Jennifer Lang (1), Maud Beneton (2), Yannick Parmentier (2), Claire Denizot (2), Ludwig Vincent (2), Marylore Chenel (2), Kayode Ogungbenro (1) and Aleksandra Galetin (1)
(1) Centre of Applied Pharmacokinetic Research, University of Manchester, United Kingdom, (2) Institut de Recherches Internationales Servier, Suresnes, France
Introduction: Ivabradine (If currents inhibitor) is indicated in stable chronic angina and heart failure. Ivabradine and its main metabolite (S18982) are both substrates of the metabolic enzyme CYP3A4, but nonlinear pharmacokinetics (PK) of S18982 was attributed to the efflux transporter P-gp in the intestine. The interplay between P-gp and CYP3A4 is often cited in the literature but in silico characterisation of this complex process is still not unequivocally successful. Combination of physiologically-based-pharmacokinetic (PBPK) and population PK modelling enables to take advantage of both top-down and bottom-up [1]. Underlying physiological structure of PBPK modelling allows investigation of the complex interplay between active transport and metabolism occurring in the gastrointestinal tract whereas the population approach provides more robustness and statistical power to the PBPK model development [2].
Objectives: The main objective is to model the interplay between P-gp and CYP3A4 using the example of ivabradine. To this end, PK after intravenous (IV) administration was described using a PBPK model and drug-specific parameter estimates were refined by means of a Bayesian method. Subsequently, the whole-body PBPK (WB-PBPK) model complexity was reduced by using the lumping method proposed by Dokoumetzidis and Aarons [3]. Finally, model prediction of oral drug absorption was evaluated against clinical data from Phase-I studies and drug-drug interaction studies with CYP3A4 and/or P-gp inhibitors.
Methods: A WB-PBPK model (n=14 organ tissues [4]) was used to describe ivabradine disposition in Phase-I clinical data in healthy male volunteers [5]. Drug-specific parameters were either measured in in vitro systems (e.g. permeability in Caco-2 cells, CYP3A4 metabolism in human liver microsomes) or predicted (i.e. blood-to-tissue partition coefficients by Rodgers and Rowland’s method [6]) and used as informative priors in statistical analyses. A global sensitivity analysis was performed to identify drug-specific parameters that significantly influence prediction of systemic drug exposure. The Bayesian estimation method using MCMC was carried out in NONMEM v7.4. Parameter posterior distributions were updated using a Gibbs sampling method and convergence of parameter posterior distributions was verified by using the Gelman-Rubin test. Using the lumping method, reduction of the WB-PBPK model was investigated in order to facilitate further analyses (e.g. parent-metabolite relationship, oral absorption…). The disposition model was then linked to a gut model to account for drug absorption and model predictive performance was assessed using four clinical studies (following ivabradine dosing alone or co-administered with ketoconazole) [7].
Results: The sensitivity analysis identified importance of the blood-to-tissue partition coefficient for muscle (Kbmuscle) and the hepatic intrinsic clearance (CLinthep) on systemic drug exposure for both parent and metabolite and were thus updated using the Bayesian method. The MCMC chains were well sampled and parameter posterior distributions reached convergence. Therefore the population-PBPK model described successfully ivabradine PK following IV administration. Model reduction from 14 to 8 compartments demonstrated that lumping allowed similar model performance and did not affect plasma prediction of drug disposition. A peripheral compartment comprised of skin, adipose tissue, bones and rest of body and a splanchnic compartment including spleen, pancreas, stomach wall and intestinal serosa were redefined. The reduced PBPK model linked to a gut model adequately described ivabradine oral absorption, metabolite formation and the magnitude of intestinal and hepatic CYP3A4 inhibition by ketoconazole.
Conclusions: Combining physiological structure of PBPK models with parameter refinement using clinical data allows the investigation of complex physiological mechanisms for which in vitro or clinical data might be sparse. This integrated population-PBPK method was satisfactorily applied to ivabradine in healthy volunteers and is useful for predicting drug exposure or drug-drug interaction risk in different populations (e.g. the elderly or children).
References:
[1] N. Tsamandouras, A. Rostami-Hodjegan, L. Aarons, Br J Clin Pharmacol 79 (2015) 48–55.
[2] T. Wendling, S. Dumitras, K. Ogungbenro, L. Aarons, J Pharmacokinet Pharmacodyn. 42 (2015) 639–657.
[3] A. Dokoumetzidis, L. Aarons, IET Syst Biol. 3 (2009) 40–51.
[4] M. Gertz, J.B. Houston, A. Galetin, Drug Metab Dispos 39 (2011) 1633–1642.
[5] I. Ragueneau, C. Laveille, R. Jochemsen, G. Resplandy, C. Funck-Brentano, P. Jaillon, Clin. Pharmacol. Ther. 64 (1998) 192–203.
[6] T. Rodgers, M. Rowland, J Pharm Sci 95 (2006) 1238–1257.
[7] N. Perdaems, H. Blasco, C. Vinson, M. Chenel, S. Whalley, F. Cazade, F. Bouzom, Clin Pharmacokinet 49 (2010) 239–258.
Reference: PAGE 28 (2019) Abstr 8946 [www.page-meeting.org/?abstract=8946]
Poster: Drug/Disease Modelling - Absorption & PBPK