A Bayesian Phase I/II Trial Design with Delayed Outcomes

• Activity Number: 404
• Type: Contributed
• Date/Time: Tuesday, July 31, 2012 : 2:00 PM to 3:50 PM
• Sponsor: Biopharmaceutical Section
• Author(s): Ick Hoon Jin*+ and Ying Yuan
• Companies: MD Anderson Cancer Center and MD Anderson Cancer Center
• Keywords: Bayesian Adaptive Clinical Design ; Dose-Finding ; Phase I/II Clinical Trial Design ; Missing Data ; Data Augmentation ; Piecewise Exponential Model

Abstract

Recently, researchers have been focusing on developing dose-finding methods incorporating both toxicity and efficacy. One such method, EffTox, proposed by Thall and Cook (2004) selects a proper dose adaptively based on trade-offs between the probabilities of efficacy and toxicity. However, in practice, it requires that the outcomes should be observed shortly after the initiation of treatment, which may not be possible for the cases of delayed outcomes. To accommodate late-onset problems, we propose the data augmentation EffTox (DA-EffTox) design. The main innovation of the design is that we treat the unobserved outcomes as missing data and impute the missing data using the Bayesian data augmentation approach. In addition, we introduce the easier way of a prior specification to EffTox method by employing a weakly informative prior proposed by Gelman et al. (2008). We evaluate the performance of the DA-EffTox through extensive simulation studies and show that the proposed design solves the issues related to delayed outcomes and owns desirable operating characteristics.