Pre-existing liver disease

Pre-existing liver disease may increase the risk of developing drug-induced hepatic injury with agents such as methotrexate, cytotoxic agents, aspirin and sodium valproate. In general, patients with liver disease are more likely to suffer ADRs. A past medical history of DILD from any medication has been shown to be a predictor of future DILD from other drugs.

Age

It is generally accepted that the elderly are at an increased risk of ADRs. There are multiple reasons for this, including higher exposure rates and decreased metabolism. Drug-induced hepatic injury is more likely to occur in elderly patients than in those under 35 years of age. Similarly, halothane hepatitis and isoniazid or chlorpromazine hepatotoxicity are more likely in patients over 40 years of age. The severity of the reactions also appears to increase with age, especially in those over 60 years. Sodium valproate toxicity, on the other hand, demonstrates an increased risk of developing serious or fatal hepatotoxicity in those under 3 years, with risk decreasing as age advances. Aspirin is an example of another drug causing hepatotoxicity, specificallyin children. Reye’s syndrome in children has been linked to the use of aspirin following a viral illness; it is life-threatening and associated with coma, seizures and liver failure. A cholestatic presentation of DILD is more common in older patients and hepatocellular damage is generally more common in younger patients.

Gender

The frequency of drug-induced hepatotoxicity is thought to be more common in females than males, particularly with halothane, isoniazid, flucloxacillin, chlorpromazine, erythromycin and nitrofurantoin. However, evidence from a recent series of more than 600 cases of DILD found that female gender was not a risk factor. There is weak evidence of a gender difference in toxicity of sodium valproate, being more common in boys before puberty and in females after puberty. Cholestatic jaundice associated with co-amoxiclav has been reported to be more common in males than females.

Genetics

Genetic differences that affect an individual’s ability to metabolise certain drugs may predispose to DILD. For example, both fast and slow acetylators may be more susceptible to isoniazid-induced liver damage. The conventional view is that rapid acetylators are at risk of increased toxic reactions due to transformation of acetylhydrazine by cytochrome P450 into a reactive metabolite; other studies suggest slow acetylation may result in toxicity due to formation of hydrazine, which is toxic in itself. When starting, isoniazid monitoring of LFTs is recommended monthly for the first 3 months, as toxicity is most likely to occur early in therapy.

Halothane-induced injury has been reported for multiple family members.

It is thought that a genetic predisposition to allergic forms of drug hypersensitivity could be a factor in some types of liver disease. Flucloxacillin liver injury has been associated with the human leukocyte antigen B*5701. Having this HLA haptotype confers an 80-fold increase in susceptibility to liver injury with flucloxacillin.

Enzyme induction

Alcohol, rifampicin and other drugs that induce cytochrome P450 isoenzyme 2El potentiate the risk of hepatotoxicity with other drugs such as paracetamol, isoniazid and halothane. The role of alcohol as a risk factor for DILD is, however, not clear-cut and acute and chronic alcohol consumption may have different effects.

Concomitant therapy with other anticonvulsants, particularly phenytoin and phenobarbital, is a risk factor for toxicity with sodium valproate, where 90% of cases of liver injury are associated with combination therapy.

Polypharmacy

A typical example of this is seen with NSAIDs. The risk of liver disease with NSAIDs is normally extremely low but is increased when NSAIDs are used with other hepatotoxic drugs.

Concurrent diseases and pregnancy

Pre-existing renal disease, diabetes, pregnancy and poor nutrition may all affect the ability of the liver to metabolise drugs effectively and may put the patient at risk of developing liver damage. Table 15.3 summarises the host factors that may predispose a patient to drug hepatotoxicity.

Table 15.3 Examples of host factors that predispose to drug hepatotoxicity

Necrosis

Necrosis is characterised by cytotoxic cellular breakdown (hepatocellular destruction) and is generally associated with a poor prognosis. Drugs commonly associated with DILD have a variable propensity to cause hepatic necrosis. For example, hepatic necrosis has been reported in 3% of cases with co-amoxiclav DILD compared to 89% in individuals with halothane-induced liver injury cases.

Paracetamol causes hepatic necrosis when its normal metabolic pathway is saturated. Subsequent metabolism occurs by an alternative pathway that produces a toxic metabolite which covalently binds to liver cell proteins and causes necrosis.

