Hypoxia-inducible factors in cancer: an overview of major findings from meta-analyses

Introduction

Hypoxia, which refers to a low oxygen condition, is closely associated with the development and progression of cancer (1-3). Hypoxia-inducible factors (HIFs) are important proteins for the regulation of molecular response on hypoxia (4). HIFs consist of two subunits (i.e., α and β). The α subunit is expressed according to the oxygen conditions and determines the transcriptional activity of HIF. In details, the degradation of HIF-1α is enhanced and suppressed in the normoxic and hypoxic conditions, respectively; and high and low expression of HIF-1α increases and decreases the HIF activity, respectively. HIF-1α family contains 3 members (HIF-1α, HIF-2α, and HIF-3α) (5-8). By comparison, the β subunit is constitutively expressed in the nucleus.

Among the HIF-1α family members, HIF-1α is the most widely studied in human cancer (9,10). HIF-1α gene, which is located at the chromosome 14q21-24, consists of 15 exons and 14 introns, codes the cDNA of 3,919 bps, and produces the protein of 826 amino acids. HIF-1α can transactivate more than 70 target genes and is a master regulator of erythropoiesis, blood vessel formation, cell metabolism, and genetic stability. There are two major HIF-1α gene polymorphisms (C1772T and G1790A). Both of them are located at the exon 12 of the HIF-1α gene within the oxygen-dependent degradation domain. HIF-1α C1772T (rs11549465) mutation refers to an amino acid substitution from proline to serine at codon 582 (Pro582Ser or P582S). HIF-1α G1790A (rs11549467) mutation refers to an amino acid substitution from alanine to threonine at codon 588 (Ala588Thr or A588T).

To the best of our knowledge, numerous studies and meta-analyses have explored the role of HIF-1α gene polymorphism and protein expression in cancer. By comparison, less evidence has been accumulated regarding the role of HIF-2α and HIF-3α in cancer. In this paper, we have conducted an overview of meta-analyses to provide more comprehensive recognition of evidence regarding the role of HIFs in cancer.

Methods

Registration

Our study protocol was registered in PROSPERO database. The registration number was CRD42016037401.

Search strategy

We identified the relevant meta-analysis papers via the PubMed and EMBASE databases. We also manually identified the relevant meta-analysis papers. Search items were: “(hypoxia inducible factor) OR HIF” AND “(((cancer) OR tumor) OR neoplasm) OR carcinoma” AND “(meta analysis)”. The last search date was April 6, 2016.

Eligibility criteria

Only meta-analysis papers regarding the role of HIF in cancer were eligible for our study. Duplicates, comments or editorials, narrative reviews, original articles, and irrelevant meta-analysis papers were excluded. Publication language or date was not limited.

Data extraction

We primarily extracted the data from the eligible meta-analysis papers, as follows: first author, publication year, journal, country, databases which were employed for each meta-analysis, date when each meta-analysis was conducted, type of cancer, HIF gene polymorphism or protein expression, number of studies which were included in each meta-analysis, and results of each meta-analysis. If the statistical analyses were performed by using both fixed- and random-effects models, only the results by a random-effects model would be considered.

Evaluation of heterogeneity

If the results were heterogeneous among two or more meta-analyses, we would further identify the reliability according to the following criteria.

First, the number of eligible studies should be considered. A meta-analysis with a larger number of eligible studies would be more reliable.

Second, if the number of eligible studies was similar among them, the number of participants would be considered. A meta-analysis with a larger number of participants would be more reliable.

Third, if the eligible studies were completely overlapped among them, the methods of meta-analysis would be considered. A meta-analysis using a random-effect model would be more reliable.

Fourth, if the controversy or uncertainty remained according to the above-mentioned criteria, the original studies would be extracted and a meta-analysis might be updated. We might also contact with the authors or journal editors, if necessary.

Results

After excluding the irrelevant papers, a total of 55 meta-analysis papers were included in our study (Figure 1). Among them, 53 papers were written by Chinese researchers, 1 paper by UK researchers, and 1 paper by Bangladeshi researchers. The last search date for each meta-analysis ranged from 2009 to 2016. Results of meta-analyses were summarized according to the location of cancer.

Figure 1 The flowchart of inclusion.

Overall cancer

A total of 13 meta-analysis papers explored the role of HIF in overall cancer regardless of location of cancer (11-23) (Table S1). Among them, 5 papers explored both HIF-1α rs11549465 (1772 C/T) and rs11549467 (1790 G/A) polymorphisms (11,13,15,18,22), 5 papers explored HIF-1α rs11549465 (1772 C/T) polymorphism alone (12,14,17,19,21), and 3 papers explored HIF-1α rs11549467 (1790 G/A) polymorphism alone (16,20,23).

Risk

Nine papers explored the association of HIF-1α rs11549465 (1772 C/T) polymorphism with the risk of overall cancer (11,12,14,15,17-19,21,22). All of them demonstrated that HIF-1α rs11549465 (1772 C/T) polymorphism was significantly associated with the risk of overall cancer (11,12,14,15,17-19,21,22).

Seven papers explored the association of HIF-1α rs11549467 (1790 G/A) polymorphism with the risk of overall cancer (11,15,16,18,20,22,23). Six of them demonstrated that HIF-1α rs11549467 (1790 G/A) polymorphism was significantly associated with the risk of overall cancer (11,15,16,18,20,23). But another paper did not show any significant association between them (22). The meta-analyses by Liu P (16) and Zhou Y (23) had a larger number of included studies than those by Yang X (18), Ye Y (20), Anam MT (11), Zhao T (22), and Liu J (15) (26 and 26 versus 24, 21, 19, 12, and 6). Thus, we should support a significant association between HIF-1α rs11549467 (1790 G/A) polymorphism and the risk of overall cancer.

