Drug-induced liver injury，drug disposition and metabolite profiling

Wei TANG

(Department of Drug Metabolism and Pharmacokinetics，Shanghai Chem Partner，Shanghai201203，China)

Drug-induced liver injury，drug disposition and metabolite profiling

Wei TANG

(Department of Drug Metabolism and Pharmacokinetics，Shanghai Chem Partner，Shanghai201203，China)

Drug-induced liver injury(DILI)is a leading cause responsible for the failure of drug development and for the withdrawal of commercial drug products.The high frequency of DILI is due in part to the physiology of the liver，since in many cases elimination of drug molecules from the body is dependent on hepatic clearance via either metabolism or biliary excretion.Many of the mechanistic details underlying DILI remain poorly defined in spite of extensive studies of the pathogenesis.In this regard，metabolomics may become a powerful tool for investigation of DILI，leading to better mechanistic understanding and biomarkers identification.

liver injury;pharmacokinetics;metabolite analysis

Drug-induced liver injury(DILI)accounts for a high percentage of clinical liver failures，poses a major obstacle to drug development and contributes significantly to the withdrawal of commercial drug products［1-2］.Arecent example involves lumiracoxib whose marketing authorization was revoked by the regulatory agencies in the EU，Australia and Canada，following several case reports of serious treatment-related hepatotoxicity that led to liver transplantation or fatality［3-5］.Although epidemiological studies of DILI have identified females and elders as the most susceptible［6-7］，these risk factors are likely dependent on a specific offending drug in question.This is because the term DILI combines a number of different hepatic reactions following the insult by agents with wide varieties of indications and chemical structures［8-9］.In this context，aspirin-and valproic acid-induced liver injury occurs more frequently in young children than in adult patients［10-11］.The high frequency of DILI can be attributed in part to the physiological function of the liver in that the entire drug load absorbed following oral administration is subject to metabolism and/or biliary excretion before entering into the systemic circulation.Molecules that have escaped from the first-pass hepatic extraction circulate back to the liver for eventual elimination unless other organs，such as the kidneys，participate in the clearance of the drug.Metabolism of drug molecules sometimes produces chemically reactive species capable of modifying covalently hepatic proteins and/or nucleic acids.The parent drug and metabolitesmay also inhibit hepatic transporters and therefore interfere with hepatic uptake and/or efflux of bile acids and bilirubin.These events lead potentially to perturbation of endogenous cellular metabolism，which can be either the cause or consequence of DILI.This manuscript intends to provide a brief overview of the proposed mechanisms underlying DILI and to discuss possible application of metabolomics in studies of DILI.

1 Drug disposition and DILI

DILI may result from drug overdose，consistent with the presence of a threshold below which use of the agent is deemed safe or low risk.Such hepatotoxicity is likely associated with the direct cytotoxic effect produced by an offending drug and/or its metabolites，and acetaminophen(APAP)in this regard may serve as an example.The drug is an over-the-counter analgesic generally well tolerated and effective within the recommended dose range.On the other hand，severe hepatocellular damage occurs if a large quantity of APAP is administered in a single dose or over a short interval，with the potential of progressing to liver failure and consequently fatality［12-13］.The hepatotoxicity exhibits a rapid onset of symptoms，usually in a time frame of 48 h after drug ingestion;liver enzyme elevation and hyperbilirubinemia become apparent within 12 to 36 h，while abnormality of hepatic functions peaks on the third day.Pathological characteristics of DILI include extensive centrilobular necrosis，consistent with the hypothesis implicating a toxic metabolite in the pathogenesis of cellular damage.Patients suffering from APAP-related liver injury have generally experienced reasona-ble prognosis，exhibiting a 60%-80%transplant-free survival rate when N-acetylcysteine(NAC)is administered promptly［14］.Because NAC is capable of scavenging chemically reactive electrophiles，its effectiveness as an antidote for management of APAP over dose reinforces the idea that reactive species formed during APAP metabolism is a culprit responsible for the DILI.

DILI can also be'diosyncratic'，characterized by the lack of a clear dose-response relation and delayed onset of clinical symptoms after weeks to months of exposure to the offending drug.Hepatotoxicity of this type usually escapes the detection by preclinical and clinical safety investigations and occurs in a small subset of treated subjects in spite of tolerability shown by the majority of the patient population.For example，use of diclofenac，a nonsteroid anti-inflammatory drug(NSAID)，is associated with rare but severe hepatotoxicity，clinical symptoms of which include jaundice and elevated serum transaminase activity［15-16］.However，these features of toxicity are normally not evident until after 1-3 months into diclofenac treatment，and the incidence ranges from 1 to 5 per 100 000 exposed patients，with fatality at about 10%.The data from liver biopsies revealed extensive cell necrosis and resemblance to viral hepatitis or chronic hepatitis accompanied by cholestasis.Diclofenac-induced liver toxicity has been termed a hepatocellular form of injury attributed to metabolic instead of immune idiosyncrasy，since rechallenge of patients with the drug rarely resulted in rapid adverse responses［17-19］.Putative toxic metabolites identified for the drug include diclofenac acyl glucuronide and quinonei mine derivatives［20］.Reactive quinone imine species were also detected for lumiracoxib［21］.Another example of idiosyncratic DILI involves flucloxacillin，the use of which sometimes causes cholestatic liver injury featuring pruritus and prolonged jaundice［22］.An estimated rate of the DILI is about 1 per 15 000 treatment courses，while delayed onset of clinical symptoms could take up to 45 d.In other words，the toxicity in some cases dose not become evident until several weeks after cessation of the antibiotic.Pathological findings include canalicular cholestasis，eosinophilia，interlobular biliary epithelial tissue degeneration and severe bile duct damage，with a minimum to absence of hepatocyte necrosis［21］.The adverse reactions to flucloxacillin were immediate in patients re-challenged inadvertently with the antibiotic，implicating therefore an immune component in pathogenesis of flucloxacillin-induced hepatotoxicity［23］.

