Introduction

Hepatotoxicity is damage caused by exposure to a drug or non-pharmacological agents. Risk factors include idiosyncrasy, age, gender, alcohol consumption, smoking, concomitant use of other drugs, previous or underlying liver disease, genetic and environmental. ^1-3 Although most lipophilic drugs can cause hepatotoxicity, ⁴ antibiotics, nonsteroidal anti-inflammatory drugs (NSAIDs) and anticonvulsants are the pharmacological groups which are the most frequent causes. ^1,5-9. Among drugs administered intravenously, antibiotics and drugs to treat neoplasia are the groups most associated with liver toxicity.¹⁰

Hepatotoxicity can be classified into intrinsic reactions and idiosyncratic reactions:

Intrinsic hepatotoxicity is less common than idiosyncratic hepatotoxicity^12,18-20. Liver histology is ideal for defining patterns of liver toxicity, but in clinical practice the majority of hepatotoxic damage is classified according to biochemical tests ²¹. According to the international consensus of the Council for International Organizations of Medical Sciences (CIOMS), liver damage is present when liver enzymes are over two times the upper limit of normal (ULN). On the other hand, types of injuries are classified in ^12,22,23:

Hepatotoxicity is related to mitochondrial dysfunction, inhibition of cellular respiration or alteration in β oxidation of fatty acids^26,27. These result in apoptosis, necrosis, autophagy and, therefore, cell death^28,29. The main clinical-pathological manifestations of hepatotoxicity and its histological findings are:

These manifestations are accompanied by nonspecific signs and symptoms such as fever, fatigue, nausea, abdominal pain, jaundice, dark urine, pruritus, ascites, encephalopathy and increased transaminases^16,32,33.

Although some 1,100 drugs, excluding substances of abuse and natural products, have been associated with hepatotoxicity ¹⁹, identification of this adverse event is a complex process. Therefore, a meticulous investigation is required, aimed at identifying any substance and ruling out other causes of liver disease^3,8,34. In addition, liver biopsy is fundamental for identifying hepatotoxicity ³⁵. The chronological relationship between exposure to the suspect agent and the hepatotoxic reaction is key. To establish the likelihood that a drug is associated with hepatotoxicity, clinical scales such as the Roussel Uclaf Causality Assessment Method (RUCAM) and the Maria & Victorino (M&V) clinical scale have been developed. It is considered that the RUCAM scale’s content and criterion validity make it most appropriate and that it generates results compatible with medical judgment and expert opinion on hepatotoxicity. Nevertheless, due to its high cost of application, its usefulness in clinical practice is limited ^36-38.

In the absence of a specific pharmacotherapy, treatment of hepatotoxicity is based on suspension of the suspect medication, treatment of symptoms and follow-up laboratory tests³⁹. However, the use of N-acetylcysteine ​​as an antidote for acetaminophen toxicity and hepatotoxicity due to phenytoin and carbamazepine, and the use of carnitine to treat valproic acid toxicity are exceptions⁴⁰.

An updated list of hepatotoxic drugs and associated factors could help optimize identification and prevention of this adverse event. Therefore, the objectives of this review were to prepare an updated list of drugs associated with hepatotoxicity and identify, according to scientific evidence, the drugs most likely to cause hepatotoxicity. In addition, this review will systematize and specify key information such as type of injury, probability of occurrence, pathophysiological mechanisms, clinical and pathological manifestations, variation in liver enzyme levels, intrinsic or idiosyncratic reaction, risk factors and clinical outcomes.

Method

Information Analysis

Two independent reviewers determined eligibility of articles and extracted information from them while discrepancies between them were resolved by a third reviewer. The title, author, year of publication, type of study, related pharmacological group and compliance with inclusion criteria for each of the references found was recorded in a database in Excel 2010 for Windows®. In addition, the pharmacological group, ATC code (Anatomical, Therapeutic, Chemical classification), probability of occurrence of hepatotoxicity, type of injury, and probability of occurrence of that type of lesion were tabulated for each of the hepatotoxic drugs found. Mechanisms of hepatotoxicity, risk factors, clinical manifestations, management, outcome, measurements of liver enzymes and medication dosages were also recorded. Means and standard deviations were calculated for numerical data such as liver enzyme values ​​(aspartate aminotransferase [AST], ALT, FA and total bilirubin [TB]) and dosages ​​of drugs administered.

