External validation of population pharmacokinetic models of oxcarbazepine active metabolite in Chinese children with epilepsy.

Abstract

Objective: To assess the predictive performance of published population pharmacokinetic models of the oxcarbazepine (OXC) active metabolite, 10, 11-dihydro-10-monohydroxycarbazepine (MHD), using external data sets in Chinese children with epilepsy.

Method: A total of 231 concentrations from 185 Chinese children with epilepsy were used for external validation. PubMed, Embase, and Web of Science were searched for published PPK models of OXC active metabolite MHD in children. Seven models were searched and labeled A to G based on the year of publication. Prediction error, visual predictive check, and normal prediction distribution error tests were employed to assess the model's extrapolation performance. The Bayesian prediction method was applied to ascertain the influence of prior concentrations on the model's predictive performance.

Result: A total of seven MHD PPK models were retrieved. Models A, D, E, and G exhibited good predictive performance in prediction-based diagnostics and visual predictive checks. The normalized predictive distribution error test shows that none of the models is suitable to describe our data. Bayesian prediction significantly improved the prediction performance of all the models with one prior observation.

Conclusion: The published MHD PPK models showed extensive variation in predictive performance when extrapolated in Chinese children with epilepsy. Bayesian forecasting substantially improved the predictive performance of the model and might facilitate the customization of OXC dosing.