Time-dependent pharmacokinetics of cyclosporine (NeoralR) in de novo renal transplant patients

Abstract

A model for the large scale temporal trend in the oral bioavailability of microemulsion cyclosporine (Neoral) (CsA) is established, with dependence on post-(renal) transplantation day (PTD). Twenty de novo adult renal transplant recipients were monitored for CsA administered orally q12 h. A model development group (11 patients, 315 blood concentration samples) was screened at 2 h (C(2); n = 92), 3 h (C(3); n = 56) and at predose troughs (C(min); n = 167) over periods of up to 75 days. The final model was tested in nine patients with C(min) (n = 580) monitored across 4-5 years. The doses varied between 100 and 538 mg with an apparent hyperbolic trend in C(2)/dose vs. PTD. A nonlinear mixed effects modelling (NONMEM) approach was used to obtain population and individual patient one-compartment pharmacokinetic (PK) parameters for oral CsA, which carry implicit the bioavailability (F). In the final PK model (PK-f) the F was modelled via a simple function for the temporal (days) trend of the bioavailability after transplantation as, F(f) = 1-alpha * exp(-lambda * PTD) resulting in a 28% reduction in the unexplained intra-individual variability. The population PK-f parameters were, for apparent clearance [mean, 95% confidence interval (interindividual CV%)] Cl/F(f) = 17.0 (13.8-20.2) L/h (27%), apparent central compartment volume of distribution, V/F(f) = 134 L (108-160) (28%), and lambda = 0.037/day (0.005-0.069) (120%). The absorption rate k(a) and the parameter alpha were approximated iteratively as 4/h and 0.62 respectively. The PK-f was structurally superior to the base model in explaining part of the within subject (occasion) variability and predicting the exposure surrogates C(2) and C(3). Also, the PK-f was better than the base model with Bayesian fitting of individual profiles in that group. The PTD-dependent relative bioavailability model provides a rational means of steering dose titration of CsA in de novo renal transplantation patients by removing the large scale PK adjustment signal, either through nomograms or as a Bayesian prior.