Abstract:
|
A Bayesian randomized phase III group sequential enrichment design to facilitate precision medicine is proposed. An experimental treatment is compared to a control based on survival time, using early response to assist with adaptive variable selection and enrichment. Initially, patients are enrolled under broad eligibility criteria. At each interim decision, submodels for regression of survival time and response on baseline covariates and treatment are fit. Variable selection is used to identify a covariate subvector characterizing treatment-sensitive patients and determine a personalized benefit index for making comparative decisions. Enrollment is restricted to the most recently identified treatment sensitive patients. Decision cutoffs are calibrated to control overall type I error and account for adaptive enrollment restriction. Simulations show that the proposed design reliably identifies a sensitive subpopulation, has much higher generalized power than several existing enrichment designs and a conventional group sequential design, and is robust.
|