HLA-A31 strongly associates with carbamazepine-induced adverse drug reactions but not with carbamazepine-induced lymphocyte proliferation in a Japanese population
This article is corrected by:
- Errata: Corrigendum Volume 40, Issue 1, 87, Article first published online: 16 January 2013
Hiroyuki Niihara, M.D., Department of Dermatology, Faculty of Medicine, Shimane University, 89-1 Enya-cho, Izumo, Shimane 693-8501, Japan. Email: email@example.com
Carbamazepine (CBZ) is the most frequent culprit drug for severe cutaneous adverse drug reactions (ADR), such as Stevens–Johnson syndrome (SJS), toxic epidermal necrolysis (TEN) and drug-induced hypersensitivity syndrome (DIHS). A strong association between human leukocyte antigen (HLA)-B*1502 and CBZ-induced SJS/TEN has been reported in Han Chinese, Thai, Malaysian and Indian populations, but not in Caucasian or Japanese populations. Recent studies showed an association between HLA-A*3101 and CBZ-induced ADR in Caucasian and Japanese populations. We conducted a case–control study to determine HLA genotyping of patients with CBZ-induced ADR in a Japanese population. Fifteen patients with CBZ-induced ADR and 33 subjects who had taken CBZ for more than 3 months without evidence of any ADR as a control were enrolled. In addition, the results of a CBZ-induced lymphocyte stimulation test were compared between the groups. A strong association was found between HLA-A31 and CBZ-induced ADR (P < 0.001), and a weak association was found between HLA-A11 and HLA-B51 with CBZ-induced ADR. No HLA-B*1502 was found in either patients or control subjects. The mean CBZ-induced lymphocyte stimulation index was significantly high in patients with CBZ-induced ADR compared with CBZ-tolerant patients (P < 0.001); however, no significant difference was seen between HLA-A31-positive subjects and HLA-A31-negative subjects in either group. These findings suggest that HLA-A31 is strongly associated with CBZ-induced ADR in the Japanese, but does not determine CBZ-induced lymphocyte proliferation.
Carbamazepine (CBZ) is an antiepileptic drug that has been widely used for treating not only seizures, but also neuropathic pain. CBZ is the most frequent culprit drug for severe cutaneous adverse drug reactions (ADR), including Stevens–Johnson syndrome (SJS), toxic epidermal necrolysis (TEN) and drug-induced hypersensitivity syndrome (DIHS). Since a strong association between human leukocyte antigen (HLA)-B*1502 and CBZ-induced SJS/TEN was reported in Han Chinese residing in Taiwan,1 intensive studies have focused on the association between HLA class I allele and ADR.2–16 The results are summarized in Table 1. An association between HLA-B*1502 and CBZ-induced SJS/TEN has been confirmed in Han Chinese in Hong Kong and in Chinese, Thai, Malaysian and Indian populations,5,6,8,9,11–14 but an association has not been seen in Caucasians.3,17,18 In Japanese studies, no CBZ-induced SJS/TEN patients carried HLA-B*1502.7,10 In addition, no HLA-B*1502 carriers were detected in drug-unspecified SJS/TEN patients in a Japanese population.19,20 Instead, HLA-B*1511 was found to be associated with CBZ-induced SJS/TEN patients in the Japanese.10 Interestingly, HLA-B*1502 was found to be specific to CBZ-induced SJS/TEN, and no association was seen in patients with CBZ-induced hypersensitivity syndrome (HSS) or maculopapular eruption (MPE) in Han Chinese residing in Taiwan.2 In addition, no association with HLA-B*1502 was confirmed in Caucasian patients with HSS.16 An association between HLA-A*3101 and CBZ-induced ADR was recently reported in both Japanese and Europeans by genome-wide approaches.15,16 These observations indicate a diversity in HLA class I association with CBZ-induced ADR.
