- Email: [email protected]
Possible role of HLA in hepatotoxicity
Journal of Hepatolog.v 1994; 20:336-342 Printed in Denmark. All rights reserved Munksgaard. Copenhagen
Copyright © Journalof Hepatology 1994 Journal of Hepatology ISSN 0168-8278
Possible role of HLA in hepatotoxicity An exploratory study in 71 patients with drug-induced idiosyncratic hepatitis Alain Berson ~, Eric Fr6neaux ~, Dominique Larrey ~, Virginia Lepage 2, Corinne D o u a y 2, Christine Mallet 2, Bernard F r o m e n t y ~, Jean-Pierre B e n h a m o u I and Dominique Pessayre 1 ~UnitO de Recherche de Physiopathologie HOpatique ( 1 N S E R M U24), Hdpital Beaujon, Clichy, France and 'Laboratoire d'HistocornpatibilitO et d'lmmunogOnOtique, H6pital Saint-Louis, Paris, France
(Received 12 January 1992)
Possible associations between particular human leucocyte antigen molecules and immunoallergic hepatitis have been suggested previously (HLA-All in halothane hepatitis, HLA-DR6 and DR2 in nitrofurantoin hepatitis, HLA-B8 in clometacin hepatitis). In this study the HLA haplotype was determined in 71 patients with idiosyncratic hepatitis due to different drugs. The prevalence of HLA-A11 was twice as high in the 71 patients in the study (23%) as in controls (12%o), but p-values were not significant when corrections were made for the large number of comparisons (n=39). The prevalences of HLA-DR2, DR6, and B8 were similar in the 71 patients and in controls. When hepatitis due to particular drugs was considered, HLA-AI 1 was found to be present in six of 12 patients (50%) with hepatitis caused by tricyclic antidepressants, and three of four patients (75%) with diclofenac hepatitis, compared to 12% in controls. HLA-DR6 was present in four of five patients (80%) with chlorpromazine hepatitis, compared to 22% in controls. In conclusion, the HLA phenotype does not contribute significantly to idiosyncratic drug-induced hepatitis considered collectively. Possible associations between some HLA molecules and the hepatotoxicity of certain drugs require further confirmation. © Journal of Hepatology. Key words: Drug-induced hepatitis; Genetic predisposition; HLA phenotype; Idiosyncratic hepatitis; Immunoallergic
hepatitis; Major histocompatibility complex molecules.
Drug-induced hepatitis occurs in a few patients receiving therapeutic doses, in an apparently unpredictable fashion (1). The mechanisms of hepatotoxicity and the reasons for the susceptibility of these subjects remain largely unknown (1). While hepatitis induced by some drugs (or their metabolites) may be caused by a toxic mechanism, hepatitis induced by many other drugs appears to have an immunological basis (1-3). There is some evidence suggesting the involvement of immunological reactions in hepatitis associated with these drugs: 1) Hepatitis is frequently associated with symptoms of hypersensitivity such as fever, rash and eosinophilia; 2) The disease recurs more quickly after a drug rechallenge than after the first administration; 3) Liver lesions often include infiltration of the liver parenchyma by immunocompetent
cells; and 4) Specific antibodies directed against cell constituents are detected in some cases. The distinction between toxic and immunoallergic hepatitis is difficult with some drugs, however. It is possible that these drugs may produce toxic hepatitis in some patients, and immunoallergic hepatitis in other patients (1-3). Major histocompatibility complex molecules (i.e. class I and class II HLA molecules) are involved in the presentation of peptides from the antigen to immunocompetent cells and in the regulation of the immune response (4). Considering the polymorphism of HLA molecules in the population, a relationship could therefore exist between the particular HLA haplotype of a subject and his propensity to develop immune reactions. In support of this, several immunoallergic extrahepatic drug reactions have
Correspondence to: Alain Berson, M.D., Inserm U24, H6pital Beaujon, 92118 Clichy, France.
337
HLA HAPLOTYPE IN DRUG-INDUCED HEPATITIS
been related to the HLA phenotype. Thus, hydralazineinduced lupus erythematosus is linked to the HLA-DR4 antigen (5) and HLA-DR3 is associated with an increased incidence of renal side-effects due to either gold salts or penicillamine in patients with rheumatoid arthritis (6). Agranulocytosis induced by levamisole occurs more frequently in patients with the HLA-B27 molecule (7). Patients with toxic epidermal necrosis have a high prevalence of the HLA-B12 antigen (8). Much less information is available on the possible role of HLA molecules in drug-induced hepatitis. Although several possible associations between HLA phenotypes and drug-induced hepatitis have been reported (see Discussion), the difficulty of collecting a group of subjects which is large enough, combined with the necessity of correcting p values for many comparisons, explains why even great differences in prevalence have remained statistically insignificant (9-11). Theoretically, four relationships are possible between the HLA phenotype and the occurrence of drug-induced idiosyncratic hepatitis: 1) Certain HLA molecules might cause a predisposition to drug-induced immunoallergic hepatitis, whatever the immunogenic drug. 2) Some HLA molecules might cause a predisposition to immunoallergic hepatitis for a finite number of drug(s), but not influence hepatitis due to other potentially immunogenic drugs. 3) The HLA phenotype may cause a predisposition towards particular types of evolution (fulminant hepatitis, prolonged cholestasis). 4) The HLA phenotype may have no relationship with the occurrence and the outcome of druginduced immunoallergic hepatitis. The aim of the present study was to attempt to answer these questions by assessing the overall influence of the HLA phenotype on drug hepatotoxicity and its outcome in a relatively large group of patients. The HLA phenotype was determined in 71 Caucasian patients with idiosyncratic drug-induced hepatitis, observed in the same hepatology unit, from June 1986 to June 1991, and compared to results in a similar, local, Caucasian population.
virus (26 patients) in all patients who underwent these tests. There was no circumstantial evidence for non-A, non-B hepatitis, in particular no drug addiction, no blood transfusion, and no surgery within the last 6 months. Serological tests for hepatitis C virus were performed in patients seen after November 1989 (11 patients) and were negative in all. Ultrasonographic examination of the liver and biliary tract was normal in most of the tested patients or showed a slight enlargement consistent with drug-induced hepatitis. In contrast, there was a temporal relationship between the onset of drug administration and the onset of drug hepatitis in all patients, and a close temporal relationship between the discontinuation of treatment and the improvement or disappearance of liver dysfunction. A liver biopsy was performed in 50 patients. In all cases, the lesions were either suggestive of, or compatible with, a diagnosis of drug-induced hepatitis. In several cases, associated manifestations of hypersensitivity (rash, fever, eosinophilia) further supported a diagnosis of drug-induced hepatitis. In one case (patient 60), the diagnosis was further confirmed by the presence of serum anti-mitochondrial antibodies of type M6, an autoantibody specifically associated with iproniazid hepatitis (12).
HLA typing Peripheral blood mononuclear cells were isolated by centrifugation through a Ficoll-Hypaque gradient. HLA typing was performed using the standard NIH lymphocytotoxicity technique at 22°C. HLA-A and B typing was performed on peripheral blood mononuclear cells. HLADR and DQ typing was performed on a peripheral blood suspension enriched in B cells isolated from blood by means of antibody-coated microspheres, a technique which mostly prevents typing errors (13). The control population consisted of a similar Caucasian population, from the same geographical area, typed using local sera and core set sera from the 9th and the 10th International Histocompatibility Workshops.
Statistical analysis Methods
Subjects The causative drug and the main clinical and biological data for each patient are indicated in Table 1. The diagnosis was considered definite or probable if the following criteria were met. There was no past history of liver or biliary tract disease, no alcohol abuse and no evidence of another cause of hepatitis. In particular, serological tests were negative for recent infection due to hepatitis A virus (43 patients), hepatitis B virus (71 patients), Epstein-Barr virus (31 patients), cytomegalovirus (30 patients) and herpes simplex
Comparisons of antigen frequencies in the 71 patients and in the control population were made using the Chisquare test, with Yates' correction for low numbers, p corrected values (pcorr) were obtained by mutiplyingp values by the number of alleles tested (n=39), since multiple comparisons were made.
