Drug Effects
Michael S. Torbenson, MD
15.1 OVERVIEW
The acronym DILI, which stands for “drug-induced liver injury”, is commonly used in the literature on drug reactions. In the United States, the most common categories of drugs associated with DILI in adults are acetaminophen, antibiotics, central nervous system agents, antihypertensive agents, and antidiabetic agents.1 Dietary supplements are also important causes of DILI, being found in about 10% of cases.1 In contrast to adults, the most common agents causing DILI in children are antibiotics and central nervous system agents.2 DILI can be broadly classified by the main mechanism of injury: direct toxin, allergic (hypersensitivity) drug reaction, or an idiosyncratic drug reaction. Idiosyncratic drug reactions are the most common type seen in routine surgical pathology, as allergic reactions and direct toxins are often readily identified by clinical findings and laboratory testing.
When evaluating a liver biopsy for a possible drug reaction, one of the most important points to remember is that essentially no findings are pathognomonic for a drug reaction. In addition, there can be a wide range of histological findings for any given single drug. Tamoxifen-associated DILI nicely illustrates this point, being associated with acute hepatitis, massive liver necrosis, peliosis hepatitis, steatosis, steatohepatitis, and cholestasis.3 Although some medications tend to have a typical pattern of injury, the findings in DILI overall tend to be neither very sensitive nor specific, and it is unwise to rely solely on memorizing a few or even many patterns that are illustrated in books or review articles. Instead, the best approach is to have a high degree of suspicion and to consider drug reactions in your differential in all cases, but especially in those cases that have patterns of injury that do not fit for the known clinical and serological findings. In fact, histological patterns that “just don’t fit” for other entities can often be a useful clue to DILI.
When correlating the histological findings with the patient’s mediations, the goal is to identify medications that are the most likely to have caused the injury, but it’s not uncommon to have multiple potentially offending agents. In this setting, the overall pattern of injury can provide some guidance, but is not foolproof because of the variability of injury patterns for any one drug and the large degree of overlap seen with different DILI causing agents. Thus, correlation with other clinical findings is often critical, including prior reactions to drugs in the same class and temporal associations. In general, the most
straightforward temporal associations are with direct toxins such as acetaminophen because toxicity is dose-dependent and based on both recent usage and the amount of intake. In contrast, for idiosyncratic drug reactions, a compatible exposure history includes only that the drug is in current usage or was recently discontinued. Most idiosyncratic drug reactions occur within the first several weeks following exposure, a very helpful observation, but there are exceptions where a drug is used for months to years before hepatic injury develops.
straightforward temporal associations are with direct toxins such as acetaminophen because toxicity is dose-dependent and based on both recent usage and the amount of intake. In contrast, for idiosyncratic drug reactions, a compatible exposure history includes only that the drug is in current usage or was recently discontinued. Most idiosyncratic drug reactions occur within the first several weeks following exposure, a very helpful observation, but there are exceptions where a drug is used for months to years before hepatic injury develops.
Resources for the patterns of DILI associated with drugs are numerous. Books can be an excellent starting point, but the list of potential agents for any given pattern of injury seems to be ever growing, so other important resources include searching PubMed or commercial databases. Another helpful resource is the LiverTox website, http://livertox.nlm.nih.gov/, a free resource hosted by the NIDDK and the NLM of the United States.
A diagnosis of DILI is made by identifying a compatible history of drug exposure, finding compatible histological findings, and excluding other potential causes. However, in most idiosyncratic DILI cases, the diagnosis should be considered as “most consistent with” a drug reaction and not a proven drug reaction, as subsequent testing can sometimes identify other causes. As an example, both acute hepatitis E and acute hepatitis C can closely mimic drug reactions and are sometimes only retrospectively identified, after a presumed diagnosis of DILI has been made based on the then available clinical and histological findings.1,4
15.2 ISOLATED HYPERAMMONEMIA
Isolated elevations in serum ammonia can result from drug toxicity to the liver, in particular valproic acid5 or infusion of high-dose 5- fluorouracil for chemotherapy.6 The liver biopsy often looks essentially normal, other than minimal fatty change (Fig. 15.1). The differential includes urea cycle defects7 or portosystemic shunts,8 but the clinical findings in these entities are distinct.
