Abstract
Angiogenesis is crucial for development and metastasis of tumors, and vascular endothelial growth factor (VEGF) is a key mediator of this process. The importance of VEGF in tumorigenesis and tumor progression makes it an attractive target for the development of anticancer therapies. Inhibition of angiogenesis has shown promising clinical efficacy; however, not all patients treated with antiangiogenic agents derive benefit from them. Some patients are predisposed to refractory disease, whereas others develop resistance after initial response. Patients may also have different severity of drug-related adverse events. Optimization of drug administration based on disease status and individual responsiveness is important in limiting the treatment failure and minimization of side-effects. Single nucleotide polymorphisms (SNP) in VEGF may alter VEGF protein concentrations, influence the process of angiogenesis, and may relate to interindividual variation in the risk and progression of selected tumors, and their resistance to treatments. This review examines the role of SNPs in the VEGF gene as predictive and prognostic markers for major solid tumors, including the breast, non-small cell lung, colorectal, and prostate cancers. Selected VEGF SNPs seem to be associated with risk of these cancers; however, there is lack of unanimity in findings, in part influenced by differences in study design and analysis. [Mol Cancer Ther 2009;8(9):2496–508]
Introduction
Antiangiogenic therapies have improved the clinical outcome of selected tumors by inhibiting tumor growth via restriction of new blood vessel formation. As the tumor grows, the lack of sufficient blood supply creates a hypoxic environment, stimulating the release of factors like hypoxia inducible factor (HIF)-1α, which induces angiogenesis by activating the transcription of vascular endothelial growth factor (VEGF; ref. 1). VEGF and its receptors are frequently overexpressed in human tumors, including the breast, non-small cell lung, colorectal, and prostate cancers (2–5). VEGF and associated signaling pathways have been the target for many newly developed anticancer drugs. These agents have shown promising efficacy; however, they are often very expensive, and some patients experience drug resistance, limited activity, and severe toxicities. There is a strong need for identification of markers enabling a priori selection of patients who are likely to benefit from these agents. These markers might also carry prognostic value and may help in decisions about the overall treatment approach to manage more or less aggressive tumors with varying degrees of angiogenic involvement. This review evaluates the role of VEGF single nucleotide polymorphisms (SNP) in predicting the cancer susceptibility, progression, and anti-VEGF therapeutic response in subjects with major solid tumors including breast, non-small cell lung, colorectal, and prostate cancers. For these tumors anti-VEGF or anti-VEGFR treatments are either approved or in advance stages of clinical development. We also did a meta-analysis, using the methods of DerSimonian and Laird (6) to test the homogeneity of odds ratios (OR) across studies, and for calculation of pooled ORs and their confidence intervals (CI). Standard error estimates for individual ORs were calculated using exact methods from published genotype frequencies or were derived from published CIs.
VEGF Gene
The VEGF gene is located on chromosome 6 at location 6p21.1 (7). Its coding region spans approximately 14 kilobases and consists of eight exons (8, 9). Numerous SNPs in the promoter, 5′-, and 3′-untranslated regions (UTR) are present in VEGF. Some of the more frequent SNPs are shown in Fig. 1, and their frequencies in control population are summarized in Table 1. The 5′- and 3′-UTR contain key regulatory elements that are sensitive to hypoxia (10), and contribute to high variability in VEGF production among tissues (11). For example, 634G > C SNP in the 5′-UTR of VEGF affects the protein translation efficiency (12), and 936C > T SNP in the 3′-UTR influences the circulating plasma concentrations (13) and tumor tissue expression of VEGF (14). However, it is likely that only a small number of these polymorphisms and haplotypes (linearly linked SNPs) actually have a functional effect on VEGF translation, whereas others act as proxies (15).
Structure of VEGF gene and position of VEGF SNPs relative to translation start site. Italic positions indicate alternate names. Dashed lines indicate the region consisting of coding sequence (CDS) and seven introns (adapted from ref. 29).
