Notice: Wiley Online Library will be unavailable on Saturday 30th July 2016 from 08:00-11:00 BST / 03:00-06:00 EST / 15:00-18:00 SGT for essential maintenance. Apologies for the inconvenience.
School of Clinical Medical Sciences (Hepatology), Medical School, University of Newcastle, Newcastle-upon-Tyne, United Kingdom
School of Clinical Medical Sciences (Hepatology), Medical School, University of Newcastle, Newcastle-upon-Tyne, Floor 4, William Leech Building, The Medical School, Framlington Place, Newcastle-upon-Tyne, NE2 4HH, United Kingdom
Diclofenac is a nonsteroidal anti-inflammatory drug that causes rare but serious hepatotoxicity, the mechanism of which is unclear. The purpose of the present study was to explore the potential role played by the immune processes. Antibodies to diclofenac metabolite-modified liver protein adducts were detected in the sera of seven out of seven patients with diclofenac-induced hepatotoxicity, 12 of 20 subjects on diclofenac without hepatotoxicity, and none of four healthy controls. The antibodies recognized adducts expressed in livers from rats treated with multiple doses of diclofenac, but not in those given single doses. In addition, several potential diclofenac adducts were identified in the liver of a patient with diclofenac-induced hepatic failure, but not from a normal human donor liver, by immunoblotting with an adduct-selective rabbit antiserum. To determine whether or not polymorphisms in genes encoding cytokine-related proteins influence susceptibility to hepatotoxicity, genotyping for the polymorphisms -627 in the interleukin (IL)-10 gene, -590 in the IL-4 gene, and codon 551 in the IL-4 receptor (IL-4R) were performed on DNA from 24 patients on diclofenac with hepatotoxicity, 48 subjects on diclofenac without hepatotoxicity, and healthy controls. The frequencies of the variant alleles for IL-10 and IL-4 were higher in patients (OR [odds ratio]: 2.8 for IL-10; 2.6 for IL-4; 5.3 for IL-10 + IL-4) compared with healthy controls and subjects on diclofenac without hepatotoxicity (OR: 2.8 for IL-10; 1.2 for IL-4; 5.0 for IL-10 + IL-4). In conclusion, the observed polymorphisms, resulting in low IL-10 and high IL-4 gene transcription, could favor a T helper (Th)-2 mediated antibody response to neoantigenic stimulation associated with disease susceptibility. HEPATOLOGY 2004;39:1430–1440.)
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
Diclofenac is a widely used nonsteroidal anti-inflammatory drug that can cause rare but potentially serious hepatotoxicity. Approximately 3.6 per 100,000 users of diclofenac develop severe liver injury1 with an 8% case fatality rate.2 Because of its common use, diclofenac hepatotoxicity has been one of the most common causes of hepatic adverse drug reactions, with 180 confirmed cases reported to the U.S. Food and Drug Administration during the first 3 years of marketing.2 Although the precise mechanism of diclofenac-induced hepatotoxicity is unknown, both metabolic2 and immunological3 mechanisms have been thought to contribute to liver injury. A common mechanism that may underlie either immunomediated liver injury or metabolic idiosyncrasy is the formation of drug-modified protein adducts.4 For several drugs, such as halothane and dihydralazine, antibodies directed against specific hepatocellular proteins have been identified.5 In each of these cases, covalent adducts of reactive drug metabolites and hepatocellular proteins have been implicated in triggering an immunological response. Studies undertaken with specific polyclonal antisera produced by immunization of rabbits with synthetic diclofenac–protein conjugates have shown that diclofenac–protein adducts are expressed in the livers of rats treated with the drug in vivo and also by rat and human hepatocytes incubated with diclofenac in vitro.6–9 Adducts appear to be localized on the bile canalicular plasma membrane of hepatocytes.9, 10 Diclofenac-treated hepatocytes carry antigenic determinants that are recognized by T cell and non–T cell–enriched splenocytes derived from diclofenac/keyhole limpet hemocyanin (KLH)–immunized mice, resulting in immunomediated destruction of target hepatocytes.11 Antibodies directed against 4′-hydroxydiclofenac glucuronide have been detected in the serum of a patient with diclofenac-induced immune hemolytic anemia,12 raising the possibility that humoral immune mechanisms might also be involved in diclofenac hepatotoxicity.
