Development of a Bayesian estimator for the therapeutic drug monitoring of mycophenolate mofetil in children with idiopathic nephrotic syndrome

Abstract

The use of mycophenolate mofetil (MMF) in children with idiopathic nephrotic syndrome (INS) is increasing. However, the clinical benefit of its monitoring has been scarcely studied, and little is known about its pharmacokinetics in this context. The objectives of the present study were: (i) to study and model the pharmacokinetics of mycophenolic acid (MPA; the active moiety of MMF) in paediatric patients with INS given MMF, at all stages of the disease; (ii) to develop a Bayesian estimator (MAP-BE) for individual inter-dose area under the concentration-time curve (AUC) prediction in this population, using a limited blood sampling strategy (LSS). Full-pharmacokinetic (PK) profiles of MPA collected in paediatric inpatients with INS already treated with a maintenance immunosuppressive therapy based on MMF (with no calcineurin inhibitors; CNI) were studied. A classical iterative two-stage (ITS) method was applied to model the data and develop MAP-BEs using a one-compartment open model where the absorption is described by a double gamma law allowing the description of a potential enterohepatic recirculation. The performance of the MAP-BE developed for individual exposure assessment was evaluated by the bias and precision of predicted AUCs with respect to measured, trapezoidal AUCs (reference value), and by the proportion of predicted AUCs with absolute error >20%. These PK tools were tested in an independent group of patients. Sixty PK profiles of MPA from children receiving MMF in association to corticosteroids or given alone were included in the study. Forty-five of these PK profiles were used to develop a PK model and a MAP-BE, and 15 for their validation. In the building group, the PK model fitted accurately the PK profiles of MPA: mean residual error of modelled vs. reference AUC was m±SD=-0.015±0.092 (range: -0.153 to 0.204). The MAP-BE which allowed the estimation of MPA AUC on the basis of a 20 min-60 min-180 min LSS was then developed. In the independent group of patients, its mean residual error vs. reference AUCs was m±SD=-0.036±0.145 (range: -0.205 to 0.189). Thus, a PK model and its derived MAP-BE for MMF (without any associated CNI) when given to children with INS have been developed. Clinical trials using these PK tools could test the potential impact of the therapeutic drug monitoring of MMF based on the AUC on the clinical evolution of INS.