A Meta-Analysis of the Association between DNMT1 Polymorphisms and Cancer Risk

Previous studies have examined the associations of DNA methyltransferase 1 (DNMT1) polymorphisms, including single nucleotide polymorphisms rs16999593 (T/C), rs2228611 (G/A), and rs2228612 (A/G), with cancer risk. However, the results are inconclusive. The aim of this meta-analysis is to elucidate the associations between DNMT1 polymorphisms and cancer susceptibility. The PubMed, Embase, Web of Science, and Chinese National Knowledge Infrastructure databases were searched systematically to identify potentially eligible reports. Odd ratios and 95% confidence intervals were used to evaluate the strength of association between three DNMT1 polymorphisms and cancer risk. A total of 16 studies were finally included in the meta-analysis, namely, nine studies of 3378 cases and 4244 controls for rs16999593, 11 studies of 3643 cases and 3866 controls for rs2228611, and three studies of 1343 cases and 1309 controls for rs2228612. The DNMT1 rs2228612 (A/G) polymorphism was significantly related to cancer risk in the recessive model. The meta-analysis also suggested that DNMT1 rs16999593 (T/C) may be associated with gastric cancer, while rs2228611 (G/A) may be associated with breast cancer. In future research, large-scale and well-designed studies are required to verify these findings.


Introduction
DNA methylation is one of the most commonly occurring epigenetic events in the mammalian genome. DNA methyltransferases (DNMTs) are critical to establishing and maintaining DNA methylation patterns by converting cytosine residues to 5-methylcytosine (5mC) in cytosine-guanine (CpG) dinucleotides [1,2]. The DNMT family comprises three active forms: DNMT1, DNMT3A, and DNMT3B. DNMT1 is the major enzyme responsible for methylation maintenance [3,4], while DNMT3A and DNMT3B are thought to function in de novo methylation rather than in methylation maintenance [3,4].
DNMT1 is located on human chromosome 19p13.2 and encodes a protein comprising 1632 amino acids, which may be implicated in occurrence progression and prognosis of the cancer. DNMT1 consists of three major structural elements: an N-terminal regulatory domain that is necessary for the localization of DNMT1; a C-terminal catalytic domain that is involved in the binding of substrates; and a central linker that contains repeated glycine-lysine dipeptides [5]. Genetic variation caused by single nucleotide polymorphisms (SNPs) is the most common form of altered gene structure. The most commonly studied DNMT1 SNPs are rs16999593 (T/C), rs2228611 (G/A), and rs2228612 (A/G), which are present in coding regions and, therefore, may influence DNMT1 expression.

BioMed Research International
Until now, no meta-analysis has been carried out to investigate the relation of the three DNMT1 polymorphisms (rs16999593 (T/C), rs2228611 (G/A), and rs2228612 (A/G)) with risk of cancer. Individual studies have lacked the ability to obtain overall reliable conclusions because of limited sample sizes and/or variations in ethnicities. To obtain further insights into the roles of DNMT1 polymorphisms in carcinogenesis, we mainly performed a meta-analysis on the associations between these three SNPs and cancer risk.

Search Strategy.
We systematically searched the PubMed, Embase, Web of Science, and Chinese National Knowledge Infrastructure databases using different combinations of the search terms "DNMT1 or DNA methyltransferase 1," "polymorphism or mutation or variant," and "cancer or neoplasm or tumor." The search was last updated on 06 June 2016. When overlapping data were found, only the largest and latest study was selected. We contacted the authors and requested their specific raw data when the data provided in the published article were not sufficient.

Inclusion and Exclusion
Criteria. Studies were included when they met the following criteria: case-control study; the subject was the association of DNMT1 polymorphisms (rs16999593 (T/C), rs2228611 (G/A), and rs2228612 (A/G)) with risk of cancer; and essential information on genotype or allele frequencies was available to assess the odds ratios (ORs) and 95% confidence intervals (CIs). Exclusion criteria included review articles; systematic reviews and metaanalyses; animal studies; sample size less than 100; and scarce or insufficient information on genotype or allele frequencies for the rs16999593 (T/C), rs2228611 (G/A), or rs2228612 (A/G) polymorphisms of DNMT1 despite us contacting the authors.

