Introduction

Because the liver is the major site of drug metabolism, it is also the major target of drug-induced injury. Despite rigorous preclinical and clinical toxicologic studies and safety analyses in clinical trials, the frequency of drug hepatotoxicity has remained relatively unchanged during the years, and drug-induced liver injury (DILI) is still the main reason for removal of a drug from the market. Furthermore, adverse chemical reactions are not confined to pharmaceutical drugs (i.e., the drug itself and its excipients used for classic therapeutic purposes): dietary supplements and herbal medicines, often used as self-medications, also represent potential hepatotoxins. Various environmental toxins and recreational drugs, such as alcohol, illicit drugs (e.g., cocaine, heroin, ecstasy), criminal poisons, and industrial toxins (e.g., natural toxins, mushrooms, industrial chemicals, pesticides), can also give rise to hepatotoxicity. The list of putative offending drugs and toxins is extremely long and evolves with time. Although all families of therapeutic agents can potentially be involved, a small group of drugs represent the most frequently incriminated molecules (Box 48.1). The circumstances of exposure to these various forms of liver toxins are listed in Box 48.2, with the exception of alcohol injury, which is discussed in Chapter 49.

Drug hepatotoxicity can be classified as either intrinsic or idiosyncratic. Intrinsic hepatotoxicity is predictable, dose dependent, and often characteristic of a particular agent when consumed in large quantities. Examples are ingestion of acetaminophen (paracetamol) and exposure to carbon tetrachloride or chloroform. Hepatotoxicity occurs in most of those who are exposed and starts shortly after some threshold for toxicity is reached. The mechanism of intrinsic injury can be direct, through damage to cells and organelles, or indirect, through conversion of a xenobiotic into an active toxin or through an immune-mediated mechanism. Idiosyncratic hepatotoxicity, by far the more frequent form of hepatotoxicity, involves unpredictable reactions that occur without warning and are unrelated to dose; they occur in particular hosts, depending on individual genetic variations in the metabolism of drugs and on environmental factors. Because the formation of reactive metabolites is a frequent mechanism of idiosyncratic reactions, the hepatotoxicity is highly dependent on the metabolic capacity of the host. In idiosyncratic toxicity, variable latency periods can be observed, from a few days to more than a year.

Diagnosis of a toxic liver injury or DILI is challenging, and it is often a diagnosis of exclusion and probability. Assessment of causality is difficult because almost all drugs and toxins are potentially hepatotoxic because of frequent idiosyncrasies and unpredictable toxicity, whereas predictable and dose-dependent liver toxins are rare. From a clinical or a pathologic point of view, any pattern of hepatic injury may be encountered and may mimic other liver diseases, and a single drug can induce different lesions in different patients. The clinical presentation is usually acute and largely reversible, but chronic disease can occur. The time of onset of liver dysfunction varies depending on the drug and the patient and can be long after the first ingestion of drug. Severe cases can occur and include mainly fulminant hepatitis: DILI is one of the major causes of acute liver failure with viral hepatitis. Significant fibrosis and cirrhosis may develop, and even hepatic tumors. In addition, hepatocytes are usually the target in DILI, but some drugs may target endothelial cells, cholangiocytes, hepatic stellate cells, and/or Kupffer cells.

The Council for International Organizations of Medical Sciences (CIOMS) has proposed consensus criteria for terminology in DILI, based on biologic tests, chronology, and availability of liver biopsy. Six categories of DILI can be defined in this system: hepatocellular injury, cholestatic injury, mixed injury, acute injury, chronic injury, and chronic liver disease (Table 48.1).

The histologic examination of a liver biopsy specimen is not always performed, either because the toxic episode may quickly resolve or because the information provided is disappointing. Indeed, specific lesions are rare in this field. However, when analyzed by a specialist, the liver biopsy can often provide useful information for positive and differential diagnosis of DILI (Box 48.3). The lesions can be classified as to pattern of injury, a topic that is discussed in detail later in this chapter.

DILI should be included in the differential diagnosis in cases with any hepatic laboratory abnormalities or hepatic dysfunction, but the assessment of causality of a drug in liver disease is often difficult. Discussion between clinicians and pathologists is especially important in this field of hepatology.

