Exploring the tumor micro-environment in primary and metastatic tumors of different ovarian cancer histotypes

Abstract

Ovarian cancer is a highly heterogeneous disease consisting of at least five different histological subtypes with varying clinical features, cells of origin, molecular composition, risk factors, and treatments. While most single-cell studies have focused on High grade serous ovarian cancer, a comprehensive landscape of the constituent cell types and their interactions within the tumor microenvironment are yet to be established in the different ovarian cancer histotypes. Further characterization of tumor progression, metastasis, and various histotypes are also needed to connect molecular signatures to pathological grading for personalized diagnosis and tailored treatment. In this study, we leveraged high-resolution single-cell RNA sequencing technology to elucidate the cellular compositions on 21 solid tumor samples collected from 12 patients with six ovarian cancer histotypes and both primary (ovaries) and metastatic (omentum, rectum) sites. The diverse collection allowed us to deconstruct the histotypes and tumor site-specific expression patterns of cells in the tumor, and identify key marker genes and ligand-receptor pairs that are active in the ovarian tumor microenvironment. Our findings can be used in improving precision disease stratification and optimizing treatment options.

Keywords: cancer associated fibroblasts; cell type annotation; ligand-receptor analysis; metastatic tumor site; ovarian cancer histotypes; primary tumor site; single-cell tumor profiling; tumor micro-environment.

FIGURE 1

Experiment design and 2D reduced representation of all cells included in the study, annotated by major cell lineage, predicted cancer subtype, and cell cycle phase. (A) Profiling ovarian cancer tumor samples of different using droplet single cell RNA-seq. (B) All cell types projected on UMAP divided by Epithelia, Immune, and Stroma subpopulations. (C) Predicted cancer subtype projected on UMAP. (D) Cell cycle assignment projected on UMAP.

FIGURE 2

Cellular sub-types in the Immune, Epithelia, and Stroma. (A–C) Subclusters of major immune cell types: T cells, B cells, and macrophages, respectively. (D,E) Subcluster annotation for epithelial and embryonic stem cells (ESC), respectively. (F–H) Subcluster annotation for major cell-types in the stroma: fibroblasts, mesenchymal stem cells, and endothelial cells, respectively.

FIGURE 3

Cell composition by tumor site, T cell infiltration, and histotypes; fractions of immune, stromal, and epithelial cells are explored using immunohistochemistry. (A) Heatmap of major cell type composition (left) and sub cell type (right) for all patient samples. The column z-scores are calculated from cellular compositional percentages within each sample; the rows are split by site and T cell infiltration status. (B) Heatmap of cell type subclusters composition percentage for all patient samples. The values are column z-scores normalizing the percentage and the rows are split by histotypes. (C) Dot plot of histotype markers expression in epithelial (EP) and embryonic stem (ES) cells. The expression in the dot plot is the averaged scaled log normalized TP10k value.

FIGURE 4

Ligand-receptor (LR) interactions predicted by CellPhoneDB using a customized cancer database. (A) Total number of interactions between all cell subtypes. (B) Counts of significant Ligand-receptor pairs for all cell type subclusters stratified by sample, the columns are grouped by the cell lineage of the first interactor.