Mechanisms of Drug-Induced Liver Injury

MECHANISMS OF DRUG-INDUCED LIVER INJURY

The general mechanism by which most drugs induce liver injury is based on the unusual susceptibility of individual patients. For some drugs, the idiosyncratic reaction is immunologically mediated and for others metabolic idiosyncrasy may be responsible. Even though such classification of drug-induced liver injury is simplistic and the molecular basis of the idiosyn-cratic drug-induced liver injury is poorly understood, some key events have emerged as being particularly important.

METABOLIC IDIOSYNCRASY

The suggestion of dose dependence in some cases of drug-induced liver injury indicates that a host-dependent idiosyncrasy in the metabolism or excretion of these drugs may be responsible for hepa-totoxicity. Although several xenobiotics are trans-formed by the cytochrome P450 system (CYPs) into stable metabolites, many others are oxidised into unstable, chemically reactive intermediates. These reactive intermediates attack hepatic constituents such as unsaturated lipids, proteins or DNA and can lead to liver cell death (Pessayre, 1995). The abun-dance of CYPs in the liver explains the major role of these metabolites in drug-induced hepatotoxic-ity. Furthermore, the centrilobular location of most CYPs accounts for the pericentral location of these lesions. When small amounts of reactive metabolites are formed, glutathione serves as a decoy target, spar-ing critical hepatic macromolecules. However, when large amounts of the reactive metabolite are formed, the formation of glutathione conjugates exceeds the capacity of the liver to synthesise glutathione. The resulting depletion of glutathione together with direct covalent binding of the metabolite protein thiols has serious consequences. The oxidation of protein thiol groups results in the formation of disulphur bonds between different molecules of actin, result-ing in destruction of the microfilamentous network beneath the plasma membrane (Mirabelli et al., 1988). The depletion of protein thiol groups also decreases the activity of calcium translocases resulting in increases in intracellular Ca²⁺ that further damages the cytoskeleton (Bellomo and Orrenius, 1985). These and other effects of oxidative stress lead to the swelling and disruption on intracellular organelles ultimately resulting in hepatocyte necrosis.

Although it was initially thought that the toxic-ity of reactive metabolites only caused cell necro-sis, this idea has been challenged in recent years (Pessayre et al., 1999). It is now clear that the exten-sive formation of reactive metabolites can cause apop-tosis, necrosis or both (Fau et al., 1997; Shi et al., 1998). Several compounds, such as acetaminophen and cocaine, transformed into reactive metabolites have been shown to cause DNA fragmentation of hepatocytes indicative of apoptosis (Shen et al., 1992; Cascales et al., 1994). The cellular mechanisms caus-ing metabolite-induced apoptosis have been studied with germander, a medicinal plant used in weight control diets, the widespread use of which led to an epidemic of hepatitis in France (Larrey et al., 1992b). Germander contains furano diterpenoids, which are activated by CYP 3A into electrophilic metabo-lites (Lekehal et al., 1996). Extensive formation of glutathione depletion, which in combination with covalent binding of the metabolites, results in protein thiol oxidation (Lekehal et al., 1996). The oxida-tion of protein thiols inactivates plasma membrane calcium translocases and increases the permeability of the mitochondrial inner membrane (the mitochon-drial membrane permeability transit or MMPT), which through the release of cytochrome C leads to the acti-vation of caspases (Fagian et al., 1990). Caspases are cysteine proteases that cut proteins after an aspartate residue and are the major executioners of apopto-sis (Thornberry and Lazebnik, 1998). Caspase acti-vation in conjunction with increased intra-cellular calcium activates calcium-dependent endonucleases, which cut the DNA between nucleosomes, eventually resulting in apoptosis (Fau et al., 1997). Germander-induced apoptotic hepatocyte death is prevented by troleandomycin, which inhibits its metabolic activa-tion by CYP 3A4 or by preventing the depletion of glutathione with cysteine (Fau et al., 1997).

Factors Influencing Direct Toxicity Due to Reactive Metabolites

Hepatotoxicity from the reactive metabolites of drugs is a significant problem with drugs where the forma-tion of reactive metabolites is low enough to ensure the absence of hepatotoxicity in most recipients (and therefore allowing the marketing of the drug) but is high enough to lead to ‘idiosyncratic’ toxicity in some ‘susceptible’ subjects. The reason for susceptibility could be either genetically determined or acquired.

