Design evaluation and optimisation in crossover pharmacokinetic studies analysed by nonlinear mixed effects models

Area of Science:

• Pharmacokinetics
• Statistical Modeling
• Clinical Trial Design

Background:

• Crossover designs are standard for bioequivalence and drug interaction studies.
• Nonlinear mixed effects (NLME) models offer an alternative to noncompartmental analysis for these trials.
• Designing efficient crossover trials requires robust statistical methods.

Purpose of the Study:

• To extend the population Fisher information matrix (PFIM) for designing pharmacokinetic crossover trials within the NLME framework.
• To incorporate within-subject variability and discrete covariates into the trial design.
• To provide a method for calculating statistical power and determining the required sample size for crossover studies.

Main Methods:

• Linearization of the NLME model around the random effect expectation.
• Extension of the PFIM to account for within-subject variability and covariates.
• Calculation of expected standard errors for treatment effects.
• Simulation studies to validate the proposed methodology.
• Application to a real-world pharmacokinetic study and implementation in R (PFIM v3.2).

Main Results:

• The developed method allows for precise calculation of statistical power for comparison or equivalence tests.
• The approach accurately determines the optimal number of subjects needed for a given power.
• Simulations confirm the relevance and accuracy of the proposed design extension.
• The methodology was successfully applied to design a amoxicillin pharmacokinetic study in piglets.

Conclusions:

• The extended PFIM provides a valuable tool for designing efficient and statistically sound pharmacokinetic crossover trials.
• This method enhances the planning of bioequivalence and drug interaction studies by optimizing sample size and power.
• The implementation in the PFIM R package (v3.2) makes this advanced design methodology accessible to researchers.