CURRENT OPINION Drug Safety 2001; 24 (7): 483-490

0114-5916/01/0007-0483/$22.00/0

© Adis International Limited. All rights reserved.

Drug-Induced Liver Disorders

Implications for Drug Development and Regulation
Neil Kaplowitz
USC Research Center for Liver Diseases, Keck School of Medicine, University of Southern
California, Los Angeles, California, USA

Abstract Drug-induced hepatotoxicity is a frequent cause of liver disease. Although

often presenting as acute hepatitis and/or cholestasis, virtually any clinical-
pathological pattern of acute or chronic liver disease can occur. Most reactions
occur in a small proportion of the population using a particular drug. Each drug
associated with hepatotoxicity tends to have a characteristic signature regarding
latency and pattern of injury. The mechanism can be drug metabolism-dependent
or related to the chemical properties of the parent drug. The former are immune
mediated or due to metabolic idiosyncrasy. Monitoring serum ALT levels is of
unproven effectiveness but should be considered when there is an increased risk
of delayed onset serious hepatitis-like reactions. The key for the future is im-
proved identification of toxic potential in preclinical studies, clinical trials and
postmarketing experience. The elucidation of the genetic and environmental
mechanisms contributing to delayed idiosyncratic reactions is a major barrier to
overcome in this field.

Drug-induced liver disease represents a major facts and opinions in this review was obtained from
problem for drug development and safety. Hun- the authors’ files and a literature search on drug-
dreds of drugs have been implicated in causing induced liver disorders since 1994 to provide the
liver disease at least on rare occasions and in the most up to date information.
US, drug-induced liver disease is the most common
cause of acute liver failure, accounting for one-third 1. Clinical Overview
to one-half of cases.[1,2] In addition to representing
an important diagnostic/therapeutic challenge for Drug-induced liver disease can mimic all forms
physicians caring for patients presenting with liver of acute and chronic hepatobiliary diseases (table
disorders, the frequency and economic impact of I).[3,4] However, the predominant clinical presenta-
this long standing problem is a major challenge for tions resemble acute icteric hepatitis or cholestatic
the pharmaceutical industry and regulatory bodies. liver disease. The former is of grave significance
Recent problems with bromfenac and troglitazone as the mortality approximates 10% irrespective of
have once again put the spotlight on this problem. the causative drug.[3] This type of reaction is ac-
The purpose of this review is to call attention to companied by systemic symptoms, jaundice, mark-
this important problem and the issues it raises for edly elevated serum transaminase levels and, in the
clinicians, pharmaceutical industry and regulatory more severe cases, coagulopathy and encephalop-
bodies. The material which forms the basis of the athy indicative of acute (fulminant) liver failure.
484 Kaplowitz

Table I. Spectrum of hepatic manifestations of drug-induced liver disorders

Acute hepatitis Isoniazid, paracetamol (acetaminophen), troglitazone
Chronic hepatitisa Nitofurantoin, methyldopa, diclofenac, minocyline
Acute cholestasis Amoxicillin-clavulanic acid, erythromycin, sulindac, chlorpromazine
Mixed hepatitis/cholestasis or atypical hepatitis Phenytoin, sulfonamides
Chronic cholestasisa Chlorpromazine
Nonalcoholic steatohepatitis Amiodarone
Fibrosis/cirrhosis Methotrexate
Microvesicular fatty liver Valproic acid (sodium valproate), nucleoside reverse transcriptase
inhibitors
Veno-occlusive disease Busulfan, cyclophosphamide
a Drugs which cause chronic disease more frequently cause acute disease.

