CHAPTER 8 Drugs and Toxins
Introduction
This chapter deals with the pathology of the important liver lesions attributed to drugs and toxins, with their recognition, and with their differential diagnosis. There are many hundreds of hepatotoxic drugs and other chemicals, and new reports of adverse drug reactions appear regularly in the literature under the acronym ‘DILI’ (drug-induced liver injury). Details of the lesions attributed to each of the many individual drugs are not considered appropriate for a bench book of this kind. If required, the information can be found in one of several comprehensive reviews.1–5 There is also a regularly updated bibliography in the French literature.6 The pathologist reporting a liver biopsy should, however, bear in mind that a drug cannot be exonerated simply because an adverse reaction has not been reported; there is always a first time.
Chemical injury is not confined to drugs listed in pharmacopoeias. Herbal medicines,7,8 illicit drugs,9–17 criminally administered poisons,18 industrial chemicals,19–22 vitamins23,24 and foods25,26 have all been held responsible for liver disease. Drugs used for the treatment of liver disease have themselves been suspected of causing liver damage.27
In his foreword to the second edition of Stricker’s Drug-Induced Hepatic Injury2 Zimmerman wrote: ‘… virtually all known acute and chronic hepatic lesions can result from drug injury.’ This important observation implies that drugs should be considered as a possible cause of any liver lesion found on biopsy, but some lesions are more often produced by drugs than others. Hepatocellular necrosis, hepatitis and cholestasis in particular should arouse a greater degree of suspicion, especially if no other cause has been found. Also, some groups of drugs are associated with particular kinds of injury; non-steroidal anti-inflammatory drugs (NSAIDs), for example, are often associated with hepatocellular injury, while neuroleptic drugs mostly cause cholestasis. However, these are generalisations and a drug which causes a dose-related hepatocellular necrosis in one patient may cause non-dose-related hepatitis, cholestasis or granulomas in another.28,29
Classification and mechanisms
Drugs may be regarded as producing liver injury in two main ways (Table 8.1). Intrinsic (predictable) hepatotoxins are those which predictably produce liver damage when taken in sufficient quantities. The type of damage is often characteristic of a particular drug; for example, the typical result of paracetamol (acetaminophen) overdose is hepatocellular necrosis and steatosis. Intrinsic hepatotoxicity can often be studied in laboratory animals. This type of hepatotoxicity is also frequently zonal in distribution; examples of this are the perivenular lesions of paracetamol and carbon tetrachloride and the periportal necrosis seen in phosphorus toxicity. The mechanism of intrinsic hepatotoxicity can be direct or indirect; in the former, the chemical or its metabolites causes structural damage to cells and organelles, while in indirect intrinsic hepatotoxicity the chemical interferes with a specific metabolic pathway or cell component.
Table 8.1 Examples of liver lesions due to drugs and toxins
| Lesion | Example of substance |
|---|---|
| Intrinsic hepatotoxicity | |
| Idiosyncratic hepatotoxicity | |
The commoner kind of drug-related liver damage is idiosyncratic (unpredictable). Only a small proportion of patients on a particular drug is affected, so that the adverse reaction is not detected in initial human trials. Many different mechanisms for idiosyncratic hepatotoxicity have now been elucidated. They include individual genetic variation in the metabolism of drugs, and the development of immune reactions to a drug or its metabolites.30 The immune reactions may be directed to neoantigens produced by the binding of reactive metabolites to hepatic drug metabolising enzymes of the P450 system.31,32 In some instances the distinction between an idiosyncratic and intrinsic drug reaction is difficult to make. Typical idiosyncratic damage may follow a small dose of the offending drug, and cannot easily be studied in the laboratory. With the exception of a few drugs shown to cause liver damage in patients using a particular metabolic pathway, idiosyncratic drug injury is unpredictable in the sense that the susceptibility of individual patients cannot be tested before the drug is given.
