The spatial architecture of tumor immunity: Emerging predictive biomarkers across cancer

Tumor immunity is shaped not only by the cellular composition of the tumor microenvironment (TME), but also by its spatial architecture. Increasing evidence shows that immune function cannot be inferred from cell abundance alone, because the positioning and coordination of immune and stromal cells determine whether local immunity is activated, excluded, or suppressed. Advances in multiplex imaging, spatial proteomics, spatial transcriptomics, and AI-assisted tissue analysis now enable high-resolution mapping of these spatial states across clinically relevant scales. These approaches reveal that tumors are organized as recurrent microarchitectures that can function either as immune engines or as suppressive brakes. Immune-supportive structures, including mature tertiary lymphoid structures, dendritic cell-centered immune triads, high endothelial venule-associated recruitment corridors, and effector-tumor interfaces, promote antigen presentation, lymphocyte recruitment, and cytotoxic engagement. In contrast, exclusionary and suppressive architectures, such as CAF-mediated stromal barriers, SPP1 + macrophage-CAF niches, stalled T-cell zones, Treg-enriched suppressive niches, and NLRP3 + macrophage-rich regions, restrict infiltration and sustain immune dysfunction. Across cancer types, such spatial readouts are emerging as robust biomarkers. In this Review, we summarize recent progress in spatial tumor immunology and propose a conceptual framework for interpreting tumor immunity from composition to cellular neighborhoods. We further discuss how these spatial architectures shape tumor heterogeneity, therapeutic response, and resistance, and how they may be translated into actionable biomarkers for patient stratification, treatment monitoring, and next-generation precision immunotherapy.