Abstract
Both the R(+) and the S(−) enantiomers of carvedilol were metabolized in human liver microsomes primarily to 4′- (4OHC) and 5′- (5OHC) hydroxyphenyl, 8-hydroxy carbazolyl (8OHC) andO-desmethyl (ODMC) derivatives. The S(−) enantiomer was metabolized faster than the R(+) enantiomer although the same P450 enzymes seemed to be involved in each case. A combination of multivariate correlation analysis, the use of selective inhibitors of P450, and microsomes from human lymphoblastoid cells expressing various human P450s enabled phenotyping of the enzymes involved in the oxidative metabolism of carvedilol. CYP2D6 was primarily responsible for 4OHC and 5OHC production, although considerable activity was observed in a CYP2D6 poor metabolizer liver and the variability of these activities across a human liver bank was not high. There was some evidence that CYP2E1, CYP2C9, and CYP3A4 were also involved in the production of these metabolites. CYP1A2 was primarily responsible for the 8OHC pathway with additional contributions from CYP3A4. The ODMC was clearly associated with CYP2C9 with some evidence for the partial involvement of CYP2D6, CYP1A2, and CYP2E1. With its complex P450 phenotype pattern and the known contribution of non-oxidative pathways of elimination, the activity (or lack of activity) of any particular P450 would have a limited influence on the disposition of carvedilol in an individual.
Carvedilol1(Coreg/Kredex, SmithKline Beecham Pharmaceuticals, Brentford, UK; Dilatrend/Eucardic, Boehringer Mannheim GmbH, Mannheim, Germany) is a β-adrenoceptor antagonist with vasodilating activity based on α1-blockade, available for the treatment of hypertension and congestive heart failure. This drug is used clinically as a racemic mixture of R(+)- and S(−)-enantiomers, and maximum plasma concentrations after oral administration of 50 mg racemic carvedilol were reported to be 74 ng/ml (0.18 μM) (R(+)-carvedilol) and 29 ng/ml (0.07 μM) (S(−)-carvedilol) (1).
Previous studies have shown that carvedilol is extensively metabolized in man giving products from both oxidation (fig.1) and conjugation pathways (2). It was clear that P450 mono-oxygenases play an important role in the oxidative biotransformation of carvedilol, and the objective of this study was to identify the specific enzymes involved.
The specific P450 enzyme(s) involved in the biotransformation of a drug can be the defining characteristic of its pharmacokinetic behavior, and drug interactions can occur when there is competition between two or more drugs for oxidation by the same P450 enzyme (3). Pharmacokinetic and clinical issues may also occur in individuals who are deficient in a specific P450 enzyme. For example, 5–10% of Caucasians are deficient in CYP2D6 and 3–5% of Caucasians and 20% of Far Eastern populations lack functional CYP2C19 (4). Thus, it is useful to delineate the involvement of particular P450 enzymes in metabolic pathways of new pharmaceuticals to explain and forecast interindividual pharmacokinetic variability and help predict drug-drug interactions.
To identify which P450 enzymes are involved in the in vitrometabolism of carvedilol, a combination of selective inhibitors of human P450 enzymes, correlation analysis, and heterologous expression systems was used. Chemicals characterized as inhibitors of individual P450 enzymes include furafylline, sulfaphenazole, quinidine, and ketoconazole which are selective inhibitors of CYP1A2 (5), CYP2C9 (6), CYP2D6 (7, 8) and CYP3A (6, 9), respectively. Correlation analysis with known activities using specific substrates for the P450 enzymes in a human liver bank can help determine which P450 enzymes are responsible for the metabolism of a compound (10). Lymphoblastoid cells transfected with human P450 cDNAs can provide a source of individual enzymes which can demonstrate which P450 enzymes can contribute to the metabolism of a compound.
Materials and Methods
Chemicals.
R(+)-carvedilol purity 98.10% and S(−)-carvedilol purity 99.87% and SK&F 108410 (BM 14.225) purity 99.54% (internal standard) were obtained from Chemical Development, SmithKline Beecham Pharmaceuticals. Metabolite standards 4′-hydroxyphenyl carvedilol (4OHC), 5′-hydroxyphenyl carvedilol (5OHC), 1-hydroxyphenyl carvedilol (1OHC), 8-hydroxy carbazolyl carvedilol (8OHC), and O-desmethyl carvedilol (ODMC) were supplied by Boehringer Mannheim GMBH, and were of purity greater than 94% (by HPLC). Ketoconazole and quinidine sulfate were obtained from Sigma Chemical Company (St Louis, MO). Sulfaphenazole and furafylline were obtained from Ultrafine Chemicals (Salford, UK).
