Clinical utility of multiparameter flow cytometry in the diagnosis of 1013 patients with suspected myelodysplastic syndrome

Correlation to cytomorphology, cytogenetics, and clinical data

Authors


Abstract

BACKGROUND:

The diagnosis and classification of myelodysplastic syndromes (MDS) is based on cytomorphology (CM) and cytogenetics (CG). Multiparameter flow cytometry (MFC) may add important diagnostic information.

METHODS:

To evaluate the potential role of MFC in the diagnostic setting of MDS, the authors analyzed the results from 1013 patients with suspected MDS by using CM, CG, and MFC in parallel.

RESULTS:

Concordance between CM and MFC was 82% for diagnostic results in 788 patients who had unequivocal CM results. An additional 225 patients had only minor dysplastic features identified by CM, including 51 patients (22.7%) who had clear evidence of MDS by MFC. Twelve patients who had no indication of MDS identified by CM had MDS-typical CG aberrations; in 6 of those patients (50%), MFC revealed MDS characteristics. In another 11 of 23 patients (47.8%) who had minor dysplastic features identified by CM and MDS-typical CG aberrations, MFC revealed MDS characteristics. The percentages of blasts determined by CM and by MFC were strongly correlated (P < .001). The frequency of aberrantly expressed antigens differed significantly between patients rated by CM as MDS (highest frequencies), suspected MDS, and no MDS (lowest frequencies). In various patients, MFC identified MDS-typical aberrant antigen expression in cell compartments that were not rated dysplastic by CM. The numbers of aberrantly expressed antigens were correlated with International Prognostic Scoring System scores and overall survival.

CONCLUSIONS:

The current analysis clearly demonstrated an increased diagnostic yield with MFC when added to CM and CG in patients with suspected MDS. Cancer 2010. © 2010 American Cancer Society.

Myelodysplastic syndromes (MDS) comprise a heterogeneous group of clonal diseases that originate in malignant hematopoietic stem cells.1 Clinical manifestations of MDS result from peripheral blood cytopenias and include fatigue, infections, and bleeding.1 The diagnosis of MDS relies on the evaluation of dysplastic findings in bone marrow (BM) cells. Acquired chromosomal aberrations are detected in 50% of patients.2-4

Diagnostic criteria and classification of MDS have been based on the French-American-British (FAB) classification for 2 decades5 with standardized criteria for dysplasia.6 Cytogenetic (CG) analysis has been included in the standard diagnostic workup and classification.7-9 In addition to cytomorphology (CM) and CG, the number of cytopenias and transfusion requirements have been included in the powerful International Prognostic Scoring System (IPSS)10 and World Health Organization (WHO) classification-based Prognostic Scoring System,11 respectively, for prognosis.

It has been demonstrated that both FAB and the WHO classification identify clinically relevant subgroups.12 In patients with chromosomal aberrations, CG may contribute substantially to establish a diagnosis of MDS if CM results are equivocal. However, in a large number of patients with cytopenias, the diagnosis may not be made, and repeat evaluations are recommended.

Because of these diagnostic difficulties, it is worthwhile to test the power of multiparameter flow cytometry (MFC). MFC repeatedly has demonstrated the ability to identify the aberrant expression of antigens that are considered dysplastic features.13-16 These findings have been described in MDS and do not occur or occur only infrequently in patients who have cytopenias that are not considered MDS. The objective of the current study was to estimate the potential diagnostic value of MFC.

MATERIALS AND METHODS

Patients

In total, 1013 BM samples from patients with suspected MDS were sent to MLL Munich Leukemia Laboratory between August 2005 and January 2009 and were analyzed in parallel by using CM, CG, and MFC. The suspicion of MDS was raised when the physician sent the sample and was based on peripheral blood count and differential as well as clinical criteria. Patients who had hematologic malignancies other than MDS were excluded after the respective diagnosis was confirmed. Thus, in the current cohort, all patients had in common an initial suspicion of MDS. All 3 diagnostic methods were performed independently from each other. CM was performed by T.H., CG was performed by C.H., and MFC was performed by W.K. The median patient age was 69.8 years (range, 1.8-88.7 years), and the ratio of men to women was 539:474. Median values and ranges for peripheral blood counts were as follows: median white blood cell count, 4400/μL (range, 600-117,600/μL); median hemoglobin, 10.7 g/dL (range, 4.0-20.2 g/dL); and median thrombocyte count, 157,500/μL, (range, 2000-1195,000/μL).

Cytomorphology and Cytogenetics

CM assessment was based on May-Grunwald-Giemsa stains, myeloperoxidase reactions, and nonspecific esterase using alpha-naphtyl-acetate17-19 and was used for diagnosis according to FAB20-22 and WHO7, 8 criteria. CG analyses were performed according to standard protocols. Classification was according to the International System for Human Cytogenetic Nomenclature.23-25 Complex aberrant karyotypes were defined by ≥3 clonal chromosome aberrations.8, 26 Fluorescence in situ hybridization was used according to standard procedures27 to clarify all difficult cases.

Multiparameter Flow Cytometry

Samples were processed by Ficoll gradient centrifugation followed by ammonium chloride-based erythrocyte lysis to isolate mononuclear cells. MFC was performed by applying 5-fold stainings and using the antibodies outlined in Table 1 (selected based on previous studies).13-15, 28, 29 Antibodies were purchased from Immunotech (Marseilles, France) except for carcinoembryonic antigen-related cell adhesion molecule 6/nonspecific cross-reacting antigen (cluster of differentiation [CD] 66c [CD66c]) (Becton Dickinson, Heidelberg, Germany). Antibody combinations were added to 106 mononuclear cells (volume, 100 μL) and incubated for 10 minutes. After the addition of 2 mL lysing solution, samples were incubated for additional 10 minutes, then washed twice in phosphate-buffered saline (PBS), and resuspended in 0.5 mL PBS. FC500 flow cytometers were used (Beckman Coulter, Miami, Fla), and 20,000 events were acquired. Cytomics CXP Software (Beckman Coulter) was used for data analysis.

