Abstract

Objectives

E-cadherin, epithelial calcium-dependent cell adhesion protein, has been identified as a marker of immature erythroid precursors in recent years. However, the specificity of E-cadherin in bone marrow specimens for erythroblasts vs myeloblasts or other early hematopoietic precursors in acute myeloid leukemia (AML) and myelodysplastic syndromes (MDS) has not been fully elucidated.

Methods

We analyzed 105 cases of AML and MDS to evaluate the specificity of E-cadherin.

Results

Of 84 cases of AML, including cases with megakaryocytic, erythroid, monocytic, and granulocytic differentiation, all five acute erythroleukemia cases were positive, as well as one case of megakaryoblastic leukemia that showed coexpression of glycophorin A. In addition, we demonstrate that a panel of three markers, E-cadherin, CD117, and CD34, is effective in identifying lineage-specific myeloblasts in cases of MDS where left-shifted erythroid hyperplasia may complicate morphologic assessment of myeloblasts.

Conclusions

In marrow specimens, E-cadherin is a useful marker for erythroid differentation.

E-cadherin (epithelial calcium-dependent adhesion protein) is a transmembrane protein expressed primarily in epithelial cells.1 In normal epithelial tissues, e-cadherin acts to establish and maintain cell-to-cell adhesion and is also a central player in the morphogenesis of cells during migration.2 Because its expression pattern is limited to cells of epithelial origin, E-cadherin is frequently used as a marker to identify cells undergoing or having undergone epithelial differentiation.3 Interestingly, loss of E-cadherin, in some solid tumor subtypes, is thought to enable these neoplastic cells to metastasize more readily.4

In recent years, E-cadherin also has been identified as a protein expressed in hematopoietic erythroid precursors in a developmentally regulated pattern; it is present in early erythroblasts and gradually downregulated as erythroid cells mature.5,6 E-cadherin has also been shown to be critical in erythroid development because its inhibition results in disruption of erythroid maturation.5 It is speculated that E-cadherin may play a role in the development of erythroid cells by augmenting specific cellular interactions and thus help establish an appropriate cellular microenvironment necessary for erythroid maturation.

More recently, E-cadherin has been investigated in cases of erythroleukemia and shown to be expressed in many such cases in the neoplastic erythroblasts.7 However, the true specificity of E-cadherin for blasts with erythroid differentiation vs myeloid differentiation in cases of acute myeloid leukemia (AML) and myelodysplastic syndromes (MDS) is not known. We analyzed 84 cases of AML and 21 cases of MDS to determine the specificity of E-cadherin for blasts with erythroid differentiation. Our results show that in the bone marrow, E-cadherin is highly specific and limited to cells with erythroid differentiation. Furthermore, in cases of MDS, we demonstrate that a panel of immunostains including E-cadherin as well as CD117, a transmembrane protein also known to mark immature myeloblasts and erythroblasts, in conjunction with CD34, allows for effective enumeration of myeloblasts.

Materials and Methods

Patient Cohort

Eighty-four cases of AML and 21 cases of MDS diagnosed at Stanford University Medical Center (Palo Alto, CA) with available histologic material were reviewed and classified according to 2008 World Health Organization criteria.8 Clinical data, peripheral blood smears, bone marrow aspirates, trephine biopsies, and results of karyotype and NPM1 and CEBPA mutation analysis and/or flow cytometry studies were reviewed. Exon 12 NPM1 insertion mutations and mutations in CEBPA were detected and analyzed as previously described.9 Blast percentages were based on a morphologic count of at least 200 cells in the bone marrow aspirate smears. This study was approved by Stanford University’s institutional review board.

Immunohistochemistry

Immunohistochemical staining was performed as previously described.10 In brief, 4-μm Bouin fixed, decalcified (Formical-4 solution, Decal, Tallman, NY) paraffin-embedded bone marrow core biopsy specimens were stained using BenchMarkXT (Ventana, Illkirch, France) or Leica-Bond Max processors (Leica Biosystems, Wetzlar, Germany). Antibodies to E-cadherin (Invitrogen; clone 4A2C7, Thermo Fisher Scientific, Waltham, MA) at a 1:25 dilution, CD34 (clone MY10, BD Biosciences, San Jose, CA) at a 1:40 dilution, glycophorin C (GPC) (clone Ret40f, Dako, Carpinteria, CA) at a 1:100 dilution, Linker for activation of T cells (LAT) at a 1:100 dilution (clone LAT-1, Dako), and CD117 (rabbit polyclonal A4502, Dako) at a 1:200 dilution were used.11 Percent staining of antibodies on cells was performed by manually counting individual cells stained vs nuclei on H&E-stained core biopsy bone marrow sections in three random high-power fields.

