Androgen Receptor Expression in Epithelial and Stromal Cells of Prostatic Carcinoma and Benign Prostatic Hyperplasia

BACKGROUND: Prostatic carcinoma (PCa) derives from prostatic epithelial cells. However stromal microenvironment, associated with malignant epithelium, also plays a role in prostatic carcinogenesis. Alterations in prostatic stromal cells contribute to the loss of growth control in epithelial cells that lead to progression of PCa. AIM: To analyse the differences between Androgen Receptor (AR) expression in both epithelial and stromal cells in PCa and the surrounding benign prostatic hyperplasia (BPH) and to compare the results with tumour grade. MATERIAL AND METHODS: Samples from 70 cases of radical prostatectomy specimens were used. The expression and intensity of the signal for AR was analysed in the epithelial and stromal cells of PCa and BPH, and the data was quantified using histological score (H-score). RESULTS: AR showed significantly lower expression in both epithelial and stromal cells of PCa compared to BPH. In PCa a significant positive correlation of AR expression was found between stromal and epithelial cells of PCa. AR expression showed a correlation between the stromal cells of PCa and tumour grade. CONCLUSION: AR expression is reduced in epithelial and stromal cells of PCa. Expression of AR in stromal cells of PCa significantly correlates with tumour grade.


Introduction
Prostatic carcinoma (P Ca) is the most frequently diagnosed malignancy and the second leading cause of cancer-related death in men in industrialized countries [1]. Androgens play a vital role in growth, differentiation and maintenance of prostate tissue via Androgen Recept or (A R). AR in stromal cells contributes to the development and growth of prostate during fetal stages as well as during prostate carcinogenesis and cancer progression [1]. Researchers have mainly focused on studying epithelial A R ex pression whereas there is limited dat a concerning stromal AR expression.
AR expression represents a potential prognostic marker for prostatic carcinoma, but more studies are needed in order to prove the usefulness of this factor in the future.

Materials and Methods
We reviewed 70 patients who underwent radical prostatectomy for prostatic carcinoma in the Pathology Laboratory in Clinical Hospital Acibadem -Sistina, Skopje, Macedonia between January 2010 and July 2015. All cases pres ented with localized disease without the lymph node metastatic involvement and all cases were assigned to acinar types of prostatic adenocarcinoma. Representative samples were chosen from the periphery of P Ca that contained relatively equal amounts of tumour and surrounding benign prostatic hyperplasia (BP H).
The immunohistochemical analysis was performed using Androgen Receptor (AR) mous e antibody clone A R441 (DAKO) isotype IgG1 kappa dilution 1: 50. For visualization of the antigen-antibody complex DAKO REALTM En VisionTM Detection System. Peroxidase/DAB+, Rabbit/Mouse was used.
To achieve a relevant analysis of the immunostaining, we analysed the positive signal and the intensity of the signal. We found the positive nuclear signal in stromal and epithelial cells of prostatic carcinoma and normal prostate tissue. The signal was detected on small magnification (x 40) in order to find an area with the most intense signal in the first step. Then on large magnification (x 400), we counted at least 100 epithelial and stromal cells. The number of positive nuclei was shown as a percentage of the total number of counted nuclei. The intensity of the signal was graded on a scale from 0 to 3 where the 0=no signal, 1= weak signal, 2=moderat e signal and 3=strong signal. Additionally a histological score (H-score) was constructed that measures the intensity and distribution of the signal using the formula: [3 (strong signal) x (percentage of cells with strong signal)] + [2 (moderate signal) x (percentage of cells with moderate signal)] + [1 (weak signal) x (percentage of cells with weak signal)]. The histological score ranges from 0 to 300 [2,3]. The data were compared bet ween stromal and epithelial cells in PCa and the surrounding BPH.
Additionally, the data were c ompared wit h tumour grade using the Gleason score retrieved from the medical records. Gleason score was grouped in Gleason prognostic grade groups as follo ws: Gleason prognostic grade group 1 -Gleason sum 1-6; Gleason prognostic grade group 2 -Gleason score 3+4; Gleason prognostic grade group 3 -Gleason score 4+3; Gleason prognostic grade group 4 -Gleason score 4+4 and Gleason prognostic grade group 5 -Gleason sum 9-10.
For statistical analyses of the data the following methods were used: descriptive methods (average, median), methods of testing significance of differenc es among analyzed paramet ers (Chi-square test, Student t test, Wilcoxon matched test, Analysis of Variance, Kruskal-Wallis ANOVA) and methods of determining correlation among designated parameters (Spearman coefficient of rang correlation and Pearson coefficient of correlation). Statistically significant values were determined to be p < 0.05 and highly statistically significant values were determined to be p < 0.01.

