Current Concepts in Liver Disease: An Update, Part 2
Moderator: Dr. Linda D. Ferrell
Section 4 -
The Pathology of Drug-Induced Liver Injury
David E. Kleiner
Laboratory of Pathology, National Cancer Institute
The Problem of Drug-Induced Liver Injury
Discerning the presence drug-induced liver injury (DILI) on biopsy is one of the most challenging and
frustrating problems in all of hepatic pathology. Reported patterns of DILI in the liver encompass all
of non-drug-related pathology and more. Clinical information on the patient's drug history is often
sketchy or may involve multiple medications, the use of herbal remedies is often unreported and
occupational or household toxic exposures may further complicate the picture. Laboratory data are
frequently incomplete and baseline values that pre-date the DILI may not even exist. The literature of
hepatotoxicity is replete with case reports and small series that may or may not include biopsy data and
for which no single search strategy will identify all of the relevant articles. One can easily spend
hours trying to evaluate a single case.
Nevertheless, the diagnosis of DILI and identification of the offending agent(s) is of critical
importance, not only for a particular patient but also for patients in general. The literature evidence
suggests that DILI is a significant cause of hepatitis and acute liver failure. A large study from
France noted that DILI is responsible for approximate 10% of all cases of acute hepatitis . Lee and
his coworkers in the Acute Liver Failure Study Group have reported that DILI is the leading cause of
acute liver failure (ALF)
in the U.S., with acetaminophen responsible for half of DILI-related ALF
A study of the UNOS transplant database demonstrated similar findings, with DILI responsible for 15% of
ALF going to transplant. Half of those cases were attributed to acetaminophen, with isoniazid,
propylthiouracil, phenytoin and valproate accounting for the next 21% . Hepatotoxicity, either
demonstrated in pre-clinical studies, clinical trials or post-marketing surveillance, is the single most
common reason for a drug to be withdrawn from clinical use by the FDA or by drug companies.
Troglitazone, fialuridine, bromfenac and trovafloxacin have all fallen recently from this flaw.
Sometimes the question is not whether a particular agent caused DILI, but whether the agent may be
safely continued, i.e. is the level of injury mild enough that one can risk continuing a medication that
is beneficial and for which no good alternatives exist. The most common example for this situation is
methotrexate, where the drug may be given until the patient develops bridging fibrosis. A related
example occurs when multiple potential agents are involved, not an uncommon situation in older or
hospitalized patients, and a biopsy is performed to help sort out the possible causes of liver injury.
Pathology and Pathophysiology
Hepatotoxic agents are classified in several ways, but a common categorization is to divide them into
intrinsic and idiosyncratic hepatotoxins. Intrinsic hepatotoxins cause injury in a dose-dependent,
reproducible manner, whereas the injury caused by idiosyncratic hepatotoxins is unpredictable and not
dose-dependent. Most DILI is idiosyncratic, although a few medications, such as acetaminophen, are
intrinsic hepatotoxins. Because of the dose-dependency of intrinsic hepatotoxins, toxicity is easier to
control and study. Intrinsic hepatotoxins are further sub-classified as either direct, in which the
agent itself is the poison, or indirect, in which the agent is metabolized reproducibly to a toxic
substance. Finally, toxicity is classified by the target of the agent, usually either the hepatocyte,
resulting in hepatocyte necrosis (usually along zonal lines), or the cholangiocyte, resulting in duct
destruction. Acetaminophen is an example of an indirect intrinsic hepatotoxin that causes zone 3
necrosis of hepatocytes.
Idiosyncratic hepatotoxins are generally subdivided by mechanism of action into metabolic and
immunologic hepatotoxins. Metabolic hepatotoxins are presumably converted to toxic metabolites only in
sensitive individuals, leading to clinical evidence of injury only rarely. Similarly, only certain
individuals develop "drug allergy" to immunologic hepatotoxins, leading to hepatic injury that resembles
autoimmune hepatitis or cholangitis. Phenytoin and the sulfonamides are examples of immunologic
hepatotoxins, while isoniazid and valproic acid are examples of metabolic hepatotoxins.