Frequencies of selected VEGF polymorphisms in control subjects
VEGF SNPs in Breast Cancer
VEGF has shown to be important for development, invasiveness, and metastasis of breast cancer in both preclinical and clinical settings (16). Higher levels of VEGF have shown to be related with adverse prognosis, decreased overall survival, and resistance to hormonal therapy (17–19). High microvessel density (MVD), a marker of VEGF expression and local activity, increases the likelihood of metastatic disease (20) and acts as a prognostic indicator for relapse-free and overall survival in node-negative breast cancer patients (21). These data suggest that SNPs in VEGF may play a substantial role in development and progression of breast cancer; studies evaluating this hypothesis are summarized in Table 2.
Summary of case-control studies assessing the VEGF polymorphisms in relation to risk of breast cancer
A case-control study, with 500 subjects in each group, evaluating the role of the germline VEGF 936C > T SNP in Austrian individuals, found that subjects with variant 936T allele were at reduced risk of breast cancer (22). However, the 936C > T SNP was not related to prognostic factors such as tumor size, histological grades, lymph node status, estrogen and progesterone receptor status, or age at diagnosis (22). In a small subset of 21 healthy, nonsmoking postmenopausal women, the 936T allele was associated with significantly lower plasma concentrations of VEGF, which may partially explain the reduced risk of cancer associated with this allele (22). A smaller study from the UK, including 144 cases and 263 controls, evaluated the effect of SNPs in the promoter regions of various cytokines, including the VEGF -1154G > A transition, on risk and prognosis of breast cancer. The -1154G > A allele was not associated with risk of breast cancer; however, carriers of heterozygous -1154AG genotype displayed a trend toward favorable prognosis (23). A larger case-control study including 571 patients with familial breast cancer from Germany and Poland, and 974 patients with sporadic breast cancer from Sweden failed to find a relation between VEGF SNPs (-2578C > A, -1154G > A, -634G > C, and 936C > T) or their haplotypes and risk of breast cancer (24). This finding is in contrast with the earlier report (22) suggesting the protective role of 936T allele. However, patients with the -634CC genotype and -2578C/-634 C haplotype had larger, more histologically advanced tumors, whereas -2578AA and -2578A/-634G had low histological grade tumors, suggesting that these SNPs might act as genetic marker for tumor aggressiveness (24).
Two different publications (25, 26) evaluated data from the Shanghai Breast Cancer Study, a large population-based case-control study to evaluate the impact of germline VEGF -1498C > T, -634G > C, and 936C > T SNPs on survival and risk of breast cancer in Chinese women. On the basis of data from 1,133 cases and 1,233 age-matched controls, the -1498 CC allele was associated with shorter overall and disease-free survival [age-adjusted hazard ratio (HR) of 1.5 for both; ref. 25], but it had no association with risk of breast cancer (26). The carriers of the -634GG allele had significantly shorter overall survival than noncarriers with an age-adjusted HR of 1.6, but the disease-free survival and risk were not different between these groups (25, 26); whereas previously the complementary allele -634CC was shown to be related with advanced tumors (24). The 936C > T SNP was not related to overall or disease-free survival (25), but the 936TT genotype carriers were at reduced risk of cancer (OR = 0.65; ref.26), agreeing with Krippl and colleagues (22) despite differences in ethnicity. The subset analysis based on menopausal status explained that the protective role of the 936TT genotype was restricted to premenopausal women (OR = 0.45). Haplotype analysis showed that -1498T/-634C/936C was related with longer overall survival (HR = 0.57), and premenopausal women with -1498T/-634G/936T were at reduced risk of breast cancer (OR = 0.47) compared with individuals with the reference -1498C/-634G/936T haplotype (25, 26).