Most intrinsic hepatotoxins produce liver damage within a few hours or days, whereas in the idiosyncratic type of injury there is often a latent period of many days, weeks or months33 before liver disease becomes apparent. The latent period tends to shorten with repeated adminstration of the drug. Because of the latent period and the tendency for idiosyncratic injury to mimic non-drug-related liver diseases, clinicians and pathologists need to be alert to the possibility of idiosyncratic drug injury if diagnostic errors are to be avoided. The clinician may be helped by a clinical scale or scoring system,34 and the pathologist by a suspicious or characteristic pattern of injury. Conclusive proof that a particular drug or combination of drugs is responsible is often impossible to obtain, although re-challenge (usually inadvertent) can provide strong circumstantial evidence. Liver injury may follow inadvertent re-challenge many years after a first episode.35 Biochemical evidence of improvement after drug withdrawal is occasionally supported by a return to normal histology.33
Morphological categories
The categories described below represent the main changes attributed to drugs and toxins, apart from alcohol-related liver damage (Ch 7), neoplasms (Ch 11) and vascular lesions (Ch 12). A mixture of lesions may be found in the same liver; amiodarone, for example, produces both phospholipidosis and steatohepatitis, but by different mechanisms.36 As already indicated, a single drug may give rise to different forms of hepatotoxicity in different patients. Phenylbutazone, for example, can cause necrosis, cholestasis, granuloma formation or combinations of these,37 while the NSAIDs nimesulide and diclofenac can cause either severe hepatitis or cholestasis.38,39
Adaptation
Not all changes seen under the microscope necessarily represent liver damage. The increase in endoplasmic reticulum produced by long-term treatment with anticonvulsant drugs is commonly regarded as an adaptive phenomenon.40,41 By light microscopy, this increase is seen as an abundance of pale-staining cytoplasm in hepatocytes (see Figs 4.8 and 8.1), which is difficult to distinguish from simple abundance of glycogen on a haematoxylin–eosin (H&E)-stained section.
Non-hepatitic liver-cell damage
One of the commonest manifestations of intrinsic hepatotoxicity is steatosis. As discussed in Chapter 7, this may be macrovesicular or microvesicular. Macrovesicular steatosis, in which the nucleus of the hepatocyte is displaced by one or more fat vacuoles easily visible by light microscopy, is produced by chlorinated hydrocarbons and methotrexate, for example. It is common in patients on total parenteral nutrition,42,43 although underlying disease may also contribute to the liver changes.44 In patients treated with gold compounds for rheumatoid arthritis, intralobular lipogranulomas (focal accumulations of lipid-containing macrophages) have been found to contain gold pigment in the form of fine black or brown granules. These were also seen within portal lipid droplets.45
Causes of the more serious microvesicular steatosis46 (Fig. 8.2) include treatment with the anticonvulsant drug valproate47 and with the nucleoside analogue fialuridine.48 This leads to the combination of microvesicular steatosis with mitochondrial abnormalities, found also in Reye’s syndrome (see Ch. 13). Similar changes are reported after zidovudine,49 didanosine (see Fig. 8.3) and other nucleoside reverse transcriptase inhibitors in highly active antiretroviral therapy (HAART) for AIDS.50–52 In the microvesicular form of steatosis the fat within the hepatocytes is finely divided and is not always obvious with conventional stains. The hepatocyte nuclei remain in their normal central location, in contrast to macrovesicular steatosis. There is a variable degree of associated hepatocellular necrosis.
Figure 8.2 Microvesicular steatosis.
(The section was kindly provided by Professor B. C. Portmann, London, UK.)
Several drugs, among them amiodarone36 and trimethoprim-sulphamethoxazole (co-trimoxazole),53 are causes of acquired phospholipidosis. Similar changes have been reported in patients receiving total parenteral nutrition.54 Lamellar inclusions are seen within hepatocytes and other cells by electron microscopy (see Fig. 17.3). Light microscopy of conventionally stained sections is not diagnostic.
In acute arsenic intoxication, a striking increase in hepatocyte mitoses has been reported, accompanied by ballooning, cholestasis and mild inflammation.18 Markers of cell proliferation were also markedly increased.
