Background Polymorphisms in human leucocyte antigen (HLA) class I genes have been found to be associated with cold medicine (CM)-related Stevens-Johnson syndrome (SJS) and toxic epidermal necrolysis (TEN) with severe ocular complications (SOC). Because ethnic differences in genetic predisposition to SJS/TEN among different populations have been proposed, we focused on Thai patients and investigated the association between HLA class I genotypes and CM-SJS/TEN with SOC.
Methods This multicentre case–control study was conducted between September 2014 and August 2017. Seventy-one Thai patients with SJS/TEN with SOC and 159 healthy Thai controls were enrolled. HLA typing was performed. Genetic relationships were analysed using Fisher’s exact test.
Results Of 71 patients with SJS/TEN with SOC (28 male, 43 female), 49 (69%) had a history of taking cold medications prior to SJS/TEN onset. The mean age of onset was 26.7±17.1 years (range, 2–77 years). HLA-B*44:03 (OR, 7.2, p=5.5×10-6, pc=1.1×10-4) and HLA-C*07:01 (OR, 6.1, p=7.1×10-6, pc=1.1×10-4) showed significant positive associations with Thai patients with CM-SJS/TEN with SOC. Additionally, 17 of 49 patients with CM-SJS/TEN with SOC (34.7%) significantly harboured the HLA-B*44:03 and HLA-C*07:01 haplotype compared with only 11 of 159 healthy controls (6.9%) (OR=7.1, p=5.5×10-6).
Conclusions HLA-B*44:03-HLA-C*07:01 haplotype is a potential risk factor for CM-SJS/TEN with SOC in the Thai population. This study supports that HLA-B*44:03 might be a common marker for CM-SJS/TEN with SOC in Eurasia populations, including European, Indian, Japanese and Thai.
- Stevens-Johnson syndrome
- cold medicine
- severe ocular complications
- human leucocyte antigen
Statistics from Altmetric.com
Stevens-Johnson syndrome (SJS) and toxic epidermal necrolysis (TEN) are spectrums of fatal vesiculobullous diseases characterised by an acute cutaneous eruption that involves the skin and at least two sites of mucosal membranes, particularly oral and ocular mucosae.1 These conditions are dominantly caused by drugs; however, they can be induced by some certain infections such as mycoplasma and herpes viruses, or other unidentified aetiologies.2 3 Approximately 50%–89% of patients with SJS/TEN develop ocular complications, potentially leading to devastating consequences such as corneal damage and loss of vision.4–6
Due to the rarity of SJS/TEN, individual genetic susceptibility has been suspected to have an important role for developing this devastating reaction. In 1987, Roujeau et al 7 reported that human leucocyte antigen (HLA)-B*12 was weakly associated with oxicam-related and sulfonamide-related TEN in Europeans. Their work pointed out the significance of polymorphisms in HLA to the pathogenesis of SJS/TEN. Later, in the era of sequence-based typing and deep sequencing of four-digit resolution, HLA-B*15:02 allele showed a strong association with carbamazepine-related SJS in Han Chinese, Malaysian, Thai and Indian populations.8 9 Another well-established association is HLA-B*58:01 and allopurinol-related reactions in Han Chinese, Thai, Japanese, Korean and European.10 Associations between HLA genotypes and causative drugs, and associations between HLA genotypes and ocular complications in patients with SJS/TEN, have been discovered in both Caucasian and Asian.11 12
Because of the high prevalence of cold medicine (CM) use among patients with SJS/TEN with severe ocular complications (SOC) in many countries, subgroup analyses were consequently conducted. Two HLA genotypes were identified, consisting of HLA-B*44:03, which had a strong association with CM-related SJS/TEN with SOC (CM-SJS/TEN with SOC) in trans-ethnic populations including Japanese, Indian and Brazilian, and HLA*02:06, which demonstrated a significant association only in Japanese and Korean, not in Indian and Brazilian.13 14 As the strength of HLA associations with specific drug-induced SJS/TEN has been found to be related to the prevalence of the susceptibility allele in each population,9 the predisposing HLA allele should be individually identified. Therefore, in this current study, we aim to explore HLA genotypes and the associations between HLA and CM-SJS/TEN with SOC in Thai population.
