Kawasaki disease: Is it caused by an infectious agent?

Kawasaki disease: Is it caused by an infectious agent?

Lloyd, Allyson J


Historical background

Kawasaki disease (KD) was described initially by Tomisaku Kawasaki, a Japanese paediatrician, in 1967. In his report, he outlined 50 cases of what he termed `mucocutaneous lymph node syndrome’, observed in young children between 1961 and 1967.’ Originally thought to be a benign, self-limiting childhood disease, by late 1970 research revealed that up to 10 Japanese children had died of the illness. This prompted further study of this new disease, which culminated in the publication of Kawasaki and his coworkers’ findings during 1974.2 This resulted in the worldwide recognition of the condition, then renamed Kawasaki disease.

At this point, it should be noted that KD is also known as Kawasaki syndrome, reflecting the varied clinical signs and symptoms associated with the illness; however, both terms are interchangeable and there is no clinical differentiation made between the two. Since 1974, cases of KD have been reported worldwide,3 and the illness is now recognised as an important cause of child morbidity, having replaced acute rheumatic fever as the most common cause of acquired heart disease in children in the developed world.4


KD is an acute systemic febrile illness, with a duration of six to eight weeks, usually affecting children under five years of age.5 The causative agent remains unknown and currently no direct diagnostic test exists for the illness. Diagnosis is based on clinical presentation of the patient, coupled with a differential diagnosis. The current diagnostic criteria for KD are presented in Table 1.

A variety of other clinical features often can be seen in cases of KD, although these additional signs and symptoms are not included in the diagnostic criteria. Approximately one quarter of patients with KD have concurrent aseptic meningitis, and many patients experience arthralgia during the acute phase of the illness. This may be accompanied by arthritis of the wrists, knees and occasionally the hips. Gastrointestinal symptoms – including hepatitis, diarrhoea and vomiting, pancreatitis and acute hydrops of the gall bladder – also have been reported.’

Additional clinical features include pneumonitis, thrombocytosis and otitis media.6 Cardiac-related complications – including myocarditis, pericarditis and occasionally mitral-valve regurgitation – also have been reported in cases of KD;6 however, perhaps the most significant clinical feature associated with KD is the development of life-threatening coronary artery abnormalities.

Most deaths associated with KD occur during the third or fourth week of the illness, invariably when the patients’ general clinical condition has improved. These deaths are usually due to myocardial infarction, secondary to the formation of coronary artery aneurysms.2 Untreated KD can result in the development of such abnormalities, with most aneurysms forming within four weeks of the onset of symptoms. KD also may affect other medium-sized arteries throughout the body, resulting in acute vasculitis.5

Atypical disease

A small number of cases do not fulfil all the required diagnostic criteria, and these are regarded as atypical or incomplete presentations of the illness. Such cases are found most commonly in younger infants, who are unfortunately also at most risk of developing coronary artery abnormalities.7 Diagnosis in these cases is difficult and is usually based on evidence of a prolonged episode of fever, coupled with the presence of coronary artery aneurysms detected by echocardiography.5


Early studies showed that antimicrobial agents had little influence on the clinical course of KD, including the development of coronary artery abnormalities.8 Following extensive research, the combined administration of aspirin and intravenous gamma-globulin (IVGG) now is acknowledged as the preferred treatment, with additional therapies used in special circumstances.

Aspirin is used for its anti-inflammatory and antithrombotic effects, leading to a reduced incidence of fatal myocardial infarctions caused by thrombosis of the coronary arteries. However, there are potential side effects to the use of aspirin in young children, including the risk of gastrointestinal haemorrhage and the development of Reye’s syndrome. If such complications occur, the use of alternative therapies, such as dipyridamole, are indicated.9

Administration of IVGG results in a significant decrease in both the incidence and severity of aneurysm formation.4 In addition, it helps to reduce inflammation and contributes to relief of the general symptoms of KD; however, the means by which IVGG actually achieves these effects has yet to be determined.5

In the 33 years since KD was first recognised, much clinical and epidemiological research into this puzzling disease has been performed. To date, however, no causative agent has been proven consistently to be associated directly with this illness. Here, we examine the evidence pertaining to the general concept that KD is caused by an infectious agent.

Is Kawasaki disease caused by an infectious agent?

Potential bacterial aetiological agents

Excluding toxin-producing strains of Staphylococcus aureus, which will be reviewed separately, two bacterial agents have been proposed as potential causes of KD.

