Bioactivation and bioinactivation of drugs and drug metabolites: Relevance to adverse drug reactions

~)

Pergamon

Toxic. in Vitro Vol. 8, No. 4, pp. 613-621, 1994

0887-2333(94)E0053-V

Copyright © 1994ElsevierScienceLtd Printed in Great Britain.All rights reserved 0887-2333/94$7.00+ 0.00

Session 2 BIOACTIVATION AND BIOINACTIVATION OF DRUGS AND D R U G METABOLITES: RELEVANCE TO ADVERSE D R U G REACTIONS B. K. PARK*, M. PIRMOHAMED,M. D. TINGLE,S. MADDEN and N. R. KITTERINGHAM Department of Pharmacology and Therapeutics, University of Liverpool, PO Box 147, Liverpool, Merseyside L69 3BX, UK Summary--Adverse drug reactions that cannot be predicted from the pharmacological properties of the drug and which are not easily reproduced in laboratory animals are a major complication of drug therapy. It is necessary to investigate the mechanisms of such reactions in order to (1) define structural features within a given drug moleculewhich are responsiblefor causing toxicity and (2) to identify those individuals who are particularly sensitiveto a givendrug reaction. In theory, drug toxicity may arise by direct toxicity, genotoxicity or immune-mediated toxicity caused by either parent drug or chemical. In this respect chemically reactive metabolites are of particular importance and the balance between bioactivation and bioinactivation pathways of drug metabolism will be a critical factor in both the type and extent of toxicity. We have therefore developed in vitro techniques that incorporate human cells for the detection and characterization of stable, chemically reactive and cytotoxic metabolites. In such experiments bioactivation (by CYP1A, CYP2D6, CYP3A, etc.) can be investigated by use of a liver bank, while lymphocytes provide accessible human cells, which can be obtained from both patients and volunteers, genotyped and/or phenotyped for particular drug-metabolizing enzymes (eg. glutathione transferase/~). The relevance of in vitro experiments to drug toxicity observed in humans will be illustrated by reference to studies with anticonvulsants and antimalarials.

Introduction Adverse drug reactions represent a large clinical problem and account for a great deal of patient morbidity and significant patient mortality, and therefore deny certain patient groups effective drug therapy. It has been estimated that 2-3% of hospital patients are admitted as a direct result of an adverse drug reaction, while 10-20% of patients develop some form of drug toxicity prolonging their stay in hospital (D'Arcy, 1986). Adverse reactions to drugs may take many different forms ranging from effects on a single physiological/biochemical parameter to multiple organ failure, and can be classified from a clinical perspective as follows (Park et al., 1992): Type A: These reactions are predictable in terms of the known pharmacology of the drug and are usually dose dependent. Examples of this type of reaction include hypoglycaemia with oral hypoglycaemics, hypotension with anti-hypertensives and bleeding with anticoagulants. Such reactions should be anticipated and can often be eliminated by dose reduction. Type B (idiosyncratic reactions): These are unpredictable from a knowledge of the basic pharmacology

*To whom correspondence should be addressed.

of the drug and do not show any simple dose-response relationship, in that is there is lack of correlation between the dose and risk of toxicity. These reactions occur in only a small percentage of the population, but are often serious and account for many drug-induced deaths, and yet the mechanisms are poorly understood. Such reactions are not detected by preclinical toxicology testing in animals and cannot usually be reproduced in animal models. Type C: Reactions associated with long-term drug therapy, examples of which include benzodiazepine dependence and analgesic nephropathy. These reactions are well described and can be anticipated. Type D: Delayed effects such as carcinogenicity and teratogenicity. It is thought that such toxicities are precluded by the extensive programme of preclinical mutagenicity and carcinogenicity studies that a new chemical entity must undergo before a product licence is granted. Although such a classification is useful from a clinical viewpoint in that it may help with the diagnosis of an adverse drug reaction, it does not necessarily provide any understanding of the underlying mechanisms involved. Broadly speaking the mechanisms involved in adverse reactions can be divided into two types:

613

B. K. PARKet al.

614

Pharmacological: These can be rationalized in terms of either the primary or secondary pharmacology of the drug (and its metabolites) and are usually dose dependent and reversible. Chemical: These can be rationalized in terms of either the physicochemical properties or chemical reactivity of the drug or (more commonly) drug metabolite. Such reactions are dependent on the relative rates of accumulation and clearance of the toxic species and do not necessarily show a simple relationship with dose. During the past 20 years our understanding of drug toxicity has been increased by an appreciation of the role of drug metabolism in various toxicological processes. Much of this knowledge is based on animal work, but more recently evidence has been forthcoming from both in vitro and in vivo human studies. Both the rate and route of drug metabolism may influence the type and frequency of toxicity. The rate of drug metabolism will influence the net accumulation of a drug at its various sites of action and is therefore likely to be a common factor in both pharmacological and chemical toxicities. The route(s) of metabolism will determine the nature of chemical entities that are present in both target and non-target tissues and will therefore be an important determinant in chemical toxicities. The balance between bioactivation and bioinactivation may play a central role in the relationship between drug metabolism and drug toxicity. Of particular importance in this respect metabolites chemically reactive are short-lived (Park et a/., 1992) which, if not rapidly detoxified, can react with essential macromolecules such as proteins and nucleic acids and could, in theory, cause various types of chemical toxicity, including cellular necrosis teratogenicity, carcinogenicity, and hypersensitivity (Fig. 1). Because of the complexities of human metabolism such reactions may affect only a subsection of the population and may not be predictable

from preclinical

evalu-

ation.

