Journal of Hepatology 1999; 31: 1098-l 105
`Printed in Denmark
`. All rights reserved
`. Copenhagen
`Munksgaard
`
`Copyright 0 European Association
`for the Study of the Liver 1999
`
`Journal of Hepatology
`ISSN 0168-8278
`
`Special Article
`Antecedent liver disease and drug toxici&
`
`Steven Schenker, Ralston R. Martin and Anastacio M. Hoyumpa
`The University of Texas Health Science Center, Department of Medicine, Division of Gastroenterology and Nutrition, San Antonio, Texas, USA
`
`0 NE OF THE MORE common questions asked of hepa-
`
`tologists concerns the risks and proper use of po-
`tentially hepatotoxic drugs in patients with underlying
`liver disease. A typical example
`is a patient with tu-
`berculosis and abnormal
`liver tests (perhaps due to al-
`cohol and/or hepatitis C) who requires treatment with
`isoniazid, rifampicin and/or pyrazinamide. Should all
`these drugs be used, should the combination be modi-
`fied, and/or should the dosages be changed? Are there
`special monitoring considerations
`in such a patient? At
`what level of increased
`liver test abnormality
`should
`the drugs be stopped? The purpose of this illustrative
`case is not to focus on this specific patient, but to pro-
`vide a forum for a general discussion of this dilemma.
`While it is generally known that most drugs are me-
`tabolized by the liver and many are excreted by it, and
`thus liver dysfunction may require adjustment of drug
`dosage in such patients,
`the implications of this for
`overall drug hepatotoxicity have not been formalized.
`This is, perhaps, best explained by the lack of meaning-
`ful published data. Nevertheless,
`in a number of in-
`stances involving the use of therapeutic agents with po-
`tential liver toxicity (pyrazinamide,
`tolcapone,
`troglita-
`zone), prior
`liver dysfunction
`has been cited as a
`contraindication. Our aim in this article is to amalga-
`mate and interpret available information on this topic.
`A number of key concepts serve as guidelines for
`this analysis. These can generally be catalogued under
`the headings of mechanisms of drug-induced
`liver in-
`jury, summation of toxic effects and difficulties in early
`diagnosis.
`
`Correspondence: Steven Schenker, The University of Texas
`Health Science Center, Department of Medicine, Division
`of Gastroenterology
`and Nutrition, 7703 Floyd Curl
`Drive, San Antonio, Texas, 78229 3900, USA.
`Tel: 210 567 4878. Fax: 210 567 1976.
`e-mail: mundym@uthscsa.edu
`
`1098
`
`Mechanisms of Drug-Induced Liver Injury
`Dose-dependent
`hepatotoxicity
`First, hepatotoxic drug reactions can be divided into
`those which are predictable
`(i.e., dose-dependent)
`and
`those which are idiosyncratic
`(i.e. dose-independent)
`(1). The predictable group
`is small, inasmuch as the
`dose of the drug recommended on the basis of initial
`studies
`is aimed at achieving concentrations
`of the
`agent (or its toxic metabolites) below (and ideally well
`below) the dangerous
`level. Where such precautions
`may break down is when due to age, genetically altered
`metabolism,
`interaction with another drug or with
`underlying
`liver disease, normal
`removal of the drug
`is impaired,
`resulting over a period of time in toxic
`concentrations.
`Such alterations
`in drug disposition
`are termed pharmacokinetic
`changes. Some examples
`of this are seen in Table 1, as with perhexiline maleate
`
`TABLE 1
`Dosage-deuendent
`
`Drug
`
`Acetaminophen
`Tetracycline
`Methotrexate
`
`Perhexiline
`Amiodarone
`Cyclosporine
`Cyclophosphamide
`Valproic acid (?)
`Oral contraceptives
`Aspirin
`(?)
`Niacin
`Pyrizinamide
`Bromfenac
`Naltrexone
`*
`**
`+
`
`(?)