Clinicopathological features

One paper explored the association of HIF-1α rs11549465 (1772 C/T) polymorphism with the clinicopathological features of overall cancer (13). It demonstrated that HIF-1α rs11549465 (1772 C/T) polymorphism was significantly associated with the lymph node metastasis and histological grade of overall cancer, but not the tumor size or stage (13).

One paper explored the association of HIF-1α rs11549467 (1790 G/A) polymorphism with the clinicopathological features of overall cancer (13). It demonstrated that HIF-1α rs11549467 (1790 G/A) polymorphism was significantly associated with the lymph node metastasis and tumor size of overall cancer, but not the histological grade or tumor stage (13).

Head and neck cancer

A total of 4 meta-analysis papers explored the role of HIF in head and neck cancer (12,14,16,23) (Table S2). Among them, 2 papers explored HIF-1α rs11549465 (1772 C/T) polymorphism alone (12,14), and another 2 papers explored HIF-1α rs11549467 (1790 G/A) polymorphism alone (16,23).

Risk

Two papers explored the association of HIF-1α rs11549465 (1772 C/T) polymorphism with the risk of head and neck cancer (12,14). One of them demonstrated that HIF-1α rs11549465 (1772 C/T) polymorphism was significantly associated with the risk of head and neck cancer (12). But another paper did not show any significant association between them (14). The meta-analysis by He P (12) had a larger number of included studies than that by Li Y (14) (5 versus 1). Thus, we should support a significant association between HIF-1α rs11549465 (1772 C/T) polymorphism and the risk of head and neck cancer.

Two papers explored the association of HIF-1α rs11549467 (1790 G/A) polymorphism with the risk of head and neck cancer (16,23). One of them demonstrated that HIF-1α rs11549467 (1790 G/A) polymorphism was significantly associated with the risk of head and neck cancer (23). But another paper did not show any significant association between them (16). The meta-analysis by Zhou Y (23) had a larger number of included studies than that by Liu P (16) (6 versus 1). Thus, we should support a significant association between HIF-1α rs11549467 (1790 G/A) polymorphism and the risk of head and neck cancer.

Glioma

A total of two meta-analysis papers explored the role of HIF in glioma (14,24) (Table S3). Among them, one paper explored HIF-1α rs11549465 (1772 C/T) polymorphism alone (14), and another paper explored HIF-1α expression alone (24).

Risk

One paper explored the association of HIF-1α rs11549465 (1772 C/T) polymorphism with the risk of glioma (14). It demonstrated that HIF-1α rs11549465 (1772 C/T) polymorphism was significantly associated with the risk of glioma (14).

Clinicopathological features

One paper explored the association of HIF-1α expression with the clinicopathological features of glioma (24). It demonstrated that HIF-1α expression was significantly associated with the tumor stage of glioma (24).

Oral cancer

A total of ten meta-analysis papers explored the role of HIF in oral cancer (14,16,18-20,25-29) (Table S4). Among them, four papers explored both HIF-1α rs11549465 (1772 C/T) and rs11549467 (1790 G/A) polymorphisms (18,27-29), three papers explored HIF-1α rs11549465 (1772 C/T) polymorphism alone (14,19,25), two papers explored HIF-1α rs11549467 (1790 G/A) polymorphism alone (16,20), and one paper explored both HIF-1α and HIF-2α protein expressions (26).

Risk

Seven papers explored the association of HIF-1α rs11549465 (1772 C/T) polymorphism with the risk of oral cancer (14,18,19,25,27-29). All of them did not show any significant association between them (14,18,19,25,27-29).

Six papers explored the association of HIF-1α rs11549467 (1790 G/A) polymorphism with the risk of oral cancer (16,18,20,27-29). Four of them demonstrated that HIF-1α rs11549467 (1790 G/A) polymorphism was significantly associated with the risk of oral cancer (16,20,27,28). But another two papers did not show any significant association between them (18,29). The meta-analyses by Sun X (27) and Yan Q (28) had a larger number of included studies than those by Liu P (16), Yang X (Plos One, 2013) (18), Yang X (Tumour Biol, 2014) (29), and Ye Y (20). Thus, we should support a significant association between HIF-1α rs11549467 (1790 G/A) polymorphism and the risk of oral cancer.

Prognosis

One paper explored the association of HIF-1α and HIF-2α protein expression with the prognosis of oral cancer (26). It demonstrated that neither HIF-1α nor HIF-2α protein expression was significantly associated with the survival of oral cancer (26).

Oropharyngeal cancer

Only one paper explored the role of HIF in oropharyngeal cancer (30) (Table S5). It explored the association of HIF-1α expression with the prognosis of oropharyngeal cancer (30). It demonstrated that HIF-1α expression was significantly associated with the survival of oropharyngeal cancer (30).

Nasopharyngeal cancer

Only one paper explored the role of HIF in nasopharyngeal cancer (31) (Table S6). It explored the association of HIF-1α expression with the risk and clinicopathological features of nasopharyngeal cancer (31). It demonstrated that HIF-1α expression was significantly associated with the risk, lymph node metastasis, and clinical stage of nasopharyngeal cancer (31).

Lung cancer

A total of 12 meta-analysis papers explored the role of HIF in lung cancer (11,12,14,16,18,23,25,28,32-35) (Table S7). Among them, 4 papers explored both HIF-1α rs11549465 (1772 C/T) and rs11549467 (1790 G/A) polymorphisms (11,18,28,33), 3 papers explored HIF-1α rs11549465 (1772 C/T) polymorphism alone (12,14,25), 2 papers explored HIF-1α rs11549467 (1790 G/A) polymorphism alone (16,23), 1 paper explored both HIF-1α and HIF-2α protein expressions (32), and 2 papers explored HIF-1α protein expression alone (34,35).

