*2.2. Immunohistochemistry*

Immunohistochemistry was carried out using anti-AKR1C3 antibody (at a dilution of 1:200; Abcam (ab49680, Abcam plc, Cambridge, UK)). As a positive control of anti-AKR1C3 antibody, we used surgical specimens of breast cancer (estrogen receptor (+) and progesterone receptor (+)) [24]. Immunohistostainings were performed using Ventana Discovery Ultra system (Roche diagnostics) as an automatic immunohistostaining apparatus. All specimens were evaluated by two urological pathologists (S.S. and T.Y.). AKR1C3 immunostaining of benign epithelium was relatively homogenous, whereas that of cancer epithelium was heterogeneous. Thus, the strongest immunostaining intensity of AKR1C3 was compared between benign epithelium and cancer cells at each spot. In order to compare AKR1C3 immunostaining in consecutive specimens of cancer cells in each individual, and to evaluate the correlation of AKR1C3 immunostaining of cancer cells with PSA progression-free survival (PFS) after RP, the pathologists evaluated each of the staining proportion and intensity, and the sum of these evaluation scores was considered as the total score (TS). "Proportion score (PS)" was evaluated according to the expression rate of stained tumor cells as: <1% (score 0), 1%–10% (score 1), 11%–33% (score 2), 34%–66% (score 3), and >67% (score 4). "Intensity score (IS)" was evaluated as none (score 0), weak (score 1), intermediate (score 2), and strong (score 3) in most immunostained cells. The Gleason score (GS) of hematoxylin and eosin staining was also evaluated by the urological pathologists.
