Is MTHFD1 polymorphism rs 2236225 (c.1958G>A) associated with the susceptibility of NSCL/P? A systematic review and meta-analysis

Aims: To investigate the association between the methylenetetrahydrofolate dehydrogenase 1 (MTHFD1) polymorphism rs 2236225 (c.1958G>A) and susceptibility to non-syndromic cleft of the lip and/or palate (NSCL/P). Methods: An extensive literature review has been conducted using PubMed, Web of Science, Cochrane Library, Google Scholar, the China National Knowledge Infrastructure (CNKI), and Wanfang Database for eligible researches. The terms for searching were “cleft lip OR cleft palate OR CLP OR CL/P OR oral facial cleft OR OFC” AND “methylenetetrahydrofolate dehydrogenase (NADP+ dependent) 1 OR methenyltetrahydrofolate cyclohydrolase formyltetrahydrofolate synthetase OR MTHFD1 OR MTHFD”. Two independent researchers screened, evaluated and extracted the data of included studies. The pooled odds ratios (OR) with 95% confidence intervals (95% CI) were calculated by random effects model under five gene models. Subgroup, sensitivity analysis and publication bias were also assessed. Results: Ten case-control studies have been included in the systematic review and eight studies have been considered for the meta-analysis. Overall, the MTHFD1 polymorphism rs2236225 and the risk of NSCL/P showed pooled OR (95% CI) of 1.02 (0.86-1.21) under allelic model. A higher degree of heterogeneity was observed in Asian countries (I 2 = 75.6%) compared to non-Asian countries (I 2 = 48.9%). Similar consequence appeared in the subgroup of children (I 2 = 78.6%) compared with that of mothers (I 2 = 0.0%). There was no significant difference in the publication bias by the Begg’s funnel plot (P = 0.711) and Egger’s regression test (P = 0.746). Conclusion: Our assessment suggested there was no significant association between the MTHFD1 polymorphism rs 2236225 (c.1958G>A) and the susceptibility to NSCL/P. Further investigations using a large sample size and a more advanced technique should be adopted to reach a more precise conclusion in the future.


Introduction
Cleft of the lip and/or palate (CL/P) is one of the most common facial malformations 1-3 and a societal burden, affecting the patient ability to eat and speak and influencing social integration 4 . Nonsyndromic CL/P, accounting for about 70% of CL/P, is considered closely related to genetic and environmental factors 5 . Recent studies suggested that using folic acid could reduce the rates of oral clefts 6,7 and single nucleotide polymorphisms of some genes such as MTHFR 8,9 ,MTR 40 and MTRR involved in the metabolism of folic acid have been associated to high risk of NSCL/P 8,9 . Methylenetetrahydrofolate dehydrogenase 1 (MTHFD1), a key gene associated with three sequential enzymatic reactions in the metabolism of folic acid, might play a potential role in the risk of NSCL/P, especially the polymorphism rs2236225 (c.1958G>A) 10 . Indeed, different observations that linked the polymorphism rs2236225 to the risk of NSCL/P have been reported 11,12 . The suggestion of a link between rs2236225 polymorphism and susceptibility to NSCL/P might be result of the limitations in sample size, different ethnic populations and other environmental factors. Therefore, we conducted a systematic review and meta-analysis of eligible case-control studies to reveal a more precise connection between the MTHFD1 polymorphism rs2236225 and the risk of NSCL/P.

