Expression Levels of Genes that Regulate Metastasis and Angiogenesis Correlate with Advanced Pathological Stage of Renal Cell Carcinoma

Abstract

We examined the expression levels of a number of metastasis-related genes to determine the relationship of these levels to the development of metastasis in renal cell carcinoma. Gene expression was examined in 46 formalin-fixed, paraffin-embedded, archival specimens of primary organ-confined, clear-cell, renal cell carcinoma from patients who had undergone radical nephrectomy. Twenty samples were from patients who did not have metastasis after a median of 48 months; 26 were from patients with either synchronous or metachronous metastases. Microvessel density was assessed by anti-CD-34 immunohistochemical analysis. The expression levels of basic fibroblast growth factor (bFGF), vascular endothelial growth factor (VEGF), interleukin-8 (IL-8), matrix metalloproteinases (MMP)-2 and -9, and E-cadherin were examined at the periphery of the tumor by a colorimetric in situ mRNA. The expression levels of bFGF, VEGF, IL-8, MMP-2, and MMP-9 were significantly higher in primary renal tumors from patients with either synchronous or metachronous metastases than those who were disease-free at a median of 48 months of follow-up. Multivariate analysis of disease-free survival showed that the ratio of MMP-9 to E-cadherin (P = 0.012) and the expression level of bFGF expression (P = 0.045), were independent predictors for the development of metastases. The expression levels of bFGF, VEGF, and IL-8 did not correlate with microvessel density, which in itself was not a significant predictor of progression (P = 0.21). In summary, expression levels of genes that regulate metastasis angiogenesis can predict the metastatic potential in individual patients with organ-confined clear-cell renal carcinoma.

Renal cell carcinoma (RCC) is the eighth most common solid malignancy and the eleventh leading cause of cancer mortality in the United States, with more than 25,000 new cases and 12,000 deaths predicted in the year 2000. 1 Typically, RCC is a highly vascular neoplasm with an unpredictable pattern of recurrence. Several pathological features of the primary tumor have been evaluated as prognostic factors. Advanced pathological stage has been the most powerful marker associated with, or predicting for, the development of metastasis. 2,3 Other pathological factors associated with metastases include high tumor grade, 4,5 abnormal DNA ploidy, 6,7 nuclear morphometry, 8 and tumor proliferation (Ki67, proliferating cell nuclear antigen). 9-12 Although these markers have improved our ability to predict the prognosis of RCC, they have nevertheless provided only general guidelines with which we can counsel patients with this disease. Clearly, there is a need for more precise prognostic markers for predicting the risk of developing metastatic RCC.

Because RCC is one of the most highly vascularized solid malignancies, it seems logical that factors that regulate the process of angiogenesis and invasion would be integral to its pathogenesis. The expression level of several genes, including the pro-angiogenic factors basic fibroblast growth factor (bFGF) 13-16 and vascular endothelial growth factor (VEGF), 16-19 the extracellular matrix-degrading matrix metalloproteinases (MMP-2 and MMP-9) 20,21 and the cell-to-cell adhesion molecule E-cadherin 22 have been studied individually and shown to correlate with the metastatic potential of RCC. Most of these correlative studies, however, have reached the inevitable conclusion that the expression of a given gene is necessary but insufficient to account for the multistep process of metastasis. 23

We have developed a colorimetric in situ hybridization technique 24,25 to study the expression levels of multiple genes that regulate distinct steps of the metastatic cascade in several neoplasms, including colon, 26,27 gastric, 28 pancreatic, 29 prostate, 30 and lung cancers. 31 We extended this research to clear cell carcinoma of the kidney, analyzing the relationship of angiogenic factor expression with microvessel density (MVD), a reflection of tumor angiogenesis, as well as the expression of MMP-2 and MMP-9, as it relates to E-cadherin, ie, to the degree of cell detachment, invasion, and metastasis. We hypothesized that the expression levels of one or more factors associated with tumor invasion or angiogenesis would identify aggressive RCC associated with a high risk of metastasis. We found that the expression of angiogenic factors did not correlate with MVD, whereas the expression level of bFGF and the relative expression level of the two MMPs as related to the expression level of E-cadherin did predict the invasive and metastatic potential of individual organ-confined clear-cell renal neoplasms. Moreover, these factors may become relevant targets for novel therapeutic strategies.

