Full text 2026

Architectural Refuges: Mapping Spatial Heterogeneity and Niche-Mediated Drug Resistance in Gastric and Esophageal Adenocarcinomas

Thangaretnam K, Islam MO, Lv J, et al.

Full text

Loading PDF… Expand reader Download

Abstract

Resistance to systemic therapy remains the defining challenge in the management of gastric cancer (GC) and esophageal adenocarcinoma (EAC). While genomic drivers of resistance are well characterized, traditional bulk profiling has failed to capture the physical rules governing tumor survival within the complex tissue ecosystem. Emerging data from 2024-2025, leveraging high-resolution spatial transcriptomics and multi-omics, have recontextualized resistance as a phenomenon of "spatial privilege" rather than solely an intrinsic cellular fate. This review summarizes recent evidence to define "architectural refuges": distinct spatial niches that physically shield malignant clones from cytotoxic and targeted agents. We delineate three critical resistance domains common to upper gastrointestinal adenocarcinomas: (1) The "Excluded" Niche, where specific cancer-associated fibroblast (CAF) subpopulations (iCAFs vs. myCAFs) and stiffened extracellular matrix create hypovascular zones that limit drug delivery; (2) the "Immune-Tolerant" Niche, characterized by the spatial exclusion of CD8+ T cells and the recruitment of suppressive myeloid populations via the MIF/CD74 and USP14 axes; and (3) the "Metabolic" Niche, where mitochondrial heterogeneity and lipid metabolic symbiosis establish nutrient-deprived niches that select for stem-like, dormant states. By mapping these conserved spatial determinants from primary GEJ tumors to peritoneal and distant metastases, we argue that overcoming resistance requires an advancement: moving beyond targeting individual mutations to dismantling the multicellular architecture that sustains malignancy.

Keywords

Drug resistance spatial niche Tumor Microenvironment Cancer-associated Fibroblasts Metabolic Reprogramming Gastroesophageal Adenocarcinoma Spatial Transcriptomics