One concept may have multiple terms, and one term can harbor multiple concepts. The concept of erythroid cells undergoing proliferation analogous to granulocytes in leukemia has been described as erythroleukemia, erythremic myelosis, Di Guglielmo syndrome, M6 acute myeloid leukemia (AML), acute erythroleukemia (AEL), and pure erythroid leukemia.51 Additionally, erythroid predominance has been associated with various disorders including benign conditions and myelodysplastic syndromes (MDS). In MDS, the issue of erythroid predominance is subject to debate as the clinical approach and underlying pathophysiology are questionable. Although the 2016 revision of the World Health Organization (WHO) eliminated the non-erythroid blast count (NEBC) rule that advised enumerating marrow myeloblast percentages from non-erythroid cells in MDS with erythroid predominance (MDS-E), studies in favor and against this rule are still being published (Online Supplementary Table S1).96 To study the published contradictions regarding the NEBC rule, we retrospectively investigated a cohort of 280 patients from our institutional registry classified into appropriate subgroups by marrow erythroid cell- and myeloblast percentages following the WHO 2008 and 2016 criteria (Figure 1A). We performed survival analysis with censoring of patients undergoing stem cell transplantation or induction chemotherapy, evaluated the performance of the clinical risk scores with and without applying the NEBC rule using Harrell’s concordance index C, and questioned current definition of erythroid predominance. Our data show that MDS-E comprise both indolent and aggressive subtypes and that erythroid predominance can be a transient condition. We conclude that the NEBC rule is of no value based on its prognostic irrelevance and the inter- and intra-patient variety in MDS-E. Instead, we suggest refining the current definition of erythroid predominance: a relative increase in erythroid cells of at least 50% of total marrow cells.
On observing that MDS-E have a comparable outcome as MDS-NE despite lower myeloblast percentages, investigators stated that MDS-E patients have a poor prognosis that is inadequately recognized by myeloblast percentages from total marrow cells.86 Still, one may question whether MDS-E behave more aggressively than MDS-NE and as predicted by myeloblast percentages. In our study cohort of 280 patients (see details in Online Supplementary Appendix), MDS-E patients have a longer leukemia-free survival (LFS) and overall survival (OS) time than MDS-NE, irrespective of which WHO criteria are used (Figure 1B). Whereas the median percentage of marrow myeloblasts is lower in MDS-E than MDS-NE (WHO 2008: 1% vs. 4%, P<0.001; WHO 2016: 2% vs. 4%, P=0.013), these percentages have prognostic value for overall survival (OS) and leukemia-free survival (LFS) in MDS-E and MDS-NE (Online Supplementary Table S2). To study whether erythroid predominance leads to a misinterpretation of myeloblast percentages such that the prognosis of MDS-E is underestimated, we compared the outcome of WHO 2016 MDS-E and MDS-NE patients stratified by marrow myeloblast percentages. Interestingly, MDS-E and MDS-NE with corresponding myeloblast percentages show comparable outcomes, except for MDS-E with >10% myeloblasts who have a longer OS time (median not reached, P=0.025) (Figure 1C). In spite of our small sample size, these observations challenge the presumption that marrow myeloblast percentages lose their prognostic value in the presence of erythroid predominance. Rather, MDS-E may represent an indolent disorder that is accurately captured by low myeloblast percentages.
