Aripiprazole is a new antipsychotic developed in Japan, and is prescribed for schizophrenia as a dopamine-serotonin system stabilizer.1,2) The action of aripiprazole is different from that of other typical and atypical antipsychotics; that is, aripiprazole is a potent partial agonist at dopamine D2 and serotonin (5-HT 1A ) receptors and antagonist at 5-HT 2A receptor.3,4) The results of in vitro studies indicated that aripiprazole is mainly metabolized by human cytochrome P450 isozymes CYP3A4 and CYP2D6.5) It has been reported that CYP3A4 and CYP2D6 are metabolic enzymes for numerous compounds, and also that there are many compounds that inhibit these enzymes. Although individual differences in hepatic levels of CYP3A4 enzyme protein have been reported to vary as much as 40-fold, 6) almost no gene mutations affecting CYP3A4 metabolic activity or ethnic differences have been reported. For CYP2D6, however, a number of polymorphisms and the existence of ethnic differences in the types and distribution of polymorphisms have been reported.6,7) The purpose of the present study was to estimate the population pharmacokinetic parameters of aripiprazole in healthy Japanese males, and to evaluate the effects of CYP2D6 polymorphisms and CYP3A4 inhibition on the pharmacokinetics of aripiprazole, and also its unknown (residual) interindividual variability.In the present study, we used the nonlinear mixed-effects model (NONMEM) method designed for the estimation of population pharmacokinetic parameters.8) This method pools data from all individuals but explicitly models and handles the complicated error structure arising from proper accounting of interindividual (h) and intraindividual (e) random effects. The first-order estimation method is the first estimation method available with NONMEM: it handles the error structure of data using first-order Taylor-series expansion in the random effect h, evaluated at the expected values (i.e.
hϭ0).9) On the other hand, a newer analysis method designed for the estimation of population parameters, the firstorder conditional estimation (FOCE) method, is available with NONMEM software. 9) That is, FOCE uses first-order expansions for values of the h, but these values are the conditional (Bayesian) estimates of the h, rather than zero. The FOCE method needs much (20-40-fold) longer computational time than the first-order method, but generally gives precise parameter estimates, especially for the analysis of multiple-dose pharmacokinetic data with multiple compartment models. [9][10][11][12] In the present study, therefore, we used the FOCE method to estimate the population pharmacokinetic parameters of aripiprazole, where a two-compartment model with first-order input was applied for data obtained from single-and multiple-dose clinical trials.
MATERIALS AND METHODS
Pharmacokinetic DataPlasma aripiprazole concentration data for population pharmacokinetic analysis were obtained in two previous studies. 13,14) Briefly, 68 healthy Japanese male subjects, aged 20-32 years old (mean: 23.1) and w...