This was a single center observational prospective study with a convenience sample consisting of patients diagnosed with MIS-C between one month and 18 years of age, with confirmed SARS-CoV-2 infection (molecular test), and who required hospitalization in PICU. Patients were recruited from a tertiary Service in Belém, state of Pará, Brazil, between May 1, 2020, and December 31, 2023. All patients included in this study fulfilled the MIS-C case definition of World Health Organization (WHO) (4). Patients with similar presentation but with positive microbiology other than SARS-CoV-2 and those with immunosuppression status were excluded. Ethics approvals were obtained from the institutional review board of the Institution. We obtained a waiver of written informed consent to conduct this observational study.

Recognizing the unprecedented challenges posed by MIS-C and the need for comprehensive biomarker analysis, a task force was organized to support both the clinical and research aspects of this study. In practical terms, despite the designation as a single-center study, an informal network of pediatric intensivists in the region collaborated to facilitate active case identification. As MIS-C cases were identified in other local PICUs, the attending physicians promptly contacted the lead investigator to expedite the transfer of patients meeting MIS-C criteria to the primary institution for consistent clinical and laboratory management. This approach ensured a streamlined and uniform protocol for patient assessment and sample collection. The biomarker analyses were conducted within this centralized framework, though specific aspects required external partnerships for optimized processing. The oxidative stress measurements were conducted in collaboration with the Laboratory of In Vitro and Microbiology Assays at the Federal University of Pará, and cytokine level assessments were performed at the Evandro Chagas Institute’s Immunology Laboratory. These collaborations aimed to enhance the diagnostic and prognostic understanding of MIS-C, especially considering the notably higher mortality rates observed in our region compared to global data. As for the ethics approval process, the institutional review board approval was obtained solely from the primary hospital, as this institution served as the formal site of data collection, patient care, and biomarker analyses.

The PICU fulfil international criteria (23) for quaternary care levels, characterized by having a high level of resources and access to Mechanical Ventilation (MV), Renal Replacement Therapy (RRT), vasoactive/inotropic medications, and treatments targeted to MIS-C (Intravenous Immunoglobulin [IVIG] and steroids). They also serve as training centers for medical and health professionals.

Patients were splitted into two groups based on the main outcome, i.e., in-hospital mortality up to 28^th day: survivors and non-survivors. Presence of cardiogenic shock, critical disease, and length of stay at the Hospital were evaluated. The 28-day mortality endpoint was assessed according to current pattern of outcome measures of clinical studies in order not to achieve an ambiguous measure (24). All outcomes and the worst features were recorded and compared between the 2 groups at admission and on the third day.

Clinical and laboratory data as well as demographic and epidemiological characteristics were collected from electronic records of the patients in PICU. Intensive support characteristics were also assessed sequentially until the end of hospitalization. The patients presented critical or non-critical disease according to the multiple organ dysfunction syndrome (MODS) but at least two organic dysfunctions should be present (25–28).

The presence of critical illness was determined by the finding of MODS which was defined as the simultaneous occurrence of three or more organ dysfunctions according to previously published criteria (25–28). Some patients in our study were hospitalized in the PICU only for monitoring, and they were not considered for classification in critical illness. The definition and categorization of comorbidities were based on the classification by Feudtner et al., named Pediatric Complex Chronic Conditions (PCCC) (29).

In the current study, circulatory shock was defined by altered levels of consciousness, decreased urine output (< 1 mL/kg/h or 12 mL/m²/h), and lactic acidosis, with arterial hypotension being a late sign of shock in children. Meanwhile, cardiogenic shock was defined by the presence of an ejection fraction less than 55%, as assessed by a pediatric echocardiographer using the Teichholz method, troponin elevation to twice the reference value, and the presence of clinical signs of circulatory shock (30, 31). Severely immunocompromised patients: HIV carriers; patients on immunosuppressive therapy post-solid organ or bone marrow transplant; neutropenic patients (below 500 neutrophils/mm³ or below 1000/mm³ with a downward trend); patients undergoing chemotherapy in the last 28 days or other immunosuppressive therapies; patients on immunosuppressive dose of corticosteroid (equivalent to 2mg/kg/day of prednisone for more than 2 weeks in those under 10 kg or 20mg/day for more than 2 weeks or 40mg of prednisone for ≥ 10 days in those over 10 kg) (32).

Blood samples were drawn by venipuncture from all patients on day 1 and 3 of the admission, before the initiation of immunomodulatory therapy. For complete blood count parameters, 2 mL of blood was taken into in tubes containing ethylenediaminetetraacetic acid (EDTA). The plasma obtained was stored at − 80°C. Next, cytokines and oxidative stress biomarkers were measured.