Flow Cytometry

Flow cytometry was performed as previously described using a FACSCanto II (BD Biosciences) cytometer with commercially available antibodies.12 Blasts were gated by characteristic CD45+(dim)/low-side scatter. Events reactive with each monoclonal antibody were determined by setting thresholds with isotypic controls.

Statistical Analysis

Fisher exact test, Student t test, and linear regression analysis were performed using XLSTAT software (Addinsoft, New York, NY). Differences between groups were considered statistically significant if P values were less than .05 in a two-tailed test.

Results

E-Cadherin Is Specific for Erythroid Differentiation in AML

Although E-cadherin has been identified as a marker of erythroblasts, the true specificity of E-cadherin for erythroblasts vs malignant myeloblasts is unknown. We studied 84 cases, including diverse subtypes, of AML to determine the specificity of E-cadherin for malignant blasts. Of these 84 cases, only six were positive; five cases of erythroleukemia were positive for E-cadherin Image 1 and one megakaryoblastic leukemia in a patient with Down syndrome was also positive Image 2. Of the five cases of erythroleukemia, four were of the erythroid/myeloid subtype and one was a case of pure erythroleukemia. The megakaryoblastic leukemia case that was positive for E-cadherin staining had malignant blasts that were negative for GPC on immunohistochemistry (data not shown). This case demonstrated other evidence of erythroid differentiation, specifically glycophorin A (GPA) expression on flow cytometry (Image 2). In all instances with positive E-cadherin staining, more than 50% of immature cells were positive with variable but generally moderate staining (Image 1 and Image 2). In all other AML cases, E-cadherin positive cells were less than 10% of immature cells seen Table 1.

Image 1

Immunohistochemical staining of E-cadherin in acute myeloid leukemia (AML). A, Wright-Giemsa–stained bone marrow aspirate smear of AML with myelodysplasia-related changes (AML-MRC). B, H&E-staining of AML-MRC biopsy specimen. C, E-cadherin staining in AML-MRC. Blasts are uniformly negative. Inset is a small erythroid island. Arrow denotes faint nuclear staining in a blast, but not membranous staining. D, Wright-Giemsa–stained bone marrow aspirate smear of a case of pure erythroleukemia. E, H&E staining of pure erythroleukemia. F, E-cadherin staining in a case of pure erythroleukemia. (All images, ×1,000.)

Image 1

Immunohistochemical staining of E-cadherin in acute myeloid leukemia (AML). A, Wright-Giemsa–stained bone marrow aspirate smear of AML with myelodysplasia-related changes (AML-MRC). B, H&E-staining of AML-MRC biopsy specimen. C, E-cadherin staining in AML-MRC. Blasts are uniformly negative. Inset is a small erythroid island. Arrow denotes faint nuclear staining in a blast, but not membranous staining. D, Wright-Giemsa–stained bone marrow aspirate smear of a case of pure erythroleukemia. E, H&E staining of pure erythroleukemia. F, E-cadherin staining in a case of pure erythroleukemia. (All images, ×1,000.)

Image 2

Megakaryoblastic leukemia associated with Down syndrome with partial erythroid differentiation showing E-cadherin expression. A, H&E-stained bone marrow core biopsy specimen. B, E-cadherin staining. C, CD117 staining. D, Linker for activation of T-cells staining. E, Flow cytometry plot demonstrating glycophorin A (GPA) and CD41a expression on blasts. (All images, ×1,000.)

Image 2

Megakaryoblastic leukemia associated with Down syndrome with partial erythroid differentiation showing E-cadherin expression. A, H&E-stained bone marrow core biopsy specimen. B, E-cadherin staining. C, CD117 staining. D, Linker for activation of T-cells staining. E, Flow cytometry plot demonstrating glycophorin A (GPA) and CD41a expression on blasts. (All images, ×1,000.)

