Response Rate, Event-Free Survival (EFS), and Overall Survival in Higher-Risk Myelodysplastic Syndromes (MDS): U.S. Food and Drug Administration (FDA) Trial-Level and Patient-Level Analyses

Background: Therapeutic options for patients with higher-risk MDS are limited. To facilitate the development of novel therapeutics, various definitions of response rate and EFS have been suggested as clinical benefit endpoints for regulatory approvals. However, the relationship between response rate, EFS, and OS in higher-risk MDS has not been conclusively established. To evaluate the relationship between these endpoints, we conducted trial-level and patient-level analyses of higher-risk MDS trials submitted to the FDA.

Methods: We searched for trials submitted with New Drug Applications and Investigational New Drug Applications for the treatment of MDS between 2000 and 2024. Criteria for inclusion were randomized, controlled, multicenter trials that enrolled patients with higher-risk MDS (defined as IPSS-R intermediate risk or higher or IPSS intermediate-2 risk or higher). Patients with chronic myelomonocytic leukemia, acute myeloid leukemia, or baseline bone marrow blasts ≥ 20% were excluded from the analyses. The estimated odds ratios (OR - ratio of odds of response in treatment to odds of response in controls) of CR and CR+PR and hazard ratios (HR - ratios of hazard of observing events in treatment to hazard of observing events in controls at any time) for EFS and OS were calculated for each study using logistic or proportional hazards regression models as appropriate. OS estimates by response were obtained based on the Kaplan-Meier method. Response rates were defined as the best overall response - as defined in the protocol - at any time; response criteria included IWG 2000, IWG 2006, and custom criteria. EFS was defined as time from randomization to treatment failure (defined as date of transformation to AML, given availability of this information across trials) or death from any cause. Associations between treatment effects at the trial-level were evaluated using weighted linear regression analyses on the log-scale (weighted by sample size of each randomized comparison). Coefficient of determination (R²) and 95% confidence interval (CI) were calculated to measure the strength of associations. Only trials with active controls were included in the trial-level analyses of CR and CR+PR vs. OS, while all identified trials were included in the EFS vs. OS analysis. An exploratory patient-level responder analysis was also performed including patients from all identified trials to compare OS between responders and non-responders, regardless of treatment assignment in the pooled dataset.

Results: We identified 9 trials with a total of 2053 patients and 6 experimental agents for treatment of higher-risk MDS. Two trials lacked active controls. Across all trials, the population characteristics were: median age 71 years (range 19-94); 66% male, 34% female; 75% White, 16% Asian, 2% Black, 7% Other. The association between OS and CR at the trial-level was weak (R² = 0.40, 95% CI 0.08 - 0.87) and the association between OS and CR+PR was also weak (R² = 0.33, 95% CI 0.13 – 0.85). For the OS vs. EFS analysis, a moderate association was observed (R² = 0.82, 95% CI 0.31 – 0.96). In the patient-level responder analysis, patients who achieved a CR response had better OS compared with PR response (CR vs. PR: HR = 0.70, 95% CI 0.45 – 1.15), HI response (CR vs. HI: HR 0.56, 95% CI 0.44 – 0.70), and no response (CR vs. NR: HR = 0.33, 95% CI 0.27, 0.39).

Conclusions: On a trial-level, the meta-analysis of randomized, controlled trials in higher-risk MDS suggests a weak association between OS and CR rate, and a similarly weak association between OS and CR+PR rate. A moderate association between EFS and OS was observed and CIs were wide; potential confounders include differences in treatment post AML progression, hematopoietic stem cell transplantation, and/or censoring across trials. A patient-level responder analysis showed that CR responders have better OS compared to PR responders, HI responders, and non-responders. Our results are limited by the small number of trials, lack of an active comparator in two trials, and non-standardized response criteria across trials. However, the results suggest that CR remains the response endpoint associated with greatest long-term benefit and that EFS, as defined, is not an established surrogate for OS, but is worthy of further study.