Frédéric Gaspar 1,2, Jean Terrier 4,5, Youssef Dali 2,8, Pauline Gosselin 4,5, Pierre Fontana5,6, Jean-Luc Reny 4,5,7, Monia Guidi 1,2,3, Chantal Csajka 1,2
1 Center for Research and Innovation in Clinical Pharmaceutical Sciences, Lausanne University Hospital and University of Lausanne, Lausanne, Switzerland 2 Institute of Pharmaceutical Sciences of Western Switzerland, University of Geneva, University of Lausanne, Geneva & Lausanne, Switzerland 3 School of Pharmaceutical Sciences, University of Geneva, Geneva, Switzerland 4 Division of General Internal Medicine, Geneva University Hospitals, Geneva, Switzerland 5 Geneva Platelet Group, Faculty of Medicine, University of Geneva, Geneva, Switzerland 6Clinical Pharmacology and Toxicology Service, Anesthesiology, Pharmacology and Intensive Care Department, Geneva University Hospitals, Geneva, Switzerland 7 Division of Angiology and Haemostasis, Geneva University Hospitals, Geneva, Switzerland 8Service of Clinical Pharmacology, Lausanne University Hospital and University of Lausanne, Lausanne, Switzerland
Objectives: The requirement for individualized dosing of some drugs has been recognised for decades [1]. Nowadays, population pharmacokinetic (PopPK) models are widely used to support precision dosing but this approach includes the requirement for at least one time-sensitive blood sampling and specific bioassays for concentration measurements [2]. Recent advances in physiologically-based pharmacokinetic (PBPK) models may be an alternative but there is a lack of evidence in clinical practice [3]. In this paper, we evaluate the prediction of individual drug exposure with both PopPK and PBPK modeling approaches, illustrated with ticagrelor, a widely prescribed antiplatelet agent, and its active metabolite AR-C124910XX (AM).
Methods: A top-down popPK model was built for ticagrelor and its AM using full PK profiles collected from a single-dose pharmacokinetic study performed in the University Hospitals of Geneva (NONMEM®). Several covariates (age, body weight, height, body mass index, ethnicities and CYP3A4/3A5 phenotype) were tested on Ticagrelor and AM PK parameters. The final popPK model was used to estimate subject exposure over a dosing interval (AUCNONMEM,pred) that was compared to observed individual exposure (AUCobs, calculated with Pheonix®) . Meanwhile, a bottom-up PBPK model for ticagrelor and its AM was used to predict individual drug exposure for the same patients. A virtual twin profile for each participant was constructed by individualizing the Simcyp® healthy volunteer population file for ethnicity, gender, age, height, weight and CYP3A4/5 phenotype. Predicted AUCSIMCYP,pred were then compared to AUCobs. Two validation criteria were applied to measure bias, a mean forecast error (MFE) of 0.5 to 2.0, a criterion commonly used in PBPK modelling, and a MFE 0.8 to 1.25, largely used for bioequivalence studies.
Results: A total of 148 plasma ticagrelor and 145 AM concentrations were obtained from 19 volunteers. A two-compartment model with zero- and first-order mixed absorption and linear elimination best described ticagrelor and AM data. An additional compartment was used to describe AM data, assuming an equal central volume of distribution for drug and metabolite. Ticagrelor clearance was 48.8 L/h (between subject variability CV% 19.1%), central and peripheral volumes of distribution were 24.5 L (CV% 14.9%) and 270 L, respectively, zero-and first-order absorption rate constant were 0.825 h-1 and 0.391 h-1 respectively, inter-compartmental clearance was 55.9 L/h, metabolic rate constant from drug to metabolite was 1.03 h-1 (CV% 32.2%) and AM clearance 29.6 L/h. Among the variables tested on drug and metabolite parameters, only CYP3A4/3A5 phenotype had a significant influence on parent drug clearance (P=0.0012), indicating a 10.2% lower elimination in poor metabolizers and 29.6% higher elimination in extensive metabolizers compared to normal metabolizers. The virtual twins model construction in Simcyp® showed that 100% of Ticagrelor AUCSIMCYP,pred were included in the MFE 0.5 to 2.0 range (MFE=1.31 IQR [1.05-1.45]) and 68% in the stricter range MFE 0.8 to 1.25. Similar results were obtained with AM, with 100% and 58% AUCSIMCYP,pred predictions lying within the larger and stricter criterion, respectively (MFE=1.13 IQR [0.99-1.23]). As expected for PopPK, 100% of ticagrelor (MFE 0.95 IQR [0.89-1.01]) and AM (0.92 [0.86-0.98]) AUCNONMEM, pred fell within the MFE 0.5 to 2.0 criterion and and 90% and 95% for ticagrelor and AM, respectively, within the stricter range.
Conclusions: The PopPK analysis highlights the interpatient variability in ticagrelor pharmacokinetics and the influence of CYP3A4 activity on its clearance, underlining the importance of taking into account genetic polymorphism in treatments’ personalization, although no dose adjustment is currently recommended for ticagrelor. This preliminary study evaluating the predictability of ticagrelor exposure based on PBPK modeling indicates that this approach is worth further exploring for drug individualization to circumvent some limitations inherent to the more traditional PopPK approach. To be clinically relevant in daily practice, it will be however necessary to define for which drug it may be applicable, considering its variability and its therapeutic window and to develop a robust PBPK model integrating influencing covariates. For such purpose, a dual PopPK and PBPK development may be the right way to follow.
References:
[1] Jelliffe RW, Schumitzky A, Bayard D, Milman M, Van Guilder M, Wang X, et al. Model-based, goal-oriented, individualised drug therapy. Linkage of population modelling, new ‘multiple model’ dosage design, Bayesian feedback and individualised target goals. Clin Pharmacokinet 1998; 34: 57– 77..
[2] Polasek, T. M., Tucker, G. T., Sorich, M. J., Wiese, M. D., Mohan, T., Rostami-Hodjegan, A., Korprasertthaworn, P., Perera, V., and Rowland, A. (2018) Prediction of olanzapine exposure in individual patients using physiologically based pharmacokinetic modelling and simulation. Br J Clin Pharmacol, 84: 462– 476. doi: 10.1111/bcp.13480.
[3] Marsousi, N., et al., Usefulness of PBPK Modeling in Incorporation of Clinical Conditions in Personalized Medicine. J Pharm Sci, 2017. 106(9): p. 2380-2391.
Reference: PAGE 29 (2021) Abstr 9790 [www.page-meeting.org/?abstract=9790]
Poster: Drug/Disease Modelling - Other Topics