Validation of the Molecular International Prognostic Scoring System (IPSS-M) in Patients (Pts) with Myelodysplastic Syndromes/Neoplasms (MDS) Who Underwent Allogenic Stem Cell Transplantation (HSCT)

TK and JPB Co-first; RK and AMZ are Co-senior authors.

Introduction

MDS are genetically heterogenous neoplasms with diverse prognoses. While hypomethylating agents (HMA) can improve outcomes, HSCT remains the only potentially curative therapy. The IPSS-M incorporates clinical and molecular features to improve outcome prediction beyond traditional risk stratification tools. The IPSS-M has been validated in studies that either variably included pts who underwent HSCT, were limited by sample size, or were single-institution analyses. We aimed to validate the performance of the IPSS-M in predicting outcomes among pts with MDS who were treated with HMAs and subsequently underwent HSCT.

Methods

The VALIDATE database includes MDS pts treated with frontline HMA from 14 specialized centers. Only pts with available molecular data pre-HMA therapy and subsequently underwent HSCT were included. Time-to-event analyses used Kaplan-Meier estimator and the log-rank test. Overall survival (OS) was measured from time of HSCT. Multivariate-adjusted survival analysis used Cox proportional hazards. We compared the different scoring scores by Harrell’s c-index. This study was supported by an independent research grant from AbbVie.

Results

A total of 350 pts met eligibility (Panel A). Median age was 63 years (IQR: 58-68). MDS with excess blasts 1/2 (66%) was the most common type. Overall, 74% of pts received HMA monotherapy (27% decitabine; 73% azacitidine), while 26% received HMA in a combined therapy. The most prevalent mutations were TP53 (29%), ASXL1 (21%), RUNX1 (13%), STAG2 (12%), SRSF2 (12%), and DNMT3A (11%); 53% of pts had disease harboring >1 gene mutation. Only 44% had TP53 CN-LOH and 40% had MLL-PTD testing. The most common donor types were matched unrelated donors (MUD, 46%), haploidentical donors (21%), and matched related donors (MRD, 17%). Myeloablative, non-myeloablative, and reduced-intensity conditioning were used in 38%, 27%, and 27% of the pts, respectively. Only 14% of pts received post-HSCT maintenance therapy.

Our cohort was enriched with very high risk (41%) and high risk (30%) IPSS-R groups at time of diagnosis, while very low (1%), low (9%), and intermediate (20%) IPSS-R risk groups accounted for remaining 29% of pts. According to the IPSS-M, pts were classified into very low (2%), low (10%), moderate low (13%), moderate high (14%), high (27%) and very high (34%) risk groups. The IPSS-M re-stratified 198 (57%) pts, of whom 77 (39%) pts were up-staged and 121 (61%) pts were down-staged. More specifically, 35% and 26% of pts with intermediate risk IPSS-R MDS, and 32% and 35% of pts with high risk IPSS-R MDS were up-staged and down-staged, respectively. Importantly, 45% of pts with very high risk IPSS-R MDS were down-staged.

Median follow-up time was 20 (IQR:8-36) months (mo). The median OS for the entire cohort was 32 mo (IQR:10-Not reached [NR]). The IPSS-M groups showed significantly different OS (p-value= 0.001, Panel B) with median OS (mo, 95%CI) of: very low (NR, 11.5-NR), low (55, 22-NR), moderate low (48, 25-NR), moderate high (35, 22-NR), high (29, 23-NR), and very high (20, 13-28). The 5-year OS (%) were as following: very low (56%), low (47%), moderate low (37%), moderate high (40), high (43%), and very high (17%). IPSS-M risk (HR: 1.3, 95% CI:1.1-1.5) was significantly associated with OS in a multivariable model (HR, 95%CI) adjusted for age (1.01, 1.00-1.03), sex (1.2, 0.9-1.7), type of HMA used before HSCT (1.0, 0.9-1.1), number of HMA cycles (1.0, 0.9-1.05), donor type (1.0, 0.9-1.1), and conditioning regimen used (1.3, 1.1-1.5).

For OS, the IPSS-M significantly stratified pts assigned to the very high risk IPSS-R subgroup (p=0.009). However, the IPSS-M showed comparable performance to IPSS-R with c-index (95%CI): 0.597 (0.555-0.640) vs. 0.599 (0.554-0.643), respectively, for OS. When IPSS-M/IPSS-R were analyzed as continuous scores, the c-index (95%CI) for OS were: 0.606 (0.561-0.651) vs. 0.611 (0.559-0.662). When focusing on the higher risk group defined as IPSS ≥.1.5, the IPSS-M (c-index: 0.584) still had performance comparable to the IPSS-R (c-index: 0.606).

Conclusions

To our knowledge, this is the largest reported cohort in which IPSS-M performance was evaluated among pts with HMA-treated MDS and subsequently underwent HSCT. While the IPSS-M subgroups showed significant OS differences, the IPSS-M did not seem to substantially improve prediction when compared to the IPSS-R.