CD137 agonism enhances anti-PD1 induced activation of expanded CD8+ T cell clones in a neoadjuvant pancreatic cancer clinical trial

Abstract

Successful pancreatic ductal adenocarcinoma (PDAC) immunotherapy requires therapeutic combinations that induce quality T cells. Tumor microenvironment (TME) analysis following therapeutic interventions can identify response mechanisms, informing design of effective combinations. We provide a reference single-cell dataset from tumor-infiltrating leukocytes (TILs) from a human neoadjuvant clinical trial comparing the granulocyte-macrophage colony-stimulating factor (GM-CSF)-secreting allogeneic PDAC vaccine GVAX alone, in combination with anti-PD1 or with both anti-PD1 and CD137 agonist. Treatment with GVAX and anti-PD-1 led to increased CD8⁺ T cell activation and expression of cytoskeletal and extracellular matrix (ECM)-interacting components. Addition of CD137 agonist increased abundance of clonally expanded CD8⁺ T cells and increased immunosuppressive TREM2 signaling in tumor associated macrophages (TAMs), identified by comparison of ligand-receptor networks, corresponding to changes in metabolism and ECM interactions. These findings associate therapy with GVAX, anti-PD1, and CD137 agonist with enhanced CD8⁺ T cell function while inducing alternative immunosuppressive pathways in patients with PDAC.

Keywords: Health sciences; Internal medicine; Medical specialty; Medicine; Oncology.

Graphical abstract

Figure 1

Summary of single-cell RNA-sequencing of PDAC-infiltrating leukocytes (A) UMAP depicting Leiden clustering of the dataset. (B) Dot plot showing genes and their expression levels used for cell annotations across clusters in the dataset. (C) UMAP summarizing the cell types annotated within the dataset based upon differential expression of marker genes between subclusters observed in each of the major clusters from C, displayed in Figure S1. (D) Barplots showing cellular proportions across treatment arms. Panels C and D were generated scatterHatch colors and patterns.

Figure 2

Differential expression of CD8⁺ T cells across treatment arms (A) Heatmap of the normalized Variance Stabilizing Transformation (VST) expression of differentially expressed genes from the following comparisons indicated on the right of heatmap: Arm A vs. B, Arm B vs. C, Arm A vs. C, and genes present in both the Arm AvB and AvC comparisons. Genes are grouped into the bins from which comparison they arose. (B) Gene set enrichment barplot from GOMF of Arms B vs. C. Positive normalized enrichment indicates higher presence in Arm B samples, while negative normalized enrichment scores indicate higher presence in Arm C samples. (C) Violin plot of percent CA surrounding each LA by treatment arm. Overlayed points represent each measured LA. (D) Violin plot of percent CA of the entire tissue section for each patient. Overlayed points represent each measured slide.

Figure 3

Addition of anti-PD1 and CD137 agonist to Cy/GVAX is associated with hyperabundant and clonal tumor-infiltrating CD8⁺ effector T cells with improved function (A) UMAP of the major T cell cluster from scRNA-seq (subset from Figure 2A). (B) UMAP of the cells from panel A colored by T cell expansion status calculated from matched scTCR-seq. (C) Stacked barplots showing the proportions of T cells from each Arm belonging to hyperexpanded, large, medium, small, or singlet clones. Overlayed numbers are the proportion of T cells belonging to the clone types. (D and E) Boxplots showing the number of expanded T cell clones (D) and contracted T cell clones (E) in the peripheral blood after therapy by Arm. Boxplots display the 25^th percentile (top), median (middle), and 75^th percentile (bottom) T cell clone numbers within each Arm with whiskers extending to 1.5 times the interquartile range and overlayed with individual sample measurements as points. Wilcoxon test p-values are displayed above. (F–H) Volcano plots of differential expression analyses using a MAST test of expanded CD8⁺ T cells comparing Arms A and B (F), Arms A and C (G), and Arms B and C (H). LFC: log fold change, Padj: Bonferroni adjusted p value.

Figure 4

Combination immunotherapy approaches elicit alterations in the interaction of CD8⁺ T cells and TAMs with other cells in the TME (A) Workflow of cellular communication inference using Domino and comparing the proportions of patients with active signal receipt between treatment groups. Briefly, the Domino method infers active signaling based on identifying correlation of receptor expression with downstream transcription factor regulons and coordinated gene expression of theses receptors with cognate ligands expressed by other cells, indicating a candidate cell-cell signaling event. Inferred interactions are tallied across patients in each trial arm, and a Fisher’s exact test is employed to rank the extent to which ligand-receptor cellular interactions differ between treatment arms. (B) Ranking of receptors in CD8⁺ T cells prioritized from our differential ligand-receptor analysis between patients from treatment arm A relative to patients in both treatment arms B and C. The values shaded blocks represent p-values from a Fisher’s Exact Test comparing the frequency of patients with active signal receipt as determined using Domino, with darker color denoting lower p-values. The colored bars show the frequency of active signal receipt among all patients (black), patients in arm A (light red), or arms B and C (teal) tallied as described in A. (C–F) Circos plots are used to show the ligands in the cell types found to signal with the receptors in CD8⁺ T cells in trial arm A from the differential ligand-receptor analysis. The width of chords corresponds to the mean normalized expression of the ligand gene within a cell type in the data for all patients from trial arm A, representative of the putative strength of signaling from that ligand in the given cell type to the receptor in CD8⁺ T cells. Because each receptor may have multiple ligands, chords and the inner annotation are colored based on all candidate ligand genes that can signal to a receptor. The circle is divided into cell types, indicated by the labels for each outer arc. Circos plots are divided by each candidate receptor prioritized from the differential analysis in B, with ligands associated with (C) CCR7, (D) IL7R, (E) CD96, and (F) TNFRSF4. (G) Ranked signal receipt in TAMs between the patients in treatment arms A and B relative to patients from treatment arm C. The colored bars show the frequency of signaling among all patients (black), patients from arms A and B (purple), or patients from arm C (light blue). (H) Circos plot from the single-cell data for patients in trial arm C of ligands associated with TREM2 and with (I) PECAM1.

Figure 5

Differential expression of TAMs across treatment arms (A) Heatmap of normalized VST expression of differentially expressed genes from the following comparisons indicated on the right of heatmap: Arm A vs. B, Arm B vs. C, Arm A vs. C, and genes present in both the Arm AvB and AvC comparisons. Genes are grouped into the bins from which comparison they arose. (B) Gene set enrichment barplot comparing Arms A and B. Positive normalized enrichment indicates higher presence in Arm A samples, negative normalized enrichment scores indicate higher presence in Arm B samples. (C) Gene set enrichment barplot comparing Arms B and C. Positive normalized enrichment indicates higher presence in Arm B samples, negative normalized enrichment scores indicate higher presence in Arm C samples.