Table 1. Panel of Monoclonal Antibodies Used for Immunophenotyping
CombinationFITCPEECDPC5PC7
  1. FITC indicates fluorescein isothiocyanate; PE, phycoerythrin; ECD, phycoerythrin-Texas Red; PC5, phycoerythrin-cyanine 5; PC7, phycoerythrin-cyanine 7; MoAb, monoclonal antibody; CD, cluster of differentiation; CD11b, integrin α M; CD13, alanine aminopeptidase; HLA-DR, human leukocyte antigen-D related; CD16, human neutrophil antigen 1; CD45, protein tyrosine phosphatase, receptor type, C; CD71, transferrin receptor 1; CD235a, glycophorin A; CD19, protein encoded by the CD19 B-lymphocyte antigen gene; CD2, lymphocyte function-associated antigen 2; CD61, protein encoded by the integrin beta 3 gene; CD5, type I transmembrane protein; CD14, monocyte differentiation antigen CD14; CD4, protein encoded by the T-cell surface glycoprotein CD4 gene; CD15, 3-fucosyl-N-acetyl-lactosamine; CD66c, carcinoembryonic antigen-related cell adhesion molecule 6 (nonspecific cross-reacting antigen); CD3, T-cell coreceptor; CD64, integral membrane glycoprotein; CD65, fucosylated carbohydrate cell-surface antigen; CD56, neural cell adhesion molecule; CD10, common acute lymphocytic leukemia antigen; CD33, 67-kDa membrane glycoprotein; CD36, collagen type 1 receptor; CD38, cyclic adenine diphosphate hydrolase; CD34, cell surface antigen, glycoprotein 105-120; CD7, glycoprotein 40.

1CD11bCD13HLA-DRCD16CD45
 MoAbBear1Immu103.44Immu3573G8J33
2CD71CD235aCD19CD2CD45
 MoAbYDJ1.2.211E4B-7-6J3-11939C1.5J33
3CD61CD5CD14CD4CD45
 MoAbSZ21BL1ARMO5213B8.2J33
4CD15CD66cCD3CD64CD45
 MoAb80H5B6.2UCHT122J33
5CD65CD56CD10CD33CD45
 MoAb88H7N901ALB1D3HL60.251J33
6CD36CD38CD34CD7CD45
 MoAbFA6-152T165818H8.1J33

Gating Strategy

Protein tyrosine phosphatase, receptor type, C (CD45)-side scatter (SSC) gating was performed to identify cellular compartments to allow the separate evaluation of antigen expression for each compartment.30 The lack of expression of an antigen is indicated by “−,” dim expression is indicated by “(+),” any expression is indicated by “+,” and strong expression is indicated by “++.” A broader range of expression is indicated by 2 respective values separated by a virgule, ie, expression ranging from “(+)” to “+” is indicated by “(+)/+.” Granulocytes were identified as SSC+/(+)CD45(+), monocytes were identified as SSC(+)CD45+, myeloid blasts were identified as SSC([+])CD45(+), and erythroid cells were identified as CD45− glycophorin A (CD235a)++. It should be recognized that, because a DNA-binding dye was not used, there is no guarantee that the erythroid cells identified are truly nucleated. By applying back-gating, monocytes also were identified as monocyte differentiation antigen CD14 (CD14)+, and blasts were separated from the protein encoded by the CD19 B-lymphocyte antigen gene (CD19)-positive B-lymphoid progenitors, and in addition, were demonstrated to express cell surface antigen-glycoprotein 105-120 (CD34), or human leukocyte antigen-D related (HLA-DR), or both. The percentage of blasts was calculated by using the total amount of mononuclear cells after sample preparation as the denominator. According to previous reports,13-15, 28, 29 the following features were evaluated.

Granulocytes were evaluated for SSC signal, abnormal alanine aminopeptidase (CD13)/human neutrophil antigen 1 (CD16) and integrin α M (CD11b)/CD16 expression patterns, neural cell adhesion molecule (CD56) coexpression, 67-kDa membrane glycoprotein (CD33) negativity, and integral membrane glycoprotein (CD64) negativity. The CD13/CD16 expression pattern was considered abnormal if a deviation of normal granulocytic maturation (CD13+/CD16−→CD13[+]/CD16−→CD13−/CD16[+]→CD13−/CD16+→CD13+/CD16++) was observed (Fig. 1). Also, the CD11b-/CD16 expression pattern was considered abnormal if a deviation of normal granulocytic maturation (CD11b−/CD16−→CD11b+/CD16[+]→CD11b+/CD16+→CD11b++/CD16++) was observed. Deviation from normal was defined as a difference that amounted to at least a half-log signal intensity in 1 parameter, although it must be recognized that the assessment is always at least partially subjective.

Figure 1.

Cluster of differentiation 13 (alanine aminopeptidase) (CD13)/human neutrophil antigen (CD16) expression patterns are shown (Left) in granulocytes from normal bone marrow (BM) and (Right) in myelodysplastic syndrome (MDS) PC5 indicates phycoerythrin-cyanine 5; PE, phycoerythrin.

Monocytes were evaluated for CD11b negativity, HLA-DR negativity, CD13 negativity, CD16 coexpression, CD56 coexpression, aberrant lymphocyte function-associated antigen 2 (CD2) coexpression.

Myeloid blasts

Myeloid blasts were evaluated for the percentage of BM myeloid blasts, coexpression of CD11b, type I transmembrane protein (CD5), CD56, transferrin receptor 1(CD7), CD15, and CD64 and HLA-DR negativity.

Erythroid cells

Erythoroid cells were evaluated for homogeneously strong transferrin receptor 1 (CD71) expression and CD71 negativity. Examples of aberrant antigen expression are shown in Figures 1, 2, and 3.

Figure 2.

(Left) Monocytes in normal bone marrow (BM) revealed no expression of cluster of differentiation 56 (CD56) (neural cell adhesion molecule). (Right) In myelodysplastic syndrome (MDS), strong CD56 expression was observed. CD33 indicates 67-kDa membrane glycoprotein; PC5, phycoerythrin-cyanine 5; PE, phycoerythrin.

Figure 3.

(Left) In normal bone marrow (BM), the majority of erythrocytes had strong expression of cluster of differentiation 71 (CD71) (transferrin receptor 1) with some degree of heterogeneity for dim expression. (Right) In myelodysplastic syndrome (MDS), negative CD71 expression may be present in the majority of erythrocytes. CD235a indicates glycophorin A; PE, phycoerythrin; FITC, fluorescein isothiocyanate.

