Drug hepatotoxicity: an integrated clinical and mechanistic investigation

Intraperitoneal administration of single doses of several (NSAIDs diclofenac, sulindac and ibuprofen) to mice and rats was shown specifically induce dose-dependent expression of hepatic haem oxygenase-1 (H0-1), but not other enzymes which share antioxidant responsive element (ARE) sequences (i.e. NADP(H):quinone reductase, GSTs and UGTs). Induction of HO-1 mRNA in mouse liver correlated temporally with acute hepatic injury, as assessed by enzymatic activities of serum transaminases and sorbitol dehydrogenase. Of the three NSAIDs investigated ibuprofen (which was not hepatotoxic) was the least potent inducer of HO. Induction of HO and liver injury were not attributed to altered glutathione homeostasis activity was evident when animals were pre-treated in vivo with an anti-TNF antibody, before administration of diclofenac. Although this suggests that TNF plays a role in induction of HO-1 and toxicity, elevated level of TNF were not detected at any time point in serum or liver tissue of diclofenac treated mice. Dose-dependant induction of HO-1 mRNA was observed after exposure of cultured rat hepatocytes to diclofenac, suggesting that the modifications induced by diclofenac and/or metabolites inside the hepatocyte itself are responsible.

Summary: The possible involvement of polymorphisms in CYP 2C9 and IL-10 in diclofenac-induced hepatotoxicity was investigated. The known major genetic variants in CYP2C9 were found not to play a major role in diclofenac-induced hepatotoxicity. The entire CYP2C9 gene and -1 to -1000 of its upstream sequence were also sequenced from 4 patients with the most severe diclofenac hepatotoxicity. No novel polymorphisms were detected, implying that polymorphisms in CYP2C9 are unlikely to contribute to diclofenac hepatotoxicity. However, investigation of the kinetics of 4'-hydroylation of diclofenac hepatotoxicity of diclofenac revealed that the Km was 5-fold higher for rare *3 variant (cloned and expressed in yeast) was by the rare *2 variant of the wild type *1 enzyme. This suggests that metabolism might be affected by the 2C9 genotype. In addition, an increased frequency of the -592 (C to A) polymorphism in the IL-10 promoter was defected (by Partner 4) in patients with diclofenac hepatotoxicity, compared to controls (OR: 2.84 [range 1.2-6.7]). This raises the possibility that antibody-medicated mechanisms are involve in disease pathogenesis

Summary: Clinical studies explored possible mechanisms of hepatotoxicity in humans. In vitro and in vivo studies showed that CYP 1A2 was the main enzyme involved in the metabolism of tacrine in man. Clinical studies showed that the risk of tacrine-induced hepatotoxicity in Alzheimer's patients was significantly increased in cases with an associated deficiency in glutathions transferases GST mu and theta. These and other studies suggest that tacrine-induced hepatotoxicity hepatotoxicity in Alzheimer's patients was significantly increased in cases with an associated deficiency in glutathions transferases GST mu and theta. These and other studies suggest that tacrine-induced hepatotoxicity may result from the combined presence of two different factors: -high CYPIA2 activity, resulting in an important production of possibly toxic metabolites -deficiencies in glutathione transferase activities resulting in the accumulation of these metabolites in the liver.

Summary: Precision-cut liver slices were used to explore development of novel markers of zone-specific liver toxicity caused by drugs that include paracetamol, fotemustine and tacrine. As markers for cellular stress, methods to detect heat shock proteins (HSPs), P53-up regulation, aldehydes, DNA-laddering (apoptosis), GST-isoenzymes, and several other enzymes were evaluated. Except for the aldehydes (as markers of lipid peroxidation), GST-isoenzymes and some of the other enzymes, none of the other markers proved sufficiently sensitive and/or reproducible. However, sensitive targets for reactive electrophilic intermediates formed by cytochromes P450 and flavine mono-oxygenases (FMP's), at least in the case of thiourea containing hepatotoxins, appeared to be the microsomal GST and Cyt P450 1A2. Interaction with both macromolecular targets could be detected easily by measuring enzyme activities.

Summary: Human cytochrome P450 isozymes that catalyze generation of hydroxylated metabolites from the NSAID diclofenac were studied using fully characterized human hepatic microsomes from a liver bank, isozyme-selective inhibitors, human hepatocytes and engineered cells expressing defined human CYPs. It was concluded that >99.5% of hydroxylation at position 4' (the major metabolic route) was catalyzed by CYP2C9. Based on relative CYP specific activities and abundance in hepatocytes, it was proposed that, at concentrations close to that found in plasma, 5-hydroxylation of diclofenac is catalyzed by CYP2C19 = CYP2C8 > CYP2B6, while 3' -hydroxylation is catalyzed by CYP2C19>CYP2C9. No evidence for formation of 3' methoxy diclofenac derivative was found, suggesting this metabolite could be if extraheptic origin. These studies provide the essential scientific underpinning required for investigations of the mechanism of hepatotoxicity of diclofenac.

