Optimizing sample size in relative bioavailability trials using a Bayesian decision-theoretic framework

Bioequivalence (BE) trials are sometimes preceded by a pilot relative bioavailability (BA) trial to investigate whether the test formulation is sufficiently similar to the reference. The geometric mean ratio and its confidence bounds provide guidance as to how the BE trial can be appropriately sized to attain sufficient power. The aim of this work is to optimize the sample size of a pilot BA trial in order to minimize the overall sample size for the combination of pilot and pivotal trials. This is done through specification of a gain function associated with any of two possible outcomes of the trial; that is, abandon further development of the test formulation or proceed to a pivotal BE trial. The gain functions will be constructed on the basis of sample size considerations only, because subject numbers are indicative of both the cost and the feasibility of a clinical trial. Using simulations, it is demonstrated that for drugs with high intrasubject variability, the BA trial should be sufficiently sized to avoid erroneous decision making and to control the overall cost. In contrast, when the intrasubject variability of the pharmacokinetic (PK) parameters is low, not conducting the BA trial should be considered. It is concluded that the rather typical practice of conducting small pilot trials is unlikely to be a cost-effective approach.