Vascular Endothelial Growth Factor Levels in Ovarian Cyst Fluid Correlate with Malignancy

INTRODUCTION

Ovarian cancer is one of the most common gynecological neoplasms. Each year, 26,000 women develop ovarian cancer, but only 25% of them are diagnosed when the tumor is confined to the ovary (1) . As a result, fifty percent of these patients die within five years. There is, therefore, a need for new approaches in the early diagnosis, prognostic evaluation, and treatment of this insidious and lethal disease.

During the early stages of tumor progression, cancer cells acquire the capacity to stimulate angiogenesis and secrete growth factors that promote endothelial migration, proliferation, capillary tube formation, and extracellular matrix degradation (2) . The newly formed vessels carry oxygen and nutrients to the growing tumor, dispose of its metabolic waste products and secrete factors that further promote the proliferation and invasive behavior of tumor cells through paracrine mechanisms (3 , 4) . They also function as a port of entry for the spread of metastatic cells to distant sites. The hypothesis that tumor growth is angiogenesis dependent is particularly relevant for ovarian cancer, which becomes very large in spite of the small size of the organ in which it develops. Histological studies have confirmed that ovarian cancer is richly vascularized and have shown a correlation between microvascular density and tumor aggressiveness (5) .

Because of the importance of angiogenesis in tumor growth, the identification of growth factors responsible for ovarian cancer-related angiogenesis may have important implications for the diagnostic and prognostic evaluation of this disease. These factors and their signaling mechanisms may also be potential targets of adjuvant therapy (6) . Among the known endothelial growth factors, VEGF3 has emerged as a critical regulator of the angiogenic process ,(7) . VEGF is an M_r 32,000–45,000 homodimer produced by a variety of cell types including cancer cells. VEGF has endothelial target specificity and is secreted as four alternatively spliced forms, VEGF₁₂₁, VEGF₁₆₅, VEGF₁₈₉, and VEGF₂₁₀. VEGF promotes endothelial migration, proliferation, protease activity, and capillary tube formation (7) . VEGF is also known as vascular permeability factor because of its potent effect on the permeability of postcapillary venules (8) . The deletion of even a single VEGF allele in mice results in defective vasculogenesis and intrauterine death of the embryos (9 , 10) . Similarly, the deletion of genes for the VEGF receptors flk-1 and flt-1 causes severe vascular defects that are incompatible with the survival of the mouse embryo (11 , 12) . VEGF contributes also to angiogenic responses in the adult and has been shown to mediate tumor angiogenesis in experimental animals (13) . The safety and efficacy of a humanized anti-VEGF antibody is presently being tested in cancer patients (14) .

Because VEGF has been implicated as a regulator of angiogenesis in ovarian cancer (15) , production of VEGF may reflect the angiogenic activity of this neoplasm. Ovarian cancers generate fluid-filled cysts that contain secretory products of cancer cells. On this basis, we hypothesized that cyst fluid could be used to quantitatively evaluate VEGF production in ovarian lesions. In this study, we measured VEGF in ovarian cyst fluid using a highly sensitive ELISA. We also measured bFGF which, like VEGF, has been proposed as a regulator of tumor angiogenesis (16 , 17) . Our data demonstrate that malignant ovarian cysts have markedly elevated levels of VEGF. Benign ovarian cysts have either undetectable or low levels of VEGF, whereas borderline tumors secrete low to intermediate amounts of VEGF. bFGF levels in malignant cysts are either undetectable or very low, and no significant differences are found in bFGF levels among malignant, benign, borderline, and functional cysts. These findings indicate that VEGF levels in ovarian cyst fluid may represent a useful biomarker of angiogenesis and tumor progression. They also support the idea that VEGF plays an important role in ovarian cancer-related angiogenesis and may represent an important target of antiangiogenic therapy.

MATERIALS AND METHODS

Fluid Collection.

Fluid was collected from ovarian cysts in the surgical pathology laboratories of the Medical College of Pennsylvania Hospital (Philadelphia PA) and of the H. L. Moffitt Cancer Center and Research Institute (Tampa, FL). Each operative specimen was obtained with the patient’s informed consent. Fluid was obtained by puncturing the cyst wall with an 18-gauge needle mounted on a 10-ml syringe. If a tumor contained more than one cyst, fluid was collected from the dominant cyst. After collection, the syringe was labeled with the patient name and surgical pathology number and stored at −70°C. Before each assay, the fluids were thawed and centrifuged at 2000 rpm for 10 min. The supernatants were used for the assay. Excess supernatant fluid was aliquoted and stored at −70°C. Ovarian fluids were evaluated for VEGF, bFGF, and total protein.

