Expansion of tumor-infiltrating lymphocytes (TIL) from human pancreatic tumors

The development of novel therapeutic approaches for pancreatic adenocarcinoma is paramount to the successful treatment and recovery from this aggressive cancer. In the age of personalized cancer medicine, adoptive cell therapy (ACT) using tumor-infiltrating lymphocytes (TIL) represents a unique opportunity to exploit both the diversity and specificity of a patient’s immune system. The CD8⁺ T cells within this population represent an endogenous polyclonal repertoire of TCRs specific and comprehensive for the array of tumor-associated antigens (TAAs) and unique neoantigens present [35]. Within the appropriate setting, these infiltrates are capable effectors of a targeted anti-tumor response that can be both durable and complete [4].

In the present study, we explored the feasibility of expanding TIL from resected pancreatic adenocarcinomas. ACT is dependent upon both the quantity and quality of the lymphocytes recovered from the original source. Importantly, we showed that TIL can be expanded from surgically resected pancreatic tumors and REP’d to clinically relevant metrics. Given the inherent constraint of only one in every five patients with pancreatic cancer as a candidate for surgery for curative intent, it is significant that nearly all of the tumors in this study contained lymphocytes capable of in vitro propagation. In addition, a recent study has also demonstrated the feasibility of expanding functional TIL from pancreatic tumor samples [36].

Excess (not required for pathologic analysis), surgically resected pancreatic adenocarcinomas received in the laboratory for TIL generation were characteristically small in size, restricting the supply of starting material crucial to culturing TIL from tumor fragments. It has previously been reported that the density of infiltrating lymphocytes is also inferior in GI tumors when compared to melanoma, leading to an increased time required for expansion [28]. These drawbacks are important to consider given that one central concern of ACT is the necessary delay between surgery and infusion. Pretreatment of patients, such as with immune checkpoint inhibitor therapies, prior to surgical resection in order to enrich tumors for specific lymphocyte populations should be investigated as a means to overcome this limitation.

T cells localized to the pancreatic tumor microenvironment have long been characterized as both intrinsically and extrinsically suppressed, frequently attributed to the presence of T_regs [14, 37, 38]. Importantly, in the current study after expansion in IL-2, pancreatic TIL were predominantly activated T cells, before and after REP, and presented markers of antigen experience, including CD69 and CD45RO. Furthermore, we demonstrated that high doses of IL-2 in vitro did not polarize the pancreatic TIL cultures toward a T_reg phenotype, but may have influenced the observed activated phenotype. A relatively high T_eff to T_reg ratio was established, which is thought to be critical for effective immunosurveillance of tumors [37, 38]. This result highlights the importance of releasing pancreatic TIL from their naturally inhibitory tumor microenvironment and supports the notion that pancreatic TIL have a plastic phenotype that can be polarized toward an effector function. Therefore, our data on pancreatic TIL support previous reports that TIL from GI tumors resemble the activated state of TIL derived from melanomas [28].

We found that the inhibitory molecule PD-1 was expressed at substantial levels on minimally cultured pancreatic TIL, within the range of PD-1 expression reported on TIL from other GI tumors [28]. The presence of PD-1, which is thought to be preferentially expressed on a comprehensive repertoire of tumor-reactive TIL, confirmed that these T cells were antigen-experienced and demonstrated the need to isolate these TIL from potential sources of ligand inhibition [23, 30]. Moreover, we were able to capitalize on the surface expression of PD-1 through the addition of a blocking antibody to the TIL culture and observed a substantial increase in the expansion of TIL. This data suggest that disrupting PD-1 signaling during ex vivo TIL culture may selectively expand tumor-reactive T cells, but this requires further investigation before any conclusions can be drawn. Additionally, PD-1 expression on TIL is known to be transient during culture with IL-2, as we also observed (Fig. 2b), potentially highlighting the limited window to employ this strategy for augmenting TIL growth [30].

We also investigated the effects of an agonistic 4-1BB antibody on its ability to improve the yield of pancreatic TIL as we have previously observed this in TIL derived from melanoma [39]. TIL expanded in media supplemented with anti-4-1BB demonstrated a significant increase in the absolute number of TIL produced on a per fragment basis. Furthermore, these TIL were predominantly CD8⁺ T cells with a 32-fold difference in yield of this population compared to control cultures. As 4-1BB positive TIL are thought to correspond to those T cells undergoing recent TCR engagement, it is possible that this selective expansion of CD8⁺ lymphocytes represents the population of tumor resident TIL specific for expressed tumor antigens on the surface of pancreatic adenocarcinoma [28].

It is widely accepted that CD8⁺ T cells are instrumental for an anti-tumor response, but the involvement of CD4⁺ T cells is far less clear and potentially disruptive [31, 40, 41]. In one murine study, the presence of TNF? and IL-17-producing CD4⁺ T cells in pancreatic cancer was associated with relatively aggressive disease [42]. For this reason, we chose to isolate the CD8⁺ pancreatic TIL to use as effectors in tumor reactivity assays. A major difficulty for the translation of TIL therapy into pancreatic patients is the lack of autologous tumor for the evaluation of tumor-specific reactivity. This required a reliance on HLA-matched pancreatic tumor lines, known for their relative lack of shared antigens, as TIL targets [35]. Importantly, we were able to demonstrate a significant and specific, HLA restricted immune response to matched tumor targets using this strategy. Previous studies have shown that an immune response toward shared pancreatic cancer antigens can be elicited, and our results confirm that TIL expanded from pancreatic tumors are functional when re-stimulated with HLA-matched tumor targets [7, 13, 43–48]. A recent study demonstrated that TIL expanded from pancreatic tumors recognized shared pancreatic tumor antigens, including NY-ESO-1, survivin, and mesothelin [36]. While reactivity to shared antigens can be measured in a subset of pancreatic TIL samples, it has been shown that patient with metastatic GI malignancies have unique mutations, and anti-tumor T cell responses are targeted to neoantigens specific to each individual patient [35]. Moving forward, it will be important to develop effective methods to establish autologous tumor targets. This would also allow for more complete evaluation of the TIL compartment, including CD4⁺ TIL, which have been previously demonstrated to recognize autologous tumor [41].