Low differentiated microvascular density and low expression of platelet-derived growth factor-BB (PDGF-BB) predict distant metastasis and poor prognosis in clear cell renal cell carcinoma

Authors


Correspondence: Xin Yao, Department of Genitourinary Oncology, Tianjin Medical University Cancer Institute and Hospital, Huan-Hu-Xi Road, Ti-Yuan-Bei, He Xi District, Tianjin 300060, China.

e-mail: yaoxin1969@hotmail.com

Abstract

Objective

  • To examine the prognostic significance of the expression of platelet-derived growth factor-BB (PDGF-BB) and differentiated microvascular density (MVD) in patients with clear cell renal cell carcinoma (ccRCC).

Patients and Methods

  • We used the vascular marker cluster of differentiation 34 (CD34) to identify tumour blood vessels.
  • The expression of PDGF-BB and CD34 was detected by immunohistochemistry (IHC) in tissue microarrays (TMAs) from 100 ccRCCs.
  • Prognostic effects of individual parameters were calculated using Cox regression models and Harrell's concordance index (c-index).

Results

  • Higher grade and more advanced stage ccRCCs had significantly less PDGF-BB expression and differentiated MVD (P < 0.05).
  • Higher PDGF-BB expression was an independent prognostic factor for longer survival, and moreover, the final model built by the addition of PDGF-BB expression improved the predictive accuracy for disease-free survival (c-index 0.707) compared with the clinicopathological-based model (c-index 0.695).
  • PDGF-BB expression was positively associated with differentiated MVD assessed by Spearman correlation and factor analysis (r = 0.634, P < 0.001).

Conclusion

  • PDGF-BB is as a novel and promising prognostic marker and antiangiogenic therapeutic target for the treatment of ccRCC.

Introduction

Clear cell RCC (ccRCC), a hypervascular tumour, is the most common subtype of adult kidney cancer [1]. The biological and clinical behaviour of ccRCC is notoriously unpredictable based on its histological features. The metastatic potential of even low-grade, early stage ccRCC has been reported and there can be major differences in the clinical outcomes of patients with the same tumour stage. Although many new markers have been identified to have prognostic capacity for ccRCC, none of them has gained widespread acceptance as a clinical tool. The identification of additional independent prognostic markers is urgently needed to help in more accurately identifying aggressive ccRCC.

Angiogenesis, the generation of new blood vessels from pre-existing microvasculature, is an essential process for tumour growth and metastasis. The assessment of various aspects of tumour microvascularity, e.g. quantity, structure and function, might provide an indication of angiogenic activity, which may help in the discovery of an acceptable clinical tool to predict metastasis and for prognosis in ccRCC. Microvascular density (MVD) defined as the number of small vessels in a specific tumoral area, is a quantitative index used to describe tumour angiogenesis. The assessment of tumour MVD has been reported as a prognosticator in several human malignancies. Blood vessels of all sizes are normally supported by mural cells. In the microvascular bed, mural cells are termed ‘pericytes’. Their presence and communication with endothelial cells promotes many aspects of vessel stability and permeability control, including endothelial cell–cell adhesion and tight junctions [2]. Pericyte recruitment and attachment with endothelial cells requires local production of platelet-derived growth factor BB (PDGF-BB), close interaction with endothelial cells, particularly structured cell surface and matrix components to establish functionally stable vessels [3]. So the expression of PDGF-BB can be used to gauge pericyte coverage [3].

Consequently, we characterised angiogenesis by evaluating tumour vasculature quantitatively and qualitatively in 100 patients with ccRCC. We evaluated the relation between differentiated MVD marked by cluster of differentiation 34 (CD34) and the expression of PDGF-BB, and their correlations and prognostic significances.

