Endpoint Exposure-Response Analyses in the Presence of Concurrent Dose Modification During Clinical Trials

Exposure-response (E-R) analyses are essential for dose selection in drug development, yet conventional endpoint E-R approaches often overlook concurrent clinical events such as adverse event (AE)-driven dose modifications (DMs), potentially leading to biased or misleading conclusions. In this study, we developed a framework to quantify the impact of AE-driven DMs on endpoint E-R relationships and to explore strategies to improve their accuracy. Using duvelisib as an exemplar, a drug known for frequent DMs due to Grade ≥ 3 infections, pneumonia, or transaminase elevations, we evaluated E-R relationships under three scenarios: ground truth, conventional E-R (based on planned dose exposures), and DM-adjusted E-R (accounting for AE-induced DMs and resultant exposure changes). Our analyses showed that conventional E-R analyses often deviated substantially from the ground truth at frequent DMs, particularly for AEs with delayed onset. Early-onset AEs and their resultant DMs could significantly distort E-R relationships for subsequent AEs. DM-adjusted E-R analyses better approximate the ground truth, especially for late-occurring AEs, but may introduce a risk of overcorrection of early-occurring AEs. These findings highlight the critical need to incorporate the timing and nature of AEs into E-R analyses to ensure robust interpretation, particularly in settings where DMs are frequent.