This study was conducted at two university referral centres in Thailand, including Mahidol University (MU; Ramathibodi Hospital and Siriraj Hospital) and Chulalongkorn University (CU; King Chulalongkorn Memorial Hospital), which independently enrolled patients and healthy volunteers, and Kyoto Prefectural University of Medicine in Japan, in which HLA analysis was performed. All experimental processes complied with the principles set forth in the Helsinki Declaration. Protocol explanation and written informed consent were carried out in all participant groups prior to starting the experimental procedures.
Thai patients who were diagnosed with SJS/TEN either in acute, subacute or chronic phase between September 2014 and August 2017 were recruited from two centres. The diagnostic criteria of SJS/TEN were based on a confirmed history of acute onset of high fever, and skin eruption with at least two sites of serious mucocutaneous involvement including the oral mucosa and the ocular surface.15 Thai healthy volunteers were used as controls.
Demographic data including ethnicity, culprit drugs and ocular features were collected. Causative agents were identified on the basis of documentation from dermatologists and immunologists. CM was defined as the drug that patients took for relieving cold symptoms, including non-steroidal anti-inflammatory drugs (NSAIDS), acetaminophen and other multi-ingredient cold medications. Patients were classified as having SOC if the following manifestations were detected: severe conjunctivitis, pseudomembrane, and epithelial defect on the ocular surface in the acute stage and/or ocular sequelae such as dry eye, trichiasis, symblepharon and conjunctival invasion into the cornea in the chronic stage.13
Subject DNA extraction was obtained from whole peripheral blood using the PAXgene Blood DNA Kits (Qiagen, Hilden, Germany) or from saliva using Oragene DNA (Kyodou International, Kanagawa, Japan). As originally described in our previous study, PCR assays were carried out and followed by hybridisation with sequence-specific oligonucleotide probes using commercially available bead-based typing kits (Wakunaga Pharmaceutical, Hiroshima, Japan) and Luminex 100 (Luminex, Austin, Texas, USA). Genotype determination and data analyses were performed automatically using WAKFlow typing software (Wakunaga Pharmaceutical) according to the manufacturer’s instructions.14
To identify the associations between HLA class I, SJS/TEN with SOC and CM-related SJS/TEN with SOC, candidate HLAs were initially screened from the MU group data. Then, a second analysis of only a certain selection of identified HLAs from the MU group was done in the CU group. A final analysis was performed using combined data from the two groups, MU and CU groups. Carrier and gene frequencies of individual HLA alleles of patients were compared with those of the controls based on the dominant model using the Fisher’s exact test. Carrier frequency is defined as ‘the frequency of the person with the allele at the population level’, and gene frequency is defined as ‘the frequency of alleles at the population level’. The OR with a 95% CI was calculated using JMP V.11 software (SAS Institute Japan, Tokyo, Japan). The nominal p values for each allele were subsequently corrected (corrected p value) by using the Bonferroni correction. A corrected p value of <0.05 was considered statistically significant.
Seventy-one patients with SJS/TEN with SOC (28 male, 43 female) were collected from two centres (56 patients from MU and 15 patients from CU). The age of onset ranged from 2 to 77 years, with a mean age of 26.7±17.1 years. One hundred and fifty-nine healthy controls were recruited from two centres (60 controls from MU and 99 controls from CU).