Patients and Methods

Patients with ccRCC were identified from the files of Tianjin Cancer Institute and Hospital (Tianjin, China). Specimens were collected from 100 patients with ccRCC who underwent nephrectomy between 1998 and 2008. Patients with a history of other types of malignancy, preoperative distant metastases, lymph node involvement, or venous tumour thrombus, and those with inflammatory disorders were excluded. The pathological staging for each patient was corrected according to American Joint Cancer Committee/Union Internationale Contre le Cancer (AJCC/UICC) TNM 7th edition staging classification. ccRCC tissue microarrays (TMAs) were constructed as previously described [4] using Beecher instruments (Silver Springs, MD, USA). All tissue samples were histologically re-evaluated by two pathologists ‘blinded’ to the prognostic factors and/or clinical outcomes. Survival data were obtained by reviewing the hospital records.

Immunohistochemistry (IHC)

Briefly, after deparaffinization, slides were steam pre-treated in a citrate buffer at pH 6.0 for 30 min. The endogenous peroxidase activity and endogenous biotin were blocked with 3% H2O2 and protein-block buffer, respectively. The TMA sections were then incubated at room temperature for 30 min with one of the primary antibodies: mouse anti-human CD34 monoclonal antibody (mAb; 1:50, Dako, Carpinteria, CA, USA), or rabbit anti-human PDGF-BB mAb (1:30, Sigma, Saint Louis, MO, USA). Normal mouse IgG1 was used as a substitute for the primary antibody in the negative controls. Confirmed PDGF-BB positive and CD34 positive ccRCC tissue slices were used as positive controls. After washes using Tris-buffered saline (TBS) with 0.1% Tween 20, the sections were incubated at room temperature with biotinylated rabbit anti-mouse secondary antibody for 15 min followed by TBS washes. The sections were then incubated with streptavidin-biotin complex for 15 min. Staining was carried out using 3,3′-diaminobenzidine and H2O2. All sections were counter-stained with haematoxylin.

Staining Analysis

Differentiated MVD was determined by IHC staining of CD34. Briefly, three separately located tumour areas with the highest density of discrete microvessels (‘hotspots’) within each TMA core were selected for quantification of blood vessels at a magnification of × 200 with a Nikon (Melville, NY, USA) microscope. Therefore, a total of nine ‘hotspots’ were evaluated from the three TMA cores for each patient. The generally accepted criteria for determining a vessel profile [5] were used, MVD was derived by counting each vessel identified within the selected areas, including any stained endothelial cell or endothelial cell cluster that was separate from adjacent microvessels. Vessel lumens were not required for identifying a structure as a microvessel. Microvessels in necrotic or sclerotic areas within a tumour and intimately adjacent areas of unaffected kidney tissue were not considered in vessel evaluations. Vessels with thick smooth muscle walls or diameters of >8 × diameter of a single red blood cell were also excluded. The mean value of the vessel counts in the selected ‘hotspots’ was retained as the final MVD value.

For PDGF-BB assessment, the staining intensity and the proportion of stained tumour cells were analysed, as the consensus is that both variables should be quantitatively analysed to evaluate the expression level of growth factors in correlation with angiogenesis within the tumour nodule [6, 7], positivity was expressed as an index for each case. Staining was considered immunoreactive when brown granules were identified in the cytoplasm or nucleus of tumour cells. According to one of the established methods [6, 7], staining intensity was scored as zero (none), 1 (weak), 2 (moderate), or 3 (strong), the proportion of positively stained tumour cells in lesions was scored as zero (0%), 1 (1–25%), 2 (26–50%), 3 (51–75%), or 4 (76–100%), then we counted the sum of the two scores and tentatively defined 2, 3, 4, 5, or 6 as the threshold level for high expression vs low expression of PDGF-BB. A score of 4 had the most significant prognostic effect on progression-free survival, thus when the score was <4, the section was considered to have low expression of PDGF-BB, whereas ≥4 was considered as high expression.

