Identification of Crohn's disease-associated molecular signatures by integrative bioinformatics analysis and experimental validation.

Background: Crohn's disease (CD) is a chronic inflammatory bowel disease with rising global incidence. Its multifactorial etiology involves genetic, environmental, microbial, and immunological factors. Current diagnostic methods are invasive and costly, underscoring the need for improved biomarkers and non-invasive tools for early detection and monitoring. Methods: Using bioinformatics and machine learning, we analyzed datasets from the Gene Expression Omnibus (GEO) to identify diagnostic biomarkers and immune infiltration patterns. Multi-omics integration and immune profiling were employed. Differential gene expression, Weighted Gene Co-expression Network Analysis (WGCNA), and Mendelian randomization (MR) identified causal genes. Machine learning (12 algorithms, 113 combinations) developed diagnostic models, validated via Receiver Operating Characteristic (ROC) analysis. Protein-protein interaction network, immune infiltration, and gene enrichment analysis identified biological pathways or functions of key genes. Single-cell sequencing validated key findings. Key genes were experimentally validated through immunohistochemistry (IHC), quantitative polymerase chain reaction (qPCR) and Western blot (WB). Results: A total of 502 differentially expressed genes (DEGs) were identified, with WGCNA highlighting a key module enriched in immune pathways (IL-17, NF-κB). Machine learning selected 96 genes, and MR identified Fc gamma receptor IIa (FCGR2A) (OR = 1.38, p=0.043) and Serpin family B member 9 (SERPINB9) (OR = 1.27, p=0.001) as candidate biomarkers with evidence supporting potential causal associations with CD. FCGR2A and SERPINB9 were identified as candidate diagnostic biomarkers with Area Under the Curves (AUCs) of 0.788(95% CI, 0.722-0.853) and 0.789(95% CI, 0.707-0.871) and correlated with neutrophil and M1 macrophage infiltration. Single-cell analysis revealed cell-specific expression: FCGR2A in neutrophils/monocytes and SERPINB9 in B cells. IHC, qPCR and WB confirmed their upregulation in CD tissues. A nomogram integrating both genes showed promising predictive performance (C-index=0.819) and net benefit in decision curve analysis. Conclusion: FCGR2A and SERPINB9 represent candidate molecular biomarkers associated with CD, although FCGR2A largely reflects neutrophil-associated inflammation. The two-gene model showed preliminary discriminatory performance, but independent prospective validation and comparison with established clinical biomarkers are required before clinical application.