Prognostic impact of SF3B1 mutation and multilineage dysplasia in myelodysplastic syndromes with ring sideroblasts: a Mayo Clinic study of 170 informative cases

Abstract

The revised 4^th edition of the World Health Organization (WHO4R) classification lists myelodysplastic syndromes with ring sideroblasts (MDS-RS) as a separate entity with single lineage (MDS-RS-SLD) or multilineage (MDS-RS-MLD) dysplasia. The more recent International Consensus Classification (ICC) distinguishes between MDS with SF3B1 mutation (MDS-SF3B1) and MDS-RS without SF3B1 mutation; the latter is instead included under the category of MDS not otherwise specified. The current study includes 170 Mayo Clinic patients with WHO4R-defined MDS-RS, including MDS-RS-SLD (N=83) and MDS-RSMLD (N=87); a subset of 145 patients were also evaluable for the presence of SF3B1 and other mutations, including 126 with (87%) and 19 (13%) without SF3B1 mutation. Median overall survival for all 170 patients was 6.6 years with 5- and 10-year survival rates of 59% and 25%, respectively. A significant difference in overall survival was apparent between MDS-RS-MLD and MDS-RS-SLD (P<0.01) but not between MDS-RS with and without SF3B1 mutation (P=0.36). Multivariable analysis confirmed the independent prognostic contribution of MLD (hazard ratio=1.8, 95% confidence interval: 1.1-2.8; P=0.01) and also identified age (P<0.01), transfusion need at diagnosis (P<0.01), and abnormal karyotype (P<0.01), as additional risk factors; the impact from SF3B1 or other mutations was not significant. Leukemia-free survival was independently affected by abnormal karyotype (P<0.01), RUNX1 (P=0.02) and IDH1 (P=0.01) mutations, but not by MLD or SF3B1 mutation. Exclusion of patients not meeting ICC-criteria for MDS-SF3B1 did not change the observations on overall survival. MLD-based, as opposed to SF3B1 mutation-based, disease classification for MDS-RS might be prognostically more relevant.

Introduction

The entity of refractory anemia with ring sideroblasts (MDS-RS) has been well codified for several decades.¹ According to the 2016/17 World Health Organization (WHO) classification system (revised 4^th edition; WHO4R),² myelodysplastic syndrome with ring sideroblasts (MDS-RS) was listed as a subcategory of MDS, primarily characterized by the presence of ≥15% ring sideroblasts, in bone marrow (BM) erythroid precursors; additional diagnostic criteria included the absence of ≥5% BM myeloblasts, among nucleated BM cells, or ≥1% peripheral blood (PB) myeloblasts, among PB leukocytes, Auer rods, and diagnostic criteria for MDS with isolated del(5q); of note, the presence of SF3B1 mutation was used to lower the diagnostically required ring sideroblasts threshold to 5%. The new 2022 International Consensus Classification (ICC-2022) of Myeloid Neoplasms and Acute Leukemias, which represents revision of the WHO4R document, considered SF3B1 mutation over and above ring sideroblasts in defining a more homogeneous group and have thus replaced the term MDS-RS with MDS-SF3B1;³ diagnostic criteria for the latter required the presence of SF3B1 mutation (≥variant allele frequency [VAF] 10%), absence of multi-hit TP53 or RUNX1 mutation, absence of isolated del(5q), -7/ del(7q), abn3q26.2, or complex karyotype, and absence of BM/PB myeloblasts ≥5%/2%, but did not include ring sideroblast percentage. MDS-RS without SF3B1 mutation has accordingly been relocated to the subcategory of MDS-not otherwise specified (MDS-NOS), regardless of the percentage of BM ring sideroblasts. MDS-NOS, according to ICC-2022, includes MDS with single lineage (SLD), multilineage (MLD), or no dysplasia.³ The rationale stated for these changes included the assumption that genetic risk stratification superseded the effect from morphologic distinction between MDS-RS-SLD (Figure 1) and MDS-RS-MLD (Figure 2). The current study examined these assumptions in a retrospective cohort of 170 patients with WHO4R-defined MDS-RS, in the context of presenting features and impact on survival.