Group A beta-haemolytic streptococcal infection can result in complications which give rise to symptoms similar to those encountered during KD, including acquired heart disorders. As a result, early research involved attempts to isolate this organism from cases of KD, either by direct isolation or via serological methods and were successful in a minority of cases.10 However, the results suggested that group A beta-haemolytic streptococcal infection occurred concurrently with KD, rather than causing the illness directly.

Propionibacterium aches, a skin commensal, also has been proposed as a potential causative agent. During a case-control study, P. aches was isolated from about a third of patients with KD. The same P. aches serotype also was isolated from house-dust mites present in six patients’ homes; whereas the organism was not isolated from age-matched controls or their environments. During the same study, when animal models were inoculated with the same P. aches serotype, various inflammatory responses, including coronary arteritis and myocarditis, were observed.11 Therefore, it was proposed that this P. acnes serotype could have a causative role in KD, with the house-dust mite a potential vector. Unfortunately, these results have never been replicated.

The development of molecular techniques has helped in the search for the agent responsible for KD. Many studies have used polymerase chain reaction (PCR) techniques to amplify the highly conserved bacterial chromosomal 16S ribosomal RNA (rRNA) sequences common to all known members of the bacterial domain, thus enabling the identification of previously uncultivable bacteria. However, when used on a variety of samples from patients with KD, this technique failed to detect any of these sequences, and effectively ruled out a bacterial cause for KD.12

Potential viral aetiological agents

KD shares many clinical symptoms with a range of virus infections, and, as a result, investigators have suggested that the aetiological agent is viral in origin. To further support this, a recent editorial13 proposed that cases of KD may have been misdiagnosed clinically as measles by general practitioners.

Increased serum-interferon levels and lymphocyte abnormalities, seen in patients during the acute phase of the illness, further strengthen the evidence for a viral aetiological agent. Increased number of activated T-helper cells and decreased numbers of T-suppressor cells, seen in these patients, are especially suggestive of a retroviral agent.

Further support was provided by the demonstration of reverse transcriptase activity in lymphocytes harvested from patients with KD;14-16 however, additional study has failed to confirm a retroviral presence,17 and others have been unable to demonstrate such reverse transcriptase activity.18

Other viruses have been investigated and presented as potential factors in the development of KD, including parvovirus B19(19) and Epstein-Barr virus;20 however, due to lack of corroborating evidence, these are no longer thought to be involved in the development of the disease.

Potential Rickettsial aetiological agents

During the late 1970s and early 1980s, a series of letters and concise articles addressed the possibility that KD is caused by members of the Rickettsiaceae family. One such letter21 outlined a case of KD that was diagnosed in London. Paired sera taken during the investigation of this case demonstrated significant complement fixation test antibody titres against Coxiella burnetii, a member of the Rickettsiaceae.

In an earlier communication from Australia,22 the presence of Rickettsia-like bodies, observed by electron microscopy, were reported in the spleen of a child who had died following development of KD. Unfortunately, in this case, paired sera taken two weeks before death and at post-mortem examination failed to demonstrate significant antibody titres against C. burnetii.

On review of these two cases, investigators considered the possibility that KD was caused by either an unusual variant of C. burnetii, infecting only infants, or another antigenically related organism; the latter being supported by a brief investigation23 carried out at the Pasteur Institute, Paris.

Sera from two children with KD were examined, alongside samples taken from apparently healthy children, for antibodies against Ehrlichia canis. This organism is a member of the Rickettsiaceae and has been linked to haemorrhagic diseases in dogs. The sera from the control cases were negative for E. canis antibodies; whereas, those from the two cases of KD demonstrated significant titres. However, due to the small number of cases presented and the lack of subsequent research, the possibility that members of the Rickettsiaceae are potential causative agents of KD has been discounted.

Chlamydia pneumoniae as a potential aetiological agent

Recently, C. pneumoniae has been proposed as a potential aetiological agent. A common respiratory pathogen, it has been linked to the development of endocarditis and myocarditis in children,24 as well as being implicated in increasing the risk of atherosclerosis and heart attacks in adults.25 Such clinical manifestations, coupled with evidence suggesting that a pathogen with a respiratory route of entry may be responsible for KD,5 makes C. pneumoniae a plausible potential infectious agent.

Two studies have supported an association between KD and chlamydial infection. In a case-control study,26 chlamydia-like organisms were isolated from urine samples taken from nine patients with KD; however, isolate identification was based only on inclusion-body morphology, rather than on chlamydia– specific diagnostic methods.

In another more recent study,27 C. pneumoniae was detected by immunochemical techniques in stored heart tissue samples taken from two children who had died after contracting KD. However, this study did not examine similar tissue samples taken from control patients, and reported an association between KD and C. pneumoniae, based solely upon these two cases.

