Meta- and Pooled Analyses of the Methylenetetrahydrofolate Reductase C677T and A1298C Polymorphisms and Gastric Cancer Risk: A Huge-GSEC Review

¹ Institute of Hygiene, Università Cattolica del Sacro Cuore, Rome, Italy
² Genetic Epidemiology Cluster, International Agency for Research on Cancer, Lyon, France
³ Department of Medicine II, Klinikum rechts der Isar, Technical University of München, München, Germany
⁴ Shanghai Institute of Digestive Disease, Shanghai Jiao-Tong University School of Medicine, Shanghai, China
⁵ Department of Epidemiology, Jiangsu Province Institute of Cancer Research, Nanjing, China
⁶ Division of Medical Oncology, Azienda Ospedale "San Salvatore," Pesaro, Italy
⁷ Andalusian School of Public Health, Granada, Spain
⁸ Department of Etiology and Carcinogenesis, Cancer Institute, Chinese Academy of Medical Sciences and Peking Union Medical College, Beijing, China
⁹ Mexico National Institute of Public Health, Cuernavaca, Mexico
¹⁰ Department of Epidemiology, Cancer Institute, Chinese Academy of Medical Sciences, Beijing, China
¹¹ Department of Epidemiology and Statistics, Nanjing Medical University School of Public Health, Nanjing, China
¹² Nutritional Epidemiology Branch, Division of Cancer Epidemiology and Genetics, National Cancer Institute, Bethesda, MD
¹³ Department of Epidemiology, UNT Health Science Center, Fort Worth, TX
¹⁴ Department of Epidemiology and Biostatistics, Erasmus Medical Center, Rotterdam, The Netherlands
¹⁵ School of Public Health, University of Pittsburgh/University of Pittsburgh Cancer Institute, Pittsburgh, PA

Correspondence to Dr. Stefania Boccia, Genetic Epidemiology and Molecular Biology Unit, Institute of Hygiene, Università Cattolica del Sacro Cuore, L.go F.Vito 1, 00168 Rome, Italy (e-mail: sboccia@rm.unicatt.it ).

Received for publication July 25, 2007. Accepted for publication October 29, 2007.

Methylenetetrahydrofolate reductase (MTHFR) is a key enzyme in the metabolism of folate, whose role in gastric carcinogenesis is controversial. The authors performed a meta-analysis and individual data pooled analysis of case-control studies that examined the association between C677T and A1298C polymorphisms (the former being associated with low folate serum levels) and gastric cancer (meta-analyses: 16 studies, 2,727 cases and 4,640 controls for C677T and seven studies, 1,223 cases and 2,015 controls for A1298C; pooled analyses: nine studies, 1,540 cases and 2,577 controls for C677T and five studies, 1,146 cases and 1,549 controls for A1298C). An increased risk was found for MTHFR 677 TT in the meta-analysis (odds ratio (OR) = 1.52, 95% confidence interval (CI): 1.31, 1.77) and pooled analysis (OR = 1.49, 95% CI: 1.14, 1.95). No association resulted for MTHFR 1298 CC (meta-OR = 0.94, 95% CI: 0.65, 1.35; pooled OR = 0.90, 95% CI: 0.69, 1.34). Results from the pooled analysis of four studies on C677T stratified according to folate levels showed an increased risk for individuals with low (OR = 2.05, 95% CI: 1.13, 3.72) versus high (OR = 0.95, 95% CI: 0.54, 1.67) folate levels. Overall, these findings support the hypothesis that folate plays a role in gastric carcinogenesis.

