Correspondence to: Prof. Satoshi Kokura, Department of Molecular Gastroenterology and Hepatology, Graduate School of Medical Science, Kyoto Prefectural University of Medicine, Kyoto 602-8566, Japan, Tel.: +81-75-251-5519, Fax: +81-75-251-0710, E-mail: email@example.com
A core challenge in administering immune-based treatments for cancer is the establishment of easily accessible immunological assays that can predict patients' clinical responses to immunotherapy. In this study, our aim was to predict the therapeutic effects of adoptive T-cell therapy in patients with advanced pancreatic cancer. To do this, we evaluated whole blood cytokine levels and peripheral regulatory T cells (Tregs) in 46 patients with unresectable or recurrent pancreatic cancer who received adoptive T-cell therapy at 2-week intervals. To test immune function, venous blood was obtained from patients before the start of therapy and 2 weeks after the 4th treatment. Whole blood interferon (IFN)-α levels (after stimulation with the Sendai virus) were evaluated, as well as the levels of 9 cytokines stimulated with phytohemagglutinin [interleukin (IL)−2, IL-4, IL-5, IL-10, IL-12(p70), IL-13, tumor necrosis factor-α, IFN-γ, and granulocyte-monocyte colony-stimulating factor]. Peripheral Tregs were analyzed by flow cytometry. Using the obtained data, we then observed the relationship between these immunological parameters and clinical outcome of patients. We found that the whole blood production of IFN-γ, IL-2, IL-4, IL-5 and IL-13 significantly increased after adoptive T-cell therapy, whereas the number of peripheral Tregs did not change. Multivariate Cox proportional hazards analyses indicated that the number of peripheral Tregs before receiving adoptive T-cell therapy and the change in IFN-γ levels after adoptive T-cell therapy were independent variables predicting overall survival. The findings of this study indicate that the assay of whole blood IFN-γ production offers promise for evaluating the clinical response of patients to cancer immunotherapy.
Pancreatic adenocarcinoma is an aggressive malignancy that has a proclivity to lead to early metastasis and relatively poor outcome. The prognoses of patients with pancreatic adenocarcinoma are dismal, with 5-year survivals for all stages being less than 5%[1, 2]. These poor prognoses are partially due to the combination of late detection, (most patients are first diagnosed with locally advanced or metastatic disease), and the wide use of treatments that consist of relatively ineffective chemotherapeutic regimens. Gemcitabine is currently approved for treating patients with metastatic pancreatic cancer. For patients with metastatic disease, gemcitabine administration resulted in median survival being prolonged by only 1.24 months; a combination therapy with erlotinib and gemcitabine increased the median survival by a mere 2 weeks. On the basis of these outcomes, it is painstakingly clear that new treatment modalities for advanced pancreatic cancer are urgently required.
Preclinical and clinical studies of immune-based treatments including the use of peptide vaccines,[5-7] DNA vaccines,[8, 9] dendritic cells,[10, 11] monoclonal antibodies,[12, 13] adoptive T-cells,[14, 15] and chimeric antigen receptors[16, 17] have been investigated and are currently under study in pancreatic cancer patients. In immunotherapy studies, such as those using cancer vaccines, monitoring patients' immune response is important to determine if the intended immune response is being induced and to optimize the dose and schedule required in early-phase clinical trials. Moreover, in later-phase clinical trials, it is important to provide data so that clinical efficacy parameters and the types and magnitudes of immune responses can be compared. In antigen-specific immunotherapies, such as those involving cancer vaccines, antigen-specific immune monitoring methods, including the enzyme-linked immunospot (ELISPOT) assay or the human leukocyte antigen (HLA)-peptide multimer assay, are widely used. However, with few exceptions, there have been poor correlations between the magnitudes of the immunizing-antigen-specific T-cell responses that are measured by these immunological assays and the improvements in clinical outcome.[18, 19] When the tumor-rejection antigen is known, monitoring antigen specific responses is clearly important, but it is not enough to predict clinical response. Thus, the development of new assay systems that can predict clinical outcome more accurately is urgently needed in cancer immunotherapy.
The objective of this study was to evaluate whole blood cytokine levels and peripheral regulatory T cells (Tregs) to predict the therapeutic effects of adoptive T-cell therapy in patients with advanced pancreatic cancer.
