Role of the VEGF ligand to receptor ratio in the progression of mismatch repair-proficient colorectal cancer

Angiogenesis, the process of developing new blood vessels from pre-existing vascular networks, is now a well-described mechanism leading to the initiation and maintenance of tumours, and the promotion of metastasis at secondary sites [1]. Hypoxia is a major activator of angiogenesis in tumours [2]; the hypoxic state of cells promotes the up-regulation of a variety of cytokines and tumour suppressors, such as p53 and also of hypoxia-inducible factor 1-alpha, primarily known for its ability to activate Vascular Endothelial Growth Factor (VEGF) expression [3].

The VEGF family of ligands and receptors includes VEGF-A, VEGF-B, VEGF-C, VEGF-D, platelet derived growth factor (PlGF) and VEGFR1, VEGFR2, VEGFR3 and neuropilin NP1 and NP2 [4]. The best characterized of the VEGF family members is VEGF-A, whose binding to VEGFR2 (FLK1) is the predominant mechanism through which tumour cells promote angiogenesis. VEGF-A/VEGFR2 binding activates RAS/RAF-1/MEK/ERK phosphorylation as well as signalling through PI3K/pAKT. In response to signalling activity, up-regulation of downstream effectors such as mdm2, p53, p27, endothelial nitric oxide, and Bcl-2 can occur as well as inhibition of pro-apoptotic proteins caspase-9 and APAF-1. The consequences of this binding are increased vascular permeability, enhanced endothelial cell proliferation as well as increased survival, migration and invasion of tumour cells. Although significantly less is known about VEGFR1 (FLT1), it appears to function as a negative regulator of angiogenesis [5]. VEGF-A is expressed on vascular cells and binds to VEGFR1 with an affinity that is much higher than that for VEGFR2. However, VEGFA seems to induce much weaker tyrosine kinase activity in VEGFR1 possibly because of an inhibitory sequence in the juxtamembrane domain that represses VEGFR1 activity [6]. In keeping with this observation, a model for VEGFR1 has been developed whereby it could act as a decoy receptor to modulate angiogenesis through its ability to sequester VEGFA thereby reducing signaling through VEGFR2. VEGF-B has also been found to bind to VEGFR1, although the role of this interaction remains to be completely elucidated. VEGFR3 is the specific receptor for VEGF-C and -D and is predominantly found on lymphatic, but also to a lesser extent, on vascular endothelial cells and also on tumour cells [7]. Interestingly, VEGF-C along with VEGF-A and a variety of pro-angiogenic cytokines have been shown to be released from tumour associated macrophages, whose infiltration is thought to be, at least in part, responsible for the angiogenic switch in tumours whereby the balance of pro- and anti-angiogeneic factors favour a pro-angiogenic phenotype [8‐10].

In 1971, the pioneering work by Folkman and colleagues led to the hypothesis that anti-angiogenic compounds could be successfully applied as anti-cancer therapies [11, 12]. In fact, blocking of VEGF has been shown to lead to normalization of the vasculature, thus increasing the efficacy of both radiotherapy (by increasing the partial oxygen pressure of cells) and also the delivery of chemotherapeutic agents to target cells (by decreasing vascular permeability) [13]. Currently, the humanized monoclonal antibody Bevacizumab approved for the treatment of patients with metastatic colorectal cancer has been successful in improving overall survival times in several randomized controlled studies while other approaches such as the use of tyrosine kinase inhibitors continue to be investigated [14, 15]. VEGFR1 immunoreactivity in tumour cells has been correlated with poor prognosis, metastasis and recurrence in a variety of tumour types including breast and lung cancers [16‐18]. Inhibitors of VEGFR1 activity, such as VEGFR1 antibodies or soluble VEGFR1 traps have been developed for preclinical and clinical evaluation and have been shown to suppress tumour growth by inhibiting expression of VEGF on both tumour and stromal cells [5].

Although several studies have evaluated one or more of these VEGF ligands or their receptors by immunohistochemistry and their potential prognostic value, still lacking is a comprehensive analysis performed on a large number of tumours from patients with full clinico-pathological data taking into consideration the different expression ratios between the VEGF ligands and their receptors. Such an evaluation may provide a more profound understanding of the involvement of these angiogenic proteins in colorectal tumour progression, particularly considering the known differences in binding affinities of VEGF ligands to their receptors. The aim of this study was therefore to elucidate the prognostic role of the VEGF ligand to receptor ratios and their effects in tumour progression and metastasis on 387 patients with mismatch repair-proficient colorectal cancers.

This work appears to be the first to evaluate in a single study the immunohistochemical importance of four VEGF ligands with their corresponding receptors in an expression ratio in colorectal cancer. The findings here support a role, not only for VEGF-A, VEGFR1 and VEGFR2 in tumour progression but most importantly of a potential prognostic role of VEGFR1 expression in mismatch repair-proficient colorectal cancer.

