Endometrial cancer (EC) is a prevalent gynecologic cancer, with worldwide increasing incidence and disease-associated mortality. N-glycosylation, a critical post-translational modification, has been implicated in cancer progression and immune response modulation. We aimed to elucidate the role of N-glycosylation-related genes on EC cell heterogeneity, prognosis, and immunotherapy response. Data from single-cell RNA sequencing (scRNA) of five patients with EC were acquired from the Gene Expression Omnibus (GEO) database. Nonnegative matrix factorization (NMF) was used to identify cell subtypes related to N-glycosylation from a scRNA matrix. Subsequently, a consensus prognostic signature by integrating 101 combinations of 10 machine learning algorithms. The response to immunotherapy in EC was further examined by multiple algorithms. Our findings identified 11,020 differentially expressed genes (DEGs), of which 34 N-glycosylation-related DEGs were remarkably associated with overall survival (OS) in EC. Single-cell RNA sequencing analysis revealed 30,233 cells divided into eight clusters, with T cells and epithelial cells showing distinct functional characteristics. NMF clustering further classified malignant cells into four subtypes: N-glycosylation-C0, Glycosphingolipid-C1, O-GalNAc-C2, and Elongation-C3. The O-GalNAc-C2 subtype exhibited the highest metabolic pathway activity and activation of transcription factors SOX4, JUND, and FOS. Additionally, cell-cell interaction networks highlighted the MK signaling pathway as a critical mediator of intercellular communication. An integrated machine learning framework generated a prognostic model comprising eight DEGs (LAMC2, KRT7, IL32, KRT18, SERPINA1, PGR, AKAP12, EDN2), achieving an average C-index of 0.712 in training and validation cohorts. A low-risk score implies more significant immune cell infiltration and better response to immunotherapy. Our study underscores the role of N-glycosylation-related genes in EC prognosis and immunotherapy response prediction, and may provide a basis for the development of targeted therapies and personalized treatment strategies.
Read full abstract