Portal de Periódicos da CAPES

Drug-induced liver injury: Hy’s rule revisited

2006; Wiley; Volume: 79; Issue: 6 Linguagem: Inglês

10.1016/j.clpt.2006.02.012

1532-6535

E BJORNSSON,

Liver Disease Diagnosis and Treatment

Clinical Pharmacology & Therapeutics (2006) 79, 521–528; doi: 10.1016/j.clpt.2006.02.012 Drug-induced liver injury is a potential complication of many drugs. This is to be expected because the liver plays a central role in drug metabolism. Only a very few drugs causing drug-induced liver injury have a predictable, dose-dependent toxic mechanism of action, the most widely recognized being acetaminophen (INN, paracetamol) and, to some extent, methotrexate. When acetaminophen overdose is excluded, in most cases drug-induced liver injuries are due to rare, unpredictable reactions of commonly used drugs. These unpredictable reactions are mostly either idiosyncratic or immune-mediated, or both, but may also, in some cases, have a dose-related component. Current definition of drug-induced liver injury is dependent on nonspecific elevations in liver test results, in particular, increases in aminotransferase levels. Histologic manifestations of drug-induced liver injury can mimic most other known liver diseases, and there are no pathognomonic histologic features of drug-induced liver injury. In most cases of drug-induced liver injury, liver biopsy findings are "compatible with" drug-induced liver injury, and the histologic examination often tells more about the extent of the injury than the cause. Consideration of the course of events and the results of dechallenge, as well as the results of other tests, is necessary to achieve a high diagnostic accuracy of drug-induced liver injury. The combination of high serum aminotransferase levels and jaundice was reported by the legendary pioneer on the study of drug-induced liver injury Hyman Zimmerman to result in a mortality of 10% to 50% for different drugs. These observations have been named Hy's rule. Hy's rule, defined as ALT level greater than or equal to 3 times the upper limit of normal (ULN) plus serum bilirubin level greater than or equal to 2 times the ULN, has been advocated by the US Food and Drug Administration for use in the assessment of the hepatotoxicity of newly developed drugs. Recently, this rule has been validated for the first time by two different groups in a large number of patients with suspected drug-induced liver injury. Drug-induced liver injury is one of the most common reasons for termination of drug development of otherwise promising therapeutic agents after preclinical studies.1 Drug-induced hepatotoxicity has in recent years been a subject of increasing interest because of withdrawal of a number of drugs shortly after being put onto the market.1,2 Drug-induced liver injury has become the major reason for withdrawal of drugs after release.1,2 The number of patients included in clinical trials usually ranges from several hundred up to 8000,3 and because only 1 in 10,000 or 1 in 100,000 patients who receive treatment may have drug-induced liver injury, these phase III trials are generally underpowered to detect uncommon, but in some cases life-threatening, adverse liver reactions. Reports on the frequency of adverse drug liver reactions as a proportion of all adverse reactions to drugs range from 4.2% in New Zealand4 and 5.9% in Denmark5 to approximately 10% of adverse reactions involving the liver in the United States.6 Acute hepatic injury caused by drugs has been reported to occur in between 5% and 10% of patients hospitalized because of jaundice.7,8,9,10 Thus drug-induced liver injury is not a common cause of jaundice. However, drugs are the most common cause of fulminant hepatic failure in both the United States and Europe.11,12,13 Acetaminophen overdose has been reported to be the single most common cause of acute liver failure in the western world,11,12,13 and idiosyncratic drug reactions were the presumptive causes in 13% to 17% of cases of acute liver failure in the United States and Sweden.11,12 Early recognition of drugs as a cause of acute liver dysfunction is important to decrease the risk of morbidity and mortality. However, the prognosis for patients with acute liver failure, who by definition are initially seen with coagulopathy and encephalopathy, caused by idiosyncratic drug reactions is usually poor, with approximately 60% to 80% mortality without liver transplantation.11,12,13 However, among all cases with suspected drug-induced liver injury with either elevated aminotransferase levels or elevated alkaline phosphatase levels, or both, only a minority of patients have concomitant jaundice, and the proportion with concomitant acute liver failure is even smaller.14,15 In a recent study from Sweden, among 3841 notifications of adverse hepatic reactions received by the Swedish Adverse Drug Reactions Advisory Committee from 1970 to 2004 and considered to have at least a possible causal relationship, 873 (23%) patients had concomitant jaundice.14 It is very important for the clinician always to include drug-induced liver injury as a differential diagnosis in the diagnostic workup of either jaundice