COVID-19 Positive Versus Negative Complete Kawasaki Disease: A Study from the International Kawasaki Disease Registry

Abstract

To determine clinical differences for children with complete Kawasaki disease (KD) with and without evidence of preceding SARS-CoV-2 infection. From January 2020, contemporaneous patients with complete KD criteria were classified as either SARS-CoV-2 positive (KDCOVID+; confirmed household exposure, positive PCR and/or serology) or SARS-CoV-2 negative (KDCOVID−; negative testing and no exposure) and compared. Of 744 patients in the International Kawasaki Disease Registry, 52 were KDCOVID− and 61 were KDCOVID+. KDCOVID+ patients were older (median 5.5 vs. 3.7 years; p < 0.001), and all additionally met diagnostic criteria for multisystem inflammatory syndrome in children (MIS-C). They were more likely to have abdominal pain (60% vs. 35%; p = 0.008) and headache (38% vs. 10%; p < 0.001) and had significantly higher CRP, troponin, and BUN/creatinine, and lower hemoglobin, platelets, and lymphocytes. KDCOVID+ patients were more likely to have shock (41% vs. 6%; p < 0.001), ICU admission (62% vs. 10%; p < 0.001), lower left ventricular ejection fraction (mean lowest LVEF 53% vs. 60%; p < 0.001), and to have received inotropic support (60% vs. 10%; p < 0.001). Both groups received IVIG (2 doses in 22% vs. 18%; p = 0.63), but KDCOVID+ were more likely to have received steroids (85% vs. 35%; p < 0.001) and anakinra (60% vs. 10%; p = 0.002). KDCOVID− patients were more likely to have medium/large coronary artery aneurysms (CAA, 12% vs. 0%; p = 0.01). KDCOVID+ patients differ from KDCOVID−, have more severe disease, and greater evidence of myocardial involvement and cardiovascular dysfunction rather than CAA. These patients may be a distinct KD phenotype in the presence of a prevalent specific trigger.

Introduction

Understanding of the SARS-CoV-2 (COVID-19) virus and its pathophysiology has evolved rapidly since the initial reports in late 2019 from Wuhan, China. Although the pediatric population initially presented with asymptomatic or mild disease with acute SARS-CoV-2 infection, reports of a delayed multisystem inflammatory syndrome (MIS-C) [1] began proliferating from around the world. Initial reports showed that MIS-C was associated with a more severe presentation than prior acute SARS-CoV-2 infections, and that there was clinical overlap with other immune-mediated diseases, specifically Kawasaki disease (KD) [2–7].

KD is a medium vessel vasculitis of childhood that is the leading cause of acquired heart disease in children in high-income countries, and its etiology remains largely unknown. Epidemiologic studies have reported outbreaks of disease and temporal associations with infectious agents, which has led to the etiologic hypothesis that KD is caused by an infectious or environmental agent that triggers a dysregulated inflammatory cascade in susceptible individuals [8–10]. The overlap of clinical features between MIS-C and KD has led to speculations regarding possible shared pathophysiology, particularly since individual patients may meet diagnostic criteria for both conditions [11–14].

We sought to compare a subgroup of patients with MIS-C who also met criteria for complete KD with patients who had complete KD without an associated SARS-CoV-2 infection. The purpose of this study is to distinguish the clinical characteristics between two contemporaneous and distinct groups (MIS-C with KD phenotype vs. complete KD) to further our understanding of both MIS-C and KD.

Methods

Study Population

The International Kawasaki Disease Registry (IKDR) originated in 2013 with the aim of determining factors associated with cardiac outcomes for patients who had coronary artery aneurysms after KD [15, 16]. Beginning in May 2020 with the first reports of MIS-C and reports of clinical similarities with KD [4, 17–21], the IKDR recruited additional international sites and prospectively enrolled patients as of January 2020 with acute KD, MIS-C and cardiac complications of acute SARS-CoV-2 infection.

Inclusion criteria for the prospective Registry are as follows:

Outpatient assessment or inpatient admission to a participating site starting January 1, 2020, and any of:

1. Patients for whom there has been a request for echocardiography or pediatric cardiology consultation for evaluation and the patient has or has had confirmed or suspected SARS-CoV-2 infection.

