85 – Biomarkers and Endpoint Validation 1
Statistics to Aid Decision Making for Early-Phase Cancer Trials with Time-to-Event Endpoints
Shaoyi Li
Celgene Corporation
As shown in many publications, the high failure rate of confirmatory trials may be due to the discordance in the endpoints adopted in Phase I/II trials. The commonly-used endpoint in early-phase cancer trials is the tumor response as opposed to survival in late-phase. While a robust tumor response is expected to translate into survival benefit, it may not always be true. The present study was designed to address this issue by adding a survival endpoint to the dose expansion phase. With limitation of small budget, a single-arm design is preferred. In this case, two issues need to be addressed: (a) regulatory guidance indicates that the interpretation of time-to-event single-arm data without a randomized reference is problematic; (b) the trial is too short which leads to a lot of survival data being censored when it comes to the decision point. Based on discussions with the clinical team, a simulation approach was proposed to project the Kaplan-Meier curve longer term. A reference curve was also simulated based on the input from key opinion leaders and publications. The final simulation result is helpful to aid decision making for early-phase cancer trials with time-to-event endpoints such as survival.