Journal of Hepatology 2000; 32 (suppl. 1): 77-88 Printed in Denmark AN rights reserved Munksgaard Copenhagen
Copyright 0 European Association for the Study of the Liver 2000 Journal of Hepatology ISSN 0169-5185
Drug-induced liver diseases Dominique
Service d’Ht!patogastro-entPrologie et Transplantation h$atique.
H6pital Saint-Eloi, Montpellier. France
Drug-induced liver injuries make up a persisting and challenging problem for physicians, health agencies and pharmaceutical firms. The clinical expression is polymorphous, acute hepatitis being predominant. The diagnosis is frequently difficult because of the ahsence of specific signs in most cases and mainly relies on the exclusion of other causes. The diagnosis should he particularly evoked in patients over 50 yr who are taking many drugs, after viral infections have been ruled out. Acute hepatocellular hepatitis is particularly severe because of the risk of fulminant hepatitis or of a more insidious course leading to cirrhosis.
Cross hepatotoxicity can sometimes occur. One should avoid re-administration of not only the causative agents but also of other drugs belonging to the same family or having a related chemical structure. The prediction of the hepatotoxicity of new drugs must be improved. Investigations would be particularly useful for drugs having critical chemical structures and belonging to families with an established history of hepatotoxicity.
DEVELOPMENT of new diagnostic and therapeutic methods is regularly improving the management and prognosis of most diseases but it is also associated with the occurrence of new iatrogenic diseases. A recent survey done in the Hepatogastroenterology Unit of Montpellier School of Medicine showed that 9% of patients were admitted because of disorders having an iatrogenic origin (1). Drug-induced liver injuries were one of the major problems (1). Liver toxicity is permanently developing and remains the first cause of drug-induced death and withdrawal of drugs from the pharmaceutical market (2-4). Despite improvement in toxicological studies and in the safety analysis of clinical trials, the frequency of drug hepatotoxicity has not decreased in the last 10 yr (5). The spectrum of liver damage caused by drugs is very broad (24). Indeed, all cells present in the liver can be affected by drugs, as shown in Fig. 1. Almost the entire spectrum of liver injuries can be reproduced by drugs. This explains the concern of physicians, health authorities and pharmaceutical companies about drug hepatotoxicity. The purpose of this article is to summarize
the most important aspects of drug hepatotoxicity at the beginning of 2000 and to discuss future directions for management and prevention in this field.
Key words: Drug-induced medicines.
Drug hepatotoxicity: past and present The toxic effects of drugs on the liver have remained ignored or, at least, underestimated for a long time. For some well-known drugs the discovery of hepatotoxicity occurred several decades after their first use; for instance, 100 yr for aspirin, 40 yr for papaverine and 25 yr for amiodarone (2-6). The hepatotoxicity of herbal medicines in Western countries has been recog-
Hepatocyte acute hepatitis - cholestasis l chronic hepatitis l cirrhosis l steato*1s l phosphollpldosis l steatohepatitis l granulomatous hepatitis
Correspondence: D. Larrey, Service d’Htpato-Gastroenterologie, HBpital Saint Eloi, 2 avenue Bertin Sans, 34295 Montpellier cedex 5, France. Tel: +33 4 67 33 70 61. Fax: +33 4 6752 3897. e-mail: [email protected]
acute and chronic cholangitis sclerosing cholangitis
Endothelial cell - ~eno-cccIus~ve disease l sinusoidal dilation l pellosis hepatltis l Budd Chiari syndrome
It0 cell l
Fig. 1. Cells involved in drug-induced hepatotoxicity.
nized for about 10 yr (7) and we are just understanding that excipients, combined with classical drugs, could also be implicated (8). The review of the past and present major aspects of drug hepatotoxicity will be mainly focused on the epidemiology, some mechanisms and the factors modulating hepatotoxicity, the diagnostic procedures, the principal clinicopathological patterns and the main categories of drugs or other xenobiotics which have been involved. Epidemiological aspects
The epidemiology of drug hepatotoxicity remains poorly documented despite the creation of specific Drug Safety Departments in most countries 15-20 yr ago. These organizations were set up to a) elaborate analytical methods for diagnosis; b) record adverse events and analyze their prevalence and incidence; c) establish the mechanisms or contributing factors involved in order to propose measures adapted for the suspected drug; d) withdraw drugs, place restrictions on drugs, associate drugs with specific controls; e) and finally, disseminate information about all of the above. Despite the many efforts made worldwide, the results have been disappointing (9). There are at least three reasons for this relative failure (9). First, the diagnosis of drug hepatotoxicity remains difficult, as we will see below. Second, there is a limited number of epidemiological studies. Clinical or toxicological studies performed at the pre-marketing stage provide some prospective data with which to evaluate the prevalence of biological abnormalities or more clinically relevant events (2-4). But most available information comes from the spontaneous reporting of individual cases to the Drug Safety Authorities. Thus, many cases remain ignored and we are only aware of the tip of the iceberg. There are very few studies assessing the relative prevalence of drug-induced liver injury compared to other causes, all of them being based on the retrospective analysis of large series. A IO-yr study in the liver unit of Hdpital Beaujon, Paris, revealed that among all adult patients admitted with acute hepatitis, 10% of cases were related to drug toxicity (9). The prevalence of drug hepatotoxicity exceeded 40% for patients over 50 yr of age. In another French national survey, made in 1983, which collected 980 cases of hepatitis due to drugs, 63% of the patients were women and most were over 50 yr old (10). Similarly, the Danish Board of Adverse Reaction to Drugs recorded 572 cases of drug-induced hepatitis during 1968-78 (6% of all drug-related side effects), and most cases occurred in women over 50 yr old (11). Drugs have been estimated to cause around 78
