🤖 AI Summary
This study addresses the challenge of accurately estimating heterogeneous mediation effects and identifying interpretable patient subgroups with distinct mediation pathways in clinical trials with censored outcomes. The authors propose the M-survival learner framework, which integrates causal inference, survival analysis, and machine learning to handle high-dimensional covariates and censored data. The method introduces a survival-data–oriented criterion for detecting heterogeneous mediation effects and establishes a theoretically grounded mechanism for interpretable subgroup identification, thereby supporting biomarker-informed accelerated drug approval decisions. Empirical evaluations on both a Phase III HIV clinical trial dataset and simulation studies demonstrate its strong finite-sample performance, offering reliable evidence for regulatory decision-making.
📝 Abstract
Mediation analysis is a useful tool to evaluate surrogate endpoints in clinical trials. We propose a novel method, the M-survival learner, for estimating heterogeneous indirect treatment effects in the presence of censored outcomes. The proposed approach enables the identification of interpretable patient subgroups characterized by distinct mediation pathways. To distinguish heterogeneous from homogeneous mediation effects, we introduce a new statistical criterion specifically designed for survival data. The method provides a principled framework for evaluating heterogeneity in surrogate biomarker performance across patient populations, offering evidence to support accelerated approval drug. By explicitly assessing subgroup-specific surrogate validity, the proposed approach addresses key regulatory concerns regarding the reliability of surrogate endpoints. We further establish theoretical properties of the method to justify its statistical guarantees. We apply the approach to data from a Phase III randomized clinical trial of HIV treatment, demonstrating its practical utility in real-world settings. Extensive simulation studies further evaluate and demonstrate its finite-sample performance.