PDGFRα/β and VEGFR2 polymorphisms in colorectal cancer: incidence and implications in clinical outcome

Colorectal cancer (CRC) is the third most common tumour in the world, with over 1.2 million new cases diagnosed every year, and is responsible for about 8% of cancer related deaths [1]. Approximately one third of patients present metastatic disease at diagnosis, and about 40% of those with early-stage tumors will eventually relapse at some point over the course of the disease [2, 3]. Although prognosis has greatly improved over the past decades due to significant surgical and medical advances, once the tumor has progressed beyond surgical resectability, the disease is essentially incurable and median survival ranges from 14 to 24 months with best available systemic therapy [4]. Development of new more effective agents is thus actively pursued.

Angiogenesis has become a major target in colorectal cancer therapy. Bevacizumab, a humanized monoclonal antibody against the vascular endothelial growth factor A (VEGF-A), was the first antiangiogenic agent to demonstrate efficacy in CRC. In the pivotal study by Hurwitz et al., the addition of this agent to irinotecan-based combination cytotoxic therapy significantly improved survival compared to irinotecan-based chemotherapy alone in patients with advanced CRC [5]. Subsequently, bevacizumab has been tested in combination with other chemotherapy regimens with more modest results [3, 4]. More recently, a benefit in survival has been also reported in patients with advanced CRC with two new promising antiangiogenic drugs: aflibercept (a VEGF trap) in combination with FOLFIRI (folinic acid, 5-fluoruracil and irinotecan) following progression to oxaliplatin-based therapy [6], and regorafenib (a novel tyrosine kinase inhibitor targeting VEGFR, PDGFR, FGFR, RET, KIT and TIE2) as single-agent therapy in patients who had progressed to all standard therapies [7]. These results clearly illustrate angiogenesis inhibition is to play a major role in the management of this disease.

Angiogenesis is a highly controlled process under physiological conditions, such as embryonal development, postnatal growth and wound healing, but is also a critical driver of tumor growth and progression [8]. It is tightly regulated by a complex equilibrium among different pro- and antiangiogenic factors secreted both by tumor cells and by cells of the tumor microenvironment (pericytes, endothelial, mesenchymal or immune cells). VEGF and their receptors represent one of the best validated pathways involved in angiogenesis [9, 10]. VEGF stimulates both proliferation and migration of endothelial cells, enhances microvascular permeability, and is essential for revascularization during tumor formation. It is commonly over-expressed in human tumors, and this is often associated with increased vascular density and more aggressive clinical behavior. VEGF-A and its main receptor, VEGFR2/KDR, are key members of this family and common targets of antiangiogenic agents [11, 12].

Platelet-derived growth factor (PDGF) and their receptors (PDGFR-α, PDGFR-β and PDGFR-αβ) play also a critical role in angiogenesis regulation by exerting important control functions in mesenchymal cells during development [13]. PDGF is expressed by endothelial cells and acts in a paracrine manner by recruiting PDGFR-expressing cells, such as pericytes and smooth muscle cells, to the developing vessels, thus improving pericyte coverage and vessel function. PDGF signaling promotes cell migration, survival and proliferation and indirectly regulates angiogenesis by inducing VEGF transcription and secretion [10, 13, 14]. Mutations involving up-regulation of PDGF and/or PDGFR, as well as PDGFR-dependent growth stimulation, have been documented in a number of solid tumors and hematological malignancies, suggesting a likely role of this pathway in carcinogenesis [10, 15]. Moreover, agents antagonizing PDGFR-mediated signaling have also demonstrated antineoplastic activity in preclinical models and in clinical trials, including some conducted in patients with CRC (i.e. regorafenib) [7]. Nevertheless, several other drugs also targeting these pathways (i.e. sunitinib, sorafenib) [16, 17] have failed to prove a significant positive impact on the outcome of patients with CRC. The biological grounds for these discordant results are not well understood.

Therefore, and in spite of their undeniable success, only a small proportion of patients do actually benefit from antiangiogenic agents, and reliable tools to prospectively identify which patients are more likely to benefit are scarce. In this scenario, efforts to unravel the intricate molecular pathways governing tumor angiogenesis are certainly needed for progress to be made. In the present study, we sought to evaluate the incidence of genetic polymorphisms of some of the key players of angiogenesis, such as VEGFR-2, PDGFR-α and PDGFR-β, and their potential influence in CRC biology. With this purpose we sequenced the tyrosine kinase domains of these receptors in 8 CRC cell lines (T84, LOVO, LS174T, HT29, LS180, SW48, SW480, COLO205) and in 92 tumor samples of patients with colorectal adenocarcinoma. Correlations of encountered genetic variables with protein expression in cell lines, as well as with clinicopathological features and survival of these patients were also analyzed to assess their potential biological and clinical implications.