 Figure 15.1 Isolated hypermoniaemia. This patient was taking valoproic acid. The liver biopsy shows minimal fat and inflammation but is otherwise essentially normal in appearance. |
 Figure 15.2 Resolving hepatitis. The clinical history was that of previously elevated enzymes that were significantly lower at the time of biopsy. The liver biopsy shows only minimal inflammation, but numerous pigmented Kupffer cells are present in the lobules. |
15.3 RESOLVING PATTERN OF HEPATITIS
In most cases of idiosyncratic DILI, the hepatitic injury diminishes after stopping the drug, but the liver enzymes may take months to completely normalize. DILI with significant cholestasis tends to resolve more slowly than cases that are purely hepatitic. This clinical situation, of improved but persistent biochemical abnormalities despite stopping a drug, can often prompt a liver biopsy. Biopsies in this setting typically show a resolving hepatitis pattern, but sometimes can show findings that suggest an alternative diagnosis to DILI. The resolving hepatitis pattern, like most patterns, is not etiologically specific, but the histological differential always includes resolving DILI. The main finding is in the lobules, with a mild prominence in Kupffer cells, often small Kupffer cell aggregates that are easiest to seen on periodic acid-Schiff (PAS)-D stains, and minimal lobular inflammation (Fig. 15.2). Rare acidophil bodies may be present, and patchy minimal cholestasis can also be present. The portal tracts often show minimal to mild lymphocytic inflammation and scattered ceroid-laden macrophages.
15.4 DIRECT TOXINS
Direct toxins by definition lead to liver injury in a reproducible and dose-dependent fashion. The most common cause is acetaminophen, but other causes include mushroom poisoning and miscellaneous household and industrial chemicals.
 Figure 15.3 Acetaminophen toxicity. There is extensive hepatocellular necrosis in Zone 3, with sparing of the Zone 1 hepatocytes. Inflammation is minimal. |
Toxins cause direct hepatocyte necrosis with little inflammation. In most cases, the necrosis tends to have a Zone 3 pattern (Fig. 15.3). The dead hepatocytes are typically a deep eosinophilic color and lose their nuclei. Although the hepatocytes in Zone 3 will be dead, the Kupffer cells and endothelial cells in the same areas are often still intact and present in their normal locations. When there is severe injury, the necrosis is often panacinar, and no zonation will be evident. Other zonal patterns of necrosis are less common. A Zone 1 pattern of necrosis has been associated with toxic injury from phosphorous, ferrous sulfate, and cocaine.9,10 Hepatitis A, while not a toxic injury, can rarely show a Zones 1 and 2 pattern of necrosis with sparing of Zone 3 (Fig. 15.4). A Zone 2 pattern of necrosis has been associated with Beryllium toxicity. With any of these patterns, those hepatocytes that survive often show mild small- and medium-sized droplet fatty change and cholestasis. Depending on the time interval between injury and biopsy, a reactive bile ductular proliferation can also be seen in later biopsies. At these later time points, the liver can also show mild portal inflammation and mild Kupffer cell hyperplasia. The Kupffer cells and the proliferating bile ductules can also show significant reactive iron accumulation.
Acetaminophen toxicity
Acetaminophen toxicity is the most common etiology of direct liver injury, causing up to 50% of cases of acute liver failure in the United States and the United Kingdom.11,12 Acetaminophen toxicity tends to have very high serum alanine aminotransferase levels (ALT; median ALT levels of about 4,000 IU/L) with relative low
bilirubin (median levels of about 4 mg/dL). This stands in contrast to idiosyncratic DILI which tends to have lower ALT levels (median of approximately 500 IU/L) but higher bilirubin levels (median around 20 mg/dL).
bilirubin (median levels of about 4 mg/dL). This stands in contrast to idiosyncratic DILI which tends to have lower ALT levels (median of approximately 500 IU/L) but higher bilirubin levels (median around 20 mg/dL).
