Breast cancer stromal compartment, may influence responsiveness to chemotherapy. Our aim was to detect a stromal cell signature (using a direct approach of microdissected stromal cells) associated with response to neoadjuvant chemotherapy (neoCT) in locally advanced breast cancer (LABC). The tumor samples were collected from 44 patients with LABC (29 estrogen receptor (ER) positive and 15 ER negative) before the start of any treatment. Neoadjuvant chemotherapy consisted of doxorubicin and cyclophosphamide, followed by paclitaxel. Response was defined as downstaging to maximum ypT1a-b/ypN0. The stromal cells, mainly composed of fibroblast and immune cells, were microdissected from fresh frozen tumor samples and gene expression profile was determined using Agilent SurePrint G3 Human Gene Expression microarrays. Expression levels were compared using MeV (MultiExperiment Viewer) software, applying SAM (significance analysis of microarrays). To classify samples according to tumor response, the order of median based on confidence statements (MedOr) was used, and to identify gene sets correlated with the phenotype downstaging, gene set enrichment analysis (GSEA). Nine patients presented disease downstaging. Eleven sequences (FDR 17) were differentially expressed, all of which (except H2AFJ) more expressed in responsive tumors, including PTCHD1 and genes involved in abnormal cytotoxic T cell physiology, TOX, LY75, and SH2D1A. The following four pairs of markers could correctly classify all tumor samples according to response: PTCHD1/PDXDC2P, LOC100506731/NEURL4, SH2D1A/ENST00000478672, and TOX/H2AFJ. Gene sets correlated with tumor downstaging (FDR < 0.01) were mainly involved in immune response or lymphocyte activation, including CD47, LCK, NCK1, CD24, CD3E, ZAP70, FOXP3, and CD74, among others. In locally advanced breast cancer, stromal cells may present specific features of immune response that may be associated with chemotherapy response.
Keywords: breast cancer; chemotherapy neoadjuvant; gene expression; stromal cells.