Materials and Methods

Surgical Specimens

Forty-six formalin-fixed, paraffin-embedded, archival surgical specimens of primary organ-confined clear-cell renal carcinoma from patients treated at The University of Texas M. D. Anderson Cancer Center were selected for study. All patients had undergone radical nephrectomy between the years 1990 and 1995. Twenty patients were disease-free at a median follow-up of 48 months (range, 24 to 60 months); the other 26 patients had either synchronous metastases (n = 11) at the time of nephrectomy or metachronous metastases (n = 15) at a median of 8 months after nephrectomy. The histology of the primary tumor was reviewed by a pathologist (AEN) who confirmed that all of the specimens were clear cell RCC. All specimens had intact mRNA as determined by a positive reaction with a poly(dT)₂₀ probe 26 and were evaluated for expression of the metastasis-related genes.

Tumor Size, Fuhrman’s Grade, and DNA Ploidy

The size of the neoplasm in the nephrectomy specimen was determined immediately after resection. Fuhrman’s grade was assessed in all specimens by one pathologist using established criteria (Table 1) ▶ . 4 Immediately after nephrectomy, a single needle specimen of kidney was obtained from viable portions of the pathological specimens. Flow cytometric analysis was performed using acridine orange staining. 32 Any specimen with a diploid index (number of aneuploid cells divided by the number of true diploid cells) greater than or less than 1.0 was considered aneuploid (Table 1) ▶ .

Table 1.

Clinicopathological Data

Case no.	Tumor size* (cm)	Furhman’s nuclear grade	DNA ploidy†	Metastasis‡
1	7.5	2	Aneuploid	N
2	6.0	3	Diploid	N
3	5.0	2	Diploid	N
4	3.5	2	Aneuploid	N
5	9.0	2	Diploid	N
6	10.0	4	Aneuploid	Y
7	8.0	3	Diploid	Y
8	4.0	4	Aneuploid	Y
9	9.5	3	Diploid	Y
10	10.0	3	Aneuploid	Y
11	6.0	2	Diploid	N
12	8.5	4	Aneuploid	Y
13	4.5	2	Aneuploid	Y
14	8.0	3	Aneuploid	N
15	8.5	4	Aneuploid	Y
16	2.0	2	Diploid	N
17	6.5	3	Aneuploid	Y
18	5.0	3	Diploid	Y
19	9.0	4	Aneuploid	Y
20	1.8	2	Diploid	N
21	4.0	3	Diploid	N
22	4.4	2	Diploid	N
23	5.0	3	Aneuploid	N
24	5.8	3	Diploid	Y
25	12.0	2	Aneuploid	Y
26	11.5	3	Diploid	N
27	8.0	3	Aneuploid	Y
28	9.0	2	Aneuploid	Y
29	10.0	3	Aneuploid	Y
30	6.3	2	Diploid	N
31	5.9	2	Aneuploid	N
32	7.5	2	Diploid	Y
33	8.5	3	Aneuploid	Y
34	5.0	2	Diploid	N
35	12.2	3	Aneuploid	Y
36	4.0	3	Diploid	N
37	5.5	2	Diploid	N
38	10.5	3	Diploid	Y
39	5.5	4	Aneuploid	N
40	11.2	3	Aneuploid	Y
41	5.5	3	Diploid	Y
42	5.3	2	Aneuploid	N
43	5.7	2	Diploid	Y
44	9.4	3	Aneuploid	Y
45	6.0	4	Aneuploid	Y
46	5.2	4	Aneuploid	Y

*Tumor size (largest transverse diameter in centimeters) was assessed pathologically after resection.