Positivity and negativity, respectively, of the respective antigens were determined by comparison with the respective isotype controls using 20% as the cutoff. The SSC signal in granulocytes was assessed qualitatively as either reduced or not reduced based on the CD45/SSC dot plot. For a marker with which to describe objectively the reduced SSC signal in granulocytes, the granulocyte-to-lymphocyte ratio of the mean SSC signal as acquired on a linear scale was calculated for each case (the SSC granulocyte:lymphocyte ratio).

Statistical Analysis

Dichotomous variables were compared using chi-square tests, and continuous variables were analyzed with the Student t test. Spearman rank correlation was used to analyze correlations between continuous parameters. Survival was analyzed using the Kaplan-Meier method,31 and differences were analyzed using the log-rank test. All calculations were performed using SPSS software (version 14.0.1; SPSS Inc., Chicago, Ill). All reported P values are 2-sided.

Study Conduct

Patients provided informed consent to participate in the current diagnostic procedures and the evaluation at MLL Munich Leukemia Laboratory after they were advised about the purpose and investigational nature of the study and of the potential risks. The study design adhered to the Declaration of Helsinki.

RESULTS

Cytomorphologic and Cytogenetic Characterization

Classification by CM is detailed in Table 2. The highest percentages of results that were in agreement with MDS by MFC were observed in the MDS subgroups with blast counts >10%, as anticipated. The lowest percentages of results that were in agreement with MDS by MFC were observed in the MDS subgroup of patients who had refractory cytopenia with multilineage dysplasia (RCMD); however, the percentages still clearly were higher than the respective percentages observed in patients with suspected MDS according to CM results and in patients with no MDS. CG results are detailed in Table 3. Fluorescence in situ hybridization was performed in 443 cases.27

Table 2. Diagnostic Results of Multiparameter Flow Cytometry in Cytomorphologically Defined Myelodysplastic Syndrome Subgroups
CM ResultTotal No. of PatientsNo. of Patients With Results in Agreement With MDS by MFC Independent of CM Results (%)
  1. CM indicates cytomorphology; MDS, myelodysplastic syndrome; MFC, multiparameter flow cytometry; RA, refractory anemia; RARS, refractory anemia with ring sideroblasts; RCMD, refractory cytopenia with multilineage dysplasia; RCMD-RS, refractory cytopenia with multilineage dysplasia and ring sideroblasts; RAEB-1, refractory anemia with 5% to 9% excess blasts; RAEB-2, refractory anemia with 10% to 19% excess blasts; −, negative; CMML, chronic myelomonocytic leukemia; MDS-u; unclassifiable myelodysplastic syndrome; AML, acute myeloid leukemia; MPS, myeloproliferative syndrome.

RA3122 (71)
RARS2716 (59.3)
RCMD6423 (35.9)
RCMD-RS4932 (65.3)
RAEB-1133104 (78.2)
RAEB-28178 (96.3)
5q– Syndrome2418 (75)
CMML6562 (95.4)
MDS-u1512 (80)
MDS/AML66 (100)
MDS/MPS169 (56.3)
Suspected MDS22551 (22.7)
Reactive condition26613 (4.9)
Normal findings110 (0)
Table 3. Diagnostic Results of Multiparameter Flow Cytometry in Cytogenetically Defined Myelodysplastic Syndrome Subgroups
Cytogenetic ResultTotal No. of PatientsNo. of Patients With Results in Agreement With MDS by MFC Independent of CM Result (%)
  1. MFC indicates multiparameter flow cytometry; MDS, myelodysplastic syndrome; CM, cytomorphology; Del, deletion.

Normal karyotype768257 (33.5)
Del(5q)4333 (76.7)
Aberrations of chromosome 71414 (100)
Trisomy 83025 (83.3)
Del(20q)2118 (85.7)
Complex karyotype2319 (82.6)
Loss of Y-chromosome4322 (51.2)
Other aberrations7158 (81.7)

Comparison of Diagnostic Results Obtained by Cytomorphology and Multiparameter Flow Cytometry

There was 82% percent concordance between CM and MFC for diagnostic results (646 of 788) in patients with definite CM results. In detail, 511 patients were classified with MDS according to CM, including 382 patients (74.8%) who had antigen expression features according to MFC in agreement with MDS. Of 277 patients who had CM results indicating no MDS, only 13 patients (4.7%) had MDS-typical features according to MFC. The pattern of these MDS-typical features did not differ from the overall pattern observed in the total cohort. Two hundred twenty-five patients had some dysplastic features on CM but did not fulfill the diagnostic criteria for MDS; however, 51 of those patients (22.7%) had MDS-typical findings on MFC. Clinical follow-up of these patients is not available to an extent that would allow reasonable clinical validation.

Comparison of Cytogenetics and Diagnostic Results Obtained by Multiparameter Flow Cytometry

Two hundred forty-five patients had an aberrant karyotype, and 189 (77.1%) of those patients had MDS according to the MFC results (Table 3).

Comparison of Diagnostic Results Obtained by Multiparameter Flow Cytometry, Cytomorphology, and Cytogenetics

The numbers of patients who had results that were rated in agreement and not in agreement with MDS on MFC are listed separately according to the CM results (no MDS, MDS, or suspected MDS), and each category is broken down according to aberrant CG results and normal CG results in Table 4. Overall, 1.3% of patients had MFC results that were in agreement with MDS in the absence of evidence for MDS on CM, and 50% of those patients had an aberrant karyotype. In addition, 5.1% of all patients had MFC results that were in agreement with MDS, whereas the CM results indicated suspected MDS, and slightly less than 33% of those patients had an aberrant karyotype. Thus, 6.4% of all patients had MFC results that were in agreement with MDS without a clear diagnosis of MDS by CM, and 33% of those patients had an aberrant karyotype.

Table 4. Diagnostic Results of Multiparameter Flow Cytometry in Cytomorphologically Defined Myelodysplastic Syndrome Subgroups Separated According to the Presence of Cytogenetic Abnormalities
CM ResultMFC Result in Agreement With MDS Independent of CM Result: No. of Patients (%)
NoYes
  1. MFC indicates multiparameter flow cytometry; MDS, myelodysplastic syndrome; CM, cytomorphology.