Table 1. Reported HLA associated with CBZ-induced cutaneous ADR
|B*1502||Han Chinese (Taiwan)||SJS/TEN||44/44|| 1|
|Han Chinese (Taiwan)||SJS/TEN||59/60|| 2|
|Asian in Europe||SJS/TEN|| 4/4|| 3|
|Caucasians||SJS/TEN|| 0/8|| 3|
|Caucasians||DIHS|| 0/56|| 4|
|Han Chinese (Hong Kong)||SJS/TEN|| 4/4|| 5|
|Thai||SJS|| 6/6|| 6|
|Thai||MPE|| 0/5|| 6|
|Japanese||ADR|| 0/22|| 7|
|Indians||SJS|| 6/8|| 8|
|Han Chinese (Central China)||SJS/TEN|| 8/8||11|
|Han Chinese (Central China)||MPE|| 3/28||11|
|Han Chinese (Southern China)||SJS/TEN|| 9/9||13|
|Han Chinese (Southern China)||MPE||10/39||13|
|Han Chinese (Taiwan)||HSS|| 0/13|| 2|
|A*3101||Han Chinese (Taiwan)||SJS/TEN|| 1/60|| 2|
|Han Chinese (Taiwan)||HSS|| 2/13|| 2|
|Han Chinese (Taiwan)||MPE|| 6/18|| 2|
|Caucasian (Northern European)||SJS|| 5/12||16|
|Caucasian (Northern European)||HSS||10/27||16|
|Caucasian (Northern European)||MPE||23/106||16|
We conducted a case–control study to determine HLA types associated with CBZ-induced ADR in a Japanese population. In addition, CBZ-induced lymphocyte proliferation was evaluated to determine whether HLA types are associated with lymphocyte activation, because T-cell-mediated allergic reaction is likely to be involved in the pathogenesis of ADR.19
All patients were recruited from Shimane University Hospital between April 2005 and February 2011. These included 15 patients with CBZ-induced ADR and 33 patients who had been receiving CBZ for more than 3 months without drug eruption. The backgrounds of the patients with ADR are shown in Table 2. CBZ-induced ADR was determined by medical history indicating that symptoms occurred within 3 months after starting CBZ administration and that the symptoms resolved upon withdrawal of this drug. The diagnosis was confirmed by a positive stimulation index (>180%) in the CBZ-induced lymphocyte stimulation test (n = 12), positive patch testing (n = 4) and/or challenge test (n = 1). Diagnoses of SJS/TEN and DIHS were made according to the diagnostic criteria established by Roujeau and Shiohara.21,22 All patients were interviewed by investigators regarding the histories of their biological parents and grandparents. Those with both biological parents and both sets of grandparents born in Japan were classified as native Japanese. This study was approved by the ethics committee of Shimane University Faculty of Medicine (approval no. 221).
Table 2. Clinical characteristics of CBZ-induced ADR patients
| 1||42/F||TEN||Trigeminal neuralgia||32||398||52||8185||ND||ND|
| 2||68/F||SJS||Peripheral neuropathy||13||132||15||950||+||ND|
| 3||74/M||SJS||Peripheral neuropathy||32||137||9||952||−||+|
| 7||67/F||DIHS||Neuropathic pain||20||1336||19||0||+||ND|
| 8||49/M||DIHS||Peripheral neuropathy||47||642||78||6800||ND||ND|
| 9||51/M||DIHS||Peripheral neuropathy||75||246||24||2200||ND||ND|
DNA was extracted from peripheral blood. Low-resolution HLA typing was performed using the reverse sequence-specific oligonucleotide with polymerase chain reaction (PCR-rSSO) method, which is also called the Luminex method.23,24 The area of exons 2 and 3 of the HLA-A, -B and -DR genes was amplified using HLA-A, -B and -DR locus-specific primers. The amplicon sizes of HLA-A and HLA-B loci were 400–500 bp (exon 2) and 300–400 bp (exon 3), and that of HLA-DRB1 locus was 250–300 bp (exon 2). Amplified PCR products were treated with alkali and turned to single chains. After being neutralized, PCR products were hybridized with a carboxylated fluorescent microbead-coated voluntary sequenced oligonucleotide primer. After a centrifugal wash of the microbead mixtures, reaction outcomes from biotin-labeled PCR amplicons were measured by the Luminex 100 flow cytometer (GMI, Inc., Ramsey, Mn, USA), which is equipped with two types of lasers. The bead populations were detected and identified using the 635-nm laser. The phycoerythrin fluorescence of the streptavidin–phycoerythrin–biotin-labeled amplicons (Genosearch ver. 2) that had hybridized to the oligobeads was quantitated using the 532-nm laser. The median fluorescence intensity (MFI) of phycoerythrin was used to quantify the amount of DNA bound to the oligobeads. The measured data were read using dedicated software. The fluorescence intensity of negative controls was subtracted as background from each of the MFI values to determine the true intensity. The preset cut-off value for each fluorescing oligobead set was used to discriminate between positive and negative controls. The sequence-specific oligonucleotide probe (SSOP) hybridization results were matched with the Pattern File database using Genosearch HLA typing software to determine the HLA alleles.
