The Evaluation of Proof-of-Concept Trial Design for Compound Selection

Traditionally, clinical proof-of-concept (POC) trials were designed as mini-phase III trials with a lower power and/or an intermediate endpoint. And go/no-go decisions were made based on a specification of the Type I error. This article considers a POC trial a tool to select the compound with a desired efficacy profile and introduces the approach to compute the distribution of the treatment effect given a go decision. The methodology is then applied to three POC design options that range from a direct comparison to indirect comparisons utilizing historical data. Using the tools proposed, we are able to compare the performance of the designs and impact of sample sizes with respect to the probability of meeting the target product profile given a go decision or no-go decision. Furthermore, for given a design we highlight the trade-off for the go/nogo decision and propose an approach to search for the optimal decision criterion that maximizes a utility defined by an economic scale. We illustrate the approach and results with an example for the development of a treatment for rheumatoid arthritis.