The cost of next-generation sequencing technologies is rapidly declining, making RNA-seq-based gene expression profiling (GEP) an affordable technique for predicting receptor expression status and intrinsic subtypes in breast cancer patients. Based on the expression levels of co-expressed genes, GEP-based receptor-status prediction can classify clinical subtypes more accurately than can immunohistochemistry (IHC). Using data from The Cancer Genome Atlas Breast Invasive Carcinoma (TCGA BRCA) and Molecular Taxonomy of Breast Cancer International Consortium (METABRIC) datasets, we identified common predictor genes found in both datasets and performed receptor-status prediction based on these genes. By assessing the survival outcomes of patients classified using GEP- or IHC-based receptor status, we compared the prognostic value of the two methods. We found that GEP-based HR prediction provided higher concordance with the intrinsic subtypes and a stronger association with treatment outcomes than did IHC-based hormone receptor (HR) status. GEP-based prediction improved the identification of patients who could benefit from hormone therapy, even in patients with non-luminal breast cancer. We also confirmed that non-matching subgroup classification affected the survival of breast cancer patients and that this could be largely overcome by GEP-based receptor-status prediction. In conclusion, GEP-based prediction provides more reliable classification of HR status, improving therapeutic decision making for breast cancer patients.
Keywords: LASSO regression; breast cancer; gene expression profile; hormone receptor-status prediction; intrinsic subtype.