Risk

Seven papers explored the association of HIF-1α rs11549465 (1772 C/T) polymorphism with the risk of lung cancer (11,12,14,18,25,28,33). Four of them demonstrated that HIF-1α rs11549465 (1772 C/T) polymorphism was significantly associated with the risk of lung cancer (11,18,28,33). But another 3 papers did not show any significant association between them (12,14,25). The meta-analyses by He P (12), Hu X (25), Li Y (14), Yan Q (28), and Yang X (18) had a larger number of included studies than those by Anam MT (11) and Liao S (33) (3, 3, 3, 3, and 3 versus 2 and 2). Among the meta-analyses by He P (12), Hu X (25), Li Y (14), Yan Q (28), and Yang X (18), the included studies were completely identical (Table S8). Only the meta-analysis by Yang X employed a random-effect model (18). Thus, we should support a significant association between HIF-1α rs11549465 (1772 C/T) polymorphism and the risk of lung cancer.

Six papers explored the association of HIF-1α rs11549467 (1790 G/A) polymorphism with the risk of lung cancer (11,16,18,23,28,33). All of them demonstrated that HIF-1α rs11549467 (1790 G/A) polymorphism was significantly associated with the risk of lung cancer (11,16,18,23,28,33).

Clinicopathological features

One paper explored the association of HIF-1α protein expression with the clinicopathological features of lung cancer (34). It demonstrated that HIF-1α protein expression was significantly associated with the stage, pathological type, diameter, lymph node metastasis, and differentiation of lung cancer (34).

Prognosis

Three papers explored the association of HIF-1α protein expression with the prognosis of lung cancer (32,34,35). All of them demonstrated that HIF-1α protein expression was significantly associated with the survival of lung cancer (32,34,35).

One paper explored the association of HIF-2α protein expression with the prognosis of lung cancer (32). It demonstrated that HIF-2α protein expression was significantly associated with the survival of lung cancer (32).

Breast cancer

A total of 17 meta-analysis papers explored the role of HIF in breast cancer (11-14,16-20,22,23,25,28,36-39) (Table S9). Among them, 5 papers explored both HIF-1α rs11549465 (1772 C/T) and rs11549467 (1790 G/A) polymorphisms (11,18,22,28,39), 7 papers explored HIF-1α rs11549465 (1772 C/T) polymorphism alone (12-14,17,19,25,36), 3 papers explored HIF-1α rs11549467 (1790 G/A) polymorphism alone (16,20,23), and 2 papers explored HIF-1α protein expression alone (37,38).

Risk

Eleven papers explored the association of HIF-1α rs11549465 (1772 C/T) polymorphism with the risk of breast cancer (11,12,14,17-19,22,25,28,36,39). Three of them demonstrated that HIF-1α rs11549465 (1772 C/T) polymorphism was significantly associated with the risk of breast cancer (14,18,39). But another 8 papers did not show any significant association between them (11,12,17,19,22,25,28,36). The meta-analyses by He P (12), Ren HT (36), Wu G (17), and Yan Q (28) had a larger number of included studies than those by Hu X (Tumour Biol, 2014) (25), Li Y (14), Yang X (18), Ye Y (19), Zhao T (22), and Anam MT (11) (6, 6, 6, and 6 versus 5, 5, 5, 3, 3, and 2). An abstract paper by Yin W did not report the number of included studies (39). Thus, we should not support any significant association between HIF-1α rs11549465 (1772 C/T) polymorphism and the risk of breast cancer.

Eight papers explored the association of HIF-1α rs11549467 (1790 G/A) polymorphism with the risk of breast cancer (11,16,18,20,22,23,28,39). Two of them demonstrated that HIF-1α rs11549467 (1790 G/A) polymorphism was significantly associated with the risk of breast cancer (11,22). But another 6 papers did not show any significant association between them (16,18,20,23,28,39). The meta-analysis by Yan Q (28) had a larger number of included studies than those by Liu P (16), Yang X (18), Zhou Y (23), Anam MT (11), Ye Y (20), and Zhao T (22) (4 versus 3, 3, 3, 2, 2, and 2). An abstract paper by Yin W did not report the number of included studies (39). Thus, we should not support any significant association between HIF-1α rs11549467 (1790 G/A) polymorphism and the risk of breast cancer.

Clinicopathological features

One paper explored the association of HIF-1α rs11549465 (1772 C/T) polymorphism with the clinicopathological features of breast cancer (13). It did not show any significant association of HIF-1α rs11549465 (1772 C/T) polymorphism with the lymph node metastasis or histological grade of breast cancer (13).

One paper explored the association of HIF-1α protein expression with the clinicopathological features of breast cancer (37). It demonstrated that HIF-1α protein expression was significantly associated with the pathological differentiation, regional invasive extension, axillary lymph node status, and clinical stage of breast cancer (37).

Prognosis

Two papers explored the association of HIF-1α protein expression with the prognosis of breast cancer (37,38). Both of them demonstrated that HIF-1α protein expression was significantly associated with the survival of breast cancer (37,38).

Digestive cancer

A total of 33 meta-analysis papers explored the role of HIF in digestive cancer (Table S10). Among them, 6 papers explored both HIF-1α rs11549465 (1772 C/T) and rs11549467 (1790 G/A) polymorphisms (15,27-29,40,41), 11 papers explored HIF-1α rs11549465 (1772 C/T)polymorphism alone (11-14,17-19,22,25,42,43), 3 papers explored HIF-1α rs11549467 (1790 G/A) polymorphism alone (16,20,23), 1 paper explored both HIF-1α and HIF-2α protein expressions (44), 10 papers explored HIF-1α protein expression alone (45-54), and 2 papers explored HIF-2α protein expression alone (55,56).