`
`drug-induced
`
`liver disease
`
`Risk factors
`
`liver
`
`young age
`
`toxic derivative+
`total dose*,
`total dose (renal dysfunction)
`total dose*, alcohol, diabetes, prior
`disease*
`total dose*, slow hydroxylator**
`drug
`levels over time*
`drug
`levels*, P450 3A phenotype**
`total dose*
`toxic derivative**+,
`total dose*,
`total dose* (adenomas)
`level*, rheumatoid
`drug
`total dose
`total dose, ? drug/drug
`total dose
`total dose*, prior
`
`diseases
`
`interaction
`
`liver disease*
`
`hepatic metabolism.
`abnormal
`genetic effects.
`induction.
`
`of drug-
`from Farrell GC. Management
`(with permission)
`(Modified
`induced
`liver disease, Chap 8. In: Drug
`Induced Liver Disease, Chur-
`chill Livingstone,
`London,
`1994, pp 163).
`
`APOTEX EXHIBIT 1041
`Apotex v. Alkermes
`IPR2025-00514
`
`

`

`in slow hydroxylators
`whose toxicity is seen primarily
`(a genetic effect) or with acetaminophen
`or other
`agents in chronic alcoholic patients
`in whom there is
`enhanced metabolism
`to a toxic derivative of usually
`safe doses of the drug. It should be appreciated
`that
`dosage and duration of therapy are not the only fac-
`tors in drug toxicity of such predictable
`toxins, and
`that other
`individual metabolic variations must also
`play a part. These metabolic aberrations
`(a form of
`superimposed
`idiosyncracy) probably account for the
`infrequent clinical manifestations of toxicity with these
`drugs. A more comprehensive
`list of agents which may
`accumulate
`in the body in the presence of liver disease
`and which require dosage adjustment
`to prevent toxic
`effects has been compiled (2). The presence of underly-
`ing liver diseases may indeed predispose
`to greater
`dose-dependent drug toxicity, if the drug dosage is not
`appropriately
`adjusted downward, and if the margin of
`safety between
`therapeutic
`and
`toxic concentrations
`(the toxic threshold)
`is small. This effect of antecedent
`liver disease presupposes
`that the drug is eliminated
`primarily by the liver, that it is the parent drug or one
`of the early metabolites
`that is toxic, that the liver is
`sufficiently damaged
`to impair drug elimination and
`that the injury affects the metabolic pathway relevant
`to the drug in question. Otherwise, if a distal metabo-
`lite is toxic, liver dysfunction may actually be protective
`in this type of drug-induced
`liver injury by decreasing
`the formation of such a derivative. Thus, underlying
`liver disease would be conceptually anticipated
`to en-
`hance drug-induced
`liver disease only for dose-depend-
`ent liver toxins with a low therapeutic
`index and only
`under special clinical/pharmacokinetic
`circumstances.
`This may partly account
`for the lack of reports of
`drug-induced
`liver injury in patients with prior stable
`hepatic damage. The paucity of such reports may be
`related also to the instinctive
`reluctance
`to prescribe
`potentially hepatotoxic drugs to patients with liver dis-
`ease
`liver disease often results in a low
`Second, chronic
`serum albumin
`concentration.
`Inasmuch
`as many
`drugs in serum are highly protein bound, a lower pro-
`tein with decreased drug binding may result in greater
`distribution of free (unbound) drugs to their target or-
`gans with subsequent
`tissue toxicity (3). An example of
`such adverse effects may be the greater sedative effect
`of diazepam
`in cirrhotic patients (4), in part at least,
`due to a greater penetration of the free drug into the
`brain (5). Similar observations have been reported
`for
`triazolam (6). Thus, decreased protein binding of drugs
`may be another adverse effect of underlying chronic
`liver disease on predictable drug toxicity. However, al-
`though
`the greater concentrations
`of unbound benzo-
`
`Antecedent
`
`liver injury and drug toxicity
`
`diazepine may affect cerebral function adversely, there
`is only scanty evidence that such a process with other
`drugs contributes
`to damage of hepatocytes.
`Thus, in patients with liver disease, hypoalbuminem-
`ia may play a role in aspirin-induced hepatic damage.
