Keywords: precision medicine, biomarkers, phase III trials, study design
Many precision medicine efforts aim to identify subsets of patients who benefit from targeted therapies. These subsets of patients are defined by predictive biomarkers, which are characterized by an interaction between biomarker status and treatment benefit. Two study designs that are commonly employed in evaluating such biomarker/therapy combinations are the biomarker-stratified design and the enrichment design. In the biomarker-stratified design, patients are randomized to either the experimental or standard therapy in both the biomarker positive and negative groups. In the enrichment design, patients in the biomarker positive group only are randomized to treatment, while patients in the biomarker negative group are not enrolled to the study. This talk will focus on statistical and practical issues involved with both designs. The effect of misclassification of the biomarker in the trial will also be discussed.