Profiling of patients with glioma reveals the dominant immunosuppressive axis is refractory to immune function restoration

Abstract

In order to prioritize available immune therapeutics, immune profiling across glioma grades was conducted, followed by preclinical determinations of therapeutic effect in immune-competent mice harboring gliomas. T cells and myeloid cells were isolated from the blood of healthy donors and the blood and tumors from patients with glioma and profiled for the expression of immunomodulatory targets with an available therapeutic. Murine glioma models were used to assess therapeutic efficacy of agents targeting the most frequently expressed immune targets. In patients with glioma, the A2aR/CD73/CD39 pathway was most frequently expressed, followed by the PD-1 pathway. CD73 expression was upregulated on immune cells by 2-hydroxyglutarate in IDH1 mutant glioma patients. In murine glioma models, adenosine receptor inhibitors demonstrated a modest therapeutic response; however, the addition of other inhibitors of the adenosine pathway did not further enhance this therapeutic effect. Although adenosine receptor inhibitors could recover immunological effector functions in T cells, immune recovery was impaired in the presence of gliomas, indicating that irreversible immune exhaustion limits the effectiveness of adenosine pathway inhibitors in patients with glioma. This study illustrates vetting steps that should be considered before clinical trial implementation for immunotherapy-resistant cancers, including testing an agent's ability to restore immunological function in the context of intended use.

Keywords: Brain cancer; Cancer immunotherapy; Immunology; Oncology.

Figure 1. Schema demonstrating the dual processing of matched tumor and blood from patients with glioma.

Glioma-infiltrating immune cells were isolated from a Percoll density gradient and further purified with magnetic myelin removal beads. Peripheral blood mononuclear cells (PBMCs) from both patients with glioma and healthy donors were isolated from the Histopaque 1077 density gradient. Both cell populations were subsequently lineage typed into myeloid (CD11b⁺) or T cells (CD3⁺). Thereafter, the cells were stained for the expression of immune modulator targets for which there was an available therapeutic.

Figure 2. Heatmaps demonstrating the frequency of CD8⁺ and CD4⁺ T cell expression for designated immune markers in patients with glioma.

Flow cytometry was conducted for each immune marker relative to its associated isotype control, as described in Supplemental Figure 1. Data are shown for CD8⁺ TILs (A) and PBMCs (B) as well as CD4⁺ TILs (C) and PBMCs (D). The percentage of positive cells was then converted to a heatmap. Each row represents an analysis of a single patient. These are not matched rows, because, in some instances, the tumor or the peripheral blood was not available for the same patient.

Figure 3. Heatmaps demonstrating the mean fluorescent intensity of immune markers expressed on CD11b⁺ myeloid cells in patients with glioma.

Flow cytometry was conducted for each immune marker relative to its associated isotype control, as described in Supplemental Figure 1. The mean fluorescent intensity (MFI) was then converted to a heatmap. Data are shown for CD11b⁺ GIMs (A) and PBMCs (B). Each row represents the analysis of a single patient. These are not matched rows, because, in some instances, the tumor or the peripheral blood was not available for the same patient.

Figure 4. Clinical, genetic, and pathological features that were significantly associated with immune regulatory markers.

(A) Glioma-infiltrating immune cells compared with matched patient blood across all tumor grades. PD-1, LAG-3, CD39, and CD160 were significantly upregulated in CD4⁺ TILs compared with CD4⁺ T cells isolated from patient blood. In CD8⁺ TILs, PD-1, LAG-3, and CD39 were significantly upregulated compared with the matched patient blood. In the myeloid CD11b⁺ cell subset, CD80, B7-H4, and A2aR showed significantly increased expression in GIMs compared with in the matched PBMCs. (B) GBM-infiltrating immune cells compared with matched patient blood in GBM. PD-1, CD160, and CTLA-4 were significantly upregulated in CD4⁺ TILs compared with CD4⁺ T cells isolated from GBM patient blood. In CD8⁺ TILs, PD-1, LAG-3, and CD39 were significantly upregulated compared with in the GBM patient blood. In the myeloid CD11b⁺ cell subset, only A2aR showed significantly increased expression in GIMs compared with in the CD11b⁺ cells isolated from the GBM patient peripheral blood. (C) Glioma-infiltrating immune cells compared with blood from healthy controls. LAG-3, CD39, and PD-1 were significantly upregulated in CD8⁺ T cells isolated from the tumor tissue from patients with glioma compared with in healthy donor (hD) blood. In the myeloid CD11b⁺ cell subset, only A2aR showed significantly increased expression in GIMs compared with in CD11b⁺ cells isolated from healthy donor blood. (D) GBM-infiltrating immune cells compared with blood from healthy donors. PD-1 is the only marker that was significantly upregulated in CD4⁺ and CD8⁺ TILs compared with in healthy donor blood. (E) CD73 was upregulated in CD4⁺ T cells isolated from the peripheral blood from patients with glioma carrying an IDH mutation compared with that of IDH WT patients. Biomarker expression values were compared using a Mann-Whitney test. P values were adjusted to control for the false discovery rate of multiple comparison using Bonferroni’s correction. (F) CD73 expression levels in CD4⁺ and CD8⁺ PBMCs isolated from healthy donors and treated with different concentrations of 2HG. The experiment was performed in duplicate or triplicate using samples from 3 different donors. A mixed-effects model with dosage as a fixed effect, donor as a random effect, and a donor-dosage interaction was fit to the data. The null hypothesis is no shift in mean percentage across doses. P values were computed using parametric bootstrap quantiles. P values were corrected for multiple testing using Bonferroni.

