Induction of cytochrome P450 3A (CYP3A) has been suggested as a mechanism of action of ursodeoxycholic acid (UDCA) in cholestasis. CYP3A is of key importance in human drug metabolism, being involved in presystemic extraction of more than 50% of all drugs currently available and of various endogenous compounds. Therefore, we compared the induction potential of UDCA with that of the prototypical inducer rifampicin in a human model study with the CYP3A substrates budesonide and cortisol. Twelve patients with early-stage primary biliary cirrhosis and eight healthy volunteers were treated with UDCA (15 mg/kg daily) for 3 weeks and subsequently with rifampicin (600 mg/d) for 1 week. Extensive pharmacokinetic profiling of oral budesonide (3 mg) was performed by determination of budesonide and phase I metabolites (6β-hydroxybudesonide, 16α-hydroxyprednisolone) in plasma and urine at baseline and at the end of each treatment. In parallel, urinary 6β-hydroxycortisol, a validated marker of CYP3A induction, was determined. UDCA did not affect biotransformation of budesonide and urinary excretion of 6β-hydroxycortisol either in patients or in healthy volunteers. Ratios of areas under plasma concentration-time curves (AUC0-12 h during UDCA/AUC0-12 h before UDCA) of both metabolites were not higher than those of budesonide itself. In contrast, administration of rifampicin markedly induced CYP3A metabolism, resulting in abolished budesonide plasma levels and high urinary excretion of 6β-hydroxycortisol. Metabolite formation was enhanced by rifampicin, but not by UDCA (e.g., AUC16α-hydroxyprednisolone/AUCbudesonide in patients: baseline, 8.6 ± 3.9; UDCA, 10.7 ± 7.1; rifampicin, 527.0 ± 248.7). In conclusion, UDCA is not a relevant inducer of CYP3A enzymes in humans. (HEPATOLOGY 2005;41:595–602.)
Ursodeoxycholic acid (UDCA) is a dihydroxy bile acid (3α,7β-dihydroxy-5β-cholanic acid) that is used for the treatment of a variety of chronic cholestatic liver diseases.1–3 UDCA is the only drug approved by the US Food and Drug Administration for the treatment of primary biliary cirrhosis (PBC). In PBC, UDCA has been shown to improve serum liver test results,4–9 to delay histological progression,9, 10 and to prolong transplant-free survival,11, 12 although the latter aspect has become a matter of discussion.13–15 Potential mechanisms of action of UDCA in cholestatic syndromes increasingly have been unraveled during the past decade and include (1) protection of injured cholangiocytes against toxic effects of bile acids, (2) stimulation of impaired biliary secretion, and (3) antiapoptotic effects.1–3 More recently, stimulation of detoxification of hydrophobic bile acids via induction of cytochrome P450 3A (CYP3A) has been discussed as an additional mechanism of action of UDCA in cholestasis.
CYP3A enzymes are the most abundant and predominant drug-metabolizing enzymes in humans.16 The CYP3A subfamily is responsible for biotransformation of more than 50% of drugs in clinical use, including macrolide antibiotics, benzodiazepines, calcium channel blockers, immunosuppressants, and chemotherapeutic agents.17 These enzymes are expressed in various tissues. In humans, approximately 60% of the total hepatic cytochrome P450 enzymes belong to the CYP3A subfamily.18 In adults, the CYP3A subfamily consists of CYP3A4, which is expressed in all livers, and polymorphic CYP3A5, which is expressed in approximately 25% of livers.19 Moreover, CYP3A is the principle cytochrome P450 enzyme found in the small intestinal epithelia of human adults.20
UDCA was first shown to induce hepatic CYP3A in murine liver microsomes.21, 22 In a human hepatoma cell line, UDCA activated the nuclear pregnane X receptor/steroid and xenobiotic receptor, a known transcriptional modulator of CYP3A expression, more efficaciously than most other bile acids, including cholic acid, deoxycholic acid, and the taurine conjugate of UDCA.23 In primary human hepatocytes, a 10-fold increase in activity of CYP3A-dependent testosterone 6β-hydroxylase over control was reported for UDCA (100 μmol/L) in comparison with a substantially lower effect of several other bile acids and with an almost 20-fold increase for rifampicin (10 μmol/L).23 In four patients with gallstone disease, UDCA treatment (1000 mg/d for 3 weeks) increased plasma levels of 4β-hydroxycholesterol by 45% (23.0 ± 3.8 ng/mL vs. 33.6 ± 6.8 ng/mL).24 Because insect cell microsomes with recombinant human CYP3A4 converted cholesterol into 4β-hydroxycholesterol, that observation has been interpreted as induction of CYP3A4 metabolism. However, the effect of UDCA on CYP3A has not been studied in a comprehensive human model study, so that the clinical relevance of these preliminary observations remains unclear.
