Immune Checkpoint Inhibitor-induced Reinvigoration of Tumor-infiltrating CD8⁺ T Cells is Determined by Their Differentiation Status in Glioblastoma

This article is featured in Highlights of This Issue, p. 2361

Glioblastoma (GBM, WHO grade IV glioma) is the most common primary malignant brain tumor in adults presenting aggressive progression with poor prognosis (1). The current standard treatment is maximal surgical resection of the tumor, followed by radiotherapy and temozolomide chemotherapy. However, the current protocol is far from successful, as almost all cases of GBM ultimately recur following treatment, with a median overall survival of approximately 14–15 months after diagnosis (1–3). Although cancer therapies have improved significantly over the years, no effective treatment is available that can overcome the limitations of the current standard treatment and improve the survival of patients with GBM (4).

Immune checkpoint inhibitors (ICI) recently ushered in a new era of cancer immunotherapy. Immune checkpoint receptors are normally expressed on activated T cells to prevent excessive immune responses. However, following chronic antigen exposure during chronic viral infection or cancer, the effector T cells differentiate into exhausted T cells (5, 6). Exhausted T cells are in a dysfunctional state, expressing high levels of immune checkpoint receptors, including programmed cell death protein 1 (PD-1), cytotoxic T-lymphocyte antigen 4 (CTLA-4), T-cell immunoglobulin domain and mucin domain protein 3 (Tim-3), and lymphocyte activation gene 3 protein (LAG-3; ref. 7). The antitumor functions of exhausted T cells can be restored by blocking immune checkpoint receptors. In particular, mAbs against PD-1 have been approved for therapeutic use for the treatment of various tumors, including melanoma, non–small cell lung cancer, head and neck cancer, renal cell carcinoma, and Hodgkin lymphoma. Moreover, recent findings suggest that ICIs can induce durable remission that lasts for several years (8–12).

In the case of GBM, the therapeutic effect of ICIs is being tested in various settings (13). In a preclinical setting, the therapeutic effect of ICIs in a murine model with orthotopically transplanted GBM demonstrated an improved survival rate (14–16). However, the objective response rate of nivolumab (anti-PD-1) in patients with recurrent GBM was low in a recent clinical trial (17–19). To overcome the limitation of using ICIs in GBM, it is necessary to better understand the phenotypic and functional characteristics of tumor-infiltrating lymphocytes (TIL), CD8⁺ cytotoxic T cells.

In this study, we investigated the phenotype and differentiation status of CD8⁺ TILs from surgically resected GBM tissues. We also examined whether blocking the PD-1/PD-L1 pathway can restore the functions of exhausted CD8⁺ TILs ex vivo. We found that PD-1⁺CD8⁺ TILs cells exhibit more terminally differentiated phenotypes represented by Eomes^hiT-bet^lo than their PD-1⁻ counterparts. Importantly, terminal differentiation of CD8⁺ TILs is associated with poor reinvigoration of CD8⁺ TILs upon ex vivo treatment with anti-PD-1 and/or anti-CTLA-4 treatment, indicating that ICI-induced reinvigoration of CD8⁺ TILs is determined by their differentiation status in GBM.

Ninety-eight patients who underwent surgical resection of GBM were enrolled in the study from May 2015 to November 2018 at Severance Hospital (Seoul, Republic of Korea). Demographic and clinical information is provided in Supplementary Table S1. Fresh tumor tissues and paired peripheral blood were collected on the day of resection. None of the patients received chemotherapy or radiotherapy before surgery. This study was approved by the Institutional Review Board, and all enrolled patients agreed to participate in the study by providing informed consent. This study was conducted in accordance with Declaration of Helsinki.

Peripheral blood mononuclear cells (PBMC) were isolated from the whole blood by density gradient centrifugation (Lymphocyte Separation Medium). We performed meticulous mechanical and enzymatic dissociation using a gentleMACS dissociator and the Human Tumor Dissociation Kit (Miltenyi Biotec) following the manufacturer's instructions. Isolated single-cell suspension from tumors was filtered through a 100-μm pore cell strainer. TILs were separated from myelin by centrifugation in Percoll (GE Healthcare). Isolated TILs were cryopreserved until further use.

