A meta-analysis of hypoxia inducible factor 1-alpha (HIF1A) gene polymorphisms: association with cancers

Background Hypoxia inducible factor 1-alpha (HIF1A) is a transcription factor that plays important role in regulating cascade of reactions. In this study, the effect of rs11549465 (1772 C/T) and rs11549467 (1790 G/A) polymorphisms of HIF1A gene and its association with cancers were investigated through meta-analysis. Methods Meta-analysis of genome wide association studies of HIF1A 1772 C/T polymorphism were conducted on 22 case-control studies of sample size 19024 and for 1790 G/A polymorphism 19 case-control studies were included with sample size 10654. Genotype and allelic frequency compared between cases and controls together with further subgroup analyses were carried out by cancer type and ethnicity. Results Meta-analysis from this study indicated that HIF1A 1772 C/T polymorphism is significantly associated with overall cancer risk. T allele and genotype TT are significantly associated with increasing overall cancer risk; odds ratios (OR) dominant model [TT + CT vs. CC: OR 1.30, 95 % CI (1.06-1.59), p-value: 0.0115], and T allele vs. C allele: OR 1.32, 95 % CI (1.07-1.63), p-value: 0.0098. Also, HIF1A 1790 G/A polymorphism, analyses showed that A allele and genotype AA are significantly associated with increasing overall cancer risk; odds ratios (OR) homozygote comparison [AA vs. GG: OR 5.10, 95 % CI (3.12-8.33), p-value: <0.0001], heterozygote comparison [GA vs. GG: OR 1.74, 95 % CI (1.20-2.52), p-value: 0.0033], dominant model [AA + GA vs. GG: OR 1.82, 95 % CI (1.26-2.62), p-value: 0.0014], recessive model [AA vs. GA + GG: OR 3.79, 95 % CI (2.34-6.15), p-value: <0.0001] and A allele vs. G allele: OR 1.82, 95 % CI (1.31-2.52), p-value: 0.0003. Conclusion In detail meta-analysis indicated that both the polymorphisms 1772 C/T and 1790 G/A are significantly associated with overall cancer risk. The subgroup analyses showed that lung cancer is significantly associated with both polymorphisms. Although the 1772 C/T polymorphism is significantly associated with decreasing risk of renal cell carcinoma but the 1790 G/A polymorphism has shown to significantly increase the cancer risk in both Caucasian and Asian population. Thus, HIF1A could be a useful prognostic marker for cancers early predisposition. Electronic supplementary material The online version of this article (doi:10.1186/s40364-015-0054-z) contains supplementary material, which is available to authorized users.


Background
Cancer is the second leading cause of morbidity and mortality worldwide [1]. One major feature of cancer is uncontrolled cell proliferation, which can then invade adjacent parts of the body and spread to other organs, the latter process is referred as metastases, which are the major cause of death from cancer [2]. The most common causes of cancer deaths are due to cancers of the: lung (1.59 million deaths), liver (745,000 deaths), stomach (723,000 deaths), colorectal (694,000 deaths), breast (521,000 deaths) and esophageal (400,000 deaths) [1,2]. Alongside, metabolic alterations and tumor hypoxia have consistently been identified as classical features with aggressive malignancy [3,4]. Hypoxia regulates tumor cell phenotype mainly by altering genes that are sensitive to oxygen pressure [5]. However, the exact mechanism of carcinogenesis is yet to be elucidated. In recent years, an increasing number of studies have focused on understanding the relationship between genetic factors and cancer risk [3,4]. Through the years, it has become well accepted that single nucleotide polymorphisms (SNPs) are the most common and effective type of genetic variations studied in association with disease susceptibility and are the markers of many complex diseases [6].
Hypoxia inducible factor 1α (HIF1A), is a transcription factor that has major impacts in the process of development and progression of cancers [7]. HIF1A regulates the expression of over 100 genes that control the major cellular functions including apoptosis, cell proliferation, glucose metabolism, erythropoiesis, iron metabolism and angiogenesis. It is a master regulator of oxygen homeostasis [7]. In the scientific community, HIF1A has been a research focus and a number of SNPs within HIF1A gene have been identified in association with cancers, with the most widely studied polymorphisms are C1772T (rs11549465) and G1790A (rs11549467) polymorphisms . These two SNPs are located within the same domain (ODD/ pVHL) in exon 12 of the HIF1A gene [8,9]. Recently a meta-analysis has revealed that C1772T is not in substantial linkage disequilibrium (LD) with G1790A [38]. A number of studies have suggested that these two nonsynonymous mutations might alter the transcriptional activity of HIF1A gene by causing structural changes with varied stability, which in turn, might influence the downstream target genes expression and regulation [8,9,38]. In the recent years, a good number of studies have investigated the impact of HIF1A polymorphisms on cancer risk in different populations; however reported results varied across studies and remain inconclusive . In this study, the effect of rs11549465 (1772 C/T) and rs11549467 (1790 G/A) polymorphisms of HIF1A gene and its association with cancers were investigated systematically through metaanalysis.

Eligibility criteria
Two authors independently investigated titles and abstracts of all the articles. Irrelevant and incompatible studies were excluded primarily. For final review, criteria's for further study elimination were: if (1) the study population was not defined completely; (2) it is not a case-control study; (3) not a genome wide association study; (4) incomplete information of allele frequency; and (5) the year of study conducted was not specified. Also, reviews, editorials, meta-analysis and non-human researches were excluded. Only case-control studies, genome wide association study (GWAS) and human researches were considered for the final review. Further, the references of the selected studies were screened carefully for incorporation of additional relevant studies. Only English language articles were considered for this study. Discrepancies and difficulties were discussed with corresponding authors where necessary. Following information were extracted from each study: (1) authors name, (2) year of study, (3) ethnicity of the study subjects, (4) cancer type and (5) allelic frequency (Fig. 1).