High-resolution HLA-B genotyping was determined using the PCR sequence-based typing (SBT) method.25 Exons 1–5 of the HLA-B gene were amplified using a HLA-B locus-specific primer by PCR using DNA extract kits (Qiagen, Tokyo, Japan). By using amplified products, the areas of exons 2–4 were each sequenced using a sequence kit (Abbott Japan, Tokyo, Japan). The sequenced products of exons 2–4 were directly read using a DNA analyzer (ABI 3730 DNA analyzer; Life Technologies, Carlsbad, CA, USA) and the type of gene was determined.
Drug-induced lymphocyte stimulation test (DLST)
Peripheral blood mononuclear cells (PBMC) were isolated from whole blood and in vitro proliferation assays were performed as previously described by Pichler and Tilch.26 The CBZ used for DLST assays were unmodified compounds dissolved in culture medium and they were sonicated to solve in the medium. Cultures were performed in triplicate at 37°C and 5% CO2 for 3 days. As positive and negative controls, cells in triplicate were also incubated in the presence of 5 μg/mL phytohemagglutinin and in the absence of these agents, respectively. Twenty-four hours before harvesting, 1 μCi 3H-thymidine (Amersham, Arlington Heights, IL, USA) was added. After harvesting, radioactivity was measured in a liquid scintillation counter (Pharmacia LKB Nuclear, Gaithersburg, MD, USA) and the results were expressed as the stimulation index (SI) (%); SI was calculated as follows: SI (%) = counts per minute (c.p.m.) with drug/c.p.m. without drug × 100. An SI (%) of more than 180 was regarded as positive based on previous studies preformed in Japan.27–31 The DLST was also performed in control patients who were taking CBZ for more than 3 months without any clinical symptoms. DLST of the patients with CBZ-induced ADR was tested several times after consultation and the maximal SI was presented. The maximal SI was observed at day 56 ± 61 (mean ± SD) from onset.
Statistical analysis of the differences in each allele frequency among patients with ADR and control subjects was performed by Fisher’s exact test. The strength of association was estimated by calculating the odds ratio (OR). The OR was determined using Haldane’s modification, which adds 0.5 to all cells to accommodate possible zero counts. Differences in SI of DLST, the mean measurement day and total systemic steroid between subject groups were compared by the Mann–Whitney U-test. SAS ver. 9.2 software was used for statistical analysis. All reported P-values were two-sided. Values of P < 0.05 were considered to be statistically significant.
Table 3 shows HLA DNA typing of CBZ-induced ADR patients. One of the three SJS/TEN patients had A31, eight of the nine DIHS patients had A31, and one of the three MPE/erythema exsudativum multiforme had A31 when low-resolution HLA DNA typing was performed. Comparing the frequency of each type between the CBZ-induced ADR patients and the CBZ-tolerant patients, we found that the OR of A11, A31 and B51 were individually significantly high in the CBZ-induced ADR patients, as shown in Table 4. In particular, the OR of A31 was the highest (P = 0.001). Although the P-value of A11 was more than 0.05, we considered that the OR of A11 was significantly high because the 95% confidence interval (CI) of A11 did not range across 1.000. On the contrary, the OR of A2 was significantly low in the CBZ-induced ADR patients (P = 0.04). Table 5 shows the high-resolution HLA-B typing and their frequencies in the CBZ-induced ADR patients and CBZ-tolerant patients together with those reported for a general Japanese population.32 The HLA-B*5101 genotype appeared significantly higher in the CBZ-induced ADR patients (P = 0.031). The OR of HLA-B*5101 was 4.900 and the 95% CI was 1.219–19.689. None of the 15 CBZ-induced ADR patients, including three SJS/TEN patients and 33 CBZ-tolerant patients, possessed the HLA-B*1502 genotype.
