Adaptive cellular therapy (ACT) is a procedure that aims to first expand T cells in vitro and then re-infuse the expanded T cell pool back into patients for cancer treatment [53]. Compared to peripheral blood of cancer patients, tumor infiltrated lymphocytes (TILs) are enriched in tumor-specific T cells and can be easily expanded in vitro by tumor cells with the presence of growth factors like interleukin-2 [54, 55]. This practice can generate tumor-reactive T cells with a broad range of tumor reactivity, without the knowledge of tumor antigen identities. With the improvement of culturing technology, the degree of expansion and quality control has been greatly enhanced. Isolating and expanding TILs for ACT is a very efficacious treatment strategy in melanoma [56]. However, the number of TILs that can be successfully recovered from the vast majority of solid tumors is very limited, especially for those cancers with few TILs. In addition, the majority of TILs display exhausted or dysfunctional phenotypes, which might cause the poor persistence of expansion tumor-specific T cell clones upon intravenous infusion [57, 58]. Therefore, the current approach of expanding TILs for ACT is mainly practiced in melanoma patients.

Genetically, engineering of lymphocytes is a new approach that aims to eliminate the obstacle posed by many tumors with a limited number of tumor-reactive T cells for ACT [52, 59, 60]. This strategy involves transducing immune cells with genes that redirect T cells to recognize cancer cells. The specificity of T cells can be redirected by the incorporation of genes encoding either conventional alpha-beta TCRs or chimeric antigen receptors (CARs) [61]. In this case, T cells from patient blood can be directly used as a source for ACT. CARs are constructed by linking the variable regions of the antibody heavy and light chains to intracellular signaling chains, such as CD3-zeta. The new generation of CARs is also composed of costimulatory domains of CD28 and/or CD137 to promote T cell expansion [62]. Because CARs are derived from antibodies, recognition of tumor-associated antigens (TAAs) by CARs is strong and is not restricted by major histocompatibility complex (MHC) [62].

However, a major hurdle for CAR-T therapy in human cancer is the identification requirement of appropriate tumor antigens that are exclusively expressed on the cancer cells, but not normal self-tissues. Most of the currently identified tumor-specific antigens are self-antigens that are normally expressed in early fetal development and that are aberrantly expressed during malignancy [63]. Examples include NY-ESO1 and the MAGE family antigens. The phenomenon known as “off-tumor, on-target,” where CAR-T cells recognize non-cancer cells expressing the tumor antigen, is responsible for severe immune-mediated toxicities that have limited the applicability of this treatment strategy [64, 65]. Therefore, careful monitoring and screening of targets for CAR-T therapy is extremely important. As a result, current successes of CAR-T therapy in clinic are mainly limited to certain types of lymphoma/leukemia [66]. Testing the feasibility of this approach can only be carried out in clinical trials, as preclinical models have proven to be insufficiently predictive of both efficacy and toxicity in humans. Whole-genome sequencing of cancer cells is generating abundant information about specific mutations in tumor cells, which may lead to the identification of tumor-specific antigens, also called neoantigens [67]. Innovative ways of generating antigen receptors that recognize these, including CARs that directly recognize intracellular molecules presented by MHC, may generate T cells with even greater specificities for tumor cells. It is worth noting that another potential obstacle for ACT is that the majority of the inoculated T cells die before reaching the cancer site, which can be a challenging obstacle for patients with solid cancers [68]. Therefore, selection of T cell subsets with better capacity for survival and proliferation is a critical step in ACT, and methods to selectively enrich central memory and stem cell memory T cell subsets from human lymphocytes may enable more effective anticancer responses in humans, similar to those observed in mouse models [68–70]. Although the range of CARs currently available is sufficient to cover most types of malignancy, tumor cells can lose the expression of TAAs to evade immune attack during ACT [60]. Therefore, using several CAR genes that target multiple TAAs simultaneously may be needed for future ACT to better accommodate the heterogeneity in human cancers.

Animal models of pancreatic cancer have shown encouraging results with the use of ACT [71, 72], and clinical trials using CAR engineered T cells for pancreatic cancer are currently ongoing in many cancer centers (NCT01897415, NCT02465983) [33]. The recently completed PDAC genomic analysis by Bailey et al. led to a deeper understanding of the molecular evolution of PDAC and to the identification of a specific immunogenic PDAC subtype [73]. This new and long awaited information may open the way to new and more accurate therapeutic targets for ACT.

Vaccine is an active therapeutic approach aiming to mobilize the immune system to generate or amplify tumor-specific immune response to combat cancer [74]. The primary mechanism for therapeutic cancer vaccines is to increase the presentation of tumor-associated antigens (TAAs) to the immune system, so as to mount a potent immune response against tumors. Cancer vaccines attempt to copy the achievements made in vaccinations against pathogens though more work is necessary to bring it to fruition. Based on the formats utilized, cancer vaccine can be classified into three major categories: protein/peptide vaccines, whole cell vaccines, and DNA vaccines [51].