 Figure 15.4 Acute hepatitis A. The liver shows extensive necrosis in Zones 1 and 2, with sparing of Zone 3 hepatocytes. |
Toxicity occurs both with intentional overdose as part of suicide attempts (approximately 40% of cases) and with unintentional overdoses. Unintentional overdoses most commonly occur when there is significant alcohol use or in the setting of chronic pain, with patients using medications that contain both narcotics and acetaminophen.11 In general, hepatic injury requires an exposure of more than 7.5 g, with severe injury seen at levels of 15 to 25 g. However, these general toxic thresholds can be lowered by the coexistence of fatty liver disease,13 chronic alcohol consumption,11 sleep apnea,14 or use of drugs that stimulate the P-450 enzyme system, including carbamazepine, dimetidine, isoniazid, and phenytoin.11 Treatment for acetaminophen toxicity with N-acetylcysteine is very effective when given within the first 24 hours of presentation.
15.5 ALLERGIC TYPE DRUG REACTIONS
Allergic drug reactions in general are rarely biopsied because the clinical findings tend to be easily recognized, including hives, wheezing, and peripheral eosinophilia. However, in some cases, the clinical findings can be mild or obscured by other comorbid conditions, leading to a liver biopsy in order to evaluate elevated liver enzymes. In these cases, the drug exposure is typically recent, usually within the preceding few days or weeks.
The main biopsy findings are increased eosinophils in the portal tract, lobules, and often the sinusoids. Both the portal inflammation and the lobules also commonly show mild lymphocytic inflammation, and occasional apoptotic bodies are found in the lobules.
Eosinophil-rich inflammation is the main histological finding in allergic DILI of the liver. However, diagnoses of drug reactions can be wrong when they are based solely on finding eosinophils in the biopsy because eosinophils are a common component of many other inflammatory conditions of the liver, including viral hepatitis, autoimmune hepatitis, acute cellular rejection, and primary biliary cirrhosis. In some of these cases, the eosinophils can even be focally prominent, for example, with a small cluster of cells in a portal tract. However, eosinophils are not the dominant finding in these cases, lacking the diffuse nature and density of eosinophils seen in an allergic type drug reaction. In addition, an eosinophilic drug reaction will lack the histological changes that typify these other diseases. Finally, the differential also includes a peripheral eosinophilia without an eosinophil-rich hepatitis. In these cases, the eosinophils are essentially all located in the sinusoids, and there is minimal or absent lobular injury.
DRESS syndrome
Rare drug reactions develop a cluster of severe systemic conditions called the DRESS syndrome, an acronym that stands for “drug reaction with eosinophilia and systemic symptoms.” The DRESS syndrome typically develops three weeks or more after starting the triggering medication and can have a mortality of up to 10%.15 Symptoms can last for weeks after discontinuing the triggering medication. The symptoms tend to include a core set of findings consisting of a fever greater than 38°C, rash, lymphadenopathy in at least two locations, and peripheral blood findings of eosinophilia, thrombocytopenia, and either lymphocytosis or lymphopenia. The lymphocytes also commonly show cytological atypia. The liver is involved in about 80% of cases, with primarily aspartate aminotransferase (AST) and ALT elevations. Biopsies show marked portal and lobular eosinophilia (Fig. 15.5). The list of drugs that can lead to the DRESS syndrome grows continually, but in one comprehensive review of the literature,16 the most common reported drugs were carbamazepine (27% of cases), allopurinol (11%), lamotrigine (6%), phenobarbital (6%), nevirapine (5%), phenytoin (4%), and abacavir (3%), with many more drugs at lower frequencies.
15.6 IDIOSYNCRATIC DRUG REACTIONS
Idiosyncratic drug reactions are the most common form of DILI encountered in surgical pathology. This
type of DILI by definition is not dose related and cannot be predicted on an individual level, making the diagnosis a clinical challenge. A diagnosis of idiosyncratic DILI requires exclusion of viral hepatitis (A, B, C, and E) and autoimmune hepatitis. Several broad histological patterns are seen with idiosyncratic DILI, which are described in more detail later. These patterns are not exclusive, and a biopsy can show elements of more than one of these patterns.
type of DILI by definition is not dose related and cannot be predicted on an individual level, making the diagnosis a clinical challenge. A diagnosis of idiosyncratic DILI requires exclusion of viral hepatitis (A, B, C, and E) and autoimmune hepatitis. Several broad histological patterns are seen with idiosyncratic DILI, which are described in more detail later. These patterns are not exclusive, and a biopsy can show elements of more than one of these patterns.
