Co‐assessment of cytoplasmic and nuclear androgen receptor location in prostate specimens: potential implications for prostate cancer development and prognosis

Abstract

To address, by co-assessing cytoplasmic and nuclear androgen receptor (AR) expression in prostate tissues, the contribution of the AR throughout the stages of prostate cancer (PC) and its value as a marker for predicting biochemical recurrence (BCR) after radical prostatectomy (RP). Archival prostate specimens from patients who were cancer-free (43), with hormone-sensitive prostate cancer (HSPC, 62), and with androgen-independent prostate cancer (AIPC, 30) were used to construct tissue microarrays (total 135). Prostatic intraepithelial neoplasia (PIN) and non-neoplastic tissues (NA) found adjacent to HSPC were also included. Nuclear and cytoplasmic AR expression was scored by two observers using a composite scale, after immunohistochemical detection of the AR. The nuclear/cytoplasmic AR expression ratio was also calculated. Univariate Kaplan-Meier plots, and multivariate Cox and survival-tree analyses, were then used to assess the ability of the AR to predict BCR in the patients with HSPC. There was markedly greater nuclear AR staining intensity in NA than in normal prostate tissues from cancer-free patients. Cytoplasmic AR expression was highest in AIPC and markedly more than in HSPC. The nuclear/cytoplasmic AR expression ratio was highest in NA and PIN. In univariate analyses, a low nuclear AR, low cytoplasmic AR, and a high nuclear/cytoplasmic AR expression ratio were associated with BCR. Although cytoplasmic AR was an independent predictor of BCR in a Cox multivariate model (hazard ratio 2.736, 95% confidence interval 1.228-6.091, P = 0.014), survival-tree analyses suggested a complex relationship between AR expression and clinicopathological features. We propose that increased nuclear AR expression might be a precursor to PC and that cytoplasmic AR could contribute to the AIPC phenotype. The predictive ability of the AR might be closely linked to clinicopathological features.