Anatomic Pathology / Updated Approach to Leukemia Cutis
An Updated Approach to the Diagnosis of Myeloid
Leukemia Cutis
Danielle M. P. Cronin, MD,1 Tracy I. George, MD,1 and Uma N. Sundram, MD, PhD1,2
Key Words: Myeloid leukemia cutis; Myeloid sarcoma; Immunohistochemistry; CD34; Flow cytometry; World Health Organization
classification; Blastic plasmacytoid dendritic cell neoplasm; CD4+/CD56+ hematodermic neoplasm
DOI: 10.1309/AJCP6GR8BDEXPKHR
Abstract
The diagnosis of myeloid leukemia cutis can be
difficult, particularly in the context of an initial skin
biopsy with a malignant hematopoietic neoplasm. We
studied the immunohistochemical characteristics of 33
cases of myeloid leukemia cutis diagnosed at Stanford
University Medical Center, Stanford, CA, 1996-2007,
and compared them with the corresponding bone
marrow blast immunophenotype and World Health
Organization classification (2008). In the skin, CD43
marked 97% of cases (32/33), myeloperoxidase marked
42% (14/33), CD68 marked 94% (31/33), CD163
marked 25% (7/28), and CD56 marked 47% (14/30).
CD34 and CD117 were predominantly negative. In
19 cases in which myeloperoxidase was negative, all
marked with CD68 and CD43. The flow cytometric
immunophenotype of the leukemic blasts in the bone
marrow was discordant with the immunohistochemical
profile in the skin in all cases, showing loss or gain
of at least 1 antigen. Given the immunophenotypic
differences between skin and bone marrow blasts, we
provide an updated immunohistochemical approach to
the diagnosis of myeloid leukemia cutis.
© American Society for Clinical Pathology
Downloaded from https://academic.oup.com/ajcp/article/132/1/101/1766248 by guest on 05 May 2021
Leukemia cutis (LC) refers to the specific infiltration of
the skin by neoplastic leukemic cells, most often in conjunction
with systemic leukemia.1,2 There is debate about the appropri-
ate diagnostic terminology for LC, and various authors have
used historic, related, and/or overlapping terms, including
chloroma, extramedullary myeloid tumor, granulocytic sarco-
ma, and myeloid sarcoma. The term leukemia cutis is perhaps
the most broad and is frequently used in the dermatopathology
literature. In studies using biopsy-confirmed cases of LC, the
prevalence of this disease is 2% to 3% in the setting of sys-
temic acute leukemia. Skin involvement in myeloid leukemia
is associated with monocytic differentiation, central nervous
system involvement, and aneuploidy of chromosome 8.2-4
Very rarely, LC may precede clinically detectable systemic
leukemia (so-called aleukemic leukemia cutis).1
Clinically and histologically, myeloid LC can be con-
fused with a wide array of skin conditions. Clinically, lesions
of myeloid LC are characterized by single or multiple papules,
nodules, or plaques that may range from violaceous to red-
brown.5 The clinical differential diagnosis is broad, as a wide
range of neoplastic, inflammatory, and infectious skin lesions
may be associated with hematologic malignancies and their
treatment. In addition, myeloid LC lesions are histologically
heterogeneous but typically characterized by a primarily inter-
stitial infiltrate of leukemic cells involving the dermis, some-
times extending into the subcutis. Perivascular and/or periad-
nexal arrangements can also be seen.6 The blasts are medium
sized to large with smooth chromatin, prominent nucleoli,
and increased mitotic activity.7 Among cutaneous neoplastic
hematologic conditions, the histologic differential diagnosis
includes lymphomas with high-grade morphologic features
(B- or T-lineage lymphomas, CD30+ lymphoproliferative
Am J Clin Pathol 2009;132:101-110 101
101 DOI: 10.1309/AJCP6GR8BDEXPKHR 101
Cronin et al / Updated Approach to Leukemia Cutis

disorders), blastic plasmacytoid dendritic cell neoplasm (also
known as agranular CD4+/CD56+ hematodermic neoplasm),
mast cell sarcoma, and high-grade plasma cell neoplasms.
In the absence of established systemic acute leukemia,
confirming the diagnosis in cases of suspected myeloid LC
can be difficult. In all cases, judicious use of immunohis-
tochemical staining is of paramount importance.8-13 With this
in mind, we examined a series of cases of myeloid LC by
immunohistochemical staining and compared these findings
with the flow cytometric and cytogenetic findings in the cor-
responding bone marrow specimens. Our goal was to identify
an appropriate panel of commercially available immunohis-
marrow biopsies, and the presence or absence of intervening
therapy. This study was approved by the Stanford University
Institutional Review Board.
Immunohistochemical Studies
H&E-stained sections and appropriate immunohis-
tochemical studies performed at the time of diagnosis were
reviewed to confirm the original findings. Additional immu-
nohistochemical studies were performed on formalin-fixed,
paraffin-embedded tissue using a Ventana Benchmark semi-
automated stainer (Ventana Medical Systems North America,
Tucson, AZ) or an automated DAKO polymer-based detec-

Downloaded from https://academic.oup.com/ajcp/article/132/1/101/1766248 by guest on 05 May 2021

