**3. Results**

### *3.1. AKR1C3 Immunostaining of Cancer Epithelium Is Significantly Stronger than That of Benign Epithelia in Patients with Localized Hormone-Naïve Prostate Cancer*

Clinical and pathological features are demonstrated and the results of statistical analysis of correlation between demographic features and AKR1C3 expression are presented as *p*-values in Table 1. Representative immunostaining of AKR1C3 is presented in Figure 1 (Figures S1–S4, Supplementary Materials). The distribution of AKR1C3 immunostaining scores are presented in Figure 2. AKR1C3 immunostaining was significantly stronger in cancer epithelia than in benign ones within the same spots (*p* < 0.0001). No correlation was observed between GS and AKR1C3 immunostaining in each spot (Table 1). These results suggested that AKR1C3 might play a role in PCa occurrence.

**Figure 1.** Representative immunostainings of aldo-keto reductase family 1 member C3 (AKR1C3): (**a**) score 0 (none staining), (**b**) score 1 (weak staining), (**c**) score 2 (intermediate staining), and (**d**) score 3 (strong staining).

**Figure 2.** Difference of AKR1C3 immunostaining score between benign and cancer epithelia in the same individuals. AKR1C3 immunostaining was significantly stronger in the cancer epithelia than in the benign ones at the same spots (*p* < 0.0001, Pearson's chi-squared test).


**Table 1.** Clinicopathological features of tissue-microarray (TMA) specimens with both benign epithelium and cancer cells in the same spot.

*p*-values indicate correlation of expression intensity with AKR1C3 total score; † Kruskal–Wallis test, †† Pearson's chi-squared test.

### *3.2. AKR1C3 Immunostaining of Cancer Cells Is Statistically Associated with PSA Progression-Free Survival after Radical Prostatectomy*

The distribution of TS of AKR1C3 immunostaining in cancer cells from RP specimens is listed in Table 2. RP cases were dichotomized according to the median TS of the AKR1C3 immunostainings as: AKR1C3 TS ≤ 2 and AKR1C3 TS ≥ 3. AKR1C3 immunostainings and PSA PFS after RP were statistically correlated, and cases with a high AKR1C3 immunostaining TS had lower PSA PFS than those with a low AKR1C3 immunostaining TS (*p* = 0.042) (Figure 3). In order to evaluate prognostic factors for PSA PFS after RP, cox proportional hazards regression analysis was conducted with PSA at diagnosis, Gleason grade group, and AKR1C3 expression. AKR1C3 expression was an independent risk factor of PSA failure among our cohorts (*p* = 0.032, hazard ratio = 2.19) (Table 3). These results showed that AKR1C3 expression of cancer cells may be a prognostic marker of patients who received RP.


**Table 2.** AKR1C3) total score distribution of TMA specimens.


**Figure 3.** Kaplan–Meier survival curves revealed that the AKR1C3 positive group (TS ≥ 3) had a significantly lower PSA PFS rate than the negative group (TS ≤ 2) (*p* = 0.042, log-rank test).



HR: hazard ratio, CI: confidence interval, and ††† Wald test.


We obtained HNPC and CRPC tissues from the same patients in 11 cases; clinical and pathological characteristics are presented in Table 4. CRPC tissues revealed significantly stronger AKR1C3 immunostaining than hormone-naïve tissues in the same cases (*p* = 0.0234, Wilcoxon signed-rank test; Table 4). Interestingly, the longitudinal specimens at hormone-naïve, hormone-sensitive, and castration-resistant states were evaluated in one patient. Immunostainings of AKR1C3 are presented in Figure 4; AKR1C3 was gradually up-regulated with disease progression. These results implied that up-regulation of AKR1C3 might be required for the progression to CRPC in some cases, and it could be a therapeutic target for this complicated disease.


*J. Clin. Med.* **2019**, *8*, 601

**Figure 4.** (**<sup>a</sup>**–**<sup>c</sup>**) AKR1C3 immunostaining of HNPC, Hormone-sensitive prostate cancer (HSPC), and CRPC specimens in the same case (case 9). CAB (leuprolide acetate + bicalutamide) was initiated for case 9 after diagnosis (Figure 3a): (**a**) HNPC (biopsy) specimens at diagnosis, PSA: 29.9 ng/mL (normal reference range 0-4.0 ng/mL), AKR1C3: Proportion score (PS), 1; Intensity score (IS), 1; Total score (TS), 2; (**b**) HSPC specimens on day 45 after commencing CAB, PSA: 2.89 ng/mL, AKR1C3: PS, 2; IS, 2; TS, 4; (**c**) CRPC specimens, PSA: 46.74 ng/mL, AKR1C3: PS, 4; IS, 3; TS 7. After CAB initiation, transurethral lithotomy (TUL) and TUR-P were performed due to repeated urinary retention resulting from bladder stone (Figure 4b). After 1.5 years of bicalutamide, 2 months of flutamide, and 3 months of ethinylestradiol, together with continuous luteinizing hormone-releasing hormone (LHRH) agonist administration, TUR-P was performed due to urinary retention caused by enlargement of the local tumor (Figure 4c). The immunostaining results suggested increased expression of AKR1C3 in PCa tissues with disease progression.
