Sidney D. Nelson

Isoniazid Liver Injury: Clinical Spectrum, Pathology, and Probable Pathogenesis

The clinical spectrum of isoniazid-induced liver injury seems to be clinically, biochemically, and histologically indistinguishable from viral hepatitis, except that the injury occurs primarily in persons older than 35 years. A possible relation between susceptibility of patients to isoniazid liver injury and rapid metabolism (acetylation) of the drug has been found. Examination of isoniazid metabolites showed that patients with rapid acetylator phenotype hydrolyze much more isoniazid to isonicotinic acid and the free hydrazine moiety than do slow acetylators. The hydrazine moiety liberated from isoniazid is primarily acetylhydrazine, and studies in animals show this metabolite to be converted to a potent acylating agent that produces liver necrosis. It seems likely that formation of chemically reactive metabolites is also the biochemical event initiating isoniazid liver injury in man. Recognition of the seriousness of isoniazid hepatic injury, not readily accepted at first, has led to revisions in the uses of isoniazid prophylaxis.

Managing the challenge of chemically reactive metabolites in drug development

The normal metabolism of drugs can generate metabolites that have intrinsic chemical reactivity towards cellular molecules, and therefore have the potential to alter biological function and initiate serious adverse drug reactions. Here, we present an assessment of the current approaches used for the evaluation of chemically reactive metabolites. We also describe how these approaches are being used within the pharmaceutical industry to assess and minimize the potential of drug candidates to cause toxicity. At early stages of drug discovery, iteration between medicinal chemistry and drug metabolism can eliminate perceived reactive metabolite-mediated chemical liabilities without compromising pharmacological activity or the need for extensive safety evaluation beyond standard practices. In the future, reactive metabolite evaluation may also be useful during clinical development for improving clinical risk assessment and risk management. Currently, there remains a huge gap in our understanding of the basic mechanisms that underlie chemical stress-mediated adverse reactions in humans. This Review summarizes our views on this complex topic, and includes insights into practices considered by the pharmaceutical industry.

Oxidation of acetaminophen to N-acetyl-p-aminobenzoquinone imine by Human CYP3A4

We have investigated: (a) the formation of N-acetyl-p-aminobenzoquinone imine (NAPQI) from acetaminophen (APAP) by reconstituted human liver CYP3A4, (b) the kinetics of NAPQI formation in microsomes prepared from four human livers varying in CYP1A2, 2E1 and 3A4 content determined by Western blot analysis, (c) the contribution of CYP3A4 to the total formation of NAPQI from 0.1 mM APAP in human liver microsomes using troleandomycin as a specific inhibitor, and (d) the relationship between the contribution of CYP3A4 to NAPQI formation and relative CYP3A4 content. The Km of CYP3A4 for APAP was found to be approximately 0.15 mM, similar to concentrations observed in humans after therapeutic doses of the drug. The kinetics of formation of NAPQI in human liver microsomes were complex; the lower Km was similar to that found for reconstituted CYP3A4. The contribution of CYP3A4 to total NAPQI formation varied from 1 to 20% among livers, and correlated with the relative CYP3A4 content, r2 = 0.88, P < 0.05. Our findings indicate that CYP3A4, the major P450 isoform in human liver and enterocytes, contributes appreciably to the formation of the cytotoxic metabolite NAPQI at therapeutically relevant concentrations of APAP and suggest that APAP may be a previously unrecognized inhibitor of this enzyme.

Pennyroyal toxicity: measurement of toxic metabolite levels in two cases and review of the literature.