Cytotoxic effects of drug molecules and metabolites are attributable to their insult on mitochondria［24-25］.The organelle consists of an outer and inner membrane，each of which expresses ion channels，fatty acid carriers and electron-transport complexes;its matrix contains high levels of proteins and is slightly alkaline and negatively charged.This structural feature of mitochondria appears in favor of concentrating lipophilic，cationic molecules，some of which can be toxins capable of interfering with mitochondrial respiration and/or fatty acid β-oxidation(FAO).Both the respiratory and FAO reactions are essential for mitochondria to function as an energy supplier，generating ATP in support of hepatocyte survival;therefore，their disruptions are likely to result in'nergy shortage'within the affected cells.Excessive levels of free fatty acids due to FAO inhibition are also deleterious through a separate pathway(vide infra).A pathological observation of hepatic steatosis would suggest an overproduction of triglyceride secondary to the impairment of FAO.During mitochondrial respiration，reactive oxygen species are produced as by-products whose neutralization is dependent on superoxide dismutase(SOD)，glutathione(GSH)and glutathione reductase(GR).An imbalance of this detoxification process resulting from the loss of SOD and GR enzyme activities，GSH depletion and/or formation of reactive drug metabolites leads to oxidative stress to the organelle，triggering mitochondrial membrane permeability transition and perturbed calcium homeostasis.In addition，drug molecules and their metabolites may interfere directly with the mitochondrial DNA(mtDNA)replication process，impairing the synthesis of mtDNA-encoded respiratory polypeptides and consequently the respiratory reaction.Mitochondrial dysfunction associated with ATP depletion leads to hepatocyte necrosis characterized by rupture of plasma membrane and leakage of intracellular constituents，some of which are stimuli of the innate immune system in the liver［26］.Subsequent release of cytokines and chemokines from activated immune cells including Kupffer，NK and NK T cells further exacerbate hepatic tissue injury［27-28］.

Alternatively，mitochondria may function as a mediator in the cascade of cellular damage as the organelle，not only supports cell survival but also regulates the programmed cell death machinery，namely，apoptosis.In this context，cytotoxic bile acids and free fatty acids accumulate when their uptake，metabolism and/or efflux are perturbed by drug molecules or metabolites.For example，inhibition of the bile salt export pump(BSEP)leads to retention of bile acids in hepatocytes，or alternatively，activation of pregnane X receptor(PXR)and liver X receptor(LXR)up-regulates the uptake transporter CD36，resulting in over-expression of the transporter protein and increased influx of fatty acids in to hepatocytes［29-30］.Excessive bile acids and fatty acids in the liver stimulate interaction of the death receptors with Fas ligand(FasL)，TNF-related apoptosis-inducing ligand(TRAIL)and tumor necrosis factor-α(TNF-α)，respectively.Subsequent activation of caspase-8 mediates the cleavage of Bid in the BH3 family to a truncated Bid(tBid)，followed by trans location of the tBid to the surface of mitochondria，provoking mitochondrial membrane leakage，cytochrome c efflux and eventually apoptotic cell death.Additional paths of apoptosis involve endoplasmic reticulum stress-related unfolded protein response and lysosomal release of the proteases cathepsins，with the former in some cases induced by fatty acids.Both routes lead to mitochondrial membrane permeabilization and consequently hepatocyte apoptosis.Apoptotic cell death promotes inflammation via activation of Kupffer cells，further aggravating liver tissue injury，such as hepatic fibrosis.

In addition to altering protein functions，alkylation of hepatic proteins by reactive metabolites has the potential to trigger antigen production and consequently DILI.This is the so-called'hapten hypothesis'，according to which it is essential for small molecule drugs and their metabolites to form protein adducts in order to invoke immune responses［31-32］.In this context，drugprotein adducts are taken up and processed by antigen presenting cells，and the resulting immunogenic adducts are presented in association with the major histocompatibility complex(MHC).This leads to proliferation of CD4+and/or CD8+T-cells that are cytotoxic to hepatocytes once self-tolerance is overwhelmed.The extent of protein modification and specific protein modified are likely important factors dictating whether immunological reactions are to take place.Alternatively，drug molecules and metabolites could interact directly with MHC and T-cell receptor to sensitize the immune system and activate T-cells［33-34］.DILI of immunological nature usually exhibits clinical symptoms of fever，rash and eosinophilia.Although characterized by a delayed onset of responses upon initial exposure to an offending drug，DILI occurs upon re-challenge with either the implicated agent or a different drug molecule whose metabolism produces a structurally similar reactive species［35-36］.Antibodies recognizing drug-related antigens sometimes are detectable in the circulation of patients with liver injury.Polymorphic expression of MHC is attributable to DILI.In addition，environmental factors such as inflammation and poly-pharmacotherapy are likely to tip the balance between tolerance and immunity in favor of the latter，contributing to the idiosyncrasy of immune-mediated hepatitis.

