Adaptive Bayesian compound designs for dose finding studies

Abstract

We consider the problem of how to efficiently and safely design dose finding studies. Bayesian sequential design methodology is discussed with new and current utility functions for the estimation of a maximum tolerated dose (MTD). In particular, we explore widely adopted approaches such as the continual reassessment method and minimizing the variance of the estimate of an MTD. New utility functions are constructed in the Bayesian framework and evaluated against current approaches. Such utilities can be computationally intensive, so we explore importance sampling (to re-weight posterior samples) to reduce computing time. Further, as such studies are generally first-in-man, the safety of patients is paramount. We therefore explore how to incorporate safety considerations into utility functions such that only safe and well predicted doses are selected. The performance of proposed and current utility functions is investigated via simulation studies. The paper concludes with a discussion of extensions that could be included into the methodology presented.