Mechanism-based Pharmacokinetic Models

Martin Bergstrand, Mats Karlsson

Clinical pharmacokinetic experiments typically measure drug concentrations in plasma only. As a consequence, pharmacokinetic models, used in drug development, aim to describe observations of drug concentration in plasma with minimum model complexity. Such models have limited capacity for extrapolations and to predict concentration-time profiles in tissues and organs. Also, mechanistic insight about drug disposition dependence of factors related to individual organs and tissues may not be possible to incorporate in a fully satisfactory manner. Physiologically-based pharmacokinetic (PBPK) models, which have a structure based on anatomy, can provide predictions in tissues and organs. However, because of their complexity, such models are traditionally not used for analysing clinical data. We have demonstrated that PBPK model parameters can be estimated based on clinical observations and are currently investigating the possibilities for improving such a combined “bottom-up and top down approach”. We have showed that for a relatively simple PBPK model such analyses can become feasible by using informative prior information about physiology and drug-related parameters. To further improve such an approach we are combining information about (co‑)variability in organ and tissue properties from a data base representing physiological values for about 30000 subjects, tissue composition models and models for relating drug molecular properties and in vitro data to expected behaviour in tissues and organs.

A mechanism-based approach has also been applied to better understand the processes involved in oral absorption. Several factors influencing oral absorption vary along the GI tract, e.g. pH, active influx and efflux transporters, gut wall metabolism. Models characterizing gastro intestinal transit and absorption properties along the GI region can be especially useful in the case of modified release formulations and/or substances with slow dissolution rate. Mechanism based models have been used for prospective population predictions of plasma concentrations based on in vitro dissolution data.