Abstract
Purpose: Carcinoembryonic antigen (CEA) is a candidate target for cellular immunotherapy of pancreatic cancer. In this study, we have characterized the antigen-specific function of autologous cytotoxic T lymphocytes (CTL) specific for the HLA-A2–restricted peptide, pCEA691-699, isolated from the peripheral T-cell repertoire of pancreatic cancer patients and sought to determine if ex vivo PD-L1 and TIM-3 blockade could enhance CTL function.
Experimental Design: CD8+ T-cell lines were generated from peripheral blood mononuclear cells of 18 HLA-A2+ patients with pancreatic cancer and from 15 healthy controls. In vitro peptide-specific responses were evaluated by flow cytometry after staining for intracellular cytokine production and carboxy fluorescein succinimydyl ester cytotoxicity assays using pancreatic cancer cell lines as targets.
Results: Cytokine-secreting functional CEA691-specific CTL lines were successfully generated from 10 of 18 pancreatic cancer patients, with two CTL lines able to recognize and kill both CEA691 peptide–loaded T2 cells and CEA+ HLA-A2+ pancreatic cancer cell lines. In the presence of ex vivo PD-L1 blockade, functional CEA691-specific CD8+ T-cell responses, including IFNγ secretion and proliferation, were enhanced, and this effect was more pronounced on Ag-specific T cells isolated from tumor draining lymph nodes.
Conclusions: These data demonstrate that CEA691-specific CTL can be readily expanded from the self-restricted T-cell repertoire of pancreatic cancer patients and that their function can be enhanced by PD-L1 blockade. Clin Cancer Res; 23(20); 6178–89. ©2017 AACR.
Translational Relevance
This article describes the isolation of functional CEA691-specific CD8+ T cells from the peripheral blood and/or draining lymph nodes of 18 consecutive patients with carcinoma of the pancreas. We demonstrate for the first time that the antigen-specific function of these self-restricted tumor-reactive T cells can be enhanced by PD1/PDL1 pathway blockade. These findings support the clinical development of T-cell–mediated therapies in combination with checkpoint inhibitors for patients with an extremely poor prognosis malignancy.
Introduction
Pancreatic cancer remains a highly aggressive and difficult-to-treat malignancy. As such, it is one of the leading causes of cancer deaths worldwide (1). At the time of diagnosis, approximately 30% of patients have locally advanced disease and a further 50% already have evidence of metastatic disease. A minority of patients (approximately 15%) are eligible for potentially curative surgery (such as pancreatico-duodenectomy, or the Whipple procedure), but the majority will die from recurrent disease (2, 3). Current 5-year overall survival (OS) rates are in the order of 5% with only marginal improvements in prognosis having been achieved in the last few decades (4–6).
For patients with unresectable pancreatic cancer, gemcitabine- and folfirinox-based chemotherapeutic regimens are the treatment of choice; however, efficacy is limited, and drug resistance and disease relapse are very common (7–9). There is therefore an urgent need to develop alternative therapeutic approaches for the treatment of pancreatic cancer.
Targeted immunotherapeutic approaches aim to improve the therapeutic index by maximizing tumor cell death while minimizing side effects. Promising results have recently been demonstrated for various advanced cancers, including melanoma, metastatic non–small cell lung cancer, renal cell cancer, and ovarian cancer (10) using a number of strategies including monoclonal antibodies, checkpoint inhibitors, vaccination, and genetically modified immune cells (11–13). Cytotoxic T lymphocytes (CTL) play a pivotal role in cellular immunity and, by recognizing tumor-associated antigen (TAA)–derived short peptide epitopes presented by MHC-class I molecules, they can mediate target cell killing and induce protective antitumor immune responses (14).
Although a number of TAAs have been identified in the context of pancreatic cancer, including mesothelin (15) and the carcinoembryonic antigen (CEA; ref. 16), both of which are widely expressed in pancreatic cancer cells, pancreatic tumors are generally considered to be nonimmunogenic and resistant to immunotherapies (10, 17, 18). In addition, pancreatic cancer cells actively contribute to local immune suppression in the tumor microenvironment through the production of anti-inflammatory cytokines such as TGFβ, IL10, and IL6 and/or the expression of negative regulatory molecules (18). Specifically, pancreatic cancer cells express high levels of programmed death ligand-1 (PD-L1; ref. 19), an immunosuppressive molecule that, upon engagement with its receptor programmed death-1 (PD-1) on the surface of CTLs, delivers inhibitory signals impairing T-cell effector function (20). In murine models, it has been shown that anti–PD-L1 and anti–CTLA-4 blocking antibodies, commonly referred to as checkpoint inhibitors, improved T-cell–mediated antitumor immunity and prolonged survival (21–23). Although supporting evidence from human clinical trials is currently lacking (24), it is expected that T-cell–based immunotherapies in combination with checkpoint inhibitors may circumvent the immunosuppressive properties of the pancreatic tumor microenvironment, and such combinatorial approaches are likely to be required.
In this study, we examined the ex vivo functional and phenotypic properties of CEA-specific T cells isolated from 18 consecutive HLA-A2+ pancreatic cancer patients.