2. Patients with a confirmed or suspected diagnosis of complete or incomplete KD, or KD shock syndrome.

3. Patients with confirmed or suspected MIS-C associated with or assumed to be associated with SARS-CoV-2.

KD was defined according to criteria specified in the American Heart Association Scientific Statement of 2017 [22], and MIS-C was defined according to criteria specified by the United States Centers for Disease Control and Prevention [1].

Measurements

Participating sites abstracted data from medical records and entered patient demographics, clinical features and course, and management into a secure REDCap database maintained by the IKDR Data Coordinating Centre (DCC). All entered data were reviewed and queries were sent to the sites for clarifications and corrections. All the queries were resolved prior to patient inclusion in data analyses. Sites submitted deidentified copies of serial laboratory values, electrocardiogram tracings and cardiac imaging reports. For these variables, data were abstracted and entered in the REDCap database by personnel at the DCC.

Data Analysis

For the purposes of this analysis, patients were classified into two groups according to their COVID status—those with a recent history of SARS-CoV-2 exposure/infection (first degree relative with confirmed infection, or first degree relative with compatible symptoms who was a recent contact of a person with confirmed infection), together with either positive SARS-CoV-2 PCR or serology (COVID+); and those with no recent exposure history together with negative PCR and/or serology (COVID−). Patients with insufficient data to confirm this distinction were excluded. We then identified patients in each COVID group who had complete clinical criteria for acute KD (fever for at least 5 days, together with at least 4 of the following symptoms: non-purulent conjunctivitis, oral changes, cervical lymphadenopathy (≥ 1.5 cm diameter), polymorphous rash, and erythema and edema of the hands and feet), creating two groups for comparison (KDCOVID+ and KDCOVID−).

Data are described as frequencies, medians with values at the 25th and 50th percentiles, and means with standard deviation as appropriate. KDCOVID+ versus KDCOVID− groups were compared using Fishers exact tests, χ² tests, Mantel–Haenszel χ² tests, Kruskal–Wallis analysis of variance and Students t tests as appropriate to the level of measurement and distribution of the values of the variables. The level of statistical significance was set at 0.05.

Regulatory and Ethics Approval

The DCC is located at the Hospital for Sick Children, Toronto, and maintains Local Research Ethics Board approval for the DCC. All participating sites have approved data sharing agreements in place with the DCC and obtain and maintain Local Institutional Review Board’s approval for their site’s participation. All patients/parents either provided written informed consent/assent or data are submitted with an approved waiver of consent as per the Local Institutional Review Board requirements. All data are submitted and maintained without patient identifiers, and data security/confidentiality measures are maintained as per institutional requirements. The IKDR is governed by bylaws and a steering committee.

Results

Study Population

From January 1, 2020, through April 12, 2021, 744 patients were reported to the IKDR. Data regarding COVID status were missing for 103 patients and deemed inconclusive for 36 patients. Of the remaining 605 patients, 440 were deemed to be COVID+ as defined above, of whom the following were excluded from analysis: 215 patients with 0 or 1 KD clinical criteria, 162 patients with 2 or 3 criteria, and an additional 2 patients whose KD clinical criteria were unknown. This left 61 patients who were confirmed COVID+ and who met complete criteria for KD (KDCOVID+). Of the 165 patients who were deemed to be confirmed COVID− as defined above, 29 patients had 0 or 1 KD clinical criteria and 84 had 2 or 3 criteria and were excluded. This left 52 patients who were confirmed COVID− and who met complete criteria for KD (KDCOVID−) (Fig. 1). Of note, all KDCOVID+ patients also met diagnostic criteria for MIS-C and, therefore, represent MIS-C with KD phenotype.

Fig. 1.