Genetic factors influencing the development of drug hepatotoxicity can be grouped into factors affecting the amount of the reactive metabolite—and therefore protein adduct formed—and factors affecting the immune response to the adducts. The possibility that polymorphisms in the cytochrome P450 gene CYP2C9, which encodes the enzyme responsible for the 4′-hydroxylation of diclofenac, may be associated with diclofenac hepatotoxicity has recently been investigated.13 Possession of variant CYP2C9 alleles was not a risk factor for the development of hepatotoxicity.13 The pattern and magnitude of the immune response to drug–protein adducts is likely to vary among individuals. A major determinant of interindividual variation in immune reactions is likely to be in the production of immunoregulatory cytokines, such as interleukin (IL)-10 and IL-4, both of which are encoded by polymorphic genes. IL-10 has diverse immunomodulating effects, including the inhibition of antigen-specific activation, and facilitates cytokine production through Th-0, Th-1 (IL-2 and interferon-α), and Th-2 (IL-4, IL-5) CD4+ lymphocytes.14 It may also exert anti-inflammatory and antifibrotic effects in the liver.15, 16 Several polymorphisms have been identified in the promoter region of the IL-10 gene. A C-to-A substitution at position -627 is present in 21%–23% of healthy individuals and is in linkage disequilibrium with polymorphisms at positions -854, -1117, and microsatellite loci IL10.G and IL10.R.17–19In vitro studies and studies in systemic lupus erythematosis, bronchial asthma, and alcoholic liver disease indicate that the -627*A allele is associated with lower transcriptional activity as well as decreased IL-10 secretion.18–21
IL-4 is the signature cytokine of Th-2 CD4+ cells, which are primarily responsible for humoral immunity.22 IL-4 induces B cells to differentiate and stimulates production of immunoglobulin (Ig) E as well as non–complement-fixing IgG isotypes such as IgG4.22 IL-4 is also important for the development of Th-2 cells.22 In contrast, IL-4 down-regulates cytokine production by Th-1 cells and inhibits their effector functions.22 A C-to-T exchange has been identified at position -590 in the promoter region of IL-4, and the variant allele has been associated with increased transcriptional activity and enhanced IgE secretion.23 This polymorphism is associated with bronchial asthma, atopic dermatitis, and rheumatoid arthritis.14–26 In addition, the gene encoding the IL-4 receptor (IL-4R) α subunit is polymorphic.27 A novel IL-4 receptor α allele in which a G-to-A substitution at position 1902 results in a change from glutamine to arginine at position 551 in the IL-4Rα protein has been identified, with the variant allele associated with enhanced signalling activity28 and more common in patients with atopy.29
The potential role of immune mechanisms in the pathogenesis of diclofenac-induced hepatotoxicity has been explored in this study. We have investigated whether or not diclofenac-modified hepatic proteins could be detected in a patient with diclofenac-induced hepatitis and whether or not antibodies to diclofenac-modified hepatic proteins occur in serum from patients with diclofenac hepatotoxicity. We have also assessed the role of allelic variants in the genes encoding IL-10, IL-4, and the IL-4R as potential susceptibility factors.
Sera were collected from seven patients recruited from Addenbrookes Hospital, Cambridge, UK; Freeman Hospital, Newcastle-upon-Tyne, UK; and the Karolinska Institute, Stockholm, Sweden who were suffering from diclofenac-induced hepatotoxicity. Sera were also collected from 20 patients on diclofenac without hepatotoxicity and four healthy control individuals not taking diclofenac. An explanted liver was collected from a patient with diclofenac-induced hepatic failure who underwent liver transplantation, and a liver biopsy was also taken from the donor liver.
