Identification of programmed cell death and mitochondria correlated biomarkers for Kawasaki disease by integrated bioinformatics and machine-learning algorithms with RT-qPCR verification.

Background: The interplay between programmed cell death and mitochondrial dysfunction is a central mechanism in Kawasaki disease pathogenesis, yet the specific molecular connectors within this network are poorly defined. This study aimed to identify and validate integrated biomarkers that bridge these processes, offering new diagnostic and mechanistic insights. Methods: Using the GSE73461 dataset as a training set, we identified Kawasaki disease-related differentially expressed genes. These were intersected with curated programmed cell death-related and mitochondrial-related gene sets to obtain candidates. Subtype-specific analysis was performed to clarify which programmed cell death processes were most closely linked to mitochondrial dysfunction. A protein-protein interaction network was constructed, and hub genes were screened using multiple algorithms. Biomarker selection was refined via two machine learning models. Diagnostic performance was evaluated using the independent GSE68004 validation set, receiver operating characteristic analysis, and a composite nomogram. Immune infiltration, functional enrichment, regulatory network construction, drug prediction, and molecular docking were also conducted. Finally, biomarker expression was validated in clinical blood samples using reverse transcription quantitative polymerase chain reaction. Results: We identified 63 candidate genes. Five biomarkers-CD177, Matrix Metallopeptidase 9, Nuclear Factor Erythroid 2, Colony Stimulating Factor 3 Receptor, and Suppressor Of Cytokine Signaling 3-were consistently upregulated in Kawasaki disease and showed high diagnostic accuracy (area under the curve >0.94 in both datasets). These biomarkers are functionally anchored in a network where specific programmed cell death subtypes, strongly correlated with mitochondrial genes, converge on inflammatory pathways. A diagnostic nomogram integrating all five biomarkers achieved an area under the curve of 0.986. Clinical validation confirmed significant upregulation of CD177, Matrix Metallopeptidase 9, and Suppressor Of Cytokine Signaling 3, while Nuclear Factor Erythroid 2 and Colony Stimulating Factor 3 Receptor exhibited no significant differential expression. All five biomarkers were enriched in the Fc gamma receptor-mediated phagocytosis pathway and correlated cohesively with key immune cells. Decitabine and Ibuprofen were highlighted as potential therapeutic candidates. Conclusion: This study defines a novel five-biomarker panel that functionally links mitochondrial dysfunction, dysregulated programmed cell death, and immune activation in Kawasaki disease, providing an integrated, mechanism-based framework for improving diagnosis and guiding future targeted therapies.