Steatosis

In steatosis, hepatocytes become filled with small droplets of lipid (microvesicular fatty liver) or occasionally with lipid droplets that are much larger (macrovesicular fatty liver).Tetracyclines are thought to cause steatosis by interfering with synthesis of lipoproteins that normally remove triglycerides from the liver.

Cholestasis

Some drugs injure bile ducts and cause partial or complete obstruction of the common bile duct, resulting in retention of bile acids and the condition known as cholestasis. Cholestasis caused by anabolic and contraceptive steroids is due to inhibition of bilirubin excretion from the hepatocyte into the bile.

The penicillins, although commonly associated with allergic drug reactions, are a very rare cause of liver disease. The isoxazoyl group present in the synthetic β-lactamase resistant oxypenicillins has been implicated as a cause of liver injury. Acute cholestatic hepatitis has increasingly been reported during treatment with flucloxacillin, and in some countries this has become the most important cause of drug-induced cholestatic hepatitis. The incidence appears to be about twice that of the related isoxazoyl penicillins cloxacillin and dicloxacillin. Moreover, there is likely to be underreporting due to a delay in onset of up to 42 days after stopping treatment. Female sex, age over 55 years, longer courses and high daily doses also seem to be associated with a higher risk of liver reaction to flucloxacillin.

Rifampicin causes hyperbilirubinaemia by inhibiting uptake of bilirubin by the hepatocyte as well as inhibiting bilirubin excretion into bile. This is generally not an indication for interrupting rifampicin therapy, although liver function will need to be closely monitored. Other therapeutic agents affect sinusoidal or endothelial cells, which may result in veno-occlusive disease or fibrosis. Vitamin A affects the fat storing cells, causing toxicity that leads to fibrosis.

Hepatocellular necrosis

In severe cases, acute hepatocellular necrosis presents with jaundice and LFT abnormalities, including a modestly raised alkaline phosphatase and a markedly elevated alanine aminotransferase level of up to 200 times the upper limit of the reference range. Prolongation of the prothrombin time occurs but depends on the severity of the injury, increasing dramatically in severe cases. Microscopy reveals necrosis of the hepatocytes in a characteristic pattern.

Steatosis

Steatosis (fatty liver) is the accumulation of fat droplets within liver cells and is associated with abnormal LFTs, although the elevation of alanine aminotransferase is not as high as that seen in acute hepatocellular necrosis. Hyperammonia, hypoglycaemia, acidosis and clotting factor deficiency may also be present. Histologically, the liver damage resembles the acute fatty liver of pregnancy. Fat distribution within the hepatocyte is either microvesicular, as occurs with tetracycline, aspirin and sodium valproate, or macrovesicular where the hepatocyte cell nucleus is displaced to the periphery by a single large fat droplet. This type of damage occurs typically with steroids, methotrexate, alcohol and amiodarone.

A less severe, more chronic form of fatty liver, steatohepatitis, also occurs. Steatohepatitis differs from diffuse fattychange. Notably, the clinical symptoms and biochemistry resemble chronic parenchymal disease and the histology is similar to that seen in alcoholic hepatitis. Amiodarone is an example of a drug that can cause chronic steatohepatitis associated with phospholipidosis.

Cholestasis

Cholestasis without hepatitis is associated with a raised bilirubin and a normal or minimally raised alanine aminotransferase level. No inflammation or hepatocellular necrosis is seen. In contrast, cholestasis associated with hepatitis presents with raised bilirubin, alanine aminotransferase and alkaline phosphatase levels and a certain amount of liver damage.

Granulomatous hepatitis

Granulomatous hepatitis occurs with modestly elevated LFTs and, usually, normal synthetic liver function. Histology reveals granulomas and tissue eosinophilia.

Acute hepatitis

Acute hepatitis resembles viral hepatitis with LFTs raised in proportion to the severity of the hepatocellular damage. The best indicator of severity is the prothrombin time. Histologically, necrosis and cellular degeneration are seen in combination with an inflammatory infiltrate.

Chronic active hepatitis

Chronic active hepatitis may present as an acute injury or progress to cirrhosis. Serum transaminases are usually raised and albumin is low. The histology resembles that of autoimmune chronic active hepatitis and is associated with circulating autoantibodies. Methyldopa is an example of a drug that can cause chronic active hepatitis.

