Gastroenterology 2014;146:914 REVIEWS IN BASIC AND ......Overall, DILI accounts for

Gastroenterology 2014;146:914–928

REVIEWSAND

PERSPECTIVES

REVIEWS IN BASIC AND CLINICAL GASTROENTEROLOGYAND HEPATOLOGY

Robert F. Schwabe and John W. Wiley, Section Editors

Pathogenesis of Idiosyncratic Drug-Induced Liver Injury andClinical PerspectivesRobert J. Fontana

University of Michigan Medical Center, Ann Arbor, Michigan

Idiosyncratic drug-induced liver injury (DILI) is a rare dis-ease that develops independently of drug dose, route, orduration of administration. Furthermore, idiosyncratic DILIis not a single disease entity but rather a spectrum of rarediseases with varying clinical, histological, and laboratoryfeatures. The pathogenesis of DILI is not fully understood.Standardization of the DILI nomenclature and methods toassess causality, along with the information provided by theLiverTox Web site, will harmonize and accelerate researchonDILI. Studies of newserumbiomarkers such as glutamatedehydrogenase, high mobility group box protein 1, andmicroRNA-122 could provide information for use in diag-nosis and prognosis and provide important insights into themechanisms of the pathogenesis of DILI. Single nucleotidepolymorphisms in theHLA region have been associatedwithidiosyncratic hepatotoxicity attributed to flucloxacillin,ximelagatran, lapatinib, and amoxicillin-clavulanate. How-ever, genome-wide association studies of pooled cases havenot associated any genetic factors with idiosyncratic DILI.Whole genome and whole exome sequencing analyses areunder way to study cases of DILI attributed to a singlemedication. Serum proteomic, transcriptome, and metab-olome as well as intestinal microbiome analyses will in-crease our understanding of the mechanisms of thisdisorder. Further improvements to in vitro and in vivo testsystems should advance our understanding of the causes,risk factors, and mechanisms of idiosyncratic DILI.

Keywords: Acetaminophen; Liver Failure; Side Effect;Epidemiology.

here is a growing interest in developing “personal-

Abbreviations used in this paper: ADR, adverse drug reaction; ALF, acuteliver failure; APAP, acetaminophen; ALT, alanine aminotransferase; DILI,drug-induced liver injury; DILIN, Drug-Induced Liver Injury Network;GWAS, genome-wide association studies; HIV, human immunodeficiencyvirus; HMGB-1, high mobility group box protein 1; IL, interleukin; K-18,keratin 18; SDH, sorbitol dehydrogenase.

© 2014 by the AGA Institute0016-5085/$36.00

http://dx.doi.org/10.1053/j.gastro.2013.12.032

Tized medicine” wherein a specific drug or treatmentis offered to a given patient based on its predicted efficacyderived from host genomic data and/or tissue-specific geneexpression.1 Examples of personalized medicine includeinterleukin (IL)-28B genotyping before interferon therapyfor hepatitis C virus infection and use of CD117 expression ingastrointestinal stromal tumors to guide decisions regardingchemotherapy.2–4 Genomic and transcriptomic approachesmay also improve patient safety by avoiding use of poten-tially hazardous drugs in susceptible patients. For example,avoidance of abacavir therapy in HLA-B*5701–positivepatients with HIV infection has reduced the incidence of a

potentially severe hypersensitivity reaction from 15% tonearly 0%.5,6 Although severe adverse drug reactions (ADRs)such as drug-induced acute liver failure (ALF) are very un-common, the inability to reliably identify high-risk patientshas prevented many promising drugs from gaining regula-tory approval and led to the removal of other drugs from themarketplace.7 Currently, the role of host genetic, immuno-logic, and metabolic factors as well as drug and environ-mental influences on the pathogenesis of idiosyncraticdrug-induced liver injury (DILI) is poorly understood. Thisis in part attributable to the lack of reliable in vitro and in vivotest systems to study DILI and the difficulty in reliablydiagnosing and tracking patients with DILI.8,9 The aim of thisreview is to summarize recent advances in the epidemiologyand diagnosis of idiosyncratic DILI, the development of sen-sitive and specific biomarkers for DILI, and insights gleanedfrom pharmacogenetic studies. As our understanding of therole of the immune system in idiosyncratic DILI evolves,studies of other host factors such as the gut microbiome willhopefully further improve our understanding of the causesand mechanisms of idiosyncratic DILI.

Advances in the Epidemiology ofIdiosyncratic DILI

Intrinsic and “idiosyncratic” DILI are commonly believedto arise by different pathophysiologicalmechanisms. Intrinsichepatotoxins such as acetaminophen (APAP) are typicallydose dependent and have reproducible animal models thathelp inform our understanding of the pathways leading tohepatocyte injury.10 In contrast, most instances of DILI seenin clinical practice are termed “idiosyncratic” (ie, a mixture ofcharacteristics unique to that patient) because they are notclearly related to the dose, route, or duration of drugadministration (Figure 1). The aim of this review is to providean update on advances in research on idiosyncratic DILI.


Figure 1. Factors implicated in the pathogenesis of idiosyn-cratic DILI. (A) Drug factors have not been reliably associatedwith liver injury in preclinical test systems or in patients withDILI. However, drug-drug or drug-disease interactions couldalter the concentration of a drug or reactive metabolite at acellular level involved in the initiation, maintenance, or reso-lution of liver injury. (B) Clinical host risk factors such as age,body weight, and body mass index have only rarely beenimplicated in the pathogenesis of DILI. However, recent GWAShave shown consistent associations between various singlenucleotide polymorphisms in the HLA region and susceptibilityto idiosyncratic DILI. (C) The microenvironment and macro-environment vary greatly among patients receiving medica-tions. However, coffee, alcohol consumption, and diet havenot been identified as bona fide risk factors for DILI. The recentdevelopment of powerful transcriptomic, metabolomic, andmicrobiome methods may improve our understanding ofenvironmental factors in the pathogenesis of DILI usingadvanced bioinformatics and systems biology approaches.

April 2014 Drug-Induced Liver Injury 915

REVIEW

SAN

DPE

RSPE

CTIVES

Overall, DILI accounts for <1% of cases of acute liverinjury seen by most gastroenterologists in the UnitedStates.11,12 Nonetheless, idiosyncratic DILI is a leading causeof ALF in the United States and is likely underdiagnosedbecause of the need to exclude other more common causesof liver injury and show improvement after drug discon-tinuation or “dechallenge.”13 Furthermore, idiosyncraticDILI attributed to a specific drug may present with variablelaboratory, clinical, and histopathological features, making iteven more difficult to reliably diagnose and study (Table 1).Until an objective and reliable confirmatory test is devel-oped, idiosyncratic DILI will remain a “clinical diagnosis ofexclusion” that requires a high index of suspicion.11

Studies of the epidemiology of idiosyncratic DILI havelargely been retrospective case series with highly variableestimates of the incidence and natural history.14–16 Therecent adaptation of electronic medical records into routinemedical practice has created a unique opportunity to trackand study various rare ADRs.17,18 Identification of idiosyn-cratic cases of DILI from administrative databases usingInternational Classification of Diseases, Ninth Revision(ICD-9) diagnostic coding has proven to be labor intensive

with low sensitivity and specificity.14,19 However, recentstudies that use natural language processing algorithms thatcan search for key words in a text field such as “hepato-toxicity” or “toxic hepatitis” have shown improved sensi-tivity and specificity for DILI.19 In addition, the linking ofclinical, laboratory, and pathology databases with text-searching algorithms may allow for more real-time identi-fication of cases of idiosyncratic DILI.20,21

Registries of Idiosyncratic DILIIn 2004, the Drug-Induced Liver Injury Network (DILIN)

was established by the National Institutes of Health toimprove our understanding of the causes, mechanisms, andoutcomes of idiosyncratic DILI in adults and children.22

Similar multicenter networks have been established inSpain, Iceland, the United Kingdom, Europe, Japan, China, andKorea.23–27 These networks are leading efforts to developstandardized nomenclature, grading systems, and causalityassessment methods in research on DILI.9,28,29 Harmoniza-tion of the approach to DILI phenotyping and causalityassessment will hopefully provide an increased number ofcases of DILI for pooling in genetic association studies(Supplementary Table 1).28–31 In addition, the National In-stitutes of Health in conjunction with the National Library ofMedicine has developed a comprehensive, multilayered, andinteractive database of the published literature of humandrug hepatotoxicity.32 The LiverTox Web site (http://www.livertox.nih.gov) has concise overview sections on DILI phe-notypes, severity grading, and likelihood scales. In addition,chapters that summarize the clinical and laboratory featuresof DILI associated with more than 650 individual drugs areavailable, along with illustrative cases and annotated refer-ences with available hyperlinks to the full PubMed reference.The LiverTox Web site has already increased awareness ofDILI and will likely prove to be a valuable resource for basic,translational, and clinical research into the pathogenesis ofDILI for years to come.

