Regency D
Iadapt: An R Package and Shiny App for Simulation and Implementation of Early Phase Dose-Finding Designs Incorporating Toxicity and Continuous Efficacy Outcomes (303964)
Cody Chiuzan, Columbia University
Laura Cosgrove, Columbia University
*Alyssa M Vanderbeek, Columbia University
Keywords: Clinical trial design, simulation, immunotherapy, early-phase, Shiny app
Early-phase statistical design is of critical importance in clinical trials, but often depends on strict assumptions that may or may not be satisfied by the tested agents. In oncology especially, one of these assumptions is a monotone dose-efficacy curve; as dose increases, efficacy improves as well. While reasonable for cytotoxic agents or other drug classes, cancer immunotherapies have different assumptions and mechanisms of actions, i.e., reduced toxicity profile and non-monotonic dose-response trends. Chiuzan et al. developed an early phase two-stage adaptive design that allows for non-monotone dose-efficacy relationship by incorporating evaluation of efficacy amongst all doses determined to be of acceptable toxicity. Rather than rule-based methods, the design uses a likelihood framework to assess safety and employs randomization to skew patient allocation to the most promising/efficacious doses. Here we present an R package and Shiny app that enable the user to assess the design performance under various scenarios and also facilitates trial implementation/conduct in real time.