Therefore, we initiated a clinical trial to study the effect of UDCA on human CYP3A metabolism in patients with early stage PBC as well as in healthy volunteers. We analyzed CYP3A metabolism by (1) CYP3A-dependent25 pharmacokinetics of oral budesonide, a glucocorticosteroid recently being suggested for combined medical treatment of early-stage PBC,26, 27 and by (2) urinary 6β-hydroxycortisol, a noninvasive, validated marker of CYP3A induction.28
UDCA, ursodeoxycholic acid; PBC, primary biliary cirrhosis; CYP3A, cytochrome P450 3A; Cmax, peak plasma concentration; AUC, area under the plasma concentration-time curve; AUC0-12 h, area under the plasma concentration-time curve over the course of 12 hours; t1/2, terminal elimination half-life; CLoral, apparent oral clearance.
Patients and Methods
Twelve patients with early-stage PBC (11 women and 1 man; mean age, 51.8 ± 10.4 yr; mean weight, 62.7 ± 11.3 kg; mean Mayo risk score, 4.1 ± 0.7; mean serum bilirubin level, 0.69 ± 0.32 mg/dL [normal, <1.00]; mean alkaline phosphatase level, 247 ± 251 U/L [normal, <135]; mean alanine aminotransferase level, 91 ± 114 U/L [normal, <35]) and 8 healthy volunteers (all men; mean age, 44.4 ± 8.6 yr; mean weight, 78.3 ± 5.0 kg) were enrolled in a controlled clinical trial. All individuals were white persons with a body mass index of 18 to 30 kg/m2 and a body weight range of 50 to 100 kg. Inclusion criteria in PBC patients were (1) positive antimitochondrial antibody test results, (2) an alkaline phosphatase or γ-glutamyltransferase level of more than 1.5-fold the upper limit of normal at initial diagnosis, and (3) histologically proven noncirrhotic liver disease compatible with PBC stages I-II within 5 years before inclusion and no morphological signs of portal hypertension on abdominal ultrasound at study entry. Exclusion criteria in patients and healthy volunteers were (1) cardiac, renal, gastrointestinal, and other findings that may interfere with the tolerability or pharmacokinetics of the study drugs; (2) excessive alcohol consumption (>35 g/d in men and >25 g/d in women); (3) smoking in healthy volunteers or heavy smoking in patients (>10 cigarettes/d); (4) human immunodeficiency virus or hepatitis C virus infection; (5) administration of glucocorticosteroids within 6 weeks before the first study day; (6) use of drugs during the 4 weeks before the first administration or during the trial that may influence biotransformation of budesonide29; (7) intake of grapefruit or grapefruit juice within 1 week before administration of the first study drug; and (8) administration of agents interfering with gastrointestinal absorption and motility within the 3 days before the start of the study. Patients were not eligible to participate if they displayed any hepatic disease other than PBC. Intake of UDCA in the 6 weeks before the first study day precluded participation. Furthermore, healthy volunteers were excluded if they had taken any medication within the 2 weeks before or during the conduct of the study.
