Exposures associated with the onset of Kawasaki disease in infancy from the Japan Environment and Children’s Study

Kawasaki disease (KD) is an acute systemic vasculitis that mainly affects infants and young children. The etiology of KD has been discussed for several decades; however, no reproducible risk factors have yet been proven. We used the Japan Environment and Children’s Study data to explore the association between the causal effects of exposure during the fetal and neonatal periods and KD onset. The Japan Environment and Children’s Study, a nationwide birth cohort study, has followed approximately 100,000 children since 2011. We obtained data on exposures and outcomes from the first trimester to 12 months after birth. Finally, we included 90,486 children who were followed for 12 months. Among them, 343 children developed KD. Multivariate logistic regression revealed that insufficient intake of folic acid during pregnancy (odds ratio [OR], 1.37; 95% CI 1.08–1.74), maternal thyroid disease during pregnancy (OR, 2.03; 95% CI 1.04–3.94), and presence of siblings (OR, 1.33; 95% CI 1.06–1.67) were associated with KD onset in infancy. In this study, we identified three exposures as risk factors for KD. Further well-designed studies are needed to confirm a causal relationship between these exposures and KD onset.


Scientific Reports
| (2021) 11:13309 | https://doi.org/10.1038/s41598-021-92669-z www.nature.com/scientificreports/ cohort. Unfortunately, these studies did not prospectively investigate fetal exposure. None of the risk factors reported in these studies have been verified in large-scale prospective studies. The Japan Environment and Children's Study (JECS) is a large prospective Japanese birth cohort study of approximately 100,000 children 24,25 . Its purpose is to prospectively evaluate the effects of exposures during the fetal stage and early childhood on pediatric health and development. The age at KD onset in Japan shows a monomorphic distribution with a peak at 9-11 months 26 . Although various factors have been implicated in the onset of KD, we hypothesized that exposures during the fetal and neonatal periods are associated with KD onset in infancy. We used data from the JECS to explore the association between the causal effects of exposures during the fetal and neonatal periods and KD onset. Our goal was to identify prenatal and perinatal exposures associated with early onset of KD using the JECS database. Figure 1 shows a flowchart of participant selection in our study. In total, 104,062 fetal records were registered during the study period. Finally, we included 90,486 children as the analysis population. Self-administered questionnaires at 12 months reported KD development in 343 of 90,486 children (0.38%; 95% CI 0.34-0.42%).

Results
Of 343 children with KD, detailed clinical information for 312 children (91.0%) was obtained from the KDspecific questionnaire. The mean age at KD onset was 6.9 ± 3.0 months (median, 7 months; range, 0-13 months). A total of 221 children (70.8%) fulfilled ≥ 5 major symptoms, and 91 children (29.2%) had incomplete KD. The   Table 1 shows the parental characteristics of participants. Neither a maternal nor paternal history of KD was significantly associated with KD onset in these children. However, a paternal history of allergic disease was significantly associated with KD onset. However, the paternal medical history was missing for approximately 50% of participants. Other maternal physical characteristics and socioeconomic status were not significantly associated with KD onset. Tables 2 and 3 show the maternal prenatal findings and perinatal backgrounds of participants. Maternal complications of thyroid disease during pregnancy were significantly associated with KD onset in children. However, only nine mothers (2.6%) in the KD group had thyroid disease. Presence of siblings of the affected child was significantly associated with KD onset. The frequency of folic acid supplementation of mothers during pregnancy was also significantly associated with KD onset. Table 4 shows the results of a multivariable logistic regression analysis. Maternal thyroid disease during pregnancy, insufficient intake of folic acid during pregnancy, and presence of siblings of the affected child were risk factors associated with KD onset in infancy.

