Evaluation of Potential Surrogate Endpoints

Abstract

Valid surrogate endpoints can make clinical trials more efficient, allowing for more trials to be conducted and more rapid development of effective treatments. Identifying useful surrogates is a statistically challenging but extremely valuable endeavor. This work develops statistical methods for the evaluation and comparison of biomarkers as correlates of protection (CoP). Methods herein were developed with a focus on a time-to-event clinical endpoint and possible time-varying effects of treatment, an important and thus far neglected topic is CoP evaluation. We propose a novel Weibull model and three methods of estimation for use in CoP evaluation; simulations and real data examples demonstrate the characteristics of these methods.