Table 3. HLA DNA typing of CBZ-induced ADR patients
Table 4. Statistical analysis in HLA typing of CBZ-induced ADR patients and CBZ-tolerant patients
Table 5. Statistical analysis of HLA-B DNA typing of CBZ-induced ADR patients and CBZ-tolerant patients
We also investigated the CBZ-induced proliferation of PBMC in each patient. The mean SI of CBZ-induced ADR patients (382.1 ± 295.1%, n = 15) was significantly high compared with that of CBZ-tolerant patients (125.3 ± 29.5%, n = 32, P < 0.001). Table 6 shows a comparison of DLST values, mean measurement day from onset and mean systemic steroid dose from onset to DLST measurement day between subjects with and without the HLA-A31 allele in CBZ-induced ADR patients. DLST values were also compared between subjects with and without the HLA-A31 allele in CBZ-tolerant patients. The mean SI was not significantly different between subjects with and without the HLA-A31 allele in both CBZ-induced ADR patients and CBZ-tolerant patients. The mean measurement day and the mean systemic steroid dose were not significantly different between subjects with and without the HLA-A31 allele in CBZ-induced ADR patients. No significant difference was seen in DLST values between subjects with or without the HLA-A11 allele or between subjects with or without A51 in CBZ-induced ADR patients and CBZ-tolerant patients (data not shown).
Table 6. Comparison of DLST between subjects with or without HLA-A31 allele in CBZ-induced ADR patients and CBZ-tolerant patients
|CBZ-induced ADR patients||(+)||10||302.8 ± 140.5||0.147||72 ± 70||0.111||3128 ± 2440||0.212|
|(−)||5||540.6 ± 461.7||25 ± 16||1827 ± 3577|
|CBZ-tolerant patients||(+)||4||103.3 ± 27.8||0.167|| || || || |
|(−)||28||128.5 ± 28.8|| || || || |
On the basis of previous reports of HLA associated with CBZ-induced ADR in a Japanese population,7,15 we confirmed the association between HLA-A*3101 and CBZ-induced ADR, especially CBZ-induced DIHS. HLA-B*1502 was not found in either CBZ-induced ADR patients or CBZ-tolerant patients, compatible with previous results obtained from Japanese populations.7,10,15,19,20 Altogether, HLA-B*1502 is strongly associated with SJS/TEN in Asians, but not in Japanese or Caucasians. On the other hand, HLA-A*3101 is well associated with SJS/TEN and DIHS in Japanese and Caucasians and, though to a somewhat lesser extent, associated with HSS/MPE in Asians.
