PSMA expression: a potential ally for the pathologist in prostate cancer diagnosis

Prostate cancer (PCa) patients are risk-stratified on the basis of clinical stage and PSA level at diagnosis and the Gleason Score (GS) in prostate biopsy. However, these parameters are not completely accurate in discriminating between high- and low-risk disease, creating a need for a reliable marker to determine aggressiveness. Prostate-specific membrane antigen (PSMA) appears to fulfill this need. We analyzed 79 prostate biopsies and 28 prostatectomies to assess whether PSMA expression detected by immunohistochemistry is related to GS. PSMA expression was correlated with GS in both sample types (biopsies, P < 0.0001 and prostatectomy samples, P = 0.007). We observed lower PSMA expression in Gleason pattern 3 than Gleason pattern 4, suggesting that this biomarker could be useful to distinguish between these entities (p < 0.0001). The best cut-off value of 45% immunopositivity was determined by receiver operating characteristic (ROC) curve analysis. In Gleason pattern 3 vs. Gleason pattern 4 and 5, PSMA sensitivity was 84.1% (95% CI 76.5%-91.7%) and specificity was 95.2% (95% CI 90.6%-99.8%), with an area under the curve of 93.1 (95% CI 88.8–97.4). Our results suggest that PSMA represents a potential ally for the pathologist in the diagnostic work-up of PCa to overcome long-standing morphological classification limits.

or T2b) who require active treatment 4 . Patients with the highest GS score (8)(9)(10) are classified as Grade Group 4 (GS 8) or Grade Group 5 (GS 9 and 10) 4 . Thus, there is clearly an urgent need for more accurate and reliable prognostic markers capable of distinguishing patients who require intensive treatment from who are candidates for a watch-and-wait approach.
Prostate-specific membrane antigen (PSMA) is a non-soluble type 2 integral membrane protein with carboxypeptidase activity, expressed on the apical surface of endothelial cells [5][6][7][8][9][10] . It is weakly expressed in normal prostate tissue but strongly upregulated in prostate cancer 10 . In actual fact, tissue expression of this antigen is not fully prostate-specific as it is also expressed in the neovasculature of numerous solid malignancies 11 . PSMA overexpression is associated with higher PCa grade and androgen deprivation, further increasing in metastatic disease and when castration resistance sets in. This suggests that PSMA plays a functional role in PCa progression, but the correlation with GS and serum PSA is not well established as yet 12,13 .
The aim of the present study was to assess whether PCa patients can be stratified on the basis of PSMA expression and, in particular, to establish whether its expression is related to GS and serum PSA values at diagnosis.

Results
The analysis of PSMA expression was feasible in the overall series of biopsies and prostatectomies. Nineteen bioptic specimens were classified as Grade Group 1 (GS 6), 35 as Grade Group 2-3 (GS 7), 25 as Grade . With regard to the 28 prostatectomies, 6 lesions were Grade Group 1 (GS 6), 19 were Grade Group 2-3 (GS 7) and 3 were Grade Group 4-5. The median PSMA expression in terms of percentage and H-score differed significantly and positively correlated with GS in both biopsies (P < 0.0001) and prostatectomies (P = 0.007) ( Fig. 1) ( Table 1). The same correlation in terms of PSMA staining intensity was observed in both biopsies and prostatectomy specimens (P < 0.0001 and P = 0.006, respectively) ( Table 2). The median value of PSMA expression was similar in the small series of Grade Group 2 and Grade Group 3 lesions analyzed (Table 3). Table 4 reports median PSMA expression values in the different Gleason patterns according to cellular morphology and differentiation. Lower expression of PSMA was observed in Gleason pattern 3 with respect to Gleason patterns 4 and 5, both of which showed higher PSMA expression (P < 0.0001) ( Table 4). Absent or weak PSMA expression was observed in the normal and benign tissue analyzed (Table 4). In addition, stronger PSMA staining intensity was more frequently observed in the less differentiated Gleason patterns 4 and 5 than in Gleason pattern 3 (p < 0.0001) ( Table 5).
The sensitivity and specificity of PSMA expression according to the best cut-off value of 45% determined by the Receiver Operating Characteristic (ROC) curve analysis were 84.1% (95% confidence intervals [CI] Median values of PSMA expression %* (range)   Table 2. PSMA staining intensity in the different Grade Groups. * 0 absent, 1 weak, 2 moderate, 3 strong PSMA staining intensity.  Similar PSMA expression (%, H-score and staining intensity) was observed in biopsies and prostatectomies of the 23 patients for whom data on prostatectomy were also available (Tables 6 and 7).
PSMA expression was positively correlated with basal PSA serum value with respect to both percentage and H-score (Table 8). The median PSA value was 7.85 ng/mL (range 3.2-452.4 ng/mL). No correlation was found between PSMA expression and age. Figure 2 shows 68 Ga-PSMA PET/CT images of in vivo PSMA expression in a 46-year-old male with biochemical relapse after radical prostatectomy for high-risk PCa (Gleason 4 + 4). Figure 2D shows PSMA expression in PCa cells in lymph node biopsy of the same patient.

