Pancreatic cancer has the worst prognosis and lowest survival rate among all types of cancers and thus, there exists a strong need for novel therapeutic strategies. Chimeric antigen receptor (CAR)-modified T cells present a new potential option after successful FDA-approval in hematologic malignancies, however, current CAR T cell clinical trials in pancreatic cancer failed to improve survival and were unable to demonstrate any significant response. The physical and environmental barriers created by the distinct tumor microenvironment (TME) as a result of the desmoplastic reaction in pancreatic cancer present major hurdles for CAR T cells as a viable therapeutic option in this tumor entity. Cancer cells and cancer-associated fibroblasts express extracellular matrix molecules, enzymes, and growth factors, which can attenuate CAR T cell infiltration and efficacy. Recent efforts demonstrate a niche shift where targeting the TME along CAR T cell therapy is believed or hoped to provide a substantial clinical added value to improve overall survival. This review summarizes therapeutic approaches targeting the TME and their effect on CAR T cells as well as their outcome in preclinical and clinical trials in pancreatic cancer.

A domain that is often neglected in the assessment of chimeric antigen receptor (CAR) functionality is the extracellular spacer module. However, several studies have elucidated that membrane proximal epitopes are best targeted through CARs comprising long spacers, while short spacer CARs exhibit highest activity on distal epitopes. This finding can be explained by the requirement to have an optimal distance between the effector T cell and target cell. Commonly used long spacer domains are the CH2-CH3 domains of IgG molecules. However, CARs containing these spacers generally show inferior in vivo efficacy in mouse models compared to their observed in vitro activity, which is linked to unspecific Fcγ-Receptor binding and can be abolished by mutating the respective regions. Here, we first assessed a CAR therapy targeting membrane proximal CD20 using such a modified long IgG1 spacer. However, despite these mutations, this construct failed to unfold its observed in vitro cytotoxic potential in an in vivo model, while a shorter but less structured CD8α spacer CAR showed complete tumor clearance. Given the shortage of well-described long spacer domains with a favorable functionality profile, we designed a novel class of CAR spacers with similar attributes to IgG spacers but without unspecific off-target binding, derived from the Sialic acid-binding immunoglobulin-type lectins (Siglecs). Of five constructs tested, a Siglec-4 derived spacer showed highest cytotoxic potential and similar performance to a CD8α spacer in a CD20 specific CAR setting. In a pancreatic ductal adenocarcinoma model, a Siglec-4 spacer CAR targeting a membrane proximal (TSPAN8) epitope was efficiently engaged in vitro, while a membrane distal (CD66c) epitope did not activate the T cell. Transfer of the TSPAN8 specific Siglec-4 spacer CAR to an in vivo setting maintained the excellent tumor killing characteristics being indistinguishable from a TSPAN8 CD8α spacer CAR while outperforming an IgG4 long spacer CAR and, at the same time, showing an advantageous central memory CAR T cell phenotype with lower release of inflammatory cytokines. In summary, we developed a novel spacer that combines cytotoxic potential with an advantageous T cell and cytokine release phenotype, which make this an interesting candidate for future clinical applications.