Population physiologically-based pharmacokinetic (PBPK) models are powerful tools in both drug development and regulatory science. However, the adoption of PBPK models has been hampered by their complexity, uncertainties in pharmacokinetic (PK) parameters, difficulties in optimizing to individual experimental data, and lack of an integrated software platform. There is a critical need to develop a comprehensive and robust PBPK model-based approach to population PK analysis. Without such an approach, it will remain a challenge to integrate information from human physiology, drug properties and interactions, and drug formulation characteristics to make predictions and ensure bioequivalence for the large and diverse patient populations relevant to most generic drugs. The long-term goal of this research program is to harness the predictive power of population-based PBPK models to make personalized PK predictions to ensure both the efficacy and safety of drugs across diverse patient populations. The overall objective here, which is the next step in pursuit of that goal, is to develop a robust and efficient algorithm for analyzing population PK data using PBPK models. The central hypothesis is that a hierarchical Bayesian statistical approach using Markov Chain Monte Carlo, when enhanced by efficiency improvements through parallelization and algorithmic changes, global sensitivity analysis, and user interface-focused features, will substantially improve both the end user experience and the accuracy, speed, and robustness of resulting model predictions. These enhancements, while previously investigated individually, have never been combined in a population PK context. The platform on which these enhancements will be made is MCSim, an open source software application ideally-suited for population PBPK analyses. We anticipate that the proposed research will result in wider application of population PBPK models in drug development and regulatory reviews, resulting in enhanced drug efficacy and safety in the population.
Population physiologically-based pharmacokinetic (pop-PBPK) models are powerful tools that allow scientists to better understand the variability in drug concentrations among individuals in a population. The aim of this study is to incorporate state-of-the-art modeling and statistical algorithms into an integrated software platform for pop-PBPK analysis, so as to improve the reliability, efficiency, and usability of these models and their resulting predictions. We expect that this platform will lead to more widespread adoption of the pop-PBPK approach and provide scientists with an effective tool to evaluate and improve the efficacy and safety of drugs.