Profiling human breast tumor biopsies. gene expression changes associated with ERBB2 status and prognosis, and possible implications for molecular breast cancer classification in the clinic

Breast cancer is the most common malignancy in women in Western countries, and has proven to be a very heterogeneous disease both on the biological and clinical level. Traditionally, breast cancer classification included histopathological and clinical parameters such. In addition, few molecular markers such as hormone receptors and ERBB2 status are used to classify breast cancer patients and make treatment decisions in clinical routine. The latter, ERBB2 (HER2/Neu), has been shown to be amplified and overexpressed in 15 to 30% of breast cancers, and is associated with poor prognosis and an increased likelihood of metastasis. However, not all patients with ERBB2-positive tumors develop metastasis and despite intensive research efforts, the biological mechanisms underlying the oncogenicity of ERBB2 are still not fully understood. Regardless, ERBB2 and its other family members have become a promising field for targeted therapy with monoclonal antibodies or tyrosinekinase inhibitors, some of which are already successfully being used in the clinic. More recently, gene expression-based approaches suggested that they could be superior to classical classification systems or at least add significant new information. In this context, microarrays and qrt-PCR have emerged as key technologies allowing analysis of up to thousands of genes simultaneously. Together with various bioinformatics techniques, complex relationships in the data can be explored which potentially can provide the basis for personalized treatment based on individual molecular finger prints to enhance the treatment efficacy and decrease the risk of side effects. The aims of this study were to investigate the differences between ERBB2+/- breast tumor samples on the gene expression level and characterize the molecular phenotype associated with ERBB2 status, to evaluate possible downstream effects associated with ERBB2 signaling, to identify relevant subgroups or genes associated with outcome according to ERBB2 status, and to explore putative clinical implications towards a molecular classification of breast cancer in clinical routine. During the project, several cell lines and various fresh frozen breast tumor samples were comprehensively profiled using “home-made” cDNA microarrays, qrt-PCR and selected immunoassays. Various supervised and unsupervised statistical methods were applied and demonstrated that tumor samples can be classified into molecularly distinct and clinically relevant subgroups based on variations in gene expression patterns. In these analyses, proliferation emerged a key determinant of breast cancer prognosis and was particularly suitable to identify subgroups of patients with favorable outcome. The prognostic value of established markers such as ER and ERBB2, in contrast, was strongly related to proliferation. Of interest, this further applied for the recently described “intrinsic subtypes” (eg. “luminal A/B”, “basal”), the “Amsterdam 70-gene signature” as well as the 21-gene “recurrence score” and “wound response signature”. Thus, proliferation seems to be one of the most downstream players of all of these markers in terms of prognosis, and potentially could be easily assessed in clinical routine using few selected markers such as the transcription factor E2F1 by art-PCR as shown in this study. Moreover, proliferation and in particular E2F1 as a transcription factor also appear to be a good candidates markers for chemotherapy response: genes such as TK1, TOP2A and TYMS are regulated by E2F1 or associated with proliferation and are involved as direct targets or in metabolism of 5-FU and anthracycline-based therapy. Indeed, several studies demonstrated that high proliferating tumors responded better to chemotherapy. Several differentially expressed genes were identified when comparing ERBB2+/- tumor samples. Interestingly, many of these mapped to chromosome 17q12-21 which harbors the “ERBB2-amplicon”, indicating the underlying amplification was responsible for this finding and one of the main drivers for the discrimination of ERBB2 status. In addition, several other differentially expressed genes associated with ERBB2 status were indentified, some of which might potentially be regulated by ERBB2 or arise through association with phenotypic features of the disease such as the cell of origin. Among these were genes related to ER status, genes involved in metastasis and invasion (e.g. proteases, S100P/A4), angiogenesis (e.g. VEGF, IL8, HIF1A), proliferation (BIRC5, TK1, CDC2, CDKN1A) and cell adhesion (e.g. CEACAM6, CDH1). Thus, gene expression changes associated with ERBB2 status observed in breast tumor biopsies appear to reflect different aspects of the disease. These include the underlying genetic changes (eg. the “ERBB2 amplicon”), the cellular phenotype which itself appears to be combination of genetic changes and possibly the cell of origin, as well as the interplay between tumor tissue and stroma, vasculature and immune response. Clinically, they result in an aggressive phenotype through regulation of key processes relevant to cancer such as proliferation, invasion, angiogenesis and cell adhesion. Data from the literature demonstrated, that ERBB2 indeed can regulated and interfere with these processes, and inhibition of ERBB2 using a monoclonal antibody directed against the extracellular leads to prolonged survival. While consistency starts to emerge with respect to classification of low-risk patients, efforts continue to refine and further explore the high-proliferation, poor-prognosis subtypes. In this context, our analysis also further focused on ERBB2+ tumors in particular, where proteases emerged as the most relevant genes with regards to prognosis. Of note, our results indicated that apparently only taken in conjunction, uPA and ERBB2 overexpression determine the aggressive breast cancer phenotype whereas apparently the proteolytic activity of uPA alone is insufficient to determine the most metastatic of breast cancer phenotypes and perhaps requires the proliferation and survival advantages provided by ERBB2 overexpression. Finally, our study of paired core biopsy (CB) and surgical sample showed that quantitative expression levels detected in CBs were highly comparable to their paired surgical samples indicating that the CB is representative of the main tumor in terms of gene expression profile. However, gene-by-gene analysis demonstrated higher expression level of PAI-1, COX-2, uPAR and MMP1 in the surgical specimen as compared to their paired CB. These genes, potentially attributable to a wound healing process induced by the CB procedure, have been associated with tumor invasion and angiogenesis and therefore might impact the clinical interpretation with respect to tumor aggressiveness and subsequent treatment decisions. This is an important finding since today ultrasound-guided core biopsy is a well established method in routine breast cancer diagnosis. Taken together, gene-expressed based molecular classification of tumor samples is feasible even in smallest amounts of tissue, and seems to complement or even outperform traditional classification systems in certain aspects. The heterogeneous nature of the breast cancer, however, will likely require developing “individualized gene signatures” rather than one signature for all patients since the prognostic value of individual markers or gene signatures depends on the biological context of the tumor. Thus, patient stratification might be necessary to further optimize current molecular breast cancer classification. The use of “supervised approaches” in terms of assessing the contribution of individual biological processes or pathways constitutes an interesting approach. Together these are valuable strategies towards a biology driven classification and molecular understanding of breast cancer, and facilitating the clinical interpretation with respect to the development of appropriate treatment strategies in the future. In terms of routine clinical assessment, the accurate quantification of at least the proliferation status and potentially other markers should be encouraged and incorporated into the current breast cancer classification system. Several data suggest that, for example, poor prognostic patients with high proliferation levels should receive systemic chemotherapy on top of endocrine treatment for ER+, and ERBB2targeted therapy for ERBB2+ tumors. In contrast, tumors with low proliferation levels would likely not require aggressive treatment and could potentially be spared from unnecessary toxicity. Since these tumors are ER+ endocrine therapy might be sufficient.