15-20% of all cases of fulminant and subfulminant hepatitis in Western countries (12) and 10% of all cases in Japan (13). The risk of fulminant course is around 20% in patients with drug-induced hepatitis with jaundice, which is much greater than the risk in patients suffering from acute viral hepatitis with jaundice (1%). Furthermore, in 70% of the patients with drug-induced hepatitis with encephalopathy, the disease runs a subfulminant cause (12). For chronic liver diseases, such as chronic hepatitis or cirrhosis, drugs appear to be rarely involved, being the cause in less than of 1% of cases (24). The third difficulty is the wide variability of hepatotoxicity prevalence from one drug to another. Drugs with a high prevalence (over 1%) are quickly eliminated before marketing. The higher frequency for marketed drugs is around l%, the major examples being isoniazid, chlorpromazine and tacrine (24,6). In contrast, for other drugs, the risk of hepatotoxicity is exceedingly low, below l/100000 or even l/l 000000, for instance, with anti-histaminic compounds or penicillin (2-4,6). For most drugs, the risk of toxicity ranges from l/ 10 000 to 1/100000. This explains probably why toxicity has not been detected during therapeutic trials done to obtain marketing authorization. Most trials include 1000-3000 patients (9). Therefore, the first cases of hepatotoxicity are generally described within l-2 yr after starting marketing when a sufficient number of patients have been exposed to the new drug (9). The overall number of drugs liable to be toxic to the liver exceeds 1100 (5). To this, one should add chemical agents in herbal remedies, illegal drugs such as cocaine and some amphetamines (Ecstasy) and even excipients used with conventional drugs, as recently demonstrated. Mechanisms of hepatotoxicity
Despite many advances in the last 20 yr, hepatotoxicity mechanisms of most drugs remain unknown (2-4,14). A single drug may have several toxic effects on the liver, or may produce either toxic or allergic hepatitis in different patients. The formation of reactive metabolite is relatively frequent (24,14,15). Despite several protective mechanisms, it may lead to the covalent binding of electrophilic metabolites to proteins, the mutation of lipid peroxidation by free radicals, and the depletion and/or oxidation of glutathione. There follow a number of structural and functional lesions, including a sustained increase in cytosolic calcium, eventually leading to cell death (24,14,15). The formation of reactive metabolites may lead to two major types of hepatitis in humans (24,14,15): 1) toxic hepatitis, which can occur predictably after massive overdoses,
Drug-induced liver diseases
e.g. paracetamol, and 2) immunoallergic hepatitis, in which the drug triggers an adverse immune response directed against the liver. The main clinical features are the dose independence, the association with hypersensitivity reactions such as fever, chills, skin rash, hypereosinophilia and immunoallergic thrombopenia, the shortened delay upon rechallenge and the occasional presence of serum autoantibodies. These autoantibodies have been observed in cases of hepatitis caused by halothane, tienilic acid, dihydralazine, anticonvulsants, papaverine and nitrofurantoin (15-20). The following mechanism has been postulated (4,15): 1) The drug is first metabolized into a reactive metabolite, which binds to the enzyme that generates it. 2) This produces a neoantigen, which, once presented to the immune system, may trigger an immune response characterized by the production of antibodies recognizing both the natural and/or the modified protein. 3) Rechallenge leads to increased neoantigen production, a situation in which the presence of antibodies may induce cytolysis. Recently, several lines of evidence have suggested that apoptosis may also contribute to hepatotoxicity (4). Microvesicular steatosis induced by several drugs, including non-steroidal anti-inflammatory drugs, some tricyclic antidepressants, tetracycline and acetylsalicylic acid, is caused by an inhibition at different stages of the mitochondrial beta-oxidation of fatty acids (21). Other mechanisms of hepatotoxicity are summarized in Table 1. Factors modulating hepatotoxicity
The risk of drug hepatotoxicity is influenced by various acquired and genetic factors summarized in Tables 2 and 3 (2-4,22). Acquired factors - Age above 60 yr is a promoting
factor for isoniazid and nitrofurantoin hepatitis, whereas children are more susceptible to toxicity from valproic acid and salicylates (2,3). In particular, the latter drug more frequently induces microvesicular steatosis and Reye’s syndrome (24). Gender is also involved, women being more frequently susceptible to methyldopa and nitrofurantoin hepatotoxicity in contrast to men who are more prone to azathioprine-induced injury (2,3). Quality of nutrition can affect hepatotoxicity in different ways. For instance, obesity promotes halothane hepatotoxicity, whereas fasting and denutrition facilitate paracetamol hepatitis, probably by depleting glutathione stock in hepatocytes (24). Pregnancy also appears to be an influencing factor. For instance, most tetracycline-induced severe hepatitis has been observed in pregnant women receiving the drug by the intravenous route (24). Experimentally, pregnant mice are particularly susceptible to paracetamol hepatotoxicity, probably because of an overuse of glutathione by the fetus and the placenta (23). Chronic alcohol abuse facilitates paracetamol hepatotoxicity, probably by complex mechanisms involving the induction of critical enzymes such as cytochrome P-450, (CYP2El) forming toxic metabolites from paracetamol, and the lowered resistance to these metabolites because of glutathione depletion (24). Drug interactions can also contribute to drug hepatotoxicity in different ways. For instance, the role of enzyme induction, leading to an increased formation of toxic metabolites from one drug, is demonstrated with the rifampicin-isoniazid combination in which rifampicin facilitates the transformation of isoniazid into toxic metabolites (24). Enzyme induction by phenobarbital can induce the hepatotoxicity of antidepressants by a similar mechanism (24). In contrast, enzyme inhibition can also be implicated. This is dem-
of hepatotoxicity Liver injuries
Acute cholestasis Macrovacuolar steatosis Microvesicular steatosis
Inhibition Decreased Inhibition
Phospholipidosis Chronic hepatitis Vanishing bile duct syndrome
Inhibition of lysosomal phospholipases Metabolite-mediated immune reaction Autoimmune destruction of small bile ducts. Abnormal multidrug resistance protein system?
