Kawasaki disease (KD) is an acute self-limited systemic vasculitis of childhood of unknown aetiology that presents predominantly in infants and children under 5 years old. Its diagnosis is based on clinical criteria. Its prognosis depends on the extent of cardiac involvement. Without treatment, up to 20%–25% of patients develop coronary aneurysms.1

The aetiology of KD is unknown despite years of investigation. Clinical, laboratory and epidemiological features suggest an infectious origin or trigger. To date, many studies have failed to identify an infectious aetiological agent or demonstrate an association with exposure to drugs or a response to super-antigen. Besides, there is evidence of an important genetic predisposition for KD. Some genome-wide association studies (GWAS) have identified some biologically plausible loci involved in inflammation, immune responses and cardiovascular status.2–4 Due to the clustering of cases in space and/or time, among other patterns, a reasonable hypothesis is that KD may be caused by a trigger (whether it is an infectious agent is still under debate) that causes the disease or elicits a response in genetically predisposed individuals, particularly those of Asian descent.1 However, little is known about the impact of the human microbiome in the development of KD.

The term microbiome refers to the community of commensal, symbiotic and pathogenic microorganisms that share human body space.5 Several studies have linked the human microbiome with some diseases or the predisposition to develop them. Some studies have aimed at establishing an association between the dysregulation of the gut microbiome and multiple autoimmune diseases. Animal models suggest that the gut microbiome contains organisms that direct both pro-inflammatory and anti-inflammatory immune responses depending on the immunologic context, genetic makeup or sex of the host and the overall microbial community structure.6

The role of human microbiome in rheumatology has been studied in diseases such as rheumatoid arthritis (RA) or systemic lupus erythematosus (SLE). In the case of RA, evidence has been found of an association between the increased presence of Porphyromonas gingivalis, an oral anaerobe involved in the development of periodontitis, and positive anticitrullinated peptide antibodies.7 Gut microbial perturbations are present in RA too, with substantial expansion of Prevotella copri in untreated patients with new-onset RA.8 Systemic lupus erythematosus is also characterized by microbial perturbations. One of the earliest studies found decreased gut bacterial diversity and a lower Firmicutes to Bacteroidetes ratio in patients with SLE.9 As occurs in RA, there is evidence of translocation of intestinal bacteria in SLE, with detection of Enterococcus gallinarum in liver biopsies of patients with active SLE.10

Although the involvement of dysregulation of the intestinal microbiome in the pathogenesis of autoimmune diseases has been widely studied, the role of the respiratory microbiome in autoimmunity is not as well established. Although no infectious causative agent has been identified, some epidemiological studies have documented infections by various microorganisms in many KD patients. Recently, KD-like cases have been described in association with SARS-CoV-2 infection.11,12 Besides, the fact that there have been KD epidemics in Japan, its higher incidence in spring and winter and its occurrence predominantly in children aged 6 months to 5 years, support the hypothesis of an infectious aetiology or trigger.1

To our knowledge, studies on the role of the microbiome in KD are limited. Some studies sought to establish a link between the intestinal and pharyngeal flora with KD and found no significant differences between cases and controls.13,14 In a study on the intestinal microbiome in KD published in 2005, the faecal microbiome of 28 KD patients was analysed, evincing an increase of Streptoccocus spp. in the acute phase of the disease and Ruminococcus bacteria in the subacute phase.15 Recent advances allowed the discovery that Th17 and Treg cell differentiation is regulated by short chain fatty acids (SCFAs), in particular butyrate, produced by the gut microbiota. This finding provided a mechanistic link between dysbiosis, defined as changes in the composition of the gut microbiota, and various inflammatory diseases. Recent studies have tried to support this hypothesis, finding that faecal concentrations of butyrate were significantly reduced in KD patients, providing important information regarding the association between dysbiosis and dysregulated immune responses in KD.15–17

The contribution of viruses to the pathogenesis of KD has been suggested by ultrastructural studies that found cytoplasmic inclusion bodies containing RNA of viral origin in the bronchial epithelia of deceased KD patients.18–20 Other studies have found that one or more respiratory viruses were detected in nasopharyngeal aspirates of approximately half of KD patients.21

Some works have reported a possible role of fungal infections in the development of KD. Recent research22,23 has suggested that the agent causing KD may be an environmental agent carried through tropospheric winds. The authors found aerosolized toxins from several Candida species in specimens collected in the troposphere at times of high KD incidence. In Japan, the burden of cases often appears to be associated with wind currents originating in densely cultivated cereal croplands of north-eastern China.24,25 In addition, Sato et al. observed that Candida albicans-derived substances, such as C. albicans water-soluble fraction, induced coronary arteritis in mice.26

Most of the data shows that the composition of the gut microbiome in KD patients differs from that in healthy subjects.16 A plausible hypothesis for the pathogenesis of KD would be that, in genetically predisposed individuals with a favourable microbiome environment, an external trigger could facilitate the development of the disease. As far as we know, there are no studies on the characterization of the microbiome found in respiratory exudates in KD.

The main objective of this study was to compare differences in the nasopharyngeal bacterial microbiota and respiratory viruses between KD patients and healthy controls as a first step to delve deeper into the role of the microbiome and previous infections in the pathogenesis of KD.

We conducted a case-control study including children (cases) diagnosed with KD treated in a tertiary care hospital (n = 11) matched with healthy children (controls) managed in the same hospital and who underwent routine laboratory tests previous to minor surgery or general paediatrics follow-up visits (n = 11). One of the cases was eventually excluded due to poor sample quality. The final study was performed in 10 cases and 10 controls. Samples were collected between February 2016 and May 2018. Cases and controls were matched for age, sex and seasonality. Additional information on the methods can be found in the Supplementary material.