Patients and Methods
Patients and blood sampling
Forty-six consecutive patients with unresectable or recurrent pancreatic cancer were treated at Hyakmanben Clinic (Kyoto, Japan) between October 2008 and February 2011. Of the 46 patients, 11 had locally advanced disease, 27 had metastatic disease, and the remaining 8 had recurrent disease. Patients received adoptive T-cell therapy at 2-week interval; 25 received antiCD3-stimulated lymphokine-activated killer (CD3-LAK) therapy, and 21 received RetroNectin® (CH296)-induced T-cell (RIT) therapy. The patients received adoptive T-cell therapy until they were unable to receive further treatment because of poor general conditions or if they refused any further treatment. For the testing of immune function, venous blood was obtained from the patients before receiving therapy (baseline) and during the follow up, which occurred after 4 cycles of adoptive T-cell therapy. The first follow-up blood sampling was conducted immediately before the 5th adoptive T-cell therapy (2 weeks after the 4th adoptive T-cell therapy, Fig.1).
This study was approved by the institutional review board and was conducted in accordance with the Declaration of Helsinki and the Ethical Guidelines for Clinical Research (the Ministry of Health, Labor and Welfare, Japan). Written informed consent was obtained from all patients.
Preparation of CD3-LAK cells and RIT cells
For the preparation of CD3-LAK cells, peripheral blood (30–40 mL) was taken from the patients. Mononuclear cells were separated and resuspended in a CultiLife215 bag (Takara Bio, Otsu, Japan) that was precoated with anti-CD3 (Janssen Pharmaceuticals, Tokyo, Japan). Cells were then cultured in serum-free media, GT-T505 (Takara Bio), which was supplemented with 1% heat-inactivated plasma and 1,000 U/mL of recombinant interleukin (IL)−2 (Nipro Corporation, Osaka, Japan). On days 3 and 5, GT-T505 media supplemented with 1% heat-inactivated plasma and 1,000 U/mL of recombinant IL-2 was added. On day 7, the cells were transferred to a CultiLife Eva bag (Takara Bio), and then GT-T505 media supplemented with 1% heat-inactivated plasma and 1,000 U/mL of recombinant IL-2 was added. On days 9 and 11, GT-T505 media supplemented with 1% heat-inactivated plasma and 200 U/mL of recombinant IL-2 was added. On days 13–14, the cells were harvested and resuspended in 100 mL of saline with 0.25% human albumin to create the final cell product.
For the preparation of the RIT cells, peripheral blood (30–40 mL) was taken from the patients. Mononuclear cells were separated and resuspended in a CultiLife215 bag that was precoated with both anti-CD3 and RetroNectin (CH296) (Takara Bio), which is a recombinant fragment of human fibronectin. Cells were then cultured in serum-free media, GT-T551 (Takara Bio), which was supplemented with 1% heat-inactivated plasma and 200 U/mL of recombinant IL-2. On day 4, the cells were transferred to a CultiLife Eva bag, and GT-T551 media that was supplemented with 1% heat-inactivated plasma and 200 U/mL of IL-2 was added. On day 7, GT-T551 media containing 200U/mL of IL-2 was added. On days 10 to 12, the cells were harvested and resuspended in 100 mL of saline containing 0.25% human albumin to create the final cell product. Prior studies have suggested that the RetroNectin® (CH296) stimulation method for ex vivo T-cell expansion is a useful basic technology for adoptive T-cell therapy as it can be expanded to create preferentially naive T cells that are likely to persist in vivo for long periods and also because they can accumulate in lymph nodes.
These cell products were assessed for their viability and to confirm the lack of contamination with bacteria, fungi, and/or endotoxins.