Cryopreserved PBMCs and TILs were thawed and stained using the LIVE/DEAD Fixable Red Dead Cell Stain Kit (Life Technologies). The cells were then washed once and stained with fluorochrome-conjugated antibodies in the dark at 4°C for 30 minutes. For the staining of CTLA-4, T-box transcription factors T-bet and Eomesodermin (Eomes), cells were fixed and permeabilized using Foxp3 Staining Buffer Kit (eBioscience) following the manufacturer's instructions. To detect tumor antigen-specific T cells, PBMCs, or TILs from HLA-A2⁺ patients (n = 11) were pretreated with protein tyrosine kinase inhibitor (dasatinib, Axon Medchem) at a final concentration of 50 nmol/L in PBS at 37°C for 30 minutes. The cells were then stained with phycoerythrin-labeled HLA-A*0201 dextramer: NY-ESO-1_157-165 minimal epitope (SLLMWITQV/HLA-A*0201; Immunedex) for 20 minutes at room temperature and then washed twice. This was followed by live/dead, surface staining, and the intracellular protein staining protocol described above. Multicolor flow cytometry was performed on an LSR II instrument (BD Biosciences). Data were analyzed using FlowJo software (Treestar).

Multicolor flow cytometry was performed using the following fluorochrome-conjugated mAbs: anti-CD8 (SK1), anti-CD3 (UCTH1), anti-CD45RA (HI100), anti-CD14 (MϕP9), anti-CD19 (HIB19), anti-CD4 (SK3), anti-HLA-A2 (BB7.2), anti-T-bet (O4-46), anti-NKG2D (1D11), and anti-IFNγ (B27) from BD Biosciences; anti-PD-1 (EH.12.2H7), anti-Tim-3 (F38-2E2), anti-CTLA-4 (L3D10), and anti-CCR7 (G043H7) from BioLegend; anti-LAG-3 (3DS223H), anti-Eomes (WD1928), and anti-TNFα (Mab11) from eBioscience.

Cryopreserved TILs (n = 67) were thawed and suspended in RPMI1640 containing 10% FBS and rested at 37°C in a 5% CO₂ incubator for 8 hours. TILs were labeled with CellTrace Violet (CTV; Thermo Fisher Scientific). A total of 2 × 10⁵ cells in 200 μL medium were cultured in each well of a 96-well round-bottom culture plate and stimulated with soluble anti-CD3 antibody (1 ng/mL, OKT-3, eBioscience) in the presence of 10 μg/mL of antibodies blocking immune checkpoint receptors [anti-PD-1 (EH12.2H7) and anti-CTLA-4 (L3D10)] or isotype control (mIgG₁, MOPC-21; all from BioLegend).

After 108 hours of culture in the 5% CO₂ incubator, cells were harvested and stained with the following fluorochrome-conjugated mAbs: anti-CD8 (RPA-T8), anti-CD3 (HIT3a, BioLegend), anti-CD4 (SK3), anti-CD14 (MϕP9), anti-CD19 (HIB19, eBioscience), and 7-aminoactinomycin D (eBioscience). CTV^loCD8⁺ T cells were counted as proliferated cells. To further evaluate proliferating CTV^loCD8⁺ T cells, the mitotic index was calculated by dividing mitotic events by the absolute number of precursor cells based on the number of cells in each mitotic division. We counted the number of divided cells up to the eighth mitotic division based on the fluorescence intensity of CTV. Next, the stimulation index was determined by dividing the mitotic index of blocking antibody-treated samples by that of the isotype-treated samples (20, 21).

During surgical resection of GBM, tumor tissues were obtained and frozen (n = 60). Small pieces of tumor tissues were homogenized in the lysis buffer using Precellys 24 homogenizer (Bertin Technologies), and the homogenates were used to extract total RNA using GeneAll Ribospin (GeneAll Biotechnology). Complementary DNA was synthesized using a ReverTra Ace qPCR RT Master Mix with gDNA remover (Toyobo). TaqMan Gene Expression Assays (Applied Biosystems) were used to determine the mRNA levels of target genes, including TGFB1, TGFB2, FOXP3, PTGS2, PTGES, KLRK1, MICA, MICB, ULBP1, ULBP2, ULBP3, RAET1E, IL10, IDO1, CD274, CXCL10, STAT1, and ACTB. SYBR Green real-time PCR was performed to determine the mRNA levels of target genes, including CXCL9, HLA-DRA, and IFNG. The data are presented as relative gene expression compared with β-actin (2^C_t^{(β-actin)−C}_t^{(target gene)}).