I cytotoxicity

gemtoxicity

1 sensitisation

Fig. 1. Scheme illustrating the relationship between drug metabolism and toxicity. The central role of chemically reactive metabolites is indicated and it is the balance between their formation (bioactivation) and removal (bioinactivation) that determines the degree of toxicity observed.

Use of in vitro test systems to explore mechanisms of adverse drug reactions There is increasing interest in the use of mammalian cells for in vitro toxicity assays as alternatives to conventional animal tests. There are many examples of this type of assay that can use endpoints that reflect interference with cell proliferation, altered cell function or simply a loss of cell integrity. Such tests may have the disadvantage that they do not contain the capacity for oxidative metabolism that is required for the bioactivation of many compounds. In bacterial mutagenicity assays, this shortcoming has been overcome by the addition of liver homogenate 9000g supernatant, prepared from Aroclorpretreated rats, and an NADPH-regenerating system (Maron and Ames, 1983). To investigate the hypothesis that chemically reactive metabolites are indeed responsible for certain adverse drug reactions in humans we have used in vitro test systems based largely on human cells and tissues. We have developed methods with which to explore the chemical mechanisms of drug toxicity in order to: (1) determine the functional group(s) within the molecule that are responsible for the toxicity; and (2) identify those (biological) factors that predispose certain individuals to drug toxicity and may also determine cell-directed toxicity. To achieve these goals we have incorporated certain operational characteristics into the test systems wherever possible. These are: (1) full chemical analysis (quantitative and qualitative) of drug and drug metabolites, including both stable and unstable (chemically reactive) metabolites; (2) use of human cells and tissues that are genotyped and phenotyped for variation in enzyme levels/activity. For this purpose we routinely use a genotyped human liver bank for the investigation of bioactivation, although more recently human lymphoblastoid cell lines have been used additionally for this purpose. Drug activation has also been investigated in peripheral blood cells, such as polymorphonuclear leucocytes, where appropriate; (3) peripheral blood cells are used as targets for drug toxicity. Such cells may simply be used as a convenient target cell in cytotoxicity assays or may be used as specific target cells in the investigation of drug-induced blood dyscrasias. Cells may be obtained from volunteers when dealing with chemical aspects of drug toxicity or from patients when attempting to analyse host factors involved in toxicity. Human lymphocytes are widely used as target cells in toxicity assays because they are readily available cells that are robust and can therefore be easily isolated. handled and incorporated into metabolism/toxicity assay systems. We have used lymphocytes for two purposes. First, simply as target cells that contain low levels of drug activating enzymes (e.g. cytochrome P-450) but relatively high levels of enzymes involved in drug (metabolite) bioinactivation (e.g. epoxide hydrolase and glutathione transferase p), which

Bioactivation and bioinactivation of drugs

=J

GENOTYPEDHUMAN LIVERBANK

RADIOLABELLED oRuG

0'

¢

microsomes

Isolated blood cells

615

b

TOXIC METABOLITES

% STABLE REACTIVE METABOLITESMETABOLITES HPLC

cytotoxicity genotoxicity

covalent binding

Fig. 2. Generalized scheme for investigation of the mechanisms of adverse drug reaction in vitro. can be characterized by both phenotyping and genotyping. Secondly, as peripheral markers of individual risk factors (pharmacological, metabolic, immunological) in patients who have suffered an adverse drug reaction; (4) a functional measure of toxicity based on cell viability. In assays where lymphocytes have been used as target cells, cell death serves as an indirect marker for the formation of chemically reactive metabolites. Cell viability may be measured by, for example, the ability of the cell to exclude supravital stains, such as trypan blue. Figure 2 shows a generalized scheme of the experimental approach adopted. This approach will be illustrated with reference to the mechanisms of a number of diverse adverse drug reactions that we have investigated recently.

