Abstract P5-06-02: A gene expression meta-analysis identifies microenvironment differences in cellular composition and cell-cell interactions associated with breast cancer invasion

Abstract

Abstract Background: The invasion of malignant cells into surrounding stroma is the histological hallmark differentiating invasive ductal carcinoma (IDC) from ductal carcinoma in situ (DCIS). The associated variations in the cellular composition and interactions in the tumor microenvironment, and how those mediate invasion are poorly understood. Systematic analyses of gene expression differences between DCIS and IDC are rare, confounded by cellular heterogeneity, and limited to few cases unable to distinguish stromal from epithelial contributions or compare pure DCIS from those synchronous with IDC. To address these problems we have assembled multiple datasets from publicly available gene expression studies of microdissected DCIS and IDC samples. Methods: The transcriptome from 192 regions microdissected from 140 cases was assembled from 5 published studies and included DCIS (83 epithelial, 22 stromal) and IDC (67 epithelial, 20 stromal) cases. Differences in infiltration of 10 immune cell types were measured using expression signatures. Expression subtypes were determined from the PAM50 signatures of epithelial regions using the genefu package. Gene Set Enrichment Analysis (GSEA) was performed using the gene sets in the Hallmark and Reactome collections. Cell-cell interactions between stromal and epithelial regions were measured using the co-expression of 1175 curated pairs of ligand-receptor genes and compared between DCIS and IDC using a permutation test. The corresponding interacting cell types were identified using RNA sequencing of 5444 single nuclei extracted from fine needle aspirates of two excised DCIS specimens. Results: Of the measured immune cell types, B cells showed a progressive depletion from normal to DCIS to IDC stroma. Pure IDC regions were significantly less likely to be the Luminal A subtype compared to synchronous IDC/DCIS and pure DCIS (p=0.0008). IFN signaling was higher in IDC epithelium suggesting possible differences in immune visibility between these two diseases. Higher calcium and glutamate signaling is observed in the DCIS stroma compared to IDC, calcium signaling has been implicated as both essential to normal breast function and as a positive effector of IDC proliferation. The expression-based measurement of stromal-epithelial interactions identified 99 and 115 ligand-receptor interactions enriched in DCIS and IDC, respectively (FDR<0.1). To precisely identify the corresponding interacting cell types, the expression profile of cells within the DCIS microenvironment was determined using single-nucleus RNA sequencing from two DCIS patients. The analysis identified 9 cell types, including luminal, basal, macrophage, adipocyte and endothelial, and 54% of the candidate cell-cell interactions could be mapped to at least one cell type pair. Based on these cell type mappings, interactions between luminal cells and fibroblasts were gained in IDC while those involving the vascular endothelium were lost, including interactions between CD24 and P-selectin, an interaction involved in leukocyte recruitment. Increased macrophage autocrine interactions were identified in both IDC stroma and epithelium through urokinase-urokinase receptor gene co-expression, an interaction previously associated with a transition to the M2 phenotype. Conclusions: The meta-analysis combined with novel computational methods, improves our ability to characterize the micro-environment of DCIS specimens, typically hard to study. Changes in cellular dynamics involving both immune and non-immune cell types suggest that mechanisms other than direct immune escape accompany progression. Citation Format: Adam Officer, Andre Dempsey, Farnaz Hasteh, Michal Slyper, Asa Segerstolpe, Joanna Klughammer, Judit Jane-Valbuena, Orit Rozenblatt-Rosen, Lyndsay Murrow, Zev Gartner, Aviv Regev, Christina Yau, Pablo Tamayo, Olivier Harismendy. A gene expression meta-analysis identifies microenvironment differences in cellular composition and cell-cell interactions associated with breast cancer invasion [abstract]. In: Proceedings of the 2021 San Antonio Breast Cancer Symposium; 2021 Dec 7-10; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2022;82(4 Suppl):Abstract nr P5-06-02.