A Series of Genes for Predicting Responses to Anti-Tumor Necrosis Factor α Therapy in Crohn's Disease

Background: Patients with Crohn's disease (CD) experience severely reduced quality of life, particularly those who do not respond to conventional therapies. Antitumor necrosis factor (TNF)α is commonly used as first-line therapy; however, many patients remain unresponsive to this treatment, and the identification of response predictors could facilitate the improvement of therapeutic strategies. Methods: We screened Gene Expression Omnibus (GEO) microarray cohorts with different anti-TNFα responses in patients with CD (discovery cohort) and explored the hub genes. The finding was confirmed in independent validation cohorts, and multiple algorithms and in vitro cellular models were performed to further validate the core predictor. Results: We screened four discovery datasets. Differentially expressed genes between anti-TNFα responders and nonresponders were confirmed in each cohort. Gene ontology enrichment revealed that innate immunity was involved in the anti-TNFα response in patients with CD. Prediction analysis of microarrays provided the minimum misclassification of genes, and the constructed network containing the hub genes supported the core status of TLR2. Furthermore, GSEA also supports TLR2 as the core predictor. The top hub genes were then validated in the validation cohort (GSE159034; p < 0.05). Furthermore, ROC analyses demonstrated the significant predictive value of TLR2 (AUC: 0.829), TREM1 (AUC: 0.844), and CXCR1 (AUC: 0.841). Moreover, TLR2 expression in monocytes affected the immune-epithelial inflammatory response and epithelial barrier during lipopolysaccharide-induced inflammation (p < 0.05). Conclusion: Bioinformatics and experimental research identified TLR2, TREM1, CXCR1, FPR1, and FPR2 as promising candidates for predicting the anti-TNFα response in patients with Crohn's disease and especially TLR2 as a core predictor.