Current preclinical tests for hepatotoxicity are inadequate, reflecting our limited understanding of the mechanisms of drug toxicity. In particular, “hypersensitivity” and “idiosyncratic” reactions remain poorly understood and probably affect individuals possessing a rare combination of genetic and nongenetic factors that lead to drug toxicity in a given environmental setting. Many meetings on this topic have been organized by the Center for Drug Evaluation and Research (CDER) of the U.S. Food and Drug Administration (FDA) since 2001, and collection of data in regional or national registries (e.g., Spain, France, United States) or databanks has been encouraged. The goal of the Drug-Induced Liver Injury Network (DILIN; https://dilin.dcri.duke.edu/) is to collect clinical data, genomic DNA, and liver tissues from patients who have experienced idiosyncratic drug reactions, in an effort to determine pathogeny.¹ Recent studies from these groups have focused on the specific group of DILI cases with features of autoimmune hepatitis (AIH),² which could explain recurrent DILI with different drugs in a same patient.³

Despite much effort, hepatotoxicity remains a problem for many existing drugs, as well as those in development. This has a major economic impact because hepatotoxicity is the most frequent cause of postmarketing withdrawal of new medications.

This chapter reviews the main clinical features of toxin- and drug-induced disorders and describes the main pathologic patterns attributed to drugs and toxins. In the last part, some frequently prescribed hepatotoxic drugs are described more extensively to illustrate the variety of clinicopathologic presentations of liver injury.

For an extensive and detailed description of hepatotoxicity produced by a larger number of individual drugs, the reader is referred to a variety of specialized publications.^4–21 Synthetic and actualized data can be obtained from specialized Web sites such as Hepatox, which was developed in France by Dr. Biour and is hosted on the Web site of the Association Française pour l’Etude du Foie (http://www.afef.asso.fr/liens/Hepatox/index.phtml), or from the Uppsala Monitoring Centre of the World Health Organization (WHO) database (http://who-umc.org).

Epidemiology

DILI accounts for approximately 10% of cases of acute hepatitis in adults and for more than 40% of cases in individuals older than 50 years.¹³ In various series, it has accounted for 10% to 20% of cases of fulminant and subfulminant hepatitis²² and for 2% to 5% of patients hospitalized for jaundice. The risk of a fulminant course is much greater for DILI (20%) than for viral acute hepatitis (1%). On the other hand, drugs are less often incriminated in chronic hepatitis or cirrhosis (<1% of cases).

However, it is probable that the real incidence of DILI is much higher because of unrecognized and benign presentations. In children, DILI is less frequent, but it is also an underrecognized cause of pediatric liver disease, and large series are rare. Children could represent 5% to 8.7% of DILI patients.²³

A French study found an annual incidence of DILI of almost 14 cases per 100,000 population, a rate 16 times higher than that based on spontaneous reports.²⁴ The most common causes of DILI are the analgesic drug acetaminophen and several antiinfectious agents such as antibiotics from different families (e.g., amoxicillin/clavulanic acid [AMC], erythromycin, minocycline), as well as antituberculosis drugs (especially isoniazid [INH]), psychotropic agents (e.g., chlorpromazine), anticonvulsants (e.g., valproic acid), anesthetics, oral contraceptives, lipid-lowering agents, antiinflammatory agents (e.g. diclofenac, disulfiram), and cardiovascular agents (e.g., amiodarone), as shown in large series^16,25,26 (see Box 48.1). In children, apart from acetaminophen, antimicrobial and central nervous system agents are the most commonly implicated drug classes, representing 50% and 40% of cases, respectively, as demonstrated in the recent DILIN series.²⁷ Another pediatric series from India emphasizes the role of antituberculosis drugs (e.g., INH, rifampicin [RFP], ethambutol) and anticonvulsants (phenytoin and carbamazepine),²³ pointing out the geographic specificity for certain drug toxicities. In this Indian pediatric population, hypersensitivity features such as skin rashes, eosinophilia, fever, lymphadenopathy, and Stevens-Johnson syndrome were frequently seen (41%) and were associated with a better outcome.