Genetic Factors

The amount of reactive metabolite formed depends on a particular isoenzyme the hepatic level of which may vary between individuals. Genetic polymorphisms of drug-metabolising enzymes may contribute to an indi-vidual’s risk to an ADR. Polymorphism in debriso-quine oxidation (CYP 2D6) leads to the accumula-tion of perhexiline resulting in liver injury in poor metabolisers (Morgan et al., 1984) and increases the formation of reactive metabolites leading to chlor-promazine hepatotoxicity in extensive metabolisers (Watson et al., 1988). Polymorphism in mepheny-toin hydroxylation (CYP 2C19) may predispose poor metabolisers to atrium (phenobarbital, febar-bamate and difebarbamate)-induced hepatotoxicity (Horsmans et al., 1984). Recent studies have shown susceptibility to isoniazid-induced hepatotoxicity was increased with the possession of both NAT2 geno-type associated with slow acetylation and CYP 2E1 genotype leading to increased activity (Huang et al., 2002, 2003).

Acquired Factors

Individual susceptibility to hepatotoxicity due to reac-tive metabolites may also be related to physiologi-cal, nutritional or therapeutic modifications in drug metabolism. For example, fasting leads to glycogen depletion and decreased glucuronidation, the deple-tion of glutathione and the induction of CYP 2E1 leading to an increased risk of paracetamol-induced liver injury (Price, Miller and Jollow, 1987; Whit-comb and Block, 1994). Acquired factors enhancing the rate of biotransformation of a drug to its reactive metabolites through the induction of CYP P450 isoen-zymes play an important role in increasing the direct toxicity. Alcohol is a potent inducer of CYP 2E1 and to a lesser extent CYP 3A4. Subjects who consume alcohol regularly may therefore have increased the bioactivation of paracetamol (which is metabolised by CYP 2E1 and CYP 3A4), resulting in hepato-toxicity at conventional ‘therapeutic’ doses (Zimmer-man and Maddrey, 1995). In individuals with heavy alcohol intake, this is compounded by the reduced glutathione synthesis and low glutathione stores due to the inhibition of glutathione synthetase and ethanol-related oxidative stress, respectively. Isoniazid also increases the toxicity of paracetamol by inducing CYP 2E1, whereas rifampicin, another microsomal enzyme inducer, increases the risk of hepatotoxic-ity due to isoniazid (Moulding, Redeker and Kanel, 1991; Pessayre et al., 1977). Anticonvulsants (pheny-toin, carbamazepine and phenobarbital) induce CYP 3A4 and can also enhance the toxic effects of parac-etamol (Bray et al., 1992). As an alternative mech-anism of drug interaction leading to an increased risk of paracetamol-induced liver injury, zidovudine competes for glucuronidation of the toxic metabo-lite, thus reducing its excretion (Shriner and Goetz, 1992). Drug accumulation can result from metabolic inhibition caused by another drug. For instance, trole-andomycin increases the risk of cholestasis with oral contraceptives by inhibiting the CYP 3A responsible for oestrogen oxidation (Miguet et al., 1980).

The presence of underlying liver disease may predispose to dose-dependent drug toxicity, especially if the margin between therapeutic and toxic concen-trations is small (Schenker, Martin and Hoyumpa, 1999). It is generally believed that pre-existing liver disease would neither induce nor worsen the idiosyn-cratic hepatotoxicity, although this issue has not been studied adequately. However, a recent study demon-strated a higher incidence of hepatotoxicity as well as more severe liver injury secondary to antituberculo-sis agents in hepatitis B virus (HBV) carriers when compared with non-carriers and with HBV carriers who did not receive antituberculosis therapy (Wong et al., 2000).

IMMUNOLOGIC IDIOSYNCRASY

The clinico-pathologic features of some idiosyncratic drug reactions suggest that immunological mecha-nisms could play an important role in the patho-genesis of drug hepatotoxicity. These include (a) a fever, rash, lymphadenopathy, eosinophilia and involvement of other organs; (b) hepatic inflamma-tory infiltrates; (c) low frequency (<1/1000 users); (d) delay in appearance of the disease (2 weeks to several months); and (e) accelerated onset after rechal-lenge (Beaune and Lecoeur, 1997; Robin et al., 1997). In hepatitis, secondary to sulphonamides, phenytoin and nitrofurantoin, the liver is implicated as part of a systemic hypersensitivity reaction, and evidence for immunological responsiveness to these drugs can be obtained by in vitro rechallenge with the drug or its metabolite (Spielberg et al., 1981; Shear and Spielberg, 1988; Rieder et al., 1989). Interestingly, the immune response may not be directed at the drug per se but at compounds arising because of its metabolism. Drug hepatotoxicity may therefore be the result of both metabolic and immunological idiosyncrasy. In this respect, the superimposition of CYP P450 and the immune system in the liver have potential disadvantages. The covalent binding of the reactive metabolites to ‘self’ proteins results in the formation of neo-antigens that ‘mislead’ the immune system into mounting an immune response against hepatocytes.