The pathology of injury can help to classify the type of hepatotoxin. In general, many agents that
cause zonal necrosis are intrinsic hepatotoxins, while agents that cause hepatitis-like necroinflammatory
injury or hepatocellular cholestasis are idiosyncratic hepatotoxins. There are exceptions of
course—halothane is usually considered an idiosyncratic hepatotoxin yet is associated with zonal
A Systematic Approach to the Pathology of DILI
Given the complexity of the problem, one must have a systematic approach to examining the liver biopsy
followed by a systematic evaluation of the clinical information. Reviewing the liver biopsy first,
without considering the history, is ideal, because that keeps the mind free of biases and leaves one
alert to seeing all of the features. As a practical matter this may be difficult. Cases of potential
DILI are frequently associated with anxious clinical staff, and one may receive a phone call or an e-mail
on the case before the biopsy is even performed. These contacts can be used as an opportunity to assess
the extent of the pre-biopsy clinical work-up as well as to make decisions about processing the biopsy
(i.e. whether or not to freeze a piece or fix a bit in glutaraldehyde). Decisions can also be made at
the grossing bench to cut additional sections for special stains or immunohistochemistry. At a minimum,
one should ask for a Masson trichrome and iron stain in addition to the routine stains, as well as a
reticulin if there is evidence of portal hypertension. Cut additional unstained slides to be held in
reserve, particularly if the practice in your lab is to level through most of the biopsy in the first
While it sounds obvious, the importance of careful systematic examination of biopsy cannot be
overestimated. It is easy to become distracted by one or another feature and overlook key pathologic
changes that may help distinguish drug-injury from non-drug-injury. It is vitally important to note what
is happening where—some patterns of liver injury follow zonal or acinar lines while other injury patterns
are irregular or random. Table 1 outlines the features and changes one should evaluate before deciding
on the pattern of injury.
Table 1: Basic Histologic Features of (Drug-Induced) Liver Injury
Once a careful microscopic examination has catalogued the lesions, one should try to define the
pattern of injury. This is an extremely useful approach, since any individual hepatotoxic agent is
usually associated with only a few types of injury and because the non-DILI pathologic differential is
driven by the pattern of injury. Determination of the pattern of injury is critical when there is
pre-existing hepatic disease since it allows one to separate features of disease from features of injury
(sometimes). For example, a biopsy that shows chronic hepatitis with zone 3 coagulative necrosis is not
explained by underlying hepatitis C infection, since zone 3 necrosis is not part of the pattern of injury
caused by hepatitis C. Table 2 shows one way to classify the patterns of
injury in liver disease, along with the non-DILI differential and some typical implicated drugs. Please
note that the table is not meant be exhaustive, either in terms of the non-DILI differential or the
associated drugs. Certain patterns of injury are unusual in typical practice, and so DILI should be
considered first until excluded. These patterns include zonal necrosis, acute cholestasis, mixed
hepatocellular-cholestatic injury, microvesicular steatosis, veno-occlusive disease/sinusoidal
obstruction syndrome, peliosis, and nodular regenerative hyperplasia. Certain individual features have
also been associated more often with DILI, including granulomas, large numbers of eosinophils, and
canalicular/hepatocellular cholestasis. Odd mixed patterns or unusual features, such as cholestasis with
steatohepatitis should also trigger a search for a causative agent.
|Feature ||Sub-feature ||What to Look For|
|Inflammation ||Overall ||Degree, Lobular vs. Portal vs. Periportal, Zonal distribution, Confluent/bridging inflammation, General character|
| ||Granulomas ||Location (Peri-ductal, Portal, Lobular), Type (microgranulomas, necrotizing vs. non-necrotizing)|
| ||Plasma cells ||Clustering at interface, in lobules|
| ||Eosinophils ||More than expected numbers, presence in lobular inflammation|
| ||Neutrophils ||Association with ductular proliferation, coagulative necrosis, Mallory's hyalin|
|Hepatocellular Injury ||Coagulative necrosis ||Sheets of necrotic (rather than apoptotic) hepatocytes with or without areas of collapse/drop-out, note zonality|
| ||Apoptosis ||Numbers and location. Apoptosis may be seen at the edges of zones of coagulative necrosis|
| ||Ballooning ||Location and other associations (steatosis, cholestasis, Mallory's hyalin)|
|Fibrosis ||Overall ||Degree, age (recently deposited collagen may be pale blue or red-blue compared to portal area collagen), location (portal, peri-portal, pericentral)|