In an open-label study with fixed treatment order, a single oral dose of 3 mg budesonide was administered on study day 1. Thereafter, each patient was treated with UDCA capsules (15 mg/kg daily, divided into a morning and an evening dose) for 3 weeks (days 2-22) and subsequently with rifampicin dragées (600 mg/d, evening dose) for 1 week (days 23-29). Administration of 3 mg budesonide was repeated on study days 22 and 29. Pharmacokinetic profiling of budesonide was performed identically on the three study days (days 1, 22, and 29) by determination of budesonide and its phase I metabolites in plasma and urine before and during the 12 hours after drug intake. Volunteers were hospitalized the evening before each study day. After an overnight fast, a standardized lunch was served 4 hours after ingestion of budesonide. Blood samples were collected just before and 2, 2.5, 3, 3.5, 4, 4.5, 5, 5.5, 6, 7, 8, 10, and 12 hours after administration of budesonide. Blood samples (ethylenediaminetetraacetic acid tubes) were centrifuged immediately, and plasma was stored at −20°C until analysis. Before receiving budesonide, each participant emptied the bladder to collect urine during the interval 0 to 12 hours after dosing. Aliquots of urine also were frozen at −20°C until analysis. On each study day urinary 6β-hydroxycortisol, a noninvasive marker of CYP3A induction,28 was measured. Additionally, blood samples for bile acids were obtained. To ensure compliance during this trial, each healthy volunteer had to report to the study site for each drug intake (mouth check). The patients were called by telephone by the investigator. Furthermore, patients had to use a dispenser and fill in a medication diary. The study protocol was approved by the local medical ethics committees. The study was conducted in accordance with the ethical guidelines of the Declaration of Helsinki and the International Conference on Harmonization guidelines for good clinical practice. Written informed consent was obtained from each participant.
Concentrations of budesonide, 6β-hydroxybudesonide, and 16α-hydroxyprednisolone in plasma and urine were determined by validated liquid chromatography tandem mass spectrometry. After extraction from the matrix, budesonide and its metabolites were quantified using a triple-stage mass spectrometer SCIEX API III PLUS (SCIEX; Thornhill, Ontario, Canada). The chromatography column was coupled using a heated nebulizer interface to an atmospheric pressure ionization chamber of the mass spectrometer. For determination of 16α-hydroxyprednisolone, a turbo ion spray interface was used instead. The lower limit of quantification in plasma (urine) was 0.1 ng/mL (0.5 ng/mL) for budesonide and 6β-hydroxybudesonide and 0.4 ng/mL (2 ng/mL) for 16α-hydroxyprednisolone. Between-day and within-day coefficients of variation of quality controls were less than 15%. Determination of 6β-hydroxycortisol in urine was performed by a validated high-performance liquid chromatography method with ultraviolet detection.30 The lower limit of quantification was 20.4 ng/mL. Concentrations of bile acids in serum were measured by capillary gas chromatography.31 UDCA levels were expressed as a percentage of total serum bile acids.
Budesonide and Metabolite Kinetics.
Standard model-independent methods were used to determine the pharmacokinetic parameters of interest (WinNonlin version 3.1A; Pharsight Corp., Mountain View, CA). Peak plasma concentration (Cmax), and time of Cmax were taken directly from the plasma concentration-time curves. Area under the plasma concentration-time curve over the course of 12 hours (AUC0-12 h) was determined by a combination of linear and logarithmic trapezoidal methods with extrapolation to infinity (AUC0-∞). Terminal elimination half-life (t1/2) was calculated from the final slope of the log-linear concentration-time curve by least-squares linear regression. Apparent oral clearance (CLoral) of budesonide was derived from the equation CLoral = dose / AUC0-∞ and was normalized for body weight. AUC ratios (AUC0-12 h during induction / AUC0-12 h before induction) of budesonide and each metabolite, respectively, were calculated to characterize the extent of induction. Ratios of metabolite formation (AUCMet/AUCbudesonide, where Met is the metabolite) such as AUC0-12 h of 6β-hydroxybudesonide to AUC0-12 h of budesonide were used as indices of CYP3A activity. Urinary recoveries of the analytes were based on the cumulative amount of the analyte excreted during the 12-hour collection period (Ae0-12 h), and were expressed as a percentage of the budesonide dose administered. Apparent oral clearances of metabolites were calculated by dividing the amount of analyte excreted during the 12-hour collection period of the respective metabolite by AUC0-12 h of budesonide in plasma.
