Even with the rapid development of surgical techniques and a broader understanding of oncology over the past few decades, pancreatic ductal adenocarcinoma (PDAC) remains at a 5-year survival rate of 5–6% (1,2). Only approximately 15–20% of patients with PDAC are suitable for surgical resection, which is the best possible treatment for good oncologic outcome (1,2). For most pancreatic cancers that are locally advanced or meta-static, the National Comprehensive Cancer Network (NCCN) recommends chemotherapy (1–4). However, only a limited subgroup of these patients will really benefit from chemotherapy; instead, patients often have severe adverse events, and even a more progressive disease (1–4). Therefore, it is necessary to select the patients with unresectable PDAC that could have a better prognosis or benefit from chemotherapy. Different tumor biological characteristics may contribute to better outcomes in this subgroup of patients. Unfortunately, there is still a lack of a simple and effective marker to identify these patients.

Generally, PDAC, especially advanced cases, is accompanied by obstruction of the pancreatic duct and varying degrees of chronic pancreatitis, which contribute to a unique tumor microenvironment comprising a specific network of multiple immune and inflammatory factors (5,6). It was reported that a dynamic imbalance involving CD4⁺ T cells and CD8⁺ T cells contributed to tumor immune escape and failure of immune surveillance during pancreatic tumorigenesis (5–7). T regulatory (Treg) cells, one of the main subsets of CD4⁺ T helper cells, were reported to be recruited by the pancreatic tumor microenvironment from a pre-invasive stage to invasive and metastatic PDAC and played a crucial role in immunosuppression (6,7). Previously, we found that peripheral Treg levels were significantly increased in patients with PDAC compared with that of healthy donors and patients with benign pancreatic disease; furthermore, for patients with PDAC who underwent a radical resection, the percentage of peripheral Tregs was regarded as an independent prognostic factor (8). Although it has been widely confirmed that infiltrating immune cells in tumor tissue have prognostic values (5–7), whether the systemic quantifiable immune parameters also have prognostic value is an interesting question. The authors consider that the data of circulating immune parameters are more widely available than the data of infiltrating immune cells. Thus, if the circulating immune parameters have prognostic value, the makers would be more convenient and accessible to monitor tumor progression. Therefore, the present study focused on circulatory immune cells to explore prognostic markers, based on retrospective analysis. In addition, tissue samples obtained through EUS-FNA in unresectable pancreatic cancer patiens are rare and limited and often insufficient for tumor microenvironment analysis. In the present study, although the results are not direct evidence, consideration should be given that PDCA caused the abnormal distribution of peripheral T lymphocyte subsets.

Chemotherapy is a primary treatment for advanced or metastatic PDAC; however, a new perspective has emerged that the cytotoxic effects of chemotherapy not only destroys tumor cells, but also impairs immune cells that subsequently damage the antitumor response and even promote tumor growth and metastasis (9–11). In this study, we aimed to observe the alteration and distribution of peripheral T cell subsets in patients with unresectable PDAC, evaluate the effect of chemotherapy on T cell-related immune system, and discover some effective factors that could predict systemic chemotherapy response in patients with unresectable PDAC.

PDAC contains abundant stroma elements surrounding cancer cells, and thereby has a unique biologic profile including chemotherapy resistance and poor prognosis (12). The PDAC specific tumor microenvironment lacks oxygen and blood supply, which contributes to metabolic reprogramming and immune cell adaptation (12–14). Emerging evidence demonstrates that tumor infiltrating T cells such as Tregs, CD8⁺ T cells and Th17 cells were associated with prognosis in various human cancers (15,16). In a previous study, we investigated the expression of tumor infiltrating cells within the specimens of resected pancreatic cancer and found that more infiltration of CD4⁺ and CD8⁺ T lymphocytes indicated a better prognosis after surgical resection; in contrast, the primary immunosuppressive factor Tregs and other T helper cell subsets could hardly be detected in PDAC tissues (16). In the present study, the percentage of peripheral Tregs and CD4/CD8 ratio before treatment was associated with overall survival of patients, exhibiting a circulating T lymphocyte signature in patients with advanced PDAC cases; thus, this may serve as an effective marker for the prognosis of this subgroup of patients before any treatment is done.

