Dose Finding in Late-Phase Drug Development

Abstract

A drug may fail or raise concerns that must be addressed in later-phase trials because of an unacceptable toxicity profile, even if the drug meets expectations for efficacy. The problem could be due to the late onset of unacceptable toxicities that were not observed in early-phase trials. We explore methodologies to find appropriate doses in situations where a drug meets the primary efficacy objective but concerns about toxicity remain. In this manuscript, we propose a general framework to design a good phase IV trial that is designed to optimize the treatment regimen. In the first step, we learn from the existing data about the dose-response and dose-toxicity relationships and further to explore and establish the relationship using a statistical model. Noting the limitations of the exploratory analyses, these analyses are not sufficient to allow us to make definitive conclusions. However, the prediction obtained from the model, either estimated efficacy metrics or safety parameters for some doses, can be incorporated in the design of the phase IV trial. In the second step, we further propose and compare design options, including options that could effectively incorporate information from these exploratory analyses, for clinical trials to find optimal doses, including trials to fulfill postmarketing commitments or requirements.