Table 1

E-Cadherin Staining in Cases of AML

E-Cadherin in Combination With CD34 and CD117 Can Be Helpful in Qualifying Myeloblasts vs Erythroblasts

Cases of MDS often have an erythroid hyperplasia with a left-shift that can make enumeration of myeloblasts difficult. Although CD34 can be used to enumerate blasts on bone marrow core biopsy sections, in some cases, blasts can be CD34 negative, but will retain CD117 expression. However, CD117 also marks immature erythroid cells and can give a falsely elevated blast percentage, if assumed to be a direct indication of blast numbers in cases of MDS with erythroid hyperplasia and left-shift.

Given the specificity of E-cadherin for immature cells of erythroid lineage, we wanted to determine if E-cadherin in conjunction with CD117 as well as CD34 could be used to enumerate myeloblasts vs early erythroid precursors in cases of MDS, and be useful in confirming or establishing blast counts. We stained core bone marrow biopsy specimens in 21 MDS cases with CD34, E-cadherin, CD117, and GPC. We assessed the staining patterns of these markers, and correlated them with morphologic blast percentages. Specifically, we correlated morphologic blast percentages with CD34 percentages, as well as morphologic blast percentages with CD117 percentages minus E-cadherin percentages (abbreviated as CD117% – E-cad%); CD117 bright mast cells were not counted among the CD117 percentages.

In almost all of our 21 MDS cases (6 RCMD, 6 RAEB-1, and 9 RAEB-2) a subset of immature mononuclear cells with morphologic features of early erythroid precursors was positive for E-cadherin as well as CD117 Image 3 and Image 4. However, E-cadherin and CD117 primarily stained the most immature erythroid cells unlike GPC, which stained more mature erythroid precursors and non-nucleated RBCs (Image 3). In fact, in many cases, compared with the E-cadherin or CD117 staining, the GPC staining patterns in these erythroid islands were near-negative images of each other (Image 3).

Image 3

E-cadherin, CD117, glycophorin C (GPC), and CD34 staining in an erythroid island in a case of myelodysplastic syndrome (MDS). A, H&E-stained bone marrow core biopsy specimen of a case of MDS. Note the immature and maturing erythroid precursors. B, CD34 staining in this erythroid island. C, CD117 staining that marks immature erythroid precursors. D, E-cadherin stains the most immature erythroid cells but not more mature erythroid precursors. E, GPC staining can be seen to highlight more mature erythroid precursors and nonnucleated RBCs, but is either negative or faintly positive on the most immature erythroid precursors. (All images, ×1,000.)

Image 3

E-cadherin, CD117, glycophorin C (GPC), and CD34 staining in an erythroid island in a case of myelodysplastic syndrome (MDS). A, H&E-stained bone marrow core biopsy specimen of a case of MDS. Note the immature and maturing erythroid precursors. B, CD34 staining in this erythroid island. C, CD117 staining that marks immature erythroid precursors. D, E-cadherin stains the most immature erythroid cells but not more mature erythroid precursors. E, GPC staining can be seen to highlight more mature erythroid precursors and nonnucleated RBCs, but is either negative or faintly positive on the most immature erythroid precursors. (All images, ×1,000.)

Image 4

CD117 and E-cadherin staining in a case of refractory anemia with excess blasts-2 (RAEB-2) in which blasts were only partially CD34 positive. A, H&E-stained section of a case of RAEB-2. B, CD34 staining highlights two blasts side by side. C, CD117 staining. D, E-cadherin staining. E, glycophorin C staining. (All images, ×1,000.)

Image 4

CD117 and E-cadherin staining in a case of refractory anemia with excess blasts-2 (RAEB-2) in which blasts were only partially CD34 positive. A, H&E-stained section of a case of RAEB-2. B, CD34 staining highlights two blasts side by side. C, CD117 staining. D, E-cadherin staining. E, glycophorin C staining. (All images, ×1,000.)

When assessing the significance of CD34 percentages and the CD117% – E-cad% values in cases of MDS, a positive correlation between morphologic blast percentages and CD34 percentages was seen on linear regression analysis (R2 = 0.645) Figure 1A. Although in most of the 21 cases, subclassifying cases of MDS based on CD34 percentages would have led to appropriate stratification (RCMD vs RAEB-1 or RAEB-2), in 3 of 21 cases the CD34 percentages were an underestimation and if used alone, would have resulted in a lower-grade MDS than the final diagnosis Table 2 (Image 4 and Figure 1).

