Androgen receptor gene expression in primary breast cancer

Abstract

We studied androgen receptor (AR) gene expression in primary breast cancer (BC) to determine associations with clinical characteristics and outcomes in the I-SPY 1 study. AR was evaluated in I-SPY 1 (n = 149) using expression microarrays. Associations of AR with clinical and tumor features were determined using the Wilcoxon rank sum test (two-level factors) or the Kruskal-Wallis test (multi-level factors). We identified an optimal AR cut-point to maximize recurrence-free survival (RFS) differences between AR biomarker stratified groups, and assessed the association between the AR stratified groups and RFS using the Cox proportional hazard model. Pearson correlations between AR and selected genes were determined in I-SPY 1, METABRIC (n = 1992), and TCGA (n = 817). AR was lower in triple negative BC vs. hormone receptor positive (HR+)/HER2- and HER2+ disease (p < 0.00001), and lower in basal-like BC (p < 0.00001). AR was higher in grade I/II vs. III tumors (p < 0.00001), in patients >age 50 (p = 0.05), and in node negative disease (p = 0.006). Higher AR was associated with better RFS (p = 0.0007), which remained significant after receptor subtype adjustment (p = 0.01). AR correlated with expression of luminal, HER2, and steroid hormone genes. AR expression was related to clinicopathologic features, intrinsic subtype, and correlated with improved outcome.

Keywords: Breast cancer; Tumour biomarkers.

Conflict of interest statement

Competing interestsHope S. Rugo: research funding to the institution (UCSF) from Pfizer. Pfizer was not involved in the conceptualization, design, data collection, analysis, decision to publish, or preparation of this manuscript and study. Neelima Vidula, Christina Yau, and Denise Wolf do not have any competing interests.

© The Author(s) 2019.

Figures

Fig. 1. AR expression according to HR/HER2 subtype, intrinsic subtype, histological grade, and nodal status in I-SPY 1.

Box plots show median-centered normalized AR gene expression levels stratified by a HR/HER2 status; b expression-based intrinsic subtype; c histological grade; and d nodal status. The line within boxes indicate the median AR expression; and boxes span the inter-quartile range (IQR). Whiskers span 1st quartile −1.5 × IQR and 3rd quartile +1.5 IQR; and outliers are represented by points.

Fig. 2. Association of AR expression with recurrence-free survival (RFS).

I-SPY 1 patients were divided based on AR gene expression level into High vs. Low groups using an optimal cut point of −0.89. Kaplan–Meier survival plots of the AR High (gold) vs. Low (blue)groups within a all, b HR + HER2−, c HER2+, and d HR-HER2− patients are shown.