Cholestatic disease, although not usually life (up to 12 months). Intermediate latency is charac-
threatening, presents with jaundice, disproportion- teristic of hypersensitivity reactions. These tend to
ate increased serum alkaline phosphatase levels be associated with fever, rash and eosinophilia and
and pruritus; cholestatic reactions tend to resolve a rapid positive rechallenge.[3,4] Hepatotoxicity of
very slowly (i.e. months versus weeks for hepatitis) sulindac,[9] phenytoin[10] and amoxicillin-clavulanic
and on rare occasion lead to vanishing bile duct acid[11] are typical examples. It is important to rec-
disease and biliary cirrhosis.[5,6] Individual drugs ognise that these reactions may occur up to 3 to 4
tend to exhibit a consistent pattern or signature of weeks after a 1 to 2 week course of medication (e.g.
the reaction with characteristic latency and clinical amoxicillin-clavulanic acid). The long latency type
presentation. However, some drugs may show sev- of reaction is characteristically not associated with
eral patterns: e.g. nimesulide can cause a short la- features of hypersensitivity and the response to re-
tency hypersensitivity-mediated cholestatic injury challenge is variable. Thus, one assumes that these
and a delayed idiosyncratic acute hepatitis-like re- events reflect some type of delayed change in the
action.[7] Other types of reactions are listed in table metabolism of the drug or the response to injury
I with some examples of causative drugs. (repair or regeneration). Drugs associated with de-
layed reactions include isoniazid[12] and troglita-
2. Pathogenesis zone.[13] However, in the case of isoniazid, the la-
tency may be shortened when used in combination
Adverse hepatic events due to drugs can be con- with rifampicin (rifampin).[14] Also, as noted above,
sidered simply as predictable (high incidence) or some drugs can produce either a short latency hy-
unpredictable (low incidence). The bulk are un- persensitivity or long latency idiosyncratic reac-
predictable and can be viewed as either immune- tion.[7]
mediated hypersensitivity reactions or as idiosyn- An interesting feature of these low frequency
cratic. Latency between the initiation of therapy and unpredictable reactions, either hypersensitivity or
the onset of liver disease provides some clue as to idiosyncratic, is that they often occur on a back-
the pathogenesis. Early onset within a few days ground, higher rate of mild, asymptomatic, and
(particularly if there has been no previous expo- usually transient liver injury which is detected as
sure) is strong evidence for direct toxicity of the abnormal biochemical tests, particularly serum
drug or its metabolite which is characteristic of pre- ALT. Generally, the biochemical abnormality de-
dictable reactions; paracetamol (acetaminophen) fined as an ALT level >3 times the upper limit of
overdose is an example.[8] Unpredictable reactions normal (ULN) may occur 10 to 20 more frequently
manifested as overt or symptomatic disease can oc- than overt disease. Thus, in the majority of patients
cur with intermediate (1 to 8 weeks) or long latency with an increased ALT level some type of adapta-

© Adis International Limited. All rights reserved. Drug Safety 2001; 24 (7)
Regulatory Implications of Drug-Induced Liver Disorders 485

