Abstract
Throughout recent years, there has been a rapidly increasing interest regarding the evaluation of so-called targeted therapies. These therapies are assumed to show a greater benefit in a pre-specified subgroup of patients—commonly identified by a predictive biomarker—as compared to the total patient population of interest. This situation has led to the necessity to develop biostatistical methods allowing an efficient evaluation of such treatments. Among others, adaptive enrichment designs have been proposed as a solution. These designs allow the selection of the most promising patient population based on an efficacy analysis at interim and restricting recruitment to these patients afterwards. As has recently been shown, the performance of the applied interim decision rule in such a design plays a crucial role in ensuring a successful trial. In this work, we investigate the situation when the primary outcome of the trial is a binary variable. Optimal decision rules are derived which incorporate the uncertainty about the treatment effects. These optimal decision rules are evaluated with respect to their performance in an adaptive enrichment design in terms of correct selection probability and power, and are compared to proposed ad hoc decision rules. Our methods are illustrated by means of a clinical trial example.
Highlights
Throughout the recent years, triggered by an increasingly more profound understanding of disease mechanisms, clinical researchers have come to the conclusion that the assumption of a homogenous treatment effect throughout the patient population of interest does not always hold true
If the uncertainty about treatment effects can be modeled in terms of prior distributions, it is possible to determine an optimal decision rule
In this subsection we provide some optimal decision rules which were derived for some specific parameter situations
Summary
Throughout the recent years, triggered by an increasingly more profound understanding of disease mechanisms, clinical researchers have come to the conclusion that the assumption of a homogenous treatment effect throughout the patient population of interest does not always hold true. In order to control the nominal significance level, adaptive enrichment designs address this issue by incorporating adjustment methods for multiple testing Since it crucially influences the properties and outcome of the trial, the role of the applied interim decision rule should not be undervalued. This article aims to further investigate the role of the applied decision rule in an adaptive enrichment design in case of a binary outcome variable, which has not yet been covered in the literature. Another important issue concerning the determination of a decision rule with desirable properties is the generally common uncertainty about treatment effects.
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