An unusual form of cell injury is produced by cyanamide, used in alcohol aversion therapy.55–57 Periportal hepatocytes contain large, pale-staining cytoplasmic inclusion bodies, giving the cells a superficial resemblance to the ground-glass cells of chronic type B hepatitis (see Figs 4.8 and 9.13). The inclusions are, however, orcein-negative and diastase–periodic acid–Schiff positive.
Hepatocellular necrosis
Hepatocellular necrosis without the diffuse inflammatory lesion of hepatitis is usually a consequence of the intrinsic type of hepatotoxicity. A common example is suicidal or accidental overdose with the analgesic paracetamol (acetaminophen).58 Jaundice develops after an interval of days, during which available glutathione, which reacts with a toxic metabolite, is used up. The necrosis – like that of shock or heatstroke (see Fig. 12.2) – is most severe in perivenular regions (acinar zones 3) and is accompanied by little or no inflammation (Fig. 8.4). Kupffer cells contain brown ceroid pigment. Portal tracts usually remain normal. Complete recovery is possible. While most paracetamol-induced necrosis follows suicidal overdose, it is occasionally found in habitual drinkers taking large doses in the high therapeutic range.59
Figure 8.4 Hepatocellular necrosis due to paracetamol (acetaminophen).
Confluent necrosis with little inflammation is seen in a perivenular area. (Needle biopsy, H&E.)
Hepatocellular necrosis, sometimes accompanied by steatosis, is also a feature of cocaine intoxication,9,11,13 of glue sniffing and of solvent abuse.14,60 In most instances the necrosis is perivenular and mid-zonal (in acinar zones 3 and 2), but periportal (zone 1) necrosis has been reported in a cocaine user.61 ‘Ecstasy’ (3,4-methylenedioxymethamphetamine, MDMA) can cause a hepatitic lesion of the kind described in the next section,15,16,62 but there may also be confluent hepatocellular necrosis as a result of concurrent hyperthermia.17 Other agents capable of causing confluent necrosis include industrial hydrochlorofluorocarbons.19
Acute drug-induced hepatitis
A large number of drugs of different chemical structure and with widely differing pharmacological actions occasionally give rise to acute hepatitis, and any drug should be regarded as a potential offender. Acute drug-induced hepatotoxicity is of the idiosyncratic type. The histological lesion is very like that of acute viral hepatitis, and often indistinguishable from it (Figs 8.5, 8.6). Incriminated substances include antituberculous drugs,63 NSAIDs, anaesthetics,64,65 herbal remedies8 and many others.
Figure 8.6 Drug-induced liver injury: hepatitic type.
There is a severe lobular hepatitis with disruption of liver-cell plates and apoptosis, attributed to ecstasy (3,4-methylenedioxymethamphetamine). The patient is the second of the two reported by Fidler and colleagues.16 (Needle biopsy, H&E.)
The hepatitis ranges in severity from a mild inflammatory lesion, sometimes combined with a cholestatic reaction (see under Cholestasis, below), to severe and even fatal disease.66 In milder cases, removal of the drug usually leads to rapid improvement. Later uncommon outcomes also include cirrhosis and the development of autoimmune hepatitis.67
Differential diagnosis
The possibility of drug idiosyncrasy should be considered in all patients with acute hepatitis, because in many cases the histological appearances are identical to those of viral hepatitis. A higher than usual degree of suspicion should be aroused when the hepatitis is histologically unusual (Table 8.2). Well-demarcated centrilobular confluent necrosis (Fig. 8.5) is common. There may be a very mild lobular hepatitis together with canalicular cholestasis. The portal inflammatory reaction may be poorly developed or even absent. Conversely, the portal infiltrate may be unusually rich in neutrophils or eosinophils, although the latter are neither proof of drug aetiology nor necessary for its diagnosis. The presence of epithelioid-cell granulomas increases the likelihood that drug idiosyncrasy is the correct diagnosis.
Table 8.2 Features sometimes associated with drug-induced hepatitis