Several reports of the etiologies and outcomes of idio-syncratic DILI from around the world have recently beenpublished (Table 2).24,25,27,33,34 The laboratory profile ofDILI at presentation is defined by the ratio of serum alanineaminotransferase (ALT) to alkaline phosphatase levels (ie,R value ¼ [ALT/upper limit of normal]/[alkaline phospha-tase /upper limit of normal]) and can be categorized ashepatocellular (R > 5), mixed (R ¼ 2–5), or cholestatic(R < 2). Similarities across the DILI cohorts include theproportion of female patients (49%–65%), the median ageof the subjects (48–55 years), and the proportion of patientswith acute hepatocellular injury at presentation (42%–58%). Although the DILIN and Spanish networks tend toenroll sicker patients who are more likely to be hospitalized,the proportion of patients who have died or required livertransplantation is remarkably similar. Antibiotics are themost commonly identified drug class leading to idiosyn-cratic DILI, but the specific implicated agents differ sub-stantially. In addition, DILIN has reported a significantproportion of cases attributed to herbal and dietary sup-plements, increasing from 7% in 2004–2005 to 20% in

http://www.livertox.nih.gov


Table 1.Clinicopathological Presentations of Idiosyncratic DILI

Phenotype Histological features (examples) Proposed mechanism

Acute fatty liver with lactic acidosis Microvesicular hepatic steatosis and/or other tissueinvolvement (didanosine, fialuridine)

Severe acute mitochondrial injury

Acute hepatic necrosis Collapse and necrosis of liver parenchyma (isoniazid, aspirin,niacin)

? Reactive metabolite and/or immuneactivation

Autoimmune-like hepatitis Plasma cells and interface hepatitis with detectableautoantibodies (nitrofurantoin, minocycline)

? Autoimmunity

Bland cholestasis Balloon hepatocytes with minimal inflammation (anabolicsteroids)

? Inhibition of bile salt export pump orother biliary transporters

Cholestatic hepatitis Balloon hepatocytes with inflammation, predominance ofelevated serum alkaline phosphatase levels (phenytoin,amoxicillin-clavulanate)

? Immune-mediated and/or reactivemetabolite

Fibrosis/cirrhosis Hepatic collagenization with minimal inflammation(methotrexate, amiodarone)

? Stellate cell activation/chronicendothelial cell injury

Immunoallergic hepatitis Skin rash, fever, eosinophilia (trimethoprim-sulfamethoxazole) Drug hypersensitivity/allergyNodular regeneration Microscopic or macroscopic liver nodules (azathioprine,

oxaliplatin)? Chronic injury to endothelial cells

Nonalcoholic fatty liver Macrosteatosis and microsteatosis, hepatocyte ballooning,and periportal inflammation (tamoxifen)

? Chronic mitochondrial injury oraltered lipid metabolism

Sinusoidal obstruction syndrome Inflammation with obliteration of central veins (busulfan) ? Severe, acute endothelial cell injuryVanishing bile duct syndrome Paucity of interlobular bile ducts (amoxicillin-clavulanate,

sulfonamides)? Immune-mediated cholangiocyte

injury

916 Fontana Gastroenterology Vol. 146, No. 4

REVIEWSAND

PERSPECTIVES

2010–2011.35 The most frequently implicated herbal anddietary supplements in DILIN are bodybuilding supple-ments and weight loss products that contain green teaextract.36,37

Table 2.Presenting Clinical Features and Outcomes in Prospec

Feature DILIN US (N ¼ 300)32 Spain (

Study design Multicenter (8 sites)2004–2008

Multicen1994–

Causality method Expert opinion RUCAMDuration of follow-up (mo) 6–24Mean age (y) 48Female 60Race

White 79Black 11Asian 4Other 6

Type of liver injuryHepatocellular 57Mixed/cholestatic 20/23 2

Jaundice 69Liver biopsy 50Hospitalized 60Died or underwent transplantation 10Chronic 14Suspected drugs

Antibiotics 45Psychotropic drugs 15Herbal and dietary supplements 9Hypolipidemic agents 3

NOTE. All values are percentages unless otherwise noted.RUCAM, Roussel Uclaf Causality Assessment Method; CIOMSNA, not available.

DILIN is prospectively following all study subjects for aminimum of 6 months after enrollment.22,33 A competingcause of liver injury has been identified in up to 15% ofpatients in DILIN, including previously unsuspected acute

tive Studies of DILI

N ¼ 446)23Japan

(N ¼ 1676)26Iceland

(N ¼ 96)24France

(N ¼ 34)33

ter (32 sites)2005

Multicenter1997–2006

Population based2010–2012

Population based1997–2001

RUCAM RUCAM CIOMS3 NA 3 3

53 55 55 5549 57 56 65

100 0 100 1000

1000

58 59 42 532/20 20/21 26/32 26/2171 NA 27 2925 NA 11 NA53 NA 23 127 3.7 11 6

10 8.4 7 NA

32 14 37 2517 10 7 220 17.1 16 03 0 3 13

, Council for International Organizations of Medical Sciences;


REVIEW

SAN

DPE

RSPE

CTIVES

hepatitis C virus and hepatitis E virus infection.38–40 DILINhas also confirmed the association of high serum amino-transferase levels at presentation in jaundiced patients withDILI with a greater likelihood of early death/transplant (ie,Hy’s law ¼ serum ALT level >5 times the upper limit ofnormal with elevated total bilirubin level).41–43 However,the observation that nearly 50% of the deaths of patients inDILIN were attributed to an underlying nonhepatic medicalcondition such as malignancy or heart failure is also animportant observation.41 In addition, DILIN and othergroups have confirmed the greater likelihood of patientswith a severe cholestatic liver injury developing chronicliver injury during follow-up.42,43 These findings highlightthe need for careful clinical and laboratory monitoring ofpatients with DILI and also provide the rationale for aclinical trial to improve patient outcomes.44–46

Figure 2. Potential mechanism(s) involved in the pathogenesisof idiosyncratic DILI. It is plausible that several cellular mech-anismsmay be involved in the pathogenesis of DILI. Onemajorhypothesis of the pathogenesis of idiosyncratic DILI is theinadvertent generation of a reactive metabolite or parent drug-protein complex that can directly or indirectly mediate damageto intracellular proteins and/or organelles, resulting in theinitiation of “danger” signals. Patients with DILI may beuniquely susceptible to developing liver injury from reduceddetoxification, adaptation, or tolerance pathways that wouldnormally rescue damaged hepatocytes or an increased likeli-hood of “bioactivation” pathways. In the hapten hypothesis,the drug-protein or metabolite-protein adduct leads to inad-vertent activation of the adaptive immune system. Alterna-tively, in nonimmune mechanisms, damage-associatedmolecular pathway proteins such as HMGB-1, heat shockproteins, or cellular DNA released from necrotic cells lead tothe recruitment of localized tissue injury. Cytokines, chemo-kines, and costimulatory molecules may play an integral role inmacrophage activation and magnification of the damage-associated molecular pattern response via their impact ondrug-metabolizing enzyme activity, the density of HLA mole-cules on antigen-presenting cells, the ability of the presentedantigens to activate T cells, and the ability of activated T cellsto cause hepatocyte death. ER, endoplasmic reticulum; BSEP,bile salt export pump.

Epidemiology of Idiosyncratic DILI:Mechanistic Insights

Although drugs undergo extensive safety testing inin vitro test systems and various animal species beforeclinical development, this testing frequently fails to identifypotentially hepatotoxic drugs. In addition, the low overallincidence of idiosyncratic DILI with most available drugs ofonly 1 in 10,000 to 1 in 100,000 patient-years preventsmost hepatotoxic drugs from being identified in clinicaltrials. Recent studies have suggested that drugs that areadministered at a daily dose >50 to 100 mg/day withgreater lipophilicity are more prone to cause DILI comparedwith agents given at a lower daily dose with less lip-ophilicity.47,48 Possible explanations for these simple butintriguing observations include the fact that drugs given inhigh daily doses may lead to higher intrahepatic levels of theparent drug or a metabolite involved in the pathogenesis ofDILI. In addition, lipophilic drugs require greater meta-bolism to be eliminated from the body, which may increasetheir likelihood of causing liver damage. It is also possiblethat extensively metabolized drugs may induce covalentlybound haptens, which can elicit an adverse adaptive im-mune response in a genetically susceptible patient.49,50

However, the daily dose of a medication, its lipophilicity,and the extent of hepatic metabolism are inadequate fea-tures to reliably predict the risk of DILI from individualdrugs. In addition, studies of chemoinformatics have failedto identify chemical moieties in drugs that are more proneto lead to idiosyncratic hepatotoxicity.51,52

In Western patients with DILI, antibacterial antibioticsand psychoactive drugs are the most frequently implicatedtherapeutic drug classes (Table 2). Although antibacterialantibiotics are used by millions of Americans each day, theyare usually only taken for a few days or weeks. This sug-gests that an active or recent infection may increase thesusceptibility for an ADR via aberrant innate or adaptiveimmune pathways or alterations in the gut microbiome insusceptible patients.46 In support of the “danger hypothesis”(Figure 2), administration of tumor necrosis factor to he-patocyte cultures as well as in animal models of DILI can

potentiate the hepatotoxicity of multiple drugs.53 In addi-tion, the greater frequency of DILI in immunosuppressedliver transplant recipients (1 in 100 patient-years)compared with the general population (1 in 10,000patient-years) suggests that specific patient populationsmay be at increased risk for idiosyncratic DILI.21 However, aplausible explanation as to why the majority of patients withan acute or chronic illness who receive multiple medica-tions, including antibiotics, do not develop DILI remainsunclear.

A recent population-based study from Iceland providesthe best estimates of the incidence of idiosyncratic DILI in a


REVIEWSAND

PERSPECTIVES

Western country.25 In this study, all suspected episodes ofDILI regardless of severity among the 250,000 inhabitantsof Iceland over a 2-year period were reviewed. The crudeoverall annual incidence of idiosyncratic DILI was 19.1 casesper 100,000 population, which is remarkably similar to therate of 13.9 cases per 100,000 reported from France.34

These rates are likely higher than prior estimates becauseof the prospective nature of these studies, the inclusion ofsubclinical cases, and the ability to systematically canvassan entire population. Although no differences in sex werenoted, there was a strong association of the risk of DILI withpatient age, which varied from 9 per 100,000 (age 15–29years) to 41 per 100,000 (age >70 years). Althoughamoxicillin-clavulanate was the most frequently implicateddrug, the estimate of the risk of DILI was 1 per 2350 users;this was substantially lower than that observed withazathioprine (1 per 133 users) and infliximab (1 per 148users). These data lend support to the notion that somedrugs used in clinical practice are intrinsically more hepa-totoxic than others. Although DILIN is not a population-based study, amoxicillin-clavulanate is the most commonlyimplicated agent (>120 cases in DILIN) but is prescribed tomore than 70 million Americans each year. In contrast,isoniazid (>50 cases in DILIN and <200,000 prescriptionsper year) and nitrofurantoin (>50 cases of DILIN and<500,000 prescriptions each year) appear to be over-represented in the DILIN and ALF Study Group databases.44

Similarly, the infrequent reporting of statins, beta-blockers,and calcium channel blockers in DILI registries despite theirwidespread use suggests that these drug classes are prob-ably not as intrinsically hepatotoxic as previously sug-gested.54,55 The LiverTox Web site has developed acategorization of the likelihood of DILI from a specific agentbased on the frequency of bona fide cases in the literaturethat varies from category A (“Well known with > 50 casesdescribed,” such as amoxicillin-clavulanate and phenytoin)to category E (“Despite extensive use, no evidence of liverinjury,” such as felodipine or propranolol).

