CARBAMAZEPINE POPULATION PHARMACOKINETICS IN CHILDREN - MIXED-EFFECT MODELS

The aim of the authors' study was to investigate the factors affecting carbamazepine (CBZ) clearance (CL) in children with epilepsy. The fac tors evaluated were total body weight (TBW), age, dose, sex, and pheno barbital (PB) and valproic acid (VA) comedication. A total of 387 stea dy-state serum concentration samples was analyzed. These were collecte d during CBZ therapy from 201 children, aged 1-14 years and weighting 9-78 kg. Population CL was calculated by using NONMEM, with a one-comp artment model with first-order absorption and elimination. The absorpt ion rate, bioavailability, and volume of distribution were set at valu es found in the literature. The model found best to describe the data was CL = (0.0122 TEW + 0.0467 Dose) Age(0.331) (1.289 PB). The interin dividual variability in CL had a variation coefficient (CV) of 11.8%, and the residual error, described by using an additive model, was 1.5 mg/l. The results show that CL increases linearly with TEW and nonline arly with age; thus older children have a lower CL with respect to TEW than do younger ones. Likewise CL was seen to increase with the incre ase in the CBZ dose, suggesting a dose-dependent autoinduction of CBZ metabolism. Concomitant PB administration affected CL; however, sex an d VA comedication did not affect it significantly. The final regressio n model for CL was validated in a different group of 74 children. The standarized prediction error (SPE) was not significantly different fro m zero (SPE = 0.028), indicating that the model proposed for CL can be used to make accurate dosage recommendations. With these population e stimates, CBZ doses that would be suitable for pediatric patients of d ifferent ages are proposed.