Other groups have developed methods to identify gene expression signatures that reflect CIN and analyze associations with relapse,32 prognosis,33 and survival outcomes29 (Table 1). Tumor grade, which reflects the differentiation and proliferation potential of tumor cells, is a routinely used histological measure. Based on the expression of the top 25 genes in a 25-gene expression assay known as CIN25, grade 1 and grade 2 breast tumors from three data sets were stratified to high or low scores.29 The genes within the panel were selected based on strong associations between altered gene expression and tumor aneuploidy. A higher CIN25 score was associated with a worse clinical outcome for patients with either grade 1 or grade 2 tumors. Similarly, another study using only four genes (CIN4) found a significant association with the proliferation marker Ki67 in grade 2 tumors. CIN4 was used to further stratify patients with grade 2 tumors into good and poor prognosis groups.33 Higher CIN4 was associated with worse recurrence-free survival. Data from these studies, summarized in Table 1, highlight the differences in CIN and the outcomes observed between breast cancer subtypes.
Patterns and Frequency of Genomic Rearrangements in Breast Cancer
An increase in genomic instability has been linked to a concomitant increase in the frequency of gene rearrangements or fusions.2,34 Next-generation sequencing strategies have led to genomic and transcriptomic analysis of large cohorts across cancer types as well as detailed analysis of the patterns of genomic rearrangements in breast cancer.34–36 Some breast cancers showed genome-wide rearrangements, whereas others were reported as clusters in regions of amplification. Array-based comparative genomic hybridization (array CGH) studies of copy number alterations have previously defined structural changes in the genome in terms of gains or losses of specific chromosomal regions such as 1q/16 for low-grade ER+ tumors, commonly amplified sites such as 8p11–12 (FGFR1), 8q24 (MYC), 11q13 (CCND1), and 17q12 (ERBB2) for high-grade ER+ cancers, and numerous low amplitude gains and losses for TNBCs.28,37–40The presence of 8p11, 8q24, 17q12, and 17q24 amplicons in high-grade ER+ breast cancer has been demonstrated to be associated with poor outcome in multiple studies.41,42
Using next-generation technologies, quantification of the number of rearrangements occurring within chromosomes (intrachromosomal rearrangements such as duplications, inversions, amplifications, and deletions) and also those occurring between different chromosomes (interchromosomal rearrangements) has been depicted by Kwei et al39 using circos plots for breast cancer. Circos plots are circular illustrations for visualizing the structural relationships between regions of chromosomes. These comparisons clearly show distinct patterns related to breast cancer subtypes.39 Low-grade ER+ breast tumors generally display few rearrangements and amplifications, whereas high-grade ER+ breast cancers and TNBCs display a large amount of both large- and small-scale rearrangements, especially increased frequency of intrachromosomal rearrangements such as tandem duplications in TNBCs. This implies that subtype-specific differences in genetic instability may mechanistically contribute to different gene expression patterns observed in breast cancer subtypes.36,39,43
By analyzing 24 breast cancer genomes by paired-end sequencing, Stephens et al36 showed that intrachromosomal alterations are much more prevalent than anticipated across a broad spectrum of molecular subtypes based on ER/PR/ERBB2 status. Tandem duplications were the most commonly observed rearrangement. This feature was largely underappreciated using prior array CGH profiles. Strikingly, there was a large variation between genomes, with some harboring almost no duplications and others having hundreds, ranging from 3 kb to >1 Mb of duplicated segments.36 When subtype analysis was performed, unfavorable ER− cancers harbored comparatively more tandem duplications than ER+ subtypes. In this study, tandem duplications were not found to be associated with breast cancers arising in the setting of germline mutations in the DNA repair genes BRCA1/2. Notably, another group that analyzed tandem duplications in ovarian cancers also reported a lack of correlation of duplications with BRCA1/2mutations.44 This emphasizes the likelihood of other unknown defects in DNA repair/maintenance in TNBCs, perhaps contributing to the increase in tandem duplications and the relatively poor prognosis of this breast cancer subtype.
Although it has been shown that tandem duplications are the most commonly observed rearrangement in breast cancer genomes, the frequency of specific tandem duplications is currently unknown as these alterations are not easily detected with standard techniques such as array CGH or FISH. Next-generation sequencing techniques will achieve improved clarification of the genes and gene regions that are frequently involved in such duplications.