Bayesian Leveraging of Historical and Concurrent Data to Assess the Contribution of a New Molecular Entity with a Delayed Effect in a Combination Survival Trial

• Activity Number: 251 - SPEED: Biopharmaceutical Methods and Application I, Part 2
• Type: Contributed
• Date/Time: Monday, July 29, 2019 : 2:00 PM to 2:45 PM
• Sponsor: Biopharmaceutical Section
• Author(s): Samson Ghebremariam* and Lisa Hampson and Amy Racine-Poon and Beat Neuenschwander and Bharani Dharan and Kalyanee Appanna
• Companies: Novartis Pharmaceutical Corporation and Novartis Pharmaceutical Corporation and Novartis Pharmaceutical Corporation and Novartis Pharmaceutical Corporation and Novartis Pharmaceuticals and Novartis Pharmaceutical Corporation
• Keywords: Robust; Bayesian; Meta-analytic; Piece-wise; Hierarchical-model; Mixture-priors

Abstract

When testing a new therapy combining two new molecular entities (NMEs), one must demonstrate not only that the combination therapy is superior to standard of care, but also that each component contributes to its efficacy. In this presentation, we consider the design and analysis of a future 3-arm clinical trial randomizing patients between a combination of immuno-therapy (IO) (experimental IO + a NME), monotherapy (experimental IO) and standard of care (SOC IO). The primary endpoint is a time-to-event outcome and baseline hazards are assumed to be piece-wise constant.

There was a biological rationale to suggest that survival on SOC IO may be similar to survival on experimental IO. Furthermore, there was also historical subject-level data available on experimental IO in the intended indication and line of therapy. We adopted a robust Bayesian meta-analytic approach using a hierarchical model with mixture priors [1] to leverage the external historical data on SOC IO and internal concurrent data on experimental IO (co-data). We also explored whether incorporating the co-data could compensate for randomizing fewer patients to the experimental IO in the new trial.