How multisystem inflammatory syndrome in children discriminated from Kawasaki disease: a differentiating score based on an inception cohort study

Background About 25–50% of multisystem inflammatory syndrome in children (MIS-C) patients meet the criteria for diagnosis of Kawasaki disease (KD). The differentiation of both conditions is so challenging on clinical practice as the management of both is time dependant and precise diagnosis is fundamental. Method Data were collected from children < 18 years old hospitalized with MIS-C or KD. Patient demographics, clinical, and laboratory data were compared, and a discrimination score was created to assist in clinical differentiation. Results 72 patients with MIS-C and 18 with KD were included in the study. Patients with MIS-C had a higher prevalence of abdominal pain (p = 0.02), vomiting (p = 0.03), and cervical lymphadenopathy (p = 0.02) compared with KD cases. MIS-C patients had higher liver enzymes (aspartate aminotransferase (AST) (p = 0.04), alanine aminotransferase (ALT) (p = 0.03), serum creatinine (p = 0.03), and lower platelet count nadir (p = 0.02) than KD. Four variables were detected in the regression analysis model, and the independent predictors were utilized to generate a scoring model that distinguished MIS-C from KD with an area under the curve of 0.70. Conclusion This study constructed a prediction model for differentiation of MIS-C from KD based on clinical and laboratory profiles. This model will be valuable to guide clinicians in the treatment decisions. Key Points • Children with MIS-C are more likely to have gastrointestinal symptoms, cervical lymphadenopathy, and respiratory involvement than KD patients. • Elevated liver enzymes and lower platelet count are more pronounced laboratory findings in MIS-C than KD. • This study constructed a prediction model for differentiation of MIS-C from KD based on clinical and laboratory profiles. This model will be valuable to guide clinicians in the treatment decisions.


Introduction
As the COVID-19 pandemic had spread across the world, a novel hyperinflammatory state dubbed as multisystem inflammatory syndrome in children (MIS-C) had emerged. MIS-C is an immune activation syndrome but not an acute infectious process, and patients do not currently have active infection [1].
The main issue is that about 25-50% of MIS-C patients meet the criteria for diagnosis of Kawasaki disease (KD) and resembles its presentation. Furthermore, there was no evidence of prior severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2) exposure in these patients; hence, it might be so difficult to discriminate them from those with KD [2]. The differentiation of both conditions is so challenging on clinical practice as the clinical management of both is time dependant and precise diagnosis is fundamental [3,4].
According to the center for disease control and prevention (CDC), MIS-C is defined as an individual < 21 years with fever, laboratory evidence of inflammation, with multisystem (≥ 2) organ involvement (cardiac, mucocutaneous, respiratory, hematologic), and no alternative diagnoses. In addition to the positive tests for SARS-CoV-2 infection by polymerase chain reactant (PCR), serology, antigen test; or COVID-19 exposure within the 4 weeks prior to the onset of symptoms [5].
Meanwhile, KD predominantly affects children between 6 months and 5 years of age. The prevalence is higher in children of Asian descent. It is a disease of 5 days fever or more and at least 4 of the following: rash, cervical lymphadenopathy, bilateral conjunctival injection, oral mucosal changes, and peripheral extremity changes [6].
This hyperinflammatory state and KD are overlapping in many clinical presentations such as high fever, diffuse skin rashes, mucous membrane changes, conjunctivitis, extremities erythema/edema, and cervical lymphadenopathy [7][8][9]. However, apparent differences in clinical and laboratory findings have been observed between the two syndromes. For example, gastrointestinal (GI) symptoms (vomiting, abdominal pain, or diarrhea) and neurological complications are more noticeable in MIS-C [10][11][12].
Cardiac presentations of the two diseases are quite diverse, the main complication of KD is coronary artery aneurysms in < 10% of cases [13,14]. By contrast, coronary artery dilation in MIS-C has an incidence rate of 14-36% [15,16]. Meanwhile, the most common cardiovascular complications in MIS-C are ventricular dysfunction, myocarditis, and pericardial effusion. Besides, 33-87% of MIS-C patients experienced shock and hemodynamic instability that required inotropic support, while 2-7% of KD children developed shock [17,18].
Pulmonary symptoms are usually absent in KD, but in MIS-C, respiratory involvement usually consists of respiratory insufficiency, pleural effusions, and pulmonary infiltrates [19,20]. In addition, hypercoagulable state occurs in both entities but is more pronounced in MIS-C [10,21].
Regarding the dissimilarities in laboratory features of MIS-C and KD, the inflammatory marker elevation observed in KD cases is moderate compared to MIS-C [22,23]. Moreover, lymphopenia and thrombocytopenia are present in 37-81% of MIS-C patients [19,24,25], whereas they are rare in KD. Another characteristic of MIS-C is a marked increase in the biochemical markers of cardiac injury such as troponin T and B-type natriuretic peptide [4,26].
Studies of both KD and MIS-C about the disease immunopathogenesis showed increased interferon-γ (IFN-γ)associated cytokines, interleukin-1 (IL-1), IL-6, IL-18, and tumor necrosis factor (TNF) signaling pathways [7,27,28]. However, the subtle difference between MIS-C and KD presentation might be related to the different pathogenesis of the two diseases; compared with MIS-C patients, IL-1, IL-17A, and anti-endothelial cell antibodies are significantly higher in KD with the lacking of polyclonal proliferation of TCR Vβ 21.3 + activated CD4 + and CD8 + T cells [29,30].
The distinction between MIS-C from KD can be challenging for health care practitioners, as prompt recognition of both patients is crucial for appropriate treatment. The current study underpins the demographic, clinical, and laboratory differences between both conditions that can assist in establishing the diagnosis and optimal management. In addition to a simple scoring model that will be valuable for distinguishing MIS-C from KD and guide clinicians decisions in future studies.

