Statistical approaches for the integration of external controls in a cystic fibrosis clinical trial: a simulation and...

Area of Science:

• Clinical Trials
• Pharmacoepidemiology
• Biostatistics

Background:

• Drug development for rare diseases like cystic fibrosis faces challenges in patient recruitment for clinical trials.
• External controls from historical trials can mitigate the need for large prospective studies, aiding regulatory approvals for rare disease interventions.

Purpose of the Study:

• To evaluate statistical methods for incorporating external controls into clinical trials for cystic fibrosis treatments.
• To compare the performance of inverse probability weighting, Bayesian modeling with power priors, and hierarchical Bayesian modeling with commensurate priors.

Main Methods:

• The study employed simulation studies and analysis of real clinical trial data.
• Three statistical methods were assessed: inverse probability weighting, Bayesian modeling with propensity-score-based power priors, and hierarchical Bayesian modeling with commensurate priors.

Main Results:

• All evaluated methods demonstrated minimal bias (less than 4%) in treatment effect estimation.
• Type I error rates were generally below 5%.
• The commensurate prior method exhibited superior performance with real clinical trial data, while inverse probability weighting showed sensitivity to covariate prevalence differences.

Conclusions:

• Incorporating external controls offers a viable strategy to reduce clinical trial sizes for rare diseases, potentially accelerating therapeutic development.
• The commensurate prior Bayesian method is a promising approach for utilizing external controls effectively in clinical trial settings.
• This methodology can significantly advance the pipeline for new treatments targeting rare diseases.