Can knowledge of the relationship between biomarkers and clinical endpoints help us to optimize an early development program and improve the probability of selecting the right dose in Phase 3?
Our client approached us hoping to expedite dose-finding with biomarkers in Phase 1b, and to design an optimal Phase 2b clinical endpoint trial to maximize probability of correct Phase 3 dose selection.
The Cytel Solution:
Our strategy resulted in the paper subsequently published by Musser et al, 2012, according to which:
Phase 1b trials with typical sample sizes complete more quickly with biomarker endpoints.
No matter the level of correlation between surrogate and clinical endpoints, variability in Phase 2 requires large sample size.
Step 1: We built a statistical model to capture the known relationship between 2 biomarkers and the clinical endpoint. Using this model – a Bayesian trivariate normal distribution – we were able to simulate the sequence of clinical trials: two Phase 1b trials and one Phase 2b trial.
Step 2: We used powerful simulation tools to assess outcomes of various Phase 1b and 2b trial designs. This enabled us to design a clinical trial with a view to optimizing development time.
Step 3: We summarized our results and presented them to the clinical development team, and subsequently to oversight management, both of which readily accepted them.
Our simulations showed that the sponsor could eliminate an entire biomarker trial, improving development time by 6 to 9 months.