T cell response is largely controlled by an array of cellular surface signaling molecules, also known as cosignaling molecules [89]. Modulating these cosignaling pathways increases anticancer immunity, either through the amplification of costimulatory pathways or blockade of negative signals, also known as immune checkpoints [90]. The major immune checkpoints under clinical investigation include at least CTLA-4, PD-1, TIM-3, LAG3, and TIGIT [91, 92]. Many of the ligands for immune checkpoints are upregulated at cancer sites and contribute to the induction of tumor-specific T cell exhaustion/dysfunction at cancer sites [91, 93]. Using monoclonal antibodies or fusion proteins are the main strategy to block or amplify cosignaling pathways. The immunomodulation strategy strives to promote or liberate internal anticancer immunity in a patient with an established cancer [94]. One of the advantages of this therapeutic strategy is that immunomodulation does not require the knowledge of specific cancer antigens but rather focuses on the manipulation of known leukocyte receptors. These provide several potential therapeutic targets that are characterized by a broad spectrum of antigen diversity that could ultimately avoid the mechanism of cancer immune evasion, caused by mutation of cancer-specific antigens [95].

Targeting immune checkpoints has been a major breakthrough in cancer treatment in recent years [96]. CTLA-4 is transiently expressed on the T cell surface upon activation and attenuates ongoing T cell response by competing ligands B7-1 (CD80) and B7-2 (CD86) with the costimulatory receptor CD28 [97, 98]. In addition, CTLA-4 also transduces a suppressive signal to T cell via the recruitment of phosphatases SHP-2 and PP2A [90]. Ipilimumab, an anti-CTLA-4 mAb, is the first FDA-approved immunotherapy drug to treat patients with late-stage melanoma [99, 100]. Administration of ipilimumab activates T cells systemically, leading to extensive antitumor immunity and therefore a survival benefit in 10–15% of patients with advanced metastatic melanoma. Furthermore, this antitumor response significantly increases overall patient survival in advanced melanoma cases [99]. However, antitumor activity is frequently accompanied by significant immune-related adverse events. PD-1 is another inducible immune checkpoint on T cells that suppresses T cell response upon interaction with its two ligands, PD-L1 (B7-H1) and PD-L2 (B7-DC) [36, 101]. The PD-1 pathway is heavily involved in the immunosuppressive cancer microenvironment: PD-1 is highly expressed in TILs while the ligand PD-L1 is found on tumor cells, tumor-associated DCs, macrophages, and fibroblasts [94, 102]. Targeting the PD-1 pathway has elicited durable antitumor responses and long-term remissions in patients with a broad spectrum of cancers. The objective response rates varies in different cancer types, with bladder cancer, melanoma, mismatch repair-deficient colorectal cancer and certain hematopoietic malignancies among the most responsive cancer types [102]. Compared to CTLA-4 blockade, the antitumor efficacy of PD-1 blockade is higher, and the side effect is significantly milder and manageable [49, 103–105]. Currently, PD-1/PD-L1 mAbs have been approved by FDA to treat late-stage melanoma, non-small cell lung cancer, and kidney cancer [102]. It is anticipated that PD-1 mAb will be approved for treating more cancer types and become the frontline therapy for future cancer treatment.

Ipilimumab alone, or in combination with peptide vaccine, did not have any clinical benefit in treating PDAC patients. In a phase II trial of 27 patients with advanced PDAC, single-agent ipilimumab failed to detect any responder by response evaluation criteria in solid tumors [50]. However, a significant delayed response in one subject of this trial suggests that this immunotherapeutic approach to PDAC deserves further exploration [50]. With tremendous success in many cancer types, early trials of anti-PD-1 mAb alone showed no effect in treating patients with advanced PDAC, though the number of patients in the study was small [49]. PD-1 mAb alone is ineffective in treating cancers with few neoantigen loads [49]. PDAC happens to be a low immunogenic cancer [31]. It is not surprising that targeting immune checkpoints alone is incapable of launching an effective anticancer immunity in PDAC patients. Therefore, additional procedures are needed to increase the number of TILs surrounding PDAC cancers, so as to prime PDAC for immune checkpoint therapy [106].

The low immunogenicity and unique stromal structure of PDAC cancer poses a great challenge for immunotherapy [22]. The disappointing outcomes in clinical trials using single-agent immunotherapy propel the launch of combinatory approaches. Combination therapy targets more than one aspect and can be classified as the combination of two different arms of immunotherapeutic approaches, or the combination of immunotherapy with traditional therapy (chemotherapy or radiotherapy).