Materials and Methods
Patients and samples
This study was approved by the Central London Research Ethics Committee (Study no. 06/Q0512/106) and conducted in accordance with the Declaration of Helsinki. Written, informed consent was obtained from all patients.
Peripheral blood samples were collected from pancreatic cancer patients at three central hospitals: University College London Hospitals NHS Foundation Trust (UCLH); Royal Free London Hospital NHS Foundation Trust (RFH); and Charing Cross Hospital - Imperial College Healthcare NHS Foundation Trust. Detailed patient demographics and tumor characteristics are summarized in Table 1. In all cases, the diagnosis of pancreatic carcinoma was confirmed by standard cytopathology or histopathology after biopsy, and the clinical stage was assigned using staging criteria described in the World Health Organization histologic classification of tumors of the exocrine pancreas (25). Anonymized peripheral blood mononuclear cells (PBMC) were obtained from the National Blood Service from healthy controls.
Patient demographic information (PBMC samples, N = 18; LN samples, N = 3)
PBMCs were isolated by density gradient centrifugation using standard methodology (Ficoll, Lymphoprep-Apogent Discoveries). Where possible, lymph node (LN) samples were collected from patients undergoing surgery and then mechanically disrupted to generate a single-cell suspension prior to PBMC isolation.
Patient HLA-A2 status was determined by flow cytometric analysis after staining with a phycoerythrin (PE)-conjugated, anti–HLA-A2 antibody (clone BB7.2; BD BioSciences).
Peptides and tetramers
The HLA-A2–restricted peptides CEA691 (IMIGVLVGV), CMV pp65 (NLVPMVATV), and Telomerase540 (ILAKFLHWL) were synthesized by Mimotopes Pty Limited and dissolved in PBS at a stock concentration of 2 μmol/L prior to use. APC-labeled pCEA691/HLA-A*0201 tetramers (TCMetrix) were used to detect CEA-specific T cells. Other HLA-A2–restricted peptides used in this study are detailed in Supplementary Table S1 and were also synthesized by Mimotopes Pty Ltd.
Short-term T-cell expansion cultures
Short-term primary T-cell lines were generated as previously described (26). Briefly, isolated PBMCs were cultured at 1.5 × 106 cells/mL in normal growth medium (NGM) consisting of RPMI 1640 (Invitrogen) supplemented with 2 mmol/L glutamine, 1% penicillin/streptomycin (Sigma-Aldrich), and 10% heat-inactivated FCS (BioWest). Peptide stimulation was performed at a final concentration of 2 μmol/L, in the presence of rhIL2 (20 U/mL), and the cells were harvested after 9 to 10 days of culture.
Establishment of primary T-cell lines
PBMCs were expanded over four rounds of peptide-specific stimulation, with cells being analyzed by flow cytometry at the end of rounds one and four. Briefly, 3 × 106 PBMCs were initially resuspended in NGM at 1.5 × 106 cells/mL in a 24-well plate, in the presence of rhIL2 (20 U/mL; Roche), rhIL7 (2 ng/mL; R&D Systems), rhIL15 (5 ng/mL; R&D Systems), and rmIL21 (0.5 ng/mL; R&D Systems). Peptide-specific stimulation was performed by adding pCEA691 or CMV pp65 directly into specific wells, at a final concentration of 10 μmol/L. After a first round of 7 to 9 days of duration, cells were moved to restimulation. Three rounds of restimulation were performed as described below, each of which was 7- to 9-day duration.
Restimulation was conducted by resuspending 5 × 105 cells in 2 mL of cytokine-supplemented NGM in a new 24-well plate, and stimulating them with irradiated (70 Gy) T2 cells (2 × 105) pulsed with the appropriate peptide, at a final concentration of 10 μmol/L. Irradiated (35 Gy) autologous PBMCs or PBMCs from healthy HLA-A2+ donors obtained from the National Blood Service were used as feeder cells (2 × 106).
Cell lines
Six pancreatic cancer cell lines (MiaPaca-2, PK-45, Panc-1, KLM-1, Bx-Pc-3, and PK-1) were obtained from PIKEN BioResourcecentre (PIKEN BRC). FITC-CEA antibody (clone B1.1; BD Sciences) was used to detect CEA expression on the surface of these cell lines, which was analyzed by flow cytometry. Among them, Panc-1 (HLA-A2+, CEA+), MiaPaca2 (HLA-A2−, CEA−), PK-1 (HLA-A2+, CEA+++), and Bx-Pc-3 (HLA-A2+, CEA+++) were used as target cells in cytotoxicity assays. Apart from MiaPaca2, all cell lines were maintained in NGM containing RPMI 1640 (Invitrogen) supplemented with 2 mmol/L glutamine, 1% penicillin plus streptomycin (Sigma-Aldrich), and 10% heat-inactivated FCS (BioWest). MiaPaca2 cells were cultured in DMEM (Invitrogen) with 1% penicillin plus streptomycin and 10% heat-inactivated FCS. The HLA-A2–positive T2 cell line was loaded with specific peptides and used as target cells were indicated. TAP-deficient T2 cells have impaired presentation of HLA molecules with endogenous peptide, but can be efficiently loaded with exogenously peptides (27). The T2 cell line was maintained in NGM.
