Optimal Cost-Effective Designs of Phase II Proof of Concept Trials and Associated Go–No Go Decisions

Abstract

This manuscript discusses optimal cost-effective designs for Phase II proof of concept (PoC) trials. Unlike a confirmatory registration trial, a PoC trial is exploratory in nature, and sponsors of such trials have the liberty to choose the type I error rate and the power. The decision is largely driven by the perceived probability of having a truly active treatment per patient exposure (a surrogate measure to development cost), which is naturally captured in an efficiency score to be defined in this manuscript. Optimization of the score function leads to type I error rate and power (and therefore sample size) for the trial that is most cost-effective. This in turn leads to cost-effective go–no go criteria for development decisions. The idea is applied to derive optimal trial-level, program-level, and franchise-level design strategies. The study is not meant to provide any general conclusion because the settings used are largely simplified for illustrative purposes. However, through the examples provided herein, a reader should be able to gain useful insight into these design problems and apply them to the design of their own PoC trials.

Key Words:

• Confirmation trial
• Oncology
• Power
• Treatment effect
• Type I error rate

ACKNOWLEDGMENTS

During the preparation of this manuscript, we received insightful comments from our colleagues Drs. Frank Liu, Larry Gould, Linda Sun, and Jason Clark that led to an improved presentation.

Notes

∗Both authors contributed equally to this manuscript.