Background: This study investigated a framework that leverages the relationship between biomarkers and a target clinical endpoint to optimize an early development plan.
Methods: Different biomarker designs were assessed for proof of concept (PoC) and dose finding (DF) to improve phase 2b (Ph2b) design as well as phase 3 (Ph3) dose choice. A case study using a Bayesian trivariate normal distribution model for 2 biomarkers and a clinically relevant endpoint was utilized with simulation to assess performance characteristics.
Results: We found the following: (1) at typical sample sizes for early development trials, biomarkers appear useful for PoC but not for clinical endpoint DF; and (2) even with large amounts of prior information and near perfect correlation between biomarkers and clinical endpoints, Ph2b variability is only overcome by increased Ph2b sample sizes to improve Ph3 dose choice.
Conclusions: For highly variable clinical endpoints, the fastest path should be to demonstrate PoC by biomarkers and then go directly to Ph2b to measure the target clinical endpoint.
Keywords: biomarkers; early development plan; phase 3 dose choice.