Analyzing OS in the Presence of Secondary Treatment Using Marginal Structural Model (MSM) and Rank-Preserving Structural Failure Time Model (RPSFT)

• Activity Number: 172
• Type: Contributed
• Date/Time: Monday, August 4, 2014 : 10:30 AM to 12:20 PM
• Sponsor: Biopharmaceutical Section
• Author(s): Yong Zhang*+ and Hongtao Zhang and Kalyanee Appanna and Can Cai and Kaushal Mishra
• Companies: Novartis and and Novartis and Novartis and Novartis
• Keywords: PFS ; OS ; secondary treatment ; MSM ; RPSFT

Abstract

In oncology clinical trials, progression-free survival (PFS) and overall survival (OS) are usually the two most important endpoints. While a beneficial effect on OS is the ultimate goal, PFS serves as a reasonably good surrogate for OS, which makes accelerated approval possible. Our work focuses on how to correctly estimate treatment effects in OS with the existence of a secondary treatment. When patients discontinue study treatment that could be censored for PFS or have non-fatal disease progression, they would normally initiate secondary salvage treatment in order to prolong survival. As a result, the effect of primary treatment on OS is confounded by the secondary treatment effect. We consider two types of causal models: marginal structural Cox models (MSM) and structural nested models (SNM). We examine their performances through simulation studies and implement the methods using data from a clinical trial. We also discuss the choice of models under different scenarios.