Overall digestive cancer

A total of 8 meta-analysis papers explored the role of HIF in overall digestive cancer regardless of location of digestive cancer (17,20,27,29,40-43) (Table S10). Among them, 4 papers explored both HIF-1α rs11549465 (1772 C/T) and rs11549467 (1790 G/A) polymorphisms (27,29,40,41), 3 papers explored HIF-1α rs11549465 (1772 C/T)polymorphism alone (17,42,43), and 1 paper explored HIF-1α rs11549467 (1790 G/A) polymorphism alone (20).

Risk

Seven papers explored the association of HIF-1α rs11549465 (1772 C/T) polymorphism with the risk of overall digestive cancer (17,27,29,40-43). Four of them demonstrated that HIF-1α rs11549465 (1772 C/T) polymorphism was significantly associated with the risk of overall digestive cancer (29,40,42,43). But another 3 papers did not show any significant association between them (17,27,41). The meta-analysis by Sun X (27) had a larger number of included studies than those by Yang X (29), Ni Z (40), Wu G (17), Xu JJ (Genet Mol Res, 2014) (41), Xu J (Genet Mol Res, 2014) (43), and Xu J (Genet Test Mol Biomarkers, 2013) (42) (13 versus 12, 10, 9, 8, 6, and 6). Thus, we should not support any significant association between HIF-1α rs11549465 (1772 C/T) polymorphism and the risk of overall digestive cancer.

Five papers explored the association of HIF-1α rs11549467 (1790 G/A) polymorphism with the risk of overall digestive cancer (20,27,29,40,41). All of them demonstrated that HIF-1α rs11549467 (1790 G/A) polymorphism was significantly associated with the risk of overall digestive cancer (20,27,29,40,41).

Esophageal cancer

A total of 5 meta-analysis papers explored the role of HIF in esophageal cancer (14,42,47,49,50) (Table S10). Among them, 2 papers explored HIF-1α rs11549465 (1772 C/T) polymorphism alone (14,42) and 3 papers explored HIF-1α protein expression alone (47,49,50).

Risk

Two papers explored the association of HIF-1α rs11549465 (1772 C/T) polymorphism with the risk of esophageal cancer (14,42). Both of them did not show any significant association between them (14,42).

Two papers explored the association of HIF-1α protein expression with the risk of esophageal cancer (47,50). Both of them demonstrated that HIF-1α protein expression was significantly associated with the risk of esophageal cancer (47,50).

Clinicopathological features

Three papers explored the association of HIF-1α protein expression with the clinicopathological features of esophageal cancer (47,49,50). All of them demonstrated that HIF-1α protein expression was significantly associated with the lymphoma node metastasis of esophageal cancer (47,49,50).

Prognosis

Two papers explored the association of HIF-1α protein expression with the prognosis of esophageal cancer (47,49). Both of them demonstrated that HIF-1α protein expression was significantly associated with the survival of esophageal cancer (47,49).

Gastric cancer

A total of 8 meta-analysis papers explored the role of HIF in gastric cancer (14,16,42,46,48,52,54) (Table S10). Among them, 2 papers explored HIF-1α rs11549465 (1772 C/T)polymorphism alone (14,42), 1 paper explored HIF-1α rs11549467 (1790 G/A) polymorphism alone (16), 4 papers explored HIF-1α protein expression alone (46,48,52,54), and 1 paper explored HIF-2α expression alone (56).

Risk

Two papers explored the association of HIF-1α rs11549465 (1772 C/T) polymorphism with the risk of gastric cancer (14,42). One of them demonstrated that HIF-1α rs11549465 (1772 C/T) polymorphism was significantly associated with the risk of gastric cancer (42). But another paper did not show any significant association between them (14). The number of included studies was similar between the two meta-analysis papers by Li Y (14) and Xu J (42) (1 versus 1). The included study was also identical between the two meta-analysis papers (Table S11). After learning the results from the original study (Li K, et al. Biochem Genet, 2009) (57), we should not support any significant association between HIF-1α rs11549465 (1772 C/T) polymorphism and the risk of gastric cancer.

One paper explored the association of HIF-1α rs11549467 (1790 G/A) polymorphism with the risk of gastric cancer (16). It demonstrated that HIF-1α rs11549467 (1790 G/A) polymorphism was significantly associated with the risk of gastric cancer (16).

Clinicopathological features

Three papers explored the association of HIF-1α protein expression with the clinicopathological features of gastric cancer (46,48,54). All of them demonstrated that HIF-1α protein expression was significantly associated with the depth of invasion, lymphatic invasion, vascular invasion, and TNM stage of gastric cancer (46,48,54).

One paper explored the association of HIF-2α protein expression with the clinicopathological features of gastric cancer (56). It demonstrated that HIF-2α protein expression was significantly associated with the tumor infiltration, lymphatic metastasis, and TNM stage of gastric cancer (56).

Prognosis

Four papers explored the association of HIF-1α protein expression with the prognosis of gastric cancer (46,48,52,54). All of them demonstrated that HIF-1α protein expression was significantly associated with the survival of gastric cancer (46,48,52,54).

One paper explored the association of HIF-2α protein expression with the prognosis of gastric cancer (56). It demonstrated that HIF-2α protein expression was significantly associated with the survival of gastric cancer (56).

Colorectal cancer

A total of 15 meta-analysis papers explored the role of HIF in colorectal cancer (11-16,18,19,22,25,27-29,42,44) (Table S10). Among them, 4 papers explored both HIF-1α rs11549465 (1772 C/T) and rs11549467 (1790 G/A) polymorphisms (15,27-29), 9 papers explored HIF-1α rs11549465 (1772 C/T) polymorphism alone (11-14,18,19,22,25,42), 1 paper explored HIF-1α rs11549467 (1790 G/A) polymorphism alone (16), and 1 paper explored both HIF-1α and HIF-2α protein expressions (44).

