Sensitivity Analyses Adjusting Treatment Crossover for Survival Estimate: A Case Study

• Activity Number: 233
• Type: Topic Contributed
• Date/Time: Monday, August 10, 2015 : 2:00 PM to 3:50 PM
• Sponsor: International Chinese Statistical Association
• Author(s): Kathy Zhang and Ying Tian and Alan Rong* and Hua Yu and Xuesong Guan
• Companies: Medivation, Inc. and Biogen and Astellas and Biomarin and Amgen
• Keywords: treatment crossover ; oncology trials ; overall survial

Abstract

In oncology clinical trials, overall survival (OS) is widely accepted as a direct measure of clinical benefit and is considered the most reliable efficacy endpoint. However, demonstration of survival improvement is often confounded by subsequent cancer therapies received after trial treatment, which may result in an underestimate of the true treatment effect on OS. In this presentation, we discuss an example of a phase III randomized trial comparing an investigational treatment with standard treatment in patients with metastatic colorectal cancer. Compared to the results from the study primary model, the treatment effect on OS adjusting for the use of subsequent cancer therapies was estimated using sensitivity analyses based on different statistical methods. Results from models including Branson&Whitehead method, rank preserving structural failure time model, time-dependent cox model and Law's method will be presented. Statistical Considerations and model comparisons will be further discussed.