Xiaohong Xu, Jun Fan, Li Peng, Zhengkai Xiang, D J Luo, Chenggong Ma, Bo Huang, Xiu Nie, Xiaochuan Dong
Background Neoadjuvant immunochemotherapy (nICT) has emerged as a promising strategy for locally advanced gastric cancer (LAGC), yet clinical responses remain heterogeneous and reliable predictive biomarkers are lacking. A comprehensive dissection of the tumor microenvironment (TME) is essential to uncover determinants of therapeutic efficacy and enable precision immunotherapy. Methods We performed digital spatial profiling (DSP) using the NanoString GeoMx platform on pretreatment endoscopic biopsies from 19 LAGC patients treated with tislelizumab plus SOX chemotherapy. Multiplex fluorescence staining (PanCK, CD45, CD68) enabled compartment-specific transcriptomic analysis of tumor center regions, immune cell infiltration area, and other stromal regions. Findings were integrated with TCGA-STAD data and validated in an independent cohort (n = 20) by immunohistochemistry (IHC) for NOTUM, SERPINA3, CD8, and FOXP3. Results Spatial profiling revealed distinct transcriptional programs across tumor-center regions (TC), immune cell infiltration area (MA), and other stromal regions (OTHER) compartments. High tumor-intrinsic expression of NOTUM, NKD1, and SERPINA3, together with elevated CD8 + T cell infiltration and a reduced Treg/CD3 + ratio within the TME, robustly associated with major pathological response (MPR). These spatial biomarkers were orthogonally validated by IHC in an independent cohort. In TCGA-STAD, a six-gene signature (NOTUM, APOA2, SERPINA3, NKD1, GGH, BPIFB1) correlated with prolonged survival and favorable immune infiltration, with conserved immune-modulatory patterns across multiple cancer types. Conclusions This study identifies NOTUM, SERPINA3, and CD8 + T cell density as spatially resolved, clinically actionable predictors of nICT response in LAGC. Our findings underscore the power of spatial TME interrogation to uncover novel biomarkers and guide personalized immunotherapeutic strategies.