ELISA.

VEGF and bFGF were measured by capture ELISA (R & D Systems, Minneapolis, MN) using a Biotek microplate reader. The sensitivity of the assays was <9 pg/ml for bFGF and <1 pg/ml for VEGF. Fluids were diluted 1:10 to 1:100 using dilution buffer and measured in microtiter plates according to the manufacturer’s recommendations. Standard curves were prepared with human recombinant VEGF and bFGF (R & D Systems). Endothelial Cell Growth Supplement (Sigma Chemical Co., St. Louis, MO) was used as a positive control for the bFGF ELISA. Each mg of Endothelial Cell Growth Supplement contained 2.3 ng of bFGF. Total proteins were measured by the Bradford method using the Bio-Rad Protein Assay (Bio-Rad, Burlingame, CA).

All samples were measured without prior knowledge of the diagnosis. Differences among experimental groups were evaluated by ANOVA, followed by Student-Newman-Keuls test. For evaluation of tumor recurrence in patients with malignant cysts, differences were assessed by Student’s t test. Statistical significance was set at P < 0.05.

Immunohistochemical Studies.

Representative tumor sections mounted on Fisher Plus slides (Fisher) were baked at 60°C overnight, deparaffinized in xylene, and rehydrated in graded ethanols. Endogenous peroxidase activity was quenched with 3% hydrogen peroxide in methanol. Antigen unmasking was performed in a microwave oven using sodium citrate buffer (pH 6.0). The sections were blocked in a 20% solution of goat serum (Sigma) in PBS and reacted with either primary rabbit anti-human VEGF antibody (Santa Cruz Biotechnology, Inc., Santa Cruz, CA) or nonimmune rabbit IgG (Sigma). The antibodies were diluted to 0.1 μg/ml with 1% goat serum in PBS. To demonstrate the specificity of the VEGF antibody reaction, the primary antibody was incubated overnight at 4°C with the corresponding peptide (Santa Cruz Biotechnology, Inc.), according to the manufacturer’s protocol. The neutralized antibody was then used as a negative control. The sections were then incubated in biotinylated goat anti-rabbit IgG (Dako, Carpinteria, CA; 1:100 in 1% goat serum/PBS), reacted with avidin-biotin complex (Vector ABC Elite), developed with diaminobenzidine, and counterstained with Meyer’s hematoxylin.

RESULTS

ELISA of Ovarian Cyst Fluid.

VEGF and bFGF levels were evaluated in 23 patients with benign cysts, 8 patients with functional cysts, 5 patients with borderline tumors, and 13 patients with malignant tumors. Data on the various patients are listed in Tables 1<$REFLINK> 2<$REFLINK> 3<$REFLINK> 4<$REFLINK> . Benign lesions consisted of 9 simple cysts, 11 serous cystadenomas, 1 mucinous cystadenoma, and 1 paratubal cyst. This group included also one hydrosalpynx, which grossly appeared as an ovarian cyst at the time of fluid collection. Five corpus luteum cysts, two follicular cysts, and one endometriotic cyst were grouped together as functional cysts. The term “functional cyst” was used in a broad sense to include cysts composed of tissues that physiologically produce a variety of growth factors including VEGF during the female menstrual cycle (7) . Borderline lesions included two serous and three mucinous tumors of low malignant potential. Malignant tumors consisted of nine serous cystadenocarcinomas, two endometrioid carcinomas, one mucinous cystadenocarcinomas, and one granulosa cell tumor.

There was a marked difference in VEGF levels between malignant cysts (38.5 ± 8.2 ng/ml) and benign (1.6 ± 0.4 ng/ml), borderline (5.7 ± 1.5 ng/ml), or functional (3.8 ± 2.0 ng/ml) cysts (Fig. 1)<$REFLINK> . Malignant neoplasms had an average 24-fold increase in VEGF over benign lesions and a 6-fold increase over borderline tumors. VEGF levels in malignant tumors ranged from 7.3 to 97.5 ng/ml. Benign cysts had either undetectable or relatively low levels of VEGF. The highest concentration of VEGF among the nonmalignant lesions was detected in a case of cystic endometriosis. Borderline tumors exhibited an average 3-fold increase in VEGF concentration over benign lesions. VEGF levels in these tumors, however, were low to intermediate and overlapped with those of the VEGF-producing benign and functional cysts. There was a slight difference in total proteins between benign and malignant tumors, which, however, could not account for the dramatically elevated levels of VEGF in the latter (Fig. 2)<$REFLINK> .