Results
Overall group of 71 subjects The study group included 47 females and 24 males, 18-87 years old (mean, 46 years). The main causative
A. B E R S O N et al.
338
TABLE 1 Clinical data and individual H L A phenotype in 71 patients with idiosyncratic drug-induced hepatitis Case
Age (year)/ sex
Causative drug
Max. serum bilirubin ~mol/I)
Max. ALT d
Max. AP~
Histological lesion c
1 2 3 4 5 6 7
39/F 45/M 26/F 50/F 44/F 30/F 60/F
Amineptine Amineptine Amineptine Amineptine Amineptine Amineptine Amineptine
150 240 65 8 6 132 7
25 10 7.4 35 64 5 8
2 3.5 1 2.5 1.2 1.1 1
8 9 10 11 12
35/M 30/M 40/M 24/F 19/F
Chlorpromazine Chlorpromazine Chlorpromazine Chlorpromazine Chlorpromazine
412 117 750 I 18 7
2.6 7 5 10 4.2
13 14 15 16 17
47/M 37/M 35/F 43/M 59/F
Diclofenac Diclofenac Diclofenac Diclofenac Diclofenac
5 16 7 55 38
18 19 20
55/F 37/F 34/F
Amitriptyline Amitriptyline Amitriptyline
21 22 23
76/F 77/F 76/M
24 25
H L A phenotype A
B
DPC AMH AMH AMH -g AMH -
2/11 3/32 11/19 3/I 1 11/I/3/10
13/37 7/35 5/21 15/70 22/8/37 7/27
6/7 1/2 1/7 3/4 5/4/1/4
1/2 1/1/2 2/3 3/3/1/3
6 3.8 3.4 5 1.4
DPC AMH DPC ACH -
1/1/9 9/19 2/9/11
17/21 17/35 12/17 8/7/35
6/5/7 I/6 3/6 6/-
-/2/3 I/3/1/-
4 3.8 15 11 25
1 1 2 2 2.3
AMH AMH AHH AHH -
2/11 1/3 11/28 -/3/11
7/18 7/8 16/17 -/8/35
2/7 3/7 6/7 4/5 3/5
1/2 2/3 1/2 /2/3
9 40 68
3.5 2.8 2.5
1.3 3.2 l.l
AMH DPC ACH
1/I 1 28/11 1/9
5/14 12/17 5/12
3/7 2/5 3/4
2/I/3 2/3
THA W THA" THA ~
75 28 15
45 7.6 6.5
1.1 2 1
AHH AHH
2/I/1/11
21/27 5/35 7/17
4/5 5/6 6/-
2/1/1/-
64/F 65/F
Allopurinol Allopurinol
9 10
3 2.7
1 1
-
2/9 2/9
12/41 12/-
1/2 2/7
1/1/2
26 27
27/F 38/M
Amodiaquine Amodiaquine
135 17
5 1.3
AHH
2/32 2/10
7/5/21
4/4/-
3/-/-
28 29
48/M 51/M
Amoxicillin Clavulanic acid Amoxicillin Clavulanic acid
250 82
I 3.5
ACH ACH
2/9 2/-
13/40 7/22
2/4 I/2
1/I/-
30 31
26/M 65/F
Clomipramine Clomipramine
532 353
20 41
1 2.6
AHH r CAH
2/3 2/32
5/27 44/-
5/8 I/7
3/3/-
32 33
25/M 54/M
Dapsone Dapsone
73 32
18 16
4.2 2
AHH
2/3 -/-
7/8 8/-
2/2/6
1/2/3
34 35
59/F 53/M
Dextro-propoxyphene Dextro-propoxyphene
12 380
8 8
1 2.1
AHH
28/2/3
5/17 5/35
3/5 4/5
2/3 3/-
36 37
50/M 53/F
Erythromycin Erythromycin
15 31
1.4 14
2.3 5
AMH AMH
2/11 19/28
15/35 13/-
4/8 6/7
3/1/2
38 39
60/M 74/F
Exifone Exifone
5 18
1.1 1.4
AHH r AHH
2/33 2/3
14/35 7/15
10/2/4
1/3/-
40 41
61/M 40/F
Glafenine Glafenine
5.5 41
1 1.75
AHH AHH
2/11 1/28
21/35 12/16
3/7/-
2/3 2/-
42 43
41/M 18/F
lndomethacin Indomethacin
247 4
37 12
1.2 1
AMH AHH
I/9 2/33
14/17 5/8
-/2/3
3/1/2
44 45
32/F 48/F
Ketoconazole Ketoconazole
33 28
35 20
1.5 1.2
AHH -
2/19 29/-
14/18 12/-
-/7/-
-/2/-
46 47
52/F 61/F
Pirprofen Pirprofen
6 27
1 1. I
-
3/1/9
13/8/13
2/7 3/7
1/2 2/-
48 49
39/F 32/F
Plethoryl b Plethoryl b
251 115
24 31
2.5 1.5
AHH AHH
2/9 11/29
18/35 5/35
5/6 2/5
3/I/3
50 51
41/F 31/F
Acenocoumarol Acetylsalicylic acid
32 42
13 68
1.6 2.7
AHH
2/29 2/28
7/35 21/27
6/4/6
1/3/-
52
51/M
Atrium c
12
47.5
2
AMH
1/10
5/38
4/6
1/3
51 465 24 46
62 3 2 5.5
5.6 3.2
DR
DQ
HLA HAPLOTYPE IN DRUG-INDUCED HEPATITIS
339
TABLE 1. Cont. Case
53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71
Age (year)/ sex
Causative drug
42/F 87/F 65/F 59/F 36/F 26/M 21/F 58/F 30/F 60/F 55/M 35/M 40/F 64/F 40/F 23/F 68/F 62/F 50/M
Captopril Clometacin Clotiazepam Cyamemazine Dihydralazine Ethionamide Fipexide Iproniazid Isotretinoine Levomepromazine Lisinopril Nilutamide Oxomemazine Papaverine Phenytoin Piroxicam Quinidine Ranitidine Viloxazine
Max. serum bilirubin (/.tmol/l)
Max. ALTd
518 372 58 172 323 337 47 6 10 48 1200 18 460 13 36 173 72 575 165
40 8.8 25.7 13.2 29 8 37 32 9 11 64 13 25 11 28 10.6 11.5 3 55
Max. APd 5 1.6 1.3 3.1 2 1 2.8 2 2 1.1 1.5 1 5 1 1.3 1 5.6 6 1.5
Histological lesion" AHH r AHH f AHH AHH AHH AHH r DPC CAH AHH AHH ACH AMH
HLA phenotype A
B
2/9 3/9 1/29 1/2 1/2 2/28 2/9/11 I0/1/2 10/9/2/9 29/2/3 11/29 10/29 2/3 1/9
35/40 7/16 12/41 17/18 5/16 5/21/70 5/16 42/8/12 8/41 35/16/40 7/35 5/27 12/16/17 7/44 17/35
DR
1/6 2/5 1/7 1/7 7/4/5 4/2/6 5/3/5 3/6/1/2 2/5 1/5 7/4/2/4 5/7
DQ
1/1/3 1/2 1/2 3/3/3/1/3/2/3 2/1/1/I/1/3 2/3/3/2/3
a THA: Tetrahydroaminoacridine. b Plethoryl: drug containing vitamin A, cyclovalone, tiratricol, and two antioxidants, butylated hydroxytoluene and butylated hydroxyanisole, c Atrium: drug containing febarbamate, difebarbamate and phenobarbital. ~ ALT=Serum alanine aminotransferase expressed in x upper limit of normal; AP=serum alkaline phosphatase expressed in x upper limit of normal, c AHH=acute hepatocellular hepatitis; ACH=acute cholestatic hepatitis; AMH=acute mixed hepatocellular and cholestatic hepatitis; CAH=chronic active hepatitis; DPC=drug-induced prolonged cholestasis, r Fulminant or subfulminant hepatitis, g Not determined.
drugs were tricyclic antidepressants (12/71; 17%), nonsteroid anti-inflammatory agents (11/71; 15%), antiinfectious or antiparasitic drugs (11/71; 15%), phenothiazines (8/71; 11%) and antalgic agents (6/71; 8%). The respective frequencies o f the histocompatibility antigens in patients and in the control group are given in Table 2. Frequencies o f H L A - A , B, D R and D Q molecules in patients and in controls were not statistically different except for H L A - A l l (Table 2). The frequency o f H L A - A 1 1 was higher in the patients (23%) than in controls (12%) when p was not corrected (Table 2) for multiple comparisons (Chi-square=7.35; p=0.0067). However, this difference was not significant when the p value was multiplied by the large n u m b e r (n=39) o f antigens tested (pcorr>0.05). The relative risk for H L A A-11 (defined by the calculated chance o f getting the disease for subjects carrying this specific H L A molecule in comparison with subjects who do not carry the same marker) was 2.17.
