Molecular gene expression and genome wide profiling in tamoxifen-resistant breast cancer.

Abstract

Oestrogen receptor positive (ER+) breast cancers (BC) are heterogeneous in both their clinical behaviour and response to therapy. The ER and Progesterone (PgR) are currently the best predictors of response to the anti-oestrogen tamoxifen, yet up to 40% of ER+ breast cancer will relapse despite tamoxifen treatment. New prognostic biomarkers and further biological understanding of tamoxifen resistance (TR) are required. There has been an explosion of greater understanding since the arrival of cutting-edge gene and genomic profiling technology. The two major aims of this research are to develop stable gene signatures that are effective at distinguishing „prognostic‟ groups and, when tested directly for response to tamoxifen, a set of „predictive‟ markers. In order to establish cellular pathways responsible for TR, tissue at relapse while on tamoxifen is preferred. However, in practice, this is difficult to obtain. Hence, in this study, I have established TR derivatives of breast cancer cell lines, T47D and ZR75-1, and analysed their gene-expression by microarray. MAGEA2 and EGLN3 were 4.0 and 3.8 fold upregulated respectively in TR cell lines. For MAGEA2- and EGLN3-overexpressing lines, the proliferation and growth rates in tamoxifen-containing media were significantly higher (p-value <0.001 and p<0.05, respectively) than for control cells. I have investigated possible downstream targets for each protein which may contribute to the mechanism of resistance. Immunohistochemistry validation was performed on a cohort of 196 tamoxifen-treated primary breast tumour tissues: MAGEA2 and EGLN3 were found to be valuable predictive (Positive predictive value of 89%, and 85%, with high sensitivity 38% and 42% respectively) biomarkers for TR in primary breast tumours. In the human breast tumour arm of this study, 25 frozen samples with known response to tamoxifen were analysed on both SNP6.0 and expression EXON arrays. The integrated analysis suggested that 5 genes (OPCML, OR10G7, SNF1LK2, PALM and ZBTB-16) are good predictors of TR, with high negative predictor values (68%, 71%, 59% and 73% respectively for the last 4 genes). Significant regions of copy number variation (CNV) were identified at chromosomes 8q24, 17q21-22 and 11q23-25. The application of this high-resolution approach should lead to a better understanding of the roles of complex genetic alterations in TR.