Examples of combinatory approaches that combine two different arms of immunotherapy could include the combination of cancer vaccine with immune checkpoint blockade or the simultaneous use of both active and passive immune therapy. Immune checkpoint inhibitors alone are not effective in the treatment of PDAC, much due to the lack of tumor-infiltrating T cells at tumor sites [23]. Cancer vaccine is known to be a very efficient method of expanding tumor-reactive T cells, while blockade of immune checkpoints will further promote antitumor immune responses at tumor sites. In fact, several preclinical studies exist that demonstrate the synergistic role of cancer vaccine therapy, which is responsible for stimulation of the immune system, and the use of immune checkpoint blockade, which allows for the unopposed effector function of cytotoxic T cells [107–109]. Consistently, clinical examination of resected PDAC tumors demonstrated that vaccine therapy can alter the immunosuppressive cancer microenvironment [106]. The majority of PDAC patients receiving GVAX vaccine had vaccine-induced intratumoral tertiary lymphoid aggregates in resected tumors, accompanied with increased intratumoral T_eff/T_reg ratios [106]. As such, a phase Ib, open-label randomized study demonstrated the feasibility and safety of an approach based on the combination of Ipilimumab with GVAX in patients with previously treated PDAC [47]. One of the most interesting aspects of this study was the difference in 12-month OS: 27% vs. 7% and the median OS: 5.7 vs. 3.6 months (HR = 0.51; P = 0.072), respectively, for combination therapy vs. monotherapy (i.e., Ipilimumab alone).

Given that PD-1/PD-L1 blockade is safer and more effective than CTLA-4 blockade in the treatment of many cancers, it is interesting to see how the combination of GVAX with PD-1/PD-L1 blockade performs in the treatment of PDAC [104]. Interestingly, PD-L1 expression was observed in all these lymphoid aggregates in GVAX-treated PDAC patients [106]. Currently, a phase I/II study with GVAX and anti-PD-1 mAb (nivolumab) has started to recruit patients with PDAC (NCT02451982). Similarly, a randomized phase II trial of GVAX and CRS-207 with or without nivolumab has also launched (NCT02243371).

Combinational therapy involving PD-1/PD-L1 blockade has also been investigated with chemotherapy or radiotherapy in PDAC [110]. These approaches are based on the observation that chemotherapy or radiotherapy can kill cancer cells to increase the supply of tumor antigens for presentation, so as to promote tumor-reactive immune responses [111–113]. In addition, many conventional cancer treatments in chemotherapy and radiotherapy have immune potentiating mechanisms of action, such as the elimination of immunosuppressive cells, including T_reg and MDSC (Zitvogel L, JCI 2008). A phase I trial (NCT02303990) of pembrolizumab (anti-PD-1 mAb) with the combination of hypofractionated radiotherapy has started to treat patients with metastatic pancreatic cancer. In another phase I study of PDAC (NCT02546531), PD-1 blockade (pembrolizumab) is proposed to be combined with gemcitabine and defactinib, an inhibitor of focal adhesion kinase (FAK), which promotes stromal fibrosis. Because immune effector cells are also sensitive to chemotherapy and radiotherapy, early phase clinical investigations into optimizing dose and schedule in patients are necessary.

Despite recent advancements in PDAC treatment modalities, modest success has been achieved and the curative goal remains unmet. Surgical resection remains to be the only potential cure for early-stage PDAC. Immunotherapy emerges as a promising treatment for metastatic PDAC, with the potential of targeting disseminated disease as well as preventing cancer recurrence. With the technological advancement in genome sequencing, neoantigens in PDAC will be identified as better targets for vaccine therapy or ACT. Together with further interrogation of the PDAC microenvironment, it is promising that more PDAC immunosuppressive mechanisms, by which PDAC evades immune attack, will be revealed for future immune interventions.

Early clinical trials in immunotherapy also demonstrated that the complexity of the PDAC microenvironment and the formidable immunosuppressive nature of this cancer might require a combination of different therapeutic strategies [110]. These therapies need to be able to simultaneously target the stroma-cell population, where the tumor cells locate, as well as the cytotoxic T lymphocytes (manipulating different immune checkpoint inhibitors) or directly the tumor cells (traditional chemotherapeutic, vaccination, ACT). For instance, besides cancer vaccine, other therapeutic approaches, including chemotherapy, radiotherapy, and ACT may prime PDAC to become susceptible for immune checkpoint inhibitor therapy. Moving forward, the focus of modern clinical immunotherapy will be to identify the most efficacious, synergistic therapy that is able to obtain the maximum antitumor activity with the least systemic toxicity. Finally, it is imperative to identify reliable biomarkers to predict tumor susceptibility to immunotherapy in clinic, to identify those patients that are more likely to benefit from this unique therapeutic approach.