For the genetic association studies, 24 patients (19 female, 79%), aged 24 to 70 (mean: 50.8) years who had suffered diclofenac hepatotoxicity between 1990 and 1999 were identified from the records of the Freeman Hospital and adverse drug reaction reports to the Committee on Safety of Medicines, UK. From the case notes, demography, medical history, physical findings, full blood count, liver function tests (albumin, bilirubin, alanine aminotransferase, alkaline phosphatase, and γ-glutamyl transferase), hepatitis B and C serology, auto-antibody screen, liver ultrasound scans, and liver biopsy reports were extracted. The clinical details of these patients are summarized in Table 1. Six patients presented with jaundice, three presented with hepatic failure (numbers 10, 11, and 20), and the rest had nonspecific symptoms associated with raised liver enzymes. Two patients (numbers 16 and 17) had peripheral eosinophilia in association with their hepatic adverse drug reaction. The causal relationship of liver injury to diclofenac was established using international consensus criteria as described previously.30, 31 When patients were taking more than one drug, causality assessment was performed with regard to each drug, and the best fit was determined.
Table 1. Demographic and Clinical Details of Patients With Hepatotoxicity Recruited Into the Genotyping Study
Forty-eight Caucasian subjects (35 female, 75%), aged 22 to 77 (mean: 52.3) years who were taking diclofenac for 0.3 to 20 (mean: 4) years without developing hepatotoxicity were recruited from the outpatient clinic of the Freeman Hospital. The underlying diagnoses were rheumatoid arthritis in 34 patients, psoriatic arthritis in seven, osteoarthritis in four, ankylosing spodylosis in two, and juvenile arthritis in one. Healthy Caucasian controls were recruited from the same area of England. The study was approved by the Newcastle-upon-Tyne Joint Ethics committee, and all subjects gave informed consent.
Animal Dosing Experiments.
In the single dosing studies, male Sprague-Dawley rats (250 g, n = 4–7 per group) received single intraperitoneal (IP) doses of diclofenac at 200 mg/kg (dissolved in normal saline, 0.5 mL/rat). Control rats received equivalent amounts of normal saline. Animals were killed at 6 hours and liver subcellular fractions were prepared by differential centrifugation. Some groups of animals (n = 3–6) were pretreated with inducers or inhibitors of drug metabolizing enzymes before administration of diclofenac at 100 mg/kg. Phenobarbital was given IP at a daily dose of 75 mg/kg/d in saline for 4 days; rats were then treated on day 5. Borneol was administered IP as a single dose of 750 mg/kg in tricaprylin, while galactosamine-HCl was administered as a single IP dose of 600 mg/kg in saline, following which rats received an IP dose of diclofenac after 30 minutes. In multiple dosing studies, male Sprague-Dawley rats (250 g, n = 4–7 per group) received daily doses of diclofenac at 30 mg/kg (dissolved in normal saline, 0.5 mL/rat) for 5 days. Animals were then killed 6 hours after the final dose and liver subcellular fractions were prepared.
Human liver fractions were prepared from an explanted liver from a patient with diclofenac-induced hepatic failure and a liver biopsy from a normal donor liver. Both animal and human liver fractions were prepared in the same manner; all steps took place at 4°C. Livers were minced in three volumes of ice-cold sucrose buffer (0.25 M sucrose, 15 mM Tris-HCl, 0.1 mM ethylenediaminetetraacetic acid pH 6.8) and subsequently homogenized using five strokes of a Potter homogenizer (Wheaton Science Products, Milleville, NJ) at 800 rpm. Homogenates were then filtered through two layers of muslin and nuclear (600 × gav pellet), mitochondrial (10,000 × gav pellet), microsomal (100,000 × gav pellet) and cytosolic fractions (100,000 x gav supernatant) were prepared by sequential centrifugation, with each pellet washed twice and resuspended in sucrose buffer using a Dounce homogenizer (Wheaton Science Products) followed by further centrifugation.42 The resulting fractions were then aliquoted and stored at −70°C.
Immunoblotting was carried out using rabbit polyclonal anti-diclofenac KLH antisera10 and sera from patients with diclofenac hepatotoxicity, subjects on diclofenac without hepatotoxicity, and controls. The sera (dilution 1:1000 for both diclofenac antibody and patients sera) were used to probe nuclear fractions from diclofenac-treated and control rats (50 μg/lane) as described by Wade et al.10 Anti–rabbit and anti–human IgG horseradish peroxidase secondary antibodies were used at 1:10,000 dilution followed by enhanced chemoluminescence development.