If MDS-E do not behave per se aggressively, only improved risk stratification would justify the continued use of the NEBC rule. However, the NEBC rule does neither increase the performance for predicting LFS and OS times of marrow myeloblast percentages nor of the IPSS-R and WHO criteria (Online Supplementary Table S2). Still, we elaborate on the impact of the NEBC rule on risk stratification for sake of comparison with other studies (Table 1). When enumerating myeloblast percentages from total marrow cells, risk distribution by the IPSS-R was comparable between MDS-E and MDS-NE (Online Supplementary Table S3A). In contrast, the WHO 2008 and 2016 criteria reflect the favorable prognosis of MDS-E, as fewer MDS-E than MDS-NE patients are classified as refractory anemia with excess blasts type 2 (6% vs. 16%, P<0.001) and, despite inclusion of AEL, excess blasts type 2 (8% vs. 35%, P<0.001), respectively (Figure 1D). When using the NEBC rule, a proportion of the MDS-E patients are upgraded within the IPSS-R and WHO criteria, respectively (Figure 1E). We found no difference between the outcome of upgraded MDS-E patients and MDS-E and MDS-NE patients remaining classified within initial categories (Figure 1F). These data contradict the presumption that the NEBC rule identifies an unfavorable MDS-E subgroup within distinct risk categories. Interestingly, Calvo et al. recommend the use of the NEBC rule for better risk stratification of MDS in general.9 Note that this recommendation is based on a marginally increased concordance probability estimate as a reflection of performance of the IPSS-R for predicting LFS and OS time. We observe an increase and decrease in concordance of 2% for predicting LFS and OS time, respectively, when using the NEBC rule in all MDS patients (Online Supplementary Table S2). Accordingly, our data do not support the use of the NEBC rule in any of the MDS patients. Although myeloid neoplasms with erythroid predominance are generally typified by poor karyotypes, multilineage dysplasia, pancytopenia and increased cellularity, the WHO does no longer define AEL separately.
Questioning whether AEL and MDS-E should be defined as unique entities, we searched for distinctive clinicopathological features across available publications (Table 1) and our patients. Despite poor peripheral blood counts (Online Supplementary Table S3B), AEL patients have a longer OS than MDS patients with excess blasts (median not reached vs. 8 months, P=0.005) and AML patients (median not reached vs. 13 months, P=0.046) (data not shown). This suggests that AEL patients do not have a poor prognosis that justifies promotion to AML, supporting its inclusion within the MDS spectrum. Compared to MDS-NE, our MDS-E patients had an increased incidence of the low-risk cytogenetic abnormality del(20q) and higher proportions of ring sideroblasts, suggestive for SF3B1 as underlying mechanisms (Online Supplementary Table S3B). The favorable prognostic profile of MDS-E is in contrast with the literature, which may indicate the heterogeneity underlying erythroid predominance possibly due to the non-specificity of a relative increase in marrow erythroid cells.
The last question that should be asked is whether a relative threshold of 50% marrow erythroid cells sufficiently defines erythroid predominance. First, we studied the kinetics of 50% marrow erythroid cells using repeated marrow aspirations of patients who were not treated with induction chemotherapy or autologous or allogeneic transplanted. Whereas an equal number of MDS-E patients have normalization or persistence of erythroid predominance at the time of follow up, most MDS-NE patients maintain a non-erythroid predominant marrow (Figure 2A). This suggests that erythroid predominance can be a transient condition that may be restored naturally, whereas its onset during the disease is uncommon. Second, we investigated the prognostic value of marrow erythroid cell percentages following Bennett et al. In contrast to the 50% threshold, we observed that extremely low and high marrow erythroid cell percentages have prognostic value: MDS with ≤15% marrow erythroid cells and MDS-E with ≥80% marrow erythroid cells have a poorer outcome than other MDS (OS: 16 vs. 47 months, P=0.006; LFS: median not reached, P=0.049) and MDS-E with <80% marrow erythroid cells (OS: 5 vs. 38 months, P=0.024), respectively (Figure 2B). Finally, with the aim of differentiating between erythroid predominance and myeloid hypoplasia, we related myeloid/erythroid (M/E) ratios to marrow- and peripheral blood cell counts using K-means clustering. Based on statistically significant M/E ratios, marrow myeloblast and erythroid cell percentages, blood erythroblast percentages and white blood cell and neutrophil counts, we identified four clusters of which two represented erythroid predominance (Figure 2C and Online Supplementary Table S4). Survival analysis suggests that erythroid predominance clusters comprised an indolent and aggressive subtype. Patients with an aggressive subtype had a lower age at the time of diagnosis (P=0.012) and a higher percentage of circulating erythroblasts (P=0.024) than patients with an indolent subtype (Figure 2D). These explorative analyses reveal the heterogeneity underlying MDS-E and suggest parameters for refining the current definition of erythroid predominance.
