Abstract Kawasaki disease (KD) remains the leading cause of acquired heart disease in children. While intravenous immunoglobulin (IVIG) represents standard therapy, approximately 10–20% of patients exhibit treatment refractoriness associated with significantly elevated coronary artery lesion risk. Current risk stratification relies on clinical parameters—fever duration or nonspecific inflammatory markers—which fail to capture distinct immunopathological features associated with treatment failure. To bridge this gap, we employed a convergent strategy integrating transcriptomic discovery from two independent cohorts with prospective protein-level and cellular validation in a clinical cohort. Three machine learning models—LASSO regression, Random Forest, and XGBoost—independently identified TREM1 (Triggering Receptor Expressed on Myeloid cells-1) as a core resistance predictor. Prospective validation in 117 KD patients demonstrated that plasma soluble TREM-1 (sTREM-1) was markedly elevated in IVIG-resistant cases (median 1247 vs. 856 pg/mL, P < 0.001) and effectively differentiated KD from other febrile illnesses. Mechanistically, sTREM-1 elevation showed a moderate positive correlation with myeloperoxidase-DNA complexes (ρ = 0.612, P < 0.001), indicating that TREM-1-associated neutrophil hyperactivation is associated with excessive neutrophil extracellular trap (NET) formation and vascular injury. External validation (GSE63881) confirmed TREM1 upregulation (2.1-fold) in resistant patients. Incorporating sTREM-1 into a multivariable prediction model achieved area under curve 0.884 using Firth’s penalized logistic regression (optimism-corrected AUC 0.871) in this derivation cohort, significantly outperforming traditional clinical scores. These findings suggest TREM-1-associated NETosis as a potential contributor to the pathobiology of IVIG resistance. As a TRIPOD Type 1b derivation study, external clinical validation in independent multi-center cohorts is required prior to clinical implementation.
Wang et al. (Mon,) studied this question.