tochemical stains that may be helpful in the diagnosis of
myeloid LC in paraffin-embedded tissues. We also correlated
these findings with the current World Health Organization
(WHO) classification of acute myeloid leukemia.9,14
Materials and Methods
Case Selection
The pathology database of Stanford University Medical
Center, Stanford, CA, was searched for cases of myeloid LC.
For the study, 33 cases of myeloid LC diagnosed between
1996 and 2007 had paraffin-embedded materials available.
The cases selected all met the common dermatopathology
definition of LC, ie, they all had specific blastic infiltrates of
varying densities.1 Corresponding bone marrow aspirate, cyto-
genetic and/or molecular data, and flow cytometric data were
obtained when available, and the diagnosis was classified by
using the 2008 WHO classification of tumors of hematopoi-
etic and lymphoid tissues.14 Other clinical data obtained about
the patients included age, sex, location of the biopsy, clinical
appearance of the lesions, duration between skin and bone
tion system (DAKO North America, Carpinteria, CA), per the
manufacturer’s protocol as previously described.15 Primary
antibodies directed against CD3, CD20, CD34, CD43, CD56,
CD68, CD117, CD163, and myeloperoxidase (MPO) were
used. The antibody sources, dilutions, epitope retrieval proce-
dures used before incubation with the primary antibody, auto-
mated stainer used, and staining patterns are summarized in
zTable 1z. We use 2 protocols for CD34 and CD117 staining
at our institution: a standard protocol titered against vascular
endothelium (CD34 standard) or mast cells (CD117 standard)
and a bone marrow protocol optimized for the detection of
bone marrow blasts (CD34 BM and CD117 BM) (Table 1).
Staining for all markers was defined as follows: positive,
moderate to intense staining of at least 5% of lesional cells;
and negative, faint staining of fewer than 5% of lesional cells
to no staining of lesional cells.15 Diffuse faint staining was not
seen in any of the cases. For the purposes of this study, we
evaluated immunohistochemical staining in malignant infil-
trates only and did not include reactive conditions occurring
in patients with leukemia.
Positive and negative control samples were present in
all cases for all antibodies. Lymphocytes served as positive
internal control samples for CD20, CD3, and CD43; basal

zTable 1z
Antibodies Used: Pretreatment Conditions, Dilutions, Source, and Staining Pattern*

Antigen Clone Pretreatment Antibody Dilution Source Automated Stainer Staining Pattern

MPO Poly None 1:8,000 DAKO Ventana Cytoplasm

CD43 L60 HIER 1:2,000 BD Ventana Membrane
CD68 KP1 HIER 1:1,600 DAKO Ventana Membrane and cytoplasm
CD56 123C3 HIER 1:20 Zymed DAKO Membrane
CD117 (standard) Poly HIER 1:200 DAKO Ventana Membrane
CD117 (BM) Poly HIER 1:200 DAKO DAKO Membrane
CD34 (standard) MY10 HIER 1:10 BD Ventana Membrane
CD34 (BM) MY10 HIER 1:20 BD DAKO Membrane
CD163 10D6 HIER 1:100 Vector Ventana Membrane and cytoplasm
CD20 L26 HIER 1:1,000 DAKO Ventana Membrane
CD3 Poly HIER 1:200 Cell Marque Ventana Membrane
CD4 IFG HIER 1:20 Novocastra Ventana Membrane and cytoplasm

BM, bone marrow protocol (see text for details); HIER, heat-induced epitope retrieval; MPO, myeloperoxidase; Poly, polyclonal antibody.
* BD, Becton Dickinson, Franklin Lakes, NJ; Cell Marque, Rocklin, CA; DAKO North America, Carpinteria, CA; Novocastra Laboratories, Newcastle upon Tyne, England;

Vector Laboratories, Burlingame, CA; Ventana Medical Systems North America, Tucson, AZ; Zymed Laboratories, South San Francisco, CA.

