Evaluation of T-Cell Compartment By Complex Multiparameter Flow Cytometry Reveals Distinct Patterns of T-Cell Exhaustion in DLBCL, FL and HL Patients

Abstract

Listen

Translate article

Background: Immune checkpoint (IC) blockade has revolutionized the treatment of chemo-refractory solid tumors and has demonstrated promising results for the treatment of resistant blood cancers, including lymphoma. However, clinical response to PD1 blockade depends on the subtype of lymphoma, with durable responses observed in 70 % of Hodgkin lymphoma (HL) patients versus only 36 to 40% of Diffuse Large B-Cell Lymphoma (DLBCL) and Follicular Lymphoma (FL) patients(1-3). The biological aspects that hinder the efficacy of immunotherapy in DLBCL and FL patient are not well-known. Primary tumor resistance to PD1 blockade could be driven by inherent genomic alterations or by ineffective anti-tumor T cell responses. ICs expressed at the surface of tumor-infiltrating-lymphocytes (TILs) use distinct suppressive mechanisms to enforce T-cell dysfunction. The collective expression of ICs defines multiple subsets of exhausted T cells during cancer(4). Since multiple of these ICs collaborate to repress anti-tumor T cell functions, we hypothesized that TILs from DLBCL and FL patients express high levels of ICs and present a more advanced exhausted phenotype compared to TILs from HL patients. Methods: We studied lymphoma samples collected at diagnosis and frozen as a viable cell suspension from 57 patients with HL, FL or DLBCL. We performed an extended immunophenotype of TILs through flow cytometry using maturation markers, exhaustion markers, homing and chemokine receptors. The results were analyzed with FlowJo. Boolean gating was used to assess IC co-expression. Statistical analyses were performed using Prism. Results: Our data indicated that CD8 TILs from DLBCL patients express higher levels of classical ICs such as PD1 (1059 vs 1015 and 648), Tim3 (1747 vs 868 and 918,) and Lag3 (2603 vs 1013 and 1031) compared to FL and HL patients (p-value < 0.01), while the expression of 2B4, BTLA and CD160 were similar. Boolean analysis revealed that DLBCL and FL had a higher frequency of CD8T cells that co-expressed 6-7 ICs, compared to HL, suggesting a more advanced state of exhaustion (19% and 26% vs 14%, p-value < 0.05). This correlated with an increased frequency of PD-1highCD39high T cells in DLBCL and FL patients compared to HL (32% and 24% vs 14%, p-value < 0.005), which correlated with increased frequency of EOMEShigh cells in both DLBCL and FL (84% and 76% vs 55%, p-value < 0.005). These markers identify populations of cells with phenotypic and transcriptomic profiles of terminal exhaustion with reduced cytokine production, proliferative capacity and persistence (5, 6). In contrast, stem-like exhausted CD8 T cells are defined by intermediate expression of PD-1 and expression of the chemokine receptor CXCR5, and are driven by the transcription factor TCF-1(7, 8). We observed that DLBCL exhibited significantly less progenitor Tim3-/CXCR5+ cells compared to FL and HL (27% vs 52% and 40%, p-value < 0.01). Further, the frequency and mean fluorescent intensity of TCF-1 was significantly reduced in DLBCL compared to FL and HL (43% / 495 ± 64 vs 70% / 752 ± 42 and 61% / 670 ± 74, p-value < 0.01). This data correlated with lower frequency of CD27+ CD8 T cells in DLBCL vs FL and HL (82% vs 96% and 96%, p-value < 0.01). Conclusion: Collectively, our data indicate that TILs from DLBCL and FL patients express a whole array of ICs that have non-redundant suppressive functions. Further, these patients present subsets of cells with increased functional exhaustion, subsets demonstrated to be unresponsive to PD-1 blockade(7). Further, DLBCL patients lack protective stem-like T-cell subsets which were recently demonstrated to correlate with clinical outcome(9). These results enlighten our knowledge of the cellular defects precluding potent anti-tumor T cell functions in DLBCL and FL and propose novel targets for immunotherapies in PD-1-unresponsive lymphomas. Research funding: This work is supported by the Canadian Institutes of Health Research (PJT-155996), Canadian Cancer Society and Cole Foundation (grant #705478).