Dose selection in seamless phase II/III clinical trials based on efficacy and toxicity

Seamless phase II/III clinical trials are attractive in development of new drugs because they accelerate the drug development process. Seamless phase II/III trials are carried out in two stages. After stage 1 (phase II stage), an interim analysis is performed and a decision is made on whether to proceed to stage 2 (phase III stage). If the decision is to continue with further testing, some dose selection procedure is used to determine the set of doses to be tested in stage 2. Methodology exists for the analysis of such trials that allows complete flexibility of the choice of doses that continue to the second stage. There is very little work, however, on optimizing the selection of the doses. This is a challenging problem as it requires incorporation of the dose-response relationship, of the observed safety profile and of the planned analysis method. In this thesis we propose a dose-selection procedure for binary outcomes in adaptive seamless phase II/III clinical trials that incorporates the dose response relationship, and explicitly incorporates both efficacy and toxicity. The choice of the doses to continue to stage 2 is made by comparing the predictive power of the potential sets of doses which might continue to stage 2