Association between HLA genotypes and SJS/TEN with SOC
We first analysed data from the MU group (56 patients with SJS/TEN with SOC and 60 healthy controls) and found that HLA-A*33:03 (carrier frequency: OR=3.6, p=0.0026, pc=0.024; gene frequency: OR=3.5, p=0.00087, pc=0.0078), HLA-B*18:01 (carrier frequency: OR=15.6, p=0.011; gene frequency: OR=14.7, p=0.012), HLA-B*40:01 (carrier frequency: OR=0.28, p=0.035; gene frequency: OR=0.30, p=0.043), HLA-B*44:03 (carrier frequency: OR=4.3, p=0.0052; gene frequency: OR=4.1, p=0.0029, pc=0.041) and HLA-C*07:01 (carrier frequency: OR=4.2, p=0.0037, pc=0.045; gene frequency: OR=4.0, p=0.0024, pc=0.029) were associated with SJS/TEN with SOC. After Bonferroni correction, the associations between HLA-A*33:03, HLA-B*44:03 and HLA-C*07:01 and SJS/TEN with SOC remained significant (online supplementary table 1). Previously identified HLAs from this initial MU group analysis were then tested in the CU group (15 patients with SJS/TEN with SOC and 99 healthy controls), revealing that only HLA-B*44:03 (carrier frequency: OR=6.8, p=0.017; gene frequency: OR=5.9, p=0.019) and HLA-C*07:01 (gene frequency: OR=4.8, p=0.017) were associated with SJS/TEN with SOC (online supplementary table 2). On final analysis of combined data between the MU and CU groups, we still found significant associations between SJS/TEN with SOC and HLA-A*33:03 (carrier frequency: OR=2.6, p=0.0028, pc=0.033; gene frequency: OR=2.4, p=8.7×10-4, pc=0.010), HLA-B*44:03 (carrier frequency: OR=6.0, p=6.3×10-6, pc=1.1×10-4; gene frequency: OR=5.7, p=3.3×10-6, pc=5.9×10-5) and HLA-C*07:01 (carrier frequency: OR=4.9, p=1.5×10-5, pc=2.3×10-4; gene frequency: OR=4.7, p=4.5×10-6, pc=6.7×10-5), while HLA-B*27:04 (carrier frequency: OR=0.0065, p=0.0066; gene frequency: OR=0.069, p=0.0076) and HLA-C*12:02 (carrier frequency: OR=0.11, p=0.0093; gene frequency: OR=0.11, p=0.011) demonstrated significantly negative associations with SJS/TEN with SOC. After multiple testing corrections within the HLA locus, only HLA-A*33:03, HLA-B*44:03 and HLA-C*07:01 revealed significant associations with SJS/TEN with SOC (table 1).
Association between HLA genotypes and CM-related SJS/TENwithSOC
Forty-nine of 71 patients with SJS/TEN with SOC (69%) had a history of taking CM prior to the onset of SJS/TEN, including 18 acetaminophen (36.7%), 10 NSAIDs (20.4%), and 4 multi-ingredient cold medications (8.2%), 1 combination of both acetaminophen and NSAIDs (2%), and 17 unspecified cold medications (34.7%). We first analysed data from the MU group (40 patients with CM-SJS/TEN with SOC and 60 healthy controls) and found that HLA-A*33:03 (carrier frequency: OR=3.3, p=0.012; gene frequency: OR=3.5, p=0.0016, pc=0.018), HLA-B*18:01 (carrier frequency: OR=18.7, p=0.0087; gene frequency: OR=17.6, p=0.0095), HLAB*40:01 (carrier frequency: OR=0.19, p=0.024; gene frequency: OR=0.21, p=0.030), HLA-B*44:03 (carrier frequency: OR=4.9, p=0.0041; gene frequency: OR=4.8, p=0.0020, pc=0.026) and HLA-C*07:01 (carrier frequency: OR=5.1, p=0.0015, pc=0.022; gene frequency: OR=4.7, p=0.00079, pc=0.012) were associated with CM-SJS/TEN with SOC. After Bonferroni correction, the associations between HLA-A*33:03, HLA-B*44:03 and HLA-C*07:01 and CM-SJS/TEN with SOC remained significant (online supplementary table 3). These previously identified HLAs were then tested in the CU group (9 patients with CM-SJS/TEN with SOC and 99 healthy controls), which revealed that only HLA-B*44:03 (carrier frequency: OR=9.4, p=0.019; gene frequency: OR=7.7, p=0.021) and HLA-C*07:01 (carrier frequency: OR=5.7, p=0.048; gene frequency: OR=6.8, p=0.011) were associated with CM-SJS/TEN with SOC (online supplementary table 4). On final analysis of combined data between the MU and CU groups (49 patients with CM-SJS/TEN with SOC and 159 healthy controls), we found significant positive associations between CM-SJS/TEN with SOC and HLA-A*33:03 (carrier frequency: OR=2.3, p=0.027; gene frequency: OR=2.4, p=0.0030), HLA-B*44:03 (carrier frequency: OR=7.2, p=5.5×10-6, pc=1.1×10-4; gene frequency: OR=6.7, p=1.5×10-6, pc=2.9×10-5) and HLA-C*07:01 (carrier frequency: OR=6.1, p=7.1×10-6, pc=1.1×10-4; gene frequency: OR=5.9, p=8.9×10-7, pc=1.4×10-5), while HLA-A*24:02 (carrier frequency: OR=0.24, p=0.017; gene frequency: OR=0.23, p=0.011), HLA-B*27:04 (carrier frequency: OR=0.094, p=0.024; gene frequency: OR=0.10, p=0.027) and HLA-B*40:01 (carrier frequency: OR=0.18, p=0.012; gene frequency: OR=0.18, p=0.0064) demonstrated significantly negative associations with CM-SJS/TEN with SOC. After multiple testing corrections within the HLA locus, only HLA-B*44:03 and HLA-C*07:01 revealed significant associations with CM-SJS/TEN with SOC (table 2).
