Pharmacokinetic Modeling and Development of Bayesian Estimators in Kidney Transplant Patients Receiving the Tacrolimus Once-Daily Formulation

Background: The once-daily formulation of tacrolimus has been reported to exhibit the same efficacy and safety profile as compared with the immediate-release form administered twice daily. However, as a result of differences in their pharmacokinetic (PK) profile, the PK models or Bayesian estimators (MAP-BE) previously developed for the immediate-release formulation cannot be used for the new once-daily formulation. Using the PK information obtained from a Phase II trial, the aim of this study was to explore the feasibility of developing a PK model and a MAP-BE able to estimate, on the basis of a routinely applicable limited sampling strategy, tacrolimus individual PK parameters and AUC(0-24h) in de novo renal transplant patients given the once-daily formulation. Methods: Twelve de novo kidney transplant recipients receiving once-daily tacrolimus as part of their immunosuppressive regimen provided full PK profiles (17 concentration time points over 24 hours) on Days 14 and 42 posttransplantation. On the basis of a one-compartment open model with absorption described as following a double gamma distribution, a classic iterative two-stage method was applied to develop MAP-BEs. All the limited sampling strategies with a maximum of three sampling times within 4 hours postdose were tested for Bayesian forecasting with the aim of accurately estimating the AUC(0-24h). Results: Once-daily tacrolimus exhibited a high interpatient PK variability with coefficients of variation of 34.3% and 36.2% for AUC(0-24h)/dose (mg/kg) on Days 14 and 42, respectively. Regression analysis between C-0 and AUC(0-24h) yielded r(2) = 0.68 and 0.76 at these two periods, respectively. The iterative two-stage approach led to the development of a different MAP-BE for each posttransplantation period, which allowed estimation of once-daily tacrolimus pharmacokinetics and AUC(0-24h) on the basis of a C-0-C-1h-C-3h sampling schedule. The mean bias of the Bayesian versus reference (trapezoidal) AUCs was 4.2% +/- 6.1% (range, -11.8% to +11.2%; root mean square error = 7.1%) on Day 14 and 0.2% +/- 7.9% (range, -12.9% to +14.1%; root mean square error = 7.8%) on Day 42. Conclusion: A PK model and Bayesian estimators allowing estimation of tacrolimus AUC(0-24h) based on a routinely applicable limited sampling strategy were developed for once-daily tacrolimus in renal transplantation. Further validation in independent groups of patients is required to confirm their applicability for optimizing the monitoring of once-daily tacrolimus in routine clinical practice or to conduct observational or comparative therapeutic drug monitoring clinical trials.