|Home | About | Journals | Submit | Contact Us | Français|
Several compelling lines of evidence suggest an important influence of genetic variation in susceptibility to Kawasaki disease (KD), an acute vasculitis that causes coronary artery aneurysms in children. We performed a family-based genotyping study to test for association between KD and 58 genes involved in cardiovascular disease and inflammation. By analysis of a cohort of 209 KD trios using the transmission disequilibrium test, we documented the asymmetric transmission of five alleles including the interleukin-4 (IL-4) C(−589)T allele (P = 0.03). Asymmetric transmission of the IL-4 C(−589)T was replicated in a second, independent cohort of 60 trios (P = 0.05, combined P = 0.002). Haplotypes of alleles in IL-4, colony-stimulating factor 2 (CSF2), IL-13, and transcription factor 7 (TCF7), all located in the interleukin gene cluster on 5q31, were also asymmetrically transmitted. The reported associations of KD with atopic dermatitis and allergy, elevated serum IgE levels, eosinophilia, and increased circulating numbers of monocyte/macrophages expressing the low-affinity IgE receptor (FCεR2) may be related to effects of IL-4. Thus, the largest family-based genotyping study of KD patients to date suggests that genetic variation in the IL-4 gene, or regions linked to IL-4, plays an important role in KD pathogenesis and disease susceptibility.
Kawasaki disease (KD) is an acute, self-limited vasculitis that is now the leading cause of acquired heart disease in children in the US and Japan. 1,2 The vascular inflammation results in permanent damage to coronary arteries with the development of aneurysms in up to 25% of untreated children.3 Since the original description over 30 years ago by Tomisaku Kawasaki, a pediatrician in Japan,4,5 investigators have searched unsuccessfully for a causative infectious agent.6 Several lines of evidence point to an important role of genetic variation in KD susceptibility. For example, while KD has been reported in most ethnic groups, it is over-represented among Asians and Asian/American populations and is 1.5 times more common in males.7–9 In Hawaii, the annual incidence for Japanese-Americans is 145/100 000 children <5 years of age, which is similar to the incidence for Japanese living in Japan and approximately 10-fold higher than the rate for Caucasian Americans.10,11 Asians in San Diego County, where active surveillance for KD has been performed since 1994, have a 2.7-fold increased risk as compared to all other ethnic groups.12 In Japan, siblings of an index case have a 10-fold increased risk of KD compared to the general population.13 In addition, the incidence of KD is two-fold higher in Japanese parents of children with KD.14 Genetic association studies in KD have been performed in small cohorts, but none have been replicated to date.15–20 A recent study used a genome-wide scan in 82 Japanese KD sibling pairs and identified a polymorphism in the CD40 ligand (TNFSF5 [MIM300386]) gene on Xq26 that was associated with aneurysm formation in male infants with KD.21
To search for genes that influence susceptibility to KD, we evaluated genetic variation in candidate genes known to be involved in inflammation and cardiovascular disease in children diagnosed with KD and their families (Figure 1 and Supplemental Table 1). We chose a candidate-gene association study, rather than a genome-wide scan, because of the small number of families with multiple affected members and strong biologic reasons to choose specific genes as candidates. Given that KD affects multiple ethnic groups with potentially different risk allele frequencies, we chose a family-based study design of affected trios analyzed by the transmission disequilibrium test (TDT).22 Our analysis of polymorphisms in 58 candidate genes (Supplemental Table 1) in KD children and their parents is the most comprehensive, family-based association study of KD patients reported to date. We present evidence that genetic variation in interleukin-4 (IL-4) (IL4 [MIM147780]) is implicated in susceptibility to KD. Further investigation of this gene may lead to new insights into KD pathogenesis.
All 95 polymorphisms in 58 genes were in Hardy–Weinberg equilibrium in parents and KD cases in Cohorts 1 and 2 (data not shown). Differential transmission was observed for five polymorphic alleles in Cohort 1 at a significance level of 0.05 or less (Table 1). These five alleles were from five different genes, which encode the following proteins (i) paraoxonase 1 (PON1 [MIM168820]), an enzyme involved in lipid peroxidation, (ii) G-protein-coupled receptor 2-like (GRK4 [MIM137026]), an intracellular kinase that inactivates several members of the G-protein-coupled 7 transmembrane-spanning receptor family, (iii) IL-4, a cytokine implicated in the T-helper (Th2) immune response, (iv) Transforming growth factor-β (TGFB [MIM190180]), a pleiotropic growth factor associated with acute inflammation, and (v) group-specific component for vitamin D binding (GC [MIM139200]), a precursor for a potent macrophage-activating factor. Of these five alleles, only the C allele of the IL-4 promoter polymorphism C(−589) T was again preferentially transmitted in Cohort 2 (P = 0.05). An analysis of this allele in the 269 trios combined from both cohorts resulted in a significance level of 0.002. Owing to the consistent difference in KD incidence between males and females (male predominance 1.5 : 1.0), we stratified the sample of 269 trios by the sex of the affected child and reran the TDT. The trend was the same for both sexes: 39 vs 24 (P<0.04) for girls and 63 vs 41 (P<0.02) for boys.