Because of the unique tumor microenvironment of PDAC, contributed to by a consistent deprivation of oxygen and nutrients, there will be a persistence of endoplasmic reticulum (ER) stress, which is a mechanism of cell self-protection within the tumor microenvironment of PDAC (17,18). Given that certain chemotherapeutic agents were also stress inducers, there existed a sustained stress status and subsequent tumor metabolism reprogramming in advanced PDAC (19–21). Prolonged ER stress promotes cell apoptosis, but, conversely, might induce cancer cells to acquire more aggressive biologic features to resist chemotherapeutic attack; furthermore, the percentage and types of immune cells recruited by the tumor microenvironment also adapted accordingly (13,14,19–21). In particular, T cell dysfunction owing to nutrient deprivation and tumor metabolism alteration was regarded as a major issue of concern in antitumor immune response (22,23). Immune responses to malignant cells can be categorized as locoregional or systemic. In situ, the immune contexture would be essential to accurately define the impact of the local host-immune reaction in tumor (24). It has been well reported that the infiltrating immune cells, in tumor site, play important roles in tumor progression. Unfortunately, in most of the patients with unresectable PDAC, a sufficient amount of tumor tissues to detect intratumoral infiltration of immune cells is difficult to obtain, especially for monitoring chemotherapy-related immune status. Systemic immunity to tumors, as measured in the peripheral circulation, is difficult to demonstrate and tumor-specific responses are particularly elusive. The role of systemic immunity in tumor progression deserves further study. The available evidence suggests that both local and systemic antitumor immunity is compromised in patients with cancer. Many researchers have investigated the correlation of circulating immune cells and tumor infiltrating lymphocytes (TIL) in regard to distribution and function characterization. Circulating T cells obtained from patients with cancer are either biased in cytokine profile or otherwise functionally compromised. Furthermore, dysfunction in circulating lymphocytes was linked to the extent of dysfunction seen in paired TIL and to the disease stage and/or activity (25,26). It remains controversial whether systemic chemotherapy impairs the immune system. A limited number of patients with pancreatic cancer exhibited a favorable response to systemic chemotherapy, and only limited available options including gemcitabine-based or fluoropyrimidine-based regimens were recommended (9,27). Accordingly, the balance between benefit and risk of chemotherapy in advanced PDAC should be well evaluated. Therefore, if the circulating immune parameters have prognostic value, the markers would be more convenient and accessible to monitor tumor progression. In the present study, the changes in circulating Tregs and CD4⁺/CD8⁺ T cell ratio before and after chemotherapy discriminate the populations that benefit from chemotherapy, to some extent, and show the divergence in adaption of T cell immune status responsive to systemic chemotherapy in patients with high PDAC tumor burden. Furthermore, decreased Tregs was detected more in PR and SD cases. Moreover, the increased amount of immunosuppressive T cells found in PD cases also indicated that PDAC response to chemotherapy was associated with immune adaption of T cell subsets following a local or systemic stress response or metabolism reprogramming.

The chemotherapeutic agent is a two-edged sword having both immune suppressing and promoting effects. CD3⁺ T cells contain various T cell subsets, such as CD4⁺ T cells, CD8⁺ T cells and Tregs. Although our results found that the percentage of CD3⁺ T cells increased after two cycles of chemotherapy, suggesting chemotherapy does not lead to total T cell ratio decrease, the change of other T cell subsets need to be further verified. It was shown that some chemo-agents could induce T-cell infiltration into pancreatic cancer. Tsuchikawa et al (28) reported that Foxp3⁺ T cell infiltration was significantly lower in the neoadjuvant chemoradiation therapy cases, while the numbers of CD4⁺ and CD8⁺ T cell infiltration had no statistical difference in the tumor microenvironment. Other studies found that CD4⁺ and CD8⁺ T cells were significantly increased after neoadjuvant chemotherapy (29). However, for circulating T cells, although it was reported that circulating CD4⁺ and CD8⁺ T cells were changed after chemotherapy in other cancers such as breast (30) and cervical cancer patients (31), the relationship between circulating T cells and chemotherapy was still unclear in pancreatic cancer. In the present study, gemcitabine-based chemotherapy did not exhibit an obviously impaired effect on human T cell immune system, but underlined some insights that high tumor burden PDAC and continued chemotherapy could result in differentiation of T cell subsets.