`In patients hospitalized with alcoholic liver disease, the
`unbound plasma salicylate concentration
`after a single
`oral dose of aspirin (1.2 g) was increased (7), while the
`urinary excretion of total or unbound aspirin was not
`affected. There was no comment as to whether the liver
`disease became worse. Another study confirmed
`the in-
`crease in unbound plasma concentration of aspirin, in
`inverse proportion
`to the serum albumin,
`in patients
`with acute rheumatic
`fever. The ratio of unbound
`to
`bound salicylic acid was directly related to the rise in
`transaminases
`(AST, ALT) (8). Moreover, others have
`shown that aspirin, through activation of its metabolite
`into salicyl-CoA,
`leads to sequestration of extra mito-
`chondrial CoA and eventual impairment of beta oxi-
`dation of fatty acids to produce hepatic microvesicular
`stea.tosis (9). Thus, hypoalbuminemia
`in liver disease
`increases the risk of aspirin hepatotoxicity. We are un-
`aware of other drug-induced
`liver damage related
`to
`drug binding. It is conceivable
`that a similar adverse
`phenomenon may affect patients with other causes of
`hypoalbuminemia
`as occurs
`in malnutrition
`and
`chronic protein
`loss through
`the kidneys or the intes-
`tine.
`
`Toxicity by altered pharmacodynamics
`In contrast
`to altered drug removal or binding, some
`drug toxicity may be exerted at the tissue (receptor)
`level. Such effects may be caused by altered pharmaco-
`dynamics. For instance, in the presence of chronic liver
`disease, it is known that the brain is more sensitive to
`the sedative effects of benzodiazepines, morphine and
`possibly other soporific agents (4). Moreover,
`the kid-
`ney in such individuals
`is more sensitive
`to prosta-
`glandin
`depletion
`from
`non-steroidal
`analgesics
`(10,l l), and to aminoglycosides, which tend to be more
`toxic to renal tubules of cirrhotic patients
`(12). The
`precise mechanisms of these adverse effects are not
`clear. While these effects of drugs in cirrhosis are not
`exerted on the liver, they still fall into the category of
`adverse drug effects in the presence of underlying cir-
`rhosis. A general compilation of these agents has re-
`cently been published (2).
`to toxins may
`A similar type of hepatic sensitivity
`be seen also in patients with liver disease. This is most
`likely to be in the realm of hepatic response to poten-
`tial toxins. Although
`little studied, conceptually, a dis-
`eased liver may exhibit impaired Kupffer cell function
`in detoxifying endotoxin, may have aberrant
`intercell
`
`1099
`
`

`

`S. Schenker et al.
`
`TABLE 2
`in liver disease
`Drug pharmacodynamics
`Sensitivity of brain in chronic liver disease to benzodiazepines,
`morphine, etc.
`Sensitivity of kidneys to prostaglandin depletion and to
`aminoglycosides
`? Sensitivity of diseased liver to respond to “toxins”
`l ? Endotoxin
`l ? Kupffer cell function (tumor necrosis factor)
`l ? Intercell signaling
`l ? Cytokines
`l ? Mitochondrial
`
`integrity (GSH)
`
`signaling and abnormal cytokine release or may have
`depleted mitochondrial
`reduced glutathione
`(GSH)
`stores with adverse effects on mitochondrial
`integrity
`and function
`(Table 2). Examples of this are the en-
`hancement of acetaminophen,
`alcohol, endotoxin and
`tumor necrosis
`factor
`toxicity
`to mitochondria
`de-
`pleted of antioxidants
`(GSH, S-adenosyl methionine),
`a condition which may prevail in patients with chronic
`alcoholic
`liver disease (13-18). The possible contri-
`bution of malnutrition
`(i.e., with acetaminophen
`toxic-
`ity) (19), or of increased hepatic iron stores (favoring
`oxidative stress) (20) in such patients needs investiga-
`tion. Increase
`in hepatotoxicity
`due to endotoxin
`(a
`possible mediator of alcoholic liver disease) by estrogen
`may be another example of such enhanced sensitivity
`of the liver to ethanol
`in women (21). Data regarding
`this aspect of drug toxicity with underlying
`liver dis-
`ease are surprisingly scanty and are needed.
`
`injury
`Idiosyncratic (unpredictable)
`liver in-
`The overwhelming majority of drug-induced
`jury
`is not dose-dependent.