Inferences of the Pathogenesis ofDILI From Liver Histology

DILI is a well-known imitator of most forms of acute andchronic liver injury (Table 1). Furthermore, some drugs,such as valproate, may cause differing patterns of liverinjury in individual patients that can vary from hepaticsteatosis to massive necrosis.56 A careful review of liverhistopathology in 249 patients in DILIN indicated that 5liver injury patterns accounted for 83% of the cases.57 Inaddition, poorer outcomes were associated with higher de-grees of necrosis, microvesicular steatosis, and a ductularreaction, whereas subjects with intrahepatic eosinophilsand/or granulomas tended to fare better. The prognosticuse of these histological features is consistent with priorreports and suggests that peripheral or intrahepatic eosin-ophilia is protective.58,59

Other studies have suggested potential histopathologicaldifferences in patients with idiopathic and drug-induced

autoimmune hepatitis.60 Similarly, studies of lymphocytesubsets in liver tissue samples have shown that subjectswith DILI are less likely to have natural killer cells, CD4þ

T-helper cells, and B cells compared with patients withacute viral hepatitis (Foureau et al, manuscript in submis-sion). The predominance of CD8þ cytotoxic T cells in thelivers of patients with idiosyncratic DILI is consistent withthe hypothesis that the intrahepatic generation of neo-antigens from the drug or its metabolite may lead to therecruitment and activation of T lymphocytes that caninitiate or perpetuate the liver injury.61 In contrast,neutrophil and macrophage infiltration into the liver isbelieved to be a late event in patients with APAP overdoseand treated with other drugs that directly injure pericentralhepatocytes.62 Similarly, patients with bland cholestasis dueto use of anabolic steroids may develop liver injury throughdirect toxicity to bile salt or other drug transporters.63

Drugs that disrupt mitochondrial function via depletionof mitochondrial DNA or triggering of outer mitochondrialmembrane permeabilization can lead to variable patterns ofliver tissue injury. Acute impairment of mitochondrialfunction is a distinctive and frequently dramatic clinicalsyndrome characterized by small droplets of fat (micro-vesicular) that accumulate in the hepatocyte cytosol fromimpaired beta oxidation of fatty acids as reported withtetracycline and valproic acid.56,64 In contrast, drugs thatlead to partial but chronic depletion of mitochondrialfunction can lead to accumulation of large fat droplets(macrovesicular) that are often eccentrically located in thecell. Drugs associated with the latter subacute or chronicphenotype include tamoxifen and various dideoxynucleo-side analogues used to treat patients with human immu-nodeficiency virus (HIV) infection.65,66 Other drugs, such asoxaliplatin, can damage liver endothelial cells and lead tonodular regenerative hyperplasia and portal hypertensionin some patients.67,68

In Vitro Test Systems to Study theMechanisms of DILI

Extensive in vitro and animal toxicology testing isrequired in drug development, but these approaches haveconsistently failed to predict the development of variousADRs, including DILI. In 2007, the US National Academy ofSciences suggested greater use of human biological testsystems to improve detection of and provide mechanisticinsight into human ADRs.69 In addition, the need forpathway analysis of toxicological mechanisms was empha-sized using metabolomic, transcriptomic, and genomic ap-proaches. Currently, potential drug-drug interactions can bereliably predicted from in vitro inhibition of known phase 1and 2 polymorphisms in cultured human hepatocytes.Recently, several chimeric mice strains expressing highlydifferentiated human hepatocytes that can be used in cellculture experiments as well as human hepatoma cell lineswith highly differentiated cellular function have beendeveloped.70,71 In addition, the discovery that human andmouse fibroblasts and somatic cells can be preprogrammed


REVIEW

SAN

DPE

RSPE

CTIVES

into inducible pluripotent stem cells has generated a greatdeal of interest in the use of this novel technology to studydrug hepatotoxicity, liver regeneration, and various geneticdiseases.72–74 Derivation of hepatocytes from induciblepluripotent stem cells of patients who develop DILI couldallow for a long-term and renewable source of cells to studythe mechanisms of idiosyncratic DILI in affected patientscompared with treated controls.75

Advances in the development of in vitro culture methodsof T cells implicated in various hypersensitivity drug re-actions have also recently been reported, including tech-niques to expand the number of available T cells derived fromthe peripheral blood of patients with allergic drug re-actions.76 Because antigens are processed and presented onHLA molecules of antigen-presenting cells to the T-cell re-ceptor on T cells, additional studies of T-cell physiology cannow be conducted (Figure 3). Patients with known hyper-sensitivity to piperacillin, which by itself is unlikely to stim-ulate an immune response, have shown alterations in thebinding site of albumin that can lead to highly immunogenicdrug metabolite–albumin conjugates.77,78 In addition, modi-fications of the lymphocyte proliferation assay have beendeveloped with readouts of T-cell cytokine expression andmicroarrays rather than pure cellular proliferation.79

Whether these modified lymphocyte transformation assays

Figure 3. Proposed role of aberrant immunity in the pathogeneserum proteins under normal physiological circumstances fordrug-protein conjugate will not elicit a host immune response.alleles that are ubiquitously expressed may be uniquely prediprotein conjugate activate an antigen presenting cell such as adrug-protein conjugates in these subjects can then inadvertentlend-organ damage. MHC, major histocompatibility complex.

will improve drug-specific diagnoses or provide mechanisticinsights into the pathogenesis of DILI remains to bedetermined.

Biomarkers for the Pathogenesis ofDILI and Outcomes

Biomarkers are analytes from blood, urine, or otherbiological samples that may provide insight into theseverity, cause, or outcome of an episode of DILI. In addi-tion, biomarkers may improve the speed or accuracy ofdiagnosing DILI.80 Ideally, direct examination of liver tissuewould provide the greatest insight into the pathophysio-logical steps involved in idiosyncratic DILI but is impracticalfor obvious reasons. The serum biomarkers most commonlyused to detect and manage most forms of acute and chronicliver injury are serum ALT, alkaline phosphatase, and totalbilirubin levels. Serum ALT is more liver specific than serumaspartate aminotransferase but is not etiology specific, andlevels can also be elevated in subjects with extensive hepaticglycogen, hepatocyte autophagy, and hepatic steatosis.81,82

Furthermore, monitoring for elevated serum ALT levels insubjects receiving a potentially hepatotoxic drug such asisoniazid has consistently failed to identify subjects at riskfor developing DILI compared with the larger group of

sis of DILI. Drugs are small molecules capable of binding totransport, metabolism, and elimination. In most instances, aHowever, a minority of individuals with specific class II HLAsposed to have the native drug-protein or drug-metabolite-dendritic cell or macrophage. The processing and handling ofy activate T-cell receptors, which may proliferate and mediate


REVIEWSAND

PERSPECTIVES

patients who “adapt” with normalization of ALT levelsduring continued treatment.83,84 Similarly, serum alkalinephosphatase levels are not liver specific and may be spuri-ously elevated in other disease states.80 Total bilirubinlevels are an insensitive marker for most forms of liverdisease, increasing only when there has been extensive liverdamage or via direct inhibition of biliary transporters.Fractionation of total bilirubin levels can help excludebenign elevations due to intravascular hemolysis and ge-netic polymorphisms in uridine glucuronyl transferase ac-tivity (Gilbert syndrome), which are present in 1% to 5% ofthe general population.

New Serum BiomarkersOngoing efforts to identify new biomarkers for DILI

include the large-scale initiative of the Safer and FasterEvidence-based Translation (SAFE-T) Consortium inEurope.85 Proposed biomarker candidates include serummarkers of liver injury (sorbitol dehydrogenase [SDH],glutathione S-transferase a) and mitochondrial dysfunctionfrom disrupted hepatocytes (glutamate dehydrogenase)(Table 3). In addition, the discovery of circulating microRNAs(miRNAs) in the serum has shown the novel tissue specificityof miR-122 for liver injury. Assessment of full-length keratin-18 (K-18) and highmobility group box protein 1 (HMGB-1) inthe serum has been shown to be a sensitive biomarker fornecrotic cell death, while caspase cleaved K-18 fragments arenoted in subjectswith ongoing apoptosis. However, neither ofthese markers are liver disease specific. M-30 is a serumprotein that selectively recognizes caspase cleaved neo-epitopes of K-18 released from hepatocytes undergoing

Table 3.Proposed Biomarkers for DILI

Analyte Source and significance

Liver injury markersSDH Hepatocyte-specific injuryGlutathione S-transferase a Centrilobular liver damage and kidney

Bile acids Synthesized in the liver; disruption of hexcretion

Glutamate dehydrogenase Mitochondrial disruptionSerum cytokine profiles Intra and extrahepatic originmiRNAsmiR-122miR-192

Liver-specific release from damagedhepatocytes

Mechanistic biomarkersHMGB-1 Necrosis markerAcetylated HMGB-1 Innate immune activation markerCytokeratin-18 fragments Marker of caspase cleaved proteins in

cell deathM-30 Apoptosis markerM-65 Total apoptosis and necrosis markerSerum Cys-APAP adducts Marker of APAP overdose

MetabolomicsUrine or serum metabolome Amount and type of endogenous subs

apoptotic death, while serum M-65 reflects total hepatocytedeath (apoptosis and necrosis).86,87 An index of serum M-30levels in combination with other laboratory parameters wasrecently shown to be superior to the King’s College criteriaand Model for End-Stage Liver Disease (MELD) score inpredicting spontaneous survival in ALF.88,89

