Elisabet Størset (1, 3), Christine Staatz (2), Stefanie Hennig (2), Troels K. Bergmann (2, 4), Anders Åsberg (1), Stein Bergan (1, 3), Sara Bremer (3), Karsten Midtvedt (3), Nick Holford (5)
(1) School of Pharmacy, University of Oslo, Norway, (2) School of Pharmacy, University of Queensland, Australia, (3) Oslo University Hospital, Rikshospitalet, Norway, (4) Department of Pharmacology, Aarhus University, Denmark, (5) Department of Pharmacology and Clinical Pharmacology, University of Auckland, New Zealand
Objectives: Two independent population pharmacokinetic models have been developed for tacrolimus in kidney transplant recipients in Brisbane [1] and Oslo [2]. Both included “time after transplantation” as an empirical covariate. The objectives of this study were to develop a new population pharmacokinetic model based on combined data from Brisbane and Oslo, using a theory based approach to covariate recognition and to evaluate the models’ abilities to predict internal and external tacrolimus concentrations.
Methods: Model building and Bayesian forecasting was performed using NONMEM 7.2. A total of 3100 tacrolimus whole blood concentrations were obtained from 242 patients. Tacrolimus whole blood concentrations were predicted using literature values of maximum binding capacity to erythrocytes (Bmax), binding affinity (Kd) [3] and hematocrit. Body size metrics (total body weight, normal fat mass, fat free mass) were evaluated using allometric scaling. CYP3A5 expression was included as a covariate on both bioavailability and clearance. External evaluation was performed by retrospective Bayesian forecasting of observed concentrations in 30 new patients the first two weeks after kidney transplantation.
Results: Relating pharmacokinetics to implicit plasma concentrations rather than whole blood concentrations improved the model markedly (ΔOFV -193). Bioavailability decreased as a function of prednisolone dose, best described by an Emax model (Predmax=-61 %, Pred50=19 mg), theoretically resulting from induction of intestinal CYP3A/P-gp [4]. Fat free mass was the best body size metric using theory based allometry. In CYP3A5 expressers, as predicted from theory, clearance was higher (32 %) and bioavailability was lower (15 %). The combined model showed superior predictive performance of external concentrations compared with the two previous empirically based models.
Conclusions: Theory based population pharmacokinetic modeling of tacrolimus improves predictive performance in external patients.
References:
[1] Bergmann, T. et al. Prediction correction: quick fix for VPC misdiagnosis in a tacrolimus popPK model. PAGANZ 2012. Available from: http://www.paganz.org/abstracts/prediction-correction-quick-fix-for-vpc-misdiagnosis-in-a-tacrolimus-poppk-model/. Accessed January 2013
[2] Storset, E. et al. Population pharmakokinetics of tacrolimus to aid individualized dosing in kidney transplant recipients. PAGE 2012, abstract II-252012. Available from: http://www.page-meeting.org/default.asp?abstract=2306. Accessed January 2013
[3] Jusko, W. J. et al. Pharmacokinetics of tacrolimus in liver transplant patients. Clin Pharmaco Ther 57, 281-290 (1995).
[4] Lam, S. et al. Corticosteroid interactions with cyclosporine, tacrolimus, mycophenolate, and sirolimus: fact or fiction? Ann Pharmacother 42, 1037-1047 (2008).
Reference: PAGE 22 (2013) Abstr 2808 [www.page-meeting.org/?abstract=2808]
Poster: Model evaluation