Statistical Analyses

The median value of MVD in the entire cohort was used as the threshold for vessel counts. The expression level of PDGF-BB was defined as low and high according to the sum of the two scores. The disease-free survival (DFS) was defined as: from the date of radical nephrectomy to the date that tumour distant metastasis was confirmed. The overall survival (OS) was investigated and defined as: from the date of radical nephrectomy to the date of death or censored at the date of last follow-up. The association between CD34+ MVD or PDGF-BB expression and various clinicopathological parameters was assessed using the chi-squared test or Fisher's exact test, respectively. The Student's independent samples t-test and one-way anova were used to compare means in groups. The correlation between CD34+ MVD and PDGF-BB expression was calculated using Spearman's rank correlation coefficients. The survival curves were estimated by the Kaplan–Meier method and compared by the log-rank test. Cox proportional hazard regression models were fitted for multivariate analyses. The predictive accuracy was evaluated using Harrell's concordance index (c-index). A clinicopathological base model was built, consisting of all variables that had a significant independent prognostic value in the multivariable Cox proportional hazard regression model. The predictive accuracy of the clinicopathological base model was compared on the addition of PDGF-BB expression. For internal validation of the predictive accuracy, bootstrap resampling was performed. All of the statistical tests and P-values were two-tailed with P < 0.05 considered to indicate statistical significance.

Results

Patients Characteristics

In all, 100 patients were enrolled in this study, including 58 men and 42 women. The median (range) age was 55.9 (23–82) years. The mean (range) follow-up was 89.7 (5–178) months. In all, 28 patients died by the end of the clinical follow-up: 25 were disease-related, with a median (range) survival time of 27.6 (9–52) months, and three were attributable to causes other than tumour relapse.

Quantification of MVD and Expression of PDGF-BB

The median (range) MVD was 100.14 (26–194) microvessels/high-power field (HPF). The median value of MVD in the entire cohort was used as the threshold value. In the low-MVD group the mean (sd) MVD was 76.35 (17.81) microvessels/HPF, in the high-MVD group it was 128.06 (26.54) microvessels/HPF. IHC analyses of representative cases in the low- and high-MVD groups are shown in Fig. 1a and b.

Figure 1.

IHC analysis of TMAs of ccRCC representing different values of MVD and PDGF-BB expression: (A) Low MVD; (B) High MVD; (C) Low PDGF-BB expression; (D) High PDGF-BB expression.

Immunoreactivity for PDGF-BB was present diffusely or focally in the cytoplasm and nucleus of malignant cells, of some inflammatory cells, and of cells in the surrounding stroma (Fig. 1c and d), but not in normal epithelial cells. According to the high/low expression of PDGF-BB, as measured by a summation score of staining intensity and proportion of positive-staining cells, cases were classified into two types: Type 1, low expression of PDGF-BB and Type 2, high expression of PDGF-BB. IHC analyses of representative cases in Type 1and Type 2 are shown in Fig. 1c and d.

Correlations between MVD, Expression of PDGF-BB and Other Clinicopathological Factors

As shown in Table 1, there were no significant associations between MVD and age and sex (both P > 0.05). High MVD was associated with tumour diameter (P < 0.001), combined tumour pathological stage (P = 0.038) and combined Fuhrman grade (P = 0.005). The MVD was higher in the stage pT1–T2 group than in the stage ≥ pT3 group (P = 0.001). When we compared the MVD between different grade ccRCCs, there were no significant differences in the MVD between Grade 1 vs 2 (P = 0.272) and Grade 3 vs 4 (P = 0.119), but the MVD in Grade 3 was significantly less than in Grade 1 and Grade 2 (P < 0.001), the difference was also significant between Grade 1 + 2 and 3 + 4 (P < 0.001). Patients who had distant metastases had a low MVD pattern, which consisted of large vessels.