An increasing segment of the U.S. population is seeking alternatives to traditional Western allopathic medicine. In 1990, Americans made an estimated 425 million visits to providers of unconventional therapies [1]. One particularly popular alternative is herbal medication. Herbal medicines are promoted as more natural and therefore safer than conventional over-the-counter and prescription medicines, but many may be more dangerous than conventional pharmaceutical agents [2]. Both over-the-counter and prescription medicines in the United States must be extensively tested and certified before the Food and Drug Administration approves them for indicated uses. In contrast, herbal preparations are not subjected to such scrutiny before being promoted and sold. Pennyroyal is one widely available herbal medicine that can be life threatening after ingestion. Pennyroyal, an herb consisting of the leaves of either Mentha pulegium or Hedeoma pulegioides, primarily contains pulegone (Figure 1) plus smaller amounts of several other monoterpenes that are encountered in mint species [3]. Pennyroyal is commonly available in health food stores. Since Roman times, herbalists have recommended the herb as an abortifacient [4]. Although no evidence supports its efficacy in this regard [5], many herbal books continue to cite the use of pennyroyal for this purpose [6, 7], despite reports of centrilobular hepatic necrosis and death in connection with its use [8-11]. Pennyroyal is also advocated as a pesticide, primarily for controlling fleas on domestic pets and in the home [12]. Hepatotoxicity and other cellular damage have been reported in a household pet treated with pennyroyal oil [13]. Figure 1. Structures of the major monoterpene in pennyroyal, (R)-(+)-pulegone, and its major proximate toxic metabolite, menthofuran. Pennyroyal poisoning continues to occur regularly. Although pennyroyal ingestion can be fatal, cases of poisoning have only been sporadically documented in the modern medical literature, and none has involved the use of recent analytic techniques to measure pennyroyal metabolite levels. We report four recent cases of pennyroyal toxicity, two of which had laboratory confirmation of pulegone or its major toxic metabolite menthofuran [14]. We also review all of the published clinical case data from earlier reports and place our four cases in the context of reported signs and symptoms of toxicity. Methods Quantification of Pulegone and Menthofuran Plasma samples were acidified and extracted with diethylether after internal standards were added. To identify and quantify pulegone and menthofuran [14, 15], we compared their gas chromatographic retention times and mass spectra with those of known standards. Gas chromatographic analysis was done on a Hewlett-Packard (Palo Alto, California) Model 5980. Chromatography was done on a 30 m 0.320 mm Wall Coated Open Tubular DB-5 fused silica capillary column (J & W Scientific, Folsom, California). Electron-impact mass spectrometry was done using a VG-7070H double-focusing instrument (Manchester, United Kingdom) that was equipped with a Hewlett-Packard Model 5980 Series II gas chromatograph and was electronically linked to a Mass Spectrometry Service data system (Manchester, United Kingdom). Identification of Protein-Bound Pennyroyal Metabolites by Western Blot Analysis Liver microsomes were prepared from a liver sample obtained from patient 1 and from a human liver sample obtained from the University of Washington School of Pharmacy Liver Bank. To provide a positive control, a portion of the latter sample was incubated with menthofuran and an NADPH (reduced nicotinamide adenine dinucleotide phosphate)-regenerating system. Microsomal proteins were subjected to electrophoresis in sodium dodecyl sulfate-polyacrylamide gels, and the protein bands were transferred to nitrocellulose solid support. Protein adducts were detected using a primary antibody obtained from rabbits immunized with chemically synthesized oxidative metabolites of menthofuran coupled to metallothionein and with a secondary antibody of horseradish peroxidase-conjugated goat antirabbit IgG (Pierce Chemical Co., Rockford, Illinois). The membranes were developed using 0.004% nickel chloride, 0.0075% hydrogen peroxide, and 0.05% 3,3-diaminobenzidine tetrahydrochloride. Poison Center Reporting We selected patients by reviewing all cases involving pennyroyal ingestion for which the San Francisco Bay Area Regional Poison Control Center was primarily consulted during a 2-year period. We included all medically treated or symptomatic cases and excluded two other cases initially reported to a collaborating regional poison control center for which our service provided a secondary consultation. All data were collected by telephone and were recorded on a standard American Association Poison Control Center Cooperative Poison Center report form at the time of the initial consultation. The poison information specialist obtained all