Introduction: Predicting resistance to intravenous immunoglobulin (IVIG) remains a major challenge in Kawasaki disease (KD), particularly in non-Asian populations. The Kawanet score was developed as the first European score specifically designed for Western populations and showed good internal performance in its derivation cohort; however, its standalone performance has been only partially explored in independent European cohorts.
Objectives: To evaluate the predictive performance of the Kawanet score for IVIG resistance and coronary outcomes in a European real-world cohort of children with KD.
Methods: We conducted a retrospective monocentric observational study including children diagnosed with KD and followed at Bicêtre Hospital (2015–2025). Clinical, biological, therapeutic and cardiac ultrasound data were collected. The Kawanet score was reconstructed according to original criteria in patients with complete data. IVIG resistance was defined as the need for second-line therapy (second IVIG, anakinra, anti-TNF or cyclophosphamide). Diagnostic performance was assessed using sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), likelihood ratios, Fisher’s exact test, Spearman correlation, and univariate logistic regression.
Results: Fifty-six patients were included; 74% had complete KD. The Kawanet score classified 51.8% as positive (≥2). IVIG resistance occurred in 44.6%. Overall, 35.7% developed aneurysms during follow-up. At the standard cut-off (≥2), Kawanet showed sensitivity 76.9% and specificity 26.9% for IVIG resistance (PPV 34.5%, NPV 70.0%), with AUC 0.61 (95%CI 0.42–0.80). Exploratory ROC analysis suggested an optimal cut-off of 2.5 (sensitivity 38.5%, specificity 88.5%). No significant association was observed between Kawanet positivity and IVIG resistance (Fisher’s OR 1.22, p=1.00; Spearman ρ=0.19, p=0.24) or coronary aneurysms (Fisher p=0.65; Spearman ρ=−0.007, p=0.98; univariate OR 0.94, p=0.92; AUC 0.50).Higher CRP at admission predicted IVIG resistance (OR 1.009, p=0.017; AUC 0.71), with exploratory cut-off 182.5 mg/L (sensitivity 52.4%, specificity 86.2%). Hyponatremia was also associated with IVIG resistance (OR 0.80, p=0.029; AUC 0.71), with cut-off 135.5 mmol/L(sensitivity 78.9%, specificity 51.9%).
Discussion: In our cohort, the Kawanet score showed limited standalone performance for predicting IVIG resistance and coronary outcomes. CRP elevation and hyponatremia emerged as independent predictors, supporting the need for further validation and improved biomarker-based stratification in European KD cohorts.
External validation of the Kawanet score in a French cohort of children with Kawasaki disease: a pilot study
E Barbieri