^†DNA ploidy was determined by acridine orange flow cytometry. Any specimen with a diploid index greater than or less than 1.0 was considered aneuploid.

^‡Metastasis = development of metastases. Y, yes; N, no.

CD-34 Immunohistochemical Analysis

CD-34 immunohistochemical analysis was performed as previously described. 33 Briefly, paraffin-embedded sections of kidney tumors (4 to 5 μm) were treated sequentially with xylene and ethanol to remove the paraffin. All sections were treated with pepsin (Biomeda Corp., Foster City, CA) for 15 minutes at 37°C. The sections were washed three times with phosphate-buffered saline (PBS), and endogenous peroxidase was blocked by the use of 3% hydrogen peroxide in methanol for 12 minutes. The samples were then washed three times with PBS and incubated with a protein-blocking solution (PBS containing 1% normal goat serum and 1% horse serum) for 20 minutes at room temperature. Sections were processed for indirect immunoperoxidase assay, in which the primary antibody is primary polyclonal rabbit anti-human CD-34, and then developed using the Multilink system (Biogenex, San Ramon, CA). All specimens were stained within a single session. The sections were then washed three times with distilled water, counterstained with aqueous hematoxylin, washed, mounted with Permount, and examined using a bright-field microscope (Figure 1) ▶ .

Figure 1.

In situ hybridization analysis of angiogenic factors in nonmetastatic and metastatic RCC (original magnification, ×200). Hybridization with a hyperbiotinylated poly(dT)₂₀ probe confirmed the integrity and lack of mRNA degradation. A positive reaction in this assay stains red. The expression intensity of each factor was determined in both cells at the periphery of the tumor and divided by the expression in normal proximal tubular epithelium. This value was normalized to the expression of poly(dT)₂₀ in neoplastic and normal tissue. Note higher expression of bFGF and VEGF relative to poly(dT) in metastatic primary tumor. In addition, an example of anti-CD-34 immunohistochemistry in organ-confined renal primary tumors from both nonmetastatic and metastatic groups. Note similar vascular density in both tumors.

MVD was determined by light microscopy using the procedure of Weidner and colleagues. 34 Areas with the most intensely stained blood from adjacent microvessels, tumor cells, or other stromal cells were each considered a single countable microvessel. The images were projected and recorded by digitizing the image using a cooled imaging camera (Optronics Tec 470; Optronics Engineering, Goletha, CA) linked to a computer and a digital printer (Sony Corp., Tokyo, Japan). Results were expressed as the number of microvessels identified within selected fields (×40).

Oligonucleotide Probes

Specific antisense oligonucleotide DNA probes were designed complementary to the transcripts of four metastasis-related genes, bFGF, 35,36 VEGF, 37 interleukin (IL)-8, 38 MMP-2, 39 MMP-9, 40 and E-cadherin 41 (Table 2) ▶ , based on the reports of cDNA sequences. The specificity of the oligonucleotide sequences was initially determined by a GenBank/European Molecular Biology Laboratory database search (using the Genetics Computer Group, Madison, WI) based on the FastA algorithm, 42 which showed 100% homology with the target gene sequences. The specificity of each sequence was also confirmed by Northern blot analysis. A poly(dT)₂₀ oligonucleotide was used to verify the integrity of the mRNA in each sample. All DNA probes were synthesized with six biotin molecules (hyperbiotinylated) at the 3′ end via direct coupling using phosphormidine chemistry 43,44 (Research Genetics, Huntsville, AL). The lyophilized probes were reconstituted to a 1 μg/ml stock solution in 10 mmol/L Tris-HCL (Research Genetics) immediately before use. The working dilutions of each probe are shown in Table 2 ▶ .

Table 2.

