Background
Kawasaki disease (KD) is the leading cause of acquired heart disease in children [
1,
2]. Nearly one quarter of untreated KD children will go on to develop coronary aneurysms. The classic KD presentation is a systemic vasculitis with minimum of 5 days of fever and 4 of the 5 following clinical criteria; conjunctivitis, rash, distal extremity swelling, oral mucous membrane inflammation, and non-generalized lymphadenopathy [
3]. Both incomplete forms, presenting with less than 4 criteria, and atypical variants, presenting classically but with an additional clinical finding not typically seen with KD, have a similar coronary arteritis propensity to classic KD presentations [
4]. Treatment with intravenous immunoglobulin (IVIG) given prior to the 10
th day of illness helps to prevent the development of coronary aneurysms [
5].
Several studies have supported a role for B cell responses in the pathogenesis of KD. Trans-placental passage of maternal antibodies is thought to be protective and explain the paucity of cases in infancy [
6]. Recent studies support a protective role of breastfeeding [
7]. Genome-wide genetic marker association studies show that specific polymorphisms in CD40 and in the B lymphoid tyrosine kinase genes associate with KD [
8,
9]. IgA+ plasma cells are shown within inflammatory infiltrates in the vessel walls of aneurysms from KD children [
10,
11], and depleted in the peripheral blood compartment [
12]. These IgA+ cell rich aneurysmal infiltrates are proposed to be specific responses to an infiltrating pathogen.
IL-21, produced mainly by T cells and Natural Killer cells [
13,
14], has recently been shown to be a specific marker in KD in a Korean cohort of children [
15]. In this study, IL-21 was elevated compared to a control group of children with prolonged fevers thought to be due to mononucleosis. This finding further supports a role of B cell activation in KD, since IL-21 modulates immunoglobulin isotype switching and is involved in the differentiation of both naïve and memory B cells into mature plasma cells [
16]. Only one other published study, focusing on the chemokine IP-10 (CXCL10), has included IL-21 levels in there published data [
17]. Unfortunately, the cytokine bead array used was comparably much less sensitive than the published ELISA-based data [
15], making direct comparison difficult.
We wished to assess if IL-21 could be used as a marker of KD in a North American cohort of febrile children to distinguish KD from other febrile presentations. We recruited 12 KD and 60 control subjects between 9 months and six years of age through an urban Emergency Department. Within this manuscript we detail our results that show that IL-21 levels were non-specifically elevated across all febrile children, irrespective of KD diagnosis.
Methods
Enrollment
This is a single site observational (unmatched case–control) study of children with suspected KD compared to other febrile children. Subjects consisted of children aged nine months to six years of age who presented to the Emergency Department of Women and Children’s Hospital of Buffalo (WCHOB) between March 2014 and March 2015. Patients in this age group with a planned blood draw as part of their emergency room evaluation were eligible if they had fever (>100.9 °F or 38.2 °C) at home (within 24 h of seeking care) or confirmed in the emergency room and at least one of the following symptoms: rash, mucous membrane changes, extremity changes, conjunctivitis or a single isolated enlarged lymph node. Patients were also eligible for enrollment if they were sent to the emergency room with the specific concern for Kawasaki disease. After parental informed consent was obtained, during clinical blood collection, an additional 5–10 ml of blood was collected. This study successfully enrolled 72 subjects. University at Buffalo’s Human Research Protection Program institutional review board approval was obtained prior to the initiation of the study and has been modified and renewed on a yearly basis (MODCR00000185).
Notable exclusions to prevent effects of excessive blood draws included prior study enrollment within two months, chronic or active blood borne infection such as HIV, HBV or HCV, chronic anemia, excessive blood loss or previous multiple blood draws within 8 weeks prior to study enrollment. Symptoms occurring at ED presentation, vital signs, and demographic information were recorded on the study data collection form. Access to the medical record was included in the consent to allow for review of individual medical records to assign enrolled subjects as KD or controls. Assignment to the KD group required the following: admittion to the hospital, an infectious disease consultant agreed with the diagnosis of KD, and the subject having received immunoglobulin therapy. Notably, there were no controversies in assignment between the primary team and infectious disease consultant, no control subject received IVIG, and most control subjects were not admitted.