Keywords: epidemiology; folic acid; genetic predisposition to disease; meta-analysis; MTHFR; stomach neoplasms
Abbreviations: CI, confidence interval; GSEC, Genetic Susceptibility to Environmental Carcinogens; HWE, Hardy-Weinberg equilibrium; MTHFR, methylenetetrahydrofolate reductase

The 5,10-methylenetetrahydrofolate reductase gene (MTHFR) maps to chromosome 1p36.3 (1). The complementary DNA sequence is 2.2-kb long and contains 11 exons (1). The gene product is a 77-kD protein, although a smaller isoform of approximately 70 kD has been observed in some tissues such as liver (2). MTHFR plays a central role in folate metabolism, together with other enzymes, by irreversibly catalyzing the conversion of 5,10-methylenetetrahydrofolate to 5-methyltetrahydrofolate, the primary circulating form of folate and a cosubstrate for homocysteine methylation to methionine (figure 1). In humans, folate plays the fundamental role of providing methyl groups for de novo deoxynucleotide synthesis and for intracellular methylation reactions (2, 3).

click to view

Figure 1: The folate pathway. Modified from Hung et al. (83). dUMP, deoxyuridine monophosphate; DHF, dihydrofolate; THF, tetrahydrofolate; MTR, methionine synthase; SAM, S-adenosyl-L-methionine; TS, thymidylate synthase; dTMP, deoxythymidine monophosphate; MTHFR, methylentrahydrofolate reductase.

MTHFR enzyme function may influence cancer risk in two ways. The substrate of MTHFR enzyme, 5,10-methylenetetrahydrofolate, is involved in the conversion of deoxyuridylate monophosphate to deoxythymidylate monophosphate, and low levels of 5,10-methylenetetrahydrofolate would lead to an increased deoxyuridylate monophosphate/deoxythymidylate monophosphate ratio. In this situation, increased incorporation of uracil into DNA in place of thymine may follow, resulting in an increased chance of point mutations and DNA/chromosome breakage (3). A less active form of MTHFR would lead, all other factors being equal, to an accumulation of 5,10-methylenetetrahydrofolate, thus a lower deoxyuridylate monophosphate/deoxythymidylate monophosphate ratio, and a presumably lower cancer risk (3).

The second way in which impaired MTHFR activity might influence cancer risk is determined by the level of S-adenosyl-L-methionine, the common donor of methyl that is necessary for maintenance of the methylation patterns in DNA. Changes in methylation modify DNA conformation and gene expression. A less active form of MTHFR leads to lower S-adenosyl-L-methionine levels and consequently to hypomethylation; this phenomenon would be expected to increase the risk of some cancers (4) (figure 1). Similarly, low folate intake may modify cancer risk by inducing uracil misincorporation during DNA synthesis, leading to chromosomal damage, DNA strand breaks and impaired DNA repair, and DNA hypomethylation (5).

Twenty-nine rare mutations of the MTHFR gene have been described in homocystinuric patients, resulting in very low enzymatic activity (6), whereas two common polymorphisms are present in healthy individuals with lower enzyme activity: C→T in exon 4 at nucleotide 677, leading to Ala222Val (7); and A→C in exon 7 at nucleotide 1298, leading to Glu429Ala (8, 9). These polymorphisms are located 2.1-kb apart and have been investigated in association with the risk of gastric and other cancers (10). Three additional polymorphisms have been described, T1059C, T1317C, and G1793A (9, 11, 12). The T1059C polymorphism has been reported to be associated with increased neural tube defects in an Iowa population (11), while T1317C is a silent change with no effect on plasma homocysteine and folate concentrations (9). The variant allele of the G1793A polymorphism is least frequent among Ashkenazi Jewish individuals (1.3 percent) compared with Caucasians (6.9 percent) (12); it has been reported that individuals with the heterozygous genotype for the variant allele, compared with individuals with the wild genotype, have borderline or deficient folate concentrations (13).