Subjects
This an inception cohort study was conducted on a group of children who attended to Mansoura University Children's Hospital (MUCH) from January 2020 to December 2021.
1. All children with a diagnosis of either complete or incomplete KD according to the American Heart Association criteria (AHA) diagnostic criteria [6] with negative link to COVID-19 infection (negative SARS-CoV-2 PCR, serology, antigen test, and negative contact history). 2. All children with CDC's MIS-C case definition [5]. 3. Patient whose parents signed a consent to be included in the study.
Exclusion criteria: Patients who did not fulfil any of the above-mentioned definitions were excluded from the study.

Data collection
Data were collected from our medical files and interpreted with respect to the followings: 1. Demographic: age, sex, contact history, and duration of symptoms before admission in days. 2. Clinical features and course: fever, GI, central nervous system (CNS), respiratory, cardiovascular, or mucocutaneous involvement. Besides, disease complications, intensive care unit (ICU) admission, its duration, treatments details (anti-platelets, intravenous immunoglobulins (IVIG), steroids), and outcomes. 3. Laboratory findings at the peak of the disease: complete blood count (CBC), erythrocyte sedimentation rate (ESR), C-reactive protein (CRP), alanine aminotransferase (ALT), aspartate aminotransferase (AST), serum ferritin, serum creatinine, D-dimer, SARS-CoV-2 PCR, and SARS-CoV-2 antibody test. However, cardiac biomarkers were not routinely done for all cases so no highlights about them in this article. 4. In addition to imaging, such as echocardiography (ECHO) and high resolution computed tomography (HRCT) chest.

Statistical analysis
Data were analyzed using the IBM-Statistical Package of Social Science (SPSS) program for Windows (version 22). The normality of data was first tested. Qualitative data were described using numbers and percentages. Association between categorical variables was tested using the chisquare test, while the Fischer exact test was used when the expected cell count less than 5. Continuous variables were presented as mean ± SD for normally distributed data and median (min-max) for non-normal data. The two groups were compared with the Student t-test for normal data and the Mann Whitney test for non-normal data. The ability of each variable to differentiate MIS-C from KD was estimated with sensitivity, specificity analysis, and the best cut-off detection of continuous variables. Significant variables entered into the logistic regression model to predict the most significant determinants and to control for possible interactions and confounding effects. Odds ratio (OR) were calculated with 95% confidence intervals. Sensitivity and specificity at different cut-off points were tested by the receiver operating characteristic (ROC) curve. Besides, the estimated probability of a child to have MIS-C versus KD based on the weighted regression results was calculated at the score and cut-off points. For all the above-mentioned statistical tests, the results were considered significant when p ≤ 0.05.

