Abstract Primary central nervous system (CNS) lymphoma (PCNSL) is a highly aggressive, brain–restricted B-cell malignancy with limited therapeutic options and poor survival. Unlike glioblastoma (GBM), PCNSL shows sensitivity to immune checkpoint inhibitors in ∼30% of patients; however, the biological basis for this difference remains unclear. Moreover, its relative genetic homogeneity suggests tumor–immune interactions as key drivers of disease behavior. We hypothesized that dissecting immune regulation in PCNSL could inform immune manipulation within the CNS and guide potential therapeutic strategies. To test this, we compared PCNSL and GBM to define shared and disease-specific immunosuppressive mechanisms within the brain tumor microenvironment (TME). Using an in-house single-cell RNA sequencing (scRNA-seq) dataset of 25 PCNSL tumors, representing the largest scRNA-seq dataset of ∼90, 000 cells in PCNSL to date, alongside 110 publicly available GBM tumors, we analyzed cell-type composition, immune checkpoint expression, and ligand–receptor interactions to define immune regulatory networks relevant to therapeutic response. Both tumors exhibit profoundly immunosuppressive TMEs, with compositional analyses uncovering distinct underlying pathways. Although ∼90% of tumor-associated macrophage (TAM) subtypes are shared, GBM shows significantly higher infiltration of blood–derived TAMs, whereas PCNSL is enriched for border–associated TAMs. PCNSL also uniquely harbors a microglial subset enriched for extracellular matrix remodeling and invasion-associated genes (MMP2, MMP9, MMP14), consistent with PCNSL's multifocal, whole-brain disease pattern. T-cell analysis corroborates distinct immunosuppression mechanisms, with GBM driven by reduced non-exhausted CD8+ T cells, while PCNSL showed near absence of natural killer cells and an increase in exhausted CD4+ T cells. Furthermore, interaction mapping revealed GBM dominated by tumor–tumor interactions, while PCNSL was enriched in TAM–TAM/T cell interactions. PCNSL also exhibited higher immune checkpoint engagement, though galectin-3 and galectin-9 emerged as dominant ligands in both diseases. Notably, PD-1–PD-L1/PD-L2 interactions, as well as BAFF and APRIL signaling supporting tumor B-cell survival, were exclusive to PCNSL. Both diseases also displayed features consistent with blood–brain barrier disruption, a finding potentially underlying clinical responses to immunotherapies. In summary, these data identify TAMs as central orchestrators of immune suppression in PCNSL, simultaneously promoting B-cell survival, inhibiting T-cell responses, and facilitating tumor spread within the brain. Notably, both diseases exhibit galectin-mediated immune evasion, highlighting this pathway as a potentially targetable checkpoint. Together, this work establishes a framework providing insights into CNS immunotherapies, as PCNSL exhibits an immune-engaged TME serving as an archetype for understanding CNS immune responses, particularly in the context of CAR-T and bispecific antibody therapies, owing to its permissive CNS milieu. Citation Format: Nella Martyna. Tuczko, Colm Keane, Zachary Hawula, Melinda Burgess, Rakin Chowdhury, Lucy Bradbury, Fiona Swain. Differential Immune Suppression in PCNSL and GBM Reveals Checkpoint Dependencies Underlying CNS Therapy Response abstract. In: Proceedings of the AACR Special Conference in Cancer Research: Brain Cancer; 2026 Mar 23-25; Philadelphia, PA. Philadelphia (PA): AACR; Cancer Res 2026;86 (6Suppl): Abstract nr A005.
Tuczko et al. (Mon,) studied this question.