Integrated analysis of a large number of breast cancer samples identifies stratification strategies.
Major finding: Integrated analysis of a large number of breast cancer samples identifies stratification strategies.
Concept: Cis-acting copy number aberrations are associated with distinct clinical outcomes.
Impact: The use of genomic features to refine breast cancer subgroups may inform treatment decisions.
The high degree of heterogeneity among breast cancers requires that large numbers of patients be studied to uncover driver mutations and define disease subtypes. Curtis and colleagues subjected approximately 2,000 clinically annotated primary breast cancer samples to DNA copy number and gene expression analysis using single-nucleotide polymorphism (SNP) arrays and microarrays, respectively. The authors assessed the influence of SNPs, copy number variants (CNV), and copy number aberrations (CNA) on gene expression and found that somatically acquired tumor CNAs had the most influence on genome-wide expression variation on a per-gene basis, affecting roughly equal numbers of genes in cis or trans. Trans-acting CNAs were associated with a larger number of genes, representing subtype-specific modules linked to processes such as lymphocyte infiltration and increased mitosis that may contribute to clinical outcome. Cis-acting CNAs tended to more strongly regulate gene expression, and some genes with particularly large changes in expression represented previously unappreciated subtype-specific recurrent deletions that implicated PPP2R2A, MTAP, and MAP2K4 as putative breast cancer tumor suppressor genes. Furthermore, clustering of a discovery set of 997 breast cancers by the 1,000 most cis-regulated genes suggested 10 subgroups with distinct clinical outcomes and was reproduced by a validation set of 995 tumors. These groups included an estrogen receptor (ER)-positive subgroup with a poor outcome characterized by 2 separate amplicons at 11q13 and 11q14 (affecting CCND1 and several putative oncogenes); 2 low-CNA subgroups with a good prognosis that incorporated luminal A, ER-positive, and ER-negative tumors; and an expanded ERBB2-amplified subgroup that included both ER-negative and luminal ER-positive subtypes. Collectively, these findings demonstrate the power of integrated genomic analyses of large numbers of tumors to reveal recurring genetic lesions and refine patient stratification.