Effects of Uptake and Efflux Transporter Inhibition on Erythromycin Breath Test Results

nature publishing group Effects of Uptake and Efflux Transporter Inhibition on Erythromycin Breath Test Results LA Frassetto 1, S Poon 2, C Tsourounis 2, C Valera 2 and LZ Benet 3 The erythromycin breath test (EBT) is a standard test used to evaluate the extent of CYP3A4 activity. This study examines whether presumed changes in CYP3A4 activity are in fact related to inhibition of an uptake organic anion transporter using rifampin and inhibition of the efflux hepatic P-glycoprotein transporter using lansoprazole. Three EBT tests in healthy adults were conducted: EBT alone, with lansoprazole, and with rifampin. For all subjects, lansoprazole treatment increased respiratory 14 Cexcretionby þ 0.2570.51 met/h (P ¼ 0.07) and rifampin decreased 14 Cexcretionby 0.4470.40 met/h (Po0.001) compared with baseline. Comparing lansoprazole to rifampin, 14 C excretion increased by þ 0.6970.50 met/h (Po0.001). Only women had significant changes after drug infusion: 14 C excretion after rifampin 0.4070.36 met/h (P ¼ 0.018) and þ 0.4770.44 met/h (P ¼ 0.018) after lansoprazole. Relying on EBT without considering transporter interactions can lead to errors in interpreting the degree of CYP3A4 metabolism. The cytochrome P450 (CYP450) family is the major metabolic enzyme system in the liver. CYP3A4 alone is responsible for the metabolism of more than 50% of all drugs sold in the US that are metabolized by the P450 system. 1 Therefore, drugs or chemical substances that cause CYP3A4 inhibition or induction are likely causes for drug drug interactions. The erythromycin breath test (EBT) is the most commonly used noninvasive phenotype probe for CYP3A4. 2 However, marked variability has been found when using EBT to predict metabolism of other CYP3A4 substrates and resulting drug drug interactions. 3 Although it has been recognized recently that a small portion of this variability may be attributed to the polymorphic expression of CYP3A4, we previously hypothesized that the lack of correlation is also owing to effects of the erythromycin uptake and efflux transporters, which alter the quantity of erythromycin available for metabolism. 4 Erythromycin metabolism in the liver initially requires uptake into the cells, which is mediated by organic anion transporting polypeptides (OATPs). 5 After injection, [ 14 C-Nmethyl] erythromycin is partially broken down by CYP3A4 to its N-demethylated metabolite and 14 C-formaldehyde. 14 C- formaldehyde is then predominantly converted to 14 C- formate and finally to radiolabeled 14 C-CO 2, which is exhaled through the lungs (ref. 6; Figure 1). The majority of the unmetabolized parent drug and the N-demethylated metabolites are excreted by the efflux transporter P- glycoprotein (P-gp). An increase in EBT 14 C-CO 2 excretion is considered to be a marker for CYP3A4 induction, whereas a decrease in 14 C- CO 2 excretion is considered to be a marker for CYP3A4 inhibition. But the ability of EBT to predict the metabolism of other CYP3A4 substrates may be compromised if these substrates inhibit uptake or efflux transporters, and could cause misleading interpretation of CYP enzyme activity. We have suggested previously 4 that although the EBT is a useful mechanistic probe, its ability to predict the disposition of CYP3A4 substrates in individual patients will not be successful as erythromycin is a good substrate for both uptake and efflux transporters. Previous studies have suggested that the interaction between membrane transporters and metabolizing enzymes are important in understanding drug metabolism and elimination. 7 10 Rifampin is a potent inhibitor for most of the OATP subtypes, but following single-dose administration has minimal induction effects on CYP3A4 and does not inhibit P-gp. For example, in vitro studies in rat hepatocytes have shown that when the liver uptake transporter OATP was inhibited by rifampin, without any effect on CYP3A4, metabolism of erythromycin was decreased. 6 Lansoprazole 1 Clinical Research Center and Department of Medicine, University of California at San Francisco, San Francisco, CA, USA; 2 Department of Clinical Pharmacy, University of California at San Francisco, San Francisco, CA, USA; 3 Department of Biopharmaceutical Sciences, University of California at San Francisco, San Francisco, CA, USA. Correspondence: LZ Benet (Leslie.Benet@ucsf.edu) Received 18 October 2006; accepted 21 January 2007; published online 14 March 2007. doi:10.1038/sj.clpt.6100148 828 VOLUME 81 NUMBER 6 JUNE 2007 www.nature.com/cpt

