Leveraging Two-Stage δ Global Sensibility Analysis Method to Inform Parameter Estimation in PBPK Models

Area of Science:

• Pharmacokinetics and Pharmacometrics
• Computational Biology and Bioinformatics
• Systems Biology

Background:

• Physiologically Based Pharmacokinetic (PBPK) modeling is crucial for drug development but often lacks robust methods for sensitivity analysis and parameter estimation.
• Current practices inadequately address Global Sensitivity Analysis (GSA) and individual parameter refinement, limiting PBPK model accuracy and applicability.

Purpose of the Study:

• To establish a computational framework integrating PhysPK and Python for advanced PBPK analysis.
• To implement Two-stage delta GSA and iterative two-stage (ITS) methods for semi-mechanistic PBPK models.
• To assess the impact of parameter uncertainty and correlations on key pharmacokinetic endpoints.

Main Methods:

• Developed a framework linking PhysPK with Python for GSA and parameter estimation.
• Applied Two-stage delta GSA to identify influential parameters affecting AUC, Cmax, and Tmax.
• Utilized the ITS method for individual parameter estimation across various simulated scenarios and optimization algorithms (Nelder-Mead, Powell, BFGS).

Main Results:

• Two-stage delta GSA identified volume of distribution, clearance, and gastric emptying rate as key parameters.
• Parameter estimation performance was evaluated using AFE, AAFE, and PEE metrics.
• Most estimations achieved AFE and AAFE values between 0.8 and 1.25, with Nelder-Mead showing superior accuracy.

Conclusions:

• The integrated framework successfully combines correlation-aware GSA with individual parameter estimation in PBPK models.
• This approach enhances PBPK model simplification and supports data-constrained individual parameter estimation.
• The integration represents a significant advancement for the PhysPK platform, offering a powerful tool for pharmacokinetic modeling.