or abnormal liver test results, or both. The clinical scenario of drug-induced liver injury can mimic all kinds of known liver pathologic conditions, and recent studies have shown that drug-induced liver injury is frequently missed.16,17 Missing a diagnosis of drug-induced liver injury can result in severe morbidity or mortality from acute liver failure.11,12,13 Unfortunately, there are numerous examples of patients with severe drug-induced liver injury who have undergone biliary tract surgery for suspicion of gallstone disease with postoperative worsening of the liver disease. Continuation of drug therapy with drugs that have been suspected to cause drug-induced liver injury has also been reported to lead to chronic liver injury.18 It is also important to bear in mind that causality assessment of drug-induced liver injury can be difficult, and making a diagnosis of drug-induced liver injury is far from straightforward. Unfortunately, there are no specific diagnostic markers or tests that can be used to verify the suspected diagnosis of drug-induced liver injury. The diagnosis of drug-induced liver injury is based on "guilt by association."19 Most cases of drug-induced liver injury are suspected in patients with a close temporal association with onset and the course of the liver injury after withdrawal of the suspected agent. Improvement of liver test results after cessation of the drug (positive dechallenge), although supportive of a drug etiology, is obviously not enough for a diagnosis of drug-induced liver injury. Exclusion of other etiologies is equally important to correctly diagnose the liver injury. Information about previously documented hepatotoxicity of the drug adds to suspicion of guilt, but in the absence of such supportive evidence the clinician must consider the possibility that the current reaction might be the first recognized adverse liver reaction. Thus, well-worked-up and published case reports with careful consideration and exclusion of nondrug causes are very valuable for the documentation of previously unreported drug-induced liver injuries. Most countries in the western world have voluntary reporting of adverse drug reactions. A recent study from the United Kingdom revealed that drug-induced liver disease was wrongly implicated in almost half of alleged cases of drug-induced liver disease when subjected to a critical view.20 This critical review consisted of a systematic use of the International Consensus Criteria for causality assessment.21,22 This method of causality assessment is the most widely used method for the scoring of the probability of drug-induced liver injury by researchers and health authorities. Attempts have been made to develop a simpler scale for suspected drug-induced liver injury,23 but the International Consensus Criteria were found to be superior to these simplified criteria in a comparison of these 2 scales.24 Our group has recently analyzed suspected drug-induced liver injury associated with a fatal outcome in Sweden from 1966 to 2002.25 When the International Consensus Criteria for causality assessment21,22 were applied, 31% of reports did not fulfil these criteria for a likely relationship, and these cases had to be excluded.25 On the other hand, it is widely recognised that adverse drug reactions are significantly underreported. It is generally considered that only 10% of real cases are ultimately reported2,26 (see "Incidence of drug-induced liver injury"). The reports of fatal drug-induced liver injury might therefore only be the tip of the iceberg. However, it is possible that more severe adverse drug reactions tend be reported more frequently, although this has not been assessed in a well-designed study. According to the International Consensus Criteria, liver injury caused by drugs is classified as cholestatic, hepatocellular, or mixed-pattern.21,22 These different types of drug reactions have been shown to have different clinical features and a different prognosis.6,14,15 Although hypersensitivity features with rash, fever, and eosinophilia have been considered to be the hallmark of drug-induced liver injury and to support a role for a drug etiology of the hepatic adverse reaction, these clinical signs and symptoms are only observed in a minority of patients with a high probability of drug-induced liver injury.15,27 In two prospective studies of drug-induced liver injury, hypersensitivity features were only observed in 20% to 25% of cases.15,27 The true incidence of hepatic adverse drug reactions has until recently been obscure. However, a careful prospective survey of drug-induced liver injury in the general population in France has been reported.26 The spontaneous reporting to the French regulatory authorities was only, at the most, one sixteenth of the true incidence detected by the methods used in this prospective study.26 Spontaneous reporting was found to be at least twice as high in Sweden compared with figures from France.28,29 However, except within clinical trials that give reliable information about development of abnormal liver test results, great uncertainty exists about the occurrence of drug-induced liver injury