Main

KD is an acute systemic vasculitis that affects infants and children and is the leading cause of acquired heart disease in the pediatric age group in the United States and Japan(1, 2). Although the first reports of KD were in Japanese children(3, 4), it is now recognized in children of all races and ethnic groups. Although an infectious agent is suspected, the cause of KD remains unknown. After the acute illness, approximately 15 to 25% of untreated patients will develop damage to the coronary arteries as a result of intense inflammation in the vessel wall(3–5). The high incidence of KD in Japan and among Americans of Japanese descent suggests that these groups may be genetically predisposed to this disease(2, 6). Linkage of different genetic markers (class I and class II MHC antigens, Ig allotypes) with KD has been sought, but only weak associations have been found(7–13).

A number of immunoregulatory changes are observed in children with KD that may contribute to the pathogenesis of the disease(14, 15). Serum levels of several immune mediators are increased, including TNF-α, IL-1, IL-2, IL-6, IL-8, sIL-2R, and IFN-γ(16–19). TNF-α levels are elevated in the majority of children during the acute phase of the disease(17–19) and are highest in children in whom CAA develop(17, 18). In vitro studies using vascular endothelial cells have demonstrated that TNF-α, IL-1, and IFN-γ induce the expression of surface antigens that render the cells susceptible to lysis by IgG or IgM antibodies in the sera of children with acute KD(20, 21). TNF-α, therefore, may play an important role in the pathogenesis of the vascular injury in KD.

The stimulus in KD for the increased serum levels of TNF-α remains unknown. The gene coding for TNF-α lies within the MHC on chromosome 6(22). A number of genetic polymorphisms upstream from the coding sequence for TNF-α have been described and include the nucleotides at LT-α+250, TNF-α−1031, −863, −857, −308, and −238 (Fig. 1)(23–29). The presence of the A allele at the −308 site is associated with elevated TNF-α production in response to endotoxin in whole blood cell cultures(23). The LT-α+250 polymorphic site is approximately 3.2 kb upstream from the transcriptional start site for TNF-α gene and has been associated with higher levels of TNF-α production in stimulated, cultured monocytes (Fig. 1)(27). We tested the hypothesis that children with KD may have a higher frequency of the A allele at the LT-α+250 and the TNF-α−308 sites, each of which are associated with elevated serum levels of TNF-α after an inflammatory stimulus. In addition, we examined the genotypic and allelic frequencies in white and Japanese children with and without KD to determine whether differences in these frequencies were observed across ethnic groups.

Fig. 1
figure 1

Location of the polymorphic sites affecting TNF-α response. Gray boxes, exons; black boxes, introns; hatched area, promotor; white boxes, 5′ and 3′ untranslated regions. Not to scale.

Full size image

METHODS

Patients

Japanese patients and controls were all born in Japan, and white patients and controls were all born in the United States. Informed consent was obtained from patients or their parents or guardians, and the study was approved by the institutional review boards at the participating institutions. Human experimentation guidelines of the U.S. Department of Health and Human Services and those of the authors’ institutions were followed. Efforts were made to recruit patients with CAA so that an association between the polymorphisms and coronary artery abnormalities could be assessed. Children included in the study met the standard clinical criteria for KD(6) and had fever of ≥38°C for 5 or more days and four of the five following criteria:1) cervical lymphadenopathy >1.5 cm in diameter;2) erythematous rash;3) erythematous bulbar conjunctiva without exudate;4) erythematous mouth and pharynx, strawberry tongue, or red, cracked lips; and 5) induration of the hands and feet, erythematous palms and soles, or periungual desquamation. The control populations consisted of adult volunteers with no history of KD, autoimmune diseases, or chronic treatment with anti-inflammatory agents.

Nomenclature for polymorphisms

The LT-α+250 describes the polymorphism 250 bases downstream from the transcriptional start site within the LT-α gene (Fig. 1). The two alleles at this site have been previously designated as TNFB1 when a guanine is present and TNFB2 when an adenine is present(25, 26). The TNF-α−308 describes the polymorphism 308 bases upstream from the transcriptional start site for the TNF-α gene (Fig. 1). Polymorphisms at the TNF-α−308 locus have been previously designated as TNF1 when a guanine is present and TNF2 when an adenine is present(26). We will refer to polymorphisms at these loci as the G allele or the A allele at the LT-α+250 or the TNF-α−308 site.