DILI may present as cholestasis resulting from accumulation of bile acids in the liver.A possible trigger of the pathogenesis involves disruption by drug or metabolites of active transporters localized in the hepatocyte canalicular membrane，including BSEP，multidrug resistance proteins(MDR)and multidrug resistance-associated proteins(MRP).Because these transporters constitute the regulatory machinery governing bile components and flow rate，their compromised activity leads to the impairment of bile secretion and excessive retention of bile acids cytotoxic to hepatocytes.An elevated level of bile acid induces translocation of the cytoplasmic death receptor Fas to the plasma membrane，initiating cell death processes via interactions with the Fas ligand.Subsequent caspase 8 activation promotes mitochondrial membrane permeability transition，followed by cytochrome c release，activation of caspase3/7and eventually hepatocyte apoptosis［37-38］.Bile acids may also interfere directly with mitochondrial respiration，resulting in oxidative stress，mitochondrial dysfunction and consequently cytotoxicity［39］.Inhibition of biliary bile acid efflux by drug molecules and/or metabolites can be reversible when the inhibitors are competitive substrates of the same bile acid transporter(s).Irreversible inhibition occurs if drug metabolism generates reactive species that not only is a substrate but also capable of alkylating the transporter protein(s)involved in bile acid efflux.Consequently，structurally modified transport proteins lose their activities and fail to facilitate bile acid secretion，leading to the accumulation of bile acids in hepatocytes，hence DILI characteristic of cholestasis.

2 Deficiency of current understanding of DILI

The mechanism of DILI remains poorly defined in spite of recent progress in studies of the pathogenesis.For example，electrophilic species formed from drug metabolism have often been sought during the investigation of DILI，but proteins targeted by those reactive metabolites usually are unknown.In the few cases wherein the modified proteins are identified，their roles often remain masked in the cause of hepatotoxicity.A rat model of diclofenac hepatotoxicity may serve to illustrate a convoluted pattern and time course of cytotoxicity and covalent protein modification［40］.Pathological examination of liver samples from rats treated with diclofenac revealed swollen and apoptic hepatocytes on day 1.Such cellular injury increased progressively on days 2 to 4，and eventually focal necrosis plus reduced bile flow and bile acid secretion became evident on day 5.Protein modification was initially associated with an adduct of 110 ku identified in a separate study as dipeptidyl peptidaseⅣ［41］.This adduct was major on day 1 but diminished to an undetectable level on day 5，at which time the predominant protein adducts were at 85 and 96 ku［40］.Formation of the protein adducts is likely due to diclofenac metabolism that produces reactive acyl glucuronide and quinone imines［20］.Another example involves troglitazone，an antidiabetes agent that was withdrawn from the market as a result of se-vere DILI featuring hepatocyte necrosis，bile duct proliferation and cholestasis.Studies of troglitazone biotransformation have identified several reactive metabolites resulting from P450-mediated oxidation of the hydroxy chromane and thiazolidinedione moieties［42-43］.However，troglitazone is also subject to extensive sulfation catalyzed by the sulfotransferase SULT1A3，leading to the formation of a sulfate conjugate whose plasma exposure was 7-10times that of the parent drug［44-45］.Both troglitazone and its sulfate conjugate inhibit BSEP in vitro，exhibiting IC50values that are comparable to or lower than their respective plasma concentrations following therapeutic doses.Impairment of BSEP following troglitazone treatment could therefore cause accumulation of toxic bile acids in hepatocytes and consequently liver injury in susceptible patients.However，to what extent the parent drug or metabolites actually contribute to the DILI is far from clear.In other words，'evidence'collected by the aforementioned studies usually is not sufficient to pinpoint a'culprit'in the pathogenesis of DILI.It is highly possible that DILI is a result of accumulative/collective effects，with the contributing factors ranging from the patient disease state to genetic predisposition.Therefore，the ability to detect and monitor signals of various biochemical reactions at the onset of hepatotoxicity can aid mechanistic understanding of DILI.

Inability to predict the patient population susceptible to DILI severely hinders the effort to produce safer molecules during drug discovery and development.In the current paradigm，potential drugs are subject to extensive preclinical and clinical safety evaluations before submissions for approvals by the regulatory agencies.There are several obvious flaws in the existing preclinical safety investigations.For example，laboratory animals are homogeneous while patient populations are heterogenous in terms of their genetic make-ups and environmental exposures.Laboratory animals are healthy and free from disease but patients are likely compromised to some extent with their defense mechanisms due to illnesses.These issues impede a direct extrapolation of preclinical findings to humans，although they are supposedly addressed in theory by subsequent clinical trials of assessing safety and tolerability.These trials are usually divided into three phases with the number of subjects in each trial increasing from a dozen to thousands.In spite of these safety precautions，certain drugs approved for prescription still induce hepatotoxicity when their clinical application is extended to a large patient population.DILI of this type has sometimes been termed idiosyncratic，and the occurrence has been estimated to be 1 per 10 000 to 100 000 treatments［6，46］.Such low incidence could have contributed to the'miss'of detection by clinical trials of the flawed drug molecules，since the number of human subjects involved in those trials appears to be statistically under-powered according to the so-called rule of three［47-48］.Thus，an investigation involving 30 000 or more subjects would be necessary in order to detect DILI that takes place in 1 of every 10 000 treated patients.One approach to increase the probability of identifying'flawed'molecules during drug development is to test drug safety in animal models that reflect the most sensitive patient population［49-50］.Another complication is that DILI has been observed in association with the so-called'adaptation'developed by certain individuals during their therapy.In other words，DILI would resolve spontaneously despite continuing or reinstating treatment with the offending drug.For example，the anti-tuberculosis agent isoniazid(isonicotinylhydrazine，INH)causes hepatotoxicity featuring focal necrosis sometimes accompanied by cholestasis.DILI has exhibited a rate of incidents in the range of 0.1%-0.6%and a link to INH metabolism that results in the formation of acetylhydrazine［51］.Further oxidative bioactivation of acetylhydrazine，catalyzed by CYP2E1，produces reactive species that are capable of modifying hepatic proteins and thus potentially responsible for INH-induced hepatotoxicity.Whereas about 20%of the patients may initially experience elevation of serum liver enzymes，they are able to complete INH therapy without progressing to severe DILI［52］.Rechallenge with the drug has also been a common practice since few options are available for the treatment of latent tuberculosis or tuberculosis.An open question is therefore related to what dictates which patient who would experience DILI or adaptation，or be free from any adverse reaction.Because preclinical and clinical safety studies are unlikely to eliminate DILI，an alternative is to search for toxicological biomarkers that are of utility not only for post-prescription surveillance but also for screening of patients suitable for a specific therapy.Medication should become much safer with biomarkers able to distinguish patients who can or cannot develop tolerance to drug treatment.