Assessment of Appearance of Hepatotoxicity and Type of Injury

Assessment of the appearance of hepatotoxicity and the type of injury was based on probability of occurrence. ⁴¹ Three categories were established according to the evidence found:

In cases of drugs for which several references and different types of study were available, articles with the highest level of evidence were used.

Results

The search identified 610 articles of which 402 met the inclusion criteria and were selected while 208 did not meet the inclusion criteria and were excluded. Forty-six other articles considered relevant for the review were included (Figure 1). A list of 181 drugs and 17 combined pharmaceutical forms or therapeutic regimens likely to cause hepatotoxicity was prepared. Six of these drugs (methotrexate, minocycline, vancomycin, everolimus, isoniazid, and tamoxifen) and one therapeutic regimen (isoniazid, rifampicin plus pyrazinamide) were classified as definite, 56 drugs and five combined pharmaceutical forms or therapeutic regimens were classified as probable, and 119 drugs and 11 combined dosage forms were classified as possible.

Figure 1 General results of the structured review: drugs that cause liver toxicity. 

The type of lesion caused by each drug was identified, and hepatocellular damage was found to be more common than cholestatic or mixed damage. Information found for each drug with definite probability that was tabulated included type of hepatotoxicity, type of lesion, appearance, mechanism of hepatotoxicity, risk factors, clinical manifestations and outcomes (Table 1). The drugs found were classified according to their pharmacological group and their ATC code to unify them. Drugs found to have probable probability are found in Table 2 and the drugs with probable probability are found in Table 3. Figures for liver enzymes and dosages found are in Table 4.

Table 1 Drugs with a probability of causing definite hepatotoxicity 

ALT: Alanine aminotransferase; ATC: Anatomical, Therapeutic, Chemical; DM2: diabetes mellitus type 2; HIV: human immunodeficiency virus.

Table 2 Drugs likely to cause probable hepatotoxicity 

Table 3 Medications likely to cause possible hepatotoxicity 

Table 4 Liver enzyme values ​​and dosages associated with hepatotoxic drugs 

ALT: Alanine aminotransferase; AST: aspartate aminotransferase; TB: Total bilirubin; SD: standard deviation; AP: alkaline phosphatase; avg.: average

Among antidiabetic agents, the probable probability of causing hepatotoxicity of acarbose ⁴² and troglitazone ^43,44 (withdrawn from the market) has been determined. Case reports indicate that it causes hepatocellular, cholestatic and mixed type lesions accompanied by jaundice, rashes, fevers, and other symptoms. The antiarrhythmic agents associated with probable hepatotoxicity were propafenone and amiodarone. ⁴⁵ There were more reported cases for amiodarone, and they were associated with elevated liver enzymes in 15% -55% of the patients. ⁴⁶ Patients improved upon suspension of medication, but there are reports of death associated with amiodarone⁴⁷. Antihypertensives such as enalapril increased liver enzyme levels and produced jaundice and structural changes in the liver confirmed by biopsies which led to transplantation and death⁴⁸. For methyldopa (probable), there were nine reported cases of idiosyncratic liver toxicity¹⁷. They had a pattern of hepatocellular injury, especially in women, manifested by jaundice, anorexia and nausea. In addition, liver biopsies identified necrosis and inflammatory infiltrates. ^49,50 Hepatocellular lesions accompanied by elevated liver enzymes, jaundice, fever and asthenia were found to be associated with atorvastatin and ezetimibe^51,52.

Propylthiouracil caused the death of one patient, affected women and girls, generated symptoms such as jaundice, pruritus and weight loss; necrosis, fibrosis, inflammatory infiltrate and ductopenia and was found in liver biopsies. Suspension of the drug improved the evolution of some patients⁵³. Four cases of increased liver enzyme values, weakness and jaundice were identified in patients taking methylprednisolone. Symptoms improved upon suspension of the medication⁵⁴.