In Situ Hybridization Probes: Sequence, Guanosine Cytosine (GC) Content, and Working Dilutions

Probe	Sequence (5′-3′)	GC content	Working dilution
bFGF	CGGGAAGGCGCCGCTGCCGCC	87.5%	1:200
VEGF	TGGTGATGATGTTGGACTCCTCAGTGGGC	57.7%	1:100
IL-8	CTCCACAACCCTCTGCACCC	65.0%	1:100
MMP-2	GGCCACATCTGGGTTGCGGC	87.5%	1:200
MMP-9	CCGGTCCACCTCGCTGGCGCTCCGG	57.7%	1:100
E-cadherin	TGGAGCGGGCTGGAGTCTGAACTG	65.0%	1:100
Poly(dT)20	T T T T T T T T T T T T T T T T T T T T		1:800

In Situ Hybridization

In situ hybridization was performed as described previously 25,26 with a minor modification using the Microprobe manual staining system (Fisher Scientific, Pittsburgh, PA). 45 Tissue sections (4 μm) of formalin-fixed, paraffin-embedded specimens were mounted on Probe-On slides (Fisher Scientific). The slides were placed in the Microprobe slide holder and dewaxed and dehydrated with Auto Dewaxer and Auto Alcohol (Research Genetics), followed by enzymatic digestion with pepsin. The probe was hybridized for 100 minutes at 45°C, and samples were then washed three times with 2× standard saline citrate for 2 minutes at 45°C. The samples were incubated for 30 minutes in alkaline phosphatase-labeled avidin at 45°C, briefly rinsed in 50 nmol/L Tris buffer (pH 7.6), rinsed for 1 minute with alkaline phosphatase enhancer (Biomeda Corp.) and incubated for 30 minutes with the chromogen substrate Fast RED (Research Genetics) at 45°C. A red stain indicated a positive reaction in this assay. Control for endogenous alkaline phosphatase included treatment of the samples in the absence of the biotinylated probe and the use of chromogen in the absence of any oligonucleotide probes. To check the specificity of the hybridization signal, the following controls were used: RNase pretreatment of tissue sections, a biotin-labeled sense probe, a competition assay with unlabeled sense probe, and a competition assay with unlabeled antisense probe. A markedly lower signal or no signal was obtained after all of these treatments.

Image Analysis to Quantify Intensity of the Color Reaction

Stained sections were examined using a Zeiss photomicroscope (Carl Zeiss, Inc., Thornwood, NY) equipped with a three-chip charged-coupled color camera (model DXC-960 MD, Sony Corp.). The images were analyzed using Optimas image analysis software (version 6.2; Media Cybernetics, Silver Spring, MD). The slides were prescreened by one of the investigators to determine the range in staining intensity of the slides to be analyzed. Images covering this range of staining intensities were captured electronically, a color bar (montage) was created, and a threshold value was set in the red, green, and blue modes of the color camera. All subsequent images were quantified based on this threshold. The integrated absorbance of the selected fields was determined based on its equivalence to the mean log inverse gray scale values multiplied by the area of the field. The samples were not counterstained; therefore, the absorbance was due solely to the product of the in situ hybridization reaction.

For each section, we determined the absorbance in several 2 × 2 mm zones located at the periphery of the tumor. Within each zone, five different fields were quantified and an average value derived. The intensity of staining was standardized to that of the integrated absorbance of poly(dT)₂₀ and determined by comparison with the integrated absorbance of normal proximal tubule epithelium according to the following equation: intensity = 100 × (a/b)/(c/d), where a = expression of each factor in the tumor specimen, b = expression of poly(dT)₂₀ in the tumor specimen, c = expression of factor in normal proximal tubule epithelium, and d = expression of poly(dT)₂₀ in normal proximal tubule epithelium. Examples of in situ hybridization and relative gene expression are shown in Figure 1 ▶ .

