# INFECTION-RELATED IMMUNE-MEDIATED DISEASES AND MICROBIOTA

EDITED BY : Kyung-Yil Lee, Hiromichi Hamada and Miika Kaleva Arvonen PUBLISHED IN : Frontiers in Pediatrics

#### Frontiers eBook Copyright Statement

The copyright in the text of individual articles in this eBook is the property of their respective authors or their respective institutions or funders. The copyright in graphics and images within each article may be subject to copyright of other parties. In both cases this is subject to a license granted to Frontiers. The compilation of articles constituting this eBook is the property of Frontiers.

Each article within this eBook, and the eBook itself, are published under the most recent version of the Creative Commons CC-BY licence. The version current at the date of publication of this eBook is CC-BY 4.0. If the CC-BY licence is updated, the licence granted by Frontiers is automatically updated to the new version.

When exercising any right under the CC-BY licence, Frontiers must be attributed as the original publisher of the article or eBook, as applicable.

Authors have the responsibility of ensuring that any graphics or other materials which are the property of others may be included in the CC-BY licence, but this should be checked before relying on the CC-BY licence to reproduce those materials. Any copyright notices relating to those materials must be complied with.

Copyright and source acknowledgement notices may not be removed and must be displayed in any copy, derivative work or partial copy which includes the elements in question.

All copyright, and all rights therein, are protected by national and international copyright laws. The above represents a summary only. For further information please read Frontiers' Conditions for Website Use and Copyright Statement, and the applicable CC-BY licence.

ISSN 1664-8714 ISBN 978-2-88963-744-7 DOI 10.3389/978-2-88963-744-7

#### About Frontiers

Frontiers is more than just an open-access publisher of scholarly articles: it is a pioneering approach to the world of academia, radically improving the way scholarly research is managed. The grand vision of Frontiers is a world where all people have an equal opportunity to seek, share and generate knowledge. Frontiers provides immediate and permanent online open access to all its publications, but this alone is not enough to realize our grand goals.

#### Frontiers Journal Series

The Frontiers Journal Series is a multi-tier and interdisciplinary set of open-access, online journals, promising a paradigm shift from the current review, selection and dissemination processes in academic publishing. All Frontiers journals are driven by researchers for researchers; therefore, they constitute a service to the scholarly community. At the same time, the Frontiers Journal Series operates on a revolutionary invention, the tiered publishing system, initially addressing specific communities of scholars, and gradually climbing up to broader public understanding, thus serving the interests of the lay society, too.

#### Dedication to Quality

Each Frontiers article is a landmark of the highest quality, thanks to genuinely collaborative interactions between authors and review editors, who include some of the world's best academicians. Research must be certified by peers before entering a stream of knowledge that may eventually reach the public - and shape society; therefore, Frontiers only applies the most rigorous and unbiased reviews.

Frontiers revolutionizes research publishing by freely delivering the most outstanding research, evaluated with no bias from both the academic and social point of view. By applying the most advanced information technologies, Frontiers is catapulting scholarly publishing into a new generation.

#### What are Frontiers Research Topics?

Frontiers Research Topics are very popular trademarks of the Frontiers Journals Series: they are collections of at least ten articles, all centered on a particular subject. With their unique mix of varied contributions from Original Research to Review Articles, Frontiers Research Topics unify the most influential researchers, the latest key findings and historical advances in a hot research area! Find out more on how to host your own Frontiers Research Topic or contribute to one as an author by contacting the Frontiers Editorial Office: researchtopics@frontiersin.org

# INFECTION-RELATED IMMUNE-MEDIATED DISEASES AND MICROBIOTA

Topic Editors:

Kyung-Yil Lee, The Catholic University of Korea, South Korea Hiromichi Hamada, Tokyo Women's Medical University Yachiyo Medical Center, Japan

Miika Kaleva Arvonen, Department of Pediatrics, Kuopio University Hospital, Finland

Citation: Lee, K.-Y., Hamada, H., Arvonen, M. K., eds. (2020). Infection-Related Immune-Mediated Diseases and Microbiota. Lausanne: Frontiers Media SA. doi: 10.3389/978-2-88963-744-7

# Table of Contents


Susanna Esposito, Ilaria Polinori and Donato Rigante

*34 Outcomes of Kawasaki Disease Children With Spontaneous Defervescence Within 10 Days*

Ya-Chiao Hu, Hsin-Min Liu, Ming-Tai Lin, Chun-An Chen, Shuenn-Nan Chiu, Chun-Wei Lu, Luan-Yin Chang, Jou-Kou Wang and Mei-Hwan Wu


Himanshi Chaudhary, Johnson Nameirakpam, Rajni Kumrah, Vignesh Pandiarajan, Deepti Suri, Amit Rawat and Surjit Singh


Manuela Mosconi, Valentina Rossi, Stefano Avanzini, Maria G. Faticato, Isabella Ceccherini, Francesca Lantieri, Girolamo Mattioli, Denise Larson, William Pavan, Carlotta De Filippo, Monica Di Paola, Domenico Mavilio and Duccio Cavalieri

# Editorial: Infection-Related Immune-Mediated Diseases and Microbiota

#### Kyung-Yil Lee<sup>1</sup> \*, Hiromichi Hamada<sup>2</sup> and Miika Arvonen<sup>3</sup>

*<sup>1</sup> The Catholic University of Korea, College of Medicine, Seoul, South Korea, <sup>2</sup> Department of Pediatrics, Tokyo Women's Medical University Yachiyo Medical Center, Chiba, Japan, <sup>3</sup> Department of Pediatrics, Kuopio University Hospital, Kuopio, Finland*

Keywords: microbiota, Kawasaki disease, juvenile idiopathic arthritis, Hirschsprung's disease, etiology, epidemiology, pathogenesis, multiple-drug resistant Staphylococcus aureus

**Editorial on the Research Topic**

#### **Infection-Related Immune-Mediated Diseases and Microbiota**

Microbiota may form a collaborative partnership with the host. Eubiosis, i.e., the homeostatic state of the host and symbiotic microflora, is critical for maintaining a state of well-being in the host. Commensal bacteria may prevent by colonization by external pathogens, and the mucosal immune system, such as gut-associated lymphoid tissues (GALT), is established after colonization of the gut by normal flora, suggesting that microbiota and host immune systems are closely related to each other (1). On the other hand, certain strains of microbes can elicit infectious events on occasion when they invade into the host and induce an immune reaction.

#### Edited by:

*Marzia Duse, Sapienza University of Rome, Italy*

> Reviewed by: *Riccardo Castagnoli, University of Pavia, Italy*

#### \*Correspondence:

*Kyung-Yil Lee leekyungyil@catholic.ac.kr*

#### Specialty section:

*This article was submitted to Pediatric Immunology, a section of the journal Frontiers in Pediatrics*

Received: *04 October 2019* Accepted: *02 March 2020* Published: *07 April 2020*

#### Citation:

*Lee K-Y, Hamada H and Arvonen M (2020) Editorial: Infection-Related Immune-Mediated Diseases and Microbiota. Front. Pediatr. 8:108. doi: 10.3389/fped.2020.00108*

The microbiota in an individual continuously changes after birth (2), and certain strains of microbes in ethnic groups may be influenced by environmental factors such as diet and socio-economic state. Thus, the microbiota is different in different populations and can be changed by the changing environment. The disruption of the reciprocal equilibrium between microbiota and host, that is, dysbiosis, has now become a major subject of study in various medical fields, including those of metabolic, gastroenteric, psychiatric, neurologic, and allergic diseases and cancers. Also, intestinal dysbiosis has been reported in patients with Kawasaki disease (KD) and juvenile idiopathic arthritis (JIA) (3, 4). Although the mechanisms relating dysbiosis and provocation of diseases remain elusive, over-production of toxic materials from dysbiosis-causing microbials, vulnerable invasion of microbials through wakened mucosal barriers, or the disruption of the homeostatic relationship between the microbiota and the host's immune system may be responsible for disease onset [(5), Esposito et al.].

The human gut microbiota is composed of over 1,000 species, and different strains are colonized in the small intestine and colon. Hirschsprung's disease (HD) is characterized by a defect of intestinal nerve ganglia, and occasionally patients with HD can be affected with HD-associated enterocolitis before or after operation. In the majority of cases, the causative agents of HD-associated enterocolitis are not external pathogens, and dysbiosis may be associated with intestinal inflammation (6). A different microbiome between patients with total colon resection and those with partial resection is observed, and the former group tends to have a higher risk of enterocolitis (Pini Prato et al.). Staphylococcus species are one of the main normal flora that colonize the skin and, on occasion, the vagina tract of pregnant women. Normal flora in neonates begin to colonize just after birth, though some strains may be different according to delivery method (Caesarian section or vaginal delivery). Although colonization by commensals, including multiple-drug-resistant S. aureus, may be a risk factor for subsequent infection, severe invasive infection is

**4**

rare in immune-competent hosts. It is observed that colonized multiple-drug-resistant S. aureus in neonates could be transmitted from their mothers (Lin and Yao).

Childhood immune-mediated diseases, including KD and JIA, may be associated with prevalent infections in childhood. Interestingly, the incidence of both diseases is quite different among populations; the incidence of KD is over 10–20 times higher in East Asian countries such as Japan and Korea, and the incidence of JIA is over 10 times higher in North European countries compared to children in East Asian countries (7). The discrepancy of incidence rates across the populations has been observed in other infection-related immune-mediated diseases, including type I diabetes, inflammatory bowel disease, and Behcet disease. Although genetic or environmental factors may be responsible for the finding, it is possible that children living in higherprevalence countries may have more chances of being exposed to KD or JIA pathogens, since the clinical manifestations and immune function of children with KD or JIA are nearly identical across the populations.

Based on the epidemiological and clinical characteristics of KD, it has been suggested that it may be associated with infectious agents, including viruses, especially RNA viruses (Rowley and Schulman), bacteria that can activate innate and adaptive immune systems (Nagata; Nakamura et al.), and strains of normal flora (Esposito et al.; Rhim et al.). Since KD is a self-limiting systemic inflammation, many laboratory parameters are affected during the natural course of the disease; levels of white blood cells, erythrocyte sedimentation rate, C-reactive protein, albumin, hemoglobin, and other biomarkers such as proinflammatory cytokines are up-regulated or down-regulated, and various genetic traits are related to KD susceptibility or phenotype (Chaudhary et al.). Intravenous immunoglobulin (IVIG) treatment is known to reduce the risk of coronary artery lesions (CALs), a major complication of KD. It was reported that a patient group with spontaneous defervescence had a higher rate of CALs (aneurysms) at 1 month after disease onset compared to the IVIG-treated group (Hu et al.). However, a small proportion of initially CAL-negative patients in both groups showed CALs at 1 month, suggesting that some

#### REFERENCES


KD patients may have ongoing inflammation in CALs after defervescence (8).

There are few human diseases of which the pathogenesis has been clearly proven from the era of Hippocrates to the present time. Although etiologic agents have been discovered in infectious diseases, the substances inducing inflammation and tissue cell injury in infectious diseases and infection-related immune-mediated diseases are not pathogens themselves but smaller substances derived from the infectious insults (9). Many researchers may agree on the notions that every disease has etiologic substances and that present immunological concepts have limitations in explaining the pathogenesis of many diseases. It is now known that the host immune system reacts not to only the substances derived from the infectious agents, including toxins and pathogen-associated molecular patterns (PAMPs), but also to the substances derived from host cells injured by infectious insults such as damage (danger)-associated molecular patterns (DAMPs), especially in cases of intracellular infection such as viral or intracellular pathogen infections (10). Because of the appearance of KD as a novel disease in East Asia, the discovery of the etiology and pathogenesis of KD will help to extend our understanding regarding these issues of human diseases. The pathogenesis of KD is associated with the immune reaction of the host against infectious insults. The contributions to this Research Topic present and discuss various aspects of the pathogenesis of KD, including the roles of components of the adaptive immune system such as IgA plasma cells and cytotoxic CD8 T cells against viral antigens (Rowley and Schulman), a similar/common immune process associated with the activation of T cells and innate immune cells caused by diverse pathogens (Nagata), an innate immune response including PAMPs, tolllike receptors, and complement pathways (Nakamura et al.), and types of DAMPs produced after infectious insults based on the protein-homeostasis-system hypothesis (Rhim et al.).

## AUTHOR CONTRIBUTIONS

K-YL wrote the manuscript. HH and MA have made a substantial, direct, and intellectual contribution to the work. All authors approved it for publication.

disease patients by metagenomic analysis. Front Microbiol. (2015) 6:824. doi: 10.3389/fmicb.2015.00824


**Conflict of Interest:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2020 Lee, Hamada and Arvonen. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Maternal-Infant Correlation of Multidrug-Resistant *Staphylococcus aureus* Carriage: A Prospective Cohort Study

Jialing Lin and Zhenjiang Yao\*

*Department of Epidemiology and Health Statistics, School of Public Health, Guangdong Pharmaceutical University, Guangzhou, China*

Objectives: We aim to assess the correlation of multidrug-resistant *Staphylococcus aureus* (MDR *S. aureus*) carriage between mothers and their newborn infants.

#### *Edited by:*

*Hiromichi Hamada, Tokyo Women's Medical University Yachiyo Medical Center, Japan*

#### *Reviewed by:*

*Marine Butin, Centre Hospitalier Universitaire de Lyon, France Guillermo Soza, Universidad de La Frontera, Chile Hiroyuki Shiro, Yokohama Rosai Hospital, Japan*

> *\*Correspondence: Zhenjiang Yao zhjyao2001@yahoo.com*

#### *Specialty section:*

*This article was submitted to Pediatric Infectious Diseases, a section of the journal Frontiers in Pediatrics*

*Received: 21 September 2018 Accepted: 20 November 2018 Published: 05 December 2018*

#### *Citation:*

*Lin J and Yao Z (2018) Maternal-Infant Correlation of Multidrug-Resistant Staphylococcus aureus Carriage: A Prospective Cohort Study. Front. Pediatr. 6:384. doi: 10.3389/fped.2018.00384* Materials and Methods: We conducted a prospective cohort study of mothers and their newborn infants in two hospitals in Shenzhen, China, from August to November 2015. We collected demographic and clinical information from mothers and newborn infants by face-to-face questionnaires and medical datasets. Serial swabs were collected from mothers and their newborn infants for further experiments. Maternal-infant correlation was assessed using the Poisson regression model.

Results: The prevalence of MDR *S. aureus* vaginal carriage in mothers was 4.7% (86/1834). The incidence of MDR *S. aureus* carriage in newborn infants was 1.3% (23/1834). The adjusted relative risk and 95% confidence interval of maternal-infant MDR *S. aureus* carriage was 7.63 (2.99–19.49). Six MDR *S. aureus* maternal-infant pairs were concordant. The phenotypic and molecular characteristics of MDR *S. aureus* isolates were similar between mothers and their newborn infants.

Conclusion: MDR *S. aureus* vaginal carriage in mothers was associated with an increased risk for MDR *S. aureus* carriage in their newborn infants.

Keywords: *Staphylococcus aureus*, multidrug resistant, mothers, infants, cohort

# INTRODUCTION

Staphylococcus aureus (S. aureus) is one of the most important human bacterial pathogens that cause nosocomial and community infections, and results in substantial morbidity and mortality (1). Multidrug-resistant S. aureus (MDR S. aureus) is of greater significance because of its more severe clinical outcomes (2).

Those persons who are colonized with MDR S. aureus have a higher risk for subsequent infections than non-colonized individuals (3, 4). Most infections are caused by the same MDR S. aureus strain that previously colonized the person (5). In response to these findings, there has been increasing attention to the detection of MDR S. aureus carriage, with subsequent decolonization of carriers as a potential method for the prevention of MDR S. aureus infection.

Most of the available data on the detection of MDR S. aureus carriage have been in hospitalized non-pregnant adults (6, 7). Despite the fact that MDR S. aureus outbreaks in infants have been

**7**

linked to the infected or colonized mother, there are limited data on MDR S. aureus carriage rates among mothers or on the risk of transmission of MDR S. aureus from pregnant MDR S. aureus carriers to their newborn infants (8, 9). The maternalinfant relatedness of MDR S. aureus carriage is of particular interest.

Accordingly, we determined whether vaginal carriage of MDR S. aureus in the mothers is independently correlated with MDR S. aureus carriage in their newborn infants. Antimicrobial susceptibility, virulence factors, and clonality of isolates were also evaluated. The hypothesis of this study is that there is maternal-infant vertical MDR S. aureus carriage.

# MATERIALS AND METHODS

This study was approved by the Ethics Committee of Guangdong Pharmaceutical University and was performed in accordance with the approved guidelines. Written informed consent was obtained from the mothers and their newborn infants involved in the study before enrollment.

This prospective cohort study was conducted in two hospitals in Shenzhen, China, between August and November 2015. The two hospitals, Longhua Central Hospital and Guanlan People's Hospital, are large hospitals whose obstetric services deliver between 10,000 and 15,000 neonates per year; these are two of the largest delivery hospitals in China. The target populations were mothers and their newborn infants. Chinese mothers with gestation between 35 and 40 weeks were voluntarily included. Mothers with twin or multiple gestations, cesarean section, or acute diseases were excluded. Newborn infants without sample collection were also excluded. The sample size was calculated using the power two proportions method with 4.00% proportion (approximate prevalence of S. aureus in the newborns), 4.00 ratio, 0.05 alpha, and 0.80 power. Taking a 10% loss of follow-up into consideration, the estimated minimum sample size of this study was 107 for each group.

A face-to-face questionnaire was administered by trained personnel that aimed to collect demographic information and information regarding potential factors influencing MDR S. aureus carriage during pregnancy. Medical records of the mothers and their newborn infants were reviewed by two members of the study team who were blinded to the maternal and infant MDR S. aureus carriage status.

Sterile swabs moistened with sterile saline water were used to get specimens from the vagina of the enrolled mothers prior to delivery by trained personnel. Infant specimens for MDR S. aureus were sampled from the nasal cavity, ear canal, oral cavity, and umbilicus by trained nurses immediately after birth. All specimens were then inoculated into enrichment broth tubes containing 1% tryptone, 7.5% sodium chloride, 1% mannitol, and 0.25% yeast extract.

All included mothers were informed of the research question and outcome measures by well-trained staff in person before participating. The trained personnel informed the mothers of the results of S. aureus carriage by telephone calls. Chinese mothers with gestation between 35 and 40 weeks were voluntarily included in this study. Included mothers were asked to complete face-to-face questionnaires and collected vaginal samples by trained personnel.

Swabs were inoculated into mannitol salt agar plates at 37 ± 1 ◦C for 24 h of incubation. Isolates were confirmed to be S. aureus by a combination of gram staining, catalase reaction, hemolysis test, DNase test, and coagulase tests.

Antimicrobial susceptibility testing was performed by the Kirby-Bauer disk diffusion method according to the Clinical and Laboratory Standards Institute guidelines of 2015. The following antimicrobial disks were used: cefoxitin, clindamycin, rifampicin, moxifloxacin, tobramycin, gentamicin, sulfamethoxazole-trimethoprim, linezolid, and erythromycin. Isolates were defined as resistant if they were resistant or intermediate to the antimicrobial disk. S. aureus isolates were classified as MDR if they were resistant to no less than three antibiotic classes (10).

Multilocus sequence typing (MLST) was used to type MDR S. aureus, which involved sequencing the seven housekeeping genes (11). Then, the sequence types (STs) and allelic profiles were confirmed by querying the MLST database (http://eburst. mlst.net). STs were clustered into a clonal complex (CC) by using the eBURST software program (Department of Infectious Disease Epidemiology, Imperial College London, London, UK; http://eburst.mlst.net). The presence of virulence genes including Panton-valentine leukocidin (Pvl), toxic shock syndrome toxin (Tst), exfoliative toxin A (Eta), and exfoliative toxin B (Etb) using polymerase chain reaction assays as in previous studies was determined as previously described (12, 13).

All data were entered in duplicate into the EpiData version 3.0 database (The EpiData Association, Odense Denmark). Missing data were excluded. Categorical variables were compared by Pearson's chi-squared test or the Fisher exact test when appropriate. To identify variables that might confound the correlation of MDR S. aureus carriage between mothers and newborn infants, influencing factors with a P-value < 0.2 were identified as adjustment variables using the Poisson regression model to estimate the correlations of isolates between mothers and newborn infants. Relative risks (RRs) and 95%confidence intervals (CIs) were used to assess the maternal-infant relatedness of MDR S. aureus carriage. The relationship between populations and main STs of MDR S. aureus isolates was illustrated by a minimum spanning tree (PHYLOVIZ software version 2.0; http://www.phyloviz.net). A two-sided P-value < 0.05 was considered as being of statistical significance. All analyses were performed using Stata version 14.2 (College Station, Texas, USA).

# RESULTS

Overall, 1968 mothers were preliminarily enrolled, but only 1,834 mothers and their newborn infants met the inclusion criteria and consented to participate in this study. We found that there were 133 S. aureus isolates and the prevalence of MDR S. aureus vaginal carriage in mothers was 4.7% (86/1834). There was no statistical difference of MDR S. aureus carriage in mothers from the two different hospitals (χ <sup>2</sup> = 0.029, P-value = 0.865).

There were 60 S. aureus isolates and the incidence of MDR S. aureus carriage in newborn infants was 1.3% (23/1834). There was no statistical difference of MDR S. aureus carriage in newborn infants from the two different hospitals (χ <sup>2</sup> = 0.846, P-value = 0.358).

MDR S. aureus vaginal carriage in mothers (χ <sup>2</sup> = 69.163, Pvalue < 0.001) was significantly correlated with MDR S. aureus carriage in their newborn infants. However, there were no other significant variables correlated with MDR S. aureus carriage in infants. More details can be found in **Table 1**.

After adjusting for the age of mothers, education level of mothers, average monthly income, and gender of newborn infants, MDR S. aureus vaginal carriage in mothers was still a risk factor for MDR S. aureus carriage in newborn infants [adjusted RR (aRR) = 7.63; 95%CI, 2.99–19.49; P-value < 0.001]. More details can be found in **Table 2**.

The most predominant proportion of antibiotic resistance in the 86 MDR S. aureus isolates from mothers was penicillin (97.7%), followed by erythromycin (73.3%), clindamycin (64.0%), tobramycin (38.4%), cefoxitin (26.7%), trimethoprimsulfamethoxazole (20.9%), moxifloxacin (19.8%), gentamicin (12.8%), linezolid (5.8%), and rifampicin (2.3%). The most predominant proportion of antibiotic resistance in 23 MDR S. aureus isolates from the newborn infants was penicillin (95.7%), followed by erythromycin (91.3%), clindamycin (69.6%), cefoxitin (47.8%), trimethoprim-sulfamethoxazole (30.4%), gentamicin (21.7%), tobramycin (21.7%), rifampicin (21.7%), moxifloxacin (13.0%), and linezolid (8.7%). The proportion of resistant rifampicin (P-value = 0.004) in MDR S. aureus isolates was significantly different between mothers and their newborn infants. More details can be found in **Table 3**.

The results indicated that the proportions of virulence genes of isolates in both mothers and their newborns were low. The

TABLE 2 | Correlations of multidrug-resistant *Staphylococcus aureus* carriage between mothers and their newborn infants in Shenzhen, 2015.


*MDR S. aureus, multidrug-resistant Staphylococcus aureus; RR, relative risk; aRR, adjusted relative risk; CI, confidence interval.*

*a It was adjusted for age of mothers, education level of mothers, average monthly income, and gender of newborn infants.*

TABLE 1 | Characteristics of multidrug-resistant *Staphylococcus aureus* carriage among newborn infants in Shenzhen, 2015 [n (%)].


*MDR S. aureus, multidrug-resistant Staphylococcus aureus; n, number of isolates; NA, not applicable.*

*<sup>a</sup> The P-values were calculated with the Fisher's exact test.*

*<sup>b</sup> No estimate of the P-value is provided owing to the lack of occurrence of the outcome of interest in at least one group.*

TABLE 3 | Phenotypic and genetic characteristics of multidrug-resistant *Staphylococcus aureus* isolates between mothers and their newborn infants in Shenzhen, 2015 [n (%)].


*MDR S. aureus, multidrug-resistant Staphylococcus aureus; n, number of isolates; Pvl, panton-valentine leukocidin; Tst, toxic shock syndrome toxin; Eta, exfoliative toxin A; Etb, exfoliative toxin B; NA, not applicable.*

*<sup>a</sup> The P-values were calculated with the Fisher's exact test.*

*<sup>b</sup> No estimate of the P-value is provided owing to the lack of occurrence of the outcome of interest in at least one group.*

proportion of the positive Pvl gene in MDR S. aureus isolates was significantly different between mothers and newborn infants (Fisher's exact test, P-value = 0.041). There were no significant differences of virulence genes in MDR S. aureus isolates between mothers and their newborn infants. More details can be found in **Table 3**.

The most common CC of 86 MDR S. aureus isolates in mothers was CC5 (51.2%), followed by CC7 (19.8%), CC59 (14.0%), CC88 (5.8%), CC398 (3.5%), CC20 (2.3%), CC45 (2.3%), and CC121 (1.2%). The most common CC of 23 MDR S. aureus isolates in newborn infants was CC5 (38.1%), followed by CC59 (28.6%), CC45 (9.5%), CC88 (9.5%), CC7 (4.8%), CC22 (4.8%), and CC398 (4.8%).

The minimum spanning tree demonstrated a good concordance of certain STs between mothers and their newborn infants; for example, MDR S. aureus isolates belonging to ST1, ST6, ST7, ST59, ST88, ST398, and ST965 were found in both mothers and newborn infants. More details can be found in **Figure 1**.

Overall, there were six MDR S. aureus maternalinfant pairs. These six MDR S. aureus maternal-infant pairs were concordant, with the same phenotypic and molecular characteristics. More details can be found in **Table 4**.

