Author's Reply: Drs. Rowley and Shulman call attention to the important controversy over whether Kawasaki syndrome (KS) can be caused by superantigen-producing bacteria, and more specifically toxic shock syndrome toxin-1 (TSST-1)-producing S. aureus. While the association between KS and infection by superantigen-producing bacteria is not yet accepted by all investigators, there is growing evidence to support this theory. The issues surrounding this controversy have been carefully reviewed inreference 1, and we appreciate the opportunity to offer the following comments.

A well controlled study(2) analyzing cultures in a blinded manner from multiple body sites of 16 consecutive patients with acute KS, and 15 age-matched hospitalized control patients with fever/rash demonstrated the presence of superantigen-producing bacteria in 13 of 16 acute KS patients, but only in 1 of the 15 control patients (p < 0.0001). The control patients were hospitalized during the same time period as the KS patients. Eleven of 13 toxin (superantigen)-positive cultures from patients with KS were an off-white, nonhemolytic, coagulase-positive, TSST-secreting S. aureus, and 2 of 13 were streptococci producing streptococcal pyrogenic exotoxin (SPE) B and C.

The recent report in the September 1997 issue of Pediatric Research(3) was not designed to be another controlled study but mainly to document that abundant numbers or even pure cultures of superantigen-producing S. aureus occurs in patients with KS complicated by coronary artery aneurysms. While a large, prospective, controlled study would have been preferable, such patients are relatively rare, and these were the only four cases referred to us with coronary artery complications. In the current era of high dose intravenous immunoglobin (IVIG) and acetylsalicylic acid therapy, fewer than 3% of patients develop coronary artery aneurysms. It was not our intention to provide a prospective assessment of toxin-producing S. aureus at these institutions. The importance of this report is the demonstration that the same strain of nonhemolytic toxin-producing S. aureus found in the earlier prospective series of KS(2) can also be found in KS patients who develop coronary artery disease.

The results reported in this and previous papers raise the interesting possibility that patients with both typical and atypical KS might be diagnosed and treated earlier if TSST-producing S. aureus could be recovered from such patients at the beginning of the disease process, before the autoimmune cascade has been set in motion. Furthermore, approximately 10% of KS patients are currently retreated with a second dose of IVIG because of failure to adequately respond to the first dose. It is possible that persistent carriage of this organism is associated with ongoing symptoms, persistent immune activation, and an increased risk of coronary artery disease. We agree with Rowley and Shulman that conventional antistaphylococcal therapy in patients with acute KS has no effect on the course of the illness. However, whether antibiotics, such as rifampin, that reach bacteriostatic or bactericidal levels at the surface of mucous membranes (the apparent site of infection by TSST-producing S. aureus in KS) might have an effect on patients either early in the course of their illness or in patients who do not respond to IVIG is not known.

Rowley and Shulman raise the issue of T cell Vβ skewing. While some investigators have not observed abnormalities of the T cell repertoire in acute KS, reports from three independent research groups(4–7) have demonstrated that acute KS is associated with expansion of Vβ2+ T cells, consistent with staphylococcal or streptococcal superantigen stimulation, particularly TSST-1. The failure of some groups to observe Vβ2+ T cell expansion does not rule out TSST-1 involvement as even in staphylococcal toxic shock syndrome, a disease known to be superantigen-mediated, Vβ2+ T cells are frequently not abnormally elevated(8). We agree that demonstrating an expansion of Vβ2+ T cells does not directly implicate TSST-1. Other superantigens, including exfoliative toxins and SPEB and SPEC can have similar effects. As yet unidentified superantigens may also be involved in this disease. However, our consistent ability to isolate toxin-producing S. aureus leads us to conclude that this is at least one important cause.

Other observations also support a toxin-mediated disease. The clinical features of acute KS overlap with other infectious illnesses, e.g. staphylococcal and streptococcal toxic shock syndrome, diseases known to be caused by toxin-producing bacteria. Second, the most effective therapy in KS, IVIG, contains high titered neutralizing antibodies against TSST-1, providing a rationale for its action(9). Third, a number of published and unpublished cases of toxic shock syndrome have now been found to be associated with the development of coronary artery abnormalities similar to those seen in KS(10–12). These latter data coupled with recent unpublished studies demonstrating that young rabbits infused with low concentrations of TSST-1 develop fever and vasculitis, but not shock(13), fulfills Koch's postulates on disease causation.

