Pavan Vajjah, Trevor N Johnson, Masoud Jamei, Sibylle Neuhoff, Amin Rostami-Hodjegan
Simcyp Ltd, John Street, Sheffield, S2 4SU. UK
Objective: The objective of the work was to predict the cerebrospinal fluid (CSF) concentration of carbamazepine (CBZ) in humans using the permeability-limited 4-compartment brain model in the Simcyp® Simulator V11 and the default library values for populations related values.
Methods: A total of 108 steady state serum CBZ and carbamazepine-10, 11- epoxide (CBE) and 12 single point pre-dose CSF concentrations from 14 male individuals (age 14 – 44 years, dose range 600-2000mg daily in 2-4 divided doses) were extracted from a literature report [1]. In the first stage of the modelling, the expectation maximisation algorithm within the Simcyp parameter estimation (PE) module was used to optimise the CBZ and CBE parameters using dose normalised concentrations of CBZ and CBE (a Top-Down approach). Subsequently, in the second stage, the retrograde model was used for parent drug to calculate intrinsic clearance values from oral clearance (CLpo) for CBZ metabolism by CYPs 3A4, 3A5 and 2C8, the CLpo for the metabolite was used directly as an in vivo input into the Simcyp Simulator (a combination of the Top-Down and Bottom-Up approaches). In the third stage, the model and fitted parameters were used to predict the concentration of CBZ in CSF using the permeability-limited 4-compartment brain model in the Simcyp Simulator assuming passive diffusion of CBZ into brain [2] (a Bottom-Up approach). The simulated CSF concentrations were compared with the observed values.
Results: A full PBPK model with first order absorption provided the best fit to the CBZ serum data. The metabolite data was well described by a minimal PBPK model. A combined error model was the best residual error model both parent and metabolite data. The final model was evaluated using a visual predictive check. The estimated CLpo of carbamazepine was (mean (CV%)) 6.39 (56) L/h. The estimated ka was 0.09 (63) h-1. The estimated CLpo for the metabolite was 22.27 (50) L/h. An initial analysis, in agreement with previously published experimental data [3], shows that active transporters may not influence the transfer of CBZ into CSF. The observed CSF concentrations of CBZ (in literature) fell within the 95 percentile interval of the model predictions, all predicted values were within 2.1-fold of observed values.
Conclusions: The model successfully predicted the concentration of CBZ in CSF using the three stage modelling process that combined the To-Down and Bottom-Up approaches. An application of the permeability-limited 4-compartment brain model within the Simcyp Simulator was demonstrated.
References:
[1] Johannessen SI et al., British Journal of Clinical Pharmacology, 1976, 3, 575-582
[2] Gaohau L et al., PAGE 20 (2011), Abstr 2078 [www.page-meeting.org/?abstract=2078]
[3] Owen A et al., British Journal of Clinical Pharmacology, 2001, 51, 345-349
Reference: PAGE 21 (2012) Abstr 2464 [www.page-meeting.org/?abstract=2464]
Poster: Absorption and Physiology-Based PK