## DISCUSSION

Methicillin-Resistant Staphylococcus aureus (MRSA) carriage between mothers and their infants has been reported, but no

isolates in Shenzhen, 2015. The numbers in the circles are the sequence types. The size of the circles represents the number of isolates. The distance of lines between each circle represents the degree of the relationship.

previous studies reported the proportions of MDR S. aureus in both mothers and infants. The prevalence of MRSA in MDR S. aureus among mothers and their newborns was 26.7 and 47.8%,


TABLE 4 | Characteristics of six maternal-infant pairs with multidrug-resistant *Staphylococcus aureus* isolates in Shenzhen, 2015.

*MDR S. aureus, multidrug-resistant Staphylococcus aureus; CC, clonal complex; ST, sequence type; FOX, cefoxitin; EM, erythromycin; PCN, penicillin; GM, gentamicin; CM, clindamycin; RFP, rifampicin; LZD, linezolid; MXF, moxifloxacin; TOB, tobramycin; TMP/SMX, trimethoprim-sulfamethoxazole; MDR, multidrug-resistance, resistant to no less than three antibiotic classes; Pvl, panton-valentine leukocidin; Tst, toxic shock syndrome toxin; Eta, exfoliative toxin A; Etb, exfoliative toxin B;* +*, positive;* −*, negative.*

respectively. On the contrary, the prevalence of MDR S. aureus in MRSA among mothers and their newborns were 88.5 and 91.7%, respectively. These results were similar to current observed studies that MRSA is always MDR S. aureus, but MDR S. aureus is not necessarily MRSA (7, 11). Therefore, it is significant to assess the mother-infant relationship of MDR S. aureus. An American study reported that the proportion of MDR S. aureus carriage in community residents was 6.9% (14). Other previous studies have reported that the prevalence of S. aureus vaginal carriage in mothers (pregnant women) and infants ranged from 0.96–12.6% (15–18) to 5.4–17.7% (15, 16, 18–21), respectively. Therefore, the results suggested that MDR S. aureus carriage in our target populations were moderate when compared with other countries and regions.

This prospective cohort study identified risk factors for MDR S. aureus carriage in newborn infants and assessed the risk of MDR S. aureus vaginal carriage of mothers for MDR S. aureus carriage in their newborn infants. The gender of the newborn infants was not associated with MDR S. aureus isolates in this study, which was different to previous studies (22, 23). The mechanism between gender and MDR S. aureus carriage requires further exploration. No previous studies have particularly explored the maternal-infant relatedness of MDR S. aureus carriage. Therefore, we compared our results with other previous maternal-infant Staphylococcus aureus carriage. The aRR of maternal-infant MDR S. aureus carriage in our study was 7.63. We found that it was much higher than other previous studies, which reported that the RRs of maternal-infant Staphylococcus aureus carriage ranged from 2.04 to 5.70 (21, 24–26). The results demonstrated that maternal-infant MDR S. aureus transmission is much more hazardous.

There were few significant differences on phenotypic and virulence genetic characterizations in MDR S. aureus isolates between mothers and their newborn infants. We observed good consistency on certain STs of MDR S. aureus isolates between mothers and their newborn infants, such as ST1, ST6, ST7, ST59, ST88, ST398, and ST965. These STs were also reported in other previous studies (20, 27). Furthermore, we found that six concordant maternal-infant MDR S. aureus pairs had the same phenotypic and molecular characteristics. These results could further verify the homology of maternal-infant MDR S. aureus carriage.

In the current study, we found that more than 5% of MDR S. aureus isolates were resistant to linezolid and the rate was higher in isolates from newborns. This was not usual, and it could raise major issues in the case of infection with these strains (28). Healthcare workers should pay greater attention to mothers and newborns with linezolid resistance.

This study has some limitations. First, the correlation of MDR S. aureus carriage between mothers and infants needs to be explored further because of the limited sample size in this study. Second, whole-genome sequencing would further strengthen this study. Third, the potential importance of environmental contamination in MDR S. aureus carriage would need to be further explored. Fourth, participants in this study were volunteers, recruited using a convenient sampling approach. It is possible that selection bias may have occurred. Last, we did not follow up with the newborns because of insufficient financial and human support.

In summary, we identified vertical maternal-infant transmission of MDR S. aureus carriage in newborn infants. Newborn infants born with maternal MDR S. aureus carriage appear to be at higher risk of MDR S. aureus carriage. Accordingly, routine surveillance for MDR S. aureus carriage in mothers may be indicated. Prevention measures focused on controlling the spread of MDR S. aureus in mothers could be a more effective strategy when outbreaks of MDR S. aureus in newborn infants occur. Further work should seek to elucidate the potential role of maternally derived antibodies in modifying MDR S. aureus carriage risk in newborn infants.

#### AUTHOR CONTRIBUTIONS

ZY designed the study, performed the data analysis, and revised the manuscript. JL collected the information, conducted the experiments, performed the data analysis, and wrote the manuscript.

#### REFERENCES


#### FUNDING

This work was funded by the Innovation Fund of Guangdong Science and Technology Planning Project (grant number 2014A020213013) and Guangdong Pharmaceutical University Innovation and Strength School Funding Project (No.2016). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

## ACKNOWLEDGMENTS

We greatly thank Qianting Ou and Dongxin Lin for helping us with collecting samples and conducting the experiments.


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Lin and Yao. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# The Epidemiology and Pathogenesis of Kawasaki Disease

#### Anne H. Rowley\* and Stanford T. Shulman\*

*Department of Pediatrics, Northwestern University Feinberg School of Medicine, The Ann & Robert H. Lurie Children's Hospital of Chicago, Chicago, IL, United States*

Epidemiologic and clinical features of Kawasaki Disease (KD) strongly support an infectious etiology. KD is worldwide, most prominently in Japan, Korea, and Taiwan, reflecting increased genetic susceptibility among Asian populations. In Hawaii, KD rates are 20-fold higher in Japanese ethnics than in Caucasians, intermediate in other ethnicities. The age distribution of KD, highest in children <2 yo, lower in those <6 months, is compatible with infection by a ubiquitous agent resulting in increasing immunity with age and with transplacental immunity, as with some classic viruses. The primarily winter-spring KD seasonality and well-documented Japanese epidemics with wave-like spread also support an infectious trigger. We hypothesize KD pathogenesis involves an RNA virus that usually causes asymptomatic infection but KD in a subset of genetically predisposed children. CD8 T cells, oligoclonal IgA, and upregulation of cytotoxic T cell and interferon pathway genes in the coronaries in fatal KD also support a viral etiology. Cytoplasmic inclusion bodies in ciliated bronchial epithelium identified by monoclonal antibodies made from oligoclonal IgA heavy chains also supports a viral etiology. Recent availability of "second generation" antibodies from KD peripheral blood plasmablasts may identify a specific viral antigen. Thus, we propose an unidentified ("new") RNA virus infects bronchial epithelium usually causing asymptomatic infection but KD in a subset of genetically predisposed children. The agent persists in inclusion bodies, with intermittent respiratory shedding, entering the bloodstream via macrophages targeting coronaries. Antigen-specific IgA plasma cells and CD8 T cells respond but coronaries can be damaged. IVIG may include antibody against the agent. Post infection, 97–99% of KD patients are immune to the agent, protected against recurrence. The agent can spread either from those with asymptomatic primary infection in winter-spring or from a previously infected contact who intermittently sheds the agent.

Keywords: kawasaki, pathogenesis, etiology, pediatric, coronary

## EPIDEMIOLOGY

Both epidemiologic and clinical features of Kawasaki Disease (KD) strongly support an Infectious etiology. The clinical features of KD including fever, rash, mucosal changes, conjunctival erythema, and cervical lymphadenopathy are all compatible with an infectious illness, and many common (predominantly viral) infections by necessity are included in the differential diagnosis of KD.

Kawasaki Disease (KD) is a worldwide illness, with varying incidence rates that primarily reflect the racial composition of the populations of various countries. The highest incidence of KD is in

#### Edited by:

*Miika Kaleva Arvonen, Kuopio University Hospital, Finland*

#### Reviewed by:

*Audrey Dionne, Harvard Medical School, United States Nagib Dahdah, CHU Sainte-Justine, Canada Toru Watanabe, Niigata City General Hospital, Japan Dirk Van Gysel, Onze Lieve Vrouwziekenhuis Hospital, Belgium*

#### \*Correspondence:

*Anne H. Rowley arowley@luriechildrens.org Stanford T. Shulman sshulman@northwestern.edu*

#### Specialty section:

*This article was submitted to Pediatric Infectious Diseases, a section of the journal Frontiers in Pediatrics*

Received: *12 October 2018* Accepted: *15 November 2018* Published: *11 December 2018*

#### Citation:

*Rowley AH and Shulman ST (2018) The Epidemiology and Pathogenesis of Kawasaki Disease. Front. Pediatr. 6:374. doi: 10.3389/fped.2018.00374*

**14**

Japan, and this has steadily increased with an annual rate of 308.0 per 100,000 children under 5 years reported in 2014 (1). In Japan one in 65 children develops KD by age 5 years. The second highest reported rate was 199.7 per 100,000 <5 years old in 2014 in South Korea (2), while Taiwan has the third highest rate, 82.8 per 100,000 <5 years old in 2010 (3). In countries with predominantly non-Asian populations, the usual annual rate is 10–20 per 100,000 <5 years old (4).

More than 15,979 cases of KD were reported in Japan in 2015, with local clusters occurring commonly, unlike the nationwide epidemics that occurred in 1979, 1982, and 1985–86 (1, 5). In those epidemics, there appeared to be wave-like spread from one prefecture to an adjacent one, a pattern very similar to the spread of specific viral illnesses like measles, for example, in Japan, thus strongly supporting an infectious etiology of KD. In Hawaii, with its complex multi-racial and multi-ethnic population, the overall annual KD incidence is about 50.4/100,000 <5 y/o; for Japanese ethnic children in Hawaii, the rate is about 210.5 and for Caucasians about 13.7, with intermediate rates for children of native Hawaiian, Chinese, Filipino and other Asian ancestries (6). The very striking differences in ethnic-specific rates are indicative of a very strong genetic basis of susceptibility.

The ratio of male: female KD patients approximates 1.5:1 in virtually all countries (1, 4), and severe cardiac complications of KD are even more significantly overrepresented in males. The basis of the male preponderance is unclear but similar to that observed in many infectious diseases.

Kawasaki Disease (KD) has a striking age distribution, with almost 100% of cases occurring in children, 80% in children <5 years old, and 50% in those <2 years old. In a recent Japanese survey 0.7% of cases were ≥ 10 years old (1). The age-incidence curve of KD may help to elucidate risk factors and appears compatible with a ubiquitous highly transmissible infectious agent, and is similar to that seen with respiratory syncytial virus (RSV), for example. The peak age of KD is approximately 10–11 months of life, with a relatively low incidence in the first 6 months, suggesting both the possibility of transplacental immunity as seen in many classic infectious illnesses, as well as progressively increasing degrees of immunity to the KD agent throughout childhood.

The seasonality of KD, with winter peaks in Japan and winterspring predominance in the US and many other temperate areas, is highly suggestive of a viral (probably respiratory viral) etiology (4, 7). Some reports have suggested summer and winter peaks (Beijing and Shanghai), or spring peaks (Sichuan and Hong Kong), while no clear seasonality has been seen in Hawaii (6), and winter predominance was reported from at least some Southern Hemisphere countries. Despite the observed seasonality, in most areas sporadic cases are recognized throughout the year, contrasting somewhat with the usual patterns commonly seen with many highly transmissible respiratory viral illnesses. Recurrent KD is defined as a new illness that meets KD criteria beginning at least 3-months and usually within 2 years after an initial episode of KD, when levels of inflammatory markers have completely normalized. Recurrence occurs in about 1% or fewer of all KD patients, and in up to 3% of those of Asian ethnicity (8).

During an outbreak of KD on Mikayo Island, Japan, in 1980– 1981 (a fairly isolated population of ∼80,000 at that time), 9 KD cases were diagnosed in a 1 month period, and 4 of the cases had close geographic and social contacts, supporting the possibility of direct person-to-person transmission of a KD etiologic agent (9). While there is limited other direct evidence to indicate that KD can be transmitted from person to person, for example in a daycare setting, much circumstantial evidence supports an infectious etiology with genetically susceptible individuals manifesting the clinical features of KD and others having trivial or no symptoms. Simultaneous or sequential cases in siblings, twins, or other contacts are reported, especially during Japanese outbreaks (10). In Japan, secondary sibling cases occur at rates substantially higher than the general childhood population. Sibling cases are reported more frequently in twins than in nontwins, suggesting both genetic susceptibility and person to person transmission. Japanese family data suggest that sibling cases tend to cluster either on the same day as the index case or 7 days later (11).

History of increased frequency of antecedent respiratory illnesses in KD compared to controls was documented in the 1980's in several outbreak investigations (12, 13). Together with the epidemiologic features noted above, the clinical features characteristic of KD also strongly suggest that an infectious agent, perhaps one that has not yet been identified as a human pathogen, is etiologically related to KD.

## PATHOGENESIS

The epidemiologic features of KD described above strongly support infection with a ubiquitous agent that usually results in asymptomatic infection, but causes KD in a small subset of genetically predisposed children. The occurrence of epidemics and geographic wave-like spread of KD during epidemics supports a presently unknown single agent or closely related group of agents as the etiology. The failure of KD patients to respond to antibiotic therapy makes a viral etiology more likely than a bacterial cause. Moreover, the prevalence of CD8 T cells in the inflammatory infiltrate and the upregulation of cytotoxic T cell and interferon pathway genes in the coronary arteries of children who have died of KD are very suggestive of a viral etiology (14, 15).

We discovered an oligoclonal IgA response in the coronary arteries of children who died from KD, and we made "first generation" KD synthetic antibodies using oligoclonal IgA heavy chains with random light chains (16–19). These "first generation" antibodies detected antigen residing in ciliated bronchial epithelium in KD lung and in a subset of macrophages in KD but not in infant control tissues by immunohistochemistry; the antigen in lung localized to intracytoplasmic inclusion bodies that were identified using stains for protein and for RNA (20–22). The inclusion bodies were identified in children from the US and Japan using a single monoclonal antibody, strongly suggesting a single infectious agent as the cause (20, 22, 23). The inclusion bodies could also be identified in some KD children who died as late as months to years after onset (21). Further investigations of

acute phase KD lung samples showed upregulation of interferon pathway genes and virus-like particles in close proximity to the inclusion bodies by transmission electron microscopy (23). However, these antibodies did not identify the specific antigen by Western blot and immunoprecipitation assays, likely because of a lack of cognate heavy and light chain partners in these "first generation" antibodies. This problem has been recently overcome by preparing "second generation" antibodies from acute KD peripheral blood plasmablasts, which include cognate light and heavy chain partners and show great promise in identifying specific antigen (24).

One theory presently favored by some is that KD can result from infection with any of a wide range of infectious agents in a genetically predisposed host, and some investigators propose an immune defect in KD children. We believe that these theories fail to explain epidemiologic findings in KD. If multiple agents can cause KD, epidemics would either not be observed or specific known infectious agents would be identified by careful epidemiologic study as being associated with the epidemics. In fact, there has been an absence of association of KD with known infectious agents during epidemics and outbreaks despite careful study by epidemiologists in Japan, in the US at the Centers for Disease Control, and in other nations (5, 12, 13, 25). If any of multiple agents can cause KD, a substantially higher recurrence rate than the observed 1–3% in the US and Japan would be likely. Because the vast majority of patients do not develop other health problems following KD, an immune defect seems highly unlikely.

Our studies demonstrating an antigen-driven IgA immune response in acute KD and the presence of KD antigen in intracytoplasmic inclusion bodies in KD bronchial epithelium lead us to put forth the following model of KD pathogenesis (**Figure 1**). We propose that a presently unidentified (likely "new") RNA virus infects ciliated bronchial epithelium, causing asymptomatic infection in most individuals and KD in a small subset of genetically predisposed children. Children <6 months of age are less susceptible because of passive maternal antibody. The virus can result in sporadic cases of KD or in outbreaks. The agent can remain persistent in cytoplasmic inclusion bodies, with intermittent shedding into the respiratory tract of previously infected individuals. It can enter the bloodstream via macrophages and target particularly the coronary arteries and also other sites. Antigen-specific IgA plasma cells (17, 19, 20, 22, 23) and CD8 T cells (14, 15, 26) respond to the infection, but coronary arteries can be damaged. The provision of specific antibodies directed at the ubiquitous KD agent could explain the efficacy of intravenous gammaglobulin (IVIG) in the treatment of KD. These specific antibodies are present in IVIG because most adult donors were asymptomatically infected during young childhood, which accounts for the reduced prevalence in older children and the rarity of KD in adults. After infection, 97– 99% of KD patients are immune to the agent and do not have a recurrence of KD. The agent can be spread through the population either from community contacts with asymptomatic primary infection particularly in the winter-spring, or from a close contact who had been previously infected and then intermittently sheds the agent, resulting in cases during other seasons. We believe that our model, although speculative, fits clinical and epidemiologic findings in KD much better than other currently proposed speculative models.

Identification of the etiology of KD is the most important research goal in the field. With this information, a diagnostic

test can be developed, therapy improved, and prevention become possible. Hopefully, in the near future, the etiology can be discovered using synthetic antibodies derived from KD patients' B cell immune response to the triggering agent.

## REFERENCES


## AUTHOR CONTRIBUTIONS

AR and SS contributed equally to conceiving the topics covered and in authoring the work.


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Rowley and Shulman. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Causes of Kawasaki Disease—From Past to Present

#### Satoru Nagata\*

Departments of Pediatrics, Tokyo Women's Medical University, Tokyo, Japan

Kawasaki disease (KD) is a multisystem vasculitis that primarily affects the coronary arteries of young children. The causes of KD remain a mystery. It is suspected that some sort of infectious agent is involved because KD has epidemicity and seasonality. That said, the incidence of the disease is high among Japanese people, so it can be speculated that the hosts may have some sort of genetic characteristic that leaves them susceptible to KD. Various theories regarding the etiology have been asserted, such as the infectious vasculitis theory, autoantigen theory, superantigen theory, and RNA virus theory; however, none of them have been able to overcome this epidemicity. Taking into consideration the knowledge gained from previous reports, the best scenario explaining the pathogenesis is "individuals with certain genetic backgrounds are affected by microorganisms which trigger KD." In this article, the pathogenesis of KD is discussed with a focus on the microorganisms mentioned above, along with the previous and current hypotheses as well as my own opinion.

#### Edited by:

Hiromichi Hamada, Tokyo Women's Medical University Yachiyo Medical Center, Japan

#### Reviewed by:

Antonio Condino-Neto, University of São Paulo, Brazil Hermann Girschick, Vivantes Klinikum, Germany

> \*Correspondence: Satoru Nagata nagata.satoru@twmu.ac.jp

#### Specialty section:

This article was submitted to Pediatric Immunology, a section of the journal Frontiers in Pediatrics

Received: 31 October 2018 Accepted: 16 January 2019 Published: 05 February 2019

#### Citation:

Nagata S (2019) Causes of Kawasaki Disease—From Past to Present. Front. Pediatr. 7:18. doi: 10.3389/fped.2019.00018 Keywords: Kawasaki disease, etiology, superantigens, heat-shock proteins, epidemicity

# INTRODUCTION

The etiology of Kawasaki disease (KD) has remained a mystery since Dr. Tomisaku Kawasaki proposed the disease in 1967. A number of epidemiological and clinical observations suggest that KD is caused by an infectious agent, with suggestions ranging from Staphylococci, Streptococci, Mycoplasma, or Chlamydia (1–4), to viruses such as adenovirus, parvovirus, or Epstein-Barr virus (5–7). It is suspected that infection is involved because KD has epidemicity and seasonality. That said, the incidence of the disease is high among Japanese people, so it can be speculated that hosts may have some sort of genetic characteristic that leaves them susceptible to KD. Taking into consideration the knowledge gained from previous reports, the best scenario explaining the pathogenesis is "individuals with certain genetic backgrounds are affected by certain microorganisms which trigger KD." This article provides an overview along with the newly acquired knowledge by Nagata et al. (8), with a focus on +pathogenic organisms.

#### Possible Pathogens

Various pathogens have been proposed as the trigger, but none have been decisively established. One of the reasons is that there are a variety of items that must be explained to determine that the etiology is certain [**Table 1**, (8)]. In particular, the epidemiology, wherein "the incidence is high among Japanese people," shall be most difficult to explain.

#### TABLE 1 | Epidemiological conditions that the pathogen of Kawasaki disease must meet (8).


Such pathogens must overcome the following challenges, in order, from low to high level of difficulty: (1) detected at a high frequency in a patient; (2) symptoms can be explained; (3) coronary artery lesions (CALs) can be explained; and (4) epidemiological conditions can be explained.

# Various Theories Regarding the Etiology

A common point of view among infectious theories at an early stage is that a pathogen is recognized as an antigen-presenting cell (APC), with factors such as tumor necrosis factor (TNF) α, interleukin (IL)-6, vascular endothelial growth factor (VEGF) (9) produced by macrophage and T cells, etc. activated by macrophages causing vasculitis, leading to the formation of the pathology (**Figure 1A**). Although there are many pediatric diseases mainly involving vasculitis, few diseases besides KD cause CALs. Therefore, many researchers assume the existence of an autoantigen that becomes the target of these attack factors in the components of the vessel wall of small and medium-sized arteries such as the coronary artery (**Figure 1B**). However, the autoantigen involved remains uncertain.

#### Superantigen Theory

According to previous reports on Staphylococcus, hemolytic streptococcus, and Yersinia, the generation of superantigens is involved in the pathology of KD in each report (10–15) (**Figure 1C**). The primary symptoms of KD such as fever, oral cavity findings, and exanthema as well as aggravation of the serum inflammatory reaction are similar to those of diseases related to superantigens. However, the formation of CALs, as the most typical characteristic of KD, cannot be observed in diseases related to superantigens (16).

#### RNA Virus Theory

Rowley et al. have reported that the postmortem examination of KD patients revealed the invasion of many mononucleoses, CD8 positive T-cells, and IgA producing plasma cells (17), proposing a hypothesis that viruses invading mainly from the respiratory apparatus stimulate CD8 positive T-cells and B cells in the organ lymph nodes and differentiate them into IgA producing plasma cells, with these cell groups giving rise to vasculitis (**Figure 1D**). Recent studies have shown that respiratory viruses are detected by a PCR in up to half of KD patients (18, 19) and an ultrastructural search suggested the possibility of a respiratory virus in autopsy specimens (20, 21); however, no virus has ever been repeatedly confirmed in such studies (22).

# Superantigen + Heat-Shock Protein Theory

The authors have hypothesized the gastrointestinal (GI) tract microbiota could be involved in KD because of the following reasons: (1). GI tract is constantly exposed to a milieu of microorganisms, various antigens, and other agents, (2). It is the largest lymphoid tissue in the body, and (3). It has not been fully investigated due to some technical problems. Therefore, we focused on the mucous membrane of the GI tract with a vast area invaded by antigens and enhanced mucosal immunity, and conducted a biopsy on the mucous membrane of the duodenum of infants with KD, suggesting that certain types of antigens may have invaded that significantly activate the immunologic system of the host (23). We suspected that the possible antigens causing the disease may be a virulent alpha hemolytic streptococcus and Staphylococcus, which produce superantigens using the T-cell receptors of Vβ2 repertoire of the GI tract mucosa of infant patients 1 (24, 25). However, the problem we faced is that superantigens cannot explain the formation of coronary artery lesions and differences in race among patients with KD. In 2005, the authors conducted the following analysis (8), hinting at the theory that "superantigens cause the explosive activation of T-cells, producing autoreactive cytotoxic T-cells, which attack autoantibodies expressed on vascular endothelial cells, resulting in vasculitis" (16).

#### Knowledge Obtained From the Analysis by the Authors

First, culture supernatants of bacteria isolated from the oral cavity or duodenal mucosa of 19 children with KD were added to peripheral blood mononuclear cells of the same host, from which we selected those promoting significant cell proliferation using <sup>3</sup>H-thymidine uptake (21–23). As a result, six kinds (18 strains) of gram-positive cocci and five kinds (13 strains) of gramnegative bacteria which remarkably proliferate the peripheral blood mononuclear cells of the same patient child were isolated from the oral mucosa/duodenal mucosa of the patient. In 12 of 19 patients, both gram-positive cocci and gram-negative bacteria were isolated, including patients complicated with CALs. These bacteria showed remarkable tolerance to antibiotics. An examination of the superantigen activity by flow cytometry revealed that the isolated gram-positive cocci induced a marked increase in T cells using the Vβ2 repertoire of the host peripheral blood (26, 27). Furthermore, we detected the reactive protein between the co-culture supernatant with mononuclear cells and the host serum using the Western blotting method (28) (**Figure 2**). In the gram-negative bacteria, while a large amount of bacteria-specific heat-shock protein (HSP) 60 was generated in the acute phase of KD, the production of a large amount of human HSP60 was induced in the peripheral blood of the

infant patients. In addition, both gram-positive cocci and gramnegative bacteria caused the secretion of inflammatory cytokine: interferon gamma (IFNγ) and TNFα, in the peripheral blood mononuclear cells of the hosts. On the other hand, gramnegative bacteria induced the production of anti-inflammatory cytokine, IL-10 in the peripheral blood mononuclear cells of the patients (29).

