The myelodysplastic syndromes (MDS) constitute a heterogeneous group of clonal hematopoietic stem cell disorders. They are characterized by abnormal bone marrow (BM) differentiation, and possibly by apoptosis, they culminate in peripheral blood cytopenia(s), with a substantial risk of transformation into overt acute myeloblastic leukemia (AML). The diagnosis of MDS involves a multidisciplinary approach and depends mainly on morphological criteria, as proposed by the FAB group (1), and cytogenetics. The diagnosis is difficult to make, and is often made late, and it is subjective. Immunophenotyping (IP) is not widely accepted within the current diagnostic armamentarium for the diagnosis of MDS. IP is only used in the late acute leukemia phase of MDS. Compared with AML, much less is known regarding MDS immunophenotypic surface antigen patterns for diagnosis, staging, and prognosis (2).
CD44 antigen is a highly glycosylated transmembrane protein that displays multiple isoforms generated by alternative splicing (3). The so-called standard isoform CD44s is the most common antigen and it is expressed on most mononuclear BM precursors (4, 5), especially on CD34+ hematopoietic progenitor cells (6, 7). It is also strongly expressed on peripheral granulocytic blood cells as they exit the vascular space (8–10), on erythrocytes, and in a soluble form in plasma. An interesting feature of CD44 is its transient decrease in granulocytic cells in the stage preceding their adherence to the vascular endothelium. CD44 plays an important role in lymphomyelopoiesis, lymphocyte homing, and tumor metastasis. The addition of anti-CD44 monoclonal antibodies (mAbs) fully inhibits in vitro long-term hematopoiesis on preestablished stroma (11, 12). As an adhesion molecule, CD44 mediates the adhesion of progenitor cells to hyaluronic acid (13, 14) and mature peripheral granulocytes to the endothelium. The expression of CD44, and several of its variants, is increased on lymphocytes and monocytes in states of inflammation and malignancy (15). The increased concentration of soluble CD44 in plasma, mostly its 6v or 9v variants, has been linked to an unfavorable clinical course in patients with solid organ malignancies (16, 17), lymphoma, myeloma (18), and B-cell chronic lymphocytic leukemia (B-CLL). The increased expression of its 6v variant on blasts in AML has also been linked to poor prognosis (19) and its soluble form in plasma exacerbates MDS (20). Finally, ligation of CD44 antibodies has reversed blockage of differentiation in human AML (21).
In previous reports, the expression of membrane-bound CD44 in patients with MDS/AML was investigated solely on blast cells in the BM (19), except for a few reports on peripheral blood myelomonocytic populations (22). The significant role of CD44 in hematopoiesis prompted us to investigate its expression on the myeloid BM cells of early and late-stage MDS patients. Two triple-surface marker assays incorporating CD44/CD33/CD66 and CD33/CD34/HLA-DR were used, the former to evaluate the expression of CD44 at different maturational stages of the myeloid cell line and the latter to assess the number of blasts.
MATERIALS AND METHODS
Thirty-eight MDS patients and 13 controls with normal BM were studied using flow cytometry-assisted IP (Table 1). All the MDS BM samples included in this study were analyzed consecutively in our flow cytometry laboratory over a period of 2.5 years (August 1997 to December 1999). All MDS patients (FAB classification: 20 RA, 8 RAEB, 10 RAEB-T) and controls (patients with various malignant or other diseases whose routine workup included a BM aspirate) gave informed consent. RAEB-T patients had more than 20% blasts in a dysplastic BM. None of the patients had received chemotherapy during the previous 3 months, nor did they receive hematopoietic growth factors.
Table 1. Data on MDS Patients and Controls
Average age (range)
BM aspirates, anticoagulated with EDTA, were diluted in RPMI and processed within 24 h. The diagnosis of MDS was based on a combination of morphological evaluation of the BM and blood, clinical elements, cytogenetic analysis using G-banding direct karyotyping when available, and fetal hemoglobin quantification.
Surface Antigen Labeling
Simultaneous three-color staining was used in two combinations (CD33/CD34/HLA-DR and CD44/CD33/CD66) on all BM samples. CD45 was not added to the combinations in the current study, although it was added to a broader panel (data not shown). One aliquot of BM was stained directly with HLA-DR peridinin chlorophyll protein (PerCP; clone L243, Becton-Dickinson, San Jose, CA), CD33 fluorescein isothiocyanate (FITC; clone WM-54, Dako, Copenhagen, Denmark), and CD34 phycoerythrin (PE; clone 8G12, Becton-Dickinson) conjugated mAbs. This was followed by lysis of red blood cells. A second aliquot was first lysed, washed, and then stained with CD66 RPE-CY5 (Cy5 dye coupled to R-phycoerythrin; 23,24), anti CD66 abce (clone Kat4C, Dako), CD33 PE (clone P 67.6, Becton-Dickinson), and CD44 FITC (clone Leu 44, Becton-Dickinson) conjugated mAbs. Flow cytometry analysis was performed using an XL-MCL flow cytometer (Beckman-Coulter, Hialeah, FL).