The present study evaluated the incidence of VEGFR2, PDGFRα and PDGFRβ TK domain genetic variants in different CRC cell lines (T84, LOVO, LS174T, HT29, LS180, SW48, SW480, COLO205) and in tumor samples of 92 patients diagnosed of colorectal adenocarcinoma. Four SNPs were identified, three in PDGFRα [exon 12 (A12): c.1701A>G, rs1873778; exon 13 (A13): c.1809G>A, rs10028020; and exon 17 (A17): c.2439+58C>A, rs2412559] and one in PDGFRβ [exon 19 (B19): c.2601A>G, rs246395]. SNP B19, present in 4 CRC cell lines (LS174T, LS180, SW48, COLO205) and in 58% of patients, had a substantial impact on overall survival, with 5-year survival rates of 51% for patients with PDGFR B19 wild type tumors versus 17% for those harboring the SNP variant (c.2601A>G). This is the first study to analyze the PDGFR genotype in a series of human colorectal cancer and its correlation with different clinicopathological features, and to demonstrate a significant association of a PDGFR SNP with patients’ outcome.

Angiogenesis is a complex process controlled by a number of interconnected signaling pathways, among which PDGF and their receptors play a critical role. Moreover, PDGFR has been the target for many newly developed anticancer drugs, some of them with proven efficacy in CRC (i.e. regorafenib) [7] and some that have failed to demonstrate a benefit in patients with this tumor type (i.e. sunitinib, sorafenib) [16, 17]. Despite this, however, only few studies have analyzed the clinical implications of PDGF/PDGFR expression in colorectal cancer. In this regard, Schimanski and cols reported that specific receptor tyrosine kinases (TK) were overexpressed in K-ras mutated CRC [22]. In particular, VEGFR1, VEGFR2 and PDGFRα expression, documented in 95%, 46% and 62% of tested samples, respectively, were significantly linked to K-ras codon 12 or 13 mutations. Whether this could translate into a higher likelihood of responding to TK inhibitors, however, is a matter of speculation. On the other hand, Wheler et al. reported, in a series of 99 human colorectal carcinomas, that co-expression of PDGFRα/β, observed in 57% of tumor samples, was significantly associated with lymphatic metastasis (P=0.007) and advanced tumor stage (P=0.03) [23]. Similarly, high PDGFRβ tumor stromal expression significantly correlated with more aggressive clinical behavior in patients with breast cancer, including high histopathological grade, estrogen receptor negativity, high HER2 expression and shorter survival [24].

Nevertheless, PDGFR genetic variants had never been previously assessed in CRC patients. In our study, four genetic variants were identified, all of them corresponding to SNPs previously reported in public databases. Three of them were silent mutations (A12, A13 and B19) and the other one was an intronic insertion (A17). PDGFRα exon 12 SNP (rs1873778), present in homozygosis in all CRC cell lines and 100% of analyzed tumor samples, has been also described in other neoplasias although in a smaller proportion of patients, including KIT and FLT3 mutation-negative core binding factor (CBFL) acute myeloid leukemias (14% of 35 patients) [25], cervical adenosquamous carcinomas (30% of 30 patients) [26] and gliomas (7% of 86 patients) [27]. In this last study, no association was found between the presence of this mutation and PDGFRα tissue expression. Our results are in agreement with the distribution reported for a European Caucasian population at the NCBI website (http://www.ncbi.nlm.nih.gov/sites/entrez/), being the G-allele the most frequently encountered (p=0.98). PDGFRα exon 13 SNP (rs10028020), detected in heterozygosis in 2 (LS174T and LS180) of the 8 cell lines examined and in 18% of tumor samples, was associated with poorer tumor differentiation but no significant correlation was found with survival. This polymorphism had been first reported also in heterozygosis by Trojani et al. in 34% of CBFL acute leukemias [25], although potential association of this genotype with clinical features or patient′s outcome was not explored by these authors. Finally, neither PDGFRα exon 17 SNP (rs2412559), identified in all of our patients, nor PDGFRβ exon 19 SNP (rs246395), present in 58% of them, had been previously described in human cancers. PDGFR B19 SNP has been reported to be present in the general population with a frequency of 37%, and was more commonly encountered in our study population among colon primary tumors (64%) than in tumors of rectal origin (35%). Of note, and despite not being an activating mutation, the B19 SNP was found to be a significant prognostic factor (HR: 2.89, p=0.029) independent of tumor stage or patient′s age. This negative effect on patient′s survival did not differ according to primary tumor location (data not shown).

