Performance of the Molecular International Prognostic Scoring System (IPSS-M) in Myelodysplastic Syndromes Treated with Hypomethylating Agent-Based Therapy: A Multicenter Analysis

Background: Patients with myelodysplastic syndromes (MDS) undergo risk stratification with various prognostication tools, such as the International Prognostic Scoring System (IPSS) and Revised IPSS (IPSS-R). The Molecular IPSS (IPSS-M) was developed incorporating genomic data with existing clinical and cytogenetic characteristics. Though IPSS-M has been shown to improve prognostication in all MDS patients at baseline and validity as a decision tool for timing of allogeneic stem cell transplantation (alloSCT), its utility specifically in patients undergoing hypomethylating agent (HMA) therapy remains unknown. The benefit of HMA therapy is mainly driven by restoring effective hematopoiesis among cytopenic patients, and it is unclear if there is benefit of starting treatment earlier based merely on higher disease risk. Here, we assess the efficacy of IPSS-M in HMA-treated patients with MDS. Methods: We retrospectively evaluated all MDS patients treated with frontline HMA-based therapy at multiple cancer centers in the United States and Europe. Patient characteristics and bone marrow (BM) data were assessed at diagnosis. Median overall survival (OS) and leukemia-free survival (LFS) were analyzed using the Kaplan-Meier method with log-rank tests comparing groups. Model discrimination was evaluated using the Harrell concordance index (c-index). Results: A total of 1838 MDS patients underwent treatment with HMA and had available IPSS-M risk stratification data. At diagnosis, the median age was 70 years (range: 19-94) and 1189 (65%) were male. Patients generally had high risk disease, including therapy-related MDS in 422/1511 (28%), complex cytogenetics in 582 (32%), and TP53 mutations in 444 (28%). By IPSS-R, 1180 patients (64%) had high- or very high-risk disease. By IPSS-M, 1506 patients (82%) had higher-risk disease (moderate high-, high-, and very high-risk). A total of 1542/1811 patients (85%) were treated with HMA monotherapy. By IPSS-R, the median OS was 60.0 months (mos) (95% CI: 46.0, not estimable (NE)) in 15 very low-, 71.2 mos (95% CI: 63.4, 114.8) in 152 low-, 38.0 mos (95% CI: 33.5, 44.8) in 482 intermediate-, 27.3 mos (95% CI: 24.3, 32.0), in 521 high-, and 14.8 mos (95% CI: 13.9, 16.3) in 659 very high-risk patients (p<0.0001) with a c-index of 0.661. Risk stratification by IPSS-M provided similar results: not reached (95% CI: 88.0, NE) in 12 very low-, 66.5 mos (95% CI: 62.2, 131.3) in 159 low-, 53.2 mos (95% CI: 45.4, 70.0) in 161 moderate low-, 36.5 mos (95% CI: 30.7, 46.9) in 238 moderate high-, 26.1 mos (95% CI: 23.5, 30.9) in 529 high-, and 15.8 mos (95% CI: 14.8, 17.0) in 739 very high-risk patients (p<0.0001). The c-index of IPSS-M in all HMA-treated MDS patients was 0.669. Among patients with MDS who were more likely to receive HMA per standard-of-care treatment (IPSS-R score >3.5), the IPSS-M remained statistically significant (p<0.0001) though less discerning, demonstrated by a c-index of 0.639. When evaluating LFS, comparable results were obtained with IPSS-R (p<0.0001, c-index 0.650) and IPSS-M (p<0.0001, c-index 0.657) for the entire cohort. For OS among subgroups, IPSS-M can well separate patients to different risk groups (p<0.001), but c-indices remained below 0.7 (normal karyotype: 0.660, complex cytogenetics: 0.604, therapy-related MDS: 0.638, TP53 mutated: 0.587, wild-type TP53: 0.656, and alloSCT recipients: 0.643). Patients were stratified into lower-risk and higher-risk groups by IPSS-R (IPSS-R score ≤3.5 and >3.5, respectively) and IPSS-M. The median OS was 66.5 mos (95% CI: 62.2, 93.4) in IPSS-R lower-, 22.7 mos (95% CI: 21.7, 23.9) in IPSS-R higher-, 64.3 mos (95% CI: 59.3. 87.4) in IPSS-M lower-, and 21.0 mos (95% CI: 19.5, 22.3) in IPSS-M higher-risk patients. Twenty-nine (2%) patients were upstaged from IPSS-R lower- to IPSS-M higher-risk disease with a median OS of 35.6 mos (95% CI: 31.1, NE), and 35 (2%) patients were downstaged from IPSS-R higher- to IPSS-M lower-risk disease with a median OS of 43.0 mos (95% CI: 29.5, 71.4). Conclusions: In this group of HMA-treated patients with MDS, the IPSS-M did not provide additional prognostic power over the IPSS-R, though median OS remained inversely proportional to IPSS-M risk category. Concordance indices for survival remained below 0.7 amongst all subgroups. Further development and validation of prognostic scoring systems in HMA-treated MDS patients are warranted.