Alexander H. Morrison | Katelyn T. Byrne | Robert H. Vonderheide

Highlights
Pancreatic cancer remains a lethal tumor that is difficult to treat and, unfortunately, immune therapies that have garnered FDA approval in other tumors have shown little efficacy to date in this tumor. These therapies include checkpoint antibodies and engineered T cell infusions.

A formidable problem in developing effective immunotherapy for PDA is the striking immunosuppressive and ‘immune-privileged’ tumor microenvironment. Few patients exhibit robust T cell infiltration in the tumor microenvironment, although when this does occur patient survival is prolonged.

Major clinical efforts, justified by preclinical models, are now aimed at combination immune therapies that address multiple immune vulnerabilities in PDA in a nonredundant fashion.

Generation of stronger adaptive immunity with vaccines and immune agonists may be necessary before antibodies against CTLA-4, PD-1, or PD-L1 will be effective.

Pancreatic cancer is the third-leading cause of cancer mortality in the USA, recently surpassing breast cancer. A key component of pancreatic cancer’s lethality is its acquired immune privilege, which is driven by an immunosuppressive microenvironment, poor T cell infiltration, and a low mutational burden. Although immunotherapies such as checkpoint blockade or engineered T cells have yet to demonstrate efficacy, a growing body of evidence suggests that orthogonal combinations of these and other strategies could unlock immunotherapy in pancreatic cancer. In this Review article, we discuss promising immunotherapies currently under investigation in pancreatic cancer and provide a roadmap for the development of prevention vaccines for this and other cancers.

Challenge of Pancreatic Cancer
In early 2017 pancreatic cancer surpassed breast cancer to become the third-leading cause of cancer-related mortality in the USA, behind only colon and lung cancer [1]. Unlike many other cancers, pancreatic cancer is increasing in both incidence and mortality, and it is predicted to become the second-leading cause of cancer-related death by 2030 [1]. Current therapies are severely lacking; recently approved combination chemotherapies such as FOLFIRINOX and gemcitabine/nab-paclitaxel improve median survival by only 2–4 months and are associated with significant, toxic side effects [2, 3]. Encouragingly, a few long-term survivors are beginning to be observed after such treatment, but the 5-year survival rate – although improving – remains a grim 8% [4]. Moreover, for a variety of complex and unfortunate reasons, including limited geographical access to trial sites, restrictive eligibility criteria, and patient decisions [5], 95% of pancreatic cancer patients do not enroll in trials of investigative therapies.

Immunotherapy has had remarkable efficacy in many malignancies [6, 7, 8, 9, 10] but has not yet translated to pancreatic ductal adenocarcinoma (PDA). Immune checkpoint blockade seems to have minimal activity and, despite promising Phase I data, whole-cell therapeutic vaccines have demonstrated no effect in late-stage trials [6, 11, 12]. There are many reasons for these failures, but key contributors are the immunosuppressive tumor microenvironment (TME) – characterized by typically poor infiltration of effector T cells and prominent myeloid inflammation [13, 14, 15] – and a low mutational burden predicted to generate very few immunogenic antigens [16, 17]. Promisingly, a small subset of patients present with tumors exhibiting high effector T cell infiltration and have longer overall survival [18, 19, 20], suggesting the potential for effective treatment of PDA with immunotherapy. Investigations into an increasingly diverse array of immunotherapies and subsequent rational combinations with other therapeutic approaches are likely to hold the most promise for patients with PDA. Moreover, the development of prevention vaccines for PDA is now within reach and could transform the way PDA is treated by targeting malignant cells before the immunosuppressive TME is fully established, thereby obviating the need for toxic immunotherapies.
In this Review article, we discuss the major immunotherapies and combinations that are being investigated in pancreatic cancer, both clinically and preclinically, with an eye toward the most promising approaches (Figure 1). We then discuss cancer prevention vaccines and the rationale for investigating the use of these vaccines in the quest to cure pancreatic cancer.