Figure 1

A, Linear regression analysis of CD34 vs morphologic blasts. B, Linear regression analysis of CD117 – E-cadherin (E-cad) vs morphologic blasts.

Figure 1

A, Linear regression analysis of CD34 vs morphologic blasts. B, Linear regression analysis of CD117 – E-cadherin (E-cad) vs morphologic blasts.

Table 2

Percentage of Cells Staining for CD34, CD117, or E-Cadherin in Core Biopsy Sectionsa

We also saw a positive correlation between morphologic blast percentages and CD117% – E-cad% (R2 = 0.505) (Figure 1B). Again as with CD34 percentages, subclassifying cases of MDS based on CD117% – E-cad% would have led to appropriate stratification (RCMD vs RAEB-1 or RAEB-2), in most cases of MDS. However, in 4 of 21 cases, the CD117% – E-cad% was an overestimation, and if used alone, would have led to a higher-grade MDS than the final diagnosis. Conversely, in 1 of the 21 cases, the CD117% – E-cad% was variable enough that, if based on the CD117% – E-cad%, the final diagnosis would have been a lower-grade MDS.

Interestingly, in two cases of MDS, in which only a subset of blasts were positive for CD34 (4% and 7% of cells, respectively), the CD117% – E-cad% (11% and 12% of all cells, respectively) was a more accurate enumeration of blasts (15% and 12% on morphology) (Image 4, Table 2). In two other MDS cases, E-cadherin percentages exceeded CD117 percentages, resulting in a CD117% – E-cad% of less than 0 (−4% and −3%); both cases, however, had less than 1% blasts. Finally, all 21 cases of MDS were correctly stratified, and were based on either the CD34 percentage or CD117% – E-cad%.

Discussion

The expression of E-cadherin in erythroid cells is developmentally controlled, present in early erythroid cells but not in later more mature forms. Recently, E-cadherin was identified as a protein expressed in a majority of cases of erythroleukemia; however, no study to date has assessed the expression of E-cadherin broadly and systematically in myeloid blasts.

Our work demonstrates that in the bone marrow, E-cadherin is specific to the erythroid lineage and is only expressed in blasts in AML when there is erythroid differentiation. In addition, we demonstrate the usefulness of E-cadherin in combination with CD34 and CD117 in evaluating myeloblasts vs erythroblasts in cases of MDS, in which erythroid hyperplasia and left-shifted forms can complicate enumeration and evaluation of myeloid blast percentages.

In our series of myeloid neoplasms, all five cases of erythroleukemia were positive (>50% of cells staining) for E-cadherin. These results were consistent with those of Liu et al,7 who demonstrated that 13 of 14 cases of pure erythroleukemia were positive for E-cadherin. In addition, like Liu et al,7 we saw variability in E-cadherin staining among erythroblasts, presumably because of dysregulation of normal erythroid maturation. However, in a separate study, Acs and LiVolsi13 found that all of their cases of erythroleukemia were negative for surface membrane expression of E-cadherin, though most of their cases of erythroleukemia did in fact express E-cadherin, albeit internally, in a Golgi distribution pattern, but not strongly at the cell surface. The differences between the findings of Acs and LiVolsi13 and others as well as ours could be the result of sensitivity, differences in antibody preparations, or differences in fixation and/or decalcification.

Of the 12 leukemias with megakaryocytic differentiation, representing two acute megakaryoblastic leukemias as subtypes of AML, not otherwise specified, and 10 myeloid proliferations related to Down syndrome, one had expression of E-cadherin. This case, though negative for GPC on immunohistochemistry, also showed other evidence of significant erythroid differentiation as demonstrated by GPA reactivity on flow cytometry (Image 2). E-cadherin expression on this AML with megakaryocytic differentiation is not entirely surprising because it is well known that megakaryocytes and erythroid cells share a common pluripotent stem cell progenitor.14

We also found that E-cadherin is specific to erythroid precursors and primarily expressed in those cases that are also CD117 positive (Image 3), consistent with prior reports.15 Indeed we also found that in some cases of MDS in which CD34 is only partially expressed on myeloid blasts, CD117 and E-cadherin can be used in combination to obtain a more reasonable estimate of myeloid blast numbers (specifically CD117% – E-cad%).

