Written informed consent was obtained from all patients, and the study was approved by the institutional review board of Sun Yat-sen University. For real-time RT-PCR analysis, we collected 46 paired samples of ccRCCs and adjacent normal tissues from patients who underwent radical nephrectomy between February 2008 and December 2009. The 46 patients included 40 men and 6 women with a median age of 50 years (range, 37-75 years). The fresh tissues were immediately immersed in RNAlater (Qiagen; Germany) after surgical resection, stored at 4°C overnight to allow thorough penetration of the tissue, and then frozen at -80°C. In addition, we performed an immunohistochemical assay of 211 paraffin-embedded samples of ccRCCs and 107 adjacent normal renal tissue samples collected from patients between 1999 through 2007. The characteristics of these 211 patients are listed in Table 1. None of the patients underwent radiotherapy or chemotherapy before surgery. The histological and clinical diagnosis of the tumors in all these patients was performed by the Cancer Center of Sun Yat-sen University. The disease stage of each patient was classified or reclassified according to the 2002 American Joint Committee on Cancer (AJCC) staging system 9.

An immunohistochemical assay was performed to examine DUSP-9 expression in the 211 ccRCC samples and 107 paired samples of adjacent normal renal tissue. All procedures were performed using classical protocols. In brief, paraffin-embedded specimens were cut into 5-μm sections and baked at 65°C for 30 min. The sections were deparaffinized with xylene and rehydrated, submerged into 0.01 M citrate buffer (pH 6.0) antigen retrieval buffer, and then microwaved for antigen retrieval. They were then treated with 3% hydrogen peroxide in methanol to quench the endogenous peroxidase activity, which was followed by incubation with 10% bovine serum albumin to block nonspecific binding. The DUSP-9 protein was detected by using a mouse monoclonal antibody against DUSP-9 (Abcam; Cambridge, MA, USA). The specimens were incubated overnight at 4°C with anti-DUSP-9 antibody (1:250). The negative control for immunohistochemical analysis was obtained by replacing the primary antibodies with an antibody diluent. After being washed in phosphate buffered saline (PBS), the sections were treated with MaxVision™ HRP-Polymer anti-Mouse IHC Kit (Maixin Bio; Fujian, China) at 37°C for 15-20 min. The tissue sections were immersed in 3-amino-9-ethyl carbazole, counterstained with Mayer's hematoxylin, dehydrated, and finally mounted in Crystal Mount.

The formalin-fixed, paraffin-embedded sections were reviewed for the degree of immunostaining and scored by 2 independent observers. The proportion of cells expressing DUSP-9 varied from 0% to 100%, and the intensity of staining varied from weak to strong. The proportion of DUSP-9- expressing tumor cells was scored as follows: 0, no positive cells; 1, <5%; 2, 6%-25%; 3, 26%-50%; 4, 51%-75%; and 5, >75% according to Tsuchiya et al. The staining intensity was graded according to the mean optical density 101112: 0, no staining; 1, weak staining (light yellow); 2, moderate staining (yellow brown); and 3, strong staining (brown). Staining index was calculated as the multiplication of staining intensity score and the proportion of DUSP-9-positive tumor cells. We evaluated DUSP-9 expression in benign kidney tissue and malignant lesions on the basis of the staining index values, with scores of 0, 1, 2, 3, 4, 5, 6, 8, 9, 10, 12, and 15. The cutoff values for DUSP-9 expression were chosen on the basis of a measure of heterogeneity in overall survival rates, which was calculated using the log-rank test. An optimal cutoff value was identified: a staining index score of ≥5 was considered as high DUSP-9 expression, whereas a staining index score of ≤4 was considered as low DUSP-9 expression.

All statistical analysis was carried out with the SPSS 17.0 statistical software package. In the real-time RT-PCR and immunohistochemical assays, paired-sample t tests were used to analyze the significance of the differences in mRNA and protein expression between ccRCCs and the adjacent normal tissues. The χ²-test for proportion was used to analyze the relationship between DUSP-9 expression and clinical significance. Survival curves were plotted by the Kaplan-Meier method and compared by the log-rank test. We determined that the assumption of proportional hazards was met in all Cox regression models. The significance of various variables for survival was analyzed by the Cox proportional hazards model in multivariate analysis. p < 0.05 was considered to be statistically significant.

