2010 - Berlin - Germany

PAGE 2010: Methodology- PBPK
Jörg Lippert

Separating individual physiological variability from drug related properties using PBPK Modeling with PK-Sim® and MoBi® – Theophylline

Michael Block, Linus Görlitz, Clara Happ, Rolf Burghaus, Jörg Lippert

[1] Systems Biology and Computational Solutions, Bayer Technology Services GmbH, 51368 Leverkusen, Germany; [2] Clinical Pharmacokinetics Modeling & Simulation, Bayer Schering Pharma AG, Wuppertal, Germany

Objectives: Physiologically-based pharmacokinetic and pharmacodynamic (PBPK/PD) modeling and classical compartimental methods using NLME or Bayesian Markov Chain Monte Carlo simulation have been considered as complementary approaches to PKPD. But only Bayesian PBPK/PD as a completely knowledge-based approach allows a systematic identification of highly valuable drug-independent information about individual physiological processes. In using this approach every experiment will lead to an improvement of understanding of substance-specific behavior and its interaction with individual physiological processes.

Methods: The classical theophylline example [1] was used to develop and evaluate the integrated statistified PBPK/PD approach. All modeling was done in the systems biology software platform consisting of PK-Sim® and MoBi®. Prior knowledge about anatomical and physiological structure, population parameters and their variability was automatically loaded from PK-Sim®’s built-in database [2]. Global, compound specific parameters, e.g. lipophilicity, and individual physiological parameters like intestinal transit time were modeled. We implemented Metropolis Sampling [3] in our platform which allows sampling nonstandard posterior distributions for all parameters of interest.

Results: The use of physiologically proper prior information allows fitting of high-dimensional PBPK-models. The statistified PBPK gives a clear identification of aimed parameter values and automatically identifies the subset of relevant parameters. The resulting pharmacokinetic curves show a superior fit compared to [4].

Conclusion: PK-Sim® and MoBi® greatly facilitate combined PBPK/PD modeling and statistical analysis by statistical methods. In contrast to compartmental models where physiological properties and drug properties are represented by joint parameters and intra- and inter-individual variation is only indirectly represented by substance-specific parameters and variabilities, Bayesian PBPK/PD directly deconvolutes variability of physiological processes. This allows interpretability of results and generates knowledge about individual physiological processes that can be applied even in the context of further substances and will lead to advancements in interpretation and prediction of drug trials.

[1] Davidian M, Giltinan DM. (1995) Nonlinear Models for Repeated Measurement Data, Chapman & Hall
[2] Willmann S, Lippert J, Sevestre M, Solodenko J, Fois F, Schmitt W. (2003) PK-Sim: A physiologically based pharmacokinetic ‘whole-body’ model. Biosilico. 1(4):121-24
[3] Hastings WK. (1970) Monte Carlo Sampling Methods Using Markov Chains and Their Applications. Biometrika. 57(1):97-109
[4] Tornøe C, Agersø H, Niclas Jonsson E, Madsen H, Nielsen H. (2004) Non-linear mixed-effects pharmacokinetic/pharmacodynamic modelling in NLME using differential equations. Computer Methods and Programs in Biomedicine. 76:31-40

Reference: PAGE 19 (2010) Abstr 1860 [www.page-meeting.org/?abstract=1860]
Poster: Methodology- PBPK