Cytotoxicity assays
A carboxy fluorescein succinimydyl ester (CFSE) cytotoxicity assay was used to determine the antigen-specific cytotoxicity of expanded T-cell lines. pCEA691 peptide–loaded T2 cells or HLA-A2–positive pancreatic cancer cell lines (known to express CEA, CEA+) were used as target cells. T2 cells pulsed with irrelevant peptides and HLA-A2–negative pancreatic cancer cell lines were used as control target cells. Target cells (1 × 106) were suspended in PBS/1%FCS at a concentration of 106/mL and then stained with CFSE. For sensitive targets, 0.5 μL of CFSE stock solution (5 mol/L) was added to 1 mL of cell suspension, whereas for control targets, 0.5 μL of diluted CFSE at 500 μmol/L was used. After 4 minutes' incubation at room temperature, 9 mL of PBS/1%FCS was added to stop the reaction. The cells were washed with PBS, and then resuspended at 5 × 104 cells/mL prior to setting up cocultures with the effector cells. T cells (effectors) were added to round-bottomed 96-well plates to obtain a total volume of 200 μL/well (28). Various effector:target (E:T) ratios were tested, including 100:1, 50:1, 20:1, 10:1, 5:1, 2.5:1, 1.25:1, and 0.625:1, respectively. Assay plates were incubated for 4 hours at 37°C, 5% CO2. Cells were washed in PBS prior to FACS analysis (FACSCalibur). For peptide titration assays, CFSE-stained T2 cells were loaded with variable concentrations of peptides, at 10−5, 10−6, 10−7, 10−8, 10-9, 10−10, 10−11, and 10−12 mol/L, respectively (29).
In vitro PD-L1 and TIM-3 blockade
PBMC from 11 pancreatic cancer patients (where sufficient cells were available) and LN-derived lymphocytes from 1 pancreatic cancer patient were cultured in the presence of CEA691 peptides and rhIL2 as described above, at a concentration of 2 × 106/mL in 200 μL of NGM. On day one, anti–PD-L1 and anti–TIM-3 antibodies (Mouse IgG, eBioscience) were added to the wells, either separately, or in combination, at a final concentration of 10 μg/mL. After 7 days of incubation at 37°C, the cells were harvested for functional analysis using intracellular cytokine staining.
Flow cytometry
The following antibody–fluorochrome combinations were used: CD3-PE-Cy7, CD8-Horizon v450, CD4-Horizon v500, IFNγ–FITC, PD-1–PE, CD45RO-BV650 (all from BD Biosciences); CD62L-APC-Cy7 (eBioscience); LAG-3–FITC (R&D Systems); and TIM-3-AF700 (eBioscience). Ex vivo surface staining was performed on 1 × 106 freshly isolated PBMC. Briefly, one microliter of a 1:50 dilution of each antibody was added to the cells and incubated for 30 minutes, at 4°C, in the dark. Cells were washed twice with PBS/1% FCS and then resuspended in 200 μL PBS/1% FCS for data acquisition. Flow cytometry data acquisition was performed using a FACSCalibur. Propidium iodide (10 μg/mL) was added immediately prior to acquisition to discriminate dead cells from viable cells. Data analysis was performed using FlowJo (Treestar Inc.; version v10).
Intracellular cytokine staining assay
Intracellular cytokine staining was performed on cultured cells, either after short-term stimulation or after four rounds of antigen-specific stimulation during the primary T-cell line establishment protocol (described above). Upon harvest, cultured cells were restimulated with 10 μmol/L of relevant peptide for a further 5 hours in the presence of 10 μg/mL Brefeldin A. Cells stimulated with an irrelevant peptide were used as negative controls, and cells stimulated with PMA (50 ng/mL) + Ionomycin (500 ng/mL) were used as positive controls. Cells were surface stained with anti-CD3, anti-CD4, and anti-CD8 antibodies, as described above, then permeabilized and fixed using FACS fix/perm solution (Invitrogen) prior to staining for intracellular cytokines with FITC-conjugated anti-IFNγ, for 20 minutes, at 4°C in the dark. Cells were washed with PBS/1% FCS and then resuspended in 200 μL PBS/1% FCS.
An immunological “response/responder” was defined as a 2-fold increase in the frequency of cytokine-producing cells in relation to that observed with the irrelevant peptide (Telomerase 540). For example, if the frequency of IFNγ-producing CD8+ T cells induced by CEA691 doubled that stimulated by control peptide at the end of 4 rounds of pCEA691-specific stimulation, the response was defined as positive (i.e., a “responder”).
Statistical analysis
Statistical analysis was conducted using the SPSS software (SPSS for windows, version 21). Data sets were first tested for parametric distribution using the Skewness–Kurtosis and the homogeneity of variance tests. For parametric data, the T test was used to determine statistical significance; for nonparametric data distributions, the Mann–Whitney U test was applied. When comparing data sets between more than two groups, the one-way ANOVA test (for parametric data) or Kruskal–Wallis test (for nonparametric data) was used. Whenever an overall P value was statistically significant, post hoc pairwise comparisons were performed with the Tukey Honest Significant Difference method. P values < 0.05 were considered statistically significant. For survival analyses, the Mann–Whitney U test was applied.