#### The Pathogenesis of KD Presumed From the Results of Our Analysis (Figure 3)

At first, once gram-positive cocci with superantigen activity infect the host from the mouth, they stimulate gram-negative bacteria in the upper GI tract and promote the production of HSP60. The superantigen stimulates Th1 cells causing inflammation and also induces the production of autoreactive B cells and cytotoxic T cells. With vascular endothelial cells, the stimulation of HSP60 produced by gram-negative bacteria promotes the production of human HSP 60, which plays a cytoprotective role in the nucleus and cytoplasm and is arranged so as to penetrate the membrane on the surface of vascular endothelial cells. Since part of human HSP 60 has a molecular structure derived from bacteria and the other part has molecular structure specific to human beings, human HSP 60 becomes a target of autoantibodies and cytotoxic T cells produced by

autoreactive B cells generated by superantigens. The membrane of vascular endothelial cells is destroyed, causing CALs (30– 34). Subsequently, human HSP60 is extracellularly released and the molecular structure derived from bacteria activates Th1

cells, while the molecular structure specific to humans activates regulatory T cells. The former further contributes to vasculitis through the secretion of IFNγ, with the latter suppressing the excessive activation of Th1 cells through the secretion of anti-inflammatory cytokine, IL-10 (35). Epidemiological and immunopathological studies have suggested that HSP60 autoantibodies can cross-react with bacterial and self-HSP 60 and induce cytotoxic damage of stressed endothelial cells, resulting in coronary atherosclerosis (31–34). These theories appear applicable to the pathogenesis of coronary lesions of KD considering the high HSP 60 expression in endothelial cells together with the serological detection of bacterial and self-HSP in patients with KD (36–38). Vascular surface-expressed HSP60 the transfer of which from the cytoplasm or mitochondria could be induced by bacterial HSP60 stimulation, can be recognized by circulating anti-HSP autoantibodies or cytotoxic T lymphocytes targeting autoantigens. The incidence of coronary lesions in patients with KD may depend upon how strongly causative agents can induce the initial immune activation that elicits autoreactive T cells and importantly, the number of self-HSP molecules they can evoke from the cytoplasm or mitochondria to the vascular surface. Gram-negative microbes appeared to trigger more self-HSP than Gram-positive cocci. Actually, Gramnegative microbes such as N. mucosa coexisting with Grampositive cocci have been isolated in KD patients with vascular involvement.

# Why the Incidence Is High in Japanese People

The amino acid sequences of Neisseria HSP60 221–231 and 255– 269 had high homology (60 and 50%, respectively) with self-HSP 60 246–256 and 280–294, which are core protein epitopes with high capability to bind to human class II molecules, to induce the production of IFNγ and IL-10, respectively (29). These sequences were demonstrated to have high affinity with HLA-DRB1 <sup>∗</sup> 0401, the gene products of which are recognized by HLA-DR4, which is detected by DR-peptide binding assays. Therefore, the higher incidence of KD is due to the affinity between the binding site with MHC class II molecules on bacterial HSP60 and HLA-DR4 (39–41) (**Figure 4**). The frequency of having this subclass of MHC class II of HLA-DR4 varies between races and has been reported as being typically high in Japanese people (42). This may account for the high incidence of KD among Japanese. Korea and Taiwan have the second and third highest annual incidences of KD in the world, which may support this hypothesis (43, 44).

Regional differences in the risk-allele frequencies of some susceptibility single nucleotide polymorphisms (SNPs) have been identified in genes such as caspase-3 (CASP3) and inositol 1,4,5 triphosphate kinase-C(ITPKC) (45, 46); however, none of the associated parameters proved to be informative in predicting the onset of KD or the development of coronary artery complications (47, 48).

## Antimicrobial Therapy to KD

The background of our study was based on the hypothesized mechanisms underlying the efficacy of intravenous immunoglobulin (IVIG), which include neutralization of the etiologic agents as well as the immunomodulation of T cell regulation and a reduction in the productions of inflammatory cytokines, such as TNFα (49). The levels of IL-6, IFNγ, and TNFα have been reported to be significantly increased before IVIG treatment. While those of IL-6, IL-10, and IFNγ rapidly

decreased after treatment (50). The level of TNFα significantly reduced after treatment in KD patients without coronary artery lesions (CALs); however, it was still high in those with CALs and in patients with IVIG-resistant disease (51). Corticosteroids have been considered for such patients as a representative adjunctive therapy that has the potential to non-specifically reduce inflammatory cytokine production; however, such treatment also has the potential to induce hypercoagulopathy at the time of thrombus formation in CALs. Most the other agents used for adjunctive therapies, such as single infusion of infliximab and cyclosporine, are administered because they are effective for reducing inflammatory cytokine production; however, these are nothing more than symptomatic treatments.

In our study, we found particular Gram-negative microbes producing HSP 60 and several Gram-positive cocci possessing superantigenic properties on the surface of the GI tract that might be involved in the onset of KD. We showed that these organisms were all resistant to commonly used antibiotics, except for sulfamethoxazole trimethoprim (SMX-TMP). We used SMX-TMP for seven cases of KD that were unresponsive to IVIG and studied the antipyretic potency of this treatment. In six out of the seven cases, antipyretic potency was observed without side effects within 2 days of the initial administration (52). Antimicrobial therapy using SMX-TMP may therefore represent a novel strategy for treating cases of KD that are unresponsive to IVIG.

#### REFERENCES


#### Future Prospects

The key to solving the mystery of the pathology of KD is likely to be hidden in CALs as the primary lesion. Aneurysmorrhaphy has interesting characteristics in that the intima and the adventitia are invaded before the media, along with the fact that the blood vessels distributing outside the organ are invaded, although the main characteristic is the destruction of the media forming the framework. We intend to detect possible pathogens on the upper intestinal mucosa or in the peripheral blood using a highly sensitive microbial analytical system based on reverse transcription-quantitative PCR (53). Going forward, we think it is desirable to launch a multicenter collaborative project involving the registration, preservation and search for autopsy specimens, etc.

#### CONCLUSION

Currently, among the numerous etiologies of KD, the most credible theory is that bacterial infection triggers KD. However, we should also consider the possibility of fungi and new types of viruses, which are pathogens that have not yet been paid attention to Rowley (54). To this end, the application of new technologies is expected, such as comprehensive analyses and microarrays. In addition, we would like to emphasize that therapeutic validation should be conducted for possible pathogens, including confirming the effectiveness of antimicrobial agents, etc., in order to make the etiology theory useful in practice.

## AUTHOR CONTRIBUTIONS

The author confirms being the sole contributor of this work and has approved it for publication.

## FUNDING

This work was supported by Secom Science and Technology Foundation.

## ACKNOWLEDGMENTS

I wish to express extreme gratitude to Prof. Yuichiro Yamashiro and Dr. Yoshikazu Ohtsuka, Juntendo University Graduate School of Medicine, Tokyo for offering many valuable advices.


54. Rowley AH. Is Kawasaki disease an infectious disorder? Int J Rheum Dis. (2018) 21:21–5. doi: 10.1111/1756-185X.13213

**Conflict of Interest Statement:** The author declares that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

The handling editor declared a shared affiliation, though no other collaboration, with the author SN at time of review.

Copyright © 2019 Nagata. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# The Gut Microbiota-Host Partnership as a Potential Driver of Kawasaki Syndrome

#### Susanna Esposito<sup>1</sup> \*, Ilaria Polinori <sup>1</sup> and Donato Rigante2,3,4

<sup>1</sup> Pediatric Clinic, Department of Surgical and Biomedical Sciences, Università degli Studi di Perugia, Perugia, Italy, <sup>2</sup> Institute of Pediatrics, IRCCS, Rome, Italy, <sup>3</sup> Fondazione Policlinico Universitario A. Gemelli, IRCCS, Rome, Italy, <sup>4</sup> Università Cattolica Sacro Cuore, Rome, Italy

Kawasaki syndrome (KS) is a necrotizing vasculitis of small- and medium-sized vessels mostly affecting children under 5 years of age; a host of clinical and epidemiological data supports the notion that KS might result from an infectious disease. However, many efforts have failed to identify a potentially universal trigger of KS. The contribution of the intestinal microbial community—called the "microbiota"—to KS has been evaluated by an increasing number of studies, though limited to small cohorts of patients. Differences in the microbiota composition were found in children with KS, both its acute and non-acute phase, with abnormal colonization by Streptococcus species in the intestinal tract and a wider presence of Gram-positive cocci in jejunal biopsies. In particular, a higher number of Gram-positive cocci (of the genera Streptococcus and Staphylococcus), Eubacterium, Peptostreptococcus, and HSP60-producing Gram-negative microbes have been found in the stools of KS children, and their effects on the antigenic repertoire of specific T cells and Vβ2 T cell expansion have been assessed. Conversely, Lactobacilli were lacking in most children with KS compared with other febrile illnesses and healthy controls. All studies available to date have confirmed that an imbalance in the gut microbiota might indirectly interfere with the normal function of innate and adaptive immunity, and that variable microbiota interactions with environmental factors, mainly infectious agents, might selectively drive the development of KS in genetically susceptible children. Further investigations of the intestinal microflora in larger cohorts of KS patients will provide clues to disentangle the pathogenesis of this disease and probably indicate disease-modifying agents or more rational KS-specific therapies.

Keywords: Kawasaki syndrome, infection, innovative biotechnologies, microbiota, personalized medicine, child

# INTRODUCTION

The most insidious primary vasculitis in childhood is Kawasaki syndrome (KS), an acute multi-systemic illness which predominantly affects children under 5 years of age (1). Currently, this disorder of unknown etiology remains the main cause of acquired heart disease among children living in developed countries, where rheumatic fever has been surpassed (2, 3). This condition was originally called "mucocutaneous lymph node syndrome" by Dr. Tomisaku Kawasaki, who was its discoverer, and was thought to be a benign children's disease; nowadays, the illness has

#### Edited by:

Kyung-Yil Lee, The Catholic University of Korea, South Korea

#### Reviewed by:

Valerio Iebba, Istituto Pasteur Italia, Italy Jeong Jin Yu, University of Ulsan College of Medicine, South Korea

\*Correspondence: Susanna Esposito susanna.esposito@unimi.it

#### Specialty section:

This article was submitted to Pediatric Infectious Diseases, a section of the journal Frontiers in Pediatrics

Received: 03 February 2019 Accepted: 15 March 2019 Published: 05 April 2019

#### Citation:

Esposito S, Polinori I and Rigante D (2019) The Gut Microbiota-Host Partnership as a Potential Driver of Kawasaki Syndrome. Front. Pediatr. 7:124. doi: 10.3389/fped.2019.00124

**25**

been described worldwide in children of every ethnicity following the presence of fever persisting (at least) 5 days together with (at least) 4 of the 5 following signs: bilateral conjunctival injection, oropharyngeal inflammation, abnormalities of hands and feet, polymorphous exanthema, and non-purulent cervical lymphadenopathy, usually unilateral (4). Several years after the first description, fatalities occurred among children with KS younger than 2 years living in Japan, prompting clinicians to reconsider KS's long-term risks related to systemic and necrotizing effects on the vascular endothelium of small- and medium-sized arteries, which have been acknowledged in all guidelines related to the management of KS (5, 6). The most relevant sequelae of KS include variable degrees of damage within coronary arteries in combination with angina, myocardial infarction, ischemic cardiomyopathy, and sudden death; these complications should be preventable with a timely treatment of high-dose intravenous immunoglobulin (IVIG), which is the recommended therapeutic strategy in KS (7). Higher acute phase reactants and younger age at onset of KS are nodal points in determining, respectively, a failure in the response to IVIG and an increased occurrence of coronary artery abnormalities (8). The prediction of IVIG resistance is also crucial in KS patients, as recognizing these high-risk children should consent to start an intensified treatment protocol combined with IVIG to prevent coronary injuries (9).

KS incidence varies widely among different ethnic groups; for instance, in the United Kingdom it has stabilized and remains low at 2.8 per 100,000 population under the age of 20 years. However, general practitioners should be aware that the condition occurs throughout childhood and across the seasons, and—given the potential cardiovascular sequelae—KS should be considered in all children with persistent fever, even in older children and adolescents (10). KS is most prominently recognized in Japan, Korea, and Taiwan, reflecting increased genetic susceptibility among Asian populations. A recent study reported an incidence of ∼240 per 100,000 children under 4 years of age in Japan (11). There is still much controversy about the etiology of KS, though epidemiologic and clinical data suggest that KS might originate from an abnormal response to undisclosed infectious diseases in genetically susceptible children (12). There is no agreement whether KS-related infectious agents are of viral, bacterial or fungal origin (13), and the underlying immune mechanisms behind KS have not been completely highlighted, remaining only partially known. The absence of a proven unambiguous cause of KS has induced the scientific community to pay attention to other environmental hypothetical triggers, and in particular the composition of the resident intestinal flora as a potential contributor to KS has been evaluated by different research groups.

The main aim of this review was to analyze the relationship between the microbial community, or the "microbiota," and the overall impact of bacterial or viral infections in the potential development of KS. Scientific papers have been searched from the electronic databases of PubMed until January 2019; the retrieving words were "Kawasaki disease," "Kawasaki syndrome," "microbiota," and "microbiome"; additional reports were identified and analyzed through the specific references cited in the retrieved papers. Only papers published in English and those showing evidence-based data were included in our evaluation.

# EVOLUTION OF THE MICROBIOTA IN CHILDREN

The microbiota, a microbial community of trillions of microorganisms and at least 1,000 different bacterial species, some eukaryotic fungi and viruses, and which covers every surface of the human body, plays a contributory role in many infections, immune-mediated disorders, rheumatologic diseases, and disorders of the nervous system. The microbiome, on the other hand, is the collection of the whole genome sequences of those microorganisms, consisting of more than 5,000,000 genes (14, 15). In particular, the gut microbiota is strictly linked to the chronological age of each individual and modulates host physiology and metabolism through different mechanisms.

Each stage of human life is characterized by a specific intestinal microbial composition: the microbiota that initially colonizes the fetus' intestinal tube consists of aerobic organisms such as Enterococcus and Streptococcus, is then gradually replaced by anaerobes such as Bifidobacterium and Lactobacillus and finally reaches the adult composition dominated by Bacteroides and Firmicutes (16, 17). The fetal microbiota is prone to be conditioned by the type of delivery; the mother's vaginal flora is a relevant source of Lactobacillus, Prevotella, and Bifidobacterium. Conversely, a cesarean delivery delays contact with these species, producing a similar-to-skin flora, dominated by Staphylococci (18). The feeding regimens and food supplements also play a role in modifying the resident flora; a greater complexity is normally seen in infants fed with formula, rather than in breastfed babies who have an "adult-like" structured microbiota with a population rich with Bifidobacteria, Lactobacilli, and Bacteroides (18). The infant gut microbiota is variable in composition over time and highly changeable during the first year of life, being influenced by specific bacteria to which a baby happens to be exposed, as shown by the resemblance of infants' stool microbial community with mothers' milk and vaginal samples (19). Thereafter, the infant's intestinal tract progresses toward an extremely dense colonization, ending with a mixture of microbes that is broadly very similar to an adult's intestine. During adulthood, the gut microbiota becomes stable and this intestinal homeostasis remains in equilibrium with the host. Food habits influence the composition of the whole intestinal microbiota, as testified by the lower prevalence of Bacteroides in those suffering from malnutrition and by different microbiota changes occurring in children with diet-related diseases, such as allergies and obesity (19).

Arumugam et al. identified three distinct enterotypes, namely Bacteroides, Prevotella, and Ruminococcus, which reflect many individual alimentary profiles: Bacteroides correlates to a highfat or high-protein regimen, whereas Prevotella is associated with higher consumption of fibers and simple sugars (20). More recent data have consented to unify Bacteroides and Ruminococcus enterotypes due to the large similarity between the two (21). It is well-established that early events of birth, environmental factors during infancy, sex hormones, diet, body weight, and use of antibiotics can undoubtedly differentiate the composition of the microbiota (22, 23). There are no ideal culture methods to give us a real overview of the complete intestinal flora, though molecular methods, e.g., DNA microarrays with comprehensive coverage of most bacterial taxa represented in the available database of small subunit ribosomal RNA gene sequences, should allow the characterization of most taxonomic groups of the intestinal bacteria (24, 25).

## ROLE OF THE BACTERIAL FLORA IN CHILDHOOD DISEASES

The ancient symbiosis between the human gastrointestinal tract and its resident microbiota involves diverse reciprocal interactions between the microbiota itself and the host, with relevant consequences for human health and physiology. The quality of the microbial flora has an impact on the maintenance of health and also on prevention of diseases. Indeed, there are numerous roles carried out by the intestinal ecosystem, as the stimulation of angiogenesis, control of host fat storage and protection against other pathogens. In particular, the microbiota influences the formation and progress of regulation of both innate and adaptive immunity in close interaction with the intestinal mucosal immune system. The intestinal mucosa may be considered as an immunological niche as it hosts a complex immune-functional organ comprised by T cell subpopulations, neutrophils, macrophage-dendritic cells, enterocytes (that possess tight intercellular junctions) and their related anti- and pro-inflammatory cytokines as well as several other mediators of inflammation or antimicrobial peptides, defensins, and secretory immunoglobulin A (IgA) (26). The intestinal microbiota may also have direct or indirect effects on the natural course of viral infections, interacting with viral particles and leading to differences in either pathogenicity or anti-viral immune response through recruitment and activation of several T cell subpopulations (18). A large amount of data has also depicted the relevance of gut microbiota-immune system cross-talk in several diseases, and indeed an "imbalance" of the intestinal flora has been shown in patients with atopic diseases and various noninfectious diseases, including metabolic disorders, chronic inflammatory bowel disease (IBD), irritable bowel syndrome, pancreatic diseases, atherosclerosis, and rheumatoid arthritis (27, 28). Furthermore, the gut microbiota, interacting with pattern recognition receptors (PRRs), signaling receptors that can recognize molecular structures of pathogens and activate the cascade of innate immunity, plays a crucial role in maintaining the homeostasis of the innate immunity responses in the gut, and leaks in the intestinal mucosal barrier lead to the translocation of bacterial products into portal circulation which promotes systemic inflammation (29). In addition, the microbiota can maintain a segregation between intestinal mucosa and bacteria via PRRs, though pathogens might usurp innate signals to their advantage (30).

Several immunologic, metabolic and nutritional processes are normally controlled by the intestinal microbiota, and changes in the local microbial communities have been linked to chronic low-grade inflammation (31). Alexander et al. demonstrated that the microbiota has specific effects on adaptive immunity, such as the induction of regulatory effector CD4+ cells and production of cytokines and antimicrobial factors, influencing the individual response to various environmental stimuli, as seen in IBD, Crohn's disease and ulcerative colitis (32). In a recent work, Schwiertz et al. have shown microbiota changes characterized by decreased numbers of Faecalibacterium praunsitzii and increased numbers of Escherichia coli in IBD (33). Additionally, in celiac disease De Palma et al. have described a difference in the microbiota composition with a reduction of Bifidobacterium, Clostridium histolyticum, Clostridium lituseburense, Faecalibacterium prausnitzii, and an abundance of Bacteroides and Prevotella strains (34). A significantly higher biodiversity in coeliac children's duodenal mucosa was demonstrated by Schippa et al. who also highlighted that the possible pathophysiological role of such microbial differences needs further characterization (35). In addition, many immune phenomena were shown to deteriorate under the effect of changes in the microbiota, as revealed by the correlation among intestinal bacterial overgrowth, increased permeability, and development of non-alcoholic steatohepatitis (36).

Much interest has also been paid to the role of the microbiome in the development of "sterile" inflammation, and recent studies have proved that either depletion of the microbiota or changes in the diet and in the gut microbiome might lead to the improvement of inflammasome-mediated manifestations of autoinflammatory disorders, which are caused by dysregulation of specific components of innate immunity (37). These diseases can be subdivided into monogenic and multifactorial disorders, with the former being caused by mutations of genes involved in the regulation of the innate immune system and the latter by a combination of genetic background and environmental factors (38–40). Given the evidence for the role of the intestinal microbiota in the inflammatory state of IBD, atopic diseases and numerous non-infectious diseases, it has been speculated that intestinal microbial agents might also play a trigger role in the development of other inflammatory disorders which do not have a clearly defined etiology, such as KS.

# UNSOLVED PROBLEMS ABOUT THE ETIOLOGY OF KAWASAKI SYNDROME

The etiology of KS remains obscure, although clinical and epidemiological features suggest a primary infectious cause. Indeed, a self-limited and generally nonrecurring illness that manifests itself by fever, rash, mucositis, conjunctival injection, and cervical adenopathy fits well with an infection. More precisely, clinical features of KS resemble some peculiar infectious diseases, such as streptococcal infections, staphylococcal toxic shock syndrome, and atypical measles (12, 13). Additionally, age distribution, winter-spring seasonality, high rates of KS in siblings, and occurrence of community outbreaks are suggestive of a transmissible childhood disease, though many efforts with conventional bacterial and viral cultures and serological methods as well as animal inoculation studies have failed to identify a final unique infectious agent of KS (41). Reported non-infectious factors associated with KS include carpet shampoo, preexisting eczema, environmental pollution, and house dust mites (42). The higher incidence of KS in Asian populations, presence of familial clustering, and elevated risk of recurrence vs. the risk of a first episode in KS-naïve children are all strong evidence of a genetic contribution to KS susceptibility, probably in association with an infectious trigger (43, 44). However, KS does not appear easily transmittable and does not respond to any antibiotics: these characteristics should contradict a primitively infectious pathogenesis of the illness.

The most current theory about the pathogenesis of KS is that the disease results from an exaggerated immune response toward environmental stimuli occurring in a genetically susceptible child (13). The prominent role played by the immune system in KS is confirmed by the high number of studies revealing the activation of neutrophils and multiple immune cells with overproduction of pro-inflammatory cytokines, including tumor necrosis factor (TNF)-α (45, 46). Manlhiot et al. have very recently proposed a new pathogenetic model of KS in which the disease risk is determined by concurrent interacting processes: genetic susceptibility, habitual exposure to allergens, atmospheric biological particles, and infectious agents (47). A study exploring the immune responses during the acute phase of KS showed increased levels of lipopolysaccharide (LPS) bound to the surfaces of circulating neutrophils via CD14 receptor and found markedly increased levels of the soluble CD14 in the plasma (48). Furthermore, Chen et al. (49) demonstrated that NLRP3 inflammasome activation is associated with the development of coronary arteritis in a mouse model of KS, and that infusion of visfatin, a major injurious adipokine, can activate NLRP3 inflammasome and increase interleukin (IL)- 1 production, leading to enhanced endothelial dysfunction. These novel molecular mechanisms of vasculitis mediated by inflammasome activation open more specifically the road to innate immunity pathways in the interpretation of KS (49, 50).

# THE COMPLEX RELATIONSHIP OF KAWASAKI SYNDROME AND INFECTIONS

Unchallengeable proof that an infection is the starting point of KS is not available. Different epidemiologic studies have supported this hypothesis, based on documented infections by various microorganisms in many cases of KS. Further clues are the occurrence of KS in epidemics, higher incidence during spring and winter and early age at which the disease might be acquired, that is 6 months to 5 years (5). Striking perturbations of many immune pathways occur during the acute phase of KS, which determines a multi-cytokine cascade in the vascular endothelium with focal disruption of small- and mediumsized vessels. However, the exact key steps leading to coronary arteritis are still far from being clarified, though endothelial cell activation; prolonged start-up of neutrophils, CD68<sup>+</sup>

monocyte/macrophages and CD8<sup>+</sup> lymphocytes; production of oxygen intermediates and lysosomal enzymes; and oligoclonal IgA response by plasma cells all appear to be involved (45, 46, 51–53).

The contribution of viruses to KS has been suggested by ultrastructural studies which found cytoplasmic inclusion bodies containing RNA of viral origin in the bronchial epithelia (54) as well as by studies revealing the detection of intracytoplasmic inclusion bodies containing viral proteins and nucleic acid aggregates (55, 56). Recent investigations have supported the hypothesis that immune responses in KS are oligoclonal rather than polyclonal (as found typically in superantigen-driven responses), and that IgA plasma cells play a crucial role. Indeed, higher levels of IgA have been found in the vasculature of patients with previous KS, indicating an antigen-driven response against an etiologic agent which might have a respiratory or gastrointestinal portal of entry (57).

About half of all KS patients might have one or more respiratory viruses detected by polymerase chain reaction, without any particular predominance, but a positive respiratory viral test or presence of respiratory symptoms at the time of presentation should not be used to exclude the diagnosis of KS (58). Other studies have investigated a potential relationship of KS with coronaviruses, though without finding definite proof of cause and effect (59, 60). Many viruses have been suggested to be implicated in the pathogenesis of KS, such as adenovirus, parvovirus B19, rotavirus, H1N1 influenza virus, Epstein-Barr virus, herpesvirus 6, coxsackie B3 virus, parainfluenza virus type 3, measles virus, dengue virus, human immunodeficiency virus, and varicella-zoster virus, but no significant differences emerged from case-controlled studies (61–69). One of the most difficult infections that are harder to distinguish from KS is caused by adenovirus, due to its frequent incidental detection in KS patients and frequent laboratory finding of increased inflammatory markers (70). In particular, adenovirus has been detected in 8.8 and 25% of cases with complete and incomplete KS, respectively, by Jaggi et al. (71). There are reports that have emphasized the possible relationship of cytomegalovirus (in one child from Turkey) and human bocavirus 1 (in a French cohort of 32 patients) with KS through serologic tests and molecular techniques, but this link might be only casual and misleading (72, 73).

Regarding the bacterial origin of KS, it is debated if the infectious agents might be conventional bacteria or bacteria with superantigen activity. Superantigens (SAgs) are the most powerful T cell mitogens ever discovered, with potent immunostimulatory actions that simultaneously activate the major histocompatibility complex class II molecules and T cell receptors, leading to massive activation of various immune cells. Many studies have investigated the involvement of SAgs in KS. One of the first studies showed the selective expansion of T cell receptor (TCR) Vβ2-bearing T cells in the peripheral blood of children during the acute phase of KS (74). In addition, Leung et al. isolated Staphylococci and Streptococci that produced SAgs (toxic shock syndrome toxin-1 or TSST-1, SST-1, SEB, SEC, SPEB, and SPEC) from the throat, rectum and groin in 25 of 45 patients with untreated KS (56%) in comparison with 13 of 37 control patients (35%) (75).