Results
AR immunoreactivity was exclusively nuclear and was detected in tumour epithelial cells, epithelial cells of benign glands, peritumoral stromal cells and intraglandular stroma in BP H ( Fig.1A and Fig. 1B ).

Figure 1: A) Expression of Androgen Receptor in epithelial and stromal cells in prostatic carcinoma (Androgen Receptor x 400); B) Expression of Androgen Receptor in epithelial and stromal cells in benign prostatic tissue (Androgen Receptor x 400); C) Heterogenous signal in epithelial cells of prostatic carcinoma in poorly differentiated tumors (Androgen Receptor x 600)
The expression of AR in the epithelial cells of PCa is significantly (Wilcoxon Matched Pairs Z = 3.1 p = 0.002) lower compared to the expression of A R in BPH (med 257.5 vs . 267). We registered significantly lower (t = 9.5 p < 0.001) average values of A R expression in P Ca compared to BP H (114.5 vs . 161) in the stromal cells. The average value of A R expression in the epithelial cells of PCa is 253.73 ± 22.9 and is significantly higher than the average value of AR expression in stromal cells of PCa with a score of 114.2 ± 37.3 (t = 34.7, p < 0.01). The value of A R expression in the epithelial cells of BPH is significantly (Wilcoxon Matched Pairs Z = 7.3 p < 0.01) higher than the expression of AR in stromal cells of BPH (med 267 vs. 161).  These data show t hat there was a drop in A R expression in both stromal and epithelial cells in prostatic carcinoma compartment but the drop of A R expression was more pronounc ed in the stromal compartment of prostatic carcinoma.
We studied the correlation of AR ex pression between t he epithelial and stromal cells of P Ca and BPH. The value of Pearson c oefficient of linear correlation shows that expression of A R in the stromal cells of PCa significantly positively correlates with A R expression in the epithelial cells of P Ca (r = 0.046 p < 0.01). The correlation bet ween AR expression in the epithelial and stromal cells of BPH is positive but statistically not significant (r = 0.22 p > 0.05). A positive but not significant correlation exists between AR expression in epithelial cells of PCa and epithelial cells of BPH (R = 0.056, p = 0.6). Also, AR ex pression between stromal cells of PCa and BP H shows positive but not significant correlation (R = 0.094, p = 0.44).
The analyses of P Ca in relation to tumor grade (Gleason score) according to the new Gleason score grouping showed that 12 cases belonged to Gleason prognostic grade group 1 (Gleas on sum 2-6), 23 cases belonged to Gleason prognostic grade group 2 (Gleason score 3+ 4), 24 cases belonged to Gleason prognostic grade group 3 (Gleason score 4+ 3), 3 cases belonged to Gleason prognostic grade group 4 (Gleason score 4+4), and 8 cases belonged t o Gleason prognostic grade group 5 (Gleason sum 9 -10) ( Table 2). There was no significant statistical differenc e in the average values of AR expression and Gleason prognostic grade group (F = 0.25, p = 0.9) in the epithelial cells of PCa (Table 3). The average value of AR expression in the stroma of P Ca showed a statistically significant differenc e compared to Gleas on grade group (F = 4.33, p = 0.0036). In the Gleason prognostic grade group 1 the average value was 144.0 ± 41 that was significantly higher compared to Gleason prognostic grade groups 3, 4 and 5 (p = 0.033; p = 0.0034; p = 0.0085 consecutively) ( Table 4). We also analysed the correlation between Gleason grade group and AR in bot h the epithelial and stromal compartments of P Ca. Tumor differentiation ex pressed through Gleason score did not show significant correlation with AR expression in the epithelial cells of P Ca (R = -0.15, p = 0.02). The correlation bet ween Gleason score and A R expression in the stromal cells of PCa was negative and statistically significant (R = -0.44. p < 0.01) which means that as a tumour showed less differentiation AR expression in the stromal cells of prostatic carcinoma decreased significantly and vic e versa.