Aberrant cytogenetics  
 No MDS17 (1.7)6 (0.6)
 MDS22 (2.2)168 (16.6)
 Suspected MDS17 (1.7)15 (1.5)
Normal cytogenetics  
 No MDS247 (24.4)7 (0.7)
 MDS107 (10.6)214 (21.1)
 Suspected MDS157 (15.5)36 (3.6)

Multiparameter Flow Cytometry in Cytogenetically Aberrant Cases Not Classified as Myelodysplastic Syndrome by Cytomorphology

Loss of the Y chromosome was not considered MDS-related, because it occurs in healthy, older individuals. Thus, although these patients are listed in Table 5, they are not included in the totals. In 12 patients, CM gave no indication of MDS, but MDS-typical CG aberrations were present (Table 5). The MFC results from 6 of 12 patients (50%) were in agreement with MDS. In another 23 patients, CM identified dysplastic features that were not sufficient to diagnose MDS, but CG revealed aberrations. MFC results from 11 of 23 patients (47.8%) were in agreement with MDS. It is noteworthy that patients who had loss of the Y chromosome had MFC results that were in agreement with MDS only if CM identified dysplastic features that were not sufficient to diagnose MDS, whereas all patients who had loss of the Y chromosome and no indication of MDS on CM had MFC results that were not in agreement with MDS (Table 5).

Table 5. Diagnostic Results of Multiparameter Flow Cytometry in Patients With Cytogenetically Aberrant Results Not Classified as Myelodysplastic Syndrome by Cytomorphology
Cytogenetic ResultSuspected MDS by CMNo MDS by CM
Total No. of PatientsNo. of Patients With Results in Agreement With MDS by MFCTotal No. of PatientsNo. of Patients With Results in Agreement With MDS by MFC
  1. MDS indicates myelodysplastic syndrome; CM, cytomorphology; MFC, multiparameter flow cytometry; Del, deletion.

Del(5q)5310
Aberrations of chromosome 71133
Trisomy 83131
Del(20q)4310
Complex karyotype1011
Loss of Y chromosome54100
Other aberrations9331

Comparison of Myeloid Blasts Counts by Cytomorphology and Multiparameter Flow Cytometry

The mean ± standard deviation percentages of myeloid blasts determined by CM versus MFC were 4.67% ± 4.18% versus 3.78% ± 2.97% (Spearman rho [r], 0.362; P < .001) (Fig. 4). A higher blast count on CM in some patients resulted from the presence of monocytoid cells in which monocytic characteristics were featured that had to be included in the blast counts according to WHO criteria but that did not express progenitor markers according to the MFC results. However, these cells clearly were myeloid/monoblastic or promonocytic blasts that were detected by CM and cytochemistry (for an example, see Fig. 5). Accordingly, in patients who had aberrant CD56 expression on monocytes, a higher blast count (as determined by CM) was observed compared with patients who did not have aberrant CD56 expression (6.1% ± 4.6% vs 4.1% ± 3.9%; P < .001), whereas the respective differences in blast counts determined by MFC clearly were smaller (4.1% ± 3.2% vs 3.6% ± 2.9%; P = .014).

Figure 4.

The correlation of blast counts obtained by cytomorphology (CM) and by multiparameter flow cytometry (MFC) is depicted. The height of each column indicates the numbers of patients. The mean ± standard deviation percentages of myeloid blasts determined by CM versus MFC were 4.67% ± 4.18% versus 3.78% ± 2.97%, respectively (Spearman rank correlation, 0.362; P < .001).

Figure 5.

A bone marrow sample from a patient with myelodysplastic syndrome is shown in which multiparameter flow cytometry yielded (A) a myeloid blast count of 6% (red) and (A,B) 14% monocytic cells (purple), whereas (C) cytomorphology yielded a myeloid blast count of 18%. SSC indicates side scatter; CD, cluster of differentiation; CD45, protein tyrosine phosphatase, receptor type, C; PC7, proprotein convertase 7; CD33, 67-kDa membrane glycoprotein; PC5, phycoerythrin-cyanine 5; CD56, neural cell adhesion molecule; PE, phycoerythrin.

Characteristics in Patients With Myelodysplastic Syndromes Defined by Multiparameter Flow Cytometry

Aberrant antigen expression in myeloid blasts

Aberrant antigen expression in myeloid blasts was most frequent in patients who had MDS compared with patients who did not have MDS or who had suspected MDS according to the CM results (Table 6). In patients who had CG aberrations, an aberrant expression of the following antigens was observed more frequently compared with their expression in patients who had normal CG results: CD11b (7.3% vs 3.3%; P = .010), CD56 (4.1% vs 1.4%; P = .018), and CD7 (3.7% vs 1.4%; P = .036).

Table 6. Correlation of Aberrant Antigen Expression in Myeloid Blasts with Blast Count and Dysplastic Features in Cytomorphologya
MFC FindingsCytomorphologic Findings: No. of Patients (%)P
No MDS, n=277MDS, n=511Suspected MDS, n=225
  • MFC indicates multiparameter flow cytometry; MDS, myelodysplastic syndrome; CD, cluster of differentiation; CD11b, integrin α M; +, positive; HLA-DR, human leukocyte antigen-D related; −, negative; NS, nonsignificant; CD5, type I transmembrane protein; CD56, neural cell adhesion molecule; CD7, glycoprotein 40; CD15, 3-fucosyl-N-acetyl-lactosamine; CD64, integral membrane glycoprotein; SD, standard deviation.

  • a

    Numbers of patients were compared using chi-square tests, and mean values were determined using Student t tests.

  • b

    Comparison of columns 1 and 2.

  • c

    Comparison of columns 1 and 3.

