Multi-omic profiling of lung and liver tumor microenvironments of metastatic pancreatic cancer reveals site-specific immune regulatory pathways

Abstract

Background: The majority of pancreatic ductal adenocarcinomas (PDAC) are diagnosed at the metastatic stage, and standard therapies have limited activity with a dismal 5-year survival rate of only 8%. The liver and lung are the most common sites of PDAC metastasis, and each have been differentially associated with prognoses and responses to systemic therapies. A deeper understanding of the molecular and cellular landscape within the tumor microenvironment (TME) metastasis at these different sites is critical to informing future therapeutic strategies against metastatic PDAC.

Results: By leveraging combined mass cytometry, immunohistochemistry, and RNA sequencing, we identify key regulatory pathways that distinguish the liver and lung TMEs in a preclinical mouse model of metastatic PDAC. We demonstrate that the lung TME generally exhibits higher levels of immune infiltration, immune activation, and pro-immune signaling pathways, whereas multiple immune-suppressive pathways are emphasized in the liver TME. We then perform further validation of these preclinical findings in paired human lung and liver metastatic samples using immunohistochemistry from PDAC rapid autopsy specimens. Finally, in silico validation with transfer learning between our mouse model and TCGA datasets further demonstrates that many of the site-associated features are detectable even in the context of different primary tumors.

Conclusions: Determining the distinctive immune-suppressive features in multiple liver and lung TME datasets provides further insight into the tissue specificity of molecular and cellular pathways, suggesting a potential mechanism underlying the discordant clinical responses that are often observed in metastatic diseases.

Fig. 1

Global immune profiling of metastatic liver and lung TME with CyTOF. a Mice were injected with KrasG12D and TP53R172H-driven pancreatic cancer cells (KPC) into the liver by the hemispleen method or into the lung by intravenous tail vein injection. Normal lung and livers (day 0, “D0”) and day 21 (“D21”) from the day of injection were harvested and barcoded with a CD45 antibody conjugated to a unique metal. One sample from each group was then combined into a 4-plex batch to be stained with the full CyTOF panel (Table S1). b Heatmap of normalized marker expression for FlowSOM clustering of the dataset and c UMAP visualization of the clusters are shown. Two thousand events per sample are represented. Immune cell type proportions as a percentage of total CD45+ cells from each of the four groups are shown as d stacked barplots for the entire group and e mean values for each group as radar plots (left) and relative values scaled for each immune cell subtype for all mice as a heatmap (right). Groups are annotated by color within the radar plot or as horizontal bars above the heatmap. FDR-adjusted p values compared using edgeR for KPC-bearing vs. normal conditions in the liver (LV_KpcNl) and lung (LG_KpcNl) as well as lung vs. liver in KPC-bearing (KPC_LgLv) and normal (NL_LgLv) conditions are annotated as an adjacent heatmap to the right

Fig. 2

T cell profiling of metastatic liver and lung TME with CyTOF. a CD3+ subset of the dataset was re-clustered using both subtyping and functional markers. Select T cell clusters are represented with mean values for each group as radar plots (top). Separate radar plots with different axis ranges are used to facilitate the visualization of low-level subtypes. Relative values scaled for each immune cell subtype for all mice as a heatmap (bottom). Groups are annotated by color within the radar plot or as horizontal bars above the heatmap. FDR-adjusted p values compared using edgeR for KPC-bearing vs. normal conditions in the liver (LV_KpcNl) and lung (LG_KpcNl) as well as lung vs. liver in KPC-bearing (KPC_LgLv) and normal (NL_LgLv) conditions are annotated as an adjacent heatmap to the right. b Proportion of LAG3 positive CD8 T cells (top) and NK cells (bottom) as assessed by fluorescent flow cytometry comparing between KPC-bearing liver and lung is shown (n = 6). Two-tailed t test, ***p < 0.005. c Sample clonality values derived from TCRseq analysis of T cells within normal (gray) livers and lungs and KPC-bearing (red) livers and lungs are compared. Wilcoxon-Mann-Whitney, *p < 0.05; ***< 0.005

Fig. 3

Approach for non-immune compartment profiling of metastatic liver and lung TME by RNAseq analysis. a KPC-bearing lung and liver are analyzed along with KPC cells from in vitro culture in 2D, normal lung, and normal liver as controls. All tissue samples are enzymatically dissociated into single cells. KPC-bearing liver and lung samples are further processed using negative magnetic bead selection to obtain the non-immune cells. b PCA plotting of the RNAseq dataset shows how individual samples cluster. c Expression of key genes identifying cells of pancreatic origin (KPC), tissue-specific parenchyma, and immune cells are shown in a heatmap to demonstrate the quality of the sample preparation process

Fig. 4

Heatmap of expression profiles for a select set of chemokine and immune regulatory genes shown for every sample. All gene expression differential analyses were performed based on the negative binomial distribution using DESeq2. FDR-adjusted p values *< 0.05, **< 0.01, and ***< 0.005 for comparison between the liver and lung TME sites are shown next to each gene name

Fig. 5

Immune cell types in lung and liver TCGA samples deconvolved by MIXTURE algorithm. Both tumor and adjacent normal samples in the TCGA databases for liver hepatocellular carcinoma (LIHC; normal n = 50, tumor n = 366) and lung adenocarcinoma and squamous cell carcinomas (LUAD and LUSC; normal n = 105, tumor n = 992) are assessed for computed relative abundances of immune cell types (coefficients). Summarizing boxplots are shown. FDR-adjusted p values for Tukey HSD *< 0.1, **< 0.01, and ***< 0.001

Fig. 6

Differential expression of select gene sets comparing lung and liver TCGA datasets. a Volcano plot of − log10 FDR-adjusted p values by DESeq2 are shown over log2-fold differences between lung and liver sites for four sets of genes (as analyzed for mouse and human metastatic pancreatic cancer samples). Gene set analysis was performed using the Wilcoxon Gene Set Test (GST) (FDR-adjusted p values noted). b Barcode plots representing the results of the gene set enrichment analyses