Unlike VEGF, bFGF was nonmeasurable or very low in malignant cysts except in a serous cystadenocarcinoma (2.5 ng/ml). Although small amounts of bFGF were found in the majority of these cysts, bFGF levels did not correlate with malignancy (Fig. 3)<$REFLINK> . In fact, the highest concentration of bFGF (7 ng/ml) was found in a simple serous cyst.

Immunohistochemical Localization of VEGF.

Immunostaining of histological sections confirmed the presence of VEGF at the tissue level. VEGF localized to the cell cytoplasm of tumor cells in both solid and papillary areas (Fig. 4A)<$REFLINK> . VEGF staining correlated with malignant transformation because benign cysts, which were occasionally found next to malignant cells, were lined by a VEGF-negative epithelium. Stromal elements such as fibroblasts, endothelial cells, and vascular smooth muscle cells exhibited variable immunoreactivity. Plasma cells and leukocytes were positive. Borderline serous tumors stained for VEGF particularly in regions with papillary projections and epithelial tufting (Fig. 4C)<$REFLINK> . Borderline mucinous tumors exhibited a distinctively focal pattern of immunoreactivity. In these tumors, cystic glandular formations lined by stratified atypical epithelium stained for VEGF, whereas the well-differentiated, mucin-secreting epithelium of adjacent cysts was negative (Fig. 4D)<$REFLINK> . The epithelium of benign serous cysts and serous cystadenomas was either negative or weakly to moderately positive (Fig. 4E)<$REFLINK> . Among the functional cysts, the corpus luteum cysts and an endometriotic cyst showed the strongest staining.

Relationship between VEGF Levels and Tumor Recurrence.

Twelve patients with malignant cysts and 5 patients with borderline cysts were followed up to 3 years after initial surgery. Patients with malignant cysts were diagnosed at stages I (n = 2), III (n = 7), or IV (n = 3), whereas patients with borderline cysts were diagnosed at stages I (n = 4) or II (n = 1). Six patients with malignant cysts had evidence of disease at follow-up. They included all stage IV patients, two stage III patients, and one patient initially diagnosed at stage I. There was no evidence of disease at follow-up in patients with borderline tumors. VEGF levels in the primary tumor of patients with evidence of disease were significantly higher than those of patients with no evidence of tumor recurrence (Fig. 5)<$REFLINK> .

DISCUSSION

The results of this study demonstrate that malignant ovarian cysts contain markedly elevated levels of VEGF as compared with benign, borderline, or functional cysts. This finding supports the idea that VEGF plays an important role in ovarian-related tumor angiogenesis and indicates that high VEGF levels in ovarian cyst fluid may represent an indicator of malignancy and tumor progression in the ovary. Using in situ hybridization, Abu-Jawdeh et al. (18) found strong expression of VEGF in 29 of 29 ovarian cancers and 2 of 8 borderline tumors. They also measured by an immunofluorometric method VEGF in the cyst fluid of two cancer patients and found it to be elevated in relation to benign patients. Paley et al. (19) observed overexpression of VEGF by in situ hybridization in 29 of 68 ovarian cancers and reported that median disease-free survival for patients with positive VEGF was 22 months, compared with >120 months for the VEGF-negative group. Our findings confirm and expand these observations and establish ELISA as a rapid and sensitive method for measuring VEGF in the fluid of ovarian cystic lesions. The range of VEGF levels found in our series may reflect differences in the angiogenic capacity and, possibly, biological behavior of different ovarian neoplasms.

Dvorak et al. (8) have shown that VEGF is a potent vasopermeability factor. In ovarian cancer, VEGF may cause accumulation of cyst fluid by increasing the permeability of the tumor microvasculature (20) . A similar mechanism may explain the massive accumulation of ascitic fluid, which follows the i.p. spread of this tumor. This idea is supported by experimental studies in guinea pigs that demonstrate a link between VEGF production, angiogenesis, increased vascular permeability, and i.p. growth of tumors (21) .

VEGF may not only function as an angiogenic and vasopermeability factor but also as an autocrine factor for the tumor cells themselves. Although it is known as an endothelium-specific mitogen, VEGF may potentially influence the behavior of ovarian cancer cells because these cells may express KDR, which is a high affinity VEGF receptor (22) . It is, however, not clear what effects, if any, VEGF may have on ovarian cancer cells. Boocock et al. (22) tested the responsiveness of ovarian cancer cell lines to VEGF but were unable to demonstrate a mitogenic effect. The observation that VEGF promotes in vitro papillogenesis by rabbit ovarian surface epithelial cells has raised the possibility that this growth factor may contribute to the morphogenesis of papillae in ovarian cancer (23) . In addition, a close correlation has been demonstrated between VEGF expression and tumor cell proliferation in human ovarian cancer (24) .