3-Results
A total of 90 patients were included of which 72 cases (80%) with MIS-C, 18 children (20%) with KD. Table 1 exhibits a description of demographics, clinical features, and outcomes of both categories. History of contact with probable or confirmed COVID-19 cases was negative in 60% of MIS-C cases. Among patients with MIS-C, 65.3% were male, whereas they represented 77.8% of cases in KD group. Most of MIS-C cases were older than 6 years old (54.2%). About 40.3% MIS-C children were admitted to ICU, which was higher than for KD (5.6%; p = 0.005). Patients with MIS-C generally had longer median ICU days (7 days) compared with patients with KD (5 days; p < 0.015). Twenty-one patients died in MIS-C cases (29%), but no deaths among patients with KD. The most frequently reported clinical findings were fever (90%), vomiting (50%), and skin rash (44.4%) in both groups.
Diarrhea, cardiovascular (myocardial dysfunction, coronary artery dilation, valve involvement, and shock/hypotension), neurologic (headache, seizures, and disturbed consciousness), or renal presentations were more likely to occur in MIS-C patients compared to KD but did not attain a statistically significant difference. Management options for each group were displayed at the end of Table 1.
A summary of laboratory difference between KD and MIS-C is shown in Table 2. On comparison with KD, patients with MIS-C had higher liver enzymes (AST (p = 0.04), ALT (p = 0.038)), serum creatinine (p = 0.036), and lower platelet count nadir (p = 0.025). Despite of lower lymphocyte count, higher peak CRP, and serum ferritin in MIS-C patients than KD, the difference did not reach a statistical significance. Analysis of sensitivity, specificity, negative predictive value, positive predictive value, and accuracy of each significant parameter was shown in Table 3.
Four variables were detected in the univariate logistic analysis including abdominal pain, vomiting, cervical lymphadenopathy, and platelet nadir, which were the significant predictors of MIS-C from KD. However, on multivariate analysis, vomiting and cervical lymphadenopathy were detected as the independent factors for discriminating MIS-C from KD (Table 4).
Significant variables were assigned cut-off and scoring points based on the β coefficient (Table 5). These variables were utilized to generate a scoring model that yielded an AUC of 0.70 (95% CI 0.55-0.80). At the cut-off of 4 or more points, the sensitivity, specificity, and accuracy were 57%, 85.4%, and 57.8%, respectively, to distinguish MIS-C from KD in the scoring model (Table 6) (Fig. 2). Table 7 displays the likelihood of a child having MIS-C compared with KD. A cut-off greater than or equal to 4 suggests > 99% probability that this patient has MIS-C not KD.