Among the antibiotics, idiosyncratic reactions were identified in association with vancomycin ⁵⁵ and minocycline^17,33,55,56. Minocycline affected women between 16 and 57 years of age who had been diagnosed with autoimmune hepatitis. Rifampicin caused hepatocellular lesions and especially affected women^57,58. The following antibiotics were classified as probable causes of hepatotoxicity: nitrofurantoin (12% frequency of cases, idiosyncratic), ^59,60, flucoxacillin (11 cases, idiosyncratic)⁶¹, telithromycin (hepatocellular lesions with elevated transaminases and fever), ⁶² ciprofloxacin and trovafloxacin (withdrawal from the market). In general, the outcomes varied from favorable evolutions to liver transplantation and death of the patient.

Liver damage associated with the antifungal agents itraconazole, fluconazole and ketoconazole improved with suspension of the medications. ^63-66 Antiretroviral agents, especially reverse transcriptase inhibitors, nucleoside analogues and protease inhibitors, can cause dose-dependent hepatotoxicity. ⁶⁷ Cases reported with efavirenz and nevirapine had elevated transaminases and an incidence between 1% and 14%. ⁹ Coinfection with hepatitis B or C virus can increase the level of hepatotoxicity associated with antiretroviral treatments. ^68,69

Chemotherapy has increased life expectancy, but it can cause liver damage ranging from steatosis and steatohepatitis to cirrhosis. ^70,71 The probability of hepatotoxicity for tamoxifen, everolimus and methotrexate is definite. Medications such as flutamide, etoposide, imatinib, ipilimumab, oxaliplatin, temozolomide, thioguanine, glatiramer, azathioprine, and infliximab were classified as probable causes of hepatotoxicity.

NSAIDs were identified as an important group that can cause liver damage, mainly idiosyncratic, in cases of abuse or overdose. ^1,72,73,74 Risk factors identified included age, female gender, chronic alcohol consumption, concomitant drugs, underlying diseases, obesity, DM2 and insulin resistance. ⁷² Causative agents include diclofenac, lumiracoxib and nimesulide. Acetaminophen is widely recognized as an intrinsic hepatotoxic substance due to a metabolite that causes hepatic necrosis,. Fourteen case reports characterized by hepatocellular lesions were identified. When managed with N-acetylcysteine ​​and prednisone patients improved. ^75,76

Halothane was the general anesthetic most likely to cause liver toxicity. Genetic predisposition, repeated doses, obesity and advanced age were some risk factors. Women are more likely to suffer liver damage including hepatocellular lesions, increased liver enzymes, necrosis, fever, jaundice and fatigue. ^77-79

Among the anticonvulsants, valproic acid had the largest number of cases of hepatotoxicity (hepatocellular type) which manifested with elevated transaminases, abdominal pain, jaundice and anorexia. In addition, microvesicular and macrovesicular steatosis, necrosis and inflammatory infiltrate were identified in liver biopsies. This medication can cause liver damage in people under 30 years of age^80,81. Carbamazepine cases identified were mainly of the mixed type with the formation of granulomas^82,83. Lamotrigine generated cases of idiosyncratic hepatotoxicity that required liver transplantation^84,85.

We found 17 combined pharmaceutical forms or therapeutic regimens (drugs used simultaneously) capable of causing idiosyncratic hepatocellular damage. Among them is the combination isoniazid, rifampicin and pyrazinamide (definite probability). Hepatotoxicity manifested with elevated liver enzymes, abdominal pain, jaundice, asthenia, nausea, vomiting and necrosis and has been confirmed by liver biopsies. ^86-88 In the case of the combined pharmaceutical forms of antibiotics, such as trimethoprim/sulfamethoxazole and amoxicillin/clavulanic acid, cases of hepatotoxicity were identified as idiosyncratic and classified as probable. ⁵⁷ Hepatotoxicity occurred mainly in men and caused jaundice and pruritus. In some cases caused by amoxicillin/clavulanic acid, the outcome was liver transplantation or death. ⁸⁹ The antiretroviral regimen of ritonavir, indinavir, darunavir and fosamprenavir was associated with hepatocellular damage and necrosis. ^68,90 Case reports of antineoplastic agents 6-thioguanine, daunomycin, and cytosine arabinose used as part of the therapeutic regimen for myeloid leukemia in children have shown hepatomegaly, cirrhosis, and veno-occlusive disease. ⁹¹