associated with the clinical use of drugs.30 The incidence of drug-induced liver injury in a prospective survey in a French community was approximately 14 per 100,000 inhabitants.26 Because these cases were detected in individuals with symptoms, this incidence is more likely to be an underestimation than an overestimation of drug-induced liver injury–associated cases. Most other studies on the incidence of drug-induced liver injury have been retrospective cohort studies with uncertain completeness of exclusion of nondrug causes.31,32,33,34,35,36,37 In a case-control study, risks for drug-induced liver injury ranged from 1 per 100,000 prescriptions for ampicillin and diclofenac to 14 cases per 100,000 prescriptions for erythromycin.33 The impact of drug-induced liver injury in a clinical hepatology practice has been studied in a retrospective study from a tertiary referral center in Michigan. In this study only 32 cases of drug-induced liver injury cases were found among 4039 hepatology referrals, and with exclusion of fulminant drug-induced liver injury–associated hepatitis, this represented 0.8% of all hepatology patients.38 Among patients seen in an outpatient hepatology clinic for the first time during a 10-year period in Sweden, a drug-induced liver injury diagnosis was made in 6% of all cases, in 3% of patients referred for evaluations of liver test results, and in 3% of patients undergoing follow-up after hospitalization.39 In consecutive patients hospitalized in the United Kingdom and evaluated by laboratory test results, 13 (8.8%) drug-induced liver injury cases were observed in 147 patients with liver enzyme elevations among 1964 admissions, giving an incidence of 0.7% of inpatients.40 In a recent study from Zürich, of 4209 cases at risk of drug-induced liver injury (on the basis of the International Consensus Criteria) among medical inpatients, 88 cases of drug-induced liver injury were found, but 31 of these did not have documented normal liver test results at admission and thus were considered to represent prevalent cases.16 The remaining 57 suspected cases of drug-induced liver injury gave a crude incidence of 1.4%, but liver injury was not mentioned among diagnoses or in the physician's discharge letter in 52% to 68% of cases.16 In this study, neither polymorbidity nor polypharmacy predisposed to drug-induced liver injury.16 In contrast, a population-based case-control study using the United Kingdom–based General Practice Research Database demonstrated that the combination of 2 or more hepatotoxic drugs increased the risk for drug-induced liver injury by a factor of 6.41 Predicting the susceptibility for drug-induced liver injury in an individual patient is almost impossible. The characteristics of drug-induced liver injury obviously depend on the drug involved, its elimination routes, and acquired and genetic factors.6,42 Deficiencies in certain cytochrome P450 enzymes have been reported in susceptible individuals, and formation of reactive metabolites in these individuals can provoke either immune or idiosyncratic reactions, or both.6,43 Genetic polymorphism has been demonstrated in immune processes of certain interleukins, favoring pathologic immune reactions in patients with previous diclofenac hepatotoxicity.44 Amoxicillin and clavulanic acid–induced hepatotoxicity has been associated with certain human leukocyte antigen (HLA) haplotypes.45,46 Increasing age has been shown to increase the risk for halothane-, isoniazid-, nitrofurantoin-, and floxacillin (INN, flucloxacillin)–associated hepatitis.6,42,47,48 Sex has also been shown to be of importance. Women are more susceptible to drug-induced liver injury associated with floxacillin,42,48 isoniazid,6 nitrofurantoin,6 and erythromycin,41 whereas men are more at risk of azathioprine-induced liver injury.47 In general, drug-induced liver injury has been reported to occur more often in female than in male patients in most5,14,26,49 but not all studies.15 However, Sgro et al26 observed similar drug-induced liver injury incidence rates in the two sexes until age 49 years, whereas drug-induced liver injury became more than twice as high in women after that age, suggesting an increased risk of drug-induced liver injury associated with menopause. Because most idiosyncratic drug reactions are probably genetically determined, the general assumption among hepatologists has been that most drugs with drug-induced liver injury–associated potential can be used safely in patients with an underlying liver disease.6,50,51 However, if patients with chronic liver disease happen to have an underlying genetic defect that increases the risk of drug-induced liver injury, a liver injury might be expected to cause more severe morbidity. Interestingly, Chalasani et al52 found that patients with elevated baseline levels of aminotransferases were not at an increased risk for the development of hepatotoxicity while receiving treatment with statins compared with a control group with preexisting liver disease, probably mostly consisting of nonalcoholic fatty liver disease. Similarly, in a study of the clinical characteristics and outcomes of patients with