As with CD34 percentages, CD117% – E-cad% approximated morphologic blast percentages but was not perfectly correlated. Overall, therefore, the usefulness of E-cadherin in conjunction with CD34 and CD117 in cases of MDS may not necessarily be in obtaining an absolute count for blast percentages. It is more useful for assisting in specific cases in which blast enumeration may be difficult morphologically, and where CD34 is only variably expressed on blasts. Finally, if used together, either the CD34 percentages or CD117% – E-cad% correctly stratified all MDS cases studied herein.

Our results demonstrate the usefulness of E-cadherin, not only as a marker specifically of erythroid differentiation, but also as a useful marker in conjunction with CD34 and CD117 for enumeration of myeloblasts vs erythroblasts.

This study was supported through funding from the Stanford University Department of Pathology.

References

1.
van Roy
F
Berx
G
.
The cell-cell adhesion molecule E-cadherin
.
Cell Mol Life Sci
 .
2008
;
65
:
3756
3788
.
2.
Chen
H
Paradies
NE
Fedor-Chaiken
M
et al
.
E-cadherin mediates adhesion and suppresses cell motility via distinct mechanisms
.
J Cell Sci
 .
1997
;
110
:
345
356
.
3.
Pecina-Slaus
N
.
Tumor suppressor gene E-cadherin and its role in normal and malignant cells
.
Cancer Cell Int
 .
2003
;
3
:
17
.
4.
Canel
M
Serrels
A
Frame
MC
et al
.
E-cadherin-integrin crosstalk in cancer invasion and metastasis
.
J Cell Sci
 .
2013
;
126
:
393
401
.
5.
Armeanu
S
Müller
CA
Klein
G
.
Involvement of E-cadherin in the development of erythroid cells
.
Hematology
 .
2000
;
5
:
307
316
.
6.
Tidswell
M
Pachynski
R
Wu
SW
et al
.
Structure-function analysis of the integrin beta 7 subunit: identification of domains involved in adhesion to MAdCAM-1
.
J Immunol
 .
1997
;
159
:
1497
1505
.
7.
Liu
W
Hasserjian
RP
Hu
Y
et al
.
Pure erythroid leukemia: a reassessment of the entity using the 2008 World Health Organization classification
.
Mod Pathol
 .
2011
;
24
:
375
383
.
8.
Swerdlow
SH
International Agency for Research on Cancer, World Health Organization
.
WHO Classification of Tumours of Haematopoietic and Lymphoid Tissues
 .
Lyon, France
:
IARC Press
;
2008
.
9.
Weinberg
OK
Seetharam
M
Ren
L
et al
.
Clinical characterization of acute myeloid leukemia with myelodysplasia-related changes as defined by the 2008 WHO classification system
.
Blood
 .
2009
;
113
:
1906
1908
.
10.
Ohgami
RS
Zhao
S
Ohgami
JK
et al
.
TdT+T-lymphoblastic populations are increased in Castleman disease, in Castleman disease in association with follicular dendritic cell tumors, and in angioimmunoblastic T-cell lymphoma
.
Am J Surg Pathol
 .
2012
;
36
:
1619
1628
.
11.
Ohgami
RS
Arber
DA
Zehnder
JL
et al
.
Indolent T-lymphoblastic proliferation (iT-LBP): a review of clinical and pathologic features and distinction from malignant T-lymphoblastic lymphoma
.
Adv Anat Pathol
 .
2013
;
20
:
137
140
.
12.
Ohgami
RS
Ohgami
JK
Pereira
IT
et al
.
Refining the diagnosis of T-cell large granular lymphocytic leukemia by combining distinct patterns of antigen expression with T-cell clonality studies
.
Leukemia
 .
2011
;
25
:
1439
1443
.
13.
Acs
G
LiVolsi
VA
.
Loss of membrane expression of E-cadherin in leukemic erythroblasts
.
Arch Pathol Lab Med
 .
2001
;
125
:
198
201
.
14.
Klimchenko
O
Mori
M
Distefano
A
et al
.
A common bipotent progenitor generates the erythroid and megakaryocyte lineages in embryonic stem cell-derived primitive hematopoiesis
.
Blood
 .
2009
;
114
:
1506
1517
.
15.
Buhring
HJ
Muller
T
Herbst
R
et al
.
The adhesion molecule E-cadherin and a surface antigen recognized by the antibody 9C4 are selectively expressed on erythroid cells of defined maturational stages
.
Leukemia
 .
1996
;
10
:
106
116
.