Fibrosis

In patients with fibrosis, the serum transaminase levels may be only slightly raised, and are not good predictors of hepatic damage. Microscopy shows deposition of fibrous tissues. Fibrosis may proceed to cirrhosis. Such damage may be seen with long-term methotrexate use.

Vascular disorders

A variety of drugs can cause veno-occlusive disease, which is characterised by non-thrombotic narrowing of small centrilobular veins, and is typically caused by cytotoxic agents and some herbal remedies. Use of oral contraceptives or cytotoxic agents may exacerbate an underlying thrombotic disorder and increase the risk of the Budd–Chiari syndrome (obstruction of the large veins) developing.

Tumours

Drugs have been associated with a variety of hepatic tumours. The drugs most commonly linked to malignancy are the oral contraceptives, anabolic steroids and danazol.

Acute hepatocellular necrosis

In acute hepatocellular necrosis caused by paracetamol, early symptoms include anorexia, nausea and vomiting, malaise and lethargy. Abdominal pain may be the first indication of liver damage but is not usually apparent for 24–48 h. A period of apparent recovery precedes the development of jaundice and production of dark urine. If the liver injury is severe, deterioration follows, with repeated vomiting, hypoglycaemia, metabolic acidosis, bruising and bleeding, drowsiness and hepatic encephalopathy. Oliguria (diminished urine output) and anuria (complete cessation of urine production) may result from acute tubular necrosis. Renal failure may occur even in the absence of severe liver disease. In addition to acute renal failure, myocardial injury and pancreatitis have also been reported. In fatal cases, death from acute hepatic failure occurs between 4 and 18 days after ingestion.

Steatosis

A patient presenting with steatosis generally shows fatigue, nausea, vomiting, hypoglycaemia and confusion. Jaundice is present in severe cases.

Acute hepatitis

Acute hepatitis may present with a prodromal illness with non-specific symptoms or include features of drug allergy followed by anorexia, nausea and vomiting, dark urine, pale stools and jaundice. Jaundice tends to be present in severe cases. Weight loss may also be a feature of acute hepatitis. Fatalities occur in 5–30% of jaundiced patients. Acute hepatitis is second only to paracetamol self-poisoning as a cause of DILD.

Chronic active hepatitis

Drug-induced chronic active hepatitis may present with tiredness, lethargy and malaise, in a manner similar to other types of chronic liver disease. The symptoms may evolve over many months. Gastro-intestinal symptoms are usually present, and patients may show one or more complications of severe liver disease, including ascites, bleeding oesophageal varices or hepatic encephalopathy. If the ADR has an allergic component, a skin rash and other extrahepatic features of a drug allergy such as lymphadenopathy, evidence of bone marrow suppression (particularly petechial haemorrhages) may be present.

Cholestasis

The main clinical feature of pure cholestasis is severe pruritus, with or without other features, according to the severity, such as dark urine, pale stools and jaundice.

Drug-induced cholestatic hepatitis usually presents with gastro-intestinal symptoms following an influenza-like illness. Abdominal pain with typical features of cholestasis then occurs. The pruritus is generally less severe than with pure cholestasis.

Veno-occlusive disease

Veno-occlusive disease may present with painful hepatomegaly, ascites and jaundice along with other features of liver insufficiency. It has been reported following chemotherapy with drugs such as cyclophosphamide, doxorubicin and dacarbazine. It has also been reported as a common complication of bone marrow transplantation.

Hepatic tumours

In general, patients with hepatic tumours present in a similar manner. Abdominal pain may or may not be reported, together with a feeling of fullness after eating. Weight loss, fatigue, anorexia, nausea and, occasionally, vomiting can occur, especially in advanced cases.

Biochemical tests

Routine LFTs are measured, which generally include total bilirubin, alanine transaminase and alkaline phosphatase. Impairment of the synthetic function of the liver is detected by total protein, albumin and the prothrombin time. Other biochemical tests may include measurement of γ-glutamyl transpeptidase which may be elevated in all forms of liver disease, including drug-induced disease. α-Fetoprotein may be measured to exclude malignancy. Conjugated bilirubin may be measured to establish if there is biliary obstruction.