Heparins are a widely used class of biological agents thatare frequently associated with mild and nonprogressive el-evations of serum ALT levels but rarely, if ever, lead toclinically significant liver injury.90 A recent study of 48healthy men given a heparin formulation for 5 days showedasymptomatic elevations in serum aspartate aminotrans-ferase and ALT levels in 90% of the treated patients.91 Inaddition, significant elevations in the levels of serum SDH,glutamate dehydrogenase, miR-122, and HMGB-1 in both itsnative and acetylated form were noted.92 However, serumK-18 fragment levels indicative of apoptosis remainedunchanged. These data suggest that heparins cause a self-limited and mild hepatocyte necrosis with secondaryactivation of the innate immune system. HMGB-1, a damage-associated molecular pattern protein, can bind to Toll-likereceptor 4 and initiate an innate immune response at itssite of release in the liver.93 The detection of acetylatedHMGB-1 is suggestive of activated innate immune cells,which may be linked to tissue repair. However, the reasonfor a lack of progressive liver injury despite continueddosing with heparins remains unclear.94

Biomarkers for APAP HepatotoxicityAnother study that used therapeutic doses of APAP for 5

days in healthy volunteers showed that nearly 30% of the

Test performance to date

Earlier marker of acute liver injurydamage Early marker of acute liver (serum) and kidney

(urine) injuryepatic More sensitive than bilirubin for excretory abnormalities

Increased in some patients with chronic liver diseaseAssociated with prognosis

Released into plasma with acute and chronic injury;validation ongoing

Not liver specificAcetylation requires mass spectroscopy for detection

apoptotic Not liver specific

Ability to distinguish therapeutic dosing from drugoverdose being tested

tances Exploratory with substantial drug, dietary, environmental,and microbiome influences; bioinformatics for datareduction ongoing


REVIEW

SAN

DPE

RSPE

CTIVES

treated patients experienced an increase in serum ALTlevels.95 There was also an increase in serum SDH, gluta-mate dehydrogenase, and HMGB-1 levels.95,96 A panel ofthese early biomarkers for DILI was recently tested in pa-tients presenting to the hospital after an APAP overdosewith initially normal serum aminotransferase levels.97 Acombination of miR-122, HMGB-1, and K-18 levels identifiedthe development of liver injury with a higher degree of ac-curacy than the initial serum ALT level, internationalnormalized ratio, and plasma APAP level. In addition, acet-ylated HMGB-1 levels reflective of immune activation werenoted later in the course of APAP overdose and associatedwith a poorer outcome in an independent cohort of 78patients.98

Several studies have shown that detection of APAP-protein adducts in serum can confirm a diagnosis of APAPhepatotoxicity. The premise of this test is based on theformation of covalent adducts between intrahepatocyteproteins and the major reactive metabolite of APAP formedin the liver, N-acetyl-p-benzoquinone imine. Because serumAPAP-protein adducts have a longer half-life in serumcompared with the parent drug or its metabolites, this testmay be particularly informative in patients presenting lateafter an unintentional APAP overdose.99 In addition, 15% ofpatients with indeterminate ALF also had detectable APAP-protein adducts.100 However, low levels of APAP-proteinadducts have also been detected in the blood of patientstaking therapeutic doses of APAP over 5 days.101 Thedevelopment of tests to detect circulating drug-protein ad-ducts in other patients with idiosyncratic DILI would bedesirable. However, idiosyncratic DILI due to most drugsoccurs at lower daily doses, and toxic metabolites impli-cated in the pathogenesis of DILI are believed to represent asmaller fraction of the total testable drug in the serumcompared with APAP, making it technically difficult toidentify and quantify drug-protein adducts.102

Serum ProteomicsProteomic studies that simultaneously identify and

quantify thousands of proteins use high-pressure liquidchromatography and tandem mass spectroscopy. Withadvanced bioinformatics software, the protein(s) and/orpathways involved in the pathogenesis of DILI can then bestudied. The serum proteomic profiles of 74 patients inDILIN who had a baseline sample collected within 2 weeksof onset of DILI and 40 healthy controls were recentlyanalyzed using a label-free, mass spectrometry quantitativeapproach.103 Several proteins were expressed at a higherlevel in subjects with hepatocellular versus cholestatic DILI,including fructose-bisphosphate aldolase B. This hepaticenzyme correlated with serum ALT and aspartate amino-transferase levels at baseline and returned to normal duringfollow-up. Of note, autoantibodies to this protein havepreviously been reported in patients with troglitazonehepatotoxicity.104 In addition, elevations in levels of apoli-poprotein E, an abundant lipoprotein of triglyceride-richchylomicrons, had the greatest ability to distinguish pa-tients with DILI from controls. A longitudinal analysis of

21 patients with baseline and 6-month samples showed thatexpression of 53 priority 1 proteins either increased ordecreased over time, including components of inflamma-tory, immune system activation, and several hepatotoxicity-specific pathways. A proteomics platform in combinationwith metabolomics was recently used to distinguish patientswho developed ximelagatran hepatotoxicity from unaffectedcontrols.105 In that study, biomarkers that predicted pa-tients at risk for elevation of ALT levels included apolipo-proteins A-II, A-IV, and E. These provocative data suggestthat further proteomic studies are indicated particularly insamples obtained before or shortly after the onset of DILI.

Serum Cytokines and ChemokinesSerum cytokine and chemokine levels may also prove to

be useful diagnostic or prognostic biomarkers for DILI.106

DILIN recently completed an analysis of 27 immune analy-tes in 78 subjects who were enrolled within 2 weeks ofonset of DILI and 6 months after enrollment.107 Disparatepatterns of immune responses were evident, and low valuesof IL-9, IL-17, platelet-derived growth factor BB, andRANTES combined with serum albumin were predictive ofearly death. Lower levels of the cytokines associated withinnate immunity (ie, low IL-9 and IL-17 levels) were asso-ciated with a poorer prognosis, suggesting a role for thesebiomarkers and immune pathways in the pathogenesis ofDILI. These data are consistent with recent studies showinga role of the Th17 adaptive immunity pathway in thepathogenesis of idiosyncratic DILI.108 However, a studyfrom the ALF Study Group failed to show a difference incirculating IL-17 levels of patients with ALF due to idio-syncratic DILI compared with those with APAP overdose.109

Transcriptomics and MetabolomicsTranscriptomics represents the detection and quantifi-

cation of transcribed genes or messenger RNA in the serumand other fluids. In contrast, metabolomics represents thestudy of endogenous small molecules and metabolites in theserum and urine that reflect normal physiological anddiseased states. Transcriptomic studies require the collec-tion of RNA from specialized tubes and then detection ofrelative gene expression levels using oligo array chips andquantitation of messenger RNA expression using polymer-ase chain reaction. In metabolomic studies, samples areanalyzed using nuclear magnetic resonance spectroscopy ormass spectral techniques to simultaneously detect andquantify thousands of endogenous metabolites.

A recent study of 8 hepatotoxicants administered to ratsshowed that transcriptomes present in peripheral bloodcorrelated with direct measures of these genes in theliver.110 In fact, these studies suggested that peripheralblood transcriptomic data might be more sensitive to liverinjury than traditional liver injury tests such as serum ALTand that unique signatures for individual drugs, includingAPAP hepatotoxicity, could be determined.111 Furthermore,human whole blood transcriptome data from patients withovert APAP overdose could differentiate patients with


REVIEWSAND

PERSPECTIVES

toxicity from those without.111 However, marked down-regulation of genes involved in oxidative phosphorylationand mitochondrial function as well as metabolomic changesin the blood and urine were observed in patients receiving atherapeutic dose of APAP compared with those with overtAPAP hepatotoxicity.112,113

Advantages of metabolomic and transcriptomic studiesin biomarker discovery include the large amount of datathat can be generated and quantified from small samplealiquots. Furthermore, pathways analysis can be conductedusing bioinformatics platforms. Finally, the relationshipbetween changes in gene expression from the tran-scriptomic platforms can be integrated with the physiolog-ical changes detected from metabolomics. However, thetranscriptome in whole blood may chiefly reflect the tran-scriptional activity of lymphocytes, and it appears that theseapproaches may not be able to distinguish the pharmaco-logical effect of a drug like APAP from a toxic dose.107 Inaddition, serum and urine metabolomes in patients can beinfluenced by dietary and environmental factors, includingthe gut microbiome.114 Therefore, further studies of thedrivers of interindividual differences in the metabolome andtranscriptome are needed as part of a systems biologyapproach to DILI susceptibility.115,116

miRNAsmiRNAs are small regulatory, noncoding RNAs that are 18

to 25 nucleotides in length and can be detected in micro-vesicles in the serum. Although a number of miRNAs arewidely expressed, certain miRNAs appear to be tissue spe-cific. Liver-specific miRNAs include miR-122, miR-21, andmiR-192, and these molecules are believed to repress a set oftarget messenger RNA and thereby regulate specific cellularproteins and cell phenotypes. Liver-derivedmiRNAsmay be ahighly sensitive and specific biomarker of APAP hepatotox-icity that parallel ALT levels but increase earlier in the courseof liver injury.117 In addition, they appear to have prognosticsignificance for patients with APAP overdose in need oftransplant versus those more likely to survive.118,119 How-ever, serummiR-122 levels have not been directly correlatedto hepatic expression in humans. Nonetheless, their shorthalf-life, liver tissue specificity, and quantifiabilitymake themattractive biomarkers for severe acute liver injury.