Traditional biomarkers for detecting DILI include serum aminotransferase and bilirubin levels［53-54］.The former reflects hepatocyte damages that lead to the leakage of cellular enzymes such as alanine and aspartate aminotransferases(GPT and GOT)into the circulation，whereas the latter measures the hepatobiliary excretory function.These assays are accessible on a routine basis，but the readout does not necessarily predict whether an individual would be able to tolerate a specific pharmacotherapy.In addition，elevation of serum enzyme levels sometimes may result from non-hepatic injury.Pathological data are important with respect to characterization of tissue/cellular injury and elucidation of the mechanisms underlying DILI but for obvious reasons they are not qualified as biomarkers.In addition，a particular pattern of liver injury identified by histological examination can sometimes be subject to different interpretation about its cause since there are limited options for tissues to respond upon insults of distinctive natures［55］.With the advance of technology such as nuclear magnetic resonance spectrometry(NMR)and mass spectrometry(MS)，profiling endogenous and exogenous metabolites in serum and urine in principle should be able to provide a specific'fingerprint'for individuals examined，enabling phenotyping and monitoring the patient before，during and after drug treatment.In this context，metabolomics may represent a platform allowing for examination of the multi-organ functional integrity without invasive tissue sampling.

3 Metabolomics in studes of DILI

Metabolomics and metabonomics have been referred to studies of endogenous metabolites with different emphases［56-57］.However，use of the two terms in many cases has become increasingly interchangeable.A recent proposal is that metabolomics be expanded to include the measurement of xenobiotic metabolites，such as those derived from drugs and environmental chemicals，in order to provide a comprehensive description of the human metabolome［58］.The rationale is that to define a chemical based on whether it is purely of endogenous or xenobiotic origin can sometimes be very difficult，if not impossible，and broadening the coverage of metabolomics would avoid the dilemma when dealing with a metabolite that is partially exogenous and partially endogenous.Examples of such metabolites include those formed during drug metabolism via conjugation reactions with endogenous glucuronic acid，GSH and amino acids［59］.In addition，identification and quantification of endogenous and drug metabolites often share the same sample sources，instrumentation platforms and data analysis software，although the requirement of sensitivity for detection may vary.On that basis，metabolomics is adapted in this manuscript to describe metabolite profiling regardless of the origin of metabolites.

The application of metabolomics to studies of DILI is based on premises that①profiles of endogenous metabolites correspond to responses of a biological system to internal and/or external stimuli;and②profiles of drug metabolites represent the overall exposure of patients to drug-related substances，including those that are chemically reactive and capable of modifying proteins and nucleic acids.Quantitative measurements of endogenous metabolites should make molecular phenotypes a linker between pathological and biochemical changes under either internal or external stimulation.Profiling drug metabolites，on the other hand，may reveal toxins formed during drug metabolism and correlate patient susceptibility to DILI with their exposure to toxins.Drugs that generate potentially toxic metabolites include APAP，diclofenac，lumiracoxib，troglitazone，valproic acid，terbinafine and fialuridine［21，59］.Profiling of both endogenous and drug metabolites can therefore be complementary for the investigation of mechanisms underlying DILI，because detection and quantification of exogenous and endogenous metabolites should provide insight at a systems biology level into cellular networks under stress of drug toxicities.For example，metabolite profiling revealed reduced levels of citrate，2-oxoglutarate and succinate in urine of rats or mice administered with APAP，bromobenzene，aroclor 1254(polychlorinated biphenyls;PCB)and 2，3，7，8-tetrachlorodibenzo-p-dioxin(TCDD)，respectively［60-62］.Those endogenous metabolites are intermediates in the tricarboxylic acid(TCA)cycle，and their decreased concentrations in rat urine were accompanied by increased concentrations of 3-D-hydroxybutyrate，glucose，pyruvate，acetate and lactate in rodent plasma following treatment with the hepatotoxins.These data were rationalized in the context that impairment of FAO could account for the failure of using pyruvate by the TCA cycle and reinforce the idea that mitochondria often serve as either a target or a mediator during the pathogenesis of DILI.In the case of APAP，DILI has been attributed to NAPQ formation catalyzed primarily by CYP2E1 with minor contribution from CYP1A2 and CYP3A［63］.Metabolomics studies provided additional mechanistic details by comparing APAP metabolism in CYP2e1-null and wild type mice［64］.Thus，concentrations of NAPQI related thiol conjugates in mouse urine were significantly higher in wild-type than in CYP2e1-null animals when APAP dose was low and non-toxic to the rodents(10 mg·kg-1).On the other hand，urinary thiol conjugates exhibited similar abundance in both strains of mice following a dose of APAP(400 mg·kg-1)that was toxic to wildtype mice but not to CYP2e1-null mice.These seemingly paradoxical findings suggest a switch from CYP2E1-toCYP1A2/3A-catalyzed formation of NAPQI in CYP2e1-null mice，and a possible contribution to APAP cytotoxicity from oxidative stress generated by CYP2E1-mediated APAP metabolism.Such a conjecture is consistent with novel metabolites identified with elevated concentrations in wild-type mice following a toxic dose of APAP(400 mg·kg-1)，including APAP dimer，3-methoxy-APAP glucuronide，and a benzothiazine derivative.Formation of these APAP metabolites is likely mediated by reactive oxygen species，hence indicating cellular oxidative stress.