CD11b+3 (1.1)27 (5.3)13 (5.8).009
HLA-DR−0 (0)4 (0.8)0 (0)NS
CD5+0 (0)9 (1.8)0 (0).012
CD56+0 (0)17 (3.3)4 (1.8).007
CD7+1 (0.4)18 (3.5)1 (0.4).002
CD15+1 (0.4)10 (2)6 (2.7)NS
CD64+1 (0.4)12 (2.3)6 (2.7)NS
No. of aberrant antigens per patient, mean±SD0.0±0.2bc0.2±0.6b0.1±0.6c.001b
.003c
% Blasts, mean±SD2.8±1.6bc4.6±3.7b3.1±1.7c.001b
    .041c

Considering CG subgroups, a blast count >5% was observed more often (compared with the remaining patients) in those who had aberrations of chromosome 7 and in those who had a complex aberrant karyotype, respectively (7 of 14 patients [50%] vs 167 of 999 patients [16.7%]; P = .005 and 14 of 23 patients [60.9%] vs 160 of 990 patients [16.1%]; P < .001). Accordingly, the mean (±standard deviation) blast counts were higher in the former patients (5.8 ± 3.3 vs 3.8 ± 3.0; P = .039 and 8.3 ± 5.5 vs 3.7 ± 2.8; P < .001). Patients with 20q deletion (del[20q]) or del(5q), respectively, as the sole aberration had CD11b expression more often than others (4 of 21 patients [19%] vs 39 of 992 patients [3.9%], respectively; P = .010; and 5 of 43 patients [11.6%] vs 38 of 970 patients [3.9%], respectively; P = .010).

Aberrant antigen expression in granulocytes

In granulocytes, aberrant antigen expression was observed most frequently in patients who had MDS confirmed by CM compared with patients who had no MDS or who had suspected MDS according to the CM results (Table 7). Aberrant antigen expression was not correlated strongly with dysgranulopoiesis by CM. Thus, in 406 patients without dysgranulopoiesis according to CM results, dysplastic features were observed with the following frequencies: aberrant CD13/CD16, 104 patients (25.6%); aberrant CD11b/CD16, 62 patients (15.3%); CD56 expression, 38 patients (9.4%); CD33 negativity, 44 patients (10.8%); and CD64 negativity, 2 patients (0.5%). Aberrant expression of ≥2 antigens in granulocytes was observed in 16 of 31 patients (51.6%) and in 15 of 27 patients (55.6%) with refractory anemia (RA) and RA with ring sideroblasts (RARS), respectively.

Table 7. Correlation of Aberrant Antigen Expression in Granulocytes With Cytomorphologya
MFC FindingsCytomorphologic Findings: No. of Patients (%)P
No MDS, n=277MDS, n=511Suspected MDS, n=225
  • MFC indicates multiparameter flow cytometry; MDS, myelodysplastic syndrome; CD, cluster of differentiation; CD13, alanine aminopeptidase; CD16, human neutrophil antigen 1; CD11b, integrin α M; CD56, neural cell adhesion molecule; +, positive; CD33, transmembrane receptor; −, negative; CD64, integral membrane glycoprotein; NS, nonsignificant; SD, standard deviation; SCC, side scatter.

  • a

    Numbers of patients were compared using chi-square tests, and mean values were determined using Student t tests.

Abnormal CD13/CD1625 (9)219 (42.9)54 (24)<.001
Abnormal CD11b/CD169 (3.2)143 (28)25 (11.1)<.001
CD56+10 (3.6)90 (17.6)23 (10.2)<.001
CD33−18 (6.5)53 (10.4)19 (8.4)NS
CD64−0 (0)14 (2.7)8 (3.6).011
No. of aberrant antigens per patient, mean±SD0.0±0.2   
 1 0.2±0.6 <.001
 2  0.1±0.6.003
Reduced SSC signal14 (5.1)286 (56)42 (18.7)<.001
SSC ratio of granulocytes to lymphocytes, mean±SD7.47±1.09   
 1 6.55±2.32 <.001
 2  7.38±1.17NS

The SSC granulocyte:lymphocyte ratio was lower in patients with MDS than in patients with no MDS (6.55 ± 2.32 vs 7.47 ± 1.09, respectively; P < .001). Furthermore, no respective significant differences were observed between patients with or without cytomorphologically identified dysgranulopoiesis (6.99 ± 2.20 vs 6.98 ± 1.27, respectively; P value not significant.).

In patients who had CG aberrations, aberrant expression of the following antigens was observed more frequently compared with patients who had a normal karyotype: CD13/CD16 (42.9% vs 25.1%, respectively; P < .001), CD11b/CD16 (24.1% vs 15.4%, respectively; P = .003), CD56 (20.8% vs 9.4%, respectively; P < .001), CD33 negativity (12.7% vs 7.7%, respectively; P = .020), and CD64 negativity (4.9% vs 1.3%, respectively; P = .002). Patients who had aberrant CG results had a lower SSC granulocyte:lymphocyte ratio (6.33 ± 1.26 vs 7.20 ± 2.00, respectively; P < .001).

In patients who had aberrations of chromosome 7 only and in patients who had complex aberrant karyotypes, distinct aberrant antigen expression was observed more frequently (for this comparison and for the comparisons described below, the remaining patients were used as comparison groups): CD13/CD16, 9 of 14 patients (64.3%) versus 289 of 999 patients (28.9%), respectively (P = .007) and 12 of 23 patients (52.2%) versus 286 of 990 patients (28.9%), respectively, (P = .020).

Patients who had 5q− alone also had other specific markers: CD56 was present in 10 of 43 patients (23.3%) versus 113 of 970 patients without 5q− (11.6%; P = .031). Among those with trisomy 8, aberrant CD13/CD16 expression was observed in 16 of 30 patients (53.3%) versus 282 of 983 patients without trisomy 8 (28.7%; P = .007); aberrant CD11b/CD16 expression was observed in 13 of 30 patients (43.3%) versus 164 of 983 patients (16.7%), respectively (P = .001); and CD33 negativity was observed in 9 of 30 patients (30%) versus 81 of 983 patients (8.2%), respectively (P = .001). Furthermore, the following CG subgroups had a lower SSC granulocyte:lymphocyte ratios (trisomy 8, 6.00 ± 1.08 vs 7.02 ± 1.89 without trisomy 8 [P < .001]; 20q− alone, 6.25 ± 1.13 vs 7.00 ± 1.89 without 20q− alone [P = .007]; 5q− alone, 6.03 ± 1.20 vs 7.03 ± 1.90 without 5q− alone [P < .001]; and a complex aberrant karyotype, 5.83 ± 1.28 vs 7.01 ± 1.89 without a complex aberrant karyotype [P < .001]).