Discussion
We conducted a comprehensive survey of demographic factors, medical and obstetric history, physical and mental health, lifestyle, environmental exposure, dwelling conditions, and socioeconomic status from the first trimester to 12 months after birth in the JECS nationwide birth cohort. Insufficient intake of folic acid during pregnancy, maternal complications of thyroid disease during pregnancy, and presence of siblings of the affected child were associated with KD in infancy. A novel finding in this study is the association between folic acid supplementation during pregnancy and the risk of KD. Folic acid supplementation during pregnancy reduces the risk of congenital malformations, such as neural tube defects 27,28 ; it acts as a cofactor in the nucleotide biosynthesis pathway and methylation reaction system 29 . Amarasekera et al. 30 reported that the neonatal cord blood concentration of folic acid is positively correlated with the maternal blood concentration of folic acid.
In addition, they found that maternal folic acid supplementation upregulated DNA methylation and genomic imprinting in the CD4-positive cells of offspring 30 . Another study showed that folic acid suppresses gene expression of inflammatory cytokines, such as interleukin-1β, tumor necrosis factor-α, and monocyte chemoattractant protein-1 in lipopolysaccharide-activated macrophages 31 . These inflammatory cytokines play essential roles in the pathophysiology of KD. Therefore, we speculated that sufficient exposure to folic acid in utero might affect the development of immunological function in offspring and consequently decrease the risk of KD onset. Because the estimated folic acid from daily meals was not different between the KD group and the control group, further study is necessary to clarify the mechanism by which onset of KD in early infancy can be prevented by maternal folic acid supplementation. We found that maternal complications of thyroid disease during pregnancy were associated with a significantly increased risk of KD onset in infancy. Belkaibech et al. 23 reported that maternal autoimmune diseases, particularly autoimmune thyroid disease, are risk factors for KD in offspring. They reported that the risk of KD onset at 1 year of age was 10.98-fold higher in the presence of maternal autoimmune thyroiditis (hazard ratio, 10.98; 95% CI 9.42-12.81) 23 . In our study, the risk of KD onset in infancy was 2.03-fold higher in the presence of maternal thyroid disease (odds ratio, 2.03; 95% CI 1.04-3.94). We believe that this difference may be influenced by the inclusion of all thyroid diseases in our study; we did not limit our participants to those with autoimmune thyroiditis. Concerning pathophysiology, Svensson et al. 32 reported increased production of inflammatory cytokines and abnormal lymphocyte subsets in the cord blood of offspring with maternal autoimmune thyroiditis. We speculate that these immunological changes in offspring may be related to KD onset in infancy.
Notably, however, the number of mothers at risk was relatively low in both studies; there were six in the study by Belkaibech et al. 23 and nine in ours. Under these conditions, generalizability of the study results has not been fully assured because of a lack of statistical power. Further study is required to investigate the association between maternal thyroid disease and KD onset in offspring.
The presence of siblings significantly increased the risk of KD onset in infancy. However, whether KD is induced by infectious disease remains controversial. Various infectious agents, such as group A streptococci, S. aureus, Y. pseudotuberculosis, respiratory viruses, Epstein-Barr virus, and human adenovirus have been reported as triggers for KD onset [5][6][7][8][9][10] . Infants with siblings may be at a higher risk of exposure to infectious agents compared with infants without siblings 33,34 . Therefore, we speculate that infants with siblings are at an increased risk of KD onset. Unfortunately, we did not collect data regarding the number, timing, and pathogens of infectious episodes in this study; therefore, we cannot verify whether infants with siblings are at a higher risk of infection compared with infants without siblings.
Previous large cohort studies have revealed that a higher socioeconomic status and maternal smoking in early childhood increase the risk of KD onset 20,21 . Our study showed no significant association between these factors and KD onset. We investigated maternal smoking habits during pregnancy, whereas Yorifuji et al. 20 investigated maternal smoking habits at 6 months of age. Differences in the timing of smoking made it difficult to compare the two studies. The income categories used in our study and previous studies were similar. In a previous study, only one income category (> 10 million JPY) was associated with a significantly increased risk of KD; there were no differences in the other income categories 21 . The number of people in this category was small, and significant differences should be interpreted with caution.
This study has several limitations. First, the JECS includes only 15 regions in Japan; therefore, this study might include sampling bias. However, the prevalence of KD onset in infancy is similar to that in a Japanese nationwide study (317 per 100,000 births from 2011 to 2014) 35,36 . Therefore, we believe that the influence of sampling bias in this study is minimal. Second, approximately 10,000 children dropped out of the JECS. This attrition may have had an impact on the exact outcome. Third, approximately 50% of fathers in JECS did not complete the questionnaires regarding paternal backgrounds and exposures. Fourth, we did not obtain details on the frequency or dose of supplementation during pregnancy. Therefore, we could not study the causal effect between the dose of folic www.nature.com/scientificreports/ acid supplementation and KD onset in this study. However, from the perspective of habituation, we have shown that daily intake of folic acid might reduce the risk of KD onset. Additional analyses are needed to explore the true causal relationship between folic acid supplementation and KD onset. Finally, in this study, we analyzed the risk of KD onset up to 1 year of age because the risk of KD onset after 1 year of age could not be estimated from the JECS database. In the future, we intend to conduct an analysis to estimate the time-dependent risk of KD onset in children up to 5 years of age.
In conclusion, we found that insufficient intake of folic acid during pregnancy, maternal complications of thyroid disease during pregnancy, and presence of siblings of the affected child were independent risk factors associated with KD onset in infancy. Further well-designed studies are needed to determine the causal relationship of these exposures to KD onset.