Hepatitis due to particular drugs Because the prevalence of H L A molecules in the overall group and in the control p o p u l a t i o n was not statistically different when using p values corrected for the large number o f comparisons, distribution o f H L A molecules was not c o m p a r e d statistically in patients with hepatitis due to a particular drug and in the control population. Nevertheless, certain associations may be emphasized (Table 3).
H L A - A l l was present in 6 of 12 patients (50%) with hepatitis caused by tricyclic antidepressants (four o f seven patients with amineptine hepatitis, two o f three patients with amitriptyline hepatitis, and neither o f two patients with clomipramine hepatitis) and in three o f four patients (75%) with diclofenac hepatitis, c o m p a r e d to a prevalence o f 12% in the control population (Table 3). H L A - D R 6 was present in four o f five patients (80%) with chlorpromazine hepatitis, c o m p a r e d to a prevalence o f 22% in the control p o p u l a t i o n (Table 3).
Outcome D r u g hepatitis was fulminant or subfulminant in five patients (Table I). A particular H L A phenotype was not associated with the course o f liver injury in these five patients. Drug-induced prolonged cholestasis, defined as the persistence o f jaundice for more than 6 months or the persistence o f high serum alkaline phosphatase and g a m m a - g l u t a m y l transpeptidase activity for more than 1 year despite withdrawal of the drug (1), was observed in five patients (Table 1). Three o f these five patients (60%) had the HLA-B17 antigen, c o m p a r e d to 9% in the control population.
Discussion
The H L A phenotype was determined in 71 Caucasian patients with idiosyncratic drug-induced hepatitis (Table
A. BERSON et al.
340 TABLE 2 Prevalence of HLA molecules in the 71 patients with drug-induced hepatitis and in the control population Phenotype
% in patients
% in controls
p
HLA-A AI A2 A3 A9 (A23, A24) A10 (A25, A26, A34, A66) AI 1 A19 (A29, 30, 31, 32, 33, 74) A28
(n=69) 24.6 46.4 18.8 23.2 8.7 23.2 23.2 10.1
(n=2163) 26.4 49.4 24.7 22.3 11.6 12.2 28.7 9.2
NS a NS NS NS NS <0.01 b NS NS
HLA-B B5 (B51, B52) B7 B8 BI2 (B44, B45) B13 BI4 (B64, B65) BI5 B16 (B38, B39) BI7 (B57, B58) B18 B21 (B49, B50) B22 (B54, B55, B56) B27 B35 B37 B40 (B60, B61) B41 B42 B70 (B71, B72)
(n = 70) 21.4 20 14.3 17.1 7.1 5.7 4.3 10.0 15.7 5.7 I0.0 2.9 7.1 22.9 2.9 4.3 4.3 1.4 2.9
(n= 2132) 16 21.7 18.3 23.8 5.7 7.3 8 5.9 9.3 10.7 5.8 5.6 3.2 19.9 3.2 11.7 1.8 0.4 0.8
NS NS NS NS NS NS NS NS NS NS NS NS NS NS NS NS NS NS NS
HLA-DR DR1 DR2 DR3 DR4 DR5 (DRI 1, DRI2) DR6 (DRI3, DR14) DR7 DR8 DR10
(n=69) 17.4 26.1 17.4 26. I 27.5 24.6 27.5 2.9 1.4
(n= 1926) 18.1 29.1 22.6 23.8 27.1 21.8 22.6 5.9 1.6
NS NS NS NS NS NS NS NS NS
HLA-DQ DQ1 DQ2 DQ3
(n=67) 47.8 40.3 50.7
(n= 1926) 64.5 34.4 53.5
NS NS NS
" N S = n o t significant, b p value became insignificant when corrected for the large number of comparisons performed.
TABLE 3 Prevalence of some HLA molecules in patients with hepatitis due to some particular drugs in the present study and in previous studies Drug (number of patients)
Study
HLA
Prevalence in patients
Prevalence in controls
Tricyclic antidepressants (7 amineptine, 3 amitriptyline, 2 clomipramine) Diclofenac (n=4) Halothane (n= 17) Chlorpromazine (n = 5) Nitrofurantoin (n=9)
This study This study Eade et al. (9) This study Stricker et al. (I0) Pariente et al. (11)
AI 1 AI 1 AI 1 DR6 DR6 DR2 B8
50% 75% 29% 80% 56% 56% 70%
12% 12% 11% 22% 29% 29% 16%
Clometacin (n= 10)
HLA HAPLOTYPE IN DRUG-INDUCED HEPATITIS 1). Most of these cases were probably or possibly mediated by immunoallergic mechanisms (1-3). There were no significant differences between the prevalence of class I and class II HLA molecules in the patients in the study and in the control population, when the p values obtained in the Chi-square test (Table 2) were corrected for the large number of comparisons. Thus, there was no statistical evidence that certain HLA molecules may cause a predisposition to drug-induced immunoallergic hepatitis regardless of the immunogenic drug administered. When individual agents were considered, however, some results may be emphasized (Table 3), despite the absence of statistical significance in the group. Thus, HLA-All was present in six of 12 patients (50%) with hepatitis caused by tricyclic antidepressants and in three of four patients (75%) with diclofenac hepatitis, compared to a prevalence of 12% in the control population (Table 3). Likewise, HLA-A11 has been found in 29% of patients with halothane hepatitis, compared to 11% in the control group (9) (Table 3). HLA-DR6 was present in four of five patients (80%) with chlorpromazine hepatitis, compared to a prevalence of 22% in the control population (Table 3). Similarly, the prevalence of HLA-DR6 (and HLADR2) was 56% for both molecules in patients with nitrofurantoin hepatitis, compared to 29% in the control group (10) (Table 3). The prevalence of HLA-B8 was 70% in patients with hepatitis due to clometacin, versus 16% in the control population (11) (Table 3). These observations taken together might suggest that some HLA molecules cause a predisposition to hepatotoxicity of a finite number of drugs. However, due to the small number of patients in each group (Table 3), and the large number of comparisons performed, these differences were not significant, and must be confirmed by further studies. Many drugs which produce immunoallergic hepatitis, including amineptine (14), amitriptyline (15), chlorpromazine (16) and halothane (17), are transformed into reactive electrophilic metabolites that covalently bind to nucleophilic groups of hepatic proteins. It is thought that the modifications produced by the covalent binding of foreign molecules to proteins ("modified self") may trigger an immune response directed against hepatic proteins in a few subjects (1,I 8-20). Thus patients with halothane hepatitis exhibit antibodies directed against hepatic proteins that have been trifluoroacetylated by a reactive metabolite of halothane (18). Patients with tienilic acidor dihydralazine-induced hepatitis exhibit autoantibodies that are specifically directed against the particular forms of cytochrome P-450 (P-450 2C and 1A2, respectively) which transform these drugs into chemically reactive metabolites that covalently bind to the protein moiety of these cytochromes P-450 (19,20).
