Validation of the diagnostic potential of 17 immunophenotypic core markers defined by the ELN-iMDS-Flow Working Group in an independent cohort of patients with MDS

Myelodysplastic neoplasms (MDS) are a group of heterogeneous clonal hematopoietic stem cell disorders with complex diagnostic challenges.¹ Extensive diagnostic criteria published by the World Health Organization (WHO), combining hematological, morphological, cytogenetic, and molecular features aid clinicians in establishing an early diagnosis.² Despite these well-defined criteria, the diagnostic process remains challenging, especially in low-risk MDS.

Flow cytometric (FC) evaluation of the bone marrow, supported and recommended by the WHO, is an additional tool in this diagnostic process.^3-5 Several FC-based methods have been published and validated.⁶ The Ogata Score, consisting of four parameters, and the Integrated Flow Score (iFS), consisting of 44 parameters, are well-established methods for FC evaluation in the diagnostic work-up of MDS (Supporting Information S1: Table S1).^7-9

However, a survey amongst 200+ laboratories revealed a need to harmonize and optimize MDS-FC protocols to ensure broad clinical application.^9-11 Therefore, the European LeukemiaNet/International Myelodysplastic Syndrome Flow Cytometry Working Group (ELN-iMDS) performed a multicenter study in which they identified 17 immunophenotypic core markers (ELN Scoring System) that, when aberrant, are independently indicative of MDS and/or chronic myelomonocytic leukemia (CMML) (Supporting Information S1: Table S2). A cut-off of ≥3 aberrancies resulted in 80% concordance with cytomorphology and having >3% myeloid progenitor cells (MPC) had a Positive Predictive Value (PPV) of 98%.¹²

Here, we validated the ELN Scoring System and the >3% MPC cut-off compared to an integrated diagnostic approach, including cytomorphology, cytogenetics, next-generation sequencing (NGS), and two flow cytometry-based scores, that is, Ogata score and iFS, in an independent cohort. We specifically focused on low-risk MDS cases, as these are considered the major diagnostic challenge.¹³

In total, 180 MDS patients and 54 pathological controls (PC) were enrolled in this study (Supporting Information S1: Figure S1 and Supporting Information S1: Tables S3 and S4); 76% of the MDS patients were considered low-risk based on bone marrow blast percentage (defined as <5%), and 62% by using IPSS-M (defined as “Very Low,” “Low,” and “Moderate Low”).

First, we assessed the aberrancies according to the ELN Scoring System. MDS patients had a median of five aberrancies (range: 1–11); PC two aberrancies (range: 0–6). Having ≥3 aberrant markers resulted in an overall diagnostic accuracy of 87% as compared to an integrated diagnostic approach (p < 0.001). The iFS and Ogata Score had an overall accuracy of 91% and 71%, respectively (Figure 1A). The sensitivity of the ELN Scoring System (90%) was similar to the iFS (91%) and significantly higher than the Ogata Score (66%) (McNemar-test; p < 0.001). The specificity was lower in comparison to both the iFS and the Ogata Score, as demonstrated by a higher percentage of false positive cases (24%) compared to 7% in the iFS and 11% in the Ogata Score (McNemar-test; p = 0.012 and p = 0.092, respectively).

Next, we assessed the value of the ELN Scoring System in low-risk MDS. Our data show that the ELN Scoring System remained highly sensitive in low-risk MDS (87%), which is equal to the iFS (87%) and an improvement over the Ogata Score (54%) (Figure 1B,C). In high-risk MDS, sensitivity was >90% for all three FC scoring methods.

In some patients, a disconcordance between the results of the ELN Scoring System and the integrated diagnostic approach was observed (Supporting Information S1: Table S5). Eighteen MDS patients had a false negative result when using the ELN Scoring system. Applying the Ogata score, one of these patients would be correctly indicated as MDS, the others remained negative. The iFS proved to be more accurate, classifying eight of these patients as “consistent with MDS,” seven patients as “limited signs of dysplasia” and three patients as “no signs of dysplasia.”

On the other hand, 13/54 PC scored false positive according to the ELN Scoring System. Using the Ogata Score, 10/13 PC correctly scored negative. Applying the iFS, one PC scored “no signs of dysplasia,” nine “limited signs of dysplasia,” and three “consistent with MDS.” In the iFS, only the latter three cases would be considered FC-positive for MDS.

Next, we explored the contribution of individual markers by examining the frequency of aberrancies (Figure 2 and Supporting Information S1: Table S6). An increased CD71 coefficient of variation (CV; erythroid) was most common in MDS (85%) and PC (39%). CD117 (MPC), sideward light scatter (SSC; neutrophils), and CD13/CD16 maturation pattern of neutrophils were also aberrant in >50% of the MDS patients. In contrast, CD7, CD5, CD56 on MPC, the percentage of neutrophils, and CD33 expression were aberrant in a minority of MDS patients. Notably, these aberrancies were exclusively seen in MDS patients and not in PC (Figure 2). The MPC cut-off of >3% was exclusively observed in MDS patients (67/180), both low- and high-risk, resulting in a PPV of 100%. This implies that the proposed cut-off is useful in patients with minimal dysplasia and patients with overt disease.