Figure 5. Targeting A2aR in murine glioma models.

(A) A2aR expression was upregulated on CD8⁺ T cells (P = 0.0046), with a trend in CD4⁺ T cells (P = 0.077) isolated from the blood of GL261 tumor-bearing mice (n = 6) compared with those in cells from healthy control mice (n = 6). There was no difference in A2aR expression levels in CD11b⁺ myeloid cells between GL261 tumor-bearing mice (n = 6) and healthy control mice (n = 5) (P = 0.895). P values were calculated using the unpaired 2-tailed t test. (B) Profiling of ex vivo tumor-infiltrating lymphocytes (TILs) demonstrated that both CD4⁺ and CD8⁺ T cell populations expressed A2aR (n = 7). (C) A2aR expression level was similar on CD11b⁺ myeloid cells isolated from GL261 tumor-bearing mice (n = 6) and healthy control mice (n = 5) (P = 0.18). The unpaired 2-tailed t test was used to calculate significance. (D) Survival curve of C57BL/6 mice intracranially implanted with GL261 WT cells and treated with 10 mg/kg SCH58261 (n = 10) or vehicle control (n = 10) for 21 days or until mice showed neurological symptoms of brain tumors. Median survival duration of vehicle control mice was 18.5 days versus 21.5 days with SCH58261 (P = 0.0114). P values were calculated using the log-rank test. (E) Survival curve of Ntv-A mice treated with 10 mg/kg SCH58261 (n = 12) or vehicle control (n = 12) for 21 days. Median survival duration of vehicle control mice was 44.5 days versus 63 days with SCH58261 (P = 0.2292). P values were calculated using the log-rank test. (F) Survival curve of C57BL/6 mice intracranially injected with CD73-overexpressing GL261 cells and treated for 21 days with 60 mg/kg vipadenant (n = 10) or vehicle control (n = 10). Median survival duration of vehicle control mice was 21 days versus 27 days with vipadenant (P = 0.0002). P values were calculated using the log-rank test.

Figure 6. Targeting the adenosine pathway in murine glioma models.

(A) C57BL/6 mice with intracranially implanted CD73-expressing GL261 cells treated with combinatorial anti–PD-1 and SCH58261. (B) Survival curve of C57BL/6 mice intracranially implanted with CD73-overexpressing GL261 cells and treated with anti–PD-1 (days 7, 9, and 11), SCH58261 (daily on days 3–21), vehicle control, or isotype control. The median survival duration of the DMSO + IgG control group was 22 days. Anti–PD-1 increased this to 26 days relative to the control (P = 0.0216), and the A2aR inhibitor increased it to 25 days (P = 0.0310); however, the combination of A2aR and anti–PD-1 was not additive or synergistic for enhanced survival. (C) C57BL6/J mice with intracranially implanted CD73-expressing GL261 cells treated with anti-CD73, the CD39 inhibitor POM-1, the adenosine receptor inhibitor vipadenant, or a combination. (D) Survival curve of C57BL/6 mice intracranially implanted with CD73-overexpressing GL261 cells and treated with IgG + vehicle control (n = 8), IgG + vipadenant (n = 8), anti-CD73 + control (n = 8), anti-CD73 + vipadenant (n = 8), IgG + vehicle control + POM-1 (n = 8), or anti-CD73 + vipadenant + POM-1 (n = 8). Vipadenant or vehicle control was administered daily for 21 days, starting on day 3 (60 mg/kg, i.p.); anti-CD73 or IgG control (200 μg/mouse, i.v.) was administered on days 3, 6, 10, 13, 17, and 20; and POM-1 or PBS control (5 mg/kg; i.p.) was administered daily from day 3 to day 13. The median survival duration of the control IgG-treated group was 27 days, anti-CD73 + control was 32 days, POM + control was 29 days, vipadenant + control was 30 days, and CD73 + vipadenant + POM was 31 days. P values were calculated using the log-rank test.

Figure 7. T cell immune suppression is refractory to adenosine inhibitors in the presence of glioma.

Human T cells were stimulated with CD3 and CD28, and the percentage of intracellular IFN-γ of CD4⁺ and CD8⁺ T cells was quantified with flow cytometry. The percentage of IFN-γ in the DMSO control was set at 100% as the baseline. The cells were treated with the adenosine receptor inhibitors vipadenant and SCH58261, which can modestly increase the production of IFN-γ in both CD4⁺ and CD8⁺ T cells. The adenosine receptor agonist NECA was used to induce immune suppression in these T cell populations and, in combination with adenosine receptor inhibitors, to recover immunological function, as measured by IFN-γ production. However, in the presence of supernatants from U87 gliomas and during coculture, both CD4⁺ and CD8⁺ T cells were refractory to immunological functional restoration. The experiment was performed in technical duplicate using PBMCs from 3 different donors. Paired t tests on the average replicates were performed to calculate P values. P values were Bonferroni’s multiplicity corrected within each cell type.