Methods

The current study was approved by the Mayo Clinic institutional review board. Study patients were selected from institutional databases based on retrospective review of clinical and laboratory information and confirmation of MDS-RS diagnosis, according to WHO4R criteria.² Conventional methods were used for cytogenetic and next-generation sequencing (NGS) studies. Targeted exome sequencing included the following genes: TET2, ASXL1, DNMT3A, IDH1, IDH2, TP53, SRSF2, SF3B1, SH2B3, NPM1, FLT3, U2AF1, ZRSR2, JAK2, CSF3R, MPL, MFSD11, CEBPA, SETBP1, ZRSR2, RUNX1, IKZF1, CALR, KRAS, NRAS, CBL, PTPN11, STAG2, BCOR, and GATA2, and was performed on diagnostic BM specimens. Wright-Giemsa stain was used to stain PB and BM smears. Prussian blue iron stain was used to detect iron reserves and siderotic granulation in BM smears. The characteristics of dysplasia were classified using established criteria.⁴ Dysplasia in more than 10% of cells involving two or three lineages was described as MLD whereas dysplasia in more than 10% of erythroid-lineage cells was characterized as SLD. Erythroblasts with at least five siderotic granules encompassing at least a third of the nuclear circumference were classified as ring sideroblasts. Treatment administered was at the discretion of the treating physician and in concert with standard practice and included erythropoiesis stimulating agents (ESA), often as first-line and several other drugs, often as second- or third-line treatment, including luspatercept (N=41), lenalidomide (N=26), and hypomethylating agents (HMA; N=38). Non-parametric tests were used to compare the distributions of continuous variables, whereas the χ² test was used to compare the distributions of nominal variables. The time from diagnosis to death or last follow-up was used to calculate overall survival. Leukemia-free survival was calculated from the time of diagnosis to the date of leukemia transformation, death or the last day of follow-up. For univariate comparisons, the Kaplan-Meier approach was utilized in time-to-event analysis. For univariate and multivariate analysis of overall and leukemia-free survival, the Cox proportional hazard regression model was utilized. P value ≤0.05 was considered significant. JMP Pro 16.0.0 software package, SAS Institute, Cary, NC was utilized for all analyses.

Figure 1.Myelodysplastic syndrome with ring sideroblasts and single lineage dysplasia. (A, B) The bone marrow aspirate smear and core biopsy, respectively, demonstrate progressive maturation of the granulocytic and erythroid series with erythroid hyperplasia. Erythroid dysplasia is not striking and scattered late-stage nucleated red blood cells show coarsely basophilic stippling which corresponds with ring sideroblasts (A). There is no dysplasia in the granulocytic and megakaryocytic series and no increase in blasts. Wright-Giemsa stain, magnification 1,000x (A); hematoxylin and eosin stain, magnification 600x (B). (C) Iron stain reveals increased ring sideroblasts. Prussian-blue stain, magnification 1,200x.

Results

Phenotypic and genotypic comparisons between myelodysplastic syndrome with ring sideroblasts multilineage and myelodysplastic syndrome with ring sideroblasts single lineage

