Almost all genomic studies of breast cancer have focused on well-established tumours because it is technically challenging to study the earliest mutational events occurring in human breast epithelial cells. To address this we created a unique dataset of epithelial samples ductoscopically obtained from ducts leading to breast carcinomas and matched samples from ducts on the opposite side of the nipple. Here, we demonstrate that perturbations in mRNA abundance, with increasing proximity to tumour, cannot be explained by copy number aberrations. Rather, we find a possibility of field cancerization surrounding the primary tumour by constructing a classifier that evaluates where epithelial samples were obtained relative to a tumour (cross-validated micro-averaged AUC = 0.74). We implement a spectral co-clustering algorithm to define biclusters. Relating to over-represented bicluster pathways, we further validate two genes with tissue microarrays and in vitro experiments. We highlight evidence suggesting that bicluster perturbation occurs early in tumour development.

Purpose: In the quest for new targets, genomes of ductal carcinoma in situ (DCIS) and infiltrating duct carcinoma (IDC) have been compared previously; however, genomic alterations associated with cancer progression were difficult to identify. We hypothesized that significant events can be detected by comparing lesions with a broader range of behavior: from pure DCIS to IDC associated with lymph node metastasis. Experimental Design: Array comparative genomic hybridization, calibrated by self-self hybridization tests, was used to study 6 cases of pure DCIS and 17 cases of DCIS paired with IDC where 8 tumors had spread to the local lymph nodes. Results: Pure DCIS exhibited a marginally higher degree of genomic complexity than DCIS and IDC components of invasive tumors. The latter two showed similarity between tumors and between components of the same tumor with several regions detected preferentially compared with pure DCIS. IDC associated with lymph node metastases showed similarity of genomic profiles as a group. Gain on 17q22-24.2 was associated with higher histologic grade, large IDC size, lymphatic/vascular invasion, and lymph node metastasis (P < 0.05).Conclusions: Our findings suggest that DCIS and IDC are associated with specific genomic events. DCIS associated with IDC is genomically similar to the invasive component and therefore may represent either a clone with high invasive potential or invasive cancer spreading through the ducts. Specifically, gain on 17q22-24.2 is a candidate region for further testing as a predictor of invasion when detected in DCIS and predictor of nodal metastasis when detected in DCIS or IDC.

We prospectively evaluated a series of 254 breast cancers by quantitative polymerase chain reaction (PCR) and immunohistochemistry using 3 antibodies: HercepTest, CB11, and TAB250. DNA was extracted from a 10-micron tumor section for PCR, and 4-micron serial sections were taken from the same block for immunohistochemistry. The immunohistochemical results were scored using a semiquantitative immunohistochemical system. A positive tumor by immunohistochemistry had a score of 5 or more. The manufacturer's recommended scoring system was used for the HercepTest. Tumors were positive for gene amplification if the ratio of the HER2/neu gene to control gene after normalization was 2 or more. Of 254 cases, 61 showed gene amplification. For immunohistochemistry, 23% of tumors were positive with CB11, 27% with TAB250, and 37% with the HercepTest. Results for each antibody were compared with PCR results. The overall concordance for the HercepTest was 82%, which was significantly lower than that for CB11 (88%) or TAB250 (87%). The specificity for the HercepTest was 80% compared with 90% for TAB250 and 93% for CB11, while the positive predictive value for the HercepTest was 57% compared with 71% and 76% for TAB250 and CB11, respectively.