HLA-A*0201(+) T98G GBM cell line was purchased from Korean Cell Line Bank. PBMCs from HLA-A*0201(+) donors were used to generate NY-ESO-1₁₅₇–specific CD8⁺ T-cell lines. Briefly, CD8⁺ T cells were isolated by CD8⁺ T Cell Isolation Kit (Miltenyi Biotec) and NY-ESO-1₁₅₇–specific CD8⁺ T cells were enriched with HLA-A*0201:NY-ESO-1_157-165 (SLLMWITQV) dextramer (Immunedex). Enriched NY-ESO-1₁₅₇–specific CD8⁺ T cells were maintained in IL2 (200 IU/mL, PeproTech), IL7 (10 ng/mL, PeproTech), and IL15 (100 ng/mL, PeproTech)-containing media. Anti-CD3 (Miltenyi Biotec) was used to expand or stimulate CD8⁺ T-cell lines.

In coculture assays, target cells (T98G) was labeled with PKH26 dye (Sigma-Aldrich), and pulsed with 10 μg/mL NY-ESO-1_157-165 peptide (SLLMWITQV; JPT). Cytotoxicity assays and intracellular cytokine staining (ICS) were performed in 96-well flat-bottom culture plates precoated with recombinant human CD80 and CD86 Fc protein (10 μg/mL, Sino Biological Inc). T98G and NY-ESO-1₁₅₇–specific CD8⁺ T cells (1:1 effector:target ratio) were cocultured in the presence of anti-PD-1 [EH12.2H7] and/or anti-CTLA-4 [L3D10] or isotype control (mIgG₁, MOPC-21; 10 μg/mL, all from BioLegend). Following 6 hours of coculture, TO-PRO-3-iodide (Thermo Fisher Scientific) was added to the coculture at a final concentration of 0.5 μmol/L, and the cells were immediately analyzed by flow cytometry. TO-PRO-3⁺ cells in PKH26⁺ cells were considered as dead target cells. Apart from cytotoxicity assays, ICS was performed in the presence of brefeldin A (GolgiPlub, BD Biosciences) and monensin (GolgiStop, BD Biosciences). After 6 hours of coculture, live/dead, surface and intracellular staining were performed as described above.

Statistical analyses were performed using Prism software 6 (GraphPad Software). The D'Agostino & Pearson omnibus normality test was used to test for a normal distribution of continuous data. The independent samples t test or Mann–Whitney U test was used to compare the continuous variables. We used a paired t test or Wilcoxon matched-pairs signed rank test to compare the continuous variables of paired groups. To compare three or more groups, data showing the Gaussian distribution were analyzed using one-way ANOVA followed by the Holm–Sidak multiple comparison test. For the non-Gaussian distributed data, the Kruskall–Wallis test was performed followed by Dunn multiple comparisons test. Spearman or Pearson correlation tests were performed to assess the significance of the statistical correlation. Categorical variables were analyzed by the Pearson χ² test with SPSS software (IBM Corp.). All tests of significance were two-tailed and P ≤ 0.05 considered significant.

First, we analyzed the expression of immune checkpoint receptors on CD8⁺ TILs compared with peripheral blood CD8⁺ T cells by examining PD-1, CTLA-4, Tim-3, and LAG-3 after excluding CD45RA⁺CCR7⁺ naïve cells (Supplementary Fig. S1). CD8⁺ TILs exhibited significantly higher percentages of PD-1⁺, CTLA-4⁺, Tim-3⁺, and LAG-3⁺ cells compared with peripheral blood CD8⁺ T cells (Fig. 1A and B). Among immune checkpoint receptors, PD-1 was predominantly expressed by CD8⁺ TILs in terms of the percentage of expressing cells. We also examined tumor antigen-specific CD8⁺ T cells using an HLA-A*0201, NY-ESO-1₁₅₇ dextramer. The percentage of NY-ESO-1₁₅₇–specific CD8⁺ T cells was significantly higher among CD8⁺ TILs than peripheral blood CD8⁺ T cells (Fig. 1C and D). NY-ESO-1₁₅₇–specific CD8⁺ TILs exhibited significantly higher percentages of PD-1⁺, CTLA-4⁺, and Tim-3⁺ cells compared with peripheral blood CD8⁺ T cells, whereas the percentage of LAG-3⁺ cells was not different (Fig. 1E).

