Genetic association of hypoxia inducible factor 1-alpha (HIF1A) Pro582Ser polymorphism with risk of diabetes and diabetic complications

Diabetes is an age-related chronic disease associated with a number of complications, emerging as one of the major causes of morbidity and mortality worldwide. Several studies indicated that hypoxia-inducible factor 1-alpha (HIF1A) genetic polymorphisms may be associated with diabetes and diabetic complications. However, this association remains ambiguous. Thus, we performed a meta-analysis to provide more precise conclusion on this issue. Odds ratios (OR) with corresponding 95% confidence intervals (CI) were applied to assess the strength of the relationships. There was a protective association between HIF1A Pro582Ser polymorphism and diabetes under the heterozygous genetic model (OR = 0.70, 95% CI = 0.55-0.91; P = 0.007). Similar associations were observed in diabetic complications risk under the allelic (OR = 0.69, 95% CI = 0.57-0.83; P < 0.001), homozygous (OR = 0.51, 95% CI = 0.30-0.87; P = 0.014), recessive (OR = 0.73, 95% CI = 0.59-0.90; P = 0.004) and dominant (OR = 0.40, 95% CI = 0.25-0.65; P < 0.001) genetic models. No effects of the HIF1A Ala588Thr polymorphism were found in risk of diabetes and diabetic complications. Taken together, these findings revealed the protective effect of HIF1A Pro582Ser polymorphism against diabetes and diabetic complications.

Clinical and experimental studies indicate that hyperglycemia suggests a state of pseudohypoxia and activates HIF-1α activity for adaptation of hypoxia [15,16]. In addition, hyperglycemia may impair the stabilization and transactivation of HIF-1α [17][18][19]. It has been postulated that the function of HIF-1a is repressed by hyperglycemia leading to the loss of cellular adaptation to hypoxia in diabetes, which suggests a mechanism in the pathophysiology of diabetes and diabetic complications [20][21][22][23].
The gene HIF1A (for HIF-1α) carries two common nonsynonymous single nucleotide polymorphisms (SNP) in exon 12, Pro582Ser (rs11549465) and Ala588Thr (rs11549467), which both exhibit higher transcriptional activity of HIF1A [24,25]. Previous studies indicated that HIF1A Pro582Ser and Ala588Thr may be associated with diabetes and diabetic complications. Yamada et al. [26] firstly reported the HIF1A Pro582Ser polymorphism exerted a protective effect in the occurrence of diabetes, but no correlation with diabetic complications in a Japanese population. While Ekberg et al. [27] identified the protective effect of HIF1A Pro582Ser on the development of severe diabetic retinopathy with risk reduction of 95%. Several other studies also focused on the associations of HIF1A Pro582Ser and Ala588Thr polymorphisms with diabetes and diabetic complications, including type 1 and type 2 diabetes, diabetic nephropathy, diabetic retinopathy and diabetic foot ulcers [28][29][30][31][32][33]. Nevertheless, the results of these studies are conflicting. To obtain accurate conclusion, we conducted a comprehensive metaanalysis based on the controversial results from various independent case-control studies.

Description of eligible studies
The current meta-analysis was conducted according to the guidelines of the "Preferred Reporting Items for Systematic reviews and Meta-Analyses'' (PRISMA) statement. As depicted in the flow diagram (Figure 1), the initial literature screening yielded 145 articles, and a total of 22 articles were excluded due to duplicate publication. Then, 94 articles were removed from screening according to titles and/or abstracts. Finally, based on the study inclusion criteria, 8 articles [26][27][28][29][30][31][32][33] involving 11 eligible studies for the association of HIF1A Pro582Ser and Ala588Thr polymorphisms with diabetes and diabetic complications were included in our meta-analysis. All the included articles were all conducted with case-control design and the sample sizes varied from 145 to 1165. A total of 5 and 6 eligible studies were identified for diabetes and diabetic complications, respectively. The general characteristics of the studies included in the meta-analysis were presented in Table 1.
Quantitative synthesis of the association between HIF1A Pro582Ser polymorphism and the risk of diabetic complications The results of meta-analysis and heterogeneity test between HIF1A Pro582Ser polymorphism and the risk of diabetic complications were summarized in detail in Table 2 and Figure 5. The pooled analysis indicated that the HIF1A Pro582Ser polymorphism was also significantly associated with a decreased risk of diabetic complications under the allelic (OR = 0.69, 95% CI = 0.57-0.83; P < 0.001), homozygous (OR = 0.51, 95% CI = 0.30-0.87; P = 0.014), recessive (OR = 0.73, 95%
Similarly, Galbraith plot analysis and sensitivity analysis were used to detect the possible sources of heterogeneity ( Figure 6). Galbraith plot analysis revealed that the Pichu et al. study was the outlier ( Figure 6A), which was consistent with the results of sensitivity analysis ( Figure 6B). Interestingly, the significant heterogeneity was eliminated after omitting the study by Pichu et al. in the meta-analysis under the heterozygous genetic model (Pheterogeneity = 0.423, I 2 = 0%) ( Table 3 and Figure 6C).