| ||Perisinusoidal ||"Chicken-wire" pattern, may be periportal in chronic cholestasis, pericentral in steatohepatitis, central venous hypertension and NRH|
|Cholestasis ||Hepatocellular/Canalicular ||Often seen together, usually zone 3, must be distinguished from lipofuscin pigment (very common) and iron (also common)|
| ||Cholangiolar ||Bile plugs in dilated cholangioles at edges of portal areas|
| ||Ductal ||Bile plugs in main duct of portal area|
| ||Cholatestasis ||Pallor and microvacuolation of periportal hepatocytes, seen in chronic cholestasis, may be copper positive|
| ||Duct injury ||Infiltration of ducts by inflammatory cells, reactive atypia and necrosis of ductal epithelial cells|
| ||Ductal paucity ||Disappearance of main ducts from portal areas, may need cytokeratin stain to confirm|
|Steatosis ||Macrovesicular ||Zonal location, degree|
| ||Microvesicular ||Patches of finely vacuolated cells with central nuclei, note degree and location|
|Vascular ||Occlusion/ absence ||Narrowing or obliteration of central veins (Masson is a good stain for this purpose), missing or narrowed portal veins|
| ||Hemorrhage ||Often secondary to distal obstruction of veins|
| ||Sinusoidal dilatation ||Expanded sinusoidal spaces associated with compressed/atrophic hepatocytes, also peliosis hepatis|
| ||Nodular transformation ||Best seen on reticulin stain, expanded, 2-cell thick cords alternating with compressed atrophic cords|
|Cytoplasmic alterations ||"Ground-glass" change ||Fine homogenization of hepatocyte cytoplasm, may not fill cell completely, may be PAS(+) or (-)|
| ||Mallory's hyalin ||Note location and degree, may be related to steatohepatitis or chronic cholestasis and may be associated with particular drugs|
|Other findings ||Stellate cells ||Lipid-laden stellate cells seen as small round "holes" in sinuses, closer examination will show lipoblast-like morphology|
|Special stains and Immuno ||Masson trichrome ||Helps to distinguish true fibrosis (dark/solid blue or green) from zones of collapse (red/pink to pale blue or green), veno-occlusive lesions easier to detect, bile ducts may stand out from inflammation|
| ||Iron ||While iron can be injurious on its own, iron stains help to distinguish the three main pigments: iron (blue), lipofuscin (dirty brown), and bile (green to greenish-brown). Canalicular/hepatocellular bile may be easier to detect|
| ||Copper ||Helps to define chronic cholestasis|
| ||PAS/diastase ||Distinguishes glycoprotein inclusions (positive), abnormal glycogen inclusions (seen with cyanomid) and negative inclusions such as Mallory's hyalin and megamitochondria|
| ||Reticulin ||Indispensable for detecting nodular transformation|
| ||Ubiquitin or p63 ||Highlights/confirms the presence of Mallory's hyalin|
| ||Cytokeratin 7 ||Highlights ducts, ductular reaction, cholestatic hepatocytes|
| ||Cytokeratin 19 ||More restricted pattern than CK7-just ducts and ductular reaction|
Once the pathologic pattern of injury is identified, the clinical history and laboratory data can be
reviewed to identify potential agents and to exclude non-DILI causes of hepatic injury. In particular it
is important to identify the date of onset as accurately as possible. When did the liver-associated
enzymes begin to rise, when did the patient develop symptoms or signs of liver disease? Clinicians
frequently use the ratio of ALT to alkaline phosphatase (both as a ratio to their upper limits of normal)
to classify the injury as hepatitic (ALT/AP > 5), mixed (ALT/AP >2 and < 5) or cholestatic
(ALT/AP <2). Although there is no absolute correlation between this classification and observed
pathology, much of the literature of DILI classifies drugs by these criteria. The hardest piece of
information to obtain is often the medication history. Patients are often poor historians when it comes
to remembering when they began taking certain medication and for how long. It is critical to compare the
onset of injury to when medication was started—it is not possible for DILI to start before the drug is
administered. Many drugs, particularly those causing autoimmune or metabolic idiosyncratic injury,
require weeks to months of administration before demonstrating toxicity. A few drugs may cause injury
even though they are no longer being taken. Several patients in the fialuridine-HBV trial developed
hepatic failure several weeks after the medication was stopped. The combination drug
amoxicillin-clavulanic acid may cause jaundice days to weeks after the course of therapy is over. In
researching the medication history it is important to include uses of over-the-counter medications and
natural products (herbals). Herbals are an increasing source of hepatotoxicity because of their
increased use and poor safety evaluation. Some herbal products represent poorly characterized mixes of
multiple individual compounds making identification of the causative agent that much more difficult.
Patients should be encouraged to bring packages and pill bottles for review whenever possible.