A study evaluating the impact of four previously studied (24) VEGF SNPs (-2578C > A, -1154G > A, -634G > C, and 936C > T) in a Caucasian population from a different geographic location (United States), including 501 cases and 504 controls, also failed to find any association between these SNPs and the risk of breast cancer (27), disregarding the spread of disease. However, subgroup analysis in patients with invasive (n = 380) and in situ (n = 107) breast cancer showed higher risk of invasive, but not of in situ, cancer in carriers of -2578CC (OR = 1.46) and -1154GG (1.64) genotypes (27). Conversely, the 936C allele was not associated with invasive cancer, but was related to reduced risk of insitu breast cancer (OR = 0.59; ref. 27). Notably, this finding is in contrast with the majority of previous studies in which patients with the 936T allele had lower risk of breast cancer (22, 26). The -634G > C SNP had no association with risk of either in situ or invasive breast cancer. These results show the possible differences in genetic markers based on stage of disease (i.e., localized or metastatic). Haplotype analysis showed that one or two copies of -2578A/-1154A/-634G resulted in lower risk of invasive breast cancer compared with zero-copy carriers (OR = 0.69 and 0.62, respectively). Assessment of promoter diplotypes suggested no significant association with risk of invasive breast cancer (27). Another case-control study in 500 Caucasian patients and 500 matched controls analyzed the effect of germline VEGF -1498C > T, -634G > C, -7C > T, and 936C > T SNPs on risk and severity of breast cancer (assessed by tumor size, tumor grade, nodal involvement, vascular invasion, and estrogen receptor status), survival of patients, circulating levels of VEGF, and VEGF expression in tumor (28). Singularly, none of these SNPs were related to risk, but patients carrying the -1498T/-634C/-7C/936C haplotype were at lower risk of breast cancer than noncarriers. Except for -7C > T transition, which predicted longer overall survival, no other SNPs were related with survival (28). The serum and plasma VEGF levels and tumor VEGF expression were not related with any of these four SNPs (28). A large case-control study including 804 breast cancer patients and 804 age-matched healthy volunteers from Austria found that germline VEGF -2578C > A, -2489C > T, -1498C > T, -634G > C, -7C > T, 936C > T, and 1612G > A SNPs or their haplotypes were not related with risk of breast cancer (29). However the -634C allele was related with small tumor size (29), contrasting an earlier report (24) showing -634CC allele's relation with high tumor aggressiveness. A small case-control study with 60 patients and 60 age-matched healthy volunteers found no carriers of the 936TT genotype in two groups, and the frequency of the 936CT genotype was significantly higher in patients compared with controls (30). The 936C > T SNP had no association with clinicopathological characteristics of the patients (30).
An epidemiological study with predominantly Caucasian subjects, including 715 controls and 520 breast cancer cases, evaluated the impact of alleles in seven genes, including the VEGF -2578C > A, -1498C > T, -634G > C, -1154G > A, and936C > T SNPs on risk of breast cancer. The -2578AA and -1498CC genotypes had higher risk of breast cancer, independent of nongenetic risk factors estimated by the Gail model (respective adjusted OR = 1.99 and 2.01), whereas other VEGF SNPs had no association with disease risk (31). Finally, a case-control study in Polish Caucasian subjects, including 319 breast cancers, 146 ovarian cancers, and 290 unaffected controls, investigated the effect of the 936C > T SNP in modifying the hereditary risk of breast and ovarian cancer in women harboring a mutated BRCA1 gene (32). Carriers of the VEGF 936 T allele and BRCA1 mutation had reduced risk of breast cancer (OR = 0.63 for CT+TT genotype) but the risk of ovarian cancer was not affected (32). These findings suggest a protective role for the 936T allele in women at risk for breast cancer carrying BRCA1 mutations.
Recently two studies (33, 34) evaluated the relationship of VEGF plasma levels or somatic VEGF and VEGFR-2 SNPs with treatment efficacy or toxicity. One of them compared the effect of paclitaxel with paclitaxel and bevacizumab combination in metastatic breast cancer by enrolling 183 patients in each arm (33). The VEGF -2578AA had longer median overall survival (HR 0.58) compared with the -2578CA+CC genotype in the combination arm (33). Similarly -1154A allele had longer overall survival (HR 0.62) in the combination arm, which increased additively with increase in number of A alleles (33). VEGF -634CC and -1498TT had significantly less likelihood of developing grade 3 or 4 hypertension in the combination arm compared with -634 GC+GG and -1498 CT+CC, respectively (33). The VEGF 936C > T, VEGFR-2 889G > A, and VEGFR-2 1416A > T had no significant relationship with either the efficacy or toxicity. Another study enrolled 56 breast cancer patients to examine the role of plasma VEGF levels as a predictor of treatment outcome with bevacizumab and vinorelbine therapy (34). Lower levels of baseline VEGF predicted longer progression free survival (34).