Results
CEA691-specific CTL responses can be generated from PBMCs of pancreatic cancer patients
We previously investigated antigen-specific CTL responses generated by short-term culture of PBMCs isolated from 13 pancreatic cancer patients (Supplementary Table S1) and 10 healthy controls. These were stimulated with 3 CEA HLA-A2–restricted peptides (CEA605-613 YLSGANLNL, CEA691-699 IMIGVLVGV, and CEA694-702 GVLVGVALI) and one irrelevant peptide Telomerase540-548 ILAKFLHWL over 9 to 10 days. Antigen-specific IFNγ production was most frequently observed in CEA691-responsive CD8+ T cells, with a higher percentage of both patients and healthy controls generating functional responses than to other tested epitopes (Supplementary Fig. S1A). In responders, the frequency of IFNγ-producing cells within the CD3+CD8+ T-cell subset was not significantly higher in pancreatic cancer patients than in healthy controls (Supplementary Fig. S1B and S1C).These data were in agreement with previous findings indicating that immune tolerance to particular self-peptides may be incomplete (30).
Here, we demonstrate in Fig. 1A the gating strategy for analysis of PBMCs from a single pancreatic cancer patient (CA11) after four rounds of expansion in vitro. An increase in the percentage of antigen-specific IFNγ-producing CD8+ T cells was observed. After the 4th round of peptide stimulation, IFNγ-producing CD8+ T cells were detectable in 5 of 15 healthy controls and 10 of 18 pancreatic cancer patients, with 4 examples of each shown in Fig. 1B (healthy controls H01, H04, H07, H13) and C (cancer patients CA07, CA11, CA12, CA18), respectively.
CEA691-specific T cells isolated from pancreatic cancer patients and healthy controls produce type 1 cytokines. A, A typical example of T cells expanded during stimulation with pCEA691-loaded T2 cells. Between the second (2R) and fourth rounds (4R) of stimulation, the percentage of INFγ−producing CD8+ T cells gradually increased. By the end of the 4th round stimulation, 77.9% of CD8+ T cells were INFγ−secreting, compared with 2% after 2 rounds. Dot blots representing IFNγ and TNFα−producing CD8 T cells cultured with CEA691-pulsed T2 cells and control peptide–loaded T2 cells after 4 rounds of stimulation (Gated on CD8 T cells) in 4 of 15 healthy controls (B) and 4 of 18 cancer patients (C) are shown. The percentages of IFNγ+TNFα+ CD8 T cells are shown in the upper-right quadrants.
Pancreatic cancer disease progression is associated with impaired CEA691-specific CTL responses
Six patients recruited for this study were classified as having stage IV pancreatic cancer (metastatic disease at presentation), whereas the rest were classified stages II (N = 10) and III (N = 2) pancreatic cancer (Fig. 2A). CEA691-specific CTL responses were detected in 9 of 12 patients with stage II–III disease, whereas only 1 of 6 patients with stage IV disease had demonstrable functional CEA691-specific T-cell responses as determined by IFNγ secretion upon four-round antigen-specific stimulation (Fig. 2B). Relative frequencies of CEA691-specific CD8+ T cells were also significantly lower than that in patients with earlier stage disease II–III (Fig. 2B). Pancreatic cancer patients who had been treated with chemotherapy (Fig. 2C) and patients with inoperable tumors (Fig. 2D) also had lower frequencies of CEA691-specific CTLs compared with the nonchemo and surgical patient groups, respectively.
Higher frequencies of CEA691-specific CD8+ T-cell responses are detected in earlier stages of pancreatic cancer. A, Pie chart illustrates the frequency of pancreatic cancer patients at different disease stages (II–IV). Frequencies of IFNγ+ cells within the CD8+ T-cell population are shown for patients stratified according to (B) disease stage, (C) prior administration of chemotherapy, and (D) submission to surgery. Each symbol represents one individual, and horizontal bars represent median. P values < 0.5 were considered statistically significant and are shown in the graphs.
No statistically significant difference in median OS from date of recruitment to the study was observed between patients who generated functional CD8+ T-cell responses to CEA-691-699 and those with no responses. Median OS was 20.5 months (range, 1–31 months) for responders versus 14 months (range, 1–24 months) for nonresponders, P = 0.079 (Table 1). However, the median OS of the 3 patients with more than 20% of functional CD8+ T cells at the end of stimulation (CA07, CA11, and CA18) was significantly longer than the other patients (28 vs. 14 months, P = 0.004). When only patients treated with surgical resection were analyzed (n = 10), those with functional CEA-specific CTL had a significantly longer median OS of 38.5 months (23–59 months, n = 6) compared with 22 months (6–31 months, n = 4), P = 0.039.
CEA691-specific CTL from pancreatic cancer patients can recognize and kill pancreatic cancer cell lines in vitro
T-cell lines were generated by stimulating the PBMCs of pancreatic cancer patients using pCEA691-loaded T2 cells for four rounds. As described above, over 30% of CD8+ T cells from 3 patients (CA07, CA11, and CA18) produced cytokines in response to pCEA691 restimulation after four rounds of expansion (Fig. 1C). To examine their ability to recognize and induce tumor cell death, we utilized CFSE killing assays from T-cell lines generated from CA07, CA 11, and CA18.