Table 1. MVD and PDGF-BB expression according to gender, age, tumour diameter, pathological stage, and nuclear grade in 100 patients with ccRCC
Clinicopathological variableNo. of casesPDGF-BB expression, nPMean (sd) MVD, microvessel/HPFP
LowHigh
  1. *P < 0.05; P < 0.001.
Gender:   0.278 0.592
Male582731 100.84 (30.26) 
Female421527 99.15 (39.10) 
Age, years:   0.462 0.879
≤50341618 103.96 (34.09) 
>50662640 98.16 (34.16) 
Tumour diameter, cm:   0.266 <0.001
≤7471730 112.91 (31.57) 
>7532528 88.80 (32.42) 
Stage:   0.007* 0.038*
pT1 + pT2541638 110.59 (31.79) 
pT3 + pT4462620 87.86 (32.86) 
Grade:   0.004* 0.005*
G1 + G2662145 109.21 (34.72) 
G3 + G4342113 82.52 (24.98) 

For PDGF-BB expression, briefly, there was significantly less PDGF-BB expression in ccRCCs with a more advanced stage and nuclear Fuhrman grade (both P < 0.05). There was a correlation between PDGF-BB expression and age, sex and tumour diameter of the 100 ccRCCs (Table 1).

Correlations between MVD and the Expression of PDGF-BB

The relationship between differentiated MVD and PDGF-BB expression was of interest. Regression analysis of the Spearman's correlation coefficient showed a strong relationship between MVD and the PDGF-BB expression. A correlation coefficient of 0.634 was calculated, with a P < 0.001, meaning that a high differentiated MVD in the tumour was usually accompanied by high PDGF-BB expression.

Prognostic Significance of differentiated MVD, PDGF-BB and other Clinicopathological Factors

The Kaplan–Meier survival analysis method, using the aforementioned threshold values, showed that low differentiated MVD and low PDGF-BB expression significantly predicted a decreased period of DFS and OS (Fig. 2).

Figure 2.

Kaplan–Meier curves for DFS and OS according to differentiated MVD and PDGF-BB expression.

Table 2 reports the HRs, as obtained by Cox proportional hazards regression analysis, at the multivariate level. Multivariate analysis included all variables significantly associated with patient survival in univariate analysis. Tumour pathological stage, Fuhrman grade, haemoglobin, lactate dehydrogenase and PDGF-BB expression emerged as independent prognostic factors for both DFS and OS at a multivariate level (P < 0.05). And for DFS, corrected calcium was also an independent prognostic factor. Differentiated MVD was found not to be an independent prognostic parameter.

Table 2. Observed DFS and OS with univariate (log-rank test) and multivariate (Cox proportional hazard model) analysis
VariableDFSOS
UnivariateMultivariateUnivariateMultivariate
PHR (95%CI)PPHR (95%CI)P
  1. *P < 0.05; P < 0.001; Ref, reference category; LLN, lower limit of laboratory's normal range; ULN, upper limit of laboratory's normal range; LDH, lactate dehydrogenase.
Stage:<0.001 0.004*<0.001 0.015*
pT1 + pT2 1 (Ref)  1 (Ref) 
pT3 + pT4 3.112 (1.448–6.734)  3.173 (1.254–8.026) 
Grade:<0.001 0.020*<0.001 0.017*
G1 + G2 1 (Ref)  1 (Ref) 
G3 + G4 2.332 (1.141–4.764)  2.863 (1.209–6.779) 
Haemoglobin:0.002* 0.010*0.007* 0.021*
≥LLN 1 (Ref)  1 (Ref) 
<LLN 0.401 (0.200–0.801)  0.395 (0.180–0.868) 
Calcium, mg/dL:<0.001 0.016*0.031* 0.727
≤11 1 (Ref)  1 (Ref) 
>11 2.345 (1.172–4.691)  1.148 (0.528–2.498) 
LDH:<0.001 0.003*<0.001 0.011*
≤1.5×ULN 1 (Ref)  1 (Ref) 
>1.5×ULN 2.959 (1.432–6.114)  2.974 (1.281–6.904) 
MVD:0.032* 0.2650.011* 0.098
High 1 (Ref)  1 (Ref) 
Low 0.656 (0.313–1.375)  0.462 (0.185–1.154) 
Expression of PDGF-BB:<0.001 <0.001<0.001 0.018*
High 1 (Ref)  1 (Ref) 
Low 0.249 (0.114–0.542)  0.365 (0.158–0.841) 