Sclerosing cholangitis Veno-occlusive disease Persinusoidal fibrosis
Biliary ischemia caused by arterial lesions Metabolite-mediated endothelial lesions Activation of Ito cells
of biliary secretion secretion of lipoproteins of fatty acid mitochondrial
D. Larrey TABLE
to drug hepatotoxicity
>60 yr Children
Isoniazid, nitrofurantoin Valproic acid, salicylates
abuse Enzyme Enzyme
in in in in in in
CYP 2C19 NAT2 sulfoxidation glutathione synthetase GSTase type T detoxication of reactive metabolites
on&rated by the troleandomycin-estrogen interaction. Troleandomycin inhibits estrogen metabolism by blocking cytochrome P-450 (CUP 3A4), creating an estrogen overdose with cholestatic effects (4,14). Extrahepatic diseases can also contribute to drug hepatotoxicity. For instance, hyperthyroidism promotes halothane hepatitis and HIV infection the hepatotoxicity of cotrimoxazole (24). Genetic factors (22) The de$ciency in cytochrome P-450 06 (CYP 2D6) ap-
pears to be a major determinant of perhexiline hepatotoxicity. Indeed, more than 75% of patients with perhexiline hepatotoxicity are CYP 2D6 deficient (24). This enzyme deficiency is present in 668”/ of Caucasian populations and is transmitted as an autosomal recessive trait (25,26). However, CYP 2D6 does not appear to be involved in the hepatotoxicity of many other drugs metabolized by this enzyme (27). The deficiency in cytochrome P-450 2C19 (CUP 2C19) may be involved in Atrium@ hepatotoxicity (28). Atrium is a complex drug combining febarbamate, difebarbamate and phenobarbital. A recent study com80
Rifampicin - isoniazid Troleandomycin - estrogens
Perhexiline Atrium Sulfonamides, dihydralazine Chlorpromazine Paracetamol Tacrine? Halothane, phenytoin Carbamazepine, amineptine Sulfonamides Halothane, tricyclic Chlorpromazine
to drug hepatotoxicity
_ Deficiency in CYP 2D6 - Deficiency - Deficiency Deficiency _ Deficiency _ Deficiency _ Deficiency
prising a small number of patients with a previous history of Atrium hepatitis showed that all of them had a partial or complete deficiency in CYP 2C19, whereas the prevalence of this deficiency in the control population was only around 3-5% in Caucasian populations (26, 29). This example, however, needs to be confirmed by the study of a larger number of patients. The deficiency in acetylation capacity related to an inactive acetyltransferase 2 is incriminated in the hepatotoxicity of sulphonamides and hydralazine (4). It is transmitted as an autosomal recessive trait (26). The high frequency of the slow acetylation phenotype, in most populations, suggests that this deficiency may contribute to but is not sufficient for the toxicity of these two types of compounds (4). The de$ciency in sulfoxidation has been incriminated in one study assessing patients with chlorpromazine hepatitis. However, the method used to analyze sulfoxidation is not reproducible, making this conclusion questionable (4,26). Further studies using reproducible tests for sulfoxidation polymorphism are required. The dejciency in glutathione synthetase is an uncommon condition (l/10000), responsible for a syndrome
Drug-induced liver diseases
comprising oxoprolinuria and hemolytic anemia (4,30). By using an in vitro lymphotoxicity assay, Spielberg and colleagues have shown that the deficient subjects may be more susceptible to paracetamol hepatotoxicity (30). Other defkiencies - By using the in vitro lymphotoxicity assay developed by Spielberg et al., several groups have shown deficiencies in the detoxification capacity of reactive metabolites in patients with drug-induced hepatitis (4). This deficiency could also be observed in some members of their family, suggesting a genetic defect. Such observations have been made for halothane, phenytoin, carbamazepine, amineptine and sulfonamides (4, 31). The precise defects involved in these mechanisms have still not been identified. Glutathione S-transferases - It has been proposed that tacrine hepatotoxicity could be promoted by a deficiency in glutathione S-transferase of type T (32). However, another study (33) did not confirmed this. Genetic variations in the immune system could be also involved in drug hepatotoxicity. Indeed, an association has been observed between several HLA haplotypes and some drugs (Table 3) (34). Diagnosis In the last 20 yr, several analytical methods have been proposed to assess the causality of a given drug in the occurrence of liver injury (24). In 1990, an international consensus group proposed definitions of adverse reactions and criteria for assessing causality of druginduced liver diseases to standardize the evaluation of drug hepatotoxicity by physicians, health authorities of different countries and pharmaceutical manufacturers (35). This method is now largely used in Europe and North America. The principles of the international method and other previous methods are the same. The causality assessment relies on chronological and clinical criteria to allow elimination of other causes and to demonstrate the role of the offending drug (Table 4) (24,35). Chronological criteria (35) The first criterion is the time interval between the beginning of the suspected treatment and the onset of liver injury; this varies widely. It is considered suggestive when the interval is between 1 week and 3 months. A shorter duration (1 or 2 days) may be observed in patients who have been previously exposed to the compound and have been sensitized. A delay of between 3 months and 1 yr remains compatible but is less common. A delay above 1 yr is very uncommon and makes the role of the suspected drug very unlikely in the case of acute hepatitis.
Chronological criteria _ Interval between the beginning of the treatment and the onset of liver injury: 1 week-3 months ~ Regression of liver abnormalities after withdrawal of the treatment _ Relapse of liver abnormalities after accidental readministration of the offending drug Clinical criteria
Elimination of other causes _ Previous hepatic or biliary disease ~ Alcohol abuse Viral hepatitis (HAV, HBV, HCV, HDV CMV, Epstein-Barr Herpes, HGV (?), TTV (?)) _ Biliary obstruction (ultrasonography etc.) _ Autoimmune hepatitis/cholangitis _ Liver ischemia _ Wilson’s disease _ Bacterial infection (Listeria, Campylobacter, Salmonella)
Positive clinical criteria - Age >50 yr ~ Intake of many drugs ~ Intake of a known hepatotoxic agent ~ Specific serum autoantibodies: anti M6, anti LKM2, lA2, anti CYP 2El _ Drug analysis in blood: paracetamol, vitamin A ~ Liver biopsy: drug deposit (vitamin A), microvesicular eosinophil infiltration, centrilobular necrosis
The second criterion is the disappearance of liver abnormalities after withdrawal of the treatment. It is very suggestive when clinical features disappear within a few days and when aminotransferases decrease by more than 50% in a week. Usually, complete recovery is obtained within a few weeks. The third criterion is a relapse of liver abnormalities after an accidental readministration of the offending drug; this is a very good diagnostic criterion. However, this re-exposure should not be done on purpose because it can be very dangerous, particularly in immunoallergic hepatitis. In this situation, the readministration of a single tablet can occasionally provoke a fulminant hepatitis. Clinical criteria (24) Clinical criteria are based on the exclusion of other causes which might explain the liver injury and on the presence of features tending towards drug causality. Eliminating or negative criteria - Analytical features vary according to the type of liver injury. For acute hepatitis, it is important to look for a history of liver or biliary disease, alcohol abuse and epidemiological circumstances compatible with viral infection (drug addiction, blood transfusion, recent surgery, travel in an endemic area). Appropriate serological analysis should be performed for major viral hepatitis (hepatitis A, B, C, D, E), and in some circumstances for cytomegalo81
