Overexpression of G6PD Represents a Potential Prognostic Factor in Clear Cell Renal Cell Carcinoma

TCGA ccRCC data mining

Published mRNA expression data of 72 normal kidney tissues and 521 ccRCC cases (with Fuhrman tumor grade information) were downloaded from TCGA (https://tcga-data.nci.nih.gov/tcga/). Details for each patient that have clinical information regarding their Fuhrman grade, survival status and time to follow-up were also extracted from TCGA.

Tissue microarray and patients

75 human ccRCC tissues along with relevant normal adjacent tissue microarray sections with patient's basic parameters and overall survival information were purchased from Shanghai Outdo Biotech Co., LTD. (HKid-CRCC150CS-01). In addition, 74 ccRCC samples with clinical information (obtained from the Department of Pathology, First and Second Affiliated Hospital of Kunming Medical University with the informed consent and the approval from the Research Ethics Committee of Kunming Medical University) were also analyzed. In total, G6PD protein expression levels of tumor and relevant normal adjacent tissues were analyzed in 149 ccRCC patients. These patients were staged according to the TNM classification system of malignant tumors (7th) ¹⁵. The detailed characteristics of patients are listed in Table 1.

Immunohistochemistry

For immunological histological chemistry (IHC) staining, all specimen were subjected in turn to rehydration, antigen retrieval and endogenous peroxidase activity block in 3% H₂O₂, then a Pap pen was used to create a hydrophobic barrier around each tissue sections. After blocked with 5% bovine serum albumin (BSA), rabbit monoclonal anti-G6PD antibody (ab133525, Abcam, Cambridge, U.K.) or isotype control polyclonal rabbit IgG (X0936, DAKO A/S, Glostrup, Denmark) was used at recommended concentration of the instruction for IHC staining at 4°C overnight. Subsequently, Envision Detection Kit (GK500705, DAKO A/S) was used for the bound antibodies detection, and the sections were then counterstained with hematoxylin.

Immunohistologic analysis

A total of 10 random visual fields close to the center of each tumor core at 400 × magnification were examined under microscopy, and the immunostaining was evaluated by two independent investigators in a double-blinded fashion. G6PD expression was determined semi-quantitatively based on the staining intensity and the percentage of cells stained ¹³. The staining intensity was rated as follows: 1 point denoted as weak intensity; 2 points as moderate intensity; 3 points as strong intensity. The percentage of positive cells was rated as follows: 0 points as 0~25%; 2 points as 26~50%; 3 points as 51~75%; 4 points as 76~100%. The expression level of G6PD was scored as the product of staining intensity multiplied by the percentage of stained cells. Based on the final scores, specimens were categorized into negative (-), ≤5% cells were stained, regardless of intensity; weak (+), 1~4 points; moderate (++), 5~8 points; and strong (+++), 9~12 points. For the following statistical analysis, the negative and weak groups (- and +) were considered to be G6PD low expression and the moderate and strong groups (++ and +++) were considered to be G6PD high expression.

Statistical analysis

Statistical analyses were carried out using SPSS 17.0 (SPSS, Chicago, IL). For TCGA analysis, one-way ANOVA was employed to measure the statistical difference between various groups. Patients with clinical outcomes were separated into two groups based on the mean levels of G6PD mRNA expression. Log-rank test was used to measure the statistical difference between the high G6PD and low G6PD groups for Kaplan-Meier curves. For immunohistologic analysis, the correlation between G6PD expression and clinicopathological features of the patient was calculated using the χ² test or Fisher exact test as appropriate. The Kaplan-Meier method was applied to assess the effect of G6PD overexpression on the ccRCC patient survival and the significance was evaluated by the log-rank test. Variables with p<0.05 at the univariate analysis were joined into in multivariate survival analysis by employing the Cox proportional hazards model. Differences were considered to be statistically significant at p<0.05.

Data mining of the expression profile and prognostic value of G6PD in ccRCC patients

mRNA expression levels of G6PD and clinical information of 529 ccRCC cases were extracted from TCGA. From the validation dataset of these cases, we found that compared with normal renal tissues, G6PD gene was significantly upregulated in Fuhrman grade 3 and 4 (G3/4) of ccRCC specimens. Significant differences in G6PD expression were also observed between ccRCC G1/2 and G3/4 (p<0.001, ANOVA, Figure 1A). Moreover, G6PD showed a stepwise increment from G1 to G4 in ccRCC (Figure 1B). These findings prompted us to justify whether the G6PD expression was of significance in this current cohort of ccRCCs. Using the mean value of G6PD mRNA expression levels as cutoff, all 529 ccRCC cases were assigned into G6PD-low group (n=368) and G6PD-high group (n=161). Time to last follow-up was plotted in a Kaplan-Meier curve according to the different groups. These results showed that ccRCC patients who owned higher expression levels of G6PD but with a shorter survival rate (p<0.0001, log-rank test, Figure 1C), indicating that G6PD mRNA overexpression were correlated with poor outcomes in ccRCC.