Plasma levels of seven cytokines (IL-2, IL-4, IL-6, IL-10, TNF, IFN-γ and IL-17A) were analyzed after filtering proteins with measurements > 30% below the detection threshold, by CBA according to the instruction of BD™ Cytometric Bead Array (CBA) Human Th1/Th2/Th17 Cytokine Kit (Becton & Dickinson, CA, USA). Measurement by flow cytometry was performed using a FACSCanto™ II cytometer (Becton & Dickinson, CA, USA). Plasma levels of the respective cytokines were expressed in pg/ml.

The nitrate (NO₃ ⁻) present in the serum samples was converted to nitrite (NO₂ ⁻) with nitrate reductase, and the nitrite concentration was determined using the Griess method (33). In brief, 100 μL of the supernatant samples were incubated with an equal volume of Griess reagent for 10 min at room temperature. The absorbance was measured on a plate scanner (Spectra Max 250; Molecular Devices, Menlo Park, CA, USA) at 550 nm. The nitrite concentration was determined using a standard curve generated using sodium nitrite (NaNO₂). Nitrite production was expressed per μM.

The MDA was used for the reaction of thiobarbituric acid reactive substances (TBARS) performed according to the adapted form (34) of a previously proposed method (35). An aliquot of 1 mL of the reagent (TBA 10 nM) and 0.5 mL of the sample was added to each test tube. Then, the tubes were placed in a water bath at 94°C for 1 h. After this procedure, the samples were cooled under running water for about 15 minutes, and then 4 mL of butyl alcohol was added to each sample. Subsequently, the samples were mixed on a vortex shaker, in order to obtain the maximum extraction of MDA into the organic phase. Finally, the tubes were centrifuged at 2,500 rpm for 10 minutes. A volume of 3 mL of supernatant was pipetted to carry out spectrophotometric reading at 535 mm. Results were expressed in nmol/mL^-1.

The trolox equivalent antioxidant capacity (TEAC) is a sensitive and reliable marker for detecting in vivo oxidative stress markers that may not be detectable through the measurement of a single specific antioxidant (36). TEAC level of the plasma serum samples was measured using a previously developed method (37). In this assay, 7 mM of 2,2-azinobis, 3-ethylbenzothiazoline, 6-sulfonate (ABTS) were incubated with 2.45 mM of potassium persulfate and ABTS-potassium persulfate (1: 0.5, v/v), and the mixture was allowed to stand in the dark at room temperature for 12–16 h before use. For the study, the blue green ABTS+ solution was diluted with ethanol 95% (v/v) until the absorbance reached 0.70 ± 0.02 at 734 nm. Then, 10 μL of the plasma or trolox standard was mixed with 1 mL of ABTS+ solution, and a decrease in absorbance at 734 nm was recorded after 4 min for all samples. The absorbance of the mixture was monitored at 734 nm after 6 min. For the blank, 10 μL of water instead of the sample was used and each sample was measured in triplicate. Total antioxidant potential of plasma serum was expressed as μmol/mL of TEAC and was calculated through a calibration curve plotted with different amounts of trolox (38).

Catalase (CAT) activity was determined according to a previously validated method (39). Blood samples were hemolyzed into ice water (1: 3) and then diluted in a Tris based buffer (Tris 1 M/EDTA 5 mM, pH 8.0). To verify the decay of hydrogen peroxide (H₂O₂), aliquots of the diluted samples were added to 900 μL of reaction solution (Tris base, H₂O₂ 30% and ultrapure water, pH 8) (40). The decrease of H₂O₂ concentration was established at λ = 240 nm at 25°C for 60 seconds. CAT activity was defined as the activity required to degrade 1 mol of H₂O₂ for 60 seconds (pH 8 and 25°C) and was expressed as U/mg protein. The molar extinction coefficient of H₂O₂ used for the calculation was 39.4 cm (2)/mol. The enzymatic activity data obtained in CAT assays were normalized by the total protein concentrations, using the commercial kit (Doles, Brazil).

Determination of intracellular GSH levels was based on the ability of GSH to reduce 5,5-dithiobis-2-nitrobenzoic acid (DTNB) to nitrobenzoic acid (TNB), which was quantified by spectrophotometry at 412 nm. The methodology described by Ellman (41) as adapted for this determination, and GSH concentrations, were expressed in µmol/mL. This assay was adapted (42) for use in a microtitre plate using a microplate spectrophotometer system, Spectra MAX 250 (Molecular Devices, Union City, CA, USA).

The cohort characteristics were summarized using median (interquartile range) for continuous variables, and absolute numbers and frequencies (%) for categorical variables. Comparisons between groups were made using χ² or Fisher exact tests for categorical variables, and Mann–Whitney U was used for continuous data. For comparison between day 1 and day 3 the Wilcoxon signed-ranks test was used. Holm-Bonferroni correction for multiple comparisons was employed adjusting the significance level when patients were compared between two groups (survivors and non-survivors).

