Background: There is a clinical need for novel oral anticoagulants with predictable pharmacokinetics and pharmacodynamics. Rivaroxaban is an oral direct Factor Xa (FXa) inhibitor in clinical development for the prevention and treatment of thromboembolic disorders. This analysis was performed to characterize the population pharmacokinetics and pharmacodynamics of rivaroxaban in patients participating in two phase II, double-blind, randomized, active-comparator-controlled studies of twice-daily rivaroxaban for the prevention of venous thromboembolism after total hip- or knee-replacement surgery.
Methods: Sparse blood samples were taken from all patients participating in the studies (n = 1009). In addition, a subset of patients in the hip study (n = 36) underwent full profiling. Rivaroxaban plasma concentrations, FXa activity and the prothrombin time were determined. Nonlinear mixed-effects modelling was used to model the population pharmacokinetics and pharmacodynamics of rivaroxaban.
Results: An oral one-compartment model described the population pharmacokinetics of rivaroxaban well. On the first postoperative day only, categorization of patients as slow or fast absorbers as a tool to address variability in absorption improved the fit of the model. Clearance of rivaroxaban was lower and more variable on the first postoperative day, and so time was factored into the model. Overall, the only major difference between the models for the hip study and the knee study was that clearance was 26% lower in the knee study, resulting in approximately 30% higher exposure. Residual variability in the models was moderate (37% and 34% in the hip and knee studies, respectively). Plasma concentrations of rivaroxaban increased dose dependently. Pharmacokinetic parameters that were estimated using the models agreed closely with results from full-profile patients in the hip study, demonstrating that rivaroxaban pharmacokinetics are predictable. The pharmacokinetics of rivaroxaban were affected by expected covariates: age affected clearance in the hip study only, haematocrit (on the first postoperative day only) and gender affected clearance in the knee study only, and renal function affected clearance in both studies. Bodyweight affected the volume of distribution in both studies. However, the effects of covariates on the pharmacokinetics of rivaroxaban were generally small, and predictions of 'extreme' case scenarios suggested that fixed dosing of rivaroxaban was likely to be possible. FXa activity and the prothrombin time were both affected by surgery, probably because of perioperative bleeding and intravenous administration of fluids; therefore, time was included in the pharmacodynamic models. In both studies, FXa activity correlated with rivaroxaban plasma concentrations following a maximum effect model, whereas prothrombin time prolongation correlated following a linear model with intercept. The slope of the prothrombin time prolongation correlation was 3.2 seconds/(100 microg/L) in the hip study and 4.2 seconds/(100 microg/L) in the knee study. Both pharmacodynamic models in both studies demonstrated low residual variability of approximately 10%.
Conclusion: This population analysis in patients undergoing major orthopaedic surgery demonstrated that rivaroxaban has predictable, dose-dependent pharmacokinetics that were well described by an oral one-compartment model and affected by expected covariates. Rivaroxaban exposure could be assessed using the prothrombin time, if necessary, but not the international normalized ratio. The findings suggested that fixed dosing of rivaroxaban may be possible in patients undergoing major orthopaedic surgery.