Construction of a molecular inflammatory predictive model with histone modification-related genes and identification of CAMK2D as a potential response signature to infliximab in ulcerative colitis

Abstract

Background: Ulcerative colitis (UC) is a lifelong inflammatory disease affecting the rectum and colon with numerous treatment options that require an individualized treatment plan. Histone modifications regulate chromosome structure and gene expression, resulting in effects on inflammatory and immune responses. However, the relationship between histone modification-related genes and UC remains unclear. Methods: Transcriptomic data from GSE59071 and GSE66407 were obtained from the Gene Expression Omnibus (GEO), encompassing colonic biopsy expression profiles of UC patients in inflamed and non-inflamed status. Differentially expressed gene (DEG) analyses, functional enrichment analyses, weighted gene co-expression network analysis (WGCNA), and random forest were performed to identify histone modification-related core genes associated with UC inflammation. Features were screened through the least absolute shrinkage and selection operator (LASSO) and support vector machine‐recursive feature elimination (SVM‐RFE), establishing a molecular inflammatory predictive model using logistic regression. The model was validated in the GSE107499 dataset, and the performance of the features was assessed using receiver operating characteristic (ROC) and calibration curves. Immunohistochemistry (IHC) staining of colonic biopsy tissues from UC patients treated with infliximab was used to further confirm the clinical application value. Univariate logistic regression on GSE14580 highlighted features linked to infliximab response. Results: A total of 253 histone modification-related DEGs were identified between inflammatory and non-inflammatory patients with UC. Seven key genes (IL-1β, MSL3, HDAC7, IRF4, CAMK2D, AUTS2, and PADI2) were selected using WGCNA and random forest. Through univariate logistic regression, three core genes (CAMK2D, AUTS2, and IL-1β) were further incorporated to construct the molecular inflammatory predictive model. The area under the curve (AUC) of the model was 0.943 in the independent validation dataset. A significant association between CAMK2D protein expression and infliximab response was observed, which was validated in another independent verification set of GSE14580 from the GEO database. Conclusion: The molecular inflammatory predictive model based on CAMK2D, AUTS2, and IL-1β could reliably distinguish the mucosal inflammatory status of UC patients. We further revealed that CAMK2D was a predictive marker of infliximab response. These findings are expected to provide a new evidence base for personalized treatment and management strategies for UC patients.