T2 cells loaded with 1 mmol/L (10−6 mol/L) pCEA691 peptide (CFSEhi) or irrelevant peptide-loaded T2 cells (CFSElo) were cocultured with CTL at different E:T ratios ranging from 100:1 to 0.6:1. Unstimulated HLA-A2+ PBMCs were used as control effector cells (Supplementary Fig. S2A). The results demonstrate that CEA691-specific CTLs generated from all these three patients (CA07, 11, and 18) were able to recognize and kill CEA691-loaded T2 cells (Supplementary Fig. S2B). Moreover, T cells from CA11 were able to kill T2 cells loaded with 10−9 mol/L CEA691 (1 nmol/L), at an E:T ratio = 20:1, in peptide titration assay, in which T cell lines were stimulated with T2 cells loaded with 10−5 to 10−12 mol/L (10 mmol/L to 1 pmol/L) CEA691 peptide for 4 hours (E:T = 20:1; Supplementary Fig. S2C and S2D). The T-cell line generated from CA11 was shown to have moderate-to-high avidity with recognition of nanomolar (10−9 mol/L) concentrations of CEA, which are comparable with the concentration of TAA expression on the surface of tumor cells (31).
Subsequently, we investigated the cytotoxic effector function of T-cell lines following stimulation with pancreatic cancer cell lines in vitro. The expression of CEA and HLA-A2 in six pancreatic cancer cell lines was determined after staining with specific monoclonal antibodies and analyzed by flow cytometry. MiaPaca-2 was negative for both CEA and HLA-A2; the Panc-1 cell line was HLA-A2 positive, but expressed low levels of CEA; Bx-Pc-3 and PK-1 were HLA-A2–positive cell lines with high expression of CEA (Fig. 3A). Therefore, MiaPaca-2 was used as the negative control for both HLA-A2 and CEA, and PK-1, Bx-Bc-3 (HLA-A2+ and CEA+), together with Panc-1 (HLA-A2+ and CEA low) were used as specific targets. Figure 3B shows the relative reduction in the frequencies of PK-1 after coculture with the CTL line (from patient CA11) and MiaPaca-2 cells, at different E:T ratios. T-cell lines generated from CA07 and CA11 patients lysed PK-1 and Bx-Pc-3 cancer cell lines, but not Panc-1, demonstrating their CEA691-specific cytotoxicity (Fig. 3C). No cytotoxicity against pancreatic cancer cell lines was observed using unstimulated PBMCs from the same donors (data not shown). These data suggest that functional CEA-specific CTLs can be generated in pancreatic cancer patients and that these cells may have the potential to control tumor growth.
Cytotoxic activity of CTL lines against pancreatic cancer cell lines. A, CEA and HLA-A2 expression of six pancreatic cancer cell lines. The results were analyzed by FACS and presented in histogram (top). Isotype antibodies were used to determine the background. Percentage and MFI of HLA-A2/CEA expressions by different pancreatic cancer cell lines was also shown (bottom). N.B. The unstained cells were gated out. B, FACS analysis of CA11 T-cell killing activity in response to recognize and kill PK-1 cell line (HLA-A2+, CEA+, labeled with high dose CFSE), and MiaPaca-2 cell line (HLA-A2−, CEA−, labeled with low-dose CFSE), at different E:T ratios. C, The percentage of relative killing of PK-1, Panc-1, and Bx-Pc-3 by CTLs from CA07 or CA11, compared with MiaPaca-2, at different E:T ratios. All the experiments were repeated twice, and the mean of the results was shown in the figure.
Inhibitory pathways play a role in the modulation of antigen-specific CTL responses in pancreatic cancer patients
Possible explanations for the failure of antitumor T-cell responses in patients include inadequate T-cell priming and insufficient duration of the effector phrase, both of which can be regulated by the coinhibitory receptors, including CTLA-4, PD-1, TIM-3, and LAG-3 (32). For example, PD-1/PD-L1 signaling inhibits T-cell function via suppressing TCR-dependent activation of both CD4+ and CD8+ T cells, particularly, through the inhibition of their proliferation and cytokine production, including IFNγ, TNFα, and IL2 (33). PD-L1 expression by tumor cells has been associated with a poorer prognosis or advanced disease (20, 34–36). About 80% of pancreatic cancer cases express PD-L1, of which 20% have upregulated expression of PD-L1 (compared with normal pancreatic tissue) and tend to be highly invasive and recurrent (37, 38).
In order to investigate the relationship between CEA691-specific CTL responses and the expression of negative regulatory molecules on T cells, ex vivo phenotypic examination of pancreatic cancer patients' peripheral T cells was performed prior to stimulation with CEA691-peptide. As shown in Fig. 4A, the frequency of PD-1, TIM-3, and LAG-3 expressing CD8+ T cells was significantly higher in the nonresponder group of pancreatic cancer patients than in the healthy controls.
