Distinguishing Kawasaki Disease from Febrile Infectious Disease Using Gene Pair Signatures

Area of Science:

• Pediatric Rheumatology
• Molecular Diagnostics
• Bioinformatics

Background:

• Kawasaki disease (KD) is a childhood vasculitis often misdiagnosed due to reliance on clinical criteria.
• Current diagnostic methods lack specific molecular markers, leading to potential delays and misdiagnosis.
• Distinguishing KD from other febrile infectious (FI) diseases remains a clinical challenge.

Purpose of the Study:

• To identify robust gene pair signatures for differentiating KD from bacterial and viral FI diseases.
• To develop and validate a molecular classifier for improved KD diagnosis.
• To establish objective biomarkers for KD detection.

Main Methods:

• Utilized a relative-expression-based method (k-TSP) and resampling framework on a cohort of 808 childhood patients.
• Data was divided into discovery (n=224), validation-1 (n=197), and validation-2 (n=387) sets.
• Developed a top-ranked gene pair classifier (TRGP) using the top seven gene pair signatures.

Main Results:

• Identified 60 biologically relevant gene pairs distinguishing KD from FI.
• The TRGP classifier demonstrated high performance: Discovery set AUROC 0.947, Validation-1 AUROC 0.955.
• Validation-2 showed promising predictive performance with an AUROC of 0.796, indicating robustness across diverse datasets.

Conclusions:

• Gene pair signatures serve as reliable biomarkers for distinguishing KD from FI.
• The developed TRGP classifier offers a fast, simple, and effective molecular approach to enhance KD diagnosis.
• Objective molecular markers are crucial for improving the accuracy and timeliness of KD diagnosis in children.