All other reagents were purchased from BDH (Poole, Dorset, UK), Fisher Scientific Equipment Ltd. (Loughborough, UK), May and Baker (Dagenham, Essex, UK), or Sigma Chemical Company and were of the purest grade available.
Human Liver Tissue.
Samples of human livers (N = 27) were obtained from Vitron Inc. (Tucson, AZ) and the International Institute for the Advancement of Medicine (IIAM), Exton, PA. In all cases, the cause of death was not a result of any known biochemical deficiency in the liver, although several patients had taken or were administered drugs known to affect liver enzyme levels shortly before death. Microsomes derived from human B lymphoblastoid cells transfected with human cDNA for CYP1A1, CYP1A2, CYP2A6, CYP2B6, CYP2C9 co-expressed with P450 reductase, CYP2D6-Val, CYP3A4 co-expressed with P450 reductase, CYP2E1 co-expressed with P450 reductase, human P450 reductase, and control for native activity-containing vector were purchased from Gentest Corporation (Woburn, MA). Human liver microsomes were prepared by differential centrifugation.
Incubation of Carvedilol with Human Liver Microsomes and Microsomes from lymphoblastoid cells transfected with human P450 cDNA.
All incubations were carried out under similar conditions at 37°C in 50 mM potassium phosphate buffer (pH 7.4), using a NADPH generating system comprising NADP, glucose 6-phosphate, and glucose 6-phosphate dehydrogenase. Each incubation contained human liver microsomes at a final concentration of approximately 0.5 mg microsomal protein/ml or microsomes from lymphoblastoid cells transfected with human P450 cDNA at a final concentration of approximately 2 mg microsomal protein/ml except for CYP2D6 where 0.4 mg/ml was used. R(+)- or S(−)-carvedilol (final concentration of 0.2–400 μM) was solubilized with acetonitrile, and in a typical microsomal incubation the final concentration of acetonitrile did not exceed 2% (w/v). After a 5-min pre-incubation, the reaction was initiated by the addition of a pre-warmed NADPH-generating system. The reaction was terminated after 10 min by adding 250 μl acetonitrile containing 3uM SK&F 108410 (internal standard) and 1 mg/ml ascorbic acid to prevent breakdown of 1-hydroxy carvedilol. After centrifugation, the supernatant was removed and analyzed by HPLC.
HPLC of Carvedilol Incubations.
Incubates were analyzed on a Hewlett Packard 1090A or a Merck-Hitachi (Poole, Dorset, UK) L6200 HPLC system. Detection was by fluorescence using either a Hewlett Packard (Cheadle Heath, Stockport, Cheshire, UK) 1046A or a Perkin Elmer (Beaconsfield, Bucks, UK) LC 240 fluorescence detector at either λex278 nm, λem320 nm or λex330 nm, λem380 nm.
The HPLC method developed was based on that of Schaefer (11). Aliquots of each sample (50–100 μl) were injected onto a Supelco ABZ column (5 μm, 4.6 mm × 15 cm) maintained at a temperature of approximately 40°C with a flow rate of 1.0 ml·min-1. Elution conditions were a linear gradient of 75% solvent A (0.1 M ammonium acetate, pH 5.0):25% solvent B (acetonitrile:water (80:20 v:v) to either 60% A:40% B or 63.4% A: 36.6% B over 35 min, followed by a second linear gradient to 0% A:100% B at 37 min, followed by isocratic 100% B until 40 min and finally a linear gradient of 0% solvent A : 100% solvent B to 75% A:25% B until 45 min.
The fluorescent peaks of interest on the chromatogram were integrated and expressed as the area under each peak. Rates of formation of carvedilol metabolites were evaluated from calibration lines for each metabolite constructed by determining the peak area ratios of known concentrations of authentic standards at a constant concentration of the internal standard.
Inhibition of R(+)- and S(−)-Carvedilol Metabolism.