102 Am J Clin Pathol 2009;132:101-110 © American Society for Clinical Pathology

102 DOI: 10.1309/AJCP6GR8BDEXPKHR

melanocytes for CD117; cutaneous vessels for CD34; and
cutaneous nerves for CD56. Neutrophils in tonsils served as
positive external control samples for MPO and histiocytes
in tonsils for CD68 and CD163. The epidermis (except for
basal melanocytes) served as a negative internal control
sample for all antibodies.
Results
The mean age of the 33 patients selected for this study
was 53 years (range, 22 days to 90 years). There was a slight
male predominance, with 20 males and 13 females. Using
available data, the results of the bone marrow specimens for
29 patients were examined and the lesions classified accord-
ing to published WHO criteria zTable 2z.14 The correspond-
ing clinical information, bone marrow biopsy diagnosis (29
patients), and cytogenetic and/or molecular information (18
patients) is shown in zTable 3z. As in previous studies, a
subset of cases (5/29) showed monocytic differentiation,2,6,13
and 5 of 18 cases with available cytogenetic information had
numeric abnormalities of chromosome 8.2,16,17 In addition, 7
cases were associated with myelodysplasia-related changes,
and 3 cases had normal bone marrow findings. Outside of
these findings, WHO subtype and/or cytogenetic specific cor-
relations with development of myeloid LC were not seen.
The histologic findings of myeloid LC were relatively
uniform and similar to previously published reports.7,13
Sections showed an unaffected epidermis with a grenz zone
zImage 1z. The proliferation was composed primarily of
an interstitial infiltrate of immature malignant cells with
increased nuclear/cytoplasmic ratios. Rarely, there was fol-
liculotropism, but, in general, the infiltrate was arranged in
periadnexal and/or perivascular configurations. High-power
examination revealed enlarged nuclei with finely dispersed
chromatin, variably prominent nucleoli, and, frequently, irreg-
ular nuclear contours. The histologic characteristics could not
be used to further classify the lesions.
The results of the immunohistochemical studies are
shown in zTable 4z and Image 1. CD43 was expressed in
nearly all cases (32/33 [97%]). We also found that a high
percentage of cases expressed CD68, regardless of their WHO
classification and MPO status (31/33 [94%]). In contrast, the
antibody for CD163 marked only 7 (25%) of 28 cases. We
found that MPO was present in 14 (42%) of 33 cases. The 19
cases that did not express MPO, however, all expressed CD68
and CD43. CD56 was expressed in 14 (47%) of 30 cases and
was also not associated with a specific WHO classification.
In the group consisting of cases of acute monocytic leukemia,
we found that CD163 expression was overrepresented: 3 of
the 7 cases found to be positive overall were in this group. In
addition, of the remaining 4 cases positive for CD163, 2 were
associated with chronic myelomonocytic leukemia or acute
© American Society for Clinical Pathology
Anatomic Pathology / Original Article
zTable 2z
Bone Marrow Diagnoses by the 2008 World Health
Organization Classification
Diagnosis No. of Cases
Acute monoblastic/monocytic leukemia 5
Acute myelogenous leukemia, not otherwise specified 4
Acute myelogenous leukemia with 11q23 abnormalities 2
Acute myelogenous leukemia with maturation 1
Acute myelogenous leukemia with myelodysplasia- 7
related changes
Acute myelogenous leukemia without maturation 2
Acute promyelocytic leukemia 1
Atypical chronic myeloid leukemia 1
Chronic myeloid leukemia 1
Chronic myelomonocytic leukemia 1
Downloaded from https://academic.oup.com/ajcp/article/132/1/101/1766248 by guest on 05 May 2021
Normal bone marrow 3
Refractory anemia with excess blasts 1
Unknown 4
Total 33
myeloid leukemia arising in a patient with a history of chronic
myelomonocytic leukemia (cases 23 and 32). This finding
further confirms the relative specificity of CD163 expression
for monocytic differentiation.
It is interesting that CD34 was expressed in only 2 (6%)
of 31 cases tested. CD117 was also not expressed often and
was present in 3 (10%) of 30 cases. Approximately half of the
cases were tested with CD117 (standard) and CD34 (standard),
while the remainder were tested with CD117 (bone marrow)
and CD34 (bone marrow). For both antibodies, a similar
number tested positively with either protocol and constituted
a minority of cases (overall 7%-10%), regardless of whether
the standard or bone marrow protocol was applied, with CD34
expressed by 1 (6%) of 16 and 1 (7%) of 15 cases and CD117
expressed by 1 (7%) of 14 and 2 (13%) of 16 cases by the
standard and bone marrow protocols, respectively.
To further examine the relationship between the bone
marrow immunophenotype and that of the blasts of myeloid
LC, we compared the flow cytometric data with data obtained
from immunohistochemical analysis of biopsy specimens of
myeloid LC zTable 5z. In all cases, the immunohistochemical
profile of the skin blasts was discordant from the bone marrow
flow cytometric antigen expression profile, even in tempo-
rally concurrent studies with similar blast cell morphologic
features. For example, 8 of 13 cases were found to express
CD34 by flow cytometry, but only 1 matched case expressed
CD34 by immunohistochemical analysis. Similarly, 8 cases
expressed CD117 by flow cytometry, and only 2 matched
cases expressed CD117 by immunohistochemical analysis.
Ten cases expressed MPO by flow cytometry, and only 6
matched cases showed expression by immunohistochemical
analysis. Six cases expressed CD56 by flow cytometry, and
3 matched cases expressed CD56 by immunohistochemical
analysis. In addition, several cases showed gain of expression
Am J Clin Pathol 2009;132:101-110 103
103 DOI: 10.1309/AJCP6GR8BDEXPKHR 103
Cronin et al / Updated Approach to Leukemia Cutis

zTable 3z
Demographic, Clinical, Bone Marrow Biopsy, and Cytogenetic Data

Case No./ Clinical Time of Diagnostic Interim

Sex/Age (y)* Appearance Site Bone Marrow Biopsy Therapy

1/F/88 Nonhealing biopsy site Unknown Unknown Unknown

2/F/6 wk Infiltrative violet plaques Back 5 mo after skin biopsy No
  
3/F/63 Firm, lobulated violaceous nodule Posterior part 2 wk before skin biopsy Unknown
   of arm   
4/M/62 Pruritic, reddish-brown nodules Chest Concurrent† No
5/M/61 New nodules on skin Chest 2 wk before skin biopsy Yes
6/M/49 Erythematous, blanchable excoriated papules Chest, back 1 y after skin biopsy Yes
7/M/24 Rapidly growing, firm dermal nodules Left arm 5 mo before skin biopsy Yes
8/M/59 Erythematous papules Inguinal 2 mo before skin biopsy Yes
  

Downloaded from https://academic.oup.com/ajcp/article/132/1/101/1766248 by guest on 05 May 2021

9/M/69 Widespread, tumid, red plaques Neck Unknown Unknown
10/F/90 Unknown Abdomen Unknown Unknown
11/F/68 Indurated plaques and nodules Scalp Unknown Unknown
12/M/73 Papule Hand 9 mo before skin biopsy Yes
13/M/65 Acneiform lesions Shoulder Concurrent† No
14/M/75 Indurated erythematous to violaceous nodules Arm, leg Concurrent† No
15/M/65 Single purpuric lesion Leg 2 mo before skin biopsy Yes
  
16/M/5 mo Violaceous papule/nodules Back 1 mo before skin biopsy Yes
17/M/57 2-5 mm granulomatous papules to 1-cm nodules Shoulder Concurrent† No
18/M/53 Mild, violaceous papules and nodules Chest, arm 8 mo before skin biopsy Yes
19/M/72 Purpuric eruption Face, arm Concurrent † No
20/F/34 Infiltrative, crusted scalp lesions Scalp 2 mo before skin biopsy Yes
21/F/58 Inflamed, indurated, discrete papules and nodules Epigastric 6 mo after skin biopsy Unknown
22/M/37 Ulcerative papules Scrotum 1 mo before skin biopsy Yes
23/M/75 Multiple lesions Arm Concurrent † No
  
24/F/60 Unknown Neck Concurrent† No
25/M/64 Abdominal rash Abdomen Preexisting atypical CML Unknown
   (limited information)
26/F/56 Blue dermal nodules Right arm Preexisting AML (limited info) Unknown
   chest
27/F/27 Asymptomatic 1.5-cm, bluish-red dermal nodules Lateral part 14 mo before skin biopsy Yes
   of chest
28/M/35 Tender subcutaneous nodules Calf 8 mo before skin biopsy Yes
29/F/64 Generalized, itchy, and urticarial papules Chest Unknown Unknown
30/F/22 d Annular plaques Scalp Normal concurrent† marrow No
  
