A team of Japanese researchers has scored 179 genes tied to how breast cancer cells stay attached to each other, and found that the scores say something about who is likely to do badly. The paper, by Ishii H, Nishiyama K, Nakahara A, Tamori S, Ohno S, Sasaki K and colleagues, appeared in PLOS One on June 29.
Breast cancer is the most common cancer among women worldwide, and it is not one disease. The PAM50 system sorts it into six subtypes: Luminal A, Luminal B, Her2-enriched, Normal-like, Claudin-low and Basal-like. Each behaves differently and responds differently to treatment. Even so, patients within the same subtype can have very different outcomes, and the authors say the gene panels now used to guide treatment, such as Oncotype DX, MammaPrint and EndoPredict, were built mainly for early-stage luminal disease in patients on endocrine therapy.
The team took a different route. Earlier work had mostly studied single adhesion molecules, such as E-cadherin. The Japanese group looked at the whole set of genes filed under the Gene Ontology term “cell-cell junction organization”, and asked what they do together inside each subtype.
How the signatures were built
They drew on two large public datasets, METABRIC and The Cancer Genome Atlas. Using survival models, they picked out the junction genes that predicted disease-specific survival in each subtype and folded them into a score. They then tested the scores on an independent group of patients from the SCAN-B cohort, which covered 3,052 samples with usable data.
Higher scores went with worse survival across subtypes. The effect was strongest in Luminal B and Basal-like tumors. In those two aggressive subtypes, a high score also predicted more recurrences in the first five years after diagnosis.
Tumors with high scores showed more activity in pathways tied to the extracellular matrix, the scaffolding around cells. Single-cell sequencing added detail. In Luminal B, the relevant genes, including PARD6B and CDH3, were switched on mainly in tumor epithelial cells. The Basal-like signature drew on two cell types: epithelial cells, with genes such as MARVELD2, and endothelial cells, with genes such as RAMP2.
The authors conclude that the signatures sort patients by prognosis and capture features of both the tumor cells and their surroundings that differ by subtype.
The work is a lab analysis of existing patient data, not a clinical trial, and it does not test any treatment. Funding came from the Tokyo University of Science, the JST Moonshot R&D program and several Japanese research grants. The authors declared no competing interests.
Consult your doctor for medical advice.
Sources
- Cell–cell junction gene signatures as subtype-specific prognostic biomarkers in breast cancer
- Cell-cell junction gene signatures as subtype-specific prognostic biomarkers in breast cancer.
