Spatial transcriptomics unveils immune cellular ecosystems associated with patient survival in diffuse large B-cell lymphoma

Diffuse large B-cell Lymphoma (DLBCL) is the most prevalent subtype of non-Hodgkin's lymphoma for which current therapeutic strategies remain insufficient, in part owing to heterogeneity in tumor biology and the immune microenvironment. The diffuse nature of DLBCL represents a challenge to elucidate how malignant and immune cells are spatially organized within the tumor microenvironment (TME), and how this organization impacts immune function and clinical outcome. Here, we performed a pilot spatial transcriptomics analysis of primary DLBCL tissue sections to resolve spatial gene expression patterns and cell-cell interactions, with key cellular features validated at the protein level. We identified six recurrent, spatially organized cellular ecosystems (Cell-Eco) defined by distinct immune compositions, transcriptional programs, and neighborhood architectures. Notably, ecosystems with similar immune cell abundance exhibited divergent functional states and opposite clinical associations, demonstrating that spatial context and local interactions, rather than cell-type frequency alone, shape immune function within the DLBCL TME. Building on these spatial ecosystems, we derived Cell-Eco gene signatures that generated prognostic scores and robustly stratified patient survival in large, independent DLBCL cohorts comprising more than 1000 cases. Across ecosystems, tumor-associated macrophages have emerged as the dominant spatial partners of malignant B cells, highlighting their central role in structuring the immune microenvironment. Together, these findings establish a spatially informed framework for DLBCL immune organization and demonstrate the prognostic relevance of tumor cellular architecture.