Matsubara et al. conducted a case-control study with a serological approach based on enzyme-linked immunosorbent assay and measured serum antibodies against staphylococcal enterotoxins, including TSST-1, and streptococcal pyrogenic exotoxin (SPE), such as SPE-A. They showed that KS patients had significant elevation of IgM antibodies against one or more of five SAgs throughout the first to the fourth disease week (76). In a further paper the same authors analyzed the studies regarding the role of SAgs produced by Staphylococcus aureus and Streptococcus pyogenes in the pathogenesis of KS, finding numerous SAgs implicated, which brought the total number of the known staphylococcal SAgs to over 20 and streptococcal SAgs to 12 (77).

The most recent work that analyzed the relationship between SAgs and KS examined different SAg derived from Streptococcus pyogenes in the stools of patients with KS. Stool specimens were obtained from 36 patients with KS during the acute phase and 26 age-matched healthy children. The authors examined genes related to five Sags—SPE-A, SPE-C, SPE-G, SPE-J, and TSST-1 using polymerase chain reaction; throat and stool cultures were assessed to evaluate the presence of Streptococcus pyogenes and Staphylococcus aureus in KS patients. They reported that at least one of the SAg-related genes was detected more frequently in the stools of children with KS (78). Furthermore, among 358 patients with KS, 54 developed concurrent pneumonia and 12 of these (22.2%) had high titers of anti-Mycoplasma pneumoniae antibody (>1:640), suggesting a potential role of Mycoplasma pneumoniae in KS and the importance of anti-Mycoplasma treatment (79).

Another study has reported a possible role of fungal infections in the development of KS in Japan, San Diego, and the island of Hawaii through an active role of tropospheric wind patterns leaving from Central Asia, in which toxins of Candida species are aerosolized (80). In addition, due to the observation that Candida albicans-derived substances, such as Candida albicans watersoluble fraction (CAWS), induce a coronary arteritis in mice similar to that observed in KS, Sato et al. evaluated the role of a free-β-glucan diet on CAWS-induced vasculitis and found that quality of diet might affect the progression of systemic vasculitis (81). CAWS should act as a pathogen-associated molecular pattern in mice and activate lectin pathway of complement by binding to mannose-binding lectin and inducing an acute inflammatory reaction in the vascular system (82, 83).

# WHAT IS KNOWN ABOUT MICROBIOTA AND KAWASAKI SYNDROME

There are three main reasons that led us to postulate a role of the microbiota in KS. The first is the unsatisfactory microbiological data that limited the association between infections and KS and a lack of evidence of any clear relationship between one or more pathogens with this illness. Second, the most frequent bacteria or viruses associated with KS have a higher prevalence in the overall pediatric population, but only a limited number of children will develop the disease. Third, the association of genetic predisposition and environmental factors in the pathogenetic process of KS makes KS itself similar to other multifactorial diseases.

Currently, the majority of data has found that the composition of the gut microbiota in KS patients differs from healthy subjects. Lee et al. have postulated that the immune system should lose tolerance toward a part of the resident intestinal flora and that environmental factors, i.e., a Western lifestyle or improved public hygiene systems, could transform the commensal flora into a pathogen one, as observed in different gastrointestinal disorders (84).

A recent study about the resident flora and KS was focused on throat flora. Horita et al. (85) assumed that throat flora could be a reservoir of microorganisms triggering KS and, following this hypothesis, they investigated throat microorganisms of KS patients for their content and individual SAg activity. In particular, they collected throat swabs at the start of IVIG infusion in 21 KS patients and compared them with non-KS controls (displaying other febrile illnesses). The results showed no difference in the throat flora between KS and febrile controls even in the mean mitogenic activity of bacteria isolated (85).

The gastrointestinal tract could be one of the primary sites of entry of bacterial toxins in children with KS, and a perturbation in the intestinal microbiota has been linked to the disease's pathophysiology in another study by Yamashiro et al. (86) who investigated the microflora of the small intestine in 15 Japanese patients with KS. The range of bacterial species isolated from jejunal biopsies was characterized by a wider variety of Gram-positive cocci in the acute phase of KS. Notably, 5 kinds of streptococci (Streptococcus salivarius, Streptococcus mitis, Streptococcus oralis, Streptococcus sangius, and Gemella haemolysans) and 2 kinds of staphylococci (Staphylococcus capitis and Staphylococcus hyicus) were isolated only from KS patients, suggesting that some antigens inducing a delayed-type hypersensitivity reaction in the mucosa might inundate the body by breaching the barrier of the small intestinal mucosa of KS patients (86). In fact, Nagata et al. (87) investigated cell surface phenotypes of mononuclear cells and enterocytes in the jejunal mucosa of 16 Japanese patients with KS and in 10 patients with diarrhea due to cow's milk protein intolerance. Both HLA-DR+CD3+ and DR+CD4+ cells were significantly increased, and CD8+ cells significantly reduced in the lamina propria of KS patients during the acute phase compared with patients with cow's milk protein intolerance. These cell patterns returned to normal in the convalescent phase of KS. The authors concluded that a delayed-type hypersensitivity reaction was indeed present in the small intestinal mucosa of KS patients (87).

Due to the fact that the gastrointestinal tract is the largest interface between microbial factors and their host, containing the largest proportion of bacteria and the largest amount of lymphoid tissue in the body, it was hypothesized by Eladawy et al. that the intestinal milieu could be altered in children with KS, and indeed KS patients have a higher incidence of gastrointestinal symptoms and complications (88). Specifically, Takeshita et al. (89) evaluated 20 patients with KS, 20 patients with acute febrile diseases and 20 healthy children, finding that the incidence of Lactobacilli isolated from KS patients (2/20, 10%) was significantly lower (p < 0.001) than in the other cohorts. Moreover, no significant differences in the presence of Staphylococcus, Streptococcus, Enterococcus, Enterobacteriaceae, Bifidobacterium, Clostridium, Veillonella, or Bacteroides were found among the three groups. Conversely, the presence of Eubacterium and Peptostreptococcus was significantly higher in KS patients than in patients with other febrile diseases (p < 0.01 and p < 0.05, respectively), though no significant differences were observed between KS patients and healthy children (89). This observation confirmed that the majority of Lactobacillus species are anti-inflammatory and beneficial for health, particularly during the first years of life, due to their action in protecting against colitis, reducing pro-inflammatory cytokines such as TNF-α, IL-1, or IL-6, and increasing the subsets of regulatory T cells (90). In fact, several studies have largely described the specific action of Lactobacilli in maintaining the epithelial homeostasis of the gut and their striking anti-inflammatory potential (91, 92).

Nagata et al. (93) also studied the role of the gut microbiota in KS pathogenesis via SAgs and heat shock proteins (Hsps) produced by gut bacteria: the authors evaluated Sags and Hsps released by microorganisms isolated from the jejunal mucosa of 19 children with KS compared with 15 age-matched healthy controls, identifying 13 strains of Gram-negative microbes from patients with KS, which produced a large amount of Hsp60, having the power to induce an over-secretion of proinflammatory cytokines. They also identified 18 strains of Grampositive cocci with SAg properties which induced the expansion of Vβ2 T cells in vitro (93). This microbiological analysis disclosed different causative bacteria with a final common pathway of immune activation, which might contribute to the final development of KS.

In order to evaluate the differential microbiota composition of KS patients, Kinumaki et al. (94) performed a metagenomic analysis using non-culture-based methods on feces. Their study included 28 KS patients (15 males and 13 females, aged 1– 114 months, with a median age of 25 months); the time of admission to the study was defined as the acute phase, while 4–6 months after the onset of KS was considered as the nonacute phase. The authors collected a total of 56 samples−28 samples each for both acute and non-acute phases. It was demonstrated that the genera Ruminococcus, Roseburia, and Faecalibacterium were mostly predominant during the nonacute phase, while a higher presence of Streptococcus spp., including Streptococcus pneumoniae, pseudopneumoniae, mitis, oralis, gordonii, and sanguinis, was detected in the fecal samples during the acute phase (94).

This novel interpretation of a disease can be shared by other conditions, as liver cirrhosis and Sjögren syndrome, in which major changes of gut microbiota with higher proportions of Streptococcus spp. have been demonstrated (95, 96). These findings taken as a whole suggest that many other immune-mediated disorders are likely to be connected to an abnormal bacterial colonization of the intestinal tract, with a main role for Streptococcus spp., and that changes in the gut microflora composition might promote systemic and extra-intestinal inflammation. Therefore, all TABLE 1 | Perturbation in the intestinal microbiota of patients with Kawasaki syndrome.


<sup>a</sup>Higher in KS patients than in other febrile illness (no differences were observed between KS patients and healthy children).

<sup>b</sup>Higher in KS patients during non-acute phase of the disease.

<sup>c</sup>Higher in KS patients during acute phase of the disease.

presented studies confirm the concept that an imbalance in the gut microbiota might directly or indirectly interfere with the normal functions of the immune system, and that the interaction with other environmental factors, mainly infectious agents, might lead to the final development of KS.

**Table 1** shows the characteristics of the microbiota in children with KS, as emerging by the most relevant studies dedicated to this issue.

# CONCLUSIVE REMARKS AND FUTURE PERSPECTIVES

As no etiologic agent has ever been accused of being directly involved in the etiology of KS, basic research evaluating the pathogenic mechanisms of this disease will probably provide better therapies and probably consent to identify the most vulnerable hosts and protect them. A potential relationship between eubiosis and the protean functions of immunity has been contemplated by different studies, and conversely a relationship should exist between dysbiosis and immunity dysfunction shown by various diseases (97). It is strengthened that heterogeneity and abnormalities in the intestinal microflora composition may trigger or contribute to the development of specific diseases, and an increasing amount of research and microbiologic observations have led to a role of the intestinal microbiota for KS to be postulated. While we are becoming convinced that a role of the microbiota is likely, we do not know exactly the basic mechanism of how the microbiota should act. The principal obstacle of today's medical literature is to understand if the microbiota modification during KS can cause the disease or if it is a mark and a consequence of the disease itself, and if modulating the microbiota/microbiome might represent a future target of therapy in KS. This raises the hypothesis of targeting intestinal microflora in order to restore eubiosis through the rational use of antibiotics, xenobiotics, probiotics and nutrients. While multicenter trials and registries may allow us to improve the general outcome of KS, further in-depth analysis in larger cohorts of affected children will probably unravel the tangled pathogenesis of this mysterious disorder and

#### REFERENCES


find new targets for identifying disease-modifying agents or more specific therapies.

#### AUTHOR CONTRIBUTIONS

SE, IP, and DR contributed to all stages of the preparation of this manuscript, including conception and writing. All authors approved the final submitted version of the manuscript.

year of life. Cell Host Microbe. (2015) 17:690–703. doi: 10.1016/j.chom.201 5.04.004


hypothesis of a "new" virus associated with Kawasaki disease. J Infect Dis. (2011) 203:1021–30. doi: 10.1093/infdis/jiq136


Pediatr Res. (1993) 33:557–63. doi: 10.1203/00006450-19930600 0-00004


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Esposito, Polinori and Rigante. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Outcomes of Kawasaki Disease Children With Spontaneous Defervescence Within 10 Days

Ya-Chiao Hu, Hsin-Min Liu, Ming-Tai Lin\*, Chun-An Chen, Shuenn-Nan Chiu, Chun-Wei Lu, Luan-Yin Chang, Jou-Kou Wang and Mei-Hwan Wu

Department of Pediatrics, National Taiwan University Hospital and Medical College, National Taiwan University, Taipei, Taiwan

Background: Kawasaki disease (KD) is one of the most common vasculitis in childhood. Intravenous γ-immunoglobulin (IVIG) is recommended to be administrated within 10 days after fever onset. However, some patients didn't have IVIG therapies because of atypical disease presentations or spontaneous defervescence. We aimed to evaluate the coronary outcomes of the KD patients who didn't receive IVIG and had defervescence within 10 days.

#### Edited by:

Miika Kaleva Arvonen, Independent Researcher, Kuopio, Finland

#### Reviewed by:

Guillermo Soza, Universidad de La Frontera, Chile Laura Ferreras-Antolín, NHS England, United Kingdom

> \*Correspondence: Ming-Tai Lin mingtailin@ntu.edu.tw

#### Specialty section:

This article was submitted to Pediatric Infectious Diseases, a section of the journal Frontiers in Pediatrics

Received: 31 December 2018 Accepted: 03 April 2019 Published: 24 April 2019

#### Citation:

Hu Y-C, Liu H-M, Lin M-T, Chen C-A, Chiu S-N, Lu C-W, Chang L-Y, Wang J-K and Wu M-H (2019) Outcomes of Kawasaki Disease Children With Spontaneous Defervescence Within 10 Days. Front. Pediatr. 7:158. doi: 10.3389/fped.2019.00158 Methods: We retrospectively reviewed the KD patients in NTUCH between 2008 and 2015. The patients with a diagnosis of KD and had a febrile length between 5 and 10 days were enrolled. Days of fever, clinical symptoms, laboratory data at the acute stage, and series of coronary artery measurements within a minimum of 3 months after disease onset were recorded. Risk factors associated with coronary lesions 1 month after KD onset were also analyzed.

Results: Two hundred ninety-three eligible KD patients were enrolled (Male: 55.1%, mean age of onset: 1.8 years old). Thirty-seven patients had spontaneous defervescence without IVIG treatment. The incidence of coronary aneurysms at the 4th week after disease onset was higher in spontaneously defervesced KD patients than those treated with IVIG (18.9% vs. 5.1%, p = 0.002). Interestingly, of the 238 KD patients without coronary lesions at their acute phase, percentages of emerging coronary aneurysms became significantly higher if they didn't have IVIG therapies due to spontaneous defervescence (4/31), compared with those who received IVIG (3/208). Further analysis showed the development of coronary lesions at 1 month after disease onset was associated with younger age (<12 months old, p = 0.024), and leukocytosis (WBC > 17,000/cumm, p = 0.031).

Conclusions: 18.9% of KD patients with spontaneous defervescence had coronary aneurysms. Even without initial coronary lesions, such patients were still riskier to develop coronary aneurysms, compared with KD patients who received IVIG therapies. Such findings address the importance of refining the strategy for use of IVIG in the spontaneously defervesced KD patients within 10 days after fever onset, at least in those with age younger than 1 year and those with leukocytosis.

Keywords: Kawasaki disease, spontaneous defervescence, coronary artery lesions, immunoglobulin, risk factors

# INTRODUCTION

Kawasaki disease (KD) is the most common pediatric systemic vasculitis. The incidence in Taiwan is the third highest globally, just lower than that in Japan and Korea (1, 2). The state-ofthe-art therapy recommends the intravenous immunoglobulin (IVIG) therapy during the acute stage, which is effective to decrease the coronary complications from 20 to 5% (3). However, within 10 days, some KD patients experienced spontaneous defervescence without IVIG administration. The reasons why they didn't receive IVIG were such as incomplete presentations and consideration of the cost of medication. Although the 2012 guideline of Japanese Society of Pediatric Cardiology and Cardiac Surgery (JSPCCS) recommended clinicians may refrain from IVIG in cases of less severe KD or spontaneous defervescence, based on the considerations detailed in certain scoring systems (4), the risk-stratified management of KD patients who had spontaneous defervescence remains uncertain. Therefore, we aimed to assess the midterm coronary outcomes of Kawasaki disease children with spontaneous defervescence within 10 days and explore the risk factors associated with the development of coronary arterial lesions (CAL) in the subgroup of KD patients.

# METHODS

# Study Populations

This study was approved by an ethics committee of National Taiwan University Hospital. In the study, we retrospectively reviewed the institutional databases for patients with a diagnosis of Kawasaki disease at National Taiwan University Children Hospital between January 2008 and December 2015. The diagnosis of KD in this study was made based on clinical criteria for KD (5). Algorithm for the evaluation of incomplete KD patients from the 2004 American Heart Association (AHA) KD guideline (5) was applied to aid the diagnosis. We excluded patients who had defervescence <5 days and over 10 days, who received more than one dose of IVIG, who have <2 principal clinical features, who lost to follow-ups and who had alternative diagnosis eventually, such as adenovirus infection, herpetic gingivostomatitis, and scarlet fever. All patients enrolled in the study received a minimum follow-up period of 3 months in order to evaluate the coronary outcome from the acute phase to the convalescent period. We collected the patients' laboratory data at the acute and subacute phase including hemogram, the serum level of C-reactive protein, sodium, albumin, aspartate transaminase (AST), and alanine transaminase (ALT). Data of urinalysis, urine culture, viral culture, and serum antistreptolysin O at acute phase were also collected if the patient had received the tests. All of the patients received echocardiography during the acute phase and the subacute phase after fever onset. The frequency of the following echocardiography varied according to the CAL severity. Their laboratory data collected at diagnosis and during follow-ups were reviewed.

## Diagnosis Definition

Based on 2004 guideline of American Heart Association for KD (5), complete KD is diagnosed in the presence of fever for at least 5 days with at least 4 of the 5 principal clinical features. In the presence of <4 principal clinical features, the diagnosis of KD can be made when coronary artery disease is detected by 2D echocardiography. Patients who have fever ≥5 days but do not have sufficient principal clinical findings may be diagnosed with incomplete KD under the support of other clinical, laboratory and echocardiographic findings. In the current study, we defined the patients with a diagnosis of either complete or incomplete KD who had spontaneous alleviation of fever without IVIG treatment within 5–10 days after the onset of disease as "defervesced KD" (dKD).

## Coronary Measurement

The coronary diameters of the left main coronary artery (LMCA), left anterior descending artery (LAD), and right coronary artery (RCA) were collected. The coronary Z score is adjusted by the body surface area and using the reference established from data of Taiwanese children (6). Coronary artery lesion (CAL) is defined as a Z score of coronary diameters ≥ +2.5. (1) On the basis of previous research, persistent CAL for more than 4 weeks after fever onset is defined as coronary artery aneurysm (CA) in this study (7, 8). Coronary aneurysms were sub-classified based on their internal diameter as small (+2.5 ≤ Z < +5.0), medium (+5.0 ≤ Z < 10), and giant (Z ≥ +10.0) (3). CALs, CAs, and the regression were diagnosed based on 2D echocardiography.

# Data Analysis

Patient data were expressed as counts, percentages, medians with interquartile ranges (IQRs) for non-normal distribution data or mean with standard deviation (SD) for normal distribution data. All data analysis was conducted by SPSS statistics (version 22.0) for Windows. We used Chi-square tests and Fisher's exact tests for comparison of categorical variables. The Mann-Whitney U test and Wilcoxon rank-sum test were used for comparison of continuous variables. We set P <0.05 as a statistical significance in this study.

# RESULTS

# Patient Characteristics

A total of 293 cases with KD diagnosis were enrolled in the study. The mean (±SD) age of the enrolled patients was 1.8±1.6 years. Thirty-seven dKD patients (i.e., spontaneous defervescence within 10 days, 12.6%) were identified. The demographics, clinical symptoms, and laboratory features for KD patients treated with IVIG and dKD patients are presented in **Table 1**. The frequency of incomplete KD in the dKD patients was 51.4%, higher than that in the IVIG-treated group (37%, p = 0.11)

Furthermore, of the 37 dKD patients, 18 (48.6%) patients were diagnosed as complete KD. Twelve patients had 4 principle clinical features together with a positive echocardiogram. Two patients had 4 principle clinical features and ≥3 laboratory findings described in the 2004 AHA guideline. Five patients fulfilled only 3 principle features and all of them had a positive echocardiogram. The dKD group presented less number of the five principal clinical features compared to the IVIG group [mean (±SD): 3.4 (±0.8) vs. 3.7 (±0.8), p = 0.04]. The dKD group TABLE 1 | Clinical and laboratory features of KD patients with fever from 5 to 10 days.


dKD: KD patients who had spontaneous alleviation of fever without IVIG treatment within 5–10 days after the onset of disease. Values are expressed as percentages (%) and <sup>a</sup>Mean ± SD; SD, standard deviation; BCG, bacille Calmette–Guerin; WBC, White blood cell; CRP, C-reactive protein; AST, Aspartate aminotransferase; Hb, hemoglobin; PLT, platelet; CAL, Coronary artery lesion; CA, coronary artery aneurysm.

\*Statically significance is defined as p < 0.05.

had a significantly lower incidence of conjunctival injection and BCG scar reactivation than the IVIG group. We analyzed the laboratory data collected at the first blood test during the acute phase, platelet count was significantly higher and serum level of C-reactive protein was lower in dKD patients, compared with KD patients with IVIG therapies (**Table 1**).

#### Coronary Outcomes

Of the 293 cases, 54 (18.4%) patients were found to have CAL at their acute phase and 20 (6.8%) developed CA at the 4th week after disease onset. No difference of maximal coronary Z scores or percentages of CAL were noted between IVIG group [mean (±SD):1.65 (±0.99)] and dKD group [1.49 (±0.98), p = 0.14] at their acute phase. At 1 month after disease onset, the maximal coronary Z score of dKD group is relatively, not significantly, larger than that of the IVIG group. However, at this time point (1 month), the dKD patients had a higher incidence of CA than the IVIG group (18.9 vs. 5.1%, p = 0.002) (**Figure 1**). Compared with the 256 IVIG-treated KD patients, the 37 dKD patients had their first echocardiography at an average of 8.4 days after fever onset, significantly later than the 256 IVIG-treated KD patients (5.7 days, p = 0.001). We did univariate analysis to explore the risk factors of coronary artery lesion at 1 month after KD onset (defined as "coronary aneurysm" in the current study) and found age, incomplete KD, IVIG treatment, changes in lips and oral cavity, white blood cell count, platelet count, and levels of albumin were associated with the development of CA (**Supplement Table 1**). To avoid interaction of the above risk factors, we conducted multivariate logistic regression and found IVIG treatment (p = 0.008) as well as levels of albumin (p = 0.002) were the only two independent risk factors for the occurrence of coronary aneurysms.

Of the 37 KD patients, 6 (16.2%), 7 (18.9%), and 2 (5.4%), respectively, had coronary dilatation at 2, 4, and 12 weeks after disease onset (**Figure 1**, **Supplement Figure 1**). The most frequently affected coronary arteries was LMCA (n = 8, 80%), followed by RCA (n = 6, 60%) On the basis of the definition of CA in the current study, 7 (18.9%) patients had CA (small, n = 6, median, n = 1). Of the 31 dKD patients without initial CAL, four patients (4/31, 12.9%, **Figure 1**) had coronary progression (to coronary aneurysms) during the first month after KD onset. Of the KD patients without CAL at their acute phase (**Figure 1**), percentages of emerging coronary aneurysms (at one month after disease onset) became significantly higher if they didn't

have IVIG therapies due to spontaneous defervescence (4/31), compared with those who received IVIG (3/208, p = 0.006, **Figure 1**). Coronary aneurysms of the 4 patients regressed at 2.7, 6, 6.8, and 7.3 months after disease onset.

# Risk Factors Associated With Coronary Artery Lesions in Patients With Defervescence Between 5 and 10 Days

Previous studies have demonstrated that coronary severity at 1 month after KD onset was most crucial to the late coronary outcomes. In the current study, 7 patients (18.9%, M:F = 4:3) in the dKD group had coronary lesions at 1 month after fever onset (defined as coronary aneurysms in this study). **Table 2** presents the demographic symptoms and laboratory data of dKD patients with (7) and without (30) coronary aneurysms. Univariate analysis showed the development of coronary aneurysms in the dKD group was associated with age (p = 0.01) and serum white blood cell count (p = 0.03). The percentage of younger than 1 year of age in the 7 dKD patients with coronary aneurysms was 71.4% (5/7), significantly higher than that in the 30 dKD patients without (5/30, p = 0.009). The 6 dKD patients with CAL at their acute phase were relatively, not significantly, risky to have coronary aneurysms, compared with 31 dKD patients without (p = 0.07). Furthermore, 57% dKD patients with CA (4/7) had WBC count over 17.0 kµL, which was significantly more than the dKD patient without CA (4/30, p = 0.027).

To avoid interaction between age and WBC count, we conducted multivariate logistic regression to identify the independent risk factors of developing coronary aneurysms in the 37 dKD patients. The results showed that age [younger than 1 year, odds ratio: 24.6, 95% confidence interval (CI) = 1.5–399, p = 0.024] and serum white blood cell count more than 17.0 kµL (odds ratio: 16.1, 95% CI = 1.3–198, p = 0.031), rather than CAL at the acute phase, were both independently associated with the occurrence of coronary aneurysms in the dKD patients.

TABLE 2 | Comparison between spontaneous defervesced KD with and without coronary artery aneurysm 1 month after disease onset.


Values are expressed as medians (IQRs) or number and percentages (%); WBC, White blood cell; Seg, Neutrophil segment; CRP, C-reactive protein; AST, Aspartate aminotransferase; Hb, hemoglobin; PLT, platelet.

\*Statically significance is defined as p < 0.05.

# DISCUSSION

IVIG therapy remained the standard treatment for KD, which can effectively reduce the risk of CAL development. However, in the cases with spontaneous defervescence, treatment guidelines were still uncertain because of short of well-established data of their clinical outcomes. In this retrospective cohort, we found that 12.6% of KD patients had spontaneous defervescence within 10 days and 18.9% of such patients (dKD) suffered from coronary aneurysms, significantly higher than the percentages of coronary aneurysms in the IVIG group (5.1%, p = 0.002, **Table 1**, **Figure 1**). Such findings reinforce the importance of IVIG in the treatment of KD. We also demonstrated, in the KD patients without CAL at their acute phase, percentages of emerging coronary aneurysms (at 1 month after disease onset) became significantly higher if they didn't have IVIG therapies due to spontaneous defervescence (4/31), compared with those who received IVIG (3/208). Such observation had never been described before and drove us to look for the potential risk factors of developing coronary aneurysms in the dKD patients.