DNA staining was performed on EDTA anticoagulated BM samples with DNA prep (Beckman-Coulter; 25). Apart from giving independent information about cell cycle phases of the BM and DNA aneuploidy, the DNA cell cycle parameters also served as a quality marker to ensure that BM samples were diluted with a minimal amount (“contamination”) of blood. BM samples containing more then 95% of cells in G1 phase were not included in the study. They were considered to be too diluted with peripheral blood (26), which made CD44 levels more difficult to interpret (Table 2).
Main gates, including total white blood cell (WBC) and granulocyte gates, were drawn for each sample (Fig. 1a) and forward scatter (FSC) was plotted against side scatter (SSC). The three-color CD33/CD34/HLA-DR combination was used for blast analysis. The percent of CD34+ cells was estimated using a CD34/SSC scattergram (Fig. 1b) gated on total WBC (Fig. 1a). HLA-DR positivity of granulocytes was assessed by analyzing CD33+ cells in the granulocyte gate for HLA-DR (mean fluorescence intensity [MFI] and percentage).
For delineating myeloid subpopulations using the triple-color CD44/CD33/CD66 mAb assay, we separated all WBCs on a CD66 versus SSC scattergram (23; Fig. 1c). Three cell populations were clearly defined: lymphocytes, which stain negative for CD66 (CD66-), with a low SSC; granulocytes, which express high values of CD66 (CD66+++), with a varying intensity of SSC; and monocytes and immature myeloid cells, both of which express intermediate amounts of CD66 (CD66+), with a low SSC. We investigated CD44 expression on the myeloid cell populations by determining both MFI and the percentage of cells expressing CD44 in the CD66+++ and CD66+ gates (Fig. 1d). The cutoff value for CD44 was chosen as the lower limit of peripheral blood CD44 myeloid cell positivity. We suggest that each laboratory establish its own normal range regarding antigen fluorescence intensity levels.
DNA cell cycle analysis of BM cells was performed using Multicycle software (Phoenix Flow Systems, San Diego, CA).
All sample list modes were interpreted with the built-in System II software for surface antigen labeling, and by Multicycle software for cell cycle analysis.
The nonparametric Kruskal-Wallis test (27) was used to compare each group's parameters.
CD34 in MDS BM
Decreased expression of CD34 was found in RA BM samples (0.7% ± 0.4%) whereas increased expression of CD34 was found in RAEB (5.7% ± 3.1%) and RAEB-T (8.7% ± 6.2%) BM samples. Control BM samples expressed CD34 at 0.9% ± 0.4%. (Reported values in the literature refer to levels up to 1.5% in normal BM samples .)
Association of HLA-DR Expression With Stage of MDS Disease
Gated granulocytes expressed an increasing percentage of HLA-DR+ cells with progression of disease. HLA-DR percentage was similar in RA patients (5.0% ± 3.3%) as in normals (4.2% ± 2.7%), but was elevated in RAEB (14.9% ± 12.9%) and in RAEB-T patients (23.1% ± 19.0%; P = 0.001). MFI results showed a similar pattern (P = 0.006; Table 3).
Table 3. HLA-DR Percentage on CD33+ Granulocytes
HLA-DR+(%) on CD33+Granulocytes
HLA MFI on CD33+Granulocytes
Normal (N = 13)
4.2 ± 2.7
0.2 ± 0.05
RA (N = 20)
5.0 ± 3.3
0.2 ± 0.1
RAEB (N = 8)
14.9 ± 12.9
0.5 ± 0.3
RAEB-T (N = 10)
23.1 ± 19.0
0.8 ± 0.8
Association of CD66+++ and CD66+ Expression With Stage of Disease
Two myeloid populations were defined by CD66 versus SSC: a bright CD66+++-expressing population, with intermediate staining for CD33, compatible with maturing and mature myeloid cells, and medium intensity CD66+-expressing cells, brightly positive for CD33, compatible with monocytes and immature myeloid cells. The percentage of CD66+++ cells was significantly decreased in patients with RAEB and RAEB-T (47.0% ± 25.0% and 30.0% ± 25.0% respectively), compared with RA and control BM (74.0% ± 12.0%). On the other hand, CD66+-expressing cells increased in patients with RAEB and RAEB-T (15.5% ± 15.3%, 22.4% ±19.8%, respectively) compared with RA patients and controls (4.8% ± 2.6%, 4.2% ± 1.1%, respectively; Table 4).
