Study selection and characteristics in our meta-analysis
We established databases according to the extracted information from each article. Using the keywords, we identified 353 articles from PubMed, 56 from CNKI and 362 from Wanfang databases. 547 articles were excluded after reading abstract section and 224 articles were left for full article evaluation. Among them, 38 article were about systematic analysis/meta-analysis/review; 24 just only were offered case group; 16 articles were duplicated with including other papers; 34 had no original numbers for case/control groups, only showed total numbers; 27 article were about H-type hypertension; 4 were related to aortic hypertension and 26 were hypertension-in-pregnancy (Figure 1). After above review, 54 articles about 55 case-control studies were left, in which 5 case-control studies were not consistent with HWE and were excluded, finally, 50 case-control studies from 54 different literatures were included in our current analysis. All essential information was listed in Table 1. Table 1 showed first author, publishing year, race, the numbers of cases and controls, HWE, genotype numbers in cases/controls, study design and genotype methods. So in our study, there are 10 European case-control studies, 36 Asian case-control studies and 4 Mixed. The T frequency in Asians was 45.55%, in Europeans was (47.88%), and in Mixed was (47.38), which did not exist statistically significant (P > 0.05). The distribution of genotypes in all the controls was agreement with HWE. In addition, we checked the Minor Allele Frequency (MAF) reported for the seven main worldwide populations in the 1000 Genomes Browser (https://www.ncbi.nlm.nih.gov/snp/rs1801133): Global (0.335); European (0.345); East Asian (0.328); South Asian (0.167); African (0.123); African American (0.125); Asian (0.265) (Figure 2A). In order to observe the frequency of T-allele and C-allele both in case and control groups, we analyzed and found the frequency between case and control was pretty much the same (Figure 2B). Finally, we analyzed the trend of rs1801133 polymorphism from TCGA database, TT (AA) frequency was relatively low than other genotypes (Figure 2C). This polymorphism is associated with coronary artery, rather than aorta artery left ventricle and tibial artery (https://www.gtexportal.org/home/) (Figure 2D).
Quantitative data synthesis
Table 2 showed that the summary odds ratios of MTHFR based on10533 EH cases and 11743 matched controls, we observed increased association between the MTHFR rs1801133 polymorphism and EH in total population (for example: T-allele vs. C-allele: OR=1.37, 95%CI=1.24-1.52, Ph<0.001, p<0.001, I-squared=82.1%, Figure 3A). Then, subgroup by ethnicity analysis, similar trend was also observed (T-allele vs. C-allele: OR=1.46, 95%CI=1.29-1.67, Ph <0.001, p<0.001, I-squared=84.8% for Asian; OR=1.27, 95%CI=1.07-1.51, Ph =0.008, p=0.007, I-squared=59.7% for European; OR=1.51, 95%CI=1.30-1.74, Ph <0.001, p<0.001, I-squared=86.9% for China and OR=1.21, 95%CI=1.06-1.37, Ph <0.001, p=0.003, I-squared=66.4% for non-China, Figure 3B). In order to analyze the source of control and find the source of heterogeneity, HB and PB subgroups were calculated, significant increased relationships were shown (T-allele vs. C-allele: OR=1.49, 95%CI=1.28-1.75, Ph <0.001, p<0.001, I-squared=85.4% for HB; T-allele vs. C-allele: OR=1.27, 95%CI=1.12-1.44, Ph <0.001, p<0.001, I-squared=77.2% for PB, Figure 3C). Different methods for detecting this polymorphism were applied in all including studies, we tried to find whether positive results may be observed, finally, some significant findings were found, such as PCR (T-allele vs. C-allele: OR=1.48, 95%CI=1.14-1.91, Ph <0.001, p=0.003, I-squared=85.1%), PCR-RFLP (T-allele vs. C-allele: OR=1.49, 95%CI=1.30-1.72, Ph <0.001, p<0.001, I-squared=65.1%) and HRM(T-allele vs. C-allele: OR=1.32, 95%CI=1.15-1.51, Ph <0.001, p<0.001, I-squared=47.5%) (Figure 3D, E).
Publication bias and sensitive analysis
The Begg’s funnel plot and Egger’s test were performed to access the publication bias of literature. Significantly obvious evidence of publication bias was detected in five genetic model analysis (such as Figure 4 A, B about T-allele vs. C-allele) (Table 3).
To delete studies which may influence the power and stability of the whole study, the sensitive analysis was applied, finally, no sensitive case-control studies were found for this SNP in five models (such as Figure 4C about T-allele vs. C-allele).
The analysis showed significant relationship for allele model (T-allele vs. C-allele) for the ethnicity, source of control and genotype methods with a regression coefficient of 0.001, 0.004, 0.010 and 0.002, respectively, rather than publication year, which means the heterogeneity from rs1801133 polymorphism in EH may be from the ethnicity, source of control and genotype methods subgroups (Figure 5A-E).
Gene-gene network diagram and interaction of online website
String online server indicated that MTHFR gene interacts with numerous genes. The most ten significant genes from network of gene-gene interaction has been listed in Figure 6A,B. The ten genes are: methionine (MTR); thymidylate synthase (TYMS); C-1-tetrahydrofolate synthase (MTHFD1); serine hydroxymethyltransferase 1 (SHMT1); serine hydroxymethyltransferase 2 (SHMT2); bifunctional methylenetetrahydrofolate dehydrogenase (MTHFD2); probable bifunctional methylenetetrahydrofolate dehydrogenase (MTHFD2L); aminomethyltransferase (AMT) and methionine synthase reductase (MTRR).