Clinical outcomes linked to expression of gene subsets for protein hormones and their cognate receptors from LCM-procured breast carcinoma cells

Abstract

Purpose

Certain peptide hormones and/or their cognate receptors influencing normal cellular pathways also have been detected in breast cancers. The hypothesis is that gene subsets of these regulatory molecules predict risk of breast carcinoma recurrence in patients with primary disease.

Methods

Gene expression levels of 61 hormones and 81 receptors were determined by microarray with LCM-procured carcinoma cells of 247 de-identified biopsies. Univariable and multivariable Cox regressions were determined using expression levels of each hormone/receptor gene, individually or as a pair.

Results

Molecular signatures for ER+/PR+, ER−/PR−, and ER− carcinoma cells deciphered by LASSO were externally validated at HRs (CI) of 2.8 (1.84–4.4), 1.53 (1.01–2.3), and 1.72 (1.15–2.56), respectively. Using LCM-procured breast carcinoma cells, a 16-gene molecular signature was derived for ER+/PR+ biopsies, whereas a 10-gene signature was deciphered for ER−/PR− cancers. Four genes, POMC, CALCR, AVPR1A, and GH1, of this 10-gene signature were identified in a 6-gene molecular signature for ER− specimens.

Conclusions

Applying these signatures, Kaplan–Meier plots definitively identified a cohort of patients with either ER−/PR− or ER− carcinomas that exhibited low risk of recurrence. In contrast, the ER+/PR+ signature identified a cohort of patients with high risk of breast cancer recurrence. Each of the three molecular signatures predicted clinical outcomes of breast cancer patients with greater accuracy than observed with either single-gene analysis or by ER/PR protein content alone. Collectively, our results suggest that gene expression profiles of breast carcinomas with suspected poor prognosis (ER−/PR−) have identified a subset of patients with decreased risk of recurrence.

http://ift.tt/2fkysSu