Spectrum of cutaneous adverse reactions to aromatic antiepileptic drugs and human leukocyte antigen genotypes in Thai patients and meta-analysis

Aromatic antiepileptic drugs (AEDs)-induced cutaneous adverse drug reactions (cADRs) add up to the limited use of the AEDs in the treatment and prevention of seizures. Human leukocyte antigen-B (HLA-B) alleles have been linked to AEDs-induced cADRs. We investigated the association between cADRs (including Stevens–Johnson syndrome; SJS/toxic epidermal necrolysis; TEN, drug reaction with eosinophilia and systemic symptoms; DRESS, and Maculopapular eruption; MPE) caused by AEDs (phenytoin, carbamazepine, lamotrigine, phenobarbital and oxcarbazepine) and HLA-B alleles in Thai population. Through the case-control study, 166 patients with AEDs-induced cADRs, 426 AEDs-tolerant patients (AEDs-tolerant controls), and 470 healthy subjects (Thai population) were collected. The HLA genotypes were detected using the polymerase chain reaction-sequence specific oligonucleotide probe (PCR-SSOP) method. We also performed a meta-analysis with these data and other populations. The carrier rate of HLA-B*15:02 was significantly different between AEDs-induced cADRs group and AEDs-tolerant group (Odds ratio; OR 4.28, 95% Confidence interval; CI 2.64–6.95, p < 0.001), AEDs-induced cADRs group and Thai population (OR 2.15, 95%CI 1.41–3.29, p < 0.001). In meta-analysis showed the strong association HLA-B*15:02 with AEDs-induced cADRs (OR 4.77, 95%CI 1.79–12.73, p < 0.001). Furthermore, HLA-B*15:02 was associated with SJS/TEN induced by AEDs (OR 10.28, 95%CI 6.50–16.28, p < 0.001) Phenytoin (OR 4.12, 95%CI 1.77–9.59, p = 0.001) and carbamazepine (OR 137.69, 95%CI 50.97–371.98, p < 0.001). This study demonstrated that genetic association for AEDs-induced cADRs was phenotype-specific. A strong association between HLA-B*15:02 and AEDs-induced SJS/TEN was demonstrated with an OR of 10.79 (95%CI 5.50–21.16, p < 0.001) when compared with AEDs-tolerant group. On the other hand, the carrier rates of HLA-B*08:01, HLA-B*13:01, and HLA-B*56:02 were significantly higher in the DRESS group compared with the AEDs-tolerant group (p = 0.029, 0.007, and 0.017, respectively). The HLA-B*15:02 allele may represent a risk factor for AEDs-induced cADRs.

MPE, the mild delayed type cADRs, presents as a widespread, symmetrically distributed rash composed of pink-to-red macules and papules that may coalesce to form plaques. Redness of skin without blister, pruritus and low-grade fever may occur in MPE, but does not involve the internal organs [4]. SJS and TEN are severe life-threatening delayed type cADRs characterized by febrile illness in acute phases followed by the predominant involvement of skin and mucous membrane necrosis and systemic symptoms. The degree of skin detachment expressed in terms of the % body surface area affected classifies the SJS and TEN into 3 types: <10% body surface area is affected in SJS, 10-30% body surface area is affected in SJS-TEN overlap, and >30% body surface area is affected in TEN [5]. DRESS has a wide range of clinical characteristics, such as cutaneous eruption, fever, lymphadenopathy, hepatitis, and hematologic abnormalities with eosinophilia and atypical lymphocytes [6]. Severe cADRs has a crucial burden on healthcare system with the mortality rates reaching up to 40% [7]. Alleles of the human leukocyte antigen (HLA) genes have been associated with cADRs, and the screening of HLA alleles prior to the administration of AEDs can prevent the occurrence of cADRs.
The HLA-B genotype is a crucial immunological risk factor for cADRs. In 2004, Chung et al. reported a strong association between the HLA-B*15:02 allele and CBZ-induced SJS/TEN in Han Chinese [8]. In 2013, a systematic review and meta-analysis in Thai and Malaysian populations validated the association between the HLA-B*15:02 allele and CBZ-induced SJS/TEN [9]. In Thai population, the HLA-B*15:02 allele has a frequency of 8.16% which is higher when compared to other ethnicities: 4.87% in Asians, 2% in African-American, and rarely observed in East Asian populations (Japanese and Korean), Caucasian, and Hispanic/North Americans [10,11]. The aim of this study was to extend the investigation of HLA-B susceptibility to AEDs-induced cADRs in Thai patients.