Taerine hepatotoxicity The cognition activator tacrine (1,2,3,4-tetrahydro9-amino acridine) has been shown to cause dosedependent and reversible elevations in serum transaminase levels in 40-50% of patients undergoing therapy (Farlow et al., 1992; Forsyth et al., 1989; O'Brien et al., 1991), although frank hepatotoxicity such as hepatitis and jaundice is rare (Farlow et al., 1992; O'Brien et al., 199l). Although the mechanism of tacrine hepatotoxicity in vivo is unknown it is unlikely to be caused by tacrine directly since it is known to undergo rapid and extensive metabolism (Pool et al., 1992). A number of drugs are known to cause hepatotoxicity by mechanisms that involve the formation of chemically reactive metabolites; they include paracetamol, sodium valproate, halothane and alcohol. In vitro investigations have demonstrated the potential for tacrine to undergo metabolic activation to chemically reactive and cytotoxic metabolites (Madden et al., 1993; W o o l f e t al., 1993). Using peripheral blood lymphocytes as target cells and human liver microsomes as the metabolitegenerating system, NADPH-dependent formation of cytotoxic metabolites from tacrine has been demon-

strated. Similarly, in the presence of both N A D P H and liver microsomes, tacrine was metabolized to a species that remained irreversibly bound to the microsomai protein, which is indicative of the generation of chemically reactive species (Madden et al., 1993). By the addition of chemical modifiers to the in vitro systems it was possible to gain further information as to the nature of the metabolites being generated. Incorporation of the metabolic modifiers reduced glutathione and ascorbic acid, resulted in inhibition of the formation of both cytotoxic and protein-reactive metabolites. Furthermore, it was found that sulfhydryl but not amine nucleophiles inhibited covalent binding. Coupled with the lack of inhibition seen in the presence of the epoxide hydrolase inhibitor cyclohexene oxide we were able to conclude that the reactive metabolites generated from tacrine were electrophilic in nature and that they were unlikely to be epoxides (Madden et al., 1993). The nature of the enzyme involved in the oxidative biotransformations of tacrine was determined by an investigation of the effects of cytochrome P-450 (CYP) isoenzyme specific inhibitors on the bioactivation and metabolism of tacrine. The greatest inhibitory effect was seen in the presence of the CYPIA2 inhibitor enoxacin. Thus, it would seem that CYP1A2 is the predominant isoform of cytochrome P-450 involved in the bioactivation (and indeed overall metabolism) of tacrine. Throughout the course of these investigations we noted a strong correlation between the formation of 7-hydroxytacrine and the formation of reactive metabolites. Furthermore, when 7-hydroxytacrine was incubated with hepatic microsomes, generation of reactive metabolite(s) was the only route of metabolism evident. From these results we have postulated that tacrine undergoes sequential metabolism to form the reactive metabolites, with 7-hydroxylation being the initial step. Reactive metabolites are often difficult to identify as a result of their inherent chemical instability, and

616

B.K. PARK et al.

therefore indirect means have to be used to assess their formation. One such method involves the use of thiol-containing nucleophiles to act as trapping agents. In our system we have used three such nucleophiles, glutathione, N-acetyl cysteine and mercaptoethanol, in mierosomal incubations. All three thiols reduced covalent binding associated with both tacrine and 7-hydroxytacrine; however, only mercaptoethanol had any effect on the metabolic profile. In incubations with either tacrine or 7-hydroxytacrine in the presence of mercaptoethanol, an additional metabolite was generated, postulated to be a thioether adduct. Formation of this metabolite showed a close stoichiometric relationship with the observed decrease in covalent binding. We have therefore proposed that tacrine undergoes CYP1A2dependent hydroxylation at the 7 position to form 7-hydroxytacrine. This is followed by a 2-electron oxidation (again mediated by CYPIA2) to yield a reactive quinone methide. Reactive quinone methides are generated from a variety of anticancer drugs that are known to cause redox cycling and generation of oxygen radicals. In fact the cardiotoxicity associated with anthracyclines is thought to result from the inability of the myocytes to defend themselves against this redox cycling (Doroshow et al., 1980). One measure of redox cycling is a depletion in the levels of cellular glutathione. Dogterom et al. (1988) have demonstrated cytotoxicity and depletion of glutathione in rat hepatocytes incubated with tacrine. Thus, quinone methide-induced redox cycling may also play a role in tacrine-associated toxicity.

Dapsone haemotoxieity Dapsone (4,4'-diamino diphenyl sulfone) is used to treat a wide variety of diseases, including leprosy, inflammatory disorders involving polymorphonuclear leucocyte infiltration, rheumatoid arthritis, dermatitis herpetiformis as well as Pneumocystis carinii and Toxoplasma gondii. Administration of dapsone is associated with both dose-dependent toxicity towards red cells, usually in the form of methaemoglobinaemia, and idiosyncratic white cell toxicity, such as agranulocytosis (Zuidema et al., 1986). The toxic effects of dapsone are thought to be mediated by way of its hydroxylamine metabolite, formed either by cytochrome P-450- or myeloperoxidase-catalysed oxidation of the drug (Glader and Conrad, 1973; Uetrecht et al., 1988). In order to explore the role of metabolism in the toxicity of dapsone, we have developed an in vitro two-compartment model in which human target cells, either white or red blood cells, are separated from a drug metabolizing system by a semi-permeable membrane (Riley et al., 1990; Tingle et al., 1990). Using this model, we have demonstrated the ability of human liver microsomes to generate dapsone hydroxylamine in one compartment with toxicity towards human target cells in a second compartment. Furthermore, using