Metabolomics also has the potential to monitor toxicity development in real time without invasive tissue sampling.This can be particularly advantageous when the technique is applied in conjunction with other'omics'approaches.In a study of APAP-induced hepatotoxicity in mice，plasma concentrations of lactate，acetate，3-D-hydroxybutyrate and lipids were elevated from 15 to 240 min，whereas the plasma pyruvate level decreased initially at 15-30 min and then increased at 60 to 240 min following a toxic dose at 500 mg·kg-1［65-66］.These changes mirror proteomic and genomic analyses of APAP hepatotoxicity in mice in that decreases were apparent as early as 15 min post dosing in the expression of ATP synthase subunits and proteins in the fatty acid β-oxidation pathway［67］.The time sequence of changes is such that the alteration of mitochondrial protein expressions preceded gene responses to APAP treatment，suggesting that gene regulations either exacerbate toxicity or begin cell recovery，but are unlikely the target of initial insult.Together with system biology，metabolomics may therefore advance mechanistic understanding of DILI and facilitate the search for biomarkers characteristic of high molecular resolution in terms of their sensitivity and specificity towards a drug-related AE.One such example involves studies by NMR of rats dosed with an investigational drug for HIV treatment［68］.The compound caused hepatotoxicity in rats featuring microvesicular steatosis，centrilobular hypertrophy，single cell and focal necrosis.Analyses of the urine samples from treated rats revealed a decreased concentration of intermediate metabolites(ie，citrate，2-oxoglutarate and succinate)in the TCA cycle，but significant amounts of dicarboxylic acids such as sebacic and suberic acids.The appearance of dicarboxylic acids is likely due to inhibition of mitochondrial β-oxidation that leads to elevated microsomal ω-oxidation of fatty acids.Impairment of fatty acid β-oxidation also matched the histological detection of lipid accumulation in the liver.This case study suggests that increased dicarboxylic acids together with decreased TCA cycle metabolites are urinary molecular signatures suitable for early identification of hepatic βoxidation dysfunction with improved sensitivity and specificity.

Contributing factors to the onset of DILI may include poor nutrition，disease state，genetics and their combination.For example，choline deficiency is known to cause damage to the liver，since the nutrient is a precursor of methionine and S-adenosylmethionine that is vital for methylation of nucleic acids and proteins.In a clinical study of nutrition，healthy human subjects were fed on a choline-adequate(550 mg/70 kg/d)diet for 10 d to establish baseline，and then on choline-deficient(＜50 mg/70 kg/d)diet for up to 42 days.Their blood was collected at the end of each treatment for GC-MS and LC-MS analyses of plasma［69］.Low dietary choline intake significantly decreased metabolites in the metabolic pathways of choline and methionine.The most significant finding was that about 52%of the participants developed fatty liver following a period of feeding on choline-deficient food，and their plasma metabolite profiles measured at the baseline were already distinguishable from those who did not experience liver injury.In other words，metabolic phenotypes in this case presaged hepatic dysfunction resulting from choline deficiency，and molecular biomarkers discriminating the two phenotypes were those products in metabolic pathways of choline，carnitine，keto acid，and amino acids［69］.This study illustrates that metabolomics should be able to profile the individual metabolic statue that is important to drug treatment.Design of a regimen specific to the patient in question is termed personalized medication.A'proof of concept'study based on metabolomics to predict individual responses to drug treatment was demonstrated with a rat model of APAP hepatotoxicity，wherein urine samples were collected before and after drug treatment，and analyzed by NMR［70］.Multivariate analyses of NMRs pectra showed a correlation between the ratio of APAP glucuronide to the parent drug and NMR signal intensities in the(5.06-5.14)×10-6region of the pre-dose spectra，and a link between low degree of liver tissue damage and large presence of taurine in pre-dose urine.It follows that the concentration of taurine in urine reflects both the ability of liver to protect itself and the availability of inorganic sulfate that is a precursor essential for sulfation reactions.The sulfate in taurine may serve to indicate the capacity of the liver to detoxify NAPQI via sulfation following its formation from APAP bioactivation.In a follow-up study with human subjects administered APAP，patients who experienced GPT elevation were designated as responders and those who did not were nonresponders［71］.Analysis of urine samples by NMR clearly distinguished responders from nonresponders 4-5 d preceding the occurrence of peak serum GPT level in responders，demonstrating the power of metabolomics as a tool in serving post dose surveillance.In addition，the study illustrates that combined profiles of endogenous metabolites，drug molecule plus drug metabolites are more robust than either alone for'prediction'of the DILI，with the metabolite markers including NAPQI thiol conjugates，glycine，methylhistidine，and trimethylamine oxide.A large quantity of NAPQI thiol conjugates found in urine of responders is consistent with the established mechanism of APAP hepatotoxicity，in which NAPQI triggers the cascade of cytotoxic events.High levels of urinary glycine in responders may be rationalized in the context that those patients were'defective'in their hepatic uptake of the amino acid and consequently vulnerable to DILI since glycine has been shown to be hepatoprotective［71］.A link between responders and the amount of methyl-histidine/trimethyl amine oxide in their urine remains speculative，but one of the characteristics associated with'omic'approaches is pattern recognition，such that the response pattern itself is useful even if some of the response signals cannot be deciphered to the level of satisfaction according to existing biochemical concepts.