We demonstrated that, among the 37 dKD patients, younger age (<1 year old) and leukocytosis were associated with the development of coronary aneurysms. Downie et al. reported that males, age <1-year-old and higher platelet count were associated with increased odds of CA formation for patients with delayed or no treatment compared to KD patients with prompt IVIG (9). Younger KD patients were thought of being less likely to have a complete presentation of KD and also suffered from an increased risk of CAL (10–12). For example, Salgado et al. showed percentages of CAL and incomplete KD were significantly higher in the KD patients younger than 6 months old, compared with those were older. Moreover, 18.6% of patients <6 months old with normal echocardiogram initially developed CAL within 8 weeks of diagnosis (13). Our data presented similar findings and younger age remained an important risk factor of CAL even though the patient had spontaneous defervescence. Taken the above together, for the KD patients younger than 1 year old, we may recommend the use of IVIG even if they have defervescence without IVIG therapy because of increased risk of CAL.

Spontaneous defervescence without IVIG is not new for the KD patients. Depending on the definition of spontaneous defervescence, the percentages ranged from 7.3 to 20.9% (7, 14, 15). Takahashi et al enrolled patients who had spontaneously defervescence within 7 days, rather than 10 days as our cohort, and their results showed 7.3% of 968 KD patients had fever persistent < 7 days without an infusion of IVIG. If we applied the same enrollment criteria as Takahashi et al. (14), to our current cohort, the percentage of spontaneous defervescence would decrease from 12.6 to 10.5%. Spontaneous defervescence in KD patients usually causes hesitation of physicians and parents on the use of IVIG. However, Takahashi et al. showed 11.2% of KD patients with spontaneous defervescence within 7 days after disease onset may have recurrent fever 3–7 days later and predispose to develop coronary artery lesions (14). The current study further innovatively points out, even without initial CAL and recurrence of fever, such subgroup of KD patients still carried a significantly higher risk of late development of coronary aneurysms, compared with those had IVIG therapies (4/31 vs. 3/208, p = 0.006). Both age (younger than 1 year) and leukocytosis were independently associated with the development of coronary aneurysms in the dKD patients. In summary, it's time to refine the strategy for use of IVIG in the spontaneously defervesced KD patients within 10 days after fever onset, at least in those with age younger than 1 year and those with leukocytosis.

Studies in the 1990s (16, 17) demonstrated, in the KD patients without IVIG treatment, levels of TNF alpha and IL-2 continued increasing even at the second or third week after fever onset. Until 2–3 months later, levels of the two cytokines decreased gradually to the normal ranges. TNF alpha and IL-2 were two of the key inflammatory cytokines in the pathogenesis of KD and associated with recruitment of immune cell population, T cell activation, and CAL development (18). Together the above, for the KD patients with spontaneous resolution of fever within 10 days, delayed normalization or even increase in levels of cytokines, like TNF alpha and IL-2, may accompany higher risk to the development of coronary lesions.

There were several limitations in this study. First, there was selection bias especially when we enrolled the KD patients with atypical disease presentations. To minimize the bias, we used the evaluation algorithm for suspected incomplete KD according to AHA guideline (3) and traced their clinical presentation as well as laboratory data in the subacute phase to aid diagnosis. Second, the number of dKD patients was small which made the statistical analysis limited. Third, information bias might exist, since the ultrasound technicians were not blind to tentative or previous diagnosis. Finally, because of the observational nature of our study, large-scaled prospective studies are necessary to determine the efficacy and necessity of IVIG in dKD patients who had higher risk to develop coronary abnormalities.

# CONCLUSIONS

18.9% of KD patients with spontaneous defervescence had coronary aneurysms. Even without initial coronary lesions, such patients were still riskier to develop coronary aneurysms, compared with KD patients who received IVIG therapies. Such findings address the importance of refining the strategy for use of IVIG in the spontaneously defervesced KD patients within 10 days after fever onset, at least in those with age younger than 1 year and those with leukocytosis.

# AUTHOR CONTRIBUTIONS

M-TL and M-HW contributed to the conception and design of the study. H-ML and L-YC organized the database. Y-CH and M-TL performed the statistical analysis. Y-CH wrote the first draft of the manuscript. C-AC, S-NC, C-WL, and J-KW took care of patients and collected data. M-TL had full data access and is accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. All authors contributed to manuscript revision, read and approved the submitted version.

## SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fped. 2019.00158/full#supplementary-material

# REFERENCES


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Hu, Liu, Lin, Chen, Chiu, Lu, Chang, Wang and Wu. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# A Presumed Etiology of Kawasaki Disease Based on Epidemiological Comparison With Infectious or Immune-Mediated Diseases

Jung-Woo Rhim1,2, Hyun Mi Kang1,2, Ji-Whan Han<sup>1</sup> and Kyung-Yil Lee1,2 \*

<sup>1</sup> Department of Pediatrics, College of Medicine, The Catholic University of Korea, Seoul, South Korea, <sup>2</sup> Department of Pediatrics, Daejeon St. Mary's Hospital, The Catholic University of Korea, Daejeon, South Korea

Background: Kawasaki disease (KD) may be associated with infection of unknown pathogen(s). For predicting of the etiology of KD, we evaluated epidemiological characteristics in KD, common infectious diseases and immune-mediated diseases in childhood.

Edited by:

Claudio Pignata, University of Naples Federico II, Italy

#### Reviewed by:

Elma Inés Nievas, Independent Researcher, Mendoza, Argentina Piero Pavone, Università degli Studi di Catania, Italy

> \*Correspondence: Kyung-Yil Lee leekyungyil@catholic.ac.kr

#### Specialty section:

This article was submitted to Pediatric Immunology, a section of the journal Frontiers in Pediatrics

Received: 30 October 2018 Accepted: 30 April 2019 Published: 21 May 2019

#### Citation:

Rhim J-W, Kang HM, Han J-W and Lee K-Y (2019) A Presumed Etiology of Kawasaki Disease Based on Epidemiological Comparison With Infectious or Immune-Mediated Diseases. Front. Pediatr. 7:202. doi: 10.3389/fped.2019.00202 Methods: We respectively, reviewed the data of patients with KD, influenza, aseptic meningitis, exanthem subitum (ES), Mycoplasma pneumoniae (MP) pneumonia, acute pyelonephritis (APN), Henoch-Schönlein purpura (HSP), acute poststreptococcal glomerulonephritis (APSGN), and childhood asthma. We compared and interpreted epidemiological data across the groups.

Results: In age distribution, KD, APN, and ES showed a similar pattern in that majority of patients were infants or young children, and other diseases showed a relatively even age-distribution which had a peak age, mainly 5–6 years, with bell-shape patterns. In annual-case pattern, there were epidemic years in aseptic meningitis and MP pneumonia, and the fluctuated annual cases were seen in other diseases. The trends of decreasing cases were seen in APSGN, HSP, and childhood asthma in recent years. In seasonal frequency, influenza or aseptic meningitis occurred in mainly winter or summer season, respectively. HSP and APSGN cases had less in summer, and KD, APN, and ES showed relatively even occurrence throughout a year without significant seasonal variations.

Conclusions: Our results suggest that KD agents may be associated with normal flora that are influenced by environmental changes, since pathogens of APN and ES could be regarded as normal flora that originate from the host itself or ubiquitously existing human reservoirs.

Keywords: Kawasaki disease, etiology, epidemiology, acute pyelonephritis, exanthem subitum, microbiota

# INTRODUCTION

Kawasaki disease (KD) disease was a novel disease that appeared in East Asian countries in the order of Japan, South Korea, Taiwan, and China with a 5–10 year time-gap (1–4). After the first case report in the 1960s in Japan, KD has been reported in over 60 countries around the world (5). However, the incidence of KD is different across various populations; currently, the incidence in Western countries is lower by one tenth to one twentieth compared to East Asian countries, and has plateaued during recent decades. Whereas, the number of KD patients in East Asian countries has slowly and steadily increased after its first emergence (6). In these countries, KD occurs mostly in children between 6 months and 4 years of age, and has become a nationwide endemic disease within 2 decades after its first appearance. These findings suggest that KD spreads slowly to other regions within a nation and neighboring countries, which is contrasting to common infectious disease epidemics. KD pathogen(s) should satisfy with the epidemiological characteristics of KD.

The clinical manifestations of Kawasaki disease share similarities in some aspects of viral diseases, bacterial diseases, or infection-related immune mediated diseases such as systemic juvenile idiopathic rheumatoid arthritis or acute rheumatic fever (ARF). However, extensive studies searching for the etiologic agent(s) have turned up as failures until the present time (7). Common childhood infectious diseases and infection-related immune-mediated diseases, including KD, childhood idiopathic thrombocytopenia, and Henoch-Schönlein purpura (HSP) have low prevalence in older children and adults. Thus, these childhood diseases have been hypothesized to be associated with infections caused by unknown pathogens (8, 9). Since KD is an acute self-limiting systemic inflammation that involves multiple organs, it has been proposed that there are etiologic substances that induce systemic inflammation and target cell injuries, including coronary artery lesions (CALs), during or after the infection with KD pathogen (s) (10). Also, it has been proposed that KD pathogens may be a species of the host's normal flora based on clinical and epidemiological characteristics of KD (11). The epidemiological characteristics of a disease may aid in predicting the etiologic agent(s) of the disease. The identification of KD etiology is essential for understanding the disease, and the development of diagnostic tools and adequate treatment modalities.

In this study, to predict the etiology of KD, we evaluated and compared epidemiological characteristics between KD and common infectious diseases or immune-mediated diseases in children. After evaluating epidemiological parameters focused on age distribution, annual-case pattern and monthly-case pattern, we found that the epidemiological profiles in KD were similar to those in acute pyelonephritis (APN) or exanthema subitum (ES). We discussed the implications of these findings in KD epidemiology in Korea.

## MATERIALS AND METHODS

The subjects of this study were collected from the patients (0– 15 years of age) who were admitted at The Catholic University of Korea Daejeon St. Mary's Hospital from January 1987 to December 2016. We evaluated the epidemiological characteristics of patients diagnosed with Mycoplasma pneumoniae (MP) pneumonia, influenza, aseptic meningitis, APN, ES, acute poststreptococcal glomerulonephritis (APSGN), HSP, childhood asthma (or recurrent wheezing episode), and KD. As an exception, the subjects diagnosed with influenza in this study were selected from outpatients who were positive for influenza in rapid diagnostic testing during the winter of influenza seasons. The diagnosis or selection criteria in each disease were referenced from other publications (12–20). Although the study period and the number of patients in each disease were not identical, we reevaluated the data that were used for previously published papers or collected new data for some diseases such as ES and HSP. Age, sex, age distribution pattern, annual-case pattern and monthly-case pattern were analyzed in each disease. For mean age, <12 months of age was regarded as 0 years of age for statistical analyses, except ES. In age distribution pattern, we focused on age predilection in infancy and younger children. In annual-case pattern, we searched whether the pattern shown cyclic epidemics during the study periods. As for seasonal variation, the disease was regarded as having seasonality when the number of seasonal cases showed over 10 or 15% difference in the number of cases between the highest season and the lowest season.

# Ethics Statement

The written informed consents were obtained from the parents/caregivers of all children for the medical records to be used in this study at time of admission. The study was approved by the Institutional Review Board of The Catholic University of Korea Daejeon St. Mary's Hospital (DC18RESI0100).

# RESULTS

A total of 7,832 patients diagnosed with 9 diseases, including KD, 5 infectious diseases, and 3 immune-mediated diseases were evaluated. The study period in the majority of diseases was over 10 years; for infectious diseases, MP pneumonia (2003– 2012, n = 779) (12), influenza (2010–2017, n = 2,163) [(13, 14), unpublished observation], aseptic meningitis (1987–2003, n = 2,201) (15), APN (2005–2015, n = 320) (16) and ES (2005– 2016, n = 429), and as for infection-related immune diseases, APSGN (1987–2013, n = 99) (17), HSP (1987–2015, n = 515) (unpublished observation) and childhood asthma (or recurrent wheezing episode) (2003–2014, n = 384) (18), and KD (1987– 2016, n = 942) (19, 20) (**Table 1**).

## Age, Sex, and Age Distribution in KD and Other Diseases

The male-to-female (M:F) ratio, mean age, and peak ages are shown in **Table 1**.

Male predominance was observed in all the diseases except ES (0.95:1), but the M:F ratios were somewhat different across the diseases. The highest M:F ratio was seen in APSGN (2.3:1), and the lower M:F ratios were seen in ES, MP pneumonia and influenza (1.1:1, respectively), and KD showed 1.7:1 ratio in this series. The lowest mean age was noted in order of ES (mean 1 year of age), APN (1.5 years), and KD (2.2 years) and the higher mean age was noted in APSGN (8.3 years) and HSP (6.5 years). In age distribution, KD, APN, and ES showed a similar pattern in that the majority of patients were infants and young children (0–4 years), and after this age period the prevalence decreased


\*Outpatients; APSGN, acute poststreptoccocal glomerulonephritis.

dramatically. There were only a few patients >3 years of age in ES and a few patients >5 years in KD, however there was relatively an even distribution with female predominance after >2 years in APN. A bell-shape distribution pattern with a peak prevalence, mainly in the 4–6 years age range was observed in other diseases, however peak ages were slightly different across the diseases (**Figure 1**). There was a trend where the peak ages were changed in the epidemic diseases such as MP pneumonia and influenza over time [(12), unpublished observation].

# Annual-Case Patterns in KD and Other Diseases

Cyclic epidemics caused by possibly new strains of pathogens, such as macrolide-resistant strains were noted in MP pneumonia and in aseptic meningitis caused by mainly enteroviruses; all had similar patterns with nationwide epidemics or outbreaks in Korea (12, 15). Influenza occurred every winter and early spring, but the number of cases and peak epidemic months were different in 2009 pandemic and seasonal influenza, across seasonal influenza after 2009 pandemic [(14), unpublished observation]. Relatively an even, but fluctuating annual cases were noted in KD, APN, ES, HSP, and childhood asthma, except APSGN (**Figure 2**). The mean annual cases in each disease are shown **Table 1**, and the results may be helpful to estimate the incidence of each disease in our city, since nationwide KD epidemiological studies in Korea have been performed every 3 years (21, 22). There was a trend of decreasing number of cases in the recent years or decades in APSGN, HSP and childhood asthma, whereas an increasing trend was noted in APN, compared to the past years or decades (**Figure 2**).

# Monthly or Seasonal-Case Pattern in KD and Other Diseases

A higher number of cases were seen in the fall and winter in MP pneumonia. Aseptic meningitis was noted mainly during the summer whereas influenza mainly during winter and early spring seasons. There was a trend showing a lower number of cases in the summer in HSP, and a higher number of cases in the winter for APSGN. A higher number of cases were seen during the fall and spring for childhood asthma. Although KD affected more patients during summer than any other season in this series, KD had no difference in seasonal variation according to the definition of the seasonality. Also, no seasonality was seen in APN and ES (**Table 2** and **Figure 3**).

# DISCUSSION

It is hypothesized that KD may have an etiologic agent(s), although many studies searching for the etiology of KD have failed. Etiologic agents in infectious diseases originate from an external source and invade into the host. However, at present time, they could be classified as exogenous pathogens and the endogenous pathogens; the former is generally accepted as agents that are newly introduced to humans from other animal species or other places, whereas the latter originates from the host in the forms of normal flora (or commensals) or those becoming a normal flora from initially being an exogenous pathogen. For examples, measles in immune-innocent populations in the past and acquired immunodeficiency syndrome (AIDS) were caused by new external pathogens. On the other hand, the pathogens causing APN or acute otitis media, such as E. coli or Streptococcus pneumoniae, could be categorized as endogenous pathogens. It is well-known that a newly introduced disease by an exogenous pathogen has affected all aged persons in a population and spread to other populations in relatively a short-time period of time after its first emergence. Over time, infected persons and groups have immunity to the pathogens, and young children and infants would remain susceptible, especially during cyclic epidemic viral infections. Accordingly, epidemiologic characteristics of infectious diseases or infection-related diseases in the populations at a given time may differ.

In the present study, we found that epidemiological characteristics in KD were most similar to those in APN or ES among the 8 evaluated diseases. In age distribution, these 3 diseases had an age predilection in infancy and young children; 79.7% of patients in APN, 95.8% in ES, and 50.1% in KD were 0– 1 years of age, and 87.5% of patients in APN, near 100% in ES, and 91.7% in KD were 0–4 years of age. In the view of epidemiological and clinical aspects, an infected infant with pathogens from these diseases are difficult to disperse the pathogens to other infants. Other diseases showed a relatively even age-distribution

throughout childhood, though peak ages were slightly different across the diseases. This suggests that younger children may have less chance on exposure to the pathogens and older children and young adult groups may have immunity or tolerability to the pathogens causing the diseases.

In the annual-case pattern, cyclic epidemics were noted in MP pneumonia and aseptic meningitis, and marked fluctuated cases with a different peak week were noted in influenza. The annual-case pattern in KD was relatively even, but fluctuations in annual cases were noted as well as in APSGN, HSP and childhood

#### TABLE 2 | Seasonality in KD and other diseases.


APSGN, acute poststreptococcal glomerulonephritis.

asthma, and APN and ES in infectious diseases. APN may be the most common systemic bacterial infection in early childhood in developed countries (23). A majority of uropathogens causing APN in infants are E. coli or other uropathogens such as Klebsiella spp., Proteus spp., and Enterococcus spp., and they may originate from the intestinal commensals in the host. It is possible that after colonization of a strain of E. coli or other uropathoegns transferred from other persons, the pathogens invade into the host on occasional events and elicit immune reactions (24). Although KD and APN affect mainly infants and young children, older children and adults are also affected (25). Moreover, recurrent cases are not uncommon in the both diseases. These findings suggest that immature immune function in early childhood may be associated with both diseases, and pathogens may be multiple in KD (11). ES is an acute systemic viral infection, and is characterized by sudden appearances of

generalized maculopapular or morbilliform rashes just after defervescence. The etiologic agent is a species belonging to the human Herpes virus group, herpes virus type 6 or rarely type 7. Herpes viruses have a characteristic of latent infections after initial infection. ES may have no differences in age predilection, possibly sero-prevalence rate, and clinical manifestations across populations around the world and have occurred throughout every year with no seasonal variations (26, 27). Herpes virus groups become latent in the host after the initial infection as shown in herpes zoster, herpes labialis, and reactivation of cytomegalovirus and Epstein-Barr virus (EBV) in depressed immune state of the host (28). It was reported that EBV infection or herpes virus 6 infection was related to certain clinical manifestations in KD such as otorrhea or BCG inoculation site inflammation (29, 30). It is a reasonable presumption that etiologic viruses in ES may have been introduced to humans a long-time ago, and may be coexisting within healthy human reservoirs including family members of infected infants with ES, and can be activated at any time. On occasion, the latent viruses in healthy carriers who may have a transient immune disturbance, can be reactivated and be spread to infants who have no immunity to the viruses.

In the present study, monthly case or seasonal case patterns in KD were also the most similar to those in APN or ES. Similar to this study, Nagao et al. (31) reported that the superannual periodicity of KD was the most similar to ES among seven childhood infectious diseases including ES, herpangina, hand-foot-mouth disease, chicken pox, pharyngoconjunctival fever, erythema infectiosum, and GAS infection. They suggested that the KD pathogen is transmitted through close contact and persists asymptomatically in most hosts, likely ES (31), and this suggestion is accordance with our presumed characteristics of KD pathogens. Although the rate of KD has slowly increased with slight seasonal predominance in nationwide studies in Japan and Korea (22, 32), KD has occurred with annual fluctuations with seasonal variation each year in a given location as shown in this study. These data suggest that KD appears as local outbreaks rather than as a simultaneous nationwide epidemic as shown in viral or MP infections in Korea.

We have proposed that KD pathogens may be a species of normal flora of the host (10, 11). The disturbance of microbiota of the host, i.e., dysbiosis, has been reported in a variety of diseases, including obesity, autism spectrum disorders, allergic diseases, cancers, and autoimmune diseases such as inflammatory bowel disease, JIA and KD (33–38). It has been known that microbiota differ across ethnic groups along with different cultural environments such as diets, antibiotic use, and possibly genetic factors (39–41). Kinumaki et al. reported that Streptococcus spp. in intestinal microbiota markedly increased during the acute phase of KD, and suggested that KD-related streptococci may be involved in the pathogenesis of KD (38). Recently, Esposito et al. reviewed the possible role of the microbiota-host partnership on etiology and pathogenesis of KD (42). Because normal flora spread via colonization in individuals in populations, a new disease that is associated with normal flora may take a long period of time during which to spread, and would show different incidences across populations at given times as shown in KD epidemiology in Asian countries. Since pathogens exist in healthy human reservoirs as normal flora, the diseases may occur mainly in immunologically or genetically susceptible groups such as young children group that have developing immune system and microbiota (43, 44). Also, the epidemiology in these diseases can change over time along with an increasing number of people who obtain pathogens as normal flora in the populations, since normal flora that have adapted to the host may be less virulent compared to initial external pathogens. For example, scarlet fever caused by Group A betahemolytic streptococcus (GAS) was a severe disease in the past, with two notorious complications: acute rheumatic fever (ARF) and APSGN. Now, scarlet fever, ARF and APSGN have become rare diseases with a milder clinical phenotype in the developed countries (17, 45). However, pharyngitis caused by GAS without complications is not uncommon, and a relatively high proportion of GAS carriers among healthy children have been reported in the developed countries (46, 47). In addition, GAS strains have been susceptible to penicillin throughout the antibiotic era, with few genetic variations (48). Given that the evolutional purpose of external pathogens may be to become a species of normal flora in the host, these findings suggest that GAS strains may be changing to be part of normal flora in humans, though ARF still occurs in small populations (49). Thus, it is natural that initially severe diseases take on milder phenotypes over time, as shown in scarlet fever, pandemic influenza, and AIDS (13, 45, 50). It was also reported that recently diagnosed KD patients showed a less severe clinical phenotype, manifesting a higher incidence of incomplete KD and a lower incidence of CALs and less activated laboratory values such as C-reactive protein and albumin compared to patients who developed the disease in earlier periods (51).

Pathogenesis of KD and APN as well as other infectious and immune-mediated diseases remains to be further studied. APN has a focus in renal cortical parenchyma, where replicated uropathogens, byproducts from APN agents such as toxins and pathogen-associated molecular patterns (PAMPs), substances from injured host cells, and those from activated immune cells are produced. When these diverse substances spread from the focus into systemic circulation, clinical manifestations such as fever, tissue cell injury and/or bacteremia begin to occur, and the host immune system may respond to the substances. The majority of patients infected with APN pathogens may be asymptomatic and their disease are self-limited if systemic spread did not occur and the substances in the focus were controlled as a localized inflammation (24). On the other hand, ARF or APSGN are classic representatives of infection-related immune-mediated disease. In ARF, after 1–4 weeks after initial GAS infection when symptoms and signs of the initial bacterial infection (pharyngitis) subside, acute onset of fever and other symptoms, such as carditis, arthritis, and rarely skin rash (erythema marginatum), develop (49). The majority of patients with ARF or APSGN show a self-limited clinical course, although severely affected patients have long-term complications such as severe rheumatic heart disease or chronic renal failure, similar to giant coronary artery aneurysms in KD. Only a small proportion of patients with GAS infection are affected with ARF or APSGN. Some patients with ARF or APSGN may have preceding asymptomatic GAS infections, and other group streptococci, other bacteria or viruses are associated with postinfectious glomerulonephritis or heart tissue inflammation, including myocarditis and possibly heart valve diseases, without evidence of GAS infection (52, 53). It is believed that there are etiologic substances that induce various clinical manifestations in KD as well as in ARF and APSGN after initial infection (10). These substances include not only those originating from the pathogens but also those from the injured host cells, including damage-associated molecular patterns (DAMPs) and those from immune cells activated by infectious insults. It have been proposed that these substances bind to specific target cells of organs such as heart and joints in ARF, kidneys in APSGN, or coronary arteries or other organs in KD. The host immune system may control these substances based on their size and biochemical properties (the proteinhomeostasis-system hypothesis) (10, 54, 55).

Rhim et al. A Presumed Etiology of KD

Based on the data from epidemiological and clinical studies of KD, the pathogenesis of KD could be explained as follows. Along with economic growth and westernization in East Asian countries in order of Japan, South Korea, Taiwan, and China, a species in normal flora (microbiota) in people may be affected by environmental changes such as improved hygiene or increased consumption of westernized foods including red meats (10, 11). These species (KD agents) slowly spread via colonization as normal flora to other people, including the parents and guardians of KD patients, and are eventually colonized in infants and young children predisposed to KD. On occasion, colonized KD agents invade the host via unknown events; if they are intestinal commensals, a mechanism similar to APN (via hematogeneous route in young infants) may be activated (24). Invading KD pathogens make a focus (replication site) elsewhere within the host, possibly near the upper respiratory or intestinal tracts (portal of entry) or in the target organ via systemic or local circulation. The majority of patients infected by the KD agents may be asymptomatic and have self-limited disease. However, during or after recovery from this infection, similar to ARF or APSGN, KD develops when the substances from a focus spread abruptly into systemic circulation (10). There are various incubation periods in ARF and APSGN after initial GAS infection, and no intact pathogens or structural components of the pathogen have been detected in the pathologic lesions in ARF, APSGN, KD, and other infection-related immune-mediated diseases. Furthermore, GAS strains can reside in intracellular compartments in host cells such as tonsillar epithelial cells and macrophages (56, 57), and small extracellular bacteria such as mycoplasma species can proliferate within cells and spread to other organ cells (58). Therefore, it is possible that the etiologic substances may originate from injured host cells, including a kind of DAMPs, or from immune cells exposed to infectious insults, than from various pathogens. Immune/repair system of the host, including immunoglobulins (IgG, IgM, and IgM) and platelets, begin to control the substances and take part in tissue cell repair (59). Thus, the prognoses of KD and other immunemediated diseases depend on the ability of the host immune system to control the diverse substances from target cells injured by initial insults. Early immune modulators (corticosteroids or intravenous immunoglobulin) may act on hyperactive immune reactions performed by the non-specific adaptive immune cells (T cells and B cells) in the early stage of each disease (10, 54, 60). However, despite early treatment, a few patients with KD experience severe or giant aneurysms, and these patients may have insufficient immune function to control the substances originating from injured target cells (61). Young children with immature immune function may be prone to invasion of KD

#### REFERENCES


pathogens as occurs in APN in infants, and the probability of invasion of KD agents in young children may be similar across different populations (20). Since microbiota may differ across populations, marked differences in racial incidence in KD or other immune-mediated diseases such as HSP, inflammatory bowel disease, and Bechet disease may be associated with the colonization state of the different etiologic agents in the populations (20).