virus, Epstein-Barr virus and herpes viruses. The usefulness of testing virus G and virus TT remains to be established. A potential liver ischemia related to heart dysfunction should be ruled out, particularly in the elderly. Biliary obstruction should be eliminated by ultrasonography or other appropriate examinations. One should also rule out autoimmune hepatitis or cholangitis, some bacterial infections which may mimic acute hepatitis, such as infection by Campylobacter, Salmonella and Listeria. Finally, Wilson’s disease should be ruled out in young patients. Positive criteria ~ The positive clinical criteria for drug hepatotoxicity are listed in Table 4. The presence of specific serum autoantibodies, for instance antimitochondrial type 6, anti-LKM2, anti-CYP lA2 and anti CYP 2El is an important diagnostic marker (1620). Drug analysis in blood and liver tissue may be useful, for instance for paracetamol and vitamin A overdoses. The presence of hypersensitivity manifestations, albeit not completely specific, is a positive argument not only for the involvement of a drug but also for an immunoallergic mechanism. Finally, liver biopsy may also contribute to the diagnosis by showing the presence of drug deposits (vitamin A) or lesions suggestive of drug reactions, for instance microvesicular steatosis, eosinophil infiltration or centrilobular necrosis. Causality assessment (2-4,3.5) At the end of these enquiries, the diagnosis is more or less evident. The diagnosis is very likely in rare circumstances: clear drug overdose (paracetamol); relapse after accidental readministration; presence of specific features for drug hepatitis. The diagnosis is compatible in many cases: liver injury exhibits no specific criteria. The history is chronologically suggestive and other causes of liver injury have been reasonably ruled out. The diagnosis frequently remains doubtful: liver in-
in the diagnosis
Nonspecific clinical features Treated disease itself leading to liver abnormalities tion) ~ Intake of several hepatotoxic drugs (combined agents) Compounds considered safe (herbal remedies) - Drug prescription difficult to analyze: automedication masked information (illegal compounds) - forgotten information (elderly) -- Fulminant hepatitis
jury has no specificity and some information regarding chronological or clinical data is missing. It is noteworthy that fulminant hepatitis is always classified in this group since it is impossible to assess the complete recovery after drug withdrawal. The diagnosis is incompatible: when another cause has been demonstrated (viral infection); when chronology is not compatible (when the treatment has been started while symptoms were already present or there is a long delay above 15 days between the end of the treatment and the onset of liver injury). However, there are two exceptions. With halothane and its derivatives, hepatitis generally occurs 3 weeks after the first exposure. Similarly, for the clavulanic acid-amoxicillin combination, acute hepatitis frequently occurs 3-4 weeks after the discontinuation of the treatment. Main diagnostic [email protected]
The main diagnostic difficulties are shown in Table 5.
Liver biopsy Liver biopsy is not necessary in most cases. However, it is indicated 1) to eliminate other causes of liver injury; 2) to show lesions suggestive of drug-induced liver injury; 3) to define lesions for drugs with so far unknown hepatotoxicity. Clinicopathological manifestations Drug toxicity to the liver can reproduce practically the whole spectrum of liver disease (Fig. 1). However, acute hepatitis represents the most frequent presentation (90% of cases) (24). Acute hepatitis can be classified into three groups using biochemical criteria based on the serum activity of alanine aminotransferase (ALT) and alkaline phosphatase (AP), expressed as the number of time of the upper limit of normal and their ratio (R) (35). This classification has the advantage of separating types of hepatitis with quite different courses and prognostic features. Acute hepatocellular hepatitis (2-4) Acute hepatocellular hepatitis is defined by ALT above 2N or ALT/AP z-5 (35). Acute hepatocellular hepatitis generally has no specific features and mimics acute viral hepatitis. Liver injury may remain asymptomatic, revealed only by an increase in ALT or by nonspecific symptoms such as asthenia or anorexia. Jaundice is inconstant. The main biological feature is the marked increase in aminotransferases. The major pathological finding is liver cell necrosis generally associated with inflammatory infiltration. The presence of eosinophils in the infiltrate and the centrilobular predominance of the lesions argue for drug hepatotoxicity.
Drug-induced liver diseases TABLE
Main drugs responsible
for acute hepatocellular
Main conventional drugs ~ without hypersensitivity: paracetamol, isoniazid, pyrazinamide, ketoconazole, valproic acid ~ with hypersensitivity: non steroidal inflammatory drugs (almost all drugs), sulfonamides, almost all antidepressants (tricyclic, iproniazid, fluoxetine), halothane and derivatives New causative drugs Psychotropic and neurotropic drugs (tacrine, paroxetine, sertraline, tolcapone, topiramate, riluzole), anti-HIV (didanosine, zidovudine, zalcitabine, stavudine, ritonavir, indinavir, saquinavir), anti-mycotics (terbinafine), cytokines and growth factors (interleukin 2, interleukin 12, interleukin 2F, interleukin 3, interleukin 6, G-CSF) Herbal medicines Pyrrolizidine alkaloids (crotaloria, senecio) Atractylis gun?m$era L., germander, Chinese herbal preparations, chaparral leaf, senna, skullcap, valerian, mentha containing Pennyroyal oil IlIegal compounds Cocaine, Ecstasy@ (dioxyamphetamine) Excipients Sodium saccharinate (dihydroergoscristine, chlordemethyldiazepam, prednisolone 20 mg), polysorbate (E-Ferol syndrome, IV amiodarone), propylene glycol (Multivitamin IV solution) Chemical agents Carbon tetrachloride, trichloroethylene, dimethylformamide, vinyl chloride * For a more complete
listing, see references
Hepatitis may be associated with hypersensitivity, which suggests an immunoallergic mechanism. Several hundred drugs can induce this type of hepatitis (24). The main ones are indicated in Table 6. In most instances, discontinuation of the treatment is followed by a quick improvement of symptoms and a complete recovery within l-3 months. Sometimes, however, acute hepatitis may be followed by fulminant or subfulminant hepatitis. The course is then very severe with a spontaneous mortality around 90% (12,13). The only treatment is emergency liver transplantation. The risk of developing fulminant or subfulminant hepatitis is particularly high when drug administration is continued despite the occurrence of jaundice (12). The course of the disease may be more insidious with the progressive development of chronic hepatitis or even cirrhosis. Several drugs can induce this type of hepatitis (2-4). It may be also produced by herbal medicines (7), illegal compounds, chemical agents (24), and even some excipients (8) (see Table 6). Acute cholestatic hepatitis (2-4,36,37) Acute cholestatic hepatitis is characterized by an isolated increase of serum alkaline phosphatase above 2N or by a ratio 12, with two subtypes: pure cholestasis and cholestatic hepatitis (35).