The reason for the diversity of HLA association in CBZ-induced ADR among races is unclear. Similar diversity in HLA associated with rheumatoid arthritis (RA) has been observed. Since Isomäki et al.33 first reported an association between RA and HL-A27 by mixed lymphocyte culture in 1974, associations between RA and HLA-DR4 have been reported in various races34,35 but not in Spanish, Abrahamidae and Indian, in which an association with HLA-DR1 and DR10 was shown.36,37 Compared with the amino acid sequence of HLA-DRB1 in HLA-DR4, -DR1 and -DR10, common amino acid sequences were found in positions 67–74 in the HLA-DRB1 molecule.38 The common amino acid sequence is situated in the third super-variable area of the DRβ chain and constitutes a part of the α-helix,38 which plays an integral role in antigen presentation. Thus, this common structure has been considered to be involved in the development of RA. Accordingly, we compared the amino acid sequences of HLA-A*3101, HLA-B*1502, and HLA-A*240201 which is one of the major Japanese HLA alleles. Table 7 shows the amino acid sequences of positions 61–80 in the α1-helix structure of HLA-B*1502, HLA-A*3101 and HLA-A*240201, whose areas have a huge variety of amino acid sequences, although other areas have relatively conserved amino acid sequences. Although six amino acid compositions are common between HLA-B*1502 and HLA-A*3101 (nos. 61, 64, 68, 72, 75 and 78), each amino acid sequence of HLA-A*240201 is also the same. These same amino acid compositions are commonly preserved in other HLA types and would not affect structural difference among the types. In addition, we found no common amino acid compositions of amino acids with polar characters (nos. 71, 80) or non-polar characters (nos. 62, 63, 65, 66, 67, 69, 70, 73, 74, 76, 77 and 79), which can affect 3-D conformation, between HLA-B*1502 and HLA-A*3101. Furthermore, no single amino acid was commonly present in alleles of both HLA-A*3101 and HLA-B*1502, except the amino acids present at the 61, 64, 68, 72, 75 and 78 positions. Therefore, HLA-B*1502 and HLA-A*3101 have no structural commonality for the common antigen presentation. We found no 3-D commonality between the two HLA from amino acid sequence, and significant difference in DLST values between subjects with and without the HLA-A31 allele in CBZ-induced ADR patients and CBZ-tolerant patients, which can mean severe ADR are independent of HLA structure and antigen presentation.
Table 7. Amino acid sequence of α1-helix structure in human leukocyte antigen
Another possibility for an association between the two HLA types and ADR is a linkage disequilibrium phenomenon in the HLA locus. Near the HLA gene, several inflammatory cytokine genes are mapped, such as g-interferon and tumor necrosis factor-β.39 The genes located in these areas are highly polymorphic, some involving single-nucleotide polymorphism. If a disease-sensitivity gene exists in close association with a HLA gene, the disease seems to be caused by the HLA type. An association between HLA-B51 and Behcet’s disease is such an example.40 Thus, a pathogenic gene for CBZ-induced ADR might be strongly connected to HLA-B*1502 in Han Chinese and Asians, but the gene might be connected to HLA-A*3101 in Europeans and Japanese. However, a recent detailed genome-wide association study concerning CBZ-induced ADR indicated that the CBZ-induced ADR gene is located at the HLA locus area; thus, it is not likely that another gene with polymorphisms causes CBZ-induced ADR.
A second possible reason is that HLA-B*1502 is associated with SJS/TEN, but not with HSS/DIHS or MPS, whereas HLA-A*3101 is associated with HSS/DIHS, but not with SJS/TEN. HLA-B*1502 was found to be specific to CBZ-induced SJS/TEN, and no association was seen in patients with CBZ-induced HSS or MPE in Han Chinese residing in Taiwan.2 In addition, no association with HLA-B*1502 was confirmed in Caucasian patients with HSS.16 Recently, an association between HLA-A*3101 and CBZ-induced ADR was reported in both the Japanese and Europeans by genome-wide approaches.15,16 In the present study, we found an association between HLA-A31 and DIHS, but only one of three patients with SJS/TEN had HLA-A31, supporting this hypothesis.
Human leukocyte antigen is well documented to be associated with some chronic inflammatory diseases and autoimmune diseases; for instance, HLA-B27 is strongly associated with Reiter syndrome and ankylosing spondylitis. Thereby, the HLA molecule plays some role in the pathogenesis by modulating the immune system. In the present study, we also tested the association between HLA-A31 and SI of DLST. We failed to demonstrate the HLA-A31-associated enhancement of lymphocyte proliferation (Table 6), although we were able to confirm strong lymphocyte activation with CBZ in the patient group.
We found that HLA-A11 and HLA-A51 are weakly associated with CBZ-induced ADR patients. Because the number of cases was small in the present study, we cannot confirm the presence of an association. We need to evaluate more cases to ascertain an association between other HLA alleles and CBZ-induced ADR patients.
In the present study, we confirmed a strong association between HLA-A31 and CBZ-induced ADR in a Japanese population. However, HLA-A31 does not determine CBZ-induced lymphocyte proliferation.
This work was supported in part by Health and Labor Sciences Research Grants (Research on Intractable Diseases) from the Japanese Ministry of Health, Labor and Welfare.