disulfiram-associated liver injury, we observed no difference in the incidence of death or liver transplantation or in the degree of abnormality of the liver test results related to pretreatment enzyme levels.53 Thus there was no difference in outcome in patients with normal baseline liver enzyme levels in comparison with patients with abnormal liver test results before the start of disulfiram therapy.53 This finding supports the view that with few exceptions54 there is currently no evidence that patients with an underlying liver injury are more susceptible to drug-induced liver injury.6,50,51,52 Hepatotoxicity induced by certain drugs in patients with human immunodeficiency virus has been reported to be more common in patients coinfected with chronic hepatitis C virus and B virus.55,56 However, as has been pointed out,51 these studies have not included a control group with human immunodeficiency virus together with either hepatitis C virus or hepatitis B virus, or both, not treated with the suspected drug.55,56 Limited data exist on long-term follow-up of patients who have had clinically significant drug-induced liver injury. Recently, 2 studies have been published with a short follow-up in 2 large cohorts of patients with suspected drug-induced liver injury.14,15 Furthermore, the occurrence of drug-induced liver injury associated with fatal outcome in Sweden has been reported from our group,25 as well as in reports of suspected hepatic adverse drug reactions with fatal outcome received by the World Health Organization Collaborating Center for International Drug Monitoring.57 In general, the outcome of severe drug-induced liver injury has been considered to be an all-or-none phenomenon. Thus the liver injury can lead to death or liver transplantation or total recovery. The prognosis for patients with acute liver failure caused by idiosyncratic drug reactions is usually poor, with approximately 60% to 80% mortality without liver transplantation.11,12,13 Although the general wisdom among hepatologists is that if the patient with drug-induced liver injury recovers clinically and biochemically without liver transplantation, a total recovery is to be expected, surprisingly little information exists on the long-term outcome of patients with drug-induced liver injury. The only published study on long-term outcome of drug-induced liver injury is a retrospective study in patients identified by histologic database.18 A total of 13 of 33 patients with a median of 5 years' follow-up had persistent significant abnormalities in either liver test results or imaging studies, or both.18 However, whether these abnormalities lead to clinically significant liver morbidity or mortality is unclear, and more information is needed on potential development of chronic liver injury caused by drugs and its possible consequences. The hepatocellular type of drug-induced liver injury is considered to have greater risk to result in acute liver failure than the cholestatic and mixed types of drug-induced liver injury.6 According to Zimmerman,6 the immediate prognosis of cholestatic drug-induced liver injury is generally good, and deaths usually result from the underlying disease or other components of the cholestatic syndrome rather than the hepatic injury. However, several studies have shown that the cholestatic type of drug-induced liver injury does not have such a good prognosis as originally considered.6 In a recent study acute serious liver disease associated with drugs had similar case fatality rates in those with hepatocellular and cholestatic types of liver disease.27 Among patients with suspected drug-induced liver injury and concomitant jaundice, the hepatocellular type of jaundice carried a 12.7% mortality, but patients with cholestatic and mixed liver injury also had significant mortality with 7.8% and 2.4% mortality, respectively.14 Furthermore, in a prospective study in patients with suspected drug-induced liver injury (with and without jaundice), patients with the hepatocellular type of drug-induced liver injury had a 10% mortality and those with the cholestatic type of drug-induced liver injury and mixed pattern had mortality of 6% and 2%, respectively.15 Chronic primary biliary cirrhosis–like drug-induced liver injury has also been described,20,58,59 and a vanishing bile duct syndrome with a poor prognosis resulting from drug-induced liver injury, although rare, is well recognized.60 A Swedish study of fatal reactions to drugs in general (not only caused by drug-induced liver injury) revealed a remarkably constant fatal reaction frequency during the period of 1966 to 1975.61 Hepatic adverse reactions accounted for 8.3% of all fatal reactions to drugs.61 In a recent study from our group we found that fatal hepatic adverse reactions accounted for 7.6% of all fatal reactions to drugs,25 which is close to the early figures from Sweden and remarkably similar to a 7.4% mortality from hepatic adverse drug reactions out of all fatal drug reactions in New Zealand.4 In our recent study the frequency of drug-induced liver injury associated with fatal outcome out of all hepatic adverse reactions was also remarkably similar