Metabolic activation of the tricyclic antidepressant amineptine by human liver cytochrome P-450

Structural alert/reactive metabolite concept as applied in medicinal chemistry to mitigate the risk of idiosyncratic drug toxicity: a perspective based on the critical examination of trends in the top 200 drugs marketed in the United States

Because of a preconceived notion that eliminating reactive metabolite (RM) formation with new drug candidates could mitigate the risk of idiosyncratic drug toxicity, the potential for RM formation is routinely examined as part of lead optimization efforts in drug discovery. Likewise, avoidance of "structural alerts" is almost a norm in drug design. However, there is a growing concern that the perceived safety hazards associated with structural alerts and/or RM screening tools as standalone predictors of toxicity risks may be over exaggerated. In addition, the multifactorial nature of idiosyncratic toxicity is now well recognized based upon observations that mechanisms other than RM formation (e.g., mitochondrial toxicity and inhibition of bile salt export pump (BSEP)) also can account for certain target organ toxicities. Hence, fundamental questions arise such as: When is a molecule that contains a structural alert (RM positive or negative) a cause for concern? Could the molecule in its parent form exert toxicity? Can a low dose drug candidate truly mitigate metabolism-dependent and -independent idiosyncratic toxicity risks? In an effort to address these questions, we have retrospectively examined 68 drugs (recalled or associated with a black box warning due to idiosyncratic toxicity) and the top 200 drugs (prescription and sales) in the United States in 2009 for trends in physiochemical characteristics, daily doses, presence of structural alerts, evidence for RM formation as well as toxicity mechanism(s) potentially mediated by parent drugs. Collectively, our analysis revealed that a significant proportion (∼78-86%) of drugs associated with toxicity contained structural alerts and evidence indicating that RM formation as a causative factor for toxicity has been presented in 62-69% of these molecules. In several cases, mitochondrial toxicity and BSEP inhibition mediated by parent drugs were also noted as potential causative factors. Most drugs were administered at daily doses exceeding several hundred milligrams. There was no obvious link between idiosyncratic toxicity and physicochemical properties such as molecular weight, lipophilicity, etc. Approximately half of the top 200 drugs for 2009 (prescription and sales) also contained one or more alerts in their chemical architecture, and many were found to be RM-positive. Several instances of BSEP and mitochondrial liabilities were also noted with agents in the top 200 category. However, with relatively few exceptions, the vast majority of these drugs are rarely associated with idiosyncratic toxicity, despite years of patient use. The major differentiating factor appeared to be the daily dose; most of the drugs in the top 200 list are administered at low daily doses. In addition, competing detoxication pathways and/or alternate nonmetabolic clearance routes provided suitable justifications for the safety records of RM-positive drugs in the top 200 category. Thus, while RM elimination may be a useful and pragmatic starting point in mitigating idiosyncratic toxicity risks, our analysis suggests a need for a more integrated screening paradigm for chemical hazard identification in drug discovery. Thus, in addition to a detailed assessment of RM formation potential (in relationship to the overall elimination mechanisms of the compound(s)) for lead compounds, effects on cellular health (e.g., cytotoxicity assays), BSEP inhibition, and mitochondrial toxicity are the recommended suite of assays to characterize compound liabilities. However, the prospective use of such data in compound selection will require further validation of the cellular assays using marketed agents. Until we gain a better understanding of the pathophysiological mechanisms associated with idiosyncratic toxicities, improving pharmacokinetics and intrinsic potency as means of decreasing the dose size and the associated "body burden" of the parent drug and its metabolites will remain an overarching goal in drug discovery.

Mitochondrial involvement in drug-induced liver injury

Mitochondrial dysfunction is a major mechanism of liver injury. A parent drug or its reactive metabolite can trigger outer mitochondrial membrane permeabilization or rupture due to mitochondrial permeability transition. The latter can severely deplete ATP and cause liver cell necrosis, or it can instead lead to apoptosis by releasing cytochrome c, which activates caspases in the cytosol. Necrosis and apoptosis can trigger cytolytic hepatitis resulting in lethal fulminant hepatitis in some patients. Other drugs severely inhibit mitochondrial function and trigger extensive microvesicular steatosis, hypoglycaemia, coma, and death. Milder and more prolonged forms of drug-induced mitochondrial dysfunction can also cause macrovacuolar steatosis. Although this is a benign liver lesion in the short-term, it can progress to steatohepatitis and then to cirrhosis. Patient susceptibility to drug-induced mitochondrial dysfunction and liver injury can sometimes be explained by genetic or acquired variations in drug metabolism and/or elimination that increase the concentration of the toxic species (parent drug or metabolite). Susceptibility may also be increased by the presence of another condition, which also impairs mitochondrial function, such as an inborn mitochondrial cytopathy, beta-oxidation defect, certain viral infections, pregnancy, or the obesity-associated metabolic syndrome. Liver injury due to mitochondrial dysfunction can have important consequences for pharmaceutical companies. It has led to the interruption of clinical trials, the recall of several drugs after marketing, or the introduction of severe black box warnings by drug agencies. Pharmaceutical companies should systematically investigate mitochondrial effects during lead selection or preclinical safety studies.