We used the ROC (Receiver Operating Characteristic) analysis to choose the optimal cut-point associated with mortality outcome. Survival rate in days was calculated, and Kaplan-Meier curves were plotted. We performed a univariate and multivariate Cox regression “time - to first - event” analysis to determine risk factors for survival. In the regression model we adjusted for a risk of mortality using age under one year of age and presence of comorbidity simultaneously. For these analyses the adopted significance level was set at two-sided p <0.05. The ROC curve analysis was performed to determine the effectiveness of IL-17A as a biomarker to predict the likelihood of death in patients. Data analyses were conducted using SPSS (Statistical Package for the Social Sciences, Chicago, IL) version 28.0.

There were 109 patients admitted at PICU with suspected MIS-C during the study period. A total of 41 patients (median age of 55 months) were included ( Figure 1 ). Thirty-three (80.5%) were male, nine (22%) were up to one year of age, and 30 (73.2%) presented PCCC. The median time from SARS-CoV-2 exposure to symptom onset was 23 days (IQR: 15-34, days). Critical disease and presence of more than three organ dysfunctions occurred in 21 (52.1%) and 26 (63.4%) patients, respectively. Invasive mechanical ventilation was required in 22 (53.7%), and deaths occurred in 16 (39%) patients ( Table 1 ).

The assessment of non-specific laboratory tests between day 1 and day 3 revealed significant improvement in lymphocytes and platelets as well as in inflammatory markers, such as CRP, ESR and ferritin ( Table 2 ). The analysis of oxidative stress markers demonstrated significant differences between Day 1 and Day 3. Levels of TEAC and GSH exhibited a notable upward trend, with TEAC (mmol/L) increasing significantly from 1.67 to 2.06 (p < 0.001) and GSH (µmol/mL) rising from 34 to 37 (p = 0.02). These changes are indicative of enhanced oxidative defense mechanisms and potentially reflect increased metabolic activity during the hospitalization period. Conversely, MDA (mmol/mL) levels decreased significantly from 3.62 to 1.96 (p = 0.002), catalase (U/mg protein) levels declined from 0.042 to 0.035 (p < 0.001), and nitrite (µM/mg protein) levels were markedly reduced from 80.5 to 54.3 (p < 0.001). These reductions suggest a diminishing inflammatory and oxidative state, likely corresponding to clinical improvement and a positive therapeutic response. In contrast, cytokine levels did not exhibit significant differences between Day 1 and Day 3, as demonstrated in Table 2 .

There were 25 survivors and 16 non-survivors. Regarding cytokines, non-survivors presented higher levels of IL-17A, IL-6 and IL-4 compared to survivors on day 1 (p<0.001, p=0.001, and p=0.001, respectively) and day 3 (p<0.001, p=0.002 and p<0.001, respectively). Regarding oxidative stress markers, non-survivors presented higher levels of plasma nitrite in both days (p<0.001), and lower levels of glutathione reductase on day 3 (p<0.001) compared with survivors, as shown in Table 3 . The elapsed time between admission and the initiation of immunomodulatory therapy did not reveal difference (p=0.575) between non-survivors (median 3.5 days, IQR 3.2-4.5) and survivors (median 3.5 days, IQR 3.6-5.2).

Cardiogenic shock was present in 17 (41%) patients; levels of IL-4 were significantly higher on day 3 in the group with cardiogenic shock compared to the group without cardiogenic shock (p=0.001), with no other differences among the other inflammatory biomarkers ( Table 4 ).

In relation to the length of stay in PICU, 23 (56%) stayed longer than 7 days; none of the inflammatory biomarkers stood out in relation to this outcome ( Table 4 ). We also carried out a comparison between critical and non-critical patients, however the results were similar to the mortality outcome.

Patients with higher levels of IL-17A (≥ 19.71 pg/mL) on day 1, higher levels of IL-4 (≥ 11.89 pg/mL) on day 1, higher levels of IL-4 (≥14.96 pg/mL) on day 3 and higher levels of NO (≥75.36 µM/mg protein) on day 1 presented shorter survival time (p=0.004, p=0.002, p=0.001, p=0.009, respectively). On the other hand, patients with lower levels of GSH (< 0.036 µmol/mL) on day 3 (p=0.009), presented shorter survival time, as showed in Figures 2 – 6 .

After multivariate analysis, IL-17A was the only factor associated with shorter survival time on day 1 either in crude (HR 1.03, CI95%1.004-1.057, p=0.022) or adjusted models (HR 1.043, CI95%1.013-1.075, p=0.012) as shown in Table 5 .

As cut-point values of inflammatory biomarkers for mortality have not yet been described in MIS-C patients, we chose to perform ROC analysis to identify the best possible cut-point associated with mortality outcome after we plotted the Kaplan-Meyer curves and performed univariate and multivariate Cox regression. We observed that the cut-off value for the IL-17A was 14.32 pg/ml with sensitivity of 99% and specificity of 80%, as showed in Figure 7 .