Major depression may require antidepressant treatment for several years. This necessitates consideration of the long-term effects of antidepressants on multiple clinical endpoints. The antidepressants paroxetine and fluvoxamine are potent in vitro inhibitors of CYP2D6 and CYP1A2 isozymes, respectively. CYP2D6 and CYP1A2 are important for the clearance of 30 or more frequently used medications. Moreover, CYP1A2 also contributes to metabolism of 17beta-estradiol and metabolic activation of environmental procarcinogens (e.g., arylamines in cigarette smoke). The aim of this study was to assess the determinants of interindividual variability and extent of CYP2D6 and CYP1A2 inhibition during paroxetine and fluvoxamine treatment. Healthy volunteers and patients received caffeine (100 mg) and dextromethorphan (30 mg) at baseline and at steady state of paroxetine (10-20 mg/day, 5-74 days, N = 13) or fluvoxamine (50-100 mg/day, 5-43 days, N = 8). The caffeine metabolic ratio (CMR) and the log O-demethylation ratio (ODMR) of dextromethorphan in overnight urine were used as in vivo indices of the CYP1A2 and CYP2D6 isozyme activities, respectively. All subjects had an extensive metabolizer phenotype for CYP2D6. After fluvoxamine treatment, baseline CMR 5.1 +/- 1.4 (mean +/- SD) decreased to 2.7 +/- 1.1 (p < 0.01). Paroxetine did not have a significant effect on CMR (p > 0.05). In seven of eight subjects in the fluvoxamine group, posttreatment CMR was comparable with the minimum CMR value (2.0) attainable in nonsmoking healthy volunteers. After paroxetine treatment, log ODMR changed from a baseline value of -2.28 +/- 0.37 to -1.13 +/- 0.44, indicating significant inhibition of CYP2D6 (p < 0.001). Subjects' CYP2D6 phenotype did not change after paroxetine treatment. Fluvoxamine had no significant effect on log ODMR (p > 0.05). The extent of inhibition of CYP2D6 and CYP1A2 by paroxetine and fluvoxamine, respectively, displayed a positive correlation with baseline enzyme activity (p < 0.05). In addition, a negative association was found between the plasma paroxetine concentration and the CYP2D6 activity after paroxetine treatment (r = -0.47, p < 0.05). These data indicate that paroxetine and fluvoxamine treatment with minimum clinically effective doses significantly inhibit CYP2D6 and CYP1A2, respectively. The extent of inhibition of CYP2D6 by paroxetine and of CYP1A2 by fluvoxamine is dependent in part on the baseline enzyme activity. The interindividual variability in CYP2D6 inhibition by paroxetine can also be explained by variability in plasma paroxetine concentration. Most patients treated with fluvoxamine (50-100 mg/day) will reach population minimums for CYP1A2 activity. These results have potential implications for interindividual variability in the risk for drug-drug interactions mediated by CYP2D6 and CYP1A2 as well as for the disposition of 17beta-estradiol and environmental procarcinogens.