Kern et al. showed that four parameters are significantly correlated to CMML but not to MDS, that is, low SSC of neutrophils, aberrant CD33 expression on neutrophils, aberrant percentage of monocytes, and CD13 expression on monocytes (Supporting Information S1: Table S2).¹² When we excluded these markers from the ELN Scoring System, the modified 13 marker Scoring System showed significantly improved specificity (from 76% to 96%; McNemar-test; p < 0.001); however, at the cost of a decrease in sensitivity from 90% to 81%. As a result, the overall accuracy of the modified score (84%) was similar to the original ELN Scoring System.

To summarize, we confirmed the utility of the ELN Scoring System in an independent cohort of MDS, particularly in low-risk MDS. The ELN Scoring system was superior to the Ogata Score and had, despite significantly fewer parameters, a minimal loss of diagnostic accuracy compared to the iFS.

A limitation of the ELN Scoring System is the false-positivity rate; 24% in our cohort and 22% in the original study.¹² An explanation may be that the ELN Scoring System counts the number of marker aberrancies, irrespective of which cell subsets are affected (MPC, neutrophils, monocytes, or erythroid), while the iFS requires two or more aberrancies in at least two separate cell subsets for a positive result. These differences result in patients with multiple dysplastic features in a single cell subset, or patients with a single dysplastic feature in multiple cell subsets, being classified as an MDS when using the ELN Scoring System, but not when applying the iFS. Analysis of a set of healthy donors might further elucidate false positivity. Furthermore, we showed that CMML-specific markers contributed to false-positivity; removing these markers improved specificity, yet not the overall accuracy.

The detection of MDS-associated immunophenotypes in cytopenic controls has been observed previously and may imply an increased risk for the development of MDS.^{6, 12, 14, 15} Dysplastic features detected by flow cytometry may precede dysplasia by morphological evaluation of the bone marrow. In our opinion, in case of persistent (pan)cytopenia, disconcordant results warrant early repetition of a bone marrow biopsy, in combination with NGS to confirm or exclude clonal cytopenia of unknown significance in these patients.

The lower number of markers analyzed in the ELN Scoring System, compared to the iFS, did not result in higher false-negativity (Figure 1). Notably, a general limitation of flow cytometry is the difficulty in evaluating dysmegakaryopoiesis; dysmegakaryopoiesis was present in most of the false negative cases by morphological evaluation, albeit not as sole dysplastic lineage.

Interestingly, we showed that certain aberrancies are exclusively found in MDS, primarily high-risk. This implies that high-risk MDS patients may have distinct immunophenotypes compared to their low-risk counterparts, especially when risk stratification is performed by IPSS-M, for example, CD7 and CD56 on MPC (Figure 2). Our findings support the ongoing efforts within the MDS community to identify distinct immunophenotypic profiles within the heterogenous landscape of MDS and understand the role of immune dysregulation within the bone marrow, for example, the work of the International Integrative Innovative Immunology for MDS consortium (i4MDS).¹⁶

Diagnostic criteria in MDS are subject to constant developments. Flow cytometry could be the next addition to international diagnostic guidelines, as already shown for CMML monocyte subset analysis.^{2, 17} In daily clinical practice, the ELN scoring system can be specifically useful in cases with inconclusive cytomorphology. It provides an objective and quick tool to assess immunophenotypic aberrancies in MDS. The high sensitivity in low-risk cases, according to both blast percentage and IPSS-M, highlights the applicability of the ELN Scoring system in healthcare institutions with (un)availability of NGS.

The hurdle for broad implementation of MDS-FC in clinical practice may further decrease due to ongoing technical advancements. Single tube panels such as in spectral flow cytometry, may lead to the development of standardized diagnostic tube(s) for MDS.¹⁸ As demonstrated by Duetz et al., the development of computational FC data analysis further supports objective and fast data interpretation.¹⁹

In conclusion, we confirmed the value of FC evaluation in the diagnostic work-up of cytopenic patients with a possible MDS diagnosis, especially in the diagnostically challenging low-risk MDS cases. The ELN Scoring System provided similar diagnostic accuracy as the much more elaborate iFS. Its simplicity in combination with the extensive recommendations published should allow easy and broad implementation in laboratories.⁹ It should be stressed that FC should not be used as a stand-alone tool, but always as part of an integrated diagnostic approach.

Arjan A. van de Loosdrecht designed this study. Arjan A. van de Loosdrecht and Theresia M. Westers supervised the analysis of FC data. Theresia M. Westers and Coen R. Veenstra analyzed and interpreted the data. Coen R. Veenstra drafted the paper. Arjan A. van de Loosdrecht, Theresia M. Westers, and Canan Alhan critically reviewed the manuscript and provided corrections. Arjan A. van de Loosdrecht is the head of the MDS group, supervising hematologist and senior author of this work.

The authors declare no conflict of interest.

All research was performed according to the Declaration of Helsinki and approved by the Medical Ethics Committee of the VU University Medical Center, Amsterdam, The Netherlands (research ethics protocols: VUmc 2014-100, VUmc 2019-3448).

This research received no funding.