An Analytic Framework Characterizes the Biological Processes That Shape Copy Number-Based Genome Instability Patterns in Breast Cancer.

Copy number (CN) alterations (CNA) accumulate nonrandomly within cancer genomes reflecting specific DNA damage and repair events. Higher-order patterning of CNAs can illuminate the types and determinants of genome instability (GI), as well as their clinical relevance, highlighting the need to develop analytic frameworks to capture such patterns. To address this issue, we collated a literature-curated compendium of predefined CN-based GI scores and extracted de novo CN signatures. Application to 2,763 breast cancer genomes from The Cancer Genome Atlas and METABRIC revealed the complementarity of various GI scores and their differences across immunohistochemical subtypes. Of the eight CN signatures identified, three were associated with distinct characteristics of homologous recombination deficiency and showed differential activity between cases with BRCA1 versus BRCA2 loss. Segments assigned to an HER2+ enriched signature strongly overlapped regions of chromothripsis and circular extrachromosomal DNA, suggesting that a common mutational process contributes to these phenotypes. CN "quiet" diploid and tetraploid genomes were apparent, with the latter group capturing a unique subset of whole-genome doubled tumors enriched for PIK3CA, MAP3K1, and CDH1 mutations. Finally, combining CN signatures with tumor microenvironment analyses, patients with quiet genomes and low macrophage infiltration showed remarkably better survival outcomes. Collectively, these findings demonstrate the value of deep interrogation of scores and signatures in characterizing the biological processes and clinical implications underlying CN-based GI. The publicly available web portal will facilitate similar analyses across pan-cancer genomes. Development of an extensive framework and a web portal for quantifying copy number patterns provides biological insights into breast cancer and genomic instability and supplies tools for future studies across cancer types.