Summary: The role of mitochondrial injury in drug-induced liver injury was explored. Premature oxidative aging of mitochondrial DNA was implicated in damage to human mitochondria by ethanol. Ethanol abuse was shown to increase the mitochondrial formation of reactive oxygen species, which may damage mitochondrial by ethanol. Ethanol abuse was shown to increase the mitochondrial formation of reactive oxygen species, which may damage mitochondrial DNA and cause mitochondrial DNA mutations. At least one, and often several, mitochondrial DNA deletions were observed in 85% of alcoholics with microvesicular steatosis, while only 3% of controls of similar ages exhibited one mitochondrial DNA deletion. Similar effects occurred in Wilson's disease, where the mitochondrial accumulation of copper increases formation of reactive oxygen species in hepatic mitochondria. Furthermore, inhibition of beta-oxidation enzymes by glucocorticoids was observed. Glucocorticoids were shown to inhibit mitochondrial matrix acyl CoA dehydrogenases, to decrease the beta-oxidation of medium chain and short chain fatty acids, and to cause microvesicular steatosis in mice.

Summary: A novel in vitro system for generation of protein adducts in human liver microsomes was developed and validated. Using this approach, antibody reactivity to possible human liver microsomal adducts could not be detected in patients with tacrine hepatotoxicity, although autoantibodies to specific P450 isozymes and microsomal epoxide hydrolase were detected. An immunoprecipitation approach, based upon biotinylation of liver subcellular protein fractions, was developed for identification of conformation-dependent protein adducts. Using this novel methodology, microsomal epoxide hydrolase was shown to be a major conformation-dependant neoantigen in halothane hepatitis. Availability of these new approaches should improve our ability to explore the role played by immune process in drug hepatotoxicity.

Summary: Drug-induced cellular immune responses were studied. Assays of proliferation of patients' peripheral blood mononuclear cells in vitro, when cultured in the presence of the relevant drugs plus indomethacin (a potent inhibitor of prostaglandin-producing suppresser T cells), were developed for diagnosis of immunoallergic drug hepatotoxicities. The potential use of defined drug metabolites to increase assay sensitivity was evaluated, while CD69 was evaluated as an early activation marker which could reduce the time required to run the assay. In addition, cellular immune responses towards diclofenmodified hepatic protein adducts were studied in mice treated with the drug. Lymphocytes from draining lymph nodes of animals immunized with synthetic diclofenac-mouse serum albumin, or with diclofenac-ovalbumin conjugates, were isolated and cultured, and shown to undergo adduct-specific proliferation in vitro. Treatment of unimmunized mice with diclifenac for 7 days was shown to generate protein adducts (notably 110 kDa adducts) that could be detected using adduct-specific rabbit antisera. However, liver injury was not observed when stimulated cells were adoptively transferred into drug-related recipient mice. If the reasons for this can be understood, it may be possible to develop reliable animal models of immunoallergic drug hepatotoxicity.

Summary: of mitochondrial respiration was demonstrated. This occurs because tacrine is protonated in the acidic intermembranous space of mitochondria membrane potential and ATP formation, and cause toxicity, in rat hepatocytes and human lymphocytes. It was shown that the toxicity was observed in hepatic non-parenchymal cells, implying that oxidative metabolism is not involved. Ultrastructural studies revealed that tacrine-induced aggregation of ribosomes, which has been reported previously by others, was not evident at concentrations <0.25 mM and so can be excluded as a direct cause of this toxicity, and moreover no overt effects on mitochondria were observed. Tacrine hepatotoxicity was not associated with increased lipid peroxidation and was not prevented by anti-oxidants. However, tacrine was found to cause changes in intracellular pH and its regulation.

Summary: Expression and distribution of major phase I and phase II drug-metabolizing enzymes in freshly isolated biliary epithelial cells (BEC) lining the entire human biliary tract was assessed. Expression of CYPs and of microsomal epoxide hydrolase (mEH) was 5 to 20-fold lower in BEC than in hepatocytes and further more BEC had higher ratios of CYP3A5/CYP3A4, and CYP1A1 vs. CYP1A2, compared with hepatocytes. alpha-GST was highly expressed in both BEC and hepatocytes, while pi-GSTwas restricted to BEC. In about 50% of individuals mu-GST was expressed in hepatocytes and at lower levels in BEC. Immunohistochemical studies revealed that CYP2E1, CYP3A, mEH, and alpha, mu, and pi-GSTs, were expressed heterogeneously in the biliary tract, except that small intrahepatic bile ducts were devoid of CYP3A and alpha-GST. These marked differences in enzyme distribution in enzyme distribution in the anatomical segments of the biliary tract could explain the specific distribution of cholangiopathies due to drugs such as flucloxacillin.