In conclusion, VEGF -634G > C (24, 26–29, 31), -7C > T (26, 28, 29), and -2489C > A (26, 29) SNPs were not related with risk of breast cancer. VEGF -1154G > A had no relation with overall risk of breast cancer (23, 24, 27, 31), however the G allele was related with increased risk of invasive disease (23, 27, 31). VEGF -1498C > T was shown to have no relation (26, 28, 29) or predicted higher risk of breast cancer (31). Results for VEGF -2578C > A were inconsistent, i.e., no association with risk (23, 24, 27, 29) and higher risk of invasive disease with the C allele (23, 27, 31) or A allele (31). For the 936C > T SNP, three studies showed a protective role of the 936T allele (22, 26, 32), four failed to find any correlation with risk (24, 28, 29, 31), and one found 936 C as a protective allele for in situ breast cancer (27). Differences in geographic location of the studied population did not explain the differences in the role of the 936C > T SNP. Results of meta-analysis (Table 2) show that, except for 936 CC versus (CT or TT), ORs for all other genotype comparisons were homogenous to be combined across studies. Despite a small number of studies that disagree with the others, the pooled analyses for these genotypes seem not to have found any significant associations with breast cancer risk.
VEGF SNPs in Non-Small Cell Lung Cancer
High VEGF mRNA or protein expression in tumor (35) or in serum have shown to be associated with advanced tumor stage (IIIB and IV; ref. 36), shorter survival (37), and prognosis of non-small cell lung cancer (NSCLC; ref. 38). VEGF mRNA and protein expression were found to be significantly higher in adenocarcinomas than in squamous-cell carcinomas (35). These observations suggest that SNPs in the VEGF gene may affect the prognosis or clinical outcome of NSCLC by altering the VEGF levels; studies evaluating this hypothesis are summarized in Table 3.
Summary of case-control studies assessing the VEGF polymorphisms in relation to risk of NSCLC
A small study on DNA samples extracted from tumor and normal lung tissues from 36 surgically resected patients, examined the effect of VEGF -2578C > A, -634G > C, -1154G > A, and 936C > T SNPs on tumor VEGF expression and vascular density (14).
The -2578 CC, -634 GG, and -1154 AA and GA genotypes were associated with low VEGF expression, whereas high VEGF levels were detected in samples carrying the -2578 CA, -634 GC, and -1154 GG genotypes (14). The MVD in tumors bearing the -2578 CC and -634 GG genotypes was significantly lower compared with the -2578 CA and -634 CC+GC genotypes, respectively. VEGF -1154G > A and 936C > T were not related with tumor VEGF expression and MVD (14). A large case-control study on 1,900 NSCLC Caucasian patients and 1,458 healthy controls failed to find a correlation between germline VEGF -1498C > T, -634C > G, and 936C > T SNPs taken singularly on risk of NSCLC after adjusting for age, gender, smoking status, pack-years of smoking, and years since smoking cessation (39). However, male patients bearing the -634CC+CG genotype and -1498T/-634G/936C haplotype, respectively, had higher and reduced risk of developing lung adenocarcinoma (39). The lack of association in female patients was unclear and was not explained by differences in smoking status between genders (39). Another study on 462 patients with early-stage NSCLC treated with surgical resection, examined the impact of germline 936C > T, -1498C > T, and -634G > C SNPs on survival (3). Carriers of -634C and 936T alleles had longer 5-year overall survival than noncarriers, with respective survival of 61% versus 51%, and 67% versus 54% (3). While considering the combined effect of 936C > T and -634G > C SNPs, an increased number of pooled variant alleles resulted in significantly longer survival, with adjusted HR for one, two, and three variant allele(s) being 0.76, 0.61, and 0.61, respectively, compared with the zero variant allele (3). The -1498C > T SNP had no association with overall survival in these patients (3).