Urinary 6β-Hydroxycortisol and Pharmacodynamic Analysis.
Induction of CYP3A enzymes by UDCA and rifampicin was evaluated on study days 1, 22, and 29 by measuring the cumulative amount of 6β-hydroxycortisol excreted into urine over the course of 12 hours. In all patients, severity of fatigue and pruritus were assessed during hospitalization using a 40-question fatigue severity score (maximum score, 160) and an itch numeric rating scale (no pruritus, 1; maximum, 10).32
The study had an 80% power to detect a 20% difference in the plasma AUC of budesonide between study day 1 and study day 22 with a P value of less than .05 in each population. Sample size calculation was based on the coefficient of variation of the budesonide AUC in a previous study of patients with PBC stages I-II.27 The effect of UDCA and rifampicin on CYP3A metabolism was analyzed by use of the software package GraphPad Instat (Version 3.05, GraphPad Software, Inc., San Diego, CA). Differences between patients and healthy volunteers at baseline were tested for significance by the Mann-Whitney U test. To compare between study days, a parametric ANOVA was taken for all pharmacokinetic parameters, except that of time of Cmax and t1/2, for which a nonparametric ANOVA was chosen. Pharmacodynamic parameters were evaluated by nonparametric ANOVA. Results of parametric parameters are given as mean ± SD, and those of nonparametric parameters are given as median, with 95% CIs in parentheses. In addition, proving a lack or existence of any significant drug interaction was handled as an equivalence problem. Ninety percent CIs of the log-transformed parameters AUC0-12 h and Cmax of budesonide before and during induction were derived from the residual variance in multifactor ANOVA. Significant interaction was concluded if these 90% CIs were outside the equivalence range of 0.8 to 1.25 (AUC0-12 h) and 0.7 to 1.43 (Cmax), respectively.
All patients with early-stage PBC and all healthy volunteers completed the study according to the protocol with excellent compliance. Prior treatment with UDCA had been withdrawn over the course of 6 weeks in 8 of 12 patients. UDCA levels in serum were low on study day 1 (patients, 1.3% ± 2.9%; healthy volunteers, 4.6% ± 5.9%) and increased markedly until study day 22 (patients, 86.9% ± 4.2%; healthy volunteers, 80.8% ± 4.9%). On study day 29, after 1 week of UDCA washout, UDCA levels had not yet dropped to baseline values (patients, 25.9% ± 17.2%; healthy volunteers, 26.1% ± 5.6%).
Pharmacokinetics of Budesonide.
Pharmacokinetic parameters of the CYP3A probe drug, budesonide, are given in Table 1 for comparison of baseline with effects of UDCA or rifampicin. Plasma concentration-time curves of budesonide on the 3 consecutive study days are shown in Fig. 1. CYP3A metabolism at baseline was impaired in patients with early-stage PBC when compared with that of healthy volunteers, with mean CLoral of budesonide in patients being 63% of that of healthy volunteers, and mean apparent oral clearances of metabolites being 53% (6β-hydroxybudesonide) and 47% (16α-hydroxyprednisolone) of that of healthy volunteers, respectively. Administration of UDCA did not significantly affect metabolism of budesonide via CYP3A. AUC ratios (AUC0-12 h on day 22/AUC0-12 h on day 1) of both phase I metabolites of budesonide (e.g., in PBC patients, 6β-hydroxybudesonide, 0.87 ± 0.33; 16α-hydroxyprednisolone, 0.87 ± 0.32) were not significantly higher than that of budesonide itself (0.74 ± 0.25; Fig. 2). Metabolite formation was not significantly enhanced by UDCA (Table 2). Correspondingly, t1/2 of budesonide was not altered by intake of the bile acid. However, 90% CIs of both AUC0-12 h ratio and Cmax ratio of budesonide during UDCA relative to baseline were not completely within the equivalence range (0.63-0.96 and 0.53-110.4 for patients, respectively; 0.68-101.0 and 0.61-0.95 for healthy volunteers, respectively).