Based on this study, we conclude that there is no reason to use the NEBC rule. First, MDS-E are not a uniformly aggressive entity but are accurately diagnosed by myeloblast percentages from total bone marrow cells. Second, the NEBC rule does not improve risk stratification in either MDS-E or MDS in general. To prevent illegitimately upgrading MDS patients with indolent erythroid predominance, we support the decision of the WHO to reject the NEBC rule from clinical practice. We realize that this study is limited by its retrospective nature, sample size and referral bias, with patients selected from our tertiary-care center. However, we question whether the effect of such bias will be to underestimate the value of the NEBC rule. We expect that differences in reasoning and methodology might explain the controversy between our conclusions and previous studies.986 In the future, we wish not only for prospective population-based studies to achieve an evidence-based approach towards MDS-E, but also for functional and genomic studies to connect clinical heterogeneity of erythroid predominance and the underlying pathophysiology.1210
References
- Dameshek W, Baldini M. The Di Guglielmo syndrome. Blood. 1958; 13(2):192-194. PubMedGoogle Scholar
- Dameshek W. The DiGuglielmo syndrome revisited. Blood. 1969; 34(5):567-572. PubMedGoogle Scholar
- Bain BJ. Di Guglielmo and his syndromes. Br J Haematol. 2003; 120(6):939-943. PubMedhttps://doi.org/10.1046/j.1365-2141.2003.04181.xGoogle Scholar
- Arber DA, Orazi A, Hasserjian R. The 2016 revision to the World Health Organization classification of myeloid neoplasms and acute leukemia. Blood. 2016; 127(20):2391-2405. PubMedhttps://doi.org/10.1182/blood-2016-03-643544Google Scholar
- Vardiman JW, Thiele J, Arber DA. The 2008 revision of the World Health Organization (WHO) classification of myeloid neoplasms and acute leukemia: rationale and important changes. Blood. 2009; 114(5):937-951. PubMedhttps://doi.org/10.1182/blood-2009-03-209262Google Scholar
- Wang SA, Tang G, Fadare O. Erythroid-predominant myelodysplastic syndromes: enumeration of blasts from nonerythroid rather than total marrow cells provides superior risk stratification. Mod Pathol. 2008; 21(11):1394-1402. PubMedGoogle Scholar
- Bennett JM, Tuechler H, Aul C, Strupp C, Germing U. Dysplastic erythroid precursors in the myelodysplastic syndromes and the acute myeloid leukemias: Is there biologic significance? (How should blasts be counted?). Leuk Res. 2016; 47:63-69. Google Scholar
- Arenillas L, Calvo X, Luno E. Considering Bone Marrow Blasts From Nonerythroid Cellularity Improves the Prognostic Evaluation of Myelodysplastic Syndromes. J Clin Oncol. 2016; 34(27):3284-3292. PubMedhttps://doi.org/10.1200/JCO.2016.66.9705Google Scholar
- Calvo X, Arenillas L, Luno E. Enumerating bone marrow blasts from nonerythroid cellularity improves outcome prediction in myelodysplastic syndromes and permits a better definition of the intermediate risk category of the Revised International Prognostic Scoring System (IPSS-R). Am J Hematol. 2017; 92(7):614-621. Google Scholar
- Park S, Picard F, Dreyfus F. Erythroleukemia: a need for a new definition. Leukemia. 2002; 16(8):1399-1401. PubMedhttps://doi.org/10.1038/sj.leu.2402549Google Scholar
- Grossmann V, Bacher U, Haferlach C. Acute erythroid leukemia (AEL) can be separated into distinct prognostic subsets based on cytogenetic and molecular genetic characteristics. Leukemia. 2013; 27(9):1940-1943. PubMedhttps://doi.org/10.1038/leu.2013.144Google Scholar
- Cervera N, Carbuccia N, Garnier S. Molecular characterization of acute erythroid leukemia (M6-AML) using targeted next-generation sequencing. Leukemia. 2016; 30(4):966-970. Google Scholar