Abstract 6470: Spatial analysis of the tumor microenvironment in diffuse large B-cell lymphoma reveals clinically relevant recurrent cellular neighborhoods

Abstract Introduction: The tumor microenvironment (TME) of diffuse large B-cell lymphoma (DLBCL) has been recently dissected and different TME-based subtypes of DLBCL proposed (Kotlov et al. Cancer Discov. 2021, Steen et al. Cancer Cell. 2021). Previous studies have, however, not systematically explored the spatial organization of the DLBCL TME and its association with biological and clinical factors. Materials and Methods: We performed multiplex immunofluorescence using a 12-plex antibody panel on diagnostic tumor samples from 99 patients with DLBCL. We characterized B cells, T cells, tumor-associated macrophages (TAMs), and immune checkpoint molecules. We processed images using Ilastik and CellProfiler softwares, and segmented nuclei with a pretrained deep-learning model. We phenotyped the cells using the Phenograph algorithm combined with manual thresholding. We identified recurrent cellular neighborhoods (RCNs) and performed cell-cell interaction analyses using the scimap package (Python 3.9). We correlated the findings with biological and clinical characteristics and survival. Results: We analyzed over 1.2 x 106 single cells with a median of 12 354 cells per patient. We identified 17 cell phenotypes, including various subgroups of B cells, T cells, and TAMs. Of the cell types, B cells were the most abundant (mean 55%), followed by CD4+ helper and CD8+ cytotoxic T cells (means 14% and 10%), and M2-like TAMs (mean 8%) with great heterogeneity between the samples. We identified 10 different RCNs with distinct cellular compositions across DLBCLs. The most common RCN was an immune-poor RCN (30% of the cells), followed by an RCN with high B cell content and a moderate immune cell infiltrate (22% of the cells). We also identified different T cell and TAM-rich RCNs. The RCNs subdivided DLBCLs into immune-poor areas and areas with diverse immune cell infiltrates. Notably, HLA-ABC positive lymphomas consisted of larger proportions of CD4+ T cell-rich RCNs (p < 0.05). Defined by the shortest average distance, immune-poor RCNs were located close to the RCNs that had high B cell content combined with moderate immune cell infiltrates but far from the T cell-rich RCNs. However, when CD8+ T cell-rich and immune-poor RCNs were close to each other, the outcome was favorable (progression-free survival (PFS); p < 0.05), whereas the proximity of PD-L1+ B cell-rich and CD8+ T cell-rich RCNs was associated with poor survival (PFS; p < 0.05). Finally, we correlated RCN proportions with lymphoma microenvironment (LME) and genetic subtypes. Immune-poor RCNs were common in the depleted LME subtype, whereas PD-L1+ M2-like TAM-rich RCNs were more common in the inflamed LME. In contrast, PD-L1+ M2-like TAM-rich RCNs were rare in EZB-type DLBCLs. Conclusions: Our data show that DLBCL TME is organized into RCNs that are associated with distinct molecular and subtypic features and patient outcomes. Citation Format: Matias Autio, Suvi-Katri Leivonen, Leo Meriranta, Marja-Liisa Karjalainen-Lindsberg, Teijo Pellinen, Sirpa Leppä. Spatial analysis of the tumor microenvironment in diffuse large B-cell lymphoma reveals clinically relevant recurrent cellular neighborhoods [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2025; Part 1 (Regular Abstracts); 2025 Apr 25-30; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2025;85(8_Suppl_1):Abstract nr 6470.