Whole blood cytokine assays
Methods for quantifying IFN-α production in whole human blood have been described previously. Briefly, heparinized peripheral blood was cultured with Sendai virus (500 HA/mL) within 8 h after the blood was withdrawn. The blood–virus mixture was incubated at 37°C for 20 h, and IFN-α activity in the supernatants was quantified by a bioassay. Other cytokines were measured according to procedures described previously.[22, 23] Heparinized whole blood was diluted fourfold with Eagle's minimal essential medium (Nissui Pharmaceutical Co, Tokyo, Japan) and stimulated with phytohemagglutinin (25 μg/mL). Samples were incubated at 37°C for 48 h after which the supernatants were harvested by centrifugation at 800g for 10 min, and they were then stored at −80°C until the analysis. Cytokine levels in the samples were measured with a multiplex cytokine array system (Bio-Plex; Bio-Rad Laboratories, Hercules, CA) according to the manufacturer's instructions. The Multiplex Th1/Th2 bead kit (Bio-Rad Laboratories) measured various cytokines [IL-2, IL-4, IL-5, IL-10, IL-12(p70), IL-13, Tumor Necrosis Factor (TNF)-α, IFN-γ and granulocyte–monocyte colony-stimulating factor (GM-CSF)]. Data acquisition and analysis were conducted with the Bio-Plex Manager Software, version 5.0.
Flow cytometric analysis
To determine the regulatory cell phenotype, flow cytometry analysis on whole blood was performed with the following antibodies: PE-Cy™5 mouse anti-human CD4; PE mouse anti-human CD25, Alexa Fluor® 488 mouse anti-human Foxp3, and Alexa Fluor® 488 mouse IgG1κ isotype control (all purchased from BD Biosciences Pharmingen, San Diego, CA). For whole blood staining, 500 μL of whole blood was incubated with appropriate amounts of fluorochrome-labeled antibodies in the dark at room temperature for 30 min. After lysing red blood cells with BD FACS™ lysing solution, cells were treated with human Foxp3 Buffer A and buffer B as described in the manufacturer's recommended procedures for the Human Foxp3 Buffer set (BD Biosciences Pharmingen). After cells were permeabilized, anti-Foxp3 was added for 30 min at room temperature and cells were washed once. Flow cytometry was performed with a Becton Dickinson FACS Calibur flow cytometer (BD, Oxford, UK), and CellQuest Software (BD) was used for the analysis. The phenotypes of the Tregs were defined as positive for CD4, CD25 and Foxp3.
A paired t-test was used to compare the results before and after treatment. Univariate and multivariate Cox proportional hazard regression models were used to assess the possible associations between the potential predictive factors and overall survival (OS). The relationships between OS and both cytokine levels and the number and proportion of peripheral Tregs at baseline were first examined with univariate analyses. The variables that were possibly associated with the outcomes of interest (p < 0.20 after univariate analysis) and those that markedly changed (p < 0.20) after treatment were tested with a multivariate Cox proportional hazard regression analysis. OS was calculated from the first day of adoptive T-cell therapy until the last follow-up examination. P values less than 0.05 were considered significant. All statistical analyses were performed with SPSS software (version 20) for Windows (IBM Corporation, Chicago, Illinois).
Patient characteristics are summarized in Table 1. The median age was 63 years (age range: 39–81). Of the 46 patients, 11 had locally advanced disease, 27 had metastatic disease, and the remaining 8 had recurrent disease. Thirty-five patients (76.1%) had an Eastern Cooperative Oncology Group (ECOG) performance score of 1 or less. Approximately 40% of the patients had already received chemotherapy, and all, except one, had received chemotherapy that was combined with adoptive T-cell therapy. Of the 46 patients, 25 received gemcitabine monotherapy, 6 received monotherapy for S-1(an oral fluoropyrimidine), and 14 received combination therapy with gemcitabine and S-1. The median frequency of adoptive T-cell therapy was 10 times (range: 4–38).
Table 1. Clinical characteristics of pancreatic cancer patients
Effect of adoptive T-cell therapy on whole blood cytokine levels and the number of peripheral Tregs
Many cytokines had levels that substantially increased after adoptive T-cell therapy (Table 2). Among them, IFN-γ, IL-2, IL-4, IL-5 and IL-13 levels significantly increased after adoptive T-cell therapy. IL-12 (p70) levels increased after adoptive T-cell therapy, albeit insignificantly (p = 0.128).
Table 2. Cytokines production in patients before and after ACT
Values are expressed as the mean ± SD.
The paired t test was used to determine statistical differences.
ACT, adoptive T-cell therapy.
