Spatially integrated transcriptomics maps the metastatic niche in lymph nodes of gastric cancer

Lymph node (LN) metastasis is a major determinant of clinical outcome in gastric cancer (GC), yet the molecular and spatial features of metastatic LN niches remain incompletely defined. Here, we integrated digital spatial profiling (DSP) of primary GC and matched LNs, single-cell RNA sequencing (scRNA-seq), and 10x Genomics Visium spatial transcriptomics. Cross-sectional analysis of cytokeratin-positive regions across primary GC and N1-N3 metastatic lymph nodes (mLNs) identified a hybrid epithelial-mesenchymal expression pattern in N1 samples and prominent immune checkpoint, angiogenic, and receptor tyrosine kinase programs in the N3 group. Deconvolution showed enrichment of exhausted T cells, regulatory T cells, and fibroblasts in mLNs, consistent with immunosuppressive remodeling of the local microenvironment. Cell-cell communication analysis implicated the galectin-9 (LGALS9)-CD45/CD44, CLEC2D-KLRB1, and CCL19/CCL21-CCR7 axes in mLNs. Spatial analysis further revealed compartment-specific heterogeneity, including distinct proliferative, immune, and immunosuppressive programs. Integrating multiple transcriptomic datasets, we derived a 12-gene lymph node metastasis gene signature (LNMGS); high LNMGS scores were associated with poorer overall survival in the TCGA-STAD cohort. Spatial mapping of the LNMGS in primary GC identified an LNMGS-high region displaying a center-to-periphery expression gradient and tending to have a higher inferred copy-number burden. A distance-associated gene signature derived from this gradient was enriched for proliferation, invasion, angiogenesis, immune evasion, and stemness. Together, these findings provide a spatial and molecular framework for lymph node metastasis in GC and highlight potential biomarkers and therapeutic targets for further investigation.