BACKGROUND/AIM: The value of androgen receptor (AR) in breast cancer has gained renewed interest as a prognostic and treatment predictive biomarker. The aims of this work were to study the associations and the prognostic value of AR in patients from two clinical cohorts.

MATERIALS AND METHODS: Cohort 1 included 208 premenopausal, node-negative patients of whom 87% had received no adjuvant medical treatment; cohort 2 consisted of 263 patients with stage II disease who had all received 2 years of adjuvant tamoxifen. A semi-quantitative assessment of nuclear AR expression divided into five groups (0-1%, 2-10%, 11-50%, 51-75%, and 76-100%) was performed. Survival analyses, stratified by cohort, were performed using both a trend-test and a cut-off of >10% for positivity.

RESULTS: A total of 76% of all patients were AR+, and 89%, 48%, and 23% of the estrogen receptor-positive, negative, and triple-negative, respectively. In Cox regression, stratified by cohort, AR divided into five groups was a prognostic factor for 5-year distant disease-free survival with a hazard ratio of 0.86 per step in fraction score (p=0.018). With a predefined cut-off at 10%, moderate evidence of an effect remained (Hazard Ratio=0.67, p=0.077). In multivariable analysis, AR did not retain an independent prognostic value.

CONCLUSION: AR is a weak, however, not independent prognostic factor for distant metastasis. Although the prognostic value of AR may be questionable, the study identified a subset of AR-positive triple-negative patients as being potential candidates for AR-directed therapy for which further studies are warranted.

Original languageEnglish
Pages (from-to)6845-6853
Number of pages9
JournalAnticancer research
Issue number12
Publication statusPublished - 2017 Dec 1

Subject classification (UKÄ)

  • Cancer and Oncology


  • Androgen receptor status
  • breast cancer
  • endocrine treatment
  • prognosis
  • steroid receptors


Dive into the research topics of 'Androgen Receptor in Stage I-II Primary Breast Cancer -Prognostic Value and Distribution in Subgroups'. Together they form a unique fingerprint.

Cite this