Discussion
Many patients with MIS-C meet the criteria for KD that can make initial diagnosis and treatment decisions challenging. Hence, understanding and analysis of the subtle differences in presentation and laboratory markers by this comparative study can help in discrimination between both disorders.
The findings of this report were similar to the results of previously published studies that comparing MIS-C with KD patients specifically the male predominance, older age of presentation, longer, and higher rate of ICU admission in MIS-C [9,20,31].
Our study confirmed the preceding results given that MIS-C was commonly associated with GI symptoms (abdominal pain and vomiting) than those with KD [32][33][34][35]. Moreover, we observed that cervical lymphadenopathy was more pronounced in MIS-C patients rather than KD children [31,36].
As expected, respiratory involvement (dyspnea, crepitation, and pneumonia) was rarely seen in KD patients, whereas in MIS-C, it occurred in 20-80% of our patients in line with previous case series [12,16] [26] observed cardiac complications of MIS-C are common, including myocardial dysfunction, valve involvement, and coronary artery dilatation. Besides, the high rate of hypotension and shock required inotropic support in severe MIS-C [19,23,25].
Neurological complications or acute kidney injuries are more frequently reported in MIS-C patients than KD [19]. On the other hand, KD had a higher rate of dermatologic symptoms than MIS-C [36]. We supported the clinical evidence already published in the literature, but no statistical significant difference observed between the two groups regarding these organs involvement attributed to the small sample size in both groups. Compared to KD, patients with MIS-C had lower platelet count that provided the highest biomarker specificity in MIS-C discernment similar to Kaushik et al. (2020) [38] and Riphagen et al. (2020) [17] conclusions. This difference is due to the variation in pathogenesis; the pathogenesis of KD is mediated by inflammatory cells that recruit more platelets. However, in MIS-C, the viral-associated hyperinflammation induce bone marrow suppression [8,39].
Furthermore, it was noted that MIS-C patients had lower levels of leukocytes, absolute lymphocyte count, and hemoglobin than KD; nevertheless, we could not gather this dissimilarity which might be due to age and sex variability of these parameters or analytical issues that reduce the reliability of these results [40]. MIS-C patients had higher levels of ALT, AST, and serum creatinine than KD patients comparable with previously published results [31]. Systemic inflammation in MIS-C can involve multiple organs, such as liver, kidney, or heart, that lead to their biomarker elevation [40].
Both MIS-C and KD patients showed elevated inflammatory markers including CRP and serum ferritin attributed to hyperinflammatory response in children, although more raise of inflammatory markers may be observed in MIS-C  [41,42]. Our cohort exhibited elevation of these acute phase reactants in MIS-C cases, but failed to show significant difference from KD patients. Even more, we compare CRP cut-off > 99.55 mg/L as a guide to discriminate between MIS-C and KD as suggested by Ganguly et al. (2022) [43], but there was no difference between the two groups.
In regression analysis, abdominal pain, vomiting, cervical lymphadenopathy, and platelet nadir were detected as the significant predictors of MIS-C from KD. These variables were utilized to generate a scoring model that yielded an AUC of 0.70 (95% CI 0.55-0.80), and the total score was 7 when all criteria added together.   [44] who developed a diagnostic score to distinguish MIS-C from KD. In these scores, age > 5 years, GI involvement, headache, pericardial effusion, D-dimer > 607 ng/ml, low platelets, elevated CRP, and the absence of rash or mucocutaneous lesions were the main predictors of MIS-C. The sum of these scores provided a high probability of MIS-C rather than KD.
Nevertheless, their KD patients were a historic cohort compared with newly diagnosed patients with MIS-C. Besides, Godfred-Cato et al. [44] differentiated MIS-C from multiple mimicking conditions like COVID-19, KD, and TSS with much diversity of patient data analysis. Their retrospective data gathering might be associated with some missing information. They included pericardial effusion as an ECHO finding that would not be readily available in the outpatient rooms. In addition, the Kostik's score [31] incorporated D-dimer, which might not be accessible in some settings.
The emphasis of our observation relies on the fact that our KD patients were recruited during the same period of MIS-C enrolment, which reflects the real practice with a  standard patient's definition. It is a timely simple score that is based on initial clinical characteristics of patients and first tier universally available and routinely ordered investigations at all centers without the need for more specialized tests such as troponin, B-type natriuretic peptide, waiting ECHO, or other imaging modalities. These features making our score an effective differentiating tool at the point of preliminary evaluation before more costly or unavailable testing is ordered. Our report suggests a cut-off of four in the scoring model to provide a simplified tool to distinguish MIS-C from KD. Nevertheless, more stringent studies with prospective analysis on larger series are critical to validate the proposed scoring model before practical application, as this score is not intended to be diagnostic criteria for MIS-C. In addition, our study's limitations are that there were some missing data such as cardiac biomarkers, coagulable state, and small sample size. Future external validation with a power calculation using different sets of cases can improve its application, and we suggested that about 100 KD and 400 MIS-C patients are needed to get a clinically valuable scoring system.

Conclusion
Children with MIS-C are more likely to have GI symptoms, cervical lymphadenopathy, and respiratory involvement. Moreover, elevated liver enzymes and lower platelet count are pronounced in MIS-C. This study constructed a prediction model for differentiation of MIS-C from KD based on clinical and laboratory profiles. This model will be valuable to guide clinicians in the treatment decisions.
Author contribution The authors shared in data collection and the manuscript writing, and all have read and agreed to the published version of the manuscript.
Funding Open access funding provided by The Science, Technology & Innovation Funding Authority (STDF) in cooperation with The Egyptian Knowledge Bank (EKB).
Data availability All data generated during this study are in this published article.

Disclosures None.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.  Table 7 The estimated probability of MIS-C versus KD at the score and cut-off points * Calculation of the probability based on each score separately cannot be applied, as some categories are zero b Cut-off of the diagnostic score calculated by ROC curve Score points* Probability of MIS-C Probability of KD