Not all that glitters is gold: A simple clinical score to unmask adverse SF3B1WT MDS-RS with low blast count

Abstract

Abstract INTRODUCTION Myelodysplastic syndromes with ring sideroblasts (MDS-RS) are typically associated with SF3B1 mutations (SF3B1MUT) and carry a favorable prognosis. However, approximately 20% of MDS-RS are SF3B1 wildtype (SF3B1WT), often harboring adverse genetics and poor outcomes even in the case of bone marrow (BM) low blast count (LB, <5%). In such cases, the presence of RS can be misleading. Therefore, this study aimed at identifying readily accessible clinical or morphologic markers predictive of SF3B1WTto detect these “high-risk” MDS-RS-LB cases, particularly in centers without access to molecular diagnostics. PATIENTS AND METHODS We analyzed MDS-RS cases from the IPSS-M public dataset, excluding those with del(5q) and with increased blast counts (learning cohort). RS positivity was defined according to WHO criteria (≥5% if SF3B1MUT, ≥15% without any secondary RS causes if SF3B1WT). We compared clinical, morphologic, cytogenetic, and mutational features between MDS-LB-RS SF3B1MUT vs SF3B1WT using univariate and multivariable logistic regression analysis. Overall survival (OS) and progression free survival (PFS) were also compared. In cases from the MDS Unit cohort (part of the validation cohort), flow cytometry was used to confront the myeloid, lymphoid and erythroid compartments between the two groups. Additionally, in MDS-LB-RS cases with IPSS-R ≤ 3.5, erythroid response to erythropoiesis stimulating agents (ESAs) and luspatercept (Luspa) was compared adopting IWG 2018 criteria. Finally, a predictive score for SF3B1 status was developed based on clinical and morphological parameters and validated in an external multicenter cohort composed by cases from MDS Unit (University of Florence), RESMD Spanish Registry and Argentina. RESULTS A total of 457 cases with MDS-LB-RS were identified from the IPSS-M dataset. SF3B1WT cases (n=87) were younger (mean 69.8 vs 72.3 y, p=0.03), more frequently had therapy-related MDS (11% vs 4.3%, p=0.01), and had significantly worse OS (median 3.8 vs 7.2 years, p<0.0001) and PFS (78% vs 91% at 3 years, p<0.0001) compared to SF3B1MUTcases (n=370). Although anemia severity was similar, SF3B1WT cases had lower platelet counts (149 vs 264 ×10⁹/L, p<0.0001), neutrophil counts (2.4 vs 3.2 ×10⁹/L, p<0.0001), fewer RS (36% vs 49%, p=0.01), and more frequently multilineage dysplasia (MLD, 76.5% vs 40%, p=0.0001). SF3B1WT cases were also enriched for poor-risk cytogenetics (8.9% vs 1.1%, p=0.008), complex karyotype (16.4% vs 0.3%, p=0.0001), and mutations in TP53 (28% vs 4.3%, p=0.001), U2AF1 (25.8% vs 1.3%, p=0.001), and SRSF2 (30.6% vs 1.8%, p=0.002), resulting in higher IPSS-M scores (0.14 vs –1.1, p<0.0001). Multi-hit (MH, defined as per WHO5th) TP53 status was more frequent (10% vs 0.5%, p=0.0001), and frameshift variants exclusive to SF3B1WT cases (29% vs 0%, p=0.002). Flow cytometry (n=37) showed that SF3B1WT cases (n=18) had fewer CD117⁺ erythroid precursors (30.6 vs 41.3%, p=0.03) and more BM monocytic cells (8.1 vs 3.9%, p<0.0001) than SF3B1MUTcases (n=29). Additionally, SF3B1WT cases showed a trend towards lower response rates to both ESAs (25% vs 55.6%, p=0.09) and Luspa (17% vs 39%, p=0.3). Multivariable analysis confirmed younger age (OR 0.973, p=0.038), lower platelet counts (OR 0.986, p<0.001), and presence of MLD (OR 0.354,p=0.002) as independent predictors of SF3B1WT. Based on these findings, we developed a score using age, platelet count and presence of MLD to predict SF3B1WT MDS-LB-RS forms: 0.15 x Age + 5 x MLD (1/0) + 0.05 x Plt (109/L). A cutoff < 23.05 identified SF3B1WT cases with 83.3% sensitivity and 75.2% specificity (AUC 0.83). This score was validated in the external cohort (SF3B1MUT n=238, SF3B1WT n=105), showing a similar good performance (sensitivity 77.2%, 80.3% specificity, AUC 0.82). CONCLUSIONS SF3B1WT MDS-LB-RS cases display distinct features, including enrichment of MH and frameshift TP53 mutations, and are associated with worse OS and PFS. In case of RS presence, if Perls staining is routinely performed, a simple score based on age, platelet count, and MLD can efficiently predict SF3B1WT forms, especially valuable in settings without molecular diagnostics. It would be interesting to evaluate the prognostic value of the score. Infact, these patients warrant closer monitoring due to their adverse genetic profile and inferior prognosis. Furthermore, their apparent lower response to standard anemia therapies suggests a need for alternative treatment approaches.