 Figure 15.5 DRESS. The portal tracts are filled with eosinophils. |
Of note, idiosyncratic drug reactions are rarely if ever the cause of significant fibrosis. If a biopsy does show convincing fibrosis, then other causes of chronic hepatitis are more likely. Identifying a drug reaction superimposed upon another form of known chronic hepatitis, such as chronic hepatitis C, can be very challenging. These cases are approached by looking for findings atypical for the known underlying liver disease and by correlating histology with the laboratory and clinical findings. For example, an individual with known chronic hepatitis C who has long term but relative stable elevations in liver enzymes may present with a sudden increase in AST and ALT, suggesting a superimposed injury. A biopsy of the liver in this setting might show marked lobular hepatitis, further supporting the likelihood of a superimposed injury, because the vast majority of chronic hepatitis C biopsies show mild to moderate lobular inflammation.
Hepatic pattern
The hepatitic pattern is a common pattern with DILI and shows lobular and portal inflation that can range from mild to marked. Most herbal remedies also tend to show this pattern of injury (Fig. 15.6). The lobular hepatitis can be panacinar or can show a Zone 3 accentuation (Fig. 15.7). Zone 3 necrosis is common with more severe grades of lobular hepatitis. In some cases, there can be extensive necrosis with few surviving hepatocytes (Fig. 15.8). The inflammation is predominately T cells but there can be B cell aggregates in the portal tracts. Lobular cholestasis is common when there is moderate to marked lobular hepatitis. The inflammation in the portal tracts commonly includes occasional plasma cells, neutrophils, and eosinophils. Interface activity is typically present when there is moderate to marked portal inflammation.
 Figure 15.6 Hepatitic drug reaction. This drug reaction resulted from a herbal remedy. |
 Figure 15.7 Hepatitic drug reaction. The lobular hepatitis in this drug reaction (Atorvastatin) shows a clear Zone 3 accentuation with Zone 3 hepatocyte necrosis. |
The differential for a hepatitic pattern of DILI is primarily acute viral hepatitis and autoimmune hepatitis. There are no histological findings that will reliably separate these three possibilities, and a diagnosis of DILI is based on excluding other causes and having exposure to a plausible drug candidate. Histological findings such as prominent plasma cells,
interface activity, or Zone 3 accentuated inflammation are not etiologically specific and can be seen in all three of these entities. To reinforce this point, several drugs can cause DILI with a plasma cell-rich hepatitic pattern, one that often cannot be reliably separated from autoimmune hepatitis based on histological findings (Fig. 15.9). The most common of these are minocycline (acne), methyldopa (hypertension), clometacin (anti-inflammatory), and nitrofurantoin (urinary tract infections) (Table 15.1). Further complicating the workup of these cases, the DILI can start soon after beginning the medication or can start after several years of taking the medication. Furthermore, these drugs can also be associated with elevated serum antinuclear antibody (ANA) and/or antismooth muscle
antibody (ASMA) titers. Thus, potentially offending drugs need to be stopped as part of the clinical management before a diagnosis of either drug reaction or autoimmune hepatitis is secured. The presence of fibrosis strongly favors autoimmune hepatitis over one of these drug reactions. However, care must be taken to not over interpret expansion of the portal tracts by inflammation as portal fibrosis. Likewise, the Kupffer cell hyperplasia that accompanies many drug reactions can mimic pericellular fibrosis, whereas Zone 3 necrosis or bridging necrosis can mimic central vein fibrosis or bridging fibrosis, respectively.