Basolateral membrane 14 C-erythromycin QATP QATP CYP3A4 Hepatocyte 14 C-erythromycin H 14 CHO N-demethyl erythromycin Apical membrane (biliary excretion) P-gp P-gp Table 1 Patient characteristics at baseline: mean7sd (range) Item All Males Females N 16 8 8 Age (years) 2574 (22 38) 2572 (22 28) 2776 (22 38) Race (C:AA:AP) a 9:2:5 6:1:1 3:1:4 BMI (kg/m 2 ) 2373 (21 30) 2473 (21 30) 2272 (19 24) a AA, African American; AP, Asian-Pacific Islander; BMI, body mass index; C, Caucasian. 14 CO 2 Pulmanory exhalation 14 CO 2 H 14 COOH Figure 1 Hepatocyte transporters and the EBT metabolic pathway; 14 C-CO 2 production from 14 C-erythromycin. is a modest P-gp inhibitor, which at the time of our study was believed to have minimal effects on OATPs and CYP3A4. 11 Increased metabolism of erythromycin in this case would not be due to induced CYP3A4 activity, but rather owing to inhibition of the efflux transporter. Therefore, based on in vitro data, EBT would not be a specific probe for CYP3A4 if transporters were not accounted for during the EBT. Following completion of the studies reported here, and just before submission of this manuscript, Kurnik et al. 12 reported that concomitant administration of the potent investigational P-gp inhibitor tariquidar significantly increased EBT values, as we had predicted. 4 We therefore predict that in humans, when the uptake transporter (OATP) activity is inhibited by rifampin, an apparent decreased CYP3A4 activity will be observed. Conversely, when the efflux transporter (P-gp) is inhibited by lansoprazole, an apparent increased CYP3A4 activity will be observed. Based on our results from in vitro studies, neither drug should actually alter CYP3A4 activity. A better understanding of the transporter effects on EBT would be helpful to evaluate its usefulness as a mechanistic CYP3A4 probe versus its use as a predictor of the disposition of other CYP3A4 substrates. RESULTS Subjects Sixteen healthy adult subjects (eight women and eight men) were enrolled (see Table 1 for patient characteristics). There were no significant differences in age or body mass index between gender groups in this study. No adverse effects or symptoms attributable to erythromycin, lansoprazole, or rifampin occurred. All subjects completed the study according to the study protocol. EBT baseline Among all subjects, the mean baseline EBT was 2.63.70.70 14 C met/h (range, 1.32 3.94 14 C met/h). EBT values were significantly higher for female subjects; 3.0470.53 14 C met/h (range, 2.56 3.94 14 C met/h) compared with male subjects; 2.2170.61 14 C met/h (range, 1.32-2.95 14 C met/h; P ¼ 0.014). The EBT changes from baseline following the administration of lansoprazole and rifampin are shown in Figure 2 and Table 2. Lansoprazole and rifampin The mean EBT value for the 16 subjects following lansoprazole administration increased to 2.88.70.92 14 C met/h (P ¼ 0.07). The mean EBT value for the 16 subjects following rifampin administration was 2.1870.65 14 C met/h; significantly lower than the mean baseline EBT 2.6370.70 14 C met/h (Po0.001). The mean difference between EBT results with rifampin and with lansoprazole was 0.6970.50 (Po0.001). Among the eight female subjects receiving lansoprazole, the mean EBT value was 3.5170.80 14 C met/h compared with the baseline EBT value 3.0470.53 14 C met/h (P ¼ 0.018; Figure 2 and Table 2). The eight male subjects receiving lansoprazole had a mean EBT of 2.2470.49 14 C met/h, minimally different from the baseline EBT of 2.2170.62 14 C met/h (P ¼ 0.89); four male subjects had increases and four had decreases in 14 C excretion. Among the eight female subjects receiving rifampin, 14 C excretion decreased 0.4070.36 14 C met/h (P ¼ 0.018) from baseline (Figure 2 and Table 2). The male subjects also had a decrease from baseline that did not achieve statistical significance of 0.4970.45 14 C met/h (P ¼ 0.07; Figure 2 and Table 2). DISCUSSION EBT is believed to provide a noninvasive and easily quantitative means for evaluating hepatic CYP3A activity. However, predictions regarding the clearance of drugs using EBT have been conflicting, with interindividual variability of up to 40%. 