MATERIALS AND METHODS Subjects and characteristics
We recruited cases between 2011 and 2017 with AEDs-induced cADRs. The enrolled cases of 166 patients with AEDs-induced cADRs included 71 with MPE, 49 with DRESS, and 46 with SJS/TEN. Patients tolerated to various AEDs (AEDs-tolerant controls; n = 426) for at least 6 months without evidence of cutaneous adverse effects were also recruited in this study. Healthy subjects (Thai population; n = 470) who did not take AEDs were included for analysis. All individuals enrolled were unrelated Thai population. This study was approved by the Ramathibodi Hospital Ethical Review Board for Human Research. Written informed consent was obtained from all the participants.

Diagnosis of AEDs-induced cADRs
All patients were diagnosed and confirmed based on the photographs, pathological slides, clinical morphology, and medical records by a dermatologist and allergist. MPE was diagnosed as a macular and papular rash without skin detachment or organ involvement. The RegiSCAR criteria were used to diagnose and classify SCARs [12]. DRESS was characterized by acute skin rash, fever above 38°C, facial edema, enlarged lymph nodes, internal organ involvement (e.g., liver, kidney and lung), and hematological abnormalities, including lymphocytosis or lymphocytopenia, eosinophilia, and thrombocytopenia [12]. SJS was defined as skin detachment of <10% of the body surface area. TEN was defined as skin detachment >30%. The SJS-TEN overlap was defined as skin detachment of 10% to 30% [3]. In some cases, in vitro assay (such as ELISpot) was performed.
DNA extraction, HLA-A, HLA-B and CYP2C9*3 genotyping Blood samples were collected into EDTA tubes. Genomic DNA was extracted by the MagNA Pure automated compact machine (Roche Diagnostics, USA). The purity and concentration of genomic DNA were detected using the Nanodrop ND-1000 ultraviolet spectrophotometer. DNA was stored at −20°C before analysis.
HLA-A and HLA-B genotyping was performed using the polymerase chain reaction-sequence specific oligonucleotide probe (PCR-SSOP) method combined with Luminex™ Multiplex technology. Data analysis for the HLA-A and HLA-B assays was performed with software package HLA fusion 2.0.

Statistical analysis
Statistical analyses were performed using SPSS version 16.0. Means and standard deviations (SD) were calculated for all continuous data. Comparison of continuous variables among clinical characteristics between cases and controls were analyzed using the t-test. Chi-square test or Fisher's exact test was used for the comparison of HLA-A, HLA-B and CYP2C9 allele frequency in each group. Odds ratio (OR) and 95% confidence interval (CI) values were obtained. Bonferroni correction was applied to adjust for multiple comparisons (P < 0.0025) to indicate statistically significant association. For the meta-analysis, we used "metan" command of STATA program for pooling effect-size (StataCorp LLC. College Station, TX).

Cases and controls study
Pharmacogenetic study was performed in 166 patients who had AEDs-induced cADRs, 426 AEDs-tolerant controls and 470 Thai population. Table 1

Association of HLA-B alleles with AEDs-induced cADRs
The association between HLA-B alleles and AEDs-induced cADRs was evaluated by comparing the AEDs-induced cADRs group (n = 166) with the AEDs-tolerant controls group (n = 426) and the Thai population (n = 470). The HLA-B genotypes of cADRs cases induced by AEDs are shown in Table 2 Table 4. The carrier rate of the HLA-B*15:02 allele was higher in the MPE group than in the controls.   Table 6. The carrier rates of HLA-B*08:01, HLA-B*13:01, and HLA-B*56:02 were higher in the DRESS group than in the tolerant group. The carrier rates of these alleles between the two groups reached statistical significance (p = 0.029, 0.007, and 0.017, respectively). The carrier rate of HLA-B*08:01 allele was not found to be significantly higher in the DRESS cases as compared to the healthy subjects (p = 0.052). The carrier rates for HLA-B*13:01 and HLA-B*56:02 alleles were statistically higher in the DRESS group compared to the healthy population (p = 0.010 and 0.002, respectively). However, the significant associations these alleles disappeared after correction for Bonferroni correction (P < 0.0025).