Materials and methods
Identification of studies A systematic search based on the principle of evidence-based medicine 13 was performed in PubMed, Web of Science, Cochrane Library, Google Scholar, China National Knowledge Infrastructure (CNKI) and WanFang Database. The final update was made on April 5th, 2015. In line with our knowledge background, the Medical Subject Headings (MESH) terms in PubMed and the known aliases of the genes of interests in the National Center of Biotechnology Information (NCBI), the following terms were used for searching: "cleft lip OR cleft palate OR CLP OR CL/P OR oral facial cleft OR OFC" AND "methylenetetrahydrofolate dehydrogenase (NADP+ dependent) 1 OR methenyltetrahydrofolate cyclohydrolase formyltetrahydrofolate synthetase OR MTHFD1 OR MTHFD", which were slightly adjusted to optimize search results (Table S1; PubMed). We didn't limit the search depending on publication types, data and language. Of course, the review of the published literature was examined carefully and manual search was conducted to avoid missing potential data. Two of the authors (Huaxiang Zhao and Mengqi Zhang) were in charge of the search independently and a third author (Jieni Zhang) conducted a random inspection.

Inclusion and exclusion criteria
Researches included in our systematic review and meta-analysis meet the following criteria: (1) evaluating the association between the NSCL/P and MTHFD1 polymorphism rs2236225, (2) focusing on humans, (3) case-control studies. Exclusion criteria were: (1) no association between NSCL/P and MTFHD1, (2) not focusing on humans but animal models or in vitro studies, (3) duplication of previous researches, (4) not original literature such as reviews, meta-analyses, comments and editorials.

Data collection
Data from eligible studies were extracted by two independent researchers (Huaxiang Zhao and Mengqi Zhang) in accordance with the inclusion and exclusion criteria. In case of any discrepancies, the third chief author (Feng Chen) would make a further investigation or bring it into a group-discussion. A special table was used for collecting information from the selected articles and the following entries were recorded: authors (year), country, location of geography, subjects, methods for genotyping, sample size of cases/controls, descriptions of samples rolled in the study, P for HWE (Hardy-Weinberg equilibrium) of control group, whether included in the meta-analysis or not.
Methodological quality assessment A methodological quality assessment adapted from previous studies 14-16 was carried on included studies (Table S2). Cases, source of controls, sample sizes and Hardy-Weinberg equilibrium (HWE) were considered as important aspects in this systematic review. It is not a widely-used standard so its efficiency is not certain. The result does not include the judgments but the standard can be a reference for the certainty of conclusion.

Statistical analysis
The PRISMA checklist (Supplementary material S3) was used as a protocol in our meta-analysis 17 . Odd ratios (ORs) and 95% confidence intervals (CIs) were calculated to estimate the association between the susceptibility to NSCL/P and MTHFD1. Five genetic models were used in the process of pooling the OR and 95% CIs: allelic comparison (A versus G), heterozygote model (AG versus GG), homozygote model (AA versus GG), dominant model (AA + AG versus GG), recessive model (AA versus AG + GG). The significance of the pooled effects was determined by Z-test with P value less than 0.05. The Q-statistic and the I 2 test were used to evaluated; P < 0.05 in Q statistic or I 2 > 50% 18,19 , would indicate a significant heterogeneity. When P > 0.05 in Q statistic or I 2 < 50%, the fixed pooling model (Mantel-Haenszel) was conducted; if not, the

Amendments from Version 1
The following modifications have been made in the newest version: 1. All the gene acronyms were re-written italics; 2. Table S2 was an assessment standard for included articles but the assessment scores were not included in our article, we explained this matter in Materials and methods --Methodological quality assessment; 3. Page 6, Figure 1, Eligibility section, the number of full-text articles excluded with reasons were corrected with 5; 4. The publication bias analysis when < 10 articles is not statistically resolved (see Egers's paper), so a comment about this issue was added in Results --Sensitivity analysis and publication bias and the fourth paragraph in Discussion 5. In order to reduce the repeat content in the first two paragraph of Discussion and Introduction, we deleted several sentences in the Discussion; 6. In the third paragraph of Discussion, a description of the controversial conclusions of included studies was made to make the discussion more interesting. 7. Based on our result, we deeply explained our conclusion and put out suggestion about sample taking in future studies in the last paragraph. We also carried subgroup analyses in which different subjects (mothers or children), location of geography (non-Asian countries or Asian countries) were considered potential source of heterogeneity. A sensitivity analysis was conducted by omitting each study in turn to evaluate the single study's influence on the overall estimation. We used Begg's funnel plot and Egger's linear regression test to find out the publication bias of the included studies [20][21][22] . The studies with disequilibrium of HWE among control group were added into a supplementary meta-analysis as described previously 23 . Meanwhile, as for the studies included but not carried into the meta-analysis, to achieve a qualitative analysis we adopted a method described by others 24 . Results were considered significant when P < 0.05. Stata 12.0 (Stata Corp, College Station, TX, USA) was used for the analysis.