This retrospective study has some limitations. Epidemiological data observed in a single hospital may not match the nationwide data. However, we found that outbreaks of infectious diseases such as MP pneumonia, influenza, and aseptic meningitis in our city had occurred concurrently with nationwide epidemics. Demographic and some epidemiologic characteristics of KD, APN, APSGN, HSP, childhood asthma, and possibly ES were similar to the data in published papers in Korea (data not shown).

In conclusion, epidemiological characteristics of KD were similar to APN or ES in age distribution, annual case pattern, and seasonal variations. The pathogens of APN and ES could be regarded as one species of normal flora, since they may have originated from the host itself or ubiquitously existing human reservoirs. KD may also be associated with a species in the normal flora that may be influenced by environmental changes.

## ETHICS STATEMENT

The written informed consents were obtained from the parents/caregivers of all children for the medical records to be used in this study at time of admission. The study was approved by the Institutional Review Board of The Catholic University of Korea Daejeon St. Mary's Hospital (DC18RESI0100).

## AUTHOR CONTRIBUTIONS

K-YL designed the study, collected data, contributed to interpretation of results, and drafted the manuscript. J-WR contributed to data collection and drafted the initial manuscript. HK and J-WH contributed to data collection and revised the manuscript. All authors have read and approved submission of the final version of the manuscript.

#### ACKNOWLEDGMENTS

The authors would like to acknowledge our colleges on data collection during the study periods of each disease.


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Rhim, Kang, Han and Lee. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Biomarkers for Kawasaki Disease: Clinical Utility and the Challenges Ahead

Himanshi Chaudhary, Johnson Nameirakpam, Rajni Kumrah, Vignesh Pandiarajan, Deepti Suri, Amit Rawat and Surjit Singh\*

*Post Graduate Institute of Medical Education and Research, Chandigarh, India*

#### Edited by:

*Kyung-Yil Lee, The Catholic University of Korea, South Korea*

#### Reviewed by:

*Diego F. Coutinho, New York University, United States Hongryang Kil, Chungnam National University, South Korea*

> \*Correspondence: *Surjit Singh surjitsinghpgi@rediffmail.com*

#### Specialty section:

*This article was submitted to Pediatric Immunology, a section of the journal Frontiers in Pediatrics*

Received: *16 January 2019* Accepted: *28 May 2019* Published: *18 June 2019*

#### Citation:

*Chaudhary H, Nameirakpam J, Kumrah R, Pandiarajan V, Suri D, Rawat A and Singh S (2019) Biomarkers for Kawasaki Disease: Clinical Utility and the Challenges Ahead. Front. Pediatr. 7:242. doi: 10.3389/fped.2019.00242* Kawasaki disease (KD) has replaced acute rheumatic fever as the most common cause of acquired heart disease in children in the developed world and is increasingly being recognized from several developing countries. It is a systemic vasculitis with a predilection for coronary arteries. The diagnosis is based on a constellation of clinical findings that appear in a temporal sequence. Quite understandably, this can become a problem in situations wherein the clinical features are not typical. In such situations, it can be very difficult, if not impossible, to arrive at a diagnosis. Several biomarkers have been recognized in children with acute KD but none of these has reasonably high sensitivity and specificity in predicting the course of the illness. A line up of inflammatory, proteomic, gene expression and micro-RNA based biomarkers has been studied in association with KD. The commonly used inflammatory markers e.g. erythrocyte sedimentation rate (ESR), C-reactive protein (CRP), and total leucocyte counts (TLC) lack specificity for KD. Proteomic studies are based on the identification of specific proteins in serum, plasma and urine by gel electrophoresis. A host of genetic studies have identified genes associated with KD and some of these genes can predict the course and coronary outcomes in the affected individuals. Most of these tests are in the early stages of their development and some of these can predict the course, propensity to develop coronary artery sequelae, intravenous immunoglobulin (IVIg) resistance and the severity of the illness in a patient. Development of clinical criteria based on these tests will improve our diagnostic acumen and aid in early identification and prevention of cardiovascular complications.

Keywords: immunology, vasculitis, Kawasaki, biomarkers, childhood vasculitis

# INTRODUCTION

Kawasaki disease (KD) is a common childhood vasculitis. The disease was first described in Japanese children in 1967 by a Japanese pediatrician, Dr. Tomisaku Kawasaki (1). The highest incidence is seen in Japan, Korea and Taiwan (2). However, the disease is now being reported world over including several developing countries like India (3–7).

The diagnosis of KD is essentially clinical (8). The clinical features mimic many self-resolving exanthematous febrile illnesses of childhood (e.g., measles, adenoviral infection, scarlet fever, dengue fever). These illnesses share some common clinical features with KD like fever, rash, mucocutaneous manifestations, lymphadenopathy, and elevated inflammatory parameters. The diagnosis of KD can be easily missed if the cascade of clinical findings goes unrecognized especially in cases of incomplete KD. Things can get even more complicated when KD occurs in association with an infection. The associated coronary artery abnormalities (CAAs) may go undetected and can have significant long-term implications if treatment is not initiated at the right time. Unlike other vasculitides and other rheumatological disorders, there are no pathognomonic laboratory tests for diagnosis of this condition. For the treating clinician, it is often difficult to confirm a diagnosis of KD on the bedside. It is, therefore, important to establish a set of laboratory markers that are sensitive, specific, reproducible and help the treating pediatrician in arriving at a diagnosis. As intravenous immunoglobulin (IVIg) is an expensive product, it is imperative that it be used only in situations where it is definitely indicated. However, in a setting of incomplete KD, the pediatrician often faces the dilemma of under-treatment with its attendant risks of CAAs vs. using IVIg in circumstances where it may not really be indicated.

Several biomarkers have been studied in association with KD and some of these have been shown to be predictive of resistance to IVIg while others may be indicative of an increased risk of development of CAAs. This manuscript overviews some of the important biomarkers that have been studied in KD and highlights their role in the diagnosis and assessment of disease severity.

## INFLAMMATORY BIOMARKERS (TABLE 1)

These are the conventional biomarkers that mirror inflammatory activity and are not specific to KD. Erythrocyte sedimentation rate (ESR) is consistently elevated during the acute phase of KD but may be unreliable as a marker of disease activity after the administration of IVIG (9–11). There is neutrophilic leukocytosis during the acute phase of the disease and the degree of leukocytosis has been correlated with myocardial dysfunction (10, 12). Thrombocytosis, usually seen after the completion of the first week of illness, is a marker of ongoing inflammation. Persistent thrombocytosis has been linked to the development of CAAs but the association is tenuous (9, 18). C-reactive protein (CRP) is known to have a significant association with disease severity and the development of CAAs (9, 10, 12). Procalcitonin levels have been shown to be elevated during the acute phase and this rise is especially marked in children with resistance to IVIg (13, 16). Peripheral blood eosinophilia (PBE) (17) and low albumin has been associated with increased risk of IVIG resistance and coronary complications (14). Low mean platelet volume (MPV) (15), platelet distribution width (PDW) (19), and platelet-derived microparticles (PDMP) (20) have been shown to be markers of platelet activation and inflammation in acute stages of KD. However, their use as biomarkers for the disease requires replication of results across different populations.

#### Summary

Inflammatory markers are largely nonspecific as these are also elevated in many other inflammatory and infective conditions. In a clinical setting of KD, while these biomarkers can reflect ongoing inflammation, they are of limited use in arriving at a definitive diagnosis.

#### IMMUNOLOGICAL MARKERS (TABLE 2)

#### Cellular Markers Innate Immunity

Myeloid dendritic cells (mDCs), which serve as a bridge between adaptive and innate immunity, have been shown to be decreased in acute KD. Takahashi et al. (21) have suggested that decreased levels are due to either recirculation into affected tissues or

TABLE 1 | Inflammatory biomarkers of Kawasaki disease.


TABLE 2 | Immunological biomarkers of Kawasaki disease.


increased peripheral destruction in cases of acute KD. Furukawa et al. (22) studied CD14+ macrophages in KD patients and showed that they were higher in those with CAAs, thereby suggesting that absolute counts of CD14+ monocytes can be a marker of the severity of KD. These studies point toward a possible dysfunction in the innate immunity axis which could contribute to the inflammatory upregulation seen in acute stages of KD.

#### Acquired Immunity

Studies have shown a decreased number of CD8 T cells during acute stages of KD (23). Immunohistochemistry studies in coronary arteries of KD patients at autopsy have shown that CD8 T cells preferentially sequester in the coronary arteries and are responsible for the inflammatory vasculitis (24, 26). Ehara et al. (27) showed that markers of early T cell activation [CD69(+)CD8T cells] increased in acute stages and can be used as a marker of disease progression and response to IVIg. Helper T cells (Th1 and Th2) have been shown to be upregulated during acute stages (22, 25). There is an apparent imbalance of T helper 17 cells (Th17) and regulatory T cells (Tregs) in acute KD. Th17 proportions have been shown to be upregulated while Treg proportions were found to be downregulated during the acute phase of KD (33).

#### Soluble Markers

Soluble markers of inflammation have been studied at great length in association with KD. Plasma levels of Th1 (IFN-Y, IL-12) and Th2 (IL-4, IL-13) cytokines have been shown to be elevated during acute stages of KD (34). Multiple pro-inflammatory and anti-inflammatory cytokines have been studied in acute stages of KD, but none of these has been standardized as a biomarker of KD (29). Zhou et al. (28) showed a close relation between levels of vascular endothelial growth factor (VEGF), IL-6, and development of CAAs. TGF-β signaling has been implicated in the development of coronary artery aneurysms (29). Th1-associated (CXCL10) and Th2 associated chemokines (CCL2) are elevated in acute stages of KD and have been shown to decrease with IVIG (35, 36). Tumor Necrosis Factor α (TNFα) has a role in the recruitment of inflammatory cells to coronary endothelium and has been shown to have a role in the development of CAAs. TNFα blocking agents have been extensively studied as a therapy for KD (37).

#### **Summary**

Data on immunological markers in KD are derived from studies on relatively small groups of patients. The significance of these biomarkers in predicting the disease course, response to therapy and complications is not clear and needs confirmation across a different population.

# Proteomic Biomarkers (Table 3)

Several proteins have been studied in KD.

#### N-terminal Prohormone of Brain Natriuretic Peptide (NT-proBNP)

NT-proBNP has been used as a marker of myocardial damage in KD. It is a marker of cardiomyocyte stress imposed by pressure or volume overload and is increased in response to cardiac dilatation and neuro-humoral factors. According to Dahdah et al.

#### TABLE 3 | Proteomic biomarkers of Kawasaki disease.



(31), myocardial involvement in acute KD is universal from the histological and functional perspectives and hence the role of NTproBNP as a potential biomarker has been extensively studied. The interpretation of serum NT-proBNP levels in children is difficult because these are age-dependent, being highest in infancy and decreasing thereafter. The sensitivity and specificity of cut-off values as a biomarker for KD are derived from receiver operating characteristics (ROC) analysis, the upper normal limit for age, and Z-scores for age. Shiraishi et al. (32) showed that the sensitivity and specificity of diagnosing acute KD were 97.8 and 47%, respectively, at a z score > 2. A meta-analysis on NTproBNP in KD has substantiated its use as a diagnostic marker and cut-off values between 96 and 260 pg/ml have been shown to have sensitivity between 66 and 98% in identifying cases of KD (30). NT-proBNP levels are higher in patients with CAAs (values 515–1,300 pg/ml have a sensitivity and specificity of 73–95 and 61–85%, respectively) (38) and can predict IVIG resistance. A value between 629 and 1,300 pg/ml has a high sensitivity (70– 79%) as well as specificity (58–77%) for diagnosis of KD (39, 40). From our center, we have reported a cut-off at 1,025 pg/mL for NT-proBNP levels which has a sensitivity of 88% and specificity of 96% (41) in the acute phase of KD.

#### Other Cardiovascular Biomarkers

Suppression of tumorigenicity-2 (sST2) is a member of the interleukin 1 (IL-1) receptor family and reflects cardiovascular stress and fibrosis. It is elevated in acute stages of KD and the levels correlate with impaired myocardial relaxation. However, the prognostic significance in acute KD is unclear (42).

Kim et al. (43) have described cardiac troponin I (cTnI) in relation with KD and showed a significant increase in the level of cTnI in the acute stage of KD. Periostin is a matricellular protein that mediates responses associated with cardiovascular injury. Reindel et al. (44) have shown that periostin gets upregulated in coronary arteries during the acute and chronic phases of KD when compared to other febrile controls. Gamma-glutamyl transferase (GGT) and alanine transferase (ALT) have also been studied as biomarkers of cardiac inflammation. However, these are rather nonspecific and are not a reliable marker for KD per se (42).

#### Thrombospondin (TSP-1 and TSP-2)

TSP-1 and TSP-2 are proteins involved in cardiovascular inflammation and maintaining the integrity and function of cardiac structures. These have been shown to be elevated in the acute phase of KD in comparison with other febrile controls and higher values are seen in those with IVIg resistance. With a cutoff value of 31.50 ng/mL, the sensitivity of TSP-2 has been shown to be 82.35% and specificity 64.81% in predicting IVIG resistance in acute KD (45).

#### Other Proteins

Clusterin is a part of the high-density lipoproteins (HDL) and has a role in many physiologic processes including maintaining the integrity of coronary artery walls. Plasma clusterin levels have been studied in KD and values lower than 12 mg/L have been associated with the occurrence of CAAs in KD patients (46). Yu et al. (46, 55) have studied several proteins of fibrinogen cascade in KD. It was found that these were increased in patients with KD and may serve as a good biomarker of KD. The expression of nitric oxide synthases (iNOS) has been shown to correlate with the extent of coronary damage and progression of CAAs in patients with acute KD (47). Tenascin- C (TN-C) is a marker of tissue injury and inflammation and has a role in the maintenance of the extracellular matrix of cardiac tissue. Okuma et al. (48) have studied serum TN-C level in association with the risk of developing CAAs and resistance to IVIg therapy in the acute phase of KD.

#### **Summary**

Several protein biomarkers have been studied in relation to KD but most of these studies are based in small cohorts of patients at a single center. NT-proBNP is widely believed to be a useful marker for confirmation of the diagnosis of KD at the bedside. The levels of NT-proBNP in KD, however, may overlap those in other febrile illnesses with cardiac dysfunction. Other biomarkers are still in their early stages of discovery and will require validation in larger populations before the results can be used in clinical settings.

#### Urine Biomarkers

Kentsis et al. (49) studied urine protein biomarkers in relation with KD and showed an abundance of markers of tissue injury (filamin and talin), complement regulator (CSMD3), cytokine protease (meprin A), and immune pattern recognition receptor (muclin) in the urine of affected patients. However, these results need replication in larger studies before these can be used as noninvasive biomarkers in KD.

# GENETIC STUDIES IN KD (TABLE 4)

A genetic basis of KD has been strongly considered taking into account the geographical differences in the incidence and high risk of occurrence in family members. Incidence of KD in Japan, Korea and Taiwan is more than 10 times the incidence in Western countries (50–53). This difference is a pointer toward a possible genetic susceptibility or may reflect environmental or lifestyle differences amongst these populations. A higher incidence is also known amongst Japanese Americans settled in Hawaii and this is comparable with the incidence in Japan, further pointing toward a possible genetic association of the disease (75). Studies have shown that the risk of developing KD in siblings of a KD patient is 10–30-fold higher as compared to the general population (76). Two types of approaches have been utilized for studying the genetic basis of KD:


## Candidate Gene Approach

Candidate gene studies are the well-known approach to study genes associated with KD based on their function in the pathophysiology of the disease.

#### TABLE 4 | Genetic biomarkers of Kawasaki disease.


#### HLA Genes

Early genetic studies on KD were focused on HLA alleles. HLA class 1 antigens have been studied in great detail in various populations (54, 77). Significant predominant association of HLA-Bw54 has been found in Japanese KD population while single nucleotide polymorphisms (SNP) located in the HLA-E gene was suggested to be associated with KD in the Han Chinese population (77). A positive association was found between HLA-Bw51 and KD in the White and Jewish population while HLA-Bw51 was reported to be negatively associated in the Korean population. In a recent GWAS, association with HLA class II antigens peaked at the intergenic region between HLA-DQB2 and HLA-DOB (78). SNPs in HLA class III region have also been described in association with KD (79). TNF∞ expression has been shown to be elevated in association with CAAs in KD in Korean, Taiwanese, Chinese and Caucasian populations, but these results could not be replicated in studies from Japan (80).

#### Non-HLA Genes

Burns et al. (78) found an SNP in the promoter region of the IL-4 gene to be asymmetrically transmitted in children with KD. MHC-class-I-chain-related gene A (MICA) alleles were reported as biomarkers for evaluation of coronary aneurysms in KD. Lower frequency of MICA allele 276 A4 was reported in KD patients with aneurysms (81).

# Genome-Wide Approach

#### Genome-Wide Linkage Analyses (GWLS)

GWLS were used to map the genetic loci of diseases by analyses of related individuals through the logarithm of odds (LOD) score. The first GWLS on KD was done by Onouchi et al. (82) who studied 79 families of children with KD and analyzed 75 sibling pairs. They identified 10 chromosomal loci with positive linkage, among which the 12q24 region showed the most significant evidence of linkage. GWLS studies identified Inositol 1,4,5 trisphosphate 3-kinase C (ITPKC) gene in association with risk of KD (83).

#### Genome-Wide Association Studies (GWAS)

GWAS are based on the analysis of a genome-wide set of genetic variants typically SNP in affected individuals and healthy controls to see if a variant is repeatedly associated with a disease. GWAS studies on KD have identified several genes in association with the disease. The significant susceptibility genes identified include caspase-3 (CASP3) (69), calcium releaseactivated calcium modulator 1 (ORAI1) (59), ATP-binding cassette sub-family C member 4 gene(ABCC4) (84), Fc Fragment of IgG Receptor IIa (FCGR2A) (66), Transforming growth factor β pathway (TGFB2, TGFBR2, and SMAD3) (85), B lymphocyte kinase (BLK) (63), matrix metalloproteinase 11 (MMP-11) (74), and CD40 (86).

#### Genes Associated With B-Cell Signaling CD40

CD40L–CD40 interaction is known in relation to triggering and progression of acute coronary syndromes (87). Higher CD40 ligand expression on CD4 T-cells of KD patients as compared to febrile controls has been reported by a previous study. This over-expression decreased after IVIg administration (88). SNPs around CD40 showed association with KD susceptibility in Japanese (rs4813003, located 4.9 Kb downstream) and Taiwanese (rs1569723, located 4.8 Kb upstream) population. GWAS from Taiwan and Japan have also shown the association of SNPs of BLK and CD40 in the pathogenesis of KD (66).

#### B Lymphoid Tyrosine Kinase (BLK)

BLK has a role in signal transduction of B cells (64). GWAS from Taiwan and Japan have shown the association of SNPs of BLK and CD40 in the pathogenesis of KD (68). Kawasaki Disease Genetics Consortium (2013) confirmed the association of BLK with KD susceptibility in Korean and European population (89).

#### Fc Fragment of IgG Receptor IIa (FCGR2A)

FCGR2A is present on Fc region of IgG and encodes for cell surface receptor protein on phagocytic cells. GWAS done in five independent centers have confirmed the association of FCGR2A with KD susceptibility (66). Associations of SNPs rs2736340, rs4813003, rs3818298, and rs1801274 with KD were reported in a Han Chinese population (60, 66).

## T-Cell Activation Genes

#### Inositol 1,4,5-Trisphosphate 3-kinase (ITPK) Enzyme Gene

ITPKC is involved in Ca2+/NFAT pathway and acts as a potential candidate for immune activation and T-cell receptor signaling (68, 83). Association of functional SNP (rs28493229) in ITPKC with development of KD and CAA risk was first reported in Japanese and American children (83). However, no significant association was found between rs28493229 and KD risk in Taiwanese patients (90). Significant associations of C-allele of rs28493229 with KD and BCG scar reactivation in the acute stage were reported by Lin et al. (56).

#### Calcium Release-Activated Calcium Channel Protein 1 (ORAI1)

ORAI1 is a Ca2<sup>+</sup> channel protein involved in T cell regulation and inflammatory responses. Onouchi et al. (59) showed a significant non-synonymous association of SNP (rs3741596) in exon 2 of the ORAI1 gene with a high risk of KD in Japan. Another study reported a rare variant (rs141919534) in association with KD (91).

# Genes Associated With the Apoptotic Pathway

#### Caspase 3 (CASP3)

CASP3 gene is an execution-phase caspase involved in apoptosis of immune cells. Various SNPs including rs2720378, rs72689236, and rs113420705 have been reported in association with KD (69)**.** Kuo et al. found that SNP rs28493229 (ITPKC) and rs113420705 (CASP3) showed an increased risk of IVIG resistance and development of CAAs in Japanese but not in Taiwanese population (92).

#### TGF-β Signaling Pathway

The TGF-β pathway is an essential component of inflammation, T-cell activation and tissue remodeling. Genetic variations in TGF-β2, TGF-βR2, and SMAD3 are associated with CAA formation and was confirmed in replication studies (85). SNPrs6550004 in TGF-βR2 and rs1495592 in SMAD5 showed significant associations with KD in the Korean population (93).

## Gene Expression Studies

MicroRNAs (miRNAs) are 20–26 nucleotides long non-coding single-stranded RNA segments that modify post-transcriptional mRNA gene expression. Shimizu et al. (94) showed 6 miRNAs (miR-143, miR-199b-5p, miR-618, miR-223, miR-145, and miR-145) that were differentially expressed in acute KD. In another study, miRNA-200c and miRNA-371-5p have also been studied as diagnostic biomarkers of KD (54). miR-27b suppresses endothelial cell proliferation and has been studied as a therapeutic target in patients of acute KD (95).

• In a recent multicenter study by Wright et al. (96), the use of whole-blood gene expression patterns was explored and a 13 transcript blood gene expression signature has been developed to distinguish KD from other infectious and inflammatory febrile illnesses in the first week of illness.

• Whole genome sequencing in a family with KD has also shown genetic variations in the toll-like receptor-6 (TLR6) gene in their two affected children (97).

**Summary:** Most of the genetic studies have been carried out in small cohorts of patients with KD and the results are not reproducible across different populations and ethnicities. These studies require validations in larger and multinational cohorts with additional case-control sets for a better understanding of the genetic profiles.

# CONCLUSIONS

The need for a robust set of biochemical biomarkers to validate the diagnosis of KD in the clinical setting has become the need of the hour. Clinical application of these biomarkers is limited. Inflammatory parameters can, at best, facilitate confirmation of a clinical diagnosis of KD but none of these is pathognomonic of KD. Further, these markers have very low specificity for the diagnosis of KD. The newer proteomic studies

#### REFERENCES


have identified some biomarkers in association with KD but these also need validation across different populations before these can be used for confirming a diagnosis of KD. Genomewide studies, linkage studies and miRNA-based biomarkers are still in their early stage of development and fall short of being a diagnostic test for this enigmatic condition. These genetic markers are pointers toward a diagnostic panel for KD but clearly, these are early days and much more work needs to be done before a robust laboratory diagnostic test can be established.

#### AUTHOR CONTRIBUTIONS

HC, JN, RK: Literature review, interpretation of data and draft of the manuscript. VP: critical review and editing of the manuscript. DS: critical review of studies cited in the manuscript, editing of manuscript. AR: concept and design of the manuscript, critical review of studies cited in the manuscript, editing of manuscript. SS: concept and design of the manuscript, critical review, and editing of the manuscript.


disease susceptibility. PLoS ONE. (2013) 8:e72037. doi: 10.1371/journal.pone. 0072037


susceptibility and formation of coronary artery aneurysms. Nat Genet. (2008) 40:35–42. doi: 10.1038/ng.2007.59


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Chaudhary, Nameirakpam, Kumrah, Pandiarajan, Suri, Rawat and Singh. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Aetiological Significance of Infectious Stimuli in Kawasaki Disease

#### Akihiro Nakamura<sup>1</sup> \*, Kazuyuki Ikeda<sup>2</sup> and Kenji Hamaoka3,4

 *Central Research Laboratory, Graduate School of Medical Science, Kyoto Prefectural University of Medicine, Kyoto, Japan, Department of Pediatrics, Graduate School of Medical Science, Kyoto Prefectural University of Medicine, Kyoto, Japan, Pediatric Cardiology and Kawasaki Disease Center, Uji-Tokushukai Medical Center, Kyoto, Japan, <sup>4</sup> Faculty of Life and Medical Sciences, Doshisha University, Kyoto, Japan*

Kawasaki disease (KD) is a pediatric vasculitis syndrome that is often involves coronary artery lesions (e. g., coronary artery aneurysms). Although its causal factors and entire pathogenesis remain elusive, the available evidence indicates that the pathogenesis of KD is closely associated with dysregulation of immune responses to various viruses or microbes. In this short review, we address several essential aspects of the etiology of KD with respect to the immune response to infectious stimuli: 1) the role of viral infections, 2) the role of bacterial infections and the superantigen hypothesis, 3) involvement of innate immune response including pathogens/microbe-associated molecular patterns and complement pathways, and 4) the influence of genetic background on the response to infectious stimuli. Based on the clinical and experimental evidence, we discuss the possibility that a wide range of microbes and viruses could cause KD through common and distinct immune processes.

