A model-based evaluation of the effect of bosentan on bile salt disposition in sandwich-cultured human hepatocytes


Bosentan is an endothelin receptor antagonist used to treat patients with pulmonary arterial hypertension and is known to cause cholestatic liver toxicity. Bosentan is an inhibitor of the bile salt export pump (BSEP), the major efflux transporter for bile salts, but other mechanisms may also be involved. However, often high bosentan concentrations are used [1,2]. The aim of this study was to investigate the effect of therapeutically relevant bosentan concentrations on chenodeoxycholic acid (CDCA) and glycochenodeoxycholic acid (GCDCA) disposition in sandwich-cultured human hepatocytes (SCHH).


SCHH were incubated with 250 µL (Vmed) of a 10 µM CDCA solution in the presence and absence of 2.5 µM, 7.5 µM and 25 µM bosentan (n3 replicates per condition). The concentrations of CDCA and its primary metabolite, GCDCA, were determined over 24 h in each compartment (medium, cell, bile; Figure 1) with liquid chromatography coupled to tandem mass spectrometry (LC-MS/MS). This data was used to develop a model describing disposition of CDCA and GCDCA in the in vitro system.

The data was modeled in NONMEM v7.4 using first-order conditional estimation with interaction (FOCE-I) and differential equation solver ADVAN13. The mechanistic structural model was fixed based on the experimental design (Figure 1). Vcell was fixed to 1.37 µL based on literature. The parameter kflux represents the periodical contraction of the canaliculi [3]. Inter-condition variability was fixed to 0. All variability was included as residual variability. Additive and/or proportional error models were evaluated. The effect of bosentan was investigated as a categorical covariate on the following parameters: CLbile,GCDCA, CLeff,GCDCA and CLmet,GCDCA. The amount of GCDCA in bile (compartment 4; Figure 1) was calculated based on two independent measurements. Therefore, multiple imputation was performed using 2,000 out of 5.9×1071 possible matches to investigate robustness of the parameter estimations. In addition, the number of successfully minimized runs was considered to select the best out of three error models. Median parameter estimates were used to perform 10,000 simulations per bosentan condition.


The model contained saturable uptake kinetics and linear metabolic clearance for CDCA. All GCDCA related clearances were linear. Covariate testing only indicated a statistically significant effect of bosentan on CLbile,GCDCA (dOFV of covariate relative to base model is -33.0). The best model minimized successfully in 1535 out of 2000 (77%) imputation scenarios. Good agreement was observed between the model simulations and observed data for both CDCA and GCDCA in all compartments. They clearly showed decreased exposure of GCDCA in the bile (Figure 1). An overview of fixed effects parameters (median and 95% confidence interval [CI]) is depicted in Figure 1. The parameter estimates for different bosentan concentrations show a clear effect of bosentan on CLbile,GCDCA.


Our nonlinear mixed effects model confirms the inhibitory effect of bosentan on biliary clearance of GCDCA. This is an indication that BSEP inhibition is the major responsible for the altered bile salt levels.


[1]       Mano, Usui, Kamimura, Biopharm. Drug Dispos. 2007, 28, 13.

[2]       Fouassier, Kinnman, Lefèvre, Lasnier, Rey, Poupon, Elferink, Housset, J. Hepatol. 2002, 37, 184.

[3]       Guo, Yang, Brouwer, St Claire, Brouwer, J. Pharmacol. Exp. Ther. 2016, 358, 324.