A combination of extracellular matrix- and interferon-associated signatures identifies high-grade breast cancers with poor prognosis

Breast cancer (BC) is a heterogeneous disease in which the tumor microenvironment (TME) seems to impact the clinical outcome. Here, we investigated whether a combination of gene expression signatures relating to both the structural and immune TME aspects can help predict prognosis in women with high-grade BC (HGBC). Thus, we focused on a combined molecular biomarker variable that involved extracellular matrix (ECM)-associated gene expression (ECM3 signature) and interferon (IFN)-associated metagene (IFN metagene) expression. In 97 chemo-naive HGBCs from the METABRIC dataset, the dichotomous ECM3/IFN (dECIF) variable identified a group of high-risk patients (ECM3(+)/IFN- vs other; hazard ratio = 3.2, 95% confidence interval: 1.5-6.7). Notably, ECM3(+)/IFN- tumors showed low tumor-infiltrating lymphocytes, high levels of CD33-positive cells, absence of PD-1 expression, or low expression of PD-L1, as suggested by immune profiles and immune-histochemical analysis on an independent cohort of 131 HGBCs. To make our results transferable to clinical use, we refined the dECIF biomarker using reduced ECM3 and IFN signatures; notably, the prognostic value of this reduced dECIF was comparable to that of the original dECIF. After validation in a new BC cohort, reduced dECIF was translated into a robust qPCR classifier for real-world clinical use.

Automated summary

The study found that a combination of gene expression signatures related to extracellular matrix and interferon can predict prognosis for high-grade breast cancer, and the dichotomous ECM3/IFN variable can identify high-risk patients. After refinement, this biomarker was translated into a qPCR classifier for clinical use with comparable prognostic value.