3 We hypothesized that discrepancies in previous reports might be owing to membrane transporter influences on drug exposure. 4 In fact, we were able to predict changes in EBT results based on prediction of inhibition of the erythromycin uptake transporter with rifampin and the efflux transporter with lansoprazole. Previous rat in vitro hepatocyte studies and in vivo pharmacokinetic studies with erythromycin demonstrated that erythromycin is a substrate of Oatp1a4 and Oatp1b2 6 CLINICAL PHARMACOLOGY & THERAPEUTICS VOLUME 81 NUMBER 6 JUNE 2007 829

%EBT (met/h) 0.4 0.2 0 0.2 0.4 0.6 0.8 1 1.2 1.50 1.00 0.50 0.00 0.50 1.50 Male EBT+rifampin EBT+lansoprazole Female Figure 2 Changes from baseline in 14 C-CO 2 counts with lansoprazole and rifampin (n ¼ 16). and that modulating uptake and efflux transporters affects intracellular erythromycin concentrations. 6,11 Erythromycin is a good substrate for Oatps; in the presence of rifampin, hepatic intracellular drug decreases 23% and blood erythromycin area under the curve in rats increases 150%. 10 Erythromycin is also a good substrate for P-gp; adding the P-gp inhibitor GG918 in vitro increases rat intrahepatic erythromycin content by 30% and leads to a doubling of the intracellular N-demethyl erythromycin metabolite/parent ratio. 10 In this study, we chose lansoprazole to inhibit P-gp because of its convenient intravenous (i.v.) formulation, minimal side effects, short half-life, and apparent inhibitor specificity. 11 It is approved for human use and was available through the hospital research pharmacy. In in vitro studies in Caco-2 and MDR-1 transfected LLC-PK1 cells, lansoprazole is a substrate, and a P-gp inhibitor. 11 Although we did not measure lansoprazole levels, published reports of i.v. lansoprazole C max levels after a 30-mg i.v. dose approach the IC 50. 13 Greater erythromycin metabolism with lansoprazole may be due to its inhibition of P-gp-mediated erythromycin excretion, leading to increased intracellular erythromycin available for 14 C-N-demethylation. The traditional explanation might suggest that lansoprazole is inducing CYP3A4. Conversely, decreased erythromycin metabolism with rifampin may be due to its inhibition of OATP-mediated erythromycin uptake, leading to less intracellular erythromycin available for 14 C-N-demethylation and lower 14 C-CO 2 compared to baseline, as was seen in this study. After repeated doses, rifampin is an inducer of CYP3A4; the traditional explanation would predict that 14 C-CO 2 should be higher. Although we did not observe a statistically significant difference in EBT values among all subjects receiving lansoprazole, we did observe a statistically significant difference in female subjects. In contrast, Kurnik et al., 12 in a very recent paper, found an increase in 14 C-erythromycin excretion in all eight Caucasian male subjects given the highly potent and selective investigational P-gp inhibitor tariquidar. The evidence of P-gp inhibition with tariquidar and the in vitro P-gp inhibition observed with lansoprazole, which did not translate equally in vivo, suggests that P-gp differences may account for the differences in EBT values that we observed among male and female subjects. Our study was not specifically powered to look at EBT variability between male and female subjects. However, sex effects on EBT were seen after lansoprazole and rifampin administration. Previous studies have examined sex disparities as one of the reasons for causing conflicting reports in EBT. Many have suggested sex differences with higher female metabolism and clearance compared with male subjects for CYP3A and P-gp substrates. 14,15 We hypothesized that literature data may be consistent with the supposition that sex differences in the clearance of CYP3A4 substrates may be related to differences in liver P-gp activity between male and female subjects, 1,13 as had been reported by Schuetz et al. 16 However, subsequent evaluation of liver bank measurements of CYP3A4 and P-gp reported no sex-related differences in P-gp, but rather higher levels of CYP3A4 in women. 17 Our Table 2 EBT at baseline and after rifampin and lansoprazole ( 14 C met/h) All P-value Males P-value Females P-value Baseline EBT 2.6370.70 2.2170.61 3.0470.53 D Rifampin from baseline 0.4470.40 o0.001 0.4970.45 0.077 0.4070.36 0.018 D Lansoprazole from baseline +0.2570.51 0.071 +0.0370.51 0.89 +0.4770.44 0.018 D Rifampin versus lansoprazole 0.6970.50 o0.001 0.5270.22 o0.001 0.8770.64 0.006 EBT, erythromycin breath test. 830 VOLUME 81 NUMBER 6 JUNE 2007 www.nature.com/cpt

