Population estimation of valproic acid clearance in adult patients using routine clinical pharmacokinetic data

The aim of the present study was to estimate valproic acid (VPA) clearance values for adult patients with epilepsy, using serum concentrations gathered during their routine clinical care. Retrospective steady state serum concentrations data (n=534) collected from 208 adult patients receiving VPA were studied. Data were analysed according to a one-compartment model using the NONMEM program. The influence of VPA daily dose (Dose), gender, age, total body weight (TBW), and comedication with carbamazepine (CBZ), phenytoin (PHT) and phenobarbital (PB) were investigated. The results of the population pharmacokinetics analysis were validated in a group of 30 epileptic patients. The final regression model for VPA clearance (Cl) was: $¿rm Cl¿left (¿rm L/h ¿right )=0¿rm. 004¿times TBW¿times Dose ¿0.304¿¿rm ¿times 1.363¿,¿rm CBZ¿times 1. 541¿,¿rm PHT¿times 1.397¿,¿rm PB.$ The inter-individual variability in VPA clearance, described by a proportional error model, had a variation coefficient (CV) of 23.4% and the residual variability, described using an additive model, was 11.4 mg/L. These results show that VPA clearance increased linearly with TBW, but increases nonlinearly with increasing VPA daily dose. Concomitant administration of CBZ, PHT and PB led to a significant increase in VPA clearance. The model predictions in the validation group were found to have satisfactory precision and bias. In conclusion, inter-individual variability in VPA clearance can be partly explained by TBW, daily dose and bitherapy with CBZ, DPH or PB. Inclusion of these factors allows this variability to be reduced by 37.23% which may be very useful for clinicians when establishing the initial VPA dosage regimen. However, the magnitude of inter-individual plus residual variabilities, remaining in the final model, render these dosage predictions imprecise and justify the need for VPA serum level monitoring in order to individualize dosage regimens more accurately.