Study population

Clinical Characteristics

KDCOVID+ patients had a similar proportion of males (44% vs. 56%, p = 0.45) but were significantly older at presentation (median 5.5 years vs. 3.7 years, p < 0.001, Table 1). Where ethnicity was reported, KDCOVID+ patients were more likely to be of Black ethnicity, with no cases of East Asian ethnicity, and KDCOVID− patients were more likely to be of South American, White, or East Asian ethnicity. They did not differ regarding days of fever at presentation and total days of fever. However, KDCOVID+ patients were less likely to have all five KD clinical criteria, specifically less likely to have extremity changes. Regarding associated symptoms, KDCOVID+ patients were more likely to have had abdominal pain, but not vomiting or diarrhea. They were also more likely to have had myalgias and headache. They were significantly more likely to present with shock.

Table 1.

Demographic and clinical characteristics of patients with complete Kawasaki disease clinical criteria

Variable	n	KDCOVID−	n	KDCOVID+	p value
Male sex	52	29 (56%)	61	29 (44%)	0.45
Median age at presentation (25th/75th %ile; years)	52	3.7 (2.0, 6.6)	61	5.5 (3.6, 9.3)	< 0.001
Ethnicity	31		40		N/C
White		7 (23%)		10 (25%)
Black		3 (10%)		16 (40%)
South American		13 (42%)		9 (23%)
East Asian		6 (19%)		0
South Asian		2 (6%)		0
Arabic		0		1 (2%)
Median days of fever at presentation (25th/75th %ile)	50	6 (4, 8)	60	5 (4, 6)	0.07
Median total days of fever (25th/75th %ile)	50	8 (6, 10)	61	7 (5, 8)	0.12
Number of KD criteria
Four	52	35 (67%)	61	53 (87%)
Five	52	17 (33%)	61	8 (13%)	0.03
Conjunctivitis	52	51 (98%)	61	60 (98%)	1.0
Cervical lymphadenopathy	52	25 (48%)	61	29 (48%)	0.96
Rash	52	49 (94%)	61	60 (98%)	0.34
Extremity changes	52	49 (94%)	61	47 (77%)	0.02
Oral mucosal changes	52	51 (98%)	61	56 (92%)	0.22
Associated symptoms
Cough	52	11 (21%)	61	9 (15%)	0.38
Sore throat	52	8 (15%)	61	10 (16%)	0.89
Rhinorrhea	52	9 (17%)	61	6 (10%)	0.26
Abdominal pain	52	18 (35%)	61	36 (60%)	0.008
Diarrhea	52	17 (33%)	61	26 (43%)	0.28
Vomiting	52	21 (40%)	61	33 (54%)	0.15
Arthritis/arthralgias	52	3 (6%)	61	3 (5%)	1.00
Myalgias	52	3 (6%)	61	12 (20%)	< 0.05
Dyspnea	52	4 (8%)	61	10 (16%)	0.26
Anosmia	52	0	61	2 (3%)	0.50
Headache	52	5 (10%)	61	22 (38%)	< 0.001
Confusion	52	3 (6%)	61	9 (15%)	0.14
Syncope	52	0	61	2 (3%)	0.50
Presence of shock	52	3 (6%)	61	25 (41%)	< 0.001
Missed KD diagnosis	52	7 (13%)	61	11 (18%)	0.51

KD Kawasaki disease, N/C not able to calculate, %ile percentile

Laboratory Variables

Relative to KDCOVID− patients, KDCOVID+ patients had significantly higher peak C-reactive protein (CRP), ferritin, and D-dimer levels. Both groups had similar elevations in erythrocyte sedimentation rate (ESR) levels. KDCOVID+ patients more positive and higher levels of markers of cardiac injury [such as troponin and pro-brain natriuretic peptide (BNP)], and lower hemoglobin, platelets, lymphocytes, and albumin compared to KDCOVID− patients (Table 2). The KDCOVID+ group had significantly higher laboratory markers of renal impairment, including higher peak blood urea nitrogen (BUN) and serum creatinine levels. Both groups also had similar elevations in liver enzymes.

Table 2.

