Introduction

Kawasaki disease (KD) is the leading cause of acquired heart disease in children in developed countries due to self-limiting systemic vasculitis [1]. KD is often complicated by coronary artery aneurysms (CAAs), which occur in 20–25% of untreated children and 3–5% of children treated with intravenous immunoglobulin (IVIG) [2, 3]. Aneurysms can be classified as small (< 5 mm internal diameter), medium (5–8 mm internal diameter), or giant (> 8 mm internal diameter) [4]. Approximately 20% of patients with a CAA may also subsequently require invasive cardiac treatment [5]. Coronary arteritis in patients with KD is known to lead to CAA with destruction of the vascular media and infiltration of inflammatory cells, including neutrophils, macrophages, dendritic cells, IgA plasma cells, and T cells [6,7,8,9]. These studies suggest that excessive inflammatory cell infiltration is important in the pathogenesis of CAAs in patients with KD.

Although the etiology of CAA in patients with KD remains unknown, efforts have been directed to reduce the development of CAAs. Clinical risk factors for CAA include long fever duration, IVIG resistance, male gender, and incomplete KD [10,11,12,13,14,15,16,17]. Despite the clinical importance of CAA in patients with KD, very limited numbers of genome-wide association studies (GWAS) using a small number of samples were performed to search genetic susceptibility loci for CAA formation in these patients [18, 19]. However, all of the suggested genetic loci for CAA formation have not been replicated by either the original study group or other independent investigators. We also previously performed a GWAS analysis using Affymetrix SNP chip data from KD cases only by grouping 49 patients with KD and a large aneurysm (diameter ≥ 5 mm) versus 314 patients with KD without CAA and identified one CAA-associated locus, the rs17136627 SNP in the potassium intermediate/small conductance calcium-activated channel, subfamily N, member 2 (KCNN2) gene [20]. In this study, we performed a GWAS analysis using Illumina BeadChip data with a large number of case and control samples to identify additional CAA susceptibility genes in patients with KD.

Methods

Study Subjects and Genotype Data

All 1009 of our patients with KD were diagnosed by pediatricians according to the diagnostic criteria of the American Heart Association [1]. Information detailing the samples has been described in our previous studies [21, 22]. For genetic study, whole blood samples (approximately 5 ml) were collected from each patient at either acute, subacute, convalescent, or normal stage. Genomic DNA was extracted from whole blood samples using the PUREGENE DNA purification kit (Gentra Systems, Minneapolis, MN) according to the manufacturer’s instructions. Patients with KD were classified as having CAAs if the largest lumen diameter of the coronary artery was ≥ 3 mm. A total of 296 patients with KD were genotyped for the GWAS on the Illumina HumanOmni1-Quad BeadChip, following the manufacturer’s instructions (Illumina, San Francisco, CA, USA). All samples had a genotyping call rate of > 98%. To filter the SNP markers, we excluded SNPs with a call rate ≤ 95% in samples, minor allele frequency ≤ 0.01, and deviation from Hardy–Weinberg equilibrium (P ≤ 0.0001). Genotyping for the replication study in the additional 713 patients with KD was performed on the high-throughput Fluidigm SNPtype system (Fluidigm Corp., San Francisco, CA, USA) using the Fluidigm SNPType™ assay platform. This study was approved by our institutional review board (IRB) at Asan Medical Center (IRB No. 2014-0823) and informed consent was obtained from the parents of the patients with KD.

Statistical Analysis

The genetic associations among single-nucleotide polymorphisms (SNPs) were analyzed using PLINK (ver. 1.07) [23]. A Chi-square test was performed to compare allelic frequencies between the cases and the controls. Genetic associations were estimated using odds ratios (ORs) with 95% confidence intervals (CIs). From the initial GWAS analysis, 44 candidate SNPs for a replication study were chosen by selecting loci that were involved with cardiac functions or immune responses and that had significant P-values (P < 0.001). Regional association plots and local linkage disequilibrium (LD) values were generated from the GWAS results using LocusZoom software [24].

Results

A SNP (rs899162) Located 7 kb Upstream of the TIFAB Gene is Associated with CAA in Patients with KD

To identify the genetic variants associated with CAA formation in patients with KD, we initially carried out a GWAS using the Illumina Omni1-Quad BeadChip data from 248 patients with KD, including 215 KD patients without CAAs (control) and 33 patients with CAAs (diameter ≥ 5 mm) (case). We selected 44 candidate SNPs showing significant associations with CAA formation in the patients with KD for the replication study (P < 0.001) from functionally immune- or cardiovascular-related genes and examined them in an independent sample set consisting of 544 patients with KD (532 patients without CAA and 12 patients with CAA, diameter ≥ 5 mm) (Supplementary Table 1). Among 44 candidate SNPs, a SNP (rs899162) located 7 kb upstream of the TIFAB gene for CAA formation (diameter ≥ 5 mm) was replicated with significance in patients with KD (OR 3.20, 95% CI 2.02–5.05, Pcombined = 1.95 × 10−7) (Table 1). This locus contained the TIFAB and DCANP1 genes (Fig. 1), indicating that either the TIFAB and/or the DCANP1 gene is a CAA susceptibility gene. Interestingly, the risk genotype (AA or AG) of the significantly associated SNP rs899162 in the TIFAB-DCANP1 locus was enriched in KD patients with large aneurysms (diameter ≥ 5 mm) (Fig. 2).