31/F/67 Subcutaneous, nonscaly nodules Chest 5 mo before skin biopsy Yes
32/M/71 Progressive erythematous nodules Abdomen 3 mo before skin biopsy Yes
33/M/20 Nontender, blue-green nodules, 3-6 mm Arm, leg Relapse 5-6 wk before skin biopsy Yes

aCML, atypical chronic myeloid leukemia; AML, acute myeloid leukemia; CML, chronic myeloid leukemia; CMML, chronic myelomonocytic leukemia; ETO, eight twenty-one
gene; FISH, fluorescent in situ hybridization; MDS, myelodysplastic syndrome; MLL, mixed lineage leukemia gene; MRC, myelodysplasia-related changes; MPD, myelo-
proliferative disorder; NOS, not otherwise specified; PMF, primary myelofibrosis; PML-RARA, promyelocytic leukemia-retinoic acid receptor α gene; RAEB, refractory
anemia with excess blasts; WHO, World Health Organization.
* Unless otherwise indicated.
† Concurrent studies were defined as skin and diagnostic bone marrow biopsies performed within 2 weeks of each other.

of different antigens when the flow cytometric analysis of the
bone marrow was compared with antigen expression in skin.
In 1 case, we found that the skin blasts expressed CD56, even
though this was not documented in the flow cytometric data
analysis (case 33). Similarly, in 1 case, MPO was expressed
by the skin blasts but was not expressed when the bone mar-
row was analyzed by flow cytometry (case 12).
Discussion
While myeloid LC is a relatively rare manifestation of
acute leukemia, its development can pose difficult diagnostic
problems. In most cases, the patients are known to have sys-
temic leukemia with circulating blasts. In these cases, we have
found it expedient to correlate flow cytometric information of
systemic acute leukemia (obtained from bone marrow and/or
peripheral blood studies) to support the diagnosis of myeloid
LC. In some cases, however, myeloid LC may be the first
manifestation of acute leukemia, in which the bone marrow
biopsy demonstrates a precursor lesion (myeloproliferative or
myelodysplastic syndrome) or normal findings. In these cases,
definitive diagnosis and/or classification of the disorder using
current criteria requires immunohistochemical analysis of the
skin lesions and correlation with cytogenetic information.

104 Am J Clin Pathol 2009;132:101-110 © American Society for Clinical Pathology

104 DOI: 10.1309/AJCP6GR8BDEXPKHR
 Anatomic Pathology / Original Article

overrepresented in our study, the majority are not of mono-

cytic lineage, and a significant number expressed CD34 on
WHO Bone Cytogenetics/ bone marrow analysis.
Marrow Diagnosis Molecular Data A second possible explanation for discordant antigen
expression may be related to clinical therapy. In several cases
AML, NOS; arising from MPD Unknown
AML with 11q23 46,XX,+8,t(11;19)(q23;p13.3), in our study, there was interim therapy (usually systemic
   +16,+20[19] 50,idem,+8[5] chemotherapy) between the bone marrow biopsy and the diag-
AML with MRC; progression Unknown
   of CMML
nosis of LC (15 of 33 cases; Table 3). There is, therefore, a
Normal bone marrow Unknown possibility that the blasts manifesting as LC represent pheno-
AML with MRC Unknown typically distinct populations selected for by treatment or that
Acute monocytic leukemia 46,XY
Acute monocytic leukemia Complex karyotype with tetrasomy 8 the immunophenotype has been otherwise altered by therapy.
AML with MRC; arising from Unknown The phenomenon of an immunophenotypically distinct blast
   PMF

Downloaded from https://academic.oup.com/ajcp/article/132/1/101/1766248 by guest on 05 May 2021

Unknown Unknown subpopulation that persists following chemotherapy has been
Unknown Unknown attributed to persistence of “minimal residual disease,” the
Unknown Unknown
RAEB-2 46,XY idea being that this subpopulation may account for clinical
Normal bone marrow 46,XY relapse and may display a different phenotype than was char-
AML with MRC Unknown
AML with MRC 46,XY,del(1),(q21q25),del(3)
acterized before initiation of therapy.18-21
   (q23),del(18)(q22)[17]/46,XY[3] A third explanation may be that the blasts in myeloid
Acute monocytic leukemia MLL negative by FISH LC express CD34 and/or CD117, but immunohistochemical
AML with maturation Unknown
AML with MRC Monosomy 7, trisomy 8 methods are not of sufficient sensitivity for their detection.
Acute monocytic leukemia Unknown Previous studies have documented the differing sensitivities
AML, NOS (limited info) Trisomy 8
AML, NOS; arising from PMF Unknown of flow cytometric and immunohistochemical methods in
Acute promyelocytic leukemia t(15;17) positive by FISH detecting expression of some antigens. For example, CD117
AML with MRC; arising from 38-51,XY,+8,+8,+21,+2mar[cp20]
   CMML has been shown to have increased sensitivity when detected
AML with 11q23 11q23 by flow cytometric vs immunohistochemical methods.22-24
Mixed MDS/MPD, favor aCML BCR-ABL negative by FISH
Other studies, however, have shown that flow cytometric and
AML, NOS (limited information) Unknown, immunohistochemical studies generate reproducible immuno-
phenotyping in acute leukemias diagnosed by bone marrow
AML without maturation 46,XX
biopsy, especially with regard to CD34 expression.22,23,25-27
CML t(9;22)(q34;q11.2) Many of our cases expressed CD34 or CD117 in tempo-
Unknown Unknown
Normal bone marrow 46,XX 4q deletion, MLL and ETO rally concurrent bone marrow analysis but were frequently
   negative by FISH negative for these markers in the skin. In at least 1 case (case
AML without maturation 46,XX, PML-RARA negative by FISH
CMML-1 Unknown 21), immunohistochemical analysis for CD34 was performed
Acute monocytic leukemia 46,XY on the bone marrow and skin lesion and showed blast expres-
sion of CD34 in the bone marrow but not in the skin. This is
not necessarily linked to titering differences, as titers directed
specifically toward blast expression failed to demonstrate an
increase in staining in skin lesions. What we document for the
first time in our studies, therefore, is relative loss of CD34
and CD117 in skin, in comparison with temporally concurrent
We also confirmed that immunophenotypic markers bone marrow analysis. In these cases, these differences do not
normally used to define blasts in the bone marrow (ie, CD34 seem to be necessarily linked to therapeutic changes or differ-
and/or CD117) could not be used in the setting of skin lesions. ences in antigenic detection with differing methods.
These findings are similar to those previously reported in the In addition to CD34 and/or CD117 discrepancies, we
literature.10,13 In our study, only a fraction of cases of myeloid noted discordance of expression of MPO and/or CD56 in skin
LC expressed CD34 or CD117, even when temporally coinci- lesions relative to the immunophenotype of the bone mar-
dent flow cytometric analysis of the bone marrow showed that row. We saw a gain of MPO and/or CD56 in a minority of
bone marrow blasts expressed these antigens (Table 5). Several cases. While some of these findings may be related to therapy
possibilities are hypothesized to explain this discordance. and/or sensitivity differences in detection methods, it is also
One argument may be that acute leukemias that involve possible that an immunophenotypically distinct subgroup of
the skin are often of monocytic lineage, and these usually lack blasts occupies extramedullary sites, with gain or loss of cer-
expression of CD34.13 While acute monocytic leukemias are tain antigens in the process of extramedullary involvement.