Table 1. Effect of 3 Weeks of UDCA (15 mg/kg daily) and 1 Week of Rifampicin (600 mg/d) on Pharmacokinetic Parameters of Budesonide (3 mg, Single Oral Dose) in Patients with Early-Stage Primary Biliary Cirrhosis and Healthy Subjects
Study Day 1: Baseline
Study Day 22: UDCA
Study Day 29: Rifampicin
NOTE. The parameters are given as mean ± SD or as median with 95% CI in parentheses.
Abbreviations: tmax, time of peak plasma concentration; CLoral, apparent oral clearance.
P < .001 vs. baseline (ANOVA).
P < .05.
P < .01 vs. patients (Mann–Whitney U test).
Patients with early-stage primary biliary cirrhosis (n = 12)
In contrast to UDCA, rifampicin induced the metabolism of budesonide dramatically, resulting in abolished parent drug plasma levels. In 9 of 12 patients and in 7 of 8 healthy volunteers, peak plasma concentration of budesonide was below the limit of quantification on study day 29. CLoral of budesonide was enhanced by rifampicin (P < .001) more than 100-fold in patients and more than 300-fold in healthy volunteers. Mean AUC ratios (AUC0-12 h on day 29/AUC0-12 h on day 1) of the two metabolites (e.g., for PBC patients, 6β-hydroxybudesonide, 0.10 ± 0.09; 16α-hydroxyprednisolone, 0.47 ± 0.22) were more than 25-fold and 100-fold, respectively, higher than that of budesonide (0.004 ± 0.01; Fig. 2). Ratios of metabolite formation increased substantially by rifampicin (P < .001; Table 2). Ninety percent CIs of both AUC0-12 h ratio and Cmax ratio of budesonide during rifampicin relative to baseline were far outside the equivalence range (e.g., for PBC patients, 0.003-0.06 and 0.003-0.19, respectively). At baseline, mean urinary recovery of budesonide (sum of budesonide and both metabolites) was 10% and 11% of the dose administered in patients and healthy volunteers, respectively. Rifampicin, but not UDCA, resulted in a trend to lower recovery of the metabolites in urine. Unchanged budesonide was not detectable in the urine of any subject.
Urinary 6β-Hydroxycortisol Excretion and Pharmacodynamic Effects.
UDCA treatment did not affect urinary excretion of 6β-hydroxycortisol in either group, whereas rifampicin treatment resulted in a steep increase of this marker of CYP3A induction (patients, P < .001; healthy volunteers, P < .01; Fig. 3).
In patients, severity of fatigue (median [95% CI]: day 1, 16 [10-51]; day 22, 17 [8-54]; day 29, 16 [8-57]) did not decrease after 3 weeks of UDCA or 1 week of rifampicin, and pruritus was not present in most patients (median [95% CI]: day 1, 1 [1-4]; day 22, 1 [1-3]; day 29, 1 [1-3]).
This human study shows that UDCA is not a clinically relevant inducer of CYP3A enzymes. In patients with early-stage PBC and in healthy volunteers, UDCA (15 mg/kg daily) did not significantly enhance biotransformation of oral budesonide and endogenous cortisol via CYP3A. t1/2 of budesonide in plasma reflecting velocity and extent of metabolism was not affected by UDCA. As shown by extensive analysis of pharmacokinetics in plasma and urine, treatment with the bile acid did not alter the formation of two CYP3A-dependent metabolites of budesonide. Accordingly, there was no effect on urinary excretion of 6β-hydroxycortisol, which is a validated test for evaluating CYP3A induction.28 In contrast, we characterized a new, clinically important drug interaction of budesonide and the enzyme inducer rifampicin. Plasma levels of budesonide were no more detectable after treatment with rifampicin. Formation of both 6β-hydroxybudesonide and 16α-hydroxyprednisolone was enhanced by rifampicin in parallel to markedly enhanced urinary secretion of 6β-hydroxycortisol, confirming budesonide to be a good probe drug of CYP3A.