Online Supplementary Table S1 outlines the presenting clinical and laboratory features, along with postdiagnosis events, in 170 Mayo Clinic patients with WHO4R-defined MDS-RS (median age 73 years; range, 41-94; males 67%): 87 (51%) were classified as MDS-RS-MLD (median age 74 years; range, 47-94; males 71%) and 83 (49%) as MDS-RSSLD (median age 72; range, 41-89; males 63%); there was no significant difference between the two morphologic variants in age and sex distribution. Patients with MDSRS-MLD, compared to those with MDS-RS-SLD, were more likely to display lower leukocyte count (median 4.4x10⁹/L vs. 5.6x10⁹/L; P<0.01), lower absolute lymphocyte count (ALC; median 1.2x10⁹/L vs. 1.6x10⁹/L; P<0.01), lower absolute neutrophil count (ANC; median 2.6x10⁹/L vs. 3x10⁹/L; P=0.06), and lower platelet count (median 202x10⁹/L vs. 275x10⁹/L). Patients with MDS-RS-MLD were also more likely to display neutropenia (ANC <1x10⁹/L in 10% vs. 2%; P=0.02); lymphopenia (ALC <1.2x10⁹/L 47% vs. 31%; P=0.04), and thrombocytopenia (platelets <100x10⁹/L in 18% vs. 4%; P<0.01); a non-significant association for MDS-RS-MLD was also apparent for a lower hemoglobin level (P=0.19) and a higher incidence of transfusion need at diagnosis (P=0.19). Although not at a significant level, MDS-RS-SLD was associated with higher percentage of BM ring sideroblasts (P=0.11; Online Supplementary Table S1). BM and PB blast percentages were similar between the two morphologic cohorts; BM median 1 (range, 0-4) and PB median 0 (range, 0-1) (Online Supplementary Table S1). Cytogenetic information was available in 166 patients of whom 121 (73%) displayed normal karyotype, nine (5%) loss of Y chromosome only, eight (5%) sole trisomy 8, seven (4%) sole del(20q), five (3%) complex karyotype, three (2%) -7/7q- abnormality, and 13 (8%) other abnormalities. The frequency of abnormal karyotype other than LOY was 31% in MDS-RS-MLD versus 12% in MDS-RS-SLD (P<0.01); as depicted in Online Supplementary Table S1, the significant difference in abnormal karyotype distribution was mostly attributed to complex/-7/7q- abnormalities seen in eight (10%) of the 84 evaluable patients with MDS-RS-MLD and none of the 82 evaluable patients with MDS-RS-SLD (P<0.01). NGS information was available in 145 patients; mutational frequencies were 87% for SF3B1, 25% for TET2, 19% for DNMT3A, 11% for ASXL1, 5% for SRSF2, 5% for TP53, and 2% each for IDH1 and CSF3R; as depicted in Online Supplementary Table S1 and unlike the case with karyotype, there were no significant differences between the two morphologic groups of MDS-RS in regard to the distribution of these mutations.

Figure 2.Myelodysplastic syndrome with ring sideroblasts and multilineage dysplasia. (A) The bone marrow aspirate smear showed scattered dyserythropoietic cells (block arrow) characterized by nuclear budding, multinucleation or nuclear contour irregularities. The megakaryocytes also display dysplastic morphology characterized by small forms with monolobated or 2 separate distinct nuclear lobes (arrow); Wright-Giemsa stain, magnification 1,000x. (B) The bone marrow core biopsy is hypercellular with disruption of the normal tight erythroid colonies an increased dysplastic megakaryocytes (arrows); hematoxylin and eosin stain, magnification 600x. (C) Iron stain reveals numerous ring sideroblasts; Prussian-blue stain, magnification 1,200x.

Phenotypic and genotypic comparisons between SF3B1-mutated and -unmutated myelodysplastic syndrome with ring sideroblasts

NGS information, including SF3B1 mutational status, was available in 145 of the 170 study patients. Online Supplementary Table S2 outlines presenting features stratified by the presence or absence of SF3B1 mutation, including 126 (87%) with and 19 (13%) without the mutation. SF3B1 mutation was more likely to be associated with higher leukocyte (P<0.01), neutrophil (P<0.01), lymphocyte (P=0.03), monocyte (P=0.02), and platelet (P<0.01) counts and less likely to be associated with neutropenia (P=0.02), lymphopenia (P<0.01), or thrombocytopenia (P<0.01). Unlike the case with SLD versus MLD morphologic variants of MDS-RS, differences in karyotype distribution between SF3B1-mutated and -unmutated MDS-RS cases were not as pronounced while NGS-derived mutation distribution revealed significant clustering of wild-type SF3B1 with SRSF2 (26% vs. 2% mutational frequency in SF3B1-mutated cases; P<0.01), TP53 (16% vs. 3%; P=0.04), RUNX1 (10% vs. 25; P<0.01), IDH1 (16% vs. 0%; P<0.01), and U2AF1 (10% vs. 0%; P=0.01) mutations (Online Supplementary Table S2). In addition, borderline significance was apparent for higher prevalence of normal karyotype in SF3B1-mutated cases (79% vs. 58%) and that of complex karyotype in patients with wild-type SF3B1 (16% vs. 2%; Online Supplementary Table S2).