Immunoblotting was also employed to detect the presence of diclofenac-modified proteins in hepatic subcellular fractions from the patient with diclofenac-induced hepatic failure. Subcellular fractions prepared from a normal donor liver were used as controls. Methods outlined by Wade et al. were employed.10 Subcellular fractions were run at a concentration of 10 μg/lane and screened with a primary diclofenac antibody dilution of 1:1000 and a secondary antibody dilution of 1:10,000, followed by development using enhanced chemoluminescence. Desitometric scanning was performed using a CS-930 dual wavelength scanner coupled to a DR-2 data recorder (Shimadzu, Kyoto, Japan).
Preabsorption of Patients' Sera.
Patients' sera at 1/20 dilution were incubated at 4°C for 30 minutes, with continuous shaking, with control rat liver nuclear fractions. Samples were then centrifuged at 10,000g for 20 minutes, and the pellets were discarded. This procedure was repeated twice using either fresh nuclear fractions from rats that had received multiple doses of diclofenac or nuclear fractions from control rats (negative control).
Anti-diclofenac KLH antiserum was incubated for 3 hours at 1:1000 dilution with either diclofenac–rabbit serum albumin (RSA) conjugate or RSA—both of which were at 25 ng/mL—before immunoblots were probed.10 Patients' sera were incubated overnight at 4°C at 1:1000 dilution with 50 mM glycine or diclofenac and were then used to probe immunoblots.
Sections from the left lobe of the explanted liver were fixed in formol saline for 24 hours, embedded in paraffin, and then 5-μm sections were cut. Sections were dewaxed by sequential incubations in xylene (2 × 15 minutes), 100% ethanol (2 × 10 minutes), 90% ethanol (1 × 3 minutes), 75% ethanol (1 × 3 minutes), and 50% ethanol (1 × 3 minutes) and rinsed with tris-buffered saline (TBS: 0.2 M sodium chloride, 50 mM Tris-HCl pH 7.4). The slides were incubated for 30 minutes in 3% hydrogen peroxide to inhibit endogenous peroxidase activity and rinsed again with TBS. Nonspecific binding was blocked by incubation with 10% goat serum albumin (GSA) in TBS (10% GSA/TBS) for 1 hour. The sections were incubated for 2 hours at room temperature with anti-diclofenac KLH antiserum at 1:750 dilution in 10% GSA/TBS and washed with TBS (2 × 5 min). Subsequent incubations were undertaken using reagents from Vector Laboratories (Peterborough, UK). Slides were incubated for 30 minutes at room temperature with biotinylated goat anti–rabbit IgG at 1:500 dilution in 5% GSA/TBS, washed with TBS (2 × 5 min), and incubated for 30 minutes with Streptavidin–horse radish peroxidase at 1:200 dilution in 5% GSA/TBS. After two further 5-minute washes with TBS, slides were developed using 3,3′-diaminobenzidine reagent and counterstained for 10 seconds with hemotoxylin (Gill's formula). Slides were air dried and glass cover slips were mounted using DPX mounting medium.
Blood (10 mL) was collected from each subject and DNA was extracted as described by Daly et al.33 Genotyping for the -627 IL-10 polymorphism was performed as described by Grove et al.19 Genotyping for the -590 IL-4 polymorphism was performed by minor modification of the procedure described by Rosenwasser et al.23 Primers IL4P590F (5′ -GTTGTAATGCAGTCCTCCTG-3′) and IL-4P590R (5′ -ACTAGGCCTCACCTGATACG-3′) were used for polymerase chain reaction amplification. The temperature conditions consisted of 30 cycles of 94°C for 1-minute denaturation, 55°C for 1-minute annealing, and 72°C for 1-minute extension followed by 1 cycle at 70°C for 10 minutes. A 280-bp polymerase chain reaction product was digested with BsmF1 at 37°C and analyzed by electrophoresis on an agarose gel. Genotyping for the IL-4 R Q576R polymorphism was performed as described by Aithal et al.34
The significance of the differences between groups was assessed using either the χ2-test or Fisher's exact test depending on the sample size.
Detection of Diclofenac–Protein Adducts in Rat and Human Livers.
