A Model to Analyze Survival in a Clinical Trial Where the Treatment Is Primarily Expected to Delay Disease Progression

• Activity Number: 412
• Type: Topic Contributed
• Date/Time: Tuesday, August 11, 2015 : 2:00 PM to 3:50 PM
• Sponsor: Biopharmaceutical Section
• Author(s): Dianne M. Finkelstein* and David A. Schoenfeld
• Companies: Massachusetts General Hospital/Harvard University and Massachusetts General Hospital/Harvard University
• Keywords: survival ; progression ; Bayesian ; non-inferiority ; interval censored ; cancer

Abstract

Clinical trials of chronic diseases often assess efficacy by comparing treatments on the basis of two or more event-time outcomes. For example, in the case of cancer trials, treatments are compared on the basis of progression-free survival (PFS), which is the minimum of the time to progression or death. Because treatments are changed at disease progression, often trials show a significant benefit on PFS but not on overall survival, leaving regulators and providers with uncertainty about whether the drug should become the standard of care. To handle this situation, we propose a model for which the effect of treatment on post-progression survival (PPS) is different than the effect of the treatment before progression. Using a joint piecewise exponential model with two distinct hazards and a prior distribution on the effect of treatment on PPS, we propose a Bayesian non-inferiority analysis of overall survival. Our model is able to properly handle interval censored progression times. We apply the method to a breast cancer trial, showing that with plausible assumptions on the effect of treatment on PPS, the new treatment would exceed reasonable non-inferiority boundaries.