# Differentiating Multisystem Inflammatory Syndrome in Children (MIS-C) from Acute COVID-19 Using Biomarkers: Toward a Practical Clinical Scoring Model

## TL;DR

This study identifies key biomarkers to distinguish MIS-C from acute COVID-19 in children, offering a potential clinical scoring tool for rapid diagnosis.

## Contribution

A compact and reproducible biomarker panel for differentiating MIS-C from acute pediatric COVID-19 is proposed.

## Key findings

- MIS-C is marked by elevated CRP, NLR, PLR, lymphopenia, and electrolyte imbalances.
- A biomarker panel achieved high predictive accuracy (AUC = 0.95) using Random Forest and Ridge regression.
- Simplified models outperformed complex ones, showing clinical utility for emergency triage.

## Abstract

Background/Objectives: Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection in children presents with a heterogeneous clinical spectrum, whereas multisystem inflammatory syndrome in children (MIS-C) is a distinct immunological entity characterized by a hyperinflammatory phenotype and a distinct biological architecture. Identifying routine biomarkers with early discriminatory utility is essential for rapid differentiation between MIS-C and coronavirus disease 2019 (COVID-19). Methods: We conducted a retrospective comparative study of 144 pediatric patients with COVID-19 or MIS-C admitted to a single specialized medical center. The analyses integrated classical statistical methods, Benjamini–Hochberg false discovery rate correction (FDR), penalized regression models, and machine learning algorithms to identify biomarkers with discriminative value, using only routine laboratory tests. Results: MIS-C was associated with an intense inflammatory profile, characterized by increases in C-reactive protein (CRP), neutrophil-to-lymphocyte ratio (NLR), and platelet-to-lymphocyte ratio (PLR), lymphopenia, and selective electrolyte disturbances, highlighting a coherent biological architecture. In contrast, COVID-19 showed limited associations with traditional inflammatory markers. Predictive models identified a stable core of biomarkers with excellent performance in Random Forest analysis (area under the curve, AUC = 0.95), and reproducible thresholds (CRP ~3.7 mg/dL, NLR ~3.3, PLR ~376; potassium ~4.2 mmol/L). These findings were independently confirmed using penalized Ridge regression, where the reduced model achieved superior discrimination compared to the full 13-variable model (AUC = 0.93 vs. 0.89) and maintained stable performance under internal cross-validation, reinforcing the clinical relevance of this compact biomarker panel. Conclusions: MIS-C is clearly distinguished from COVID-19 by a specific and reproducible immunological signature. The identified biomarkers may represent a potential foundation for the development of simple clinical algorithms for pediatric triage and risk stratification, opening the prospect of a simplified scoring tool applicable in emergency settings.