present EBT results with lansoprazole versus rifampin suggest that sex-related differences in hepatic P-gp activity or response to inhibitors (possibly as opposed to P-gp hepatic amounts) should be re-examined in sex-related differences in the clearance of CYP3A4/P-gp dual substrates. Sex differences in hepatic OATP have been described in both human and animal models. These differences have been related to transporter mrna, protein expression, and the presence of gonadal hormones. 18 In our analysis, we found that both male and female subjects receiving rifampin had decreased 14 C-CO 2 production, and that the change did not reach statistical significance in male subjects (P ¼ 0.077), but did in female subjects (P ¼ 0.018). Overall, the decrease was highly significant in men and women (Po0.001). Limitations of the study One limitation of this study was the small sample size, which was calculated for the group as a whole. Based on these data, a larger group of male subjects would be needed to investigate the effects of various P-gp inhibitors on erythromycin metabolism during EBT. A second limitation of the study was that a concomitant CYP3A4 probe, not a substrate or inhibitor of an OATP or P- gp, was not given. Thus, it is possible that lansoprazole or rifampin could have altered CYP3A metabolism, independent of their effects on erythromycin uptake or efflux. However, 20 min is generally considered too short a time for enzyme upregulation or downregulation, although activation is a remote possibility. Subsequent to our clinical study, Maréchal et al. 19 reported that the IC 50 inhibition of CYP3A4 metabolism of a fluorescent probe, P4 BFC ¼ 7-benzyloxy- 4-trifluoromethylcoumarin, was less than that reported for P-gp inhibition, 11 although as stated earlier, our preliminary in vitro studies did not show lansoprazole to be an inhibitor of erythromycin metabolism. Furthermore, if lansoprazole were inhibiting erythromycin metabolism, we would have seen a decrease in 14 C-CO 2 production. We did not measure lansoprazole or rifampin levels in this study. The doses used were standard clinical doses for these medications. Also, we did not genotype the subjects enrolled in this study. Differences in metabolism owing to differences in transporter activity from polymorphisms may be important in evaluating EBT activity. But the purpose of this study was to show that transporter dynamics can affect EBT, which we have demonstrated without genotyping. CONCLUSION The EBT appears to be a very useful probe to evaluate mechanistic changes in CYP3A4 metabolism, used here to evaluate the effects of uptake and efflux transporters on CYP3A4 metabolism of erythromycin. We have shown in vivo in humans that inhibition of OATP decreases CYP3A4 metabolism and that P-pg inhibition increases CYP3A4 erythromycin metabolism, but only in female subjects. This latter finding raises again the question as to whether P-gp or CYP3A4 hepatic activity levels are responsible for the increased clearance of CYP3A4/P-gp substrates in female versus male subjects as also observed here with EBT, and suggests that P-gp sex differences may be responsible, as we have previously hypothesized. 1 However, the effects of transporters on EBT will compromise the ability of EBT to be a predictor of the clearance of other CYP3A4 drugs in patients. 