HIF-1α 1772 C/T and 1790 G/A Polymorphisms Are Significantly Associated with Higher Cancer Risk: An Updated Meta-Analysis from 34 Case-Control Studies

Background HIF-1 activates various genes in cancer progression and metastasis. HIF-1α 1772 C/T and 1790 G/A polymorphisms are reportedly associated with cancer risk; however, the results are inconclusive. Methodology/Principal Findings A meta-analysis of 34 studies that involved 7522 cases and 9847 controls for 1772 C/T and 24 studies that involved 4884 cases and 8154 controls for 1790 G/A was conducted to identify the association of C/T and G/A polymorphisms with cancer risk. Odds ratio (OR) and 95% confidence intervals (95% CI) were used to assess the strength of association. HIF-1α 1772 C/T and 1790 G/A polymorphisms were associated with higher cancer risk in homozygote comparison (1772C/T: TT vs. CC: OR = 2.45, 95% CI: 1.52, 3.96; P heterogeneity = 0.028; 1790G/A: AA vs. GG: OR=4.74, 95% CI: 1.78, 12.6; P heterogeneity < 0.01), dominant model (1772C/T: TT/CT vs. CC: OR = 1.27, 95% CI: 1.04, 1.55; P heterogeneity < 0.01, 1790G/A: AA/GA vs. GG: OR = 1.65, 95% CI: 1.05, 2.60; P heterogeneity < 0.01), T allele versus C allele (T vs. C: OR = 1.42, 95% CI: 1.18, 1.70; P heterogeneity < 0.01), and A allele versus G allele (A vs. G: OR = 1.83, 95% CI: 1.13, 2.96; P heterogeneity < 0.01). On a subgroup analysis, the 1772 C/T polymorphism was significantly linked to higher risks for breast cancer, lung cancer, prostate cancer, and cervical cancer, whereas the 1790 G/A polymorphism was significantly linked to higher risks for lung cancer and prostate cancer. A significantly increased cancer risk was found in both Asians and Caucasians for 1772C/T polymorphism, whereas a significantly increased cancer risk was found in Caucasians in the heterozygote comparison and recessive model for 1790G/A polymorphism. Conclusions HIF-1α 1772 C/T and 1790 G/A polymorphisms are significantly associated with higher cancer risk.


INTRODUCTION
Cancer, which results from complex interactions between genetic and environmental factors, has become a challenging health problem. An increasing number of studies have been performed in the past few years to assess the relationship between genetic variation and cancer risk [1].
Oxygen (O 2 ) concentration in tumor tissues is significantly lower than that in the surrounding normal tissues. Many studies have focused on hypoxia because of its function in maintaining tumor microenvironments [2]. Hypoxic tumor microenvironment initiates multiple cellular responses, such as proliferation and angiogenesis, triggering the development and progression of cancer. In general, hypoxia may regulate tumor cell phenotypes by altering genes that are sensitive to O 2 pressure [3]. Studies have demonstrated that HIF-1 has an important function in the development and progression of cancer by activating various genes associated with angiogenesis, cell adhesion, erythropoiesis, and glucose transportation [4]. HIF-1 is a heterodimer consisting of an oxygen-sensitive subunit HIF-1α and a constitutively expressed subunit HIF-1β; it is degraded rapidly through the von Hippel-Lindau-mediated ubiquitin-proteasome pathway under normoxia conditions [5]. Recent studies have shown that HIF-1α is overexpressed in many human cancers with advanced tumor grade, suggesting that HIF-1α acts as an independent factor of cancer prognosis [6].
The human HIF-1α gene, which is located at chromosome 14q21-24, is composed of 15 exons. It codes for a 3919 bp cDNA and produces an 826 amino acid protein. Single nucleotide polymorphisms (SNPs) in coding regions can mediate amino acid changes and affect the structure and biological activity of the translated protein [7]. The most widely studied HIF-1α polymorphisms are 1772 C/T (Pro582Ser, rs11549465) and 1790 G/A (Ala588Thr, G1790A, rs11549467), which induce proline-to-serine and alanine-to-threonine amino acid substitutions, respectively. Both polymorphic variants can significantly elevate transcriptional activity than the wild type under both hypoxic and normoxic conditions in in vitro studies [8]. Moreover, both polymorphisms are associated with increased tumor microvessel density, thereby contributing to the development and progression of cancer [9].
HIF-1α 1772 C/T and 1790 G/A genetic polymorphisms were previously suggested to be responsible for the risk for various types of cancer. However, the results of epidemiological studies are inconsistent [10][11][12]. Thus, the relationship between HIF-1α polymorphisms and cancers requires further investigation. Accordingly, we performed a meta-analysis on eligible case-control studies to produce a more powerful estimation of the association of HIF-1α 1772 C/T and 1790 G/A polymorphisms with cancer risk.