Table 4. Percentages of Positive Cells Stained for CD44 in the Gates CD66+++ and CD66+ in Patients With Different Stages of MDS and in Normal BM
Main total WBC CD66+++ (%)
Main total WBC CD66+ (%)
Normal (N = 13)
74 ± 12
64.5 ± 4.8
4.24 ± 1.1
92.6 ± 5.3
RA (N = 20)
74 ± 13
43.2 ± 9.2
4.82 ± 2.6
86.3 ± 11.3
RAEB (N = 8)
47 ± 25
57.5 ± 15
15.5 ± 15.3
90.5 ± 9
RAEB-T (N = 10)
30 ± 25
82.5 ± 13.3
22.4 ± 20
95 ± 5.7
Association of CD44 Expression on Myelomonocytic Cells With the Stage of Disease in MDS BM samples
CD66+++ and CD66+-bearing populations of controls and MDS patients in the early and late stages of their disease were analyzed for CD44 antigen expression and MFI (Fig. 2). CD44 expression in RA CD66+++ cell populations was significantly lower (43.2% ± 9.2%) compared with normal CD66+++ cells (64.5% ± 4.8%). However, its expression was near normal on RAEB CD66+++ cells (57.5% ± 15%) and was very high on RAEB-T cells (82.5% ± 13.3%; P < 0.001; Table 4, Fig. 2). CD44 MFI on CD66+++ cells revealed a similar pattern (not shown). CD44 expression on CD66+-bearing cells revealed a mild tendency to increase in patients with late-stage MDS (Table 4), but there were no major differences in its MFI between the groups (not shown).
Appearance of Multiple Immunophenotypical Myeloid Cell Populations
Of 14 late-stage MDS patients in whom CD66/SSC analyses were performed, two populations of CD66+++ cells were identified in five patients (two with RAEB, three with RAEB-T) with high-intensity CD66, in two separate clusters, each with different intensities of CD33 (Fig. 3). This pattern was not observed in any of the normal BM samples nor in RA patients. The immunophenotypically different myeloid cells could not be distinguished morphologically.
CD34 was elevated in RAEB, and even more so in RAEB-T BM samples. This finding, already described in the literature (26, 29), reflects a BM populated with blasts bearing CD34 in patients with late-stage MDS. Although of great help in diagnosing patients with late-stage MDS, it is not useful in the early detection of RA patients. Similarly, HLA-DR expression in RA patients was noncontributory, although levels increased as the disease progressed from RAEB to RAEB-T BM.
Our study shows that CD44 is expressed differentially on BM CD66+++ cells, with a reduced expression in patients with early-stage MDS. CD44 expression decreased in RA BM samples on gated CD66+++ BM cells compared with BM samples from controls and late-stage MDS patients (RAEB, RAEB-T). In patients with late-stage MDS, CD44 expression was very close to normal levels or above. However, in BM from late-stage MDS patients, CD34-expressing cells, HLA-DR–expressing CD33+ granulocytes, and CD66+-bearing myelomonocytic cells increased, making diagnosis easier. We show that low CD44 expression on CD66+++ BM myeloid cells can be used as an early diagnostic aid, along with BM morphology and cytogenetics in MDS patients.
According to Lund-Johansen and Terstappen (4), differential expression of CD44 exists in normal BM myeloid cell populations. They reported high expression on immature cells, a prominent decrease on cells of intermediate maturity, and an increase again on mature granulocytes. The observed differential expression of BM CD44 at different stages of MDS may result from a different composition of myeloid subpopulations. Decreased CD44 in RA patients may reflect accumulation of cells in the intermediate myeloid compartment (with low CD44 expression). As MDS progresses, a shift in myeloid cell composition occurs with an increase in the immature myeloid cell compartment, normally expressing high levels of CD44. This is seen with a concomitant decrease in CD66+++ expression. A similar explanation, although for different markers (CD13, CD66), was reported by Bonde et al. (23). Whether shifts in the distribution of granulocytes in other states (so-called maturation arrest) express a similar immunophenotype deserves further investigation.
Changes in CD44 expressions in MDS patients may reflect disordered expression of various cellular genes that evolve during disease progression.
Finally, McKenna and Cotter (30) suggested that there is an increased rate of apoptosis in patients with early-stage MDS, whereas escape from apoptosis was detected in patients with late-stage MDS and acute leukemia. CD44 is the main receptor for hyaluronic acid, a glycosaminoglycan widely distributed throughout the human body, and heparin is a related glycosaminoglycan that induces mature neutrophils to undergo apoptosis (31). Therefore, changes in CD44+ populations may be related to changes in programmed cell death mechanisms during the progression of MDS. Changes in surface antigen expression have been noted for other adhesion molecules (32).
Although patients with MDS harbor a stem cell malignancy from an early stage, diagnosis of MDS is often deferred because of a paucity of dysplastic features in BM and blood. Identification of dysmyelopoiesis in patients with the early stages of MDS by flow cytometry, especially when morphology is insufficient for diagnosis, may be of assistance when evaluating patients with mono/bi/pancytopenias. At Western Galilee Hospital, this assay has been included as part of the MDS evaluation protocol.
We thank Ms. Mala Weiss and Ms. Ariela Hazan for their excellent technical assistance. This study was done as part of the requirements for the M.D. degree (Y.K.), Bruce Rappaport Faculty of Medicine, Technion-Israel Institute of Technology, Haifa, Israel.