Multiomic dissection of immune-privileged DLBCL reveals divergent immune subtypes in testicular and primary central nervous system lymphoma

Abstract Introduction Diffuse large B-cell lymphomas (DLBCL) arising in immune-privileged sites such as the testis and central nervous system (CNS) are rare, aggressive subtypes with high relapse risk. Primary testicular lymphoma (PTL) and primary CNS lymphoma (PCNSL) often harbor BCR related mutations (e.g., MYD88, CD79B) and are typically of activated B-cell phenotype. Their microenvironments promote immune escape, with PD-L1 overexpression and impaired antigen presentation implicated. PTL's frequent CNS dissemination highlights the need for shared therapeutic strategies and immune profiling. Immune-based therapies may offer new treatment avenues, but their success likely depends on pre-existing immune states. We hypothesized that PTL and PCNSL differ immunologically despite shared origins, with immune-directed therapy implications. Methods We performed a multiomic analysis of PTL and PCNSL from four Australian centers (n=56; 40 PCNSL, 16 PTL). Immune gene expression was profiled using the NanoString PanCancer Immune panel. Spatial transcriptomic (GeoMx Whole Transcriptome Atlas) segmented regions using CD20, CD8, and CD68 antibody staining to resolve tumour and immune compartment-specific transcriptomics. Single-nuclei RNA sequencing (snRNA-seq) characterized cellular composition and immune-tumor interactions across representative cases. Whole-exome sequencing (WES) defined mutational landscapes. Results PTL exhibited a T cell enriched immune profile, with upregulation of CD8A, GZMK, and TCF7, and a significantly higher CD8:CD163 ratio than PCNSL (p<0.0001) indicating a more immunologically active state. PCNSL samples were macrophage enriched with upregulation of CD163, SPP1, and MSR1. Unsupervised clustering identified three groups: immune-hot PCNSL, immune-cold PCNSL, and PTL, which was intermediate in inflammatory status. Immune-cold PCNSL showed low CD8⁺ and CD68⁺ infiltration, reduced PD-L1 and CXCL9, and diminished antigen presentation, though SPP1 remained elevated. PTL and immune-hot PCNSL displayed upregulation of CD8A, CD4, PD-1, LAG3 and TIGIT, indicating greater immune engagement. Genomic profiling revealed shared driver mutations (MYD88L265P, CD79B), but immune-cold PCNSL had fewer MYD88L265P mutations (12.5% vs. 69.0% and 50.0% in immune-hot PCNSL and PTL respectively; p=0.04) and fewer MCD subtype cases. PTL had a significantly higher mutational burden than PCNSL (mean 34.18 vs 17.04 mutations/Mb; p=0.0007). Spatial transcriptomics revealed functional compartmentalization: CD8⁺ regions in PCNSL showed reduced MHC-II and antigen presentation pathways, while CD68⁺ regions upregulated Fcγ receptor and toll-like receptor signaling. A high SPP1:CXCL9 ratio in PCNSL suggested an immunosuppressive myeloid phenotype that may impair T cell-based immunotherapy responses and support macrophage-targeted interventions. CD20⁺ tumor zones in PCNSL showed reduced NF-κB and JAK-STAT signaling compared to PTL. snRNAseq confirmed distinct immune landscapes in PTL (n=11) and PCNSL (n=13), with subtype-specific macrophage profiles. PCNSL was enriched for SPP1+ macrophages, while PTL macrophages showed elevated PD-L1. Microglia removal enabled macrophage-specific analysis, confirming sustained upregulation of SPP1 independent of microglia contribution, along with increased CXCL13, and MERTK in PCNSL vs increased immunoregulatory gene expression in PTL. Gene set enrichment analysis highlighted strong activation of T cell-mediated immunity in PTL. Ongoing analyses aim to resolve T cell states and cellular interactions. Conclusion Despite shared genomic similarity and B-cell origin, PTL and PCNSL differ markedly in immune microenvironment and mutational burden. PTL features a cytotoxic, PD-1-engaged microenvironment, while immune-cold PCNSL shows immune exclusion and macrophage-driven suppression. These findings suggest that anti-PD-1 therapy may be more effective in PTL and immune-hot PCNSL, while immune-cold PCNSL may require combination approaches targeting myeloid suppression or antigen-presentation deficits. The immune contexture also informs emerging T cell directed therapies, which may be more effective in inflamed immune states. These insights reinforce immune subtyping as a basis for adapting immunotherapeutic strategies, including PD-1 inhibitors, bispecific antibodies, CAR-T cells, and macrophage-directed agents such as anti-CD47, to the distinct immune landscapes of aggressive B-cell lymphomas in immune-privileged sites.