Maximum a posteriori Bayesian estimation of oral cyclosporin pharmacokinetics in patients with stable renal transplants.

Objective: To develop a maximum a posteriori probability (MAP) Bayesian estimator for the pharmacokinetics of oral cyclosporin, based on only three timepoints, and evaluate its performance with respect to a full-profile nonlinear regression approach.

Patients: 20 adult patients with stable renal transplants given orally administered microemulsified cyclosporin and mycophenolate.

Methods: Cyclosporin was assayed by liquid chromatography-mass spectrometry. Nonlinear regression and MAP Bayesian estimation were performed using a home-made program and a previously designed pharmacokinetic model including an S-shaped absorption profile described by a gamma distribution.

Outcome Measures And Results: MAP Bayesian estimation using the best limited sampling strategy (before administration, and 1 and 3 hours after administration) was compared with nonlinear regression (taken as the reference method) for the prediction of the different pharmacokinetic parameters and exposure indices. Median relative prediction error was -0.49 and -3.42% for area under the concentration-time curve over the administration interval of 12 hours (AUC12) and estimated peak drug concentration (Cmax), respectively (nonsignificant). Relative precision was 2.00 and 4.32%, and correlation coefficient (r) was 0.985 and 0.955, for AUC12 and Cmax, respectively.

Conclusion: This paper reports preliminary results in a stable renal transplant patient population, showing that MAP Bayesian estimation can allow accurate prediction of AUC12 and Cmax with only three samples (0, 1 and 3 hours). Although these results require confirmation by further studies in other clinical settings, using other drug combinations, other analytical methods and commercially available pharmacokinetic software, the method seems promising as a tool for the therapeutic drug monitoring of cyclosporin in clinical practice or for exposure-controlled studies.