Polymorphisms in the genes for colony-stimulating factor 2 (CSF2 [MIM138960]), IL-13 (IL13 [MIM147683]), and transcription factor 7 (TCF7 [MIM189908]), all located on 5q31 (Figure 2), were among those genotyped in both cohorts. Pair-wise linkage disequilibrium (LD) between these loci and IL-4 revealed that none of these polymorphisms was in complete LD with IL4 C(−589)T (Table 2). Although there was evidence for asymmetric transmission of particular haplotypes, the significance levels were not greater than for the asymmetric transmission of the IL-4 C(−589)T allele alone (Table 3), suggesting that these haplotypes are not more specific risk factors for KD. As the IL-4RA (IL4R [MIM147781]) gene is on chromosome 16p12 and encodes the α-chain of the heterodimeric receptor for both IL-4 and IL-13, we also tested for gene–gene interactions between IL4 C(−589)T and three IL-4RA alleles (Ile50Val, Ser478Pro, and Gln551Arg) in Cohort 1. No significant interactions were identified in the 145 non-Hispanic Caucasian KD trios from Cohorts 1 and 2 (data not shown).
The distribution of the IL-4 alleles in 121 KD patients with normal coronary arteries and 127 KD patients with dilatation or aneurysms in both cohorts was not significantly different (P = 0.09), suggesting no association of the IL-4 C(−589)T polymorphism with coronary artery outcome in KD (Table 4).
Our study implicates IL-4 or regions linked to IL-4 in KD susceptibility but not coronary artery outcome. The IL-4 gene is located together with IL-13, IL-9 (IL9 [MIM146931]), and IL-5 (IL5 [MIM147850]) on chromosome 5q31, which form an interleukin gene cluster.23 The association of IL-4 C(−589) with increased KD susceptibility raises interesting questions about the role of IL-4 and Th 2-cell response in this enigmatic disease.
IL-4 is produced by activated CD4 +-Tcells, mast cells, and basophils, and affects many different immunoregulatory pathways.24 IL-4 is most prominently involved in the differentiation of naïve Th cells to Th2 effector cells, which in turn secrete IL-4, IL-5, IL-6, IL-10, and IL-13. Several intriguing correlates between IL-4 and features of KD are worthy of mention. Elevated IL-4 levels have been reported in the serum of acute KD patients.25 Intracellular cytokine staining of peripheral blood mono-nuclear cells from KD patients has suggested a Th1/Th2 imbalance with a predominance of the Th2 phenotype during the acute stage.26 IL-4 also regulates the expression of CD23 (low-affinity IgE receptor, FcRεII), which is expressed as an activation antigen on the surface of monocyte/macrophages, B cells, platelets, and eosino-phils.27 Increased expression of CD23 on B cells and monocyte/macrophages has been demonstrated in acute KD.28,29 Levels of soluble CD23 are also increased during acute KD.30 IL-4 also upregulates vascular cell adhesion molecule-1 (VCAM-1), which is a cell surface glycoprotein expressed by cytokine-activated endothelium that mediates the adhesion of monocytes and lymphocytes. Soluble VCAM-1 levels are elevated in acute KD and may participate in the process of vascular injury.31–33
Evidence for linkage of IL-4 C(−589)T to elevated serum IgE levels, predisposition to asthma, atopic dermatitis, allergy, and severity of respiratory syncytial virus infection has been reported.34–39 Serum IgE levels are elevated in acute KD28,40–42 as well as in infantile polyarteritis nodosa, a condition thought to be synonymous with fatal KD.43,44 The prevalence of allergy and atopic dermatitis is increased in KD patients compared to age- and ethnicity-matched controls, again suggesting a link to IL-4.45–47 Thus, many features of the immune response in acute KD are consistent with the effects of IL-4 and a dominant Th2 response, adding support to the current finding of the association of an IL-4 polymorphism with susceptibility to KD.
