Gene signature predictive of acquired resistance to hormone treatment in ER+ breast cancer

The aims of the study are to delineate molecular signatures predictive of primary or acquired resistance to hormone therapies (namely, neoadjuvant treatment with aromatase inhibitors and in vitro treatment with Tamoxifen), in ER+ breast cancer patients and cell lines. To identify a gene expression signature predictive of response to Anastrozole neoadjuvant treatment in post-menopausal ER+ breast cancer patients, we analyzed the transcriptome of 17 tru-cut biopsies and 13 matched surgical samples from ER+ breast tumor patients, treated for 3 months with Anastrozole, and associated their molecular profiles to response data. We identified a predictive signature of 54 genes, whose basal expression is different in responders compared to non-responders and correlates with response data. Leave-one-out cross validation was applied. Seven of these genes were able to accurately predict response on a totally independent cohort of 52 patients (p=0.0056) treated with Letrozole (Miller W et al. 2009). As far as acquired resistance to hormone treatment, it has been recently demonstrated that the corepressors NCoR and SMRT have a key role in mediating the antagonist activities of both pure antiestrogens and SERMs (Giralult et al. 2003). We observed that in Tamoxifen resistant cell clones derived from MCF7 cells by continuous exposure to the drug, NcoR was aberrantly localized in the cytoplasm, similarly to what happens in wild-type cells treated with IL-1. We then hypothesized that a mechanism of acquired resistance to the antiestrogen Tamoxifen could be based on dismissal of the N-CoR corepressor complex operated by TAB2, that acts as a shuttle by binding N-CoR and driving it from the nucleus to the cytoplasm. Downregulating TAB2 in Tamoxifen-resistant cells, by small interfering RNA, was sufficient to restore Tamoxifen sensitivity due to re-localization of NcoR to the nuclei observed by immunofluorescence and fraction studies. As a consequence, we hypothesized that a TAB2-dependent transcriptional fingerprint may better reflect estrogen signalling, thus being a more definitive predictor of breast cancer recurrence and patients’ response to hormonal therapy. Moreover, it could be useful for finding targets for therapies to revert hormone resistance in breast cancer. To further elucidate the molecular events underlying acquired resistance to Tamoxifen, we tested hormone-sensitive and hormone-resistant sublines of human breast cancer MCF-7 cells treated with Tamoxifen, 17-estradiol (E2) and a specific RNA targeting TAB2. Global gene expression changes induced by TAB2 siRNA treatment in both hormone-sensitive and hormone-resistant sublines were analyzed using Agilent Whole Genome 4X44k oligo glass arrays, with quantitative qPCR verification for some genes. Integrating different analyses applying the Rosetta Resolver error-model and the LIMMA package of R, we identified 125 TAB2-modulated genes in the hormone-resistant subline. Gene Ontology analysis, performed with DAVID, demonstrated an enrichment in genes involved in protein-DNA complex assembly. Then we applied the TAB2-related signature to predict the prognosis on an external dataset and to predict response to therapy on our Anastrozole dataset. A 71-gene signature was developed that best classified, by k-means clustering, a test set of 213 ER+ primary breast carcinomas (Miller LD et al., 2005) into prognostic subtypes with statistical significance, tested by Kaplan-Meier analysis. Furthermore, from a robust set of 64 TAB2-regulated genes, we selected 18 genes correlated with response that were able to predict response to therapy with 75% of accuracy, by unsupervised clustering, in our 16 samples dataset of neo-adjuvant Anastrozole treated primary breast cancers.