Inhibition of R(+)- and S(−)-carvedilol metabolism was investigated by incubating R(+)- or S(−)-carvedilol at a final concentration of 30 μM and approximately 0.5 mg protein/ml human liver microsomes in the absence and presence of each inhibitor. Quinidine (1 μM), selective for CYP2D6 (8), ketoconazole (1 μM), selective for CYP3A4 (6, 12), and sulfaphenazole (10 μM), selective for CYP2C9 (6,13-16), were preincubated at approximately 37°C with the R(+)- or S(−)-carvedilol, microsomes, and potassium phosphate buffer (pH 7.4) for 5 min, before adding a pre-warmed NADPH generating system to start the reaction. Furafylline (10 μM), selective for CYP1A2 (17), was preincubated at approximately 37°C with microsomes, NADPH generating system, and buffer for 10 min before prewarmed solutions of R(+)- or S(−)-carvedilol were added to start the reaction.
Data Analysis.
All graphical analysis of data was performed using nonlinear regression with weighted data (1/y) with GraFit version 3 (R. J. Leatherbarrow: Erithacus Software Ltd, Staines, UK (1992)).
Tests for statistical significance were performed using SAS/INSIGHT® Version 6 (SAS Institute Inc, Cary, NC). A multiple linear regression analysis was used to select a model to identify the cytochrome P450 enzyme(s) responsible for the variability in the metabolism of R(+)- and S(−)-carvedilol in 26 of the samples in the human liver microsomal bank. A multiple regression model was initially fitted using all known explanatory variables (cytochrome P450 enzyme activities). Type III F- or, equivalently, t-statistics were calculated for the exclusion of each explanatory variable from the model. The cytochrome P450 enzyme activity corresponding to the lowest value of the statistic was omitted if its p-value was greater than the 5% significance level and the change in the adjusted R2 value indicated that little explanatory power would be lost by its omission. TheF- or t-statistics were then calculated for the new model and again the explanatory variable corresponding to the lowest value of the statistic was excluded if its p-value was greater than the 5% significance level and little explanatory power would be lost by its exclusion. These steps were repeated until no other cytochrome P450 enzyme activity could be omitted.
Results
Microsomal preparations from each human liver used in this study were assayed for protein concentration, total cytochrome P450 concentration and the following enzyme activities; caffeine N3-demethylase (1A2), coumarin 7-hydroxylase (2A6), tolbutamide hydroxylase (2C9/8), S-mephenytoin 4-hydroxylase (2C19), bufuralol 1′-hydroxylase (2D6), lauric acid ω-1 hydroxylase (2E1), cyclosporine oxidase (3A) and lauric acid ω-hydroxylase (4A) (data not shown). This analysis confirmed a typical range of activities in this human liver bank, and these activities were used as the explanatory variables for the correlation analysis.
Fig. 2 illustrates the production of metabolites after incubation of 30 μM R(+)- or S(−)-carvedilol and human liver microsomes at approximately 37°C for 10 min. The four major metabolites were identified as 4′-hydroxy carvedilol (4OHC), 5′-hydroxy carvedilol (5OHC), 8-hydroxy carvedilol (8OHC), andO-desmethyl carvedilol (ODMC) by co-chromatography with known standards. Although 1OHC could not be seen consistently in the incubations with human liver microsomes (owing to analytical and stability issues), it was detected in several of the microsomes from the cell lines; additionally, it was observed with CYP2D6 and CYP1A2 cell line microsomes for both R(+)- and S(−)-carvedilol and with CYP1A1 for S(−)-carvedilol only. The production of these metabolites in incubations of R(+)- and S(−)-carvedilol with human liver microsomes was NADPH dependent and linear with respect to protein concentration and time. From this, it was estimated that a 10-min incubation at 0.5 mg protein/ml were suitable conditions.
The rates of production of 4OHC, 5OHC, 8OHC, and ODMC were evaluated following incubation of R(+) and S(−)-carvedilol at a range of concentrations (0.2–400 μM) with microsomes from three human livers. The plots were most consistent with a one enzyme Michaelis-Menten model (data not shown) despite the clear evidence of contributions from multiple enzymes from the other results. There was, however, considerable interindividual variation in the kinetic parameters, particularly the apparent Km (table 1).
From R(+)- and S(−)-carvedilol incubated at 50 μM (approximately half way between the median and the mode of the apparentKm values determined for all pathways in all human livers), the major metabolites, 4OHC, 5OHC, 8OHC, and ODMC were detected in most of the human livers investigated. The rates of production of 4OHC and 5OHC varied approximately 2- or 3-fold (fig. 3), but ODMC and 8OHC showed greater variability of approximately 5-fold and 10-fold, respectively. Generally, R(+)-carvedilol was metabolized at similar rates to 4OHC, 5OHC, and ODMC and at somewhat lower rates to 8OHC. For S(−)-carvedilol, the rank order of production being, in most livers, ODMC>4OHC>5OHC>8OHC. It was also apparent that S(−)-carvedilol was metabolized faster than R(+)-carvedilol (see figs. 3 and4).
