No Cell Left Unturned: Intraductal Papillary Mucinous Neoplasm Heterogeneity

In this issue of Clinical Cancer Research, Bernard and colleagues (1) performed single-cell transcriptomic analysis of low-grade, high-grade, and pancreatic ductal adenocarcinoma (PDAC) subtypes and reported the heterogeneity of the epithelial, immune, and stromal cells of intraductal papillary mucinous cystic neoplasms (IPMN), a precursor to PDAC. In their study, they analyzed 5,403 single cells from 6 patients, 2 with low-grade dysplasia (LGD-IPMN), 2 with high-grade dysplasia (HGD-IPMN), and 2 with invasive PDAC. Previous single-cell RNA sequencing (RNA-seq) work in the autochthonous pancreatic mouse model had demonstrated heterogeneity in tumor cells in advanced disease (2), and the current work now expands our understanding of the evolution of human PDAC heterogeneity during malignant progression of IPMN to invasive cancer. Figure 1 summarizes the changes in cell types from IPMN to PDAC. The authors find that there is loss of expression of tumor suppressor genes, including RAP1GAP, between LGD-IPMN compared with HGD-IPMN and PDAC. Core signaling pathways upregulated in HGD-IPMN included those involving integrins, small GTPases, Wnt/B-catenin, axonal guidance, apoptosis and G₁–S-phase regulation. PDAC lesions had additional enrichment of genes involved with the DNA damage response, TGFβ signaling and SAPK/JNK signaling. Comparison of LGD-IPMN, HGD-IPMN, and PDAC epithelial cells revealed a relative enrichment of G₂–M and S-phase proliferative cells with more advanced disease. Notably, there was a subset of LGD-IPMN epithelial cells, which demonstrated increased proliferative markers that may be associated with increased risk of malignant progression; however, the clinical relevance of this finding remains to be seen in a larger independent cohort of IPMN samples.

In addition to the heterogeneity seen in cancer cells, there is also great interest in understanding the associated immune response seen in PDAC development. The most common PDAC precursor lesion is pancreatic intraepithelial neoplasia (PanIN), which is characterized by an immunosuppressive environment with the most plentiful immune cells being myeloid-derived suppressor cells (MDSC), regulatory T cells (Treg), tumor-associated macrophages (TAM), and γδ-T cells (3). Jang and colleagues have also demonstrated that Tregs promote tumor growth in pancreatic cancer with an associated lack of cytotoxic CD8⁺ T-cell activation through direct interaction with CD11c⁺ dendritic cells (4). Altogether, these features are believed to preclude efficacy of immune checkpoint blockade. Although work has been done on understanding the immune response in PanIN to PDAC progression, there has been relatively little work done in understanding these features in IPMN to PDAC. Here, the authors have identified seven unique clusters of stromal cells, five immune subpopulations, and two fibroblast populations. Unlike the PanIN-PDAC sequence, which lose CD8⁺ T cells at early stages, IPMNs have a high proportion of cytotoxic CD8⁺ T cells in LGD-IPMNs when compared with more advanced lesions. Interestingly, MDSCs are virtually absent from LGD-IPMN and HGD-IPMN, although they are present in PDAC as has been previously reported. LGD-IPMN also have activated CD4⁺ T cells, which persist through the HGD stage. Furthermore, IPMNs show a unique subclass of myeloid cells—DC2-type dendritic cells, which are characterized by CD1c, THBD and FCER1a, and are present in both LGD-IPMN and HGD-IPMN. The presence of cytotoxic CD8⁺ T cells, activated CD4⁺ T cells, and dendritic cells in the microenvironment of IPMNs is clearly distinct from that of PanIN lesions. Accordingly, harnessing this knowledge may provide a unique opportunity for therapeutic interventions for a subset of patients. Most significantly, the profile of immune cells in IPMN suggests that this subset of pancreatic cancer precursor lesions may be responsive to immunotherapy or vaccine therapy as a chemoprevention strategy.

Finally, the dense desmoplastic stromal response indicative of pancreatic cancer has recently been demonstrated to be heterogeneous between tumors. Ohlund and colleagues previously described two populations of cancer-associated fibroblasts (CAF) in pancreatic cancer (5): myofibroblasts (myCAF) that are characterized as having high expression of smooth muscle actin (SMA) and low expression of IL6, and inflammatory CAFs (iCAF) that are characterized by low expression of SMA and high expression of IL6. RNA-seq analysis demonstrated that these CAFs have distinct transcriptional profiles with Acta2, Ctgf, and Col1a1 upregulated in myCAFs and Il6, Il11, Lif, Cxcl1, and Cxcl2 upregulated in iCAFs. Of the seven unique stromal clusters identified, Bernard and colleagues found a population of myCAFs and iCAFs. The myCAFs were clustered to LGD and HGD lesions and were characterized by reduced expression of CXCL12, ACTA2, and COL3A1, whereas iCAFs were exclusively found in the PDAC subset and were characterized by upregulation of FAP, ACTA2, COL3A1, and CXCL12, which is associated with metastatic spread in many solid cancers. Their findings suggest that similar to pancreatic cancer, iCAFs have a protumorigenic role in advanced lesions.

This study characterizes the single-cell phenotypes of epithelial, immune, and stromal compartments in IPMN, an important pancreatic cancer precursor lesion that accounts for approximately 10% of all PDAC. Prior work showed that the stroma has a complex role in PDAC pathogenesis, with the diversity of fibroblastic and immune cells having both tumor-promoting and -inhibiting effects depending on their activation state and the stage of tumorigenesis. In advanced tumors, the stroma contributes to the resistance of PDAC to conventional chemotherapy and immunotherapy. In this study, Bernard and his group have classified a unique immune infiltrate that is not characteristic of classical PDAC. Their findings open the possibility for targeting dendritic and T cells' cross-talk, which has previously not been possible in other PDAC precursors. While one limitation of this study is the generalizability of these findings given the small number of samples analyzed, it provides an important starting point for studies focused on the understanding the molecular features of IPMN. Future validation studies in additional human IPMN and PDAC tumors combined with new genetically engineered mouse models of IPMN will certainly build upon this work to understand the factors that are involved in the progression of the different histologic subtypes of IPMNs to PDAC. Moreover, the application of RNA and protein-based in situ methods to determine the spatial context of these cell types in IPMN and PDAC will likely uncover new relationships of epithelial, immune, and stromal cells that cannot be resolved with single-cell RNA-seq. In summary, this study carries great impact in the field as it establishes that there are marked differences in the tumor microenvironment between pancreatic cancer precursor lesions, and it points to the potential of using immunotherapeutic approaches to target IPMN.

D.T. Ting reports receiving commercial research grants from ACD-Biotechne, holds ownership interest in PanTher Therapeutics, and is a consultant/advisory board member for Merrimack Pharmaceuticals and EMD Millipore-Sigma. No potential conflicts of interest were disclosed by the other authors.

This work is supported by NIH grants T32DK007191-43S1 (to Y.G. Hernandez-Barco), P01 CA117969, and R01 CA215498-01 (to N. Bardeesy); the Granara-Skerry Trust, the Begg Family, SU2C-Lustgarten 2015-002, the NSF PHY-1549535, the Warshaw Institute for Pancreatic Cancer (to D.T. Ting); and the Linda J. Verville Foundation (to N. Bardeesy and D.T. Ting).