Among the 98 patients, 70 were treated with 20 mg of dexamethasone (133-mg prednisone equivalent) daily before tissue collection. We compared the expression of immune checkpoint inhibitory receptors between patients with or without steroid treatment. CD8⁺ TILs from steroid-treated patients exhibited significantly higher percentages of PD-1⁺, CTLA-4⁺, and Tim-3⁺ cells compared with those from untreated patients (Fig. 1F). This result is interesting because it was recently reported that patients with non–small cell lung cancer with corticosteroid use (≥10 mg prednisone equivalent daily) showed poor response to anti-PD-1 therapy (22).

We investigated the coexpression pattern of multiple immune checkpoint receptors on CD8⁺ TILs. CTLA-4, Tim-3, and LAG-3 tended to be more expressed on PD-1⁺CD8⁺ TILs than PD-1⁻CD8⁺ TILs (Fig. 2A). The percentage of CTLA-4⁺, Tim-3⁺, and LAG-3⁺ cells was significantly higher in PD-1⁺CD8⁺ TILs than in PD-1⁻CD8⁺ TILs (Fig. 2B). These results were confirmed in tumor antigen-specific CD8⁺ TILs. NY-ESO-1₁₅₇–specific PD-1⁺CD8⁺ TILs exhibited a significantly higher percentage of CTLA-4⁺ and Tim-3⁺ cells than NY-ESO-1₁₅₇–specific PD-1⁻CD8⁺ TILs, and a tendency for higher percentages of LAG-3⁺ cells (Fig. 2C). When we analyzed the coexpression of four immune checkpoint receptors in combinations, PD-1⁺CTLA-4⁻Tim-3⁻LAG-3⁻ cells were a predominant population among CD8⁺ TILs, and PD-1⁺CTLA-4⁺Tim-3⁻LAG-3⁻ cells were the next frequent population (Fig. 2D). When we analyzed the coexpression of immune checkpoint receptors in peripheral blood CD8⁺ T cells, the results were similar to those obtained for CD8⁺ TILs although the percentages in peripheral blood CD8⁺ T cells were lower (Supplementary Fig. S2).

Next, we examined the differentiation status of CD8⁺ TILs. We focused on the expression of transcriptional factors related to T-cell differentiation. T cells undergo a distinctive transcriptional program when exposed to persistent antigen stimulation, which subsequently drives the expression of multiple immune checkpoint receptors (7). In particular, Eomes^hiT-bet^loCD8⁺ T cells are terminally differentiated cells, whereas T-bet^hiEomes^loCD8⁺ T cells are progenitor-like cells among the exhausted CD8⁺ T cells (23–27). We found that the percentage of Eomes^hiT-bet^lo cells was significantly higher among PD-1⁺CD8⁺ TILs than PD-1⁻CD8⁺ TILs (Fig. 3A and B). In contrast, the percentage of T-bet^hiEomes^lo cells was significantly lower among PD-1⁺CD8⁺ TILs. We confirmed these results in the tumor antigen-specific CD8⁺ TILs. PD-1⁺NY-ESO-1₁₅₇–specific CD8⁺ T cells had a significantly higher percentage of Eomes^hiT-bet^lo cells than PD-1⁻NY-ESO-1₁₅₇–specific CD8⁺ T cells, and a tendency to have a lower percentage of T-bet^hiEomes^lo cells (Fig. 3C). When we examined CTLA-4⁺CD8⁺ TILs, we found that the percentages of Eomes^hiT-bet^lo and T-bet^hiEomes^lo cells were not different between CTLA-4⁺CD8⁺ and CTLA-4⁻CD8⁺ TILs (Fig. 3D). The percentages of Eomes^hi T-bet^lo and T-bet^hiEomes^lo cells among PD-1⁺CD8⁺ TILs were not significantly different in steroid-treated and untreated patients (Fig. 3E).

We evaluated the reinvigoration capacity of CD8⁺ TILs following ex vivo treatment with anti-PD-1–blocking antibodies under TCR signal stimulation by anti-CD3 antibodies. Anti-PD-1 treatment significantly increased the anti-CD3–stimulated proliferation of CD8⁺ TILs (Fig. 4A and B). However, not all patient-derived CD8⁺ TILs responded to PD-1 blockade. CD8⁺ TILs from some patients exhibited minimal reinvigoration, whereas those from other patients were robustly reinvigorated upon PD-1 blockade.

