Keywords: Ph1 dose-finding study, dose selection, benefit-risk assessment, meta-analysis
Combination therapies targeting multiple pathways is a primary approach for improving anticancer effects in the current oncology drug development. In practice, Phase I trials with multiple drug combinations are conducted in parallel to find optional combinations and doses for further development in terms of benefit-risk. But this can be challenging due to small trial size and heterogeneity between trials. In this setting, maximizing available information across different combination studies is critical for decision making from both clinical and statistical perspective. We propose a model based meta-analytic approach in benefit-risk framework to synthesize all available data to find “optimal” combination regimen(s) or dose(s) for future development. Developing model for combination regimen needs some additional consideration than single agent as monotonicity in dose toxicity and dose efficacy relationships may not always hold. The proposed approach borrows information across different combination for precise estimation and therefore leads robust decision-making. We will use a real data example to illustrate the methodology.