Role of Metabolism in Drug-Induced Idiosyncratic Hepatotoxicity

Rare adverse reactions to drugs that are of unknown etiology, or idiosyncratic reactions, can produce severe medical complications or even death in patients. Current hypotheses suggest that metabolic activation of a drug to a reactive intermediate is a necessary, yet insufficient, step in the generation of an idiosyncratic reaction. We review evidence for this hypothesis with drugs that are associated with hepatotoxicity, one of the most common types of idiosyncratic reactions in humans. We identified 21 drugs that have either been withdrawn from the U.S. market due to hepatotoxicity or have a black box warning for hepatotoxicity. Evidence for the formation of reactive metabolites was found for 5 out of 6 drugs that were withdrawn, and 8 out of 15 drugs that have black box warnings. For the other drugs, either evidence was not available or suitable studies have not been carried out. We also review evidence for reactive intermediate formation from a number of additional drugs that have been associated with idiosyncratic hepatotoxicity but do not have black box warnings. Finally, we consider the potential role that high dosages may play in these adverse reactions.

Genetic predisposition to drug-induced hepatotoxicity

Drug-induced hepatitis is uncommon and generally unpredictable. Hepatotoxicity may be related to the drug itself, or to chemically reactive metabolites which can bind covalently to hepatic macromolecules and may lead to either idiosyncratic, toxic hepatitis or to immunoallergic hepatitis. There is now evidence indicating that genetic variations in systems of biotransformation or detoxication may modulate either the toxic or sensitizing effects of some drugs. Thus, the genetic deficiency in a particular hepatic cytochrome P 450 isozyme (CYP 2D6) is involved in per-hexiline liver injury. The deficiency in CYP 2C19 might also contribute to Atrium hepatotoxicity. Slow acetylation related to N-acetyltransferase 2 deficiency contributes to sulfonamide hepatitis. The genetic deficiency in glutathione synthetase may increase the susceptibility to several drugs including acetaminophen. A constitutional deficiency in another cell defense mechanism, still not characterized, seems to increase significantly the risk of hepatotoxicity with halothane, phenytoin, carbamazepine, phenobarbital, sulfamides and amineptine.

Inhibition of mitochondrial beta-oxidation as a mechanism of hepatotoxicity

Severe and prolonged impairment of mitochondrial beta-oxidation leads to microvesicular steatosis, and, in severe forms, to liver failure, coma and death. Impairment of mitochondrial beta-oxidation may be either genetic or acquired, and different causes may add their effects to inhibit beta-oxidation severely and trigger the syndrome. Drugs and some endogenous compounds can sequester coenzyme A and/or inhibit mitochondrial beta-oxidation enzymes (aspirin, valproic acid, tetracyclines, several 2-arylpropionate anti-inflammatory drugs, amineptine and tianeptine); they may inhibit both mitochondrial beta-oxidation and oxidative phosphorylation (endogenous bile acids, amiodarone, perhexiline and diethylaminoethoxyhexestrol), or they may impair mitochondrial DNA transcription (interferon-alpha), or decrease mitochondrial DNA replication (dideoxynucleoside analogues), while other compounds (ethanol, female sex hormones) act through a combination of different mechanisms. Any investigational molecule should be screened for such effects.