Expression of negative regulatory molecules in pancreatic cancer. PBMCs were isolated from pancreatic cancer patients and healthy controls, and surface stained to assess (A) the expression of PD1, TIM-3, and LAG-3 molecules; as well as (B) the relative proportions of naïve/memory subsets (based on the expression of CD45RO and CD62L), within the CD8+ T-cell population. Pancreatic cancer patients were stratified according to their ability to mount CEA691-specific CD8+ T-cell responses (responders vs. nonresponders). Each symbol represents one individual, and horizontal bars represent mean. P values < 0.5 were considered statistically significant and are shown in the graphs. C, T cells were obtained from matching samples of peripheral blood (closed circles) and LNs (open circles) from 3 pancreatic cancer patients. Graphs illustrate the levels of negative regulatory molecule expression within the CD8+ population, both in terms of frequency and number of molecules per cell, expressed as MFI. Each symbol represents one individual, and horizontal bars represent mean. P values < 0.5 were considered statistically significant and are shown in the graph.
It is known that T-cell activation and differentiation status influences the expression of some cell surface markers, such as PD-1. In order to examine the T-cell phenotype in more detail, we identified naïve, central memory (TCM), effector memory (TEM), and end-stage/effector phenotypes based on CD62L and CD45RO expression (i.e., TCM was defined as CD62L+CD45RO+ T cells, TEM as CD62L−CD45RO+T cells, naïve T cells as CD62L+CD45RO−, and end-stage/effector T cells as CD62L−CD45RO+). As shown in Fig. 4B, there was no statistically significant difference in the percentage of gated CD4+ or CD8+ T cells at various differentiation stages between pancreatic cancer responders and nonresponders. The percentage of naïve T cells, however, in responders was observed to be higher than in nonresponders, but this difference did not achieve statistically difference (P = 0.051).
CD8+ T-cell priming and activation take place in draining LNs where, upon interaction with antigen-presenting cells, naïve T cells are primed and differentiate into fully functional antigen-specific CTLs. We were able to analyze the expression of PD-1, TIM-3, and LAG-3 in matched peripheral- and LN-derived CD8+ T cells isolated from 3 of the pancreatic cancer patients (see Table 1). The proportion of LN-derived CD8+ T cells expressing PD-1 and TIM-3 was higher than that observed in peripheral T cells, whereas more LAG-3–positive cells were identified within the peripheral T-cell repertoire than in the draining LNs (Fig. 4C, left). In addition, similar differences were observed when analyzing the expression levels of PD-1, TIM-3, and LAG-3, as determined by mean fluorescence intensity, MFI (Fig. 4C, right).
Blockade of the PD-1/PD-L1 pathway improves tumor antigen-specific CD8+ T-cell responses
To assess whether blockade of the PD-1/PD-L1–negative regulatory pathway enhanced the function of patient-derived, self-restricted CEA691-specific CTL, PBMCs from patients with pancreatic cancer (and LN cells from one patient) were cultured for 7 days in the presence of pCEA691 or the control peptide, rIL2, with or without anti–PD-L1 and/or anti–TIM-3 antibodies. In total, the frequency of peptide-specific cytokine-producing CD8+ T cells was evaluated in 11 patients with pancreatic cancer (see Table 2 for patient details). In addition, the frequency of CEA691 tetramer–positive CD8+ T cells was determined in 8 of the 11 patients (where sufficient samples were available for analysis).
As shown in Table 1, a significant increase in the frequency of CEA691 tetramer–binding CD8+ T cells was observed in cells cultured in the presence of anti–PD-L1 antibody (P = 0.030, n = 8) and the combination of anti–PD-L1+anti–TIM-3 antibodies (P = 0.045, n = 8), compared with nontreated cells. These findings suggest that PD-L1 and TIM-3 blockade can enhance the proliferation of pancreatic cancer patient–derived CEA691-specific CTLs in vitro. Similarly, the frequency of IFNγ-producing CD8+ T cells was significantly higher in the cells treated with anti–PD-L1 or anti–TIM-3, alone or in combination (Fig. 5A–C). Even though both PD-1/PD-L1 and TIM-3–regulatory pathways were observed to be involved in the modulation of CTL function, our data did not reveal a significant synergistic effect when PD-1/PD-L1 and TIM-3 pathways were blocked at the same time.
PD-1/PD-L1 and TIM-3 blockade in pancreatic cancer. A, The percentage of CEA tetramer binding CD8+ T cells after 7 days of culture with CEA691 in the presence or absence of anti–PD-L1 and/or anti–TIM-3 blocking antibodies. B, CEA691 CD8+ T cells from 11 pancreatic cancer patients were expanded for 7 days with or without anti–PD-L1 and/or anti–TIM-3 blocking antibodies, and intracellular stained to assess the levels of IFNγ production. The graph shows frequencies of IFNγ+ cells within the CD8+ T-cell population. Each symbol represents one individual, and horizontal bars represent mean. P values < 0.5 were considered statistically significant and are shown in the graph. PBMCs and MNCs isolated from LNs of (CA13) were also stimulated for 7 days by pCEA691 with or without anti–PD-L1 and/or anti–TIM-3 blocking antibodies, and assessed for levels of IFNγ production. A representative example of IFNγ production is shown in the dot plots (C). D, The graph shows the frequency of IFNγ+ cells within the CD8+ T-cell population. Experiments are duplicated, and symbols indicate mean and SD.
