Development of a prediction model for progression of coronary artery lesions in Kawasaki disease

• Published in: Pediatric research Vol. 95; no. 4; pp. 1041 - 1050
• Main Authors: Xu, Dan, Chen, Ye-Shi, Feng, Chen-Hui, Cao, Ai-Mei, Li, Xiao-Hui
• Format: Journal Article
• Language: English
• Published: United States Nature Publishing Group 01.03.2024
• Subjects: Case-Control Studies; Child; Childrens health; Clinical outcomes; Coronary Artery Disease - complications; Coronary Artery Disease - diagnostic imaging; Coronary vessels; Coronary Vessels - diagnostic imaging; Health risks; Humans; Immunoglobulins, Intravenous; Infant; Kawasaki disease; Mucocutaneous Lymph Node Syndrome - complications; Mucocutaneous Lymph Node Syndrome - diagnosis; Nomograms; Precision medicine; Prospective Studies; Retrospective Studies
• ISSN: 0031-3998; 1530-0447
• DOI: 10.1038/s41390-023-02931-5

This study aimed to identify risk factors for the progression of coronary artery lesions (CALs) in children with Kawasaki disease (KD) and to develop a nomogram prediction model. This is a retrospective case-control study in which the participants were categorized into three groups based on the changes of the maximum Z score (Zmax) of coronary arteries at the 1-month follow-up compared with the baseline Zmax: CALs-progressed, CALs-improved, and CALs-unchanged. Of total 387 patients, 65 (27%), 319 (73%), and 3 (0.7%) patients were categorized into CALs-progressed group, CALs-improved group, and CALs-unchanged group, respectively. Six independent factors associated with CALs progression were identified, including initial IVIG resistance, baseline Zmax, the number of coronary arteries involved, C-reactive protein, albumin, and soluble interleukin-2 receptor (odds ratio: 7.19, 1.51, 2.32, 1.52, 0.86, and 1.46, respectively; all P-values < 0.01). The nomogram prediction model including these six independent risk factors yielded an area under the curve (AUC) of 0.80 (95% confidence interval, 0.74 to 0.86). The accuracy of this model reached 81.7% after the Monte-Carlo Bootstrapping 1000 repetitions. The nomogram prediction model can identify children at high risk for the progression of CALs at early stages. Six independent factors associated with CALs progression were identified, including initial IVIG resistance, baseline Zmax, the number of coronary arteries involved, CRP, ALB, and sIL-2R. The prediction model we constructed can identify children at high risk for the progression of CALs at early stages and help clinicians make individualized treatment plans. Prospective, multi-centered studies with larger sample sizes are warranted to validate the power of this prediction model in children with KD.