Genotypic analysis

Whole blood (1.0 mL) was collected, and DNA was extracted using the Genomic DNA Purification Kit (Promega, Madison, WI, U.S.A.). The LT-α+250 polymorphism contains an Nco I restriction site when the G allele is present. We amplified a 782-bp fragment in a PCR mixture containing 20 ng of DNA, 20 pmol each of the primers TNFB+250-1 (5′-CCGTGCTTCGTGCTTTGGACTA-3′) and TNFB+250-2 (5′-AGAGGGGTGGATGCTTGGGTTC-3′)(25), 1 unit of Taq polymerase, 1× reaction buffer (Promega), 500 μM each of deoxy-ATP, deoxy-cytidine triphosphate, deoxy-guanosine triphosphate, and deoxy-thymidine triphosphate, and 2.5 mM of MgCl2. Reaction conditions were as follows: 35 cycles of denaturation at 94°C for 30 s, annealing at 68°C for 30 s, and extension at 74°C for 42 s. The amplified DNA was incubated with Nco I. The Nco I-treated fragments were analyzed by electrophoresis in a 1% agarose gel and visualized by ethidium bromide staining. Interpretation was as follows: a single band at 782 bp identified individuals homozygous for an adenine at the LT-α+250 locus; two bands at 586 and 196 bp identified individuals homozygous for a guanine at the LT-α+250 locus; three bands at 782, 586, and 196 bp identified individuals heterozygous at the LT-α+250 locus.

The region containing the TNF-α−308 locus was amplified using the primers TNFA-308-1 (5′-AGGCAATAGGTTTTGAGGG C CAT-3′) and TNFA-308-2 (5′-ACACTCCCCATCCTCCCTGCT-3′)(29). The TNFA-308-1 primer contains 4 bp of the Nco I recognition sequence including a mismatched cytosine as shown by the C in the TNFA-308-1 primer sequence. This mismatched cytosine allows for creation of an Nco I restriction site (CCATGG) when the G allele is present at position −308. A 116-bp PCR product was generated using the following reaction conditions: 35 cycles of denaturation at 95°C for 30 s, annealing at 64°C for 15 s, and extension at 74°C for 15 s. The amplified DNA was incubated with Nco I. The Nco I-treated fragments were analyzed by electrophoresis in an 8% polyacrylamide gel and visualized by ethidium bromide staining. Interpretation was as follows: a single band at 116 bp identified individuals homozygous for an adenine at the TNF-α−308 locus; two bands at 96 and 20 bp identified individuals homozygous for a guanine at the TNF-α−308 locus; three bands at 116, 96, and 20 bp identified individuals heterozygous at the TNF-α−308 locus.

Statistics

Statistical analysis on genotypic and allelic frequencies was performed using Fisher’s exact test. CIs for the odds ratios were determined using the method of Woolf(30). Linkage disequilibrium analysis between the two polymorphisms was performed using Fisher’s exact test on the 3 × 3 table of genotypes.

RESULTS

Patient characteristics

Forty-six white children and 39 Japanese children with KD were enrolled in the study. Fourteen (30%) of the white children and 11 (26%) of the Japanese children had evidence of CAA or dilatation by echocardiography. Of the 46 white patients, 44 were treated with i.v. Ig (2 g/kg), and 12 (27%) had persistent fever 48 h after their initial i.v. Ig infusion and required one or more additional doses. Persistent fever after i.v. Ig administration could not be assessed for the Japanese patients because a variety of different treatment regimens and i.v. Ig doses were used.

Genotypic and allelic frequencies at the LT-α+250 site

The genotypic frequencies of the LT-α+250 polymorphic site located in the first intron of the LT-α gene were compared between children with KD and a healthy control population (Table 1). There was a higher frequency of the A/A genotype among white children with KD compared with the healthy white controls (0.59 versus 0.36;p = 0.013; odds ratio, 2.5; 95% CI, 1.23 to 5.09). The frequencies of the A and G alleles at the LT-α+250 site in white children with KD were also significantly higher compared with the healthy white controls (A allele, 0.77 versus 0.61;p = 0.009; odds ratio, 2.2; 95% CI, 1.24 to 3.79;Table 2). In contrast, allelic frequencies in Japanese patients with KD were not different from Japanese controls (A allele, 0.58 versus 0.64;p = 0.51; odds ratio, 0.76; 95% CI, 0.41 to 1.31). The allelic frequencies for the white and Japanese control populations were similar to previous reports(25, 31, 32).

