Two-stage futility analysis in phase III adaptive trials with time-to-event endpoint

• Activity Number: 404
• Type: Contributed
• Date/Time: Tuesday, July 31, 2012 : 2:00 PM to 3:50 PM
• Sponsor: Biopharmaceutical Section
• Author(s): Xiaoyun (Nicole) Li*+ and Cong Chen
• Companies: Merck & Co. and Merck
• Keywords: Benefit-cost ratio ; survival analysis ; adaptive design ; oncology phase III trials

Abstract

Interim futility analysis is a critical component in oncology phase III studies due to high failure rate in spite of strong efficacy signals observed in phase II trials. For phase III trials with a time-to-event endpoint, a typical interim futility analysis is performed. We propose a two-stage futility analysis in phase III adaptive trials with a time-to-event endpoint to decide whether to continue enrollment or stop the trial completely in the early portion of the study without any requirement on minimum follow-up time. This approach is appropriate if the proportional hazard assumption is valid. However, it is rarely true in practice that proportional hazard assumption holds. We propose a two-stage interim futility design that allows a third option: pause enrollment and wait for mature follow up. To prevent operational challenge and save time, a stage-one futility analysis is performed without requirement on minimal follow-up. In case the futility bar has not crossed, we pause the study and wait for data to mature to make a final decision at stage-two. This approach mitigates the risk of stopping early for futility when it takes time for the study drug to differentiate from the control. We perform a simulation study to illustrate and compare this proposed design with conventional futility designs.