Teixeira P (1), Otero MJ (2), GarcÃa Sánchez MJ (1), Villacañas Palomares MV (2), Santos Buelga D (1)
(1) Department of Pharmacy and Pharmaceutical Technology, University of Salamanca, Spain, (2) Service of Pharmacy, University Hospital of Salamanca, Spain
Objectives: The main goal of this work was to develop an updated population pharmacokinetic model to identify and quantify the influence of demographics, clinical and treatment factors on the clearance (CL) of valproic acid (VLP) in adult patients.
Methods: Serum concentrations of VLP were taken from patients aged 15-84 years old, who had been treated with sodium valproate. These patients were included in the TDM program, conducted in the University Hospital of Salamanca over the last 20 years. After applying the inclusion/exclusion criteria previously established, the final database included 868 serum concentrations from 411 patients. Pharmacokinetic analysis was performed with NONMEM V7.2 (FOCEI) considering a one-compartment model, fixing the absorption constant and volume of distribution at 4.1 h-1 and 0.20 L/kg, respectively[1-3]. Proportional error models were assumed to describe interindividual and residual variabilities. The analysed covariates were: weight (WGT), age, gender (SEX), and concomitant treatment with carbamazepine (CBZ), clobazam (CLB), clonazepam (CLO), phenobarbital (PHB) phenytoin (PHT), lamotrigine (LTG), topiramate (TOP) and vigabatrin (VIG).
Results: The covariates with significant influence on CLVLP/F were: WGT, SEX and concomitant treatment with CBZ, PHB, PHT, LTG and VIG. LTG was excluded in the backward elimination process (p < 0.01). The covariate SEX was also eliminated in the final model because of the low clinical significance (< 15 %). The proposed final model for CLVLP/F was as follows:
CLVLP/F (L/h) = 0.58×((WGT/70)0.58)×(1.44×CBZ)×(1.23×PHB)×(1.39×PHT)×(1.44×VIG)
w2 = 0.055 (shrinkage: 16 %)
s2= 0.033 (shrinkage: 21 %)
The standard estimation error was lower than 11 % for all parameters.
VIG does not undergo biotransformation or binding to plasma proteins[2], however we found that this drug significantly increases CLVLP/F. Its inclusion/exclusion in the model did not affect the influence (type and extent) of the others covariates on the CLVLP/F. This is why this drug was maintained in the final model despite the lack of physiological justification of this fact.
Conclusions: The population pharmacokinetic model developed for VLP in adult patients includes WGT and the concomitant administration of CBZ, PHB, PHT and VIG. More studies are required to verify the causes and the true extent of the interaction discovered between VIG and VLP.
References:
[1] Dutta S, Reed RC. Distinct absorption characteristics of oral formulations of valproic acid/divalproex available in the United States. Epilepsy Res. 2007; 73:275-83.
[2] Patsalos PN. Antiepileptic drug interactions. A clinical guide. 2nd ed. Springer; London. 2013.
[3] Blanco-Serrano B, Otero MJ, Buelga DS, Garcia Sanchez MJ, Serrano J, Dominguez-Gil A. Population estimation of valproic acid clearance in adult patients using routine clinical pharmacokinetic data. Biopharm. Drug Dispos. 1999; 20:233-40.
Reference: PAGE 23 () Abstr 3243 [www.page-meeting.org/?abstract=3243]
Poster: Drug/Disease modeling - CNS