© American Society for Clinical Pathology Am J Clin Pathol 2009;132:101-110 105

105 DOI: 10.1309/AJCP6GR8BDEXPKHR 105
Cronin et al / Updated Approach to Leukemia Cutis

A B

Downloaded from https://academic.oup.com/ajcp/article/132/1/101/1766248 by guest on 05 May 2021

C D

E F

zImage 1z Leukemia cutis. A, Low power examination of H&E-stained sections shows a primarily interstitial infiltrate of cells
separated from the overlying epidermis by a grenz zone (×100). B, Higher power examination reveals a monomorphous population
of cytologically malignant cells with increased nuclear/cytoplasmic ratios and significant mitotic activity (H&E, ×400). Inset, Very high
power shows nuclei with finely dispersed chromatin, multiple nucleoli, and irregular nuclear contours (×1,000 oil immersion). C-F,
The leukemic cells express CD43 (C, ×600), CD68 (E, ×600), and CD56 (F, ×600) but are negative for myeloperoxidase (D, ×600).

106 Am J Clin Pathol 2009;132:101-110 © American Society for Clinical Pathology

106 DOI: 10.1309/AJCP6GR8BDEXPKHR
 Anatomic Pathology / Original Article

Further work in this area may yield very interesting results zTable 4z
Immunohistochemical Profile of Myeloid Leukemia Cutis*
about the pathogenesis of extramedullary involvement by
acute leukemia. Acute Monocytic
CD163 is an acute phase–regulated transmembrane pro- Antibody
All Cases Leukemia AML With MRC

tein that binds haptoglobin-hemoglobin complexes and is CD43 32/33 (97) 5/5 (100) 7/7 (100)
implicated as a hemoglobin scavenger receptor.28,29 It is CD68 31/33 (94) 5/5 (100) 7/7 (100)
thought to be relatively specific for monocytes and tissue
MPO 14/33 (42) 1/5 (20) 1/7 (14)
CD56 14/30 (47) 2/5 (40) 5/7 (71)
macrophages and has been proposed to function in the innate CD163 7/28 (25) 3/5 (60) 1/6 (17)
immune response and resolution of inflammation. Preliminary CD117 (total) 3/30 (10) 0/5 (0) 1/6 (17)
CD34 (total)
2/31 (6) 1/5 (20) 1/7 (14)
studies have shown that CD163 has more specificity for his- CD20
0/33 (0) 0/5 (0) 0/7 (0)
tiocytes than CD68, which is an organelle-specific marker CD3
0/33 (0) 0/5 (0) 0/7 (0)

that stains lysosomes. In our study, this marker was found to AML, acute myeloid leukemia; MPO, myeloperoxidase; MRC, myelodysplasia-

Downloaded from https://academic.oup.com/ajcp/article/132/1/101/1766248 by guest on 05 May 2021

related changes.
be present in only a minority of cases of myeloid LC but was * Data are given as number of cases positive/number of cases tested (percentage).

overrepresented in the group of cases of monocytic lineage.

This finding shows that while CD163 may not be helpful in
cases of unknown lineage, it may be useful in confirming
monocytic differentiation in infiltrates of myeloid LC.28
Given the discrepancy between what is usually observed
in the immunophenotyping of bone marrow specimens and
what is seen in skin lesions, we devised an updated approach
to immunophenotyping of myeloid LC that would confirm
this diagnosis zFigure 1z.30 We found that regardless of leuke-
mic phenotype or flow cytometric expression profile, a panel
of antibodies that included MPO, CD3, CD20, CD43, CD56,
CD68, and CD117 would correctly identify cases of myeloid
LC in the vast majority of cases. This approach would be
especially useful in the analysis of malignant cutaneous infil-
trates with immature morphologic features in which a hema-
tologic malignancy is suspected on morphologic grounds. In
this context, negative CD20 and CD3 results would exclude
most neoplasms of lymphocytic origin.
While MPO expression strongly supports the diagnosis
of myeloid LC in the context of malignant hematopoietic skin
infiltrates, we found that 58% of our cases were negative for
this marker.13 Cases that were MPO– (19/33) had the CD43+/
CD68+ immunophenotype. In this context, the differential
diagnosis would include myeloid LC, mast cell sarcoma, and
blastic plasmacytoid dendritic cell neoplasms. If the lesion is
MPO– and CD68–, one could consider CD20–/CD3– lym-
phomas, such as a B-cell lymphoma recurring after ritux-
imab therapy or CD30+ lymphoproliferative disorders. These
considerations can be confirmed by using CD79a or CD30,
respectively. Other markers that may be useful to identify dif-
ficult lymphoid neoplasms include PAX5 (B cells) and CD2
(T cells). The very rare case of acute lymphoblastic LC may
be identified with these additional markers.
We found CD56 and CD117 to be helpful in further
evaluation of MPO– cases that expressed CD43 and CD68.
In CD68+/MPO–/CD56– cases (9 of 19 cases), the differen-
tial diagnosis would include mast cell sarcoma and myeloid
LC. In these cases, mast cell tryptase would help distinguish

zTable 5z
Comparison of Cutaneous Immunohistochemical Expression With Bone Marrow Flow Cytometric Phenotype

