Lectin-affinity glycosylation pattern analysis of plasma extracellular vesicles: An all-in-one clinical assessment for gastric cancer diagnosis and treatment

Abstract

Extracellular vesicles (EVs) exhibit extensive glycosylation modifications, which are promising biomarkers for gastric cancer (GC). However, EV glycomics and the potential application of EV glycosylation patterns in liquid biopsy remain largely unexplored. This study aims to elucidate the heterogeneity of EV glycosylation in the initiation and progression of GC and to identify specific EV glycosylation markers for clinical assessment. We developed a novel platform for analyzing EV glycosylation patterns via a lectin microarray for EV glycosyl-lectin affinity assessment and antibody-based EV subgroup annotation. The lectin-affinity glycosylation pattern (LAGP) of plasma EVs was profiled across 84 plasma samples, encompassing cases from different stages of GC, benign gastric diseases (BD), and non-disease control (NC). We uncovered heterogeneous LAGPs in different patient groups, identified group-specific LAGPs, and employed them in modeling with the assistance of machine learning algorithms. The linear discriminant analysis (LDA) distinguished advanced GC, early GC, BD, and NC samples with 100 % accuracy. A LAGP-based nomogram was established to predict survival outcomes within 200, 300, and 500 days, achieving area under the ROC curve (AUC) of 0.793, 0.914, and 0.988, respectively. We also tested LAGP for immunotherapeutic response prediction, obtaining AUC values of 0.866–1.000 with various supervised machine-learning algorithms. In conclusion, we represented heterogeneous LAGPs during the occurrence and progression of GC and developed an all-in-one clinical assessment tool for screening cancer patients, monitoring survival outcomes, and predicting immunotherapeutic responses.