Methods
Study design, settings, and participants. In the JECS, participants were recruited across 15 regional centers from January 2011 to March 2014. The precise study design was published in 2014 24 . Briefly, pregnant women in the first trimester and their partners were asked to participate in the survey. The target sample number of the JECS was set at 100,000, which is equivalent to half the number of births in the study area. We obtained data on exposures from pregnant women and their partners using self-administered questionnaires in the first trimester; second or third trimester; and 1, 6, and 12 months after birth. The five priority outcomes of the JECS were reproduction and pregnancy complications, congenital anomalies, neuropsychiatric disorders, allergies and immune system deficiencies, and metabolic and endocrine system disorders, including KD 24 . The regular self-administered questionnaires also included items regarding the development of these outcomes. For children with KD onset, a detailed survey of KD (KD-specific questionnaire) was also sent to co-operating healthcare providers that treated participants. In this study, we excluded children who were miscarried or stillborn and those whose birth records were unavailable. We also excluded live-birth children whose parents did not respond to the questionnaire when their child was 6 months old and at 1 year of age. Finally, we excluded children whose parents stated that their child had no history of KD at 6 months but did not respond to the questionnaire when their child was 1 year of age.
The Variables. We obtained maternal information from the questionnaires at registration (the first trimester), during the second/third trimester, 1 month after birth, 6 months after birth, and 1 year after birth. We also obtained paternal information from the questionnaires at registration and information from doctors at registration (the first trimester), birth, and 1 month after birth. From each questionnaire in the first trimester, we obtained the maternal pre-pregnancy height and weight, body mass index (weight [kg]/height squared [m] 2 ), and medical history regarding KD, allergic disease, neurological/psychiatric disease, diabetes mellitus, and cancer. We also obtained the paternal medical history regarding KD and allergic disease, the method of pregnancy (spontaneous pregnancy), and the presence of siblings of the participating child. From each questionnaire in the second/third trimester, we obtained the maternal dietary intake, use of supplements, and socioeconomic status. Items related to diet included beans (g/day), green and yellow vegetables (g/day), fruits (g/day), fermented foods (cheese, yogurt, nattō, and Lactobacillus-fermented beverages) (g/day), and folic acid (µg/day) 37 . Items related to the use of supplements included folic acid, zinc, eicosapentaenoic acid, docosahexaenoic acid, and Lactobacillus-fermented beverages. We categorized zinc, eicosapentaenoic acid, docosahexaenoic acid, and Lactobacillus-fermented beverages as having been consumed at least once a month, regardless of frequency. Meanwhile, we divided folic acid into four categories on the basis of the frequency of intake: daily, at least once a week, at least once a month, and never. Items regarding socioeconomic status included the mother's and father's education history and household income. We divided education history into three categories: junior high or high school graduate, technical junior college or technical/vocational college graduate or associate degree, and bachelor's degree or beyond. Considering that the relative poverty line for a family was approximately 2 million JPY and the median household income was approximately 4 million JPY in Japan from 2010 to 2018, we divided household income into five categories: poor (< 2 million JPY), lower-middle (2 to < 4 million JPY), middle (4 to < 6 million JPY), higher-middle (6 to < 10 million JPY), and high (≥ 10 million JPY). We collected information on maternal smoking and drinking at two time points during pregnancy: in the first trimester and the second/ third trimester. We defined mothers as having a smoking/drinking habit if they answered smoking/drinking at either or both time points. After delivery, we obtained the maternal age at the child's birth, maternal complications during the pregnancy (thyroid disease, diabetes mellitus), perinatal complications (threatened abortion/ premature labor, gestational diabetes mellitus, gestational hypertension, premature membrane rupture), mode of delivery (cesarean section), preterm birth (< 36 weeks), sex of the child (male), and weight of the child at birth.