|
Abstract:
|
A basket trial evaluates one or more treatments for e?cacy among more than one cancer type in a single clinical trial. Though the treatment targets the common genetic aberration that causes di?erent cancer types, the possible heterogeneity in the treatment e?ects poses challenge in modelling. Compared to traditional designs, basket trials can reduce the time required for testing and, by pooling across cancer types, they also allow the drugs to be tested for rare cancers. Basket trials are gaining increasing importance with advancements in precision medicine. Using covariate information has shown merit for improving efficacy in classification of the baskets. We incorporate subject-level biomarker information to aid identification of responsive and nonresponsive baskets. We model subjects’ responses using a two-component Bayesian mixture model where the mixture weights depend on a measure of similarly among subjects’ biomarker values. We demonstrate the performance of this model using simulation.
|
