Association of OGG1 and MTHFR polymorphisms with age-related cataract: A systematic review and meta-analysis

Purpose To discern and confirm genetic biomarkers that help identify populations at high risk for age-related cataract (ARC). Methods A literature search was performed in the PubMed, Web of Science and China National Knowledge Internet databases for genetic association studies published before June 26, 2016 regarding ARC susceptibility. All genetic polymorphisms reported were systematically reviewed, followed by extraction of candidate genes/loci with sufficient genotype data in ≥3 studies for the meta-analysis. A random/fixed-effects model was used to calculate the pooled odds ratios and 95% confidence intervals to evaluate the associations considering multiple genetic models. Sensitivity analysis was also performed. Results A total of 144 polymorphisms in 36 genes were reported in the 61 previous genetic association studies. Thereby, three polymorphisms of two genes (8-oxoguanine DNA glycosylase-1 [OGG1]; methylenetetrahydrofolate reductase NADPH [MTHFR]) in eight studies were included in the meta-analysis. Regarding the OGG1-rs1052133, the GG (OR = 1.925; 95%CI, 1.181–3.136; p = 0.009) and CG (OR = 1.384; 95%CI, 1.171–1.636; p<0.001) genotypes indicated higher risk of ARC. For the MTHFR gene, the CC+TT genotype of rs1801133 might be protective (OR, 0.838; 95%CI, 0.710–0.989; p = 0.036), whereas the AA+CC genotype of rs1801131 indicated increased risk for the mixed subtype (OR = 1.517; 95%CI, 1.113–2.067; p = 0.008). Conclusions Polymorphisms of OGG1 and MTHFR genes are associated with ARC susceptibility and may help identify populations at high risk for ARC.


Literature search
A literature search was performed by two reviewers (Xiaohang Wu and Weiyi Lai) in the PubMed, Web of Science and China National Knowledge Internet (CNKI) electronic databases for genetic association studies concerning ARC susceptibility published before June 26, 2016. All the genetic association studies with ARC identified were systematically reviewed followed by extraction of candidate genes/loci with sufficient genotype data in !3 studies for the meta-analysis. We also manually assessed the reference lists of all the retrieved original studies, review articles and conference abstracts using the electronic databases listed above. In our literature search, combinations of items were used including cataract, polymorphism, and the full name/abbreviation of the candidate genes. The detailed search strategy in PubMed was listed as follows: ("polymorphism, genetic"[MeSH Terms] OR ("polymorphism"[All Fields] AND "genetic"[All Fields]) OR "genetic polymorphism"[All Fields] OR "polymorphism"[All Fields]) AND ("cataract"[MeSH Terms] OR "cataract"[All Fields]).
For more details, please refer to S1 Appendix in the supplementary information.

Eligibility criteria
We considered studies eligible for the meta-analysis if they fulfilled the following criteria: (1) original case-control and cohort studies that evaluated genetic association with age-related cataract susceptibility; (2) samples that consisted of unrelated individuals recruited from welldefined populations; (3) genotype and allele data of both case and control groups provided or calculable from the reported data; and (4) no identified previous systematic review or metaanalysis concerning the polymorphism locus, or the previous meta-analysis needed to be updated.
We excluded studies for the following reasons: (1) animal studies, case reports, reviews, duplicate publications or conference abstracts; (2) diagnosis of cataract not based on objective examination or medical records; and (3) studies published in languages other than English or Chinese.
Study selection, data collection and risk of bias assessment Two reviewers (Xiaohang Wu and Weiyi Lai) screened all the records independently. All disagreements were resolved through discussions with a third reviewer (Haotian Lin). After identifying all the eligible articles, two authors (Xiaohang Wu and Weiyi Lai) extracted the data and compared the results. We did not contact the authors of the eligible articles for additional data. A standardized data extraction form was used that included the first author, year of publication, population ethnicity, population characteristics (mean age and sex ratio, using the control group as the reference), definitions of case and control groups, sample sizes, involved genes and polymorphisms, allelic and genotypic counts of the case and control groups, and the genotype frequencies of different cataract subtypes when provided, for the purpose of stratified analysis. When the allelic counts were not reported, we calculated them using the genotype data. The results of Hardy-Weinberg equilibrium (HWE) test were also extracted from the control group using the chi-square test.
The quality of the included studies was assessed by two reviewers (Xiaohang Wu and Weiyi Lai) according to the Newcastle-Ottawa Scale (NOS) (available at http://www.ohri.ca/ programs/clinical_epidemiology/oxford.asp). This scale uses a domain-based system to assess the quality of a study involving selection, comparability and exposure, with scores ranging from 0 (worst) to 9 (best).