Table 1 Genotypic frequencies at the LT-α+250 and TNF-α−308 loci in white and Japanese children with KD compared with healthy control populations
Full size table
Table 2 Allelic frequencies at the LT-α+250 and TNF-α−308 loci in white and Japanese children with KD compared with healthy control populations
Full size table

Genotypic and allelic frequencies at the TNF-α−308 site and linkage disequilibrium

The genotypic and allelic frequencies of the TNF-α−308 polymorphic site located in the promoter region of the gene coding for TNF-α were not different between white and Japanese children with KD and their respective ethnically matched control populations (white, A allele, 0.09 versus 0.17;p = 0.08; odds ratio, 0.48; 95% CI, 0.21 to 1.07; Japanese, A allele, 0.00 versus 0.04;p = 0.25; odds ratio, 0.14; 95% CI, 0.01 to 2.71;Tables 1 and 2). Both control populations demonstrated allelic frequencies similar to previously published reports(23, 24, 26). Analyses revealed a significant linkage disequilibrium (p < 0.000001) and, in particular, an association between the A allele at the LT-α+250 site and the G allele at the TNF-α−308 site. Thus, an individual who has the A allele at the LT-α+250 site is more likely to have the G allele at the TNF-α−308 site than one would expect by chance alone.

Genotypic and allelic frequencies in children with KD and coronary artery abnormalities or the need for i.v. Ig retreatment

An association was found between the frequency of the A/G genotype at the TNF-α−308 locus and the presence of coronary artery abnormalities (0.36 versus 0.09;p = 0.044; odds ratio, 5.4; 95% CI, 1.07 to 27.01) in white children with KD (Table 3). No association was found between genotypic frequencies at the LT-α+250 locus and coronary artery abnormalities, nor was there an association observed between the genotypic or allelic frequencies at either the LT-α+250 locus or the TNF-α−308 locus and the need for retreatment with i.v. Ig.

Table 3 Genotypic frequencies at the LT-α+250 and TNF-α−308 loci in white and Japanese children with Kawasaki disease and coronary artery abnormalities
Full size table

DISCUSSION

The inflammatory reaction produced by a host in response to a stimulus is influenced by a number of factors, including the type of stimulus, the dose of stimulus, and genetic characteristics of the host. The stimulus for the inflammatory reaction in children with KD remains unknown. In this study white children with KD were more likely to have the A/A genotype at the LT-α+250 site compared with a healthy control population and were more likely to have the A/G genotype at the TNF-α−308 site if they developed coronary artery abnormalities. Individuals with either of these genotypes have a higher TNF-α response than individuals with the G/G genotype at either locus. Thus, development of KD and coronary artery damage in white children may be related to the magnitude of the host’s TNF-α response. In Japanese children, in whom the incidence of KD is almost 10-fold higher, no association with these polymorphisms was found. The low prevalence of the A allele in the Japanese population coupled with our small sample size may have precluded our ability to detect a difference.

We assumed a priori that there would be genetic heterogeneity between the Japanese and white populations and that different loci might influence disease susceptibility and outcome in the two racial groups. Thus, we did not choose to pool the two data sets. We did not detect a difference in genotype frequencies between the Japanese and white control samples, indicating that pooling would be possible. However, there were significant differences in genotype frequencies, and hence genotypic heterogeneity, among the cases for both the LT-α+250 (p = 0.014) and TNF-α−308 (p = 0.007) sites.

Certain polymorphisms within the promoter region of the gene coding for TNF-α are associated with elevated TNF-α levels when peripheral blood mononuclear cells are treated with concanavalin A(24). Their positions relative to the TNF-α transcriptional start site are −1031C, −863A, −857A, −308A, and −238A. The presence of the A allele at the −308 site has also been shown to result in elevated TNF-α production in response to endotoxin in whole blood cell cultures(23). In a recent study, these loci were compared between Japanese children with KD and healthy Japanese controls, and, as in the present study, no association was found between high secretor genotypes and KD in Japanese children(28). Japanese patients with KD may still have a TNF-α high secretor phenotype, despite the failure to demonstrate a relationship between these loci and susceptibility to disease. The regulation of the TNF-α response to inflammation is controlled both at transcriptional and posttranscriptional levels and each may be influenced by independent genetic factors. A major factor in the determination of TNF-α levels is stabilization of TNF-α mRNA by AU-rich domains in the 3′ untranslated region of the mRNA(33–36)). The genetics of this posttranscriptional control mechanism are unknown. It is possible that polymorphisms influencing TNF-α mRNA stability are also important in KD and that these loci are influential in the Japanese population in determining disease susceptibility. A more complete understanding of the genetic influences on disease susceptibility must await a refinement in our knowledge of the genetics of the TNF-α response.

