We evaluated patients who underwent surgical removal of RCC at the National Defense Medical College, Tokorozawa, Saitama, Japan, between 1994 and 2006, in order to determine clinicopathological factors from clinical records and pathological reports. Paraffin-embedded sections were prepared from surgical specimens removed from 54 patients who underwent radical nephrectomy (RN) and 2 who underwent partial nephrectomy (PN). Follow-up intervals for the 56 patients ranged from 1 to 162 months (median, 62.3 months). Paraffin-embedded sections of 5 metastatic lesions (2 in lymph nodes, 1 in the adrenal gland, 1 subcutaneous and 1 in the pancreas) from 5 different patients were also prepared. sEPO levels were available for 138 patients (121 patients underwent RN and 17 underwent PN), and EpoR expression was determined in surgical specimens removed from 47 of the 138 patients. All serum samples were obtained within a week prior to nephrectomy. Follow-up intervals for the 138 patients ranged from 1 to 129 months (median, 22.3 months). All patients were postoperatively evaluated for local recurrence and metastasis every 3 to 6 months for the first 5 years and every 6 to 12 months after. Follow-up examinations consisted of physical examination, chest radiography, abdominal and chest CT, blood tests and, if indicated, radionuclide bone scanning. The pathological stage was determined according to the 2002 TNM classification system, and a 3-graded system was used for nucleolar grading (13). This study was approved by the institutional review board of the National Defense Medical College, Tokorozawa, Saitama, Japan.

The clinicopathological factors of the 56 patients whose surgical specimens were evaluated in this study are listed in Table I. The patients included 40 males and 16 females between 36 and 78 years of age (median, 61). The median follow-up interval was 62 months (range, 1 to 162). Twenty patients underwent right nephrectomy and 36 patients underwent left nephrectomy. The size of the primary tumor was 6.7±4.1 cm (range, 1.3–20; median 6.3). The predominant histological type was clear cell type in 54 tumors and papillary type in 2. The paraffin-embedded sections contained both tumor and surrounding kidney tissue. Immunohistochemistry was performed as previously described (14). Briefly, paraffin-embedded sections were deparaffinized. Slides were placed in Dako Target Retrieval Solution (Dako Corp., Carpinteria, CA, USA) and heated for antigen retrieval. Endogenous peroxidase activity was quenched with Dako Peroxidase Blocking Reagent (Dako Corp.). Sections were incubated in 10% normal goat serum in phosphate-buffered saline and subsequently incubated overnight with rabbit polyclonal anti-EpoR antibody (C-20; Santa Cruz Biotechnology, Santa Cruz, CA, USA). They were then stained using a Simple Stain Max PO kit (Nichirei Corp., Tokyo, Japan). Reaction products were visualized by immersing the slides in diaminobenzidine tetrahydrochloride. The sections were counterstained with hematoxylin. Tubular epithelial cells, in which EpoR is known to be abundant (15), served as a positive internal control. Samples incubated without primary antibody were also stained using the same method and were used for baseline staining. Immunostaining results in all tumor sections were evaluated by 2 individuals (K.I. and T.A.) blinded to all clinical and pathological variables. The staining intensity of each tumor was compared with that of surrounding proximal tubules. Tumors with a staining intensity greater than that of the surrounding proximal tubules (level 2) were defined as tumors with high EpoR expression (level 3). Those with a staining intensity equal to (level 2) or less than (level 1) that of proximal tubules were tumors with low EpoR expression. Staining level was determined by reviewing the entire slide at ×200 magnification and the dominant level of EpoR expression in each tumor was determined. The sections from all 56 patients were also immunostained for Ki-67 and CD34. The primary antibody for Ki-67 (mouse monoclonal; Zymed Laboratories Inc., South San Francisco, CA, USA) was used at an appropriate dilution. Ten high-power fields (HPFs) in each slide were counted by two independent investigators whose results were averaged. The primary antibody for CD34 (monoclonal; Santa Cruz Biotechnology) was used at a dilution of 1:100, and CD34-positive neovessels in 10 HPFs were counted as previously described (16).

EPO levels in duplicate 50-ml serum samples were measured using an enzyme immunoassay kit (Diagnostic Products Corporation, Los Angeles, CA, USA) according to the manufacturer’s instructions.

Results are presented as the mean ± standard deviation. Variables of different groups were compared using the Mann-Whitney U test. The independence of fit of the categorical data was analyzed by the Chi-square test. The correlation between variables was analyzed using Spearman’s rank correlation coefficients. Survival curves were constructed using the Kaplan-Meier method, and the differences between them were assessed using the log-rank test. Cox’s proportional hazard regression model was used for univariate and multivariate analyses. Logistic regression analysis was used to identify independent factors influencing sEPO elevation. P<0.05 was considered to indicate a statistically significant difference.

Normal renal tubular epithelium clearly expressed EpoR, mainly in the cytoplasm (Fig. 1A); this localization of EpoR is consistent with previous studies (12). EpoR expression in RCC cells was also observed in the cytoplasm. The expression intensity was level 1 in 21 tumors (Fig. 1B), level 2 in 20 tumors and level 3 in 15 tumors (Fig. 1C). RCC cell assemblies in the microvessels surrounding the primary tumors frequently demonstrated high EpoR expression (Fig. 1D). All five RCC metastases demonstrated high EpoR expression (level 3) (Fig. 1E). The RCC cell assemblies in the microvessels surrounding the metastatic lesions demonstrated higher EpoR expression than the metastatic lesions (Fig. 1F).