Identification and eligibility of relevant studies
All studies published before June 26, 2013 that investigated the association of HIF-1α 1772 C/T and 1790 G/A polymorphisms with cancer risk were considered in this metaanalysis. A systematic search of literature was carried out using PubMed and Embase. The keywords used for the search were "hypoxia-inducible factor-1" or "HIF-1" concatenated with "SNP," "polymorphism," "mutation," or "variant" and "tumor," "cancer," "carcinoma," or "malignancy." Only studies with complete data on the comparison of HIF-1α 1772 C/T or 1790 G/A polymorphisms between cancer patients and controls were selected. Case reports, animal studies, review articles, editorials, abstracts, reports with incomplete data, and studies based on pedigree data were excluded.

Data extraction
Two investigators (Yang and Zhu) independently reviewed the articles to exclude irrelevant and overlapping studies. The results were compared, and disagreements were resolved by discussion and consensus. We only included the publication that reported the most extensive information when overlapping articles were found. The following data were extracted for each study: first author, year of publication, country, ethnicity, control source, cancer type, Hardy-Weinberg equilibrium, and the number of cases and controls for each genotype.

Statistical analysis
STATA (version 11.0; StataCorp, College Station, Texas, USA) was used for the meta-analysis. All genotype models for the two HIF-1α polymorphisms were evaluated. We also conducted subgroup analyses by cancer type, ethnicity, and source of control. For cancer type subgroups, we included the subgroup that contained more than three studies.
The existence of heterogeneity between studies was ascertained by Q-statistic. The pooled odds ratio (OR) was estimated with models based on fixed-effects or random-effects assumptions. A random-effects model was used when the significant Q statistic (P < 0.1) indicated the presence of heterogeneity in the studies. Otherwise, a fixed-effects model was selected. The 95% confidence interval (CI) of OR was also calculated. The distribution of genotypes in the controls was checked for Hardy-Weinberg equilibrium. Studies with controls not in the Hardy-Weinberg equilibrium were subjected to sensitivity analysis.
The publication bias among the studies was assayed. Funnel plots of the HIF-1α 1772 C/T polymorphism for T allele versus C allele and the HIF-1α 1790 G/A polymorphism for A allele versus G allele were built to search for any evidence of publication bias. An assymetric funnel plot is indicative of publication bias, whereas a symmetric funnel plot implies the absence of publication bias. Egger's test, estimated by MIX 1.7 software (Kitasato Clinical Research Center, Kitasato University, Japan), was performed to measure funnel plot asymmetry.

Summary statistics
The meta-analysis for the HIF-1α 1772 C/T polymorphism included 7522 cases and 9847 controls. The prevalence of the CC genotype was the highest, allele C was the most frequent, and the prevalence of the TT genotype was the lowest in both case and control groups.
The meta-analysis for the HIF-1α 1790 G/A polymorphism included 4884 cancer cases and 8154 controls. The prevalence of the GG genotype was the highest, allele G was the most frequent, and the prevalence of the AA genotype was the lowest in both case and control groups.

Overall analysis
Upon pooling of all eligible studies, we observed that both 1772 C/T and 1790 G/A polymorphisms were significantly associated with cancer risk in homozygote comparison (1772C/T: TT vs. CC: OR=2. 45

Subgroup analyses
Subgroup analyses were performed to investigate the effect of cancer type, ethnicity, and source of control. For cancer type, the 1772C/T polymorphism demonstrated an increased risk for breast cancer, lung cancer, prostate cancer, cervical cancer, and other cancers in various models. In the subgroup analyses of "oral cancer" and "colorectal cancer," we did not find any significant association between the 1772C/T polymorphism and cancer risk. The 1790G/A polymorphism exhibited an increased cancer risk for lung cancer in the homozygote and recessive models (AA vs. GG: OR = 5.42, 95% CI: 2.75, 10.7; P heterogeneity = 0.866; AA vs. GG/GA: OR = 7.11, 95% CI: 3.61, 14.0; P heterogeneity = 0.975; A vs. G: OR = 1.48, 95% CI: 1.09, 2.00; P heterogeneity = 0.575) and for prostate cancer (A vs. G: OR = 1.45, 95% CI: 1.00, 2.11; P heterogeneity = 0.330). We found a significant association between 1772C/T and 1790G/A polymorphisms and cancer risk in both population-based and hospital-based studies. However, ethnicity significantly affected cancer susceptibility. For the 1772C/T polymorphism, a significantly increased cancer risk was found in both Asians and Caucasians. For the 1790G/A polymorphism, a significantly increased cancer risk was found in Caucasians in the heterozygote comparison (AA vs. GG: OR = 17.4, 95% CI: 4.01, 75.3; P heterogeneity < 0.01) and recessive model (GG/GA: OR = 15.8, 95% CI: 3.42, 72.9; P heterogeneity < 0.01). However, no significant association between these polymorphisms and cancer risk was found in Asians. These results revealed that the effect of HIF-1α polymorphisms on cancer was associated with ethnicity.