tion or ‘tolerance’ occurs and in the minority there [e.g. quinones cycle electrons from nicotinamide
is a failure to do so. This issue is further compli- adenine dinucleotide phospate (NADPH) to O2
cated by the mysterious nature of the long latency generating reactive oxygen species]; or (iii) specif-
in some of the idiosyncratic reactions. ically inhibit enzymes or transporters (e.g. nucleo-
Hepatotoxicity of drugs can be principally side reverse transcriptase inhibitors block mito-
metabolism-dependent, parent drug-dependent or chondrial DNA polymerase[20] and cyclosporin
a combination of both. Metabolism takes place inhibits canalicular transporters).[21] In these cases,
largely in the liver which accounts for its suscepti- if the parent drug’s chemical properties account for
bility to drug-induced injury.[4] The metabolites direct toxicity, factors which enhance its availabil-
may be electrophilic chemicals or free radicals ity, such as decreased metabolism or export, may
which deplete reduced glutathione, covalently bind increase susceptibility.
to proteins, lipids or nucleic acids, or induce lipid With the advent of new technologies in geno-
peroxidation. The consequences include hepato- mics and proteomics, one can anticipate that new
cellular necrosis, apoptosis or sensitisation to insights into the mechanisms of susceptibility and
cytokines or inflammatory mediators produced by liver injury from drugs will be forthcoming.
nonparenchymal cells. Alternatively, the reactive
metabolites may covalently bind to or alter liver 3. Risk Factors
proteins such as cytochrome P450 enzymes lead-
ing to sensitisation and immune mediated injury. The Regardless of whether hepatotoxicity is predict-
immune phenomena nevertheless are metabolism- able (frequent) or unpredictable (rare), hypersen-
sitivity-mediated or idiosyncratic, metabolism-
dependent. Thus, the rare occurrence of immune-
dependent or chemical structure-dependent, the
mediated liver disease is often superimposed on a
interplay of genetic and environmental risk factors
higher frequency of mild injury (abnormal ALT
influences susceptibility.[22] Table II lists examples
level) suggesting that the drug has a mild toxic
of drugs and associated risk factors.
potential (e.g. phenytoin or halothane) but in rare
individuals this toxic potential leads to metabo-
lism-dependent hypersensitivity. Genetic poly- 4. Monitoring
morphisms of enzymes involved drug activation or The background incidence of drug-induced
detoxification have been implicated in the suscep- mild, reversible liver injury provides the rationale
tibility to hypersensitivity reactions to sulfona- for monitoring or surveillance. From this back-
mides[15,16] and anticonvulsants.[10,17] Presumably ground of mild injury a minority of individuals will
genetic polymorphisms of either major histocom- emerge with overt disease. Thus, by stopping the
patibility complex (MHC)–I-dependent antigen medications at the first sign of mild injury one
presentation in hepatocytes or MHC–II-dependent should prevent serious consequences. Although
antigen presentation in macrophages, which have this seems a logical approach, a number of prob-
scavenged necrotic or apoptotic hepatocytes di- lems must be considered. First, the approach only
rectly killed by the drug, may further contribute to applies to delayed reactions. Secondly, one is sac-
determining the rare occurrence of these hypersen- rificing potentially important therapy to a much
sitivity reactions[18] which most often have an in- larger number of patients than would actually de-
cidence of 1 : 1000 or less. velop overt disease. Thirdly, compliance with such
Parent drug–dependent toxicity occurs as a re- approaches is known to be poor and fourthly, the
sult of the properties of the parent drug (or meta- rate of development of overt disease from the first
bolite) to: (i) accumulate in organelles (e.g. weak appearance of elevated ALT levels needs to be
bases such as amiodarone accumulate in mitochon- gradual for monthly monitoring to be effective in
dria);[19] (ii) undergo nonspecific redox cycling preventing life-threatening disease.

© Adis International Limited. All rights reserved. Drug Safety 2001; 24 (7)
486 Kaplowitz

Table II. Risk factors for drug-induced hepatotoxicity

Drug Factors
Methotrexate Chronic alcohol abuse, obesity, diabetes mellitus, chronic hepatitis, psoriasis
Isoniazid Hepatitis B infection, hepatitis C infection, HIV infection, alcohol abuse, older
age, female gender, slow acetylator phenotype, concurrent use of rifampicin
(rifampin) or pyrazinamide
Paracetamol (acetaminophen) Chronic alcohol abuse,a fasting, concurrent use of isoniazid
Valproic acid (sodium valproate) Young age, concurrent use of other anticonvulsants, genetic defects of
mitochondrial β-oxidation and respiratory chain enzymes
Diclofenac Female gender, osteoarthritis
Anticonvulsants Genetic defect in detoxification
Sulfonamides HIV infection, slow acetylator phenotype, genetic defect in detoxification
a Controversial.[23,24]