#### Edited by:

*Hiromichi Hamada, Tokyo Women's Medical University Yachiyo Medical Center, Japan*

#### Reviewed by:

*Ying-Hsien Huang, Kaohsiung Chang Gung Memorial Hospital, Taiwan Hiroyuki Suzuki, Wakayama Medical University, Japan*

> \*Correspondence: *Akihiro Nakamura nakam993@koto.kpu-m.ac.jp*

#### Specialty section:

*This article was submitted to Pediatric Immunology, a section of the journal Frontiers in Pediatrics*

Received: *31 December 2018* Accepted: *29 May 2019* Published: *28 June 2019*

#### Citation:

*Nakamura A, Ikeda K and Hamaoka K (2019) Aetiological Significance of Infectious Stimuli in Kawasaki Disease. Front. Pediatr. 7:244. doi: 10.3389/fped.2019.00244* Keywords: Kawasaki disease, vasculitis, animal models, infection, superantigens, pattern recognition receptors

# INTRODUCTION

Kawasaki disease (KD), named after its discoverer Dr. Tomisaku Kawasaki, is a pediatric vasculitis syndrome which is characterized by clinical manifestations including fever persisting for 5 days or more, swelling of the cervical lymph nodes, conjunctival infection, changes in oral mucosa and the tongue, skin rash, and redness of the palms and soles of the feet (1, 2). Although KD shows a systemic vascular inflammation, the coronary arteries are one of the worst affected sites. Without adequate treatment in the acute phase, approximately 30% of patients exhibit coronary artery lesions (CALs) including coronary arterial dilation, stenosis, and aneurysms (3).

Treatment of KD typically features intravenous immunoglobulin (IVIG) therapy (4). In fact, IVIG has markedly decreased the mortality rate in patients with acute KD. However, a persisting concern is that the disease may impair cardiovascular health in adults with a history of KD. Furthermore, approximately 20% of acute KD patients show a low response to IVIG (5). The therapeutic resistance also results in an increased risk of CALs and future cardiovascular events.

The aetiological mechanism of KD remains unclear and the causal factors are also unknown (6). Although there is no definitive evidence that KD is an infectious disease, recent studies support the view that a dysregulated immune response to a variety of infectious stimuli is likely to contribute to KD pathogenesis (6, 7). Based on these studies, this short review explores the possible relationship between KD and the immune response to various infectious agents (**Figure 1**).

**59**


examinations of clinical specimens. The asterisks indicate experimental evidence from animal models, not clinical specimens.

# INVOLVEMENT OF VIRUSES IN KD PATHOGENESIS

The incidence of KD exhibits seasonality and outbreaks occurred in Japan in 1979, 1982, and 1986 (8, 9). This has led to the speculation that viral infection may underlie KD pathogenesis. Based on serological and polymerase chain reaction (PCR) basedanalyses of clinical specimens, at least 14 species of the virus have been reported to be relevant to KD (10). We consider here three possible candidates: Epstein-Barr virus (EBV), human adenovirus, and a putative KD-associated RNA virus.

# Epstein-Barr Virus

EBV is a type of human herpes virus. Kikuta et al. (11, 12) reported that the EBV DNA sequence was detected in 83% of KD patients and in 18% of control subjects. Chronic active EBV infection sometimes involves CALs, including coronary artery aneurysms (13, 14). Although the pathogenesis of EBVinfection-associated CALs is unclear, it has been demonstrated in vitro that deoxyuridine 5'-triphosphate nucleotidohydrolase (dUTPase), an EBV-encoded protein, stimulates monocytederived macrophages through Toll-like receptor 2 (TLR2) dependent signaling (15). This up-regulates the production of interleukin-6 (IL-6) (15), which activates endothelial cells (ECs) and platelets. Contrary evidence has also been presented on the involvement of EBV in KD (16, 17).

#### Adenovirus

It has been known that adenoviral infection exhibits seasonal pattern and some symptoms similar to that of KD (18, 19). However, semi-quantitative PCR-based investigations found no association between adenovirus and KD (20). Gene microarray of the blood samples also showed the distinct pattern between KD and adenovirus-infected patients (21). Further study is needed to clarify the involvement of adenoviral infection in KD (22).

# Putative RNA-Associated Virus

Immunoglobulin A (IgA)-producing plasma cells are observed in the affected arterial tissue of KD patients (23, 24). Rowley et al. (25, 26) detected RNA virus-like inclusion bodies in the cytoplasm of bronchoepithelial cells of KD patients. It was detected using synthetic antibodies generated from the alpha and kappa chain-encoding DNA sequences, which were cloned from the affected arterial tissue of KD patients (27). However, the putative KD-associated RNA virus has not been identified yet.

Previous studies suggest that the involvement of viruses in KD is still very controversial. However, a recent transcriptomic study reported the significant up-regulation of a set of type I interferon (INF)-induced genes closely related to cellular antiviral processes in the coronary arteries of KD patients (28). Furthermore, the increased plasma level of C-X-C motif chemokine ligand 10 (CXCL10/IP-10), a representative INFalpha2a/gamma-inducible protein was recently reported as a promising biomarker for the early acute phase of KD (29). These two studies raise the possibility that KD pathogenesis might be associated with a common immune response to viral infections.

# BACTERIAL INFECTION AND SUPERANTIGENS HYPOTHESIS

Superantigens (SAgs) are a group of proteins, which can activate approximately 20% of the T cells in the peripheral blood. SAgs stimulate these cells by forming a bridge between the Tcell receptor and the major histocompatibility complex class II (MHC-II) of antigen presenting cells in the absence of any antigenic peptide presentation (30, 31). This results in the overproduction of pro-inflammatory cytokines, including tumor necrosis factor-alpha (TNF-α), by the activated T cells (32). Human MHC-class II and co-stimulators are also expressed in the endothelial cells (33). In fact, SAgs can directly injure these cells in conjunction with T cells (34, 35) in vitro.

SAgs are produced by a wide range of organisms, including bacteria, viruses, fungi, and plants (36–38). The pathological role of SAgs has been well-studied with respect to toxic shock syndrome and scarlet fever, which are caused by Staphylococcus aureus and Streptococcus pyogenes, respectively (39, 40). KD displays some clinical similarities to these two diseases (41).

# Staphylococcus aureus and Streptococcus pyogenes

Staphylococcus aureus produces a SAg designated TTS toxin-1 (TSST-1), which induces expansion of Vβ2 T cell receptor (TCR)-positive T cells. Some early studies reported the frequent detection of TSST-1-producing S. aureus or the specific antibody to the SAg in KD-patients (42, 43). As ECs express MHC-II and co-stimulators, TSST-1 can activate human umbilical vein ECs in vitro in the presence of T cells (34). Considering that specific types of vessels (e.g., coronary arteries), are preferentially affected in KD., it might be worth investigating the pathophysiological significance of the role of ECs as a semi-professional antigenpresenting cells in KD (44, 45).

Streptococcal pyrogenic exotoxin C (SPEC) selectively activates Vβ2-bearing T-cell subsets and Vβ6.5-bearing ones (46). The number of Vβ2-/Vβ6.5-bearing T-cell subsets and anti-SPEC antibody levels are increased in the peripheral blood of acute KD patients (47). DNA fragments encoding these SAgs were also significantly more prevalent in the stool of KD patients than in that of non-KD febrile subjects (48), indicating that these bacteria-derived SAgs are involved in KD pathogenesis. Streptococcus pyogenes infection occasionally triggers autoimmune heart diseases, such as rheumatic fever. The serum level of IgM-type autoantibody to the cardiac myosin heavy chain, which is highly homologous to group A streptococcal M5 protein (49), was increased in KD patients (50). A variety of autoantibodies has also been detected in KD patients (51–53). Besides molecular mimicry, SAgs could be implicated in the activation of potentially autoreactive peripheral T and B cells (54).

Although the physiological role of circulating follicular T helper cells (cTfh) remains elusive, the cells could be stimulated by SAgs and/or other pathogen-derived molecules. Xu et al. (55) reported that IL-4 and the cTfh2 subpopulation of total cTfh cells significantly increased during the acute phase of KD. Increases in cTfh2 and IL-4 are observed in IgA-vasculitis and IgG4-related disease (IgG4-RD) (56–58). Although the general clinical features of IgG4-RD are apparently dissimilar to those of KD, IgG4-RD occasionally involves abnormalities of large- and intermediate-sized vessels, including coronary arteritis (59).

Despite the foregoing evidence, the involvement of SAgs in KD has not been definitively confirmed. Some studies have independently found no significant elevation of antibodies against S. aureus or S. pyogenes-derived SAgs in KD patients (60, 61). Although these bacterial-derived SAgs may provide a rationale to understand the pathogenesis of KD, its aetiological significance is still very controversial.

## Yersinia pseudotuberculosis

Based on the similarity in clinical manifestations and the data from some serological studies, it has been argued that Y. pseudotuberculosis (YP) infections might be involved in the pathogenesis of KD (62). Some species of YP produce YPderived mitogens (YPM), which are SAg-like virulence factors (63). Consistent with the increased prevalence of KD in Far East countries, YPM-positive pathogenic YP are also predominantly distributed in these countries and are less frequent in western countries (64). However, in a recent clinical study involving 108 Japanese KD patients, it was found that 90% of patients were negative for antibodies to YP and YPM (65). Vβ2-bearing Tcell subsets be preferentially activated in KD, whereas Vβ3, Vβ9, Vβ13.1, or Vβ13.2-bearing T-cell subsets are activated in YP infection (66).

## Mycobacterium

Reactivation of Bacillus Calmette-Guérin (BCG) scar is a wellestablished clinical manifestation in acute KD, indicating that Mycobacteria or immunologically related pathogens might be involved in KD. Although Mycobacteria have not been isolated from KD patients as a causal pathogen, antibody and CD4+/CD8<sup>+</sup> T-cell clones specific to Mycobacterium heat shock protein 65 have been detected in KD patients (67, 68).

Besides M. tuberculosis, M. leprae, and M. lepromatosis, non-tuberculous mycobacteria (NTM) can cause selflimited infections in humans. M. avium complex (MAC) is a representative NTM. The immune response to MAC infections involves peculiar macrophages, which are characterized by the co-expression of anti-inflammatory M2 macrophage markers (e.g., CD163, IL10) and markers of pro-inflammatory M1 macrophages (e.g., CCR7, IL1β) (69). These cells promote Th17 cell-differentiation (69). Intriguingly, the coronary arteries affected in acute KD also often feature marked infiltration of macrophages, which are negative for CD80 (an M1 marker), and positive for CD163 (an M2 marker) (70). Plasma level of Th17-related cytokine sets are also increased (71). Considering these recent findings and BCG scar reactivation in KD, revisiting the possible implication of mycobacteria in KD could be warranted.

## Fungi

Although there is no definitive clinical evidence suggesting that fungi are a causal factor for KD, it has been well-established that Candida albicans extracts develop KD-like experimental model of vasculitis in mice. Based on the meteorological and enviromentological study, Rodo et al. (72) recently reported that KD is associated with tropospheric winds containing Candida species as a predominant fungus. Similar investigation in Canada also suggested the implication of westerly wind-associated fungi in KD (73).

It might be possible to validate phlogogenic activity of the wind-blown microbes / molecules with established animal models for KD.

## Possible Triggers and Diagnostic Criteria of Kawasaki Disease

Except Y. pseudotuberculosis infection (65, 74, 75), few cases satisfy the six diagnostic criteria for Kawasaki disease (KD) among the infectious diseases caused by the aforementioned agents. Considering that genetic background is closely associated with the susceptibility to KD (see section Influence of Genetic Background Affecting Response to Infectious Stimuli), polymorphisms in some specific genes of infected children might affect the clinical futures of the infectious diseases (76). Moreover, it is not exclusive that unidentified agents could be involved in the onset of KD. Caution should be exercised when considering the possible causal agents on the basis of the symptom similarity.

#### AETIOLOGICAL SIGNIFICANCE OF INNATE IMMUNE RESPONSE IN KD

While innate immunity is the first line of self-defense against infectious agents, it is also accompanied by inflammatory reaction (77). Thus, its inadequate regulation gives rise to inflammatory tissue damage. Accumulating evidence indicates that KD could be associated with dysregulated innate immune response.

# Insight From Experimental Studies With Animal Model for KD

Apart from SAgs, a variety of microbes or virus-derived biomolecules (e.g., lipopolysaccharides, glucans, and nucleotides) can stimulate immune cells or other cell types, including ECs. These biomolecules are designated pathogen-associated molecular patterns (PAMPs) and microbe-associated molecular patterns (MAMPs). PAMPs and MAMPs are recognized by a type of cellular or soluble receptors, which are designated pattern recognition receptors (PRRs) (e.g., TLRs) to trigger innate immune responses, including the production of inflammatory cytokines through intracellular signaling pathways. A possible implication of these PAMPs or MAMPs in KD has been demonstrated in mouse models and more recently in clinical specimens from KD patients (78, 79).

KD-like vasculitis can be induced in rabbit, swine, and mouse models by various methods (80–83). The known genetic background and ease of genetic manipulation have made mouse models the preferred model to investigate molecular pathogenesis of KD and to discover its therapeutic targets (84–88).

In a murine model of vasculitis, Candida albicans watersoluble fraction (CAWS) is used as the inducer. Marked inflammatory change is observed in the aortic root, including the coronary arteries in CAWS-treated mice (89, 90). While CCR2 and GM-CSF play an indispensable role in its pathogenesis (89, 90), T and B cells are also involved in the vasculitis (89). Sensitivity and severity of CAWS-induced vasculitis apparently depend on mouse strain, suggesting that genetic background affects its pathogenesis (also see section Influence of Genetic Background Affecting Response to Infectious Stimuli).

Another KD-like murine coronary arteritis model involves induction by Lactobacillus casei cell wall extract (LCWE). TLR2 dependent signaling, ILβ-dependent signaling, and CD8<sup>+</sup> T cells (cytotoxic T cells) play a crucial role in its pathogenesis (91–93).

KD-like coronary arteritis can also be induced in mice by NOD1 ligand, FK565 (94, 95), a synthetic derivative of acylpeptide produced by Streptomyces olivaceogriseus (96). The experimental model of vasculitis reportedly underlies the interaction between the CCR2 expressed in monocytes and the CCL2 induced by the NOD1-dependent signaling in EC (78).

It is important to consider whether these models accurately represent the actual molecular mechanisms of KD (97). However, some evidences indicate that the experimental murine vasculitis is relevant to KD pathogenesis. The antibodies to β-glucan, another major CAWS component, was increased in KD patients (98). Regarding findings in CAWS-induced or FK565-induced mouse model, it was also reported that genetic polymorphisms in the CCR3-CCR2-CCR5 gene cluster are associated with KD susceptibility (99). Furthermore, it has been found that lipophilic substances almost identical to MAMPs derived from Bacillus cereus, Bacillus subtilis, Y. pseudotuberculosis, and S. aureus, have been detected in the serum of KD patients (79). This finding supports the possibility that KD underlies the molecular pathogenesis similar to the aforementioned mouse models. The possible involvement of Y. pseudotuberculosis and S. aureus in KD has been argued based on other clinical evidence (also see sections Staphylococcus aureus and Streptococcus pyogenes and Yersinia pseudotuberculosis). The collective findings indicate the possible relationship between PAMPs/MAMPs and the aetiological mechanism of KD vasculitis.

# Other Insight to PAMPs/MAMPs Hypothesis in KD

Regarding the implication of PRRs, Huang et al. (100) reported that the CpG sites of TLR genes were reversibly hypomethylated in the peripheral whole blood cells of acute KD patients, resulting in upregulated expression of TLRs. This transient epigenetic change supposedly potentiates the TLR-dependent innate immune response. DNA demethylation in immune cells is also observed in patients infected with mycobacteria and certain viruses (101, 102), possibly implicating these intracellular-living pathogens in KD.

While killer immunoglobulin-like receptor (KIR) expressed in natural killer cells interacts with HLA-I to suppress activation of NK cells, KIR3DL1/2, a subtype of KIR, also recognizes and KIR3DL2 engulfs pathogen-derived CpG-oligodeoxynucleotide (CpG-ODN) to activate TLR9-dependent signaling (103). A recent case-control study of HLA-A and -B genotypes in Caucasian KD patients proposed the hypothesis that high abundance of HLA ligands for KIR3DL1/2 in KD patients interferes with KIR-dependent cellular CpG-ODN sensing to impair effective clearance of pathogens during infection (104). Consequently this impaired clearance might allow expansion of PAMPs/MAMPs. Although accumulating evidence suggests that PAMPs/MAMPs are involved in KD pathogenesis, their pathophysiological significance may be more complex than is currently appreciated.

# Possible Implication of Complement Pathways

Three complement pathways (classical, lectin, and alternative pathway) are major components of the innate immune system against infectious agents. However, their excess activation causes inflammatory tissue injury. Indeed, accumulating evidence suggests that their dysregulated activation underlies the pathogenesis of vascular inflammation and aortic aneurysms (105, 106). Nevertheless, a limited number of studies had been undertaken regarding the involvement of complement pathways in KD (107, 108).

Recently Okuzaki et al. (109) found that ficolin-1, a circulating soluble PRR that is responsible for activating the lectin pathway,

was increased in acute KD. They also demonstrated that KDlike murine vasculitis is ameliorated by infusion of an inhibitory antibody to ficolin-1 (110), suggesting that the lectin pathway could participate in KD pathogenesis. The lectin pathway is triggered by activation of mannose-binding lectin-associated serine proteinases (MASPs). In addition to their role in the complement system, MASPs are involved in coagulation and EC activation (111), both of which are closely connected to KD.

Complement systems are rigorously regulated by more than ten protein species to prevent their undesirable excess activation. Genetic polymorphisms of these proteins might affect susceptibility of KD under infectious condition (106, 112).

# INFLUENCE OF GENETIC BACKGROUND AFFECTING RESPONSE TO INFECTIOUS STIMULI

While a variety of microorganisms could be causative agents of KD, the prevalence of KD in children is potentially limited, suggesting that the genetic background of an individual likely affects the disease susceptibility. A series of clinical genetic investigations have revealed KD-related gene polymorphisms in more than 20 genes (113), although its aetiological significance is elucidated only in a limited number of these polymorphisms.

Statistical genetic studies identified an SNP associated with KD susceptibility and resistance to IVIG therapy in the gene encoding inositol-triphosphate 3-kinase C (ITPKC), which suppresses T-cell activation and regulates inflammasome activity in macrophages (114, 115). This SNP destabilizes ITPKC mRNA and reduces the cellular level of ITPKC protein (115). KDassociated gene polymorphisms have been discovered in the CASP3 gene, which might also influence the down-regulation of activated immune cells (116). In addition, Onouchi et al. (114) reported KD-associated polymorphisms in the gene encoding ORAI1, a calcium release-activated calcium channel protein 1. Notably, these proteins are also involved in the calciumdependent nuclear factor-activated T cell (NFAT) pathway (117, 118).which regulates the function of T and B cells. EBV-encoded latent membrane protein 1 (LMP-1) up-regulates ORAI1 (119). Considering that EBV is a possible trigger for KD (see section Epstein-Barr Virus), it is intriguing whether the identified genetic polymorphisms could affect immune response to EBV infection.

The collective molecular genetic data indicate that most genes with KD-associated polymorphisms are responsible for the modulation of inflammatory responses, including T-cell activation, which is compatible with the postulated roles of SAg and PAMPs/MAMPs. This strengthens the possibility that the dysregulated immune response to infectious stimuli underlies the pathogenesis of KD.

Induced pluripotent stem cell technique has been successfully used to establish some human EC lines harboring genetic backgrounds of KD patients (120). Studies with these cells enable the verification of the aetiological significance of individual genetic backgrounds in KD in vitro.

## CONCLUSION

Although the etiology of KD is far from being resolved, the evidence collected so far drives the following hypothesis (**Figure 2**): Diverse pathogens could be potential causative agents of KD. However, such different infectious stimuli converge on a similar/common immune process associated with the activation of T cells, innate immune cells, and ECs. The genetic background of infected children affects the magnitude of the immune responses to develop KD in a limited number of children.

# AUTHOR CONTRIBUTIONS

The manuscript was written by AN, and it was edited by KI and KH.

#### REFERENCES


susceptibility to Kawasaki disease. Clin Exp Immunol. (2007) 150:83–90. doi: 10.1111/j.1365-2249.2007.03457.x


Kawasaki disease. Lancet. (1994) 343:1260–1. doi: 10.1016/S0140-6736(94) 92154-7


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Nakamura, Ikeda and Hamaoka. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# A Metagenomics Study on Hirschsprung's Disease Associated Enterocolitis: Biodiversity and Gut Microbial Homeostasis Depend on Resection Length and Patient's Clinical History

Alessio Pini Prato1,2 \*, Casey Bartow-McKenney <sup>3</sup> , Kelly Hudspeth4,5, Manuela Mosconi <sup>2</sup> , Valentina Rossi <sup>2</sup> , Stefano Avanzini <sup>2</sup> , Maria G. Faticato2,6, Isabella Ceccherini <sup>7</sup> , Francesca Lantieri <sup>8</sup> , Girolamo Mattioli 2,6, Denise Larson<sup>9</sup> , William Pavan<sup>9</sup> , Carlotta De Filippo<sup>10</sup>, Monica Di Paola<sup>11</sup>, Domenico Mavilio4,5 and Duccio Cavalieri <sup>11</sup> \*

<sup>1</sup> Division of Pediatric Surgery, AON SS Antonio e Biagio e Cesare Arrigo, Alessandria, Italy, <sup>2</sup> Division of Pediatric Surgery, Giannina Gaslini Institute, Genoa, Italy, <sup>3</sup> Department of Dermatology and Microbiology, University of Pennsylvania,

#### Edited by:

Miika Kaleva Arvonen, Kuopio University Hospital, Finland

#### Reviewed by:

Dipankar Ghosh, Jawaharlal Nehru University, India Andrea Quagliariello, Bambino Gesù Ospedale Pediatrico, Italy

#### \*Correspondence:

Alessio Pini Prato apini@ospedale.al.it Duccio Cavalieri duccio.cavalieri@unifi.it; cavalieri.unifi@gmail.com

#### Specialty section:

This article was submitted to Pediatric Infectious Diseases, a section of the journal Frontiers in Pediatrics

Received: 31 December 2018 Accepted: 19 July 2019 Published: 09 August 2019

#### Citation:

Pini Prato A, Bartow-McKenney C, Hudspeth K, Mosconi M, Rossi V, Avanzini S, Faticato MG, Ceccherini I, Lantieri F, Mattioli G, Larson D, Pavan W, De Filippo C, Di Paola M, Mavilio D and Cavalieri D (2019) A Metagenomics Study on Hirschsprung's Disease Associated Enterocolitis: Biodiversity and Gut Microbial Homeostasis Depend on Resection Length and Patient's Clinical History. Front. Pediatr. 7:326. doi: 10.3389/fped.2019.00326 Philadelphia, PA, United States, <sup>4</sup> Unit of Clinical and Experimental Immunology, Humanitas Clinical and Research Center, Milan, Italy, <sup>5</sup> Department of Medical Biotechnologies and Translational Medicine (BioMeTra), University of Milan, Milan, Italy, <sup>6</sup> Department of Neuroscience, Rehabilitation, Ophthalmology, Genetics and Maternal and Child Science (DINOGMI), University of Genoa, Genoa, Italy, <sup>7</sup> UOC Medical Genetics, Giannina Gaslini Institute, Genoa, Italy, <sup>8</sup> Biostatistics Section, Department of Health Science, University of Genoa, Genoa, Italy, <sup>9</sup> Genomics, Development and Disease Section, National Human Genome Research Institute (NHGRI), National Institutes of Health (NIH), Bethesda, MD, United States, <sup>10</sup> Institute of Agriculture Biology and Biotechnology, National Research Council, Pisa, Italy, <sup>11</sup> Department, of Biology, University of Florence, Firenze, Italy

Objectives: Since 2010, several researches demonstrated that microbiota dynamics correlate and can even predispose to Hirschsprung (HSCR) associated enterocolitis (HAEC). This study aims at assessing the structure of the microbiota of HSCR patients in relation to extent of aganglionosis and HAEC status.

Methods: All consecutive HSCR patients admitted to Gaslini Institute (Genova, Italy) between May 2012 and November 2014 were enrolled. Institutional review board (IRB) approval was obtained. Stools were sampled and 16S rDNA V3-V4 regions were sequenced using the Illumina-MiSeq. Taxonomy assignments were performed using QIIME RDP. Alpha diversity indexes were analyzed by Shannon and Simpson Indexes, and Phylogenetic Diversity.

Results: We enrolled 20 patients. Male to female ratio was 4:1. Six patients suffered from Total Colonic Aganglionosis (TCSA). Considering sample site (i.e., extent of aganglionosis), we confirmed the known relationship between sample site and both biodiversity and composition of intestinal microbiota. Patients with TCSA showed lower biodiversity and increased Proteobacteria/Bacteroidetes relative abundance ratio. When addressing biodiversity, composition and dynamics of TCSA patients we could not find any significant relationship with regard to HAEC occurrences.

Conclusions: The composition of HAEC predisposing microbiota is specific to each patient. We could confirm that total colon resections can change the composition of

**68**

intestinal microbiota and to dramatically reduce microbial diversity. The subsequent reduction of system robustness could expose TCSA patients to environmental microbes that might not be part of the normal microbiota. Future long-term studies should investigate both patients and their family environment, as well as their disease history.