Discussion
The clinical management of patients with PCa is normally based on serum PSA value, Gleason score and clinical stage 2 . However, these parameters are not 100% accurate in discriminating between low-and high-risk patients, indicating the need for companion biomarkers to better predict disease evolution and improve patient stratification. Although PSMA is a promising biomarker in that it is more specific than PSA, its role in PCa diagnosis and evolution has yet to be confirmed 2,3 . In fact, differences in PSMA expression in terms of cellular or tissue localization may depend on the type of antibody used or sample analyzed. It has also been shown that healthy and benign tissues display PSMA positivity in the cytoplasm and vasculature 11,21,22 . We observed a lower PSMA expression in the well-differentiated Gleason patterns 3 than 4, indicating that this biomarker could be useful to distinguish between these 2 challenging morphological entities. Despite the retrospective nature of the study and the limited number of cases analyzed, we found that PSMA expression in terms of percentage, H-score and intensity may be related to GS, reflecting tumor aggressiveness in both bioptic and radical prostatectomy specimens. This explorative work could thus form the basis for future research to assess the real diagnostic and prognostic value of PSMA. It would also be interesting to see whether PSMA is capable of distinguishing between high-grade prostate intraepithelial neoplasia (PIN), intraductal carcinoma and atypical small acinar proliferation (ASAP).
It is known that GS 3 + 4 lesions are less aggressive than GS 4 + 3 lesions, and several studies have assessed the differences between these 2 subtypes by integrating GS with other prognostic factors and biochemical progression 23,24 . In 2014, a new grading system consisting of 5 prognostically distinct Grade Groups was proposed on the basis of data from Johns Hopkins Hospital. This new classification system partitions the old Gleason score 7 into 2 distinct groups, i.e. Grade Group 2 (Gleason score 3 + 4) and Grade Group 3 (Gleason score 4 + 3), with studies confirming a significantly poorer prognosis for the latter group 4 . This finding takes on an even greater importance   when a GS 3 + 4 lesion is underestimated because the evaluation of the GS depends entirely on the pathologist's expertise 24 . In addition, it is not always easy to quantify and distinguish the proportion of Gleason patterns 3 and 4 in prostate biopsy specimens. Although we did not observe a significant difference between Grade Group 2 and 3 lesions in terms of PSMA expression, our results highlighted the promising role of PSMA in discriminating between well differentiated Gleason patterns 3 and 4 within the same sample.
Although a biopsy generally does not reflect the histology of the entire lesion, it may result in the complete removal of a tumor. Our results showed good concordance in PSMA expression between biopsy and prostatectomy. In agreement with some authors who studied PSMA expression in benign lesions and PCa, we demonstrated that PSMA expression reflects the aggressiveness of the disease 13,25,26 . We also focused our analysis on PSMA expression between Gleason patterns 3 and 4.
The positive correlation between PSMA expression and baseline PSA serum levels makes it a reliable biomarker for clinical decision making.
It can also be hypothesized that different tumor uptake of 68 Ga-PSMA may reflect PSMA expression status. This information could be used to distinguish more aggressive lesions from indolent ones in diagnosis and therapy, as tumors with high PSMA expression may have a high uptake of radiolabeled PSMA ligands. Other markers of aggressive variants of PCa such as those associated with neuroendocrine differentiation (PTEN, p53, BRCA2) could be compared with PSMA expression in patients with advanced disease with poor prognostic features or in those with visceral metastases to have a better overall picture of the biological and clinical impact of PSMA expression [27][28][29] . Furthermore, the use of 68 Ga-PSMA PET/CT as a potential biomarker for the in vivo assessment of PSMA expression could provide reliable information on prognosis and prediction of response to different antitumor agents in advanced PCa, as recently seen for 18F-choline PET/CT [30][31][32] , opening up new avenues of research into PSMA pathological-clinical correlations.
PSMA could prove to be a powerful ally for pathologists in the diagnostic workup of PCa. It could also facilitate the selection of candidates for surveillance/observation programs or local treatment (e.g. prostatectomy or radiotherapy), thus reducing healthcare costs and the risk of overtreatment. Our hypothesis, if confirmed in larger prospective trials, could make a positive impact on/lead to an important change the clinical workup of PCa patients.

