Garcia MJ, Santos Buelga D, Otero MJ, Serrano J, Blanco B, Dominguez-Gil A
Departamento de Farmacia y Tecnologma Farmaciutic, Facultad de Farmacia,
The aim of this study was to investigate the influence of coadministration of another antiepileptic drug (carbamazepine-CBZ-, phenytoin-PHT-, phenobarbital-PB-, lamotrigine-LAM-, vigabatrin-VIG- ) on valproic acid (VAL) clearance, using the program NONMEM.
A total of 535 measured serum VAL concentrations were obtained from 208 adult epileptic patients receiving Depakine tablets in mono or bitherapy. Patients were 274 females and 261 males, with ages ranged from 14 to 95 years and total body weight (TBW) ranged from 27 to 100 kg. Steady-state trough VAL concentrations, dosing history, concomitant drug administration and demographic characteristics of patients were retrospectively collected from TDM. A one-compartment model with first order absorption and elimination was used and because of the nature of data absorption rate, bioavailability and distribution volume were fixed to literature values.
The influence of covariates on VAL clearance was investigated using linear and non-linear models. Results show that clearance linearly depends on body weight (WT) and clearance increases when PHT, CBZ or PB are concomitantly administered, resulting the following final model:
CL = 0.00794 x TBW x (1 + 0.47 x CBZ) (1 + 0.659 x PHT) (1 + 0.346 x PB)
Although this model includes the influence of CBZ, PHT and PB, it was built from data of patients receiving VAL alone or VAL plus one other anticonvulsant drug, we consider therefore it shouldn’t be used in other situations.
According proportional error models for both clearance interindividual and residual variabilities, this were respectively reduced from 38 % and 18 % in the simplest model to 29.2 % and 13.78 % in the final model.
The standardised predictions errors in 36 serum VAL concentrations prospectively collected, resulting in a mean value of 0.097 (not significantly different from zero) with an standard deviation of 0.86 (close to expected value of one), indicating the validity of the final proposed model.
Reference: PAGE 6 (1997) Abstr 657 [www.page-meeting.org/?abstract=657]
Poster: poster