Aberrant antigen expression in monocytes

In patients who had both MDS and suspected MDS according to CM results, an aberrant antigen expression in monocytes was observed more frequently compared with patients who had no MDS (Table 8). CD56 coexpression was observed most often in patients with MDS (43.1%); however, it also was encountered in patients without CM evidence of MDS (9.4%). Similar percentages were observed for CD16 expression (9% vs 3.6%, respectively), whereas both CD2 coexpression (8.8% vs 0.4%, respectively) and CD13 negativity (8.6% vs 2.2%, respectively) clearly occurred more often in patients with MDS.

Table 8. Correlation of Aberrant Antigen Expression in Monocytes With Cytomorphologya
MFC FindingsCytomorphologic Findings: No. of Patients (%)P
No MDS, n=277MDS, n=511Suspected MDS, n=225
  • MFC indicates multiparameter flow cytometry; MDS, myelodysplastic syndrome; CD, cluster of differentiation; CD11b, integrin α M; −, negative; NS, nonsignificant; HLA-DR, human leukocyte antigen-D related; CD13, alanine aminopeptidase; CD16, human neutrophil antigen 1; +, positive; CD56, neural cell adhesion molecule; CD2, lymphocyte function-associated antigen 2; SD, standard deviation.

  • a

    Numbers of patients were compared with chi-square tests, and mean values were determined with Student t tests.

CD11b−1 (0.4)7 (1.4)3 (1.3)NS
HLA-DR−3 (1.1)44 (8.6)6 (2.7)<.001
CD13−6 (2.2)44 (8.6)13 (5.8).002
CD16+10 (3.6)46 (9)15 (6.7).018
CD56+26 (9.4)220 (43.1)47 (20.9)<.001
CD2+1 (0.4)45 (8.8)6 (2.7)<.001
No. of aberrant antigens per patient, mean±SD0.2±0.5  <.001
 1 0.8±0.9  
 2  0.4±0.7 

The following analyses were made using the respective remaining patients in each subgroup for comparisons. Aberrantly expressed antigens in monocytes were observed in patients with chronic myelomonocytic leukemia (CMML) more frequently (vs patients without CMML), including CD13 negativity in 9 of 65 patients (13.8%) versus 54 of 948 patients (5.7%; P = .015), CD56 expression in 53 of 65 patients (81.5%) versus 240 of 948 patients (25.3%; P < .001), and CD2 expression in 14 of 65 patients (21.5%) versus 38 of 948 patients (4%; P < .001).

Fifteen of 133 patients (11.3%) who had RA with excess blasts (RAEB) between 5% and 9% (RAEB-1) were negative for HLA-DR compared with 48 of 880 patients without RAEB-1 (5.5%; P = .019); CD13 negativity was observed in 15 of 133 patients (11.3%) versus 38 of 880 patients (4.3%), respectively (P = .020); CD16 expression was observed in 16 of 133 patients (12%) versus 55 of 880 patients (6.3%), respectively (P = .027); and CD56 expression was observed in 51 of 133 patients (38.3%) versus 242 of 880 patients (27.5%), respectively (P = .014).

Among the patients who had RAEB with 10% to 19% excess blasts (RAEB-2) (vs patients without RAEB-2), CD13 negativity was observed in 11 of 81 patients (13.6%) versus 52 of 932 patients (5.6%) (P = .013); CD56 expression was observed in 41 of 81 patients (50.6%) versus 252 of 932 (27%; P < .001); and CD2 expression was observed in 12 of 81 patients (14.8%) versus 40 of 932 patients (4.3%; P < .001).

In patients with RA (vs patients without RA), CD56 expression in monocytes was observed more often (17 of 31 patients [54.8%] vs 276 of 928 patients [28.1%]; P = .002). In both patients with RA and patients with RARS, at least 2 aberrantly expressed antigens in monocytes were observed in 6 of 31 patients (19.4%) and in 2 of 27 patients (7.4%), respectively.

In patients with CG aberrations (vs patients without CG aberrations), CD56 expression was observed more frequently (95 of 245 patients [38.8%] vs 198 of 768 patients [25.8%]; P < .001). In patients who had MDS with a complex aberrant karyotype (vs patients who had MDS with normal CG results), HLA-DR negativity was observed in 5 of 23 patients (21.7%) versus 48 of 909 patients (4.8%), respectively (P = .005). In patients who had MDS with trisomy 8 (vs patients who had MDS without trisomy 8), CD13 negativity was observed more frequently (7 of 30 patients [23.3%] vs 56 of 983 patients [5.7%]; P = .002), and CD56 expression was observed more frequently (19 of 30 patients [63.3%] vs 274 of 983 patients [27.9%]; P < .001). Finally, in patients who had MDS with aberrations of chromosome 7 (vs patients who had MDS without aberrations of chromosome 7), CD56 expression was observed in 10 of 14 patients (71.4%) versus 283 of 999 patients (28.3%; P = .001).

Aberrant antigen expression in erythrocytes

Aberrant CD71 expression was observed significantly more often in both in patients who had MDS and in patients who had suspected MDS (judged by CM) compared with patients who did not have MDS (Table 9). In patients who had RA, RARS, and RCMD/ringed sideroblasts, compared with the respective remaining patients who did not have those findings, homogeneously strong expression of CD71 was observed in 5 of 31 patients (16.1%) versus 62 of 982 patients (6.3%; P = .048), in 6 of 27 patients (22.2%) versus 61 of 986 patients (6.2%; P = .007), and in 8 of 49 patients (16.3%) versus 59 of 964 patients (6.1%; (P = .012), respectively. In patients with RAEB-1 and RAEB-2, compared with the respective remaining patients without those findings, CD71 negativity was observed in 21 of 133 patients (15.8%) versus 73 of 880 patients (8.3%; P = .009) and in 14 of 81 patients (17.3%) versus 80 of 932 patients (8.6%; (P = .015), respectively.