Table 1 Association between the TIFAB-DCANP1 locus (rs899162) and CAA formation in patients with KD
Full size table
Fig. 1
figure 1

Regional association plot of the novel TIFAB-DCANP1 (rs899162) locus for CAA susceptibility from the results of our initial genome-wide association study data analysis using 33 patients with CAAs (diameter ≥ 5 mm) versus 215 patients without a CAA. The vertical axis shows the negative logarithm of the association P-values, and the horizontal axis shows the position in mega bases. The estimated recombination rates (right y-axis) were established based on 1000 Genomes Project Asian data (November 2014). The color of each SNP is based on the linkage disequilibrium (r2) with the lead SNP rs899162 for the TIFAB-DCANP1 locus (shown in purple). CAA coronary artery aneurysm, SNP single-nucleotide polymorphism

Full size image
Fig. 2
figure 2

Distribution of the risk genotype frequency of rs899162 (risk allele: A) in the TIFAB-DCANP1 locus according to the size of the CAA. The rs899162 genotypes are shown on the top of the figure by CAA size: normal, non-CAA; small, < 5 mm internal diameter; medium, 5–8 mm internal diameter; giant, > 8 mm internal diameter. CAA coronary artery aneurysm

Full size image

Discussion

KD is the leading cause of heart disease in children due to acute febrile and systemic vasculitis of unknown origin. Here, we performed a GWAS and a replication study in an independent sample set to identify genetic loci for CAA susceptibility. We found that a SNP (rs899162) located 7 kb upstream of the TIFAB gene was significantly associated with the extreme phenotype of CAA formation (≥ 5 mm diameter) in KD patients with a large effect size (OR 3.20, 95% CI 2.02–5.05, P = 1.95 × 10−7). This locus contained the TIFAB and DCANP1 genes, suggesting that either the TIFAB or DCANP1 gene may play a role as a CAA susceptibility gene.

The TIFAB gene was identified by a homology search of the human and mouse genomic databases for TIFA with the FHA domain [25]. TIFAB is mainly expressed in B cells, dendritic cells, and macrophages [26]. In addition, TIFAB is associated with TIFA and inhibits TIFA-mediated activation of nuclear factor-kappa-B [27]. TIFAB is also known to be a negative regulator of TRAF6-induced cell proliferation and maturation in hematopoiesis [26]. TIFAB is also a haploinsufficient gene in del (5q) myelodysplastic syndrome that shows skewed proportions of hematopoietic stem/progenitor cells and altered myeloid differentiation, resulting in blood cytopenia, myeloid dysplasia, and a predisposition to acute myeloid leukemia or bone marrow failure [27]. In contrast, DCANP1, which is mainly localized in the perinucleus, is highly expressed in mature dendritic cells, whereas it is poorly expressed in immature dendritic cells and not expressed in monocytes or B cells [28]. Interestingly, unlike other febrile diseases, the number of circulating myeloid dendritic cells decreases during the acute phase of KD [29] and arterial mature myeloid dendritic cells accumulate in the coronary arterial wall with CD3+ T cells [8]. These data suggest that the dysregulation of TIFAB- or the DCANP1-mediated inflammatory response plays a role in the development of CAA formation in patients with KD.

As the TIFAB and DCANP1 genes are in same LD block of SNP rs899162, it is unknown which gene is the CAA susceptibility gene. However, the most significant CAA-associated SNP (rs899162) was located closer to TIFAB (7 kb upstream of the TIFAB gene) rather than the DCANP1 gene (11.9 kb upstream of the DCANP1 gene). In addition, a tightly linked SNP (rs1031844) of the eQTL SNP in the DCANP1 gene (rs13154143, located 456 bp upstream of the DCANP1 gene) (P = 1.22 × 10−13) was not associated (P = 0.063) with CAA formation in patients with KD (Supplementary Table 2). These data strongly suggest that TIFAB is more likely the causative susceptibility gene for CAA formation in patients with KD. However, we cannot exclude DCANP1 as a potential candidate gene for CAA formation. Although the tightly linked SNPs of the eQTL SNP in the DCANP1 gene were not associated with CAA formation, a CAA-associated SNP (rs899162) located 7 kb upstream of the TIFAB gene has been reported to be associated with expression of the DCANP1 gene in whole blood in a previous eQTL study (P = 3.52 × 10−7) [30] (Supplementary Table 2). This result may support the possibility of differential expression of DCANP1 by the CAA-associated SNP rs899162 genotype, subsequently playing a role as a susceptibility gene for CAA formation. Furthermore, to identify any potential functional coding variants in the TIFAB and DCANP1 genes associated with CAA formation in patients with KD, we also examined our whole exome sequencing data, containing 22 KD patients with CAAs (diameter ≥ 5 mm) and 126 KD patients without CAAs. However, we could not find any coding variants in the TIFAB and DCANP1 genes associated with CAA formation in the patients with KD (data not shown). This result suggests that allelic gene expression in TIFAB-DCANP1 locus plays an important role in CAA formation in patients with KD.

A limited number of GWAS was performed to identify genetic markers for coronary artery aneurysm in Kawasaki disease [18,19,20]. However, all these previous GWAS used small sample size (less than 100 cases with CAA). The strengths of our study were that we used a genome-wide SNP array chip comprising approximately 1,000,000 SNPs and relatively large sample size (a total of 216 cases with CAA) to identify a new locus that has not been previously reported. However, the P-value of rs899162 in TIFAB locus did not reach genome-wide significance, and it was not validated in other ethnic population yet.

In conclusion, we identified a SNP (rs899162) located 7 kb upstream of the TIFAB gene, which is significantly associated with CAA formation in patients with KD. This result suggests that TIFAB plays an important role in the pathogenesis of CAA in these patients. Further larger studies are required to validate whether this candidate locus plays a role as a CAA susceptibility gene. It is also necessary to validate a functional role for TIFAB and DCANP1 as well as the CAA-associated SNP (rs899162) in CAA susceptibility.