Data analysis
Meta-analysis was conducted for each of the candidate polymorphisms using a model-free approach [12]. No prior assumptions regarding genetic models were made. In brief, values of λ = 0, 0.5, and 1 indicate recessive, codominant and dominant models respectively. With λ higher than 1 or lower than 0, the overdominant model is favored. Other genetic models were also analyzed to aid the comprehensive assessment of the estimated association. The genetic models we used included allelic (a vs A), dominant (aa+Aa vs AA), recessive (aa vs Aa+AA), codominant (aa vs AA and Aa vs AA) and overdominant (aa+AA vs Aa) models, where 'a' and 'A' represent the mutant allele and the wild-type allele, respectively.
The pooled odds ratios (OR) and 95% confidence intervals (95% CI) were calculated for each polymorphism involving multiple genetic models with a fixed-effects model (the Mantel-Haenszel method) or random-effects model (the DerSimonian-Laird method) according to the interstudy heterogeneity. Heterogeneity among the included studies was assessed by the I 2based Q statistic test [13]. I 2 values of 50% or more were considered to indicate substantial heterogeneity, and the random-effects model was then used; otherwise, the fixed-effects model was used. The significance of the pooled OR was determined by Z-test, with p<0.05 considered statistically significant. The Egger's test was used to assess the publication bias with p<0.05 considered statistically significant. We also conducted sensitivity analysis to test the robustness of associations by sequentially omitting each of the included studies one at a time. All the data analysis was performed using the software STATA 13.0 (Stata Corporation, College Station, TX, USA).

Review of the genetic polymorphisms reported in association studies with ARC
Our general literature search of all genetic association studies regarding age-related cataract generated a total of 61 selected studies, involving 144 polymorphisms in 36 genes. All these involved polymorphisms and references are listed in Table 1.

Inclusion of studies for meta-analysis
On the basis of the comprehensive review, three polymorphisms in two genes were extracted (8-oxoguanine DNA glycosylase-1 [OGG1]-rs1052133; methylenetetrahydrofolate reductase NADPH [MTHFR]-rs1801131, rs1801133), which were new for meta-analysis or the previous meta-analysis needed to be updated, and had sufficient genotype data in !3 studies.

Quality assessment
Quality assessments by the NOS scores of the included observational studies are listed in Table 2. All eight eligible studies in the meta-analysis yielded scores !6, indicating relatively high methodological quality.

Genetic associations of the OGG1 gene with ARC
We used a model-free approach (details provided in Methods) to identify the best-fit genetic model. A codominant model was suggested for a λ value of 0.5. This approach revealed that the GG vs. CC genotype (OR = 1.925; 95%CI, 1.181-3.136; p = 0.009; I 2 = 76.3%; Egger's test, p = 0.469) and the CG vs. CC genotype (OR = 1.384; 95%CI, 1.171-1.636; p<0.001; I 2 = 12.1%; Egger's test, p = 0.613) were both significantly associated with an increased risk for ARC (forest plot shown in Fig 2A; sensitivity analysis in Fig 3A; funnel plot in Fig 4A). Subgroup analysis indicated that the significant associations also existed in most of the subtypes based on population ethnicity, article language and cataract morphology (see Table 3 for details). The significant associations were consistently found in some other genetic models, for example, the allelic, dominant, and recessive models for all cases and cortical subtype (for more details, refer to S1 Table). Substantial heterogeneity (I 2 >75%) existed in the overall analysis of the allelic, recessive and codominant (GG vs. CC) models. Regardless, the heterogeneity could be notably controlled in the subgroup analysis by cataract morphology, which represented the preponderant source of the heterogeneity.