Pure cholestasis is mainly characterized by jaundice, pruritus and dark urine. Biologically, there is an increase in alkaline phosphatase, conjugated bilirubin and gammaglutamyltranspeptidase. Transaminases are normal or only slightly increased. When performed, liver biopsy mainly shows bilirubin deposit in hepatocytes and dilated biliary canaliculi containing biliary pigments. These lesions predominate in the centrilobular area. Pure cholestasis is observed with a few drugs, mainly hormonal derivatives (see Table 7). The discontinuation of the causative agent is followed by a complete recovery. Acute cholestatic hepatitis - In addition to the manifestation of pure cholestasis, this type of liver injury may be associated with abdominal pain, fever and chills which can mimick acute biliary obstruction. Hypersensitivity manifestations are frequent. Cholestasis occurs with inflammatory infiltration in portal tracts. The prognosis is much better than that of hepatocellular hepatitis. After drug withdrawal, symptoms rapidly disappear and recovery occurs within a few weeks. Rarely, however, chronic cholestasis mimicking primary biliary cirrhosis may develop (3840). This particular course is observed in association with cholangitis (39,40). Several hundred drugs are able to cause cholestatic hepatitis, the main ones being indicated in Table 7. Mixed pattern acute hepatitis (24) Mixed pattern acute hepatitis is characterized by an ALT/alkaline phosphatase ratio of between 2 and 5
Main drugs responsible
for acute cholestasis*
Pure cbolestasis Oral contraceptives, estrogens, estrogens + troleandomycin thromycin, androgens, tamoxifen, azathioprine, cytarabine Acute cholestatic hepatitis Conventional drugs - Phenothiazines - NSAIDs - Macrolides - Sulfonamides - Beta-lactam antibiotics
_ tricyclic antidepressants _ carbamazepine _ amoxicillin/clavulanic acid - gold salts ~ propoxyphene
New drugs Anti-HIV: didanosine, zidovudine, Interleukins: IL2, IL& IL12
Acute cholangitis - phenothiazines, ajmaline, carbamazepine, tricyclic antidepressants, macrolides, amoxicillin/clavulanic acid, dextro-propoxyphene Chronic cholangitis _ phenothiazines, ajimaline, pressants _ macrolides, thiabendazole, * For a more complete
(35). The clinicopathological manifestations correspond to a mixture of those observed with hepatocellular and cholestatic hepatitis. Jaundice is frequent as well as manifestations mimicking biliary obstruction. The prognosis is generally very good and it is very uncommon to observe a course to fulminant hepatitis or cirrhosis. The main causative drugs are tricyclic antidepressants, NSAIDs, sulfonamides, macrolides, and propoxyphene. Mixed pattern hepatitis is frequently associated with immunoallergic manifestations. Other types
The array of clinical syndromes and liver pathology produced by drugs almost mimicks all known hepatobiliary diseases as summarized in Fig. 1 and Table 8. Some drugs almost always produce the same liver disease; for instance, acute hepatocellular hepatitis with paracetamol overdose, or immunoallergic cholestatic hepatitis with chlorpromazine. In contrast, other drugs can produce many different hepatobiliary diseases varying from one patient to another. For instance, car-
The only example of well-established treatment consists in preventing hepatitis in patients with paracetamol overdose by giving N-acetylcysteine to detoxify formed reactive metabolites (41). N-acetylcysteine should be given within the first 10 h to obtain the greatest protective effect (41). In other cases, there is no specific treatment for drug-induced liver injury (24), the main measure being to stop the administration of the offending agent. In some cases, symptomatic treatment may be useful, for instance to relieve pruritus in acute cholestasis or to compensate vitamin malabsorption in chronic cholestasis. The usefulness of corticosteroids in immunoallergic hepatitis has not been demonstrated (24). The administration of ursodeoxycholic acid has been proposed for long-lasting chronic cholangitis (4,39). However, the small number of individual cases have provided no clear evidence of efficiency.
(2-5) Chronic hepatitis antior cirrhosis Valproic acid, amiodarone, aspirin, benzarone, halothane, iproniazid, isoniazide, methotrexate, methyldopa, nitrofurantoin, vitamin A, papaverine, herbal medicines (germander, chaparral) Granulomatous hepatitis Allopurinol, carbamazepine, phenylbutazone, penicillanine,
Microvesicular steatosis NSAIDs, valproic acid, tetracycline,
Phospholipidosis and steatohepatitis Amiodarone, perhexiline, diethylaminohexestrol,
Vascular disease of the liver ~ perisinusoidal fibrosis: vitamin A, azathioprine, 6_mercaptopurine, methotrexate ~ sinusoidal dilatation and peliosis hepatitis: oral contraceptives, androgens, estrogens, azathioprine ~ veno-occlusive disease: pyrrolyzidine alkaloids, azathioprine, antineoplasic agents - Budd-Chiari syndrome: oral contraceptives, dacarbazine Tumors - Adenoma and hepatocellular carcinoma: androgens, oral contraceptives and estrogens ~ Acute and/or chronic small bile duct lesions mimicking primary biliary cirrhosis: chlorpromazine and its derivatives, tricyclic antidepressants, macrolides, clavulanic acid - amoxicillin combination, herbal medicines (germander, chaparral) ~ Acute and/or chronic large bile duct lesions mimicking primary sclerosing cholangitis: floxuridine, form01 and hypertonic saline used for surgery of hydatid cyst
bamazepine can lead to all types of acute hepatitis, granulomatous hepatitis, and cholangitis (2-4). Other syndromes with causative drugs are reviewed in detail in references 24.
Prevention begins with detection of toxicity in animal or cellular models at the preclinical stage and then safety analysis in healthy volunteers or patients included at the various steps of drug development. Despite large improvements, the present methods are not sufficient to eliminate all risks of hepatotoxicity. Therefore, we continue to observe hepatotoxicity with low frequency (
Drug-induced liver diseases
P-450 isoform (macrolides for CYP 3A, quinidine for CYP 2D6) (42,43). In contrast, the major inducers are rifampicin, barbiturates, phenytoin, and, more selectively, dexamethasone on CYP 3A, omeprazole for CYP 1Al and lA2. Control administration of drugs to patients with malnutrition who lack defenses against reactive metabolites and to alcoholic patients (24). Remember that old patients are more susceptible to drug hepatotoxicity (24). Be very careful when administering drugs to patients with HIV infection (24). They incorporate several risk factors: the co-administration of many drugs, individually susceptible to cause occasional hepatitis (antibiotics, antimycotics, anti-retroviral agents); a decreased ability to detoxify drugs, in particular because of malnutrition; a higher susceptibility to some drugs, for instance, sulfonamides and cotrimoxazole (2-4). Although genetic factors have been experimentally demonstrated, there is no easy application. Some tests are complicated to perform (in vitro cytotoxicity assay) and genotyping or phenotyping tests are not widely available (22). Finally, the follow-up of aminotransferase level is useful to detect hepatotoxicity for some drugs for which there is no alternative prescription and no test to predict toxicity (24). Accordingly, it is recommended to follow aminotransferases during the first week of treatment with anti-tuberculosis agents and during the first months of treatment for cholesterol lowering agents as well as for tacrine and riluzole (24,44).
