Validation of the Molecular International Prognostic Scoring System (IPSS-M) in Patients (Pts) with Myelodysplastic Syndromes/Neoplasms (MDS) Who Were Treated with Hypomethylating Agents (HMA)

Abstract

TK and JPB Co-first; RK and AMZ are Co-senior authors. Introduction Recognition of the growing importance of genetic mutations in prognostication for pts with MDS prompted the development of IPSS-M to improve outcome prediction. Subsequent validation studies included a heterogenous groups of pts; treated & untreated pts with lower & higher risk (HR) disease, and thus the value of IPSS-M in predicting outcomes of pts with HR-MDS treated with HMA remains unclear. We aimed to evaluate the ability of IPSS-M to predict outcomes of pts with HR-MDS who were treated with HMA in a large multicenter cohort. Methods Using the multi-center VALIDATE database, we included pts with HR-MDS (Defined as IPSS score >1 or IPSS-R score > 3.5) who were treated with any HMA in the frontline setting and had available molecular data before HMA initiation. Time-to-event analysis used Kaplan-Meier estimator and the log-rank tests. Survival was measured from time of HMA initiation. We compared the different scoring systems by Harrell's c-index. Best response to HMA was assessed using the IWG 2023 response criteria in HR pts (n=480) who had a marrow evaluation within 180 days of HMA initiation. This study was supported by an independent research grant from AbbVie. Results A total of 925 pts were included in the analysis. Median age was 68 years (IQR: 62-75) and 66% of the pts were men. MDS excess blast 1/2 accounted for 65% of the diagnosis. Overall, 76% of pts were treated with HMA monotherapy (29% decitabine and 71% azacitidine), 24% received HMA combination therapy, and 39% of pts had HSCT. The most common mutations (MT) were TP53 (32%), ASXL1 (20%), TET2 (13%) and SRSF2 (12%); 36% of pts had MDS harboring >1 gene MT. As anticipated, the cohort was enriched with high risk (26%) and very high risk (46%) IPSS-R groups. According to pre-therapy IPSS-M, pts were classified into very low (2%), low (9%), moderate low (11%), moderate high (14%), high (29%) and very high (36%) risk groups. IPSS-M reclassified 54% of pts, of whom 184 (37%) were up-staged and 317 (63%) were down-staged. Specifically, 38%/21% of pts with intermediate risk IPSS-R MDS, and 30%/36% of pts with high risk IPSS-R MDS were up-staged/down-staged, respectively. In addition, 44% of pts with very high risk IPSS-R MDS were down-staged. The median follow-up time was 19 (IQR:10-35) months (mo). Median overall survival (OS) was 19 (95% CI: 17-22) mo. Overall, 35% of the pts progressed to AML during follow up. IPSS-M groups showed significant differences for both OS and LFS (p-value: <0.0001 for both). The median OS (mo, 95%CI) based on IPSS-M categories were: low (37, 25-62), moderate low (30, 23-47), moderate high (29, 22-35), high (17, 14-20), and very high (14, 12-15) (Panel A). The 3-year LFS (%) were: low (20%), moderate low (48%), moderate high (41), high (50%), and very high (59%). IPSS-M was significantly associated with OS in a multivariable model adjusted for age, sex, type of HMA, number of HMA cycles and receiving allogeneic HSCT (HR: 1.4, 95% CI:1.2-1.5). However, IPSS-M showed comparable performance when compared with the IPSS-R with c-index (95%CI): 0.599 (0.557-0.622) vs. 0.619 (0.560-0.641) for OS. In addition, IPSS-M showed similar performance for LFS: 0.583 (0.560-0.606) vs. 0.588 (0.565-0.610) (Panel B). When IPSS-M/IPSS-R were used as continuous scores, the c-indices (95%CI) for OS and LFS were: 0.617 (0.592-0.640) vs. 0.6400 (0.616-0.664) for OS and 0.598 (0.575-0.622) vs. 0.606 (0.581-0.632) for LFS. We then analyzed the performance of IPSS-R/IPSS-M in pts with HR MDS (IPSS ≥1.5). Both IPSS-R and IPSS-M had comparable performance for OS prediction (c-index:0.548 vs. 0.566) and LFS (0.551 vs. 0.538). Based on the proposed IWG 2023 response criteria, complete remission rate (CR)/ overall response rate (ORR) were 12/47%, 9/58%, 13/47%, and 17/57% of pts assigned to moderate low, moderate high, high, and very high IPSS-M risk groups. No significant association was observed between IPSS-M and ORR (OR: 1.1, p=0.475) or CR (1.1, p=0.286). Conclusions In our multi-center large cohort of pts with HR MDS treated with frontline HMA, IPSS-M significantly stratified pts for OS and LFS. However, when compared with IPSS-R, IPSS-M did not seem to substantially improve prediction of outcomes among HMA-treated pts. Our results emphasize the high predictive prognostic power of clinical features and cytogenetic alterations among HR-MDS pts treated with HMA.