Serological markers

Serological markers for hepatitis A, B and C and other viruses such as the Epstein–Barr virus should be determined in patients with symptoms of hepatitis with appropriate risk factors to exclude an infective cause.

Radiological investigations

Radiological investigations, such as ultrasound, computed tomography, percutaneous cholangiograms and endoscopic retrograde cholangiopancreatography (ERCP), are used to look for physical obstruction of bile ducts by gallstones, masses or strictures.

Liver biopsy

Liver biopsy is seldom helpful for diagnosis but certain drugs can cause characteristic lesions, such as the distribution of microvesicular fat droplets seen with tetracyclines. Specific diagnostic tests for drug-induced disease exist for few drugs, with halothane being a notable exception.

Other causes of liver dysfunction such as autoimmune chronic active hepatitis, acute severe cholestasis, ischaemic hepatic necrosis, pregnancy-related liver disease, the Budd–Chiari syndrome, rare metabolic disorders or liver disease related to alcohol abuse must also be excluded.

Diagnosis

Drug-induced hepatic injury should be suspected in every patient with jaundice while ruling out other causes of liver disease by the clinical history and the results of investigations. The typical process in screening patients presenting with jaundice is outlined in Fig. 15.2 and the general approach to the differential diagnosis of acute hepatitis is set out in Fig. 15.3.

Fig. 15.2 General approach to the diagnosis and management of drug-induced jaundice.

Fig. 15.3 General approach to the differential diagnosis of acute hepatitis. ALT, alanine transferase; ALP, alkaline phosphatase; INR, international normalised ratio; ANA, antinuclear antibodies.

Drugs that are commonly prescribed, such as NSAIDs, antimicrobials and antihypertensive agents, are more likely to be implicated in DILD, although the frequency for the individual agents is low. Identifying the causative agent and stopping it is important in reducing the morbidity and mortality associated with DILD. Recovery normally follows discontinuation of a hepatotoxic drug. Serious toxicity or ALF may result if the drug is continued after symptoms appear or the serum transaminases rise significantly. Failure to discontinue the drug may give grounds for claims of negligence.

A detailed and thorough drug history, including use of oral contraceptives, over-the-counter medicines, vitamins, herbal preparations and illicit drug use, should be obtained. Examples of herbal and dietary preparations implicated in causing liver damage are listed in Box 15.1. Attention to the duration of treatment with a specific drug and the relationship to the onset of symptoms is important. The likelihood of a drug-related disease is greatest when the abnormality begins between 5 and 90 days after taking the first dose and within 15 days of taking the last dose. The latent period, that is, the time between starting therapy and the appearance ofsymptoms, may vary but for many drugs is sufficiently reproducible to be of some diagnostic value.

Predisposing factors for liver toxicity should also be noted. If the liver injury is accompanied by fever, rash and eosinophilia, the likelihood of drug-induced disease increases, although lack of these features does not exclude it. Unequivocal diagnosis cannot be made in most circumstances, and improvement on withdrawal of the implicated drug may provide the strongest evidence for drug-induced disease. Time for resolution of the abnormalities is dependent on the individual drug and type of liver disease. In some cases, several months may elapse.

Idiosyncratic reactions may also need to be considered and the literature consulted for previous reports. A key component of secondary prevention is the reporting of all suspected hepatic drug reactions to the appropriate monitoring agency, particularly for newer agents with fewer published cases. Many drugs are approved and licensed before the idiosyncratic reactions are identified as there is little chance of detecting the reaction in the phase III studies which typically involve 300 patients. To detect a single case of clinically significant hepatic injury due to a drug with 95% confidence, the number of patients included in the trial must be about three times the incidence of the reaction. Idiosyncratic reactions occur in about 1 in 10,000 patients so to detect the reaction, the clinical trial would have to study 30,000 patients. Practice points for diagnosing DILD are shown in Box 15.2.

Withdrawal

Once drug-induced hepatotoxicity has been recognised as a possibility, therapy should be stopped. If the patient is receiving more than one potentially hepatotoxic drug, all drugsshould be stopped. Withdrawal of the agent usually results in recovery that begins within a few days. However, LFTs may take many months or even years to return to normal. Co-amoxiclav and phenytoin are examples of drugs that have been associated with a worsening of the patient’s condition for several weeks after withdrawal, and a protracted recovery period of several months. Examples of drugs associated with chronic liver injury are shown in Box 15.3.