Tianeptine--an instance of drug-induced hepatotoxicity predicted by prospective experimental studies

We report the case of a patient who developed acute hepatitis after taking tianeptine, a new tricyclic antidepressant, for 8 weeks. Hepatitis exhibited cholangitis-like clinical features and was associated with hypersensitivity manifestations suggestive of an immuno-allergic mechanism. Histological examination showed microvesicular steatosis. The discontinuation of tianeptine administration was followed by complete recovery. Immunoallergic hepatitis and microvesicular steatosis were predicted 2 years ago from prospective experimental studies prompted by the similarity of the chemical structures of tianeptine and amineptine, another tricyclic antidepressant, well-known for its hepatotoxicity. Experimentally, tianeptine has been found to be oxidized into reactive metabolites in several rodents and human liver and to produce microvesicular steatosis probably through inhibition of mitochondrial beta-oxidation of fatty acid in mice. This case illustrates the value of prospectively assessing potential hepatotoxicity mechanisms for new compounds chemically related to drugs already known to be hepatotoxic.

Idiosyncratic reactions to antidepressants: a review of the possible mechanisms and predisposing factors

Antidepressants, a widely used group of drugs, are associated with a range of idiosyncratic reactions affecting in particular the liver, skin and both the hematological and central nervous systems. These reactions seem to be mediated by chemically reactive metabolites formed by the cytochrome P450 enzyme system, the toxicity occurring either directly or indirectly via an immune mechanism. Individual susceptibility is determined by factors, both genetic and environmental, which result in inadequate detoxication of the chemically reactive metabolite. Prevention of such reactions will depend on either the development of new compounds which are not converted to toxic metabolites or by prediction of individual susceptibility prior to drug administration.

Metabolic activation of the new tricyclic antidepressant tianeptine by human liver cytochrome P450

Incubation of [14C]tianeptine (0.5 mM) with human liver microsomes and a NADPH-generating system resulted in the in vitro covalent binding of a tianeptine metabolite to microsomal proteins. This covalent binding required oxygen and NADPH. It was decreased by piperonyl butoxide (4 mM) by 81%, and SKF 525-A (4 mM) by 87%, two relatively non-specific inhibitors of cytochrome P450, and by glutathione (4 mM) by 70%, a nucleophile. Covalent binding was decreased by 54% in the presence of troleandomycin (0.1 mM), a specific inhibitor of the glucocorticoid-inducible cytochrome P450 IIIA3, but remained unchanged in the presence of quinidine (0.1 mM) or dextromethorphan (0.1 mM), two inhibitors of cytochrome P450 IID6. Preincubation with IgG antibodies directed against cytochrome P450 IIIA3 decreased covalent binding by 65% whereas either preimmune IgG or IgG antibodies directed against P450 IA1, an isoenzyme inducible by polycyclic aromatic compounds, exhibited no significant inhibitory effect. We conclude that tianeptine is activated by human liver cytochrome P450 into a reactive metabolite. This activation is mediated in part by glucocorticoid-inducible isoenzymes but not by P450 IID6 (the isoenzyme which oxidizes debrisoquine) nor by P450 IA1 (an isoenzyme inducible by polycyclic aromatic compounds). The predictive value of this study regarding possible idiosyncratic and immunoallergic reactions in humans remains unknown.

Metabolic activation of the antidepressant tianeptine. I. Cytochrome P-450-mediated in vitro covalent binding

Incubation under air of [14C]tianeptine (0.5 mM) with a NADPH-generating system and hamster, mouse or rat liver microsomes resulted in the in vitro covalent binding of [14C]tianeptine metabolites to microsomal proteins. Covalent binding to hamster liver microsomes required NADPH and oxygen; it was decreased in the presence of the cytochrome P-450 inhibitors, carbon monoxide, piperonyl butoxide (4 mM), and SKF 525-A (4 mM) or in the presence of the nucleophile, glutathione (1 or 4 mM). In vitro covalent binding to hamster liver microsomes was not decreased in the presence of quinidine (1 microM), and was similar with microsomes from either female Dark Agouti, or female Sprague-Dawley rats. In contrast, in vitro covalent binding to hamster liver microsomes was decreased in the presence of troleandomycin (0.25 mM), while covalent binding was increased with microsomes from either hamsters, mice or rats pretreated with dexamethasone. Preincubation with IgG antibodies directed against rabbit liver glucocorticoid-inducible cytochrome P-450 3c(P-450 IIIA4) decreased in vitro covalent binding by 53 and 89%, respectively, with microsomes from control hamsters and dexamethasone-pretreated hamsters, and by 60 and 81%, respectively, with microsomes from control and dexamethasone-pretreated rats. We conclude that tianeptine is activated by hamster, mouse and rat liver cytochrome P-450 into a reactive metabolite. Metabolic activation is mediated in part by glucocorticoid-inducible isoenzymes but not by the isoenzyme metabolizing debrisoquine. In vivo studies are reported in the accompanying paper.