Risk

Twelve papers explored the association of HIF-1α rs11549465 (1772 C/T) polymorphism with the risk of colorectal cancer (11,12,14,15,18,19,22,25,27-29,42,55). All of them did not show any significant association between them (11,12,14,15,18,19,22,25,27-29,42).

Four papers explored the association of HIF-1α rs11549467 (1790 G/A) polymorphism with the risk of colorectal cancer (16,27-29). All of them did not show any significant association between them (16,27-29).

Clinicopathological features

One paper explored the association of HIF-1α rs11549465 (1772 C/T) polymorphism with the clinicopathological features of colorectal cancer (13). It did not show any significant association of HIF-1α rs11549465 (1772 C/T) polymorphism with the lymph node metastasis and histological grade of colorectal cancer (13).

One paper explored the association of HIF-1α protein expression with the clinicopathological features of colorectal cancer (44). It demonstrated that HIF-1α protein expression was significantly associated with the Dukes’ stages, lymph node status, depth of invasion, metastasis, and UICC stage of colorectal cancer, but not the differentiation grade (44).

One paper explored the association of HIF-2α protein expression with the clinicopathological features of colorectal cancer (44). It demonstrated that HIF-2α protein expression was significantly associated with the differentiation grade of colorectal cancer, but not the Dukes’ stages, lymph node status, or depth of invasion (44).

Prognosis

One paper explored the association of HIF-1α protein expression with the prognosis of colorectal cancer (44). It demonstrated that HIF-1α protein expression was significantly associated with the survival of colorectal cancer (44).

One paper explored the association of HIF-2α protein expression with the prognosis of colorectal cancer (44). It demonstrated that HIF-2α protein expression was significantly associated with the survival of colorectal cancer (44).

Pancreatic cancer

A total of 8 meta-analysis papers explored the role of HIF in pancreatic cancer (12,14,16,23,27-29,51) (Table S10). Among them, 2 papers explored both HIF-1α rs11549465 (1772 C/T) and rs11549467 (1790 G/A) polymorphisms (27,29), 2 papers explored HIF-1α rs11549465 (1772 C/T)polymorphism alone (12,14), 3 papers explored HIF-1α rs11549467 (1790 G/A) polymorphism alone (16,23,28), and 1 paper explored HIF-1α protein expression alone (51).

Risk

Four papers explored the association of HIF-1α rs11549465 (1772 C/T) polymorphism with the risk of pancreatic cancer (12,14,27,29). All of them demonstrated that HIF-1α rs11549465 (1772 C/T) polymorphism was significantly associated with the risk of pancreatic cancer (12,14,27,29).

Five papers explored the association of HIF-1α rs11549467 (1790 G/A) polymorphism with the risk of pancreatic cancer (16,23,27-29). All of them demonstrated that HIF-1α rs11549467 (1790 G/A)polymorphism was significantly associated with the risk of pancreatic cancer (16,23,27-29).

Clinicopathological features

One paper explored the association of HIF-1α protein expression with the clinicopathological features of pancreatic cancer (51). It demonstrated that HIF-1α protein expression was significantly associated with the lymph node metastasis and tumor stage of pancreatic cancer, but not the tumor size (51).

Prognosis

One paper explored the association of HIF-1α protein expression with the prognosis of pancreatic cancer (51). It demonstrated that HIF-1α protein expression was significantly associated with the survival of pancreatic cancer (51).

Hepatocellular carcinoma

A total of 5 meta-analysis papers explored the role of HIF in hepatocellular carcinoma (14,16,45,53,55) (Table S10). Among them, 1 paper explored HIF-1α rs11549465 (1772 C/T) polymorphism alone (14), 1 paper explored HIF-1α rs11549467 (1790 G/A) polymorphism alone (16), 2 papers explored HIF-1α protein expression alone (45,53), and 1 paper explored HIF-2α protein expression alone (55).

Risk

One paper explored the association of HIF-1α rs11549465 (1772 C/T) polymorphism with the risk of hepatocellular carcinoma (14). It did not show any significant association between them (14).

One paper explored the association of HIF-1α rs11549467 (1790 G/A) polymorphism with the risk of hepatocellular carcinoma (16). It demonstrated that HIF-1α rs11549467 (1790 G/A) polymorphism was significantly associated with the risk of hepatocellular carcinoma (16).

Clinicopathological features

One paper explored the association of HIF-1α protein expression with the clinicopathological features of hepatocellular carcinoma (46). It demonstrated that HIF-1α protein expression was significantly associated with the vascular invasion of hepatocellular carcinoma, but not the tumor size or differentiation, liver cirrhosis, or capsule formation (46).

One paper explored the association of HIF-2α protein expression with the clinicopathological features of hepatocellular carcinoma (55). It demonstrated that HIF-2α protein expression was significantly associated with the vein invasion, histological grade, and capsule infiltration of hepatocellular carcinoma, but not the tumor size or liver cirrhosis (55).

Prognosis

Two papers explored the association of HIF-1α protein expression with the prognosis of hepatocellular carcinoma (46,53). Both of them demonstrated that HIF-1α protein expression was significantly associated with the survival of hepatocellular carcinoma (46,53).

One paper explored the association of HIF-2α protein expression with the prognosis of hepatocellular carcinoma (55). It did not show any significant association between HIF-2α protein expression and the survival of hepatocellular carcinoma (55).

