Prediction of pharmacokinetic behaviour by combining in vivo and in vitro data in physiologically based pharmacokinetic (PBPK) model: Parameter estimation and sensitivity analysis

Abstract

A large number of failures of potential drug molecules in the late stages of the drug discovery process have been ascribed to poor in vivo pharmacokinetic properties. Physiologically based pharmacokinetic (PBPK) models can potentially be used to predict the in vivo drug disposition from the in vitro characterization of the drug molecules. One potential drawback of the existing models is the large number of parameters and the uncertainty associated with the parameter values. We present a framework for the estimation of kinetic parameters and sensitivity analysis for PBPK model. The existing model was enhanced by incorporating the mechanistic knowledge of drug disposition involving transporter proteins and metabolic enzymes. Some of the drug-specific model parameters have been estimated from the in vitro and in vivo data available in literature and rest are estimated by fitting the in vivo data and in vitro data to the model to minimize the error between model predictions and experimental values. The model predictions for tissue specific concentration profiles have been shown to agree with experimental data for rats. Further, the model predictions for plasma drug disposition in humans or animals for the five exemplary drugs agree well with the in vivo clinical trial data from literature. The selected drugs are of two different categories; cardiovascular system and diabetes. By using global sensitivity analysis we find that the parameters associated with liver, kidney, and renal excretion have the highest effect on the in vivo drug disposition. These analyses and predictions will help in making an early selection of compounds for development based on pharmacokinetic properties as well as for advancing personalized medicine which, in turn will improve therapy for specific subpopulations.