The total number of drugs liable to be toxic to the liver exceeds 1100, and this long list must be frequently updated.²⁸ The highest frequency of hepatotoxicity for marketed drugs has been approximately 1% (for tacrine), but for most drugs, the risk is low (1/10,000 to 1/100,000) or extremely low (1/100,000 to 1/1,000,000 for antihistaminic compounds or penicillin). However, evaluation of the accurate incidence and risk factors, as well as assessment of the causality of one or several toxic agents, remains a major problem in DILI.²⁹

In addition, overdoses of certain drugs are well known to be extremely toxic, not only in the context of a therapeutic misadventure (or with repeated doses, particularly in cases of excessive alcohol ingestion) but as a method of suicide. In the latter instance, acetaminophen is the drug most frequently used for suicidal overdose among adolescents and young adult women in the United States and Great Britain. When given in therapeutic doses for a period of 14 days, acetaminophen produced significant asymptomatic elevations in alanine aminotransferase (ALT) levels among healthy volunteers, suggesting that subclinical injury may be more common than previously believed. There was also a much lower incidence of acetaminophen toxicity as a cause of acute liver failure in children compared with adults, with almost half of all cases being indeterminate in origin.³⁰ Acetaminophen toxicity is reviewed more extensively later in this chapter.

The potential hepatotoxicity of herbal remedies commonly used for self-medication (i.e., alternative or “natural” treatments) and of other botanicals (e.g., the well-known mushroom poisoning from Amanita phalloides) should always be considered in the evaluation of pathology specimens. The list of confirmed or suspected hepatotoxic herbal components (e.g., Chinese herbs, germander³¹) is long, and the full extent of their toxicity remains unclear.³² Many alimentary supplements, including vitamins, minerals, and botanical extracts, are also recognized as possible cause of DILI, as reviewed by Navarro in 2009.²¹ For instance, many Herbalife products, used for nutrition or energy or to reduce stress, have been shown to commonly induce cytolysis, cholestasis, and even, in rare cases, acute liver failure.

Increased consumption of illicit drugs such as heroin, cocaine, and ecstasy, regardless of the route of administration (intranasally, intravenously, or by smoking), has increased the number of cases of hepatotoxicity³³ and potentiated other factors of liver disease. For example, daily cannabis smoking is significantly associated with progression of fibrosis in patients with chronic hepatitis C virus (HCV) infection.³⁴

Frequently, hepatotoxicity is further potentiated by the use of other drugs in combination or by alcohol intake. Therefore, when prescribing a potentially hepatotoxic drug in a patient, it is particularly important to be aware of all additional risk factors for the liver in that patient, such as alcohol, diabetes, obesity, and chronic viral hepatitis. Otherwise, the risk of DILI will increase, and also the severity of the other liver diseases can worsen. During recent decades, for example, the evolving epidemic of nonalcoholic fatty liver disease caused by metabolic syndrome has potentiated the hepatotoxic properties of certain drugs such as methotrexate—and vice versa.³⁵

Furthermore, a drug may be beneficial in the short term but harmful in the long term. For example, in individuals with human immunodeficiency virus (HIV) infection who also have hepatitis B virus (HBV) or HCV coinfection, alcohol abuse, or other hepatic risk factors, prolonged therapy with didanosine may induce chronic liver disease and may cause severe liver complications, such as variceal bleeding and portal thrombosis.³⁶

Clinical Assessment and Prediction of Hepatotoxicity

Side effects attributed to drugs and toxins in the liver are numerous and variable. Although a slight acute hepatocellular or cholestatic/mixed hepatitis is the most common presentation, DILI can mimic all forms of acute or chronic hepatitis as well as biliary or vascular hepatopathy. In addition, some liver tumors (e.g., hepatocellular adenoma, angiosarcoma) may be attributed to the long-term exposure to various drugs or toxins. Therefore, hepatologists and pathologists should consider possible drug-induced hepatotoxicity in the differential diagnosis of virtually any type of liver injury. Abnormal liver test results and nonspecific clinical signs may be present, including malaise, fatigue, abdominal discomfort, appetite loss, splenomegaly, icterus, or symptoms of acute liver failure. Signs of immunoallergic reaction, such as fever, rash, arthralgia, and eosinophilia, may be encountered but are infrequent and nonspecific. Chronologic and clinical diagnostic criteria that are useful in making the diagnosis of DILI are provided in Box 48.4. Chronologic criteria, although usually considered essential, require an accurate clinical history, which is not always possible. Many other difficulties are frequently encountered in making the diagnosis, as shown in Box 48.5. In some instances, clinical criteria³⁷ and laboratory data help to eliminate other potential causes of hepatopathy and point to drug toxicity as the only possible differential diagnosis. Additional tests, such as seric dosage of the drug, can be helpful in some cases. Liver biopsy is not mandatory in the clinical survey, but it may provide substantial information regarding the positive and differential diagnosis of DILI (see later discussion), particularly if liver disease persists, provided that pertinent clinical information is provided and the pathologist is experienced.