To further analyze differences in the degree of anti-PD-1–induced reinvigoration, we divided CD8⁺ TIL samples into two groups based on the median stimulation index value as “highly responding TILs” and “low responding TILs” groups (Fig. 4C). No significant difference was found between the two groups in demographic and clinical factors, including IDH1 mutation, MGMT promotor methylation, EGFR amplification, and preoperative steroid treatment (Table 1). In addition, we found no difference in the expression of immune checkpoint receptors on CD8⁺ TILs (Fig. 4D). Intriguingly, the percentage of Eomes^hiT-bet^lo cells among PD-1⁺CD8⁺ TILs was significantly higher in the “low responding TILs” group than in the “highly responding TILs” group, but no difference was found in the percentage of T-bet^hiEomes^lo cells (Fig. 4E). Moreover, the percentage of Eomes^hiT-bet^lo cells among PD-1⁺CD8⁺ TILs inversely correlated with the stimulation index, which represented anti-PD-1–induced increase in anti-CD3–stimulated proliferation (r = −0.6371, P < 0.001; Fig. 4F). These data indicate that the terminal differentiation of CD8⁺ TILs is associated with poor reinvigoration capacity upon PD-1 blockade.

We also divided the CD8⁺ TIL samples into two groups based on CTLA-4 blockade or combined PD-1/CTLA-4 blockade (Fig. 4G and J). As in PD-1 blockade, the percentage of Eomes^hiT-bet^lo cells among CD8⁺ TILs was higher in the “low responding TILs” group than in the “highly responding TILs” group although it did not reach statistical significance when two groups were divided on the basis of CTLA-4 blockade (Fig. 4H for CTLA-4 blockade, P = 0.0697; and Fig. 4K for combined PD-1/CTLA-4 blockade, P < 0.05). Moreover, the percentage of Eomes^hiT-bet^lo cells among CD8⁺ TILs inversely correlated with the stimulation index (Fig. 4I for CTLA-4 blockade, r = −0.3842, P < 0.05 and Fig. 4L for combined PD-1/CTLA-4 blockade, r = −0.4474, P < 0.01).

We investigated whether poor reinvigoration capacity upon PD-1 blockade was associated with the expression of immune regulation–related genes. We examined mRNA expression of TGFB1, TGFB2, IL10, CD274 (PD-L1), PTGS2, PTGES, FOXP3, ULBP1∼3, RAET1E, MICA, MICB, KLRK1 (NKG2D), IDO1, CXCL10, CXCL9, HLA-DRA, STAT1, and IFNG by qRT-PCR. Among these genes, IDO1, CXCL10, CXCL9, HLA-DRA, STAT1, and IFNG have been known as the IFNγ-related gene signature, which can predict the treatment response to PD-1 blockade therapy (28). However, we found no significant difference in the expression of those genes between the “highly responding TILs” and “low responding TILs” groups (Supplementary Fig. S3A). In addition, we analyzed the percentage of CD4⁺CD25⁺FoxP3⁺ regulatory T cells among CD4⁺ TILs and the expression level of NKG2D among CD8⁺ TILs. However, these parameters were not significantly different between the two groups (Supplementary Fig. S3B and S3C).

We examined the effect of combination immune checkpoint blockade. As CTLA-4 was the second most dominant immune checkpoint receptor after PD-1 (Fig. 2D), we examined the effect of combined PD-1/CTLA-4 blockade. First, we established in vitro coculture system of NY-ESO-1₁₅₇–specific CD8⁺ T-cell lines (Supplementary Fig. S4A) and HLA-A*0201(+) T98G GBM cells. NY-ESO-1₁₅₇–specific CD8⁺ T-cell lines were induced to express PD-1 and CTLA-4 (Supplementary Fig. S4B), and T98G GBM cells expressed PD-L1. Coculture plates were precoated with recombinant CD80 and CD86 protein. When we performed in vitro cytotoxicity assays using NY-ESO-1₁₅₇–specific CD8⁺ T-cell lines and NY-ESO-1₁₅₇ peptide-pulsed T98G GBM cells, the specific cytotoxicity was significantly increased by anti-PD-1 or anti-CTLA4 treatment, and further increased by combined treatment with anti-PD-1 and anti-CTLA4 (Supplementary Fig. S4C). In addition, the percentage of IFNγ⁺TNFα⁺ cells among NY-ESO-1₁₅₇–specific CD8⁺ T cells was increased by anti-PD-1 or anti-CTLA-4 treatment, and further increased by combined treatment with anti-PD-1 and anti-CTLA4 (Supplementary Fig. S4D).

