A Bayesian adaptive design for addressing correlated late-onset outcomes in phase I/II randomized trials of drug combinations in oncology. Issue 6 (3rd June 2022)

• Record Type: Journal Article
• Authors: Yada, Shinjo
• Is Part Of: Communications in statistics. Volume 51:Issue 6(2022)
• Page Start: 2963
• Page End: 2977
• Publication Date: 2022-06-03
• Subjects: Adaptive design -- Drug combination -- Late-onset outcomes -- Phase I/II randomized trial; Mathematical statistics -- Periodicals; Mathematical statistics -- Data processing -- Periodicals; Digital computer simulation -- Periodicals; 519.5
• DOI: 10.1080/03610918.2019.1704784
• Languages: English
• ISSNs: 0361-0918

Abstract

Abstract: An advantage of phase I/II clinical trials for anticancer drugs is that safety and efficacy can be simultaneously evaluated in a single study. However, in order to benefit from this advantage, it is necessary for the safety and efficacy variables to comprise late-onset outcomes. In addition, current cancer treatments often involve therapies combining multiple drugs. Consequently, increasing emphasis has been placed on the importance of randomized clinical trials, which randomize the drugs being tested. This study proposes specific methods for addressing late-onset outcomes for safety and efficacy in phase I/II randomized trials of anticancer drug combinations. Data from patients in cases where the evaluation of safety and efficacy is not complete are considered missing data. A model taking into consideration the missing data mechanism, relationship between safety and efficacy, and correlation between observation points was used to adaptively determine the dose combination to assign to the next cohort. Simulation studies suggested that the proposed method is capable of supporting clinical trials in practice with realistic study periods. We believe the proposed method may flexibly adapt to practical requirements and contribute to advances in drug development.