pertinent information available. Follow-up contact by telephone was continued until the clinical outcome was determined. The San Francisco Poison Control Center receives an average of 60 000 consultations each year [16]. Literature Review We identified case reports by searching MEDLINE and Index Medicus and by reviewing the reference citations of all available publications. One reference citation was supposedly a case report of pennyroyal poisoning; however, we later reviewed the original report and discovered that the case did not involve poisoning and appears to have been cited in error [17]. Only 7 of the 18 cases are from the modern medical literature; the others were reported before 1905. The cases reported before 1905 are less well documented, but we did not exclude cases for this reason. Case Reports Case 1 A 24-year-old woman repeatedly ingested pennyroyal herbal extract (pennyroyal herb, 48% to 56% in an alcohol base) and black cohosh root (Cimicifuga racemosa) extract for 2 weeks in an attempt to induce an abortion. When this was unsuccessful, she ingested additional unknown amounts of pennyroyal herbal extract and black cohosh root extract over a short period. Soon after, abdominal cramps, chills, vomiting, and syncope developed, and the patient had difficulty walking. She was placed in a cold bath 7 hours after the acute ingestion and began to manifest rigors that her roommate interpreted as a seizure. Paramedics found the patient in cardiopulmonary arrest at a time estimated to be 7.5 hours after the acute ingestion. The patient was intubated, and then cardiopulmonary resuscitation was initiated and continued for 22 minutes until the patient arrived at the hospital. On arrival at the emergency department, the patients heart rate was 120 beats/min and her blood pressure was 70/40 mm Hg while she received maximal dopamine. Her pupils were fixed and dilated. The physical examination showed coma and a rigid abdomen. A computed tomographic scan of the abdomen suggested a possible ruptured ectopic pregnancy. Initial laboratory values were the following: sodium level, 157 mmol/L; potassium level, 5.7 mmol/L; chlorine level, 107 mmol/L; bicarbonate level, 8 mmol/L; blood urea nitrogen level, 5.0 mmol/L; creatinine level, 221 mol/L; glucose level, 6.8 mmol/L; lactic acid level, 21.0 mmol/L; leukocyte count, 37.3 109/L; hemoglobin level, 89 g/L; hematocrit, 0.28; platelet count, 256 109/L; albumin level, 27 g/L; total bilirubin level, 5.1 mol/L; aspartate aminotransferase level, 0.82 kat/L; lactate dehydrogenase level, 3.54 kat/L; amylase level, 2.35 kat/L; prothrombin time, 23 seconds; partial prothrombin time, 68 seconds; and international normalized ratio, 4.1. A quantitative plasma human chorionic gonadotropin level indicated that the patient was 1 to 3 months pregnant. Arterial blood gas values (measured while the patient received 100% O2 by endotracheal tube) were the following: pH, 6.61; Pco 2, 23 mm Hg; and Po 2, 503 mm Hg. A toxicology screen was negative for alcohol, acetaminophen, and salicylates. Laboratory values 36 hours after the acute ingestion were notable for the aspartate aminotransferase level (44.53 kat/L), the alanine aminotransferase level (29.12 kat/L), and the lactate dehydrogenase level (68.18 kat/L). Throughout the initial 12 hours of hospitalization, the patients course was marked by hemodynamic shock, decreasing hematocrit, and a clinical picture consistent with disseminated intravascular coagulation. During hospitalization, the patient received 10 units of packed red blood cells and multiple units of fresh frozen plasma. Exploratory laparotomy showed a hemorrhagic, right-sided ectopic pregnancy with indications of superinfection. A substantial amount of old blood, not otherwise quantified, was found. The pregnancy was not ruptured, but there was evidence of bleeding from the end of the tube. No active bleeding was seen during surgery. A computed tomographic scan of the head was consistent with anoxic encephalopathy. The patient remained unresponsive to all stimuli. Life support was withdrawn, and the patient died 46 hours after the acute pennyroyal ingestion. Other than the anticipated brain and uterine findings, the most notable abnormalities at autopsy were found in the liver. The substantial centrilobular degeneration and necrosis of the hepatic cells were consistent with a specific toxic insult. Diffuse degenerative changes involving the proximal tubules of the kidney were also noted. The pathologist concluded that the cause of death was multiorgan failure and anoxic encephalopathy secondary to ingestion of pennyroyal and black cohosh. Blood from the heart, collected at autopsy 26 hours after death, was tested for the presence of pulegone and menthofuran. The pennyroyal and black cohosh herbal extracts that the patient had ingested were also tested for the presence of pulegone and menthofuran (Table 1). Table 1. Laboratory Values of Pennyroyal Metabolites in Case 1* Po