The rates of R(+)- and S(−-)-carvedilol metabolism were correlated with the rates of the eight P450 related enzyme activities in human liver microsomes (table 2). A number of statistically significant correlations were observed. These correlations suggested the involvement of CYP2D6 in the 4′- and 5′-hydroxylation of both enantiomers of carvedilol. CYP2D6 was also implicated in the O-demethylation, although CYP2C9 was also associated with this pathway, more strongly than CYP2D6 in the case of S(−)-carvedilol. CYP2E1 was correlated with thisO-demethylation, albeit only for R(+)-carvedilol, and this enzyme also seemed to be involved in the 5′-hydroxylation of both enantiomers. CYP1A2 demonstrated a positive relationship with the 8-hydroxylation of carvedilol for S(−)-carvedilol only and correlations were also seen with CYP3A4 for this activity. The production of each metabolite from either R(+)- or S(−)-carvedilol was highly correlated (fig. 4).
Four selective inhibitors of human P450s were available for use in this study; quinidine (2D6), ketoconazole (3A), sulfaphenazole (2C9), and furafylline (1A2). There was considerable inter-individual variation in the effect of these inhibitors on each of the major activities, and the results from each of three livers are presented in table3. Quinidine had the greatest effects on 4OHC and 5OHC, with some weak effects on 8OHC and ODMC (R(+)- carvedilol only). Ketoconazole caused little inhibition of any of the activities, except for a limited effect on 8OHC. Sulfaphenazole caused the most marked effect on ODMC, with little significant effects on the other pathways. Furafylline inhibition was primarily limited to effects on 8OHC.
The metabolism of both R(+)-and S(−)-carvedilol was slow in microsomes derived from lymphoblastoid cells transfected with human cDNA for P450 enzymes. The rates of R(+)- and S(−)-carvedilol hydroxylation and O-demethylation by the major human P450s are shown in fig.5A. 4OHC, 5OHC, and ODMC were seen in incubations from a number of cell lines. CYP2D6 showed the most activity for both enantiomers, although CYP1A2, CYP2C9, and CYP3A4 all showed some activity to one or all of these metabolic routes. In addition, CYP1A1 and CYP2C19 were active with S(−)-carvedilol. In contrast, oxidation to 8OHC was only seen to a limited extent with CYP1A1 (S(−)-carvedilol only) and CYP1A2 (both enantiomers).
These cell lines do not express the P450s equally; for example, CYP2D6 was expressed at a level of activity in excess of that in an average human liver microsomal sample. In contrast, the CYP3A4 and CYP2C9 were expressed at lower levels than those observed in human liver microsomes. Scaling the activities with a relative activity factor (18) derived from human liver microsomes, the profile of contributions can be made more reflective of the situation in an average human liver (fig. 5B). In this case, CYP2D6 is still a major enzyme in the 4OHC and 5OHC pathways, but the potential contributions from CYP3A4 and the importance of CYP2C9 in the ODMC pathway of S(−)-carvedilol can be appreciated.
Discussion
The biotransformation of carvedilol in vivo is complex and many metabolites have been identified (2, 19). In most human liver microsomes, 4′- and 5′- hydroxyphenyl carvedilol (4OHC and 5OHC),O-desmethyl carvedilol (ODMC) and 8-hydroxy carbazolyl carvedilol (8OHC) were quantifiable metabolic products for both the R(+) and S(−) enantiomers of carvedilol. The 1-hydroxy carbazolyl carvedilol was not consistently quantifiable in all the incubations in this study, but it was detected in several of the expression systems and in some human liver microsomes. The use of three different approaches (correlation analysis, selective inhibition, and P450 expression systems) enabled identification of the human P450 enzymes responsible for each of these oxidative metabolic pathways of carvedilol.
Generally the metabolism of S(−)- carvedilol in vitro was more rapid than that of the R(+) enantiomer. This is consistent within vivo observations where the plasma concentrations of the R(+) enantiomer were generally higher on dosing the racemate (1). Despite this difference in rate, the majority of the evidence suggested that the same P450 enzymes were involved in the metabolism of each enantiomer.