Data Extraction.
Two of the authors (HL and JL) independently selected the articles and extracted the original data using a standardized and consistent method. The following information was collected from each study: first author, year of publication, ethnicity of the subjects, cancer type, numbers of cases and controls, and genotyping methods. Conflicts were resolved after discussion and consensus was finally reached on all the extracted information.

Statistical Analysis.
All statistical analyses were conducted using STATA software (version 12.0; Stata Corp LP, College Station, TX). ORs and their corresponding 95% CIs were employed to assess the strength of relationships between the DNMT1 polymorphisms and cancer risk. values < 0.05 were considered as statistically significant. Heterogeneity was calculated using the statistic ( value < 0.10 indicates significant heterogeneity among studies) andsquared value. The Mantel-Haenszel fixed-effects model was used to calculate the pooled ORs when the heterogeneity of studies was not significant. Otherwise, the DerSimonian and Laird random-effects model was used. We conducted the sensitivity analysis to explore heterogeneity when significant heterogeneity existed. Subgroup analysis was applied to explore the effects of cancer type and genotyping method. In addition, Begg's test and Egger's test were performed to evaluate publication bias; values < 0.05 for Begg's and Egger's tests indicate significant publication bias.  Figure 1. A total of 215 potentially relevant articles were found by searching the four databases and after removing duplicates. Altogether 187 publications were excluded mainly due to no relevance, animal not human experiments, reviews, or meeting abstract. The 28 remaining articles were evaluated further for eligibility. Finally, 16 articles were included in the present meta-analysis [6][7][8][9][10][11][12][13][14][15][16][17][18][19][20][21].
The results for the association of DNMT1 rs2228612 (A/G) with cancer risk are summarized in Table 4 and Figure 4. Significant association with cancer risk was observed in the recessive model (GG versus AG + AA: OR = 1.29, 95% CI = 1.06-1.56, = 0.009), but no significant association was revealed in the allele analysis (G allele versus A allele: OR = 1.00, 95% CI = 0.83-1.20, = 0.980).

Heterogeneity Test and Sensitivity Analysis.
In most of the comparisons of DNMT1 rs16999593 and rs2228612 polymorphisms and one comparison of DNMT1 rs2228611, significant heterogeneity was observed. We next performed a leave-one-out sensitivity analysis. The results show that no individual study significantly affected the pooled OR (figure not shown), suggesting that the results of the meta-analysis were robust.

Publication Bias.
Begg's test and Egger's test were used to quantitatively evaluate the publication bias of the selected studies; the details are listed in Table 5. For the associations of DNMT1 rs16999593 (T/C), rs2228611 (G/A), and rs2228612 (A/G) with cancer risk, rs2228612 (A/G) showed publication bias (GG versus AA; G allele versus A allele) and rs16999593 (T/C) showed publication bias in the comparison with the recessive model.