Mechanisms of the Formation and Disposition of Reactive Metabolites That can Cause Acute Liver Injury

Acetaminophen and pulegone are just two examples for many agents that can form reactive metabolites that can cause acute liver injury. Two other classic organic compounds that have been extensively studied are carbon tetrachloride (for a recent review see Ref. 159, and for other discussions see Refs. 8 and 9) and bromobenzene (for review see Ref. 160). Different kinds of protein adducts of reactive metabolites of bromobenzene have been partially characterized [161], and specific antibodies to these adducts are now being used to isolate and identify the proteins that are modified (162). In contrast, carbon tetrachloride and other agents, such as the herbicide diquat, may form radicals that bind to and/or oxidize lipids and proteins in causing liver injury (163, 164). Therefore, the recent development [165] of antibodies to detect oxidative damage to proteins will be important in the identification and characterization of macromolecules that do not form adducts with reactive metabolites but are damaged oxidatively. Thus, some major challenges in the coming years are to identify hepatocellular macromolecules that are modified by reactive metabolites, and then approach the more difficult task of integrating this information into a time course and sequence of events leading to lethal hepatocellular injury.

Molecular Basis for Several Drug-Induced Nephropathies

A recent clinical advance has been the discovery that many drug-induced hepatic diseases result from the metabolic activation of chemically stable drugs to potent alkylating agents by the liver. In addition to the liver, however, the kidney also contains active enzyme systems capable of metabolically activating drugs and other chemicals. For this reason a systematic investigation of the possible role of metabolic activation in the pathogenesis of several drug-induced renal diseases has been undertaken. These laboratory results are reviewed in the light of the clinical spectrum of the renal injuries, and possible therapeutic implications of these new findings are briefly discussed. The potential use of these models of nephrotoxicity to probe a variety of physiologic and pathophysiologic mechanisms of renal function are noted.

The spontaneous and enzymatic reaction of N-acetyl-p-benzoquinonimine with glutathione: A stopped-flow kinetic study

The spontaneous and glutathione (GSH) transferase catalyzed reactions of GSH and N-acetyl-p-benzoquinonimine (NABQI) have been studied by stopped-flow kinetics. The spontaneous reaction was shown to be first order in NABQI, GSH and inversely proportional to the H+ concentration; e.g., at pH 7.0 and 25 degrees C the second-order rate constant was 3.2 X 10(4) M-1 s-1. Data for the enzymatic reaction gave values for Km of 27, 1.3, 7, and 7 microM and values for kappa cat of 90, 37, 5.1, and 165 s-1 for rat liver GSH transferases 1-1, 2-2, 3-3, and 7-7, respectively. Over a wide range of reactant concentrations and pH, the spontaneous reaction yields three products, namely a GSH conjugate, 3-(glutathion-S-yl)acetaminophen; a reduction product, acetaminophen; and an oxidation product, glutathione disulfide in the proportions 2:1:1. Analysis of products formed after enzymatic reaction showed that both GSH conjugation and the reduction of NABQI to acetaminophen were catalyzed to an extent characteristic of each isoenzyme. With respect to GSH conjugation, GSH transferase isoenzymes were effective in the order 7-7 greater than 2-2 greater than 1-1 greater than 3-3 greater than 4-4, and with respect to NABQI reduction these isoenzymes were effective in the order 1-1 greater than 2-2 greater than 7-7 the position of isoenzymes 3-3 and 4-4 being uncertain. Human GSH transferases delta, mu, and pi behave similarly to the homologous rat enzymes, i.e., toward conjugation in the order pi greater than delta greater than mu and the reduction delta greater than mu greater than pi (for nomenclature see W. B. Jakoby, B. Ketterer, and B. Mannervik, (1984) Biochem. Pharmacol. 33, 2539-2540). Possible mechanisms of the reaction and its effect on the toxicity of NABQI are discussed.