Testing more frequently than monthly is not pra- ing with its attendant issues of compliance and ef-
ctical, although the future development of a finger- ficacy, really justified? What rate of adverse events
stick ALT test, that could be applied in a fashion justifies monitoring? At present there are no defin-
similar to monitoring glucose, might change this itive answers to these questions. Regulatory agen-
by improving compliance and allowing more fre- cies and the pharmaceutical industry must decide
quent monitoring. In any case, monthly monitoring if any risk to life from drug-induced liver toxicity,
for delayed idiosyncratic reactions is the best ap- irrespective how rare, is justifiable with a specific
proach available, but the effectiveness of the ap- drug.
proach is assumed and not proven. Furthermore, This extremely difficult question was faced by
this should not substitute for the need to educate industry and the FDA recently with bromfenac and
patients about symptoms of hepatotoxicity such as troglitazone. In the case of the nonsteroidal anti-
fever, malaise, fatigue, gastrointestinal complaints, inflammatory drug, bromfenac, continued use of the
abdominal pain, dark urine, jaundice, pruritus, etc, drug in the face of infrequent, delayed idiosyncra-
and the need to report them to their physicians to tic severe hepatotoxicity,[25] could not be justified
insure expeditious cessation of offending agents. since many alternative treatments were available.
Despite monthly monitoring, some adverse events In the case of troglitazone, the decision to withdraw
may appear rapidly in a few weeks following a nor- was delayed and more complicated due to the im-
mal test. This was the experience with troglitazone; portant and unique therapeutic properties of the
the incidence of adverse events reported to the US drug in managing a serious medical condition, al-
Food and Drug Administration (FDA) appeared to beit with benefits which would not be evident for
decrease after monthly monitoring was recommen- many years (i.e. the effect long term control of
ded but rare cases of severe disease rapidly occur- blood sugar on complications of diabetes mellitus),
red (‘rapid risers’) within a few weeks after a nor- whereas the life-threatening complication of acute
mal ALT or mild ALT elevation [<3 times the upper liver failure occurred in the first year and with an
limit of normal (ULN)] at the previous check-up. incidence of about 1 : 20 000. It was decided that
Ultimately, the most difficult challenge to the the implementation of monthly monitoring would
application of monitoring is cost effectiveness – likely protect the users and the drug was continued.
monthly monitoring is expensive and one needs to Although this strategy may have worked to some
weigh this quantitatively against the morbidity and extent, the issue of compliance with monitoring and
mortality of adverse liver events. If hospitalisation the possibility of occasional ‘rapid risers’ meant
occurs in 1 in 10 000 patients taking a medication that the population could not be completely pro-
and acute liver failure in 1 in 100 000, is monitor- tected. At the same time, several new drugs in this

© Adis International Limited. All rights reserved. Drug Safety 2001; 24 (7)
Regulatory Implications of Drug-Induced Liver Disorders 487