Diclofenac–protein adducts were detected immunochemically using a polyclonal antiserum raised by immunizing rabbits with a synthetic diclofenac–KLH conjugate. The production and characterization of the antiserum has been described previously.10 Rabbit polyclonal antiserum recognized a major 110 kDa protein adduct expressed in the livers from rats receiving single doses of diclofenac, but not in the control rat liver (Fig. 1A). In addition, the antiserum recognized several less-abundant protein adducts (30, 144, and 200 kDa). The various diclofenac adducts exhibited dose-dependant expression in livers of diclofenac-treated rats, but not control rats, and therefore could be distinguished from non–adduct-modified liver proteins that were recognized by the antiserum in control rat liver (Fig. 1A). Recognition of the adducts by the antiserum was inhibited in the presence of diclofenac (Fig. 1B). Previously it has been shown that adduct recognition is not inhibited in the presence of equivalent concentrations of indomethacin, tolmetin, fenoprofen, naproxen, sulindac, or ibuprofen.10
Immunoblotting analysis of subcellular fractions revealed that the diclofenac adducts were present in the nuclear fraction (Supplemental Fig. 1). Immunohistochemical studies further demonstrated that diclofenac adducts were localized to the bile canalicular domain of hepatocytes (Fig. 2A).
To investigate the role played by metabolism in adduct formation, rats were pre-treated with modulators of drug metabolizing enzymes before administration of diclofenac. A marked decrease in adduct expression was observed when rats were pretreated with the glucuronidation inhibitors borneol (63% decrease) or galactosamine (67% decrease).32 Adduct formation was also decreased (by 46%, as assessed by densitometry) following pretreatment of rats with phenobarbitone (Fig. 3) which is a potent inducer of various cytochrome P450 isoforms. These results indicate that the formation of the diclofenac–protein adducts recognized by the antiserum requires glucuronidation of diclofenac into reactive acyl glucuronide metabolites.
The antidiclofenac adduct antiserum was used to probe subcellular fractions obtained from the liver of a patient with diclofenac-induced hepatic failure and from a normal (control) donor liver. The antiserum recognized several proteins that were expressed in the liver from the patient but not in the control liver. These proteins were concentrated in the nuclear fraction (Fig. 4A). The recognition of proteins of molecular mass 130, 216, and 240 kDa was abolished in the presence of RSA–diclofenac conjugate, but not by RSA alone (Fig. 4B and C), implying that they correspond to diclofenac adducts. Recognition of these putative human liver adducts was not observed when the immunoblots were probed in the absence of primary antiserum or with pre-immune serum (data not shown).
Protein adducts derived from diclofenac were also detected by immunoblotting in livers from rats treated daily with the compound for 5 days at 30 mg/kg. However, the major adduct had a molecular mass of 97 kDa (not 110 kDa, as in rats that were given a single dose), and several less-abundant adducts were also seen, including an 85 kDa adduct (Fig. 5A). Recognition of 97- kDa and 85 kDa adducts was inhibited in the presence of diclofenac (Fig. 5B).
Detection of Human Antibodies to Diclofenac Adducts.
Sera from seven patients with diclofenac hepatotoxicity, 20 subjects who received diclofenac without developing hepatotoxicity, and four healthy controls were screened against nuclear fractions from rats treated chronically with diclofenac. Details of clinical presentation and laboratory tests of the seven patients with hepatotoxicity are shown in Table 2. Three of these patients presented with jaundice, two presented with hepatic failure (numbers 25 and 27), and two presented with raised liver enzymes. Four of these patients (numbers 21–24) also participated in the genotyping study. All seven patients with the drug-associated liver injury had serum antibodies that recognized protein antigens expressed in nuclear fractions from livers of rats treated with multiple doses of diclofenac, but not in fractions from control rat liver or from livers of rats treated with a single dose of diclofenac (Fig. 6). All of the patients' sera recognized a diclofenac-induced protein antigen of 85 kDa and all but three also recognized a 97 kDa diclofenac-induced antigen. Several sera recognized additional diclofenac-induced antigens (Fig. 6, Supplemental Fig. 2, and Table 2). Interestingly, the 85 kDa and 97 kDa antigens comigrated on sodium dodecyl sulfate polyacrylamide gels with diclofenac adducts recognized by the rabbit antiserum described previously (Fig. 5). Antibodies that recognized the 85 kDa and/or 97 kDa diclofenac-induced antigen were also present in sera from subjects treated with diclofenac without hepatotoxicity, while two of these sera also recognized a 55 kDa diclofenac-induced antigen (Table 3). The incidence of antibodies to diclofenac-induced antigens was lower in subjects treated with diclofenac without hepatotoxicity (12/20) than in the patients with diclofenac hepatotoxicity (7/7; P = .07). Antibodies that recognized diclofenac-induced liver antigens were not detected in the four healthy control sera (P = .003) (Fig. 6 and Supplemental Fig. 2).