To determine the relative function of negative regulatory pathways in the inhibition of antigen-specific CTL responses in the LNs of patients with pancreatic cancer, MNCs isolated from LNs were cultured with the pCEA691 peptide in the presence or absence of anti–PD-L1 and/or anti–TIM-3 blocking antibodies. Blockade of the PD-1/PD-L1 pathway, but not of the TIM-3 pathway, led to an increase in the frequency of IFNγ-producing cells in both circulating and LN-derived CTLs, suggesting a role for the PD-1/PD-L1 axis in the modulation of IFNγ responses in patients with pancreatic cancer (Fig. 5C and D). We observed that the expression of PD-1 and TIM-3 was higher in LN-derived CD8+ T cells than that in peripheral T cells, and blockade of these inhibitory molecules led to a more significant recovery of peptide-specific IFNγ production by CD8+ T cells.
Our findings suggest that the expression level of negative regulatory molecules, in circulating and LN-derived CTLs, may influence the magnitude of the CTL-mediated antitumor response in different tissues.
Discussion
The prognosis for patients with pancreatic cancer remains extremely poor, and novel treatments such as immunotherapy are appealing alternatives to improve survival. However, there is currently little evidence from large clinical trials to demonstrate that pancreatic cancer is sensitive to checkpoint inhibition, and vaccine-based immunotherapies have largely failed to generate significant clinical results (10, 17, 18, 24). Meanwhile, some animal experiments have suggested that the combination of checkpoint inhibitors with other immunotherapies, such as therapeutic vaccines, may improve outcomes in pancreatic cancer (24). Our study has focused on exploring potential epitopes for targeted immunotherapy in pancreatic cancer patients and has evaluated the influence of blockade of PD-1/PD-L1 pathway or TIM-3 pathway in antitumor T-cell responses.
CEA is an 180 kDa glycosylated membrane protein that is overexpressed in more than 90% of pancreatic cancer cases (16). As a self-antigen, CEA-specific T cells are expected to be subject to immunological tolerance. Although previous studies demonstrated that tolerance to CEA691 is incomplete (30), little is known about the function of CEA691-specific CTL isolated from patients with pancreatic cancer. Here, we identified the most immunogenic CEA-derived epitopes using short-term in vitro PBMC stimulation followed by prolonged antigen stimulation to expand primary CEA691-specific T-cell lines. After one round of in vitro peptide stimulation, we observed that the CEA691 peptide–specific IFNγ-producing T-cell responses were more readily detectable in pancreatic cancer patients compared with healthy controls and that responding patients had higher percentages of IFNγ-producing T cells than those stimulated by other peptide epitopes. After a total of four rounds of in vitro stimulation, the CEA691-specific CTLs from pancreatic cancer patients were also able to produce TNFα and kill relevant target cells.
Despite the small sample size, we were able to demonstrate that T cells isolated from patients presenting with more advanced disease were less likely to generate effective antigen-specific cytokine-producing responses. It is possible that prolonged TAA exposure in the tumor microenvironment may induce exhaustion of antigen-specific T-cell responses. Unfortunately, we were unable to test PBMCs at different time points over the course of pancreatic cancer progression for individual patients. Thus, the evolution of phenotypic and functional changes in the anti-CEA691 T-cell responses in our patients may be related to multiple factors including prior treatment, lymphopenia in advanced patients, comorbidities including age, and/or the use of immunosuppressive drugs. Similarly, pancreatic cancer patients without demonstrable CEA691-specific T-cell responses may have presented with more aggressive and/or quickly progressing cancers, hence having more advanced disease at diagnosis. To address these issues, a prospective longitudinal study with a larger sample size is needed.
Interestingly, our preliminary findings suggest that surgical treatment, which had occurred prior to patients being evaluated in the study (and therefore prior to T-cell isolation), may also modulate CD8+ T-cell responses, suggesting different modes of treatment received may also affect TAA-specific immune responses in pancreatic cancer. The interpretation of these data should take into account that this study was not designed to look at differences in clinical outcome and that the treatment modalities used were heterogeneous. In our previous study, the influence of medical interventions on T-cell response was observed demonstrating that embolization improved AFP-specific CD4+ T-cell responses in patients with hepatocellular carcinoma (39). Again, to confirm these findings, a much larger study would be required to control for all the relevant variables.
Activation of the inhibitory PD-1/PD-L1 pathway has been linked to the poor prognosis of pancreatic cancer (40). Our data have also shown that patients not responding to pCEA691 had larger numbers of PD-1–expressing T cells in their peripheral T-cell compartment, compared with the T cells isolated from responding patients or healthy controls. Also, T cells isolated from the tumor draining LNs were enriched for PD-1– and TIM-3–expressing cells compared with T cells isolated from the peripheral blood. Not surprisingly, these findings support the existence of an immunosuppressive tumor microenvironment in pancreatic cancer patients. It is therefore likely that combination therapies with checkpoint inhibitors and antigen-specific T cells may be required to optimize immunotherapeutic approaches to pancreatic cancer.
