Identification and Validation of Signature Genes and Their Correlations with Infiltrated Immune Cells in Neonatal Sepsis.

Background And Objective: There is increasing demand to identify accurate and reliable molecular biomarkers for early diagnosis of neonatal sepsis. We aimed to identify and verify signature genes in neonatal sepsis through comprehensive bioinformatics analysis.

Methods: A Gene Expression Omnibus data set was used to identify differentially expressed genes (DEGs) in patients with neonatal sepsis and healthy controls by functional and disease enrichment analysis. Gene set enrichment analysis, screening of DEGs using 2 machine algorithms, analysis of receiver operating characteristic curves, and correlation analysis with infiltrating immune cells was performed.

Results: We identified 433 DEGs: 144 downregulated and 289 upregulated. Gene Ontology analysis identified DEGs for T cell activation, positive regulation of cytokine production, secretory granule cavity, cytoplasmic vesicle cavity, immune receptor activity, and antioxidant activity. Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis identified DEGs for hematopoietic cell lineage, cytokine-cytokine receptor interaction, and coronavirus disease (COVID-19). Disease Ontology analysis identified DEGs for hematopoietic system diseases, skin system diseases, and bacterial infectious diseases. We also gained understanding of the enrichment of various functions and pathways by gene set enrichment analysis. In the neonatal sepsis group, Gene Ontology analysis results were significant for coagulation, endocytosis, white cell migration, myeloid leukocyte-mediated immunity, and phagocytosis; KEGG analysis results were significant for chemokine signaling pathway, complement and coagulation cascade, leukocyte migration across endothelium, regulation of actin cytoskeleton, and toll-like receptor signaling pathway. We screened 2 signature DEGs (GSN and SEMA4B) using the least absolute shrinkage and selection operator and support vector machine recursive feature elimination algorithms and verified their diagnostic accuracy by receiver operating characteristic curves. We correlated GSN and SEMA4B expression levels with the infiltration levels of 22 types of immune cell.

Conclusion: GSN and SEMA4B expression accurately predicted early-stage neonatal sepsis, which is beneficial for early clinical diagnosis and treatment.