Association between HLA-A and CBZ-induced DRESS and MPE
To clarify the relationship between HLA-A*31:01 and CBZ-induced DRESS and MPE, the HLA-A was genotyped in this study. Then, we compared the frequency of HLA-A alleles between the 5 CBZinduced DRESS or 16 CBZ-induced MPE patients and the 470 normal Thai subjects. The HLA-A*31:01 allele, which was previously reported to be associated with CBZ-induced DRESS and MPE, was not found in our study. Conversely, 3 of 5 (60.00%) CBZ-induced DRESS and 6 of 16 (37.5%) CBZ-induced MPE patients were HLA-A*33:03 carriers. However, there was no significant difference of  Bold values indicate statistical significance. AEDs Aromatic epileptic drugs, HLA Human leukocyte antigen, SJS/TEN Steven Johnson syndrome and toxic epidermal necrolysis, OR odds ratio, 95%CI 95% confidence interval. *p-value less than 0.05. Association between CYP2C9*3 and PHT To date there have been relatively few studies confirming the CYP2C9*3 risk association and PHT-induced SCARs. In this study we genotyped the CYP2C9*3 in 103 PHT-induced cADRs cases. The carrier rate of the CYP2C9*3 in this study was higher in the cADRs (12/103; 11.65%), which were 22.73% of SJS/TEN (5/22), 12.50% of DRESS (5/40) and 4.89% of MPE (2/41) group than in the tolerant control group (6/105; 5.71%). There was no difference in the frequency of subjects with the CYP2C9*3 between the cADRs and tolerant control group (OR 1.79; 95%CI 0.6116-5.2339, p = 0.2881). Nevertheless, there was also a trend toward an association between PHT-induced SJS/TEN and CYP2C9*3, compared to the tolerant control group (OR 3.88; 95%CI 0.9907-15.2146, p = 0.0516). Moreover, no significant differences in CYP2C9*3 frequency were found between the PHT-induced DRESS (OR 1.88; 95%CI 0.5024-7.0773, p = 0.3472) and PHT-induced MPE (OR 0.53; 95%CI 0.0617-4.5929, p = 0.5663), when compared with tolerant control group.