ELISA tests
During peripheral blood mononuclear cell (PBMC) isolation for a separate study, diluted plasma (roughly 40% with PBS) was withdrawn and stored in a −80 °C freezer. The Human IL-21 ELISA kit (Ready-SET-Go! Kit, Affymetrix, San Diego) with a published sensitivity range of 8–1000 pg/mL was used in this study. The manufacturer’s protocol was followed and is briefly summarized. Anti-Human IL-21 capture antibody was bound to 96-well ELISA plates overnight (o/n) at 4 °C. IL-21 coated plates were washed three times (3×) then blocked (1 h, room temperature (RT)), then washed once. Human IL-21 internal standards (ranging from 15.625 pg/mL to 800 pg/mL) and diluted plasma samples were layered on the plate in duplicate and incubated (o/n, 4 °C). After washing (5×) plates were incubated (1 h, RT) with anti-Human IL-21 Biotin detection antibody. Plates were washed (5×) and incubated (30 min, RT) with Avidin-HRP enzyme. After washing (7×), plates were developed with TMB solution with 2N H2SO4 acid stop and read at 450 nm in a spectrophotometer.
Plasma was initially diluted with PBS during PBMC isolation (generally roughly 30–40% of full concentrated plasma). Raw ELISA data was converted to final concentrated plasma levels by accounting for this predilution. Measured raw IL-21 levels were divided by plasma dilution (plasma dilution = Volume Blood(1- hematocrit/100)/(Volume blood + Volume PBS added)).
Data analysis
Descriptive characteristics for study subjects were computed. Most variables were positively skewed, and are, therefore, described using medians, 2.5
th, and 97.5
th percentiles. Categorical variables were reported as proportions in percentage. Separate Mann-Whitney U tests were used to examine differences between patients with KD and controls for clinical variables (see Additional file
1: Table S1). Statistical tests were two-tailed with alpha of 0.05. Analyses were conducted with SYSTAT 13 (SYSTAT Software, 2004).
Enrollment survey and chart review was used to define number of symptoms consistent with KD. Spearman rank correlation coefficients (R) of laboratory value versus IL-21 levels and Mann-Whitney U tests comparing IL-21 levels between comparative groups were analyzed and graphed using Prism software (Graphpad, La Jolla, CA).
Discussion
KD is a childhood vasculitis of unknown etiology, difficult to diagnose with certainty. Since recent studies implicate B cells in the pathogenesis of KD, and IL-21 is involved in the differentiation of both naïve and memory B cells into mature plasma cells [
16], we were interested to see if IL-21 correlates with KD, as was previously shown [
15]. Despite the small number of samples present in this study, the IL-21 levels in the KD group assayed with ELISA were similar to previously published reports. Our samples showed ranges of IL-21 levels in both KD and control samples that were consistent in repeated experiments, suggesting no inherent bias in the way our samples were processed. Our results do not support IL-21 elevation as a specific marker of KD in heterogeneous North American populations.
The findings published herein agree with limited data in a larger study on IP-10 in a Han Chinese cohort of children [
17]. However, for the specific question of IL-21 levels, that study was hampered by a lack of sensitivity of the cytokine bead array (CBA) used. In our ELISA assay, all samples measured above our limit of detection, with a number in the thousands of pg/mL. Similarly, the data from the Korean cohort has data in the thousands of pg/mL with the majority of their samples in the measurable assay range of the ELISA. Of the 57 samples in the Han Chinese cohort assayed by CBA however, only six samples (two KD and four controls) had levels distinguishable from background levels with none of these levels being greater than 250 pg/mL. The lack of sensitivity of the CBA for IL-21 made conclusions from this study difficult to extrapolate. This lack of interpretable data necessitated our current study. We show a wide-range of IL-21 values in our samples and our median IL-21 level in the KD subgroup of 170 pg/mL was comparable to 499.5 pg/mL shown previously by ELISA in the Korean cohort of children [
15].
