The role of CXC-chemokine receptor CXCR2 and suppressor of cytokine signaling-3 (SOCS-3) in renal cell carcinoma

This is a study of 118 patients with RCC (diagnosed between 1996 and 2008) for whom archival primary tumor material at diagnosis, prior to chemotherapy, was available. In all cases, the histological diagnosis and grading were peer-reviewed by two pathologists (PK, AS) according to the principles laid down in the World Health Organization classification [21]. This study was approved by the University of Athens Medical School Ethics Committee and informed consent was obtained from each patient before study enrollment. The stage of each tumor was assigned following the guidelines from the 7th edition of TNM classification [22] and was known for 106 patients: 28 patients had stage I and 15 had stage II, 24 had stage III and 39 had stage IV disease. Follow-up information was available for 94 patients. The characteristics of patients enrolled in the present study are presented in Table 1 .

Immunostaining was performed on paraffin-embedded 4 μm sections of formalin fixed tumor tissue using the two-step peroxidase conjugated polymer technique (DAKO Envision kit, DAKO, Carpinteria, CA). The primary antibodies used are listed in Table 2. In negative controls primary antibodies were substituted with non-immune serum.

Evaluation of immunostained slides stained with IL-8, IL-6, SOCS3, CXCR2, p65/RelA, HIF-1a, p53 and VEGF was performed using light microscopy by two experienced pathologists (PK, AS) without knowledge of the clinical information and a Histo-score (H-score) based on the percentage of neoplastic cells displaying cytoplasmic immjunoreactivity multiplied by staining intensity was calculated. p65/RelA, HIF-1a and p53 were assessed in 30 random cases, whereas the remaining antibodies in the entire cohort. p-STAT3 nuclear staining and microvessel characteristics were evaluated using computerized image analysis software Image Pro software v5.1 (Media Cybernetics Inc.) on a Pentium III PC, as described previously [23]. The stained slides for CD31 and anti-pSTAT-3 were examined field by field at low magnification (×40 OLYMPUS BX51TF microscope) to identify the area showing the most intense vascularisation (i.e. the “hot spot”) and the highest H-score respectively. For CD31 the vascular hot spot area was photographed at ×200 magnification (OLYMPUS SC-30 Digital Camera) and stored as TIFF image file (2048 × 1532 pixels, RGB, 24-bit). For each countable microvessel several morphometric parameters were automatically established: major axis length (i.e. the distance between the two points along the vessel periphery that are further apart), minor axis length (i.e. the longest axis perpendicular to major axis formed by two points along the vessel periphery), perimeter, area (luminal plus endothelial cell area), Feret diameter (4 * area π), shape factor (4 π * area perimete r 2), compactness (perimete r 2 area), MVD (microvessel density, i.e. the total count of microvessels per optical field) and TVA (total vascular area, i.e. the total area occupied by microvessels). For each case the mean value of major and minor axis length, area, perimeter, Feret diameter, shape factor and compactness along with MVD and TVA were recorded for statistical analysis. In cases where the most vascularized area was not obvious, two or more optical fields were photographed and the field with the highest MVD was finally chosen for further analysis.

Western immunobloting analysis of IL-6, IL-8, CXCR2, SOCS-3, p-STAT-3, p-JAK2 and p-c-Jun expression was also performed on five RCC samples. After homogenization and fractionation of fresh frozen tumor tissue, 100 μg protein was separated on a 10% polyacrylamide gel and blotted onto nitrocellulose membranes, probed with primary antibody overnight, followed by incubation with horseradish peroxidase (HRP)-conjugated goat-anti-rabbit IgG or HRP-conjugated goat-anti-mouse IgG secondary antibody (AP132P and AP124P respectively, Chemicon, Millipore, Temecula, CA, USA). The same primary antibodies described in Table 2 were used at the following dilutions: 1:2,000 for anti–p-STAT-3, 1:200 for anti–SOCS-3, anti-CXCR2, and anti-IL-6. The anti-IL-8 antibody was diluted to a concentration of 0.1 μg/mL. The anti-p-JAK2 (sc-16566-R, Santa Cruz, 200 μg/ml) and anti-p-c-Jun (sc-822, Santa Cruz, 200 μg/ml) were diluted at 1:200. Bands were visualized using ECL chemiluminescence detection reagents (Perkin Elmer, Athens, Greece). Relative protein amounts were evaluated by a densitometric analysis using Image J software (La Jolla, CA, USA) and normalized to the corresponding Actin levels. All experiments have been performed at least 3 times and representative results of one experiment are shown.