Additionally, a haplotype constituting HLA-B*44:03 and HLA-C*07:01 was significantly found in 17 of 49 patients with CM-SJS/TEN with SOC (34.7%) compared with only 11 of 159 controls (6.9%) (p=5.5×10-6, OR=7.1), suggesting that the HLA-B*44:03-HLA-C*07:01 haplotype is a potential risk factor for CM-SJS/TEN with SOC.
The HLA gene, which locates on the short arm of chromosome 6, is known to be the most polymorphic genetic system in humans.16 As certain genes encode the major histocompatibility complex proteins on cell surfaces, they are responsible for human immune regulation.16 Since SJS/TEN has been recognised as an immune disorder elicited by drugs, HLA gene could be suspected to play a central role in the pathogenesis of the disease.9 In 1982, HLA-Bw44, which is a member of the B12 subgroup, was first found to have a significant association with ocular involvement in white Caucasians with SJS.11 A high prevalence rate of HLA-DQB1*06:01 was also reported in patients with SJS/TEN who developed vision-threatening ocular complications.17 Recently, Ueta et al 18 found a strong association between SJS/TEN with SOC and HLA-A*02:06, but not HLA class II, in Japanese patients. Because CM appears to be the most prevalent inciting agent of SJS/TEN with SOC,13 19 20 examining HLA genotypes in patients with CM-SJS/TEN with SOC was done.13 They demonstrated that HLA-A*02:06 was significantly associated with CM-SJS/TEN with SOC and this relationship was not found in other medicine-related SJS/TEN with SOC.13 In respect to the disclosure of the strong binding affinity of abacavir into the antigenic peptide-binding groove of the HLA-B*57:01 molecules that triggers the following hypersensitivity response in patients with abacavir-related SJS,21 some ingredients of CM were similarly shown to have high binding affinities to the antigenic peptide-binding groove of HLA-A*02:06 molecules using in silico docking simulations.22 This experimental finding might explain a possible mechanism of how HLA-A*02:06 relates to CM-SJS/TEN with SOC.
In our study, we investigated the association between HLA class I genotypes and SJS/TEN with SOC in Thai population. We found significant associations between HLA-B*44:03, HLA-C*07:01 and SJS/TEN with SOC in both independent and combined analyses, and these associations persisted after corrections for multiple comparisons in the combined analysis. Since the majority of our patients had a history of CM intake before the onset of SJS/TEN, we intended to analyse the association between HLA class I genotypes and CM-related SJS/TEN with SOC. Significant associations between HLA-B*44:03, HLA-C*07:01 and CM-related SJS/TEN with SOC were observed in both independent and combined analyses, and they still persisted after corrections for multiple comparisons in the combined analysis. Furthermore, because HLA-B*44:03 and HLA-C*07:01 constitute a haplotype in our subjects, we assessed the association with the haplotype and found a significant positive association between this haplotype and CM-SJS/TEN with SOC (p=5.5×10-6, OR=7.1). This observation strongly suggests that the HLA-B*44:03-HLA-C*07:01 haplotype could be a potential risk factor for Thai CM-related SJS/TEN with SOC. In trans-ethnic studies consisting of European (Brazilian) and Asian (Japanese, Indian and Korean) populations, HLA-B*44:03 showed a significant association with CM-SJS/TEN with SOC with the OR ranging from 1.66 in Japanese to 12.25 in Indian.14 These findings supported that HLA-B*44:03 might be the universal risk factor for SJS/TEN with SOC.