Cutaneous Immunohistochemical Profile

Case No. MPO CD34 CD56 CD68 CD117 Flow Cytometric Expression Profile*

2 – –
– + – HLA-DR+, partial CD14+, CD33+, CD56+, CD64+
3 – –
+ + – MPO+, CD33>CD13, CD56+, CD64+, CD65w+
8 – –
+ + – HLA-DR+, weak MPO, CD14+, CD33>CD13, CD34+, CD56+, CD64+
12 + – ND – + HLA-DR+, CD13+, CD33+, CD34+, CD117+
16 – +
– + – HLA-DR+, CD13–, CD15+, CD33+, CD34+, CD56+, CD64+
19 – –
– + – Weak MPO, CD13+, CD15+, CD33+, CD34+, CD64+
21 + –
– + + HLA-DR+, MPO+, CD15+, CD13>CD33, CD34+, CD64+, CD117+
22 + –
– + – HLA-DR–, MPO+, CD13+, CD33+, CD34–, weak CD64, CD117+
24 – –
– + – MPO+, CD13+, CD15+, CD33+, CD34+, CD117+
25 + –
– + – HLA-DR+, MPO+, CD13+, CD15+, CD33+, CD34–, CD64+, CD117+
27 + –
– + – MPO+, CD13+, CD15+, CD56+, CD33+, weak CD34, CD117+
31 + –
+ + – MPO+, CD15+, CD33+, CD34–, CD56+, weak CD64, CD117+
33 + –
+ + – MPO+, CD4+, CD11c+, CD13+, CD33+, CD34+, weak CD64+, CD117+

MPO, myeloperoxidase; ND, not done; +, positive; –, negative.

* Bolded antigens were tested by immunohistochemical and flow cytometric analyses. Three patients with normal findings by bone marrow biopsy had accompanying negative

flow cytometric studies. One patient had only very limited flow cytometric data available (data not shown).

© American Society for Clinical Pathology Am J Clin Pathol 2009;132:101-110 107

107 DOI: 10.1309/AJCP6GR8BDEXPKHR 107
Cronin et al / Updated Approach to Leukemia Cutis

CD3–/CD20– malignant neoplasm

CD43+
(32/33)*
MPO+ MPO–
(13/32) (19/32)
Myeloid leukemia
cutis CD68+ CD68–
(19/19) (0/19)
Other hematologic malignancies
CD56+ CD56– (ie, CD30+ lymphoproliferative disorders,
(10/19) (9/19) blastic plasmacytoid dendritic cell neoplasm,
Myeloid leukemia cutis, other CD3–/CD20– lymphoma/leukemia)
consider blastic plasmacytoid CD117+ (0/9) (9/9) CD117–
dendritic cell neoplasm

Downloaded from https://academic.oup.com/ajcp/article/132/1/101/1766248 by guest on 05 May 2021

Myeloid leukemia cutis, Myeloid leukemia cutis
consider mast cell sarcoma

zFigure 1z Proposed algorithm for the immunohistochemical diagnosis of leukemia cutis (LC). The numbers shown in brackets
represent the number of cases positive or negative for each marker over the number of test cases in that arm of the algorithm.
* One case (case 22) was negative for CD43, but a myeloperoxidase (MPO) stain was positive, confirming myeloid LC. The

concurrent bone marrow biopsy was diagnostic of acute promyelocytic leukemia.

between myeloid LC and mast cell sarcomas.31 In the CD68+/
MPO–/CD56+ cases (10 of 19 cases), staining for CD4 and
CD123 will often distinguish between myeloid LC and blastic
plasmacytoid dendritic cell neoplasms.32
To determine if we could find cases of blastic plasma-
cytoid dendritic cell neoplasms that had been initially classi-
fied as myeloid LC, we tested 9 of the 10 cases (cases with
material available for further study) that were CD68+/MPO–/
CD56+ with CD4. Only 1 case (case 5) had an overlapping
immunophenotype, being MPO–/CD56+/CD4+. The corre-
sponding bone marrow biopsy in case 5 confirmed myeloid
differentiation, demonstrating acute myeloid leukemia with
myelodysplasia-related changes. In addition, we noted that
expression of CD4 in this case was weak, in contrast with the
strong and uniform staining seen with blastic plasmacytoid
dendritic cell neoplasms. However, it is important to note that
differentiating between these 2 entities is not always possible
using histologic findings and paraffin-based immunohis-
tochemical stains.33 Markers more recently reported to be of
use in characterizing blastic plasmacytoid dendritic cell neo-
plasms include BDCA-2, TCL1, CLA, and MxA, although
the WHO recommends “acute leukemia of ambiguous lin-
eage” for tumors having some but not all of the immunophe-
notypic characteristics of these tumors.14,32-36 Frozen section
staining with CD4 may also prove useful in distinguishing
these entities.37 Therefore, distinguishing between myeloid
LC and blastic plasmacytoid dendritic cell neoplasms con-
fined to the skin may require extensive immunohistochemical
studies, with ambiguous cases best classified as acute leu-
kemia of ambiguous lineage. Regardless, cutaneous CD56+
hematologic neoplasms as a group are known to have a poor
prognosis.38-40
While an extensive workup may be indicated in patients
with LC, a more abbreviated panel may be sufficient for
patients who have a documented diagnosis of systemic leu-
kemia. In these patients, it is less critical to exclude other
malignancies such as nonhematologic malignancies and
lymphomas because the likelihood of a second malignancy
is relatively low. Expression of MPO alone or a combination
of CD43 and CD68 expression in the face of MPO negativ-
ity may be sufficient to identify most cases of myeloid LC in
these circumstances.
This study demonstrated that the diagnosis of myeloid
LC is best made in conjunction with bone marrow findings
and flow cytometry because markers thought to be specific
for blasts (ie, CD34 and CD117) are not detected by immu-
nohistochemical staining of myeloid LC in the vast majority
of cases. We also propose an immunophenotypic algorithm
that correctly classified all 33 cases in our series. There can
be some immunophenotypic overlap between myeloid LC
and blastic plasmacytoid dendritic cell neoplasms, which
may be rectified by examining the findings in the correspond-
ing bone marrow. We were surprised to find such disparate
immunophenotypic profiles between the skin infiltrates and
the bone marrow lesions, with loss and gain of antigens
documented. Whether this represents a true alteration in the
immunophenotype of the blasts present in myeloid LC (due to
prior chemotherapy or other factors) or simply sensitivity dif-
ferences between flow cytometric and immunohistochemical
antigen detection has yet to be determined. Because of these