An interesting result of our study was the demonstration of an average 3-fold increase of VEGF levels in borderline tumors, as compared with benign lesions. Borderline tumors are neoplasms of low malignant potential. There is, however, a subset of borderline tumors that behaves aggressively and progresses to carcinoma (25) . It is possible that this process is preceded by a switch of the borderline tumors to an angiogenic phenotype, as described during tumor progression in experimental animals (26) . Our immunohistochemical findings that VEGF expression in borderline mucinous lesions is focal and correlates with cellular atypia are compatible with this possibility. If the hypothesis of the angiogenic switch is validated in ovarian cancer, borderline tumors prone to malignant transformation may be identified by demonstrating high levels of VEGF in the cyst fluid. Interestingly, in our series, the VEGF levels of two borderline tumors were within the range of VEGF levels in malignant tumors.

Previous studies have demonstrated a correlation between angiogenesis, measured by counting microvessels on histological sections, and the tendency of malignant tumors to metastasize (27) . This observation, although reproduced in many laboratories, has been questioned by some who have not been able to confirm it (28) . This discordance may be due to sampling errors, the subjective evaluation of histological sections by different observers, or the biological variability among different tumor types. To that end, VEGF ELISA of ovarian cyst fluid represents an alternative approach for measuring tumor angiogenic activity, which is more objective than evaluation of microvascular density.

The lack of correlation between cyst fluid bFGF and malignancy does not preclude the possibility that this growth factor is involved in ovarian cancer-related angiogenesis. Indeed, other investigators have reported the presence of bFGF in this type of tumor (16 , 17) . These studies, however, were based on immunochemical analysis of tissue extracts, and no attempts were made at measuring bFGF in the fluid phase. The absence or relatively low levels of bFGF in ovarian cyst fluid may be due to the fact that bFGF is not secreted through conventional pathways because it lacks a signal peptide sequence (29) . In addition, bFGF may become sequestered in the extracellular matrix due to its heparan sulfate-binding properties (29) . Thus, although bFGF is probably involved in ovarian cancer-related angiogenesis, its molecular properties may hamper its accumulation in the cyst fluid. Our findings that the majority of ovarian malignancies have low but detectable levels of bFGF are consistent with previous reports demonstrating bFGF in cerebrospinal fluid (30) , serum (31) , and urine (32) of cancer patients. In these studies, however, bFGF levels of cancer patients were compared with those of normal individuals. In our study, bFGF levels in malignant ovarian cysts were compared with those of borderline, benign, and functional cysts. Interestingly, bFGF was found in three corpus luteum cysts, one follicular cyst, four serous cysts, three serous cystadenomas, and one hydrosalpynx.

In future studies VEGF ELISA of ovarian cyst fluid may be used for diagnostic and prognostic applications. Early cystic lesions of the ovary detected by intravaginal ultrasounds or by magnetic resonance imaging (33) may be tested for VEGF production. Measuring VEGF in these cysts could differentiate benign from malignant lesions. This approach, however, is presently limited by the concern that puncturing a cyst for fluid collection may spread a possible malignancy. Alternatively, the malignant nature of these cysts might be assessed by measuring VEGF in the circulation. VEGF ELISA of cyst fluid performed after surgery may provide information concerning the angiogenic activity and aggressiveness of the tumor. Our observation that VEGF levels are higher in ovarian cancers that tend to recur supports this possibility and warrants further investigation in a larger prospective series. Recurrence of disease after surgery could be monitored by measuring serum levels of VEGF, which are elevated and correlate with tumor burden in ovarian cancer patients (34) . VEGF levels in ovarian cyst fluid may be used to identify borderline tumors that are prone to malignant behavior. VEGF levels should be interpreted with caution in cases of endometriosis because they may be significantly elevated in these patients, as reported previously (35) , and as confirmed in our study. In addition, the age of the patient should be taken into account because VEGF regulates the vascularization of the corpus luteum during the reproductive years (36) . However, VEGF levels in the follicular and corpus luteum cysts included in our study were all below those of the malignant cysts.

More studies are needed to better understand the prognostic significance of the increased VEGF levels in ovarian cancer. Particularly important will be to increase the number of observations and determine the range of VEGF concentration in the different types of ovarian cysts. Evaluation of VEGF in the circulation will establish whether VEGF measurement can be used for early detection of ovarian cancer and for the monitoring of patients after surgery. Finally, it will be important to investigate the biological significance of VEGF production in ovarian cancer progression and determine the therapeutic efficacy of drugs that specifically target VEGF and its receptor system.