In our two cohorts, the IL-4 T(−589) allele was preferentially not transmitted to offspring with KD. The IL-4 T(−589) allele has been associated with increased promoter activity by reporter gene expression assays in a single T-cell line in vitro.48 However, in vitro assays with cell lines may not accurately reflect the effect of a polymorphism in vivo, which is influenced by complex local environments and the products of other genes. Many more polymorphisms have been described in the IL-4 gene and other genes in the interleukin cluster of 5q3134 and a complex haplotype structure described.23 Analysis of haplotypes including IL-4 IL-13, TCF7 and CSF2 in our cohort was not more informative than the IL-4 C(−589)T alone. Genotyping and haplotype transmission analysis of additional polymorphic alleles in our KD cohort may further clarify the contribution of IL-4 to KD susceptibility.
As a pilot study involving only 269 children with KD, our trial has certain limitations. Our relatively limited number of subjects prevented stratification by ethnicity, and forced us to eliminate analysis of alleles that are rare in non-Hispanic Caucasians. Association of a disease with an allelic variation may be detectable only in certain ethnic groups and could be missed in an analysis performed without ethnic stratification. Thus, future studies in a larger KD cohort should include stratification by ethnicity.
In a complex disease such as KD, multiple genes are expected to influence disease susceptibility and outcome. The finding of an association of genetic variation in the IL-4 gene and KD susceptibility is the first step toward understanding these complex genetic influences. Replication of these findings should be sought in additional, independent cohorts. A more focused examination of additional genes involved in the IL-4 signaling pathway may yield further insight into KD susceptibility.
All KD patients who met 4/5 standard clinical criteria or 3/5 criteria plus coronary artery abnormalities documented by echocardiography (Online Supplement, Table S1)49 were entered with their biologic parents into the study after obtaining informed parental consent. The Institutional Review Boards of the participating clinical centers reviewed and approved this study.
The first cohort (Cohort 1) consisted of 209 trios, while Cohort 2 consisted of an independent sample of 60 trios. A total of 220 families (170 in Cohort 1 and 50 in Cohort 2) were enrolled at two clinical centers (Boston Children’s Hospital and Children’s Hospital San Diego) at the time of hospitalization for acute KD or during a subsequent clinic evaluation. In all, 49 trios from across the United States and Canada (39 in Cohort 1 and 10 in Cohort 2) who contacted the Kawasaki Disease Research Program at the University of California San Diego were also enrolled after review of their medical histories.
Demographic and clinical data including self-reported ethnicity were collected on all subjects (Online supplement). Of the 209 children in Cohort 1, 64.1% were male and the racial/ethnic distribution was as follows: 55.0% non-Hispanic Caucasian, 17.7% Hispanic Caucasian, 13.4% Asian /Pacific Islander, 0.5% African American, and 13.4% mixed/other/unknown. In Cohort 2, 55% were male and the ethnic distribution was 50.0% non-Hispanic Caucasian, 16.6% Hispanic Caucasian, 13.3% Asian /Pacific Islander, 3.3% African American, and 16.6% mixed/other/unknown. Coronary artery status was assessed by measuring the internal luminal diameter by echocardiogram during the acute and subacute illness. In Cohorts 1 and 2, 121 patients were classified as normal and 127 patients were classified as either dilated (>2 and <3 s.d. above the mean for body surface area50) or aneurysmal (focal dilatation >3 s.d.). Data were not available from some patients.
As KD is characterized by immune-mediated damage to the blood vessel wall, we selected 76 genes that had been previously implicated in the pathogenesis of cardiovascular and inflammatory diseases and encoded proteins involved in lipid metabolism, autonomic regulation, cell adhesion, extracellular matrix remodeling, cell signaling, platelet activation, immune response, and inflammation (Figure 1, Online supplemental Table S2). The 144 polymorphisms were selected either because they had been previously associated with clinical phenotypes in cardiac or immune-mediated diseases in the published literature or because they were in strong LD with such polymorphisms. As noted below (Study Design and Genetic Analysis), 49 infrequent polymorphisms were not analyzed further for disease association due to lack of statistical power.