this model it was shown that the N-hydroxylation of dapsone can be reduced by two known inhibitors of drug oxidation, ketoconazole (Tingle et al., 1990) and cimetidine (Tingle et al., 1991), with a resultant decrease in the toxicity observed. The in vitro observation that cimetidine can reduce the haemotoxicity associated with dapsone has now been shown to apply in vivo. Co-administration of cimetidine (400 mg three times daily) with dapsone reduced methaemoglobinaemia, whilst increasing both peak concentrations and plasma AUC of dapsone after a single dose (100mg) of dapsone in volunteers (Coleman et al., 1990) and in patients on long-term dapsone therapy (50-350 mg/day) (Coleman et al., 1992). Dapsone causes rare and idiosyncratic white cell toxicities (incidence less than one in 2000), while resulting in dose-dependent red cell toxicity to some extent in all individuals who take the drug (Zuidema et al., 1986). In order to investigate this observation, we have developed a three-compartment model, in which the drug-metabolizing system is contained within a central compartment and separated from both white and red cells by semi-permeable membranes. Using this model, the cell-selective toxicity of dapsone metabolites was explored, as well as the effect that the presence of one cell type has on the toxicity observed in other cells (Tingle and Park, 1993). Dapsone was metabolized to a species capable of diffusing out of microsomes, crossing a semi-permeable membrane and causing toxicity, in the form of haemoglobin oxidation, to red cells. However, when red cells were present in the three-compartment model, no significant toxicity towards white cells was observed. In contrast, when the red cells were replaced by buffer, significant white cell toxicity was detected, which suggests that the presence of red cells can protect white cells from the effects of the toxic metabolite(s). This may be explained by the uptake and accumulation of dapsone hydroxylamine into the red cell, where it is co-oxidized with haemoglobin to form nitroso-dapsone and methaemoglobin (Fig. 3). The nitroso intermediate may be reduced back to the hydroxylamine by intracellular glutathione while the methaemoglobin is reduced back to haemoglobin by both N A D H - and NADPH-dependent methaemoglobin reductase enzymes so that a futile cycle exists (Kramer et al., 1972). This cycle will continue until the dapsone metabolites, presumably, become irreversibly bound to haemoglobin, in a manner similar to 4,4'-methylene dianiline, a close structural analogue (Bailey et al., 1990). Moreover, the ability of low concentrations of red ceils (12.5% haematocrit) to provide protection of white cells suggests that there exists a large reserve capacity within the red cells for the uptake and detoxification of dapsone hydroxylamine. In order to examine the chemical aspects involved in the haemotoxicity of dapsone, 11 structural analogues have been investigated. The analogues included 4,4'-diaminodiphenyl compounds, in

617

Bioactivation and bioinactivation of drugs which the sulfone group has been replaced, and p-substituted anilines, aniline being the simplest chemical analogue of dapsone. Metabolism of dapsone produced the most methaemoglobin formation, whilst aniline produced the least. There was a significant correlation (r -- 0.9, P < 0.001) between methaemoglobin formation and the Hammett constant, trp, which suggests that an electron-withdrawing group para to the amine is required for toxicity. There was no correlation with steric effects (molecular refractivity), lipophilic constants (n, log Ko) or with ionization (pKa).

Amodiaquineagranulocytosis Amodiaquine is a 4-aminoquinoline antimalarial that is effective against chloroquine-resistant strains of Plasmodium falciparum. However, its clinical use has been severely restricted because of associations with hepatotoxicity and agranulocytosis (Hatton et al., 1986; Neftel et al., 1986; Rouviex et al., 1989). Neither the precise mechanism nor the causative agent (parent drug or metabolite) has been identified, although both direct cytotoxic and indirect immunological mechanisms have been proposed. Work undertaken in our laboratory indicates that these adverse reactions are of an immunological nature and are a consequence of the facile oxidation of the drug to chemically reactive metabolites that function as haptens in vivo, and initiate an immune response

(Christie et a1.,1989; Maggs et al., 1987 and 1988) (Fig. 4). A combination of chemical, biochemical and immunochemical in vitro techniques were used alongside in vivo investigations, using animal models, in these studies. Specifically it has been demonstrated that amodiaquine is readily oxidized to a protein-reactive metabolite, amodiaquine quinoneimine, and that this process can be catalysed by several; biological oxidizing systems, including activated white cells and cytochrome P-450 enzymes. Electrochemical studies show that the 4-hydroxyaniline side-chain in amodiaquine is readily oxidized to a quinoneimine that can conjugate directly with sulfhydryl groups in glutathione and proteins. The identification of a 5'-glutathione conjugate as the major biliary metabolite in the rat, and the identification of drug-related antigen in liver, confirm the in vivo relevance of the in vitro findings. Amodiaquine is immunogenic in the rat when given by the oral, intraperitoneal or intramuscular routes, without the requirement for co-administration of immunological adjuvant (Clarke et al., 1990). The drug is also immunogenic in humans when given in repeated prophylactic doses, and in a retrospective study, anti-amodiaquine antibodies were detected in patients with serious adverse reactions. Importantly, it was shown that the antibody from patients recognized a synthetic drug antigen designed and synthesized with a knowledge of the chemical