The covalent binding of acetaminophen to protein. Evidence for cysteine residues as major sites of arylation in vitro

Covalent binding of the reactive metabolite of acetaminophen has been investigated in hepatic microsomal preparations from phenobarbital-pretreated mice. Low molecular weight thiols (cysteine and glutathione) were found to inhibit this binding, whereas several other amino acids which were tested did not. Bovine serum albumin (BSA), which contains a single free sulfhydryl group per molecule and which thus represents a macromolecular thiol compound, inhibited covalent binding of the reactive acetaminophen metabolite to microsomal protein in a concentration-dependent manner. The acetaminophen metabolite also became irreversibly bound to BSA in these experiments, although this binding was reduced by approx. 47% when the thiol function of BSA was selectively blocked prior to incubation. Covalent binding of the acetaminophen metabolite to bovine alpha s1-casein, a soluble protein which does not contain any cysteine residues, was found to occur to an extent of 37% of that which became bound to native BSA. These results were taken to indicate that protein thiol groups are major sites of covalent binding of the reactive metabolite of acetaminophen in vitro. The covalent binding characteristics of synthetic N-acetyl-p-benzoquinoneimine (NAPQI), the putative electrophilic intermediate produced during oxidative metabolism of acetaminophen, paralleled closely those of the reactive species generated metabolically. These findings support the contention that NAPQI is indeed the reactive arylating metabolite of acetaminophen which binds irreversibly to protein.

Dose‐dependent pharmacokinetics of acetaminophen: Evidence of glutathione depletion in humans

The time course of excretion of acetaminophen and its metabolites in urine was determined in eight healthy adults (seven men and one woman) who ingested 1 gm of the drug and collected timed urine samples for 24 hours. The mean time of peak excretion rate was 1.3 to 3.7 hours for acetaminophen, its glucuronide, sulfate, cysteine, mercapturate, and methoxy metabolites but 13.5 hours for methylthioacetaminophen. The mean half‐life of acetaminophen was 3.1 hours and the mean half‐life of the metabolites other than methylthioacetaminophen ranged from 4.1 to 5.7 hours. The half‐life of methylthiometabolite could not be determined because of its very late peak time. In a second study the effect of dose on the clearance of acetaminophen was determined in nine healthy adult subjects (eight men and one woman) who received doses of 0.5 and 3 gm acetaminophen on separate occasions, separated by 4 to 10 days. The renal clearance of acetaminophen and the formation clearances of the sulfate, glutathione, and catechol metabolites were lower (by 38%, 41%, 35%, and 46%, respectively) at the higher dose. The renal clearance of acetaminophen sulfate and glucuronide conjugates were not different between doses. In a third study (10 men), 10 gm N‐acetylcysteine was found to increase the formation clearance of the sulfate conjugate by 27% and that of the glutathione conjugate by 10%. The data suggest that the hepatic supply of reduced glutathione and 3′‐phosphoadenosine 5′‐phosphosulfate begins to be depleted over the range of 0.5 to 3 gm acetaminophen and that the depletion is overcome by the administration of N‐acetylcysteine.

Inhibition and induction of cytochrome P4502E1‐catalyzed oxidation by isoniazid in humans

We studied the effect of isoniazid administration on the cytochrome P4502E1—catalyzed elimination of chlorzoxazone and acetaminophen. Isoniazid, 300 mg daily, was administered for 7 days to a group of 10 volunteer slow acetylators. Acetaminophen, 500 mg, and chlorzoxazone, 750 mg, were administered on separate occasions before isoniazid, during the period of isoniazid administration, and after the discontinuation of isoniazid. Isoniazid inhibited the clearance of chlorzoxazone by 58%, as assessed from plasma data, and inhibited the formation of acetaminophen thioether metabolites (a measure of the formation of the hepatotoxin N‐acetyl‐p‐benzoquinone imine and catechol oxidative metabolites of acetaminophen, as determined from their recovery in urine, by 63% and 49%, respectively. Two days after the discontinuation of isoniazid, the clearance of chlorzoxazone was increased over the value before isoniazid by 56%. Acetaminophen thioether but not catechol metabolites were increased by 56% 1 day after the discontinuation of isoniazid and had returned to the pre‐isoniazid value 3 days after the discontinuation of isoniazid. We conclude that the time course of the interaction with regard to chlorzoxazone elimination and formation is compatible with an inhibition‐induction effect of isoniazid on cytochrome P4502E1. The mechanism of this biphasic effect is probably induction by protein stabilization, which results in inhibition of catalytic activity while isoniazid is present.