Dynamic profiling of immune microenvironment during pancreatic cancer development suggests early intervention and combination strategy of immunotherapy

Abstract

Background: Pancreatic ductal adenocarcinoma (PDAC) has little response to immune checkpoint inhibitors. An in-depth understanding of the immune microenvironment from a comprehensive and dynamic perspective is critical to generate effective therapeutic strategies for PDAC.

Methods: Using mass cytometry and immunohistochemistry, we explored the dynamic changes of tumor-infiltrating immune cells during the development of PDAC in a genetically engineered mouse model (Kras^G12D/+; Trp53^R172H/+; Pdx1-cre) and human specimens. PD-L1^-/- mice were crossed with Kras^G12D/+; TgfβR2^flox/flox; Ptf1a-cre mice to achieve early depletion of PD-L1 in pancreatic cancer. Combination therapy of Arginase-1 (Arg-1) inhibitor and anti-PD-1 mAb was validated in syngeneic mouse models.

Findings: Two different stages of immunosuppression with unique features were observed in both mouse model and human specimens. Early stage of immunosuppression featured highly abundant Tregs during acinar-to-ductal metaplasia, despite of a prominent and continuous presence of effector lymphocytes. The differentiation/activation branch of Ly-6C⁺ monocytes changed from a BST2⁺/MHC-II⁺ phenotype to an Arg-1⁺ phenotype over time during PDAC development. The late stage of immunosuppression thus featured the presence of a large number of myeloid suppressive cells together with a significant reduction of effector lymphocytes. Removal of PD-L1 from the beginning efficiently triggered anti-tumor immunity and significantly prolonged survival in PDAC-developing mice. Targeting Arg1⁺ macrophages with an Arg-1 inhibitor synergized with anti-PD-1 immunotherapy and led to PDAC-specific immune memory.

Interpretation: By demonstrating the coevolution of histopathology and immunology in PDAC, this study highlights the necessity and value of early intervention and combinational approach in leveraging immunotherapy to treat pancreatic cancer.

Funding: A full list of funding bodies that contributed to this study can be found in the Acknowledgements section.

Keywords: Immunotherapy; Mass cytometry; Pancreatic ductal adenocarcinoma; Tumor microenvironment; Tumor-infiltrating immune cells.

Figure 1

Intratumoral immune cell profiling by CyTOF. (a) Breeding stragety of KPC mice. (b) Five stages of PDAC development. Precancerous stages were identified by histopathology and tumor stages were divided by whether metastasis exists or not. Fresh samples were stained and detected by CyTOF. (c, d) tSNE analysis of the total immune cell populations according to some specific markers. (e, f) The immune contexture in the five developmental stages. (g) Percentage of each immune cell populations (mean percent ± SD of total immune cells), * p < 0.05, ** p < 0.01, *** p < 0.001, one-way ANOVA multiple comparisons. ES Tumor, early-stage tumor; LS Tumor, late-stage tumor.

Figure 2

Characterization of intratumoral lymphocytes. (a) tSNE plots of CD3⁺ and CD19⁺ lymphoid cell populations. (b) tSNE plots displaying dynamic change of CD8⁺ T cells, CD4⁺ non-Tregs and Tregs in the five developmental stages. (c) tSNE plots identifying T cells with markers of CD8, CD4, CD25 and FoxP3. (d, e) Percentage of CD8⁺ and CD4⁺ T cells in clusters with significant change during PDAC development (mean percent ± SD of total T cells), * p < 0.05, ** p < 0.01, *** p < 0.001, one-way ANOVA multiple comparisons. (f) tSNE plots displaying dynamic change of B cells in the five developmental stages. (g) tSNE plots identifying B cells with markers of IgD and CD43. (h) Percentage of CD43⁺IgD⁻-immature and CD43⁻IgD⁺-mature B cell clusters during PDAC development (mean percent ± SD of total CD19+ B cells), * p < 0.05, ** p < 0.01, *** p < 0.001, one-way ANOVA, multiple comparisons.

Figure 3

Characterization of intratumoral myeloid cells. (a) tSNE plots of CD11b⁺ myeloid cell populations. (b, d) Heatmap of DC and MDSC clusters with normalized expression of selected markers. (c, e) Percentage of DCs and MDSCs in each cluster (mean percent ± SD of total myeloid cells), * p < 0.05, ** p < 0.01, *** p < 0.001, one-way ANOVA multiple comparisons. (f) tSNE plots identifying distinguishable macrophage subtypes by markers of F4/80, BST2, MHC-II and Arg-1. (g) Trajectory analysis of the time course of monocytes/macrophages differentiation/activation. (h) tSNE plots of Arg-1⁺ and MHC-II⁺/BST2⁺ macrophages in the five developmental stages. (i) Percentage of MHC-II⁺/BST2⁺ macrophage AND Arg-1⁺ macrophage clusters (*p < 0.05, ** p < 0.01, *** p < 0.001, one-way ANOVA multiple comparisons.