Table 2. Proteins Identified by Sera From Seven Patients on Diclofenac With Hepatotoxicity
Table 3. Proteins Identified by Sera From 20 Subjects on Diclofenac Without Hepatotoxicity
Proteins Recognized (kd)
97, 85, 55
97, 85, 55
97, 85, 55
To test the specificity of interaction between the patients' sera and the diclofenac-induced liver protein antigens, competitive inhibition studies were undertaken in the presence of 50 mM of diclofenac. As a control for nonspecific ionic inhibition, 50 mM of glycine was also incubated with patient sera. Diclofenac inhibited antibody binding, while glycine failed to show any inhibition (Supplemental Fig. 3).
Genotyping for Cytokine Polymorphisms.
There was a higher frequency of patients with diclofenac hepatotoxicity with one or two A alleles at position -627 in the IL-10 gene (AA/AC genotype, n = 14/24) when compared with healthy controls (n = 75/227; P = .02; OR 2.84 [range: 1.20–6.69]) as well as subjects on diclofenac without hepatotoxicity (n = 16/48; P = .07; OR 2.80 [range: 1.02–7.68]) (Table 4). Similarly, there was an increased frequency of subjects with one or more T alleles at position -590 in the IL-4 gene (TT/CT genotype, n = 8/24) when compared with healthy controls (51/321; P = .04; OR 2.6 [range: 1.1–6.5]), but this was not significant compared with subjects on diclofenac without hepatotoxicity (n = 14/48; P = .78; OR 1.2 [range: 0.42–3.47]) (Table 4). Six out of 24 patients had at least one variant allele for both the IL-10 -627 polymorphism and the IL-4 -590 polymorphism when compared with 10/167 in the healthy control group (P = .007; OR 5.3 [range: 1.7–16.3]) and 3/48 in subjects on diclofenac without hepatotoxicity (P = .05; OR 5.0 [range: 1.12–22.19]). There were no significant differences in the frequency of variant genotypes between those presenting with jaundice or liver failure and those with raised liver enzymes (IL-10: 3/9 vs. 11/15, P = .09; IL-4: 2/9 vs. 6/15, P = .66; and IL-10 + IL-4: 2/9 vs. 5/15, P = .66) or between those developing hepatotoxicity after less than 3 months of diclofenac intake and those presenting after a longer period (IL-10: 9/16 vs. 5/8, P = .10; IL-4: 4/16 vs. 4/8, P = .36; and IL-10 + IL-4: 4/16 vs. 3/8, P = .65).
Table 4. Genotype Frequencies in Patients With Diclofenac Hepatotoxicity Compared With Controls
NOTE: Odds ratio for possession of IL-10-627A: 2.8 (95% CI 1.2–6.7; P = .02 [compared with healthy controls]; 2.8 (95% CI 1.02–7.68; P = .07 [compared with subjects on diclofenac]).
Odds ratio for possession of IL-4-590T: 2.6 (95% CI 1.1–6.5; P = .04 [compared with healthy controls]; 1.2 (95% CI 0.42–3.47; P = .78 [compared with subjects on diclofenac]).
Odds ratio for possession of IL-4R R variant: 1.3 (95% CI 0.54–3.18; P = .64 [compared with healthy controls]; 1.32 (95% CI 0.47–3.69; P = .61 [compared with subjects on diclofenac]).
Odds ratio for possession of IL-10-627A and IL-4-590T: 5.3 (95% CI 1.7–16.3; P = .007 [compared with healthy controls]; 5.0 (95% CI 1.12–22.19; P = .05 [compared with subjects on diclofenac]).