Individuals who are homozygous for the MTHFR 677 less frequent variant (TT) have 30 percent of the expected enzyme activity in vitro compared with those who are homozygous for the common variant (CC), whereas heterozygous carriers have 65 percent activity (2). van der Put et al. (8) evaluated MTHFR activity according to the combination of A1298C and C677T genotypes, showing that individuals who are homozygous for the wild-type MTHFR 677 allele (CC) and contemporarily homozygous for the 1298 mutant allele (CC) have 60 percent activity compared with subjects carrying the 1298AA genotype, while, in the same population, 80 percent activity was detected for the 1298 heterozygotes (AC). Enzyme activity for individuals who are heterozygous for both C677T and A1298C appears to be approximately 50–60 percent that for those without either variant (9). It has been reported that subjects who are TT homozygous for MTHFR 677 exhibit reduced folate concentrations and higher serum homocysteine levels compared with those who carry at least one 677C allele (14–16). The evidence regarding the association of the 1298 variant allele with increased folate levels is less consistent (14, 15, 17, 18).

Three recent studies reported that the MTHFR TT genotype is related to DNA hypomethylation (19, 20), particularly in individuals with reduced plasma folate concentrations (21). Inconsistent results derive from studies of the A1298C polymorphism, plasma folate, and homocysteine levels (15, 18, 22, 23).

The T allele frequency (percentage) of the MTHFR 677 polymorphism is reported to be 24.5–43.8 in Europeans, 17.6–42.4 in Asians, 21.1–39.1 in US individuals, and 12.0–23.5 in African Americans (24). The frequency of homozygosity ranges from 1 percent (95 percent confidence interval (CI): 0.2, 2.0) in US African-American populations to more than 20 percent (95 percent CI: 14.6, 26.8) in US Latinos; 5 percent (95 percent CI: 1.2, 9.6) to 30 percent (95 percent CI: 21.4, 38.9) in White populations in Europe and North America; 32.2 percent (95 percent CI: 28.3, 36.4) in Mexico; 5.8 percent (95 percent CI: 3.5, 9.6) in White Canadians in Alberta to 14.3 percent (95 percent CI: 10.9, 17.6) in those in Quebec, Canada; 0.0 percent (95 percent CI: 0.0, 1.2) in Sub-Saharan Africa; 10.7 percent in Oceania (95 percent CI: 5.5, 19.7); and 11.5 percent (95 percent CI: 10.2, 12.7) in Japanese and 16 percent (95 percent CI: 8.0, 31.0) in Chinese (24–27). For A1298C, the variant allele frequency is reported to be 14.0–40.4 in Europeans and 11.1–17.0 in Asians. The frequency of homozygosity ranges from 10 percent (95 percent CI: 9.0, 11.0) in White populations in Europe and North America to 3.5 percent (95 percent CI: 0.2, 7.2) in Asians (28).

Gastric cancer is the second most common cause of cancer mortality, with 647,000 deaths reported worldwide in 2002 (29). In many populations, particularly in high-income countries, its incidence has gradually decreased in the last decades; however, it is still the fifth most common type of cancer in Europe and the fourth internationally (30). Helicobacter pylori infection is the single most common cause of adenocarcinoma of the distal stomach (31), but it is not a necessary or a sufficient cause. The development of gastric cancer appears in fact to be the result of a complex interaction between H. pylori infection, lifestyle, and genetic factors. Among the lifestyle risk factors, tobacco smoking, a high intake of salt, and lack of food refrigeration all seem to play a major role (32). Lastly, gastric cancer risk shows a familial clustering (33).

With regard to genetic factors, several single nucleotide polymorphisms might potentially alter individual susceptibility to gastric cancer (34). Among them are polymorphisms in genes involved in the protection of gastric mucosa against damaging agents and inflammatory response, genes that influence the ability to detoxify carcinogens (metabolic genes) and are involved in oxidative damage response and DNA repair, and oncogenes (35). Genes involved in folate metabolism have also been considered to play a role in gastric cancer risk (28).

Some nutrients involved in the folate metabolic pathway (e.g., vitamins B6 and B12, methionine), alcohol (a folate antagonist), and smoking (which impairs folate level) may interact with plasma folate levels and the MTHFR polymorphisms in determining cancer risk (36, 37). It has been reported that alcohol perturbs folate metabolism by reducing folate absorption, increasing folate excretion, or inhibiting methionine synthase (38, 39). The inverse association between folate intake and plasma homocysteine levels can be modified by alcohol intake and by the MTHFR 677 but not the 1298 polymorphism (40). The inverse effect of smoking on folate status might be confounded by alcohol intake or dietary habits (41, 42), even though the association persists after adjusting for dietary folate and alcohol intake (42, 43). Additional studies have reported that elevated folate turnover in response to rapid tissue proliferation or DNA repair in aerodigestive tissues among individuals exposed to tobacco smoke might partially explain this phenomenon (44, 45).