Data retrieval
A total of 251 articles resulted from the search described above (PubMed: 86, Web of Science: 8, Google Scholar: 135, Cochrane Library: 0, CNKI: 18, Wanfang: 4). After being imported into End-Note X6 software (Thomson Corporation, Stamford), a screening process was conducted among 102 articles-that is, duplicates were removed using the 'Discard Duplicates' function as well as by handwork. Following paper selection by two independent researchers, 15 studies were then thoroughly reviewed. Of these, five studies were excluded, among which two had no control groups 25,26 , one no relation to MTFHD1 27 , and the other two presented data previously published 28,29 . Finally, 10 studies that met the criteria were included in the systematic review (Table 1)  NM ψ : Not mentioned in the study.
from eight studies were used for reference to carry out the metaanalysis 10-12,31-33,35,36 . The selection process is shown in Figure 1.

Study characteristics
Eventually, all 10 studies containing 6216 samples (2959 cases and 3257 controls) were analyzed in our review. The characteristics of every study can be seen in Table 1. To summarize briefly, there were four studies from European groups, four from Asian groups and two from South American groups, among which two studies focused on the genotype of patients' mothers only, four on children' s genotype only and four on both of them. PCR-restriction fragment length polymorphism (PCR-RFLP) was the major method of genotyping, while other techniques had been used as well.
Next we conducted the subgroup analysis using allelic A versus G model according to the location of geography and subjects (mothers or children). It turned out that there was no significant difference between Asian (OR = 1.03, 95% CI 0.75-1.40, P H = 0.003) or non-Asian population (OR = 1.06, 95% CI 0.86-1.30, P H = 0.118). However, a higher degree of heterogeneity was observed in the Asian countries compared to non-Asian countries ( Figure 4A). A similar result was observed in the subgroup analysis between mothers and children. The heterogeneity was much higher in the children group (OR = 0.99, 95% CI 0.72-1.36, P H = 0.001) than in the mothers' group (OR = 1.11, 95% CI 0.98-1.27, P H = 0.630), while no significant difference was observed in both groups ( Figure 4B).

Sensitivity analysis and publication bias
To access the influence of each individual study on the pooled ORs, a sensitivity analysis was performed by omitting each study at a time. The results of sensitivity suggests that no individual study affects the pooled ORs of the associations between MTHFD1 polymorphism rs2236225 (c.1958G>A) and NSCL/P risk under allelic model ( Figure 5).
We used the Begg's funnel plot and Egger's regression test (both used the allelic A versus G model) to estimate the publication bias.
Our results indicate that there is no significant publication bias both in the symmetry of Begg's funnel plot (P = 0.711, Figure 6) and Egger's regression test (P = 0.746). But due to the sample size (the total number of included article is less than 10), statistical analyses may not describe the publication bias precisely. The test reliability is in doubt.