Statistical Analysis

The specimens were stratified according to tumor size (≤7 cm and >7 cm), Fuhrman’s nuclear grade [low (grades 1 and 2) and high (grades 3 and 4)], and ploidy (diploid and aneuploid) (Table 1) ▶ . The angiogenic factors were compared with known prognostic factors for recurrence of organ-confined clear-cell renal cancer: tumor size, nuclear grade, and DNA ploidy. The level of gene expression among the groups was compared by the Mann-Whitney U test. 46 Univariate and multivariate analyses of the development of metastasis were conducted using the logistic regression model. 47 Statistical significance was defined as a two-sided P < 0.05.

Results

Increased expression levels of bFGF, MMP-2, MMP-9, and IL-8 correlated with tumor size. In addition, bFGF and VEGF correlated with high nuclear grade (Table 3) ▶ . MVD was not correlated with bFGF, VEGF, and IL-8 (P = 0.345, P = 0.591, and P = 0.887, respectively; Figure 2 ▶ ).

Table 3.

Correlation of Metastasis-Related Expression and MVD with Tumor Size, Nuclear Grade, and DNA Ploidy

	Size		Grade		DNA Ploidy
		<7 cm	≥7 cm	Low (1 and 2)	High (3 and 4)	Diploidy	Aneuploid
MVD*	321 ± 115	279 ± 135	331 ± 117	275 ± 131	310 ± 103	208 ± 133
Gene†
bFGF	138 ± 44	212 ± 64‡	127 ± 51	213 ± 46‡	154 ± 38	175 ± 57
VEGF	113 ± 48	133 ± 36	98 ± 26	135 ± 38‡	107 ± 44	137 ± 39
IL-8	102 ± 39	139 ± 29‡	111 ± 42	127 ± 34	146 ± 62	197 ± 77
MMP-2	119 ± 34	173 ± 56‡	151 ± 56	132 ± 81	132 ± 86	152 ± 40
MMP-9	118 ± 31	144 ± 29‡	129 ± 47	120 ± 44	108 ± 52	139 ± 29
MMP§	118 ± 44	153 ± 63‡	141 ± 45	126 ± 59	146 ± 65	120 ± 26
E-cadherin	88 ± 27	85 ± 24	92 ± 22	86 ± 28	87 ± 27	86 ± 23
MMP/E-Cadherin	1.4 ± 0.5	1.8 ± 0.5‡	1.5 ± 0.4	1.5 ± 0.5	1.7 ± 0.5	1.5 ± 0.5

*Microvessel density (×100) determined by method of Weidner et al. 34

^†The intensity of cytoplasmic staining quantified by an image analyzer was compared with poly(dT)₂₀ and then standardized by the expression of nonneoplastic proximal tubule epithelium (set at 100). This normalization was repeated for each stained specimen.

^‡P < 0.05, Student’s t-test.

^§MMP = (MMP-2 + MMP-9)/2.

Figure 2.

Correlation of MVD and the expression of angiogenic factors. The absolute value of MVD in each tumor specimen did not correlate with the relative intensity of bFGF, VEGF, or IL-8.

MVD and the expression levels of the metastasis-related genes were compared among the metastatic and nonmetastatic groups (Table 4) ▶ . Nonmetastatic primary tumors had a median MVD of 77 (range, 27 to 122 vessels per ×100 field) whereas metastatic primary tumors had a median MVD of 85 (range, 41 to 138 vessels per ×100 field; P = 0.21). Thus, MVD by itself did not correlate with metastasis. On the other hand, the expression levels of bFGF, MMP [(MMP-9 + MMP-2)/2, VEGF, and IL-8, taken as a continuous variable, did correlate with the development of metastases (P < 0.0001, P = 0.0002, P = 0.0009, and P = 0.006, respectively).

Table 4.