Keywords: metagenomics, enterocolitis, Hirschsprung, RET gene, aganglionosis

# INTRODUCTION

Enterocolitis (HAEC) is an extremely serious, life-threatening complication that can occur in children with Hirschsprung disease (HSCR) pre- and even post-operatively. Despite a number of studies, the causative agents of HAEC are still elusive. Standard culturomics technologies did not lead to the discovery of microbial pathogens causing HAEC. Novel culture independent approaches based on DNA sequencing of target genes or of the whole bacterial DNA content hold the promise to discover the causative agents and the etiology of HAEC. Since 2006 (1) metagenomics in children has been used to address the role of intestinal microbiota in the etiology of a number of diseases (1–11). Only a minority of papers addressed HSCR and HAEC (6, 12–14). In 2010, our group carried out a pilot genomic study in a single patient with HSCR and demonstrated that HAEC can correlate to changes in gut bacterial dynamics (12). Ward and co-workers later on reported a sustained abnormal microbiota in an animal model of HSCR (13). Yan in 2014 (6) and Frykman in 2015 (14) investigated HSCR patients with or without HAEC. Both authors did not find any significant difference apart from the latter who reported an altered Candida community in those with HAEC (14). Of note, Li et al. in 2016 reported a significantly different microbiome in HSCR patients with HAEC (15). Proteobacteria were significantly more represented in patients with HAEC whereas Bacteroidetes were significantly more represented in patients without. Also HSCR patients had a relatively distinct, more stable community than the HAEC and HAEC-R patients (previously settled HAEC episode), suggesting that enterocolitis may either be caused by or result in a disruption of the patient's uniquely adapted intestinal flora. The intestinal microbiota associated with enterocolitis may persist following symptom resolution and can be implicated in symptom recurrence. In another recent study on a mouse model of HSCR, the metagenomics analysis of Ednrb−/<sup>−</sup> and wild type mice showed that mutants had a distinct microbiota with respect to wild type (WT) and that the HAEC group had lower alpha diversity by Chao1 index compared with WT. Also the animals with HAEC had increased proportion of Akkermansia genus and reduced Bacteroidetes phylum compared with the NO HAEC and WT groups, suggesting Akkermansia may contribute to development of enterocolitis while Bacteroidetes may be protective (16). Finally, another study on Finnish patients showed how those with HD and HAEC had a significantly altered intestinal microbiome compared to healthy individuals, characterized by a lack of richness and pathologic expansions of taxa, particularly Enterobacteria and Bacilli (17). These initial studies did not lead to a predictive profile for HAEC (18, 19). Nonetheless, no mention was done either regarding extent of aganglionosis or diversity measures (15). Here we report results of a metagenomics study on fecal microbiota performed on HSCR patients, addressing limitations, drawbacks and potential benefits of such approach, delineating future perspectives in this field of research.

# METHODS

# Patients

All pediatric patients with HSCR consecutively admitted to Giannina Gaslini Institute (Genova, Italy) between May 2012 and November 2014 were eligible for this study. Institutional Ethical committee approval was obtained by the Review Board of Giannina Gaslini Institute on November 2009 as part of a wider research project on HSCR. A specific informed consent was signed by all participant families. Inclusion criteria were: (1) diagnosis of HSCR based on histochemical assessment of adequate rectal suction biopsies, as previously reported (20); (2) exhaustive data regarding extent of aganglionosis (adequate intraoperative histology) and regarding personal history with specific regard to previous bouts of HAEC; (3) stool sampling available both from preoperative and postoperative periods. Exclusion criteria were (1) refusal of signing the informed consent; (2) failure to pass internal quality control; (3) inadequate sampling or storage.

## Definitions

HSCR, Hirschsprung's disease; RSA, HSCR with aganglionosis extended up to the colonic splenic flexure (i.e., Rectosigmoid Aganglionosis); L-HSCR, Long HSCR with aganglionosis extended beyond the splenic flexure up to ascending colon; TCSA, HSCR with aganglionosis extended to the whole colon (i.e., Total Colonic Aganglionosis); HAEC, Enterocolitis diagnosed according to the combination of Pastor criteria (21) (to confirm the diagnosis) and Elhalaby criteria (22) (to grade HAEC severity).

#### Stools Collection, Storage, and Delivery

Spontaneous stools were collected and stored frozen (−20◦C up to 48 h and −80◦C afterwards) until shipment to the reference center for processing of fecal samples (Bethesda, Maryland, USA). Stools from Total Colonic Aganglionosis (TCSA) patients belonged to stoma bags preoperatively and from direct bowel movements postoperatively (in both cases from the ileum) whereas stools from Rectosigmoid Aganglionosis (RSA) or long (L)-HSCR belonged from bowel nursing (enema or rectal tube) preoperatively and from direct bowel movements postoperatively. Time-points for stool sampling were: (1) before surgery (Timepoint 1, preoperative hospital stay), (2) intraoperative (timepoint 2), and (3) postoperatively 7 to 10 after pull-through (timepoint 3). Only timepoints 1 and 3 (preoperative and postoperative, respectively) were assessed in this study.

#### Clinical Features and Molecular Genetics

All patients included in this study underwent a thorough phenotype assessment as previously published (23). Demographic data, phenotype results, HAEC status, surgical details, possible complications and long-term outcome were recorded and stored in a digital database according to data protection Act. Furthermore, all patients underwent sequencing of the RET gene coding portion (21 exons flanked by at least 20 bp of intronic sequences) as part of the multidisciplinary diagnostic algorithm, as previously published (24). In order to identify mutations that are potentially involved in HSCR, the results of molecular genetics, consisting of (1) putatively pathogenic RET mutations; (2) common variants among which those known to represent HSCR risk modulating factors (25), were recorded in the same database as above.

# 16S rDNA Sequencing and Processing

Amplification and sequencing of the 16S rDNA hypervariable V4 region was performed as previously described using the Illumina MiSeq platform with 150 bp paired end reads and V2 chemistry (26). Sequences were demultiplexed with FLEXBAR (27) and assembled using PEAR (28). Sequences were then quality filtered and restricted to length of 245–255 nt, which retained 9.77 million reads of the filtered 9.78 million reads. The sequences were then analyzed using scripts within the QIIME package (29). Sequences were first clustered into de novo operational taxonomic units (OTUs) defined by 97% sequence identity using UCLUST (30). A representative set of sequences for the OTUs was selected using the QIIME "most abundant" selection method. Taxonomy assignments for the representative sequences were performed using the QIIME RDP wrapper with an 80% cut-off for bootstrap confidence in assignment (31). Unclassified sequences and reads classified as Cyanobacteria were removed. The representative sequences were then aligned using PyNAST with a minimum length of alignment cutoff of 150 nucleotides and a minimum percent identity cutoff value of 75% (32). Alignments were performed against the Greengenes 13\_8 taxonomy core-set alignment sequences (33). Chimeras were removed from the aligned reads using ChimeraSlayer (34). ChimeraSlayer identified 21,593 OTUs (4.0%) out of 537,481 total OTUs as chimeric. OTUs that possessed <2 sequences and did not occur in more than one sample were removed. Taxonomy summaries were further filtered to only include OTUs that make up at least 0.5% of the total sequences. Rarefaction of the samples was performed on all samples at a depth of 7070 sequences to ensure homogeneity in sample size for downstream analysis. Phylogenetic tree construction of the aligned sequences was performed using FastTree (35). The phylogenetic tree was used for calculation of UniFrac (weighted and unweighted) beta diversity distance matrices (36).

# Statistical Analyses

The results were compared according to length of aganglionosis, timepoints, and HAEC status (HAEC episodes either experienced during the study period or before enrollment but reported in personal history (pre-HEAC). The single patient with L-HSCR could not be categorized as classic or ultralong HSCR and was excluded from this statistical analysis in order to avoid misinterpretation (**Table 1**). Diversity indexes as well as phylogenetic analysis were compared in patients with RSA and TCSA regardless of timepoints. TCSA patients that represented the core of our study underwent a further analysis based on HAEC status and timepoints of stool sampling. The R ggplot2 package was used for plotting (37). Alpha diversity index analyses of the samples were conducted using QIIME and the constructed phylogenetic tree to calculate the Shannon index (38), Simpson Index (39), and Phylogenetic Diversity (PD) (40). Pairwise comparison with Wilcoxon rank sum test was used to address Alpha diversity measures. Differences in categorical variables were addressed with chisquare or Fisher exact test, when appropriate. All tests were 2-tailed. A p-value lower than 0.05 was considered as statistically significant.

# RESULTS

# Overall Samples Distribution

During the study period, we admitted 41 consecutive HSCR patients. Due to the technical difficulties in frozen storage, only 31 of these patients provided adequate material to be sent to Bethesda (NIH, USA) for metagenomics. Of these 31 patients, only 20 provided both preoperative and postoperative specimen that passed internal quality controls. To summarize, out of 41 eligible HSCR patients, preoperative and postoperative stool samples were collected from 20 HSCR patients, 13 suffering from Rectosigmoid Aganglionosis (RSA), 1 L-HSCR and 6 Total Colonic Aganglionosis (TCSA). Median age at enrollment of these 20 patients was 16 months (ranging between 2 months and 13 years). A total of 40 stool samples (20 preoperative and 20 postoperative) were sequenced for 16S rDNA. In order to increase results reliability and to address HSCR forms with the highest risk of HAEC occurrences, we mostly focused on patients with TCSA that represented the core of our study (**Table 1**).

# Demographics of TCSA Patients

Six patients with TCSA were included. Male to female ratio was 2:1. Median age at enrollment was 15.5 months (range 9 to 26 months). RET mutations were detected in 7 out of 20 patients (35%), 4 of whom suffering from TCSA (67%). Three of these 4 nucleotide changes lead to truncating mutations and, interestingly, none was associated with HSCR cases complicated by HAEC manifestations. Two patients reported a previous history of HAEC (subject A and K), one (subject C) developed HAEC postoperatively well after postoperative timepoint (**Table 1**). RET mutations did not correlate either with protection or predisposition to HAEC (p = 0.4000). Four associated anomalies were detected in 3 patients. These included Congenital Anomalies of the Kidney and Urinary Tract


#### TABLE 1 | Patients with TCSA in our series.

The prevalence of associated anomalies is in line to what previously published by our group (20). The table summarizes data regarding RET mutations, Associated Anomalies and HAEC status of the whole series of patients who underwent metagenomics, focusing on patients with TCSA who underwent a deeper assessment. None of the patients received antibiotics close to preoperative timepoint sampling. All TCSA patients received antibiotic prophylaxis for the first 3 days postoperatively (4 to 7 days before postoperative timepoint sampling). N, number; pts, Patients; n.a, not assessed; HAEC, Hirschsprung's associated enterocolitis; RSA, RectoSigmoid Aganglionosis; L-HSCR, Long Hirschsprung form; TCSA, Total Colonic Aganglionosis; RET, Rearranged During Transfection proto-oncogene. Post Severe, Severe HAEC following definitive surgery; Pre Severe, Severe HAEC before diagnosis; CAKUT, Congenital Anomaly of the Kidney and Urinary Tract; GUT, Congenital Intestinal Malformation; EYE, Eye abnormality. \*The asterisks in the new nomenclature recommendations indicate the stop codon. In particular it refers to the protein and the stop signal (https://varnomen.hgvs.org/).

(CAKUT) (n = 2), Gastrointestinal anomalies (n = 1), and Visual Impairment (n = 1) (**Table 1**).

# Microbiota Diversity and Composition Comparing TCSA to RSA (20 Patients, 40 Samples)

#### Diversity

The diversity measures (**Figure 1**) were calculated based on the results derived by the overall assessment of specimen provided by each group (13 RSA and 6 TCSA, both timepoints). The diversity of gut microbial communities was assessed with alpha diversity measures. We calculated the number of observed species-level OTUs in each sample and found that the gut of RSA patients contained a higher number of species-level OTUs than the gut microbiota of TCSA patients (mean difference = 278 species; p <0.001). Similarly, the Shannon index (35), which measures the richness and evenness of a community, was higher in RSA patients than TCSA patients (mean difference = 1.51; p < 0.01). Phylogenetic Diversity (PD) (39), which takes into account phylogeny, further confirmed our findings that RSA patients possessed greater microbial diversity compared to TCSA ones (mean difference = 13.42; p < 0.005). However, the Simpson Index (38) did not significantly differ between RSA and TCSA patients. The Simpson index measures diversity by calculating the evenness of OTUs in each community and penalizes communities dominated by a small number of OTUs (**Figure 1**).

#### Composition

We assessed microbial composition by comparing the relative abundances of taxa at the phylum and genus level. At the phylum level differences between RSA and TCSA microbiota were even more striking, specifically with regard to the presence of Bacteroidetes (**Figure 2**). Nearly 70% of RSA patients (69%) had communities composed of over 33% Bacteroidetes whereas all TCSA communities contained <2% Bacteroidetes, which was found to be a significant difference (mean difference = 40.83%; p < 0.05). A greater presence of Proteobacteria in TCSA microbiota compared to RSA microbiota was also observed (mean difference = 32.27%; p <0.05). The remaining two classifiable phyla found in all of the samples, Firmicutes and Actinobacteria, did not significantly differ in relative abundance between the two groups.

When considering the communities at the genus level, we observed a prominent composition (>25%) of Bacteroides in the majority of the RSA patients' gut microbiota, whereas all of the TCSA patients contained <1% of Bacteroides in their gut microbiota. Similarly, we observed a significant increase of Alistipes (a genus in the same Phylum as Bacteroides, Bacteroidetes) in the gut microbiota of RSA patients when compared to the TCSA patients (mean difference = 1.76%; p < 0.05). Conversely, we found that members of the genus Enterococcus (a member of the Firmicutes Phylum) were more prevalent in the microbiota of TCSA patients (mean difference = 2.44%; p < 0.05).

#### TCSA Patients (6 Patients, 12 Samples)

Gut microbiota of 6 HSCR patients with TCSA was assessed before and after surgery (comparing timepoints) performed to restore bowel continuity and reverse the stoma (**Figure 3**). All preoperative sampling were obtained by the stoma. Three patients (Subjects L, N, and Q) never experienced

FIGURE 1 | The diversity of gut microbial communities was assessed through four alpha diversity measures. Together, these results suggest that ileal stools belonging to TCSA carry lower microbiota diversity.

FIGURE 2 | We assessed microbial composition by comparing the relative abundances of taxa at phylum level. When considering the communities at the phylum level, differences between RSA and TCSA microbiota were clearly evident. Bacteroidetes represented over 33% of all bacteria in RSA and where basically absent (<2%) in TCSA. Similarly, Proteobacteria accounted for nearly 40% of all bacteria in TCSA and for <5% in RSA. Conversely, Firmicutes and Actinobacteria were present both in RSA and TCSA without statistically significant differences.

episodes of HAEC. One patient (Subject C) experienced an HAEC episode preoperatively, while the two remaining patients experienced post-operative HAEC episodes (Subjects A and K).

We analyzed changes in alpha diversity between the two timepoints using three metrics: OTU richness (the number of observed OTUs in a sample), Shannon Diversity (a measure of OTU richness and evenness), and PD whole tree (a metric that takes phylogeny into account). Four of six subjects increased in Shannon diversity post-operatively. The two subjects that did not increase in Shannon diversity (Subjects A and L) were stable by this metric as well as number of OTUs and PD whole tree. For PD whole tree and OTU richness, most patients also trended toward an increase, though two subjects remained stable or slightly decreased by these metrics (Subject C and Subject N).

Though our sample size was not large enough to adequately power statistical comparisons, we noted that composition of the gut microbiota was highly variable, both pre- and postoperatively, regarding genus level taxonomy. Pre-operatively Subject A was predominantly colonized with Veillonella, Subjects K and Q were predominantly colonized by Enterobacteriaceae family, Subject C was predominantly colonized with Clostridium, and Subject N was predominantly colonized with Lactobacillus genus. Subject L was not dominantly colonized by a single genus level taxa and contained a more diverse mixture. For all subjects, there was a drastic shift in composition post-operatively, though we did not detect any specific trend in this regard. At a phylum level, Proteobacteria showed a striking predominance over other phyla either preoperatively or postoperatively but without relationship with HAEC occurrences or HAEC status. In fact, HEAC status, as well as genetic background and phenotype did not show clear correlations with gut microbiota both pre- and post-operatively (Spearman correlation, p = 0.40; see **Table 1**, **Figure 3**).

Subjects L, N, and Q never developed. The top graph of each panel represents alpha diversity, normalized to values of sample collected pre-operatively, with each color line representing a different metric. The dotted line on each graph is set to 1.0. The bottom panel illustrates relative abundance of taxa at genus level.

# DISCUSSION

Our study was aimed at addressing the effect of ultralong aganglionosis (TCSA) and its multiple implications, namely genetic background (higher prevalence of RET mutations), motility issues (higher prevalence of obstructive symptoms), and HAEC occurrences [higher frequency of HAEC episodes (12, 21, 22)] on gut microbiota composition. In fact, though it is still unknown if those variables are independent one from the others, RSA and TCSA must be considered separately when addressing HSCR patients.

Only a few published papers compared ileal and colonic fecal microbiota in children with HSCR-associated enterocolitis (HAEC). Available studies mostly refer to newborns or expreterm with a history of necrotizing enterocolitis or to children with inflammatory bowel diseases (41–45). The only study comparing microbiota of ileal and colonic stools in healthy subjects, is that by Zoetendal et al. in 2012 (46) but the authors included only adult subjects. Even so, our study confirmed that ileal samples have a lower microbiota diversity when compared to colonic ones, regardless of the presence of a stoma or not, in accordance to work reported by Zoetendal et al. (46). Recently, Barret et al. compared two preterm babies with different stomas (ileal and colonic) and followed them up to 7 months of age (41). They reported that Bifidobacteria and Enterobacteriaceae dominate at a genus level in the ileal stools samples (41). Our results are consistent with these findings with similar preponderances at a genus level. Of note, statistical analysis failed to show significant differences in terms of diversity and composition with relation to HAEC occurrences in TCSA patients, thus confirming the heterogeneity of gut microbiota and the difficulty in finding a clear and univocal marker for HAEC predisposition.

Nonetheless, the lower diversity and variable composition of microbiota throughout time points confirms that either with a stoma or not, patients with TCSA have different gut microbiota compared to those with RSA. This aspect could be easily expected basing on previous reports concerning microbiota diversity and composition in different bowel sites (41–46). Even so, as TCSA patients are well known to be susceptible to HAEC (12, 21, 22, 47), we speculate that either reduced diversity in TCSA patients can be merely related to a different sampling site or that this aspect has a pathogenetic role in terms of HAEC predisposition/facilitation. Alternatively, the well-known dysmotility and fecal stasis observed in TCSA patients could lead to bacterial overgrowth that can interfere with homeostasis and bacterial dynamics even more. In case of bacterial overgrowth, a potentially harmful microbial species can outcompete commensals very rapidly and lead to HAEC as a result of systemic reaction to this dysbiosis. This argument is supported by the factthat we could not observe a specific genus or phylum significantly associated with HAEC occurrence within

TCSA patients. Nonetheless, our series of patients with TCSA proved to have a significantly higher abundance of Proteobacteria when compared to RSA ones. In particular, 5 out of 12 samples belonging to TCSA patients were more than 80% composed by Proteobacteria. Conversely, Bacteroidetes were basically absent. The total loss of the colon undermines the existence of a specific organ, composed both of mammalian cells and microbial cells. The colon microbiota provides the system with a fundamental stabilizing function, of crucial importance for the robustness of the whole microbial and immune system. The colonic diversity and richness creates a resilient, reliable, and robust microbial community that can easily cope with potential insults, including the colonization from environmental and food microbiota. In this context, the absence of Bacteroidetes affects the production of short chain fatty acids (SCFAs) which are fundamental for intestinal homeostasis. Our result is thus in agreement with a recent report that showed how fecal samples from HAEC children showed a 4-fold decline in total SCFA concentration vs. non-HAEC HSCR patients (48). In particular, the authors found reduced acetate and increased butyrate in HAEC children, with 10 of 12 butyrate-producing genera as well as 3 of 4 acetateproducing genera demonstrated multi-fold expansion. Yet, we cannot demonstrate whether Proteobacteria preponderance and Bacteroidetes deficiency are linked to sampling site (ileal vs. colonic) or to HAEC predisposition, as recently suggested by Li et al. (15). Noteworthy, sampling site of their HAEC patients belonged more frequently to ileum or right colon, thus introducing a significant bias in the interpretation of their results (15). In TCSA, the ecological resilience of the microbiota to resist an insult is deeply undermined. Thus, TCSA patients have a fragile ileal microbial community, which may be extremely sensitive to dysbiosis that is otherwise harmless in a normal system. This is evident not only by the recurrence of HAEC, but also the variations of microbial compositions between individual patients.

This result is shown not only by the recurrence of HAEC but also from the extreme interindividual variability of the microbial composition. In absence of the colon TCSA patients have a highly variable microbiota, lacking fundamental species associated to heath in normal individuals, such as Bacterioidetes. The abundance of Proteobacteria also reflects the invasion from environmental communities, that are free to thrive in an environment that would be normally precluded to them by the presence of Bacterioidetes. Thus, HAEC should be seen not as a response to a pathogen, but as a response to a community that normally should not be present in the ileum, or in the colon, eliciting deleterious consequences. This explanation of our results is further supported by the most striking finding of a recent study on HAEC in a mouse model of HSCR that suggested how Akkermansia, a microorganism normally seen as protective, may contribute to development of enterocolitis while Bacteroidetes may be protective. Less abundant genera that were reduced in HAEC were Dysgonomonas and Clostridium XIVa, which may play a protective role (16). Taken as a whole our findings suggest that the use of therapies targeted at Clostridium difficile, without sufficient confirmation for Clostridium difficile overgrowth, might be detrimental in patients with HAEC following total colon resection. Even so, it is still possible that certain composition of the intestinal microbiota, as that reported by Li et al. or in our study, can predispose to HAEC in case of a susceptible genetic and immunologic background well known to be significantly more frequent in TCSA (22). In agreement with previous results (49) 65% of TCSA and 20% of RSA patients in our study have RET mutations. As recently published by our group (50), these mutations could determine abnormal expression of RET-dependent and independent pro-inflammatory programs that might predispose to HAEC occurrence. RET sequencing in our study could not be correlated to the incidence of HAEC mostly due to the limited number of HAEC-TCSA in our series of patients. We speculate that the interaction of a less diverse and compositionally peculiar gut microbiota (preponderance of Proteobacteria over Bacteroidetes) in patients with an imbalanced RET-dependent and independent immunity (regardless of the loss-of-function effect of RET mutations) could facilitate HAEC onset and/or predisposition. In this perspective, the defects observed in HSCR are not restricted to the aganglionic segment but extend to the mucosal immune system within and beyond the gastrointestinal tract, including the microbiota composition (50).

Although our study underlined the potential of metagenomics and improves the understanding of the relationship between microbiota, host, and immune system in patients with HSCR, it suffered important limitations. First of all, although all patients were sampled relatively far away from antibiotic therapy, we cannot exclude long term effects of antibiotic treatment. The microbiota composition proved to be patient-specific and likely depend on patients' personal history, as previously reported by Barrett et al. and Zoetendal et al. (41, 46). In particular, genera and phyla were heterogeneous in patients from our series and there was no specific and reproducible common pattern according to length of aganglionosis, genotype, phenotype, and HAEC status. This was evident when we addressed stools composition of TCSA patients with and without HAEC whose relative abundance of bacterial taxa at each time point were basically not comparable. In context of bacterial diversity and community dynamics, we could observe some of the most intriguing potentials of metagenomics in HSCR, all pointing to a lack of robustness in the gut microbiota. We argue this is due to the loss of the organ principally responsible for maintaining the reservoir of those microbes providing the buffer effect, with a loss of biodiversity corresponding to a loss of resilience. On this specific regard, it appears extremely important to achieve a heathy status by reconstituting this microbial buffer, in order to re-establish a minimal resilience and homeostasis. On the ground of these considerations, strategically designed fecal transplant defined by composition and the abundance enjoy exciting potential to address this issue in TCSA patients.

To conclude, our study confirmed the enormous potential of metagenomics in HSCR but underlined the importance of identifying the proper subset of patients for this powerful methodology. Based on the prevalence of HAEC, patients with TCSA represent the ideal subgroup to study HAEC susceptibility. A longitudinal long-term study on high-risk patients will presumably provide information that could be better compared, analyzed, and possibly applied or transferred to the general HSCR population. At present, we can only speculate that higher biodiversity could play a role in maintaining gut homeostasis and that its disruption could facilitate HAEC development.

#### ETHICS STATEMENT

Institutional Ethical committee approval was obtained by the Review Board of Giannina Gaslini Institute on November 2009 as part of a wider research project on HSCR.

#### AUTHOR CONTRIBUTIONS

AP formulated the idea, designed the study, and wrote and reviewed the manuscript. CB-M performed the DNA extraction and sequencing and analysis. MM coordinated HSCR patients sample acquisition, biobanking, and shipping HSCR patients. VR, IC, SA, and MF sampled stools from HSCR patients and

#### REFERENCES


coordinated patients' phenotype and genotype assessment. KH revised the manuscript drafts and supported the lab work. GM revised the manuscript draft and supervised patients data acquisition. FL performed the molecular genetics studies. DL, WP, and DM conceived the study and drafted the manuscript. CD and MD analyzed the microbiome data. DC reviewed the partners contributions, drafted the manuscript, analyzed the microbiome data, and inspired the study.

#### ACKNOWLEDGMENTS

This work was supported by the Italian Ministry of Health (MOH) Young Researchers Award, code WFR GR-2011-02347381.

#### SUPPLEMENTARY MATERIAL

Raw data are available at the following link: https://drive.google. com/drive/folders/1MinrLjrlnuufdHRo7aKm1XuvOwNJvR6W.


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Pini Prato, Bartow-McKenney, Hudspeth, Mosconi, Rossi, Avanzini, Faticato, Ceccherini, Lantieri, Mattioli, Larson, Pavan, De Filippo, Di Paola, Mavilio and Cavalieri. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.