Tissue samples from 61 PC patients in the Department of Urology of the VU University Medical Center were obtained with the approval of the VU University Institutional Review Board. Median age of the patients was 64 years (range, 53–84) and median follow-up time was 67 months (range, 11–178). None of the patients received hormonal or radiation therapy prior to surgery. Tissue sections of each specimen were stained with hematoxylin and eosin and examined by a pathologist for histopathological diagnosis and tumor grading. Subject population was increased with the addition of 40 specimens of benign prostatic hyperplasia and 95 cores from 40 cases of advanced prostate cancer (TMA PR955; US Biomax Inc., Rockville, MD, USA). The clinicopathological data of each specimen including TNM stages, Gleason scores, preoperative serum PSA and survival data were also obtained. In all, 141 cases were examined. The cohorts were classified as follows: i) non-cancer, DRE negative, prostate biopsy negative; ii) cancer, organ-confined: Gleason Score: 3+3, pT2a (absence of extracapsular extension or seminal vesicle invasion); iii) cancer, non-organ confined: Gleason Score 4+4 or higher, pT3 (presence of extracapsular extension and/or seminal vesicle invasion); and iv) cancer, metastatic: confirmed distant metastases including hormone refractoriness.

Paraffin-embedded specimens were deparaffinized, hydrated, subjected to antigen retrieval by heating the slides for 5 min in 5 mM sodium citrate. The sections were then stained for CT/CTR as previously described (14). Incubations with primary antibodies were followed by horseradish peroxidase phosphatase-conjugated secondary antibodies. The slides were then counterstained with hematoxylin.

Antibodies against CT and the peptide (50–65 aa) of human CTR have been previously validated for the IHC of prostate cancer specimens (7). Tissue sections (5 μm) were deparaffinized, hydrated and CT/CTR immunofluorescence was performed as previously described (14).

Tissue sections were incubated either in the presence of no primary antibody, no secondary antibody, or primary antibody blocked with the peptide.

Six images per section were acquired. Immunostaining was scored by two individuals independently using established methods (16) and the mean reading was taken. The staining intensity was assigned an arbitrary value, on a scale of 0–3 as follows: (−), 0; (+/−), 0.5; (+), 1; (++) 2; and (+++), 3.

An IHC index for each sample was calculated by multiplying staining intensity with the percentage of positive cells. The results were graded from 0 (negative) to 300 (all cells display strong staining intensity). Reproducibility of the analysis was verified by rescoring of randomly chosen slides. Duplicate readings gave similar results.

The samples were broadly classified as: i) negative (IHC index = 0); ii) low (IHC index <5); iii) moderate (IHC index = 5–50); iv) high (IHC index = 50–150); and v) very high (IHC index >150).

Statistical calculations were performed using Prism 5 computer program (GraphPad Software, San Diego, CA, USA). Results are generally expressed as mean IHC index ± standard error of the mean (SEM) unless otherwise stated. P<0.05 was considered to indicate a statistically significant result. One-way ANOVA and t-tests were used to compare CT/CTR IHC index across the clinical groups (cancer vs. non-cancer; organ-confined vs. metastatic; Gleason score 6 vs. >8). Receiver operating characteristics (ROC) curves were used to determine diagnostic utility of CT/CTR expression. Area under the curve (AUC) was calculated by numerical integration of each ROC curve. Cut-off points are defined as the values at which sensitivity and specificity are optimized. A two-sided P-value of <0.05 is considered to indicate a statistically significant result. Kaplan-Meier analysis was performed to compare the prognostic ability of CT/CTR expression with that of other clinicopathological parameters.

CT staining in benign acinii was predominantly localized to basal epithelium and was absent in secretory epithelium (Fig. 1). CTR expression displayed a similar profile (data not shown). Comparatively, CT/CTR immunostaining was absent/or weak in benign regions of the prostate, but was stronger in HGPIN and malignant acinii (Fig. 2A and B). Higher magnification images of a PC specimen suggest that basal epithelial cells displaying prominent nucleoli were strongly CT- or CTR-positive (Fig. 2C and D). Bone metastasis of PC displayed intense cytoplasmic CT staining, suggesting tumor cells may secrete CT after implanting in the bone (Fig. 3).

Since CT/CTR expression displayed a similar spatial localization, we evaluated whether CT and CTR are co-expressed in same cell populations. We performed double immunofluorescence-stainings of prostate TMA sections (TI95; US Biomax Inc.). A majority of cells in sections of PC of T stage 2 or 3 stained for either CT (red) or CTR (green) (Fig. 4A1–4). In contrast, a majority of CT-positive cells co-expressed CTR in T stage 4 cases as characterized by the yellow in the merged figure (Fig. 4B4 and C). These results, that CT-CTR axis is predominantly paracrine in earlier stages of PC but converts to autocrine in advanced stages, are consistent with our earlier results that activation of CT-CTR autocrine loop enhances the ability of PC cells to grow and metastasize (8,14,17).