Figure 1Clinical Status of Immunotherapies in Pancreatic Cancer. Each line represents a single class of immunotherapy. The right end of each line indicates the latest stage of clinical trials that class of compounds has reached. Filled green arrows indicate ongoing trials. Red lines indicate negative trials. Broken green arrow indicates successful trials in other malignancies. dMMR, DNA mismatch repair; RCT, randomized controlled trial; PDA, pancreatic ductal adenocarcinoma.
View Large ImageFigure ViewerDownload Hi-res imageDownload (PPT)
Single-Agent Immunotherapies in Pancreatic Cancer
Immune Checkpoint Blockade
Checkpoint blockade has resulted in remarkable successes in other cancers, including melanoma and lung cancer, but has shown little efficacy in PDA [6, 11]. Checkpoint blockade targets immune checkpoint molecules – primarily programmed cell death protein 1 (PD-1), programmed death-ligand 1 (PD-L1), and cytotoxic T lymphocyte-associated protein 4 (CTLA-4) – that negatively regulate T cell function. Inhibition of these molecules ‘takes the brakes off’ the immune system, resulting in tumor killing. The reasons for the failure of immune checkpoint blockade in PDA are multifactorial. PDA has low baseline PD-1+ T cell infiltration into the tumor [15, 21] and a paucity of neoepitopes [17, 22], both of which are predictive of response to PD-1 blockade in other solid tumors [23, 24]. In a very small subset (∼1%) of PDA patients with a high burden of microsatellite instability (MSI-high) – and therefore high neoepitope burden – PD-1 blockade is effective [25, 26] and was recently FDA approved [27, 28]. In the absence of high neoepitope burden, preclinical models have shown that therapies capable of improving T cell infiltration into the TME sensitize PDA to checkpoint blockade [29], suggesting that combinations of treatments that improve T cell trafficking with checkpoint blockade may be successful in the clinic. However, a small proportion of PDA patients have both activated T cells and a detectable neoepitope burden and yet are resistant to therapy [17]. The complete lack of efficacy of checkpoint blockade in these patients suggests that there is more to T cell responses in PDA than the PD-1–PD-L1 axis. Multiple other immune molecules can suppress T cell responses in cancer, including: TIM3, TIGIT, and LAG3, which are inhibitory receptors on T cells analogous to PD-1; VISTA, an inhibitory ligand on myeloid cells analogous to PD-L1; and CD73, an extracellular enzyme that generates the immunosuppressive and prometastatic molecule adenosine. All of these immunosuppressive pathways are highly expressed in PDA [17] and investigation of these targets may unlock checkpoint blockade in PDA.

Therapeutic Vaccines
Therapeutic vaccines have the potential to induce robust antitumor immune responses but have so far failed to deliver on their early-stage promise in pancreatic cancer. Vaccine approaches, including whole tumor cells, peptides, proteins, and recombinant constructs, aim to prime circulating tumor-specific T cells that can then eliminate tumors. In small Phase I studies, almost all of these formulations have generated tumor-specific T cell immunity in subsets of patients [30, 31, 32]. Tantalizingly, patients who generated vaccine-specific T cell immunity appeared to have superior survival in many of these small, early-stage trials. Unfortunately, results from later-stage trials did not support these early findings. A Phase III trial of a vaccine using a single peptide derived from the tumor-associated self-antigen human telomerase (hTERT) showed no survival benefit in patients with metastatic disease, even in immunologic responders [33]. Whole-cell vaccine approaches, which may broaden the immune response against both tumor-specific and shared tumor/self-antigens, have also had limited success. GVAX, a vaccine comprising irradiated allogenic PDA cells that express granulocytic-macrophage colony-stimulating factor, failed to improve survival in Phase IIb/III trials in metastatic PDA [12], even among the immunologic responders. Together these results have diminished excitement about therapeutic vaccines.
Therapeutic vaccination has the potential to be more effective in the adjuvant setting, where the volume of tumor and immunosuppressive stroma is greatly reduced. Small Phase I/II trials have shown that adjuvant vaccines to WT-1, mutant Kras, and MUC1 can generate a T cell immune response and have suggested that the potency of this response correlates with patient outcomes [34, 35, 36, 37, 38]. However, only a fraction of patients in these trials had durable responses. Moreover, the whole-cell lysate vaccine algenpantucel-L (irradiated allogenic PDA cells expressing murine alpha-1,3-galactosyltransferase) failed to improve survival in a recent Phase III trial despite similarly promising immunologic responses in early trials [39]. Emerging data suggest that the adjuvant setting may be less conducive to vaccination than was previously thought. Comparisons with healthy patients have found that the overall immunologic response to vaccination is reduced in the adjuvant setting [40] and in lung cancer the postoperative TME has been shown to be strongly immunosuppressive [41].
Despite the negative top-line results of Phase IIb/III vaccine trials, these trials contain key insights that should provide a path forward for therapeutic vaccines. Importantly, these trials have proved that vaccines can break tolerance and generate T cell immunity to tumor-associated self-antigens without obvious short-term side effects. GVAX vaccination even induced the formation of tertiary lymphoid structures and T cell infiltration in many patients [42]. Why this T cell response did not improve survival remains incompletely understood, but it may be due to suboptimal antigen selection or T cell dysfunction. The path forward for antigen selection is encouraging, as many antigenic targets have been identified and are under investigation [43] and improvements in bioinformatics tools are rapidly enabling the prediction of other high-priority antigens and neoantigens for vaccination [20]. Similarly, there is hope for rescuing T cells from dysfunction. Preclinical evidence suggests that depletion of regulatory T cells (Tregs) can reduce suppression by the TME and enable effective therapeutic vaccination [44, 45]. Moreover, GVAX-induced T cells in human PDA upregulate checkpoint molecules, including PD-1 [42], suggesting that checkpoint blockade can rescue vaccine-primed CD8 T cells from inhibition by the TME; trials combining therapeutic vaccines with checkpoint blockade are under way [46]. Finally, the vaccine delivery vector can be changed or enhanced with cytokines, as is currently under investigation in a Phase I trial using DNA electroporation to increase the immunogenicity of hTERT and IL-12 to improve the priming of the anti-hTERT T cell response in the adjuvant setting (NCT02960594i) [47].
Engineered T Cells
Engineered T cells, such as chimeric antigen receptor T cells (CAR-Ts), have shown remarkable efficacy in B cell malignancies, with response rates up to 90% [48,