class were approved and after a number of months long term exposure (e.g. 6 months or more). If the
of postmarketing experience with the alternative drug in question turns out to cause overt liver dis-
new agents, it was concluded that these agents ease in 1 in 100 patients, 300 patients would need
were probably less likely to induce severe hepato- to be treated to ensure that the adverse event is not
toxicity, leading to the withdrawal of troglitazone. missed. However, if overt liver disease occurs in
approximately 1 in 1000 patients, the number of
5. Issues for Drug Development patients needed would be 3000. This is exactly
what was seen with troglitazone: a 2- to 3-fold in-
Nonclinical assessment using animal models creased incidence ALT > 3 times ULN and 2 pa-
undoubtedly screens out most potent toxins. How- tients with jaundice in the cohort of 2500 patients
ever, liver injury at extremely high doses in animal treated with the drug.[26] Thus, the clinical trials
models does not necessarily predict problems in provided the signals for concern.
humans. Obviously, drugs that pass through animal On the other hand, if the true incidence of overt
testing and reach clinical trials have been consid- drug-induced liver disease is 1 in 10 000, there
ered of no or negligible risk to humans. Com-
would be a high likelihood of missing overt disease
pounds which are not hepatotoxic in animals usu-
in the clinical trials due to insufficient numbers of
ally do not induce serious hepatotoxicity in humans.
patients involved. Thus, the absence of hyperbili-
The occurrence of rare hypersensitivity or idiosyn-
rubinaemia and markedly elevated ALT levels in
cratic reactions in humans cannot be reliably pre-
3000 study patients predicts that the incidence of
dicted at present from preclinical studies. Thus,
overt disease should be somewhere between zero
whether or not animal testing shows some toxic
and less than 1 per 1000 and fatal disease between
propensity at extreme doses, vigilance during clin-
zero and less than 1 per 10 000 (remembering that
ical trials and postmarketing needs to be univer-
sally practised. we are concerned with this issue because overt
drug-induced cytotoxic disease carries about a 1 in
The key arena for identification of drugs with
significant hepatotoxic potential is phase I to III 10 chance of death or liver transplantation).
clinical trials. At this stage it is critical to determine A major dilemma is whether to not approve a
the incidence of abnormalities in treated versus drug with a signal of increased incidence of ALT >
control patients. The value of >3 times ULN for 3 times ULN and more than 1 case of hyperbilirub-
ALT is commonly used as a signal for concern. The inaemia with elevated ALT level versus approval
background incidence of this signal depends on the with a recommendation for monitoring. In general,
nature of the control population (obesity, diabetes it would seem prudent to require more extensive
mellitus, heart failure, chronic hepatitis C infec- premarketing assessment of safety when time-
tion) and ranges between 0.1 to 1.0%. Thus, if the tested, well-tolerated alternatives are already avail-
incidence of ALT in drug recipients is 2 to 3%, this able. The decision depends on the uniqueness or
is cause for some concern. However, even more importance of the new drug and the prediction (un-
concerning would be the occurrence of ALT >8 to proved) for the effectiveness of monitoring. The
10 times ULN, in conjunction with hyperbilirubin- latter approach should include careful postmarket-
aemia and/or symptomatic hepatitis. The rule of ing surveillance or expanded phase IV studies in-
threes applies to this situation: in order to be >95% cluding an assessment of the impact of monitoring.
sure of not missing an adverse event (e.g. overt There are situations in which a statistically signif-
icteric symptomatic disease) one needs to study icant increase in mild liver injury (ALT level >3
nearly 3-fold more patients than the true incidence. times ULN) has been observed without overt dis-
Most phase I to III trials involve of a total of ease. Under such circumstances, predictions about
2000 to 3000 patients and volunteers who receive risk of serious liver injury cannot be made and de-
a drug for variable time; a subgroup may undergo cisions concerning the need for monitoring are

© Adis International Limited. All rights reserved. Drug Safety 2001; 24 (7)
488 Kaplowitz