Methods
Case series. This exploratory retrospective study included 84 patients with acinar adenocarcinoma of the prostate followed at the Department of Urology of Santa Maria delle Croci Hospital in Ravenna (Italy) from 2013 to 2017. Twenty-three (29%) underwent both prostate biopsy and prostatectomy. Overall, 79 biopsies and 28 prostatectomies were analyzed. The study protocol was reviewed and approved by IRST and AVR (Area Vasta Romagna) Ethics Committee (approval no 1478, 11-Nov-2015). Median age was 64 years (range 46 to 84 years). All of the analyses were carried out in accordance with the relevant guidelines and regulations, and written informed consent was obtained from all study participants. Formalin-fixed paraffin-embedded samples were used for diagnosis. The histology and grading of prostate lesions were established by expert pathologists at the Ravenna hospital in accordance with International Society of Urological Pathology (ISUP) Consensus Conference guidelines 4 .
Given that, at the time of the study 68 Ga-PSMA11 PET/CT was only performed at our institute in patients with biochemical recurrence, we provide the images of a patient that underwent 68 Ga-PSMA 11 PET/CT for biochemical recurrence in relation to PSMA expression detected in the lymph node biopsy.
Immunohistochemistry. Immunostaining for PSMA expression was performed using the Ventana Benchmark XT staining system (Ventana Medical Systems, Tucson, AZ, USA) with Optiview DAB Detection Kit (Ventana Medical Systems). Tissue sections were incubated for 32 minutes with ready-to-use anti-PSMA antibody (SP29 Spring Bioscience, Pleasanton, CA, USA). Sections were automatically counterstained with hematoxylin II (Ventana Medical Systems). PCa and breast cancer tissues were used as positive and negative controls, respectively, in all of the experiments. Biomarker expression was quantified as the percentage of tumor cells with membrane staining out of the total number of tumor cells. Non malignant tissue around the tumor, when present, was also evaluated for PSMA expression. Staining intensity (i.e. 0 absent, 1 weak, 2 moderate, 3 strong) was assessed to calculate the H-score, defined as the product of the percentage of the immunopositive tumor cells and the staining intensity 33,34 . Given that there is still no established cut-off value for PSMA expression, we chose a cut-off of 45% immunopositive tumor cells on the basis of the results from ROC curve analysis. All samples were evaluated by 2 independent observers. Disagreement of >10% positive cells was resolved by consensus after joint review using a multihead microscope. Statistical analysis. Descriptive statistics are reported as counts, proportions, median values and ranges.
The Chi-square test was used to determine the strength of the association between categorical variables. The relationship between median PSMA expression values and GS Nonparametric was evaluated by ranking statistics (Wilcoxon median test), and the Kruskall-Wallis test was used to assess the relationship between PSMA expression values in the different histological patterns. Spearman's rank correlation test was used to investigate the relation between PSMA expression and PSA serum level and age. All P values were based on two-sided testing and values < 0.05 were considered statistically significant. All the statistical analyses were carried out using SAS Software, version 9.4 (SAS Institute, Cary, NC, USA). Data sharing statement. All data generated or analyzed during this study are included in this manuscript.