Our results do not support experimental data mentioned above suggesting UDCA to be a relevant inducer of CYP3A metabolism.21–24 In particular, a recent report on increased plasma levels of 4β-hydroxycholesterol in four patients with gallstone disease24 has to be interpreted with caution: (1) plasma 4β-hydroxycholesterol has not been validated as marker of CYP3A induction in contrast to urinary 6β-hydroxycortisol; (2) plasma 4β-hydroxycholesterol levels were approximately eightfold higher in patients receiving antiepileptic drug treatment than those receiving UDCA treatment,24 questioning the clinical significance of the findings in UDCA-treated patients; and (3) adequate statistical analysis in only four volunteers seem difficult to achieve.
Patients with early-stage PBC showed impaired biotransformation via CYP3A enzymes when they were compared with volunteers without liver pathological features. This may be explained by reduced CYP3A activity in chronic liver disease,33 but not by sex differences in both subject groups. Although sex has been noted as determinant of CYP3A4 expression because of twofold higher CYP3A4 levels in female compared with male human liver samples, sex did not affect oral clearance of midazolam, a classical probe for intestinal and hepatic CYP3A, in vivo.34, 35 It may be speculated that plasma levels of budesonide would have been even lower in female healthy volunteers than in male healthy volunteers, further enlarging the difference between patients and healthy volunteers. The finding at baseline is important for physicians treating patients with early-stage PBC. It may be prudent to start CYP3A substrates with a narrow therapeutic index such as antiarrhythmics or benzodiazepines in reduced dosage to avoid toxic side effects in these patients. Unfortunately, detailed pharmacokinetic investigations with those drugs are not available in early cholestatic liver disease. Our previous pharmacokinetic study with budesonide revealed that budesonide should not be administered in late-stage PBC because of unpredicted side effects.27
Rifampicin is used as second-line therapy for controlling pruritus in PBC.36–38 It is a potent inducer of intestinal and hepatic CYP3A via activation of the pregnane X receptor/steroid and xenobiotic receptor.39 Activation of this nuclear receptor explains the coordinate induction of the intestinal drug transporter P-glycoprotein.40 Moreover, rifampicin induces CYP2C, phase II metabolizing enzymes, and multidrug resistance protein 2, and it inhibits the bile salt export pump.41 The greatest effects of rifampicin in drug interaction studies are shown with orally administered drugs, because the substances have to pass through gut and liver before reaching the systemic circulation.42 Biotransformation in gut wall contributes substantially to the overall first-pass metabolism of many CYP3A substrates.43 It is noteworthy that CYP3A4 protein content is approximately three times higher in enterocyte homogenates than in paired liver homogenates.44 Budesonide is a dual substrate of CYP3A and P-glycoprotein.25, 45 Therefore, the dramatic effect of rifampicin on budesonide with a reduction of the AUC by 99% to 100% may be explained by induction of both drug transport and biotransformation. In PBC stages I-III, expression of P-glycoprotein was not impaired when compared with levels in controls.46 However, relative effects of rifampicin on transport proteins and drug metabolizing enzymes cannot be differentiated because budesonide plasma levels were too low to characterize distribution and elimination of the substance. The effect of rifampicin on other corticosteroids such as prednisolone was reported to be much lower, with a reduction of the AUC of 60%.47 When midazolam, which is a substrate of CYP3A, but not of P-glycoprotein, was administered orally after a short course of rifampicin (600 mg/d), the AUC of midazolam was decreased by 98%.48 The AUC of oral digoxin—a substrate of P-glycoprotein but not of CYP3A—decreased by 42% after 2 weeks of rifampicin administration.49
In conclusion, the results of this human study indicate that UDCA is not a clinically relevant inducer of CYP3A metabolism, rendering modulation of CYP3A unlikely as a mechanism of action of UDCA in cholestatic liver disease. In contrast, rifampicin was confirmed to be a strong inducer of CYP3A in the cholestatic liver, resulting in loss of systemic bioavailability of the model substrate oral budesonide and probably of numerous other drugs that are metabolized via CYP3A.
The expert methodological and technical support by Dr. R. Pokorny, Dr. M. Seiberling, and Mrs. G. Schwertfeger is gratefully acknowledged.