AGING
What's more, the corrected OR also revealed a significant association between HIF1A Pro582Ser polymorphism and a decreased risk of diabetic complications under the heterozygous genetic model (OR = 0.72, 95% CI = 0.57-0.91; P = 0.006) ( Table 3 and Figure 6C).

Quantitative synthesis of the association between HIF1A Ala588Thr genetic polymorphism with risk of diabetes
The results of meta-analysis and heterogeneity test between HIF1A Ala588Thr polymorphism and the risk of diabetes were summarized in detail in Table 4 and

Quantitative synthesis of the association between HIF1A Ala588Thr genetic polymorphism with risk of diabetic complications
The results of meta-analysis and heterogeneity test between HIF1A Ala588Thr polymorphism and the risk of diabetic complications were summarized in detail in Table 4 and Figure 8.

Publication bias evaluation
Publication bias of the included studies was assessed by using the Begg's funnel plot (Figure 9). For the metaanalysis of the association between HIF1A Pro582Ser polymorphism and the risk of diabetes, no evidence of significant publication bias was detected by the Begg's test (P = 0.089 for allelic genetic model; P = 0.602 for homozygous genetic model; P = 0.734 for dominant genetic model; P = 0.296 for recessive genetic model).
The P-values for Begg's test also demonstrated that there was no publication bias of meta-analysis of the association between HIF1A Pro582Ser polymorphism and the risk of diabetic complications (P > 0.1 for all genetic models).

DISCUSSION
To our knowledge, this is the first meta-analysis to explore the genetic associations between HIF1A  As a key oxygen sensor mediating cellular adaptive responses to hypoxia, HIF-1α plays a pivotal role in cellular and systemic homeostatic. The stabilization of HIF-1α is regulated by oxygen-dependent prolyl hydroxylation of proline domains located in Pro402 and Pro564, which is significant for the effect of hyperglycemia on HIF-1α [8,34]. However, HIF-1α Pro582 has not been certified as a domain for hydroxylating, and the substitution of serine for proline in this position has no essential role in HIF-1α stability [27,29,35]. Amino acids 582 is contained in a region of HIF-1α subunit which could act independently to convey inducible responses and confer transcriptional activation [36,37]. Previous studies revealed that HIF1A Pro582Ser polymorphism was a stable variant and showed increased transcriptional activity, which may offer enhanced HIF-1α activity under normoxic conditions [38,39]. It has been postulated that the enhanced activity of HIF-1α may provide increased adaptability for pseudohypoxia induced by hyperglycemia [12,19,40]. Hence, the increased    Heterogeneity among constituent studies is common in the meta-analysis of genetic association study and may affect the interpretation of the meta-analysis results [41,42]. For the meta-analysis of the role of HIF1A Pro582Ser polymorphism in the risk of diabetic complications, one of the strengths was lack of obvious heterogeneity in all genetic models except the heterozygous genetic model. In contrast, for the metaanalysis of the HIF1A Pro582Ser polymorphism in the risk of diabetes, significant heterogeneity was found in all genetic models except the heterozygous genetic model. Heterogeneity may result from the potential differences across the included studies, such as the definition of disease, ethnicity, genotyping methods, sample size. To detect the potential sources of heterogeneity, Galbraith plot analysis was firstly used to explore whether there was outlier study. Then, sensitivity analysis by omitting one individual study each time was further performed to identify the possible source of heterogeneity [42]. Galbraith plot analysis indicated the study conducted by Pichu et al. was the outlier, and sensitivity analysis also found Pichu's study was the main contributor to the significant heterogeneity. We found that HIF1A Pro582Ser genotype frequencies showed significant departure of HWE in the health control group of Pichu's study (P = 9.34*10 -7 ). In population study, migration, selection, mutation, and absence of random mating may exist when the genotype distribution of controls (disease-free subjects) deviates from HWE. Consequently, the Pichu's study departure from HWE may bias the metaanalysis results, and can explain the between-study heterogeneity. After excluding the outlier study, the AGING between-study heterogeneity can be effectively eliminated. What's more, the meta-analysis based on the corrected ORs also revealed that the HIF1A Pro582Ser polymorphism played a protective role in the risk of diabetes and diabetic complications.
Some limitations of the current meta-analysis should be admitted. Firstly, the development of diabetes and diabetic complications is affected not only by environmental factors but also multiple genetic factors, the effect of gene-to-gene interactions should be taken into account. For example, vascular endothelial growth factor, another susceptibility gene for diabetes, may interact with HIF1A gene [22]. Then, because these information was not available in the included studies, the results of our meta-analysis were all based on the crude ORs with corresponding 95% CIs. In addition, due to the small number of studies included in the meta-analysis of HIF1A Ala588Thr polymorphism with the risk of diabetes and its complications, the findings should be interpreted with caution. The small sample size may be responsible for the negative relationship between the HIF1A Ala588Thr polymorphism and diabetes and diabetic complications.
In conclusion, our meta-analysis revealed the protective role of the HIF1A Pro582Ser polymorphism against diabetes and diabetic complications. However, there was no significant association of HIF1A Ala588Thr polymorphism with the risk of diabetes and its complications. Owing to the limitations mentioned above, further studies, with larger sample sizes on the association of HIF1A genetic polymorphisms (especially HIF1A Ala588Thr) with the risk of diabetes and its complications, should be performed to confirm our findings in the future.