IH2

glucuronidation

~

i

f

o= =o

i

,/

acetylation

IH2

o=i =o

IHOH

red blood cells

GSSG ~

GSH

j-1

///..') L_(

white blood cells

haemoglobin' / ' ' - - - " ' ~ 02

Fig. 3. Mechanism of the cell selective toxicity seen with dapsone.

methaemoglobln

B. K. PARKet al.

618

OH

OH

/---

Amodiaquine [~CH2-% NH

/H

NH

Cytochrome P450 MicrosomalI I Activated white blood cells enzymes

Desethylamodiaquine

0 I Bioactivation] Amodiaquine quinoneimine

N

GS-c"2-L_ OH

/---

OH /---pr°tein'~ CH2

NH Amodiaquine-glutathione NH Arnodiaquine-protein C I ~ conjugate C I ~ conjugate Detoxication I

ANTIGEN/IMMUNOGEN FORMATION

Fig. 4. The metabolic fate of amodiaquine, showing the proposed mechanism for an immunotoxic reaction by way of the unstable quinone imine intermediate. mechanisms of bioactivation of amodiaquine, determined from both in vitro and in vivo metabolic studies (Clarke et al., 1991). It is therefore evident that the cell-directed toxicity may be explained by two separate mechanisms of bioactivation of amodiaquine. First, the formation of cell-surface antigen may lead to a secondary immune reaction, or secondly, the intracellular generation of the quinone imine may result in direct toxicity (as observed with paracetamol) by interfering with vital cell function. In vitro studies indicate that host factors that may be important determinants of toxicity include rates of glutathione conjugation and also reduction of the quinoneimine by microsomal enzymes. Thus, incubation of radiolabelled amodiaquine quinoneimine with human liver microsomes resulted in the formation of desethyl-amodiaquine as the major metabolite, produced by successive reduction and de-ethylation reactions.

Thus, it is unlikely that it will be possible to predict individual susceptibility to amodiaquine toxicity on the basis of the measurement of a single factor. However, preliminary studies from our laboratory, based on assays with either malaria parasites or human polymorphonuclear lymphocytes, indicate that it is possible to modify amodiaquine chemically and thereby prevent bioactivation without necessarily losing pharmacological activity.

Anticonvulsanthypersensitivity The aromatic anticonvulsants (phenytoin, carbamazepine and phenobarbitone) are a widely used group of drugs that can occasionally lead to unpredictable severe adverse reactions that can affect multiple organ systems and are often accompanied by hypersensitivity manifestations such as rash, fever, arthralgia and eosinophilia. A major advance in the

Bioactivation and bioinactivation of drugs investigation of anticonvulsant hypersensitivity came about with the development of the in vitro cytotoxicity assay by Speilberg (1980). In this assay, lymphocytes, which contain detoxication enzymes, are used as target ceils, and therefore serve as peripheral markers for the hepatic enzymes. A major advantage of this system is that tissues from patients suspected or known to be hypersensitive to drugs can be exposed to the drug in vitro without the risk of an in vivo rechallenge. Using such a system, it was suggested that the unique susceptibility of patients may be due to a deficiency of cellular detoxication processes (Shear et al., 1988; Spieiberg et al., 1981). In our studies (Friedmann et al., 1993; Pirmohamed et al., 1991 and 1992b, c), carbamazepine, the most widely used anticonvulsant in this country, has been used as a model compound. 10 patients with a clinical diagnosis of carbamazepine hypersensitivity were identified and investigated using an in vitro cytotoxicity assay that comprised lymphocytes and hepatic microsomes prepared from mice pretreated with phenobarbitone (Riley et al., 1988). Lymphocytes from the hypersensitive patients were found to be more susceptible to metabolites of carbamazepine generated in situ than were controls, which comprised normal healthy volunteers and patients on carbamazepine without adverse effects. No in vitro chemical cross-sensitivity was observed when the patient lymphocytes were exposed to phenytoin, which suggests that the system was highly specific, since five of the patients were on long-term phenytoin therapy without any adverse effects (Pirmohamed et al., 1991). Using microsomes prepared from a panel of 10 human livers, we were able to demonstrate that all human livers were capable of bioactivating carbamazepine to a chemically reactive metabolite (Pirmohamed et al., 1991), further reinforcing the fact that the predominant abnormality in these patients was one of detoxification rather than of bioactivation. The nature of the chemically reactive metabolite was investigated further by the addition of chemical modifiers and enzyme inhibitors. The metabolism-dependent cytotoxicity of carbamazepine was enhanced by co-incubation with trichloropropene oxide, an inhibitor of microsomal epoxide hydrolase, and reduced by the addition of purified hepatic microsomal epoxide hydrolase, but not cytosolic epoxide hydrolase. Taken collectively, these results suggest that the chemically reactive metabolite is an arene oxide and that the affected patients may have a deficiency of microsomal epoxide hydrolase (Pirmohamed et al., 1992c). In line with the initial studies undertaken by Spielberg (1980), we and others (Larrey et aL, 1989) have used induced mouse microsomes because they provide a consistent level of bioactivation and bioinactivation, and therefore overcome the problem of interindividual variability in expression of the drugmetabolizing enzymes, which is well recognized in humans. To define more precisely the enzyme activi-