According to recent reports, alcohol drinkers carrying the MTHFR 677 TT genotype had about a fivefold increased risk of gastric cancer compared with drinkers carrying the wild homozygous variant, namely, odds ratios of 5.36 (95 percent CI: 1.94, 14.83) reported by Graziano et al. (19) and 5.32 (95 percent CI: 1.66, 17.02) by Stolzenberg-Solomon et al. (46), whereas others did not show such interaction (47, 48). Additionally, Gao et al. (49) reported that smokers carrying the MTHFR 677 T allele had a 7.7-fold increased risk (OR = 7.72, 95 percent CI: 2.23, 26.79) of gastric cancer compared with nonsmokers with the CC genotype. To our knowledge, no published study has ever explored whether the effect of the MTHFR 677 TT genotype on gastric cancer is modified by individual folate intake or by plasma folate levels. Finally, the interaction between alcohol, smoking, or folate status and the MTHFR A1298C polymorphism has never been known to be tested in gastric cancer.

The effect of the combination of the two common MTHFR polymorphisms on gastric cancer was investigated by Miao et al. (50) and Boccia et al. (51). Both reported no interaction between them.

A meta-analysis of prospective studies showed an inverse association between fruit and vegetables intake, the main dietary source of folate, and gastric cancer risk, particularly after 10 or more years of follow-up (52). Discrepant results, however, recently emerged from a large European cohort study, showing no association between fresh fruit intake and gastric cancer and a slight protective effect of total vegetable intake for the intestinal histotype only (53). Results from a meta-analysis of prospective and retrospective studies specifically focusing on dietary folate intake and risk of gastric cancer also reported no clear effect of dietary folate intake, with no differences between cohort or case-control studies (54).

On the other hand, two recent meta-analyses showed that the MTHFR 677 TT genotype is associated with an increased risk of gastric cancer, suggesting an important role of folate levels and subsequent impaired chromosomal DNA synthesis and aberrant DNA methylation in gastric carcinogenesis (28, 54). However, neither meta-analysis included all published reports available when the meta-analyses were published and specifically included either eight (28) or nine (54) studies compared with 16 studies in the present meta-analysis. In addition, these two provided unadjusted overall estimates, and the results were not stratified according to potential factors affecting folate status and MTHFR polymorphisms because of the nature of already published data. We accomplished both of the last two points by also carrying out a pooled analysis of individual-level data.

With the present meta- and pooled analyses, we aimed to assess the overall effect of the MTHFR C677T and A1298C polymorphisms on gastric cancer by including all available published papers and to help clarify the interrelations between these polymorphisms with folate, alcohol, and smoking and gastric cancer risk.

We assessed the association between the MTHFR C677T and A1298C polymorphisms and gastric cancer by conducting meta-analyses of all published papers and pooled analyses of individual-level data when available.

Meta-analysis
Selection criteria.
The papers were identified by searching the MEDLINE (National Library of Medicine, National Institutes of Health, Bethesda, Maryland) and EMBASE (Elsevier, Amsterdam, the Netherlands) databases up to January 2007 using the following terms: ("methylenetetrahydrofolate reductase" or MTHFR) and (gastric or stomach) and (cancer or carcinoma), without any restriction on language. Our research produced 35 articles. A cited reference search of the retrieved articles was carried out, and publications were also identified by reviewing their bibliographies. Eligible were community-based studies that reported the frequency of the MTHFR C677T and/or A1298C polymorphisms as number of individuals with gastric cancer and controls according to the three variant genotypes of both polymorphisms. Studies whose allele frequencies in the control population deviated from Hardy-Weinberg equilibrium (HWE) at a p value of ≤0.05 were excluded from the meta-analysis. If more than one article was published from the same case series, we used the one that included the most individuals in the analysis.

