Abstract P3-04-01: Discovery of distinct sub-groups of ER positive breast cancers by DNA repair expression profiling: Implications for patient stratification and personalization of therapy

Abstract

Abstract Background: The role of DNA repair in estrogen receptor (ER) positive breast cancer is largely unknown. We hypothesized that DNA repair dysregulation could lead to evolution of aggressive phenotypes with increasing resistance to endocrine therapy. Methods: We immuno-profiled 1120 early stage ER positive breast tumours for Pol b, FEN1, APE1, XRCC1, SMUG1, PARP1, ATR, ATM, DNA-PK, BRCA1, CHK1, CHK2, TOPO2, and p53. Multivariate analysis was used to calculate a DNA repair prognostic index after controlling for other validated prognostic factors, confounders and adjuvant therapy. For additional validation, genomic expression as well as array CGH evaluation of key DNA repair genes was performed [discovery cohort (n = 128), external validation in Uppsala (n = 249) and METABRIC cohort (n = 1950)]. Neural network analysis was conducted in 249 tumours to map genetic interactions. Results: DNA repair immuno-profiling stratified ER positive breast tumours into four distinct sub-groups with increasing aggressiveness and endocrine resistance (p = 1.6 × 10 −27). Subgroup 1 had the best breast cancer specific survival, subgroups 2 & 3 have intermediate breast cancer specific survival and sub-group 4 had the worst breast cancer specific survival (Table). Survival analysis HR (95% CI)P ValuesSub-group 2 versus. Sub-group 11.6 (1.1-2.1)0.006Sub-group 3 versus. Sub-group 13.1 (2.3-4.3)1.5 × 10-12Sub-group 4 versus. Sub-group 15.7 (4.0-8.1)9.0 × 10−21 These distinct subgroups manifest progressively increasing tumour size, level of proliferation, de-differentiation, lymphovascular invasion, metastasis, altered apoptosis and increasing level of genomic instability (ps<0.001). At the genomic level DNA repair gene dysregulation remains independently associated with poor outcome and resistance (ps<0.001). Neural network analysis confirms interaction with genes involved in ER response. Conclusion: This is the first and largest study with 20 years long term follow-up data to provide compelling evidence that DNA repair dysregulation impacts ER positive breast cancer pathogenesis and drives evolution of aggressive phenotypes. Our data could transform stratification and the personalization landscape for ER positive tumours. Citation Information: Cancer Res 2013;73(24 Suppl): Abstract nr P3-04-01.