Healthy controls (n = 227)
Subjects on diclofenac (n = 48)
Cases (n = 24)
Controls (n = 321)
Subjects on diclofenac (n = 48)
Cases (n = 24)
IL-4R codon 551
Controls (n = 162)
Subjects on diclofenac (n = 48)
Cases (n = 24)
There was no significant difference in the number of patients with one or more Q551R alleles of the IL-4Rα gene (n = 9/24) when compared with healthy controls (51/162; P = .64) and the group on diclofenac without hepatotoxicity (15/48; P = .61) (Table 4). Four patients had genotyping and serum identification of antidiclofenac antibodies (numbers 21–24 in Tables 1 and 2), and three out of four possessed at least one variant IL-10 or IL-4 allele.
These results have provided evidence that diclofenac-modified proteins are formed in human liver in association with hepatotoxicity and that these adducts elicit a selective antibody response. The immunoblotting studies performed using an adduct-selective rabbit antiserum identified several proteins expressed in the explanted liver from a patient with diclofenac-induced hepatic failure that could comprise diclofenac adducts (as indicated by blocking studies using free drug). In addition, serum antibodies that recognized rat liver diclofenac–protein adducts were detected in all of the patients with diclofenac hepatotoxicity tested. The major adducts recognized by the patients' antibodies (97 kDa and 85 kDa adducts) were also recognized by the diclofenac adduct–selective rabbit antiserum. Although the evidence for diclofenac adduct formation in vivo in humans was derived from a single patient, it is reasonable to infer that adduct formation is a frequent event, because adduct specific antibodies were commonly detected in patients' sera. The nature of the diclofenac-modified target proteins recognized by the patients' antibodies remains to be determined. However, our in vivo experiments in rats indicate that this requires formation of acyl glucuronides. The latter has been previously shown to mediate protein adduct formation in hepatocytes in vivo.7 Hence acyl glucuronide formation may be responsible for protein adduct formation and ensuing antibody response in humans.
Circulating antibody directed against 4′-hydroxydiclofenac glucuronide has also been detected in diclofenac-induced immune haemolytic anemia.12 This raises the possibility that antibodies to diclofenac adducts could play a role in the mechanism of hepatotoxicity. The observation that similar antibodies were also present in sera from 60% of subjects who had not developed hepatotoxicity on diclofenac does, however, suggest that antibody production alone may not result in a clinically significant hepatotoxicity. One explanation for the appearance of antibodies in the absence of obvious hepatotoxicity is that a mild direct toxicity of the drug metabolite may be a prerequisite for immunomediated liver injury to develop in some of the affected individuals. It has been hypothesized that a mild toxic effect of the drug metabolite, which leads to the release of drug adducts, would increase the likelihood of an immunization reaction.35 Indeed, drugs such as halothane that produce immunoallergic hepatitis in a few patients also lead to a mild increase in serum transaminase activity in a much larger proportion of recipients.36 Similarly, up to 15% of patients taking diclofenac develop mild increases in serum transaminases37 and asymptomatic threefold increases of transaminases were seen in 5% of subjects during premarketing testing of the drug,2 although clinically significant hepatotoxicity is rare. This implies that multiple steps are involved in the development of diclofenac hepatotoxicity.
In the case–control study, we have shown an association between variant alleles of the IL-10 and IL-4 genes and diclofenac hepatotoxicity, providing further evidence for immune mechanisms in the pathogenesis of liver disease. The A allele at position -627 in the IL-10 gene promoter region and the T allele at position -590 in the IL-4 promoter region were more common in patients with hepatotoxicity than healthy controls as well as age- and gender-matched subjects on diclofenac without hepatotoxicity. Although there was no significant difference between the frequencies of variant IL-4 allele in patients with hepatotoxicity compared with those on diclofenac without hepatotoxicity, the combination of variant IL-10 and IL-4 alleles was more frequent in patients with hepatotoxicity compared with both groups of controls. The variant genotype would predict the production of lower levels of IL-10 and higher levels of IL-4 by stimulated cells. Low IL-10 could increase the antigen presentation of diclofenac-related neoantigens by monocytes and lead to the subsequent activation of T cells and immunomediated liver injury. High levels of IL-4 promote a Th-2–mediated immune response and induce B cell differentiation.22 The resulting enhanced production of antibodies against diclofenac-related neoantigen may increase the likelihood of liver injury. It is possible, therefore, that mild direct hepatotoxicity and the subsequent release of low levels of diclofenac–protein adducts occurs quite commonly, but that severe hepatotoxicity only occurs when these adducts elicit a strong immune response due to an individual's “immune” genotype.
