🤖 AI Summary
This study addresses the lack of systematic guidance on the influence of prior hyperparameters in Bayesian Go/No-Go decision-making for early-phase clinical trials with dual endpoints. The authors propose a calibrated bivariate prior specification framework that constructs skeptical and optimistic priors by assigning target probabilities to predefined decision regions, requiring only the selection of a central location and a precision parameter κ. Theoretically, for any κ > 0, there exists a unique scaling parameter λ₀ that achieves calibration, and κ governs the prior’s discriminative power, supporting adaptive κ strategies. Under a normal–inverse-Wishart model (with default ν₀ = 2), simulations show that at κ = 10, the Go-rate difference between optimistic and skeptical priors reaches 0.56 while maintaining a false-positive rate below 0.01. In a real lupus dataset, optimistic priors with high κ yield Go rates up to three times those of skeptical priors, even with small sample sizes.
📝 Abstract
Bayesian Go/No-Go decisions with co-primary endpoints require specifying prior distributions under the Normal-Inverse-Wishart framework; however guidance on how prior hyperparameters influence trial decisions remains limited. We propose a calibrated prior specification framework for bivariate Go/No-Go decisions. Skeptical and enthusiastic priors are calibrated so that each assigns a target probability to a clinically relevant decision region. We prove that for any prior precision $κ> 0$, a unique scale parameter $λ_0$ achieves the target calibration. Operating characteristics are evaluated across different $κ$ via simulation and applied to a phase~3 telitacicept lupus trial.The simulation result indicates $κ$ is the primary driver of prior discrimination. At $κ= 1$, the go rate difference between priors was 0.07; at $κ= 10$ it reached 0.56, with false positive rates below 0.01. Operating characteristics were robust to the degrees of freedom parameter $ν_0$ and prior correlation $ρ_0$, supporting a default of $ν_0 = 2$. In the lupus application, prior sensitivity was negligible at $κ= 1$ but at $κ= 10$ the enthusiastic go rate was three times the skeptical rate at small sample sizes. The framework reduces prior specification to two choices: the prior center and the prior precision $κ$. The identification of $κ$ as the dominant parameter, together with the cautious choice of $κ$ before the trial, motivates adaptive approaches to prior precision.