Relative mRNA Expression of Angiogenic Factors

Factor	Gene expression*		Univariate P value†	Multivariate P value‡
	Nonmetastatic tumor	Metastatic tumor
bFGF	105 (55–228)	227 (107–386)	0.0006	0.0137
VEGF	96 (49–123)	152 (68–259)	0.0072	0.1943
IL-8	98 (45–152)	139 (87–256)	0.0325	0.4188
MMP‡	102 (50–138)	142 (91–275)	0.0012	0.0901
E-Cadherin	98 (5–155)	88 (10–138)	0.5233	0.7827
MMP/E-Cadherin	1.3 (0.6–2.3)	2.0 (0.8–2.6)	<0.0001	0.0073

*The intensity of cytoplasmic staining as quantified by an image analyzer was compared with poly(dT)₂₀ and then standardized by the expression of nonneoplastic proximal tubule epithelium (set at 100). This normalization was repeated for each stained specimen.

^†Logistic regression.

^‡MMP = (MMP-2 + MMP-9)/2.

We then determined the optimal cutoff value for each of the factors by assessing the chi-square significance at multiple cutoff intervals (data not shown). Figure 3 ▶ shows the scattergrams for the expression levels. The optimal cutoff value for bFGF, VEGF, IL-8, MMP [(MMP-9 + MMP-2)/2], and E-cadherin were 150%, 130%, 120%, 140%, and 98%, respectively. By multivariate analysis (logistic regression model) using these cutoff values, the expression levels of bFGF followed by MMP were the variables most strongly associated with metastasis. However, when we included the ratio of MMP to E-cadherin into the statistical model, it was clearly the factor most significantly associated with metastasis (P = 0.0073, Table 4 ▶ ).

Figure 3.

Scattergram of angiogenic factor expression in primary tumors from nonmetastatic and metastatic groups. The optimal cutoff values for relative gene expression as a predictor for recurrence were 150%, 130%, and 120% for bFGF, VEGF, and IL-8, respectively.

The prognostic significance of the clinicopathological variables (Table 1) ▶ and the expression of the metastasis-related genes factors were analyzed by univariate and multivariate analyses. A MMP to E-cadherin ratio of ≥1.7 was the most significant independent prognostic variable for advanced pathological stage (P = 0.0122), followed by bFGF (P = 0.0453) and grade (P = 0.0931) (Table 5) ▶ .

Table 5.

Univariate and Multivariate Analysis of Tumor Size, Nuclear Grade, DNA Ploidy, and bFGF Using Logistic Regression

Parameter	Univariate analysis*	Multivariate analysis*
Size (≥7 cm)	0.0614	0.3355
Nuclear grade (3 and 4)	0.0059	0.0931
DNA ploidy (aneuploid)	0.0100	0.2230
bFGF (≥150%)	0.0006	0.0453
MMP/E-cadherin (≥1.7)	<0.0001	0.0122

*Logistic regression model.

Discussion

We used the in situ hybridization technique described herein to examine the concurrent expression of metastasis-related genes in formalin-fixed, paraffin-embedded specimens of resected primary RCC. Because metastases are produced by only by a small subpopulation of tumor cells (<1.0% of the tumor), detecting the expression levels of metastasis-related genes in this minority of tumor cells requires sensitive techniques. 48,49 We chose to use in situ hybridization because it can identify the cellular source and intratumoral heterogeneity of mRNA expression, whereas Northern blot analysis indicates only the average levels of mRNAs of all of the cells in a sample.

An increased MVD has been associated with early progression in a number of neoplasms, including breast, 34,50 colon, 51,52 and prostate cancers. 53 However, although one report has suggested a good association between MVD and outcome in RCC, 54 most studies have found MVD to be a poor predictor for metastasis. 55,56 MacLennan and Bostwick 56 examined 97 specimens using anti-factor VIII antigen. They found no correlation of MVD with clinical stage, pathological stage, tumor grade, or cancer-specific survival. Using CD-34 which detects 20 to 30% more vessels than does Factor VIII, 57 we concur with the majority of reports in finding no relationship between MVD and metastasis.

