Anna-Karin Hamberg (1), Marja-Liisa Dahl (1), Gabriella Scordo (1), Roberto Padrini (2), Martina Barban (2) and E. Niclas Jonsson (3)
(1) Department of Medical Sciences, Clinical Pharmacology, Uppsala University, Sweden, (2) Department of Pharmacology and Anaesthesiology, University of Padova, Italy and (3) Department of Pharmaceutical Biosciences, Division of Pharmacokinetics and Drug Therapy, Uppsala University, Sweden
Introduction: Racemic warfarin is the most widely prescribed anticoagulant drug for the prevention and treatment of tromboembolic disorders. Because of large interpatient variability in dose-anticoagulant effect relationship and a narrow therapeutic index,   dosing is individually titrated by repeated measurements of INR (protrombin time expressed as an international normalized ratio) to minimize the risk of serious bleeding events without compromising the anticoagulant effect. S-warfarin, the most potent enantiomer of warfarin, is metabolised by CYP2C9, and it has been suggested that genetic variation in the gene coding for this enzyme contributes significantly to the large interpatient variability in warfarin-dose requirements.
Objective: To develop a population PK/PD model for S-warfarin (S) and R-warfarin (R) and their anti-coagulant activity, and to use the model to estimate how much of the variability in INR response that is explained by variability in PK, with special emphasis on the contribution of CYP2C9 gene variation for clearance of S.
Method: The study population consisted of 57 Italian outpatients eligible for long-term warfarin anticoagulant therapy. CYP2C9 genotype was available for all subjects. Plasma concentrations of S and R and INR were measured following a 10 mg single dose and after attainment of stable maintenance dosing [median weekly dose (range): 28.75 (7.5-78.75) mg]. The analysis was performed in NONMEM in two steps. In the first step, the PK model was developed and pre-specified covariates tested. In the second step individually estimated S and R concentrations were used to drive the PD model.
Results: Disposition of S and R after oral administration was best described with a two- and a one-compartment model with first-order absorption, respectively. Estimated mean parameters (CV%) for S were: V1/F 13.9 L (32%), V2/F 3.9 L (91%) and Cl/F 0.30 L/h (32%), and for R: V/F 12.9 L (21%) and Cl/F 0.14 L/h (26%). Of the covariates tested, CYP2C9 genotype (*1/*3 and *2/*2 combined) was the only one identified as having a significant effect, with a 46% (14%) reduction in clearance for S as compared to wild-type (*1*1). The PD response was adequately described by a competitive agonist model with EC50 for S and R estimated to 0.16 mg/L and 0.35 mg/L, respectively. A transit compartment model with two parallel series of transit compartments, with mean transit times of approximately 10 and 70 h, respectively, described the time-delay in INR-response.
Discussion: The model suggests that variability in PK due to CYP2C9 polymorphism accounts for approximately 20% of the total variability in stabilised INR response in our study population. Even if the influence of genotype on the overall variability is relatively small, genotype may still be an important tool for identifying those patients most susceptible to adverse drug reactions due to impaired elimination of S, i.e. subjects with *1/*3, *2/*2, *2/*3 and *3/*3 genotypes.
Reference: PAGE 14 (2005) Abstr 755 [www.page-meeting.org/?abstract=755]
Poster: poster