We compared the clinicopathological factors of 41 patients whose tumors demonstrated low EpoR expression (levels 1 and 2) and 15 patients whose tumors demonstrated high EpoR expression (level 3) (Table I). The patients with high EpoR expression had tumors with a significantly higher pathological T (pT) stage, size and histological grade than the patients whose tumors had low EpoR expression, and those patients also had higher C-reactive protein (CRP) levels and higher percentages of metastatic disease and microvascular invasion (MVI). Patients whose tumors demonstrated high EpoR expression had a cause-specific survival (CSS) rate significantly lower than patients whose tumors demonstrated low EpoR expression (Fig. 2A). In N0M0 patients (n=41), however, disease-free survival (DFS) did not differ significantly between those with low EpoR expression and those with high EpoR expression (Fig. 2B).

The proliferative status of the cells in each of the excised tumors was evaluated by Ki-67 immunostaining (Figs. 3A–C). The number of Ki-67-positive RCC cells per HPF was 7.7±1.0 in level-1, 16.4±2.8 in level-2 and 32.6±5.3 in level-3 tumors. Tumors with higher EpoR expression levels had significantly larger numbers of Ki-67-positive cells, suggesting that EpoR expression is associated with the proliferation of RCC cells.

Level of angiogenesis was evaluated by CD34 immunostaining. The number of CD34-positive neovessels per HPF was 22.1±2.5 in level-1, 16.6±2.6 in level-2 and 19.2±3.5 in level-3 tumors. RCC specimens with low EpoR expression frequently had a large number of neovessels (Fig. 3D), and certain specimens with high EpoR expression had a small number of neovessels (Fig. 3F). There was no correlation between levels of EpoR expression and the number of CD34-positive neovessels.

According to a previous study, invasiveness of RCC was evaluated in terms of pathological growth pattern; infiltrative or expansive (17). Seven of the 21 tumors with level-1 EpoR expression were infiltrative (33%), 14 of the 20 tumors with level-2 expression were infiltrative, and 10 of the 15 tumors with level-3 expression were infiltrative (67%) (P=0.0363, Chi-square test). Infiltrative tumors contributed to a greater percentage of tumors with either level-2 or level-3 EpoR expression compared to tumors with level 1 EpoR expression. Therefore, increased EpoR expression appears to be associated with RCC invasiveness.

We also compared the clinicopathological factors of patients with low sEPO levels with those of patients with high sEPO levels (Table II). Patients with high sEPO had tumors with a significantly higher pT stage, size and histological grade, and also had higher CRP levels and higher percentages of meta-static disease and MVI (P<0.05). Patients with high sEPO had a significantly lower survival rate than those with low sEPO (Fig. 2C). The five-year survival rate for patients with low sEPO was 89%, while for patients with high sEPO it was 52.4%. In N0M0 patients (n=110), however, DFS did not differ significantly between those with low sEPO and those with high sEPO (Fig. 2D).

Of the 47 patients whose surgical specimens were evaluated for EpoR expression and whose sEPO levels were measured, the sEPO level was 17.3±6.4 mU/ml in patients whose tumors demonstrated level-1 EpoR expression, 26.0±6.4 in patients whose tumors demonstrated level-2 expression, and 36.2±7.7 in patients whose tumors demonstrated level-3 expression (P=0.0008, compared to the level-1 value; P=0.053, compared to the level-2 value). Serum EPO level increased in proportion to the increase in tissue EpoR expression level.

As EPO stimulates erythrocyte production and the hemoglobin level is reportedly a prognostic factor in RCC patients, the correlation between sEPO level and hemoglobin level was evaluated. sEPO levels inversely correlated with hemoglobin levels in male and female patients (Fig. 4). The percentage of patients with anemia was greater in patients with high sEPO levels than those with low sEPO levels.

To identify factors that influence sEPO level, we evaluated the correlation between sEPO level and various clinicopathological factors (Table III). As shown in Fig. 5, larger size, higher pT stage, distant metastasis (DM) and higher TNM stage were associated with sEPO elevation. Logistic regression analysis was used to identify independent factors associated with sEPO elevation. Univariate analysis demonstrated that tumor size, pT ≥3, lymph node metastasis (LNM), DM, presence of grade-3 component, MVI, anemia and CRP≥1 were significantly associated with sEPO elevation. In multivariate analysis, only MVI was independently associated with sEPO elevation. This result suggests that EPO-EpoR signaling might promote invasion of RCC cells into small venules or lymphatics. To confirm the association between MVI and sEPO elevation, we divided N0M0 patients with primary tumors 4–7 cm in size into 2 groups. Patients with MVI (n=15; mean tumor size, 5.5±0.2 cm) had significantly higher sEPO levels than those without MVI (n=20, mean tumor size, 5.4±0.2 cm) (sEPO level=36.9±7.9 vs. 18.6±1.0 mU/ml, respectively; P=0.0156). Although tumor size was similar in the two groups, patients with MVI had significantly higher sEPO levels than those without MVI.

As described above, patients with tumors demonstrating higher EpoR expression appeared to have higher sEPO levels than those with lower EpoR expression. It is therefore possible that tumors with increased EpoR expression might have aggressive biological activities and have an increased ability to produce EPO. When patients with elevated sEPO have tumors with increased EpoR expression, the EPO-EpoR pathway might function effectively and increase the aggressiveness of the tumors. To examine this possibility, we divided the 47 patients whose sEPO and specimen EpoR were determined into 4 groups (Fig. 6). The patients with high EpoR expression and a high sEPO level had a worse prognosis than the other patients. Their 1-, 2- and 3-year CSS rates were only 50, 33 and 17%, respectively.