Machine learning–based multi-omics analysis to identify the role of MAN1C1 in pancreatic cancer and its clinical significance as a drug target.

Abstract

e16475 Background: Exploring genes related to glycosylation modification, particularly the role of MAN1C1 in PDAC, contributes to the development of new therapeutic strategies. Methods: Single-Cell Analysis: Analyzed cell types and their glycosylation levels in the PDAC TME using scRNA-seq data (GSE212966). Dimensionality reduction and clustering were performed, combined with cell type annotation to identify tumor cell populations. GO, KEGG, and ssGSEA enrichment analyses were conducted to elucidate cellular functional states. Scoring & Interaction: Evaluated scores of glycosylation modification-related genes, with a special focus on fibroblasts. Established correlations between Cancer-Associated Fibroblasts (CAFs), CD4+ T cells, and tumor-associated macrophages to interpret potential intercellular communication patterns. Trajectory Analysis: Pseudotime analysis was used to explore tumor progression trajectories and describe potential malignant evolutionary laws of cells. Model Construction: Based on TCGA-PAAD transcriptomics (training set) and GSE28735 (validation set), 9 machine learning algorithms were utilized to screen PDAC-characteristic glycosylation genes. An optimal model was constructed using 5 genes. Experimental Validation: In vitro experiments (CCK-8, Transwell migration/invasion, scratch assay) were conducted. Results: Glycosylation & Cell Types: High-glycosylation fibroblasts exhibited stronger activity in intercellular communication and signal pathway enrichment. MAN1C1 was highly expressed in active cancer-associated fibroblasts (aCAFs), and low expression was associated with a good prognosis for patients. Clinical Prognosis: MAN1C1 expression was significantly correlated with immune infiltration levels and immune-related markers. High MAN1C1 expression was associated with poorer overall survival, demonstrating its value as a potential prognostic marker. Logistic regression, COX regression, and LASSO analysis based on TCGA-PAAD clinical subgroups confirmed that patients with high MAN1C1 expression had a worse prognosis. ROC curve analysis indicated that MAN1C1 possesses good diagnostic capability. Conclusions: The glycosylation level of fibroblasts in the PDAC tumor microenvironment is significantly elevated. As a key glycosylation-related gene, MAN1C1 expression is closely related to patient prognosis. High expression of MAN1C1 inhibits the proliferation and migration of PDAC cells and affects intercellular interactions within the tumor microenvironment. MAN1C1 may serve as a potential therapeutic target for PDAC, providing a basis for developing therapeutic strategies targeting glycosylation pathways.