The production of 4OHC and 5OHC seemed to be significantly dependent on CYP2D6. The rate of metabolism via these pathways was significantly associated with CYP2D6 in the correlation analysis, was inhibited by quinidine, and was readily detected in microsomes from cells transfected with CYP2D6 cDNA. However, the variability in this activity was not as great as is generally associated with CYP2D6 substrates. Of particular interest was the level of 4OHC and 5OHC produced by the liver encoded H103 which was approximately half that of the most active livers. H103 is genotypically a poor metabolizer of CYP2D6 (and CYP2C19), and phenotypically microsomes from this human liver have no detectable CYP2D6 activity. This result alone suggested that other P450 enzymes contribute to these hydroxylations. Multivariate correlation analysis implicated CYP2E1 and CYP2C9 in the production of 4OHC and 5OHC, and some inhibition of 4OHC was seen for S(−)-carvedilol with sulfaphenazole and furafylline, indicating a potential role for CYP2C9 and CYP1A2. From the expression systems a potential contribution from CYP3A4 was also evident for the 4OHC pathway. Overall therefore, it is clear that CYP2D6 was the primary enzyme involved in the 4OHC and 5OHC pathways, but other P450s make a significant contribution.
Although there were some limited effects of quinidine on the production of 8OHC, CYP2D6 did not appear to be involved in this pathway. In this case CYP1A2 was the most significant hepatic enzyme, with evidence of a CYP3A4 contribution being limited to the multivariate analysis. The metabolism by the CYP1A1 microsomes suggests that in vivothere could be an extrahepatic capability for this pathway.
The enzyme most clearly associated with the ODMC pathway was CYP2C9 although the incomplete inhibition with sulfaphenazole suggests the involvement of other P450s in this pathway. From the P450 expression systems and the correlation analysis data candidates include CYP2D6 and possibly CYP1A2 and CYP2E.
The three strands of evidence (correlation, inhibition, and expression) were not always entirely consistent but were generally supportive of one another. For the pathways that could be investigated in thisin vitro study CYP2D6, CYP2C9, and CYP1A2 were clearly the most important P450 enzymes. Thus, the enzymology of carvedilol metabolism is complex and the significance of any given P450 in the disposition of either enantiomer of carvedilol will depend on the balance of the P450 enzymes in an individual. Non-oxidative pathways may also be important determinants of carvedilol disposition since direct conjugation has also been described (2).
The significance of CYP2D6 in the disposition of carvedilol has been determined in vivo using poor and extensive metabolizers (19). The partial metabolic clearance of carvedilol to the 4OHC and 5OHC metabolites showed a significant reduction in the poor metabolizers; however, these metabolites were still formed in these individuals. The relatively modest difference in the plasma half life of S(−)-carvedilol (1.4-fold) and R(+)-carvedilol (1.8-fold) between the two groups was entirely consistent with only a partial role of this polymorphic enzyme in the overall elimination of carvedilol. Drugs that depend to a greater extent on CYP2D6 for their elimination can have more profound differences between poor and extensive metabolizers.. The differences in plasma half life between poor and extensive metabolizers for S(−)- and R(+)-metoprolol were 2.5- and 2.8-fold, respectively, (20) and 3.6-fold for desipramine (21).
With its complex P450 phenotype pattern and the known contribution of non-oxidative pathways of elimination, the activity (or lack of activity) of any particular P450 would have a limited influence on the disposition of carvedilol in an individual.
Acknowledgments
The authors gratefully acknowledge Paul Nelson of Statistical Sciences, SmithKline Beecham Pharmaceuticals for his contributions.
Footnotes
-
Send reprint requests to: Harriet G. Oldham, Ph.D., Drug Metabolism and Pharmacokinetics, SmithKline Beecham Pharmaceuticals, The Frythe, Welwyn, Hertfordshire AL6 9AR, England.
- Abbreviations used are::
- carvedilol
- 1-(9H-carbazol-4-yloxy)-3[[2-(2-methoxy phenoxy)ethyl]amino]-2-propanol
- P450
- cytochrome P450
- 4OHC
- 4′-hydroxyphenyl carvedilol
- 5OHC
- 5′-hydroxyphenyl carvedilol
- 1OHC
- 1-hydroxyphenyl carvedilol
- 8OHC
- 8-hydroxycarbazolyl carvedilol
- ODMC
- O-desmethyl carvedilol
- Received December 5, 1996.
- Accepted April 18, 1997.
- The American Society for Pharmacology and Experimental Therapeutics