Clinically apparent liver injury from a Food and Drug Administration (FDA)-approved drug is a rare event. However a clinician is often asked to provide an opinion as to whether elevations in liver biochemical studies (especially elevations of aminotransferases, bilirubin, or alkaline phosphatase) may have been caused by a drug.

The problem of separating a drug-related elevation in liver studies from hepatic changes related to the underlying disease process may be challenging (and often impossible). Identification is especially difficult if the patient has underlying viral hepatitis, active alcohol-induced liver disease, obesity with possible nonalcoholic fatty liver disease, or a malignancy which may have involved the liver.

Almost all therapeutic drugs in current use have been suggested as a cause of some liver alterations, most often asymptomatic elevations of aminotransferases. Awareness of drug-induced liver injury is receiving increased attention. A few drugs with proven efficacy have been withdrawn from the market following post-release approval by the FDA when recognition of even a few cases of severe liver injury have been attributed to the drug. The economic consequences resulting from the removal of an approved drug from the market or even markedly restricting its use are tremendous as are the costs expended in the development of a new agent which fails late in the preapproval process. Drug-induced injuries that lead to clinically severe liver disease garner much attention even if only a few cases are recognized (or even suggested). A challenge is to improve the methods by which the risk of drug-related hepatotoxicity from a specific drug is assessed (and hopefully predicted) so that effective plans to minimize risk are in place. The major role of acetaminophen as the cause of acute liver failure is fully established.

It is well recognized that with many drugs, transient biochemical changes (often slight to moderate increases in aminotransferases) occur in patients in whom there are no symptoms or signs suggesting liver injury. In many, actually most, of these occurrences the usually minor elevations subside to a variable extent as the liver adapts to the newly introduced agent. The mechanisms leading to these adaptations remain obscure. Determining the threshold of tolerance (risk-benefit factor) for mild to moderate changes in biochemical tests caused by drugs that are effective in the treatment of serious illnesses is a difficult but important undertaking.

Hepatic injury that occurs rarely (<1 in 10,000) is designated as an idiosyncratic reaction which has classically been defined as a response or reaction that is host dependent and dose independent. These tenets have been challenged especially as regards dose. It has been convincingly shown that drugs administered in low doses (5–10 mg) are much less likely to cause liver injury than are agents which require larger doses to achieve a desired effect. The frequency of hepatotoxicity is influenced to some extent by age, patterns of use (short courses vs long-term administration), sex, ethnicity, and increasingly recognized genetic factors that govern metabolism of the drug and the immune responses to the drug or its metabolic products. Furthermore, transport of specific drugs into the hepatocyte from the blood and from the hepatocyte into bile is dependent on genetically influenced transporters which may be affected by a therapeutic drug. Evidence of clinically apparent severe liver injury from an approved drug is rare. Clinically significant events may not be detected in the preapproval development during which a relatively limited (several thousand) and often well-characterized number of patients are exposed. Evidence of liver injury may be recognized only after release when large numbers of patients with variable backgrounds (age, other drugs, alcohol use, weight, and nutritional states to name a few) have been treated.

A challenge for further research is to identify more precisely those factors that determine why an agent well tolerated by thousands of patients may cause a devastating injury in a few. The present focus on genetic variations in drug metabolism, immune responses (especially those involving the innate immune system), and transporters is receiving much attention.

Drug-induced liver disease is usually indistinguishable (clinically and histologically) from liver injury of other causes and diagnosis depends on the awareness of the possibility, suspicion, careful history, exclusion of other alternatives, and inquisitive persistence by the clinician.

Careful attention must be given to the time of onset, duration of therapy, and nature of the hepatic injury, as well as consideration of the roles of other drugs being concurrently taken that may enhance production or accumulation of reactive intermediates or interfere with drug transport. The increasing focus of drug metabolites eliciting the innate immune responses should eventually lead to more effective ways to assess risk and aid in diagnosis.

Most clinicians prescribe a relatively limited panel of drugs. It is essential that the clinician become familiar with the hepatic risk profile and patterns of manifestation of the agents he or she uses. Difficulties in establishing a diagnosis of drug-induced injury are compounded by the status of the underlying disease, concomitant illnesses, and polypharmacy.