very difficult to support. The cholesterol-lowering pattern of disease fit and other causes are excluded
HMG-CoA reductase inhibitors (‘statins’) exem- (e.g. viral hepatitis, ischaemic hepatitis, biliary dis-
plify this problem. Some have been associated with ease). The remainder of cases are unlikely or unre-
an increase in incidence of ALT level increases ver- lated depending on the completeness of the work-
sus controls in clinical trials but extensive postmar- up and the strength of the evidence in favour of an
keting experience has shown little cause for con- alternative diagnosis. This ad hoc approach is
cern. equivalent to diagnosing as yes, no, or maybe.
Although most drugs with a propensity for hep- The presence of autoantibodies to specific
atotoxicity will be identified in preclinical testing forms of cytochrome P450 enzymes have been as-
and clinical trials, occasional drugs slip through sociated with hypersensitivity reactions to certain
because the reactions are rare and either hypersen- drugs.[18,27,28] Although of uncertain but intriguing
sitivity or delayed idiosyncratic in nature. Post- significance with respect to pathophysiology, their
marketing surveillance of newly approved drugs is presence may be helpful in the diagnosis of drug-
very important but is associated with limitations induced liver disease in these special cases (table
due to lack of compliance with reporting adverse III). However, testing for these autoantibodies is
events and lack of controls. Occasional unexplained mainly a research tool at present.
liver failure in a patient receiving a drug must be
Several groups have attempted to generate
weighed against the background incidence of liver
quantitative systems designed to generate a numer-
failure in the general population or associated with
ical score which reflects the probability of a drug
the disease being treated. For example, it is estimat-
as the cause for liver disease.[29-32] The Council for
ed that about 2000 cases of acute liver failure occur
International Organisation of Medical Sciences
in the US population of 300 000 000 (incidence <1
: 100 000). At least 10 to 20% of these cases have (CIOMS) scoring system appears to be the most
no identifiable cause, resulting in a background in- accurate[33,34] and puts numerical weight on the
cidence of 1 to 2 cases per million. Thus, the repor- factors discussed at the beginning of this section to
ting of 1 or 2 cases of acute liver failure in the first generate a composite score which reflects the prob-
million recipients of a new drug must be viewed dif- ability that liver injury is drug-induced. The advan-
ferently than 10 to 100 times that number. tage is that this system is less subjective than the
ad hoc approach. This type of scoring system per-
forms well when validated against well documented
6. Diagnosis
cases of drug-induced liver disease. Specialists,
Establishing a diagnosis of drug-induced liver pharmaceutical industry and regulatory bodies
disease in an individual case is mainly based upon should be encouraged to use this scale. It also
circumstantial evidence aided by the signature type would be reasonable to apply the scoring system to
of reaction (if known) with respect to latency and individual case reports submitted to medical jour-
clinical characteristics as well as exclusion of other
more plausible alternative causes. Additional in-
Table III. Autoantibodies in drug-induced liver disorders
formation can be gained from the response to the
Autoantibody target Drug
removal of the drug, i.e. rapid improvement in
CYP 2C9a Tienilic acid
cytotoxic reactions and slow improvement in cho-
CYP 1A2b Dihydralazine
lestatic reactions. A rechallenge with recrudes- CYP 3Ab Anticonvulsants
cence of liver abnormalities is the most definitive CYP 2E1 Halothane
evidence, i.e. highly probable, but hardly ever jus- mEH Germander
tified and not always positive in idiosyncratic a Also referred to as anti-LKM2 autoantibody.
cases. A practical approach is to consider the diag- b Also referred to as anti-LM autoantibody.
nosis probable/possible if the signature latency and CYP = cytochrome P450; mEH = microsomal epoxide hydrolase.

© Adis International Limited. All rights reserved. Drug Safety 2001; 24 (7)
Regulatory Implications of Drug-Induced Liver Disorders 489