Recently, the efficacy of anti–CTLA-4 and anti–PD-1/PD-L1 antibodies in treating several types of cancers has been demonstrated in early-phase clinical trials (41). However, these checkpoint inhibitors are considered most effective in treating cancers with high mutational loads, e.g., melanoma and lung cancer, because such cancers typically generate more neoantigens and checkpoint inhibitors reverse the inhibition of tumor-infiltrating neoantigen-specific T cells (42). On average, the neoantigen repertoire of pancreatic cancer is 1 mutation per megabase, which is much less than that observed in melanoma and lung cancer (with more than 10 mutations per megabase on average; ref. 43). In theory, therefore, pancreatic cancer should not be very sensitive to checkpoint blockade. To date, there have been a number of clinical trials assessing the safety and efficacy of checkpoint inhibitors in pancreatic cancer patients (44), but no positive results have yet been published.
However, our in vitro results suggest that the PD-1/PD-L1 pathway may be an important factor in the inhibition of antitumor functions of CD8+ T cells from pancreatic cancer patients. Following the addition of PD-L1 blocking antibodies, antigen-specific T-cell proliferation and cytokine production (IFNγ and TNFα) were improved, with a greater effect observed on T cells isolated from the tumor draining LNs than from the peripheral blood. In addition, the percentage of T cells expressing PD-1 in LNs was higher than that observed in the peripheral blood. Thus, the limited impact to date of checkpoint inhibition in pancreatic cancer may be due to the relatively small neoantigen repertoire, and/or that TAA-specific T cells less readily enter the tumor microenvironment due to associated desmoplasia. Finally, we did not observe a synergistic effect when combining PD-L1 blockade with TIM-3 blockade. This may have been because TIM-3 levels were generally low in freshly isolated T cells. Little is known about the expression and role of TIM-3 (45, 46) and LAG-3 (47) in T cells from pancreatic cancer patients. In our study, the percentage of TIM-3– and LAG-3–positive CD4+ and CD8+ T cells was higher in pancreatic cancer patients, compared with healthy donors. Although the ex vivo culture of patient-derived effector T cells with PD-1 and Tim-3 blocking antibodies has experimental limitations, which cannot reflect the complexity of in vivo environment, we believe it provides useful preliminary data on which to base further studies.
In conclusion, here we describe the in vitro stimulation and expansion of self-restricted, autologous, functional CEA691-specific T cells, isolated from the peripheral blood and draining LNs of patients with pancreatic adenocarcinoma. This raises the possibility that expansion and infusion of autologous CEA691-specific T cells may be an effective immunotherapeutic approach to pancreatic cancer. Although alternative CEA-based immunotherapies are still being designed and optimized for pancreatic cancer (12, 48, 49), including CEA-specific engineered T cells, which have been shown to eradicate pancreatic cancer tumors without inducing toxicity in mice (50, 51), in another study, the administration of autologous T lymphocytes genetically engineered to express a murine TCR specific for CEA691 to colorectal cancer patients was shown to induce severe transient colitis (52) resulting in suspension of the clinical trial, raising questions about the suitability of this particular epitope for the design of novel TCR-based therapeutic approaches. Compared with the CEA691-specific high-affinity TCR generated by others (52–54), our CEA691-specific T cell lines were isolated from patients with pancreatic cancer, and it is possible that their antitumor efficacy can be improved by the in vivo reversal of dysfunction mediated by increased expression of PD-1 and TIM-3.
Disclosure of Potential Conflicts of Interest
E.C. Morris reports receiving commercial research grants from Cell Medica, and is a consultant/advisory board member for GE, GlaxoSmithKline, and UCL Technology Fund. No potential conflicts of interest were disclosed by the other authors.
Authors' Contributions
Conception and design: Y. Chen, S. Behboudi, E.C. Morris
Development of methodology: Y. Chen, S.-A. Xue, S.P. Pereira
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): Y. Chen, G.H. Mohammad, S.P. Pereira
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): Y. Chen, E.C. Morris
Writing, review, and/or revision of the manuscript: Y. Chen, S.-A. Xue, S. Behboudi, S.P. Pereira, E.C. Morris
Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): Y. Chen, G.H. Mohammad
Study supervision: S.P. Pereira, E.C. Morris
Grant Support
E.C. Morris was supported by the UCLH NIHR Biomedical Research Centre, the CRUK UCL Experimental Cancer Medicine Centre, Bloodwise and the Medical Research Council. S.P. Pereira was supported by NIH (Grant: PO1 CA084203) and by the National Institute for Health Research University College London Hospitals Biomedical Research Centre. S.-A. Xue was supported by the LLR program grant. S. Behboudi was supported by BBSRC grant (BB/N002598/1 and BBS/E/I/00001825).
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.
Acknowledgments
We thank Mr. Kito Fusai for performing LN resections and providing clinical care for surgically treated patients.
Footnotes
Note: Supplementary data for this article are available at Clinical Cancer Research Online (http://clincancerres.aacrjournals.org/).
S.P. Pereira and E.C. Morris are joint senior authors.
- Received April 25, 2017.
- Revision received June 20, 2017.
- Accepted July 10, 2017.
- ©2017 American Association for Cancer Research.