Activation of mianserin and its metabolites by human liver microsomes

Human liver microsomes metabolise mianserin to the stable 8-hydroxymianserin, desmethylmianserin and mianserin-2-oxide and in addition to one or more chemically reactive metabolites which bind, irreversibly, to microsomal protein. The stable metabolites were isolated by HPLC and characterized by mass spectrometry. The generation of each of these metabolites showed substantial inter-individual variation between eight sets of human liver microsomes studied. Inhibition of irreversible binding was observed with SKF-525A together with concomitant decrease in the formation of 8-hydroxymianserin and desmethylmianserin but not mianserin-2-oxide. Methimazole inhibited binding and the formation of each of the metabolites at a low concentration. Quinidine did not significantly inhibit irreversible binding but did inhibit the formation of 8-hydroxymianserin. Sulphaphenazole had no effect on irreversible binding or metabolism. The irreversible binding of mianserin was inhibited by ascorbic acid, glutathione and N-acetyl cysteine, whereas N-acetyl lysine and trichloropropane oxide had no effect. The irreversible binding of mianserin, 8-hydroxymianserin and desmethylmianserin was of the same order of magnitude however significantly greater binding was observed with the desmethyl metabolite. Incubations with [10-3H/14C]mianserin showed no change in the 3H/14C ratio when irreversible binding occurred. Inhibition of irreversible binding was demonstrated with sodium cyanide at concentrations which did not inhibit total metabolism, which suggest that metabolic activation by the cytochrome P-450 enzyme system may lead to the formation of a reactive iminium intermediate that can bind to nucleophilic groups on proteins.

Genetic predisposition to drug hepatotoxicity: role in hepatitis caused by amineptine, a tricyclic antidepressant

Amineptine-induced immunoallergic hepatitis is unpredictable. It may be related to its oxidation into a reactive metabolite acting as hapten. We have looked for a possible genetic predisposition involving drug oxidation capacity and/or cell defense mechanisms in nine patients with previous amineptine hepatitis. Drug oxidation capacity was assessed using dextromethorphan, a test compound recently proposed as a substitute for debrisoquine. The eight patients tested had the extensive metabolizer phenotype. The susceptibility to amineptine metabolites was studied by an in vitro test assessing the destruction of the patients' lymphocytes by reactive metabolites generated from amineptine by a standardized oxidation microsomal system. Lymphocyte death increased with the dose of amineptine (1 to 2.5 mM); it was increased by preincubation with trichloropropene oxide, but was absent when amineptine was omitted or when the oxidation system was not operating. Mean lymphocyte death was twice higher in the nine patients with amineptine hepatitis than in 17 healthy controls. In contrast, when the test was performed with acetaminophen (3 to 10 mM), lymphocyte death was similar in controls and in patients. Basal epoxide hydrolase activity toward benzo[a]pyrene-4,5-oxide and glutathione concentration was similar in lymphocytes from controls and patients. Family studies showed an increased susceptibility to amineptine metabolites in lymphocytes from several first-degree relatives of two patients. These results show that amineptine hepatitis occurs in patients with extensive dextromethorphan oxidation capacity but with an increased susceptibility to amineptine reactive metabolites, probably related to a genetic deficiency in a cell defense mechanism.

Presence of covalently bound metabolites on rat hepatocyte plasma membrane proteins after administration of isaxonine, a drug leading to immunoallergic hepatitis in man

Isaxonine and several other drugs transformed by cytochrome P-450 into reactive metabolites apparently lead to immunoallergic hepatitis in man. Protein epitopes modified by the covalent binding of the metabolites have been proposed as possible targets for the immune response. The purpose of this work was to determine whether covalently bound metabolites are indeed present on hepatocyte plasma membrane proteins. In a first series of experiments, rats were killed 15 or 60 min after administration of [2-14C]isaxonine (0.2 mmol.kg-1 i.p.), and various fractions were prepared from isolated hepatocytes; microsomal contamination of the plasma membrane fraction was 1.2% or less. At 60 min, the amount of isaxonine metabolite covalently bound per mg of protein was similar in plasma membranes (0.42 nmole metabolite.mg protein-1) and in microsomes (0.38); both values were decreased by about 70% in rats pretreated with piperonyl butoxide, an inhibitor of cytochrome P-450. At 15 min, however, covalent binding to plasma membrane proteins (0.06 nmole metabolite.mg protein-1) was only half of that to microsomal proteins (0.12). In a second series of experiments, [2-14C] isaxonine (0.1 mM) was incubated with NADPH, hepatic microsomes and plasma membranes. The reactive isaxonine metabolite became bound extensively to microsomal proteins, but not to plasma membrane proteins. These results show that administration of isaxonine leads to the presence of isaxonine adducts on the proteins of the hepatocyte plasma membrane.(ABSTRACT TRUNCATED AT 250 WORDS)