Discussion
DNMT1, the major methyltransferase in mammals, lies in the replication fork and methylates newly synthesized DNA strands directly in S phase of DNA replication sites [22], which is essential for epigenetic inheritance. Regional aberrant DNA hypermethylation has been identified as a possible mechanism of inactivation of tumor suppressor genes [23]. Many studies have indicated that the overexpression of DNMT1 could silence vital tumor suppressor genes such as APC, P16, and RUNX3 through DNA methylation [24,25]. Therefore, DNMT1 might be implicated in the occurrence, development, and prognosis of multiple types of cancer.
Polymorphisms have been identified as a powerful tool for predicting hereditary susceptibility of some complex diseases including cancer. However, previous individual studies about the association between DNMT1 polymorphisms and cancer risk were not only limited but also inconclusive. To our knowledge, this is the first comprehensive meta-analysis investigating the possible correlations of SNPs rs16999593 (T/C), rs2228611 (G/A), and rs2228612 (A/G) in DNMT1 with risk of overall cancer and specific cancer types, which is anticipated to shed light on the role of DNMT1 polymorphisms in carcinogenesis.
SNP rs2228612 (A/G) causes an isoleucine to phenylalanine substitution at amino acid 327 in the DNMT1 protein,   which may alter the function of DNMT1 and influence its effect in the carcinogenesis. In this meta-analysis, we found that DNMT1 rs2228612 (A/G) was associated with risk of overall cancer in the recessive model. However, only three articles with small-scale studies were available for analysis; therefore, the results should to be interpreted with caution.
We did not find significant association between SNP rs16999593 (T/C) and overall cancer risk in any genetic comparison. Different types of cancer have distinct initiation and progression mechanisms, in which polymorphisms in key genes play critical roles. This meta-analysis elucidated that the DNMT1 rs16999593 (T/C) polymorphism was associated with different cancer types. In the subgroup analysis according to cancer type, the TC genotype of rs16999593 (T/C) was associated with risk of gastric cancer, but not breast cancer. Extensive evidence has suggested that DNA methylation is involved in the initiation and progression of gastric cancer and increased expression of DNMT1 had been confirmed to be related to gastric cancer [25]. The AKT-NF B and STAT3 signaling pathways can upregulate DNMT1 expression, which could cause aberrant DNA methylation of tumor suppressor genes and lead to gastric cancer [26,27]. Therefore, SNP rs16999593 (T/C), which causes a histidine to arginine substitution at 97 positions of the amino acid  [15] and infiltrating ductal beast carcinoma (IDBC) [6]. The TC genotype of rs16999593 (T/C) was related to increased TNBC risk but decreased IDBC risk. Therefore, DNMT1 might have diverse functions in different types of breast cancer. Further studies of the effects of DNMT1 polymorphisms on specific breast cancer types are still needed. In addition, only in the subgroup of sequencing, DNMT1 rs16999593 (T/C) were constantly associated with increased cancer risk in all compared genetic models, indicating that different genotype detecting methods might influence the results. According to the functional prediction tool F-SNP (http://compbio.cs.queensu.ca/F-SNP/), SNP rs2228611 (G/A) may change the regulation of DNMT1 splicing by leading to a synonymous mutation at amino acid 463. Multiple transcript variants of DNMT1 gene as a result of alternative splicing have been found. Therefore, we speculated that the rs2228611 (G/A) might influence the process of carcinogenesis by regulating the pattern of alternative splicing of DNMT1. Here, we did not find any significant associations between DNMT1 rs2228611 (G/A) with altered risk of cancer in any genetic model. However, in the subgroup analysis, individuals with the GA genotype of rs2228611 (G/A) were associated with higher risk of breast cancer in both heterozygous and dominant models. In addition, individuals with the GA genotype of rs2228611 (G/A) had decreased risk in one study on esophageal cancer [14], which was opposite to the results for breast cancer. This reverse outcome may be because various types of cancer have different mechanisms of carcinogenesis. Future studies on different types of cancer may help to better understand these findings.
Some limitations of our meta-analysis should be noted. Firstly, the number of studies was not sufficiently large, especially for subgroup analysis of DNMT1 rs2228612 (A/G). Secondly, the languages of the publications were limited to English and Chinese. Thirdly, although this meta-analysis was based on the whole population, most studies were from Chinese populations, except for one Iranian case and one Polish case for rs2228611 (G/A) and another one Caucasian case for rs2228612 (A/G), which showed the same results with the Chinese population. So this study had a certain universality, especially for Chinese population. Finally, publication bias was found in two comparisons of rs2228612 (A/G) and one comparison of rs16999593 (T/C).

Conclusion
The DNMT1 rs2228612 (A/G) GG genotype may be associated with increased risk of cancer compared with the AA + AG genotype. SNP rs16999593 (T/C) is significantly associated with gastric cancer risk while SNP rs2228611 (G/A) is significantly related to breast cancer risk. Further large-scale and well-designed investigations including different cancers are required to verify the findings of this study.