Individualisation of Leflunomide Dosing in Rheumatoid Arthritis Patients : Development of a Semi-Physiologically Based Pharmacokinetic Model

Objective: Following oral administration, leflunomide is rapidly converted to its active metabolite teriflunomide via blood born metabolism and the CYP enzymes 1A2 and 2C19. Variability in enzymatic activity may contribute to the high variability in total teriflunomide concentrations (3 – 150 mg/L) achieved at standard doses. Leflunomide itself is not detectable in the plasma and teriflunomide is highly protein bound (fu ~ 0.26%), yet the significance of free teriflunomide has been largely unexplored. Teriflunomide has a long half-life (~ 2 weeks), which is contributed to by significant entero-hepatic recycling that is mediated by ABCG2. This entero-heaptic recycling can be inhibited with the administration of cholestyramine. The aim was to build a semi-physiologically based pharmacokinetic (PK) model for leflunomide that represents the key processes influencing leflunomide and teriflunomide disposition. The strategy was to build a model around bound and unbound drug concentrations in the liver and include first-pass elimination and entero-hepatic recycling.
Methods: PK data and physiological values from the literature were used for the model development. Data were extracted from the literature using DigitizeIt. Initial model development and exploration used R Software Version 2.15.2 with the package deSolve for differential equation solving and simulation of compartmental concentrations. The final model was transferred to NONMEM for the estimation of parameter values using a meta-data set (n = 20 studies), representing mean teriflunomide concentrations after oral or IV administration of leflunomide and teriflunomide. The population parameter estimates represented inter-study variability.
Results: The final model was comprised of 14 differential equations, which described the hepatic control of unbound drug clearance, the conversion of leflunomide to teriflunomide and the enterohepatic recycling of teriflunomide. Literature based PK estimates in the R developed model described rapid conversion of leflunomide to teriflunomide, which had a dual phase of distribution followed by a half-life of approximately 2 weeks. When enterohepatic recycling was turned off, the half-life of teriflunomide decreased to approximately 1-2 days mimicking the effects of cholestyramine. Population PK estimates determined using NONMEM were similar to the literature based estimates.
Conclusion: The model described teriflunomide concentrations following teriflunomide or leflunomide administration well, and included the key physiological processes which appear to be responsible for the high inter-individual variability in teriflunomide concentrations. Future research will include using the model to describe teriflunomide concentrations following leflunomide administration within a cohort of Rheumatoid Arthritis patients, and determining the significance of the multiple patient demographics and genetic variables on kinetics.