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 plateletto-
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.
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 plateletto-
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.