Laboratory values obtained over 6 weeks from diagnosis for patients with complete Kawasaki disease clinical criteria

Variable*	n	KDCOVID−	n	KDCOVID+	p-value
Highest white blood cells (WBC, × 109/L)	43	15.9 ± 6.5	54	16.8 ± 9.1	0.59
Lowest hemoglobin (g/L)	51	102.5 ± 13.5	59	93.1 ± 13.9	< 0.001
Highest platelets (× 109/L)	51	550.4 ± 214.4	56	448.0 ± 204.8	0.01
Highest neutrophils (× 109/L)	47	9.8 ± 6.5	54	12.5 ± 7.7	0.06
Highest lymphocytes (× 109/L)	47	6.2 ± 4.1	54	4.2 ± 2.9	0.008
Highest ESR (mm/h)	44	63.5 ± 33.3	30	65.7 ± 35.0	0.78
Lowest albumin (g/L)	31	30.4 ± 8.5	24	23.9 ± 6.7	0.002
Highest C-reactive protein (CRP, mg/L)**	18	102 (19, 250)	47	177 (123, 206)	0.02
Highest highly sensitive CRP (mg/L)**	23	70 (25, 140)	9	211 (139, 237)	0.02
Highest ferritin (mcg/L)**	43	195 (116, 3840	39	636 (427, 1207)	< 0.001
Highest triglycerides (mmol/L)	35	2.4 ± 1.3	17	2.8 ± 1.4	0.4
Highest ALT (U/L)**	42	42 (23, 88)	51	36 (24, 63)	0.53
Highest AST (U/L)**	48	51 (40, 69)	53	46 (37, 77)	0.38
Lowest sodium (mmol/L)	50	134.4 ± 3.6	56	131.8 ± 3.9	< 0.001
Highest BUN (mmol/L)	26	4.7 ± 2.4	52	7.2 ± 5.5	0.009
Highest serum creatinine (μmol/L)	38	32.9 ± 12.1	54	52.9 ± 27.2	< 0.001
Highest INR	28	1.3 ± 0.3	37	1.3 ± 0.3	0.46
Highest D-dimer (mcg/mL)**	35	0.7 (0.4, 1.0)	55	2.4 (1.6, 3.2)	< 0.001
Median highest troponin-I (mcg/L)**	20	10 (10, 10)	25	126 (10, 510)	< 0.001
Median highest NT-pro-BNP (ng/L)**	20	1024 (332, 2218)	25	6080 (3700, 10682)	< 0.001

ALT alanine transaminase, AST aspartate transaminase, BUN blood urea nitrogen, CRP C-reactive protein, ESR erythrocyte sedimentation rate, g gram, INR international normalized ratio, L liter, mcg microgram, mg milligram, mmol millimole, mm/h millimeters per hour, ng nanogram, NT-pro-BNP NT pro-brain natriuretic peptide, U unit, μmol micromole

*Mean and standard deviation unless otherwise indicated

^**Median with 25th and 75th percentiles

Management and Outcomes

KDCOVID+ patients were significantly more likely to have been managed in the intensive care unit, to have received inotropic/vasoactive agents, and had a longer total length of hospital stay (Table 3). There were no deaths and no patient received ECMO in either group. While nearly everyone in both groups received intravenous immune globulin, the proportion in each group who required additional intravenous immunoglobulin (IVIG) doses was similar. A higher proportion of KDCOVID+ patients received corticosteroids and anakinra. Regarding cardiac outcomes, KDCOVID+ patients had lower left ventricular ejection fractions during their clinical course. While maximum coronary artery Z score in any coronary branch was similar between groups, the KDCOVID− patients were more likely to have higher categories of coronary artery involvement. Specifically, they were more likely to have medium or large coronary artery aneurysms (Fig. 2).

Table 3.