`It appears
`to occur
`in
`highly selected individuals with a genetic proclivity for
`generating an unusual metabolite and/or who develop
`an allergic response
`to such a derivative (1,22). Such
`reactions cannot be anticipated
`in preclinical studies
`and the susceptible
`individuals cannot be identified a
`priori. It is usually only
`in post-marketing
`surveys,
`which target
`larger populations,
`that the problem
`is
`detected. Inasmuch as the toxic effect is not dependent
`on intact (normal) hepatic metabolism,
`impairment of
`it is unlikely to promote drug toxicity. Thus, the pres-
`ence of prior hepatic dysfunction would not be ex-
`pected to induce or worsen idiosyncratic
`liver damage
`by drugs, unless the liver is more susceptible (more sen-
`sitive) to the process of damage as a result of decreased
`defense systems due to the liver injury (Table 2). The
`issue has not been adequately
`studied experimentally,
`but is conceptually
`attractive. On the other hand,
`it
`is of interest
`that increased problems with halothane
`
`1100
`
`in patients with
`anesthesia have not been reported
`prior
`liver disease undergoing portal systemic shunt
`operations. The rarity of this idiosyncratic
`reaction,
`however, makes interpretation of such lack of data dif-
`ficult. Other research
`is also needed
`to determine
`if
`prior liver damage may actually decrease the formation
`of some toxic metabolites and thus be protective.
`
`Summation of Toxic Effects
`is the incremental ef-
`Another aspect of this problem
`fect of drug-induced
`liver injury added to antecedent
`hepatic damage. Clearly, such a summation needs to
`be avoided, or at the least diagnosed early enough to
`minimize cumulative
`injury. The same principle guides
`the current recommendation
`that patients with chronic
`hepatitis C be vaccinated for hepatitis A and B, if not
`already immune to these infections (23).
`
`Diagnostic Difficulty
`liver injury
`Usually the development of drug-induced
`is heralded by the onset of new symptoms (fatigue, my-
`algias, nausea, abdominal pain and, eventually, jaun-
`dice) and abnormal
`liver tests (2). This provides an op-
`portunity
`to stop a potentially
`offensive drug and
`watch for diagnostically
`helpful
`resolution of these
`findings. With the presence of underlying
`liver disease
`such monitoring may be more difficult. Only detection
`of incremental
`symptoms and/or
`laboratory
`tests is
`helpful. Hence, recording of a good baseline and more
`frequent
`than customary clinical and biochemical
`fol-
`low up of the patient may be needed to detect early
`drug-induced
`injury.
`
`in Underlying Liver Disease
`Drug Metabolism
`A quantitative assessment of this problem is clearly rel-
`evant to proper drug dosing in patients
`treated with
`agents that exhibit dose-dependent
`drug toxicity. The
`subject of drug elimination (hence, possible dosage ad-
`justments)
`in chronic liver disease has recently been re-
`viewed (2) and will be commented on here only briefly.
`In general, hepatic drug elimination depends on
`drug binding to plasma protein, hepatic blood flow (in-
`cluding capillarization)
`and hepatic metabolism. Each
`of these factors may have a bearing on drug hepatotox-
`icity, but this will vary with the nature of the drug.
`For example, some agents with low protein binding (i.e.
`acetaminophen) will not be influenced by a low serum
`protein or the presence of other drugs which may com-
`pete for a priori lower protein binding reserve. Highly
`bound
`(>90%) agents with a low clearance, on the
`other hand, may exhibit greater penetration
`to receptor
`sites in the presence of decreased binding. Normally,
`high clearance drugs depend primarily on liver blood
`
`

`

`flow and those with a low extraction on hepatic bio-
`transformation
`(intrinsic clearance). In the presence of
`chronic
`liver disease, on the other hand, metabolism
`may become rate-limiting
`(2). High extraction agents,
`however, may be particularly
`affected by liver disease
`when the drugs are given orally. The removal of such a
`drug by the chronically diseased liver is often markedly
`decreased due
`to portosystemic
`shunting of blood
`(spontaneous or surgical), resulting in a high concen-
`tration of the drug in blood (reduction
`in first-pass ef-
`fect). An example of this is the greater analgesic/seda-
`tive effect of oral demerol (meperidine)
`in patients with
`cirrhosis as compared
`to individuals with a normal
`liver (24).