Summary: Cytotoxicity of flucloxacillin in vitro towards human biliary epithelial cells (BEC) and hepatocytes was hepatocytes was explored. Although flucloxacillin was not toxic to hepatocytes or gallbladder biliary epithelial cells, supernatants from hepatocytes that had been incubated with flucloxacillin were toxic to biliary epithelial cells in 6 to 11 preparations. This toxicity was reproduced, although to a lesser extent, when human liver microsomes or recombinant CYP3A4 expressed in yeast were used instead of hepatocytes, implying involvement of toxic metabolites (which were detected by HPLC in human hepatocyte culture supernatants). In addition, preliminary results indicted that hepatic protein adducts were generated following exposure of BEC to conditioned medium from flucloxacillin-treated hepatocytes. These various findings suggest that interactions between hepatocytes and BECs could play a key role in cholangiopathy caused by flucloxacillin and by other drugs.

Summary: An animal of NSAID-induced liver injury was produced by daily intraperitoneal administration of diclofenac or sulindac to rats for up to 7 days. Liver injury was not evident in rats treated ip with ibuprofen for up to 7 days. Mechanistic studies demonstrated that the pattern of liver injury was hepatocellular, but that this was accompanied by impairment of bile flow and by re-distribution of protein adducts from the canalicular plasma domain to cytoplasm. The protein adducts were detected immunochemically, using specific antibodies, and generation of the diclofeanc adducts was shown to involve reactive acyl glucuronides. It was proposed that the liver injury arose as a consequence of accumulation within hepatocytes of toxic bile acids and/or reactive metabolites (which can cause toxicity via an NADPH-consuming futile cycle). Drug-sensitized lymphocytes were identification in peripheral blood from patients with diclofenac-induced liver injury using in vitro lymphocyte proliferation tests, as were adduct-specific antibodies. Overall, the data obtained suggest that hepatotoxicity of many NSAIDs in humans could involve a combination of toxicity arising from protein adduct-induced impairment of canalicular transporters and immune responses directed against the adducts.

Summary: Intracellular trafficking of CYP2C11 and tienilic acid-modified CYP2C11 from the endoplasmic reticulum to the outer surface of the plasma membrane, via vesicular membrane flow, has been demonstrated in livers of rats treated with the diuretic tienilic acid. This raises the possibility that autoantibodies against CYP2C and antibodies against adduct-modified CYP 2C, which have been demonstrated in patients with tienilic acid-induced hepatotoxicity, could contribute to the process of liver injury.

Summary: Although cancer patients may be treated with both IL-2 immunotherpay and anti-cancer drugs that metabolized by cytochrome P450s, the effects of IL-2 on CYP expression had not been studied in humans. IL-2 immunotherapy was shown to decrease hepatic CYPs and monooxygenase activities in cancer patients, suggesting that drug interactions could occur, and that the interaction of IL-2 with its hepatic receptor triggers tyrosine kinase-dependant c-myc overexpression and CYP downregulation in cultured rat hepatocytes. Preventing c-myc overexpression by 5 different agents (acting at different steps) always preventing CYP down-regulation, indicating a direct or indirect role of c-myc overexpression in downregulation of CYP expression.

Summary: The role of GST-mediated metabolism as a confounding factor for hepatotoxicity (venous occlusive disease) caused by the anti-cancer drug busulfan was studied. Endothelial toxicity and cell cycle regulation was investigated using transfected cell lines derived from the endothelial cell line ECV 304, which is devoid of expression of GST alpha. Cells were transfected with the expression plasmid pTR-GST alpha, or with the respective mock transfection vector, using cationic lipids. In contrast to untransfected and mock-transfected cells, which contain only background levels of GST alpha, transfected cells express high amounts of the enzyme. FACS analysis was used as a functional parameter for investigation of endothelial toxicity of busulfan, and also to study the influence of busulfan on cell cycle characteristics. After incubation with 250mM busulfan, a busulfan-induced increase in G2-phase was detected in the mock-transfected cells that was significantly reduced in GST alphas-transfected cells compared to the control after 12 and 24 hours. It can be concluded that busulfan causes changes in cell cycle characteristics, ad these alterations are modulated by GST-expression.

Summary: A case-control study has been undertaken to determine the relationship between NSAIDs exposure in out-patients and liver injury (whether symptomatic or asymptomatic) and to analyze the type of liver injury, the main NSAID suspected and the potential risk factor. The necessary number of subjects was calculated as 87 cases and 174 controls, while 61 cases and 152 controls have been included to date. 26% of reporting came from gatroenterologists and 74% from GP. The study is due to end in December 1999. There was no difference between cases and controls concerning sex and age parameters. 15 cases of NSAIDs-related liver injury identified to date. The drugs involved were ketoprofen, diclofenac and aspirin (4 cases respectively) and flurbiprofen, meloxicam and niflumic acid (1 case respectively). The delay of the onset of the reactions was 50 days (6-170 days) and in a number of instances other drugs had been co-administered with NSAIDs (acetaminophen in 3 cases, dextropropoxyphen in 4 cases, penicillins in 2 cases and macrolide in 1 cases). Liver injuries were hepatocelluar and cholestatic in 3 cases each and mixed in 4 cases. This study has established that administration of NSAIDs is a significant cause of liver injury in the out patients studied.

Deliverables

Share this page Share this page on social networks

Download Download the content of the page