6892 ± 5667
5995 ± 5517
1641 ± 2863
3096 ± 4488
963.9 ± 1046
1183 ± 1510
155.0 ± 172.1
281.4 ± 427.6
8.829 ± 11.60
19.29 ± 29.88
108.2 ± 238.7
273.6 ± 390.7
99.38 ± 118.0
110.5 ± 102.3
IL-12 (p70) (pg/ml)
6.151 ± 10.74
16.01 ± 41.53
443.5 ± 559.8
1046 ± 1574
17.67 ± 41.14
25.80 ± 47.91
As shown in Table 3, both the number and the proportion of Tregs before treatment did not change after adoptive T-cell therapy.
Table 3. Number and proportion of Treg in patients before and after ACT
Values are expressed as the mean ± SD.
The paired t test was used to determine statistical differences.
ACT, adoptive T-cell therapy.
Number of Treg (/mm3)
26.64 ± 15.92
24.92 ± 14.91
Treg / CD4
4.537 ± 2.044
4.536 ± 2.494
Predicting OS using whole blood cytokine levels and the number of peripheral Tregs at baseline
The median overall survival after adoptive T-cell therapy among the 46 patients was 256 days. On the basis of the univariate Cox proportional hazards analysis, we noted that high levels of TNF-α (p = 0.152) and IL-4 (p = 0.187), and low numbers of peripheral Tregs (p = 0.152) at baseline appeared to be linked to longer OS. However, production levels of other cytokines were not predictors of OS (Table 4).
Table 4. Univariate Cox proportional hazard regression to determine survival
Variables at baseline
Variables were divided into two groups based on the median value [low and high].
95% CI: 95% Confidence Interval.
Number of Treg (low)
IL-12 (p70) (high)
Predictors for OS in multivariate models
Because the changes in IFN-γ and IL-13 were highly correlated (r = 0.9004, p < 0.0001), IL-13was omitted from the candidate variables in the multivariate analyses to reduce multicollinearity. Multivariate Cox proportional hazards analyses using candidate variables (i.e., gender, age, tumor location, stage, ECOG performance status, number of previous chemotherapy lines, kind of adoptive T-cell therapy, number of peripheral Tregs at baseline, whole blood TNF-α and IL-4 levels at baseline and changes in IFN-γ, IL-2, IL-4, IL-5 and IL-12 after adoptive T-cell therapy) indicated that stage [Hazard Ratio (HR), 3.487; 95% Confidence Interval [CI], 1.145–10.615; p = 0.028), performance status (HR, 24.956; 95%CI, 2.099–296.73; p = 0.011), number of peripheral Tregs at baseline (HR, 0.321; 95%CI, 0.104–0.943; p = 0.039) and change in IFN-γ levels after adoptive T-cell therapy (HR, 0.105; 95%CI, 0.013–0.850; p = 0.035) were independent variables for OS (Table 5).
Table 5. Multivariate Cox proportional hazard regression to determine survival
In this study, we demonstrated that adoptive T-cell therapy had the strong potential to enhance the whole blood production of the Th1 cytokine IFN-γ in advanced pancreatic cancer patients in whom immunosuppression and immune escape mechanisms may be present. A multivariate Cox proportional hazards analysis identified the increase of whole blood IFN-γ levels after adoptive T-cell therapy as a predictive factor of the clinical response to adoptive T-cell therapy, and this factor was independent of IFN-γ levels at baseline or the performance status of the patients. Moreover, in agreement with previous studies,[25-28] the number of peripheral Tregs at baseline was an independent prognostic marker of OS.
It is generally accepted that CD-8 T-cell responses are an important indicator of the effectiveness of cancer immunotherapies, such as cancer vaccines. Immunological assays, such as the ELISPOT assay or the HLA-peptide multimer assay, are very efficient at detecting antigen-specific CD-8 T-cells. Subsequently, they are frequently used to detect immunizing-antigen-specific T-cells after immunotherapy in an attempt to validate surrogate endpoints that correlate with clinical outcomes. However, with few exceptions, there have been poor correlations between the magnitude of the immunizing-antigen-specific T-cell responses (that were measured by these immunological assays) and improved clinical outcomes.[18, 19] The most obvious explanation for this phenomenon is that the immuno-monitoring assays measured irrelevant T-cell responses that were ineffective for tumor destruction. Additional explanations are that the assays were unable to detect how T-cells respond to multiple tumor-rejection antigens other than the immunizing antigen; and this could have resulted from immunotherapy-induced tumor lysis and endogenous priming with new tumor-derived antigens. Some recent immunotherapy trials have provided supporting evidence that the immunologic phenomena of determinant antigen or epitope spreading may play an important role in clinical responses to immunotherapy.[18, 29, 30] Butterfield et al. reported that new post-vaccine responses to non-vaccine antigens were observed in clinical responders, although nonresponders did not display reactivity to epitopes other than those used for the vaccination.[18, 30] Although ELISPOT or HLA-peptide multimer assays are quite efficient at detecting antigen-specific effector T cells, these assays are limited in their ability to assess post-immunotherapy responses to new tumor-derived antigens. To accurately predict the clinical response after immunotherapy, it is necessary to apply more comprehensive and functional assays in addition to these antigen-specific assays.