Statistical analysis was performed by a M.Sc. Biostatistician (GL). In the basic statistical analysis IL-6, IL-8, SOCS-3, CXCR2, VEGF, p65/RelA, HIF-1a, p53, p-STAT-3 expression and microvascular characteristics were treated as continuous variables. Associations with clinicopathological parameters and microvascular characteristics were tested using non-parametric tests with correction for multiple comparisons.

The set of microvascular parameters was subjected to factor analysis using the principal component extraction method. Three factors were extracted. The first factor represented microvessel caliber encompassing area, perimeter, Feret diameter and major and minor axis length. The second one represented microvessel shape (shape factor and compactness), whereas the third one represented the extent of microvascular network. The estimated factor scores were used in multivariate survival analysis.

Survival analysis was performed using death by disease as an endpoint. The effect of various clinicopathological parameters on clinical outcome was assessed by plotting survival curves according to the Kaplan-Meier method and comparing groups using the log-rank test. Numerical variables were categorized on the basis of cut-off values provided by ROC curves. Multivariate analysis was performed using stepwise forward Cox’s proportional hazard estimation model. Power estimation of the log-rank tests regarding SOCS3 and CXCR2 H-score was performed using the Freedman method for estimation of censored data. Statistical calculations were performed using the Statistical package STATA 11.0 for Windows. All results with a two-sided p level ≤0.05 were considered statistically significant, whereas a p-value between 0.05 and 0.10 was considered of borderline significance.

The expression levels by Western blot in the examined 5 cases were found to correlate with the immunohistochemical expression of IL-6, IL-8, CXCR2, p-STAT-3 and SOCS-3 (Figure 1; Additional file 1: Figure S1-5).

IL-6 and IL-8 expression were detected in 101/118 (85.6%) and 58/118 (49.15%) cases with the H-score ranging from 1-300 and 0.01-100 respectively (Figure 1A, B; Additional file 1: Figure S1, Additional file 2: Figure S2). CXCR2 was expressed in 112/118 cases with the H-score ranging from 2-285 (median value in positive cases: 80) (Figure 1C; Additional file 3: Figure S3). Immunoreactivity for all three antibodies was localized in the cytoplasm of neoplastic cells, increasing around necrosis. Endothelial cells and scattered macrophages displayed CXCR2 immunoreactivity which was also seen in the epithelial cells of distal proximal tubules and collecting ducts, albeit at lesser intensity.

IL-6 and IL-8 were coexpressed in 52/118 (44.06%) cases. Coexpression of IL-6 and CXCR2 was observed in 97/118 (82.2%) with only fifteen of CXCR2 positive cases being negative for IL-6 (15/112, 13.4%). Coexpression of IL-8 and CXCR2 was observed in 58/118 (49.15%) cases, with a significant number of CXCR2 positive cases (54/112 48.2%) being negative for IL-8. The vast majority, however, of these cases (45/54, 83.3%) expressed IL-6.

The correlations among the molecules under study are shown in Table 3. A significant positive correlation emerged between IL-6 and CXCR2 as well as between IL-8 and CXCR2. The former relationship, however, lost its statistical significance in a multivariate regression model including VEGF. Moreover, IL-6 H-score was marginally higher in the cases positive for IL-8.

The correlations between the molecules under study and clinicopathological characteristics are shown in Table 4. IL-8 expression levels were positively associated with histological grade and tumor stage (Figures 2A and 3A), the former relationship being of borderline significance. Accordingly, CXCR2 H-score increased in parallel with histological grade and marginally with tumor stage (Figure 2B and 3B). Interestingly, both IL-8 and CXCR2 H-scores were correlated with the presence (Figure 4A, B) and the total number of metastases. All other relationships of IL-6, IL-8 and CXCR2 H-score with clinicopathological features were not significant.