The future Despite many advances in the knowledge of drug-induced hepatotoxicity during the last 25 yr, many important points remain unanswered in different fields: mechanisms of toxicity, risk factors, individual susceptibility, diagnostic methods and treatment. At the turn of the millennium, the first target is the prevention of risk, which can be approached on two levels: defining the characteristics of new drugs and detecting susceptible individuals.
New drugs A critical aspect for the research and development of new drugs is to establish a relation between the molecular structure of a compound and its ability to cause either direct hepatotoxic effects or toxic effects of the immunoallergic type (45). Obviously, we are far from determining such a structure-hepatotoxicity relationship because of the complexity of the phenomena. The problem could be analyzed at different levels (Table 9).
TABLE 9 How to improve
New drugs Try to answer the following questions ~ Is the drug largely or only poorly absorbed by the digestive tract? _ How the drug is eliminated? by the liver, by other routes? _ If eliminated by the liver, by which enzyme systems: cytochrome P4.50, acetylation, conjugation? _ Is the metabolism of the drug genetically modulated: role of acetylation of type 2, CYP 2D6, CYP 2C19, glutathione S transferase, etc.? _ Has the drug a molecular structure which might predispose to hepatotoxicity (furane or nitrofurane ring, tricyclic ring, tertiary amine function, etc.)? _ Does the drug belong to a family with well-known hepatotoxicity? Hepatotoxicity screening In addition to traditional studies based on animal models and hepatocyte cultures: Precision-cut liver slices _ Detection of the formation of reactive metabolites (covalent binding to hepatocyte proteins) in standard conditions and after increasing their production by enzyme induction and/or decreasing their detoxication ability (glutathione depletion, epoxide hydrolase inhibition, etc.) Susceptible patients Acquired factors ~ consumption of many drugs _ risk of drug ~ drug interaction _ age above 50 yr ~ female sex for some drug families _ malnutrition ~ alcohol abuse Genetic factors ~ phenotyping/genotyping if the drug metabolism polymorphism (CUP 2D6, CYP 2C19)
To what extent is the drug absorbed by the digestive tract and how is it eliminated?
Drugs which act without requiring abdominal absorption are less likely to cause hepatotoxicity, with a few rare exceptions. In contrast, drugs massively absorbed in the digestive tract, reaching the liver in large amounts, represent a higher risk. The type of elimination should be taken into consideration (as well as a possible metabolic polymorphism). For instance, it is obvious that drugs metabolized by cytochrome P-450 are the most likely to produce reactive metabolites (24). This production can be enhanced by a constitutional deficiency in another metabolic pathway (for instance, the deficiency in NAT2 promotes the formation of reactive metabolites by P-450 related oxidation for sulfanomides) (22). In contrast, drugs principally eliminated by the renal route or metabolized in the liver by pathways leading to non toxic metabolites (sulfoconjugation, glucuronoconjugation, hydrolysis, etc. do not a priori appear to expose patients to hepatotoxicity.
D. Larrey Does the drug exhibit a molecular structure predisposing to the formation of reactive metabolites?
Some critical structures usually oxidized by cytochrome P-450 have been shown to have a particular tendency to produce unstable reactive metabolites. Some examples are the thiophene structure involved in tienilic acid hepatitis (45) and the furane moiety, which is present in many hepatotoxic drugs, including nitrofurantoin and ketoconazole (24), or even in some components of herbal medicines as illustrated by germander (46). Another example is a tricyclic ring. There is now evidence that this particular structure can easily lead to the formation of reactive epoxides (24). Drugs containing this structure are mainly tricyclic antidepressants and phenothiazines, but also anticonvulsants such as carbamazepine (24). Interestingly, it has been recently demonstrated that some cross hepatotoxicity can exist not only between tricyclic antidepressants or between phenothiazine derivatives but also between tricyclic antidepressants and phenothiazines and between phenothiazines and carbamazepine in some particular patients (47). These clinical observations reinforce the view that the tricyclic molecular structure makes up a promoting factor of toxicity. Does the new drug belong to a family documented hepatotoxicity?
This situation is illustrated by tricyclic antidepressants. Almost all of them have been shown to cause immunoallergic acute hepatitis: amineptine, amitriptyline, imipramine, desipramine, clomipramine, etc. (24). Consequently, pharmaceutical manufacturers of tricyclic antidepressants should be particularly aware of possible hepatotoxicity with new drugs of this class. Prospective investigations to predict potential hepatotoxicity have already been shown to be efficient, e.g. with tianeptine. Indeed, this tricyclic antidepressant is structurally very similar to amineptine, another antidepressant responsible for many cases of acute hepatitis (24). The mechanisms of amineptine hepatotoxicity have been extensively examined and involve the formation of reactive metabolites through cytochrome P-450, particularly those of family 3A for acute immunoallergic cholestatic hepatitis, and also the inhibition of mitochondrial fatty acid beta-oxidation for microvesicular steatosis (24). On this basis, similar studies were performed for tianeptine. It has been shown, both in animal models and in human hepatocyte cultures, that tianeptine could also be transformed into reactive metabolites by cytochrome P-450 3A and was an inhibitor of fatty acid mitochondrial beta-oxidation (48,49). Furthermore, microvesicular steatosis could be produced in animals (49). From these data, it was pre86
dicted that tianeptine might produce acute hepatitis with immunoallergic features associated with microvesicular steatosis in rare patients (48,49). This prediction was verified later by the occurrence of such lesions in a patient receiving the treatment (50). How can methods to determine drugs be improved?
The precise enzymatic systems involved in the elimination of new drugs should be determined, particularly when cytochrome P-450 is concerned. Hepatocyte cultures combined with specific inhibitors or inducers of individual cytochrome P-450 (probe-drugs, anti P-450 antibodies) are very useful for this (51). Further help could be offered by using heterologous expression systems for drug metabolizing enzymes, in particular humanized yeast strains to allow human P-450 expression in a tailored redox environment (52).