Analysis of G6PD expression in ccRCC tissues by immunohistologic staining

To verify the above results, the expression pattern of G6PD was further analyzed by IHC in 149 primary ccRCC specimens and relevant cancer-adjacent normal renal tissues. The G6PD levels in the tissue specimens were assessed by the measurement of the final staining scores. By using the same species of IgG antibody as internal control (Figure 2A), we found that G6PD was predominantly located in the cytoplasm of the renal cells (Figure 2B-D). Moreover, weak, moderate and strong positive expression of G6PD was detected in 49.0% (73/149), 28.2% (42/149) and 22.8% (34/149) of the ccRCC tissues (Figure 2B-D top panel), 62.4% (93/149), 30.2% (45/149) and 7.4% (11/149) of the noncancerous renal tissues (Figure 2B-D bottom panel). The moderate and strong groups were considered to be those which had high G6PD expression. Statistical analysis revealed that levels of G6PD were significantly increased in ccRCC tissues when compared with normal adjacent tissues. High expression of G6PD was detected in 51.0% (76/149) of the ccRCC specimens and 37.6% (56/149) of the adjacent samples (Figure 3A), and the score of G6PD expression in ccRCC and adjacent tissues was significantly different (p<0.05, χ² test, Figure 3B).

In addition, high expression levels of G6PD were also found in various ccRCC cell lines when compared with that in HK2 cell (a normal human renal tubular epithelial cell line) (Data not shown). These above results demonstrate that G6PD expression is significantly higher in ccRCC and might be associated with renal tumorigenesis.

Association of G6PD expression with clinicopathological features of ccRCC

To explore the role of G6PD in ccRCC, the correlations between G6PD expression levels and the clinicopathological parameters of the 149 ccRCC patients (Table 1) were analyzed. The results showed that the expression of G6PD was significantly associated with the tumor extent, lymph node metastasis, Fuhrman tumor grade, and TNM stage (p=0.031, p<0.001, p=0.001, and p=0.004, respectively; χ² test, Table 2), but not significantly related to other clinicopathological features such as gender, age at surgery, or tumor max diameter (p>0.05, χ² test, Table 2).

Moreover, we found that in the T stages 1 and 2 (T1/2) of cancer tissues, 46.7% (56/120) cases exhibited strong immunopositivity, whereas 69.0% (20/29) cases exhibited strong immunopositivity in the T3/4 cancer tissues (Figure 4A left panel), and the final score of G6PD was remarkably increased in the T3/4 cancer tissues when compared with the T1/2 tissues (p<0.05, χ² test, Figure 4A right panel). Likewise, the rate and score of G6PD expression in progressively advanced regional lymph node metastasis (p<0.001, χ² test, Figure 4B), Fuhrman tumor grade 3 and 4 (G3/4) (p<0.01, χ² test, Figure 4E), and TNM stage III and IV (III/IV) (p<0.01, χ² test, Figure 4F) was also significantly different when compared with that in each control. However, the G6PD expression levels were not correlated with distant metastasis (p>0.05, Fisher exact test, Figure 4C) or tumor max diameter in ccRCC (p>0.05, χ² test, Figure 4D). These results indicate that higher levels of G6PD might be correlated with ccRCC metastasis.

Correlation of G6PD expression with ccRCC prognosis

In order to further evaluate the value of G6PD in the prognosis of patients with ccRCC, Kaplan-Meier survival analysis and log-rank test were used to analyze the immunohistologic results. We divided the samples into high (n=54) and low (n=63) G6PD expression groups according to the IHC staining scores of G6PD. The results showed that the median survival of the patients with G6PD-high expression (1110 days) was significantly shorter than that of the patients with G6PD-low expression (p<0.001, log-rank test, Figure 5A). Similarly, both in stage I/II ccRCC and stage III/IV ccRCC, patients with higher G6PD expression had worse overall survival (p=0.013 and p=0.034, respectively; log-rank test, Figure 5B and Figure 5C).

Furthermore, the univariate cox regression analysis unraveled that tumor extent (T1/2, T3/4), lymph node metastasis (N0, ≥N1), distant metastasis (M0, M1), Fuhrman tumor grade (G1/2, G3/4), TNM stage (I/II, III/IV) as well as higher levels of G6PD expression were significant predictors of poorer outcome in ccRCC (all p<0.01, Table 3). The other clinical characteristics, such as gender (male, female), age at surgery (<60, ≥60) and tumor max diameter (<7 cm, ≥7 cm) were not statistically significant prognosis factors (all p>0.05, Table 3). When the G6PD level and other joined five factors were analyzed again using multivariate Cox hazard regression, the results showed that G6PD expression (p=0.007, Table 3), as well as Fuhrman grade (p<0.001, Table 3) and TNM stage (p=0.019, Table 3) were independent prognostic factors for ccRCC survival. In conclusion, these results demonstrate that higher expression of G6PD might be served as an independent prognostic factor of poor survival of ccRCC patients.