Figure 4

Verification of the immunotype in mouse and patient samples by IHC staining. (a, c) Representative IHC images of CD206+ M2 macrophages (a) and FoxP3+ Tregs (c) in mouse PDAC samples (400×). (b, d) Statistics of the number of M2 macrophages (b) and Tregs (d) in a 400× field, each dot represents the mean of three random field in one sample. (e) Representative IHC images of CD3, CD11b, CD11c, CD19, CD163 and FoxP3 staining using human PDAC sample (400×, n = 8 in each group). (f) Statistics of the number of T cells, myeloid cells, DCs, B cells, M2 macrophages and Tregs in a 400× field, each dot represents the mean of three random field in one sample. n = 8 in each group. Donor pancreas: pancreas from organ donors, RPC: resectable pancreatic cancer, MPC: metastatic pancreatic cancer. For (b), (d) and (f), bar indicates mean ± SD, * p < 0.05, ** p < 0.01, *** p < 0.001, one-way ANOVA, multiple comparisons.

Figure 5

Early intervention by PD-L1 knockout activates tumor immunity and prolonges mouse survival. (a) Mouse survival rates are shown, * p < 0.05, ** p < 0.01, Log-rank test. (b) Days for mice to develop a palpable pancreas tumor. N.S., no significance. (c) Upper: representative images of hematoxylin-eosin (HE) staining for 5-week precancerous pancreas and 8-week tumor tissues. Lower: quantification for different morphologies (n = 4 for KTC and n = 5 for KTC;PD-L1^−/–). (d) Representative flow cytometry dot plots of T cell and B cell in KTC;PD-L1^WT and KTC;PD-L1^−/− mice. (e) Statistics of percentage of T and B cell (n = 3 for KTC and n = 4 for KTC;PD-L1^−/–), ** p < 0.01, student t test. (f) Heatmap of CD4⁺ T cell, CD8⁺ T cell and B cell clusters in CyTOF analysis, Each column refers to one cluster and each row refers to one marker. (g) Upper left: representative flow cytometry dot plots of macrophages; Upper right: corresponding statistics, ** p < 0.01, student t test; Lower: tSNE plots of macrophages in each group. (h) Upper: representative flow cytometry histogram of TAM-expressed CD1d, CD40, CD206 and MHC-II. Lower: statistics of mean fluorescence intensity, * p < 0.05, ** p < 0.01, student t test.

Figure 6

Arginase-1 inhibitor synergizes anti-PD-1 antibody therapy. (a) Tumor growth curve of KPC syngeneic model is shown as mean±SEM of 6,7 mice per group. * p < 0.05, ** p < 0.01. (b) Mouse survival rates are shown, * p < 0.05, ** p < 0.01, Log-rank test. (c) Mice engrafted with KPC tumor cells were treated with anti-PD-1 mAb w/o Arg-1inhibitor combination therapy, 1 mouse from anti-PD-1 mAb treated group and 3 mice from combined group achieved tumor regression. After 2 months, the KPC survived mice were rechallenged subcutaneously with B16F10 (right flank) and KPC (left flank) tumors. Naïve mice were also inoculated in the same manner as control. (d) Rechallenged tumor size were measured and are shown as mean±SEM. N.S. no significance, **** p < 0.0001, student t tests. (e) Representative IHC images of Ki-67 and Cleaved-caspase3 in each group (400×). (f) Statistics of the number of Ki-67⁺ and Cleaved-caspase3⁺ cells in a 400× field. (g) tSNE plots of intratumoral immune cells in different groups. CD11b and Ly-6C were used as markers of MDSC; F4/80 and PD-L1 were used to discribe PD-L1⁺ macrophage; CD4 and CD25 represent Treg; (h) Boxplots shows statistics of corresponding immune cell types. * p < 0.05, ** p < 0.01, student t tests. (i, j) Heatmap of CD8⁺ T cell clusters, PD-L1⁻ and PD-L1⁺ macrophage clusters. Each column refers to one cluster and each row refers to one marker. Red means high expression while blue means low expression. Mac: Macrophage.