CFE 2019: Start Registration
View Submission - CMStatistics
Title: A Bayesian decision-theoretic design for a treatment-selection biomarker Authors:  Gary Rosner - Johns Hopkins University (United States) [presenting]
Zheyu Wang - Johns Hopkins University (United States)
Chenguang Wang - Johns Hopkins University (United States)
Abstract: Drug development, particularly in oncology, often focuses on developing therapies that target molecular pathways in an attempt to disrupt disease processes or alleviate symptoms. Successful drug development often relies on the ability to select appropriate patient subpopulations that are more likely to respond to the treatment. As a result, clinical studies of these targeted agents often include biomarker assessment, particularly early studies of the treatment's safety and activity. We propose a two-stage design based on a Bayesian decision-theoretic approach to achieve the dual aim of biomarker subgroup selection and efficacy demonstration. Stage 1 enrolls patients regardless of their biomarker values. An analysis at the end of stage 1 identifies a biomarker threshold based on Stage 1 data and any external information that may be available. The second stage enrolls either all patients or a biomarker-defined subset of patients, depending on the interim analysis results. We will discuss the design and its characteristics in light of the particular challenges and opportunities for clinical trial design of targeted therapies.