Maïlys De Sousa Mendes (1), Sibylle Neuhoff (1), Howard Burt(1)
(1) Simcyp UK Limited (a Certara company), Sheffield, UK
Objectives: Bidirectional transport assays can be used to obtain in vitro permeability estimates for use in PBPK models. However, the conventional analysis of these assays makes several assumptions that can impact the quality of the in-vivo prediction. For example, it assumes that sink conditions are maintained, which can be difficult to achieve experimentally, especially for highly permeable compounds. It also assumes that the unstirred water layer (UWL) doesn’t have significant impact on the observed permeability and the impact of compound ionisation is not usually studied. We to developed a model that mechanistically describes the in-vitro permeability across Caco-2 cells for metoprolol, a highly permeable drug. The impacts of ionisation and UWL on the permeability were investigated.
Methods:
In-vitro assay:
Data for the bidirectional transport of metoprolol across Caco-2 monolayers were previously generated [1]. Briefly, Caco-2 cells were seeded at a density of 1x 105 cells/well onto 12-ellTranswell® inserts and grown for 23 ±1 days prior to permeability experiments. Experiments were performed at 37°C, with apical and basolateral volumes of 0.5 and 1.5mL, respectively, and was stirred at 450 rpm (calibrated plate shaker (BMG LabTechnologies GmbH, Offenburg, Germany). The basolateral compartment was buffered to a pH of 7.;, whereas a range of buffer pH values was investigate in the apical compartment (pH 5.0, 5.5, 6.0, 6.5, 7, 7.4, 7.7 and 8.0)
Data analysis (modelling):
A mechanistic model was developed in R software (version 3.3.1) and included 5 compartments, representing apical and basolateral bulk media and unstirred water layers in addition to the cell monolayer. The fraction ionised was calculated in each compartment and for each experiment based on the drug pKa and media pH values. The total unstirred water layer thickness was predicted on the basis of stirring rate using data from Adson et al. [2]. The permeability of the ionised form ( was calculated using the permeability of the neutral form and an ionisation scalar describing the log decrease in permeability for cationic metoprolol compared to the neutral species.
Simulations:
The permeability estimates obtained were implemented in the metoprolol compound file in the Simcyp Simulator v17. The Mechanistic passive regional permeability predictor (MechPeff) model was used to predict the effective permeability observed in human (Peff,man). This Peff,man was used to predict the absorption using a first-order model and the Advanced Dissolution, Absorption and Metabolism (ADAM) model. Plasma concentration-time profiles of metoprolol after a single oral dose of 100 mg in CYP2D6 extensive metabolisers were simulated for 10 trials of 16 female subjects 18 – 40 years and compared to observed data from Sharma et al. [3].
Results:
The in vitro model was able to describe the decrease in metoprolol permeability with an increase in ionisation. The difference between the observed and predicted in-vitro concentrations was less 2-fold. The geometric mean fold error (GMFE) was 1.27 and the geometric fold bias (GMFB) was 1.02. The Ptrans,0 estimate of 40000 10-6 cm/s and ionisation scalar of 3.4 predicted an Peff,man in the jejunum I of 2.95 10-4 cm/s when applied in the metoprolol PBPK model. The mean (± SD) clinically observed in-vivo Cmax, tmax and AUCwere 0.89 ± 0.42 µmol/l, 1.69 ± 0.63 h and 4.05 ± 2.15 µmol.h/l respectively. With a first-order absorption model the mean (clinical range) predicted Cmax, tmax and AUC were 0.75 (0.64-0.85) µmol/l, 1.59 (1.48-1.66) h, 4.26 (3.37, 5.16) µmol.h/l, repectively. With the ADAM model the mean (clinical range) predicted Cmax, tmax and AUC were 0.62 (0.53-0.71) µmol/l, 2.21 (2.01-2.33) h, 4.18 (3.32, 5) µmol.h/l, repectively.
Conclusions:
The mechanistic model was able to account for the impact of metoprolol ionisation on its passive permeability in vitro. When in vitro permeability estimates were applied in the metoprolol PBPK model, the predicted in-vivo absorption was in accordance with clinical data, indicating that this approach could be used to generate robust inputs for PBPK models.
References:
[1]. Neuhoff S, Ungell A-L, Zamora I, Artursson P. Pharm Res. 2003 Aug;20(8):1141–8.
[2]. Adson A, Burton PS, Raub TJ, Barsuhn CL, Audus KL, Ho NF. J Pharm Sci. 1995 Oct;84(10):1197–204.
[3]. Sharma A, Pibarot P, Pilote S, Dumesnil JG, Arsenault M, Bélanger PM, Meibohm B, Hamelin BA.. J Pharmacol Exp Ther. 2005 Jun 1;313(3):1172–81.
Reference: PAGE 27 (2018) Abstr 8701 [www.page-meeting.org/?abstract=8701]
Poster: Drug/Disease Modelling - Absorption & PBPK