`in using dose-de-
`There are several major concerns
`pendent hepatotoxic drugs in patients with liver dis-
`ease. The first problem
`is our inability to define, with
`precision,
`the degree of impairment of liver function
`relevant to elimination of a particular drug in a given
`patient. There
`is, at present, no single equivalent of
`the creatinine clearance-like
`test (as for renal disease)
`in patients with liver disease. It is generally appreciated
`that the severity of liver disease correlates roughly with
`the Child-Pugh classification or Maddrey’s Discrim-
`inant Index, but these are gross indices for any one
`patient. Similar considerations
`apply
`to pharmaco-
`kinetic measurements
`of
`liver function
`(antipyrine,
`aminopyrine, galactose clearances) (25). Second, differ-
`ent forms of liver disease vary appreciably
`in their ef-
`fects on drug handling. For instance, acute liver disease
`often affects drug elimination
`less than chronic disease
`(cirrhosis) of apparently
`similar severity (25). Chol-
`estasis also tends to decrease drug biotransformation
`more (for many agents), as compared
`to hepatocellular
`disease. Finally, oxidative processes are generally more
`deranged
`than conjugation,
`especially with mild or
`moderate
`liver disease (25-29)
`and oxidative detoxi-
`fication varies substantially with the severity and type
`
`TABLE 3
`Selective effect of liver disease on hepatic cytochrome
`
`P-450
`
`P-450 2~19
`S-Mephentoin
`Clearance
`(ml/min)
`
`P-450 2D6
`R-Mephentoin
`Clearance
`(mUmin)
`
`(8)
`Control
`(9)
`Mild liver disease
`Moderate
`liver disease
`* pco.05
`
`(9)
`
`1987
`745*
`72*
`
`24
`21
`25
`
`JG, Hoofnagle
`Arns PA, Adedoyin A, DiBisceglie AM, Waggoner
`JH, Wilkinson GR, Branch RA. Mephenytoin
`disposition
`and serum
`bile acids as indices of hepatic
`function
`in chronic viral hepatitis. Clin
`(with permission).
`Pharm Ther 1997; 62: 527-537
`
`Antecedent
`
`liver injury and drug toxicity
`
`isozyme
`the cytochrome
`of liver disease and with
`needed for a particular drug biotransformation
`(30-
`32) (Table 3).
`Accordingly, it is a challenge to adjust dosages of such
`drugs in patients with both acute and chronic liver dis-
`ease. In mild acute disease, no change or only a modest
`alteration may be needed. In chronic hepatic disease, the
`general rule is to employ an arbitrarily
`selected dose
`(often one half), depending on Child-Pugh classification
`of severity of disease, observe for specific end points (if
`present),
`i.e., lower heart rate with propranolol or in-
`creased pulse rate with theophylline
`and/or measure
`drug blood levels (as with propranolol or theophylline)
`(2529). For example, acetaminophen
`is a safe analgesic
`in patients with modest liver disease (without chronic al-
`cohol abuse), as its metabolism
`is generally well pre-
`served in patients with mild/moderate
`liver impairment
`(33). With severe hepatic dysfunction, acetaminophen
`metabolism was decreased significantly
`in one study
`(34). In this study, using a single dose of acetaminophen,
`patients with severe liver disease (cirrhosis characterized
`by hyperbilirubinemia,
`hypoalbuminemia,
`prolonged
`prothrombin
`time, ascites, varices), but not those with
`milder disease, had significantly longer acetaminophen
`half-life and greater ratio of acetaminophen/glucuroni-
`de and acetaminophen/sulfate
`conjugates
`than healthy
`controls. However, their urinary excretion of cysteine
`and mercapturic acid conjugates - metabolites
`that re-
`flect conversion of acetaminophen
`to reactive hepato-
`toxic compounds
`- were normal. The excretion of cys-
`teine and mercapturic acid rises with increasing doses of
`acetaminophen
`(35,36). Therefore,
`there is no evidence
`to support the interdiction of acetaminophen
`in patients
`with chronic
`liver disease, provided
`low therapeutic
`doses (2 g/d) are not exceeded and alcohol is avoided.
`Nonetheless,
`it is prudent
`to monitor
`the patients’ clin-
`ical course and liver tests because the effect of long-term
`acetaminophen dosing is unknown.