Considerable supportive evidence implicating a drug as the cause of a liver injury is found if there is resolution of manifestations of liver injury (deceleration) following withdrawal. The problem in a patient who is receiving many drugs is in deciding which agent or agents is the likely culprit and should be removed when evidence of liver injury is found.

The profiles of frequency, types of reaction, and possible extent of injury for individual drugs are highly variable. Minimal biochemical abnormalities, especially minor elevations of serum aminotransferases (less than three times the upper limit of normal), do not necessarily indicate the clinical importance of liver injury or its likely occurrence. Patients with considerable elevations of aminotransferase levels accompanied by evidence of clinical jaundice are in a group at considerable risk of an adverse outcome.

The conceptual framework of differentiating categories of hepatic adverse reactions into those that are predominantly hepatocellular (elevated aminotransferase levels) and those that are predominantly cholestatic (elevated alkaline phosphatase levels) is interesting and possibly useful. With some drugs, there are elements of both patterns of injury, leading to a category of a mixed reaction.

Cholestatic disorders range from mild elevations of alkaline phosphatase levels found on routine biochemical testing to symptomatic cholestatic syndromes with jaundice and pruritus that may resemble primary biliary cirrhosis or primary sclerosing cholangitis. Hepatic steatosis (fatty liver) of both the microvesicular and macrovesicular types is frequently noted to some extent in liver biopsies and is a near-constant finding in obese and diabetic patients. Hepatic steatosis is a frequent cause of minimal biochemical changes. Drugs that cause mitochondrial injury may lead to microvesicular fat accumulation. A few drugs (eg, amiodarone) lead to an acquired phospholipidosis. Several drugs (eg, phenylbutazone and allopurinol) are established causes of hepatic granulomatous inflammation indistinguishable from those found in a variety of infections and sarcoidosis. Tumors induced or promoted by therapeutic drugs range from benign hepatic adenomas, associated with long-term use of oral contraceptives, to angiosarcomas, cholangiocarcinomas, and even hepatocellular carcinomas.

Several helpful observations regarding drug-induced liver disease include:

1. 
   

   Clinical manifestations of drug-induced hepatotoxicity are often clinically indistinguishable from those of liver disease caused by other etiologies.

   
   
2. 
   

   The risk of developing serious liver injury and even death from a drug-induced injury that is predominantly hepatocellular is considerably greater at least in the short term than that found in patients who have predominantly cholestatic injury.

   
   
3. 
   

   The appearance of clinically apparent liver disease in patients who has predominantly hepatocellular injury with clinical jaundice requires immediate attention.

Hepatic injury from a drug may have a somewhat specific “signature” as regards time of onset, type of injury, and propensity to develop severe disease. However, the signature may be blurred. Many drugs, known to cause predominantly hepatocellular injury, will on occasion present with a mixed hepatocellular and cholestatic injury or even a predominant cholestatic injury.

It may be difficult or impossible to assess the contribution of a drug to hepatic injury in a patient who has an underlying disease or a condition that itself is known to cause liver injury (eg, disseminated carcinoma, mycobacterial or fungal disease, chronic hepatitis or cirrhosis, use of alcohol, HIV infection, septicemia, and multiorgan involvement from an autoimmune process). Furthermore, in settings in which many drugs are used, often the case in the elderly, assigning attribution to any one agent is difficult.

These days herbal-induced liver injuries are being increasingly encountered. Special problems related to herbal products relate to the variable purity and dose of the herbal and an absence of a reliable way to determine how (and how often even where) the herbal was synthesized and produced. Hepatic reactions from many widely used herbal products including those to induce weight loss, build muscle, increase virility, and reduce anxiety have been clearly established as the cause of significant even fatal liver disease. Obtaining an accurate history from the patient of the use of herbal products is often extremely difficult.

The variable timing of stopping, restarting, and substitution of drugs adds to the difficulties. Dilemmas are further heightened when one or more of the drugs suspected to have caused hepatotoxicity are required in the treatment of a serious underlying illness.