The Immune Micro-Environment in Diffuse Large B Cell Lymphoma Is Characterised By an Immunosuppressive Shift in T Cell Subsets

Diffuse large B-cell Lymphoma (DLBCL) is the most frequent non-Hodgkin's lymphoma with 3 molecularly distinct subtypes based on cell of origin. Genetic alterations in DLBCL, expression of checkpoint molecules and an immunosuppressive microenvironment (ME) all contribute to escape from host anti-lymphoma immunity. The clinical success of monoclonal antibodies that engage the immune system and CAR-T cellular therapy have further highlighted the importance and therapeutic potential of the immune ME in DLBCL. Here we present data from comprehensive phenotyping of cell suspensions from diagnostic DLBCL and reactive lymph node / tonsil (RLNT) biopsies by cytometry by time of flight (CyTOF), with a focus on the T-cell compartment.

Cryopreserved samples from 6 DLBCL (5 LN, 1 spleen) at diagnosis and 5 RLNT (3 LN, 2 tonsil) were stained with a panel of metal-tagged antibodies and analysed by CyTOF2. Samples were acquired in 2 batches with the same RLNT (LN) sample with each to ensure staining consistency. Data were normalised, uploaded to Cytobank, gated to CD45+ CD3+ live single cells and exported for further analysis with Cytofkit in R.

CD3+ events were gated further into CD4+ and CD8+ subsets, which demonstrated that CD4+ T cells were the predominant phenotype in all samples. However, there was a marked skewing of the CD4:CD8 ratio, with CD4+ T cells lower as a percentage of CD3+ T cells in the DLBCL samples (55.84 v 78.18, p=0.0173*). CD8+ T cells were higher as a percentage in DLBCL (36.22 v 16.75, p=0.03*) with no difference seen in double negative (DN) T cells. CD3+ T cells were then clustered with FlowSOM and visualised according to the tSNE algorithm. A heatmap of median marker expression intensity was generated to facilitate cluster identification. This revealed a number of differences in cluster abundance between the groups, with a significant shift in differentiation away from naïve and towards an effector memory (EM) phenotype in DLBCL.

There were fewer cells in the CD27+ CD28+ CCR7+ CD45RA+ CD4+ naïve cluster in the DLBCL samples than the RLNT (p=0.0173*). Although the DLBCL samples showed an overall reduction in CD4+ T cells, the clusters of regulatory T cells (Treg: CD4+ CD25+ FOXP3+ and CD127-/low) consisted of more cells from these cases than the RLNT (p=0.0043**). Within the Treg population, the DLBCL patients had more Th1 polarised (T-bet+) Tregs and more PD-1 expressing Tregs. The Th1 Tregs predominantly secreted the suppressive cytokines IL-2, IL-10 and TGF-β on stimulation and may play a role in inhibiting Th1 responses. Conventional Th1 were not increased in DLBCL resulting in a higher Th1 Treg to Th1 ratio than in RLNT. There was a trend for RLNT samples to contribute more cells to the PD-1 high follicular helper T cell (TFH) cluster and DLBCL to the PD-1+ TIM-3+ DN cluster.

The DLBCL ME had relatively more CD8+ T cells and contributed more to the CCR7- CD45RA- CD8+ EM clusters (p=0.0173*) but the CD8+ T cells in the RNLT samples tended to a naïve CCR7+ CD45RA+ PD-1- phenotype (p=0.0519). The CD8+ EM cells enriched in the DLBCL ME expressed the cytotoxic markers granzyme and perforin and responded to stimulation with degranulation (CD107a) and cytokine production (IFNγ, TNFα, TGFβ and IL-10), not suggestive of exhaustion. It is also notable that a cluster of PD-1+ TIM-3+ CD8+ EM with reduced markers of cytotoxicity, low CD107a expression and poor cytokine production after stimulation was predominantly made up of cells from DLBCL suspensions (p=0.002**).

CyTOF analysis of the DLBCL ME has demonstrated a shift in the balance of T cell subsets and CD4:CD8 ratio with a relative abundance of immunosuppressive Tregs despite an overall reduction in the CD4+ population and a skew towards differentiation in CD4+ and CD8+ populations. The cytotoxic T cells in DLBCL tended to have an EM phenotype and express immune checkpoint molecules but remained capable of cytokine production. However, the production of IFNγ by these effector T cells may play a role in the development of inhibitory Tregs with a Th1 phenotype, which were enriched in these patients. A cluster of CD8+ EM cells expressing checkpoint molecules and displaying characteristics of exhaustion following stimulation was also seen in these DLBCL patients. These data provide new insights into the immunosuppressive nature of the DLBCL ME and provide a rationale for targeting the ME alongside existing therapeutic approaches, including CAR-T cells to improve outcomes.