The protocol was approved by the ethics committee of Kurume University (Kurume, Japan). Between 2003 and 2005, 46 patients were enrolled in the study and underwent radical prostatectomy for prostate cancer at Kurume University Hospital (Kurume, Japan). Following a full explanation of the protocol, written informed consent for the use of tissue samples was obtained from all patients prior to enrollment. Patients with clinically localized prostate cancer who underwent radical retropubic prostatectomy were enrolled, however, patients treated with hormones, irradiation or transuretheral resection prior to surgery were excluded.

Prostatectomy specimens were evaluated using the following technique, with sectioning performed at 3 mm intervals. The grade of each tumor was determined according to the Gleason system of five grades (14). In each patient, the volume of the cancer was determined using a computer-assisted image analysis system (15). Seminal vesicle invasion and positive regional lymph nodes were recorded. A tissue microarray of the prostate was constructed as previously described (16). Briefly, one donor block, which included normal and tumor regions, was selected from 10–15 blocks of formalin-fixed, paraffin-embedded prostate tissue from each patient. Tissue cylinders, with a diameter of 2 mm, were subsequently punched from six regions of normal and cancer tissue in each donor block, using a tissue microprocessor instrument (KIN-type I, AZUMAYA, Tokyo, Japan), and inserted into a recipient paraffin block. Tissue blocks from 46 prostatectomy specimens were evaluated for the expression of EphB6 and PCNA using immunohistochemistry.

Rabbit polyclonal antibody (EphB6 antibody, H-90; cat. no. sc-25461; Santa Cruz Biotechnology, Inc., Santa Cruz, CA, USA) at a dilution of 1:500 and mouse monoclonal (PCNA antibody, PC-10; cat. no. M0879; Dako Corporation, Glostrup, Denmark) IgG at a dilution of 1:200 were used to evaluate the expression of EphB6 and PCNA, respectively, by immunohistochemistry. Briefly, 5 μm thick sections of the selected paraffin blocks were de-paraffinized in xylene, rehydrated in graded alcohol, and incubated in 0.5% hydrogen peroxide/methanol for 20 min to block endogenous peroxidase activity. Antigen retrieval was conducted by boiling the sections in a microwave for 10 min using 10 mM citrate buffer (pH 6.0). The sections were subsequently incubated with anti-EphB6 or anti-PCNA antibody overnight at 4°C. Following this incubation, the sections were washed with 0.5% Tween-20/phosphate buffered saline (PBS) prior to incubation with the corresponding polyclonal peroxidase-labeled goat anti-rabbit and goat anti-mouse secondary antibodies (dilution 1:100; Histofine, Nichirei, Tokyo, Japan) for 60 min. The sections were subsequently washed with 0.5% Tween-20/PBS, and exposed to 3,3′-diaminobenzidine tetrahydrochloride solution (Dako, Carpinteria, USA) to yield an insoluble brown deposit. Finally, the sections were counterstained with hematoxylin, washed in running water, dehydrated in graded alcohol and conventionally mounted. Replacement of the primary antibodies with PBS was used as a negative control for the immunohistochemistry process.

The immunoreactivity of EphB6 and PCNA molecules was evaluated without prior knowledge of the clinicopathological findings. The staining intensity of EphB6 was scored as 0 (negative) when immunoreactivity was absent or present in <10% of cells, and as 1 (weak), 2 (moderate) and 3 (strong) when present in 10–20%, 20–50% and >50% of cells, respectively. The PCNA labeling index (LI) was determined by counting 1000 tumor cells at ×400 magnification in 10 randomly selected microscopic fields. Brown, granular nuclear staining was considered positive staining. The PCNA LI was calculated as the percentage of tumor cells with positive nuclear staining for PCNA (13).

SPSS version 19.0 for Windows (IBM SPSS, Armonk, NY, USA) was used for data analysis. P<0.05 was considered to indicate a statistically significant difference. The change in EphB6 expression in prostate cancer compared with that of corresponding normal tissue was assessed using Wilcoxon’s signed rank test. The frequency of a categorical observation was compared between different groups using the χ² and Fisher’s exact tests, and the correlation between expression status of EphB6 and other continuous variables was evaluated by Spearman’s ρ test. Mann-Whitney U and Kruskal-Wallis tests were used to compare the mean rank of continuous variables between different clinical groups and Kaplan-Meier survival analysis was used to compare the duration of prostate-specific antigen (PSA)-free survival between the different groups.

The demographic characteristics of the investigated patients and pathological features of the tumors are summarized in Table I. Patients were aged from 49–78 years (mean, 65.8 years; median, 66 years). Serum PSA was elevated in all patients, with a minimum value of 4.4 ng/ml and a maximum of 33.9 ng/ml (normal range, <4 ng/ml). The majority of tumors (56.5%) were of Gleason score 7. Cancer volume in prostatic specimens ranged from 0.2–15.2 cm³ (median, 3.7 cm³). Pathological T stages were pT2a in 16, pT2b in seven, T2c in nine, pT3a in 10 and pT3b in four patients. Nodal or distant metastasis was not detected in any of the patients.

The expression of EphB6 was evaluated in normal and prostate cancer tissues from each patient. In normal prostatic tissues, EphB6 expression was observed in 100% of investigated samples. EphB6 protein had a homogeneous cytoplasmic and membranous distribution, and the immunoreactivity was either moderate or strong (38% and 62% of samples, respectively; Fig. 1A and B). In prostate cancer tissue, EphB6 expression was detected in the majority of cases (97.8%). The distribution was also membranous and cytoplasmic, and the expression level was negative or weak in a high proportion of cases (26 cases, 56.5%) and moderate or strong in 17 (37%) and three (6.5%) cases, respectively (Fig. 1C–F). Compared with corresponding normal tissue within the same patient, prostate cancer cells showed a significantly decreased expression level of EphB6 in 26 cases and retained a similar expression level in the remaining cases (Wilcoxon signed rank test, P<0.0001).

The expression of PCNA, which is predominantly nuclear, was detected in all investigated patients. The minimum LI of PCNA was 1% and the maximum was 98%, with a median value of 22%. Representative PCNA expressions in prostate cancer are shown in Fig. 2. No significant association between the expression status of PCNA and EphB6 was observed (Spearman’s ρ=0.193, P=0.173).

The association between tumor characteristics (including Gleason score, cancer volume and pathological stage) and EphB6 or PCNA expression were evaluated (Fig. 3). The results revealed that low expression of EphB6 was significantly associated with a high volume (≥4 cm³) of cancer (P=0.015) and advanced pathological stage (pT3) (P=0.0007). However, none of the tumor characteristics were associated with PCNA expression.

The effect of EphB6 expression on biochemical progression-free (PSA-free) survival in prostate cancer patients was evaluated (Fig. 4). The minimum follow-up duration of the investigated patients was 12 months and the maximum was 120 months (median, 47.5 months). According to the Kaplan-Meier analysis, biochemical progression-free survival was reduced in patients with negative or weak expression of EphB6, compared with that of patients with mild or strong expression (hazard ratio, 2.227; 95% CI, 0.7353–6.745; log-rank, P=0.157).