108 Am J Clin Pathol 2009;132:101-110 © American Society for Clinical Pathology

108 DOI: 10.1309/AJCP6GR8BDEXPKHR

differences, caution should be used when drawing diagnostic
conclusions based on the comparison of these profiles.
From the Departments of 1Pathology and 2Dermatology, Stanford
University, Stanford, CA.
Supported by the Stanford Department of Pathology.
Address reprint requests to Dr Cronin: Dept of Pathology,
Stanford University, 300 Pasteur Dr, Lane 235, Stanford, CA
94305.
Acknowledgments: We thank Amarjeet Grewall for
laboratory assistance and Anet James for assistance with the
graphics.
References
1. Weedon D. Cutaneous infiltrates of other leukemias/
lymphomas. In: Skin Pathology. 2nd ed. New York, NY:
Churchill Livingstone; 2002:1118-1119.
2. Byrd JC, Edenfield WJ, Shields DJ, et al. Extramedullary
myeloid cell tumors in acute nonlymphocytic leukemia: a
clinical review. J Clin Oncol. 1995;13:1800-1816.
3. Agis H, Weltermann A, Fonatsch C, et al. A comparative
study on demographic, hematologic, and cytogenetic findings
and prognosis in acute myeloid leukemia with and without
leukemia cutis. Ann Hematol. 2002;81:90-95.
4. Baer MR, Barcos M, Farrell H, et al. Acute myelogenous
leukemia with leukemia cutis: eighteen cases seen between
1969 and 1986. Cancer. 1989;63:2192-2200.
5. Seymour JF, Pierce SA, Kantarjian HM, et al. Investigation of
karyotypic, morphologic, and clinical features in patients with
acute myeloid leukemia blast cells expressing the neural cell
adhesion molecule (CD56). Leukemia. 1994;8:823-826.
6. Su WP, Buechner SA, Li CY. Clinicopathologic correlations in
leukemia cutis. J Am Acad Dermatol. 1984;11:121-128.
7. Buechner SA, Li CY, Su WP. Leukemia cutis: a histopathologic
study of 42 cases. Am J Dermatopathol. 1985;7:109-119.
8. Roth MJ, Medeiros LJ, Elenitoba-Johnson K, et al.
Extramedullary myeloid cell tumors: an immunohistochemical
study of 29 cases using routinely fixed and processed
paraffin-embedded tissue sections. Arch Pathol Lab Med.
1995;119:790-798.
9. Ratnam KV, Ziesmer SC, Li CY, et al. Value of
immunohistochemistry in the diagnosis of leukemia cutis:
study of 54 cases using paraffin-section markers. J Cutan
Pathol. 1992;19:193-200.
10. Kaddu S, Zenahlik P, Beham-Schmid C, et al. Specific
cutaneous infiltrates in patients with myelogenous leukemia:
a clinicopathologic study of 26 patients with assessment of
diagnostic criteria. J Am Acad Dermatol. 1999;40:966-978.
11. Sepp N, Radaszkiewicz T, Meijer CJ, et al. Specific
skin manifestations in acute leukemia with monocytic
differentiation: a morphologic and immunohistochemical
study of 11 cases. Cancer. 1993;71:124-132.
12. Davey FR, Olson S, Kurec AS, et al. The immunophenotyping
of extramedullary myeloid cell tumors in paraffin-embedded
tissue sections. Am J Surg Pathol. 1988;12:699-707.
13. Cibull TL, Thomas AB, O’Malley DP, et al. Myeloid leukemia
cutis: a histologic and immunohistochemical review. J Cutan
Pathol. 2008;35:180-185.
© American Society for Clinical Pathology
Anatomic Pathology / Original Article
14. Swerdlow SH, Campo E, Harris NL, et al, eds. WHO
Classification of Tumours of Haematopoietic and Lymphoid
Tissues. 4th ed. Lyon, France: IARC Press; 2008. WHO
Classification of Tumours; vol 2.
15. Sundram U, Harvell JD, Rouse RV, et al. Expression of the
B-cell proliferation marker MUM1 by melanocytic lesions and
comparison with S100, gp100 (HMB45), and MelanA. Mod
Pathol. 2003;16:802-810.
16. Sen F, Zhang XX, Prieto VG, et al. Increased incidence of
trisomy 8 in acute myeloid leukemia with skin infiltration
(leukemia cutis). Diagn Mol Pathol. 2000;9:190-194.
17. Ferrara F, Cancemi D, Friso P, et al. Tetrasomy 8 and
t(1;11)(p32;q24) in acute myelo-monocytic leukemia with
extensive leukemic cutaneous involvement. Leuk Lymphoma.
1996;20:513-515.
Downloaded from https://academic.oup.com/ajcp/article/132/1/101/1766248 by guest on 05 May 2021
18. Macedo A, San Miguel JF, Vidriales MB, et al. Phenotypic
changes in acute myeloid leukaemia: implications in
the detection of minimal residual disease. J Clin Pathol.
1996;49:15-18.
19. Oelschlagel U, Nowak R, Schaub A, et al. Shift of aberrant
antigen expression at relapse or at treatment failure in acute
leukemia. Cytometry. 2000;42:247-253.
20. Baer MR, Stewart CC, Dodge RK, et al. High frequency of
immunophenotype changes in acute myeloid leukemia at
relapse: implications for residual disease detection (Cancer and