341 While all subjects probably form these reactive metabolites to some extent, only a few develop drug-induced immunoallergic hepatitis. The reasons for this susceptibility are not clear. Nevertheless, two sorts of genetic factors are thought to be involved (1). 1) Genetic metabolic defects in protective mechanisms may decrease inactivation of the reactive metabolites, and enhance covalent binding to hepatic proteins, thus increasing the initial stimulus for immunization (21-24). 2) Class II HLA molecules might be involved in the presentation of metabolite-modified peptides to T helper cells, while HLA class I molecules may present metabolite-modified peptides to cytotoxic T lymphocytes. Different HLA class I molecules have been shown to bind largely nonoverlapping sets of peptides (25). A particular HLA molecule may therefore favour presentation of some metabolite-modified peptide(s), while not presenting other modified peptides. This may explain why a particular HLA molecule may seem strongly associated with hepatitis caused by a finite number of drugs (Table 3), but not be involved in hepatitis due to several other immunogenic drugs. Interestingly, both genetic metabolic factors and the HLA phenotype may be involved in the susceptibility to a single drug. Indeed, six of nine patients with amineptineinduced hepatitis had a defect in protective mechanisms, as detected by a lymphocyte cytotoxicity assay performed in the presence of reactive metabolites of amineptine (24). In addition, four of seven other patients with amineptineinduced hepatitis (Table 1) had the HLA-All specificity. Both a genetic metabolic defect in a protective mechanism (21), and the presence of HLA-A11 (9) may be implicated in hepatitis due to halothane. In a recent study on the outcome of idiopathic autoimmune hepatitis (26), the human leukocyte antigen A1, B8 haplotype was more frequent in patients in whom treatment had failed or who had undergone transplantation than in those who entered remission during corticosteroid therapy (70% vs 41%). This suggests that the HLA haplotype may influence the outcome of some immunologically mediated liver diseases. The results in the present study of an HLA-B17 phenotype in three of the five patients with prolonged drug-induced cholestasis (60%), compared to 9 0 in the control population, may be worth mentioning, although further confirmation is needed. In summary, there appeared to be no significant link between HLA phenotype and idiosyncratic drug-induced hepatitis, when patients with hepatitis due to many different drugs were analysed collectively. The possible association of HLA molecules with hepatitis mediated by a few drugs, as suggested in the present study and in previous reports (HLA-All in hepatitis induced by amineptine, amitriptyline, diclofenac or halothane; HLA-DR6 in
342
A. BERSON et al.
hepatitis induced by c h l o r p r o m a z i n e o r n i t r o f u r a n t o i n ; H L A - B 8 in c l o m e t a c i n hepatitis), requires further c o n firmation. It is suggested that future p r o s p e c t i v e studies m i g h t only d e t e r m i n e the suspected H L A m o l e c u l e in these p a r t i c u l a r forms o f d r u g - i n d u c e d hepatitis. This w o u l d m a k e it unnecessary to m u l t i p l y p - v a l u e s by the large n u m b e r o f tested alleles (a r e q u i r e m e n t which, c o m bined with the limited n u m b e r o f patients with drug-induced hepatitis, m a k e s statistical significance a l m o s t unattainable).
References
I. Pessayre D, Larrey D. Drug-induced liver injury. In: McIntyre N, Benhamou JP, Bircher J, Rizzetto M, Rodes J, eds. Oxford Textbook of Clinical Hepatology, Vol 2. Oxford: Oxford University Press, 1991; 875-902. 2. Zimmerman HJ. Hepatotoxicity. The Adverse Effects of Drugs and Other Chemicals on the Liver. New York: Appleton Century Crofts, 1978. 3. Stricker BHCh, Spoelstra P. Drug-Induced Hepatic Injury. A Comprehensive Survey of the Literature on Adverse Drug Reactions up to January 1985. Amsterdam: Elsevier, 1985. 4. Rothbard JB, Gefter ML. Interactions between immunogenic peptides and MHC proteins. Annu Rev Immunol 1991; 9: 527-65. 5. Batchelor JR, Welsh KI, Mansilla, et al. Hydralazine-induced systemic lupus erythematosus: influence of HLA-DR and sex on susceptibility. Lancet 1980; i: 1107-9. 6. Wooley PH, Griffin J, Panayi GS, et al. HLA-DR antigens and toxic reaction to sodium aurothiomalate and D-penicillamine in patients with rheumatoid arthritis. N Engl J Med 1980; 303: 300-2. 7. Veys EM, Mielants H, Verbruggen G. Levamisole-induced adverse reactions in HLA B27-positive rheumatoid arthritis. Lancet 1978; i: 148. 8. Roujeau JC, Huynh TN, Bracq C, et al. Genetic susceptibility to toxic epidermal necrosis. Arch Dermatol 1987; 123:1171-3. 9. Eade OE, Grice D, Krawitt EL, et al. HLA A and B locus antigens in patients with unexplained hepatitis following halothane anaesthesia. Tissue Antigens 1981; 17: 428-32. 10. Stricker BHCh, Block R, Claas FHJ, Parys GEV, Desmet VJ. Hepatic injury associated with the use of nitrofurans: a clinicopathological study of 52 reported cases. Hepatology 1988; 8: 599-606. 11. Pariente A, Hamoud A, Goldfain D, et al. H~patites ~ la clom~tacine (Dup6ran®). Etude r6trospective de 30 cas. Un mod61e d'h6patite immunoallergique. Gastroenterol Clin Biol 1989; 13: 769-74.
•
12. Homberg JC, Stelly N, Andreis I, Abuaf N, Saadoun F. A new antimitochondrial antibody (anti-M6) in iproniazid-induced hepatitis. Clin Exp Immunol 1982; 47: 93-102. 13. Vartdal F, Bratlie A, Gaudernack G, Funderud S, Lea T, Thorsby E. Microcytotoxic HLA typing of cells directly isolated from blood by means of antibody-coated microspheres. Transplant Proc 1987; 19: 655-7. 14. Gen~ve J, Larrey D, Lett6ron P, et al. Metabolic activation of the tricyclic antidepressant amineptine. I. Cytochrome P-450mediated in vitro and in vivo covalent binding. Biochem Pharmacol 1987; 36: 323-9. 15. Prox A, Breyer-Pfaff U. Amitriptyline metabolites in human urine. Identification of phenols, dihydrodiols, glycols, and ketones. Drug Metab Dispos 1987; 15: 890-6. 16. Kelder PP, De Mol NJ, Hart BAT, Lambert HM. Metabolic activation of chlorpromazine by stimulated human leukocytes. Induction of covalent binding of chlorpromazine to nucleic acids and proteins. Chem Biol Interact 1991; 79: 15-30. 17. Gandolfi A J, White RD, Sipes G, Pohl LR. Bioactivation and covalent binding of halothane, hl vitro studies with [~H]- and [~4C]halothane. J Pharmacol Exp Ther 1980; 214: 721-5. 18. Kenna JG, Satoh H, Christ DD, Pohl LR. Metabolic basis for a drug hypersentivity: antibodies in sera from patients with halothane hepatitis recognize liver neoantigens that contain the trifluoroacetyl group derived from halothane. J Pharmacol Exp Ther 1988; 245:1103-9. 19. Beaune P, Dansette PM, Mansuy D, et al. Human antimicrosome antibodies appearing in a drug-induced hepatitis are directed against cytochrome P-450 hydroxylating the drug. Proc Natl Acad Sci USA 1987; 84: 551-5. 20. Bourdi M, Larrey D, Nataf J, et al. Anti-liver endoplasmic reticulum autoantibodies are directed against cytochrome P-4501A2. A specific marker of dihydralazine hepatitis. J Clin Invest 1990; 85: 1967-73. 21. Farrell G, Prendergast D, Murray M. Halothane hepatitis. Detection of a constitutional susceptibility factor. N Engl J Med 1985; 313: 1310~,. 2 2 . Shear NH, Spielberg SP, Grant DM, Tang BK, Kalow W. Differences in metabolism of sulfonamides predisposing to idiosyncratic toxicity. Ann Intern Med 1986; 105: 179-84. 23. Shear NH, Spielberg SP. Anticonvulsant hypersensitivity syndrome. In vitro assessment of risk. J Clin Invest 1988; 82: 1826-32. 24. Larrey D, Berson A, Habersetzer F, et al. Genetic predisposition to drug hepatotoxicity. Role in hepatitis caused by amineptine, a tricyclic antidepressant. Hepatology 1989; 10: 168-73. 25. Carreno BM, Anderson RW, Colignan JE, Biddison WE. HLAB37 and HLA-A2.1 molecules bind largely non-overlapping sets of peptides. Proc Natl Acad Sci USA 1990; 87: 3420~,. 26. Sanchez-Urdazpal L, Czaja A J, Van Hoek B, Krom RAF, Wiesner RH. Prognostic features and role of liver transplantation in severe corticosteroid-treated autoimmune chronic active hepatitis. Hepatology 1992; 15: 215-21.
