Extensive expression profiling studies have shown that sporadic breast cancer is composed of five clinically relevant molecular subtypes. However, although BRCA1-related tumours are known to be predominantly basal-like, there are few published data on other classes of familial breast tumours. We analysed a cohort of 75 BRCA1, BRCA2 and non-BRCA1/2 breast tumours by gene expression profiling and found that 74% BRCA1 tumours were basal-like, 73% of BRCA2 tumours were luminal A or B, and 52% non-BRCA1/2 tumours were luminal A. Thirty-four tumours were also analysed by single nucleotide polymorphism-comparative genomic hybridization (SNP-CGH) arrays. Copy number data could predict whether a tumour was basal-like or luminal with high accuracy, but could not predict its mutation class. Basal-like BRCA1 and basal-like non-BRCA1 tumours were very similar, and contained the highest number of chromosome aberrations. We identified regions of frequent gain containing potential driver genes in the basal (8q and 12p) and luminal A tumours (1q and 17q). Regions of homozygous loss associated with decreased expression of potential tumour suppressor genes were also detected, including in basal tumours (5q and 9p), and basal and luminal tumours (10q). This study highlights the heterogeneity of familial tumours and the clinical consequences for treatment and prognosis.

IntroductionMetastases to the brain from breast cancer have a high mortality, and basal-like breast cancers have a propensity for brain metastases. However, the mechanisms that allow cells to colonize the brain are unclear.MethodsWe used morphology, immunohistochemistry, gene expression and somatic mutation profiling to analyze 39 matched pairs of primary breast cancers and brain metastases, 22 unmatched brain metastases of breast cancer, 11 non-breast brain metastases and 6 autopsy cases of patients with breast cancer metastases to multiple sites, including the brain.ResultsMost brain metastases were triple negative and basal-like. The brain metastases over-expressed one or more members of the HER family and in particular HER3 was significantly over-expressed relative to matched primary tumors. Brain metastases from breast and other primary sites, and metastases to multiple organs in the autopsied cases, also contained somatic mutations in EGFR, HRAS, KRAS, NRAS or PIK3CA. This paralleled the frequent activation of AKT and MAPK pathways. In particular, activation of the MAPK pathway was increased in the brain metastases compared to the primary tumors.ConclusionsDeregulated HER family receptors, particularly HER3, and their downstream pathways are implicated in colonization of brain metastasis. The need for HER family receptors to dimerize for activation suggests that tumors may be susceptible to combinations of anti-HER family inhibitors, and may even be effective in the absence of HER2 amplification (that is, in triple negative/basal cancers). However, the presence of activating mutations in PIK3CA, HRAS, KRAS and NRAS suggests the necessity for also specifically targeting downstream molecules.

Invasive lobular carcinoma (ILC) and lobular carcinoma in situ characteristically show loss of E-cadherin expression and so immunohistochemistry for E-cadherin is being increasingly used as a tool to differentiate between lobular and ductal lesions in challenging situations. However, misinterpretation of "aberrant" positive staining may lead some to exclude a diagnosis of lobular carcinoma. E-cadherin and beta-catenin immunohistochemistry was analyzed in 25 ILCs. E-cadherin "positive" ILCs were subjected to molecular analysis including comparative genomic hybridization. Different morphologic components of case 25, showing heterogenous E-cadherin expression, were analyzed by E-cadherin gene sequencing, methylation, and DASL gene expression profiling. Four ILCs were positive for E-cadherin, but each also had neoplastic cells with aberrant staining. Two of these ILCs were positive for beta-catenin, again with some aberrantly stained neoplastic cells, and 2 were negative. The solid component of case 25 was positive for E-cadherin whereas the classic and alveolar areas were negative. All components harbored an in-frame deletion in exon 7 (867del24) of the E-cadherin gene and loss of the wild type allele. Comparative genomic hybridization demonstrated evidence of clonal evolution from E-cadherin-positive to E-cadherin-negative components. E-cadherin down-regulation seems to be through transcriptional repression via activation of transforming growth factor-beta/SMAD2 rather than methylation. Positive staining for E-cadherin should not preclude a diagnosis of lobular in favor of ductal carcinoma. Molecular evidence suggests that even when E-cadherin is expressed, the cadherin-catenin complex maybe nonfunctional. Misclassification of tumors may lead to mismanagement of patients in clinical practice, particularly in the context of in situ disease at margins.

No abstract