Hepatotoxicity screening Animal models are of interest, but they also have their limitations - in particular because of interspecies differences (53). In vitro hepatotoxicity models are currently used, in particular cultured hepatocytes, also with some limitations in their predictive value (51,53). Recent reports suggest that precision-cut liver slices may be an interesting alternative to hepatocyte culture for in vitro hepatotoxicity screening (53). A major advantage over cultured hepatocytes is that with this method, preparations can easily be made from various species, including humans (53). The detection of reactive metabolite formation by the new drug should be more systematically performed for both potential toxic and immunoallergic hepatitis. The method consists in demonstrating a covalent binding of drug metabolites to liver proteins (53). This method has been useful for predicting tianeptine hepatotoxicity (49-5 1). The sensitivity for detecting reactive metabolite formation can be increased in several ways: by increasing the production of these metabolites by nonspecific or specific inducers; by decreasing the ability to detoxify the reactive metabolites, for instance by depleting glutathione or by inhibiting epoxide hydrolases. This screening can be done using subcellular fractions (microsome preparations), cultured hepatocytes or in vivo animal models.
Detection of susceptible patients Several acquired and genetic factors can promote hepatotoxicity (as shown in Tables 2 and 3). Therefore, one should avoid creating critical situations or, if that is not possible, one should reinforce the follow-up of patients.
Drug-induced liver diseases
Genetic factors have been involved in hepatotoxicity in a relatively small number of cases, but these are consistently increasing. The role of some metabolic pathways remains to be elucidated. For instance, the polymorphism of sulfoxidation has still to be ascertained at the molecular level by reproducible tests. The real contribution of genetic variations in glutathione S transferase isoenzymes in hepatotoxicity needs further assessment. The role of some still unknown enzyme defects in detoxification mechanisms is supported by the in vitro drug cytotoxicity test developed by Spielberg et al. (30) for several drugs. It is now necessary to characterize these defects at the molecular level. The development and the wide availability of genotyping tests for drug metabolizing enzyme polymorphisms should also help to better understand the influence of genetics on drug hepatotoxicity and to prevent side effects.
References 1. Tanguy G,
Bernard V, Hyrailles V, Gosp AM, Veyrac M, Larrey D. Iatrogenic disorders inducing hospital admission. A prospective study in a hospital gastrointestinal department. 7th United European Gastroenterology Week, 13-7 Nov 1999, Rome. 2. Stricker BHCH. Drug-induced Hepatic Injury. 2nd edn. Amsterdam: Elsevier; 1992. 3. Farrell GC. Drug-induced liver disease. London: Churchill Livingstone; 1994. 4. Pessayre D, Larrey D, Biour M. Drug-induced liver injury. In: Bircher J, Benhamou JP, McIntyre N, Rizzetto M, Rod& J, editors. Oxford Textbook of Clinical Hepatology, 2nd edn, Vol. 2. Oxford: Oxford University Press; 1999. p. 1261-315. 5. Biour M, Poupon R, Grange J-D, Chazouilleres 0, Jaillon P Htpatotoxicite des mtdicaments. Gastroenterol Clin Biol; 22: 100444. 6. Zimmerman HJ. Hepatotoxicity. New York: Appleton-CenturyCrofts; 1978. 7. Larrey D. Hepatotoxicity of herbal medicine. J Hepatol 1997; 26 (suool 1): 47-51. \ 8. Negro F, Mondardini A, Palmas E Hepatotoxicity of saccharin. N Engl J Med 1994; 331: 1345. 9. Larrey D. Hepatites medicamenteuses: aspects tpidemiologiques, cliniques, diagnostiques, et physiopathologiques en 1995. Rev Med Int 1995; 16: 752-8. 10. Jean-Pastor MJ, Jouglard J. Bilan des accidents htpatiques medicamenteux recueillis par la pharmacovigilance fran$aise. Therapie 1984; 39: 493-500. M, Andreasen BP Drug-induced liver disease in 11. Dossing Denmark. An analysis of 572 cases of hepatotoxicity reported to the Danish board of adverse reaction to drugs. Stand J Gastroenter01 1982; 17: 205511. liver fail12. Bernuau J, Benhamou JP. Fulminant and subfulminant ure. In: Bircher J, Benhamou JP, McIntyre N, Rizzetto M, Rod& J, editors. Oxford Textbook of Clinical Hepatology, 2nd edn, Vol. 2. Oxford: Oxford University Press; 1999. p. 1341-72. hepatitis in Japan: etiology 13. Takahashi Y, Okuda K. Fulminant and prognostic prediction, Chin J Gastroenterol 1993; 10: 33750. of drug-induced hepatitis. In: 14. Pessayre D, Larrey D. Mechanisms Guillouzo A, editor. Liver Cells and Drugs. Vol. 164. Colloque INSERM/John Libbey Eurotext Ltd.; 1988. p. 12942. 15. Dansette PM, Bonierbale E, Minoletti C, Beaune PH, Pessayre D, Mansuy D. Drug-induced immunotoxicity. Eur J Drug Metab Pharmacokinet 1998; 23: 443-51. II