Whereas application of metabolomics is at its early stage，the approach is positioned in bridging phenotype and genotype of both endogenous and exogenous metabolites and their connections with pharmacological and toxicological responses.Together with other'omics'，metabolomics represents a powerful tool to serve investigation and early diagnoses of DILI.

Acknowledgement:The author thanks Dr.XU Qiu-wei，Safety Assessment，Merck＆Co.，for his contributions to the manuscript.

［1］Shapiro MA，Lewis JH.Causality assessment of druginduced hepatotoxicity:promises and pitfalls［J］.Clin Liver Dis，2007，11(3):477-505，Ⅴ.

［2］Abboud G，Kaplowitz N.Drug-induced liver injury［J］.Drug Saf，2007，30(4):277-294.

［3］http://www.novartis.com.au/Prexige%20press%20 release%2011%20August.pdf

［4］http://www.tga.gov.au/media/2007/070811-lumiracoxib.htm

［5］http://www.medsafe.govt.nz/hot/media/2007/prexige.asp.

［6］Navarro VJ，Senior JR.Drug-related hepatotoxicity［J］.N Engl J Med，2006，354(7):731-739.

［7］Bell LN，Chalasani N.Epidemiology of idiosyncratic drug-induced liver injury［J］.Semin Liver Dis，2009，29(4):337-347.

［8］Chalasani N，Björnsson E.Risk factors for idiosyncratic drug-induced liver injury［J］.Gastroenterology，2010，138(7):2246-2259.

［9］Lucena MI，Andrade RJ，Kaplowitz N，García-Cortes M，Fernández MC，Romero-Gomez M，et al.Phenotypic characterization of idiosyncratic drug-induced liver injury:the influence of age and sex［J］.Hepatology，2009，49(6):2001-2009.

［10］Belay ED，Bresee JS，Holman RC，Khan AS，Shahriari A，Schonberger LB.Reye's syndrome in the United States from 1981 through 1997［J］.N Engl J Med，1999，340(18):1377-1382.

［11］Dreifuss FE，Santilli N，Langer DH，Sweeney KP，Moline KA，Menander KB.Valproic acid hepatic fatalities:a retrospective review［J］.Neurology，1987，37(3):379-385.

［12］Black M.Acetaminophen hepatotoxicity［J］.Annu Rev Med，1984，35:577-593.

［13］Larson AM，Polson J，Fontana RJ，Davern TJ，Lalani E，Hynan LS，et al.Acetaminophen-induced acute liver failure:results of a United States multicenter，prospective study［J］.Hepatology，2005，42(6):1364-1372.

［14］Prescott LF.Paracetamol overdosage.Pharmacological considerations and clinical management［J］.Drugs，1983，25(3):290-314.

［15］Purcell P，Henry D，Melville G.Diclofenac hepatitis［J］.Gut，1991，32(11):1381-1385.

［16］Boelsterli UA，Zimmerman HJ，Kretz-Rommel A.Idiosyncratic liver toxicity of nonsteroidal antiinflammatory drugs:molecular mechanisms and pathology［J］.Crit Rev Toxicol，1995，25(3):207-235.

［17］Helfgott SM，Sandberg-Cook J，Zakim D，Nestler J.Diclofenac-associated hepatotoxicity［J］.JAMA，1990，264(20):2660-2662.

［18］Breen EG，McNicholl J，Cosgrove E，McCabe J，Stevens FM.Fatal hepatitis associated with diclofenac［J］.Gut，1986，27(11):1390-1393.

［19］Banks AT，Zimmerman HJ，Ishak KG，Harter JG.Diclofenac-associated hepatotoxicity:analysis of 180 cases reported to the Food and Drug Administration as adverse reactions［J］.Hepatology，1995，22(3):820-827.

［20］Tang W.The metabolism of diclofenac-enzymology and toxicology perspectives［J］.Curr Drug Metab，2003，4(4):319-329.

［21］Li Y，Slatter JG，Zhang Z，Li Y，Doss GA，Braun MP，et al.In vitro metabolic activation of lumiracoxib in rat and human liver preparations［J］.Drug Metab Dispos，2008，36(2):469-473.

［22］Russmann S，Kaye JA，Jick SS，Jick H.Risk of cholestatic liver disease associated with flucloxacillin and flucloxacillin prescribing habits in the UK:cohort study using data from the UK General Practice Research Database［J］.Br J Clin Pharmacol，2005，60(1):76-82.

［23］Lobatto S，Dijkmans BA，Mattie H，Van Hooff JP.Flucloxacillin-associated liver damage［J］.Neth J Med，1982，25(2):47-48.

［24］Scatena R，Bottoni P，Botta G，Martorana GE，Giardina B.The role of mitochondria in pharmacotoxicology:a reevaluation of an old，newly emerging topic［J］.Am J Physiol Cell Physiol，2007，293(1):C12-C21.

［25］Labbe G，Pessayre D，Fromenty B.Drug-induced liver injury through mitochondrial dysfunction:mechanisms and detection during preclinical safety studies［J］.Fundam Clin Pharmacol，2008，22(4):335-353.

［26］Malhi H，Guicciardi ME，Gores GJ.Hepatocyte death:a clear and present danger［J］.Physiol Rev，2010，90(3):1165-1194.

［27］Adams DH，Ju C，Ramaiah SK，Uetrecht J，Jaeschke H.Mechanisms of immune-mediated liver injury［J］.Toxicol Sci，2010，115(2):307-321.