4 MATERIALS AND METHODS Subjects. Healthy, nonsmoking subjects on no medications were recruited for this study. All subjects underwent a screening evaluation that included a medical history, physical exam, and blood and urine tests (hemoglobin b, hematocrit, white blood cells with differential, platelets, Na þ, K þ, HCO 3, Cl, AST, ALT, alkaline phosphatase, total bilirubin, glucose, creatinine, BUN, albumin, UA with microscopic analysis, and a serum pregnancy test for female subjects of child-bearing potential). The University of California San Francisco (UCSF) Institutional Review Board approved the study and all subjects provided written informed consent. Subjects who were sexually active consented to use two forms of birth control: condoms with each sexual encounter for men; cervical caps, diaphragms, and/or a nonhormonal intrauterine device for women. Any subject using other methods of birth control such as oral contraceptive pills, hormonal patches or vaginal rings, hormonal implants, injections, or hormonal intrauterine devices were excluded. Subjects were instructed to abstain from consuming grapefruit, grapefruit juice, and/or alcoholic beverages, over-thecounter medications and herbal products. Subjects were allowed to drink or eat caffeinated beverages/food except on study visit days. Study design. This was an open-label, randomized, crossover study. Each subject received an EBT at visits 1 3 after being admitted to the UCSF Clinical Research Center. During visit 1, the subjects received only an EBT. Each EBT was separated by at least 1 week. During visit 2, subjects were randomized to either an i.v. infusion of rifampin 600 mg or lansoprazole 30 mg over 30 min followed by an i.v. bolus of erythromycin via a separate i.v. catheter at the end of each infusion; the alternative solution was infused during visit 3. Intravenous rifampin and lansoprazole were prepared by the inpatient research pharmacy; rifampin was prepared as 600 mg in 100 ml D5W and lansoprazole was prepared in 100 ml D5W. Randomization was by odd even random number draw, adjusted so that for the 16 subjects, eight received one treatment first and eight received the alternate. Neither subjects nor study investigators were blinded to randomization. Erythromycin breath test: administration. Three micro Curies of [ 14 C-N-methyl] erythromycin alcohol (0.008 mg erythromycin in 0.5 ml alcohol) were mixed with 4.5 ml D5W just before administration. Each subject was slowly injected with this 14 C-erythromycin solution over 60 s. The breath samples were collected 20 min after the entire 14 C-erythromycin dose was injected. Patients were instructed to exhale normally into the collection balloon until the balloon was full; two balloons were collected at each test. Analysis. Each of the 20-min breath test samples obtained from study volunteers and their corresponding [ 14 C-N-methyl] erythromycin dose vials were sent to Metabolic Solutions (Amherst, NH) for analysis. Approximately 2 mmol 14 C-CO 2 were analyzed from each balloon. The radioactivity contained in the 20-min EBTs and the vials were measured with a Packard 2100TR Liquid Scintillation counter and expressed in units of disintegrations per minute (d.p.m.). CLINICAL PHARMACOLOGY & THERAPEUTICS VOLUME 81 NUMBER 6 JUNE 2007 831

The percentage of administered [ 14 C-N-methyl] erythromycin metabolized and exhaled as 14 C-CO 2 /min was calculated directly from the d.p.m. of the 20-min EBT. According to Metabolic Solutions, the EBT assumes an endogenous CO 2 production equal to 5 mmol CO 2 /m 2 body surface area (BSA)/min and 3 mci (or 6.66 10 6 d.p.m.) injected. The radioactivity administered must be corrected to account for residual amounts of [ 14 C-N-methyl] erythromycin left behind in the vials. In consideration of these factors, the following equation calculates the percent of the initial [ 14 C-N-methyl] erythromycin dose metabolized per minute from each subject s 20-min EBT: ½ 14 C-N-methylŠ erythromycin dose metabolized=minute ¼ ½ðdpm at 20 minþð5 mmol CO 2 =m 2 = min ½BSAðm 2 ÞŠÞ=2mmolCO 2 ð6:6610 6 100% d:p:m: left in vialþ The final EBT result was reported as the % [ 14 C-N-methyl] erythromycin exhaled per hour based on % [ 14 C-N-methyl] erythromycin dose metabolized per minute. This was done by using the validated equation below, developed by Wagner: 20 %½ 14 C-N-methylŠ erythromycin metabolized=h ¼ð 65:988x 2 þ 54:645x þ 0:0377Þ where x is the % dose metabolized/min after 20-min EBT. EBT interpretation. For all 16 study participants, three sets of EBT data were generated from the three GCRC study visits (visits 1 3). The EBT