Using Bayesian-PBPK modeling for assessment of inter-individual variability and subgroup stratification

Area of Science:

• Pharmacology
• Biostatistics
• Computational Biology

Background:

• Inter-individual variability in drug response and adverse effects poses challenges in drug development.
• Early identification of adverse events and patient subgroups is crucial for safety.
• Understanding pharmacokinetics and pharmacodynamics is essential for drug efficacy.

Purpose of the Study:

• To combine Bayesian statistics with physiologically-based pharmacokinetic (PBPK) models to investigate inter-individual variability.
• To demonstrate the utility of this approach in identifying clinically relevant patient subgroups.
• To support knowledge-based extrapolation for other drugs and populations.

Main Methods:

• Development of a Bayesian-PBPK approach to quantify parameter variability within patient populations.
• Utilizing Markov chain Monte Carlo algorithms for high-dimensional parameter distributions.
• Mechanistic investigation of drug distribution and action using PBPK models.

Main Results:

• Application of Bayesian-PBPK to pravastatin pharmacokinetics in a cohort of 10 patients.
• Identification of homogeneous patient subpopulations based on parameter distributions.
• Correlation analyses provided structural information for the PBPK model.

Conclusions:

• The Bayesian-PBPK approach systematically characterizes inter-individual variability by updating prior knowledge.
• Clinically relevant homogeneous subpopulations can be mechanistically identified.
• The model facilitates iterative assessment of populations by integrating information across drugs.