Genetic associations of the MTHFR gene with ARC
Both MTHFR genetic polymorphisms, C677T (rs1801133) and A1298C (rs1801131), suggested the overdominant model by the λ calculation. The results of the suggested overdominant model are shown in Table 4. Additionally, the results of other genetic models are shown in S2 Table of the supplementary information.  Fig 3C; funnel plot shown in Fig 4C). However, in the subgroup analysis, this association was found for only the cortical subtype (OR = 0.731; 95%CI, 0.566-0.945; p = 0.017; I 2 = 0.0%; Egger's test, p = 0.599). Regarding other genetic models, associations were found in the dominant and codominant models. The CT+TT vs. CC genotype (OR = 1.313; 95%CI, 1.104-1.562; p = 0.002; I 2 = 44.6%; Egger's test, p = 0.884) and CT vs. CC genotype (OR = 1.317; 95%CI, 1.095-1.584; p = 0.003; I 2 = 0.0%; Egger's test, p = 0.987) were recognized susceptible to ARC. However, the same associations existed in only the cortical subtype in subgroup analysis (for more details, refer to S2 Table). No substantial heterogeneity was found in the associations we identified for this locus.

A1298C (rs1801131)
In the suggested overdominant model, no association was found in the overall and subgroup analyses of cortical, nuclear and posterior subcapsular morphology. However, in subgroup analysis of mixed morphology, the AA+CC vs. AC genotype (OR = 1.517; 95% CI, 1.113-2.067; p = 0.008; I 2 = 34.8%) was found to be possibly associated with an increased risk for ARC.
Moreover, other genetic models revealed no association in overall and subgroup analyses of cortical, nuclear and posterior subcapsular morphology. For mixed morphology, the AC+CC vs. AA genotype (OR = 0.692; 95%CI, 0.513-0.932; p = 0.015; I 2 = 28.0%) and AC vs. AA genotype (OR = 0.657; 95%CI, 0.480-0.900; p = 0.009; I 2 = 37.7%) were both associated with decreased risk for ARC. No substantial heterogeneity was observed in the associations we assessed for this locus.

Combined genotypes
Since the association analysis of the MTHFR gene concerned 2 SNPs (C677T and A1298C), we were interested in the genetic effect of their combined genotype on ARC susceptibility. The wild genotype combination 677CC/1298AA was used as the reference group and the pooled ORs and 95% CIs of other combined genotypes were calculated.

Publication bias and sensitivity analysis
In the sensitivity analysis for all the genetic associations, the ORs were not substantially altered after removing any single studies. However, the following major associations were not robust

Genes and loci most scrutinized in previous association studies regarding ARC
Environmental and genetic factors have been confirmed contributing to the pathogenesis of ARC [69,70]. Genetic polymorphism has been recognized as a component of genetic risk for ARC and many studies have been conducted to identify the associations between genetic polymorphisms and ARC susceptibility [5][6][7][8][9][10][11]. In our systematic review, we summarized the genes/loci that have been studied by other investigators for the first time.
One intense area of study involves the genes of antioxidant enzymes that have roles in cellular defence mechanisms against oxidative stress, such as glutathione S transferase and superoxide dismutase (SOD). Oxidative stress has been well accepted as associated with age-related cataract (ARC) pathogenesis [71]. Specifically, the generation of excessive reactive oxygen OGG1 and MTHFR polymorphisms associated with age-related cataract species (ROS) leads to the abnormal degradation, cross-linking and aggregation of lens proteins, thus contributes to ARC genesis [72]. With impaired balance between the oxidative and antioxidative systems, DNA is damaged by accumulated ROS. Moreover, the DNA damage in the lens epithelium has been demonstrated to be associated with cataractogenesis [73,74]. Therefore, another robust topic in previous association analysis has concerned the DNA repair enzyme genes, such as xeroderma pigmentosum complementation group D (XPD), and X-ray cross-complementing group 1 (XRCC1). For more details, refer to Table 1.