S, Biour M, Poupon R, 16. Homberg JC, Abuaf N, Helmy-Khalil Islam S, et al. Drug-induced hepatitis associated with antiintracytoplasmic organelle autoantibodies. Hepatology 1887; 7: 133339. 17. Bourdi M, Larrey D, Nataf J, Bernuau J, Pessayre D, Iwasaki M, et al. Anti-liver endoplasmic reticulum autoantibodies are directed against human cytochrome P-450 IA2: a specific marker of dihydralazine-induced hepatitis. J Clin Invest 1990; 85: 196773. 18. Beaune PH, Dansette PM, Mansuy D, Kiffel L, Finck M, Amar C, et al. Human anti-endoplasmic reticulum auto-antibodies appearing in a drug-induced hepatitis are directed against a human liver cytochrome P-450 that hydroxylates the drug. Proc Nat1 Acad Sci USA 1987; 84: 551-5. 19. Bourdi M, Chen WQ, Peter RM, Martin JL, Buters JTM, Nelson SD, et al. Human cytochrome P-450 2El is a major autoantigen associated with halothane hepatitis. Chem Res Toxic01 1996; 9: 1159966. P-450 2El is a cell surface 20. Eliasson E, Kenna JG. Cytochrome autoantigen in halothane hepatitis. Mol Pharmacol 1996; 50: 573-82. B, Pessayre D. Impaired mitochondrial function in 21. Fromenty microvesicular steatosis. Effects of drugs, ethanol, hormones and cytokines. J Hepatol 1997; 26: 43-53. to drug-induced 22. Larrey D, Pageaux GP Genetic predisposition hepatotoxicity. J Hepatol 1997; 26: 12-21. 23. Larrey D, Letteron P, Foliot A, Descatoire V, Degott C, Gentve J, et al. Effects of pregnancy on the toxicity and metabolism of acetaminophen in mice. J Pharmacol Exp Ther 1986; 237: 28391. 24. Morgan MY, Reshef R, Shah RR, Oates NS, Smith RL, Sherlock S. Impaired oxidation of debrisoquine in patients with perhexiline liver injury. Gut 1984; 25: 1057764. 25. Larrey D, Amouyal G, Tine1 M, Letttron P, Berson A, Labbe G, et al. Polymorphism of dextromethorphan oxidation in a French population. Br J Clin Pharmacol 1987; 24: 6769. phenotyping and geno26. Gonzales FJ, Idle JR. Pharmacogenetic typing. Present status and future potential. Clin Pharmacokinet 1994; 26: 59-70. 27. Larrey D, Tine1 M, Amouyal G, Freneaux E, Berson A, FouinFortunet H, et al. Genetically determined oxidation polymorphism and drug hepatotoxicity. Study of 51 patients. J Hepatol 1989; 8: 158864. Y, Lannes D, Pessayre D, Larrey D. Possible associ28. Horsmans ation between poor metabolism of mephenytoin and hepatotoxicity caused by Atrium@. A fixed combination preparation containing phenobarbital, febarbamate and difebarbamate. J Hepatol 1994; 21: 107559. 29. De Morais SMF, Wilkinson GR, Blaisdell J, Meyer UA, Nakamura K, Goldstein JA. The major genetic defect responsible for the polymorphism of S-mephenytoin metabolism in humans. J Biol Chem 1994; 269: 15419-52. susceptibil30 Spielberg SI? In vitro assessment of pharmacogenetic ity to toxic drug metabolites in humans. Fed Proc 1984; 43: 230813. F, Tine1 M, Cstot A, Babany 31 Larrey D, Berson A, Hebersetzer G, et al. Genetic predisposition to drug hepatotoxicity. Role in hepatitis causes by amineptine, a tricyclic antidepressant. Hepatology 1989; 10: 168-73. Delomenie C, Mathelier-Fusade P, Longuemaux S, Rozenbaum W, Leynadier F, Krishnamoorthy R, et al. Glutathione S-transferase (GSTMl) null genotype and sulphonamide intolerance in acquired immunodeficiency syndrome. Pharmacogenetics 1997; 7: 519-20. 33. De Sousa M, Pirmohamed M, Kitteringham NR, Woolf T, Park BK. No association between tacrine transaminitis and the glutathione transferase 0 genotype in patients with Alzheimer’s disease. Pharmacogenetics 1998; 8: 35335. 34. Berson A, Freneaux E, Larrey D, Lepage V, Douay C, Mallet C, et al. Possible role of HLA in hepatotoxicity. An exploratory study in 71 patients with drug-induced idiosyncratic hepatitis. J Hepatol 1994; 20: 33642. _L.
D. Larrey 35. Criteria of drug-induced liver disorders. Report of an International Consensus Meeting. J Hepatol 1990; 11: 27226. 36. Zimmerman HJ, Lewis JH. Drug-induced cholestasis. Med Toxic01 1987; 2: 112. 37. Larrey D, Erlinger S. Drug-induced cholestasis. Bailliere’s Clin Gastroenterol 1988; 2: 423-52. 38. Larrey D, Pessayre D, Duhamel G. Prolonged cholestatic jaundice after ajmaline-induced acute hepatitis. J Hepatol 1986; 2: 81-7. 39. Degott C, Feldmann G, Larrey D, Durand-Schneider AM, Grange D, Machyekhi JP, et al. Drug-induced prolonged cholestasis in adults: a histological semiquantitative study demonstrating progressive ductopenia. Hepatology 1992; 17: 24451. 40. Larrey D, Michel H. Pathologie biliaire due aux mtdicaments. Gastroenterol Clin Biol 1993; 17: 599865. 41. O’Grady JG. Paracetamol-induced acute liver failure: prevention and management. J Hepatol 1997; 26: 41-6. 42. Ioannides C. Cytochromes P450. Metabolic and toxicological aspects. Boca Raton: CRC Press; 1996. 43. Maurel P Drug-metabolizing enzymes. Curr Opin Crit Care 1998; 4: 213-8. 44. Remy AJ, Camu W, Ramos J, Blanc P, Larrey D. Acute hepatitis after riluzole administration, J Hepatol 1999; 30: 527-30. 45. Mansuy D. Molecular structure and hepatotoxicity: compared data about two closely related thiophene compounds. J Hepatol 1997; 26: 22-5. 46. Fau D, Mounia L, Farrell G, Moreau A, Moulis C, Feldmann G,
et al. Diterpenoids from germander, an herbal medicine, induce apoptosis in isolated rat hepatocytes. Gastroenterology 1997; 113: 133446. Remy AJ, Larrey D, Pageaux GP, Ribstein J, Ramos J, Michel H. Cross hepatotoxicity between tricyclic antidepressants and phenothiazines. Eur J Gastroenterol Hepatol 1995; 7: 373-6. Larrey D, Tine1 M, Letttron P, Maurel E Loeper J, Belghiti J, et al. Metabolic activation of the new tricyclic antidepressant tianeptine by human liver cytochrome P-450. Biochem Pharmacol 1990; 14: 5046. Fromenty B, Freneaux E, Labbe G, Deschamps D, Larrey D, Letteron P et al. Tianeptine, a new tricyclic antidepressant metabolized by P-oxidation of its heptanoic side chain, inhibits the mitochondrial oxidation of medium and short chain fatty acids in mice. Biochem Pharmacol 1989; 38: 3743351. Le Bricquir Y, Larrey D, Blanc P, Pageaux GP, Michel H. Tianeptine an instance of drug-induced hepatotoxicity predicted by prospective experimental studies. J Hepatol 1994; 21: 771-3. Guillouzo A, Morel E Langouet S, Maheo K, Rissel M. Use of hepatocyte cultures for the study of hepatotoxic compounds. J Hepatol 1997; 26: 73-80. Pompon D, Gautier J-C, Perret A, Truan G, Urban I? Simulation of human xenobiotic metabolism in microorganisms: yeast a good compromise between E. coli and human cells, J Hepatol 1997; 26: 81-5. Ballet E Hepatotoxicity in drug development: detection, significance and solutions. J Hepatol 1997; 26: 2636.