Meta-analysis
For HIF1A 1772 C/T polymorphism 22 case-control studies were included of sample size 19024 and for 1790 G/A polymorphism 19 case-control studies were included with sample size 10654. The meta-analysis was prepared in accordance with PRISMA statement [39].

Statistical analysis
Meta-analysis of genome wide association studies (GWAS) of HIF1A were conducted for two polymorphisms, 1772 C/T and 1790 G/A using odds ratios (ORs). A slightly amended estimator of OR was used to avoid the computation of reciprocal of zeros among observed values in the calculation of the original OR [40]. Pooled ORs with 95 % CIs were calculated using random effects model (REM) incorporating the inverse variance weighted method [41]. Heterogeneity among studies was assessed using the Q statistic [42] and quantified using I^2 index [43]. Subgroup analyses were carried out by cancer type and ethnicity. The Hardy Weinberg Equilibrium (HWE) test was performed for the controls of each study. The studies with control not in HWE were supervised for sensitivity analysis. Publication bias was assessed visually by conventionally constructed funnel plot where the inverse of the standard error (1/se) of the effect estimates were plotted against the logarithm transformation of Odds Ratios [log(OR)] [44]. Furthermore, Egger's test was performed to provide quantitative evidence of publication bias [45]. "Gap: Genetic analysis package" was used to perform the Hardy Weinberg Equilibrium (HWE) test [46,47]. All analyses were conducted using "meta" package in R environment [46].

Summary measures
Odds Ratios (OR) with a 95 % confidence interval (CI) were calculated to evaluate the genotype contrasts.

Study characteristics
In the meta-analysis of the HIF1A 1772 C/T polymorphism, ten different types of cancers consisted of 22 studies with 8149 cancer cases and 10,875 controls were included. The types of cancer included in these studies were prostate cancer, colorectal cancer, renal cell carcinoma, breast cancer, lung cancer, oral squamous cell carcinoma (OSCC), head-neck cancer, cervical cancer, bladder carcinoma and pancreatic cancer. For the following cancer types: head-neck, cervical, bladder and pancreatic only one study of each were found for the final review. So, these cancer types with single studies were incorporated in subgroup analysis as Other Cancers (Table 1).
For the meta-analysis of HIF1A 1790 G/A polymorphism, 19 studies with eleven different cancer types consisted of 4681 cancer cases and 5973 controls were included. The cancer types associated with this polymorphism were: renal cancer, prostate cancer, breast cancer, lung cancer, oral squamous cell carcinoma (OSCC), head-neck cancer, gastric cancer, hepatocellular carcinoma, lymph node metastasis, pancreatic cancer and colorectal cancer. For final review, only one study of each of the following cancer types was found: head-neck cancer, gastric cancer, hepatocellular carcinoma, lymph node metastasis, pancreatic cancer and colorectal cancer. These cancer types with single studies were incorporated in subgroup analysis as Other Cancers ( Table 2).

Association of the HIF1A 1772 C/T polymorphism with cancer risk
The pooled ORs for overall cancer suggested that the HIF1A 1772 C/T polymorphism was significantly associated with increasing cancer risk for the dominant model [ (Fig. 2).

Subgroup analyses performed by cancer type
The subgroup analyses of prostate cancer, colorectal cancer, breast cancer and oral squamous-cell carcinoma suggested no significant association of the HIF1A 1772 C/T polymorphism. However, the subgroup analyses of renal cell carcinoma suggested that the HIF1A 1772 C/T polymorphism is significantly associated with lowering renal cell carcinoma risk in homozygote comparison [ Table 3).

Subgroup analyses by ethnicity group
The analyses data for the HIF1A 1772 C/T polymorphism suggested that there was no significant effect on the Caucasian population.  Table 3). The subgroup analyses of mixed ethnic groups suggested that there were no significant association between HIF1A 1772 C/T polymorphism and cancer risk ( Table 3).  (Fig. 3).

Publication bias
To investigate the evidence of publication bias of the HIF1A 1772 C/T polymorphism for T versus C allele and HIF1A 1790 G/A polymorphism for G versus A allele funnel plot were used. The conventionally constructed funnel plot (log odds ratio [log(OR] vs 1/ standard error, 1/se) of HIF1A polymorphism 1772 C/T for T vs. C allele suggested that there was evidence of publication bias (Fig. 4). Also the funnel plot of HIF1A polymorphism 1790 G/A for A vs. G allele suggested that there was evidence of publication bias (Fig. 4)

Sensitivity analysis
Studies which were not in HWE were excluded to evaluate the stability of the acquired results. The statistical significance of the results was not shifted after omitting the studies which were not in HWE which confirmed the obtained results of the meta-analysis were stable and robust.

Conclusion
Results generated from this meta-analysis indicated that both 1772 C/T and 1790 G/A polymorphisms are significantly associated with increasing overall cancer risk. The subgroup analyses by cancer type showed that both 1772 C/ T and 1790 G/A polymorphisms have significant association with lung cancer, whereas these two polymorphisms showed no significant association with prostate cancer. In oral squamous cell carcinoma (OSCC) subgroup analyses data showed that only 1790 G/A polymorphism has significant association whereas the HIF1A 1772 C/T polymorphism showed no significant association. However, the 1772 C/T polymorphism has indicated significantly decreased risk in renal cell carcinoma. Also, 1790 G/A polymorphism has increased the cancer risk significantly in both Caucasian and Asian ethnicity. Taken together all analyzed data, HIF1A could be a prognostic marker useful for early detection and diagnosis for cancers. In future, further experimental validations would be necessary to confirm the results.