This research aims to examine certain statistical properties of the design and analysis of cancer clinical trials in order to consider planned interim analysis of several endpoints simultaneously and to develop more generally applicable statistical methodology for such trials. The intent is to produce more realistic strategies for the conduct of cancer clinical trials and facilitate and make more efficient the evaluation of new cancer treatments. We will propose designs of cancer clinical trials based on more than one endpoint, including censored endpoints, and then assess their operating characteristics (i.e., their statistical properties). We will integrate the multiple endpoint methodology with group sequential methodology so that planned interim analysis of clinical trials with more than one primary endpoint has a firm basis. We also wish to extend group sequential methodology to k (greater than two)- armed trials. We plan to evaluate the new methods by comparing them to other methods using current clinical trial data as well as simulated data, in order to prospectively and retrospectively apply the methods developed to the analysis of cancer clinical trials. The statistical methodology will be both theoretical and numerical.