The ratio of VEGF-A to VEGFR1 and VEGFR2 as well as the ratio of VEGF-C/VEGFR2 demonstrated the most interesting effects of these angiogenic proteins on progression and survival. These results are similar to those reported by Hanrahan et al. who investigated VEGF ligands and their receptors at the mRNA level in normal, adenoma and colorectal carcinoma [20]. In their study, they suggest that VEGF-A and VEGF-B may be responsible for the initiation of tumour whereas VEGF-A and VEGF-C are further expressed in order to maintain disease progression. They observed a significant correlation between VEGF-A and tumour size but not with tumour stage, lymphovascular invasion or metastasis. In addition, they document a significant link between VEGFR1 expression and tumour grade and Dukes' stage and of both VEGFR1 and VEGFR2 mRNA expression and lymph node positivity. Our findings of an increased VEGF-A expression from normal tissue to tumour, but a lack of association between expression with advanced pT stage, metastasis and survival time further support a role of VEGF-A in initiation and tumour maintenance in colorectal cancer. Furthermore, the combined analysis of VEGF-A with VEGFR1 and their correlation with features of tumour progression and adverse prognosis seem to implicate in particular VEGFR1 and VEGFR2 in the progression of colorectal cancer.

Inflammatory mediators have previously been shown to have a significant effect on the process of angiogenesis through the up-regulation of certain cytokines as well as of VEGF [10, 21]. Not only does VEGF increase vascularity at sites of inflammation but its production by tumour cells results in the expression of inter-cellular adhesion molecule-1 and vascular cell adhesion molecule-1, thereby facilitating the adhesion of leukocytes to endothelial cells [22]. Our results highlight a relationship between the over-expression of VEGF-A as well as VEGFR1 and the peritumoural lymphocytic inflammatory response at the invasive tumour front. The inflammatory response at the tumour border has previously been linked to the tumour border configuration, which we recently underlined as an essential prognostic factor in colorectal cancer [23]. The presence of a conspicuous band of lymphocytes, as described by Jass and colleagues is frequently associated with the presence of a pushing tumour margin, and has been related to an increased number of CD8+ tumour infiltrating lymphocytes and to an improved survival time [24, 25]. In this study, we find that a greater VEGFR2 expression compared to VEGF-A is possibly linked to the presence of an infiltrating margin. Since an infiltrating tumour border configuration is a histomorphologic feature closely correlated to epithelial mesenchymal transition (EMT), whereby tumour cell de-differentiation and loss of cell-cell adhesion occurs at the invasive tumour front, our results may implicate VEGFR2 in this process [26].

The vast majority of the literature suggest a greater invasion and metastatic phenotype in tumours expressing these proteins [27‐33]. In particular, several groups have suggested a VEGFR1-dependent involvement in EMT. Bates and colleagues used a spheroid culture system recapitulating the structure of the colonic epithelium during EMT. Their results find a significant expression of VEGFR1, but not VEGFR2 in these cells [34]. In pancreatic cancer, Yang and coworkers also describe VEGFR1 mediated EMT [35] while in head and neck squamous cell carcinoma (HNSCC), VEGFR2 expression has been linked to vasculogenesis and budding of tumour cells into new vessels [36]. Our results additionally underline not only the expression of VEGF-A as a possible step in tumour progression of colorectal cancer, but more importantly that VEGFR1 and VEGFR2 as well as their ratios with VEGF-A to play a role in the events occurring at the invasive tumour front.

Although VEGF-C and VEGF-D are known primarily as lymphangiogenic proteins, less is known about their prognostic effect in patients with colorectal cancer. Hu and colleagues found that protein expression of VEGF-C and VEGF-D was significantly increased from normal to tumour tissues, a result which we confirm in our study. Furthermore, an increased expression of both these proteins was linked to lymph node metastasis and worse survival time [37]. Kawakami et al. report that VEGF-B and VEGF-C mRNA are significantly higher in tumours with lymph node metastases and in tumours with lymphatic invasion [38] while Onogawa and colleagues report an increased VEGF-C and VEGF-D expression at the invasive tumour front [39]. Others have found a significant association of these proteins with venous and lymphatic invasion as well as with liver metastasis. A recent report by Moehler et al. found that VEGF-D expression correlated with lymph node metastasis and interestingly, that VEGF-D expression was significantly decreased following treatment with anti-EGFR mAb both in vitro and in mouse xenograft models [40]. In our study, a lower expression ratio of VEGF-C/VEGFR2 was linked to more advanced TNM stage.

Our study has several limitations. First it is a retrospective analysis of VEGF ligand and receptor expression and therefore should be investigated in a prospective setting. Secondly, having used the tissue microarray technique, it is possible that tumour heterogeneity was not completely taken into account. As an immunohistochemistry study, inter-laboratory variation may play a role in determining the reproducibility of these findings. Also, having considered adjustment for multiple testing in this study, the associations of VEGF ligands and their receptors fall short of significance although several strong trends were observed. Therefore, our results necessitate confirmation by other, larger study groups. Finally, we were unable to randomize our patient cohort into test and validation subgroups due to the lack of statistical power that this would elicit. Nevertheless, our study may still be a basis for prospective approaches and worth to be validated in future studies.

The results of this study suggest that the VEGF ligand to receptor ratio may be an informative method for evaluating the effects of these angiogenic proteins on tumour progression. Our findings further underline a potential involvement of VEGF-A, VEGR1 and VEGFR2 in events occurring at the invasive tumour front and highlight a possible prognostic role of VEGFR1 expression in mismatch repair-proficient colorectal cancers.