Simultaneous confidence regions for multivariate bioequivalence.

Area of Science:

• Pharmacokinetics
• Biostatistics
• Drug Development

Background:

• Bioequivalence assessments often use univariate analyses for pharmacokinetic (PK) parameters like AUC and Cmax.
• This approach overlooks the multivariate nature of analyzing multiple PK parameters from the same samples.
• Regulatory bodies and practitioners frequently neglect simultaneous evaluation of PK measures.

Purpose of the Study:

• To review methods for constructing joint confidence regions for multivariate normal means.
• To investigate the utility of these methods in simultaneous bioequivalence problems through simulation.
• To address the limitations of univariate analyses in bioequivalence testing.

Main Methods:

• Review of statistical methods for joint confidence regions.
• Simulation studies to evaluate method performance under various scenarios.
• Analysis of PK parameter correlations and variances.
• Application to real-world bioequivalence trial data.

Main Results:

• Some multivariate methods perform well in idealized scenarios but struggle with real-world data complexities like unknown variance and parameter correlation.
• The study analyzes the geometric shapes of confidence regions produced by different methods.
• Methods for deriving marginal simultaneous confidence intervals for individual PK measures are discussed.

Conclusions:

• Simultaneous multivariate analysis is more appropriate for demonstrating bioequivalence of multiple PK parameters.
• Challenges exist in applying multivariate methods to real-world bioequivalence data.
• The findings inform more robust bioequivalence assessment strategies and an R package is available.