Why have other investigators been unable to confirm the presence of TSST-1-secreting S. aureus in KS? We believe that the unusual appearance and growth properties of the organism make it difficult to identify. Furthermore, unlike patients with toxic shock syndrome, this toxin-producing strain of S. aureus is present in low numbers in many patients with uncomplicated KS (the latter feature may partially account for the lack of shock in KS). In addition to the studies we have carried out inreferences 2 and 3, we have analyzed isolates of S. aureus obtained from 33 other KS patients collected between 1979 and 1995. These S. aureus isolates were submitted to P. Schlievert for toxin testing by the patients' treating physicians from throughout the United States. Isolates were obtained from diverse geographic areas in the United States including Hawaii, California, Colorado, Connecticut, Kansas, Minnesota, Illinois, New York, and Pennsylvania. Of these samples, 13 were submitted as pure cultures of S. aureus isolated by local hospital microbiology laboratories, with the remaining being primary culture plates from various body sites.

All 33 of the isolates made TSST-1. The total number of KS patients seen at these sites is unknown, so there is no denominator to calculate prevalence. However, 100% of the submitted cultures contained S. aureus producing TSST-1. These numbers differ significantly from reported data indicating 30-40% of individuals in the general population are colonized by S. aureus, but only approximately one-third of these isolates make TSST-1, indicating that about 10% of the general population is colonized by TSST-1-producing S. aureus(14, 15). In a study by Schlievert et al(14), vaginal cultures were obtained from 300 women, 205 at times other than menstruation and 95 during menstruation in 1980 and 1981. Thirty-six tested positive for vaginal S. aureus. Twelve of the 36 S. aureus (33%) were positive for production of TSST-1 as determined by reactivity with antisera raised against TSST-1(15). During the same time period, 22 cultures were obtained from the anterior nares of healthy children. Eleven of the children were colonized with S. aureus. Four of the 11 (37%)S. aureus isolates were positive for TSST-1. Using Fisher's exact probability test comparing the prevalence of S. aureus isolates that were TSST-1-positive from submitted KS isolates to the prevalence of TSST-1+ S. aureus isolates from the mucous membranes of both healthy children and young adult women resulted in p values <0.001. At the same time, four reports of group A streptococci associated with KS were made to P. Schlievert.

Based on our experience and consistent with the heterogeneity of the T cell response found in this disease, we believe that approximately 70% of patients with acute KS, as compared to 10-15% of controls, are infected or colonized with TSST-1-secreting S. aureus. The infectious agents triggering the remaining 30% of patients may vary according to geographic location and could include streptococcal superantigens, exfoliative toxin-secreting S. aureus, and Yersinia pseudotuberculosis(2, 3, 15). Viral agents are also an attractive etiologic agent, but none have surfaced as likely candidates.

Therefore we believe that acute KS is triggered by a number of etiologic agents that lead to a common clinical phenotype. The complex etiology of acute KS is consistent with the well accepted observation that toxic shock syndrome is caused by a variety of staphylococcal and streptococcal superantigens. Furthermore, only 15-25% of patients with KS develop coronary artery abnormalities and other complications of acute KS such as hydrops of the gallbladder are not uniformly seen in all patients. Studies also indicate that a subset of patients with acute KS fail to respond to IVIG therapy, again suggesting several etiologic agents may contribute to this clinical syndrome, and that certain etiologic triggers may be less responsive to IVIG than others.

In the absence of another likely etiologic agent, we believe that our hypothesis must be more carefully studied. We believe that a well controlled multicenter trial examining the relative prevalence of TSST-1-secreting S. aureus and other superantigen-producing bacteria in KS as compared to disease controls should be conducted. We are currently in the process of preparing a proposal to carry out such a study and hope that this will allow us to properly address this controversial but important area of pediatric disease.

Sincerely,

Donald Y. M. Leung, M.D., PhD

Professor and Head; Pediatric Allergy-Immunology; National Jewish Medical and Research Center

Patrick M. Schlievert, PhD

Professor of Microbiology; University of Minnesota Medical School