Rechallenge

When drug-induced hepatotoxicity has been confirmed by improvement on drug withdrawal, subsequent use in the patient is generally contraindicated. Rechallenge is not normally justified as this is potentially dangerous for the patient, although a positive rechallenge is the most definitive confirmation of drug-induced disease. Inadvertent rechallenge may occur. If the rechallenge is negative, this is usually taken to indicate that the patient may resume using the drug. Another adverse reaction on re-exposure to the drug precludes any further use.

Management

If clinical or laboratory signs of hepatic failure appear, hospitalisation is mandatory.

After withdrawal of the drug, attempts to remove it from the body are only relevant for acute hepatotoxins such as paracetamol, metals or toxic mushrooms such as Amanita phalloides (death cap).

If patients present a few hours post-ingestion, any unabsorbed drug may best be removed by gastric lavage, rather than by use of emetics.

Paracetamol-induced hepatotoxicity

Paracetamol causes a dose-related toxicity resulting in centrilobular necrosis. It normally undergoes the phase II reactions of glucuronidation and sulphation. However, paracetamol is metabolised by cytochrome P450 2E1 to N-acetyl-p-benzoquinoneimine (NABQI) if the capacity of the phase II reactions is exceeded or if cytochrome P450 2E1 is induced. After normal doses of paracetamol, NABQI is detoxified by conjugation with glutathione to produce mercaptopurine and cysteine conjugates. Following overdose, tissue stores of glutathione are depleted, allowing NABQI to accumulate and cause cell damage. Illness, starvation and alcohol deplete glutathione stores and increase the predisposition to paracetamol toxicity, while acetylcysteine and methionine provide a specific antidote by replenishing glutathione stores.

Ingestion of doses as low as 10–15 g of paracetamol have been reported to cause severe hepatocellular necrosis. Removal of unabsorbed paracetamol by gastric lavage may be worth-while if more than 150 mg/kg body weight has been taken and the patient presents within 4 h of ingestion. Activated charcoal may also be administered to reduce further absorption of paracetamol and facilitate removal of unmetabolised paracetamol from extracellular fluids. This may lessen the effect of any methionine given. A plasma paracetamol concentration should be taken as soon as possible but not within 4 h of ingestion due to the fact that a misleading and low level may be obtained because of continuing absorption and distribution of the drug. The plasma concentration measured should be compared with a standard nomogram reference line of a plot of plasma paracetamol concentration against time in hours after ingestion. This may be a semilogarithmic (Fig. 15.4) or linear (Fig. 15.5) plot. Generally, administration of intravenous acetylcysteine is the treatment of choice for paracetamol overdose when the blood paracetamol level is in the range predictive of possible or probable liver injury (see Fig. 15.4). Patients allergic to acetylcysteine may receive oral methionine.

Fig. 15.4 Semilogarithmic plot of plasma paracetamol concentration versus time in hours after ingestion.

Fig. 15.5 A linear plot of plasma paracetamol concentration versus time in hours after ingestion.

Acetylcysteine is most effective within 8 h of overdose. However, late administration in patients who present more than 16–24 h post-ingestion may be appropriate. Acetylcysteine administered at this stage will not counteract the oxidative effects of paracetamol but it may have a cytoprotective role in hepatic failure, and has been shown to reduce morbidity and mortality in patients who have already developed ALF.

Patient counselling

Patients who purchase preparations containing paracetamol should be made aware of the danger of overdosing, which may occur if other preparations containing paracetamol are taken simultaneously. Since 1994 the European guidelines on package labelling have required products containing paracetamol to warn patients of the need to avoid other products containing paracetamol. The pack size of paracetamol sold from general sales outlets has been limited to 16 tablets or capsules (32 where paracetamol is sold under the supervision of a pharmacist) with the aim of limiting availability and reducing residual stocks in the home. This appears to have reduced the incidence of ALF secondary to paracetamol overdose.

All patients should be advised of potential side effects. This information needs to be reinforced with the use of patient information leaflets.

Patients and their carers should be helped to recognise signs of liver disorder and know to report immediately symptoms such as malaise, nausea, fever and abdominal discomfort that may be significant, although non-specific during the first few weeks of any change of therapy. If these are accompanied by elevated LFTs, the drug should be discontinued.

