A0954
Title: A Bayesian adaptive design for pediatric basket trials
Authors: Yimei Li - University of Pennsylvania (United States) [presenting]
Abstract: The basket trial is a novel type of trial that treats patients with the same genetic aberration regardless of their cancer types. Pediatric basket trials have unique features that require additional considerations for incorporating the adult information. We propose a Bayesian basket design for pediatric trials with adult data (BPAD) that performs dual information borrowing: borrow information from the adult data to the pediatric trial and between the cancer types within the pediatric trial. The BPAD design also accommodates potential heterogeneous treatment effect across cancer types, by allowing each cancer type belonging to the sensitive or insensitive latent subgroups. To make a go/no-go decision for each cancer type in the interim analyses, the design adaptively update the members of the subgroups based on the accumulated pediatric and adult data and borrow information among cancer types within the same subgroup. The simulation study shows that the BPAD design has better performance than a few existing designs, yielding high power to detect the treatment effect for the sensitive cancer types and maintaining a desirable type I error rate for the insensitive cancer types.
