Molecular Subtypes of Esophageal Squamous Cell Carcinoma Driven by N- Glycosylation and Construction of a Machine Learning-Based Prognostic Model

Abstract Despite the same origin from squamous epithelial cells, patients with esophageal squamous cell carcinoma (ESCC) exhibit marked heterogeneity, posing significant challenges to clinical treatment. The development of more precise molecular subtyping strategies has become an urgent need in the ESCC field in recent years. Protein post-translational modifications (PTMs) are pivotal regulators of ESCC development and acquired therapeutic resistance. However, the functional significance and clinical implications of N-glycosylation, a major type of PTM, in this context remain poorly understood and warrant comprehensive investigation. This multi-omics study integrated single-cell, bulk RNA sequencing, and somatic mutation data to investigate N-glycosylation in esophageal squamous cell carcinoma (ESCC). scRNA-seq analysis revealed predominant enrichment of N-glycosylation pathway activity within plasma cells of the ESCC microenvironment, while malignant/epithelial cells exhibited N-glycosylation modification of key receptors (ITGA6, ITGB1, SDC4, SDC1, and EGFR) implicated in tumor progression. Bulk RNA-seq enabled stratification of ESCC into two molecular subtypes: N-glycosylation-active (NGAs) and -inactive (NGIs). Comparative analysis demonstrated NGIs harbored significantly higher KMT2D mutation frequency (P = 0.049), whereas NGAs showed increased lymph node metastasis and distinct immune infiltration patterns. Utilizing LASSO regression, we established a 5-gene prognostic signature (BTBD19, TMEM273, ALG14, GGCX, and H2AX) validated by external cohorts. Application of the risk score cutoff (-17.78) successfully stratified patients, with higher risk score groups exhibiting significantly poorer survival rate(TCGA-ESCC: P < 0.0001; GSE53624: P = 0.039). Drug sensitivity prediction based on the risk score identified three therapeutic agents (vinorelbine, thapsigargin, and docetaxel) demonstrating significant efficacy in ESCC patients; however, their sensitivity exhibited an inverse correlation with increasing risk scores. This finding provides insights for guiding precision oncology therapeutics in clinical practice.