Effects of Concomitant Antiepileptic Drugs on Serum Carbamazepine Concentration in Epileptic Patients: Quantitative Analysis Based on Extracellular Water Volume as a Transforming Factor

The effects of concomitant antiepileptic drugs on the serum carbamazepine concentration (Ct) were analyzed quantitatively. Primidone (PRM), phenobarbital (PB), phenytoin (PHT), valproic acid (VPA), zonisamide (ZNS), clonazepam (CZP), and ethosuximide (ETS) were coadministered with carbamazepine (CBZ). Routine therapeutic drug monitoring data, obtained from epileptic patients who were treated with the repetitive oral administration of CBZ fine granules/tablets, were used for the analysis. A total of 119 patients were administered CBZ alone, and 91, 39, 19, and 6 patients were coadministered one, two, three, and more than four different antiepileptic drugs, respectively. Using the data obtained from the patients administered CBZ alone, Ct could be expressed approximately as a function of the daily dose per extracellular water volume (D/VECW) as Ct=A (D/VECW)B (A, B: parameter). By comparing the regression line on logCt vs. log(D/VECW) for CBZ alone with that for CBZ plus another concomitant drug, Ct was thus found to be affected at each definite ratio by PB and PHT, but not by VPA and ZNS. We postulated a model showing that Ct is affected by each concomitant antiepileptic drug i at each definite ratio. We defined the parameter Ri(i=1, 2, …, 7) representing the effect of each concomitant antiepileptic drug on Ct. A linear polynomial expression, in which both members of this model are converted into common logarithms, was used for a multiple regression analysis. The analysis clarified that PB and PHT lowered Ct to 0.770 and 0.710 the value of CBZ alone, respectively. On the other hand, VPA and ZNS did not affect Ct. The number of patients coadministered PRM, CZP, and/or ETS was not sufficient to detect the effect on Ct based on a test of significance. In the case of the addition or discontinuation of concomitant antiepileptic drugs in the same patient, the estimated Ct values were calculated using the value of each Ri and compared with the measured Ct values. Both values were in good agreement, and thus our results appear valid.