Because of variability in both drug exposure and patient susceptibility, prediction of hepatotoxicity depends heavily on the specific type of drug used and patient characteristics (Box 48.6). Some acquired factors may enhance susceptibility to one type of drug but not another. Furthermore, various genetic factors, such as deficiency in certain isoforms of cytochrome P450 (CYP450) or in other enzymatic and metabolic pathways, may contribute to drug hepatotoxicity¹³ (Table 48.2). In most instances, potentially fatal idiosyncratic reactions cannot be reliably predicted. For example, troglitazone, which was an approved drug for the treatment of diabetes mellitus, is an idiosyncratic, directly hepatotoxic drug that led to an unacceptable rate of acute hepatic failure and was subsequently removed from the market.^38–40

Causality can be assessed with more certainty if a clear chronologic link can be demonstrated between drug intake and onset of the hepatotoxic event (Box 48.7). Components of the drug signature (e.g., pattern of liver test abnormalities, duration of latency before symptomatic presentation, presence or absence of immune-mediated hypersensitivity, response to drug withdrawal), in conjunction with certain genetic and environmental risk factors, can be formulated into a clinically based scoring system that is predictive of the likelihood of liver injury. The best validated scoring system that takes into account all of these parameters is the CIOMS/Roussel-Uclaf Causality Assessment Method (RUCAM), which nonetheless has certain imperfections.⁴¹ The Naranjo Adverse Drug Reactions Probability Scale (NADRPS) is another simple system based on similar items. Both systems produce a numerical score, indicating that the diagnosis of DILI is definite (or probable), possible, or unlikely. However, a review of 61 case reports of DILI in the PubMed database of the National Institutes of Health over the last decade indicates that in current practice, these scores are used in no more than 25% of published cases.⁴²

Although it is difficult to provide definitive proof of responsibility for a particular offending drug, and because readministration is ill advised, return to normal liver function after withdrawal of the drug is usually good supportive evidence of drug-induced toxicity. Additional tests may be performed on peripheral blood to identify a single causative agent. These may include drug dosing (e.g., acetaminophen), the double-locus sequence typing (DLST) method (which measures the patient’s lymphoproliferative response to growing doses of the suspected causative drug), or the leucocyte migration test (LMT). However, these tests are not simple to perform and not feasible in routine practice.

Prevention of Drug Hepatotoxicity

Assessment of toxicity in human hosts is performed before and after marketing of all drugs. During the early stages of drug development, preclinical studies in animals are mainly useful to detect dose-related predictable hepatotoxicity. Phase I safety studies in human volunteers test toxicity in few patients, after which more patients are exposed during efficacy testing in controlled clinical trials. However, almost 3000 patients need to be included to demonstrate a 1/1000 incidence of DILI, and many drugs that may induce idiosyncratic hepatotoxicity escape detection during preclinical safety assessment and clinical trials.

From a pharmaceutical research perspective, metabolite profiling is essential to rational drug design.⁴³ This issue is addressed, at present, to eliminate molecules that are prone to metabolic bioactivation, based on the concept that formation of electrophilic metabolites triggers covalent protein modifications and subsequent organ toxicity. A cell-based approach for testing of cell viability, mitochondrial impairment, biliary transport, and CYP450 inhibition in the presence of a drug has become useful for the evaluation of putative hepatotoxicity.^44,45 The role of mitochondria in DILI, which may be altered through direct toxicity or through immune reaction, seems central, so new drug molecules should also be screened for possible mitochondrial effects. Although such an in vitro approach is pragmatic, it has its limitations, because a linear correlation does not exist between toxicity and extent of bioactivation.

After marketing, surveillance and voluntary reporting of cases is necessary. Routine monitoring of liver enzymes does not seem useful to prevent clinically significant hepatotoxicity in the general population, but it may be interesting in high-risk patients who are taking well-known hepatotoxic drugs. At present, a few simple rules can be applied to help prevent drug hepatotoxicity (Box 48.8). In the future, advances in proteomics, metabolomics, genomics, and bioinformatics may pave the way to “personalized” pharmacotherapy in which the beneficial effect of a drug in an individual is maximized and the toxicity risk minimized.^46–48