Background: Myelodysplastic syndrome (MDS) is risk-stratified according to cytopenias, bone marrow blast percentage, and chromosomal abnormalities according to the International Prognostic Scoring System (IPSS), revised IPSS (IPSS-R), and, most recently incorporating gene mutations, the molecular IPSS (IPSS-M). These scores assign discrete risk groups that predict patient survival; however, patients are typically given a binary risk estimate of higher (HR) or lower risk (LR) disease; HR denoting IPSS Int-2/H, IPSS-R VH/H/Int(>3.5), and IPSS-M VH/H/MH. Allogeneic stem cell transplant (HCT) is well-established for eligible patients with HR IPSS-R. However, some patients are assigned a discordant risk category between the IPSS-M and IPSS-R, and optimal treatment for these patients is not well-defined. We evaluated the clinical features, mutational profile, treatment choice, and survival outcomes for these patients.Methods: Patients were retrospectively identified and risk stratified according to IPSS, IPSS-R, and IPSS-M (github: papaemmelab/IPSSM). We obtained data on disease features, genetics, treatment, and survival. Mutation co-occurrence patterns were determined using unsupervised hierarchical clustering andnetwork analysis. Louvain clustering and permutation tests were used to assessnetwork metrics. Patient survival was analyzed byCox proportional hazards regression with adjustment for age and sex. Comparison of patient groups was performed using unpaired, two-tailed t-tests.Results: We characterized 566 patients with MDS diagnosed at Massachusetts General Hospital, Boston, USA, between 2013 and 2025 with median age of 73y. The most frequent genetic alterations were complex karyotype (27.4%) and TP53 gene mutations (24.9%), followed by ASXL1, TET2, chromosome 7 alterations, 5q deletion, and SRSF2, indicative of a high-risk patient population. Clustering by genetic alteration identified two dominant, mutually exclusive groups, a cluster for TP53 and a cluster including SRSF2, TET2, ASXL1, RUNX1, and STAG2 (“STARS”); additional smaller clusters were primarily defined by single mutations, including U2AF1, SF3B1, IDH1/2, and DDX41. Binary risk class (HR vs. LR) for 24.2% (n=137/566) of patients by IPSS and 13.1% (n=74/566) by IPSS-R was changed after applying IPSS-M: 35% (n=123/351) of IPSS LR patients changed to IPSS-M HR, and 16% (n=39/246) IPSS-R LR patients changed to IPSS-M HR. Patients with escalated risk exhibited enrichment for the STARS cluster genes (92% of IPSS-R LR to IPSS-M HR; and 76% IPSS LR to IPSS-M HR), accounting for the majority of conferred molecular risk, followed by U2AF1, ETV6, PHF6, and BCOR. Conversely, 11% (n=35/320) of IPSS-R HR changed to IPSS-M LR, and these patients were enriched for DNMT3A, DDX41,and SF3B1. Most patients with biallelic TP53 alteration were already classified as having IPSS-R HR and were less commonly changed by the IPSS-M. Compared to all IPSS-R LR patients, individuals with molecularly escalated risk displayed more co-mutational complexity, most prominently with increased co-mutation between STAG2, RUNX1, and SRSF2 (“SRS”) (network edge count: p <0.001). For IPSS-R LR patients, the presence of ≥2 SRS mutations was associated with thrombocytopenia (81 vs. 125 K/μL, p<0.001), neutropenia (0.8 vs. 1.9 K/μL, p<0.001), and higher bone marrow blast percentage (6 vs 1%, p<0.001), as compared to LR patients with <2 SRS mutations, with the majority of these patients (92%) having a normal karyotype. Finally, we evaluated whether IPSS-R LR patients escalated to IPSS-M HR benefitted from an HCT-based strategy as compared to non-transplant regimens. IPSS-R LR patients aged <75 y with IPSS-M HR showed improved survival when treated with an HCT-based strategy (HR 0.084, p=0.0197). Among patients with molecularly risk escalated disease who underwent HCT, there was a trend toward improved post-HCT survival with shorter time from diagnosis to HCT (p=0.1).Conclusions: We characterized the spectrum of mutations and risk stratification in a large MDS patient cohort at a tertiary referral hospital, with findings supporting the use of IPSS-M over IPSS-R to select HR patients who may benefit from an early HCT-based treatment strategy. A mutation network involving “STARS” genes is enriched in molecularly risk-escalated patients with prominent co-mutation of STAG2, RUNX1, and SRSF2, highlighting the important role for this gene subset in MDS biology and molecular risk stratification.

This content is only available as a PDF.
2025
Sign in via your Institution