Urinary cancer

A total of 15 meta-analysis papers explored the role of HIF in urinary cancer (11,12,14,16-20,22,23,25,28,58-60) (Table S12). Among them, 5 papers explored both HIF-1α rs11549465 (1772 C/T) and rs11549467 (1790 G/A) polymorphisms (11,18,22,28,59), 5 papers explored HIF-1α rs11549465 (1772 C/T) polymorphism alone (12,14,17,19,25), 3 papers explored HIF-1α rs11549467 (1790 G/A) polymorphism alone (16,20,23), 1 paper explored both HIF-1α and HIF-2α protein expressions (58), and 1 paper explored HIF-1α protein expression alone (60).

Overall urinary cancer

One meta-analysis paper explored the association of HIF-1α rs11549465 (1772 C/T) and rs11549467 (1790 G/A) polymorphisms with the risk of overall urinary cancer (59) (Table S12). It demonstrated that neither of them was significantly associated with the risk of overall urinary cancer (59).

Prostate cancer

A total of 13 meta-analysis papers explored the role of HIF in prostate cancer (11,12,14,16-20,22,23,25,28,59) (Table S12). Among them, 5 papers explored both HIF-1α rs11549465 (1772 C/T) and rs11549467 (1790 G/A) polymorphisms (11,18,22,28,59), 5 papers explored HIF-1α rs11549465 (1772 C/T) polymorphism alone (11,18,22,28,59), and 3 papers explored HIF-1α rs11549467 (1790 G/A) polymorphism alone (16,20,23).

Risk

Ten papers explored the association of HIF-1α rs11549465 (1772 C/T) polymorphism with the risk of prostate cancer (11,12,14,17-19,22,25,59). Five of them demonstrated that HIF-1α rs11549465 (1772 C/T) polymorphism was significantly associated with the risk of prostate cancer (14,18,19,22,25). Another 5 papers did not show any significant association between them (11,12,17,28,59). The meta-analyses by Anam MT (11), He P (12), Li D (59), Wu G (17), and Yan Q (28) had a larger number of included studies than those by Hu X (25), Yang X (18), Ye Y (19), Li Y (14), and Zhao T (22) (6, 6, 6, 6, and 6 versus 5, 5, 5, 4, and 4). Thus, we should not support any significant association between HIF-1α rs11549465 (1772 C/T) polymorphism and the risk of prostate cancer.

Eight papers explored the association of HIF-1α rs11549467 (1790 G/A) polymorphism with the risk of prostate cancer (11,16,18,20,22,23,28,59). One of them demonstrated that HIF-1α rs11549467 (1790 G/A)polymorphism was significantly associated with the risk of prostate cancer (59). Another 7 papers did not show any significant association between them (11,16,18,20,22,23,28). The meta-analysis by Li D (59) had a larger number of included studies than those by Anam MT (11), Liu P (16), Yan Q (28), Ye Y (20), Yang X (18), Zhou Y (23), and Zhao T (22) (4 versus 3, 3, 3, 3, 3, 3, and 2). Thus, we should support a significant association between HIF-1α rs11549467 (1790 G/A) polymorphism and the risk of prostate cancer.

Renal cancer

A total of 13 meta-analysis papers explored the role of HIF in renal cancer (11,12,14,16,17,19,20,23,25,28,58-60) (Table S12). Among them, 3 papers explored both HIF-1α rs11549465 (1772 C/T) and rs11549467 (1790 G/A) polymorphisms (11,28,59), 5 papers explored HIF-1α rs11549465 (1772 C/T) polymorphism alone (12,14,17,19,25), 3 papers explored HIF-1α rs11549467 (1790 G/A) polymorphism alone (16,20,23), 1 paper explored both HIF-1α and HIF-2α nuclear and cytoplasmic expressions (58), and 1 paper explored HIF-1α protein expression alone (60).

Risk

Eight papers explored the association of HIF-1α rs11549465 (1772 C/T) polymorphism with the risk of renal cancer (11,12,14,17,19,25,28,59). Two of them demonstrated that HIF-1α rs11549465 (1772 C/T) polymorphism was significantly associated with the risk of renal cancer (11,12). Another 6 papers did not show any significant association between them (14,17,19,25,28,59). The meta-analyses by Hu X (25), Li D (59), Wu G (17), and Yan Q (28) had a larger number of included studies than those by Anam MT (11), He P (12), Ye Y (19), and Li Y (14) (4, 4, 4, and 4 versus 3, 3, 3, and 2). Thus, we should not support any significant association between HIF-1α rs11549465 (1772 C/T) polymorphism and the risk of renal cancer.

Six papers explored the association of HIF-1α rs11549467 (1790 G/A) polymorphism with the risk of renal cancer (11,16,20,23,28,59). Three of them demonstrated that HIF-1α rs11549467 (1790 G/A) polymorphism was significantly associated with the risk of renal cancer (11,23,28). Another 3 papers did not show any significant association between them (16,19,59). The meta-analyses by Anam MT (11), Li D (59), and Yan Q (28) had a larger number of included studies than those by Liu P (16), Zhou Y (23), and Ye Y (20) (4, 4, and 4 versus 3, 3, and 2). The included studies were completely identical among the 3 meta-analyses by Anam MT (11), Li D (59), and Yan Q (28) (Table S13). Notably, some statistical results (AA + AG vs. GG and A allele vs. G allele) were completely identical among them (11,28,59). However, the meta-analyses by Anam MT (11) and Yan Q (28) had more statistical results (AA vs. GG, GA vs. GG, and AA vs. GA + GG) than that by Li D (59). Thus, we should support a significant association between HIF-1α rs11549467 (1790 G/A) polymorphism and the risk of renal cancer.

Clinicopathological features

One paper explored the association of HIF-1α protein expression with the clinicopathological features of renal cancer (60). It demonstrated that HIF-1α protein expression was significantly associated with the lymph node metastasis and clinical and pathological stage of renal cancer (60).

