mRNA ratios of AR to ESR1 and PGR distinguish breast cancer subtypes based on public datasets and experimental models

Abstract

The role of the androgen receptor (AR) in breast cancer (BC) remains incompletely understood. Here, we conducted a meta-analysis of large-scale microarray transcriptomic datasets to evaluate whether the mRNA expression levels of the androgen receptor gene, relative to those of the estrogen receptor gene (AR/ESR1 ratio) and the progesterone receptor gene (AR/PGR ratio), can help differentiate BC tumor subtypes. Additionally, we used qRT-PCR assays to assess the mRNA levels of the AR/ESR1 and AR/PGR ratios in four cell lines representative of different BC subtypes (MCF7, BT474, MDA-MB453, and MDA-MB231), as well as in breast tissue from a small group of patients (11 cases) stratified by estrogen receptor (ER) status. Our results showed that higher AR gene expression relative to ESR1 and PGR (≥ 2.0 and ≥ 1.54, respectively) were associated with BC patients classified under the Luminal B and HER2-enriched subtypes. Positive values of AR/ESR1 and AR/PGR ratios were also observed in the ER-negative (ER-) cell line MDA-MB453, as well as in tumor tissue from ER- BC patients. Our findings confirm that higher or even positive AR/ESR1 and AR/PGR ratios may be associated with BC cases exhibiting more aggressive clinical and biological features, leading to a worse prognosis.

Keywords: Androgen receptor; Breast cancer; Estrogen receptor; Meta-analysis; Molecular subtypes; Progesterone receptor.

Fig. 1

Association of high AR/ESR1 ratio values with IHC-surrogate subtypes of worse prognosis (Luminal B and HER2-enriched). Forest plots compare Luminal A vs. Luminal B (A), Luminal A vs. HER2-enriched (B), and HER2-enriched vs. TNBC (C) IHC-surrogate subtypes. Odds ratios (ORs) for each dataset are shown as black squares, whose size reflects the relative weight of each study; horizontal lines indicate 95% confidence intervals (CIs), and red diamonds represent the pooled OR with its 95% CI.

Fig. 2

Association of high AR/ESR1 ratio values with the HER2-enriched intrinsic molecular subtype. Forest plot compares HER2-enriched vs. Basal-like intrinsic molecular subtypes. Odds ratios (ORs) for each dataset are shown as black squares, whose size reflects the relative weight of each study; horizontal lines indicate 95% confidence intervals (CIs), and red diamonds represent the pooled OR with its 95% CI.

Fig. 3

Association of high AR/PGR ratio values with IHC-surrogate subtypes of worse prognosis (Luminal B and HER2-enriched). Forest plots compare Luminal A vs. Luminal B (A) and Luminal A vs. HER2-enriched (B) IHC-surrogate subtypes. Odds ratios (ORs) for each dataset are shown as black squares, whose size reflects the relative weight of each study; horizontal lines indicate 95% confidence intervals (CIs), and red diamonds represent the pooled OR with its 95% CI.

Fig. 4

Association of high AR/PGR ratio values with the Luminal B intrinsic molecular subtype. Forest plot compares Luminal A vs. Luminal B intrinsic molecular subtypes. Odds ratios (ORs) for each dataset are shown as black squares, whose size reflects the relative weight of each study; horizontal lines indicate 95% confidence intervals (CIs), and red diamonds represent the pooled OR with its 95% CI.

Fig. 5

mRNA gene expression ratios determined by qRT-PCR. AR/ESR1 ratio in BC cell lines (A). AR/PGR ratio in BC cell lines (B). Mean gene expression ratios in BC tissue of estrogen receptor-positive (ER +) patients, estrogen receptor-negative (ER-) patients and in fibroadenomes (FA) cases (C). Statistical differences (p < 0.05) were observed only for the AR/PGR ratio when ER- cases were compared with ER + and FA cases. p values: Tukey’s multiple comparisons test.