during the whole study period. Even though the number of different drugs on the market has increased during the last decades, there does not seem to be a tendency to an increased mortality from hepatic adverse reactions. We observed a total of 2.2% of fatalities among the reported drug-induced liver injuries during the whole study period. This proportion of cases with fatal outcomes varied from 2.9% from 1973 to 1982, to 1.7% during 1983 to 1992, and to 2.1% in the last decade.25 These figures can be compared with a 4.7% frequency of fatal hepatic adverse reactions in Denmark from 1978 to 19875 and a 3.3% frequency in New Zealand during 1988 to 1994.4 Because most of the knowledge of the severity of drug-induced liver injury comes from case reports and case series,6 the prognosis of patients with drug-induced liver injury has until recently been obscure. Hyman Zimmerman, the legendary pioneer researcher in the field of drug-induced liver injury, observed that the combination of hepatocellular injury (high aminotransferase levels) and jaundice induced by a drug was associated with a poor prognosis with a fatality rate of 10% to 50% for the different drugs involved (so-called Hy's rule).6 The rule states that if drug-induced hepatocellular injury and jaundice occur at the same time without biliary obstruction, mortality of at least 10% can be expected.6,62 Hy's rule, defined as a drug-induced liver injury with a serum ALT level greater than or equal to 3 times the ULN plus a serum bilirubin level greater than or equal to 2 times the ULN, has been advocated by the US Food and Drug Administration for use in the assessment of the hepatotoxicity of newly developed drugs.63,64 According to the National Institutes of Health a new drug should be stopped in a person with previously normal liver test results if AST and ALT are more than 3 times the ULN and bilirubin is more than twice the ULN.63 Concomitant jaundice and hepatocellular injury observed in clinical trials of a new drug have been considered to raise serious concerns about safety in the postmarketing phase, when a much larger number of patients are exposed to the drug.63 Although Hy's rule is still used by the Food and Drug Administration in the assessment of the hepatotoxicity of different drugs, validation of the rule has long been requested because information about its sensitivity and specificity has until recently been largely lacking.63 We have recently looked for prognostic markers in patients with the hepatocellular type of drug-induced liver disease with concomitant jaundice (bilirubin level greater than or equal to 2 times the ULN) and found a mortality or transplantation rate of 9.2%.14 Similarly, Andrade et al15 found 11.7% mortality among patients with suspected drug-induced liver injury and concomitant jaundice. These 2 studies show remarkably similar results, and both seem to validate Hy's rule and convincingly show that hepatocellular injury with jaundice caused by drugs is a serious entity. These patients should be carefully monitored and often hospitalized and liver transplantation considered before severe encephalopathy develops. In multivariate analysis, bilirubin, age, and AST (but not ALT) were found to be independent factors associated with death or liver transplantation.14 The predictive effect of greater elevations of ALT, AST, or bilirubin levels did not increase the diagnostic accuracy of hepatocellular injury, whereas increasing the ULN of bilirubin to 5 increased the positive predictive value from 33% to 50% with an unchanged negative predictive value.14 Interestingly, the mortality rate among patients fulfilling Hy's rule because of different drugs was highly variable. The highest mortality was observed in patients with halothane-induced hepatitis (40%), whereas among 32 patients with hepatotoxicity caused by erythromycin, all survived the liver injury.14 The good prognosis in erythromycin liver injury is probably a result of milder liver injury because these patients, even though they fulfilled the Hy's rule criteria, had significantly lower bilirubin and aminotransferase levels and were also significantly younger than the nonerythromycin cases.14 To increase the knowledge in the field of drug-induced liver injury, the National Institutes of Health has recently sponsored a Drug-induced Liver Injury Network to promote research on the causes and the pathophysiologic characteristics of drug-induced liver injury, consisting of 6 university hospitals in the United States.65 In summary, although information about the prognosis of drug-induced liver injury has increased in recent years, important steps need to be taken in the research of drug-induced liver injury to improve its diagnostic accuracy. Furthermore, more studies are needed to determine its long-term prognosis and to elucidate the underlying pathophysiologic mechanisms and ideally, in the future, to identify individuals at risk for the development of drug-induced liver injury. I thank Professor Rolf Olsson for evaluation of the manuscript and for introducing me into the exciting field of drug-induced liver injury. The author has no conflict of interest.