619

ties associated with this activating system, naphthalene has been used as an in vitro probe for bioactivation and bioinactivation, particularly with respect to arene oxides (Tingle et al., 1993). In accordance with the fact that mouse liver contains less microsomal epoxide hydrolase than human liver (Guenthner, 1990), and that phenobarbitone induction has a greater effect on drug oxidation than epoxide hydrolysis (Hardwick et al., 1983), it was found that the net balance between bioactivation and bioinactivation in phenobarbital-induced mouse microsomes provides a quantitative increase in reactive metabolite production. Having gained indirect evidence that the reactive metabolite of carbamazepine was an arene oxide, further investigations were then undertaken to determine the enzyme(s) responsible for its formation. The formation of the cytotoxic and protein-reactive metabolite as well as the stable 10,1 l-epoxide could be inhibited by ketoconazole and gestodene, and was enhanced by induction with dexamethasone and phenobarbitone, indicating the involvement of the cytochrome P4503A enzyme system (Pirmohamed et al., 1992c). In summary, our studies with carbamazepine suggest that individual susceptibility to hypersensitivity may be due to a critical imbalance between bioactivation of carbamazepine to an arene oxide and its detoxication by microsomal epoxide hydrolase (Fig. 5). Such an imbalance may be systemic or tissue specific, and may be the result of genetic and/or environmental factors. Clearly, such an imbalance would lead to persistence of the chemically reactive metabolite, which could then bind to cellular macromolecules and cause either direct toxicity or a secondary immune reaction. Available evidence, mainly based on clinical symptomatology, suggests that carbamazepine idiosyncratic toxicity is immunemediated (Friedmann et al., 1993; Pirmohamed et al., 1992a). Our current efforts are directed towards determining whether the carbamazepine hypersensitive patients have a deficiency of the enzyme microsomal epoxide hydrolase. Such a deficiency may be quantitative, although this seems unlikely since we have been able to demonstrate the presence of protein in patient lymphocytes by radioimmunoassay (Guenthner et al., 1993). Alternatively, the deficiency may be qualitative with subsequent alteration in substrate specificity and/or substrate affinity. These studies will require the development of more sensitive assays for microsomal epoxide hydrolase as well as the use of molecular techniques. Conclusions In vitro systems based on human tissues and cells can provide a valuable insight into the mechanisms involved in adverse drug reactions. In order to understand the chemical mechanisms involved in toxicity it is necessary to define the chemical species responsible. However, given that unstable metabolites are often implicated in toxicity, direct chemical measurement is

B.K.PARKe t

620

al.

STABLE (10,11) EPOXIDE

C! - O

I

c-o !

N~

P450

NH2

CARBAM,~.EPINE

[ E n z y m e Induction

Microsomal epoxide hydrolase

DIHYDRODIOL

i

UNSTABLE EPOXIDE

COVALENT

~

BINDING

IMMUNE RESPONSE

CYTOTOXICITY TISSUE DAMAGE

Fig. 5. The role of epoxide metabolites in the mechanism of toxicity of carbamazepine.

not always possible, even when using an in vitro system. Such reactive intermediates can be identified by isolation of stable adducts formed from nucleophiles, which can be introduced unobtrusively into the in vitro metabolism/toxicity system. Alternatively, the chemical nature of the toxic species can be inferred from experiments in which chemical modulators, for example nucleophiles, reducing agents and enzyme inhibitors, are added to the test system and shown to perturb simultaneously both metabolism and toxicity. Host factors that influence adverse reactions can be identified by experiments in which the effects of perturbation (e.g. by induction, inhibition or simple addition) of the enzymes involved in either drug bioactivation or drug bioinactivation are monitored to assess their consequences on drug metabolism and drug toxicity in vitro. The clinical relevance of such an approach is greatly enhanced by: (1) the inclusion of tissues/cells from patients with (idiosyncratic) drug toxicity; (2) genotyping and phenotyping the cells and tissues used for known polymorphisms/variations in human drug metabolism.

Acknowledgements--BKP is a Wellcome Principal Fellow. MP is a Sir Desmond Pond Fellow (Epilepsy Research Foundation). The support of the MRC, Glaxo Group Research Ltd and Parke-Davis Pharmaceuticals is also gratefully acknowledged.

REFERENCES

Bailey E., Brooks A. G., Bird I., Farmer P. B. and Street B. (1990) Monitoring exposure to 4,4'methylenedianiline by the gas chromatography-mass spectrometry determination of adducts to haemoglobin. Analytical Biochemistry 190, 175-181.

