Pharmacokinetics of the CYP3A4 and CYP2B6 Inducer Carbamazepine and Its Drug-Drug Interaction Potential: A Physiologically Based Pharmacokinetic Modeling Approach

Abstract

The anticonvulsant carbamazepine is frequently used in the long-term therapy of epilepsy and is a known substrate and inducer of cytochrome P450 (CYP) 3A4 and CYP2B6. Carbamazepine induces the metabolism of various drugs (including its own); on the other hand, its metabolism can be affected by various CYP inhibitors and inducers. The aim of this work was to develop a physiologically based pharmacokinetic (PBPK) parent-metabolite model of carbamazepine and its metabolite carbamazepine-10,11-epoxide, including carbamazepine autoinduction, to be applied for drug-drug interaction (DDI) prediction. The model was developed in PK-Sim, using a total of 92 plasma concentration-time profiles (dosing range 50-800 mg), as well as fractions excreted unchanged in urine measurements. The carbamazepine model applies metabolism by CYP3A4 and CYP2C8 to produce carbamazepine-10,11-epoxide, metabolism by CYP2B6 and UDP-glucuronosyltransferase (UGT) 2B7 and glomerular filtration. The carbamazepine-10,11-epoxide model applies metabolism by epoxide hydroxylase 1 (EPHX1) and glomerular filtration. Good DDI performance was demonstrated by the prediction of carbamazepine DDIs with alprazolam, bupropion, erythromycin, efavirenz and simvastatin, where 14/15 DDI AUC_last ratios and 11/15 DDI C_max ratios were within the prediction success limits proposed by Guest et al. The thoroughly evaluated model will be freely available in the Open Systems Pharmacology model repository.

Keywords: carbamazepine; carbamazepine-10,11-epoxide; cytochrome P450 2B6 (CYP2B6); cytochrome P450 3A4 (CYP3A4); drug–drug interactions (DDIs); induction; physiologically based pharmacokinetic (PBPK) modeling.

Figure 1

Schematic representation of the PBPK modeling workflow. (a) PBPK modeling requires system- and drug-dependent parameters, describing the anatomical and physiological characteristics of the individual and the properties of the simulated compound, respectively. Information on the study protocol of the described clinical study is relevant as well, e.g., formulation and administration of the simulated compound. (b) The PBPK model consists of multiple compartments, representing organs of the body, which are connected via the arterial and venous blood flows. (c) The final structure of the carbamazepine parent−metabolite PBPK model. (d) Overview of the modeled DDIs. Drawings by Servier, licensed under CC BY 3.0 [34]. CYP: cytochrome P450, EPHX1: epoxide hydroxylase 1, K_d: dissociation constant, K_m: Michaelis−Menten constant, pK_a: acid dissociation constant, UGT: UDP-glucuronosyltransferase, V_max: maximum velocity.

Figure 2

Model predictions of carbamazepine (dark blue: plasma, light blue: saliva) and carbamazepine-10,11-epoxide (green) concentration−time profiles of exemplary studies after (a–e) single- and (f–i) multiple-dose administration of different carbamazepine formulations [70,71,72,73,74,75,76,77,78] in comparison to observed data. Observed data are shown as dots ± SD (if available), simulations are shown as solid lines. Detailed information about the study protocols and model simulations of all clinical studies used to evaluate the carbamazepine model performance are provided in the Supplementary Materials. bid: twice daily, cap: capsule, D: day, n: number of subjects, qd: once daily, sd: single dose, sol: solution, susp: suspension, tab: tablet, tab*: tablet with concomitant food intake, te: test dataset, tr: training dataset, XR: extended release.

Figure 3

Performance of the carbamazepine parent−metabolite PBPK model. Predicted compared to observed (a) AUC_last values and (b) C_max values of carbamazepine and carbamazepine-10,11-epoxide of all analyzed studies. The line of identity is shown as solid line; 1.25-fold deviation is shown as dotted lines; 2-fold deviation is shown as dashed lines. AUC_last: area under the plasma concentration−time curve from dosing to the last concentration measurement, C_max: maximum plasma concentration, te: test dataset, tr: training dataset.

Figure 4

Victim drug plasma concentration−time profiles of the modeled drug–drug interactions with carbamazepine as victim drug (first row: carbamazepine, second row: metabolite carbamazepine-10,11-epoxide). Predictions of the victim drug plasma concentrations during the erythromycin−carbamazepine DDI (a) without and (b) with carbamazepine pretreatment [79,80] and (c) the efavirenz−carbamazepine DDI [8] are shown in comparison to observed data. Observed data are shown as dots ± SD (if available); predictions are shown as solid lines. Details on the study protocols and model simulations of all investigated DDI studies are provided in the Supplementary Materials. md: multiple dose, n: number of individuals, sd: single dose, tab: tablet, tab*: tablet with concomitant food intake.

Figure 5

Victim drug plasma concentration−time profiles of the modeled drug–drug interactions with carbamazepine as perpetrator drug. Predictions of the victim drug plasma concentrations during the (a) carbamazepine−alprazolam DDI [81], (b) carbamazepine−simvastatin DDI [7] (c) carbamazepine−bupropion DDI [8] and (d) carbamazepine−efavirenz DDI [10] are shown in comparison to observed data. Observed data are shown as dots ± SD (if available); predictions are shown as solid lines. Details on the study protocols and model simulations of all investigated DDI studies are provided in the Supplementary Materials. md: multiple dose, n: number of individuals, sd: single dose, tab: tablet.

Figure 6

DDI performance of the carbamazepine parent−metabolite PBPK model. Predicted compared to observed (a) DDI AUC_last ratios and (b) DDI C_max ratios of all analyzed DDI studies. Dots represent the victim drug; triangles, diamonds and squares of the same color represent respective metabolites. The line of identity is shown as a straight solid line; the curved solid lines mark the prediction success limits proposed by Guest et al. [33]. A 1.25-fold deviation is shown as dotted lines; 2-fold deviation is shown as dashed lines. Details on the study protocols and all individual DDI AUC_last and DDI C_max ratios are provided in the Supplementary Materials. AUC_last: area under the plasma concentration−time curve from dosing to the last concentration measurement, C_max: maximum plasma concentration, DDI: drug–drug interaction, m: number of studies.