The association between KD and C. pneumoniae was called into doubt by the results of a case-control study28 carried out in early 1999 in San Diego County, USA. Pharyngeal specimens and samples of blood from patients with KD were found to be negative for C. pneumoniae using PCR techniques. Likewise, cultures of blood, pharyngeal secretions and urine showed no evidence of C. pneumoniae infection in these patients. As a result of these studies, the role of C. pneumoniae in the development of KD remains inconclusive.

Superantigens as potential aetiological agents

The symptoms of KD can resemble those seen in patients with scarlet fever and toxic shock syndrome (TSS), both of which are induced by bacterial exotoxins that act as superantigens, resulting in the stimulation of hyper-reactive immune responses in infected patients. As a result of these observations, it was thought that KD could be due to either an unknown aetiological agent with superantigen activity, or to toxin-producing strains of staphylococci or streptococci.

Initial studies indicated that KD was associated with superantigen-like activation of T cells, monocytes and macrophages, coupled with increases in levels of serum cytokines, especially interleukin- I (IL- 1) and tumour necrosis factor (alpha (TNF(alpha)). In particular, during the acute phase of the illness, an expansion in populations of V(beta)2- and V(beta)8-expressing T cells has been observed.29

However, other research groups have contested these findings,30,31 and failure to detect such an increase in the numbers of V(beta)-expressing T cells may be explained by the dynamic nature of the change in the ratios of specific T-cell families induced by superantigens. In patients diagnosed with KD, V(beta)2-expressing T cells are detected most consistently during the second week of the illness; however, distribution of these cells can normalise rapidly, and failure by some investigators to detect an increase in this population may be due simply to the timing of the blood sample.32

Further evidence in support of potential superantigen involvement was provided in 1993 when an American research group reported33 the detection of toxinproducing staphylococci or streptococci in 13 out of 16 patients with KD, compared with only one such isolate from 15 control subjects. They suggested that the observed expansion of V(beta)2-expressing T cells may be due to a new serotype of TSS toxin-producing S. aureus in the majority of patients with KD; and to toxin-producing streptococci in a minority.

Subsequent research was unable to demonstrate any such association between KD and toxin-producing strains of either streptococci or S. aureus, leading to suggestions that the initial study detected strains of toxin-producing bacteria that were part of the respective patients’ normal skin flora.30,34 In addition, further investigation of the immune response to KD has lead subsequently to the proposal that the agent which causes this illness is a conventional antigen rather than a superantigen.35

These conflicting findings prompted the recent suggestion that both superantigens and conventional antigens may be implicated in the pathogenesis of KD. It has been proposed that a subpopulation of T cells may be stimulated initially by a superantigen and only then become activated by conventional antigen when they invade the coronary arteries; however, the evidence supporting this theory is inconclusive.2

Epidemiological clues

Many research groups have tried to use epidemiological data to link individual cases of KD, in the hope that this would lead to the identification of the aetiological agent, and several common key areas of consistent data were identified.

KD is almost exclusively an illness of young children (80% of patients are aged less than five years) but rare in infants younger than three months. KD occurs more frequently in males, with an attack ratio of 1.5:1. Data suggest that it is more prevalent in children from middle and upper socioeconomic groups, with the highest incidence seen during the spring and winter in temperate climates.36

Epidemiological investigation identified the occurrence of KD epidemics, recurring in cycles of two to three years.37 Such epidemics exhibited a wave-like spread but did not appear to have clusters of cases within households or other common locales, suggesting that person-to-person transmission is unlikely.38 Analysis of data gathered during these epidemics, and in other sporadic outbreaks, showed that children of Japanese ancestry were at highest risk of developing KD.39 This was supported by epidemiological data gathered over a seven-year period in Hawaii, which highlighted a 21-fold difference in the incidence of KD between children of Japanese and Caucasian origin.40

A clue to why this difference could have occurred is provided by an earlier study into the relationship between human leucocyte antigens (HLAs) and KD.41 The antigen HLA-BW22 was found to occur more commonly in patients with KD than in control groups, and the antigen’s incidence is approximately 12% in individuals of Japanese origin – far higher than in other races. This suggests that susceptibility to KD may be under genetic control and linked to an HLA antigen associated with Japanese descent.