The change of T cell proportion may be altered by the chemotherapy or by the shrinkage of the tumor. Many studies reported that chemotherapy agents could change tumor microenvironment through various mechanisms, such as increasing lymphocyte infiltration, depletion of Tregs and inducing differentiation of MDSC (32). However, for the circulating lymphocytes, it was still unclear. It was reported that low Treg percentage in circulation at 1 year after pancreatic cancer resection was correlated with improved survival (33). We also previously reported that high Treg percentage in circulation predicts poor prognosis in resectable pancreatic cancer patients (8). We consider that circulating lymphocyte levels are features of immune status in pancreatic cancer patients. Our data showed that Treg cells decreased after chemotherapy in SD+PR patients (N=59), while increased in PD patients (N=41), suggested that patients with good immune status during the period of chemotherapy, which may be caused by the shrinkage of tumor, are correlated with prognosis. Metabolic features of a hypoxic microenvironment, including glycolysis and Warburg effect, might not only enable cancer cells to develop different biological properties, but may also influence T cell function and T cell differentiation (34–36). Several transcription factors and signaling pathways, such as HIF-1α, PD-1 and mTOR signaling are involved in T cell immune adaptation and differentiation (37). Hypoxia and nutrient deprivation contribute to accumulation of metabolic products and then suppress CD8⁺ T effector cells and CD4⁺ Th1, Th2 and Th17 subsets, whereas they induce immunosuppressive Treg lineage (38,39). Generally, the percentage of tumor infiltrating CD4⁺ T cells and CD8⁺ effector T cells indicates a different prognostic value in cancer (40). For CD4⁺ T cell subsets, the plasticity between regulatory T cells and Th17 lineage of T helper cells has been widely studied recently (41,42). Although Tregs always accompany poor clinical outcomes, Th17 exhibits some inconsistent prognostic significance in various human cancers (43,44). Tregs are reported at elevated frequencies in the peripheral blood and the tumor itself, and correlates with poor outcome. It was reported that Tregs can suppress immune cell activity and induce tumor cell escape from immune surveillance via secreting several cytokines. In addition, it was shown that Tregs complexed with STAT3, a transcription factor that is essential for the differentiation of Th17 cells, then activate IL-17 gene transcription to promote Th17 cell differentiation (45,46). Tregs and Th17 cells are inversely associated in the same tumors, and there is a dynamic interaction between Th17 and Tregs in the tumor microenvironment. The balance of Tregs/Th17 contributed to the shift between pro-inflammatory and anti-inflammatory response and somewhat indicates the prognosis of human cancer (47,48). It was reported that IL17⁺FOXP3⁺ T cells can be detected in tumors, which also express CD25 and TH17 specific marker RORγt and have suppressive functions. Moreover, Treg cells can transdifferentiate towards an IL-17⁺FOXP3⁺ T phenotype and eventually into FOXP3-TH17 cells (49,50). In the study, we detected several cytokines involved with the transdifferentiation between Tregs and T helper cell subsets in advanced PDAC cases. In addition, we found that the expression of IL-17A, which is characterized by the secretion of Th17, was significantly elevated in the cases whose Treg level was decreased, indicating the possibility that a shift from Tregs to Th17 might be responsible for the sensitivity to chemotherapy in patients with unresectable PDAC.

The present study included locally advanced and meta-static cases of PDAC, the basic characteristics in our group were similar with other studies (3,51). In addition, we also identified an initial CD4/CD8 ratio or Treg level before any treatment was associated with the prognosis via multivariate analyses. Therefore, we concluded the significant difference is valuable. As limitation in the present study, we found that the change of proportion in CD3 positive cells had statistical significance (P=0.035), while the main subsets of CD3⁺ cells such as CD4⁺, CD8⁺ and Treg cells remain stable after chemotherapy. However, we did not detect other T cell subsets (Th1, Th2 or Th17), which might change after chemotherapy. These results need to be further identified via expanding sample sizes and detecting more subsets of immune cells. In addition, we also found that there was no correlation between Tregs and the expression level of either IL-2 or IL-10 (data not shown), while other cytokines such as IFN-γ and IL-5, which may be associated with helper T cell subsets or Tregs, were not detected. We will continue to collect samples to validate these results and explore potential mechanisms in future.

Overall, the present study shows a circulating signature of T cell subsets that could predict the prognosis of patients with unresectable PDAC before treatment. This finding could also predict the response to gemcitabine-based chemotherapy, which could be contributed to by the plasticity of T cell subsets and their capacity to differentiate from one subset toward another lineage, depending on a high tumor burden PDAC setting or chemotherapy-related metabolic changes.