In conclusion, male subjects carrying the -634C allele seem to be at increased risk of lung adenocarcinoma (39). Considering both males and females together, -634G > C, 936C > T, and -1498C > T had no significant association with risk of disease (39), but a variant allele for these SNPs predicted longer overall survival (3). The -1498T/-634G/936C haplotype may predict the risk of NSCLC. The -2578C > A, -1154G > A, and -634G > C SNPs were associated with VEGF expression and angiogenesis, but these observations were limited by small sample size (14). There were not enough studies (Table 3) to do a meta-analysis.
VEGF SNPs in Colorectal Cancer
Like other solid tumors, overexpression of VEGF mRNA and protein has shown to be associated with tumor progression and poor prognosis of colon carcinoma (40). VEGF has shown to be an important predictor of prognosis in cases of advanced CRC (41). High MVD and VEGF expression were shown to be related with poor relapse-free and overall survival (42). Therefore, VEGF SNPs may play and important role in determining the risk, prognosis, and survival in colorectal cancer patients; studies evaluating this hypothesis are summarized in Table 4.
Summary of case-control studies assessing the VEGF polymorphisms in relation to risk of colon cancer
A study that analyzed VEGF SNPs in peripheral blood and their impact on prognosis in 125 patients with locally advanced colon cancer found that the VEGF 936CC genotype had a significantly shorter median time to relapse compared with CT or TT genotypes (2.6 versus 11.1 years; ref. 40). A case-control study on the VEGF 936C > T germline SNP in Korean colon cancer patients (n = 262) and healthy controls (n = 229) showed that the frequency of T allele-bearing genotypes (CT+TT) was higher in cases than in controls. When stratified by gender and age, the frequency of CT+TT was significantly associated with increased risk for colon cancer in women and patients younger than 55 years old, thus suggesting that the VEGF 936C > T SNP might be a genetic determinant for colon cancer in Korean patients (43). Another case-control study on the VEGF -2578C > A germline SNP was carried out in a cohort of 246 patients with colon cancer and 203 healthy controls and found that VEGF -2578C > A SNP had no association with susceptibility to colon cancer, although gender stratification indicated that the -2578CA and AA genotypes had a marginal protective effect for colon cancer in women. Therefore, the VEGF -2578C > A SNP, at least in Koreans, seems to have a borderline role in colon cancer risk (44). The analysis of two VEGF SNPs (-634G > C and 936C > T) in colorectal tissue from 465 patients and peripheral blood lymphocytes from 413 healthy controls showed that the -634 GC and CC genotypes were associated with a decreased risk of cancer, indicating a dominant role of the C allele, whereas the 936TT genotype correlated with advanced stage, high serum levels of CA19-9, and higher histologic grades (45). Haplotype analysis showed that -634C/936C and -634G/936T were associated with decreased susceptibility of colorectal cancer (45). The VEGF -1498C > T, -634G > C, and -7C > T SNPs were determined in 36 Japanese patients and showed that the -1498CC allele, but not the -634G > C and -7C > T SNPs, was predictive of poorly differentiated tumors and thereby poor prognosis (46). A larger study on 445 patients with colorectal cancer examined the VEGF -2578C > A, -634G > C, and 936C > T SNPs and showed that the survival for patients with the -634GC or CC genotypes was better than for those with the -634GG genotype (47). Furthermore, the 936CT or TT genotypes were associated with worse survival compared with the 936CC genotype, and -2578C > A was not related with survival (47). Finally, in haplotype analysis, the -2578A/-634G/936T exhibited a significantly less favorable survival when compared with the wild-type -2578C/-634G/936C haplotype (47). A case-control study examined VEGF 936C > T, -2578C > A, and -634G > C SNPs in 427 patients and 427 age- and sex-matched healthy control subjects and found no correlation between all these variants and tumor size, histological grading, positive regional lymph node metastases, or tumor stage (48). Patients treated with cetuximab were genotyped to assess the possible predictive role of SNPs of cyclin D1, cyclooxygenase 2, EGF, EGFR, IL-8, and VEGF genes in blood samples. Although a significant association was found between cyclin D1 870A > G and the EGF 61A > G SNPs and survival, VEGF 936C > T showed no association (49). Finally, a study on RNA from 25 colorectal cancer tissues showed the presence of VEGF121, VEGF145, VEGF165, and VEGF189 splice variants and an array of novel mutations within the conserved expression site of the gene. Five samples exhibited single nucleotide changes and in one sample a two-nucleotide deletion was detected. In addition to this, a one- and two-base deletion with frameshift and protein truncation in exon 3 at positions +172 and +171/172 were shown, respectively, as well as a transition mutation in exon 3 at position +248, and two transition mutations in exon 4 at positions +398 and +403. The biologic and clinical impact of these variants were not shown, however (50).