The final model built by the addition of PDGF-BB expression improved the predictive accuracy of DFS (c-index 0.707) compared with the clinicopathological-based model (c-index 0.695), which included tumour pathological stage, Fuhrman grade, haemoglobin, corrected calcium and lactate dehydrogenase. But for OS, addition of PDGF-BB expression did not improve the predictive accuracy (c-index 0.696) compared with the clinicopathological-based model (c-index 0.701), which included tumour pathological stage, Fuhrman grade, haemoglobin and lactate dehydrogenase.

Discussion

The value of MVD as a ccRCC prognosticator is controversial, according to previous reports. Joo et al. [8] reported that higher MVD was correlated with poor prognosis in ccRCC, while other researchers found the converse, where a high MVD was predictive of a good prognosis [9-11]. We think these contrary results may be due to the use of different vessels markers to evaluate MVD and the method used to count the numbers of capillaries. Vascular markers used to identify tumour blood vessels are heterogeneous; these vessels markers can reveal different characteristics of tumour blood vessels. For example, CD34 is expressed in differentiated types of endothelial cells, while CD31 is expressed in both differentiated types and undifferentiated types of endothelial cells [12]. Using these two kinds of vascular markers, we found two distinct types of microvessels: undifferentiated microvasculature (CD31+/CD34–) and differentiated microvasculature (CD34+) in ccRCC, and a higher differentiated MVD significantly correlated with lower tumour grade and longer survival [13].

The reason why poorly vascularised ccRCC tumours exhibit a poor prognosis is not clear. Delahunt et al. [9] reported that decreased MVD in high-grade tumours may also be dependent on the development of the large-diameter vascular channels that are frequently seen in large tumours. Herbst et al. [10] reported that renal cells imitate tubule differentiation, and therefore, high MVD of carcinoma tissue may be considered to reflect the normal structure, function and tissue organisation of the renal tubule system. MVD marked by CD34 represents a differentiation parameter and the diminution of MVD in relation to a high malignancy grade becomes understandable. Sabo et al. [11] reported that a high-grade solid RCC enlarges more rapidly, overcoming its vascular network and decreasing its architectural complexity. Poorly vascularised tumours then become hypoxic and necrotic. Kohler et al. [14] suggested that the decrease in MVD in high-grade RCC reflects the inability of tumour neovascularisation to keep pace with the proliferation of the tumour cells in Grade 3 tumours, and then the metabolic supply becomes insufficient, resulting in extensive necrosis, a morphological finding that is very characteristic of large RCCs. We suppose that cells in a high stage and grade RCC proliferate more rapidly, overcoming its vascular network and secondarily producing necrosis and larger vascular channels, thus decreasing its MVD, but increasing the permeability of the blood vessels. In the present study, we further compared the MVD between the four pathological stages, we found that the average MVD of pT1 was lower than the average MVD of pT2, although the P value is not so significant, we think that maybe it is because the sample size was not large enough. The MVD of pT4 stage was significantly lower than that the of pT1, pT2 and pT3 stages. The MVD of pT3 was lower than that of pT2, but there is no significant difference between the MVD of pT3 and pT1, for this unanticipated result, firstly we think that the onset of neovascularization enhances tumour growth by perfusion which allows nutrients and oxygen to enter and metabolites to exit from the tumour, secondly we think that pT3 is a transition period, in which, MVD is decreasing because of the necrosis of tumour tissues and the alteration of vascular structure.

