An ROC-Based Approach to Interim Go/No-Go Decision-Making and Design Optimization in Late Phase Clinical Trials

• Activity Number: 546
• Type: Contributed
• Date/Time: Wednesday, August 12, 2015 : 10:30 AM to 12:20 PM
• Sponsor: Biopharmaceutical Section
• Author(s): Deli Wang* and Cui Lu and Lanju Zhang and Bo Yang
• Companies: AbbVie and AbbVie and AbbVie and AbbVie Pharmaceutical Research & Development
• Keywords: Interim futility analysis ; conditional power ; predictive power ; predicted confidence interval ; receiver operating characteristics ; two-stage design

Abstract

Interim analyses can be planned to make Go/No-Go decisions in late phase clinical trials and decision quality is an issue of interest. Timing of interim analysis is often selected based on empirical experience and thresholds for interim Go/No-Go decisions are determined arbitrarily. There is no systematic research to investigate interrelationship among three commonly used statistical methods for interim decision-making, namely conditional power, predictive power, and predicted confidence interval methods. A receiver operating characteristic (ROC) based approach is proposed to evaluate interim decision-making quality of these methods and they are proved to be equivalent analytically and verified by simulations under the ROC framework. To achieve the pre-specified sensitivity and specificity for Go/No-Go decision-making at interim, the required minimum sample size for interim analysis and the threshold for each of three statistical methods can be systematically determined based on the target design parameters of the clinical trials. The application of the obtained results to design optimization is given following Simon's two-stage design approach for continuous outcome measures.