Evaluating Probability of Success in Oncology Drug Development

• Activity Number: 31
• Type: Contributed
• Date/Time: Sunday, August 4, 2013 : 2:00 PM to 3:50 PM
• Sponsor: Biopharmaceutical Section
• Author(s): Di Li*+
• Companies: Eisai Inc
• Keywords: Probability of Success ; Oncology Drug Development ; Bayesian

Abstract

Traditional oncology clinical trial design and sample size planning is usually based on achieving a targeted statistical power to detect a specified treatment effect. Trial planning based on statistical power comes with a significant amount of uncertainty relying on treatment estimates from historical data or small Phase 2 studies for novel oncology drugs. A high nominal statistical power does not guarantee a high probability of the trial achieving a successful outcome. Unlike statistical power, probability of success is the unconditional probability of a successful trial. It takes into account uncertainty of the unknown true treatment effect by incorporating a prior distribution. Probability of success can be employed to optimize the development path and enable better chance of success for a development program. In this talk, we give examples to illustrate the idea of using probability of success to inform better decision making in oncology drug development taking into account of the ultimate success of the development program when designing individual studies.