Three-gene transcriptomic panel serves as adjunctive diagnostic marker for Kawasaki disease

Original abstract

BACKGROUND: Kawasaki disease (KD) is an acute systemic vasculitis in children, yet its molecular mechanisms remain incompletely understood. This study aimed to identify downregulated candidate diagnostic biomarkers and subsequently investigate the potential functional relevance of a prioritized candidate gene in KD. METHODS: Bulk transcriptomic data from GSE73461 were used for biomarker discovery and model development, while GSE68004 was used for independent external validation. Single-cell RNA sequencing data from GSE168732 were further analyzed to investigate the cellular and regulatory characteristics of the identified genes. Differential expression analysis, WGCNA, machine learning algorithms, CellChat, and scTenifold-based perturbation analysis were performed. Functional investigation of the prioritized candidate ASCC2 was performed in TNF-α-stimulated HUVECs using siRNA-mediated knockdown. RESULTS: Three consistently downregulated genes (EPB42, RUNDC3A, and ASCC2) were identified as candidate diagnostic biomarkers for KD. The three-gene model achieved an AUC of 0.903 in the discovery cohort and 0.948 in the independent GSE68004 cohort. In clinically relevant febrile controls, the model showed moderate discrimination between KD and human adenovirus infection (AUC = 0.830) and between KD and Group A streptococcal infection (AUC = 0.769). Among the three genes, ASCC2 was prioritized for further biological investigation based on its associations with immune-cell populations. Single-cell analyses revealed altered intercellular communication and cell-type-specific ASCC2 expression patterns, while in silico perturbation analysis implicated ASCC2 in NK-cell cytotoxicity and cell-cycle-related regulatory programs. In TNF-α-stimulated HUVECs, ASCC2 knockdown reduced CCK-8-derived metabolic activity and induced G0/G1 cell-cycle arrest. CONCLUSIONS: Our findings identify EPB42, RUNDC3A, and ASCC2 as a reproducible molecular signature associated with KD. The signature robustly discriminated KD from healthy controls and showed moderate discriminatory ability against clinically relevant febrile infections, supporting its potential utility as an adjunctive molecular marker rather than a standalone diagnostic tool. Among these genes, ASCC2 was further investigated as a candidate immune-related regulator through computational and endothelial-cell functional analyses. These findings support the diagnostic potential of the three-gene panel while providing preliminary mechanistic insights into ASCC2 in KD. Key Points • EPB42, RUNDC3A, and ASCC2 were identified as a downregulated molecular signature associated with Kawasaki disease and independently validated in GSE68004. • ASCC2 was prioritized for single-cell and in silico perturbation analyses, which suggested associations with NK-cell cytotoxicity and cell-cycle-related regulatory programs. • ASCC2 depletion in TNF-α-stimulated HUVECs reduced cell viability/metabolic activity and increased G0/G1-phase accumulation under inflammatory conditions.