Discussion
CL/P is one of the most common facial malformations, affecting approximately 1.7/1000 people around the world with ethnic and geographic variation 1 . Approximately 70% of CL/P cases are considered to be non-syndromic 37,38 , and their susceptibility has been linked to the expression of various candidate genes through twin studies, familial clustering studies and genome-wide studies 39 . Recent studies suggest that using folic acid could reduce the rates of oral clefts 6,7 and genes involved in the metabolish of folic acid have been identified MTHFD1, a crucial gene associated with three sequential enzymatic reactions among 5,10-methylenetetrahydrofolate, 5,10-methenyltetrahydrofolate, 10-formyltetrahydrofolate, tetrahydrofolate, might play a potential role in NSCL/P 10 . However, controversial results about the MTHFD1 polymorphism rs2236225 (c.1958G>A) have been reported in different articles 10,12 .
In this systematic review, 10 independent case-control studies were included (eight studies for meta-analysis and two studies qualitatively analyzed) containing 6216 samples ( (2013) Mostowska et al (2006) Bufalino et al (2010) Mills et al (2008) Murthy et al (2014) Li et al (2013) ID Yuan (2013) Mostowska et (2010 (2014) Li et al (2013) Study Yuan (2013) Mostowska et (2010) Zhao (2013) Bufalino et al (2010) Mills et al (2008 (2013) Zhao (2013) Li et al (2013) ID Mills et al (2008) Bufalino et al (2010) Yuan (2013) Mostowska et (2010     NSCL/P, whether in the whole analysis of five model (A versus G, AG versus GG, AA versus GG, AA + AG versus GG, AA versus AG + GG) or in the subgroup of subjects (mothers or children) and the location of geography (non-Asian countries or Asian countries). Meanwhile, high heterogeneity was observed, which might be the reason for the genetic drift and natural selection among different ethnic groups 42 . Also, small sample size of different studies might be a possible reason for the disparate results. Our findings suggest that the MTHFD1 polymorphism rs2236225 (c.1958G>A) might not be an appropriate biomarker in predicting the susceptibility of an individual to NSCL/P. Some limitations of this systematic review and meta-analysis should be noted. Firstly, the choice of retrospective studies has its own limitations, as we may encounter selection bias and influence the results of our analysis 43 . However, a bigger size of cohort study cannot be conducted easily because of the relatively low morbidity 44 . Secondly, only 10 studies were included in our review, a small sample size that might not provide sufficient evidence to estimate the connections between the MTHFD1 polymorphisms and the risk of NSCL/P. Thirdly, the publication bias cannot be effectively analyzed because of the limited amount of included study.
NSCL/P is also associated with gene-gene and gene-environment interactions 45 . Although no correlation was observed between MTFHD1 polymorphism rs2236225 (c.1958G>A) and the risk of NSCL/P, in view of MTFHD1 gene's key role in folic acid metabolism, we cannot draw a definite conclusion that there is no association between MTFHD1 and NSCL/P's susceptibility. The use of larger sample size studies, different techniques and considering gene-gene or gene-environment interactions should be explored in future investigations. What is more, the gene samples from mother were too scarce to be representative and to explain our results. We do recommend more samples from parents in the future studies, which is significant for the early stage diagnose, as the current technology can only diagnose CLP in the midterm even later in the pregnancy.
Author contributions H Zhao, F Chen, J Lin were responsible for study conception and design of the study. H Zhao, J Zhang, M Zhang acquired the data. H, Zhao, F Chen F, Deng, L Zheng, H Zheng analyzed the data. H Zhao and J Zhang wrote the main manuscript text. Prof. J Lin and Prof. F Chen had full access to all of the data in this review and take responsibility for the integrity of the data and the accuracy of the data analysis. All authors have agreed to the final content of the manuscript.

Competing interests
No competing interests were disclosed.  none Eligibility criteria 6 Specify study characteristics (e.g., PICOS, length of follow-up) and report characteristics (e.g., years considered, language, publication status) used as criteria for eligibility, giving rationale.

5~6
Information sources 7 Describe all information sources (e.g., databases with dates of coverage, contact with study authors to identify additional studies) in the search and date last searched.

5
Search 8 Present full electronic search strategy for at least one database, including any limits used, such that it could be repeated. Table S1 Study selection 9 State the process for selecting studies (i.e., screening, eligibility, included in systematic review, and, if applicable, included in the meta-analysis).