Direct correlation between specific genotypes and antidiclofenac antibody formation could not be established in our study. The majority of patients recruited for the genetic association study were identified retrospectively after the hepatic adverse reaction had resolved and were therefore not suitable for the study of antidiclofenac antibodies. Only sera collected during an episode of diclofenac-induced hepatotoxicity were included in the immunoblotting assays, and many of these subjects did not participate in the genotyping studies. Hence only four patients (numbers 21–24 in Tables 1 and 2) had both genotyping and serum identification of antidiclofenac antibodies performed, and three of these possessed variant alleles for IL-10 and/or IL-4 genes. However, this subgroup is too small for statistical analysis. The immunoblotting methods used to identify diclofenac antibodies do not allow any estimation of the magnitude of the immune response. Hence we have not been able to compare the titres of antibody formation in subjects taking diclofenac without hepatotoxicity with those who had an adverse reaction.
It could be argued that the high frequency of variant alleles in the patient group might represent an association of cytokine polymorphism with the underlying disease (rather than hepatotoxicity) for which patients were prescribed diclofenac. The hepatotoxicity patients (Table 1) were taking diclofenac for a variety of indications. Osteoarthritis accounted for 7/24 (29%) of the cases. Immune mechanisms have not been implicated in the pathogenesis of osteoarthritis, and the disease has not been associated with any cytokine polymorphisms. Moreover, the frequency of variant alleles among patients who had osteoarthritis was similar to that in the rest of the group (IL-10: 6/7 vs. 8/17, P = .2; IL-4: 2/7 vs. 6/17, P = 1.0). In addition, the association of diclofenac hepatotoxicity with the variant IL-10 allele and the combination of the variant IL-10 and IL-4 alleles persisted when compared with a group that took diclofenac without developing hepatotoxicity. The strengths of these associations with hepatotoxicity (i.e., the odds ratios) were similar when compared with the two control groups. It should be noted that the underlying diagnoses in the group on diclofenac without hepatotoxicity were different than those in patients with diclofenac hepatotoxicity. This reflects the fact that the former group was recruited from hospital-based practice, where patients with osteoarthritis are not often referred unless it is for complications such as hepatotoxicity. In the diclofenac control group, 70% of patients were suffering from rheumatoid arthritis compared with 12% of the hepatotoxicity patients. It is possible that cytokine polymorphisms might be risk factors for rheumatoid arthritis as well as hepatotoxicity, offering an explanation for the smaller differences seen for some genotypes between cases and diclofenac controls compared with those between cases and healthy controls. In addition, the anti-inflammatory (rather than the immunoregulatory) properties of IL-10 might be important in drug-induced liver injury15 as demonstrated by the susceptibility of IL-10 “knockout” mice to acetaminophen-induced hepatotoxicity.38 Therefore, the association of cytokine polymorphisms with diclofenac hepatotoxicity demonstrated in this study may be independent of the immune mechanisms described.
The association of cytokine polymorphisms with idiosyncratic drug hepatotoxicity is a novel finding. Polymorphisms in the major histocompatibility complex molecules have previously been shown to modify susceptibility to drug-induced hepatotoxicity as they determine the efficient presentation of alkylated immunogenic peptides.38 Associations have been reported between HLA11 and diclofenac hepatotoxicity, which is consistent with an important role for the immune system in determining susceptibility to this disease.39 The current observations are based on relatively small groups and need to be confirmed in a larger study, but they point to the possibility of identifying individuals who may be at increased risk of adverse drug reactions by genotyping for immune system polymorphisms.
We are grateful to Dr. Eric Eliasson, Karolinska Institute, Stockholm, Sweden, and the Regional Centre of the Swedish Medical Products Agency in the Division of Clinical Pharmacology at Huddinge University for providing the serum samples used in this study.