HLA‐A*24:02 increase the risk of allopurinol‐induced drug reaction with eosinophilia and systemic symptoms in HLA‐B*58:01 carriers in a Korean population; a multicenter cross‐sectional case‐control study

Abstract Background HLA‐B*58:01 is a well‐known risk factor for allopurinol‐induced severe cutaneous adverse reactions (SCARs). However, only a minority of HLA‐B*58:01 carriers suffer SCARs after taking allopurinol. The aim of this study was to investigate subsidiary genetic markers that could identify those at further increased risk of developing allopurinol‐induced drug reaction with eosinophilia and systemic symptoms (DRESS) in subjects with HLA‐B*58:01. Methods Subjects with B*58:01 were enrolled (21 allopurinol‐induced DRESS and 52 allopurinol‐tolerant control). HLA‐A, ‐B, ‐C and ‐DRB1 alleles were compared. Comparison of risk between HLAs and allopurinol‐induced SCAR in separate populations was performed to support the results. Kruskal‐Wallis test, Pearson's chi‐square test, Fisher's exact test and binary logistic regression were used to analyze the risk of SCAR development. Results Frequencies of A*24:02 (71.4 vs. 17.3%, p < 0.001, odds ratio [OR] = 12.0; 95% confidence interval [CI], 3.6–39.2) were significantly higher in B*58:01 (+) DRESS than B*58:01 (+) tolerant controls. In addition, DRB1*13:02 further increased the risk of DRESS. The phenotype frequency of A*24:02/DRB1*13:02 was significantly higher in the B*58:01 (+) DRESS group than in the B*58:01 (+) tolerant controls (52.4% vs. 5.8%, p < 0.001, OR, 66.0; 95% CI, 6.1–716.2). In 2782 allopurinol user cohort, the overall prevalence of DRESS was 0.22%, which increased to 1.62% and 2.86% in the presence of B*58:01 and B*58:01/A*24:02, respectively. Conclusion The additional secondary screening with A*24:02 and DRB1*13:02 alleles may identify those at further increased risk of allopurinol‐induced DRESS in B*58:01 carriers.


| BACKGROUND
Allopurinol, a xanthine oxidase inhibitor, is recommended as the firstline urate-lowering agent in patients with normal renal function according to the 2016 updated evidence-based recommendation. [1][2][3][4] According to a recent study, allopurinol might be safer than febuxostat, especially in those with cardiovascular disease. 5 Allopurinol has been regarded as a relatively safe drug since its launch in the 1960s, but it has the potential to induce severe cutaneous adverse reactions (SCARs), such as a drug reaction with eosinophilia and systemic symptoms (DRESS) and Stevens-Johnson syndrome (SJS)/toxic epidermal necrolysis (TEN). 6,7 The known risk factors of allopurinol-induced SCARs are HLA (human leukocyte antigen)-B*58:01, Asian ethnicity, old age, female, a higher initiating dose of allopurinol, and comorbidities such as cardiovascular disease and renal impairment. 8 Out of these, B*58:01 is the strongest risk factor.
Because the distribution of HLA alleles is different in ethnic groups, 9 the risk of developing SCARs is affected by ethnicity. 10 About 12% of the general population in Korea carries B*58:01, and this likely explains why allopurinol has been the most frequently reported culprit drug of SCARs. [11][12][13] The usefulness of B*58:01 screening in preventing allopurinol-induced SCARs was successfully proven in Korean patients with renal impairment in a prospective study. 14 However, allopurinol-induced SCARs developed only in 18% of B*58:01 positive high-risk subjects with chronic renal insufficiency. 13 The majority of B*58:01 carriers can safely take allopurinol without hypersensitivity reactions. A recent retrospective cohort study with 11 cases of allopurinol-induced SCARs in Korea reported that the secondary serotype screening may help to increase the accuracy of the prediction of allopurinol-induced SCARs. 4 The aim of the study was to improve the screening method to predict the risk of allopurinol-induced SCARs among subjects with B*58:01. We identified patients who were diagnosed with 'allopurinol-induced SCARs (DRESS or SJS/TEN)' from the Korean SCAR registry. 15 A drug (allopurinol) causality evaluation was performed by allergists using the WHO (World Health Organization)-Uppsala Monitoring Center criteria. 16 The "certain" and "probable" cases were considered as acceptable for inclusion. 'Allopurinol tolerant controls' were identified subjects who had taken allopurinol for more than 90 days without any hypersensitivity symptoms. All subjects were finally enrolled when they confirmed by medical records (pre-tested HLA Hospital were used to define B*58:01 (+) general population control. 17 The first validation set was HLA-tested allopurinol user cohort among 11998 HLA-tested subjects. The history of allopurinol exposure and hypersensitivity were collected from the electronic medical records and adverse drug reaction records. The second population was the data of a previous multicenter case-control study performed in Koreans, which included 25 cases of allopurinol-induced SCARs (20 cases of DRESS and five cases of SJS/TEN). 12 Among them, the B*58:01 (+) allopurinol-induced DRESS were compared with the B*58:01 (+) general population control. All subjects were validated in order not to be overlapped between validation sets.