Of the 35 articles retrieved, 21 studies were eligible for the analysis (19, 46–50, 55–69). Five reports (55–59) were excluded either because they concerned subjects included in an expanded series (50, 60) or because they partially overlapped with another study (49) eventually selected because it gave the absolute number of individuals according to the three variant genotypes of MTHFR 677 (57–59). Finally, one study was excluded from the meta-analysis for the association between MTHFR C677T and gastric cancer (60), and one from the analysis of A1298C (46), because of deviations from HWE. One study in press at the time (not yet published) was also included (51).

The final number of articles considered for our meta-analysis of the association between MTHFR C677T and gastric cancer risk included 16 case-control studies (19, 46–51, 61–69), of which three were written in the Chinese language (49, 61, 62), comprising a total of 7,367 subjects (2,727 cases and 4,640 controls). The studies are described in table 1. Ten of 16 were population based; one was a case-control study nested in a cohort (46). Among them, seven were also included in the meta-analysis of the association between MTHFR A1298C and gastric cancer risk (48, 50, 51, 60, 61, 63, 64), for a total of 3,238 subjects (1,223 cases and 2,015 controls).

click to view

Table 1  Description of the studies included in the meta- and pooled analyses of the methylenetetrahydrofolate reductase C677T and A1298C polymorphisms and gastric cancer

Statistical analysis.
Two researchers (S. Boccia and F. Gianfagna) extracted the data from each article by using a structured sheet and entered them into a database. The followings items were considered: year and location of the study, ethnicity, characteristics of the control group, tumor site (cardia/noncardia gastric cancer), and number of individuals heterozygous and homozygous for the MTHFR 677 and 1298 variant alleles in the compared groups. Heterogeneity was tested by the Q statistic (70). In carrying out the meta-analyses, random-effects models were used (71) to take into account the possibility of heterogeneity between studies. The summary odds ratios of gastric cancer associated with the MTHFR 677 TT and CT genotypes and the MTHFR 1298 CC and CA genotypes were estimated by using the homozygous wild type for each genotype as the reference group. To determine deviation from HWE, we used Fisher's exact permutation test with a Monte Carlo technique (72). A visual inspection of Begg's funnel plot and Begg and Egger asymmetry tests (70) was used to investigate for publication bias when appropriate (73).

Because two potential causes of heterogeneity among studies were ethnicity and tumor site, we calculated separate odds ratios in subgroups of studies performed among different ethnic groups (Asian/Europeans) and in subgroups of studies including cardia and noncardia gastric cancer cases, when genotype data were tabulated according to the tumor site specified in the published papers. A heterogeneity test was then performed to test for statistically significant differences among the strata estimates.

Pooled analysis
Data collection.
The pooled analysis was performed by using the Genetic Susceptibility to Environmental Carcinogens (GSEC) database. The International Collaborative Study on GSEC (http://www.upci.upmc.edu/research/ccps/ccontrol/g_intro.html) project gathers information from both published and unpublished population-based studies on metabolic gene polymorphisms and cancer risk. The design of the GSEC study has been reported elsewhere (74). Apposite investigators were contacted and were asked to provide their data for the pooled analyses. A questionnaire was provided by e-mail to each investigator, collecting information on study design, selection and source of cases and controls, laboratory method used for genotyping, source of DNA used for the genotype analysis, and response rate for cases and controls. We contacted all authors of the identified published papers, including those whose control populations were not in HWE for the studied polymorphisms (46, 60). Of the 17 eligible data sets, we were able to obtain data from 10, with one of them (60) later excluded for the pooled analysis on MTHFR 677 and one more (46) on MTHFR 1298 because the allele frequency of the control population did not respect HWE. We finally included nine studies for MTHFR 677—four of Asians, four of Europeans, and one of Latinos—totaling 4,117 subjects (1,540 cases and 2,577 controls) (refer to table 1 for details). As for MTHFR 1298, five studies were included, totaling 2,695 subjects (1,146 cases and 1,549 controls; refer to table 1 for details).

