Serum angiopoietin-2 and soluble VEGFR-2 levels predict malignancy of ovarian neoplasm and poor prognosis in epithelial ovarian cancer

Epithelial tumors cover most neoplasms of the ovaries. Although most of them are benign or have low malignant potential, malignant ovarian neoplasms cause more deaths than other gynecological cancers together. It is crucial that the malignant forms of neoplasms are diagnosed and differentiated from benign tumors as early as possible to treat patients adequately. Cytoreductive surgery and platinum-based therapy combined with paclitaxel have become the standard first-line therapy in epithelial ovarian cancer [1]. Regardless of the high initial chemosensitivity most patients develop chemoresistance with the 5-year overall survival of only 25-35% [2]. Identification of cancer growth and dissemination mechanisms at the molecular level has led to more targeted treatments. Therefore, biomarkers predicting patient prognosis or response to specific therapies enhance the development of more personalized agents [3].

In cancer, including ovarian cancer, targeting endothelial cells of tumor blood vessels has become an emerging strategy to inhibit tumor growth [4‐6]. VEGFs (vascular endothelial growth factors) and their receptors play significant roles in tumor angiogenesis and lymphangiogenesis and are mostly specific to vascular endothelial cells [7, 8]. VEGF-A, -B, -C, -D and PLGF signal through three tyrosine kinase receptors VEGFR-1, -2 and -3, also known as Flt-1, KDR/Flk-1 and Flt-4 [7]. Both VEGFR-1 and -2 bind VEGF-A, which is the main regulator of blood vessel growth. VEGF-A also induces vessel permeability and the accumulation of malignant effusions of ascites in ovarian cancer [9]. VEGF-C and-D stimulate lymphangiogenesis through VEGFR-3 which is predominantly expressed in lymphatic endothelium [10, 11] but also exists in angiogenic sprouts [12].

Ang-1 and Ang-2 are ligands for the tyrosine kinase receptor Tie2 [13, 14]. Ang-1 is expressed by pericytes, smooth muscle cells and fibroblasts and it promotes vascular maturation in a paracrine manner by attracting pericytes and smooth muscle cells to the developing vessels and contributes to tumor dissemination and metastasis [15]. Ang-2, on the contrary, functions as an autocrine controller of endothelial cells in a context- dependent manner promoting either blood vessel growth or regression depending on the levels of other growth factors, such as VEGF-A [16, 17].

Angiogenesis related circulating proteins are referred as potential biomarkers in ovarian cancer [18]. In a previous study we have reported the role of circulating Ang-2 in predicting the prognosis of ovarian cancer [19]. However, since angiogenesis is driven by multiple pathways, measuring only one individual circulating protein of a single pathway might not be sufficient. Simultaneous evaluation of the levels of VEGF members and their receptors and angiopoietins may provide more accurate diagnostic and prognostic information. At present, cancer studies in which both the circulating levels of VEGFs, sVEGFRs and angiopoietins are measured and combined are still missing, since only individual angiogenic or lymphangiogenic growth factors and receptors have been reported previously [20, 21].

In this study we have measured the preoperative serum levels of VEGF-A, C and D, sVEGFR-1, -2 and -3 as well as Ang-1 and Ang-2 in the patients with epithelial ovarian neoplasm. The aim of this study was to find out (1) whether levels of measured growth factors and receptors differ in patients with benign, borderline or epithelial ovarian neoplasms, (2) how the measured levels predict the clinical course and survival of patients with epithelial ovarian cancer and (3) whether it is useful to combine measurements of two angiogenesis and lymphangiogenesis associated pathways. To our knowledge, this is the first study in which a panel of VEGFs and their receptors and Ang-1 and Ang-2 levels are quantified from the serum samples of the same patient population and correlated with the diagnosis and clinical outcomes of ovarian carcinoma patients.

To date this is the first study in which preoperative serum levels of a panel of growth factors and receptors of two major angiogenic pathways, VEGFs/VEGFRs and angiopoietins, are measured in the same patient population and linked to the diagnosis of the patients with ovarian neoplasm and to the clinical outcome and prognosis of ovarian cancer patients. We found that levels of VEGF, VEGF-D and both Ang-1 and Ang-2 were higher in patients with ovarian carcinoma compared to patients with benign or borderline tumors. Conversely, the level of sVEGFR-2 was lower in patients with ovarian carcinoma than in women with normal ovaries or benign neoplasms. Further, serum level of Ang-2 predicted the most significantly poor OS and Ang-2/sVEGFR-2 ratio the presence of malignant ovarian neoplasm and short RFS.

Clinical trials targeting tumor vascular supply by inhibiting VEGF or angiopoieting pathways have been reported [4‐6]. In preclinical settings dual targeting to VEGF/VEGFR and Ang-2/Tie-2 axis has shown enhanced benefits to block tumor growth [24‐26]. There are large efforts to find validated biomarkers to select patients that would benefit from antiangiogenic treatments and to follow their responses to the treatments. Gourley et al. have shown that up-regulated gene expression of proangiogenic factors has an impact on a longer progression free survival when patients are treated with bevacizumab [27]. It might be possible that circulating proangiogenic factors described in our study have potential to predict response to more personalized antiangiogenic treatments like bevacizumab or trebananib, but this clearly needs further clinical trials with pretreatment circulating levels of proangiogenic factors combined with antiangiogenic therapy. Circulating proteins associated with angiogenesis are considered the most potential biomarkers of the antiangiogenic treatments since surgical procedures are not needed in the follow-up and it is possible to monitor serial samples in routine clinical practice [18]. In the early stage of the angiogenic switch, invasive tumor cells grow along pre-existing vessels. That leads to endothelial cell activation and high Ang-2 expression resulting in endothelial cell apoptosis and regression of co-opted blood vessels. Increased intratumoral hypoxia results in continuous over-production of VEGF and initation of angiogenesis [8, 16]. In ovarian cancer, it has been demonstrated that increased hypoxia and tumor-derived VEGF further up-regulate the expression of Ang-2 in endothelial cells [17] .