Haematogenous metastasis is dependent not only on vessel density but also on the quality of vascular structure and function. Primitive tumour vessels are vulnerable to tumour cell invasion and provide a structural basis for metastasis. Theoretically, PDGF-BB would promote maturation of tumour vessels by recruiting vascular smooth muscle cells (VSMCs)/pericytes to the nascent vasculature and thus prevent metastasis. McCarty et al. [15] found that overexpression of PDGF-BB decreases colorectal and pancreatic cancer growth by increasing tumour pericyte content. The structural and functional abnormalities of vessels in advanced cancers are associated with poor VSMC/pericyte coating [16]. Unlike VSMCs, the pericytes are intimately adherent to the abluminal endothelial surface and fully embedded in a shared basement membrane. Their presence and communications with the endothelial cells promote many aspects of vessel stability and permeability control, including endothelial cell–cell adhesion and tight junctions [17]. Thus, the recruitment and attachment of pericytes is a crucial step in the formation and stabilisation of mature blood vasculatures [18, 19]. Pericyte–endothelial cell interactions are controlled by a host of different molecules, e.g. PDGF-BB, vascular endothelial growth factor (VEGF), TGD-β and angiopoietins. The local production and retention of the PDGF-BB ligand is essential for proper pericyte recruitment through investment both in normal developmental angiogenesis and in the tumour environment [20, 21]. PDGF-BB is secreted by the endothelium during the angiogenic process and activates its cognate tyrosine kinase receptor PDGFR-β on the pericytes. This interaction is required both for the longitudinal recruitment of pericytes along the advancing endothelial sprouts, as well as for pericyte proliferation and proper attachment [3]. Thus, the expression level of PDGF-BB can gauge the proliferation of pericytes and microvessel pericyte coverage index; and further evaluate the maturity of the microvasculature.

Treatment with a combination of VEGF and PDGF-specific antibodies produced only modest antitumour activity [22]. Similarly, deletion of host-derived PDGF-BB using a genetic approach did not increase the antitumour activity of a VEGF-specific antibody [22, 23]. Paradoxically, PDGF and probably other vascular remodelling factors could play an important role in normalisation of tumour vasculature, which might increase chemotherapeutic drug delivery [24, 25], and also prevent metastasis. Thus, the concept of combined therapy targeting both VEGF and PDGF warrants further preclinical validation [26]. So, based on the present findings and those of previous studies [18, 27-29], we think that tumour progression might be postponed by pruning of immature and inefficient vessels and promoting the maturation of the remaining vessels.

These data must be interpreted with caution, considering the limitations of a retrospective study on biomarkers and survival, the limited sample size and the selection bias. Nevertheless, we think that the present study can provide valuable information for ccRCC prognostication and treatment.

In conclusion, differentiated MVD reflects the number of mature microvessels and values of PDGF-BB expression can reflect the structural and functional characteristics of the tumour microvasculature. PDGF-BB protein expression was positively associated with differentiated MVD. We were able for the first time to unequivocally correlate PDGF-BB expression with angiogenesis and other clinicopathological parameters, and showed that addition of PDGF-BB expression improved prognostic accuracy. Thus, PDGF-BB appears to be a novel and promising prognostic marker, and antiangiogenic therapeutic target for the treatment of ccRCC

Acknowledgements

This work was supported by the National Natural Science Foundation of China (No. 81072090) and the Tianjin Municipal Science and Technology commission (No. 07JCYBJC17000).

Conflict of Interest

None declared.

Abbreviations
c-index

Harrell's concordance index

DFS

disease-free survival

HPF

high-power field

IHC

immunohistochemistry/immunohistochemical

mAb

monoclonal antibody

MVD

microvascular density

OS

overall survival

PDGF-BB

platelet-derived growth factor-BB

ccRCC

clear cell RCC

TMA

tissue microarray

TBS

Tris-buffered saline

VEGF

vascular endothelial growth factor

VSMC

vascular smooth muscle cells

Ancillary