`Special attention must be paid to the effects of en-
`zyme
`induction or decrease
`in drug metabolism by
`other agents in the presence of liver disease. An ex-
`ample of induction
`is the effect of chronic ethanol use
`on the formation of a toxic metabolite of acetamino-
`phen in liver and kidney (37). Since such effects are
`seen in patients with alcoholic liver disease, it is evident
`that the process of induction
`is not vitiated by the pres-
`ence of cirrhosis. Again the use of only 2 g acetamino-
`phen/day
`is felt to be safe in such patients (13,38). An
`example of impaired degradation
`(which could also be
`caused by liver disease) is the effect of ketoconazole on
`the metabolism of cisapride with accumulation of the
`parent drug, sometimes
`to cardiotoxic
`levels (39). The
`effects of liver damage on induction and inhibition of
`
`1101
`
`

`

`S. Schenker et al.
`
`to have been little
`
`drug metabolism, however, appear
`studied, so far.
`The changes in drug metabolism caused by liver dis-
`ease are an extension of the large genetic variability
`(polymorphism)
`of drug metabolism
`in normal man
`(40). Testing of various population
`groups for drug
`toxicity prior
`to drug distribution
`is a safety valve in
`defining drug dose and lack of toxicity, even over a
`wide range of metabolic activity of the liver.
`
`Idiosyncratic Drug Hepatotoxicity with Prior
`Liver Injury
`this is the most common
`As commented on earlier,
`type of drug toxicity
`to the liver and it is believed to
`depend on the presence of an unpredictable metabolic
`pathway with
`the generation of a toxic metabolite,
`which, by itself or by serving as a haptene for an im-
`mune response,
`induces
`liver injury. It seems evident
`that the presence of underlying
`liver disease should not
`promote
`this mechanism of injury which is likely genet-
`ically/immunologically
`determined.
`In fact, liver injury
`theoretically could interrupt
`the metabolite-generating
`pathway. As conceptualized
`earlier, it is possible that
`the
`idiosyncratic
`drug
`toxicity process may be en-
`hanced by the presence of prior liver disease. To our
`knowledge, however, there are no clinical examples of
`such reactions. Possible exceptions are the apparently
`increased hepatic injury with methotrexate
`in individ-
`uals with diabetes, obesity and alcoholism
`(41), and
`those given niacin with prior liver disease (42). These,
`however, are examples of possible potentiation of dose-
`dependent
`toxins. Another example
`is the worsening
`of liver damage
`in hepatitis C patients with alcohol
`consumption. More data in this area are needed, but
`will be difficult
`to obtain, as there are no animal
`models for idiosyncratic
`toxic drug reactions, and for
`most of these precise mechanism(s) of injury are uncer-
`tain.
`A more productive approach would be the develop-
`ment of techniques which can identify susceptible
`in-
`dividuals. This would require prior knowledge of the
`metabolic pathways
`for each drug
`that causes
`this
`type of injury and the use of surrogate markers
`for
`assessing this. This approach has been utilized with
`dose-dependent
`hepatotoxic
`drugs, such as perhexi-
`line maleate
`using debrisoquine
`as a metabolic
`marker (43), and with cyclosporine employing
`labeled
`erythromycin
`(44). When isoniazid toxicity was felt to
`depend fully on rapid acetylator status, this too could
`be assessed using other agents. This mechanism
`is
`now in doubt, however (45,46). The rarity of the idio-
`syncratic reactions renders genetic testing, at present,
`cost-ineffective. However,
`in patients with such reac-
`
`1102
`
`there may be merit
`tions
`toxicity
`in relatives.
`
`in studying
`
`in vitro drug
`
`Summation of Effects - the Role of Monitoring
`Since it is not possible to predict when idiosyncratic
`hepatotoxins may cause
`liver damage, and patients
`with underlying
`liver disease may require such medi-
`cations, early detection of such added injury is needed.
`This usually implies patient monitoring. There are 2
`types of surveillance - symptoms/signs and laboratory
`tests.