5~6
Data collection process 10 Describe method of data extraction from reports (e.g., piloted forms, independently, in duplicate) and any processes for obtaining and confirming data from investigators. Data items 11 List and define all variables for which data were sought (e.g., PICOS, funding sources) and any assumptions and simplifications made.

Risk of bias in individual studies
12 Describe methods used for assessing risk of bias of individual studies (including specification of whether this was done at the study or outcome level), and how this information is to be used in any data synthesis.
6 and Table S2 Summary measures 13 State the principal summary measures (e.g., risk ratio, difference in means). 6~7 Synthesis of results 14 Describe the methods of handling data and combining results of studies, if done, including measures of consistency (e.g., I 2 ) for each meta-analysis.

6~7
Risk of bias across studies 15 Specify any assessment of risk of bias that may affect the cumulative evidence (e.g., publication bias, selective reporting within studies).

Results
Study selection 17 Give numbers of studies screened, assessed for eligibility, and included in the review, with reasons for exclusions at each stage, ideally with a flow diagram.
8 and Figure 1 Study characteristics 18 For each study, present characteristics for which data were extracted (e.g., study size, PICOS, follow-up period) and provide the citations. Table 1 Risk of bias within studies 19 Present data on risk of bias of each study and, if available, any outcome level assessment (see item 12).
9~10, Table S2 Results of individual studies 20 For all outcomes considered (benefits or harms), present, for each study: (a) simple summary data for each intervention group (b) effect estimates and confidence intervals, ideally with a forest plot.

Discussion
Summary of evidence 24 Summarize the main findings including the strength of evidence for each main outcome; consider their relevance to key groups (e.g., healthcare providers, users, and policy makers).

11~12
Limitations 25 Discuss limitations at study and outcome level (e.g., risk of bias), and at review-level (e.g., incomplete retrieval of identified research, reporting bias).

12
Conclusions 26 Provide a general interpretation of the results in the context of other evidence, and implications for future research. 13

Funding
Funding 27 Describe sources of funding for the systematic review and other support (e.g., supply of data); role of funders for the systematic review.

15
I think that the Discussion could include a comment about maternal genotype effects in order to explain the negative results of this analysis.
I have read this submission. I believe that I have an appropriate level of expertise to confirm that it is of an acceptable scientific standard, however I have significant reservations, as outlined above.
No competing interests were disclosed.

Competing Interests:
Author Response 26 Nov 2015 , School of Stomatology, Peking University, China Feng Chen Thank you so much for your precious advice. We have studied your comments carefully and revised the manuscript. The point to point responds are listed as following: We have corrected all the inappropriate format of gene acronyms; Our title says "Is MTHFD1 polymorphisms rs 2236225 (c.1958G>A)…", which refers to a specific SNP in accordance with the content; The traditional meta-analysis focuses on the clinical RCTs, which aims to reveal the relevance between specific interruption and clinical effects. A complete assessment standard for the included RCT reports has been widely accepted and used. While our study material is totally different from RCTs thus there is no clear assessment standard to refer. So we do not discuss the quality analysis. But in order to provide the readers an assessment standard we provide the tables 2 in supplementary materials for reference.
We agree that the usual method to evaluate publication bias is not suitable for our study because of the limited amount, which reduces the reliability of our conclusion. We have declared that in the discussion of our updated version.
Actually the Discussion is an overview of the article, which includes and further explains the content above. While in avoidance of repetition we modified and simplified the first and second paragraph of the Discussion.
Two of our included studies only analyzed the maternal genotype, one study reported the mother and children genotype and three took samples from children and their parents. The gene samples from mother were too scarce to be representative and to explain our results. We do recommend more samples from parents in the future studies, which is significant for the early stage diagnose, as the current technology can only diagnose CLP in the midterm even later in the pregnancy.
No competing interests were disclosed.