Prostate specimens from pathology as well as TMA showed considerable variability in CT/CTR staining (Fig. 5). Specifically, benign specimens showed very weak stainings that increased to very strong in metastatic PCs.

The generated data of CT/CTR staining in benign prostates were pooled and compared with that from malignant prostates. Among 40 benign prostate specimens, 34 displayed negative staining for CT (85%) and 6 displayed low staining (15%). Among 101 primary PC specimens: 3 were negative (<3%); 25 were low (<25%); 28 were moderate (<28%); 18 were high (<18%); and 36 were very high (<36%).

Significantly higher expression of CT and CTR in PC specimens was evident when assessed either as IHC index or as percent of immunopositive cells (Fig. 6A–D; P<0.0001 for benign vs. adenocarcinoma). Interestingly, higher percentages of cells displayed CT expression than CTR expression. However, the intensity of CTR staining seemed stronger.

We then stratified the data of PC specimens in two groups: localized PC and metastatic PC. The comparison of these two data sets show that metastatic PCs displayed significantly higher CT/CTR expression than localized PCs when assessed either as IHC index or as immunopositive cell populations (Fig. 6E–H). Stratification of the data further into T stages indicated that CT/CTR IHC indices of metastatic PCs in stages 2c and 3a were significantly higher than those of localized PCs in same T stages, raising a possibility that elevated CT/CTR expression in earlier T stages may be an indicator of metastatic disease (Fig. 6I and J).

We then analyzed the results by receiver operator characteristic (ROC) curve in two ways. First, we analyzed CT IHC index data from all PC vs. all benign samples. The curve yielded an area under the curve (AUC) of 0.9936 (P<0.0001) (Fig. 7A). Choosing a cut-off of IHC index of 2.15, a specificity of 100% yielded a sensitivity of 84%, suggesting that CT/CTR IHC index >2.15 can reliably discriminate cancer patients from non-cancer patients. Next, we compared the data from localized cancer with that from metastatic cancer. The ROC curve yielded the AUC of 0.6704 (P=0.0264) (Fig. 7B). With the cut-off of 9.5, a specificity of 75% yielded the sensitivity of 53% for metastatic disease indicating that the test can be reliably used to predict clinical course of the disease. The corresponding analysis of CTR expression data also showed similar predictability (Fig. 7C and D). Both, CT and CTR IHC indices were better predictors for the cancer when compared with preoperative serum PSA levels; and for metastatic disease when compared with the Gleason score (data not shown).

We then compared CT IHC indices of benign and PC specimens with preoperative serum PSA levels of the subjects. This is because serum PSA is currently used as a marker for PC. As shown in Fig. 8A and B, CT IHC index as well as preoperative serum PSA levels of this cohort displayed significant discrimination between benign and PC patients. However, the significance of CTR IHC index between these two groups (benign vs. cancer) was higher than that of preoperative serum PSA levels. Next, we compared the ability of CT IHC index and Gleason score to discriminate local vs. metastatic disease (Fig. 8C and D). The analysis shows that CT IHC index significantly discriminated metastatic disease from localized disease, whereas the Gleason score could not. Likewise, CTR IHC index data also demonstrated similar ability (Fig. 8E–H).

Next, contingency analysis of crosstables also identified a significant association between high CT/CTR IHC indices and metastatic cancers (Table IIA and B). Thus, multiple comparative analyses of CT/CTR expression with the established clinicopathological markers/clinical outcome data demonstrate that CT/CTR expression could potentially serve as a progonostic marker of metastatic PC.

Next, we examined the prognostic ability of CT/CTR expression as well as other clinicopathological markers on metastasis-free survival of prostate cancer patients by Kaplan-Meier analysis. Data of patients were divided into two, high (>150) and low (≤150) CT/CTR IHC index groups; and survival period of the patients was considered as a clinical outcome. Likewise, the patients were also analyzed for other clinicopathological parameters as described in Fig. 9. Patients with high CT/CTR expression demonstrated significantly greater unfavorable disease course than those with low CT/CTR expression (P<0.0005; Fig. 9A and B). Among other clinicopathological parameters, only Gleason score showed prognostic significance. However, the level of significance was remarkably lower than that of CT/CTR expression [P=0.0491 vs. P=0.0005 (for CT/CTR), Fig. 9F]. All other parameters failed the test (Fig. 9C–E; Table IVA and B). We further tested the data with Cox regression analysis. In a stratified univariate analysis, a highly significant prognostic value for high CT/CTR expression was found with lower survival period; whereas in cases with low CT/CTR levels, the survival period was longer (Table IV). Among other parameters, only Gleason score displayed similar prognostic prediction with a lower significance. In multivariate Cox analyses, CT/CTR expression and Gleason scores showed significant prognostic values. Based on this analysis, CT/CTR IHC index is a significantly better prognostic indicator than Gleason score, preoperative serum PSA, age, or tumor stage (Table IV).