`ac-
`always)
`(but not
`is often
`Hepatotoxicity
`companied by symptoms of malaise, anorexia, abdomi-
`nal pain and fever (25,47). While non-specific,
`these
`symptoms should precipitate a prompt patient evalu-
`ation for drug hepatotoxicity. This is especially true in
`the first 6 months of drug use, when such reactions are
`much more
`likely (47). Such patient-generated
`re-
`porting
`involves
`the patient,
`is inexpensive and ef-
`ficient, and should be mandated. Detection of injury
`prior to the development of jaundice
`is especially use-
`ful, as onset of hyperbilirubinemia with hepatocellular
`injury implies a more serious prognosis (25,48). This is
`especially likely with underlying
`liver injury.
`The use of liver tests for monitoring drug hepatotox-
`icity is more controversial. While often advocated by
`drug manufacturers
`(perhaps partly for legal reasons),
`such an approach has a number of difficulties. First,
`mild increases in aspartate and alanine aminotransfer-
`ase (AST and ALT) may be seen with the use of a num-
`ber of therapeutic agents, i.e. propylthiouracil(49),
`ison-
`iazid (25), tacrine (50) without evidence of progressive
`liver disease and with eventual normalization of results
`while remaining on the drug. It is difficult, therefore, in
`the absence of symptoms to determine which abnormal
`results are clinically important. The usual cut off for
`concern
`in patients with previously normal
`tests is a
`three-fold
`increase in AST and ALT, and this mandates
`close follow up of tests, and often a decision about stop-
`ping the drug (51). However, this is an arbitrary number
`and must be used in the clinical context with each indi-
`vidual. In patients with a priori elevated transaminases,
`there are no guidelines as to what constitutes a signifi-
`cant increase. Extrapolating
`from data in patients with
`no liver disease, an increment of about 50-100 IU/l
`above a verified baseline, especially if sustained or actu-
`ally rising, would be a logical cause for concern. Con-
`comitant clinical symptoms would be a reinforcing argu-
`ment. The degree of increase for stopping the drug will
`depend on the baseline values, and will logically be lesser
`with higher initial values and impaired clinical condition
`of the patient. Too many individual factors enter into
`such equations to permit dogmatic recommendations
`at
`
`

`

`this time. Second, while drug hepatotoxicity usually oc-
`curs early in treatment,
`it is difficult to stipulate specific
`time intervals for testing. The usual parameters are on a
`monthly basis, but this can miss intervening abnormali-
`ties. After an uneventful 6 months on a drug, longer test-
`ing intervals would seem reasonable and more efficient.
`Third, with the rarity of clinically meaningful
`toxicity
`(O.Ol-1.0%) the cost/efficiency of such monitoring has
`been questioned
`(2552).
`In fact, even with isoniazid,
`which has a clinical toxicity frequency of about 1 .O%, re-
`cent recommendations
`have shifted away from labora-
`tory monitoring
`and toward clinical assessment (53).
`Thus, at present,
`laboratory monitoring
`for idiosyn-
`cratic drug-induced hepatotoxicity, although often used,
`is of debatable value. These considerations apply to pa-
`tients without underlying
`liver disease. In the presence
`of such dysfunction, one should logically lean more to
`carrying out laboratory
`testing in the hope of detecting
`early any incremental hepatic damage. Possibly, future
`availability of at-home enzyme testing could increase the
`cost efficacy of such a monitoring. The presence of
`underlying liver disease may even enhance patient safety
`by promoting greater vigilance on the part of both pa-
`tients and their physicians.
`Thus, to answer briefly the questions posed in the
`introduction, we believe that antituberculous
`therapy
`should not be altered
`in the presence of underlying
`liver disease
`if the indication
`for such treatment
`is
`strong. The dose of pyrazinamide
`should probably be
`kept low in patients with significant liver disease (Table
`1) or avoided, and a careful baseline of liver tests
`should be obtained
`to permit meaningful clinical and
`biochemical monitoring. Such a follow up would be
`especially important
`in the early part of therapy (first
`6 months) and the drugs should be discontinued with
`onset of clinical symptoms of liver disease and/or a sus-
`tained
`increase
`in transaminases. Clearly,
`the drugs
`should be stopped before the onset of hyperbilirubine-
`mia. The use of urine bilirubin testing by the physician
`(and perhaps educated patients) may help in this re-
`gard.
