A meta-analysis of higher-risk myelodysplastic syndrome trials to evaluate the relationship between short-term endpoints and overall survival

Background: The objective of this work was to evaluate the relationship between the response rates and median overall survival (OS) in higher-risk myelodysplastic syndrome (HR-MDS) to determine whether response rates could be used as predictors of median OS. Methods: Relevant MDS clinical trials were identified through a review of published literature. Weighted linear regression was performed with various linearizing transformations of response rates and median OS using the in-house built HR-MDS clinical trials database. Covariates of interest were evaluated using a forward inclusion, backward elimination covariate model building procedure at α=0.01 and α=0.005, respectively. Results: Twenty-five trials involving 38 cohorts were included in the meta-analysis. The analysis demonstrated that partial response (PR) or better rate (sum of complete response (CR), marrow complete response (mCR) and PR rates) was a strong predictor of median OS (adjusted R²=0.64). The median OS was 3.3 months longer (P < 0.005) with azacitidine treatment compared to treatment with other drugs for a given response rate and prior therapy status. We also have shown that the median OS of treatment naïve HR-MDS patients was 4.5 months longer (P < 0.0001) compared to that of previously treated patients for a given response rate and treatment group. Conclusion: Significant correlation between PR or better rate and median OS in HR-MDS highlights the potential to use PR or better rate as a surrogate endpoint to accelerate development of novel therapies for MDS.