NEW YORK – Researchers have mapped the ecosystem of hundreds of breast cancer tumors to find a complex landscape influenced by the tumors' genetic makeup.
By mapping the cell types in more than 480 tumor samples from the Molecular Taxonomy of Breast Cancer International Consortium (METABRIC) study, researchers from Cancer Research UK and elsewhere began to get a handle on the various types of cells found within breast tumors and how they interact. As they reported yesterday in Nature Cancer, the researchers found that features of these tumor ecosystems could be linked to patient prognosis.
"We've shown that the effects of mutations in cancer are far more wide-ranging than first thought," co-author Carlos Caldas from the Cancer Research UK Cambridge Institute said in a statement. "They affect how cancer cells interact with their neighbors and other types of cells, influencing the entire structure of the tumor."
Using imaging mass spectrometry, Caldas and his colleagues detected the presence and distribution of 37 proteins in 483 tumor samples from the METABRIC study at a sub-single-cell resolution. METABRIC study participants' tumors also underwent genomic characterization, including targeted sequencing of 173 breast cancer-linked genes and copy number, transcriptomic, and microRNA profiling.
Based on the expression of these proteins, the researchers determined the identity of the cells present and pieced together molecular tissue maps of the tumors. A clustering analysis broadly categorized the cells as tumor, stromal, and immune cells, and then as fibroblasts or myofibroblasts; as T cells, B cells, or macrophages; or as vascular smooth muscle cells.
A number of subtyping approaches exist for breast cancer, and the researchers noted that certain cell phenotypes were more prevalent among particular tumor subtypes. Luminal A tumors, for instance, were enriched for HR+ epithelial cells, while basal-like tumors were enriched for HR- Ki67+ cells, epithelial cells expressing basal cytokeratins, and a hypoxia-associated phenotype.
These various subtypes also had different cell-cell interactions, the researchers noted. For example, basal-like and integrative cluster 10 tumors, a type of subgroup, were marked by numerous homotypic relationships between epithelial and stromal cells.
Certain cell phenotypes could also be linked to somatic alterations in key breast cancer driver genes. ERBB2 gains were associated with a HER2+ cell phenotype, while TP53 mutations were associated with ER- basal cells, hypoxia-associated epithelial cells, and HR- Ki67+ epithelial cells, as expected. In addition, though, gains of CCND1 and loss of TUBD1 and ATM were linked to other cell phenotypes.
Other cells, namely epithelial cells that express carbonic anhydrase IX, a marker of hypoxia, were linked to CD274 gains and heterozygous deletion of B2M. Hypoxia has previously been linked to immune suppression, and as immune escape had been tied to resistance to immune checkpoint blockade therapy, the researchers noted that hypoxia markers could help identify patients with or who may develop resistance to those drugs.
Particular cell phenotypes were associated with patient prognosis. Cell phenotypes that expressed Ki67 and HER2 were, as expected, associated with poor outcomes, as was the cell phenotype indicating hypoxia and the presence of macrophages in the tumor microenvironment. At the same time, vascular smooth muscle cells were associated with a positive outcome.
These findings suggest that multidimensional tissue imaging combined with genomic data could help stratify patients and predict their prognosis. "At the moment, doctors only look for a few key markers to understand what type of breast cancer someone has," first author Raza Ali from the Cancer Research UK Cambridge Institute said in a statement. "But as we enter an era of personalized medicine, the more information we have about a patient's tumor, the more targeted and effective we can make their treatment."