Auxiliary variable-enriched biomarker-stratified design.

摘要: Clinical trials in the era of precision medicine require assessment biomarkers to identify appropriate subgroups patients for targeted therapy. In a biomarker-stratified design (BSD), are measured on all and used as stratification variables. However, such trial can be both inefficient costly, especially when prevalence subgroup primary interest is low cost assessing high. Efficiency improved costs reduced by using enriched designs, which interest, typically biomarker-positive patients, oversampled. We consider special type enrichment design, an auxiliary-variable-enriched (AEBSD), based some inexpensive auxiliary variable that positively correlated with true biomarker. The proposed AEBSD reduces total compared standard BSD rate biomarker positivity small positive predictive value (P P V) larger than rate. addition, AEBSD, we immediately randomize selected screening process without waiting result test, reducing treatment time. propose adaptive Bayesian method adjust assumed V while ongoing. Numerical studies example illustrate approach. An R package available.