Evaluation of tumor size response metrics to predict survival in oncology clinical trials

Model-based drug development in oncology is still lagging despite a good momentum in the clinical pharmacology and pharmacometry community in the past few years. The failure rate of late-stage oncology studies is one of the highest across therapeutic areas. The modeling of the relationship between longitudinal tumor size and overall survival has been proposed to enhance learning in early clinical studies, to predict overall survival, and to simulate clinical trials. This approach has the potential to support proof of concept, early clinical decisions, and design of late-stage trials, but it is not yet widely integrated into the oncology drug development process. In this article, we review the state of these modeling efforts and discuss several key applications of these models. We conclude by suggesting a few paths forward.