Abstract 03: Model regularization and considerations: Tuning dynamic borrowing in a hybrid clinical trial under Bayesian and frequentist framework

Abstract

Abstract There has been growing interest in leveraging existing external control data to facilitate clinical trial inferences and decision making. In recent years, the quality of real-world data (RWD) improved significantly, allowing it to be considered for being used as an external control. Borrowing by directly pooling from such external controls to underpowered randomized study has a potential to lead to biased estimates, and a dynamic borrowing framework is recommended to better control false positive outcomes (Lewis et al., 2019). However, the parameter tuning of Bayesian methods of dynamic borrowing remains challenging in order to optimize trial operating characteristics. To provide a general guidance on prior selection and parameter tuning, we implemented and evaluated multiple popular Bayesian prior settings through computational simulations and provided a described relationship to the corresponding frequentist model with regularization. Recommended prior distribution characteristics and practical considerations are proposed. Note: This abstract was not presented at the conference. Citation Format: Ruilin Li, Jiawen Zhu, Jane Fridlyand. Model regularization and considerations: Tuning dynamic borrowing in a hybrid clinical trial under Bayesian and frequentist framework [abstract]. In: Proceedings of the AACR Special Conference on Advancing Precision Medicine Drug Development: Incorporation of Real-World Data and Other Novel Strategies; Jan 9-12, 2020; San Diego, CA. Philadelphia (PA): AACR; Clin Cancer Res 2020;26(12_Suppl_1):Abstract nr 03.