All patients were enrolled in the 5 days that followed the diagnosis, and the respiratory samples were collected before initiation of treatment. They all meet the diagnostic criteria for KD established by the American Heart Association.1 We excluded patients that had received antibiotics before sample collection or without a completed informed consent form. Controls were healthy children without major comorbidities and who had not received antibiotic treatment in the 4 weeks preceding sample collection. The data sets presented in this article can be found in online repositories. The names of the repository/repositories and accession number(s) are detailed in the Figshare repository within the “Kawasaki” project: https://figshare.com/s/181b728556419d0de4c5.

To our knowledge, this is the first study that has found differences in the composition of the respiratory microbiome between KD patients and healthy controls. In 2004, Horita et al. obtained throat swab samples of 21 patients with KD and 20 controls with non-KD febrile illness. Sixty-one bacterial strains were isolated from KD patients, and 62 strains from control patients. The authors did not find any differences between KD patients and febrile controls in the bacterial species present in their throat flora.13

In our study, we obtained nasopharyngeal aspirate samples of 10 patients with KD and 10 healthy controls and studied the microbiome using NGS. We found a significant difference in the abundance of Corynebacterium, which was overrepresented in the KD group (12.25% vs 1.59%; P = 0.045). Moreover, the proportion of other genera with abundances of less than 0.5% was also higher in the case group compared to the control group (16.07% vs 7.82%; P = 0.08). A previous study that tried to isolate the causative bacterial gene from peripheral blood leukocytes of patients with KD showed a new species of the genus Corynebacterium.27Corynebacterium spp. were also present in aerosol samples collected at ground level in a location very close to Tokyo at a time of high KD incidence. Interestingly, an important number of other taxa previously associated to KD in different studies have also been identified in the metagenomic analysis of airborne samples. This might indicate that potential exposure to air with different loads of bacteria might have a role in the pathogenesis of KD, a subject that deserves further investigation.24,25,28

The genus Corynebacterium is a taxon of coryneform gram-positive bacilli or coccobacilli.29,30 Nearly all are catalase-positive, and they express a wide range of pigments and metabolic processes. Bacteria in this genus are ubiquitous and commonly present in the skin and mucosa of human beings, and most of them are saprophytes. Corynebacterium spp. have been identified in the middle ear of children with otitis media, can cause opportunistic infections in some cases, and a few species in the group, such as C. diphtheriae, are pathogenic for humans.31

Other studies of the gut microbiome have found that the beneficial Lactobacillus genus may be absent or lost from enteric flora during the acute phase of KD14 and that there seems to be a relative increase in Streptococcus spp. in the acute phase of the disease and in Ruminococcus spp. in the subacute phase.15 In our study, we found no differences in any of these 3 genera in the specimens, which could be related to the small sample size or to the fact that we studied respiratory samples rather than faecal samples.

The contribution of viruses to KD has been suggested by studies that found cytoplasmic inclusion bodies containing viral RNA in the bronchial epithelia of deceased KD patients.19 Although several studies have attempted to identify the origin of this viral RNA, no known viruses have been identified, and Rowley et al. concluded that it could be a previously unidentified, ubiquitous RNA virus that can produce persistent infection in the bronchial epithelium, thus contributing to the aetiology of KD.18,20 In another study, Turnier et al. found that nearly half (n = 93, 41.9%) of the 192 patients with KD under study had a positive respiratory viral PCR, mainly rhinovirus or enterovirus.21 Similarly, our study did not found differences between cases and controls, although rhinovirus and adenovirus, among others, where found in nearly half of the samples in both groups.

While there are obvious limitations to our study, mainly the small sample and the lack of comparison with not only healthy controls but also children with respiratory diseases, the results are intriguing if a unique aetiology for this disease is contemplated.

It is possible, though, that KD is a multifaceted disease, and that exposition to a diversity of microbial agents or a dominance of microbial agents with structural similarities might elicit the same idiosyncratic immune response. There is reasonable proof that an imbalance in the microbiome has a role in the pathogenesis of KD. Our study introduces the genus Corynebacterium as a new player in the game and demonstrates the need to conduct larger epidemiological studies focused on uncovering changes in the microbiome of patients with KD.

The current article does not assure causality but establishes the bases for conducting studies with larger samples that help us unravel the role of the Corynebacterium genus in the pathogenesis of KD. Our group is already working on collecting more samples from patients with KD, healthy controls and patients with respiratory infections to try to draw more established conclusions.

To our knowledge, this is the first study that has found differences in the composition of the respiratory microbiome between KD patients and healthy controls. The genus Corynebacterium was significantly more abundant in KD patients compared to healthy controls. These results support the hypothesis that there could be a favourable respiratory microbiome substrate for the development of the disease in genetically predisposed patients. Studies in larger samples are necessary to assess its significance and possible relationship with KD.

This work was supported by the National R+D+I Plan, Instituto de Salud Carlos III (project PI20/00517) and cofunded by European Regional Development Fund (ERDF). It was also partially funded by a 2018 Sociedad Española de Reumatología Pediatrica (SERPE) scholarship and a 2016 grant from the Fundació Daniel Bravo Andreu to the WINDBIOME project. These funding bodies had no role in study design, data collection and analysis, the decision to publish or the preparation of the manuscript.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could constitute a conflict of interest.