Copyright © Journalof Hepatology 1994 Journal of Hepatology ISSN 0168-8278
Possible role of HLA in hepatotoxicity An exploratory study in 71 patients with drug-induced idiosyncratic hepatitis Alain Berson ~, Eric Fr6neaux ~, Dominique Larrey ~, Virginia Lepage 2, Corinne D o u a y 2, Christine Mallet 2, Bernard F r o m e n t y ~, Jean-Pierre B e n h a m o u I and Dominique Pessayre 1 ~UnitO de Recherche de Physiopathologie HOpatique ( 1 N S E R M U24), Hdpital Beaujon, Clichy, France and 'Laboratoire d'HistocornpatibilitO et d'lmmunogOnOtique, H6pital Saint-Louis, Paris, France
(Received 12 January 1992)
Possible associations between particular human leucocyte antigen molecules and immunoallergic hepatitis have been suggested previously (HLA-All in halothane hepatitis, HLA-DR6 and DR2 in nitrofurantoin hepatitis, HLA-B8 in clometacin hepatitis). In this study the HLA haplotype was determined in 71 patients with idiosyncratic hepatitis due to different drugs. The prevalence of HLA-A11 was twice as high in the 71 patients in the study (23%) as in controls (12%o), but p-values were not significant when corrections were made for the large number of comparisons (n=39). The prevalences of HLA-DR2, DR6, and B8 were similar in the 71 patients and in controls. When hepatitis due to particular drugs was considered, HLA-AI 1 was found to be present in six of 12 patients (50%) with hepatitis caused by tricyclic antidepressants, and three of four patients (75%) with diclofenac hepatitis, compared to 12% in controls. HLA-DR6 was present in four of five patients (80%) with chlorpromazine hepatitis, compared to 22% in controls. In conclusion, the HLA phenotype does not contribute significantly to idiosyncratic drug-induced hepatitis considered collectively. Possible associations between some HLA molecules and the hepatotoxicity of certain drugs require further confirmation. © Journal of Hepatology. Key words: Drug-induced hepatitis; Genetic predisposition; HLA phenotype; Idiosyncratic hepatitis; Immunoallergic
hepatitis; Major histocompatibility complex molecules.
Drug-induced hepatitis occurs in a few patients receiving therapeutic doses, in an apparently unpredictable fashion (1). The mechanisms of hepatotoxicity and the reasons for the susceptibility of these subjects remain largely unknown (1). While hepatitis induced by some drugs (or their metabolites) may be caused by a toxic mechanism, hepatitis induced by many other drugs appears to have an immunological basis (1-3). There is some evidence suggesting the involvement of immunological reactions in hepatitis associated with these drugs: 1) Hepatitis is frequently associated with symptoms of hypersensitivity such as fever, rash and eosinophilia; 2) The disease recurs more quickly after a drug rechallenge than after the first administration; 3) Liver lesions often include infiltration of the liver parenchyma by immunocompetent
cells; and 4) Specific antibodies directed against cell constituents are detected in some cases. The distinction between toxic and immunoallergic hepatitis is difficult with some drugs, however. It is possible that these drugs may produce toxic hepatitis in some patients, and immunoallergic hepatitis in other patients (1-3). Major histocompatibility complex molecules (i.e. class I and class II HLA molecules) are involved in the presentation of peptides from the antigen to immunocompetent cells and in the regulation of the immune response (4). Considering the polymorphism of HLA molecules in the population, a relationship could therefore exist between the particular HLA haplotype of a subject and his propensity to develop immune reactions. In support of this, several immunoallergic extrahepatic drug reactions have
Correspondence to: Alain Berson, M.D., Inserm U24, H6pital Beaujon, 92118 Clichy, France.
337
HLA HAPLOTYPE IN DRUG-INDUCED HEPATITIS
been related to the HLA phenotype. Thus, hydralazineinduced lupus erythematosus is linked to the HLA-DR4 antigen (5) and HLA-DR3 is associated with an increased incidence of renal side-effects due to either gold salts or penicillamine in patients with rheumatoid arthritis (6). Agranulocytosis induced by levamisole occurs more frequently in patients with the HLA-B27 molecule (7). Patients with toxic epidermal necrosis have a high prevalence of the HLA-B12 antigen (8). Much less information is available on the possible role of HLA molecules in drug-induced hepatitis. Although several possible associations between HLA phenotypes and drug-induced hepatitis have been reported (see Discussion), the difficulty of collecting a group of subjects which is large enough, combined with the necessity of correcting p values for many comparisons, explains why even great differences in prevalence have remained statistically insignificant (9-11). Theoretically, four relationships are possible between the HLA phenotype and the occurrence of drug-induced idiosyncratic hepatitis: 1) Certain HLA molecules might cause a predisposition to drug-induced immunoallergic hepatitis, whatever the immunogenic drug. 2) Some HLA molecules might cause a predisposition to immunoallergic hepatitis for a finite number of drug(s), but not influence hepatitis due to other potentially immunogenic drugs. 3) The HLA phenotype may cause a predisposition towards particular types of evolution (fulminant hepatitis, prolonged cholestasis). 4) The HLA phenotype may have no relationship with the occurrence and the outcome of druginduced immunoallergic hepatitis. The aim of the present study was to attempt to answer these questions by assessing the overall influence of the HLA phenotype on drug hepatotoxicity and its outcome in a relatively large group of patients. The HLA phenotype was determined in 71 Caucasian patients with idiosyncratic drug-induced hepatitis, observed in the same hepatology unit, from June 1986 to June 1991, and compared to results in a similar, local, Caucasian population.
virus (26 patients) in all patients who underwent these tests. There was no circumstantial evidence for non-A, non-B hepatitis, in particular no drug addiction, no blood transfusion, and no surgery within the last 6 months. Serological tests for hepatitis C virus were performed in patients seen after November 1989 (11 patients) and were negative in all. Ultrasonographic examination of the liver and biliary tract was normal in most of the tested patients or showed a slight enlargement consistent with drug-induced hepatitis. In contrast, there was a temporal relationship between the onset of drug administration and the onset of drug hepatitis in all patients, and a close temporal relationship between the discontinuation of treatment and the improvement or disappearance of liver dysfunction. A liver biopsy was performed in 50 patients. In all cases, the lesions were either suggestive of, or compatible with, a diagnosis of drug-induced hepatitis. In several cases, associated manifestations of hypersensitivity (rash, fever, eosinophilia) further supported a diagnosis of drug-induced hepatitis. In one case (patient 60), the diagnosis was further confirmed by the presence of serum anti-mitochondrial antibodies of type M6, an autoantibody specifically associated with iproniazid hepatitis (12).
HLA typing Peripheral blood mononuclear cells were isolated by centrifugation through a Ficoll-Hypaque gradient. HLA typing was performed using the standard NIH lymphocytotoxicity technique at 22°C. HLA-A and B typing was performed on peripheral blood mononuclear cells. HLADR and DQ typing was performed on a peripheral blood suspension enriched in B cells isolated from blood by means of antibody-coated microspheres, a technique which mostly prevents typing errors (13). The control population consisted of a similar Caucasian population, from the same geographical area, typed using local sera and core set sera from the 9th and the 10th International Histocompatibility Workshops.
Statistical analysis Methods
Subjects The causative drug and the main clinical and biological data for each patient are indicated in Table 1. The diagnosis was considered definite or probable if the following criteria were met. There was no past history of liver or biliary tract disease, no alcohol abuse and no evidence of another cause of hepatitis. In particular, serological tests were negative for recent infection due to hepatitis A virus (43 patients), hepatitis B virus (71 patients), Epstein-Barr virus (31 patients), cytomegalovirus (30 patients) and herpes simplex
Comparisons of antigen frequencies in the 71 patients and in the control population were made using the Chisquare test, with Yates' correction for low numbers, p corrected values (pcorr) were obtained by mutiplyingp values by the number of alleles tested (n=39), since multiple comparisons were made.