Statistical analysis.
To assess the association of the MTHFR 677 TT and 1298 CC genotypes with gastric cancer, the logistic regression model was used to estimate study-specific odds ratios and 95 percent confidence intervals in each single study. Adjusted odds ratios were obtained by including age, gender, and smoking status (ever/never) as covariates. In some studies, odds ratios estimated for individual studies and numbers of cases and controls did not precisely match those reported in the publications. A pooled odds ratio was estimated by inverse-variance weighting with the random-effects model (71), taking into account the possibility of heterogeneity between studies, which was tested with Q statistics (70). We could perform stratified analyses for only MTHFR 677 TT, since MTHFR 1298 CC was available in only four studies (table 1).

Results were stratified according to ethnicity (Asians/European descendants), alcohol drinking, tobacco smoking, and folate status. Because information on pack-years of smoking was available in only four studies (47, 48, 50, 51), subjects were classified as ever (current and former) and never smokers. For alcohol drinking, individuals were categorized as ever (current and/or former) and never (<1 glass of each alcoholic beverage/month) drinkers. Information on folate serum levels was obtained from Gotze et al. (66) and gastric mucosa folate levels from Weng et al. (63). The nutrient density of folate (dietary folate intake/total caloric intake: µg/kcal x 1,000) was obtained by Zhang et al. (48) and intake of fruit and vegetables portions by Boccia et al. (51). For the analysis, subjects were categorized in two classes based on the lower quartile of each variable estimated in the control population.

For each stratified analysis, a pooled odds ratio was estimated by inverse-variance weighting with the random-effects model (71), taking into account the possibility of heterogeneity between studies, which was tested with Q statistics. A heterogeneity test was performed to assess for statistically significant differences among the pooled strata estimates.

Statistical analyses were carried out by using the STATA software package v.8.2 (Stata Corporation, College Station, Texas).

Meta-analysis of MTHFR C677T
The odds ratios in 11 of 16 studies were above the unit; among them, five studies (19, 50, 62, 65, 68) reported a significant positive association between gastric cancer and the MTHFR 677 TT genotype (table 1). The meta-analysis produced overall odds ratios of 1.52 (95 percent CI: 1.31, 1.77) and 1.17 (95 percent CI: 0.99, 1.39) for gastric cancer and the MTHFR TT (figure 2) and CT genotypes, respectively. The heterogeneity test results were 0.37 for TT and 0.01 for CT. The funnel plot (not shown) and Begg's test provided no evidence of publication bias (p = 0.72) for the MTHFR 677 TT genotype, whereas the Egger test provided a p value of 0.007.

click to view

Figure 2 : Forest plot of the odds ratios and 95% confidence intervals (CIs) of studies of the association between gastric cancer and the MTHFR C677T polymorphism (TT vs. CC). On the left, the first author of the study is followed by the publication year in parentheses. The size of the black box corresponding to each study is proportional to the sample size; the horizontal line shows the corresponding 95% CI of the odds ratio. The combined estimate is based on a random-effects model shown by the diamond. The solid vertical line represents the null result.

When stratifying the data by ethnicity, we observed odds ratios of 1.34 (95 percent CI: 0.90, 1.99) and 1.64 (95 percent CI: 1.36, 1.97) for the MTHFR 677 TT versus CC genotype in six studies of Europeans and nine studies of Asians, respectively (p for heterogeneity = 0.38). The analysis by anatomic tumor site showed that both gastric cardia cancer (11 studies) and noncardia cancer (six studies) were significantly associated with MTHFR 677 TT, with respective odds ratios of 1.51 (95 percent CI: 1.11, 2.05) and 1.57 (95 percent CI: 1.09, 2.24) (p for heterogeneity = 0.87). There was no evidence of heterogeneity in all subgroup meta-analyses performed.