With the clinical and pathologic data assembled, one can begin the process of deciding causality. A
clinical multi-disciplinary conference is a good time to review all of the data. Although there are
several published schemes for determining causality (RUCAM and other refs), a common sense approach is
often just as good. Dr. Irey, a pathologist at the AFIP, came up with such a straightforward approach
for the general assessment of environmental or medically-induced pathological injuries:
- Temporal eligibility – Did the drug
precede the injury and was there enough time for drug injury to develop.
other drugs, toxins and diseases – Have necessary clinical tests been performed to exclude other causes
of liver disease?
- Potential for injury – Is the agent known to cause liver
- Precedent for injury pattern – what is/are the reported pattern(s) of
- De-challenge/Re-challenge – Did the injury subside when the suspect
agent was stopped? If the agent was restarted did the injury return?
- Toxicological analysis – Were blood/tissue levels of drug measured? Where drug
Some level of certainty should be applied to the diagnosis, as this may be helpful in planning therapy
or deciding whether to risk re-administration of clinically-useful agents. A common scale would be to
describe the DILI as Definite, Probable, Possible, Uncertain, or Excluded. Additional testing may be
required to exclude non-DILI causes of hepatitic injury. The use of these will be dictated mainly by
clinical circumstances and judgment. In some cases, despite good clinical and pathological data, it will
not be possible to determine whether DILI has occurred simply because it is not possible or feasible to
exclude competing causes of injury. However, a systematic approach will give the clinical team its best
opportunity to settle the issue one way or another.
Summary of the Systematic Approach
- Make processing decisions prior at the
grossing bench. Leave your options open.
- Examine the biopsy in the absence of
history information. Examine all compartments and zones, noting inflammation, cholestasis,
hepatocellular injury, fibrosis, vascular changes, special stain and immunohistochemical changes.
- Try to decide on the overall pattern of injury and formulate the pathological
- Review the clinical history focusing both on the suspect
agents as well as the studies that have been done to rule out other causes of liver injury.
- Together with the clinical staff, try to assign a level of causality to the suspect
Table 2: Major Patterns of Injury Associated with DILI
|Pattern ||Characteristic features ||Non-DILI differential ||Drug Examples|
|Necroinflammatory || || |
|Acute Hepatitis ||Lobular-dominant inflammation with/without confluent or bridging necrosis; no cholestasis ||Acute viral or autoimmune hepatitis ||Isoniazid, sulfonamides, halothane, drugs associated with autoimmune hepatitis|
|Chronic Hepatitis ||Portal-dominant inflammation, interface hepatitis (also includes mononucleosis pattern), with or without portal-based fibrosis; no cholestasis ||Chronic viral or autoimmune diseases, early PBC/PSC, mononucleosis-associated hepatitis ||Nitrofurantoin, methyldopa, sulfonamides, phenytoin (mono-pattern), drugs associated with autoimmune hepatitis|
|Granulomatous ||Inflammation dominated by granulomas (usually non-necrotizing), portal or lobular ||Sarcoidosis, PBC, fungal and mycobacterial, atypical bacterial infections ||Phenytoin|
|Zonal Coagulative Necrosis ||Zone 3 or 1 coagulative necrosis, usually without significant inflammation ||Hypoxic-ischemic injury (zone 3), Herpes and adenovirus usually cause non-zonal necrosis ||Acetaminophen (zone 3), Halothane (zone 3), CCl4 (zone 3), Cocaine (zone 1 or 3)|
|Cholestatic || || |
|Acute Cholestasis ||Hepatocellular and/or canalicular cholestasis in zone 3, may show duct injury, but little inflammation ||Sepsis, acute large duct obstruction ||Erythromycin, estrogens, androgens, diazepam, diphenylhydantoin|
|Chronic Cholestasis ||Periportal cholatestasis, portal-based fibrosis, copper accumulation, duct injury and loss ||PBC, PSC, chronic large duct obstruction, autoimmune cholangitis ||Floxuridine (by hepatic artery infusion), chlorpromazine, tolbutamide|
|Mixed Hepatocellular-Cholestatic ||Acute hepatitis pattern plus zone 3 cholestasis, inflammation may be very severe with confluent necrosis ||Acute viral hepatitis ||Very common DILI pattern: antibiotics, isoniazid, nitrofurantoin, diclofenac|
