Spatial immune ecosystems and therapy resistance in large B-cell lymphoma: from single-cell multi-omics to immunotherapy-matched validation

Large B-cell lymphomas, particularly diffuse large B-cell lymphoma (DLBCL), provide a clinically important model in which tumor-intrinsic programs, immune recognition, effector-cell fitness, and tissue architecture jointly shape response and resistance to immunotherapy. Although molecular classification has refined the understanding of lymphoma biology, relapse after chemoimmunotherapy, chimeric antigen receptor (CAR) T-cell therapy, CD20×CD3 bispecific antibodies, antibody-drug conjugates (ADCs), and immune-combination strategies cannot be explained by genetics or target-antigen expression alone. Single-cell and spatial multi-omics now allow lymphoma to be interpreted as a spatial immune ecosystem composed of malignant B-cell states, exhausted or regulatory T-cell programs, myeloid suppressive hubs, stromal and vascular compartmentalization, metabolic gradients, and local communication networks. This Review synthesizes recent DLBCL and large B-cell lymphoma ecosystem studies from a cancer-immunology perspective, emphasizing how spatial immune organization may influence effector access, antigen engagement, macrophage-mediated suppression, antibody exposure, and treatment-associated spatial change. We propose a treatment-matched validation framework that connects high-dimensional discovery to formalin-fixed paraffin-embedded (FFPE)-compatible pathology assays, including reduced marker panels for CAR T-cell therapy, bispecific antibodies, ADC-containing therapy, and immune-combination trials. Throughout, we distinguish observed tissue organization from computationally inferred communication and experimentally validated mechanisms. The resulting framework is intended as a trial-design and biomarker-development roadmap, not as a current clinical allocation algorithm.