Impact of multilineage or SF3B1 mutation on overall and leukemia-free survival in revised 4th edition of the World Health Organization-defined myelodysplastic syndrome with ring sideroblasts

At a median follow-up of 5.2 years (range, 0.1-12.6) for living patients, 104 (61%) deaths, eight (5%) leukemic transformations, and seven (5%) allogeneic hematopoietic stem cell transplantations (AHSCT) were documented (Online Supplementary Table S1). There were significantly more deaths among patients with MDS-RS-MLD versus MDS-RS-SLD (72% vs. 49%; P<0.01; Online Supplementary Table S1) while this was not the case during comparison of SF3B1-mutated versus wild-type cases (63% vs. 47%; P=0.2; Online Supplementary Table S2). Median overall survival for all 170 patients was 6.6 years with 5- and 10-year survival rates of 59% and 25%, respectively (Figure 3A); the corresponding figures for MDS-RS-SLD were 7 years, 67%, and 31%; for MDS-RS-MLD, 5.5 years, 52%, and 17% (Figure 3B); for MDS-RS with SF3B1 mutation 6.8 years, 64%, and 26%; and MDS-RS with wild-type SF3B1 9.7 years, 52% and 0% (Figure 3C). Figure 3B reveals a significant difference in overall survival between MDS-RS-MLD and MDS-RS-SLD (hazard ratio [HR]=1.7, 95% confidence interval [CI]: 1.14-2.5; P<0.01) while such was not the case when comparing MDS-RS with and without SF3B1 mutation (P=0.36; Figure 3C).

The difference in overall survival between MDS-RS-MLD and MDS-RS-SLD was confirmed by multivariable analysis that included, individually, SF3B1 mutational status, age, ANC <1x10⁹/L, ALC <1.2x10⁹/L, platelets <100x10⁹/L, red cell transfusion need at diagnosis, abnormal karyotype, and TP53 mutation; significance was also sustained (HR=1.8, 95% CI: 1.1-2.8; P=0.01) in an all-inclusive multivariable analysis that included SF3B1 mutational status, karyotype, transfusion need at diagnosis, age, ANC, ALC, and platelet count, with the later four entered as continuous variables; additional independent risk factors in the latter analysis included age (P<0.01), transfusion need at diagnosis (P<0.01), and abnormal karyotype (P<0.01) and the results were not influenced by the addition of mutation information in the multivariable model: SF3B1 (P=0.48); TP53 (P=0.22); SRSF2 (P=0.97); IDH1 (P=0.93); RUNX1 (P=0.47); or U2AF1 (P=0.22).

There was borderline significance for a shorter leukemia-free survival in patients with MDS-RS-MLD versus MDS-RS-SLD (HR=3.3, 95% CI: 0.7-16.2; P=0.15) while a significant difference was apparent when comparing cases with wild-type versus mutated SF3B1 (HR=8.2, 95% CI: 1.1-58.7; P=0.03). However, the latter significance was lost during multivariable analysis that included other mutations that clustered with SF3B1 mutation including RUNX1 (HR=53.7), IDH1 (HR=54.9), and TP53 (HR=22.4); no other mutation or clinical variable (e.g., age, transfusion need, neutrophil, lymphocyte, or platelet count) displayed additional prognostic significance for leukemia-free survival. Furthermore, RUNX1 (P=0.02) and IDH1 (P=0.01), but not TP53 (P=0.35) mutations retained their significance for shortened leukemia-free survival when karyotype was added to the multivariable model with abnormal karyotype showing additional prognostic contribution (P<0.01).