A report describing an outbreak of KD in Colorado, USA, in the spring of 1982, highlighted an association between KD and exposure to carpet shampoo,42 and was followed by a series of corroborating articles.36,38,43 It was suggested that the anionic detergents found in the shampoo could provoke a hypersensitive response that manifested itself as KD, or that during the carpet cleaning process the causative agent underwent aerosolisation and subsequent inhalation. In such cases, it was suggested that the causative agent might be either infectious or allergenic in nature. House-dust mites were proposed as a potential allergen; however, it has also been suggested that housedust mites may act as vectors for potential infectious agents involved in KD.42

A subsequent study44 presented data in which no link between carpet cleaning and patients with KD could be established, placing the connection in doubt. Likewise, other proposed links with standing bodies of water,39,45 and mercury poisoning46,47 have not been established irrefutably.


During the 33 years since KD was first identified, an enormous amount of research has been conducted to determine its cause, but to no avail. Many potential agents – both infectious and chemical – have been proposed, only to be dismissed following presentation of contradictory data. In addition, the relatively small number of cases available for study outside Japan has hampered further investigation. Even during epidemics, cases may be scattered over a wide geographical area, posing difficulties for initial outbreak identification and the coordination of appropriate sample collection.

Epidemiological data provide the most consistent evidence for a potential infectious cause. Worldwide, corroborated evidence shows that KD runs a self-limiting course, rarely recurs, and occurs in epidemics that demonstrate a wave-like spread and a cyclical nature.

The disease affects young children but is rare in infants under three months; findings which may be explained by the presence of maternal antibodies that impart passively acquired immunity, but are lost over a period of time. Likewise, the relative lack of recurrence and low incidence in adulthood may be due to long-lasting immunity acquired during asymptomatic infection. All such factors are suggestive of an infectious aetiological agent, as is the nature of the immunological response provoked during acute KD.

Further supporting evidence for an infectious aetiological agent is provided by the effectiveness of IVGG in the treatment of KD. Although the exact mechanism of action is unknown, it is thought that IVGG either may cause immunological blockade of Fc receptors or contain antibodies against the specific aetiological agent, both of which would only be effective when treating an infectious aetiological agent.

Apparent lack of person-to-person transmission and paucity of case clusters within households, schools or nurseries is uncharacteristic of infectious disease, unless the infection is widely spread and only a small proportion of the infected population exhibit the full symptoms of the disease.

Such subclinical cases of KD would link with the hypothesis of long-lasting immunity, and is seen in some viral infections (e.g. that caused by EpsteinBarr virus). Therefore, is the aetiological agent of KD viral in nature? With all potential bacterial causes of KD virtually excluded – both by molecular techniques and because the disease fails to respond to antibiotic treatment, the similarity of symptoms observed during the course of KD to those seen during many viral infections further strengthens such a possibility.

However, failure to identify a causative viral agent casts doubt on such a theory. Advances in molecular diagnostic techniques have enabled the rapid identification of a number of new viruses, such as Hendra and Nipah; however, there is little progress in identifying a potential viral cause for KD after 33 years of research.

An unexplored possibility is that KD may be caused by an infectious prion protein. These transmissible agents are not typical of viruses and have never been detected using electron microscopy; however, as they do not usually exhibit inflammatory reactions and are not known to provoke antibody-mediated or other immune responses, it is unlikely that such an agent would be the cause of KD.

The role of HLA antigens in predisposing patients to develop KD remains unclear; however, evidence indicates a possible genetic aspect to disease susceptibility. The role of potential vectors, either arthropods or the house-dust mite, in the pathogenesis of KD also remains undetermined.

When all factors are taken into consideration, a multifactorial aetiology for KD would appear a distinct possibility. Indeed, the range of clinical signs and symptoms associated with the illness is more suggestive of the involvement of multiple factors, rather than a single infectious cause. The recent hypothesis proposing a combined conventional and superantigen– mediated immune response during KD provides further supportive evidence for a multifactorial aetiology.


Most of the evidence currently available indicates that the causative agent involved in KD is most likely to be infectious in nature. However, symptomatic development is probably a multifactorial event, requiring genetic predisposition, possible involvement of a vector. and lack of antibody-mediated protective immunity.

In most individuals, KD may have an asymptomatic or subclinical presentation, with only a small proportion of the population genetically or immunologically predisposed to exhibit the full range of symptoms seen in diagnosed cases. However, full understanding of the pathogenesis will be obtained only when any or all causative or contributory factors are identified. Once the manner in which these agents interact is established, improvements in both the diagnosis and treatment of this intriguing illness will be made.

Copyright Royal Society of Medicine Press Ltd. 2001

Provided by ProQuest Information and Learning Company. All rights Reserved.