Immune Cell Infiltration Impacts Chemotherapy Efficacy in B-NHL

Introduction

Microenvironmental profiling in lymphoma has identified clinically relevant subgroups through transcriptional and genomic profiling. We further enhance the characterization of DLBCL by examining the variations in cellular composition and spatial architecture and correlating these findings with clinical outcomes following chemoimmunotherapy.

Method

Using CO-Detection by indEXing (CODEX) we characterized the cellular composition of aggressive B-cell non-Hodgkin lymphoma samples from 193 patients at initial diagnosis which were consecutively treated with chemoimmunotherapy in the RICOVER-60 trial (NCT00052936), one of the landmark studies establishing R-CHOP as the standard treatment for DLBCL. Samples were assembled in tissue microarrays and stained with 54-plex antibody panel targeting microenvironmental cells along with important functional markers of malignant B cells. Genetic and clinical annotations were integrated to represent known tumor cell intrinsic features.

Results

Our analysis revealed that the cellular composition of the microenvironment in diffuse large B-cell lymphoma (DLBCL) varied significantly among samples but remained consistent between technical replicates. This heterogeneity was observed across all major non-lymphoma cell types: T cells ranged from 0.7% to 85% of all cells, with a mean frequency of 29%, followed by tumor-associated macrophages (5%), stromal cells (4%), dendritic cells (2%), and other myeloid cells (4%). We further focused on lymphoma-infiltrating T-cells, classifying them based on our recently published large T-cell Cite-seq dataset (Roider, Nat. Cell Biol. 2024). Our thorough examination of T-cell fingerprints in relation to clinical endpoints revealed among other findings that higher infiltration of cytotoxic T-cells is associated with favorable outcomes, whereas the frequency of exhausted cytotoxic T-cells correlated with poor outcomes. Additionally, we will report on the associations between DLBCL genotypes and specific T-cell phenotypes.

Beyond the cell type composition, we investigated the spatial interactions between distinct cell types and found that seven distinct cellular neighborhoods could robustly be identified across patient samples based on the 30 nearest neighboring cells. The composition and organization of cellular neighborhoods was distinct between patients, likely representing different patterns of interaction between the lymphoma and its microenvironment. Associating the observed spatial organization in cellular neighborhoods, we observed significant differences with regard to the known genetic subgroups of DLBCL.

Conclusion

Our results highlight the importance of microenvironmental factors for clinical outcome after chemoimmunotherapy in DLBCL. T cell phenotypes represent biological factors which impact clinical outcomes but are currently undervalued in clinical diagnostics since they are not represented in typical genetic tests. Additionally, our data suggest that the microenvironmental differences which have previously been described based on transcriptional profiling might also have an impact on the spatial architecture of the lymphoma. This would enable the integration into clinical diagnostic tests and might unravel the mechanisms by which the lymph node organization is disturbed in DLBCL.