Discussion
The role of stroma in prostate development, prostate function and the maintenance of tissue differentiation is well established [1]. Androgen Receptor (A R) plays a critical role in prostatic development through regulation of androgen effects on epit helial cells. AR expression is mainly localized in the mesenchymal tissue in the fetal period while A R is mainly localized in epithelial compartment in the postnatal period [1]. The mesenchymal AR initiates and controls proliferation and differentiation of epithelial cells while epithelial AR plays a role in functioning and differentiation of the prostatic gland hence stromalepithelial interactions are reciprocal in the development of mat ure prostatic tissue [1].
AR in epithelial cells of prostatic carcinoma was studied by several authors [3 -18]. Some authors found greater expression of A R in epithelial cells in well-differentiated tumours compared to moderately and poorly differentiated tumours [4][5][6][7][8]. In our study also, there was a slightly greater expression of AR in epithelial cells in better-differentiated tumours but the differenc e was statistically not significant (F = 0.25, p = 0.9). In some prospective studies, authors presented higher expression of AR to be associated with better prognosis [4,5,9] while other authors suggested higher AR expression in t he epithelial cell to be associated with worse prognosis [10][11][12][13]. However, most authors agree that a phenomenon known as A R expression heterogeneity was a consistent finding in poorly differentiated tumours [4,8,9,14,15]. This phenomenon was also obs erved in our study in poorly differentiated carcinomas (Fig. 1C).
The discrepancies in these findings can be attributed to several factors. First of all, the lack of unified criteria about the met hod used in the analyses could be an important factor. Second the specimens can also attribute to these varied results since some authors performed their analyses on biopsy specimens [4,5,9,14], some of transurethral resection specimens [5,6,16,17] and most of them on radical prostatectomy specimens [3,7,8,[10][11][12][13]18]. At the end AR expression heterogeneity complicates the matter further since the selection of analysed area is compromised. In order to eliminat e any possible miss assessments, we used only radical prostatectomy specimens and selected the area of analysis where the staining signal was strongest, and also we used the H-score t o incorporate both the staining intensity and percent age of stained cells as it is specified in the section of Mat erial and Methods. With this approach, we tried t o address these issues concerning the inc onsistencies in previous assessments of AR expression. Concerning the issue of the method used to assess AR expression, we used H-score as was recommended by some autho rs as a most valid met hod of A R expression analysis [3]. The second issue concerning material used to assess AR ex pression we only used radical prostatectomy specimens as they contain a lot of material to choos e adequate samples for analysis. In the end, the problem of het erogeneity was address by choosing areas of highest staining intensity and performing the analyses in the chos en areas as was recommended by some authors [3].
Stromal expression of A R was much less studied t han an epithelial expression of AR in P Ca [1,19]. Authors found a decline in stromal A R ex pression in the stromal cells surrounding prostatic cancer. Also, the drop of expression was higher in poorly differentiated carcinomas. Authors found a statistically significant decrease of strom al AR expression in carcinomas compared to stromal AR expression in the benign prostatic hyperplasia surrounding the prostatic carcinoma [1]. These findings are consistent with our data. In our study, the decrease in A R expression in the stromal cells in poorly differentiated carcinomas showed statistical significance. There was a negative statistically significant correlation of A R ex pression compared to Gleason grade group meaning that A R expression was significantly higher in welldifferentiated c arcinomas compared to poorly differentiated carcinomas.
AR expression belongs to the category of prognostic factors of prostatic carcinoma [20,21]. There are t hree categories of prognostic factors. The first group encompasses well established histopathological factors of prostatic carcinoma like the pathologic stage of the disease, status of surgical margins in radical prostatectomy specimens and Gleason grade. The second category is comprised of factors that are presumed to be established in the near fut ure like histologic al types of prostatic carcinoma, tumour volume, and DNA ploidy. The third category belongs to factors that do not possess enough dat a that they may represent prognostic factors shortly. These factors are genetic markers, neuroendocrine markers, proliferative markers, perineurial, vascular or lymphatic invasion, small vessel density, nuclear morphometry and A R expression of epithelial cells [20,21]. For this third category of prognostic factors, authors are welcomed to perform studies in order to prove their clinical usefulness [20,21].
AR expression in epithelial cells of prostatic carcinoma has been studied more than AR ex pression in stromal cells. Our study shows the significance of AR expression assessment in both epithelial and stromal cells because both epit helial and stromal cells may contribute to initiation and progression of prostatic carcinoma.
Further studies are needed in order to prove the clinical us efulness of this potential prognostic factor and correlations are needed with the already well establis hed prognostic factors like the Gleason grade.
In conclusion, o ur study shows a significant drop of A R expression in bot h epithelial and stomal cells of prostatic carcinoma compared to benign prostatic tissue that indicates that there is a quantitative c hange in AR expression in the malignant prostatic tissue. The drop in AR expression is more pronounced in the stromal cells. Also, this drop of A R expression continues as prostatic carcinomas evolve from well differentiated to poorly differentiated carcinomas. The drop of AR expression in the epithelial cells, as they progress to poorly differentiated carcinomas, is not statistically significant but the drop of AR expression in the stromal cells is statistically significant. This concludes that AR expression in stromal cells of prostatic carcinoma could represent one of the prognostic factors for prostatic carcinoma in the future.