Detailed methods for DNA preparation are presented in the Online supplement. Briefly, DNA was extracted from either 3 ml of blood16 or from 10 ml of mouthwash containing shed buccal cells.51 These methods yielded approximately 25–75 μg and 10–200 μg of DNA, respectively. An allele-specific amplification and detection system was developed by Roche Molecular Systems (Alameda, CA, USA),52,53 and the details are presented in the Online Supplement. Briefly, genomic regions containing polymorphic alleles were amplified by multiplex polymerase chain reaction (PCR) using 50 ng of DNA and biotinylated primer sets. PCR products were hybridized to a linear array of sequence-specific oligo-nucleotides immobilized on nylon strips and detected by colorimetric methods. Genotypes were read using software developed at Roche Molecular Systems and confirmed by visual inspection of the strips.
DNA samples with known genotypes were included as controls in each experimental run. Approximately 10% of the total study population was regenotyped and all original genotyping results were confirmed. On the five strips used for genotyping each individual, there were nine polymorphisms duplicated on different nylon strips. Data were examined for agreement between each pair of duplicate alleles and all results were consistent. Mendelian inheritance was assessed using the Pedcheck software54 and three families who did not follow Mendelian inheritance at multiple markers were eliminated from the data set leaving 209 families in Cohort 1 and 60 families in Cohort 2 for analysis. Each polymorphism was tested for Hardy–Weinberg Equilibrium separately for each ethnicity in parents and children using the Mendel software.55
Given the potential for Type I statistical errors when multiple candidate genes and alleles are tested and the desire to retain sufficient statistical power to detect associations with a sample of this size, a two-stage design was employed. In the first hypothesis-generating stage (Cohort 1), 49 alleles with a frequency estimate of less than 15% in non-Hispanic Caucasians were eliminated from further analysis, as there was insufficient power to detect differential allele transmission with a cohort of 209 trios. The remaining 95 polymorphisms in 58 genes were analyzed as described below. The level of significance was set at 0.05, which could result in observing five alleles as a false-positive test result. In the second hypothesis-testing stage, five polymorphisms identified as differentially transmitted in Cohort 1 were genotyped and analyzed in Cohort 2 using the same methods. Analysis of IL-4 C(−589)T transmission was also repeated on the combined sample of 269 trios from Cohorts 1 and 2.
The TDT22 was applied to 209 trios (Cohort 1) and 60 trios (Cohort 2) using the TDTEX software version 4.6 of the SAGE package.56 This analytic method applies the McNemar test of association in paired samples22 and computes an exact P-value using a permutation test where the alleles within the pairs are shuffled within trios. Significant preferential transmission of alleles from heterozygous parents to their affected children was interpreted as hypothesis-generating evidence implicating that gene in KD susceptibility. The TDT approach allowed families from different ethnic backgrounds to be analyzed together.
After the IL-4 C(−589) allele was found to be associated with KD, we examined LD of genes near IL-4 on chromosome 5q31.1 that were already genotyped in Cohort 1. Pair-wise LD with IL-4 C(−589)T was assessed separately for 28 Asian and 115 non-Hispanic Caucasian trios in Cohort 1 using an EM algorithm approach as programmed in the graphical overview of linkage disequilibrium (GOLD) software.57 None of these polymorphisms was in complete LD with IL4 C(−589)T, and thus, each was analyzed as a haplotype with the IL-4 polymorphism for preferential transmission to KD children in Cohort 1. The Transmit software,58 which infers haplotype frequencies using the EM algorithm, was employed. To assess gene–gene interactions between IL-4 and the IL4 receptor on Chromosome 16, an exact and doubly ordered association analysis of genotypes of KD Caucasian children in Cohort 1 was conducted using the StatExact software.59 The association of the IL-4 polymorphism with coronary artery status was analyzed using a contingency table analysis of a case–control study of 103 KD patients with normal coronary arteries (controls) and 106 KD patients with dilatation of the coronary arteries or aneurysms in Cohort 1.
This work was supported by grants from the National Institutes of Health, NIH-RO1HL69413 and K24HL074864 (awarded to JCB) and by a Grant-in-Aid from the American Heart Association, Western Affiliate #035061Y (awarded to JCB). Some of the results of this paper were obtained by using the program package SAGE which is supported by a US Public Health Service Resource Grant (RR03655) from the National Center for Research Resources. We thank John F Bastian for patient referral and collection of DNA samples. We also thank our clinical nurses Ellen McGrath, and Jennifer Foley, and the army of students including David Bronstein, Jennie Buchanan, Marina Dergun, Christina Lin, Erin Miller, and Mathew Leach without whose help we could never have collected all the DNA samples required for this study. We also thank Calvin Mano, Tracy Nguyen, and Nang Tan (RMS) for their support in producing the genotyping reagents, and Jeff Post (RMS) for the strip interpretation software used for this work.