Leukemia Group B Study 8361). Blood. 2001;97:3574-3580.
21. Langebrake C, Brinkmann I, Teigler-Schlegel A, et al.
Immunophenotypic differences between diagnosis and relapse
in childhood AML: implications for MRD monitoring.
Cytometry B Clin Cytom. 2005;63:1-9.
22. Dunphy CH, Polski JM, Evans HL, et al. Evaluation of bone
marrow specimens with acute myelogenous leukemia for
CD34, CD15, CD117, and myeloperoxidase. Arch Pathol Lab
Med. 2001;125:1063-1069.
23. Dunphy CH, O’Malley DP, Perkins SL, et al. Analysis of
immunohistochemical markers in bone marrow sections
to evaluate for myelodysplastic syndromes and acute
myeloid leukemias. Appl Immunohistochem Mol Morphol.
2007;15:154-159.
24. Arber DA, Jenkins KA. Paraffin section immunophenotyping
of acute leukemias in bone marrow specimens. Am J Clin
Pathol. 1996;106:462-468.
25. Hanson CA, Ross CW, Schnitzer B. Anti-CD34 immuno-
peroxidase staining in paraffin sections of acute leukemia:
comparison with flow cytometric immunophenotyping.
Hum Pathol. 1992;23:26-32.
26. Kanter-Lewensohn L, Hellstrom-Lindberg E, Kock Y, et
al. Analysis of CD34-positive cells in bone marrow from
patients with myelodysplastic syndromes and acute myeloid
leukemia and in normal individuals: a comparison between
FACS analysis and immunohistochemistry. Eur J Haematol.
1996;56:124-129.
27. Manaloor EJ, Neiman RS, Heilman DK, et al. Immuno-
histochemistry can be used to subtype acute myeloid leukemia
in routinely processed bone marrow biopsy specimens:
comparison with flow cytometry. Am J Clin Pathol.
2000;113:814-822.
28. Nguyen TT, Schwartz EJ, West RB, et al. Expression of CD163
(hemoglobin scavenger receptor) in normal tissues, lymphomas,
carcinomas, and sarcomas is largely restricted to the monocyte/
macrophage lineage. Am J Surg Pathol. 2005;29:617-624.
29. Sachdev R, Sundram U. Expression of CD163 in
dermatofibroma, cellular fibrous histiocytoma, and
dermatofibrosarcoma protuberans: comparison with CD68,
CD34, and factor XIIIa. J Cutan Pathol. 2006;33:353-360.
Am J Clin Pathol 2009;132:101-110 109
109 DOI: 10.1309/AJCP6GR8BDEXPKHR 109
Cronin et al / Updated Approach to Leukemia Cutis
30. Cho-Vega JH, Medeiros LJ, Prieto VG, et al. Leukemia cutis.
Am J Clin Pathol. 2008;129:130-142.
31. Sundram UN, Natkunam Y. Mast cell tryptase and
microphthalmia transcription factor effectively discriminate
cutaneous mast cell disease from myeloid leukemia cutis.
J Cutan Pathol. 2007;34:289-295.
32. Petrella T, Bagot M, Willemze R, et al. Blastic NK-cell
lymphomas (agranular CD4+CD56+ hematodermic
neoplasms): a review. Am J Clin Pathol. 2005;123:662-675.
33. Pilichowska ME, Fleming MD, Pinkus JL, et al. CD4+/
C56+ hematodermic neoplasm (“blastic natural killer cell
lymphoma”): neoplastic cells express the immature dendritic
cell marker BDCA-2 and produce interferon. Am J Clin Pathol.
2007;128:445-453.
34. Urosevic M, Conrad C, Kamarashev J, et al. CD4+CD56+
hematodermic neoplasms bear a plasmacytoid dendritic cell
phenotype. Hum Pathol. 2005;36:1020-1024.
35. Gopsca L, Banyai A, Jakab K, et al. Extensive flow cytometric
characterization of plasmacytoid dendritic cell leukemia cells.
Eur J Haematol. 2005;75:346-351.
36. Herling M, Teitell MA, Shen RR, et al. TCL1
expression in plasmacytoid dendritic cells (DC2s) and
the related CD4+CD56+ blastic tumors of skin. Blood.
2003;101:5007-5009.
110 Am J Clin Pathol 2009;132:101-110
110 DOI: 10.1309/AJCP6GR8BDEXPKHR
37. Ascani S, Massone C, Ferrara G, et al. CD4-negative variant of
CD4+/CD56+ hematodermic neoplasm: description of three
cases [published online ahead of print May 20, 2008]. J Cutan
Pathol. 2008;35:911-915.
38. Assaf C, Gellrich S, Whittaker S, et al. CD56-positive
haematological neoplasms of the skin: a multicentre study of
the Cutaneous Lymphoma Project Group of the European
Organisation for Research and Treatment of Cancer. J Clin
Pathol. 2007;60:981-989.
39. Bekkenk MV, Jansen PM, Meijer CJ, et al. CD56+
hematological neoplasms presenting in the skin: a retrospective
analysis of 23 new cases and 130 cases from the literature. Ann
Oncol. 2004;15:1097-1108.
40. Chan JK, Sin VC, Wong KF, et al. Nonnasal lymphoma
expressing the natural killer cell marker CD56: a
Downloaded from https://academic.oup.com/ajcp/article/132/1/101/1766248 by guest on 05 May 2021
clinicopathologic study of 49 cases of an uncommon aggressive
neoplasm. Blood. 1997;89:4501-4513.
© American Society for Clinical Pathology