Higher serum levels of VEGF have been measured in patients with ovarian carcinoma compared to patients with benign ovarian neoplasms [28, 29], but controversial results also exist [30]. Also, results from the effect of VEGF on the prognosis of ovarian carcinoma and on other cancers have been conflicting, although in most studies high circulating VEGF levels have predicted poor prognosis [20, 31‐35] similarly to the present study. Next to VEGF are VEGF-C and VEGF-D, which are mainly linked to lymphangiogenesis. Circulating levels of VEGF-C and –D have been less studied than levels of VEGF in cancer. So far, the results have been variable in a few cancer studies [35‐37]. To our knowledge, this is the first study reporting circulating levels of VEGF-D in patients with ovarian cancer.

Soluble VEGF receptors lack a transmembrane region of the full length receptors. sVEGFR-1 is the product of alternative mRNA splicing but it is unknown whether the sVEGFR-2 is a product of ectodomain shedding from cell-surface VEGFR-2 or a product of alternative mRNA splice variation [38]. In our study sVEGFR-1 had no significant role to distinguish benign from malignant ovarian neoplasms which was in line with previous studies [30] and did not have an effect on survival of ovarian cancer patients. In studies of other cancers the role of sVEGFR-1 as a prognostic factor has been variable [33, 35, 39]. Ebos et al. [38] have shown in preclinical models that sVEGFR-2 plasma levels decrease due to tumor derived VEGF and is the result of ligand–induced downregulation of the VEGFR-2 from the cell-surface. Decreased sVEGFR-2 levels have also been reported in clinical trials utilizing multi tyrosine kinase inhibitors such as sunitinib or sorafenib [20, 40, 41]. Interestingly, in our study circulating levels of sVEGFR-2 were lower in patients with ovarian cancer compared to those of healthy controls and low sVEGFR-2 level was also associated with the recurrence of ovarian cancer and predicted poor prognosis. This finding parallels with earlier results in an ovarian cancer animal model, in which adenoviral gene therapy with soluble VEGFRs produced high plasma level of sVEGFR-2 having significant antiangiogenic and antitumoral effects [42]. Studies concerning circulating levels of sVEGFR-3 are still limited. Although decreasing levels of sVEGFR-3 have been reported during multitargeted antiangiogenic treatment in metastatic renal cell and colorectal carcinomas [43, 44] and sVEGFR-3 has been associated with short progression free survival and poor prognosis in melanoma [45], in our study sVEGFR-3 did not have an effect on OS or RFS. Further studies with lymphangiogenesis related VEGF-C, -D and sVEGFR-3 are justified, since targeted treatments to this axis are under development [46, 47].

One purpose of this study was to evaluate the strength of angiopoietins as biomarkers in relation to the members of VEGFs/sVEGFRs pathways in serum of ovarian tumor patients. In the present study we showed that Ang-2 and sVEGFR-2 as single biomarkers were the most potential to identify healthy women or patients with benign or semimalignant ovarian neoplasms from ovarian carcinoma, although overlapping levels between benign and malignant tumors were noted. VEGF/VEGFR-2 ratio was more accurate to differentiate malignant potential of ovarian tumors than measurements of VEGF or sVEGFR-2 alone and might reflect the situation that more VEGF is available to bind full-length VEGFR-2 due to the lesser amount of soluble VEGFR-2. Interestingly, Ang-2 alone predicted most potentially ovarian carcinoma even when compared to Ang-2/sVEGFR-2 ratio. In ROC curves the role of Ang-2 as a diagnostic biomarker was supported since it yielded almost the same AUC value than Ang-2/sVEGFR-2 ratio. However, neither of the measurements reached the level of CA-125 which is the commonly used biomarker to distinguish benign and malignant ovarian neoplasms [48].

We found significant associations between common clinicopathological features of ovarian carcinoma and measured angiogenic biomarkers. Overall it was shown that angiogenic markers were associated most often with the dissemination of the disease, with the larger size of primary residual tumor and with the recurrence of the ovarian carcinoma, features that are related to angiogenesis. It was not a surprise that formation of ascites was linked very significantly to high VEGF and high VEGF/sVEGFR-2 ratio since the role of VEGF in ascites formation has been demonstrated [9].

In univariate survival analyses high Ang-2 level most significantly predicted poor OS and high Ang-2/sVEGFR-2 ratio predicted short RFS most effectively compared to other measured angiogenesis markers. These results support the findings of clinical studies in which Ang-2 [21] and sVEGFR-2 [32, 41] have had potential to predict the response to the antiangiogenic treatments as opposed to the circulating level of VEGF [49]. In this study 91% of serous ovarian carcinomas were high grade tumors and the rest 9% were low grade serous tumors. No statistical differences were noticed between those groups and angiogenic biomarker serum levels. However, when we looked only the high grade serous subgroup, OS was significantly shortened with high Ang-2, VEGF and Ang-2/VEGF level (data not shown).

We conclude that measuring circulating protein of two angiogenic pathways gives a better insight into the angiogenic profile of ovarian neoplasms and prediction of the disease outcome in the ovarian cancer patients. These results suggest that Ang-2 and Ang-2/sVEGFR-2 ratio may have potential as an angiogenic marker of decreased patient survival in clinic.