P1-03-07: Subtype-Specific Gene Expression Signatures in Peritumoral Non-Neoplastic Breast Tissue.

Abstract

Abstract Tumors and histologically normal-tissue adjacent to cancer share many biological features with healing wounds. We previously described that the majority of cancer-adjacent normal (CN) tissues show activation of wound response gene expression signatures. However, CN tissue also shows considerable heterogeneity across patients. Therefore, to characterize interindividual variation in CN tissues, we performed unsupervised clustering on gene expression data from women with invasive breast cancer or ductal carcinoma in situ of the breast. Two distinct classes of CN tissue were identified, one of which showed gene expression consistent with epithelial-to-mesenchymal transition (EMT), a process that is well known to be activated in healing wounds. Activation of this signature was present in ∼40% of cancer-adjacent normal specimens and it was also correlated with poor outcome among ER-positive patients. We next hypothesized that gene expression changes in cancer-adjacent normal tissue would be associated with breast cancer subtype and/or estrogen receptor status. To test this, we performed supervised analysis to identify subtype-associated gene expression profiles in CN. We observed that more aggressive tumor subtypes had distinct gene expression profiles in the adjacent normal tissue in supervised analysis. Specifically, Basal-like and luminal B tumors were characterized by reduced expression of claudins and cell adhesion genes, while also exhibiting higher expression of transcription factors associated with migration. These characteristic changes were validated in independent datasets by evaluating the ability of these signatures to predict subtype. It is possible to predict tumor subtype using just adjacent normal gene expression. Thus, our findings show that cancer-adjacent tissue may have value in identifying tumor subtype for cases where tumors are too small to allow genomic analyses. Furthermore, genomic field effects may have clinical implications if, as our data suggests, subtypes with higher local recurrence rates (e.g. basal-like and luminal B breast cancers) have widespread genomic alterations, even in non-neoplastic tissue. Widespread genomic changes may indicate a need for mastectomy rather than breast conserving therapy. Ongoing work is (1) assessing these signatures in association with relapse outcomes in the Polish Women's Breast Cancer Study, where more than 10 years of follow up data have been accrued on more than 200 patients, and (2) using co-culture models systems to study paracrine communication between breast cancer cells and stroma according to breast cancer subtypes. Citation Information: Cancer Res 2011;71(24 Suppl):Abstract nr P1-03-07.