The initial and crucial event underlying the so-called ‘immuno-allergic hepatitis’ is the oxida-tive metabolism of a drug by a CYP P450 enzyme resulting in the formation of reactive metabolites. Electrophilic metabolites react with and covalently bind to nucleophilic groups of patients to form protein adducts. The best-studied example is that of halothane, which is oxidised into a reactive acyl chlo-ride (CF₃COC1) by CYP P450 2E1. The metabolite reacts with the -NH₂ group of the lysine residues of proteins to form trifluoroacetylated proteins (CF₃CO-lysine proteins) (Gut, Christen and Huwyler, 1993). The reactive metabolite may also bind covalently to the CYP 2E1 protein itself (Eliasson and Kenna, 1996). The alkylation of CYP P450 proteins may lead both to anti-P450 autoantibodies and to antibodies against the modified part of the protein. Therefore, a single drug such as halothane may concomitantly give rise to both ‘immune-allergic’ and ‘autoimmune’ hepatitis.

FACTORS INFLUENCING IMMUNOLOGICALLY MEDIATED DRUG HEPATOTOXICITY

Genetic factors influencing the development of the immune-mediated drug hepatotoxicity can be grouped into factors affecting the amount of the reactive metabolite and therefore protein adduct formed and factors affecting the immune response to these adducts (Aithal, 2004). Dihydralazine hepatitis is a good example of how a ‘metabolic’ genetic factor can contribute to susceptibility to immune-mediated hepa-totoxicity. Dihydralazine is predominantly acetylated by the polymorphic N -acetyl transferase 2. In slow acetylators, the majority of the drug is available for metabolic activation by CYP 1A2 into a free radi-cal. Hence, the alkylation of hepatic proteins is more extensive, and the incidence of immune-mediated hepatitis is higher (Bourdi et al., 1994).

The second group of genetic factors influencing susceptibility to immune-mediated hepatic drug reac-tions is the genes whose products are involved in immune regulation. Genetic polymorphism in major histocompatibility complex (MHC) molecules is the most obvious example. The presence or absence of a given human leukocyte antigen (HLA) molecule may determine the efficient presentation of an alky-lated immunogenic peptide. Associations have been reported between HLA A11 and hepatotoxicity due to halothane, tricyclic antidepressants and diclofenac, HLA DR6 and liver injury secondary to chlorpro-mazine and nitrofurantoin, HLA B8 and clometacine hepatitis (Berson et al., 1994). Two case–control stud-ies involving Caucasian population have demon-strated that co-amoxiclav-induced jaundice is strongly associated with HLA DRB1*1501-DRB5*0101-DQB1*0602 haplotype (Hautekeete et al., 1999, O’Donohue et al., 2000). Subjects carrying the extended haplotype would be at nine times higher risk of developing ADR to co-amoxiclav (O’Donohue et al., 2000). More recently, genetic polymorphism in gene-encoding immunomodulatory cytokines such as interleukin-10 (IL-10) and IL-4 has been shown to influence the risk of diclofenac-induced hepatotoxic-ity (Aithal, 2004).

SPECIFIC HISTOLOGICAL TYPES OF DRUG-INDUCED LIVER INJURY

Cholestasis

From experimental models, several mechanisms have been postulated for impaired bile secretion. They are the inhibition of Na+, K+ATPase resulting in reduced uptake of bile acids, increased pericellular permeabil-ity and regurgitation into plasma of bile constituents, impaired intracellular transport due to cytoskeletal dysfunction, altered intracellular calcium homeosta-sis or altered canalicular carriers (Erlinger, 1997). A recent study demonstrated that oestrogen metabo-lites trans-inhibit the bile salt export pump in rat liver providing a molecular basis for drug-induced cholestasis (Stieger et al., 2000).

Steatosis

Microvesicular steatosis occurs in conditions char-acterised by severe impairment of the mitochondrial β-oxidation process. Drugs can sequester co-enzyme A  (aspirin  valproic  acid),  inhibit  mitochondrial  β -oxidation enzymes(tetracycline)  and,  in  addition, inhibit oxidative phosphorylation (amiodarone and  perhexiline).

When  β-oxidation  is  severely impaired, fatty acids, which are poorly oxidised by mitochondria, are mainly esterified into riglycerides and accumulated as small vesicles (Fromenty, Berson and Pessayre, 1997).