Results
Overall group of 71 subjects The study group included 47 females and 24 males, 18-87 years old (mean, 46 years). The main causative
A. B E R S O N et al.
338
TABLE 1 Clinical data and individual H L A phenotype in 71 patients with idiosyncratic drug-induced hepatitis Case
Age (year)/ sex
Causative drug
Max. serum bilirubin ~mol/I)
Max. ALT d
Max. AP~
Histological lesion c
1 2 3 4 5 6 7
39/F 45/M 26/F 50/F 44/F 30/F 60/F
Amineptine Amineptine Amineptine Amineptine Amineptine Amineptine Amineptine
150 240 65 8 6 132 7
25 10 7.4 35 64 5 8
2 3.5 1 2.5 1.2 1.1 1
8 9 10 11 12
35/M 30/M 40/M 24/F 19/F
Chlorpromazine Chlorpromazine Chlorpromazine Chlorpromazine Chlorpromazine
412 117 750 I 18 7
2.6 7 5 10 4.2
13 14 15 16 17
47/M 37/M 35/F 43/M 59/F
Diclofenac Diclofenac Diclofenac Diclofenac Diclofenac
5 16 7 55 38
18 19 20
55/F 37/F 34/F
Amitriptyline Amitriptyline Amitriptyline
21 22 23
76/F 77/F 76/M
24 25
H L A phenotype A
B
DPC AMH AMH AMH -g AMH -
2/11 3/32 11/19 3/I 1 11/I/3/10
13/37 7/35 5/21 15/70 22/8/37 7/27
6/7 1/2 1/7 3/4 5/4/1/4
1/2 1/1/2 2/3 3/3/1/3
6 3.8 3.4 5 1.4
DPC AMH DPC ACH -
1/1/9 9/19 2/9/11
17/21 17/35 12/17 8/7/35
6/5/7 I/6 3/6 6/-
-/2/3 I/3/1/-
4 3.8 15 11 25
1 1 2 2 2.3
AMH AMH AHH AHH -
2/11 1/3 11/28 -/3/11
7/18 7/8 16/17 -/8/35
2/7 3/7 6/7 4/5 3/5
1/2 2/3 1/2 /2/3
9 40 68
3.5 2.8 2.5
1.3 3.2 l.l
AMH DPC ACH
1/I 1 28/11 1/9
5/14 12/17 5/12
3/7 2/5 3/4
2/I/3 2/3
THA W THA" THA ~
75 28 15
45 7.6 6.5
1.1 2 1
AHH AHH
2/I/1/11
21/27 5/35 7/17
4/5 5/6 6/-
2/1/1/-
64/F 65/F
Allopurinol Allopurinol
9 10
3 2.7
1 1
-
2/9 2/9
12/41 12/-
1/2 2/7
1/1/2
26 27
27/F 38/M
Amodiaquine Amodiaquine
135 17
5 1.3
AHH
2/32 2/10
7/5/21
4/4/-
3/-/-
28 29
48/M 51/M
Amoxicillin Clavulanic acid Amoxicillin Clavulanic acid
250 82
I 3.5
ACH ACH
2/9 2/-
13/40 7/22
2/4 I/2
1/I/-
30 31
26/M 65/F
Clomipramine Clomipramine
532 353
20 41
1 2.6
AHH r CAH
2/3 2/32
5/27 44/-
5/8 I/7
3/3/-
32 33
25/M 54/M
Dapsone Dapsone
73 32
18 16
4.2 2
AHH
2/3 -/-
7/8 8/-
2/2/6
1/2/3
34 35
59/F 53/M
Dextro-propoxyphene Dextro-propoxyphene
12 380
8 8
1 2.1
AHH
28/2/3
5/17 5/35
3/5 4/5
2/3 3/-
36 37
50/M 53/F
Erythromycin Erythromycin
15 31
1.4 14
2.3 5
AMH AMH
2/11 19/28
15/35 13/-
4/8 6/7
3/1/2
38 39
60/M 74/F
Exifone Exifone
5 18
1.1 1.4
AHH r AHH
2/33 2/3
14/35 7/15
10/2/4
1/3/-
40 41
61/M 40/F
Glafenine Glafenine
5.5 41
1 1.75
AHH AHH
2/11 1/28
21/35 12/16
3/7/-
2/3 2/-
42 43
41/M 18/F
lndomethacin Indomethacin
247 4
37 12
1.2 1
AMH AHH
I/9 2/33
14/17 5/8
-/2/3
3/1/2
44 45
32/F 48/F
Ketoconazole Ketoconazole
33 28
35 20
1.5 1.2
AHH -
2/19 29/-
14/18 12/-
-/7/-
-/2/-
46 47
52/F 61/F
Pirprofen Pirprofen
6 27
1 1. I
-
3/1/9
13/8/13
2/7 3/7
1/2 2/-
48 49
39/F 32/F
Plethoryl b Plethoryl b
251 115
24 31
2.5 1.5
AHH AHH
2/9 11/29
18/35 5/35
5/6 2/5
3/I/3
50 51
41/F 31/F
Acenocoumarol Acetylsalicylic acid
32 42
13 68
1.6 2.7
AHH
2/29 2/28
7/35 21/27
6/4/6
1/3/-
52
51/M
Atrium c
12
47.5
2
AMH
1/10
5/38
4/6
1/3
51 465 24 46
62 3 2 5.5
5.6 3.2
DR
DQ
HLA HAPLOTYPE IN DRUG-INDUCED HEPATITIS
339
TABLE 1. Cont. Case
53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71
Age (year)/ sex
Causative drug
42/F 87/F 65/F 59/F 36/F 26/M 21/F 58/F 30/F 60/F 55/M 35/M 40/F 64/F 40/F 23/F 68/F 62/F 50/M
Captopril Clometacin Clotiazepam Cyamemazine Dihydralazine Ethionamide Fipexide Iproniazid Isotretinoine Levomepromazine Lisinopril Nilutamide Oxomemazine Papaverine Phenytoin Piroxicam Quinidine Ranitidine Viloxazine
Max. serum bilirubin (/.tmol/l)
Max. ALTd
518 372 58 172 323 337 47 6 10 48 1200 18 460 13 36 173 72 575 165
40 8.8 25.7 13.2 29 8 37 32 9 11 64 13 25 11 28 10.6 11.5 3 55
Max. APd 5 1.6 1.3 3.1 2 1 2.8 2 2 1.1 1.5 1 5 1 1.3 1 5.6 6 1.5
Histological lesion" AHH r AHH f AHH AHH AHH AHH r DPC CAH AHH AHH ACH AMH
HLA phenotype A
B
2/9 3/9 1/29 1/2 1/2 2/28 2/9/11 I0/1/2 10/9/2/9 29/2/3 11/29 10/29 2/3 1/9
35/40 7/16 12/41 17/18 5/16 5/21/70 5/16 42/8/12 8/41 35/16/40 7/35 5/27 12/16/17 7/44 17/35
DR
1/6 2/5 1/7 1/7 7/4/5 4/2/6 5/3/5 3/6/1/2 2/5 1/5 7/4/2/4 5/7
DQ
1/1/3 1/2 1/2 3/3/3/1/3/2/3 2/1/1/I/1/3 2/3/3/2/3
a THA: Tetrahydroaminoacridine. b Plethoryl: drug containing vitamin A, cyclovalone, tiratricol, and two antioxidants, butylated hydroxytoluene and butylated hydroxyanisole, c Atrium: drug containing febarbamate, difebarbamate and phenobarbital. ~ ALT=Serum alanine aminotransferase expressed in x upper limit of normal; AP=serum alkaline phosphatase expressed in x upper limit of normal, c AHH=acute hepatocellular hepatitis; ACH=acute cholestatic hepatitis; AMH=acute mixed hepatocellular and cholestatic hepatitis; CAH=chronic active hepatitis; DPC=drug-induced prolonged cholestasis, r Fulminant or subfulminant hepatitis, g Not determined.
drugs were tricyclic antidepressants (12/71; 17%), nonsteroid anti-inflammatory agents (11/71; 15%), antiinfectious or antiparasitic drugs (11/71; 15%), phenothiazines (8/71; 11%) and antalgic agents (6/71; 8%). The respective frequencies o f the histocompatibility antigens in patients and in the control group are given in Table 2. Frequencies o f H L A - A , B, D R and D Q molecules in patients and in controls were not statistically different except for H L A - A l l (Table 2). The frequency o f H L A - A 1 1 was higher in the patients (23%) than in controls (12%) when p was not corrected (Table 2) for multiple comparisons (Chi-square=7.35; p=0.0067). However, this difference was not significant when the p value was multiplied by the large n u m b e r (n=39) o f antigens tested (pcorr>0.05). The relative risk for H L A A-11 (defined by the calculated chance o f getting the disease for subjects carrying this specific H L A molecule in comparison with subjects who do not carry the same marker) was 2.17.
