Diagnostic potential of neutrophil extracellular traps in ulcerative colitis: a gene-based predictive model

Neutrophil extracellular traps (NETs), web-like structures released by neutrophils during the process NETosis, aiding in gut mucosal defense against microbial infections but potentially exacerbating inflammation and tissue damage. In Ulcerative Colitis (UC), NET formation is increased and may contribute to gut inflammation. However, the role of NETs in diagnosis of UC remains unclear. This study aims to identify NETs-related genes with diagnostic potential in UC and to develop a diagnostic predictive model based on these genes.

The transcriptome dataset of UC retrieved from the GEO database. Differential expression analysis and Gene Set Enrichment Analysis (GSEA) were performed on the training set using R software. The "CIBERSORT" algorithm was utilized to evaluate the immune cell infiltration in UC. Subsequently, differentially expressed NETs (DE-NETs) were identified by intersecting key module genes from Weighted Gene Co-Expression Network Analysis (WGCNA), NETs-related genes, and differentially expressed genes (DEGs). The diagnostic genes were identified through three machine learning algorithms (Least Absolute Shrink-age And Selection Operator (LASSO) algorithm, Random Forests (RF) and Support Vector Machine-Recursive Feature Elimination (SVM-RFE)). The results of the three algorithms were integrated to identify NETs-related diagnostic genes (NDGs). A clinical diagnostic predictive model was constructed based on NDGs, and its performance was assessed using calibration curves, Clinical Impact Curve (CIC), Decision Curve Analysis (DCA), and Receiver Operating Characteristic Curve (ROC) to evaluate its discrimination ability and clinical utility. Immunohistochemical (IHC) examination was performed on colon tissue from UC patients. Additionally, a DSS-induced UC mouse model was constructed, and NDG expression in mouse colon tissues was analyzed by qRT-PCR, Western-blot, and IHC.

The UC group exhibits significant enrichment of immune-related pathways, including the IL-17 signaling pathway, cytokine-cytokine receptor interactions, and TNF signaling pathway, as well as enhanced infiltration of immune cells, as demonstrated by GSEA and immune infiltration analysis. Fourteen genes were identified through intersecting DEGs, key module genes in WGCNA, and NETs-related genes. The identification of IL-1β, MMP-9 and CXCR2 as NDGs was using three machine learning methods. The clinical diagnostic predictive model of UC based on NDGs was constructed, and the AUC value of the model was 0.9715 in the training set, and 0.9595 and 0.9597 in external validation sets, respectively. Significant positive correlations were found between NDGs and Mast cells, Neutrophils, resting NK cells, M1 Macrophages, Tregs and Eosinophils. IHC analysis of clinical samples shows high expression of NDGs in the colon of UC patients. A clinical-associated nomogram was constructed from the IHC scores of the clinical samples, exhibiting an AUC of 100%. In DSS-induced UC mice model, NDG expression was up-regulated compared to normal control mice. A mouse-associated nomogram was constructed based on IHC scores of mouse tissues, and the AUC was also 100%.

We constructed a clinical diagnostic predictive model for UC based on IL-1β, MMP-9 and CXCR2 and validated the ability of NDGs to predict UC in the clinical patient and the animal model. The model has excellent diagnostic efficiency and can provide a new idea for clinical diagnosis of UC.