U-MET: Utility-Based Dose Optimization Approach for Multiple-Dose Randomized Trial Designs

Abstract

The advent of dose optimization has led to a paradigm shift in oncology clinical trials to find the optimal biologic dose (OBD). Phase 1/2 trials with randomized doses can facilitate additional investigation of the identified OBD in a targeted population by incorporating safety and efficacy data. We propose to compare randomized doses by extending the Bayesian optimal interval phase 1/2 (BOIN12) approach, an OBD-finding design in which a utility is used to account for risk-benefit trade-offs. Specifically, we used the BOIN12 standardized mean utilities in a hypothesis testing framework, in which frequentist and Bayesian inference were evaluated to compare the standardized mean utilities across doses and identify the OBD. We performed simulation studies and a hypothetical oncology study to compare the proposed BOIN12 utility-based extension (U-MET) design with an empirical design. The results demonstrated that the U-MET design has satisfactory operating characteristics for selecting the OBD. We recommend using the U-MET design as the primary dose comparison approach, or alternatively as supportive evidence, for optimal dose selection.