|Steatotic || || |
|Steatosis, microvesicular ||Predominantly microvesicular steatosis, inflammation variable ||Alcohol, fatty liver of pregnancy ||Valproic acid, tetracycline, azathioprine, didanosine, fialuridine|
|Steatosis, macrovesicular ||Predominantly macrovesicular steatosis without significant portal or lobular inflammation, no cholestasis ||Very common finding in general population, alcohol, obesity, diabetes ||Methotrexate, tamoxiphen, valproic acid, many organics|
|Steatohepatitis ||Zone 3 ballooning injury, sinusoidal fibrosis, Mallory bodies, variable inflammation and steatosis ||Common finding in general population, alcohol, obesity, diabetes ||Amiodarone, perhexiline maleate, tamoxiphen, methotrexate|
|Vascular || || |
|Sinusoidal Dilation/Peliosis ||Sinusoidal alterations with/ without mild lobular inflammation, sinusoidal fibrosis ||artifactual, acute congestion, bacillary angiomatosis, nearby mass lesions ||Androgens, estrogens, glucocorticoids, thioguanine, azathioprine|
|Sinusoidal Obstruction syndrome/Veno-occlusive disease/ Budd-Chiari ||Occlusion or loss of central veins, thrombosis, with or without central hemorrhage and necrosis || ||Chemotherapeutic agents, bone marrow transplant prep regimen, certain teas|
|Hepato-portal sclerosis ||Disappearance of portal veins ||Arteriohepatic dysplasia ||Arsenicals|
|Nodular regenerative hyperplasia ||Diffuse nodular transformation, with or without mild inflammation and sinusoidal fibrosis ||Collagen-vascular diseases, lymphoproliferative diseases (but perhaps because of DILI) ||Azathioprine, thioguanine, mercaptopurine, steroids|
There are a variety of publications and resources available to help you define whether DILI has
occurred. This list is not exhaustive, but indicates references that this observer has found helpful.
The books, book chapters and reviews, in particular, frequently contain lists of drugs with associated
patterns of injury.
- Farrell, GC. Drug-Induced Liver Disease.
Edinburgh : Churchill Livingstone, 1994.
- Kaplowitz, N, DeLeve, LD, eds.
Drug-Induced Liver Disease. New York: Marcel Dekker, 2003.
- Zimmerman, HJ.
Hepatotoxicity: The Adverse Effects of Drugs and Other Chemicals on the Liver, 2nd edn.
Philadelphia : Lippincott, Williams & Wilkins, 1999.
- Geller, SA, Petrovic LM. Biopsy
Interpretation of the Liver. Philadelphia : Lippincott William & Wilkins, 2004, pp. 111-124. This
book comes with a CD that has numerous pictures of DILI cases.
- Scheuer, PJ,
Lefkowitch JH. Liver Biopsy Interpretation, 7th edn. Elsevier Saunders, 2005, pp. 125-144.
- Snover , DC . Biopsy Diagnosis of the Liver. Baltimore: Williams &
Wilkins, 1992, pp. 164-177. A bit dated and out-of-print, but the chapter contains an alphabetical list
of drugs and associated patterns of injury.
- Zimmerman, HJ, Ishak KG. Hepatic
injury due to drugs and toxins. In: MacSween RNM, Burt AD, Portmann BC, Ishak KG, Scheuer PJ, Anthony
PP, eds. Pathology of the Liver. London : Churchill Livingstone, 2002.
- Black M, ed. Drug hepatotoxicity. Clin
Liver Dis 2003; 7 (2): 295-512. Entire issue devoted to drug hepatotoxicity.
- Goodman ZD. Drug hepatotoxicity. Clin Liver Dis 2002; 6: 381-397.
- Ishak KG, Zimmerman HJ. Morphologic spectrum of drug-induced hepatic disease.
Gastroenterol Clin North Am. 1995; 24: 759-786.
- Lee WM. Drug-induced
hepatotoxicity. N Engl J Med 2003; 349:474-485.
- Mohi-ud-din R, Lewis JH. Drug-
and chemical-induced cholestasis. Clin Liver Dis 2004; 8: 95-132.
- Hepatotoxicity in the Twenty-First Century. Semin Liv Pathol 2002; 22 (2): 115-206.
Entire issue devoted to drug hepatotoxicity.
- Benhamou JP. Drug-induced hepatitis: clinical aspects. In: Fillastre JP, ed. Hepatotoxicity of drugs. Rouen: University de Rouen, 1985: pp 22-30.
- Lee WM. Acetaminophen and the U.S. Acute Liver Failure Study Group: lowering the risks of hepatic failure. Hepatology. 2004; 40: 6-9.
- Ostapowicz G, Fontana RJ, Schiodt FV et al. Results of a prospective study of acute liver failure at 17 tertiary care centers in the United States. Ann Intern Med, 2002; 137: 947-954.
- Russo MW, Galanko JA, Shrestha R et al. Liver transplantation for acute liver failure from drug-induced injury in the United States. Liver Transplantation 2004; 10: 1018-1023.