(Boston)—Breast cancer (BC) is the most common cancer and the second leading cause of cancer death in women. It is a highly variable disease, defined as a malignancy of the epithelial ducts in breast tissue. Characterizing the vast heterogeneity within BC cells and their surrounding tumor microenvironment is crucial because this diversity is the primary reason for treatment resistance, disease progression and poor patient prognosis.
In a new study from Boston University Chobanian & Avedisian School of Medicine, researchers assembled a comprehensive single-cell atlas of the human breast, combining data from 138 patients and more than 600,000 single cells. They then were able to identify cell populations and cell states that previously were unclear or missed and linked these cell types to tumor features (like subtype and grade) and patient outcomes.
“This represents the largest atlas of untreated, primary breast tumors and describes, at a high-resolution, the different cancer cells and the many immune and support (stromal) cell types that live inside breast tumors,” explains corresponding author Stefano Monti, PhD, professor of medicine at the school. “This is important because previous single-cell studies of breast cancer had far fewer samples and cells, limiting what could be reliably discovered.”
The researchers gathered eight public single cell RNA sequence datasets of untreated, unsorted human breast tumors. They cleaned each dataset (filtered low quality cells and doublets) and integrated them computationally to correct technical differences. They then grouped cells into major compartments (epithelial, immune, stromal), clustered and annotated sub-populations and scored cells for functional programs (such as stemness, EMT, pathway activity). Associations between these cell states/populations and clinical features (subtype, grade, age) were then tested against patient survival.
According to the researchers, the atlas can help identify which cell types are linked to better or worse patient outcomes and which immune or stromal cells are more common in particular breast cancer subtypes. “That information can guide development of new tests to predict prognosis and suggest targets for therapies (for example, targeting specific immune or fibroblast subtypes), and may help match patients to treatments that are more likely to work for their tumor’s cellular makeup,” adds first author Andrew Chen, a PhD student in bioinformatics.
The researchers have made the atlas and code publicly available so other scientists can explore it, test new hypotheses and build on it, accelerating discovery of potential new treatments.
These findings appear online in the journal NAR Genomics and Bioinformatics .
This work was supported by grants from the National Institute of Health (NIH): U01CA243004 (G.V.D. and S.M.), the National Institute of General Medical Sciences of the NIH under award number T32GM100842 (A.C.), and by a gift from Find The Cause Breast Cancer Foundation (S.M.).
NAR Genomics and Bioinformatics
Data/statistical analysis
Cells
A highly resolved integrated single-cell atlas of human breast cancers
4-Feb-2026