Integrative analysis of copy number and gene expression data identifies potential oncogenic drivers that promote mammary tumor recurrence.

Tumor recurrence represents a significant clinical challenge in the treatment and management of breast cancer. To investigate whether copy number aberrations (CNAs) facilitate the re-emergence of tumor growth from residual disease, we performed array comparative genomic hybridization on primary and recurrent mammary tumors from an inducible mouse model of type-I insulin-like growth factor receptor driven breast cancer. This genome-wide analysis revealed primary and recurrent tumors harbored distinct CNAs with relapsed tumors containing an increased number of gene-level gains and losses. Remarkably, high-level CNAs detected in primary tumors were largely devoid of annotated cancer genes while the vast majority of recurrent tumors harbored at least one CNA containing a known oncogene or tumor suppressor. Specifically, 38% of recurrent tumors carried gains at 6qA2 and 9qA2 which encode the Met and Yap1 oncogenes, respectively. The most frequent CNA, occurring in 63% of recurrent tumors, was a focal deletion at 4qC5 involving the Cdkn2a/b tumor suppressor genes. Integrative analysis revealed positive correlations between gene copy number and mRNA expression suggesting Met, Yap1, and Cdkn2a/b may serve as potential drivers that promote tumor recurrence. Accordingly, cross-species analysis revealed gene-level murine CNAs were present in a subset of human breast cancers with high MET and YAP1 mRNA predictive of decreased relapse-free survival in basal-like breast cancers. Together, these findings indicate that tumor recurrence is facilitated by the acquisition of CNAs with oncogenic potential and provide a framework to dissect the molecular mechanisms that mediate tumor escape from dormancy.