Table 9. Correlation of Aberrant Antigen Expression in Erythrocytes With Cytomorphologya
MFC FindingsCytomorphologic Findings: No. of Patients (%)P
No MDS, n=277MDS, n=511Suspected MDS, n=225
  • MFC indicates multiparameter flow cytometry; MDS, myelodysplastic syndrome; CD, cluster of differentiation; CD71, transferrin receptor 1; SD, standard deviation.

  • a

    Numbers of patients were compared using chi-square tests, and mean values were determined using Student t tests.

CD71 homogeneously strong3 (1.1)46 (9)18 (8)<.001
CD71 negative6 (2.2)71 (13.9)17 (7.6)<.001
No. of aberrant antigens per patient, mean±SD0.0±0.2   
 1 0.3±0.4 <.001
 2  0.2±0.4<.001

In the presence of CG abnormalities (vs normal CG findings), greater frequencies of homogeneously strong CD71 expression (27 of 245 patients [11%] vs 40 of 768 patients [5.2%]; P = .003) and of CD71 negativity (32 of 245 patients [13.1%] vs 62 of 768 patients [8.1%]; P = .023) were observed. In particular, in patients who had 5q− as a sole CG abnormality, CD71 negativity was present in 9 of 43 patients (20.9%) versus 85 of 970 patients without 5q− as a sole abnormality (8.8%; P = .014).

Correlation of the Total Numbers of Aberrantly Expressed Antigens With Cytomorphology

The median total numbers of aberrantly expressed antigens in blasts, granulocytes, monocytes, and erythrocytes were 3 (range, 0-11), 1 (range, 0-8), and 0 (range, 0-7) in patients from each group with CM results of MDS, suspected MDS, and no MDS, respectively (P < .001) (Fig. 6). Using the CM result as the diagnostic gold standard, different total numbers of aberrantly expressed antigens were analyzed with regard to the sensitivity and specificity of MFC in reproducing CM results. Table 10 demonstrates that, although specificity remains at similar levels when applying higher numbers of aberrantly expressed antigens, a gradual loss of sensitivity is observed.

Figure 6.

The total numbers of aberrant antigen expression in blasts, granulocytes, monocytes, and erythrocytes are illustrated for patients who were diagnosed by cytomorphology with (A) no myelodysplastic syndrome, (B) myelodysplastic syndrome, or (C) suspected myelodysplastic syndrome. The height of each column indicates the numbers of patients.

Table 10. Sensitivity and Specificity of Multiparameter Flow Cytometry in Reproducing Cytomorphologic Diagnostic Results
MFC Criteria: No. of Aberrantly Expressed AntigensAberrantly Expressed Antigens Only, %Aberrantly Expressed Antigens or Blast Count > 5%, %Aberrantly Expressed Antigens or Blast Count > 5% or SSC Ratio (Granulocytes: Lymphocytes) > 6.3, %
SensitivitySpecificitySensitivitySpecificitySensitivitySpecificity
  1. SSC indicates side scatter; MFC, multiparameter flow cytometry.

≥275.592.679.691.385.386.3
≥357.593.665.29277.386.6
≥432.995.546.892.666.785.9
≥516.994.53591.361.685.4

A cutoff value of 6.3 for the SSC granulocyte:lymphocyte ratio was identified as the best discriminate between results that were rated as MDS and no MDS by CM. Taking into consideration an SSC granulocyte:lymphocyte ratio of >6.3 and a blast count by MFC of >5% and combining these parameters with the total number of aberrantly expressed antigens resulted in improved sensitivity (Table 10).

Relation of Aberrant Antigen Expression to the International Prognostic Scoring System

The relation of MFC results to IPSS was assessed in 431 patients with MDS who had IPSS scores available. Table 11 demonstrates that the total number of aberrantly expressed antigens gradually increases with increasing IPSS score (Spearman: r = 0.409; P < .001).

Table 11. Relation Between the Total Number of Aberrantly Expressed Antigens in All Cell Lineages and International Prognostic Scoring System Scorea
IPSS ScoreNo. of PatientsNo. of Aberrantly Expressed Antigens, Mean±SD
  • IPPS indicates International Prognostic Scoring System; SD, standard deviation.

  • a

    Data on IPSS scores were evaluable for 431 patients.

0992.26±1.65
0.51232.78±1.93
1.0852.99±1.82
1.5473.19±1.39
2.0563.34±1.81
2.5144.00±2.08
3.072.43±1.27

Relation of Aberrant Antigen Expression to Overall Survival

In 257 patients who had clinical follow-up, the relation between MFC findings and overall survival (OS) was analyzed. OS, as determined from the date of a diagnosis or a suspected diagnosis of MDS to death, amounted to 74% at 6 years. The number of aberrantly expressed antigens had a trend toward a correlation with OS (relative risk, 1.2; P = .09). Three parameters that were related to OS in univariate analyses were combined for an overall analysis of the relation between MDS-related findings by MFC and OS: Those parameters were ≥3 aberrantly expressed antigens, a blast count >5% in MFC, and an SSC granulocyte:lymphocyte ratio >6.3. Separating patients according to the presence of either versus none of these parameters resulted in significant differences in the 2-year OS rate (86% vs 100%; P = .008) (Fig. 7).

Figure 7.

This Kaplan-Meier plot illustrates overall survival for patients who had any of the 3 parameters 1) ≥3 aberrantly expressed antigens, 2) a blast count >5%, or 3) a side scatter ratio (granulocytes:lymphocytes) >6.3 (solid line) versus patients who had none of those parameters (dotted line; 10-year over survival rate, 68% vs 100%, respectively; P = .008).

DISCUSSION

Standard BM evaluation in suspected MDS includes CM, cytochemistry, and iron staining as well as a CG analysis for classification according to WHO8 and estimation of prognosis.10, 11 However, many patients who have inconclusive CM results and a normal karyotype remain9, 32, 33 for whom an MDS cannot be diagnosed or ruled out. MFC is capable of identifying dysplastic features in BM samples and, thus, was supposed to be added to the diagnostic workup in patients who have suspected MDS.13-16, 32, 34

In the current study, several aspects regarding the specificity and sensitivity of MFC for diagnosing MDS have been elucidated in agreement with previous studies.15, 16 In the current series and in the data reported by van de Loosdrecht et al,16 this also is true for the identification of dysplastic features in cell lineages that were not rated dysplastic by CM alone.

