The influence of drug–drug interaction and patients’ characteristics on valproic acid's clearance in adults with epilepsy using nonlinear mixed effects modeling

Abstract

Monitoring valproic acid (VPA) concentrations is especially challenging due to its highly variable pharmacokinetics (PK) and complex interactions with other antiepileptic drugs. We used sparse routine therapeutic drug monitoring data ( n = 200) from 129 adults with epilepsy to develop a population PK model of VPA, and determine the factors that influence its clearance (CL/F). Patients were on mono VPA therapy, or were concomitantly treated with carbamazepine, phenobarbital, topiramate (TPR), lamotrigine or benzodiazepines. A one-compartment model with first-order absorption and elimination was used to fit the concentration–time VPA data. Estimates generated by NONMEM indicated that VPA CL/F was influenced by the patients’ body weight (increases with the 0.556 exponent), VPA daily dose (if it is greater than 1000 mg/day, CL/F increases by 43%), and co-therapy with TPR (lowering CL/F for 23%). The interindividual variability in VPA CL/F was modeled with exponentional error model. The estimated coefficient of variation was 31.9%, while the residual variability was 23.8% for the proportional and 13.2 mg/l for the additive component. The model was validated in a separate set of 24 patients, and the predictive performance was evaluated, that indicated unbias and acceptable precision. This study confirms the interaction of VPA with TPR, which is presumably dependent on VPA dose.