Discussion

Antibiotics, antineoplastic agents and antituberculosis drugs were the groups of drugs identified as most likely to cause hepatotoxicity. This is in accordance with the results of previous reviews^5,23,33,57. Sufficient evidence was found for the capacities of two antibiotics, vancomycin and minocycline, to cause idiosyncratic hepatotoxicity and a type of hepatocellular lesion^{17,33,55,56,92}. They were assessed as definite. Among the other antibiotics, tetracycline was identified as being able to generate steatohepatitis, ⁹³ but the search did not yield enough information for its inclusion.

Case reports allowed identification of tamoxifen, everolimus and methotrexate as agents that can cause liver damage, with a definite probability. In the case of methotrexate, a previously published study reported increased liver enzymes, but did not identify associated methotrexate concentrations or the probability of causing hepatotoxicity. ⁹⁴ With the information found it was not possible to establish types of hepatotoxicity for these three drugs, but damage tends to be hepatocellular with elevated transaminases and outcomes that range from favorable to death.

In this review, no reports of specific cases of isoniazid hepatotoxicity were found, but some authors claim that this medication causes liver damage⁹⁵. Thus, it is important to note that the concomitant use of isoniazid, rifampicin and pyrazinamide was identified as the cause of idiosyncratic hepatotoxicity and hepatocellular damage followed by suspension of the regimen^86,87.

In this review, the likelihood for acetaminophen to cause liver toxicity was assessed as probable because only opinions of expert groups and case reports were identified. The absence of evidence from metaanalyses, systematic reviews and clinical trials did not allow it to be classified as a definite probability.

In the case of amiodarone, there are doubts about the type of hepatotoxicity it generates. Although information was found that supports idiosyncratic reactions, ^{(32, 96-98)} intrinsic reactions secondary to drug deposition in liver tissue have also been reported. ²³

There are medications that have been withdrawn from the world market or from certain countries because they are associated with the probability of causing hepatotoxicity. ⁹⁹ Some of those identified in this review are dextropropoxyphene, ketoconazole (still marketed in Colombia), nefazodone, propofol and sitaxentan.

Elevated liver enzymes were present in many of the case reports which suggests that it could be a marker for suspicion of drug hepatotoxicity. In this framework, ALT, AST, AP and TB were reported for some medications in some case reports. In addition, the available data for dosages administered doses could support the hypothesis of onset of hepatotixicty at therapeutic doses.

A limitation of this review is that it used only one database: PubMed/Medline. This may decrease the number of drugs identified as likely to cause liver toxicity and could influence assessment of the probabilities found for medications tabulated. To reduce the risk of information biases, a procedure proposed by Amariles et al. that is based on the probability of the occurrence of hepatotoxicity was used. ⁴¹ Three categories were established, definite, probable and possible, according to the types of published studies (level of evidence) for each drug. To decrease the confusion bias between and among drugs that were used at the same time, a difference is made between combined pharmaceutical forms (several drugs in a single pharmaceutical form) and therapeutic regimens.

Conclusions

We identified more than 180 drugs associated with hepatotoxicity. Of these, six have definite probabilities while most of the rest have possible probabilities. It is noteworthy that more than 50% of the drugs found are associated with idiosyncratic hepatotoxicity and that female sex is the main risk factor. The age range of people affected is broad. In addition, the elevation of liver enzymes, jaundice and fever the symptoms that occur most often. They can lead to hepatocellular lesions followed by liver necrosis. In most cases, adequate patient evolution occurs after identifying and suspending the causative agent. The consolidation of information on hepatotoxicity shows that several groups of drugs have greater evidence of being substances that cause liver toxicity. Nevertheless, precise understanding of the mechanisms that explain the development and onset of hepatotoxicity is scarce.