In conclusion, 936 T allele has shown to be associated with increased risk (43), advanced stage of disease (45), worse survival (47), and longer time to relapse (40), whereas other studies have shown no correlation with tumor size, grade, and stage (48), and no relation with survival (49) in patients with colorectal cancer. The -634 C allele was predictive of decreased risk (45) and better survival (47), and was not related with tumor size, grade, and stage (48). The VEGF -2578C > A SNP was not related with risk (44), survival (47), tumor size and stage (48), but -2578 CA and AA had a borderline protective role in Korean women (44). The results of the meta-analysis shown in Table 4 suggest that ORs for several genotypic comparisons were not homogeneous to be pooled across studies (P-value for test of homogeneity, < 0.05). The 95% CI for all the genotypic comparisons for which ORs were homogeneous and could be pooled together included one, suggesting no significant correlation of selected VEGF SNPs with risk of colorectal cancer.
VEGF SNPs in Prostate Cancer
Like some solid tumors, progressive growth and metastasis of prostate cancer is also dependent on angiogenesis. VEGF expression and vascular density were significantly higher in prostate cancer tissue compared with benign prostate hypertrophy (BPH) or normal prostate tissue (5, 51). Metastatic prostate cells had more enhanced VEGF production and tumor vascularity than cells with low metastatic potential (52), suggesting the importance of VEGF in prostate cancer metastasis. Studies evaluating the role of VEGF SNPs in prostate cancer susceptibility, progression, and metastasis are summarized in Table 5.
Summary of case-control studies assessing the VEGF polymorphisms in relation to risk of prostate cancer
A case-control study on 247 cancer patients and 263 controls from the UK examined SNPs in five cytokines, including VEGF -1154 G > A, and showed that -1154 AA genotype had a reduced risk of prostate cancer (OR = 0.45; ref. 53). Another small case-control study on the VEGF -1498C > T germline SNP in 96 prostate cancer patients and 119 controls of Taiwanese descent found that carriers of -1498T allele were at 2.3-times higher risk of prostate cancer than noncarriers (54). However, this SNP was not related with prognosis (i.e., clinical stage and pathological tumor grade) and response to hormonal therapy (54). A case-control study in Tunisian subjects evaluated the influence of germline -1154G > A, -634G > C, and 936C > T VEGF SNPs on risk and prognosis of prostate cancer in 101 patients and 100 age-matched healthy controls (55). Carriers of at least one -1154 A allele were at reduced risk (OR = 0.42) and lower susceptibility to high-grade (OR = 0.25) and advanced prostate cancer (OR = 0.37). The OR for -1154 GA and AA genotypes were 0.46 and 0.27, respectively (55), agreeing with McCarron and colleagues (53), despite differences in ethnicity. The VEGF -634 GC+CC genotypes were at increased risk of prostate cancer (OR = 1.95), and the -634GC and CC genotypes were related with high-grade tumors (OR = 3.83 and 4.89, respectively; ref. 55). Haplotype analysis showed that -1154A/-634G had reduced disease risk (OR = 0.48), high-grade tumor (OR = 0.22), and advanced disease (OR = 0.45) compared with -1154G/-634G (55). No correlation was observed between 936C > T SNP and risk or aggressiveness of tumor (55). These results suggest that both VEGF -1154G > A and -634G > C SNPs may act as an important genetic marker of risk and aggressiveness of prostate cancer in Tunisian subjects. A case-control study in 133 sporadic prostate cancer patients and 157 healthy controls of Turkish descent evaluated the germline -1498C > T SNP and found no relation with susceptibility to disease (56), which was contrary to the findings for Taiwanese subjects (54). Further, the -1498C > T SNP was not related with the tumor-node-metastasis (TNM) staging, Gleason scores, and serum prostate specific antigen (PSA) levels of patients (56). The impact of the same germline VEGF -1498C > T SNP in Japanese population was examined in a case-control study enrolling 270 patients and 252 controls. Similar to the Turkish population (56), the -1498C > T SNP was not related with disease risk, tumor grade and stage, Gleason scores, VEGF expression, MVD, and age at diagnosis (51). The rate of PSA relapse after radical prostatectomy was higher in patients with the 1498CC or CT genotype (51), but the TT genotype had significantly worse cancer-specific and overall survival (51).