Re-analysis of survival impact after exclusion of patients not meeting International Consensus Classification-2022 criteria for myelodysplastic syndromes with SF3B1 mutation

In order to address the confounding effect of revised criteria for diagnosis of MDS-SF3B1,³ we repeated the above outlined analyses after limiting the study population to patients with available NGS information (N=145) and excluding those with complex (N=5) or -7/7q- (N=3) cytogenetic abnormalities, multi-hit TP53 (N=2) or RUNX1 (N=5) mutations, and SF3B1 VAF ≥10% (N=3). After these adjustments, 130 patients were evaluable for further analysis with 63 MDS-RS-MLD and 67 MDS-RS-SLD cases; and 115 with and 15 without SF3B1 mutation. In univariate analysis, overall survival was similar between patients with and without SF3B1 mutation (P=0.92) but significantly worse in those with MLD compared to SLD (P=0.04). Multivariable analysis for overall survival demonstrated the significant difference in overall survival between MDS-RS-MLD and MDS-RS-SLD was independent of age, karyotype, and transfusion-need at diagnosis, with all four variables remaining significant: P=0.02, P<0.01, P<0.01, and P<0.01, respectively. A similar analysis for leukemia-free survival was handicapped by the small number of informative cases that included only three incidents of leukemic transformation with one of the three harboring TP53 mutation (P=0.08).

Figure 3.Overall survival among 170 patients with myelodysplastic syndrome with ring sideroblasts. (A) All patients; (B) all 170 patients stratified by single lineage versus multilineage dysplasia; and (C) a subset of 145 patients with information on SF3B1 mutation available.

Impact of multilineage dysplasia or SF3B1 mutation on treatment response

Accurate treatment information was available for luspatercept (N=41), revlimid (N=26), and hypomethylating agents (N=38). These drugs were often used after failure of prior therapy with ESA, which was historically mentioned in 121 patients, including 64 (74%) with MDS-RS-MLD and 57 (69%) with MDS-RS-SLD (P=0.48). Overall anemia response rates were 17% (7/41 patients) for luspatercept, 19% (5/26 patients) for revlimid, and 45% (17/38 patients) for HMA; the corresponding response rates in patients with MDS-RS-MLD versus MDS-RS-SLD were 16% (5/31 patients) versus 20% (2/10 patients) for luspatercept (P=0.78), 25% (4/16 patients) versus 10% (1/10 patients) for revlimid (P=0.33), and 46% (12/26 patients) versus 42% (5/12 patients) for HMA (P=0.8). A similar analysis comparing SF3B1-mutated versus wild-type patients showed luspatercept response in 13% (4/31 patients) versus 17% (1/6 patients; P=0.8), revlimid response in 26% (5/19 patients) versus 0% (0/3 patients; P=0.2), and HMA response in 32% (8/25 patients) versus 50% (3/6 patients; P=0.4), respectively. Similar data on treatment response to ESA were not available.

Discussion

The presence of ring sideroblasts in low-risk MDS has traditionally been considered as a marker of favorable prognosis in terms of both overall and leukemia-free survival, despite progressive anemia and red cell transfusion need in the majority of cases.^4-6 In a recently published natural history study of 138 patients with low-risk MDS with ≥5% ring sideroblasts, 65% of patients became red cell transfusion-dependent in their first 5 years of study enrollment with 42% deaths and 14% leukemic progressions reported during the same period.⁶ In our own previously published series of 76 patients with WHO4R-defined MDS-RS and followed for a median of 33 months, median overall survival was 46 months and leukemic transformation rate 3%.⁵ In the current study, we have expanded our WHO4R-defined MDS-RS patient population to 170 cases and the follow-up period to 5.2 years, in living patients; median overall survival was 6.6 years with 5-year and 10-year survival rates of 59% and 25%, respectively (Figure 3A); the incidence of leukemic transformation was 5%.