It is important to use structure-based assays such as HLA-peptide multimer to dissect the antitumor responses and predict clinical outcome after immunotherapy, and it is also equally important to analyze the function of effector cells. Analysis of antigen-driven cytokines expression provides a functional property of effector T cell. IFN-γ expression has been shown to be highly reproducible and sensitive enough to detect functional effector T cells. Recently, June et al. have demonstrated that the clinical response of patients' treated with genetically engineered T cells was accompanied by a delayed increase in the serum levels of cytokines (IL-6, IFN−γ, IL-8 and IL-10).[33, 34] In their study, IFN-γ peaked on day 17 after adoptive cell therapy and had the largest relative change when compared with the baseline. The demonstration by June et al. that IFN-γ plays an important role in adoptive T cell therapy emphasizes the relevance of our results.
The whole blood IFN-γ production assay used in our study had several advantages. Uno et al. reported that the value of IFN-γ production in whole blood often differs from that in peripheral blood mononuclear cells (PBMC). Serum factors such as IL-10 can affect IFN-γ production and as IFN-γ production values determined from isolated PBMC would lack information about the influence of serum on the cells; our whole blood method is likely to be a more comprehensive tool for evaluating patients' actual immune responses.[35, 36] Moreover, whole blood IFN-γ production assays are available for immune monitoring for multiple cancer immunotherapies in which the tumor antigen is unknown.
Tumors may differentiate, expand, recruit, and activate Tregs through multiple mechanisms and potentially abrogate antitumor immunity. Tregs have been shown to increase in tumors and in the peripheral blood of cancer patients and to be inversely related to the outcome of several types of human malignancies.[25-28] The prevalence of Tregs in peripheral blood and tumors is an indicator of poor survival in pancreatic cancer.[38, 39] Consistent with these findings, the results in this study demonstrated that low numbers of peripheral Tregs at the baseline correlated with prolonged OS in patients with pancreatic cancer. We expected that adoptive T-cell therapy could exert influence on the status of peripheral Tregs, but we found no consistent modulations in the number or proportion of Tregs with adoptive T-cell therapy. Therefore, it led us to conclude that the number of peripheral Tregs at baseline may not be a predictive factor of immunotherapy, but instead may be a prognostic marker.
One of the limitations of our study is that it is questionable if the increase in whole blood IFN-γ production truly reflected treatment-induced immune responses. Thus, when immunogenic antigens are known, we should use antigen-specific assays such as ELISPOT or HLA-peptide multimer assays. Doing so will allow researchers to assess the relationship between the assay results and whole blood IFN-γ levels so that the reliability of the assay as surrogate markers of clinical outcome can be validated. By combining whole blood IFN-γ assays and antigen-specific immunological assays, we believe it may be possible to construct a novel new assay system for the monitoring of patients' immune response to cancer immunotherapy.
In conclusion, we demonstrated that adoptive T-cell therapy increased the levels of whole blood Th1 cytokine, IFN-γ, in advanced pancreatic cancer patients who were in a strong immunosuppressive state. The increase in whole blood IFN-γ levels after adoptive T-cell therapy was shown to be independently related to OS in a multivariate Cox proportional hazards analysis. These findings indicated that the assay of whole blood IFN-γ levels offers promise as a convenient and efficient method for evaluating clinical responses to cancer immunotherapies.
The authors are grateful to the staff of Hyakumanben Clinic for their work on this study.