`
`Conclusions
`liver disease
`We conclude as follows. First, underlying
`should meaningfully
`impact only drug-induced hepato-
`toxicity which is predictably dose-dependent,
`and only
`with drugs with a low therapeutic
`index. Adjustment of
`drug dosage for agents eliminated by the liver should
`mitigate this problem (see Fig. 1). The same principles
`apply to predictable non-hepatic drug toxicity. Second,
`idiosyncratic drug-induced
`liver injury would theoretic-
`ally be enhanced by prior liver injury only if the liver is
`
`Antecedent
`
`liver injury and drug toxicity
`
`PRE-EXISTING LIVER DISEASE
`
`POTENTIALLY HE~ATOTOXIC DRUGS
`
`DDSE-DEPENDENT
`
`1
`ADJUST DOSAGE +
`
`CONSIDER
`PHARMACODYNAMlC
`EFFECTS (Table 2)
`
`J
`
`DOSE-INDEPENDENT
`
`-1
`NO DOSAGE ADJUSTMENT
`
`CONSIDER PLASMA BINDING ++
`
`/
`
`- LWER TEST SASEUNE c
`
`MORITOR CLINICAL COURSE
`AND LWER TESTS
`liver
`Fig. 1. Clinical approach to patients with underlying
`disease who require treatment with potentially hepatotoxic
`drugs. Use of this general algorithm requires knowledge of
`the mechanism
`(dose-dependent or not) of the drug used, its
`plasma binding and pharmacodynamic
`(tissue sensitivity)
`aspects of the drug in such patients. + Dosage may need to
`be decreased
`in amount andlor frequency.
`++ Highly
`bound drugs may be shunted to tissue sites with lower pro-
`tein binding, resulting in greater toxicity.
`
`sensitized to a toxic process. While there is some evi-
`dence for this with various agents and viruses, such data
`for idiosyncratic
`drug-induced
`liver injuries are not
`available. Future studies in this important area which
`can be assessed experimentally with agents that cause
`liver damage in animal models may alter our approach
`to drug use in patients with underlying
`liver damage.
`Such extrapolations will not be easy. The severity and
`type of biochemical
`injury will be important
`factors to
`consider. Third, the presence of underlying
`liver disease
`may enhance
`some pharmacodynamic
`(tissue sensi-
`tivity) aspects of drug toxicity (i.e., NSAID’s, amino-
`glycosides). Fourth, the main concern about drug use in
`individuals with prior liver injury is summation of unto-
`ward effects. This is akin to summation of various seda-
`tives on the brain or the compound effects of hepatitis C
`and alcohol. This, however, is not a contraindication
`to
`the use of such drugs, when needed, but rather points to
`the need for greater vigilance to detect early injury. In
`this context, the value of clinical self-monitoring
`is evi-
`dent. Laboratory
`follow up for rare (idiosyncratic)
`reac-
`tions is more tenuous. However, with underlying
`liver
`disease, one can make a more compelling case for it.
`Overall, underlying
`liver disease should not be an auto-
`matic contraindication
`to the use of potentially hepato-
`toxic drugs. Rather,
`the patient needs to be followed
`more closely to detect any incipient incremental
`liver in-
`jury. As in all patient care, the benefit/risk ratio for each
`
`1103
`
`

`

`S. Schenker et al.
`
`drug and in each patient needs to be individually as-
`sessed.
`
`in chronic
`
`liver
`
`Reference
`induced hepa-
`HJ, Maddrey WC. Toxic and drug
`1. Zimmerman
`titis. In: Schiff L, Schiff ER, editors. Diseases of the Liver, 7th
`ed, Vol 1, Chap 28. Philadelphia:
`JB Lippincott;
`1993. p. 707-
`83.
`JM. Drug administration
`JF, Brogard
`2. Westphal
`disease. Drug Safety 1997; 17: 47-73.
`of chlordia-
`toxicity
`3. Greenblatt
`DJ, Koch-Weser
`J. Clinical
`zepoxide
`and diazepam
`in relation
`to serum albumin
`concen-
`tration:
`a report
`from
`the Boston Collaborative
`Drug Surveil-
`lance Program. Eur J Clin Pharmacol
`1974; 7: 259962.
`4. Branch RA.
`Is there
`increased
`cerebral
`sensi