Christie G., Breckenridge A. M. and Park B. K. (1989) Drug-protein conjugates XVIII. Detection of antibodies towards the antimalarial amodiaquine and its quinone imine metabolite in man and the rat. Biochemical Pharmacology 38, 1451-1458. Clarke J. B., Maggs J. L., Kitteringham N. R. and Park B. K. (1990) Immunogenicity of amodiaquine in the rat. International Archives of Allergy and Applied Immunology 91, 335-342. Clarke J. B., Neftel K., Kitteringham N. R. and Park B. K. (1991) Detection of antidrug IgG antibodies in patients with adverse drug reactions to amodiaquine. International Archives of Allergy and Applied Immunology 95, 369-375. Coleman M. D., Rhodes L. E., Scott A. K., Verbov J. L., Friedmann P. S., Breckenridge A. M. and Park B. K. (1992) The use of cimetidine to reduce dapsone-dependent methaemoglobinaemia in dermatitis herpetiformis patients. British Journal of Clinical Pharmacology 34, 244-249. Coleman M. D., Scott A. K., Breckenridge A. M. and Park B. K. (1990) The use of cimetidine as a selective inhibitor of dapsone N-hydroxylation in man. British Journal of Clinical Pharmacology 30, 761-767. D'Arcy P. F. (1986) Epidemiological aspects of iatrogenic disease. In latrogenic Diseases. Edited by P. F. D'Arcy and J. P. Griffin. pp. 29-58. Oxford University Press, Oxford. Dogterom P., Nagelkerke J. F. and Mulder G. J. (1988) Hepatotoxicity of tetrahydroaminoacridine in isolated rat hepatoeytes: effects of glutathione and vitamin E. Biochemical Pharmacology 37, 2311-2313. Doroshow J. H., Locker G. Y. and Myers C. E. (1980) Enzymatic defences of the mouse heart against reactive oxygen metabolites: alterations produced by doxorubicin. Journal of Clinical Investigation 65, 128-135. Farlow M., Gracon S. I., Hershey L. A., Lewis K. W., Salowsky C. H. and Dolan-Ureno J. (1992) A controlled trial of tacrine in Alzheimer's disease, Journal of the American Medical Association 268, 2523-2529. Forsyth D. R., Sormon D. J., Morgan R. A. and Wilcock G. K. (1989) Clinical experience with and side effects of tacrine hydrochloride in Alzheimer's disease: a pilot study. Age and Ageing 18, 223-229.

Bioactivation and bioinactivation of drugs Friedmann P. S., Strickland I., Pirmohamed M. and Park B. K. (1994) Investigation of mechanisms in toxic epidermal necrolysis induced by carbamazepine. Archives of Dermatology. In press. Glader B. F. and Conrad M. E. (1973) Haemolysis by diphenylsulfones: comparative effects of DDS and hydroxylamine-DDS. Journal of Laboratory and Clinical Medicine 81, 267-272. Guenthner T. M. (1990) Epoxide hydrolases. In Conjugation Reactions in Drug Metabolism: an Integrated Approach. Edited by G. J. Mulder. pp. 365-404. Taylor & Francis, London. Guenthner T. M., Kuk J., Nguyen M., Wheeler C. W., Pirmohamed M. and Park B. K. (1993) Epoxide hydrolases: immunochemical detection in human tissues. In Human Drug Metabolism: From Molecular Biology to Man. Edited by E. H. Jeffrey. pp. 65-80. CRC Press, Boca Raton, FL. Hardwick J. P., Gonzalez F. J. and Kasper C. B. (1983) Transcriptional regulation of rat liver epoxide hydratase, NADPH-cytochrome P-450 oxidoreductase, and cytochrome P-450b genes by phenobarbital. Journal of Biological Chemistry 258, 8081-8085. Hatton C. S. R., Peto T. E. A. and Bunch C. (1986) Frequency of severe neutropenia associated with amodiaquine prophylaxis against malaria. Lancet i, 411-413. Kramer P. A., Glader B. E. and Li T. K. (1972) Mechanism of methaemoglobin formation by diphenylsulfones. Effect of 4-amino-4'-hydroxy-aminodiphenylsulfone and other p-substituted derivatives. Biochemical Pharmacology 21, 1265-1274. Larrey D., Berson A., Habersetzer F., Tinel M., Castot A., Babany G., Letteron P., Freneaux E., Loeper J., Dansette P. and Pessayre D. (1989) Genetic predisposition to drug hepatotoxicity: role in hepatitis caused by amineptine, a tricyclic antidepressant. Hepatology 10, 168-173. Madden S., Woolf T. F., Pool W. F. and Park B. K. (1993) An investigation into the formation of stable, protein-reactive and cytotoxic metabolites from tacrine in vitro: studies with human and rat liver microsomes. Biochemical Pharmacology 46, 13-20. Maggs J. L., Kitteringham N. R., Breckenridge A. M. and Park B. K. (1987) Autoxidative formation of a chemically reactive intermediate from amodiaquine, a myelotoxin and hepatotoxin in man. Biochemical Pharmacology 36, 2061-2062. Maggs J. L., Tingle M. D., Kitteringham N. R. and Park B. K. (1988) Drug-protein conjugates XIV. Mechanisms of formation of protein-arylating intermediates from amodiaquine, a myelotoxin and hepatotoxin in man. Biochemical Pharmacology 37, 303-311. Maron D. M. and Ames B. M. (1983) Revised methods for the Salmonella mutagenicity test. Mutation Research 113, 173-215. Neftel K. A., Woodtly W., Schmid M., Frick P. G. and Fehr J. (1986) Amodiaquine induced agranulocytosis and liver damage. British Medical Journal 292, 721-723. O'Brien J. T., Eagger S. and Levy R. (1991) Effect of tetrahydroaminoacridine on liver function in patients with Alzheimer's disease. Age and Ageing 20, 129-131. Park B. K., Pirmohamed M. and Kitteringham N. R. (1992) Idiosyncratic drug reactions: a mechanistic evaluation of risk factors. British Journal of Clinical Pharmacology 34, 377-395. Pirmohamed M., Graham A., Roberts P., Smith D., Chadwick D., Breckenridge A. M. and Park B. K. (1991) Carbamazepine hypersensitivity: assessment of clinical and in vitro chemical cross-reactivity with phenytoin and oxcarbazepine. British Journal of Clinical Pharmacology 32, 741-749. Pirmohamed M., Kitteringham N. R., Breckenridge A. M. and Park B. K. (1992a) Detection of an autoantibody