| DNA isolation and HLA typing
Genomic DNA was extracted from blood samples using the Mini 80 nucleic acid isolation instrument with the QuickGene DNA whole blood kit S (FUJIFILM Co., Tokyo, Japan) and QIAamp® DNA Blood

| Genetic heterogeneity between DRESS and SJS/TEN groups with HLA-B*58:01
When the DRESS and SJS/TEN groups with B*58:01 were compared, there were no significant differences in the demographic characteristics. To assess the difference in the genetic predisposition of HLA alleles according to the SCAR phenotypes, we compared the phenotype frequency of the HLA alleles presenting at least 30% and 2-folds difference between the DRESS and SJS/TEN groups: A*24:02,

| Comparison of the HLA phenotype frequency of DRESS to allopurinol-tolerant controls and B*58:01 (+) general population controls
Considering the possibility of genetic differences between DRESS and SJS/TEN, the comparison with tolerant controls were performed by each group. When comparing all HLA phenotype frequencies between DRESS and the tolerant controls, A*24:02 showed a significant difference (Table S2). After Bonferroni correction, A*24:02 still showed a significant difference (p < 0.05). The phenotype frequency of A*24:02 was significantly higher in the DRESS group than in the tolerant controls (71.4% vs. 17.3%, p < 0.001, OR 12.0, 95% CI, 3.6-39.2) ( Table 2).    Figure 2).

| Binary logistic regression to evaluate risk factors in considering clinical differences between B*58:01 (+) allopurinol-induced DRESS and tolerant controls
Binary logistic regression analysis was performed to evaluate HLAs as independent risk factor in this study group (Table S3) Note: HLA-tested allopurinol user cohort was extracted from a dataset of 11,998 HLA-tested patients in a single tertiary hospital (unpublished data). All those who did not experience allopurinol-induced DRESS were considered controls.  Various factors may be involved in the binding of drugs between HLA and T cells. 18 Structural factors for drug recognition include the T cell receptors and major histocompatibility complex (MHC) molecules of antigen presenting cells which present peptides processed from exogenous antigens or endogenous peptides. HLA is a gene complex encoding the MHC molecules in humans and is the most studied risk factor in the development of SCARs. 8,18,19 In addition, other secondary signals may also be involved in this process. 8,20 An altered peptide repertoire due to the drug binding to HLA was proposed recently as an alternative mechanism for T cell reactivity in drug hypersensitivity. 21 Crystallographic structures revealed that certain drug could be embedded underneath the altered peptides and   phenotypes. This study reported, for the first time, the possibility of HLA differences between allopurinol-induced SJS/TEN and DRESS.
However, it was not possible to draw a firm conclusion because the number of SJS/TEN patients in this study was not large enough.
SCARs sometimes result in drug-induced hypersensitivity syndrome/ TEN overlap, encompassing the clinical features of both phenotypes. 32 The results of this study suggest that it is important to analyse SCAR according to its phenotypes.
In this study, there are some limitations. First, since we obtained the allopurinol control group from pre-tested HLA subjects, relatively severe CRI patients were included to the tolerant control. These subjects might be regarded as a specific disease group, and they are frequently exposed to allopurinol because of renal impairment.
Considering that severe renal impairment is one of the key risk factors for the SCAR development, it is an important factor in the analysis of the tolerability to allopurinol. However, A*24:02 was shown to be an independent risk factor for the development of