Hepatitis due to particular drugs Because the prevalence of H L A molecules in the overall group and in the control p o p u l a t i o n was not statistically different when using p values corrected for the large number o f comparisons, distribution o f H L A molecules was not c o m p a r e d statistically in patients with hepatitis due to a particular drug and in the control population. Nevertheless, certain associations may be emphasized (Table 3).
H L A - A l l was present in 6 of 12 patients (50%) with hepatitis caused by tricyclic antidepressants (four o f seven patients with amineptine hepatitis, two o f three patients with amitriptyline hepatitis, and neither o f two patients with clomipramine hepatitis) and in three o f four patients (75%) with diclofenac hepatitis, c o m p a r e d to a prevalence o f 12% in the control population (Table 3). H L A - D R 6 was present in four o f five patients (80%) with chlorpromazine hepatitis, c o m p a r e d to a prevalence o f 22% in the control p o p u l a t i o n (Table 3).
Outcome D r u g hepatitis was fulminant or subfulminant in five patients (Table I). A particular H L A phenotype was not associated with the course o f liver injury in these five patients. Drug-induced prolonged cholestasis, defined as the persistence o f jaundice for more than 6 months or the persistence o f high serum alkaline phosphatase and g a m m a - g l u t a m y l transpeptidase activity for more than 1 year despite withdrawal of the drug (1), was observed in five patients (Table 1). Three o f these five patients (60%) had the HLA-B17 antigen, c o m p a r e d to 9% in the control population.
Discussion
The H L A phenotype was determined in 71 Caucasian patients with idiosyncratic drug-induced hepatitis (Table
A. BERSON et al.
340 TABLE 2 Prevalence of HLA molecules in the 71 patients with drug-induced hepatitis and in the control population Phenotype
% in patients
% in controls
p
HLA-A AI A2 A3 A9 (A23, A24) A10 (A25, A26, A34, A66) AI 1 A19 (A29, 30, 31, 32, 33, 74) A28
(n=69) 24.6 46.4 18.8 23.2 8.7 23.2 23.2 10.1
(n=2163) 26.4 49.4 24.7 22.3 11.6 12.2 28.7 9.2
NS a NS NS NS NS <0.01 b NS NS
HLA-B B5 (B51, B52) B7 B8 BI2 (B44, B45) B13 BI4 (B64, B65) BI5 B16 (B38, B39) BI7 (B57, B58) B18 B21 (B49, B50) B22 (B54, B55, B56) B27 B35 B37 B40 (B60, B61) B41 B42 B70 (B71, B72)
(n = 70) 21.4 20 14.3 17.1 7.1 5.7 4.3 10.0 15.7 5.7 I0.0 2.9 7.1 22.9 2.9 4.3 4.3 1.4 2.9
(n= 2132) 16 21.7 18.3 23.8 5.7 7.3 8 5.9 9.3 10.7 5.8 5.6 3.2 19.9 3.2 11.7 1.8 0.4 0.8
NS NS NS NS NS NS NS NS NS NS NS NS NS NS NS NS NS NS NS
HLA-DR DR1 DR2 DR3 DR4 DR5 (DRI 1, DRI2) DR6 (DRI3, DR14) DR7 DR8 DR10
(n=69) 17.4 26.1 17.4 26. I 27.5 24.6 27.5 2.9 1.4
(n= 1926) 18.1 29.1 22.6 23.8 27.1 21.8 22.6 5.9 1.6
NS NS NS NS NS NS NS NS NS
HLA-DQ DQ1 DQ2 DQ3
(n=67) 47.8 40.3 50.7
(n= 1926) 64.5 34.4 53.5
NS NS NS
" N S = n o t significant, b p value became insignificant when corrected for the large number of comparisons performed.
TABLE 3 Prevalence of some HLA molecules in patients with hepatitis due to some particular drugs in the present study and in previous studies Drug (number of patients)
Study
HLA
Prevalence in patients
Prevalence in controls
Tricyclic antidepressants (7 amineptine, 3 amitriptyline, 2 clomipramine) Diclofenac (n=4) Halothane (n= 17) Chlorpromazine (n = 5) Nitrofurantoin (n=9)
This study This study Eade et al. (9) This study Stricker et al. (I0) Pariente et al. (11)
AI 1 AI 1 AI 1 DR6 DR6 DR2 B8
50% 75% 29% 80% 56% 56% 70%
12% 12% 11% 22% 29% 29% 16%
Clometacin (n= 10)
HLA HAPLOTYPE IN DRUG-INDUCED HEPATITIS 1). Most of these cases were probably or possibly mediated by immunoallergic mechanisms (1-3). There were no significant differences between the prevalence of class I and class II HLA molecules in the patients in the study and in the control population, when the p values obtained in the Chi-square test (Table 2) were corrected for the large number of comparisons. Thus, there was no statistical evidence that certain HLA molecules may cause a predisposition to drug-induced immunoallergic hepatitis regardless of the immunogenic drug administered. When individual agents were considered, however, some results may be emphasized (Table 3), despite the absence of statistical significance in the group. Thus, HLA-All was present in six of 12 patients (50%) with hepatitis caused by tricyclic antidepressants and in three of four patients (75%) with diclofenac hepatitis, compared to a prevalence of 12% in the control population (Table 3). Likewise, HLA-A11 has been found in 29% of patients with halothane hepatitis, compared to 11% in the control group (9) (Table 3). HLA-DR6 was present in four of five patients (80%) with chlorpromazine hepatitis, compared to a prevalence of 22% in the control population (Table 3). Similarly, the prevalence of HLA-DR6 (and HLADR2) was 56% for both molecules in patients with nitrofurantoin hepatitis, compared to 29% in the control group (10) (Table 3). The prevalence of HLA-B8 was 70% in patients with hepatitis due to clometacin, versus 16% in the control population (11) (Table 3). These observations taken together might suggest that some HLA molecules cause a predisposition to hepatotoxicity of a finite number of drugs. However, due to the small number of patients in each group (Table 3), and the large number of comparisons performed, these differences were not significant, and must be confirmed by further studies. Many drugs which produce immunoallergic hepatitis, including amineptine (14), amitriptyline (15), chlorpromazine (16) and halothane (17), are transformed into reactive electrophilic metabolites that covalently bind to nucleophilic groups of hepatic proteins. It is thought that the modifications produced by the covalent binding of foreign molecules to proteins ("modified self") may trigger an immune response directed against hepatic proteins in a few subjects (1,I 8-20). Thus patients with halothane hepatitis exhibit antibodies directed against hepatic proteins that have been trifluoroacetylated by a reactive metabolite of halothane (18). Patients with tienilic acidor dihydralazine-induced hepatitis exhibit autoantibodies that are specifically directed against the particular forms of cytochrome P-450 (P-450 2C and 1A2, respectively) which transform these drugs into chemically reactive metabolites that covalently bind to the protein moiety of these cytochromes P-450 (19,20).