Next, we examined the effect of combined PD-1/CTLA-4 blockade using TILs from GBM. The combination further enhanced the proliferation of CD8⁺ TILs over single treatment with anti-PD-1 in the “highly responding TILs” group, but not in the “low responding TILs” group (Fig. 5A and B). When we divided the CD8⁺ TIL samples into “highly responding TILs” and “low responding TILs” groups based on CTLA-4 blockade (Fig. 4G), the combination effect of anti-CTLA-4 and anti-PD-1 antibodies was more prominent in the “highly responding TILs” group compared to the “low responding TILs” group (Fig. 5C). Taken together, these data show that the reinvigoration capacity of PD-1⁺CD8⁺ TILs is influenced by the percentage of terminally differentiated cells, represented by Eomes^hiT-bet^lo cells, and is not reversed by a combination of multiple ICIs.

Although a therapeutic approach using ICIs had breakthrough success in the treatment of cancer, such as melanoma and non–small cell lung cancer, there remain more types of cancer that are resistant to ICIs (29–32). To overcome the limitation of ICIs, we need to understand the immunologic nature of TILs in detail. The current study focused primarily on CD8⁺ TILs, which are thought to be the target of ICIs and gave us a clue as to why a varying degree of clinical responses occur with ICI treatment.

In this study, we studied not only which immune checkpoint receptors are expressed on CD8⁺ TILs in patients with GBM, but also whether their proliferation can be restored by blocking immune checkpoint receptors. Previous studies have shown increased expression of PD-1 (33–36), Tim-3 (35–37), and LAG-3 (36) on CD8⁺ TILs in patients with GBM; however, to the best of our knowledge, ex vivo functional assays to reinvigorate CD8⁺ TILs by ICI treatment have not yet been performed in GBM.

In this study, we found that PD-1⁺CD8⁺ TILs exhibited a significantly higher proportion of Eomes^hiT-bet^lo cells and lower proportion of T-bet^hiEomes^lo cells. These findings are consistent with the results of previous studies using a mouse viral infection model. During chronic LCMV infection in mice, a subset of exhausted CD8⁺ T cells expressing T-bet^hiEomes^lo had proliferative capacity and restoration potential when treated with ICIs, whereas the Eomes^hi T-bet^lo CD8⁺ T cells represented a terminally differentiated subset with limited potential for reinvigoration (23–26). The role of T-bet and Eomes in exhausted CD8⁺ T cells has been shown in patients with chronic hepatitis C virus and HIV infection (24, 27). However, the expression of T-cell transcription factors can be changed by immune checkpoint blockade. It was previously demonstrated that combination blockade of PD-1, CTLA-4, and LAG-3 increased the expression of T-bet in CD8⁺ T cells in a murine tumor model (38).

Using HLA-A*0201 dextramer loaded with NY-ESO-1_157-165 peptide, we demonstrated for the first time the characteristics of tumor-associated antigen (TAA)-specific CD8⁺ T cells in GBM. Although several TAAs have already been described, such as EGFRvIII, IL13Rα2, survivin, and NY-ESO-1 (39, 40), no previous study has detected and analyzed the immune phenotype of TAA-specific CD8⁺ T cells ex vivo in GBM. We successfully detected NY-ESO-1₁₅₇–specific CD8⁺ T cells in HLA-A*0201 patients. We demonstrated that these TAA-specific CD8⁺ T cells are enriched in GBM TILs and exhibit an exhaustion phenotype with the upregulation of immune checkpoint receptors. In addition, PD-1⁺NY-ESO-1₁₅₇–specific CD8⁺ T cells exhibited the Eomes^hiT-bet^lo terminally differentiated phenotype.

