 abortos espontáneos, 17 interrupciones sin hallazgos de
anomalías fetales, 6 interrupciones con anomalías fetales, 1 muerte fetal
sin anomalía fetal y 2 muertes fetales con anomalías fetales, además de 57
embarazos con resultado desconocido [ Oncólogo 16:1547, 2011]. Aunque
representan casos con una descripción retrospectiva, sin establecer una
relación causal entre el uso de tamoxifeno en el embarazo y el resultado
gestacional, se considera prudente suspender el tamoxifeno antes de
planificar un embarazo y evitar su uso durante todas las etapas del
embarazo. En los casos de embarazo involuntario usando tamoxifeno , se
debe considerar y discutir el riesgo materno de recaída debido a la
interrupción temprana del tratamiento. Los niños que han estado
expuestos al tamoxifeno intrauterino deben someterse a un seguimiento
a largo plazo [ Oncologist 16:1547, 2011 ; Cáncer J del sur de Asia 5:209,
2016]. No hay datos de seguridad sobre el uso de inhibidores de la
aromatasa durante el embarazo, y debe evitarse, al igual que el uso de
análogos de LHRH.

Terapia de suporte
Bisfosfonatos: Los datos de la literatura actual sugieren que el uso de
bisfosfonatos durante el embarazo no está asociado con un mayor riesgo
de efectos teratogénicos fetales, pero puede haber una mayor tasa de
complicaciones neonatales. En caso de exposición fetal a uno de estos
agentes, se recomienda monitorear la posible hipopotasemia neonatal y
síntomas cardíacos y neuromusculares [ Am J Health Syst Pharm 71:2029,
2014 ; Osteoporos Int 30:221, 2019 ].

Denosumab: El uso de denosumab , un anticuerpo monoclonal humano

que inhibe el ligando RANK, utilizado en el tratamiento de metástasis
óseas, está contraindicado en el embarazo y la lactancia. Los estudios en
animales muestran un aumento de las tasas de mortinatalidad y
mortalidad neonatal [ Reprod Toxicol 42:27, 2013 ; Biochem Biophys Res
Commun 491:614, 2017 ].

Antieméticos: el uso
de meclozina , difenhidramina y metoclopramida se considera seguro
durante el embarazo, incluido el primer trimestre [ N Engl J Med 360:2528,
2009 ; JAMA 316:1392, 2016 ]. El uso de ondansetrón durante el embarazo
se considera seguro, pero debe evitarse en el primer trimestre del
embarazo, ya que se asocia con un mayor riesgo de paladar hendido
[ JAMA 320:2429, 2018 ]. El uso de dexametasona puede afectar la
osteogénesis fetal. Debe evitarse su uso crónico en el embarazo [ Curr
Pharm Des 20:5430, 2014]. Debe evitarse el uso de antagonistas de la
neuroquinina 1 (NK1) durante el embarazo debido a la falta de datos de
seguridad.

Factor estimulante de colonias: el uso de factores estimulantes de

colonias parece ser seguro durante el embarazo. Usando datos
del Registro de Cáncer y Embarazo de la Universidad de Cooper, un
estudio evaluó retrospectivamente el papel y los posibles efectos adversos
de los factores estimulantes de colonias en pacientes diagnosticadas con
cáncer en el embarazo tratadas con quimioterapia y que usaron
o filgrastim o pegfilgrastim durante el tratamiento de quimioterapia. De
las 176 gestantes que recibieron quimioterapia, 142 no usaron factor
estimulante de colonias y 34 lo usaron. No se observó un aumento en la
incidencia de teratogenicidad, abortos o partos prematuros con el uso del
factor estimulante de colonias. Además, las pacientes que usaron estos
medicamentos tenían un menor riesgo de infecciones oportunistas
posparto. Los datos son limitados y el número de pacientes evaluados es
pequeño [ J Cancer Ther 3:157, 2012 ; Obstet Gynecol 125:197,
2015 ]. Recientemente, la Red Internacional de Cáncer, Infertilidad y
Embarazo (INCIP) publicó una serie de casos que evaluó a 42 pacientes
que recibieron factor estimulante de colonias ( G-CSF). como parte del
tratamiento del cáncer durante el embarazo. En su mayor parte, las
mujeres se vieron afectadas por el cáncer de mama. Las principales
indicaciones para el uso de G-CSF fueron: protocolo QT de dosis densa (67
%), protocolo de terapia multimedicamentosa (17 %), neutropenia febril (12
%) y prevención de la neutropenia perinatal (5 %). Las complicaciones
observadas en la serie de casos son similares a los datos de otras
publicaciones sobre mujeres embarazadas que reciben quimioterapia, lo
que en un principio no sugiere un mayor riesgo de malformaciones y
prematuridad relacionado con el uso del factor estimulante de colonias
[ Cancers (Basel) 13:1214 , 2021 ].