Abstract Gene expression studies, mainly addressed at the tumor cell compartment of breast cancers (BC) contributed to unravel the molecular heterogeneity providing clinicians with new and reliable tools for therapeutic planning. Despite many investigations have focused on genetic abnormalities initiating and driving tumorigenesis, there is now evidence that the cancer cell behavior depends not only on intrinsic features, but also on the interactions with the microenvironment. However, notwithstanding the interest in investigating tumor-stroma cross-talks, the clinical relevance of subtype-specific molecular imprint of such interactions is largely unexplored. As such, we attempted to clinically and biologically validate the subtype-specific gene signatures derived from an in vitro model recapitulating the interaction between BCs and activated stromal cells. Gene expression signatures derived from luminal (T47D, MCF7, ZR75.1), HER2+ (SkBr3+, BT474, MDA-MB-361) and basal-like (MDA-MD-468, MDA-MB-231, BT20) cell lines treated by conditioned media from normal or cancer-associated fibroblasts (CAFs) showed common- and subtype-specific pathways according to GSEA. In fact, interferon, IL-6, IL-12 and IL-23 signaling, Toll Like Receptor pathway, pathways related to inflammation and metabolism of carbohydrates were positively enriched in HER2+ and luminal subtypes. Complement cascade and Tumor Necrosis Factor pathways were positively enriched in basal and luminal tumors; no common pathway was observed in HER2+ and basal subtypes. The association between microenvironment signatures specific for luminal, HER2+ and basal BC cell lines and disease-free survival was evaluated in silico on published gene expression profiles (GEPs) from 2048 patients homogeneously treated. BCs were classified as microenvironment-positive (µENV+ve, with GEP suggesting stroma activation), or negative (µENV-ve) by correlating tumor’s GEP with the respective subtype-specific signature. Patients with luminal µENV+ve tumors had 2.5-fold higher risk of developing distant metastases (HR=2.546; 95% Confidence Interval, Cl: 1.751-3.701, P=9.84E-07), while µENV status did not affect or was only suggestive of metastases in women with HER2-enriched (HR=1.541; 95% Cl: 0.788-3.012, P=0.206) or basal tumors (HR=1.894; 95% Cl: 0.938-3.824; P=0.0747), respectively. In luminal tumors, the µENV status maintained its prognostic role (HR=2.098; CI: 1.214-3.624; P=0.00791) in multivariable analysis including size, age and genomic grade index. Validity of our in vitro model was also supported by biological endpoints such as proliferation (MTT-assay) and migration/invasion (Transwell-assay). In vitro-derived gene signatures trace a subtype-specific interaction with CAFs and add independent information to classical prognostic variables in women with luminal “microenvironment sensitive” tumors. Citation Format: Maria Grazia Daidone, Giuseppe Merlino, Patrizia Miodini, Maurizio Callari, Vera Cappelletti. A microenvironment gene signature predicts poor outcome in patients with luminal breast cancers subjected to local-regional treatment [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2017; 2017 Apr 1-5; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2017;77(13 Suppl):Abstract nr 1747. doi:10.1158/1538-7445.AM2017-1747
Read full abstract