A comparative study of the formation of chemically reactive drug metabolites by human liver microsomes

1. The metabolism of amodiaquine (A), ethinyloestradiol (E), mianserin (M), phenytoin (Ph), sulphanilamide (S) and paracetamol (Pa) to both stable and chemically reactive, i.e. irreversibly protein bound, metabolites was investigated using microsomes prepared from histologically normal human liver obtained from eight kidney donors. 2. All drugs, except amodiaquine, were metabolized by NADPH-dependent microsomal enzymes to chemically reactive metabolites. The degree of NADPH-dependent binding varied between drugs (E, 11.5 +/- 5.8% incubated drug; M, 3.0 +/- 1.9%; Ph, 0.10 +/- 0.09%; S, 0.57 +/- 0.38%; Pa, 1.2 +/- 1.2%; mean of eight livers +/- s.d.). 3. Inclusion of glutathione (1 mM) or ascorbic acid (1 mM) in the incubation reduced the NADPH-dependent binding for all substrates, indicating the involvement of electrophilic oxidation products. 4. Binding of M and Pa correlated with each other (Spearman's r = 0.86) and with total cytochrome P-450 content (r = 0.76 and 0.78 respectively). E binding also correlated with the binding of M (r = 0.79) and Pa (r = 0.81) but not with cytochrome P-450. Binding of Ph and S did not correlate with any of the other measured metabolic parameters.

Acute and chronic drug-induced hepatitis

Adverse drug reactions may mimic almost any kind of liver disease. Acute hepatitis is often due to the formation of reactive metabolites in the liver. Despite several protective mechanisms (epoxide hydrolases, conjugation with glutathione), this formation may lead to predictable toxic hepatitis after hugh overdoses (e.g. paracetamol), or to idiosyncratic toxic hepatitis after therapeutic doses (e.g. isoniazid). Both genetic factors (e.g. constitutive levels of cytochrome P-450 isoenzymes, or defects in protective mechanisms) and acquired factors (e.g. malnutrition, or chronic intake of alcohol or other microsomal enzyme inducers) may explain the unique susceptibility of some patients. Formation of chemically reactive metabolites may also lead to allergic hepatitis, probably through immunization against plasma membrane protein epitopes modified by the covalent binding of the reactive metabolites. This may be the mechanism for acute hepatitis produced by many drugs (e.g. amineptine, erythromycin derivatives, halothane, imipramine, isaxonine, alpha-methyldopa, tienilic acid, etc.). Genetic defects in several protective mechanisms (e.g. epoxide hydrolase, acetylation) may explain the unique susceptibility of some patients, possibly by increasing exposure to allergenic, metabolite-altered plasma membrane protein epitopes. Like toxic idiosyncratic hepatitis, allergic hepatitis occurs in a few patients only. Unlike toxic hepatitis, allergic hepatitis is frequently associated with fever, rash or other hypersensitivity manifestations; it may be hepatocellular, mixed or cholestatic; it promptly recurs after inadvertent drug rechallenge. Lysosomal phospholipidosis occurs frequently with three antianginal drugs (diethylaminoethoxyhexestrol, amiodarone and perhexiline). These cationic, amphiphilic drugs may form phospholipid-drug complexes within lysosomes. Such complexes resist phospholipases and accumulate within enlarged lysosomes, forming myeloid figures. This phospholipidosis has little clinical importance. In a few patients, however, it is associated with alcoholic-like liver lesions leading to overt liver disease and, at times, cirrhosis. Subjects with a deficiency in a particular isoenzyme of cytochrome P-450 poorly metabolize perhexiline and are at higher risk of developing liver lesions. Prolonged, drug-induced liver-cell necrosis may also lead to subacute hepatitis, chronic hepatitis or even cirrhosis. This usually occurs when the drug administration is continued, either because the liver disease remains undetected or because its drug aetiology is overlooked. Several autoantibodies may be present.(ABSTRACT TRUNCATED AT 400 WORDS)