Similarities and differences between the immunopathogenesis of COVID-19-related pediatric multisystem inflammatory syndrome and Kawasaki disease

Abstract

Multisystem inflammatory syndrome associated with the SARS-CoV-2 pandemic has recently been described in children (MIS-C), partially overlapping with Kawasaki disease (KD). We hypothesized that (a) MIS-C and prepandemic KD cytokine profiles may be unique and justify the clinical differences observed, and (b) SARS-CoV-2-specific immune complexes (ICs) may explain the immunopathology of MIS-C. Seventy-four children were included: 14 with MIS-C, 9 patients positive for SARS-CoV-2 by PCR without MIS-C (COVID), 14 with prepandemic KD, and 37 healthy controls (HCs). Thirty-four circulating cytokines were quantified in pretreatment serum or plasma samples and the presence of circulating SARS-CoV-2 ICs was evaluated in MIS-C patients. Compared with HCs, the MIS-C and KD groups showed most cytokines to be significantly elevated, with IFN-γ-induced response markers (including IFN-γ, IL-18, and IP-10) and inflammatory monocyte activation markers (including MCP-1, IL-1α, and IL-1RA) being the main triggers of inflammation. In linear discriminant analysis, MIS-C and KD profiles overlapped; however, a subgroup of MIS-C patients (MIS-Cplus) differentiated from the remaining MIS-C patients in IFN-γ, IL-18, GM-CSF, RANTES, IP-10, IL-1α, and SDF-1 and incipient signs of macrophage activation syndrome. Circulating SARS-CoV-2 ICs were not detected in MIS-C patients. Our findings suggest a major role for IFN-γ in the pathogenesis of MIS-C, which may be relevant for therapeutic management.

Keywords: COVID-19; Chemokines; Cytokines; Immunology; Vasculitis.

Figure 1. SARS-CoV-2–specific immunoglobulin detection.

Determination of IgG, IgA, and IgM specific for the receptor-binding domain (RBD) of the SARS-CoV-2 spike protein. Open symbols represent patients with positive PCR in nasal swab or stools. Dotted lines represent threshold for positivity. MFI, mean fluorescence intensity; MIS-C, multisystem inflammatory syndrome in children (n = 13); COVID, pediatric patients with mild COVID-19 disease without MIS-C (n = 9); HC, healthy controls (n = 37 [53 samples]).

Figure 2. Cytokine profile of the different groups.

Levels of 34 circulating cytokines were evaluated in serum or plasma. (A) Principal component analysis: every data point represents one individual. (B) Total variance explained by principal component 1 (PC1) and PC2 and variables with higher scores. (C) Representation of canonical variants (Can1 and -2) of the linear discriminant analysis (LDA) built with the cytokines represented in B. (D) Representation of Can2 and Can3 of the LDA (HCs have been removed from the plot, not from the analysis). A “+” indicates subgroup of MIS-C patients differentiated from other MIS-C and KD patients (MIS-C^plus). Open symbols represent patients who previously received immunoglobulin treatment. (E) Heatmap representing the selected cytokines (represented in panel B). Every column represents a patient. An asterisk (*) identifies the cytokines with statically significant differences between KD+MIS-C and MIS-C^plus, and a “+” identifies cytokines with statically significant differences between MIS-C and MIS-C^plus and not with KD. Color gradient represents the z score. KD, pre–SARS-CoV-2 pandemic Kawasaki disease (n = 14); MIS-C, multisystem inflammatory syndrome in children (n = 13); COVID, pediatric patients with mild COVID-19 disease without MIS-C (n = 9); HC, healthy controls (n = 37 [53 samples]).

Figure 3. Leave-one-out cross-validation of linear discriminant analysis model.

Rows represent the preclassified groups and columns the predicted categories after leave-one-out cross-validation. KD, pre–SARS-CoV-2 pandemic Kawasaki disease (n = 14); MIS-C, multisystem inflammatory syndrome in children (n = 13); COVID, pediatric patients with mild COVID-19 disease without MIS-C (n = 9); HC, healthy controls (n = 31 [53 samples]).

Figure 4. Quantitative circulating levels of selected cytokines.

Quantitative levels of cytokines with significant differences between MIS-C (n = 7) and MIS-C^plus (n = 5) are shown. Open symbols represent plasma samples and filled symbols are serum samples. Bars represent median values. Described values for IFN-γ in (i) viral infections range from 7.7 to 33.4 pg/mL (33); (ii) suspected primary hemophagocytic syndrome with or without genetic confirmation range from 147.6 to >5000 pg/mL (33) and a mean (95% CI) of 905.7 pg/mL (530.7–1280.6 pg/mL, ref. 35); and (iii) patients with MAS in systemic-onset juvenile idiopathic arthritis mean (IQR) of 15.4 pg/mL (5.1–52.6 pg/mL, ref. 36). KD, pre–SARS-CoV-2 pandemic Kawasaki disease (n = 14); MIS-C, multisystem inflammatory syndrome in children (n = 13); COVID, pediatric patients with mild COVID-19 disease without MIS-C (n = 9); HC, healthy controls (n = 37 [53 samples]).