The transcription level of DUSP-9 was determined by quantitative RT-PCR assays of 46 ccRCC tumor samples and the paired adjacent normal tissue samples. In 45 tumor samples, the mRNA level of DUSP-9 was significantly lower than that in the adjacent normal tissue sample (p < 0.001, paired-sample t tests, Figure 1).

Expression and subcellular localization of protein were determined by immunohistochemical analysis in 107 paraffin-embedded ccRCC tissues and 107 paired specimens of adjacent normal tissues (Figure 2). In normal renal tissue, specific DUSP-9 was localized mainly in the cytoplasm of renal cells in the form of yellow-brown granules (Figure 3A). The DUSP-9 protein expression in the 102 tumor tissue samples was lower than that in the adjacent normal tissue samples (p < 0.001, paired-sample t test, Figure 3B).

To further investigate the effect and the prognostic value of DUSP-9, immunohistochemical analysis was performed to assess the expression of DUSP-9 in 211 ccRCC tissue blocks. Overall, 117 of the 211 tumor samples showed low expression of DUSP-9(score ≤4), whereas 94 samples showed high expression (score ≥5). The correlation between the expression of DUSP-9 and various clinicopathological parameters are listed in Table 1. Intense expression of DUSP-9 in ccRCC samples was correlated with gender (p = 0.031), pathologic stage (p = 0.001), Fuhrman grade (p = 0.002), tumor size (p = 0.001), T stage (p = 0.001), N classification (p = 0.012), metastasis (p = 0.005), and recurrence (p < 0.001), but it was not correlated with age. Low expression of DUSP-9 was noted in 46.7%, 60.0%, and 80.5% of T1, T2, and T3/4 stage ccRCCs respectively (p = 0.001, χ2 test). Low expression of DUSP-9 was not observed in 47.1% and 70.7% of ccRCCs with size ≤7 cm and >7 cm respectively (p = 0.001, χ2 test). Low expression of DUSP-9 was seen in 52.2% and 77.8% of N0 and N1/2 stage ccRCCs respectively (p = 0.012, χ2 test). Low expression of DUSP-9 was seen in 52.1% and 82.6% of ccRCCs with or without metastasis respectively (p = 0.005, χ2 test). Low expression of DUSP-9 protein was seen in 48.6% and 96.6% of ccRCCs with or without recurrence respectively (p < 0.001, χ² test).

Kaplan-Meier analysis and the log-rank test were used to calculate the effect of the DUSP-9 expression on survival. The 5-year survival in the group of patients with high DUSP-9 expression was 97%, but it was 62.1% in the group of patients with low DUSP-9 expression (Figure 4A). The log-rank test showed that survival rates were significantly different between these 2 groups (p < 0.001). Furthermore, the relationship of DUSP-9 expression with prognosis was determined in 211 patients, which were divided into 3 subgroups depending on the pathologic stage. Patients with tumors exhibiting high DUSP-9 expression had significantly longer overall survival than those with low expression of DUSP-9 either in the stage I plus II subgroup (n = 151; log-rank, p = 0.023; Figure 4B), the stage III sub group (n = 31; log-rank, p = 0.036; Figure 4C), or the stage IV subgroup (n = 29; log-rank, p = 0.038; Figure 4D). Patients with tumors high DUSP-9 expression had significantly longer overall survival than those with low expression of DUSP-9 either in the Fuhrman grade I subgroup(n = 39; log-rank, p = 0.005; Additional file 1 Figure S1A), II subgroup (n = 123; log-rank, p < 0.0001; Additional file 1 Figure S1B), the stage III sub group (n = 34; log-rank, p < 0.001; Additional file 1 Figure S1C), or the stage IV subgroup (n = 15; log-rank, p = 0.019; Additional file 1 Figure S1D).

Univariate Cox regression analysis showed that tumor size, T stage, N stage, metastasis, Fuhrman grade and DUSP-9 expression were significantly associated with overall survival (Table 2). Furthermore, multivariate Cox regression analysis revealed that only DUSP-9 expression and Fuhrman grade were independent predictors for the overall survival of ccRCC patients (p = 0.005, respectively; Table 2), whereas the others factors were not independently related to the survival of ccRCC patients.