Genomic analyses have substantially improved our knowledge of cancer. Gene expression profiling, for example, is utilized to delineate subtypes of breast cancer, and has facilitated the derivation of predictive and prognostic signatures [1
]. However, not all of the gene expression changes observed are causal to cancer development, and global gene expression analysis alone cannot distinguish between causal and reactive changes. Corresponding alteration at the DNA level is regarded as evidence of causality; for example, gene deletion or gene silencing by methylation. Hence, examining genetic and epigenetic events in conjunction with the changes in gene expression pattern should improve the identification of causal changes that lead to disease phenotype.
Analysis of gene copy number alone has correlated breast cancer genome features with poor prognosis based on the degree of genomic instability observed [6
]. In terms of gene discovery, specific genomic regions containing important loci have been shown to be frequently gained or lost [7
]. Integrative analyses of gene dosage and gene expression in breast cancer have revealed specific genes which are deregulated at the gene expression level as a result of changes in DNA copy number. From a global perspective, studies have shown a broad range in concordance between DNA amplification and overexpression of genes. This variability is attributable to the sensitivity of the methods used in detecting gene copy number and gene expression changes as well as the number of genes examined [12
]. Conversely, when examining gene overexpression, it was found that only ~10% of the overexpression could be attributable to gene amplification [14
]. It is certain that altered gene expression can not only be attributed to disruption of regulatory/signaling cascades and downstream effects, but also to a multitude of causal genetic and epigenetic aberrations.
We reason that by examining multiple genomic dimensions simultaneously, with a dimension representing a genome wide assay measuring DNA level alterations such as gene copy number or DNA methylation, we are likely to achieve the following: (i) explain a greater fraction of the observed gene expression deregulation as compared with explaining expression deregulation using only a single dimension, (ii) improve the discovery of critical oncogenes and tumor suppressor genes (TSGs) by focusing on those genes altered simultaneously at multiple genomic dimensions, and (iii) begin to understand the complex mechanisms of dysregulation of oncogenic pathways. In this study, we demonstrate the power of an integrative genomics approach by performing multi-dimensional analyses (MDA) of the genome, epigenome, and transcriptome of breast cancer cell lines. We illustrate and demonstrate the need for integrative analysis of multiple genomic dimensions by showing the co-operative contribution of DNA mechanisms to explaining differential gene expression. Using a strategy to identify genes exhibiting congruent alteration in copy number, DNA methylation, and allelic (or loss of heterozygosity, LOH) status, which we term multiple concerted disruption (MCD) analysis, we find genes representing key nodes in pathways as well as genes which exhibit prognostic significance. In examining the neuregulin pathway, we observe the variability among samples in the mechanism of dysregulation of this commonly altered breast cancer pathway, highlighting the importance of multi-dimensional analysis of a given pathway in individual tumor samples -- in addition to the conventional approach of identifying loci simply based on frequency of disruption in a cohort. Finally, examining the subset of triple negative breast cancer cell (TNBC) lines, we show that a downstream target of FGFR2, a recently implicated oncogene in TNBC, COL1A1 is frequently affected by MCD even though in FGFR2 itself is rarely affected. Notably, this is the first such in-depth genomic, epigenomic, and transcriptomic analyses of breast cancer.