SOCS-3 expression was cytoplasmic or membranous and was detected in 111/118 (94.07%) cases (Figure 1D; Additional file 4: Figure S4). p-STAT-3 expression was nuclear and was recorded in 84/117 (71.79%) (Figure 1E; Additional file 5: Figure S5). Endothelial and inflammatory cells were also positive for SOCS3 and p-STAT-3 and therefore served as internal positive controls. Weak SOCS-3 expression was noted in distal tubules and collecting ducts. Coexpression of SOCS-3 and p-STAT-3 was observed in 79/117 (67.52%) cases. SOCS-3 H-score was marginally higher in the cases positive for p-STAT-3 (p = 0.0707).

SOCS-3 H-score was positively correlated with the presence (Figure 4C) and the total number of metastases, as well as with disease progression. An inverse correlation between p-STAT-3 H-score and histological grade (Figure 2C), the presence (Figure 4D) and the total number of metastases was also established.

VEGF H-score was positively correlated with IL-6, CXCR2 and IL-8, the latter relationship being of marginal significance (Table 3). Interestingly, although SOCS-3 and VEGF seemed to be positively correlated, when we adjusted a multivariate model including VEGF and SOCS-3 H-score along with the presence of metastasis, a parameter with which both molecules were significantly correlated, the respective relationship between VEGF and SOCS-3 failed to attain statistical significance (Table 3).

The correlations among IL-6, IL-8, CXCR-2, SOCS-3 and VEGF H-score andmicrovascular characteristics are shown in Table 5. Significant positive correlations emerged between VEGF H-score and microvessel area, TVA or Feret diameter. Moreover, CXCR2 was inversely correlated with major axis length, perimeter, area, minor axis length, Feret diameter and compactness, the latter four correlations being of marginal significance, whereas it was positively correlated with shape factor. Furthermore, SOCS-3 H-score increased in parallel with shape factor and was inversely correlated with compactness. Although, IL-8 seemed to be negatively correlated with MVD, when we adjusted a multivariate model including MVD and IL-8 H-score along with histological grade, the respective relationship between these two molecules failed to attain statistical significance.

VEGF was correlated with the presence of metastasis (Figure 4E). Moreover, minor axis length was positively correlated with histological grade (p = 0.0684, Figure 2D), being marginally higher in grades III/IV. As expected, clear cell carcinomas displayed higher MVD and TVA as well as compactness (p = 0.0001 for MVD and TVA, and p = 0.0147 for compactness) and lower levels of shape factor (p = 0.0144) when compared to the remaining histological types (Figures 5A-D and 6), consistent with the much higher vascularity of clear cell RCC, as compared to the remaining types [1, 2]. The increased compactness and lower shape factor values of microvessels in clear cell RCC illustrate the presence of collapsed vessels sections indicative of decreased intraluminal pressure (according to Bernoulli’s law) and consequently enhanced intratumoral blood flow [23].

SOCS-3 and CXCR2 were positively correlated with HIF-1a (R = 0.3675, p = 0.0498 for SOCS-3 Figure 7A and R = 0.9050, p < 0.0001 for CXCR2, Figure 7D), p65/RelA (R = 0.6204, p = 0.0003 for SOCS-3 Figure 7B and R = 0.8069, p < 0.0001 for CXCR2, Figure 7E) and p53 (R = 0.4303, p = 0.0198 for SOCS-3, Figure 7C and R = 0.8254, p < 0.0001 for CXCR2, Figure 7F) H-score. The correlations between IL-6 or IL-8 with p65/RelA, HIF-1a, p53 were not significant (p > 0.10).

Furthermore, Western blot analysis of 5 RCC cases revealed that increased expression of SOCS-3 was associated with decreased p-JAK2 and p-c-Jun expression and vice versa (Figure 8).

The results of univariate survival analysis are presented in Table 6. The parameters adversely affecting survival were advanced stage, increased CXCR2 (Figure 9A) and SOCS-3 (Figure 9B) and decreased p-STAT-3 (Figure 9C) H-score although the latter relationship was of marginal significance. The comparison of survival functions among the groups allocated by CXCR2 and SOCS3 H-score had a statistical power of 0.84 and 0.96 respectively at a significance level of 0.05.

The results of multivariate survival analysis are presented in Table 7. SOCS-3 H-score emerged as an independent predictor of adverse prognosis, along with tumor stage.