Next, we performed a functional assay to test the effectiveness of blocking the immune checkpoint receptors of TILs ex vivo. We could divide the patients into “highly responding TILs” and “low responding TILs” groups based on the proliferative response to PD-1 blockade. “Low responding TILs” had a higher proportion of Eomes^hiT-bet^lo cells, a terminally differentiated subset with limited potential for restoration, than the “highly responding TILs” (23–26). Only CD8⁺ TILs derived from the “highly responding TILs” group demonstrated a further enhanced response to the combination of PD-1 and CTLA-4 compared with PD-1 treatment alone. The findings suggest that CD8⁺ TILs of “low responding TILs” would not be reinvigorated even by combination of multiple ICIs.

In PD-1 and CTLA-4 blockade, distinct cellular mechanisms were recently reported by Wei and colleagues (41). PD-1 blockade results in the expansion of exhausted-like CD8⁺ T cells, and CTLA-4 blockade induces the expansion of Th1-like CD4⁺ effector T cells, in addition to exhausted-like CD8⁺ T cells (41). These distinct mechanisms need to be considered when anti-PD-1 and anti-CTLA-4 antibodies are combined.

Recently, Ayers and colleagues analyzed the gene expression profiles of baseline tumor samples from the pembrolizumab-treated patients with various types of cancer, although GBM was not included in this study. They reported that the IFNγ-related gene signature (IDO1, CXCL10, CXCL9, HLA-DRA, STAT1, and IFNG) could predict the treatment response to PD-1 blockade therapy (28). We applied this IFNγ-related gene signature to our GBM samples and analyzed whether the expression of these genes varied between the “highly responding TILs” and “low responding TILs” groups. However, we found no significant difference between the two groups. GBM might have a different biology from other types of cancer, which are indicated for anti-PD-1 blockade therapy and included in the study by Ayers and colleagues (28).

In line with the current trend, the use of ICIs in patients with GBM is actively being studied in clinical trials. CheckMate 143 (NCT 02017717), the first prospective clinical trial using ICIs in with GBM, recently reported results showing the safety and tolerance of anti-PD-1 (nivolumab) with or without anti-CTLA-4 (ipilimumab) in recurrent patients with GBM (17). The report also included the results of a phase III clinical trial comparing the efficacy of treatment with nivolumab alone to anti-VEGF (bevacizumab) in patients with recurrent GBM. However, nivolumab did not prolong overall survival in recurrent patients with GBM compared with bevacizumab (18, 19). The CheckMate 143 trial enrolled patients with recurrent GBM rather than newly diagnosed GBM, and was not designed to test the efficacy of ICI combination therapy. In future, the efficacy of ICIs with radiotherapy and/or temozolomide chemotherapy will need to be investigated in patients with primary GBM.

To successfully extend the application of ICIs to patients with GBM, it is important to understand the immunologic nature of TILs in detail. In this study, we conclude that tumor-infiltrating CD8⁺ T cells in patients with newly diagnosed GBM are exhausted and PD-1 blockade could revitalize the CD8⁺ TIL response. However, some patients' TILs exhibit a low response to anti-PD-1 and can be distinguished by the expression level of Eomes in PD-1⁺CD8⁺ TILs. Appropriate combination will be a key for successful treatment of GBM with anti-PD-1. Anti-PD-1 can be combined with not only other ICIs but also other immunologic agents, such as TGFβ blockers and cytokines. Although further study is needed to find modalities to turn “low responding TILs” into “highly responding TILs”, this study may provide the rationale and evidence for establishing the optimal strategies for combinational ICI treatment in patients with GBM.

No potential conflicts of interest were disclosed.

Conception and design: J. Park, M. Kwon, S.-G. Kang, S.-H. Park, E.-C. Shin

Development of methodology: J. Park, M. Kwon, K.H. Kim, S.H. Hong, J.H. Chang

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): J. Park, M. Kwon, K.H. Kim, T.-S. Kim, S.H. Hong, S.-G. Kang, J.H. Moon, E.H. Kim, J.H. Chang

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): J. Park, K.H. Kim, S.H. Hong, C.G. Kim

Writing, review, and/or revision of the manuscript: J. Park, K.H. Kim, C.G. Kim, S.-H. Park, J.H. Chang, E.-C. Shin

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): J. Park, S.H. Hong, C.G. Kim, J.H. Chang

Study supervision: S.-H. Park, J.H. Chang, E.-C. Shin

This work was supported by the National Research Foundation Grants (NRF-2017R1A2A1A17069782 and NRF-2018M3A9D3079498).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.