Pharmacogenetic Studies ofIdiosyncratic DILI

Because of its low incidence in the general population, agenetic variation in host receptors, immune response, andmetabolic pathways have been implicated in the pathogen-esis of idiosyncratic DILI. Previous genetic associationstudies have largely focused on candidate genes involved inthe uptake, metabolism, transport, or detoxification of adrug that can be used to predict drug pharmacokinetic andpharmacodynamic parameters. For example, reduced activ-ity in the NAT2 gene and increased activity in CYP2E-mediated oxidative metabolism have been implicated inisoniazid hepatotoxicity.120–122 However, these hypothesis-

driven, biologically plausible approaches have yielded onlyweak associations that are frequently not replicated in in-dependent cohorts.123

An alternative approach to identifying genetic associa-tions is to scan the entire human genome in affected casesand population controls without a specific a priori hypoth-esis. In most genome-wide association studies (GWAS), thefrequency of single nucleotide polymorphisms is at least 1%to 5% in the general population. This “discovery” platformallows for the unbiased detection of up to 1 million singlenucleotide polymorphisms that may associate with the dis-ease trait of interest. The first successful GWAS in DILIidentified a very strong association between flucloxacillin-induced liver injury and the HLA-B*5701 allele on chro-mosome 6.124 Other GWAS have identified additional HLAalleles associated with lumiracoxib, ximelagatran, andlapatinib hepatotoxicity (Supplementary Table 2).125–127

Genetic Studies of FlucloxacillinHepatotoxicity

Flucloxacillin is a parenterally administered b-lactamantibiotic that is widely used in Europe and other countriesto treat staphylococcal and streptococcal infections. Flu-cloxacillin is associated with a rare but potentially severecholestatic hepatitis that is more common in female pa-tients, in elderly patients, and after prolonged courses oftreatment.128 Daly et al performed a GWAS in 51 patientswith bona fide flucloxacillin cholestasis compared with 64ethnically matched treated controls as well as populationcontrols.124 In the initial and validation cohorts, possessionof the HLA-B*5701 allele was associated with an 80-foldincreased risk of developing DILI (P ¼ 9 � 10�19). ThisHLA allele has also been associated with abacavir hyper-sensitivity in patients with HIV infection.5 Although this isone of the strongest genetic associations ever reported froma GWAS, the high frequency of HLA-B*5701 in the generalpopulation (6%–8%) coupled with the low incidence of DILIin treated patients (1 in 10,000) leads to a low positivepredictive value (0.12%) for identifying patients at risk forflucloxacillin cholestasis but a high negative predictive value(99.99%).129 Therefore, testing for HLA-B*5701 may helpdiagnose flucloxacillin cholestasis in exposed patients whodevelop jaundice, but use of the drug should not be withheldin HLA-B*5701–positive patients.

Flucloxacillin is also an agonist of the human pregnane Xreceptor, and further studies have shown that patients withflucloxacillin cholestasis are more likely to have poly-morphisms in the pregnane X receptor promoter region.130

Other studies have shown an important role for the adaptiveimmune system, with T-cell clones of afflicted patients andpreviously unexposed HLA-B*5701–positive patientsshowing increased reactivity to flucloxacillin-albumin con-jugates in a dose-dependent manner.131 These studies alsoshowed that the reactive T-cell clones from afflicted patientshad cell surface markers that are associated with hepaticlocalization. These data confirm that flucloxacillin-proteinbinding is critical for the formation of functional T-cell


REVIEW

SAN

DPE

RSPE

CTIVES

antigens. This is in contrast to recent mechanistic studies ofabacavir and carbamazepine hypersensitivity reactionswherein the native drug itself has been shown to bind in theantigen-binding cleft of an HLA molecule and directlystimulate an immune response.132 Additional studies tofurther understand how flucloxacillin mediates liver damageand the role of hapten formation versus direct stimulation ofT cells are ongoing (Figure 3).133

Genetic Studies in Pooled PatientsWith Idiosyncratic DILI

DILIN and other groups have been collecting DNA frompatients with idiosyncratic DILI for pooled pharmacogeneticstudies.23 Hypothesizing that susceptibility to DILI may beshared across multiple drugs, a GWAS was recently under-taken in 783 white patients who experienced DILI frommorethan 200 implicated agents.134,135 Unfortunately, no genome-wide significant associations were noted, and further strati-fication of cases according to clinical phenotypes such asinjury pattern, latency, severity, drug class, and patient agedid not reveal any significant associations. The lack of GWASfindings in the pooled DILIN cases supports the notion thatgenetic determinants of risk of idiosyncratic DILI may belargely drug specific or due to rarer genetic variants notassessed on the GWAS chip. Going forward, newer tech-niques, including exome arrays that can assess for functionalgenetic variants present in 1 in 1000 to 1 in 5000 patients atmore than 250,000 loci, are being undertaken. In addition,improvements in the speed, accuracy, and costs of wholeexome and whole genome sequencing now allow for a morein-depth search of causal variants from smaller samples ofwell-phenotyped, high-causality cases attributed to a singledrug.136 The role of DNA methylation, copy number variants,and epigenetics in most forms of acute liver injury is largelyunknown but also worthy of further study.137

Mechanistic Inferences Into thePathogenesis of DILI

The strong and consistent association of susceptibility toDILI with various single nucleotide polymorphisms in theHLA region suggests that the host immune response plays akey role in the pathogenesis of DILI. HLAs are highly poly-morphic proteins that are designed to initiate immunity bypresenting pathogen-derived peptides to T cells. Poly-morphisms in HLA genes mostly map to the antigen-bindingcleft, which allows diversification of the repertoire of self-derived and pathogen-derived peptide antigens to bepresented to T cells.138 A growing number of other immu-nologically based ADRs, including dermatologic reactionsand idiosyncratic DILI, are also associating with various HLAalleles.139 In most of these instances, the implicated drugdoes not directly bind to the antigen-binding cleft of theHLA�molecule. Rather, a series of drug-protein modificationsteps or conversion of the drug to an intermediate or reactivemetabolite is required to form an immunogenic hapten.However, it is likely that other host genetic or intracellular

pathways may also be required for an ADR to develop.Furthermore, becausemany susceptible patients with a givenHLA haplotype do not develop DILI or other ADRs on drugexposure, the role of other intracellular “bioactivation” and“detoxification” pathways that may allow adaptation to occurneed to be evaluated (Figure 2). In addition, because HLApolymorphisms are ethnically restricted, the absence of agenetic association of susceptibility to DILI in one patientpopulation will not preclude a positive association in anothergroup, as recently noted in differing HLA susceptibility allelesin Han Chinese and European subjects to carbamazepinehypersensitivity reactions.140,141

The potential for the gut microbiome to affect suscepti-bility and outcome to DILI is another intriguing hypothesisworthy of further study. Early studies have shown interestingalterations in the gut microbiome of patients with obesitycompared with nonobese controls and in animal models offatty liver.142,143 Recent studies of complex biliary tract dis-orders also show that variance in microbiome content mayexceed that explained by genomic variation.144 There havebeen no studies of the gut microbiome in patients withidiosyncratic DILI, but there are interesting animal datashowing a significant impact of the gut microbiome inmediating melamine-related kidney injury.145,146

Future Directions in Researchon DILI

Over the next 5 to 10 years, additional studies of hostgenetic polymorphisms and susceptibility to idiosyncraticDILI attributed to individual agents will be completed usingnext-generation sequencing. To conduct such studies, DNAsamples collected from bona fide cases of DILI in whichcompeting viral, immunologic, and metabolic causes of liverdisease have been definitively excluded are needed, as wellas validation samples from independent cohorts. It is hopedthat mining of electronic medical records with natural lan-guage processing algorithms will improve the speed andaccuracy of the acquisition of cases of DILI and informpharmacoepidemiological studies regarding the causes ofDILI in a given population. In addition, continued effortsfrom multicenter research networks such as DILIN will helpprovide biological samples for mechanistic studies.Improved causality assessment tools, case definitions, andfurther development of a web-based portal of human hep-atotoxicity such as LiverTox will also be essential. Finally,the integration of data from divergent research platforms(ie, proteomics, transcriptomics, metabolomics, genomics)using a systems biology approach as well as data derivedfrom improved in vitro and in vivo test systems may providean unprecedented opportunity to study human drug meta-bolism and idiosyncratic DILI.

Supplementary MaterialNote: To access the supplementary material accompanyingthis article, visit the online version of Gastroenterology atwww.gastrojournal.org and at http://dx.doi.org/10.1053/j.gastro.2013.12.032.

http://www.gastrojournal.org




REVIEWSAND

PERSPECTIVES

References
1.Guttmacher AE, Collins FS. Realizing the promise ofgenomics in biomedical research. JAMA 2005;294:1399–1402.
2.Ge D, Fellay J, Thompson AJ, et al. Genetic variation inIL28B predicts hepatitis C treatment-induced viralclearance. Nature 2009;461:399–401.

3. Thompson AJ, Muir AJ, Sulkowski MS, et al. Interleukin-28B polymorphism improves viral kinetics and is thestrongest pretreatment predictor of sustained virologicresponse in genotype 1 hepatitis C virus. Gastroenter-ology 2010;139:120–129.e18.

4.Demetri GD, vonMehrenM, Blanke CD, et al. Efficacy andsafety of imatinib mesylate in advanced gastrointestinalstromal tumors. N Engl J Med 2002;347:472–480.

5.Mallal S, Nolan D, Witt C, et al. Association betweenpresence of HLA-B*5701, HLA-DR7, and HLA-DQ3 andhypersensitivity to HIV-1 reverse-transcriptase inhibitorabacavir. Lancet 2002;359:727–732.

6.Zucman D, deTruchis P, Majerholc C, et al. Prospectivescreening for human leukocyte antigen-B*5701 avoidsabacavir hypersensitivity reaction in the ethnically mixedFrench HIV population. J Acquir Immune Defic Syndr2007;45:1–3.

7.Watkins PB. Drug safety sciences and the bottleneck indrugdevelopment. ClinPharmacol Ther 2011;89:788–790.

8.Watkins P, Seeff LB. Drug-induced liver injury: summaryof a single topic clinical research conference. Hepatology2006;43:618–631.

9. Fontana RJ, Seeff LB, Andrade RJ, et al. Standardizationof nomenclature and causality assessment in drug-induced liver injury: summary of a clinical researchworkshop. Hepatology 2010;52:730–742.

