Mechanistic PBPK/PD modeling for prediction of study outcome of cancer therapies: Translating in-vitro information into valid in-vivo predictions
Michael Block (1)*, Katja Tummler (1), Michaela Meyer (1), J÷rg Lippert (1)
(1) Competence Center Systems Biology and Computational Solutions, Bayer Technology Services GmbH, 51368 Leverkusen, Germany
Objectives: Getting a valid quantitative assessment of clinical endpoints for a planned study based on preclinical observations is one of the most demanding tasks during the pharmaceutical research and development process. In order to obtain high quality predictions of study outcome it is of major importance to integrate as much prior knowledge and information as possible while performing physiological modeling. It was our aim to demonstrate how in the field of cancer research the use of mechanistic multiscale models representing the in-vitro and in-vivo situation can be applied for the estimation of efficiency of different tyrosine kinase inhibitors (e.g. Sunitinib, Imatinib).
Methods: We developed whole body physiologically based pharmacokinetic/pharmacodynamic (PBPK/PD) models for different tyrosine kinase inhibitors using the software PK-Sim« and MoBi«. Basically, pharmacokinetic (PK) models for the above mentioned chemotherapeutic agents were built to estimate the drug concentrations in the tumor site. These PK models were then coupled to a mechanistic pharmacodynamics tumor growth model, integrating detailed information to cell growth, cell cycle progression and apoptosis. The thus built multiscale tumor growth model was then parameterized using in-vitro data measured in cancer cell lines, in-vivo information derived from xenograft models as well as plasma concentration-time profiles from preclinical studies.
Results: The developed whole body PBPK/PD multiscale models for all tyrosine kinase inhibitors provide an excellent description of drug pharmacokinetics in xenograft models and humans and accurately predict the tumor growth under control and inhibition conditions. Using the mechanistic model to predict the outcome of a xenograft study leads to an excellent agreement of the simulated and experimentally determined Kaplan-Meier plots.
Conclusions: The multiscale model approach presented here provides a sophisticated mechanistic framework to obtain high quality predictions of study outcome by integrating as much prior information of different cell types (i.e. cancer types) and various compounds as possible. It offers the opportunity to systematically increase our knowledge on tumor growth and cancer therapy efficiency by translating all available in-vitro information into valid in-vivo predictions.