621

directed against a human liver microsomal protein in a patient with carbamazepine hypersensitivity. British Journal of Clinical Pharmacology 33, 183-186. Pirmohamed M., Kitteringham N. R., Breckenridge A. M. and Park B. K. (1992b) The effect of enzyme induction on the cytochrome P450-mediated bioactivation of carbamazepine by mouse liver microsomes. Biochemical Pharmacology 44, 2307-2314. Pirmohamed M., Kitteringham N. R., Guenthner T. M., Breckenridge A. M. and Park B. K. (1992c) Investigation into the formation of cytotoxic, protein reactive and stable metabolites from carbamazepine in vitro. Biochemical Pharmacology 43, 2675-1682. Pool W. F., Bjorge S. M., Chang T. and Woolf T. F. (1992) Metabolic disposition of the cognition activator in man: identification of phenol glucuronide metabolites in urine. ISSX Proceedings Fourth North American ISSX Meeting 2, 164. Riley R. J., Lambert C., Maggs J. L., Kitteringham N. R. and Park B. K. (1988) An in vitro study of the microsomal metabolism and cellular toxicity of phenytoin, sorbinil, and mianserin. British Journal of Clinical Pharmacology 26, 577-588. Riley R. J., Roberts P., Coleman M. D., Kitteringham N. R. and Park B. K. (1990) A two-compartment system to investigate drug metabolite-mediated cytotoxicity in vitro. British Journal of Clinical Pharmacology 29, 625P. Rouviex B., Coulombel L., Aymard J. P., Chau F. and Abel L. (1989) Amodiaquine induced immune agranulocytosis. British Journal of Haematology 71, 7 11. Shear N. H., Spielberg S. P., Cannon M. and Miller M. (1988) Anticonvulsant hypersensitivity syndrome: in vitro risk assessment. Journal of Clinical Investigation 82, 1826--1832. Spielberg S. P. (1980) Acetaminophen toxicity in human lymphocytes in vitro. Journal of Pharmacology and Experimental Therapeutics 213, 395-398. Tingle M. D., Coleman M. D. and Park B. K. (1990) Investigation into the role of metabolism in dapsone-induced methaemoglobinaemia using a two compartment in vitro test system. British Journal of Clinical Pharmacology 30, 829-838. Tingle M. D., Coleman M. D. and Park B. K. (1991) The effect of preincubation with cimetidine on the N-hydroxylation of dapsone by human liver microsomes. British Journal of Clinical Pharmacology 32, 120-123. Tingle M. D. and Park B. K. (1993) The use of a three compartment in vitro model to investigate the role of hepatic drug metabolism in drug-induced blood cyscrasias. British Journal of Clinical Pharmacology 36, 31-38. Tingle M. D., Pirmohamed M., Templeton E., Wilson A. S., Madden S., Kitteringham N. R. and Park B. K. (1993) An investigation of the formation of cytotoxic, genotoxic, protein-reactive and stable metabolites from naphthalene by human liver in vitro. Biochemical Pharmacology 46, 1529-1538. Uetrecht J., Zahid N., Shear N. H. and Biggar W. D. (1988) Metabolism of dapsone to a hydroxylamine by human neutrophils and mononuclear cells. Journal of Pharmacology and Experimental Therapeutics 245, 274-279. Woolf T. F., Pool W. F., Bjorge S. M., Chang T., Goel O. P., Purchase C. F., Schroeder M. C., Kunze K. L. and Trager W. F. (1993) Bioactivation and irreversible binding of the cognition activator tacrine using human and rat liver microsomal preparations: species differences. Drug Metabolism and Disposition 21, 874-882. Zuidema J., Hilbers-Moddermann E. S. M. and Merkus F. W. H. M. (1986) Clinical pharmacokinetics of dapsone. Clinical Pharmacokinetics 11, 299-315.