10.Roth RA, Ganey PE. Intrinsic versus idiosyncratic drug-induced hepatotoxicity: two villains or one?J Pharmacol Exp Ther 2010;332:692–697.

11.Fontana RJ. Approaches to the study of drug-inducedliver injury. Clin Pharmacol Ther 2010;88:416–419.

12.Galan MV, Potts JA, Silverman AL, et al. The burden ofacute non-fulminant drug-induced hepatitis in a UnitedStates tertiary care referral center. J Clin Gastroenterol2005;39:64–67.

13.Ostapowicz G, Fontana RJ, Schiodt FV, et al. Results of aprospective study of acute liver failure at 17 tertiary carecenters in the United States. Ann Intern Med 2002;137:947–954.

14.Vuppalanchi R, Liangpunsakul S, Chalasani N. Etiologyof new-onset jaundice: how often is it caused by idio-syncratic DILI in the United States? Am J Gastroenterol2007;102:558–562.

15.Meier Y, Cavallaro M, Roos M, et al. Incidence of drug-induced liver injury in medical inpatients. Eur J ClinPharmacol 2005;61:135–153.

16.Duh MS, Walker AM, Kronlund KH. Descriptive epidemi-ology of acute liver enzyme abnormalities in the generalpopulation of central Massachusetts. Pharmacoepide-miol Drug Saf 1999;8:275–283.

17.Bielinski SJ, Chai HS, Pathak J, et al. Mayo GenomeConsortia: a genotype-phenotype resource for genome-

wide association studies with an application to theanalysis of circulating bilirubin levels. Mayo Clin Proc2011;86:606–614.

18.McCarty CA, Chisholm RL, Chute CG, et al. TheeMERGE Network: a consortium of biorepositories linkedto electronic medical records data for conductinggenomic studies. BMC Med Genomics 2011;4:13.

19.Jinjuvadia K, KwanW, Fontana RJ. Searching for a needlein a haystack: use of ICD-9-CM codes in drug inducedliver injury. Am J Gastroenterol 2007;102:2437–2443.

20.Overby CL, Pathak J, Gottesman O, et al. A collaborativeapproach to developing an electronic health recordphenotyping algorithm for drug-induced liver injury. J AmMed Inform Assoc 2013;20:e243–e252.

21.Sembera S, Lammert C, Talwalkar JA, et al. Frequency,clinical presentation, and outcomes of drug-induced liverinjury after liver transplantation. Liver Transpl 2012;18:803–810.

22.Fontana RJ, Watkins PB, Bonkovsky HL, et al. Drug-Induced Liver Injury Network (DILIN) prospective study:rationale, design and conduct. Drug Saf 2009;32:55–68.

23.Fontana RJ. Drug induced Liver Injury Networks. In:Kaplowtiz N, DeLeve L, eds. Drug Induced Liver Disease.3rd ed. San Diego, CA: Elsevier, 2013:713–723.

24.Andrade RJ, Lucena MJ, Fernandez MC, et al. Druginduced liver injury: an analysis of 461 incidences sub-mitted to the Spanish registry over a 10 year period.Gastroenterology 2005;129:512–521.

25.Bjornsson ES, Bergmann OM, Bjornsson HK, et al. Inci-dence, presentation, and outcomes in patients with drug-induced liver injury in the General population of Iceland.Gastroenterology 2013;144:1419–1425.

26.Molookhia M, McKeigue P. EUDRAGENE: Europeancollaboration to establish a case-control DNA collectionfor studying the genetic basis of adverse drug reactions.Pharmacogenomics 2006;7:633–638.

27.Takikawa H, Murata Y, Horiike N, et al. Drug-inducedliver injury in Japan: an analysis of 1676 cases between1997 and 2006. Hepatol Res 2009;39:427–431.

28.Aithal GP, Watkins PB, Andrade RJ, et al. Case definitionand phenotype standardization in drug-induced liverinjury. Clin Pharmacol Ther 2011;89:806–815.

29.Agarwal V, McHutchison JG, Hoofnagle JH. Importantelements for the diagnosis of drug-induced liver injury.Clin Gastroenterol Hepatol 2010;8:463–470.

30.Rockey DC, Seeff LB, Rochon J, et al. Causality assess-ment in drug-induced liver injury using a structuredexpert opinion process: comparison to the Roussel-UclafAssessment method. Hepatology 2010;51:2117–2126.

31.Pirmohamed M, Aithal GP, Behr E, et al. The phenotypestandardization project: improving pharmacogeneticstudies of serious adverse drug reactions. Clin Pharma-col Ther 2011;89:784–785.

32.Hoofnagle JH, Serrano J, Knoben JE, et al. LiverTox: awebsite on drug-induced liver injury. Hepatology 2013;57:873–874.

33.Chalasani N, Fontana RJ, Bonkovsky HL, et al. Causes,clinical features, and outcomes from a prospective studyof drug induced liver injury in the United States.Gastroenterology 2008;135:1924–1934.

http://refhub.elsevier.com/S0016-5085(13)01846-5/sref1























































































































































REVIEW

SAN

DPE

RSPE

CTIVES

34.Sgro C, Clinard F, Quazir K, et al. Incidence of drug-induced hepatic injuries: a French population-basedstudy. Hepatology 2002;36:451–455.

35.Navarro VJ, Barnhart HX, Bonkovsky HL, et al. The risingburden of herbal and dietary supplement induced hepa-totoxicity in the U.S.A. (abstr). Hepatology 2013;58(suppl):264A.

36.Galati G, Lin A, Sultan AM, et al. Cellular and in vivohepatotoxicity caused by green tea phenolic acids andcatechins. Free Radic Biol Med 2006;40:570–580.

37.Navarro VJ, Barnhart HX, Bonkovsky HL, et al. Green teaextract containing herbal and dietary supplements arefrequently mislabelled in the US: the US DILIN experi-ence (abstr 336). Hepatology 2011;54(suppl):526A.

38.Davern TJ, Chalasani N, Fontana RJ, et al. Role of acutehepatitis E in suspected drug-induced liver injury.Gastroenterology 2011;141:1665–1672.

39.Dalton HR, Fellows HJ, Stableforth W, et al. The role ofhepatitis E virus testing in drug-induced liver injury.Aliment Pharmacol Ther 2007;26:1429–1435.

40.Hoofnagle JH, Nelson KE, Purcell RH. Hepatitis E. N EnglJ Med 2012;367:1237–1244.

41.Fontana RJ, Gu J, Barnhart H, et al. Subject race andliver biochemical profile are strongly associated withclinical outcomes and chronic liver injury: results from theDILIN Prospective Study (abstr). Hepatology 2012;56(suppl):600A.

42.Andrade RJ, Lucena MI, Kaplowitz N, et al. Outcome ofacute idiosyncratic drug-induced liver injury: long-termfollow-up in a hepatotoxicity registry. Hepatology 2006;44:1581–1588.

43.Bjornsson ES, Davidsdottir L. The long-term follow-upafter idiosyncratic drug-induced liver injury with jaundice.J Hepatol 2009;50:511–517.

44.Reuben A, Koch DG, Lee WM; Acute Liver Failure StudyGroup. Drug induced acute liver Study Group: results of aUS multicenter, prospective study. Hepatology 2010;52:2065–2076.

45.Anstee QM, Day CP. S-adenosylmethionine (SAMe)therapy in liver disease: a review of current evidence andclinical utility. J Hepatol 2012;57:1097–1109.

46.Patel SJ, Milwid JM, King KR, et al. Gap junction inhi-bition prevents drug-induced liver toxicity and fulminanthepatic failure. Nat Biotechnol 2012;30:179–183.

47.Lammert C, Einarsson S, Saha C, et al. Relationshipbetween daily dose of oral medications and idiosyncraticdrug-induced liver injury: search for signals. Hepatology2009;47:2003–2009.

48.Chen M, Borlak J, Tong W. High lipophilicity and highdaily dose of oral medication are associated with signif-icant risk for drug-induced liver injury. Hepatology 2013;58:388–396.

49.Kaplowitz N. Avoiding idiosyncratic DILI: two is betterthan one. Hepatology 2013;58:15–17.

50.Lammert C, Bjornsson E, Niklasson A, et al. Oral medi-cations with significant hepatic metabolism at higher riskfor hepatic adverse events. Hepatology 2010;51:615–620.

51.Low Y, Uehera T, Minowa Y, et al. Predicting drug-induced hepatotoxicity using QSAR and toxicogenomicsapproaches. Chem Res Toxicol 2011;24:1251–1262.

52.Rodgers AD, Zhu H, Fourches D, et al. Modeling liver-related adverse effects of drugs using nearest neighborquantitative structure-activity relationship methods.Chem Res Toxicol 2010;23:724–732.

53.Shaw PJ, Ganey PE, Roth RA. Idiosyncratic drug-induced liver injury and the role of inflammatory stresswith an emphasis on an animal model of trovafloxacinhepatotoxicity. Toxicol Sci 2010;118:7–18.

54.Lewis JH, Mortensen ME, Zweig S, et al. Efficacy andsafety of high-dose pravastatin in hypercholesterolemicpatients with well compensated chronic liver disease:results of a prospective, randomized, double-blind pla-cebo-controlled multicenter trial. Hepatology 2007;46:1453–1463.

55.FDA drug safety communication: important safety labelchanges to cholesterol lowering statin drugs. April 13,2013. http://www.fda.gov/Drugs/DrugSafety/ucm293101.htm.

56.Stewart JD, Horvath R, Baruffini E, et al. Polymerase Ggene, POLG, determines the risk of sodium-valproate-induced liver toxicity. Hepatology 2010;52:1791–1796.

57.Kleiner D, Chalasani N, Lee WML, et al. Hepatic histo-logical findings in suspected drug-induced liver injury:systematic evaluation and clinical associations. Hep-atology 2014;59:661–670.

58.Bjornsson E, Kalaitzakis E, Olsson R. The impact ofeosinophilia and hepatic necrosis on prognosis in pa-tients with drug-induced liver injury. Aliment PharmacolTher 2007;25:1411–1421.

