|Title||Biomarker Stratified Design Enriched by Auxiliary Biomarker|
|Year of Publication||2016|
In precision medicine, drugs are developed to target subgroups of patients with certain biomarkers. In a large all-comer trial, the cost of ascertaining the true biomarker status for a large number of patients is often quite costly, especially when the proportion of the desired patients is small. We propose a special type of biomarker enrichment design, Biomarker Stratified Design Enriched by Auxiliary Biomarker (ABED), in which a subgroup of patients, typically the biomarker-positive patients, can be enriched based on the value of an inexpensive auxiliary variable that is positively correlated to the true biomarker without testing the true biomarker status for all screened patients. By comparing the efficiency of the proposed design with biomarker-stratified designs (BSD) in estimating various treatment parameters, the proposed design always reduces the total cost of the trial under the same time when the prevalence rate is small and the PPV, the probability that a patient with positive auxiliary biomarker also has a positive true biomarker, is large enough. Besides, in the proposed design in most cases we can directly enroll the patients selected in the screening process into the randomized test without waiting for the result of true biomarker test, which can remarkably reduce the waiting time when the test of true biomarkers is time consuming. Since PPV plays a very important role in the proposed design, an adaptive ABED involving Bayesian method is also proposed to deal with the misspecified PPV.