Previous reports have suggested that MVD correlates strongly with the expression of VEGF 58 but less dramatically with the expression of bFGF. 59 In clear cell RCC, which is among the most vascular of all solid tumors, there seems to be a discordance between the expression of the pro-angiogenic factors bFGF, VEGF, and IL-8 and vascular density. Although discordance may be because of the presence of an unidentified angiogenic factor that is more relevant to the induction of angiogenesis in this malignancy, it is more likely because of the inherent hypervascularity of RCC and our inability to identify relevant endothelial markers for pathological neovascularization.

A number of reports have shown that the tissue or serum expression of the angiogenic factors bFGF 13-16 and VEGF 16-19 predict for survival in patients with RCC. IL-8, a chemokine that was originally identified as a chemoattractant for leukocytes 60 has recently been associated with angiogenesis. 61 IL-8 also correlates with disease progression in patients with gastric cancer 62 but has not been extensively investigated in RCC. Although all three angiogenic factors were more highly expressed in metastatic tumors than in nonmetastatic tumors, bFGF was most strongly correlated with metastasis on multivariate analysis. Therefore, despite the poor correlation between MVD and metastasis, the pro-angiogenic factor bFGF was an independent predictor for the presence of or the development of metastasis.

We previously reported that the metastatic potential of human colon, gastric, pancreatic, and prostate cancers can be predicted by the gene expression of factors that regulate cohesion (E-cadherin) and invasion and motility (MMP-2 and MMP-9). 26-31 Specifically, the up-regulation of the proteolytic enzymes MMP-2 and MMP-9 concurrent with the down-regulation of the homotypic cohesion molecule E-cadherin accurately indicated higher rates of for metastasis and disease recurrence.

E-Cadherin is a cell surface glycoprotein involved in cell-to-cell adhesion. It is localized at the epithelial junction complex and is responsible for the organization, maintenance, and morphogenesis of epithelial tissue. 63,64 Reduced levels of E-cadherin are associated with a decrease in cellular and tissue differentiation and increased histological grade of different epithelial neoplasms. 64,65 Loss of E-cadherin is also associated with an increased risk for the development of metastasis in human tumors. 66-68 In particular, Katagiri and associates 22 demonstrated the association of lower expression of E-cadherin with progression in human renal tumors.

After cells detach from the primary tumor, they must invade the host stroma if they are to metastasize. 69 Degradation of blood vessel basement components, especially type IV collagen, is one of the necessary steps in this process. 70-72 The levels of MMP-2 and MMP-9 (type IV collagenase) in human and rodent neoplasms directly correlate with invasion and metastasis, 73-76 and specific inhibitors of MMPs have been shown to inhibit tumor cell invasion. 77-79 Both Kugler and colleagues 20 and Lein and colleagues 21 have reported that the balance between MMP-2 and MMP-9 expression and between TIMP-1 and TIMP-2 expression is an essential factor in the aggressiveness of RCC. Thus a decrease in the expression of E-cadherin and an increase in collagenase type IV activity should enhance tumor cell invasion and metastasis. Our results suggest that this hypothesis is correct. A MMP:E-cadherin ratio of ≥1.7 at the periphery of the renal tumor directly correlated with progression and was the strongest predictive factor on multivariate analysis. Although the precise ratio is different for each tumor system, this value is similar to those for a number of different neoplasms. 26-31

Other prognostic factors have been shown to be useful predictors of progression-free survival in organ-confined RCC. A tumor diameter of 7 cm recently became the cutoff value between T1 and T2 tumors, and the value of this variable as a prognostic factor has been verified by others. 3 Both high Fuhrman’s nuclear grade, 4,5 and abnormal DNA ploidy 6,7 have previously been shown to be a significant predictors for progression of organ-confined tumors.

Multivariate analysis of the results with our tumor samples demonstrated that high expression of bFGF and the ratio of MMP to E-cadherin were stronger predictors for metastasis than tumor size, grade, or ploidy. Because both bFGF and MMP expression are associated with the development of metastatic RCC, these factors should be considered as relevant targets for novel therapeutic strategies.