341 While all subjects probably form these reactive metabolites to some extent, only a few develop drug-induced immunoallergic hepatitis. The reasons for this susceptibility are not clear. Nevertheless, two sorts of genetic factors are thought to be involved (1). 1) Genetic metabolic defects in protective mechanisms may decrease inactivation of the reactive metabolites, and enhance covalent binding to hepatic proteins, thus increasing the initial stimulus for immunization (21-24). 2) Class II HLA molecules might be involved in the presentation of metabolite-modified peptides to T helper cells, while HLA class I molecules may present metabolite-modified peptides to cytotoxic T lymphocytes. Different HLA class I molecules have been shown to bind largely nonoverlapping sets of peptides (25). A particular HLA molecule may therefore favour presentation of some metabolite-modified peptide(s), while not presenting other modified peptides. This may explain why a particular HLA molecule may seem strongly associated with hepatitis caused by a finite number of drugs (Table 3), but not be involved in hepatitis due to several other immunogenic drugs. Interestingly, both genetic metabolic factors and the HLA phenotype may be involved in the susceptibility to a single drug. Indeed, six of nine patients with amineptineinduced hepatitis had a defect in protective mechanisms, as detected by a lymphocyte cytotoxicity assay performed in the presence of reactive metabolites of amineptine (24). In addition, four of seven other patients with amineptineinduced hepatitis (Table 1) had the HLA-All specificity. Both a genetic metabolic defect in a protective mechanism (21), and the presence of HLA-A11 (9) may be implicated in hepatitis due to halothane. In a recent study on the outcome of idiopathic autoimmune hepatitis (26), the human leukocyte antigen A1, B8 haplotype was more frequent in patients in whom treatment had failed or who had undergone transplantation than in those who entered remission during corticosteroid therapy (70% vs 41%). This suggests that the HLA haplotype may influence the outcome of some immunologically mediated liver diseases. The results in the present study of an HLA-B17 phenotype in three of the five patients with prolonged drug-induced cholestasis (60%), compared to 9 0 in the control population, may be worth mentioning, although further confirmation is needed. In summary, there appeared to be no significant link between HLA phenotype and idiosyncratic drug-induced hepatitis, when patients with hepatitis due to many different drugs were analysed collectively. The possible association of HLA molecules with hepatitis mediated by a few drugs, as suggested in the present study and in previous reports (HLA-All in hepatitis induced by amineptine, amitriptyline, diclofenac or halothane; HLA-DR6 in
342
A. BERSON et al.
hepatitis induced by c h l o r p r o m a z i n e o r n i t r o f u r a n t o i n ; H L A - B 8 in c l o m e t a c i n hepatitis), requires further c o n firmation. It is suggested that future p r o s p e c t i v e studies m i g h t only d e t e r m i n e the suspected H L A m o l e c u l e in these p a r t i c u l a r forms o f d r u g - i n d u c e d hepatitis. This w o u l d m a k e it unnecessary to m u l t i p l y p - v a l u e s by the large n u m b e r o f tested alleles (a r e q u i r e m e n t which, c o m bined with the limited n u m b e r o f patients with drug-induced hepatitis, m a k e s statistical significance a l m o s t unattainable).
References
I. Pessayre D, Larrey D. Drug-induced liver injury. In: McIntyre N, Benhamou JP, Bircher J, Rizzetto M, Rodes J, eds. Oxford Textbook of Clinical Hepatology, Vol 2. Oxford: Oxford University Press, 1991; 875-902. 2. Zimmerman HJ. Hepatotoxicity. The Adverse Effects of Drugs and Other Chemicals on the Liver. New York: Appleton Century Crofts, 1978. 3. Stricker BHCh, Spoelstra P. Drug-Induced Hepatic Injury. A Comprehensive Survey of the Literature on Adverse Drug Reactions up to January 1985. Amsterdam: Elsevier, 1985. 4. Rothbard JB, Gefter ML. Interactions between immunogenic peptides and MHC proteins. Annu Rev Immunol 1991; 9: 527-65. 5. Batchelor JR, Welsh KI, Mansilla, et al. Hydralazine-induced systemic lupus erythematosus: influence of HLA-DR and sex on susceptibility. Lancet 1980; i: 1107-9. 6. Wooley PH, Griffin J, Panayi GS, et al. HLA-DR antigens and toxic reaction to sodium aurothiomalate and D-penicillamine in patients with rheumatoid arthritis. N Engl J Med 1980; 303: 300-2. 7. Veys EM, Mielants H, Verbruggen G. Levamisole-induced adverse reactions in HLA B27-positive rheumatoid arthritis. Lancet 1978; i: 148. 8. Roujeau JC, Huynh TN, Bracq C, et al. Genetic susceptibility to toxic epidermal necrosis. Arch Dermatol 1987; 123:1171-3. 9. Eade OE, Grice D, Krawitt EL, et al. HLA A and B locus antigens in patients with unexplained hepatitis following halothane anaesthesia. Tissue Antigens 1981; 17: 428-32. 10. Stricker BHCh, Block R, Claas FHJ, Parys GEV, Desmet VJ. Hepatic injury associated with the use of nitrofurans: a clinicopathological study of 52 reported cases. Hepatology 1988; 8: 599-606. 11. Pariente A, Hamoud A, Goldfain D, et al. H~patites ~ la clom~tacine (Dup6ran®). Etude r6trospective de 30 cas. Un mod61e d'h6patite immunoallergique. Gastroenterol Clin Biol 1989; 13: 769-74.
•
12. Homberg JC, Stelly N, Andreis I, Abuaf N, Saadoun F. A new antimitochondrial antibody (anti-M6) in iproniazid-induced hepatitis. Clin Exp Immunol 1982; 47: 93-102. 13. Vartdal F, Bratlie A, Gaudernack G, Funderud S, Lea T, Thorsby E. Microcytotoxic HLA typing of cells directly isolated from blood by means of antibody-coated microspheres. Transplant Proc 1987; 19: 655-7. 14. Gen~ve J, Larrey D, Lett6ron P, et al. Metabolic activation of the tricyclic antidepressant amineptine. I. Cytochrome P-450mediated in vitro and in vivo covalent binding. Biochem Pharmacol 1987; 36: 323-9. 15. Prox A, Breyer-Pfaff U. Amitriptyline metabolites in human urine. Identification of phenols, dihydrodiols, glycols, and ketones. Drug Metab Dispos 1987; 15: 890-6. 16. Kelder PP, De Mol NJ, Hart BAT, Lambert HM. Metabolic activation of chlorpromazine by stimulated human leukocytes. Induction of covalent binding of chlorpromazine to nucleic acids and proteins. Chem Biol Interact 1991; 79: 15-30. 17. Gandolfi A J, White RD, Sipes G, Pohl LR. Bioactivation and covalent binding of halothane, hl vitro studies with [~H]- and [~4C]halothane. J Pharmacol Exp Ther 1980; 214: 721-5. 18. Kenna JG, Satoh H, Christ DD, Pohl LR. Metabolic basis for a drug hypersentivity: antibodies in sera from patients with halothane hepatitis recognize liver neoantigens that contain the trifluoroacetyl group derived from halothane. J Pharmacol Exp Ther 1988; 245:1103-9. 19. Beaune P, Dansette PM, Mansuy D, et al. Human antimicrosome antibodies appearing in a drug-induced hepatitis are directed against cytochrome P-450 hydroxylating the drug. Proc Natl Acad Sci USA 1987; 84: 551-5. 20. Bourdi M, Larrey D, Nataf J, et al. Anti-liver endoplasmic reticulum autoantibodies are directed against cytochrome P-4501A2. A specific marker of dihydralazine hepatitis. J Clin Invest 1990; 85: 1967-73. 21. Farrell G, Prendergast D, Murray M. Halothane hepatitis. Detection of a constitutional susceptibility factor. N Engl J Med 1985; 313: 1310~,. 2 2 . Shear NH, Spielberg SP, Grant DM, Tang BK, Kalow W. Differences in metabolism of sulfonamides predisposing to idiosyncratic toxicity. Ann Intern Med 1986; 105: 179-84. 23. Shear NH, Spielberg SP. Anticonvulsant hypersensitivity syndrome. In vitro assessment of risk. J Clin Invest 1988; 82: 1826-32. 24. Larrey D, Berson A, Habersetzer F, et al. Genetic predisposition to drug hepatotoxicity. Role in hepatitis caused by amineptine, a tricyclic antidepressant. Hepatology 1989; 10: 168-73. 25. Carreno BM, Anderson RW, Colignan JE, Biddison WE. HLAB37 and HLA-A2.1 molecules bind largely non-overlapping sets of peptides. Proc Natl Acad Sci USA 1990; 87: 3420~,. 26. Sanchez-Urdazpal L, Czaja A J, Van Hoek B, Krom RAF, Wiesner RH. Prognostic features and role of liver transplantation in severe corticosteroid-treated autoimmune chronic active hepatitis. Hepatology 1992; 15: 215-21.