Most cases of MDS-RS are associated with an SF3B1 mutation,⁷ which is now considered a characteristic feature of the disease in whose presence diagnosis of MDS-RS can be made with BM ring sideroblast percentage >5% rather than the ≥15% threshold otherwise required.³ WHO4R classification also considered the number of lineages exhibiting morphologic dysplasia, in order to subclassify MDS-RS into two morphologic subcategories: MDS-RS-MLD and MDS-RS-SLD.² It is to be recalled that MDS-RS-SLD was previously recognized as refractory anemia with ringed sideroblasts (RARS) and MDS-RS-MLD as refractory cytopenia with multi-lineage dysplasia (RCMD), according to the 2008 WHO classification system (4^th edition).⁸ Microscopically, MDS-RS-SLD exhibits increased BM erythroid precursors associated with erythroid-lineage dysplasia, without associated dysplasia in granulocytes or megakaryocytes (i.e., <10% dysplastic forms; Figure 1). MDS-RS-MLD also exhibits erythroid-lineage dysplasia but also ≥10% dysplastic forms in granulocyte (e.g., nuclear hypolobulation) or megakaryocyte (e.g., micro-megakaryocytes, nuclear hypolobulation) lineages (Figure 2).²

ICC-2022 considered SF3B1 mutation over and above ring-sideroblasts in defining a more homogeneous group of low-risk MDS and have thus replaced the term MDS-RS with MDS-SF3B1.³ MDS-RS cases without SF3B1 mutations are now included in the ICC category of MDS-NOS, which also includes the subcategories of MDS-NOS with unilineage, multilineage, or no dysplasia, regardless of the percentage of BM ring sideroblasts.³ By contrast, the proposed 5^th edition of the WHO classification (WHO5) considers the distinction between SLD and MLD optional.⁹ The rationale stated for the ICC-2022 changes included the assumption that genetic risk stratification superseded the effect from morphologic distinction between MDS-RS-SLD and MDS-RS-MLD.⁴ In support of this assumption, Malcovati et al. found similar survival between SF3B1-mutated MDS patients with SLD versus MLD (P=0.4) while the same group of patients stratified by BM blast percentage at 5% resulted in significantly different survival (P<0.01).¹⁰ In the same study, the authors showed significantly longer survival for SF3B1-mutated versus unmutated cases in a spectrum of MDS subcategories, including RARS and RC-MD-RS, with the exception of those with excess blasts.¹⁰ In contrast to the aforementioned study by Malcovati et al., we have, in the past, repeatedly failed to demonstrate an independent prognostic effect from SF3B1 mutations in the context of WHO4R-defined MDS-RS. In the first (published in 2012) of several related work in MDS-RS, we examined the phenotypic and prognostic relevance of BM ring sideroblast percentage in an otherwise loosely-defined MDS-RS (MDS without excess blasts and RS% ≥1%);¹¹ we reported direct correlation of ring sideroblast percentage with age, platelet count, transfusion need and SF3B1 mutational frequency and inverse correlation with hemoglobin level, multilineage dysplasia and high-risk karyotype; more importantly, ring sideroblast percentage did not affect overall or leukemia-free survival. In a similar work published the same year (2012), we examined the prognostic interaction between SF3B1 mutation, morphology, and karyotype in MDS patients with ≥15% ring sideroblasts, regardless of BM blast content;¹² in the particular study, SF3B1 mutations did not display MLD-independent prognostic value, which was otherwise suggested in univariate analysis. Similarly, in a 2018-published study of 76 patients with MDS-RS,⁵ including 57 with MDS-RS-SLD and 19 with MDS-RS-MLD, we reported higher frequency of SF3B1 and DNMT3A mutations in the latter with no difference in overall survival, which was otherwise adversely affected by the presence of ASXL1 or absence of SF3B1 mutation. In a more recent communication,¹³ we reported similar survival in SF3B1-mutated MDS and SF3B1-mutated MDS/MPN, which otherwise shared similar mutational landscape, with the exception of higher frequency of JAK2 mutations in the latter.