Another study in Tunisian subjects determined the combined effect of four germline SNPs that are involved in the process of angiogenesis, i.e., VEGF -1154G > A, VEGF -634G > C, MMP9 -1562C > T, and TSP1 -8831A > G, in predicting the risk of prostate cancer by enrolling 101 patients and 74 controls (57). The VEGF -1154AA, VEGF -634GG, MMP9 -1562CC, and TSP1 -8831AA genotypes were considered as low-risk genotypes and were used as reference, and the remaining genotypes for each SNP were considered as high-risk genotypes (57). Patients with one, two, or three high-risk genotypes in the VEGF -1154/VEGF -634/MMP9 group had an increasingly high risk of prostate cancer (OR= 2.79, 4.57, and 7.11, respectively), compared with the zero high-risk genotype group (57). In the VEGF-1154/634/TSP1 group, patients carrying three high-risk genotypes had six times higher risk of prostate cancer (OR = 6.0) and were 20times more prone to have high-grade tumors when compared with reference groups (OR = 20.75; 57). Overall, their findings show that the gene-gene interactions between high-risk angiogenic polymorphisms may increase the risk of prostate cancer and tumor aggressiveness (57). Finally, a case-control study in 702 patients and 702 age-matched healthy controls from Austria analyzed seven VEGF SNPs (-2578C > A, -2489C > T, -1498C > T, -634G > C, -7C > T, 936C > T, and 1612G > A) and their haplotypes, and failed to find any relation between these SNPs and the risk of prostate cancer or VEGF plasma levels (58).
To conclude, studies from Austria, Japan, Taiwan, Tunisia, Turkey, and the UK evaluated the importance of VEGF SNPs in prostate cancer. The VEGF -1154 A allele seems to predict the reduced risk (53, 55), which was consistent among studies as shown by meta-analysis (Table 5), in which the pooled OR (95% CI) for -1154 AA genotype was 0.40 (0.22–0.72). The VEGF -1154A allele also imparted protection against high-grade tumor and tumor aggressiveness (55). Except for Taiwanese subjects, in which the -1498T allele predicted high risk of prostate cancer (54), the -1498C > T SNP was not related with risk (51, 56, 58) or disease prognosis (i.e., with tumor grade, stage, Gleason scores, VEGF expression, and MVD; ref. 51, 56, 58). The -634C allele was predictive of increased risk and high tumor grades (55) in Tunisian subjects but had no relation with risk in Austrian subjects (58). The results of meta-analysis showed no significant association between any other VEGF polymorphism and prostate cancer risk or clinical outcome.