nals. Although it is not perfect and may not dis- References

criminate between multiple concurrently used can- 1. Ostapowicz G, Fontana RB, Larson AM, et al. Etiology and
outcome of acute liver failure in the USA: preliminary results
didate toxins, it does provide consistency and fo- of a prospective multi-center study [abstract]. Hepatology
cuses the attention of the evaluator on most of the 1999; 30 (4): 221A
critical parameters which need to be considered in 2. Shakil A, Kramer D, Mazariegos G, et al. Acute liver failure:
clinical features, outcome analysis, and applicability of prog-
estimating the probability of causality. nostic criteria. Liver Transpl 2000; 16: 163-9
3. Zimmerman H. Drug-induced liver disease, In: Schiff E, Sorrell
M and Maddrey W, editors. Schiff’s diseases of the liver. 8th
7. Conclusions ed. Philadelphia (PA): Lippincott-Raven Publishers, 1999:
973-1064
Drug-induced liver disorders occur frequently, 4. Kaplowitz N. Drug metabolism and hepatotoxicity. In: Kap-
lowitz N, editor. Liver and biliary diseases. 2nd ed. Baltimore
may be life-threatening and can mimic all forms of (MD): Williams and Wilkins, 1996: 103-20
liver disease. Preclinical drug development screens 5. Desmet VJ. Vanishing bile duct syndrome in drug-induced liver
out potent toxins but is imperfect in identifying the disease. J Hepatology 1997; 26 Suppl. 1: 31-5
6. Degott C, Feldmann G, Larrey D, et al. Drug-induced pro-
potential for unpredictable rare hypersensitivity longed cholestasis in adults: a histological semiquantitative
and idiosyncratic reactions. Although knowledge study demonstrating progressive ductopenia. Hepatology
of the metabolism of drugs and the identification 1992; 15: 244-51
7. Van Steenberen W, Peeters P, DeBondt J, et al. Nimesulide-
of toxic metabolites may be helpful in improving induced acute hepatitis: evidence from six cases. J Hepato-
the value of preclinical studies, advancement will logy 1998; 29: 135-41
require the elucidation of the mechanism of toxic- 8. Pham T-V, Lu S, Kaplowitz N. Acetaminophen hepatotoxicity.
In: Taylor, editor. Gastrointestinal emergencies. 2nd ed. Bal-
ity in humans so as to employ animal models which timore (MD): Williams & Wilkins, 1997: 371-88
express the determinants of risk. This will allow 9. Tarazi E, Harter JG, Zimmerman HJ, et al. Sulindac-associated
deployment of transgenic and knockout models hepatic injury: analysis of 91 cases reported to the Food and
Drug Administration. Gastroenterology 1993; 104: 569-74
and other pretreatments which unmask the toxic 10. Shear N, Spielberg S. Anticonvulsant hypersensitivity syn-
potential. In the clinical development phase, drome: in vitro assessment of risk. J Clin Invest 1988; 82:
1826-32
greater attention to the identification of signals, such 11. Larrey D, Vial T, Micaleff A, et al. Hepatitis associated with
as increased frequency of ALT level elevations, amoxycillin-clavulanic acid combination. Report of 15 cases.
and more precise definition of true versus false Gut 1992; 33: 368-71
12. Thompson N, Caplin M, Hamilton M, et al. Anti-tuberculosis
positive signals is needed. In the postmarketing medication and the liver: dangers and recommendations in
phase, along with improving compliance in report- management. Eur Respir J 1995; 8: 1384-8
ing adverse events and causality assessment, stud- 13. Murphy E, Davern T, Shakil O, et al. Troglitazone-induced ful-
minant hepatic failure. Dig Dis Sci 2000; 45: 549-3
ies of the value of monitoring are of critical impor- 14. Durand F, Bernuau J, Pessayre D, et al. Deleterious influence
tance. The employment of new technologies such of pyrazinamide on the outcome of patients with fulminant or
as genomics and proteomics should lead to better subfulminant liver failure during antituberculosis treatment
including isoniazid. Hepatology 1995; 21: 929-32
understanding of the mechanisms of unpredictable 15. Rieder M, Uetrecht J, Shear N, et al. Diagnosis of sulfonamide
reactions, identification of individuals at risk, and hypersensitivity reactions by in vitro ‘rechallenge’ with hy-
droxylamine metabolites. Ann Int Med 1989; 110: 286-9
development of relevant animal model systems to
16. Rieder M, Shear N, Kanee A, et al. Prominence of slow acetyl-
screen new drugs. These new technologies will be ator phenotype among patients with sulfonamide hypersensi-
useful at multiple levels. In animal and in vitro sys- tivity reactions. Clin Pharmacol Ther 1991; 49: 13-7
17. Gennis M, Vemusi R, Burns E, et al. Familial occurrence of
tems, the identification of signature patterns of hypersensitivity to phenytoin. Am J Med 1991; 91: 631-4
gene expression (toxicogenomics) may predict 18. Robin M, LeRoy M, Descatoire V, et al. Plasma membrane
toxic potential even when overt organ damage does cytochromes P450 as neoantigens and autoimmune targets in
drug-induced hepatitis. J Hepatol 1997; 26 (1): 23-30
not occur. In the patients with ALT level abnormal- 19. Berson A, DeBeco V, Letteron P, et al. Steatohepatitis-inducing
ities or overt disease, the new technologies offer drugs cause mitochondrial dysfunction and lipid peroxidation
the hope of identifying the genetic predisposition in rat hepatocytes. Gastroenterology 1998; 114: 764-4
20. Brinkman K, Hofstede H, Burger D, et al. Adverse effects of
to rare immunological or idiosyncratic hepatotox- reverse transcriptase inhibitors: mitochondrial toxicity as
icity (pharmacogenomics). common pathway. AIDS 1998; 12: 1735-44