The LT-α+250 polymorphic site has been associated with human disease in only one other instance. Stuber et al.(25) demonstrated that the frequency of the A allele at the LT-α+250 site was higher in septic patients who died compared with survivors. The serum levels of TNF-α correlated well with the genotype; those patients with the A/A genotype had higher serum TNF-α levels whereas those with the A/G or G/G genotypes had lower serum TNF-α levels. Our study is the second report of an association between a human disease and the LT-α+250 polymorphism.

A limitation of our analyses is the small sample size coupled with the heterogeneity of the white population. We cannot exclude that population stratification alone could account for the observed differences in allele frequency. This preliminary study suggests that further testing of our hypothesis is warranted. We are currently exploring the difference in transmission frequencies of the A alleles at these polymorphic sites by the transmission disequilibrium test. By analyzing triads of affected patients and their parents, we will further test whether these alleles, or alleles in linkage disequilibrium with these alleles, are implicated in disease susceptibility or disease outcome.

An increased frequency of the TNF-α−308 A allele has been associated with poor outcome after certain infections. An association between serum TNF-α levels and mortality in meningococcal meningitis has been observed(37), and children with the A allele at the TNF-α−308 site have a worse outcome after meningococcal infection(38). Similar associations have been observed in individuals with malaria; those with higher TNF-α levels had a worse outcome(39), and the TNF-α−308 A allele site was overrepresented among patients with cerebral malaria and among those who died of malaria(40).

A genetic predisposition for KD is suggested by the higher prevalence of the disease among Japanese and Americans of Japanese descent. Several studies have demonstrated weak associations between KD and haplotypes in the MHC locus, suggesting that several genes within this locus may play a role in the pathogenesis of KD. For example, an association of the HLA antigens, HLA-Bw15 and Bw22, with KD was found in Japan(7), although not all studies support this finding(8). No association with HLA-Bw22 or class II antigens was observed in children with KD in the United States(9, 10, 12). Rather, a weak association with HLA-Bw51 and KD was observed in the United States(9) and Israel(11). The specific sites within these loci that are more common in children with KD have yet to be determined. Certain Ig allotypic markers were overrepresented among white but not Japanese KD patients(13). Taken together, no clear genetic influence on disease susceptibility emerges from these data.

Our findings suggest an association between the frequency of the A allele at the TNF-α−308 locus and the presence of CAA in white children. This raises the possibility that the vascular injury observed in children with KD is the result of a local inflammatory reaction in children genetically predisposed to have a more vigorous inflammatory response. If this association is supported by future studies, it may be possible to use genotyping to identify children who would benefit from closer monitoring and more-directed therapies to prevent vascular damage. Larger numbers of patients with and without aneurysms must be studied to more accurately assess the association between the TNF-α−308 allele and development of coronary artery abnormalities. The failure to find an association between the A allele at the TNF-α−308 locus and the presence of CAA in Japanese children may be related to our small sample size and the extremely low frequency of the A allele in the Japanese population. It may also be that the A allele at the LT-α+250 is in linkage disequilibrium in whites but not Japanese, with a more important TNF-α regulatory allele not examined in these analyses.

The mechanisms by which the polymorphic sites at positions LT-α+250 and TNF-α−308 affect TNF-α levels are unknown. The polymorphism at the TNF-α−308 site lies within the promoter region for the TNF-α gene and may alter the binding of transcription factors. Studies with the TNF-α promoter linked to a chloramphenicol acetyltransferase reporter gene have demonstrated that the A allele at TNF-α−308 site is associated with higher constitutive and inducible levels of transcription than the more common G allele(41). Kroeger et al.(42) demonstrated, furthermore, a difference in the ability of this region to bind nuclear proteins. The LT-α+250 site, on the other hand, is approximately 3.2 kb from the transcriptional start site for TNF-α. cis-acting transcriptional regulatory elements, such as enhancers, are known to act over such distances. Current studies are underway in our laboratory to identify the molecular mechanisms by which this region affects serum levels of TNF-α.

In summary, white, but not Japanese, children with KD have a higher frequency than a control population of the A/A genotype in one of two polymorphic sites in the 5′-flanking region of the gene coding for TNF-α. White children who are homozygous for A/A at the LT-α+250 site may be genetically predisposed to KD. Furthermore, white children with the A allele at the TNF-α−308 site may be at greater risk for the development of coronary artery abnormalities. These findings support the hypothesis that genetic polymorphisms that influence the host immune response may play an important role in susceptibility to KD and to disease outcome.