Management and outcomes of patients with complete Kawasaki disease diagnostic criteria

Variable	n	KDCOVID−	n	KDCOVI +	p value
Intensive care unit stay	52	5 (10%)	61	37 (62%)	< 0.001
Inotropes given	52	5 (10%)	61	36 (60%)	< 0.001
Median length of hospital stay (25th/75th %ile; days)	49	4 (3, 6)	60	7 (5, 11)	< 0.001
Management
Any IVIG	52	50 (96%)	61	60 (98%)	0.60
Two doses of IVIG	52	11 (22%)	61	11 (18%)	0.63
Intravenous corticosteroids	52	18 (35%)	61	52 (85%)	< 0.001
Oral corticosteroids	52	23 (44%)	61	38 (66%)	0.03
Infliximab	52	1 (2%)	61	1 (2%)	1.0
Anakinra	52	2 (4%)	61	16 (26%)	0.002
Echocardiography
Mean lowest LV ejection fraction (SD; %)	49	60 ± 8	58	53 ± 8	< 0.001
Median maximum coronary artery Z score (any branch; 25th/75th %ile)	46	1.39 (0.70, 2.64)	60	1.35 (0.55, 1.82)	0.37

CAA coronary artery aneurysm, IVIG intravenous immune globulin, LV left ventricle, %ile percentile, SD standard deviation

Fig. 2.

Maximum coronary artery Z score category by group. CA coronary artery, KD Kawasaki disease

Discussion

In this study of 113 patients (61 KDCOVID+, 52 KDCOVID−) who met criteria for complete KD, we found that the characteristics and clinical presentation of KDCOVID+ (MIS-C with KD phenotype) patients differed from KDCOVID− patients, and patients within the KDCOVID+ group resemble patients with MIS-C rather than complete KD.

Complete KD with COVID+ vs. COVID− 

Several aspects of the presentation for both populations were quite similar, with minor non-KD symptoms, specifically gastrointestinal symptoms, myalgia, and headache, more likely for the KDCOVID+ (MIS-C with KD phenotype) group in keeping with a concomitant diagnosis of MIS-C. It is also notable that the age of presentation for KDCOVID+ patients were older than KDCOVID−, which could represent a population who are able to communicate symptoms of headache, myalgia, and gastrointestinal symptoms. There were, however, notable distinguishing features between the groups pertaining to severity of illness during their clinical course. Patients within the KDCOVID+ population were much more likely to present with shock, to be managed in an intensive care unit, and to have received inotropic support. In addition, a larger proportion of KDCOVID+ patients receiving corticosteroids and anakinra in addition to IVIG. The increased use of anakinra in the KDCOVID+ group could suggest a more severe, IVIG-resistant clinical course but it is also notable that a few institutions suggested anakinra treatment for patients who met criteria for MIS-C who required ICU admission. The severity of illness in the KDCOVID+ patients was a considerable deviation both from the KDCOVID− patients as well as historical reports of complete KD, which document only a small percentage of KD patients (3–7%) presenting with severe symptoms [23–25]. Consistent with the severity of presentation, KDCOVID+ patients in our study also had a lower average left ventricular ejection fraction in the mildly reduced range during their hospital course. This higher likelihood of severe presentation might suggest that the trigger for KD presentation in these patients, presumably SARS-CoV-2, triggers either a more significant immune response or activates a separate immune cascade compared to other inciting triggers of classic KD. This variation in presentation is further exemplified by KDCOVID− patients having a greater likelihood of medium and large CAAs, which is more characteristically associated with KD.

Recent studies on symptom variations between clusters of cases of KD support that different exposures may trigger distinctive clinical and demographic characteristics among KD patients [26]. This suggests the possibility of a KD subset of “SARS-CoV-2-induced KD”. The opaqueness in diagnosis of KD stems from lack of definitive knowledge of the pathogenesis of KD, as it has always been defined by a constellation of clinical features. Notably, clinical studies and immunologic characterization suggest distinct differences between MIS-C and KD, although current studies of immunologic characterization are limited by small numbers of subjects [2, 27].