Search strategy and inclusion criteria
A systematical literature search was conducted in the following electronic databases: PubMed, Embase, WanFang Data, and China National Knowledge Infrastructure (CNKI) from their starting dates to December 2019. The following keywords used for the search strategy were hypoxia-inducible factor-1α gene (HIF1A) or variations (e.g.,"polymorphism", "single nucleotide polymorphism", "SNP", "variant", "mutation", "variation") in combination with diabetes and diabetic complications (e.g., "diabetes mellitus", "diabetic complications"). Additionally, other possible original publications were identified by manually searching the reference lists of the selected reviews and articles.
All the identified studies were independently evaluated by two investigators according to the inclusion criteria. The included studies met the criteria as follows: (1) studies were conducted in humans and assessed with a case-control design. (2) the association between Pro582Ser and Ala588Thr of HIF1A gene and risk of diabetes and diabetic complications was explored. (3) published in English or Chinese. (4) detailed HIF1A genotyping data were offered in case and control groups. If the two reviewers disagreed about the inclusion of a study, it was resolved by group discussion or consensus with a third reviewer.

Data extraction
For the included articles in this study, data were collected by two reviewers independently. The following information was extracted from each publication: last name of the first author, year of publication, country of the study, ethnicity of the population, mean age, gender distribution of cases and controls, allele and genetic distributions in case and control groups, total number of cases and controls.

Statistical analysis
In this meta-analysis, five genetic models were performed including the allelic (T vs. C of HIF1A Pro582Ser gene polymorphism; A vs. G of HIF1A Ala588Thr gene polymorphism), homozygous (TT vs. The strength of the association between HIF1A gene polymorphisms and diabetes and diabetic complications risk was evaluated by odds ratios (ORs) with 95% confidence intervals (CIs) according to the alleles and genotypes in case and control groups. The pooled estimates of the OR were calculated from a combination of studies in the allelic, homozygous, heterozygous, recessive and dominant models, respectively. The Z test was applied to determine the statistical significance of the pooled OR. The I 2 metric and Cochran's Q test were conducted to check the possibility of heterogeneity among the included studies. Between-study heterogeneity was considered as a statistic significance at I 2 > 50% for the I 2 test and P < 0.05 for the Q statistics [43]. If significant heterogeneity existed, the pooled OR was calculated via random effect model (the DerSimonian and Laird method). Otherwise, the fixed effect model (the Mante-Haenszel method) was used. Sensitivity analysis and Galbraith plot were conducted to explore the potential sources of heterogeneity across the studies. Potential publication bias was assessed with Begg's test [44]. All statistical analyses in our study were conducted using the software Review Manager 5.0 and STATA version 12.0 (Stata Corp, College Station, TX, USA).

CONFLICTS OF INTEREST
All of the authors declare that they have no potential conflicts of interest.