Patients who recover from drug-induced hepatotoxicity should be informed of the causative agent, warned to avoid it in the future, and advised to inform their doctor, dentist, nurse and pharmacist about the occurrence of such an event.

Parents of children commenced on sodium valproate should be warned to report side effects that may be suggestive of liver injury such as the onset of anorexia, abdominal discomfort, nausea and vomiting. Early features include drowsiness and disturbed consciousness. Fever may also be present. The time to onset is between 1 and 4 months in the majority of cases.

The challenge for all members of the health care team is to alert patients to the potentially toxic effects of drugs without creating so much concern that they fail to comply with vital medication. For the limited number of drugs presented in Table 15.8, careful monitoring of LFTs during the first 6 months of treatment is advisable, although not always practical. Thereafter, regular monitoring of LFTs is appropriate in patients who are at greater risk of hepatotoxicity. Such patients would include those with known liver disease, those taking other hepatotoxic drugs, those aged over 40 years, and heavy alcohol consumers. Surveillance should be particularly frequent in the first 2 months of treatment. In patients with no risk factors and normal pretreatment liver function, LFTs need only be repeated if fever, malaise, vomiting, jaundice or unexplained deterioration during treatment occurs.

Table 15.8 Examples of drugs where regular monitoring of liver function is recommended

a Baseline and subsequent liver function tests (LFTs) difficult to interpret in critically ill patients as LFTs will be affected by multiple factors

Since many drugs cause elevation of LFTs there may be difficulty in assessing when to stop a drug, particularly when treating an individual for tuberculosis or epilepsy. An empirical guideline is that the drug should be stopped if the levels of alanine transaminase exceed three times the upper limit of the reference range. Any clinical features of liver disease or drug allergy would require immediate discontinuation of the drug. Conversely, a raised γ-glutamyl transpeptidase level and elevated alanine transaminase level in the absence of symptoms often reflect microsomal induction and would not indicate drug-induced injury.

It should be noted that monitoring of LFTs is not a complete safeguard against hepatotoxicity, as some drug reactions develop very quickly, and the liver enzymes are an unreliable indicator of fibrosis.

Minimising the risk of DILD

Cholestatic jaundice has been reported to occur in about 1 in 6000 patients treated with co-amoxiclav. The risk of acute liver injury with co-amoxiclav is approximately six times that of amoxicillin and increases with treatment courses above 14 days. Hence, the indications for co-amoxiclav have been restricted to cover infections caused by amoxicillin-resistant β-lactam-producing infections.

Patients admitted for procedures requiring a general anaesthetic should be questioned about past exposure and any previous reactions to halothane. Halothane is well known to be associated with hepatotoxicity, particularly if patients are re-exposed. Repeated exposure to halothane within a period of less than 3 months should be avoided, while some increase in risk persists regardless of the time interval since last exposure. Unexplained jaundice or delayed-onset post-operative fever in a patient who has received halothane is an absolute contraindication to future use in that individual. Patients with a family history of halothane-related liver injury should also be treated with caution.

The individuals at greatest risk of halothane hepatitis are obese, post-menopausal women. Halothane may be present in detectable amounts even in theatres equipped with scavenging devices, and it is possible for these small concentrations to provoke a reaction in a highly sensitised individual. If electing to avoid halothane, a halothane-free circuit and operating theatre should be used. Cross-hepatotoxicity with other haloalkanes is a possibility, and enflurane should also be avoided. Isoflurane appears to be safe, as no reports of cross-sensitivity have been published. Some anaesthetists would prefer to use total intravenous anaesthesia in patients who have had a reaction to halothane.

Although hepatic ADRs are rare for most drugs, when they do occur they can cause significant morbidity and mortality. Over 600 drugs have been associated with hepatotoxicity and any new drug released on to the market may have the potential to cause hepatotoxicity. Pemoline, troglitazone and tolcapone are examples of drugs withdrawn from the market due to reports of serious hepatic reactions. These examples help to highlight the importance of post-marketing surveillance and yellow card reporting.

Appropriate selection of drugs, an awareness of predisposing factors and avoidance of toxic dose thresholds and potentially hepatotoxic drug–drug interactions will minimise the risk to patients.

Practice points for patient care and minimising the risk of DILD are outlined in Box 15.4.