The current manuscript includes the largest (N=170) single institutional cohort of WHO4R-defined MDS-RS with the objective to clarify the phenotype and genotype correlates of multilineage dysplasia and its impact on long-term survival, in the context of SF3B1 mutation and karyotype. The key observation from the current study was in regard to the independent prognostic relevance of MLD to overall survival in WHO4R-defined MDS-RS. In our contemporary study population, we were able to confirm the adverse survival impact of MLD in the context of other risk factors, including age, transfusion need at time of diagnosis, abnormal karyotype with or without inclusion of complex/-7/7q- abnormalities, previously recognized high-risk mutations, lymphopenia, neutropenia, and thrombocytopenia. By contrast, we were not able to demonstrate prognostic contribution from SF3B1 mutation, regardless of whether or not more recent criteria³ for its diagnosis were applied. This was despite the fact that wild-type SF3B1 was associated with adverse disease features, including high-risk mutations, thrombocytopenia, lymphopenia, and neutropenia (Online Supplementary Table S2). Of note, MLD was also associated with some of these adverse features and, more importantly, with adverse karyotype (Online Supplementary Table S1), but still showed an independent adverse effect on overall survival. In addition to MLD, our study highlights the prominent prognostic contribution from abnormal karyotype, both in terms of overall and leukemia-free survival, an effect that was mostly attributed to complex karyotype and -7/7q- abnormalities. The latter observation is in line with the ICC-2022 criteria for diagnosis of MDS-SF3B1, which requires exclusion of cases with the specific cytogenetic abnormalities.³ However, in the current study, abnormal karyotype other than complex karyotype or -7/7q- abnormalities remained prognostically significant for overall survival, independent of other risk factors. On the contrary, we were not able to demonstrate prognostic relevance for SF3B1 or other mutations, in regard to overall survival. The current study also found RUNX1, IDH1, and TP53 mutations to show prognostic relevance in regard to leukemia-free survival, again in line with ICC-2022 diagnostic criteria for MDS-SF3B1, which requires exclusion of cases with multihit TP53 and RUNX1 mutations.² The latter were also reported by others to be associated with leukemic progression in low-risk MDS.¹⁴

Recent developments in the treatment of MDS-RS include the introduction of new drugs, such as luspatercept.¹⁵ In the original phase III study of luspatercept versus placebo in transfusion-dependent patients with very low/low/ intermediate-risk MDS-RS who were either refractory or unlikely to respond to treatment with ESA,¹⁵ red blood cell transfusion-free period of at least 4 months was documented in 28% of study patients during weeks 1 through 48. In the particular study, response to luspatercept was not influenced by SF3B1 VAF or the total number of baseline somatic mutations.¹⁵ Similar observations regarding the lack of mutation impact on luspatercept treatment response in MDS-RS has been made by others.¹⁶ In a more recent phase III study of luspatercept versus epoetin a in transfusion-dependent and ESA-naïve patients with very low/low/intermediate risk MDS (regardless of ring sideroblast percentage), response to luspatercept was more likely in the presence of SF3B1 mutation but whether the same can be said in the context of MDS-RS is uncertain.¹⁷ In our own retrospective experience on the use of luspatercept in MDS-RS, neither SF3B1 nor other mutations appeared to effect treatment response.¹⁸ Ta k e n together, it is reasonable to question the value of SF3B1 or other mutations in predicting treatment response to currently approved drugs in MDS-RS.

Based on the observations from the current study, it is reasonable to conclude that MLD remains a powerful morphologic marker of aggressive disease in WHO4R-defined MDS-RS and is characterized peripherally by trilineage cytopenias and prognostically by shortened survival. These observations are in line with previous reports on the subject matter^19,20 but are now confirmed in a contemporary patient population that accounted for confounding influence from karyotype and mutations, including SF3B1. The current study also underscored the limited value of the SF3B1 mutation as a prognostic marker, in the context of WHO4R-defined MDS-RS, even after adjustments made to comply with criteria used in ICC-2022 for the diagnosis of MDS-SF3B1.³ These findings are consistent with those previously published by us¹² as well others.²¹ Taken together, these observations support the retention of MLD as a disease classifier in WHO4R-defined MDS-RS and suggest additional studies to clarify the role of SF3B1 mutation in a similar capacity, especially considering its promiscuity across the spectrum of myeloid neoplasms, with or without ring sideroblasts.^21,22

Footnotes

• Received November 21, 2023
• Accepted February 29, 2024

Correspondence

A. Tefferi tefferi.ayalew@mayo.edu

Disclosures

No conflicts of interest to disclose