59.Katoonizadeh A, Nevens F, Verslype C, et al. Liver regen-eration in acute severe liver impairment: a clinicopatho-logical correlation study. Liver Int 2006;26:1225–1233.

60.Suzuki A, Brunt EM, Kleiner DE, et al. The use of liverbiopsy evaluation in the discrimination of idiopathicautoimmune hepatitis versus drug-induced liver injury.Hepatology 2011;54:931–939.

61.Kaplowitz N. Drug-induced liver injury. Clin Infect Dis2004;38(suppl 2):S44–S48.

62.Holt MP, Cheng L, Ju C. Identification and character-ization of infiltrating macrophages in acetaminophen-induced liver injury. J Leukoc Biol 2008;84:1410–1421.

63.Krishnan P, Feng ZZ, Gordon S. Prolonged intrahepaticcholestasis and renal failure secondary to anabolicandrogenic steroid-enriched dietary supplements. A re-view of the literature. J Clin Gastroenterol 2009;43:672–675.

64.Robinson MJ, Rywlin AM. Tetracycline-associated fattyliver in the male. Report of an autopsied case. Am J DigDis 1970;15:857–862.

65.Bleeker-Rovers C, Kadir S, van Leusen R, et al. Hepaticsteatosis and lactic acidosis caused by stavudine in anHIV-infected patient. Neth J Med 2000;57:190–193.

66.Larosche I, Letteron P, Fromenty B, et al. Tamoxifen in-hibits topoisomerases, depletes mitochondrial DNA, andtriggers steatosis in mouse liver. J Pharmacol Exp Ther2007;321:526–535.

67.NaritaM,Oussoultzoglou E,ChenardMP, et al. Liver injurydue to chemotherapy-induced sinusoidal obstructionsyndrome is associated with sinusoidal capillarization.Ann Surg Oncol 2012;19:2230–2237.


























































































http://www.fda.gov/Drugs/DrugSafety/ucm293101.htm

http://www.fda.gov/Drugs/DrugSafety/ucm293101.htm
























































REVIEWSAND

PERSPECTIVES

68.Vieto NO, George BJ. Oxaliplatin-induced hepatocellularinjury and ototoxicity: a review of the literature and reportof unusual side effects of a commonly used chemo-therapeutic agent. J Oncol Pharm Pract 2012;18:355–359.

69.Committee on Toxicity Testing and Assessment ofEnvironmental Agents, National Research Council.Toxicity testing in the 21st century: a vision and a strat-egy. Washington, DC: National Academies Press, 2007.

70.Grompe M, Strom S. Mice with human livers. Gastroen-terology 2013;145:1209–1214.

71.McGill MR, Yan HU, Ramachandran A, et al. HepaRGcells: a human model to study mechanisms of acetamin-ophen hepatotoxicity. Hepatology 2011;53:974–982.

72.Si-Tayeb K, Fallon KN, Nagaoka M, et al. Highly efficientgeneration of human hepatocyte-like cells from inducedpluripotent stem cells. Hepatology 2010;51:297–305.

73.Song Z, Cai J, Liu Y, et al. Efficient generation ofhepatocyte-like cells from human induced pleuripotentstem cells. Cell Res 2009;19:1233–1242.

74.Guguen-Guillouzo C, Corlu A, Guillouzo A. Stem cell-derived hepatocytes and their use in toxicology. Toxi-cology 2010;270:3–9.

75.Jozefczuk W, Prigione A, Chavez L, et al. Comparativeanalysis of human embryonic stem cell and inducedpluripotent stem-cell derived hepatocyte like cells revealscurrent drawbacks and possible strategies for improveddifferentiation. Stem Cells Dev 2011;20:1259–1275.

76.Faulkner L, Martinsson K, Santoyo-Castelazo A, et al.The development of in vitro culture methods to charac-terize primary T-cell responses to drugs. Toxicol Sci2012;127:150–158.

77.El-Chaiesh S, Monshi MM, Whitaker P, et al. Character-ization of the antigen specificity of T-cell clones frompiperacillin-hypersensitive patients with cystic fibrosis.J Pharmacol Exp Ther 2012;341:597–610.

78.Chessman D, Kostenko L, Lethborg T, et al. Humanleukocyte antigen class I-restricted activation of CD8þ Tcells provides the immunogenetic basis of a systemicdrug hypersensitivity. Immunity 2008;28:822–832.

79.Porebski G, Gschwend-Azwodniak A, Pichler WJ. In vitrodiagnosis of T cell-mediated drug allergy. Clin Exp Al-lergy 2011;4:461–470.

80.Watkins PB. Biomarkers for the diagnosis and manage-ment of drug-induced liver injury. Semin Liver Dis 2009;29:394–400.

81.Green RM, Flamm S. AGA technical review on the eval-uation of liver chemistry tests. Gastroenterology 2002;123:1367–1384.

82.Sayuk GS, Elwing JE, Lisker-Melman M. Hepatic glyco-genosis: an underrecognized source of abnormal liverfunction tests? Dig Dis Sci 2007;52:936–938.

83.Senior JR. Monitoring for hepatotoxicity: what is thepredictive value of liver “function” tests? Clin PharmacolTher 2009;85:331–334.

84.Mitchell JR, Long MW, Thorgeisson UP, et al. Acetylationrates and monthly liver function tests during one year ofisoniazid preventive therapy. Chest 1975;68:181–190.

85.M’Kada H, Perazzo H, Munteanu M, et al. Real timeidentification of drug-induced liver injury (DILI) through

daily screening of ALT results: a prospective, pilot cohortstudy. PLoS One 2012;7:e42418.

86.Rutherford AE, Hynan LS, Borges CB, et al. Serumapoptosis markers in acute liver failure: a pilot study. ClinGastroenterol Hepatol 2007;5:1477–1483.

87.Bantel H, Lugering A, Hidemnan J, et al. Detection ofapoptotic caspase activation in sera from patients withchronic HCV infection is associated with fibrotic liverinjury. Hepatology 2004;40:1078–1087.

88.Rutherford A, King LY, Hynan LS, et al. Development of anaccurate index for predicting outcomes of patients withacute liver failure.Gastroenterology2012;143:1237–1243.

89.Chung RT, Stravitz T, Fontana RJ, et al. Pathogenesis ofliver injury in acute liver failure. Gastroenterology 2012;143:e1–e7.

90.Monreal M, Lafoz E, Salvador R, et al. Adverse effects ofthree different forms of heparin therapy: thrombocyto-penia, increased transaminases, and hyperkalemia. Eur JClin Pharmacol 1989;37:415–418.

91.Harrill AH, Roach J, Fier I, et al. The effects of heparins onthe liver: application of mechanistic serum biomarkers ina randomized study in healthy volunteers. Clin Pharma-col Ther 2012;92:214–220.

92.Pisetsky DS, Jiang W. Role of Toll-like receptors inHMGB1 release from macrophages. Ann N Y Acad Sci2007;1109:58–65.

93.Kubes P, Mehal WZ. Sterile inflammation in the liver.Gastroenterology 2012;143:1158–1172.

94.Uetrecht J. Mechanisms of immune-mediated liver injury.Toxicol Sci 2010;115:307–321.

95.Watkins PB, Kaplowitz N, Slattery JT, et al. Amino-transferase elevations in healthy adults receiving 4 gramsof acetaminophen daily: a randomized, controlled trial.JAMA 2006;296:87–93.

96.Wetmore BA, Brees DJ, Singh R, et al. Quantitative ana-lyses and transcriptomic profiling of circulatingmessengerRNAs as biomarkers of rat liver injury. Hepatology 2010;51:2127–2139.

97.Antoine D, Dear JW, Lewis PS, et al. Mechanistic bio-markers provide early and sensitive detection ofacetaminophen-induced acute liver injury at first pre-sentation to hospital. Hepatology 2013;58:777–787.

98.Antoine DJ, Jenkins RE, Dear JW, et al. Molecular forms ofHMGB1 and keratin-18 as mechanistic biomarkers formode of cell death and prognosis during clinical acet-aminophen hepatotoxicity. J Hepatol 2012;56:1070–1079.

99.Davern TJ, James LP, Hinson JA, et al. Measurement ofserum acetaminophen-protein adducts in patients withacute liver failure. Gastroenterology 2006;130:687–694.

100.Khandelwal N, James LP, Sanders C, et al. Unrecognizedacetaminophen toxicity as a cause of indeterminateacute liver failure. Hepatology 2011;53:567–576.

101.James LP, Simpson P, Russo M, et al. Detection ofacetaminophen-protein adducts in serum during thera-peutic exposure to acetaminophen in healthy volunteers(abstract). Hepatology 2007;46(suppl 1):812A.

102.Antoine DJ, William DP, Parke BK. Understanding therole of reactive metabolites in drug-induced hepatotox-icity. State of the science. Expert Opin Drug MetabToxicol 2008;4:1415–1427.



























































































































































REVIEW

SAN

DPE

RSPE

CTIVES

103.Bell LN, Vuppalanchi R, Watkins PB, et al. Serum pro-teomic profiling in patients with drug-induced liver injury.Aliment Pharmacol Ther 2012;35:600–612.

104.Maniratanachote R, Shibata A, Kaneko S, et al. Detectionof autoantibody to aldolase B in sera from patients withtroglitazone-induced liver dysfunction. Toxicology 2005;216:15–23.

105.Andersson U, Lindberg J, Wang S, et al. A systemsbiology approach to understanding elevated serumalanine transaminase levels in a clinical trial with xime-lagatran. Biomarkers 2009;14:572–586.

106.Shi Q, Hong H, Senior J, et al. Biomarkers for drug-induced liver injury. Exp Rev Gastroenterol Hepatol2010;4:225–234.

107.Steuerwald NM, Foureau DM, Norton HJ, et al. Profiles ofserum cytokines in acute drug-induced liver injury andtheir prognostic significance. PLoS One 2013;8:e81974.

