This study was carried out in compliance with the Global Initiative for Asthma (GINA) Global Strategy for Asthma Management and Prevention Study. Subjects were recruited from the Asthma Genome Research Center comprising the hospitals of Soonchunhyang and Chung-Ang University in Korea. All subjects provided their informed consent, and the study protocols were approved by the Institutional Review Board of each hospital.

All patients in the study had a history of dyspnea and wheezing in the past 12 months, along with one or more of the following: (1) either a ≥ 15% increase in FEV1 or a > 12% increase in FEV1 plus 200 mL follow-up inhalation of a short-acting bronchodilator, (2) <10 mg/mL PC20 methacholine, and (3) ≥20% increase in FEV1 following 2 weeks of treatment with inhaled steroids and long-acting bronchodilators. In addition, 24 common inhalant allergens were used to conduct a skin prick test. Total IgE was measured with the CAP system (Phadia, Uppsala, Sweden), and atopy was defined as having a wheal reaction equal to or greater than the histamine or 3 mm in diameter. An oral aspirin challenge (OAC) was performed with increasing doses,12 based on the updated practice parameter (2010) of the American College of Allergy, Asthma & Immunology/American Academy of Allergy, Asthma & immunology.13,14 Briefly, patients with a history of aspirin hypersensitivity were given 30 mg orally. Respiratory and nasal symptoms, blood pressure, external signs (urticaria and angiooedema), and FEV1 were documented every 30 min for a period of 1.5 h. At the end of this period, if there were no symptoms or signs suggesting an adverse reaction, increasing doses of aspirin (60 mg, 100 mg, 300 mg, and 400 mg) were administered until the patient developed a reaction. The same measurements were repeated every hour. Those having no history of aspirin hypersensitivity were started on 100 mg of aspirin, and this dose was gradually increased to 200 mg, 350 mg, and 450 mg until the patient developed a reaction. If no reaction occurred 4 h after the final dose, the test was deemed negative. Aspirin-induced bronchospasm, reflected by a decline (%) in FEV1, was calculated as the pre-challenge FEV1 minus the post-challenge FEV1 divided by the pre-challenge FEV1. Based on the OAC reactions, AERD and ATA groups were categorized as follows: (1) 15% or greater decrease in FEV1 or nasal reactions (AERD group), and (2) less than 15% decrease in FEV1 without naso-ocular or cutaneous reactions (ATA group). Peripheral venous blood was collected before the aspirin challenge.

Four common ATF6B SNPs were selected based on minor allele frequency (MAF) >0.05 and LD status, from Asian populations in the International HapMap Project database (http://hapmap.ncbi.nlm.nih.gov/index.html.en). In the case of completely tagged SNPs (r²=1), only one was selected after considering location (e.g., functionally important SNPs were preferred). SNPs were then genotyped in 93 AERD and 96 ATA subjects using a Taqman assay, with assessment of data quality based on duplicate DNAs (n = 10). SNPs that could not satisfy the following criteria were excluded: (1) a minimum call rate of 95% for the accuracy of determining genotypes, (2) no duplicate error, and (3) HWE of P-value >0.05.

We examined Lewontin's D' (|D'|) and the LD coefficient r² between all pairs of biallelic loci to determine LD among the SNPs. LD was obtained using the Haploview v4.1 program downloaded from the Broad Institute.15 Haplotypes were first estimated using PHASE software,16 and then computed by logistic analysis using the Statistical Analysis System (SAS) v9.2. Associations were analyzed using logistic models with age (continuous value), gender (male = 0, female = 1), smoking status (nonsmoker = 0, ex-smoker = 1, smoker = 2), and atopy (absence = 0, presence = 1) as covariates. The effective number of independent marker loci was calculated to correct for multiple testing using the software SNPSpD (http://genepi.qimr.edu.au/general/daleN/SNPSpD/), which is based on the spectral decomposition (SpD) of the matrices of pair-wise LD between SNPs. The beta values indicating the effect sizes of differences in mean FEV1 among the genotypes of each SNP and haplotype were analyzed using the SAS. Statistical powers were calculated using the Power for Genetic Association Analyses (PGA) software,17 which is designed to calculate statistical power and other values in case-control association studies. For these measurements, the co-dominant (1df) model with a disease prevalence value of AERD 21%,3 effective degree of freedom (EDF) 2, and an alpha error level of 5% was used.

The clinical profiles of the study subjects are summarized in Table 1. A total of 189 asthmatic subjects were divided into 2 groups (93 AERD patients and 96 ATA controls) according to their records of FEV1 decline by aspirin provocation. The percentage of FEV1 decline and positive rate of nasal polyps and rhinosinusitis in AERD patients were higher compared to ATA controls (P<0.05). However, no significant differences in blood eosinophil, PC20 methacholine, total IgE, smoking status, and positive rate of skin test were found in the subgroups of this study (P>0.05).

Four common ATF6B SNPs with minor allele frequency (MAF) > 0.1, one in the promoter region, one in the coding region, and two in the 3'-untranslated region (UTR), were selected on the basis of their frequencies in Asian populations (Chinese and Japanese) from the International HapMap Project (Table 2, Fig. 1A). All alleles were successfully genotyped with an average call rate of 99.5% as shown in Table 2. The data reveal no significant departures from Hardy-Weinberg equilibrium (HWE) in any of the loci (P>0.05, Table 2). Four SNPs constituted four haplotypes with tight LD (Fig. 1B and C). Since ATF6B-ht2 and ATF6B-ht3 were equivalent to rs8283 and rs3830076, respectively, only ATF6B-ht1 and ATF6B-ht4 were included in the association analysis (Fig. 1B).

Interestingly, two SNPs (rs2228628 and rs8111) were significantly associated with % decline of FEV1 induced by aspirin challenge in the asthmatics (P=0.008 under the recessive model, Table 3), even after corrections for multiple testing (P^corr=0.03). Minor homozygotes of rs2228628 and rs8111 showed about a 2.5-fold increase in the mean decline of FEV1 compared to the major homozygotes and heterozygotes (Table 3). In the haplotype analysis, ATF6B-ht4 was also significantly associated with FEV1 decline in the asthmatics (P=0.005 and P^corr=0.02 under the recessive model, Table 3). At rs2228628, rs8111, and ATF6B-ht4, FEV1 declines were greatest for the homozygotes of the minor alleles (Table 3). However, logistic analysis adjusted for age, gender, smoking status, and atopy as covariates revealed no significant associations between ATF6B SNPs/haplotypes and the risk of AERD (P>0.05, Table 4). In further comparison of FEV1 decline by aspirin provocation between AERD and ATA groups, the strength of association with rs2228628 and rs8111 was increased in the AERD group (P=0.006-0.007 and P^corr=0.02 under the recessive model, Table 3), but no association was observed in the ATA group.

This study employed in silico analysis to predict potential function of the significantly associated rs8111 in the 3'-UTR. The UTRScan program (http://itbtools.ba.itb.cnr.it/utrscan), using a collection of sequences and regulatory motifs of the untranslated regions of eukaryotic mRNAs, revealed that 2 regulatory elements (the cytoplasmic polyadenylation element (CPE) involved in the polyadenine tail extension and mRNA activation for protein translation, and the upstream open reading frame (uORF) involved in interference with the expression of a downstream primary ORF) were discovered to include the rs8111 polymorphic site (Supplementary Table 1).