Meta-analysis of MTHFR A1298C
All seven studies included reported odds ratios spread around the null effect (table 1). From the meta-analysis, the association between gastric cancer and MTHFR 1298 CC was 0.94 (95 percent CI: 0.65, 1.35) (figure 3); an odds ratio of 1.01 (95 percent CI: 0.86, 1.18) was found for the association with 1298 AC. There was no evidence of heterogeneity in the overall meta-analysis and in subgroup meta-analyses.

click to view

Figure 3 : Forest plot of the odds ratios and 95% confidence intervals (CIs) of studies of the association between gastric cancer and the MTHFR A1298C polymorphism (CC vs. AA). On the left, the first author of the study is followed by the publication year in parentheses. The size of the black box corresponding to each study is proportional to the sample size; the horizontal line shows the corresponding 95% CI of the odds ratio. The combined estimate is based on a random-effects model shown by the diamond. The solid vertical line represents the null result.

 
When we restricted the analysis of MTHFR A1298C to the five studies conducted among Asians (50, 60, 61, 63, 64), an overall odds ratio of 0.81 (95 percent CI: 0.43, 1.51) emerged. When the analysis was stratified by tumor site, an odds ratio of 0.99 (95 percent CI: 0.43, 2.28) resulted for cardia cancer, and an odds ratio of 0.81 (95 percent CI: 0.38, 1.74) was found for noncardia cancer (p for heterogeneity = 0.76). Pooled analyses
The study-specific adjusted odds ratios for MTHFR 677 TT are reported in table 1. Of the nine studies included in the pooled analysis, six had odds ratios above the unit; among them, three (19, 50, 51) reported a significant positive association between gastric cancer and the MTHFR 677 TT genotype (table 1). Results from the pooled analysis are shown in table 2. The overall odds ratio adjusted for age, gender, and smoking status was 1.49 (95 percent CI: 1.14, 1.95; p for heterogeneity = 0.06) for MTHFR 677 TT, whereas an odds ratio of 1.21 (95 percent CI: 0.90, 1.62; p for heterogeneity = 0.03) was detected for MTHFR 677 CT. Publication bias was not tested because of low statistical power of the tests when the number of studies is 10 or fewer (73, 75, 76).

click to view

Table 2  Odds ratios and 95% confidence intervals from the pooled analysis of the association between the MTHFR* C677T polymorphism and gastric cancer

The pooled odds ratio for MTHFR 677 TT among Asians was 1.54 (95 percent CI: 1.09, 2.15; p for heterogeneity = 0.34) and among Europeans was 1.52 (95 percent CI: 0.84, 2.76; p for heterogeneity = 0.03). The p for heterogeneity test result for Asians and Caucasians was 0.86 (table 2).

When we stratified on smoking habits, an odds ratio of 2.04 (95 percent CI: 1.27, 3.26) for MTHFR 677 TT resulted for ever smokers, whereas an odds ratio of 1.36 (95 percent CI: 1.03, 1.80) was found for never smokers, with a p for heterogeneity test result of 0.14 among them (table 2). The stratified analysis according to alcohol intake included six studies; similar risk estimates were found for ever drinkers and never drinkers (p for heterogeneity = 0.49; table 2). The stratified analysis according to estimated folate status showed an odds ratio for gastric cancer of 2.05 (95 percent CI: 1.13, 2.72) for MTHFR 677 TT individuals with low folate levels and an odds ratio of 0.95 (95 percent CI: 0.54, 1.67) for those with high folate levels (p for heterogeneity among the two estimates = 0.06) (table 2).

The study-specific adjusted odds ratios for MTHFR 1298 CC are reported in table 1. It was not possible to compute the adjusted odds ratio for the homozygous variant genotype in one study (60) because of the small number of subjects. The overall odds ratio adjusted for age, gender, and smoking status was 0.90 (95 percent CI: 0.69, 1.34; p for heterogeneity = 0.50, four studies) for MTHFR 1298 CC, whereas an odds ratio of 1.01 (95 percent CI: 0.83, 1.22; p for heterogeneity = 0.50, five studies) was found for MTHFR 1298 AC.