Sensitivity analysis
Sensitivity analysis was performed to explore the influence of an individual study on the pooled results by deleting a single study each time from the pooled analysis. The results showed that no individual study significantly affected the pooled OR because no substantial change was found (figure not shown).

Publication bias
Publication bias was assessed by Begg's funnel plot and Egger's test. Begg's funnel plot for the 1772 C/T polymorphism is shown in Figure 6 (P = 0.589 for T allele vs. C allele). Egger's test was performed for statistical analysis, and no publication bias was detected (P =0.481 for T allele vs. C allele). The results of Begg's and Egger's tests for the 1790 G/A polymorphism were P = 0.785 and P = 0.870, respectively, for A allele versus G allele (Figure 7). Overall, no publication bias was detected in the data.

Discussion
HIF-1 has an important function in cancer progression and metastasis by activating various genes that are linked to the regulation of angiogenesis, cell survival, and energy metabolism [3]. The presence of T and A variant alleles of HIF-1α 1772, namely, C/T and 1790 G/A polymorphisms, are associated with high transcriptional abilities and protein synthesis in vitro [8]. In vivo studies related these genetic variations to many aggressive clinical features of cancer, such as the ulcerative growth pattern in colorectal tumors, suggesting that HIF-1α polymorphism is associated with cancer [45]. However, studies on the association of HIF-1α 1772 C/T and 1790 G/A polymorphisms with cancer are conflicting. In 2009, Zhao [10] conducted a meta-analysis using 16 casecontrol studies and concluded that 1772 C/T is significantly associated with higher cancer risk and that 1790 G/A is only significantly associated with breast cancer. Liu [12] performed a similar meta-analysis from 22 case-control studies, including 5552 cases and 8044 controls for 1772 C/T and 3381 cases and 5830 controls for 1790 G/A, and one study evaluated cancer prognosis by polymorphism [45]. This previous study concluded that the 1790 G/A polymorphism and not the 1772 C/T polymorphism is significantly associated with cancer risk. In the present study, we performed an updated meta-analysis from 34 case-control studies that involved 7522 cases and 9847 controls for 1772 C/T polymorphism and 4884 cases and 8154 controls for 1790 G/A polymorphism.
In the present meta-analysis, we investigated the association of HIF-1α 1772 C/T and 1790 G/A polymorphisms with cancer risk. Subgroup analyses by cancer type and ethnicity were also performed. Our analyses showed that both 1772C/T and 1790G/A polymorphisms were significantly associated with cancer risk. In the subgroup study, various types of cancers, such as breast cancer, lung cancer, prostate cancer, and cervical cancer, were associated with 1772C/T, whereas only lung cancer was linked with 1790G/A. However, the odds ratio values in some of the subgroup analyses were large and lacked statistical power because of the significant heterogeneity. Ethnicity may also significantly affect cancer susceptibility. For the 1790G/A polymorphism, we did not find any association between the 1790G/A polymorphism and cancer risk in Asians. This finding can be explained by the difference in genetic background, environmental exposure, and risk factors relating to lifestyle between Asian and Caucasian populations.
Some limitations of this meta-analysis should be addressed. First, the lack of the the detailed information about environment risk factors for cancer risk from included studies limited our further evaluation of potential gene-gene and geneenvironment interactions. Second, the P value of the Hardy-Weinberg equilibrium of three included studies was less than 0.05, suggesting that these study populations were not representative of the broader target population. Despite these limitations, our meta-analysis had some strong advantages.
This meta-analysis shed light on the association between HIF-1α polymorphisms and increased risk for various cancers.  In addition, the quality of the included studies was satisfactory and met our inclusion criterion. Moreover, substantial numbers of cases and controls were pooled from different studies, which significantly increased the statistical power of the analysis. No publication bias was also found in the collected data.   In summary, this meta-analysis provided insights into the association of HIF-1α 1772 C/T and 1790 G/A gene polymorphisms with cancer risk, supporting the hypothesis that HIF-1α polymorphisms are a susceptibility marker of cancer. However, large sample studies are warranted to validate our findings, especially in some types of cancer, such as breast cancer and cervical cancer. More studies on gene-gene and gene-environment interactions should also be considered in the future to obtain a more comprehensive understanding of the association between HIF-1α polymorphisms and cancer risk.