Conclusions
Many studies have investigated the role of VEGF polymorphisms as a genetic determinant for susceptibility and outcome of breast, prostate, NSCL, and colorectal cancer. Several polymorphisms have been reported within the promoter (-2578C > A, -2489C > T, -1498C > T, and -1154G > A), 5′-UTR (-634G > C and -7C > T), and 3′-UTR (936C > T and 1612G > A) for the VEGF gene. The variant allele for -1154G > A and 936C > T results in lower VEGF expression, whereas the variant allele for -1498C > T and -7C > T results in increased concentrations of VEGF mRNA (15, 59). The functional role of -2578C > A and -634G > C is not in agreement among studies, which report low or high VEGF production for variant alleles (15). Results of association studies of these polymorphisms with risk and outcome in breast, prostate, NSCL, and colorectal cancer are also variable (Tables 2-5, Supplementary Table 6). For example the -936T allele has been shown to impart protection against breast cancer (22, 26, 32), but conversely, increases the risk of colorectal cancer (43). Similarly, the -634C allele was predictive of higher risk of NSCLC (39) and prostate cancer (55), but predicts reduced risk for colon cancer (45), and has no relation with risk of breast cancer (24, 26–29, 31). The lack of consensus about the role of these SNPs could be because of their linkage with other unknown functional SNPs in the VEGF gene or unknown SNPs in other factors of the angiogenesis pathway. Perhaps, association with haplotypes will be more predictive than individual SNPs. Increased understanding of the functionality of these polymorphisms and their role in tumor biology is essential to discern the reasons for inconsistencies in the literature (60), which may also help in designing association studies in the future. Other possible reasons are differences in study population (i.e., ethnicity and tumor type), small sample size (insufficient power), and failure to correct for multiple testing when studying the correlation with more than one SNP, which is important to avoid chance correlations. However, the likelihood of chance correlations should be assessed with consideration to the biological plausibility of any observed correlation (60).
Independently, several of these studies reported an association between VEGF SNPs and the risk of cancer (Tables 2-5); however, the findings from these studies seem to be inconsistent. We did meta-analysis to look at the combined results from studies addressing the same hypothesis. With the exception of VEGF -1154G > A, no other VEGF SNP was found to be related with a risk of studied tumors. The VEGF -1154A allele was at a significantly reduced risk of prostate cancer (OR = 0.40; 95% CI 0.22–0.72). Haplotype -2578A/-1154A/-634G and -1154A/-634G predicted a reduced risk of breast (26) and prostate cancer (55).
The VEGF-mediated angiogenesis pathway involves a complex signaling network regulated by multiple factors and it is likely that final outcome is governed by interaction and cross-talk between these signaling molecules and/or receptors rather than the VEGF gene alone. In that case, a pathway approach, which includes combined analysis of variations in different genes in the angiogenesis pathway, may provide better predictive markers. For example carriers of three high risk genotypes, i.e., VEGF -1154AA/VEGF -634GG/MMP9CC, had a 7.11-times higher risk of prostate cancer than the VEGF -1154GG/VEGF -634CC/MMP9TT group (57). Genome-wide association (GWA) studies, done to identify the candidate genes and chromosomal sites linked to cancer, showed that the locus including the VEGF gene was not associated with a risk of breast, NSCL, colon, and prostate cancer (61–71). However, our meta-analysis suggests that VEGF SNPs predict the risk of prostate cancer, indicating their role in vascular growth, at the minimum in prostate cancer; although they may not be as strongly predictive (i.e., penetrant) as other alleles in the genome. It should also be noted that there are other limitations to the GWA approach, such as multiple testing in small populations and low-density genotyping (72, 73), which may limit the identification of true susceptibility genes.
The majority of the published studies assessing the role of VEGF polymorphisms compare the cases with controls to identify the disease markers. There is a critical need to conduct more studies evaluating the relation of genotype with drug concentrations and response. These studies are equally important and helpful for identification of genetic markers to guide the choice of treatment and supportive care specific to each individual.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
Footnotes
Grant support: Intramural Research Program of the National Cancer Institute (NCI), National Institutes of Health (NIH), U.S. Department of Health and Human Services.
Note: Supplementary material for this article is available at Molecular Cancer Therapeutics Online (http://mct.aacrjournals.org/).
- Received April 1, 2009.
- Revision received July 10, 2009.
- Accepted July 21, 2009.
- © 2009 American Association for Cancer Research.