Gene expression profiling and prediction of response to hormonal neoadjuvant treatment with anastrozole in surgically resectable breast cancer.

Breast Cancer Res Treat 121(2):399-411 (2010) PMID 20428938

Aromatase inhibition (AI) is the most effective endocrine treatment for breast cancer in post-menopausal patients, but a percentage of hormone receptor-positive cancers do not benefit from such therapy: for example, about 20% of patients treated with anastrozole do not respond and it is still impossible to accurately predict sensitivity. Our main goal was to identify a robust expression signature predictive of response to neoadjuvant treatment with anastrozole in patients with ER+ breast cancer. At the same time, we addressed the question of delineating treatment effects and possible mechanisms of intrinsic resistance occurring in non-responder patients. We analyzed the transcriptome of 17 tru-cut biopsies before treatment and 13 matched surgical samples after 3 months treatment with anastrozole taken from ER+ breast tumors. Molecular profiles were related to clinical response data. Treatment with anastrozole was associated with a decreased expression of genes relating to cell proliferation and an increased expression of genes relating to inflammatory processes. There was also an enrichment of induction of T-cell anergy, positive regulation of androgen signalling, synaptic transmission and vesicle trafficking in non-responders, and of cell cycle inhibition and induction of immune response in responders. We identified an expression signature of 77 probes (54 genes) that predicted response in 100% of our cases. Five of them were able to accurately predict response on an independent dataset (P = 0.0056) of 52 ER+ breast cancers treated with letrozole. Ten fixed independent samples from the anastrozole study were also used for RT-qPCR validations. This study suggests that a relative small number of genes analysed in a pre-treatment biopsy may identify patients likely to respond to AI neoadjuvant treatment. This may have practical utility translatable to the clinics. Furthermore, it delineates novel mechanisms of intrinsic resistance to AI therapy that could be further investigated in order to explore circumventing treatments.