Thresholding of a Continuous Biomarker Based on a New Concept That Correctly Infers Efficacy in the Target Subgroup

• Activity Number: 526
• Type: Topic Contributed
• Date/Time: Wednesday, August 12, 2015 : 10:30 AM to 12:20 PM
• Sponsor: Biopharmaceutical Section
• Author(s): Yi Liu* and Jason Hsu and Szu-Yu Tang
• Companies: and The Ohio State University and Ventana Medical Systems, Inc.
• Keywords: biomarker ; modeling ; multiplicity

Abstract

Pharmaceutical companies nowadays are actively looking for biomarkers that differentiates treatment benefit of its experimental drug from standard therapy. Quite often a potential baseline biomarker can be identified based on mechanism of action. The optimal cutoff value to classify patients into target and non-target populations, however, is most often still undetermined before the planning of pivotal trials. We explore different approaches including modeling biomarker as categorical or continuous variable in establishing its relationship to efficacy. The derived efficacy in the overall and target/non-target populations involve modeling parameters as well as distribution of the biomarker value. Simultaneous inference that adjusts for multiplicity will be provided to allow change in the cutoff value for flexible decision making.