Prognosis

One paper explored the association of HIF-1α and HIF-2α nuclear and cytoplasmic expressions with the prognosis of renal cancer (58). It demonstrated that neither HIF-1α nor HIF-2α nuclear and cytoplasmic expression was significantly associated with the survival of renal cancer (58).

Bladder cancer

A total of 3 meta-analysis papers explored the role of HIF in bladder cancer (14,16,25) (Table S12). Among them, 2 papers explored HIF-1α rs11549465 (1772 C/T) polymorphism alone (14,25), and 1 paper explored HIF-1α rs11549467 (1790 G/A) polymorphism alone (16).

Risk

Two papers explored the association of HIF-1α rs11549465 (1772 C/T) polymorphism with the risk of bladder cancer (14,25). Neither of them demonstrated any significant association between HIF-1α rs11549465 (1772 C/T)polymorphism and the risk of bladder cancer (14,25).

One paper explored the association of HIF-1α rs11549467 (1790 G/A) polymorphism with the risk of bladder cancer (16). It did not show any significant association between them (16).

Gynecological cancer

A total of 12 meta-analysis papers explored the role of HIF in gynecological cancer (12-14,16,19,20,25,28,61-64) (Table S14). Among them, 1 paper explored both HIF-1α rs11549465 (1772 C/T) and rs11549467 (1790 G/A) polymorphisms (28), 8 papers explored HIF-1α rs11549465 (1772 C/T) polymorphism alone (12-14,19,25,64), 2 papers explored HIF-1α rs11549467 (1790 G/A) polymorphism alone (16,20), and 1 paper explored HIF-1α protein expression alone (61-63,65).

Overall gynecological cancer

A total of 3 meta-analysis papers explored the role of HIF in overall gynecological cancer (14,16,65) (Table S14). Among them, 1 paper explored HIF-1α rs11549465 (1772 C/T) polymorphism alone (14), 1 paper explored HIF-1α rs11549467 (1790 G/A) polymorphism alone (16), and 1 paper explored HIF-1α protein expression alone (65).

Risk

One paper explored the association of HIF-1α rs11549465 (1772 C/T) polymorphism with the risk of overall gynecological cancer (14). It demonstrated that HIF-1α rs11549465 (1772 C/T) polymorphism was significantly associated with the risk of overall gynecological cancer (14).

One paper explored the association of HIF-1α rs11549467 (1790 G/A) polymorphism with the risk of overall gynecological cancer (16). It did not show any significant association between them (16).

Clinicopathological features

One paper explored the association of HIF-1α protein expression with the clinicopathological features of overall gynecological cancer (65). It demonstrated that HIF-1α protein expression was significantly associated with the pathological and histological type, FIGO stage, and lymph node metastasis of overall gynecological cancer (65).

Prognosis

One paper explored the association of HIF-1α protein expression with the prognosis of overall gynecological cancer (65). It demonstrated that HIF-1α protein expression was significantly associated with the survival of overall gynecological cancer (65).

Ovarian cancer

A total of 3 meta-analysis papers explored the role of HIF in ovarian cancer (62,63,65). All of them explored HIF-1α protein expression alone (62,63,65) (Table S14).

Risk

One paper explored the association of HIF-1α protein expression with the risk of ovarian cancer (63). It demonstrated that HIF-1α protein expression was significantly associated with the risk of ovarian cancer (63).

Clinicopathological features

Three papers explored the association of HIF-1α protein expression with the lymph node metastasis of ovarian cancer (62,63,65). All of them demonstrated that HIF-1α protein expression was significantly associated with the lymph node metastasis of ovarian cancer (62,63,65).

Three papers explored the association of HIF-1α protein expression with the pathological type of ovarian cancer (62,63,65). Two of them demonstrated HIF-1α protein expression was significantly associated with the pathological type of ovarian cancer (62,65). But another paper did not show any significant association between them (63). The meta-analyses by Jin Y (Tumour Biol, 2014) (62) and Jin Y (PLoS One, 2015) (65) had a larger number of included studies than that by Sun C (63) (13 and 13 versus 4). Thus, we should support a significant association between HIF-1α protein expression and the pathological type of ovarian cancer.

Two papers explored the association of HIF-1α protein expression with the FIGO stage of ovarian cancer (62,65). Both of them demonstrated that HIF-1α protein expression was significantly associated with the FIGO stage of ovarian cancer (62,65).

Prognosis

Two papers explored the association of HIF-1α protein expression with the prognosis of ovarian cancer (62,65). Both of them demonstrated that HIF-1α protein expression was significantly associated with the survival of ovarian cancer (62,65).

Cervical cancer

A total of 10 meta-analysis papers explored the role of HIF in cervical cancer (12,13,18-20,25,28,61,64,65) (Table S14). Among them, 1 paper explored both HIF-1α rs11549465 (1772 C/T) and rs11549467 (1790 G/A) polymorphisms (28), 6 papers explored HIF-1α rs11549465 (1772 C/T)polymorphism alone (12,13,18,19,25,64), 1 paper explored HIF-1α rs11549467 (1790 G/A) polymorphism alone (20), and 2 papers explored HIF-1α protein expression alone (61,65).

Risk

Six papers explored the association of HIF-1α rs11549465 (1772 C/T) polymorphism with the risk of cervical cancer (12,18,19,25,28,64). Five of them demonstrated that HIF-1α rs11549465 (1772 C/T) polymorphism was significantly associated with the risk of cervical cancer (12,18,25,28,64). But another paper did not show any significant association between them (19). The meta-analysis by Zhu J (64) had a larger number of included studies than those by He P (12), Hu X (25), Yan Q (28), Yang X (18), and Ye Y (19) (4 versus 3, 3, 3, 3, and 3). Thus, we should support a significant association between HIF-1α rs11549465 (1772 C/T) polymorphism and the risk of cervical cancer.