results for visits 2 and 3 after rifampin and lansoprazole exposure were compared with baseline activity measurements from visit 1. Repeated measures analysis of variance was used to evaluate the significance of experimental results with post hoc analyses carried out using paired t-tests (Sigmastat, San Rafael, CA). Data are reported as mean7sd. ACKNOWLEDGMENTS We acknowledge the help of the staff at the UCSF General Clinical Research Center (M01-00079), without which this study could not have been accomplished. This study was supported in part by NIH Grant HD40543. CONFLICT OF INTEREST The authors declared no conflict of interest. & 2007 American Society for Clinical Pharmacology and Therapeutics 1. Cummins, C.L., Wu, C.Y. & Benet, L.Z. Sex-related differences in the clearance of cytochrome P450 3A4 substrates may be caused by P-glycoprotein. Clin. Pharmacol. Ther. 75, 474 489 (2002). 2. Watkins, P.B. et al. Erythromycin breath test as an assay of glucocorticoid-inducible liver cytochromes P-450. J. Clin. Invest. 83, 688 697 (1989). 3. Rivory, L.P. Erythromycin breath test. Clin. Pharmacol. Ther. 70, 395 399 (2001). 4. Benet, L.Z. There are no useful CYP3A probes that quantitatively predict the in vivo kinetics of other CYP3A substrates and no expectation that one will be found. Mol. Interv. 5, 79 83 (2005). 5. Baker, A.L., Kotake, A.N. & Schoeller, D.A. Clinical utility of breath tests for the assessment of hepatic function. Semin. Liver Dis. 3, 318 329 (1983). 6. Sun, H., Huang, Y., Frassetto, L. & Benet, L.Z. Effects of uremic toxins on hepatic uptake and metabolism of erythromycin. Drug Metab. Dispos. 32, 1239 1246 (2004). 7. Wu, C.Y. & Benet, L.Z. Disposition of tacrolimus in isolated perfused rat liver: influence of troleandomycin, cyclosporine, and GG918. Drug Metab. Dispos. 31, 1292 1295 (2003). 8. Lau, Y.Y, Wu, C.Y., Okochi, H. & Benet, L.Z. Ex situ inhibition of hepatic uptake and efflux significantly changes metabolism: hepatic enzyme-transporter interplay. J. Pharmacol. Exp. Ther. 308, 1040 1045 (2003). 9. Lam, J.L. & Benet, L.Z. Hepatic microsome studies are insufficient to characterize in vivo hepatic metabolic clearance and metabolic drug drug interactions: studies of digoxin metabolism in primary rat hepatocytes versus microsomes. Drug Metab. Dispos. 32, 1311 1316 (2004). 10. Lam, J.L., Okochi, H., Huang, Y. & Benet, L.Z. In vitro and in vivo correlation of hepatic transporter effects on erythromycin metabolism: characterizing the importance of transporter-enzyme interplay. Drug Metab. Dispos. 34, 1336 1344 (2006). 11. Pauli-Magnus, C., Rekersbrink, S., Klotz, U. & Fromm, M.F. Interaction of omeprazole, lansoprazole and pantoprazole with P-glycoprotein. Naunyn-Schmiedeberg s Arch. Pharmacol. 364, 551 557 (2001). 12. Kurnik, D., Wood, A.J.J. & Wilkinson, G.R. The erythromycin breath test reflects P-glycoprotein function independently of cytochrome P450 3A activity. Clin. Pharmacol. Ther. 80, 228 234 (2006). 13. Freston, J.W. et al. Evaluation of the pharmacokinetics and pharmacodynamics of intravenous lansoprazole. Aliment. Pharmacol. Ther. 19, 1111 1122 (2004). 14. Meibohm, B., Beierle, I. & Derendorf, H. How important are gender differences in pharmacokinetics? Clin. Pharmacokinet. 41, 329 342 (2002). 15. Benet, L.Z., Cummins, C.L. & Wu, C.Y. Unmasking the dynamic interplay between efflux transporters and metabolic enzymes. Int. J. Pharm. 277, 3 9 (2004). 16. Schuetz, E.G., Furuya, K.N. & Schuetz, J.D. Interindividual variation in expression of P-glycoprotein in normal human liver and secondary hepatic neoplasms. J. Pharmacol. Exp. Ther. 275, 1011 1018 (1995). 17. Wolbold, R. et al. Sex is a major determinant of CYP3A4 expression in human liver. Hepatology 38, 978 988 (2003). 18. Morris, M.E., Lee, H.J. & Predko, L.M. Gender differences in the membrane transport of endogenous and exogenous compounds. Pharmacol. Rev. 55, 229 240 (2003). 19. Maréchal, J.-D. et al. In silico and in vitro screening for inhibition of cytochrome P450 CYP3A4 by comedications commonly used by patients with cancer. Drug Metab. Dispos. 34, 534 538 (2006). 20. Wagner, D. CYP3A4 and the erythromycin breath test. Clin. Pharmacol. Ther. 63, 129 130 (1998). 832 VOLUME 81 NUMBER 6 JUNE 2007 www.nature.com/cpt