108.Hammerich L, Heymann F, Tacke F. Role of IL-17 andTh17 cells in liver diseases. Clin Dev Immunol 2011;345:803.

109.Li J, Zhu X, Liu F, et al. Cytokine and autoantibody pat-terns in acute liver failure. J Immunotoxicol 2010;7:157–164.

110.Lobenhofer EK, Auman JT, Blackshear PE, et al. Geneexpression response in target organ and whole bloodvaries as a function of target organ injury phenotype.Genome Biol 2008;9:R100.

111.Bushel PR, Heinloth AN, Li J, et al. Blood gene expres-sion signatures predict exposure levels. Proc Natl AcadSci U S A 2007;104:18211–18216.

112.Fannin RD, RussoM, O’Connell TM, et al. Acetaminophendosing of humans results in blood transcriptome andmetabolome changes consistent with impaired oxidativephosphorylation. Hepatology 2010;51:227–236.

113.Winnike JH, Li Z, Wright FA, et al. Use of pharmco-metabonomics for early prediction of acetaminophen-induced hepatotoxicity in humans. Clin Pharmacol Ther2010;88:45–50.

114.Clayton TA, Baker D, Lindon JC, et al. Pharmacometa-bonomic identification of a significant host-microbiomemetabolic interaction affecting human drug metabolism.Proc Natl Acad Sci U S A 2009;106:14728–14733.

115.O’Connell TM, Watkins PB. The application of metabo-nomics to predict drug-induced liver injury. Clin Phar-macol Ther 2010;88:384–399.

116.Robertson DG, Watkins PB, Reily MD. Metabolomics intoxicology: preclinical and clinical applications. ToxicolSci 2011;120:S146–S170.

117.Wang K, Zhang S, Marzolf B, et al. Circulating micro-RNA’s, potential biomarkers for DILI. Proc Natl Acad SciU S A 2009;106:4402–4407.

118.McGill MR, Sharpe MR, Williams CD, et al. The mecha-nism underlying acetaminophen-induced hepatotoxicityin humans and mice involves mitochondrial damage andnuclear DNA fragmentation. J Clin Invest 2012;122:1574–1583.

119.Starkey LPJ, Dear J, Platt V, et al. Circulating micro-RNA’s as potential markers of human drug-induced liverinjury. Hepatology 2011;54:1767–1776.

120.Huang YS, Chern HD, Su WJ, et al. Polymorphism of theN-acetyltransferase 2 gene as a susceptibility risk factorfor antituberculosis drug-induced hepatitis. Hepatology2002;35:883–889.

121.Huang YS, Chern HD, Su WJ, et al. Cytochrome P4502E1 genotype and the susceptibility to antituberculosisdrug-induced hepatitis. Hepatology 2003;37:924–930.

122.Hautekeete ML, Horsmans Y, Van Waeyenberge C, et al.HLA association of amoxicillin clavulanate-inducedhepatitis. Gastroenterology 1999;117:1181–1186.

123.Daly AK, Day CP. Genetic association studies in drug-induced liver injury. Semin Liver Dis 2009;29:400–411.

124.Daly AK, Donaldson PT, Bhatnagar P, et al. DILIGENStudy; International SAE Consortium. HLA-B*5701 ge-notype is a major determinant of drug-induced liver injurydue to flucloxacillin. Nat Genet 2009;41:816–819.

125.Kindmark A, Jawaid A, Harbron CG, et al. Genome-widepharmacogenetic investigation of a hepatic adverseevent without clinical signs of immunopathology sug-gests an underlying immune pathogenesis. Pharmaco-genomics J 2008;8:186–195.

126.Singer JB, Lewitzky S, Leroy E, et al. A genome-widestudy identifies HLA alleles associated with lumiracoxib-related liver injury. Nat Genet 2010;42:711–714.

127.Spraggs CF, Budde LR, Briley LP, et al. HLA-DQA1*01:01 is a major risk factor for lapatanib-induced hepato-toxicity in women with advanced breast cancer. J ClinOncol 2011;29:667–673.

128.Russmann S, Kaye JA, Jick SS, et al. Risk of cholestaticliver disease associated with flucloxacillin and fluclox-acillin prescribing habits in the UK: cohort study usingdata from the UK General Practice Research Database.Br J Clin Pharmacol 2005;60:76–82.

129.Phillips EJ, Mallal SA. HLA-B*5701 and flucloxacillinassociated drug-induced liver disease. AIDS 2013;27:491–492.

130.Andrews E, Armstrong M, Tugwood J, et al. A role for thepregnane X receptor in flucloxacillin-induced liver injury.Hepatology 2010;51:1656–1664.

131.Monshi M, Faulkner L, Gibson A, et al. Human leukocyteantigen (HLA)-B*57:01-restricted activation of drug-specific T-cells provides the immunological basis forflucloxacillin-induced liver injury. Hepatology 2013;57:727–739.

132.Illing PT, Vivain JP, Dudek NL, et al. Immune self-reactivity triggered by drug-modified HLA-peptiderepertoire. Nature 2012;486:554–558.

133.Wuillemin N, Adam J, Fontana S, et al. HLA haplotypedetermines hapten or p-I T cell reactivity to flucloxacillin.J Immunol 2013;190:4956–4964.

134.Urban TJ, Shen Y, Stolz A, et al. Limited contribution ofcommongenetic variants to riskof liver injuryduetoavarietyof drugs. Pharmacogenet Genomics 2012;22:784–795.

135.Lucena MI, Molokhia M, Shen Y, et al. Susceptibility toamoxicillin-clavulanate-induced liver injury is influencedby multiple HLA Class I and II alleles. Gastroenterology2011;131:338–347.

136.Urban TJ, Goldstein DB, Watkins PB. Genetic basis ofsusceptibility to drug-induced liver injury: what have we






















































































































































REVIEWSAND

PERSPECTIVES

learned and where do we go from here? Pharmacoge-nomics 2012;13:735–738.

137.Ventham NT, Kennedy NA, Nimmo ER, et al. Beyondgene discovery in inflammatory bowel disease: theemerging role of epigenetics. Gastroenterology 2013;145:293–308.

138.Parham P, Onta T. Population biology of antigen pre-sentation by MHC class I molecules. Science 1996;272:67–74.

139.Bharadwaj M, Iling P, Theodossis A, et al. Drug hyper-sensitivity and human leukocyte antigens of the majorhistocompatibility complex. Annu Rev Pharmacol Toxicol2012;52:401–431.

140.Chen P, Lin JJ, Lu CS, et al. Carbamazepine-inducedtoxic effects and HLA-B*5802 screening in Taiwan.N Engl J Med 2011;364:1126–1133.

141.McCormack M, Alfirevic A, Bourgeois S, et al. HLA-A*3101 and carbamazepine-induced hypersensitivity re-actions in Europeans. N Engl JMed 2011;364:1134–1143.

142.Markle JG, Frank DN,Mortin-Toth S, et al. Sex differencesin the gut microbiome drive hormone dependent regula-tion of autoimmunity. Science 2013;339:1084–1088.

143.Henao-Mejia J, Elinav E, Jin C, et al. Inflammasome-mediated dysbiosis regulates progression of NAFLD andobesity. Nature 2012;482:179–185.

144.Hirschfield GM, Chapman RW, Karlsen TH, et al. Thegenetics of complex cholestatic disorders. Gastroenter-ology 2013;144:1357–1374.

145.Holmes E, Li JV, Marchesi JR, et al. Gut microbiotacomposition and activity in relation to host metabolicphenotype and disease risk. Cell Metab 2012;16:559–564.

146.Zheng X, Zhao A, Xie G, et al. Melamine-induced renaltoxicity is mediated by the gut microbiota. Sci TranslMed 2013;5:172ra22.

Received September 12, 2013. Accepted December 11, 2013.

Reprint requestsAddress requests for reprints to: Robert J. Fontana, MD, 3912 TaubmanCenter, Ann Arbor, Michigan 48109-0362. e-mail: [email protected];fax: (734) 936-7392.

Conflicts of interestThe author discloses the following: R.J.F. has received research support fromVertex Pharmaceuticals and Gilead Sciences and served as a consultant forGlaxoSmithKline and Tibotec.

FundingR.J.F. is a National Institutes of Health–funded investigator with researchsupport as a principal investigator for the Drug-Induced Liver Injury Network(2U01-DK065184-06) and the Acute Liver Failure Study Group (DK-U-01-58369).














































mailto:[email protected]

Supplementary Table 1.Causality Assessment Methods Used in Studies of DILI

Likelihood of DILIWorld Health

Organization (4 levels)Naranjo(4 levels)

Roussel Uclaf CausalityAssessment Method

(5 levels)Maria and Victorino

(5 levels) DILIN (5 levels)

>75% Level 1 (certain) Definite Highly probable Definite Definite (1) and highlylikely (2)

50%–75% Level 2 (probable) Probable Probable Probable Probable25%–50% Level 3 (possible) Possible Possible Possible Possible<25% Level 4 (unlikely) Doubtful Not likely and excluded Not likely and excluded Unlikely

Adapted from the LiverTox Web site (http:www.livertox.nih.gov).

Supplementary Table 2.GWAS of Susceptibility for DILI

DrugNo. of cases

of DILI Control GeneMinor allele

frequency (%) Odds ratio

Flucloxacillin (b-lactam antibiotic)125 51 282 population controls HLA-B*5701 6 80Ximelagatran (thrombin inhibitor)126 74 130 treated controls DRB1*07

DQA1*028.5 4.4

4.4Lumiracoxib (cyclooxygenase-2 inhibitor)127 41 176 treated controls DRB1*1501

DQB1*060215 5.0

Lapatinib (kinase inhibitor)128 37 286 treated controls DQA1*02 21 9.0Amoxicillin-clavulanate (antibiotic)136 201 532 population controls DRB1*1501

HLA-A*0201164

2.82.3

April 2014 Drug-Induced Liver Injury 928.e1


Documents

Gastroenterology 2014;146:914 REVIEWS IN BASIC AND ......Overall, DILI accounts for