Two papers explored the association of HIF-1α rs11549467 (1790 G/A) polymorphism with the risk of cervical cancer (20,28). Neither of them showed any significant association between HIF-1α rs11549467 (1790 G/A) polymorphism and the risk of cervical cancer (20,28).

Clinicopathological features

One paper explored the association of HIF-1α rs11549465 (1772 C/T) polymorphism with the lymph node metastasis of cervical cancer (13). It did not show any significant association between them (13).

Two papers explored the association of HIF-1α protein expression with the FIGO stage of cervical cancer (61,65). Both of them demonstrated that HIF-1α protein expression was significantly associated with the FIGO stage of cervical cancer (61,65).

Two papers explored the association of HIF-1α protein expression with the histological type and lymph node metastasis of cervical cancer (61,65). One of them demonstrated that HIF-1α protein expression was significantly associated with the histological type and lymph node metastasis of cervical cancer (65). But another paper did not show any significant association between them (61). As for the histological type, the meta-analysis by Jin Y (65) had a larger number of included studies than that by Huang M (61) (6 versus 4). As for the lymph node metastasis, the meta-analysis by Jin Y (65) had a larger number of included studies than that by Huang M (61) (8 versus 5). Thus, we should support a significant association between HIF-1α protein expression and the histological type and lymph node metastasis of cervical cancer.

Prognosis

Two papers explored the association of HIF-1α protein expression with the prognosis of cervical cancer (61,65). Both of them demonstrated that HIF-1α protein expression was significantly associated with the survival of cervical cancer (61,65).

Endometrial cancer

Only one paper explored the association of HIF-1α protein expression with the clinicopathological features and prognosis of endometrial cancer (65) (Table S14). It demonstrated that HIF-1α protein expression was significantly associated with the pathological and histological type, FIGO stage, and lymph node metastasis of endometrial cancer, but not the survival (65).

Osteosarcoma

Only one paper explored the association of HIF-1α protein expression with the clinicopathological features and prognosis of osteosarcoma (66) (Table S15). It demonstrated that HIF-1α protein expression was significantly associated with the metastasis, pathologic and tumor grade, and survival of osteosarcoma, but not the histopathology, tumor size, or tumor site (66).

Conclusions

Based on our systematic search strategy, numerous meta-analyses have explored the role of HIF gene polymorphism and protein expression in various human cancers, including head and neck cancer, glioma, oral cancer, oropharyngeal cancer, nasopharyngeal cancer, lung cancer, breast cancer, esophageal cancer, gastric cancer, colorectal cancer, pancreatic cancer, hepatocellular carcinoma, prostate cancer, renal cancer, bladder cancer, ovarian cancer, cervical cancer, endometrial cancer, and osteosarcoma (Figure 2).

Figure 2 A schematic diagram of various human cancers in which the role of HIFs has been explored by meta-analyses. HIFs, hypoxia-inducible factors.

Based on the current evidence, major findings were summarized in Table 1.

First, the evidence regarding the association of HIF-1α gene polymorphism with risk of cancer suggested the following: (I) both HIF-1α rs11549465 (1772 C/T) and HIF-1α rs11549467 (1790 G/A) polymorphisms should be associated with the risk of head and neck cancer and lung cancer; (II) HIF-1α rs11549465 (1772 C/T) polymorphism, rather than HIF-1α rs11549467 (1790 G/A) polymorphism, should be associated with the risk of cervical cancer; (III) HIF-1α rs11549467 (1790 G/A) polymorphism, rather than HIF-1α rs11549465 (1772 C/T) polymorphism, should be associated with the risk of oral cancer, gastric cancer, hepatocellular carcinoma, prostate cancer, and renal cancer; and (IV) neither HIF-1α rs11549465 (1772 C/T) nor HIF-1α rs11549467 (1790 G/A) polymorphism should be associated with the risk of breast cancer, colorectal cancer, and bladder cancer.

Second, the evidence regarding the association of HIF-1α protein expression with the lymph node metastasis of cancer suggested the following: (I) both HIF-1α and HIF-2α expression were associated with the lymph node metastasis of gastric cancer; and (II) HIF-1α expression, rather than HIF-2α expression, was associated with the lymph node metastasis of colorectal cancer.

Third, the evidence regarding the association of HIF-1α protein expression alone with the lymph node metastasis of cancer suggested that HIF-1α expression was associated with the lymph node metastasis of glioma, nasopharyngeal cancer, lung cancer, breast cancer, esophageal cancer, gastric cancer, pancreatic cancer, renal cancer, ovarian cancer, and osteosarcoma.

Fourth, the evidence regarding the association of HIF-1α protein expression with the survival of cancer suggested the following: (I) both HIF-1α and HIF-2α expressions were associated with the survival of lung cancer, gastric cancer, and colorectal cancer; (II) HIF-1α expression, rather than HIF-2α expression, was associated with the survival of hepatocellular carcinoma; and (III) neither HIF-1α nor HIF-2α expression was associated with the survival of renal cancer.

Fifth, the evidence regarding the association of HIF-1α protein expression alone with the survival of cancer suggested that HIF-1α expression was associated with the survival of oropharyngeal cancer, breast cancer, esophageal cancer, pancreatic cancer, ovarian cancer, cervical cancer, and osteosarcoma, but not that of endometrial cancer.

Collectively, the impact of HIFs on the risk, clinicopathological features, and survival of various human cancers should be heterogeneous. The potential explanation might be attributed to the heterogeneity in the cancer biological behavior and effect of hypoxia across the different types of human cancers. Further studies should uncover the potential mechanisms.