［28］Martin-Murphy BV，Holt MP，Ju C.The role of damage associated molecular pattern molecules in acetaminophen-induced liver injury in mice［J］.Toxicol Lett，2010，192(3):387-394.

［29］Zhou J，Febbraio M，Wada T，Zhai Y，Kuruba R，He J，et al.Hepatic fatty acid transporter Cd36 is a common target of LXR，PXR，and PPARgamma in promoting steatosis［J］.Gastroenterology，2008，134(2):556-567.

［30］Moya M，Gómez-Lechón MJ，Castell JV，Jover R.Enhanced steatosis by nuclear receptor ligands:a study in cultured human hepatocytes and hepatoma cells with a characterized nuclear receptor expression profile［J］.Chem Biol Interact，2010，184(3):376-387.

［31］Park BK，Pirmohamed M，Kitteringham NR.Role of drug disposition in drug hypersensitivity:a chemical，molecular，and clinical perspective［J］.Chem Res Toxicol，1998，11(9):969-988.

［32］Uetrecht JP.New concepts in immunology relevant to idiosyncratic drug reactions:the'danger hypothesis'and innate immune system［J］.Chem Res Toxicol，1999，12(5):387-395.

［33］Zanni MP，von Greyerz S，Schnyder B，Brander KA，Frutig K，Hari Y，et al.HLA-restricted，processingand metabolism-independent pathway of drug recognition by human alpha beta T lymphocytes［J］.J Clin Invest，1998，102(8):1591-1598.

［34］Rieder MJ.Immune mediation of hypersensitivity adverse drug reactions:implications for therapy［J］.Expert Opin Drug Saf，2009，8(3):331-343.

［35］Sanderson JP，Naisbitt DJ，Park BK.Role of bioactivation in drug-induced hypersensitivity reactions［J］.AAPS J，2006，8(1):E55-E64.

［36］Uetrecht J.Immune-mediatedadverse drug reactions［J］.Chem Res Toxicol，2009，22(1):24-34.

［37］Jaeschke H，Gores GJ，Cederbaum AI，Hinson JA，Pessayre D，Lemasters JJ.Mechanisms of hepatotoxicity［J］.Toxicol Sci，2002，65(2):166-176.

［38］Williams DP.Toxicophores:investigations in drug safety［J］.Toxicology，2006，226(1):1-11.

［39］Palmeira CM，Rolo AP.Mitochondrially-mediated toxicity of bile acids［J］.Toxicology，2004，203(1-3):1-15.

［40］Somchit N，Wade LT，Ramsay L，Goldin RD，Kenna JG，Caldwell J.Hum Exp Toxicol，1997，16:401.

［41］Hargus SJ，Martin BM，George JW，Pohl LR.Covalent modification of rat liver dipeptidyl peptidaseⅣ(CD26)by the nonsteroidal anti-inflammatory drug diclofenac［J］.Chem Res Toxicol，1995，8(8):993-996.

［42］Yamazaki H，Shibata A，Suzuki M，Nakajima M，Shimada N，Guengerich FP，et al.Oxidation of troglitazone to a quinone-type metabolite catalyzed by cytochrome P-450 2C8 and P-450 3A4 in human liver microsomes［J］.Drug Metab Dispos，1999，27(11):1260-1266.

［43］Kassahun K，Pearson PG，Tang W，McIntosh I，Leung K，Elmore C，et al.Studies on the metabolism of troglitazone to reactive intermediates in vitro and in vivo.Evidence for novel biotransformation pathways involving quinone methide formation and thiazolidinedione ring scission［J］.Chem Res Toxicol，2001，14(1):62-70.

［44］Funk C，Pantze M，Jehle L，Ponelle C，Scheuermann G，Lazendic M，et al.Troglitazone-induced intrahepatic cholestasis by an interference with the hepatobiliary export of bile acids in male and female rats.Correlation with the gender difference in troglitazone sulfate formation and the inhibition of the canalicular bile salt export pump(Bsep)by troglitazone and troglitazone sulfate［J］.Toxicology，2001，167(1):83-98.

［45］Honma W，Shimada M，Sasano H，Ozawa S，Miyata M，Nagata K，et al.Phenol sulfotransferase，ST1A3，as the main enzyme catalyzing sulfation of troglitazone in human liver［J］.Drug Metab Dispos，2002，30(8):944-949.

［46］Larrey D.Epidemiology and individual susceptibility to adverse drug reactions affecting the liver［J］.Semin Liver Dis，2002，22(2):145-155.

［47］Hanley JA，Lippman-Hand A.If nothing goes wrong，is everything all right?Interpreting zero numerators［J］.JAMA，1983，249(13):1743-1745.

［48］Eypasch E，Lefering R，Kum CK，Troidl H.Probability of adverse events that have not yet occurred:a statistical reminder［J］.BMJ，1995，311(7005):619-620.

［49］Boelsterli UA，Hsiao CJ.The heterozygous Sod2(+/-)mouse:modeling the mitochondrial role in drug toxicity［J］.Drug Discov Today，2008，13(21-22):982-988.

［50］Tang W，Lu AY.Metabolic bioactivation and drug-related adverse effects:current status and future directions from a pharmaceutical research perspective［J］.Drug Metab Rev，2010，42(2):225-249.

［51］Preziosi P.Isoniazid:metabolic aspects and toxicological correlates［J］.Curr Drug Metab，2007，8(8):839-851.

［52］Nolan CM，Goldberg SV，Buskin SE.Hepatotoxicity associated with isoniazid preventive therapy:a 7-year survey from a public health tuberculosis clinic［J］.JAMA，1999，281(11):1014-1018.