The enumeration of BM blasts has major prognostic impact10 and also guides the WHO classification.8 We observed a significant degree of correlation and concordance; however, patients who had counts that differed between both methods also were encountered. Although a difference in sample quality always may be considered as a cause, methodologic aspects also have to be considered. First, smears for CM investigation and BM samples after Ficoll processing for MFC, per se, are different sources. Second, it is suggested that myeloid blasts should be identified in MFC by using multiple parameters, including CD45/SSC signal and the expression of HLA-DR and CD34, together with myeloid markers.35 It has to be taken into consideration, however, that cells counted as myeloid blasts by CM may fall into the monocyte category in MFC and, thus, may be identified as dysplastic monocytes, which are not easily identified by CM and need at least nonspecific-esterase evaluation. Along this line, we observed that higher percentages of blasts were determined by CM in samples that had aberrant antigen expression in monocytes. Therefore, despite the high degree of concordance, the percentages of blasts determined by CM and by MFC should be considered separately. Consequently, the prognostic value of blast counts determined by flow cytometry should be evaluated in prospective multicenter studies.

Although the majority of previous reports focused on patients with proven MDS and, in part, added control samples from patients who clearly did not have a malignancy,13-16 in the current study, a large series of 1013 patients with suspected MDS was the focus of our analyses. None of these patients had undergone a prior proof or rule out of MDS. Thus, the current series represents a perfect cohort for the assessment of the potential of MFC in the diagnostic workup of patients with unclear cytopenias and a differential diagnosis of MDS. For the first time to our knowledge, the current study allows an evaluation of the diagnostic usefulness of MFC in MDS in the context of a large cohort of patients characterized by CM and CG. In patients who had unequivocal CM findings, we observed a high degree of concordance (82%) between CM and MFC results. It is noteworthy that, in patients who had dysplastic features that were insufficient to diagnose MDS using CM alone, aberrant antigen expression was observed with MFC, although at lower frequencies compared with the frequencies observed in patients with clear-cut MDS. Furthermore, 13 of 277 patients (4.7%) who had no evidence of MDS by CM were considered in agreement with MDS by MFC, posing the question of the specificity of MFC findings or the diagnostic role of MFC in addition to CM. A definite proof of MDS could be accomplished in 6 such patients in the current series based on MDS-typical CG aberrations, arguing in favor of the significance of MFC findings (which have been debated during the revision of the WHO classification).9

Thus, the current data suggest that, in patients who have normal karyotypes, MFC also may identify individuals with MDS who have CM results that indicate either no evidence of MDS or dysplastic characteristics that are not sufficient to clearly diagnose MDS. Consequently, although the diagnosis of MDS is based on CM results according to the WHO classification,8 it is noteworthy that CM per se does not have 100% sensitivity; therefore, an analysis of the sensitivity and specificity of MFC in diagnosing MDS using CM as the gold standard is hampered, and the respective results should be interpreted accordingly. This was proven in the current analysis at least in some of the patients who had positive MFC results, negative CM results, and CG results that revealed MDS-typical chromosomal aberrations. This scenario further supports the need for clinical validation of the diagnostic use of MFC in MDS, which should be evaluated together with patient outcomes.36

It is also clear, however, that, in the current series, 129 patients who had MDS clearly proven by CM did not have dysplastic features identified by MFC. Thus, the results of this crude analysis are suggestive of a combined use of CM and MFC in the diagnostic evaluation of patients with suspected MDS.

Although we were able to reproduce some previously reported findings, like the aberrant CD56 expression in monocytes in CMML,37 the current study adds important data on the capability of MFC to identify cell compartments affected by dysplasia that are not identified by CM. This was reported previously by van de Loosdrecht et al16 and was reproduced here by the identification of an aberrant antigen expression in granulocytes in patients who had no dysgranulopoiesis according to the CM results.

One important aspect of handling data generated by MFC in MDS is the definition of criteria that are useful to define MDS by MFC. Wells et al demonstrated that the diagnostic specificity of MFC increases together with an increase in the number of aberrant markers detected, although at the cost of decreasing sensitivity.15 We made a similar observation, although a specificity of 100% could not be reached in our dataset. This difference may be because, in the current series, some patients without evidence of MDS in CM analysis had MDS proven in CG analysis, as discussed above. Furthermore, it has to be stressed that the context of the current analysis is unique, in that a large number of patients with yet unclear cytopenias and with a differential diagnosis of MDS was assessed in parallel by the 2 gold-standard diagnostic methods, ie, CM and CG, complemented by MFC. Thus, we were able to analyze both the specificity and the sensitivity of MFC findings in a routine diagnostic setting. These results are in line with a previous report on the diagnostic value of MFC in MDS that was based on a smaller series of 124 retrospectively analyzed patients in which both CM and CG were used to validate MFC.38

In the current series, a correlation between an increasing number of aberrantly expressed antigens and an increasing IPSS score was observed, and this finding confirmed previous reports in smaller series.16 In addition, in a subset of 257 patients, OS was inferior if MDS-typical findings by MFC were present. This also is in line with previous data on patients with MDS who underwent stem cell transplantation.15

In conclusion, the current results demonstrate that, in a large series of patients who had a differential diagnosis of MDS, MFC in combination with CM and CG may add important diagnostic information. Patients who have aberrant antigen expression can be identified although their CM results may indicate no features of MDS. Cell compartments that are affected by dysplasia and are not identified by CM can be identified by MFC criteria. The further comparison based on CG and clinical data strengthens the significance of the current MFC findings. Because the sensitivity of both CM and MFC for identifying MDS is <100%, we suggest considering CM and MFC in a combined approach to optimize the overall diagnostic yield. Thus, the current gold standard for diagnosing MDS, ie, CM, should be used as such but with the awareness that other methods (ie, CG and MFC) may identify patients with MDS who have results that are not rated MDS by CM. Accordingly, as the most reasonable approach to diagnosing MDS, we suggest a combination of CM, CG, and MFC. Further studies should be performed to confirm these results and especially to define standards for the application of MFC in the diagnostic workup of MDS.34, 35

CONFLICT OF INTEREST DISCLOSURES

The authors made no disclosures.

Ancillary