That the identified SNP in exon 19 of PDGFRβ may indeed have relevant biological implications is further supported by the fact that analysis of protein content in cell lines demonstrated the presence of the B19 SNP clearly correlated with higher protein levels of the PDGF receptor β, also in its phosphorylated state. PDGF pathway constitutive activation maintains highly active MEK, thus phosphorylating Bad and inhibiting apoptosis [14, 15]. Increased PDGF pathway activation has been also shown to contribute to drug resistance by activating the PI3K pathway [14, 15]. Whether or not the presence of this SNP may portend particular sensitivity to PDGFR-targeted agents is a matter of speculation but certainly deserves further investigation due to its relevant potential clinical applications.

On the contrary, no relevant findings were identified in our series regarding VEGFR2 TK domain SNP analysis. As in other solid tumors, overexpression of VEGF mRNA and protein has been associated with tumor progression and poor prognosis of colon carcinoma [28]. The VEGF-A gene is known to be highly polymorphic and harbors numerous SNPs, especially in the promoter, 5’- and 3’-untranslated regions (UTR), which contain key regulatory elements that are sensitive to hypoxia [29]. These SNPs contribute to the high variability in VEGF production among tissues and have been associated with cancer susceptibility, progression, and anti-VEGF therapeutic response in subjects with a variety of solid tumors including colorectal cancer. For example, the 936 T-allele has been associated with increased risk of CRC, advanced stage of disease and worse prognosis, whereas the 634 C allele was predictive of decreased risk and improved survival. SNPs have also been identified in the VEGF receptor genes, although the literature in this topic is still very sparse. Very recently, the VEGFR-1 319 C/A SNP, located in the promoter region of the gene, has been reported to be associated with response to therapy in a cohort of 218 CRC patients treated with different bevacizumab-containing regimens [30]. In this study by Hansen et al., response rates were significantly higher in patients homozygous for the A-allele (AA) than in patients with the C-allele genotype (CC or CA) (56% vs 39%, p=0.015). Similar results were also documented in bevacizumab-treated pancreatic cancer patients [31]. In addition, functional relevance has been demonstrated for several SNPs in the VEGFR-1 and VEGFR-2 genes, particularly SNPs 1192C/T (V2971I; rs2305948) and 1719T/A (H472Q; rs1870377). These SNPs are located in exons 7 and 11, and lead to amino acid changes potentially interfering with the receptor’s binding affinity to VEGF-A. In the current study, however, we aimed to explore potential genetic variations in the TK domain of the VEGFR-2 (exons 17 to 26), which would be expected to have relevant functional consequences. No mutations were however detected in our study population in these gene domains.

Identification of relevant SNPs in critical genes involved in angiogenesis may therefore become valuable tools in assessing risk or predicting cancer response to therapy or prognosis. However, no consensus exists at present regarding the use of any of these for clinical decisions as many studies have reported diverging, conflicting or inconclusive results. Multiple reasons may be responsible for these discrepancies, including gender and interethnic differences in the distribution of alleles, heterogeneous study populations and small sample sizes, different sources of DNA (i.e., tumor vs germline) and different methods for SNP analyses, lack of corrections for multiple testing, links to other loci in the gene or related genes responsible for the observed effect, bias due to post-transcriptional gene regulation, or simultaneous presence of somatic or epigenetic changes that may influence outcome. Prospective validation in appropriately sized and controlled studies is therefore required before these genetic variants may be used in clinical practice.

In conclusion, the present study has identified, for the first time, PDGFRβ genetic variants with relevant clinical and biological implications. In particular, the G-allele genotype of PDGFRβ exon 19 SNP (rs246395), which was commonly encountered in our series of CRC patients (58%), was associated with increased pathway activation and poorer survival. Further studies to assess the functional consequences of this genetic variant, as well as to validate its role as a prognostic marker in this disease are certainly warranted. Implications regarding its potential influence in response to PDGFR-targeted agents remain to be elucidated.

This study has identified, for the first time, PDGFRβ genetic variants with relevant clinical and biological implications in colorectal cancer. In particular, the G-allele genotype of PDGFRβ exon 19 SNP (rs246395), encountered in 58% of tumor samples from colorectal cancer patients, was associated with increased pathway activation and significantly poorer survival.