DISCUSSION
AEDs are effective for controlling seizures. Several studies have reported about the CBZ-induced SJS/TEN in Asian ancestry, including Han Chinese, Indian, Thai, Vietnamese and Malaysian [8,[13][14][15][16][17]. The HLA-B*15:02 allele was reported to have been associated with PHT-induced SJS/TEN, LTG-induced SJS/TEN, OXCinduced SJS/TEN. Because of the sample size limitation, association of the HLA-B*15:02 allele with these AEDs-induced cADRs is unclear and needs to be further investigated. The US-FDA has recommended the testing for HLA-B*15:02 in patients of Asian and HLA-A*31:01 in patients of European, Korean and Japanese ancestry prior to CBZ treatment due to the risk of developing SCARs [18]. Physicians avoid using all types of AEDs when the patient is HLA-B*15:02 or HLA-A*31:01 positive. Thus, we aim to investigate the association between HLA-A, HLA-B alleles and CYP2C9*3 genotype and AEDs-induced cADRs. We also performed a meta-analysis with these data and other populations.
In this study, we recruited cases of AEDs-induced cADRs (n = 166) and performed HLA-B allele association to find a genetic marker to screen the patients before initiating the AEDs treatment. The carrier rate of the HLA-B*15:02 and HLA-B*51:01 alleles was significantly higher in the AEDs-induced cADRs cases than the tolerant control group. HLA-B*15:02 carriers were found in 44.73% (17/38) and 40.00% (6/15) cases of CBZ and LTG-treated patients. It bears noting that the HLA-B*15:02 allele frequency was higher (66.67%) in OXC-induced cADRs, although the patient samples were too small to draw any definitive conclusions. However, there is evidence in the literature of the involvement of the HLA-B*15:02 allele in the OXC-induced SJS/TEN in populations in the Taiwan and Thailand [20]. Moreover, our data also support the Clinical Pharmacogenetics Implementation Consortium (CPIC) guideline for HLA genotype and use of CBZ and OXC [21].
In the present study, we support the hypothesis that multiple HLA alleles might be involved in AEDs-induced cADRs. Interestingly, HLA-B*51:01 was also found significantly associated with AEDs-induced cADRs with OR 2.292, but when we separated the data in each drug category, we found that HLA-B*51:01 was significantly associated with PHT-induced STS/TEN and CBZinduced DRESS compared to tolerant controls with OR of 6.19 and 7.94, respectively. SJS/TEN caused by PHT and PB was significantly associated with the HLA-B*51:01 allele in the Japanese population [22]. Moreover, the carrier rate of HLA-B*13:01 in the PB-induced SCARs was 100% (n = 4). This study first demonstrated the association between HLA-B*13:01 and PB-induced SCARs in Thai adults [23]. Further, we corroborated the findings of Li et al.
In addition, the HLA-B*15:02 allele has been reported to be specifically associated with CBZ-induced SJS/TEN in Asian Fig. 1 (Continued). C Sukasem et al. populations, and no associations have been reported for druginduced MPE and DRESS [2,8,16,25,26]. In this study, we found significant associations between HLA-B*15:02 and AEDs-induced MPE and SJS/TEN, however, this significance was not observed in DRESS cases [27]. Interestingly, previous studies reported no associations between the HLA-B*15:02 allele and the AEDsinduced MPE [14,28,29]. We believe that further studies need to be conducted to explore the molecular mechanism explaining the  HLA-B*15:11 allele, another member of the HLA-B75 serotype, was found in SJS/TEN cases due to CBZ [25,[30][31][32][33][34]. These findings suggest that HLA-B75 serotype might have a role in AEDs-induced SJS/TEN, suggesting the need for screening of other alleles of the HLA-B75 serotype in case of HLA-B*15:02 negative screening. In DRESS group, although we found positive correlations between AEDs-induced DRESS and HLA-B*08:01, HLA-B*13:01, and HLA-B*56:02 alleles. If Bonferroni correction was performed for multiple comparisons, the significance of these alleles would be lost.
In the PHT group, we found similar frequency of the HLA-B*15:02 allele between the cases and the controls (19.40% vs 17.10%). This result suggests that the patient carrying HLA-B*15:02 allele is not at risk of PHT-induced cADRs, this finding contradicts that of Dean L., who recommended avoiding PHT as an alternative for CBZ in patients positive for HLA-B*15:02 [19]. In addition to PHT, we further genotyped the CYP2C9*3 variants in 92 cases of PHT-induced cADRs and found that only 9 patients showed CYP2C9 variation (4 SJS/TEN, 4 DRESS, and 1 MPE). No signification between CYP2C9*3 and PHT-induced cADRs. There was a trend that the carrier rates of CYP2C9*3 were higher in the PHT-induced SJS/TEN group than in the tolerant control group in this study. However, previous studies demonstrated that CYP2C9*3 and having Chinese ancestry were significant risk factors of PHTinduced SJS (adjusted OR = 5.40, p = 0.0097). [35][36][37][38].
HLA-A* 31:01 also showed to increase the risk of MPE and DRESS due to CBZ in European and Han-Chinese population [21]. Therefore, we analyzed the HLA-A in patients with CBZinduced DRESS (n = 5) and CBZ-induced MPE (n = 16). Surprisingly, HLA-A*31:01 was not found in the patients with DRESS and MPE groups, suggesting that the HLA-A*31:01 allele could not be associated in the development of CBZ induced MPE and DRESS in Thai population. However, there was a trend that carriers rate of HLA-A*33:03 were higher in the MPE and DRESS groups than in the tolerant group. HLA-A*33:03 allele was detected in 60.00% of CBZ-induced DRESS (3/5) and 37.5% of CBZ-induced MPE (6/16). The HLA-A*31:01 is very low frequency (allele frequency=0.0085) but HLA-A*33:03 is the most common HLA-A alleles in Thai (allele frequency = 0.1117) [39]. In addition, a previous sequence analysis showed that HLA-A*31:01 and HLA-A*33:03 binding peptides have the same anchor residues at P2 and the C-terminus, the only difference being that HLA-A*33:03 binding peptides have two additional P2 anchor residues [40].
Finally, the meta-analysis was performed to clarify the relationship between HLA-B*15:02 and cADRs. There was a significant association between HLA-B*15:02 and AEDs-induced cADRs, and AEDs-induced SJS/TEN. while the association between HLA-B*15:02 and AEDs-induced DRESS did not reach statistical significance.
In conclusion, we found the association between HLA-B*15:02 and AEDs-induced cADRs. We also revealed an ambiguous association between the HLA-B*15:02 and the AEDs. Our results suggest that HLA-B*15:02 is not a risk allele for the cross-reactivity of AEDs-induced cADRs. We did not find any significant association between HLA-B*15:02 and PHT and PB-induced cADRs. If a patient has a positive HLA-B*15:02 genotype, a prescription for OXC needs to be carefully considered.