interface activity, or Zone 3 accentuated inflammation are not etiologically specific and can be seen in all three of these entities. To reinforce this point, several drugs can cause DILI with a plasma cell-rich hepatitic pattern, one that often cannot be reliably separated from autoimmune hepatitis based on histological findings (Fig. 15.9). The most common of these are minocycline (acne), methyldopa (hypertension), clometacin (anti-inflammatory), and nitrofurantoin (urinary tract infections) (Table 15.1). Further complicating the workup of these cases, the DILI can start soon after beginning the medication or can start after several years of taking the medication. Furthermore, these drugs can also be associated with elevated serum antinuclear antibody (ANA) and/or antismooth muscle
antibody (ASMA) titers. Thus, potentially offending drugs need to be stopped as part of the clinical management before a diagnosis of either drug reaction or autoimmune hepatitis is secured. The presence of fibrosis strongly favors autoimmune hepatitis over one of these drug reactions. However, care must be taken to not over interpret expansion of the portal tracts by inflammation as portal fibrosis. Likewise, the Kupffer cell hyperplasia that accompanies many drug reactions can mimic pericellular fibrosis, whereas Zone 3 necrosis or bridging necrosis can mimic central vein fibrosis or bridging fibrosis, respectively.
 Figure 15.8 Hepatitic drug reaction. This severe drug reaction (Isoniazid [INH]) has led to the death of nearly all of the hepatocytes, with residual inflammation and ductular proliferation in the portal tracts. |
 Figure 15.9 Plasma cell-rich drug reaction. This case (minocycline) histologically looks like autoimmune hepatitis. |
Table 15.1 Drugs associated with both an autoimmune hepatitis like pattern of injury on histology plus positive autoantibody serology | ||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Granulomatous pattern
The differential for granulomas of the liver is broad, and most cases do not represent DILI. Instead, the most common causes of hepatic granulomas are idiopathic, sarcoidosis, infection, or primary biliary cirrhosis.17 Nonetheless, DILI is an important part of the differential. In most cases of DILI, the granulomas are part of a hepatitic pattern, accompanied by mild to moderate lymphocytic inflammation of the portal tracts and lobules, although occasionally the main pattern of injury is granulomas. The granulomas can be well formed and epithelioid or can be loose and ill defined. The granulomas are not associated with fibrosis (which suggests sarcoidosis when present) and do not show necrosis (which suggests infection when present). The granulomas can be in the portal tracts or the lobules or the bile ducts, but their location does not provide a strong clue to DILI as the etiology.
As is generally true for all cases of DILI, a diagnosis of DILI with granulomatous inflammation is one of exclusion. Stains should be performed to examine for fungal and acid fast bacterial infections. Beyond organism stains, there are no immunohistochemical stains that will identify the etiology of a granuloma. All granulomas should be polarized for foreign material and serum testing for antimitochondrial antibodies is important. Other findings that would favor primary biliary cirrhosis over DILI include florid duct lesions, bile duct loss, or fibrosis.
Cholangitic pattern
The cholangitic pattern is most commonly seen with antibiotics and shows duct inflammation, bile duct injury, reactive epithelial changes, and apoptotic duct epithelial bodies. The inflammation in the portal tracts is often mixed, with predominately lymphocytes and occasional neutrophils, eosinophils, and/or plasma cells. The bile ducts also show reactive changes and can be infiltrated by lymphocytes and neutrophils.
 Figure 15.10 Bland lobular cholestasis pattern. This drug reaction (androgens) shows a canalicular cholestasis pattern, nuclear pleomorphism, and relative little inflammation. |
Cholestatic pattern
This pattern refers to cases with lobular cholestasis that do not have significant inflammation and do not show changes of downstream biliary tract disease, a pattern also called bland lobular cholestasis. The cholestasis is often mild to moderate and may show a Zone 3 predominance in some cases. The bile can be in the hepatocyte cytoplasm or in the bile canaliculi. In contrast, bile is only seen in the bile ducts proper or the proliferating ductules, when the cholestasis is severe. This pattern of injury can be seen with a wide range of medications but the classic examples are oral contraceptives or anabolic steroids (Fig. 15.10). Of note, drug-induced lobular cholestasis can persist for several months after discontinuing the drug. In other cases, DILI can initially show a mixed hepatitic and cholestatic pattern, with significant inflammation and lobular cholestasis. When the medication is stopped, the inflammation disappears more rapidly than the cholestasis, and there is a window of time when biopsies show a bland lobular cholestatic pattern.
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