Complete KD with COVID+ vs. MIS-C

The features and outcomes of KDCOVID+ (MIS-C with KD phenotype) patients in our study showed similarities with prior reports of MIS-C. First, the median age at presentation was significantly older (5.5 years) than with classic KD but comparable to the median age for pediatric patients reported with MIS-C [2–4, 17, 28]. Patients in our KDCOVID+ group also had similar laboratory findings as MIS-C patients, including greater elevations in systemic and cardiac markers (CRP, BNP and troponin), which has a strong association with MIS-C, along with significant lymphopenia and thrombocytopenia [2, 19, 29]. We also found evidence of greater renal dysfunction in our KDCOVID+ group but no difference in their liver function. As previously noted, the severity of presentation requiring admission to the intensive care unit in our study is substantially higher among the KDCOVID+ population (62% vs. 10%, p < 0.001). Prior reports of MIS-C have also shown high rates of intensive care unit admission, ranging from 50 to 85% [2, 3], due to shock and hypoxia. Although there is no established standard of care treatment strategy for MIS-C at this time to compare to our study results, the treatment of MIS-C patients has also included IVIG, steroids, and biologics [2, 30]. The treatment response of MIS-C to these various regimens has yet to be determined.

The current clinical criteria of MIS-C remain rather broad with many patients with KD also fulfilling the criteria for MIS-C except for the required concurrent or prior SARS-CoV-2 infection. The polymorphous rash within KD could be considered dermatologic involvement, along with many KD patients also presenting with GI symptoms and anemia [22]. It is imperative that we continue to study both MIS-C and KD in depth to delineate both the distinctive character of SARS-CoV-2 as a trigger and identify the immunologic cascade that follows.

Strengths and Limitations

This is a multi-institutional, multi-national database that provides data across different ethnicities and regions across a large time frame. This negates regional variations and provides a large sample size to study. We also only included patients who met the criteria for complete KD, thus excluding patients with incomplete KD. This allows us to compare clinical presentation solely based on SARS-CoV-2 status rather than being confounded by incomplete presentation of KD. Our patient groups are also bounded by the same time frame, which removes another confounding factor. However, our data were reliant on reporting from multiple centers including clinical data, echocardiogram reports and laboratory assessments. Although an exhaustive check of the data and its consistency was carried out, this adds a possibility of human error, as well as variation in definition of features and in laboratory assessment accuracy. Another limitation is that even though this is a multi-national study, most of the institutions represented are within North and Central America.

Conclusion

For pediatric patients during the COVID-19 pandemic who present with complete KD criteria, MIS-C with KD phenotype differ from complete KD and have a greater severity of illness and shows greater evidence for myocardial involvement and cardiovascular dysfunction rather than coronary artery aneurysms. It is unclear whether this may represent a distinct KD phenotype of MIS-C, or a MIS-C phenotype of KD in the presence of a prevalent specific immunologic trigger, in this case SARS-CoV-2. Ongoing investigations into environmental, genetic, and immunologic susceptibilities, pathophysiology and treatment response of MIS-C and KD will help to further our understanding of both conditions.

Author contributions

JJ, ET, BM conceptualized the idea, designed the study, performed the measurements, processed the experimental data, performed the analysis, drafted the manuscript, and designed the tables and figures. AH, ND, GR, KF, MK, MH, SM, FD, NM, MP, JB, SL, SJ, TH, JS, WO, GL, LN, EB, AB, JB, AC, KC, AD, MF, LG, MH, PJ, HK, SK, ML, RL, VM, KM, SM, KN, DP, PR, AS, PS, RS, SS, DT, MY, AY, VZ contributed to data curation and processed data. PF, CM, KR processed the experimental data and performed the analysis. All authors reviewed the manuscript.

Funding

Funding for the Data Coordinating Centre was partially provided by the CIBC World Market Chair in Child Health Research (Brian McCrindle) and the Labatt Family Heart Centre at SickKids Hospital (Brian McCrindle). Additional local funding for participation in the International Kawasaki Disease Registry (IKDR) was provided by: les Fonds BoBeau Coeur of the Ste-Justine Hospital Foundation (Nagib Dahdah), SUNY Research Seed Grant Program 2019–2020 and by the Wildermuth Memorial Foundation through ‘Variety’, the Children's Charity of Buffalo and Western New York (Mark D. Hicar), the McCance Family Foundation (Jane Newburger), R01HL143130 from the National Institutes of Health (Michael Portman).

Declarations

Conflict of interest

The authors have no conflicts of interest relevant to this article to disclose.