A Bayesian adaptive phase II clinical trial design accounting for spatial variation

Area of Science:

• Clinical Trials
• Biostatistics
• Spatial Epidemiology

Background:

• Conventional phase II trials assume patient homogeneity, overlooking significant inter-patient variability.
• Patient heterogeneity in treatment efficacy can stem from individual characteristics and unmeasured spatial exposures.
• Ignoring spatial variations can lead to suboptimal treatment effect evaluations in clinical trials.

Purpose of the Study:

• To propose a hierarchical Bayesian adaptive design for two-arm phase II trials that incorporates spatial variation and patient covariates.
• To improve the accuracy of treatment effect estimation by addressing inter-patient heterogeneity.
• To adaptively assign patients to treatments and provide robust treatment recommendations.

Main Methods:

• Utilized a hierarchical Bayesian adaptive design for phase II clinical trials.
• Treated treatment efficacy as an ordinal outcome with a utility function for category desirability.
• Employed a cumulative probit mixed model with conditional autoregressive priors for spatial effects.
• Implemented a two-stage design for adaptive patient assignment and treatment recommendations.

Main Results:

• The proposed design demonstrated good operating characteristics in simulation studies.
• The Bayesian adaptive design significantly outperformed traditional phase II designs that ignore spatial variation.
• The model effectively integrated patient-specific covariates and spatial effects for improved efficacy assessment.

Conclusions:

• The proposed hierarchical Bayesian adaptive design offers a superior approach to phase II clinical trials by accounting for spatial heterogeneity.
• This innovative design enhances the precision and reliability of treatment effect evaluations.
• The findings suggest a significant improvement over existing methods in identifying effective treatments across diverse patient populations and geographic regions.