In contrast to previous reports from Japan and Korea,14 we did not observe positive associations between HLA-A*02:06 and CM-SJS/TEN with SOC in Thai patients. This might be due to the low incidence of HLA-A*02:06 (4.4%, 7 of 159 controls) in the Thai population. Although Thailand, Korea and Japan are geographically located in the same region, in East Asia, associations between HLA genotype and CM-SJS/TEN with SOC in Thai subjects are partly different from Japanese and Korean subjects. Genetic diversity in East Asia and various frequency of specific HLA alleles among these countries may be the main reasons for the incongruous outcomes.23 Inverse associations between HLA-A*11:01 and CM-SJS/TEN with SOC were also reported in the Japanese and Brazilian.14 24 Nonetheless, there were no significant inverse associations between HLA class I and CM-related SJS/TEN with SOC in the Thai population. Therefore, HLA-A*11:01 might not be accepted as a common marker for the resistance to CM-SJS/TEN with SOC.
The aetiology of ocular involvement in SJS/TEN has not been fully established. Specific combination of genes, gene–gene interactions and certain environmental factors might interplay for the manifestation of these serious phenomena.18 It was reported that PGE2 acts at EP3 in the epidermis and mucosal epithelium, negatively regulating mucocutaneous inflammation. CMs including acetaminophen, NSAIDs and multi-ingredient cold medications downregulate PGE2, which suppresses mucocutaneous inflammation.25 We suspect that CM and some viral or microbial infections might be important keys to develop CM-SJS/TEN with SOC. When individuals with a genetic background containing SJS/TEN with SOC susceptibility factors are infected by some viral or microbial infections, they develop abnormal immune responses. The administration of CM can then downregulate the PGE2 inflammatory suppressing mechanism and might augment abnormal immune response resulting in the induction of SJS/TEN with SOC. In contrast, individuals with no genetic susceptibility factors develop a normal immune response on microbial infection, and the administration of CM has no effect outside of its normal function.
For the limitations of this study, almost all of the subjects were collected from the ophthalmology department during the chronic phase of the disease. Symptoms and signs in the acute phase were obtained from history taking and medical records. We might have missed patients with SOC who were treated and well recovered without long-term ocular sequalae. However, they are considered as a very small number and should not have significant effect on the strong HLA associations that we have identified from our study.
The results from genome-wide association study revealed the strongest association with susceptibility to CM-SJS/TEN with SOC at HLA-A region.20 Apart from that, the IKZF1 gene was identified as a marker for susceptibility to CM-SJS/TEN with SOC.20 Other single nucleotide polymorphisms including TLR3, PTGER3, IL-4R, FasL and IL-13 were also demonstrated to have significant associations with SJS/TEN with SOC in the Japanese.25 Further studies evaluating the importance of these genetic predispositions for developing SJS/TEN with SOC in different populations will strengthen our understanding of the pathophysiology of these conditions and establish a universal testing tool for early identification of high-risk patients in need of prompt evaluation and treatment from ophthalmologists. This knowledge could help practitioners improve their treatment outcomes for patients with SJS/TEN.
In summary, we demonstrated a strong association between CM-SJS/TEN with SOC and HLA-B*44:03-HLA-C*07:01 haplotype. Our findings highlight the significance of HLA genotypes to the pathogenesis of SOC in patients with CM-SJS/TEN, which may help to develop an optimal HLA screening protocol for early detection of SOC in patients with SJS/TEN. Raising public awareness of SJS/TEN due to CM use should also be encouraged to prevent the occurrence of SJS/TEN with SOC.
The authors gratefully acknowledge Chie Sotozono, Ngamjit Kasetsuwan, Usanee Reinprayoon, Vannarat Satitpittakul, Vachira Sontichai and Rungnapa Ittiwut for their assistance in data collection and data analysis.
Presented at The Third International Stevens-Johnson Syndrome Symposium, Kyoto, Japan.
Contributors MU planned the research design. PJ, KL, PhP, PC, VP, PiP and KS collected samples. SK and MU analysed the data. PJ and MU wrote the paper.
Funding This work was supported by grants-in-aid from the Ministry of Education, Culture, Sports, Science and Technology of the Japanese government (BioBank Japan Project), and by the JSPS Core-to-Core Program, A. Advanced Research Networks.
Competing interests None declared.
Patient consent Not required.
Ethics approval The study was approved by the institutional review board of all institutes of Mahidol University and Chulalongkorn University in Thailand and Kyoto Prefectural University of Medicine in Japan.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.