© Adis International Limited. All rights reserved. Drug Safety 2001; 24 (7)
490 Kaplowitz

21. Kowdley K, Keeffe E. Hepatotoxicity of transplant immuno- 30. Benichou C. Criteria of drug induced liver disorders: Report
suppressive agents. Gastroenterol Clin North Am 1995; 24: of an international consensus meeting. J Hepatol 1990; 11:
991-1001 272-6
22. DeLeve L, Kaplowitz N. Prevention and therapy of drug- 31. Danan G, Benichou C. Causality assessment of adverse reac-
induced hepatic injury. In: Wolfe MM, editor. Therapy of di- tions to drugs – I: a novel method based on the conclusions
gestive disorders. Philadelphia (PA): W.B. Saunders, Harcourt of international consensus meetings: application to drug-
Brace & Company, 2000: 334-48
induced liver injuries. J Clin Epidemiol 1993; 46: 1323-30
23. Zimmerman H, Maddrey W. Acetaminophen (paracetamol)
32. Benichou C, Danan G, Flahault A. Causality assessment of ad-
hepatotoxicity with regular intake of alcohol: analysis of in-
stances of therapeutic misadventure. Hepatology 1995; 22: verse reactions to drugs – II: an original model or validation
767-3 of drug causality assessment methods: case reports with pos-
24. Makin A, Williams R. Paracetamol hepatotoxicity and alcohol itive rechallenge. J Clin Epidemiol 1993; 46: 1331-6
consumption in deliberate and accidental overdose. Q J Med 33. Lucena M, Camargo R, Andrade R, et al. Comparison of two
2000; 93: 341-9 clinical scales for causality assessment in hepatotoxicity.
25. Moses P, Schroeder B, Alkhatib O, et al. Severe hepatotoxicity Hepatology 2001; 33: 123-30
associated with bromfenac sodium. Am J Gastroenterol 1999; 34. Kaplowitz N. Causality assessment versus guilt by association
94: 1393-6 in drug hepatotoxicity. Hepatology 2001; 33: 308-10
26. Watkins P, Whitcomb R. Hepatic dysfunction associated with
troglitazone. N Engl J Med 1998; 338: 916-7
27. Beaune P, Lecoeur S. Immunotoxicology of the liver: adverse
reactions to drugs. J Hepatol 1997; 26 Suppl. 2: 37-42 Correspondence and offprints: Dr Neil Kaplowitz, GI/Liver
28. Neuberger J. Immune mechanisms in drug hepatotoxicity. Cli
Division, Keck School of Medicine, University of Southern
Liver Dis 1998; 2: 471-82
29. Maria V, Victorino R. Development and validation of a clinical California, 2011 Zonal Avenue, HMR 101, Los Angeles, CA
scale for the diagnosis of drug-induced hepatitis. Hepatology 90033, USA.
1997; 26: 664-9 E-mail: kaplowit@hsc.usc.edu

© Adis International Limited. All rights reserved. Drug Safety 2001; 24 (7)