［53］Amacher DE.A toxicologist's guide to biomarkers of hepatic response［J］.Hum Exp Toxicol，2002，21(5):253-262.

［54］Ozer J，Ratner M，Shaw M，Bailey W，Schomaker S.The current state of serum biomarkers of hepatotoxicity［J］.Toxicology，2008，245(3):194-205.

［55］Cullen JM，Miller RT.The role of pathology in the identification of drug-induced hepatic toxicity［J］.Expert Opin Drug Metab Toxicol，2006，2(2):241-247.

［56］Nicholson JK，Connelly J，Lindon JC，Holmes E.Metabonomics:a platform for studying drug toxicity and gene function［J］.Nat Rev Drug Discov，2002，1(2):153-161.

［57］Kaddurah-Daouk R，Kristal BS，Weinshilboum RM.Metabolomics:a global biochemical approach to drug response and disease［J］.Annu Rev Pharmacol Toxicol，2008，48:653-683.

［58］Chen C，Gonzalez FJ，Idle JR.LC-MS-based metabolomics in drug metabolism［J］.Drug Metab Rev，2007，39(2-3):581-597.

［59］Tang W.Drug metabolite profiling and elucidation of drug-induced hepatotoxicity［J］.Expert Opin Drug Metab Toxicol，2007，3(3):407-420.

［60］Sun J，Schnackenberg LK，Holland RD，Schmitt TC，Cantor GH，Dragan YP，et al.Metabonomics evaluation of urine from rats given acute and chronic doses of acetaminophen using NMR and UPLC/MS［J］.J Chromatogr B Analyt Technol Biomed Life Sci，2008，871(2):328-340.

［61］Waters NJ，Waterfield CJ，Farrant RD，Holmes E，Nicholson JK.Integrated metabonomic analysis of bromobenzene-induced hepatotoxicity:novel induction of 5-oxoprolinosis［J］.J Proteome Res，2006，5(6):1448-1459.

［62］Lu C，Wang Y，Sheng Z，Liu G，Fu Z，Zhao J，et al.NMR-based metabonomic analysis of the hepatotoxicity induced by combined exposure to PCBs and TCDD in rats［J］.Toxicol Appl Pharmacol，2010，248(3):178-184.

［63］Manyike PT，Kharasch ED，Kalhorn TF，Slattery JT.Contribution of CYP2E1 and CYP3A to acetaminophen reactive metabolite formation［J］.Clin Pharmacol Ther，2000，67(3):275-282.

［64］Chen C，Krausz KW，Idle JR，Gonzalez FJ.Identification of novel toxicity-associated metabolites by metabolomics and mass isotopomer analysis of acetaminophen metabolism in wild-type and Cyp2e1-null mice［J］.J Biol Chem，2008，283(8):4543-4559.

［65］Coen M，Lenz EM，Nicholson JK，Wilson ID，Pognan F，Lindon JC.An integrated metabonomic investigation of acetaminophen toxicity in the mouse using NMR spectroscopy［J］.Chem Res Toxicol，2003，16(3):295-303.

［66］Coen M，Ruepp SU，Lindon JC，Nicholson JK，Pognan F，Lenz EM，et al.Integrated application of transcriptomics and metabonomics yields new insight into the toxicity due to paracetamol in the mouse［J］.J Pharm Biomed Anal，2004，35(1):93-105.

［67］Ruepp SU，Tonge RP，Shaw J，Wallis N，Pognan F.Genomics and proteomics analysis of acetaminophen toxicity in mouse liver［J］.Toxicol Sci，2002，65(1):135-150.

［68］Mortishire-Smith RJ，Skiles GL，Lawrence JW，Spence S，Nicholls AW，Johnson BA，et al.Use of metabonomics to identify impaired fatty acid metabolism as the mechanism of a drug-induced toxicity［J］.Chem Res Toxicol，2004，17(2):165-173.

［69］Sha W，da Costa KA，Fischer LM，Milburn MV，Lawton KA，Berger A，et al.Metabolomic profiling can predict which humans will develop liver dysfunction when deprived of dietary choline［J］.FASEB J，2010，24(8):2962-2975.

［70］Clayton TA，Lindon JC，Cloarec O，Antti H，Charuel C，Hanton G，et al.Pharmaco-metabonomic phenotyping and personalized drug treatment［J］.Nature，2006，440(7087):1073-1077.

［71］Winnike JH，Li Z，Wright FA，Macdonald JM，O'Connell TM，Watkins PB.Use of pharmaco-metabonomics for early prediction of acetaminophen-induced hepatotoxicity in humans［J］.Clin Pharmacol Ther，2010，88(1):45-51.

药物导致肝损伤、药物处置及其代谢产物分析

Wei TANG
(上海睿智化学研究有限公司药物代谢及药代动力学研究室，上海201203)

药物导致肝损伤是药物研发失败和上市药退市的一个主要原因。药物导致肝损伤的频发跟肝生理功能密切相关，因为大部分药物分子在体内的消除依赖于药物代谢或胆汁排泄的肝清除。虽然不少发病机制已有广泛的研究，但是大部分与药物导致肝损伤相关的机制仍然未明确。从这个意义上讲，代谢组学将成为研究药物导致肝损伤强有力的手段以有助于更好地了解其机制并进行相关生物标志物的鉴定。

肝损伤;代谢;药代动力学;代谢产物分析

Wei TANG，Tel:(021)51371839，E-mail:weitang1234@gmail.com

2012-05-09)

R969

A

1000-3002(2012)04-0467-09

10.3867/j.issn.1000-3002.2012.04.001

Wei TANG，Tel:(021)51371839，E-mail:weitang1234@gmail.com

(本文编辑:乔虹)