Meta-analyzed genes and loci
Basic on this review, we selected the three polymorphisms of the two different genes (OGG1-rs1052133; MTHFR-rs1801131, rs1801133) that needed to be newly or updated meta-analyzed to undertake a quantitative synthesis. Among them, the OGG1 gene encodes 8-oxoguanine glycosylase-1, a DNA repair enzyme of the base excision repair pathway that repairs oxidative DNA damage [75,76]. Methylenetetrahydrofolate reductase, encoded by the MTHFR gene, controls serum homocysteine concentration, which has been considered associated with ARC susceptibility [77,78]. Many associations were revealed from the five polymorphisms. Regarding rs1052133 in the OGG1 gene, the CG and GG genotypes were both found risky for cataractogenesis, with approximately 1.4-fold and 1.9-fold increased risks, respectively. Alternatively, wild genotype CC was protective.
Regarding the MTHFR gene, the CC+TT genotype of rs1801133 was found to be protective. In contrast, the CT genotype was shown to have an adverse effect. However, the AA+CC genotype of rs1801131 indicated higher risk for mixed morphology cataract susceptibility. Haplotype analysis revealed that the combination of 677CC/1298AC is protective against cortical cataract and posterior subcapsular cataract, and carries approximately a 0.5-fold decreased risk compared with the wild genotype combination (677CC/1298AA).
For most of our major conclusions, no substantial heterogeneity was found, which implied sound quality and consistent methodological design of the included studies. Differences in cataract morphology constitution may be the main source of substantial heterogeneity in most cases. Other potential sources of heterogeneity may derive from some uncontrolled confounding factors, for example, slightly different exclusion criteria of cases and controls, different mean ages and smoking status. Sensitivity analysis revealed that the genetic associations concerning the MTHFR gene were not as robust as the estimates for the other genes, which might be explained by the limited number of original articles.
A recently published study by Zhang et al [79] also examined the association of OGG1 polymorphism with age-related cataract (ARC) and revealed OGG1 polymorphism as a potential risk factor for ARC, in consistent with our findings. However, several discrepancies concerning analytic methodology and research findings could be found in between: (1) our literature search is more thorough, contains two more articles that double the number of cases and controls, thus, provide greater power to our conclusions; (2) we found new association between G allele and increased risk of age-related cataract (ARC) in allelic model (p = 0.008, shown in S1 Table); (3) another new association was found between GG genotype and higher ARC susceptibility, when compared with CC genotype in codominant model (p = 0.009, shown in Table 3). (4) we also found association in CG vs. CC genotype and GG+CG vs. CC genotype in nuclear subgroup in addition to cortical subtype; and (5) we used one more genetic model, the overdominant model, to gain a more comprehensive understanding of the association.

Study strengths and limitations
This systematic review and meta-analysis is the first to provide a relatively thorough summary of the genes/loci involved in previous association studies of ARC. The MTHFR gene and its two polymorphisms were also meta-analyzed for the first time. The results reveal that to a certain degree, all these three genetic polymorphisms and two genes (OGG1-rs1052133; MTHFR-rs1801131, rs1801133) are associated with ARC susceptibility and may help identify high-risk populations in the future.
The review was undertaken using a meticulous methodology. Strict inclusion criteria were adopted. Every included study was of high quality and achieved Hardy-Weinberg equilibrium. A model-free approach was used to suggest a best-fit genetic model, results of which are shown in Tables 3 and 4. Notwithstanding, other genetic models were also used to reach more comprehensive conclusions (results shown in S1 and S2 Tables).
The main limitation of the meta-analysis is that the polymorphisms of the MTHFR gene were studied in a limited number of original articles, which rendered some revealed associations less robust in the sensitivity analysis. Other limitations pertained to the published studies and to our review. Firstly, age-related cataract is a multifactorial disease. Other confounding factors such as ultraviolet light exposure and smoking may also influence the association analysis. However, the included studies did not provide detailed records of these confounding factors. Thus, the associations we found in our review may be strengthened or weakened by these confounders. We anticipate that future studies will direct greater attention to these influences when possible. Secondly, the ethnicity involved in our review is limited. For example, four of the five included studies of rs1052133 in OGG1 gene focused on the Han Chinese population, which restricts the applicability of our conclusions to a certain ethnicity. The emergence of future studies that focus on other ethnicities will facilitate the determination of the associations in other populations. Thirdly, the precise mechanisms of the genetic effects we observed remain unknown. Studies of underlying mechanisms are needed.
In summary, we consider these polymorphisms to represent new candidate biomarkers for high-risk ARC population. However, additional original research with larger sample sizes, high quality, and broader ethnicity coverage remain anticipated.