Variation in oral clearance of saquinavir is predicted by CYP3A5*1 genotype but not by enterocyte content of cytochrome P450 3A5

Objective: Saquinavir, a widely prescribed human immunodeficiency virus I protease inhibitor, has a low and variable oral bioavailability that has been attributed to extensive first-pass extraction mediated by hepatic or intestinal cytochrome P450 (CYP) 3A4 and intestinal P-glycoprotein (P-gp). The polymorphic CYP3A5 has also been shown to influence the saquinavir metabolite/parent urinary ratio, suggesting a role for CYP3A5. Methods: Twenty healthy subjects received a single oral dose of saquinavir (600 mg) with water (control) and, on a separate occasion, with Seville orange juice (a selective intestinal CYP3A4/5 inhibitor). Hepatic CYP3A4 activity was evaluated by use of the erythromycin breath test. Duodenal biopsy specimens were used to assess relative intestinal CYP3A4 and CYP3A5 protein contents. Relative P-gp content was also assessed in the biopsy specimens and in lymphocytes. Genetic polymorphisms in MDR1 (in exon 21 and 26), CYP3A5 (*1 and *3), and CYP3A4*1B were identified by direct sequencing. Saquinavir plasma concentrations were measured by tandem liquid chromatography-mass spectrometry. Pharmacokinetic parameter estimates (maximum concentration, time to reach maximum concentration, area under the concentration-time curve, apparent oral clearance [CL/F]) were computed by standard noncompartmental methods. Stepwise multiple regression analysis was used to identify the hepatic or intestinal variables that predicted variation in saquinavir pharmacokinetic measures. Results: Baseline saquinavir CL/F was not correlated with liver CYP3A4 activity (the erythromycin breath test result), intestinal CYP3A4 content, or intestinal P-gp content (r(2) = 0.08, 0.08, and 0.007, respectively; P > .2). MDR1 genotype and lymphocyte P-gp content were also not predictive. Among the 6 subjects expressing intestinal CYP3A5, the mean saquinavir CL/F was almost twice as high as for the 14 nonexpressors (36.7 L/h [95% confidence interval (CI), 18.7-54.6 L/h] and 19.3 L/h [95% CI, 11.2-27.4 L/h], respectively; P = .03). However, among the 6 CYP3A5 expressors, there was an unexpected negative correlation between CL/F and intestinal CYP3A5 content (r(2) = 0.58, P = .05). Seville orange juice decreased the mean CL/F in all 20 subjects from 24.5 L/h (95% CI, 16.7-32.3 L/h) to 14.7 L/h (95% CI, 8.4-20.6 L/h) (P = .05). The effect size did not appear to be influenced by CYP3A5 expression. Conclusions. The CYP3A5*1 genotype is associated with increased saquinavir CL/F. This does not appear to reflect intestinal CYP3A5 expression and presumably reflects the contribution of hepatic CYP3A5. The interaction with Seville orange juice in subjects not expressing CYP3A5 supports a role for intestinal CYP3A4. However, the modest nature of the interaction, combined with the inability to detect a correlation between CL/F and CYP3A4 enterocyte content, supports our recent in vitro work suggesting a smaller contribution of intestinal CYP3A4 than has been assumed.