Online Program

Return to main conference page

All Times EDT

Thursday, September 24
Thu, Sep 24, 12:00 PM - 1:15 PM
Virtual
Roundtables

TL29-Using Multi-Arm Multi-Stage Designs in Oncology Clinical Trials (302303)

*Priyam Mitra, Bristol-Myers Squibb 

Keywords: mutli-arm trials, time to event endpoint, famly wise error rate

With new discoveries in Oncology drug development and given the need for accelerated and streamlined investigation plan, innovative approaches that allow for efficient development of new assets are needed. One such approach consists of developing a seamless plan for asset development where the study is designed to answer questions of different natures in a timely manner while prioritizing patient need. A seamless study eliminates additional time between multiple phases and makes operational tasks easier. A Multi Arm Multi Stage (MAMS) trial design, based on a definitive primary and an intermediate outcome measure allows for a quicker and more efficient evaluation of multiple treatments under one study protocol. The definitive outcome measure is used as the primary endpoint in the phase III part of the study, while the intermediate outcome measure helps with screening for evidence (efficacy, safety signals, dose finding etc.) in the earlier parts of the study.

We evaluate and compare different MAMS designs that have been proposed in literature. A MAMS approach which focuses on a time to event outcome measure as the primary endpoint is discussed. This design allows for multiple arm efficacy comparison while including early stops for both efficacy and futility. We also present simulations to assess the impact of the methodology used on the operating characteristics. A detailed assessment of alpha spending across stages and arms is presented to ensure proper Type I error control. We also evaluate utilizing a time to event as well as a binary outcome measure for the intermediate decision.

Questions for discussion: 1. What is the optimal way to control Type I error in a MAMS design framework, as both multiple arm comparisons and spending across stages should be considered? 2. How does the correlation between the intermediate and final outcome measures affect the power of the study?