Department of Biostatistics Seminar/Workshop Series

Patient-Specific Dose-Finding Based on Bivariate Efficacy-Toxicity Outcomes and Prognostic Covariates

Peter Thall, PhD

Professor, Department of Biostatistics, University of Texas, M.D. Anderson Cancer Center

Wednesday, November 2, 1:30-2:30pm, MRBIII Conference Room 1220

A new Bayesian methodology for individualized dose-finding based on efficacy and toxicity is presented. Implementation requires an informative prior on effects of established patient covariates, obtained from a preliminary fit of the model to historical data. All other model parameters have non-informative priors. For each of a representative set of covariate vectors, elicited limits on the probabilities of efficacy and toxicity are used to construct bounding functions that determine the covariate-specific acceptability of each dose. An efficacy-toxicity trade-off function is established from target probability pairs elicited for a reference covariate vector. During the trial, each patient's dose is chosen from the set of acceptable doses to optimize the efficacy-toxicity trade-off for his/her specific covariates. Because selected doses are covariate-specific and the method is sequentially outcome-adaptive, (1) different patients may receive different doses at the same point in the trial, (2) the set of eligible patients may change dynamically during the trial, and (3) at the end of the trial, rather than recommending one dose for all patients, the method provides an algorithm for choosing each new patient’s dose based on his/her prognostic covariates. The method is illustrated by a phase I/II trial in acute leukemia.