IL-21 has a potential role in other inflammatory and autoimmune disorders, such as type 1 diabetes, allergy, and asthma [
22], and inflammatory disorders such as the spondyloarthropathies [
23]. Evidence supports IL-21 being involved in the transition from innate immune responses to specific T cell responses against antigens [
13] so it would not be surprising for a number of infections to lead to IL-21 elevation as these results would support.
The prior study showing specific IL-21 elevation in KD in the Korean cohort was notable in that it included only suspected mononucleosis subjects in the control group. Since mononucleosis and KD both cause prolonged fever, we did evaluate if there was a bias in our study due to length of fever (Fig.
2). IL-21 in mononucleosis has not been well studied. The Epstein-Barr Virus, the most common cause of mononucleosis, does infect B cells directly [
24] and some literature reports a delay in IL-21 production during Epstein-Barr Viral infection [
20]. In light of our current results, it is reasonable to consider that the original study may be more significant for a lack of IL-21 elevation during evaluation for mononucleosis rather than for a specific elevation associated with KD.
Genetic background may partially explain the higher incidence and differences in clinical scoring systems in Asia compared to North America [
4,
25]. These differences support that KD is more likely an immunological peri-infectious response, [
26], rather than from direct invasion of the coronaries by an infectious agent [
27]. The genetic differences may also partially explain the disparity of the results in the studies related to IL-21. Firm conclusions regarding genetic heterogeneity are hampered by the lack of a control group from a more homogenous population and lack of knowledge of the HLA status of subjects in this current study. These areas would be of interest in pursuing in future studies.
There are other significant limitations to this study. Firstly, and most obviously, we have few KD samples. However, the range of IL-21 levels between our KD group and the previously published Korean cohort are similar [
15] and the main conclusion depends more significantly on the control group. The controls (60 subjects) shows a wide range of IL-21 elevations so we believe the conclusion that there is a non-specific IL-21 elevation is sound. Secondly, most of our control group would not have been clinically confused with KD as they lacked criteria beyond fever. A more ideal control group would be febrile children with four or five KD symptoms but given an alternative diagnosis. However, due to the number of KD cases also having co-infections with other organisms, a group such as this may be even more rare than children diagnosed with KD. Additionally, a number of the included controls would llkely have fulfilled criteria for incomplete KD (only 2 or 3 symptoms with supplemental laboratory support) without their alternative diagnosis. Thirdly, timing of inflammation may significantly affect cytokine levels as early inflammation may reflect the initial assault and later inflammation may be driven by repair. Ideally, serial samples starting early in the disease course would be obtained from subjects for these types of studies. Serial studies like this are difficult to obtain in children due to safety risk of multiple blood draws and parental concern over multiple needle sticks. Also, KD is not often recognized early in the course since the symptoms can be variable and confused with a myriad of viral and non-viral illness seen by general pediatricians. In one cohort of 152 children, only 20 were included on or before day three of illness [
18]. We did attempt to address if timing of sampling affected the IL-21 levels (Fig.
2) and show no differences on that limited data set. Repetitive sampling over time, however, would better address this question.
Conclusions
The marked predilection for acquired heart disease in children warrants research into determining the underlying physiology behind this vasculitis, both in terms of improved diagnostics as well as more appropriate treatment for typical, atypical, and incomplete KD. Unfortunately, particularly for the heterogenous populations of North America, IL-21 does not seem to be the answer. Future studies would benefit from larger cohorts, serial evaluation over time, and more extensive proteomic evaluation.
Acknowledgements
The authors would like to thank Frank Carnevale for his assistance, Haiping Qiao for assistance in establishing cohort access and obtaining samples, and Meghan McLaughlin and Hakimuddin Sojar for technical support.