Breast cancers are complex cellular ecosystems where heterotypic interactions play central roles in disease progression and response to therapy. However, our knowledge of their cellular composition and organization remains limited.
To systematically profile the cellular diversity of primary breast cancers we integrated transcriptomes from over 100,000 individual cells, spanning all major molecular subtypes. This provided a high-resolution characterization of the neoplastic, immune and stromal landscape of human breast cancers. To investigate neoplastic cell heterogeneity we have developed a single cell classifier of intrinsic subtype (scSubtype) and reveal recurrent neoplastic cell transcriptional programs. Integration of immunophenotyping using CITE-Seq has revealed novel immune and stromal cell subsets which show a continuum of differentiation states with diverse predicted functions and cell surface protein expression. This has allowed us to generate a detailed taxonomy of the cells present in the breast tumour micro-environment.
Using this we show that stromal and immune niches are spatially organized in tumours, offering insights into anti-tumour immune regulation. In particular, Nanostring spatial profiling, at whole-transcriptome resolution, was used to analyse T-cell rich regions of the tumour micro-environment in a subset of matching triple-negative breast cancers. Altered T-cell subset abundance was identified by tissue domain. Integration with our cellular gene signatures allowed deconvolution of the cell-type abundances of these specific tissue regions.
Finally, our single-cell derived signatures were used to deconvolute large breast cancer cohorts. We were able to stratify them into nine clusters, termed ‘ecotypes’, with unique cellular compositions and association with clinical outcome.
This work highlights the potential of atlas-scale single-cell projects to unravel the complex cellular heterogeneity within tumours and identify novel cell types and regulatory states underlying carcinogenesis. Such insights will guide the next generation of therapies, which will likely be based upon an integrated understanding of the cell states that define a tumour ecotype and inform treatment response.