C. Meille (1), S. Benay (2), S. Kustermann (3), I. Walter (4), E. Pietilae (3), P-A. Gonsard (4), N. Kratochwil (4), A. Walz (1), A. Roth (3) and T. Lavé (1)
F. Hoffmann-La Roche Ltd, Non-Clinical Safety, (1) Modeling and Simulation, (3) Mechanistic Safety, (4) ADME, Basel, Switzerland; (2) Dpt. of Pharmacokinetics, INSERM U911 CRO2, Aix-Marseille University, Marseille, France
Objectives: Signal distribution models have been developed to describe the time dependent effects of cancer drugs in vitro [1] and in vivo [2]. Real-Time Cell Analysis (RTCA) systems allow continuous in vitro monitoring of drug effect on cancer cell count [3]. Our objective was to explore benefit of such data for refinement of signal distribution models by describing the effect of Erlotinib on A431 epidermoid human carcinoma cell line in an impedance-based RTCA assay.
Methods: At 0, 4, 8 and 16 µM Erlotinib concentrations, the cellular effect was evaluated through cell index time course of xCELLigence system [4]. In addition, in vitro Erlotinib concentration was determined at 3 time points during the incubation using LC-MS/MS. The obtained in vitro PK data were described by a single-compartment model with linear elimination and unspecific binding. An exponential model was selected to describe control cell growth. A nonlinear model associated concentration to cell kill rate signal resulting in cytostatic or cytotoxic effect. As described in the signal distribution models, the delay between Erlotinib in vitro PK and effective cell kill was reproduced by 4 transit compartments with a respective mean transit time tau. The main PK-PD model parameters were estimated using population approach with Monolix 4.1.3 software [5] by considering the well as the statistical unit.
Results: At 16 µM, a decrease of cell index was observed followed by a regrowth after 130h. A significant decrease in drug concentration was observed and the cell regrowth could be associated to this phenomenon by the PK/PD model. The PK model determined an in vitro half-life of 264 h with an inter-well variability of 7% and an unspecific binding of 34%. Drug concentrations ranging from 3 to 8 µM achieved killing rate close to the growth rate of A431 cells (0.033 h-1). Beyond 9 µM, killing rate exceeded the growth rate and reached 0.05 h-1. The mean transit time tau decreased with increasing concentrations, ranging between 2 and 17 h.
Conclusion: New insights were gained by combining modeling with rich dynamic in vitro data. It could be shown that the drug concentration determines the cytostatic or cytotoxic effect and influences the delay in the cell kill rate signal. This model based approach can help to build and test hypotheses of drug action as well as quantifying drug effect kinetics at a cellular level.
References:
[1] Lobo ED and Balthasar JP Pharmacodynamic modeling of chemotherapeutic effects: application of a transit compartment model to characterize methotrexate effects in vitro. AAPS PharmSci., 4(4), 2002.
[2] Yang J, et Al. Comparison of two pharmacodynamic transduction models for the analysis of tumor therapeutic responses in model systems. AAPS J. 2010 Mar;12(1):1-10. Epub 2009 Nov 10.
[3] Kustermann S, et Al. A label-free, impedance-based real time assay to identify drug-induced toxicities and differentiate cytostatic from cytotoxic effects. Toxicology in Vitro, [Epub ahead of print], 2012.
[4] http://www.aceabio.com
[5] Kuhn E., Lavielle M. Maximum likelihood estimation in nonlinear mixed effects models Computational Statistics and Data Analysis, 49(4), 2005
Reference: PAGE 22 () Abstr 2861 [www.page-meeting.org/?abstract=2861]
Poster: Oncology