The results from the meta-analysis of 16 studies highlighted a higher risk of developing gastric cancer for subjects carrying the MTHFR 677 TT genotype. The results were confirmed by the pooled analysis including nine studies. No association was detected from either the meta-analysis or the pooled analysis between the MTHFR 1298 CC genotype and gastric cancer. Our results were consistent with two previously published meta-analyses by Zintzaras (28) and Larsson et al. (54), which showed an increased risk of gastric cancer associated with only the MTHFR 677 TT genotype and an absence of risk for MTHFR 1298 CC. However, these two previously published meta-analyses included a smaller number of studies than ours did, and results were based on unadjusted estimates. In our pooled analysis of MTHFR 677, an increased risk of gastric cancer was observed for subjects with a low folate status compared with those with a high folate status. These results support our a priori hypothesis of a higher risk of gastric cancer for subjects carrying the variant MTHFR 677 homozygous variant who have low folate levels compared with subjects carrying the same variant but with high levels of folate.

A limitation common to both the meta- and the pooled analysis might be the presence of publication bias. In the meta-analysis of MTHFR 677 TT, we did not observe evidence of publication bias from visual inspection of the Begg's funnel plots and the results of the rank correlation statistical test. Results from Egger's regression method highlighted some publication bias; however, this method is usually more sensitive than Begg's test, reporting to provide evidence for bias (false-positive results) when in fact it is not present, especially when the number of studies is low (75, 76). We cannot rule out the possibility that the effect of MTHFR 677 TT on gastric cancer was overestimated in our meta-analysis, because negative results from small studies remained unpublished.

In the pooled analysis, we explored possible effect modification of the MTHFR 677 TT genotype on gastric cancer by stratifying on tobacco smoking and alcohol drinking, two factors that may affect folate levels. We were unable to observe any effect modification; however, in both instances, the information did not take into account the amount or duration of alcohol intake and tobacco smoking.

When the results from the pooled analysis were stratified according to folate status (available from four studies), a strong association between the MTHFR 677 TT genotype and gastric cancer was noted among subjects with a low folate status compared with a high folate status. The heterogeneity test showed that results were borderline significantly different; therefore, our result needs to be confirmed with a larger population. This result supports our hypothesis, suggesting that concomitant inadequate folate intake and impaired MTHFR activity might be important susceptibility factors for gastric cancer. A limitation is the heterogeneity regarding collection of folate information: gastric mucosa level (63), serum level (66), nutrient density of folate (48), or dietary fruit and vegetables intake (51).

To our knowledge, this pooled analysis is the first assessing the role of two common MTHFR polymorphisms in the risk of gastric cancer. In fact, the two previously published meta-analyses did not include individual-level data (28, 54); therefore, the authors were unable to calculate adjusted estimates and to stratify the results of the meta-analyses according to folate status, alcohol intake, or smoking habits. Because the data sets included information on age, gender, and cigarette smoking from all studies, it was possible to adjust for the potential confounding effect of these variables and to assess consistently the presence of gene-environment interactions for MTHFR 677, a factor that makes the pooled analysis preferable to the meta-analysis (77). The absence of publication bias and statistical heterogeneity among studies strengthens our results.

Both MTHFR C677T and A1298C can be detected by means of polymerase chain reaction (followed by restriction fragment-length polymorphism) analysis with HinfI and MboII for C677T and A1298C, respectively (7, 8). Other methods include direct DNA sequencing or TaqMan assays (48). Most studies did not report the success rate in extracting DNA from samples, the proportion of eligible subjects for whom genotyping failed, whereas 43.0 percent (7/16) of them reported the degree of genotyping reproducibility (19, 46, 48, 50, 51, 61, 68). HWE was tested in 87.5 percent (14/16) of the studies. All previously mentioned variables are important indicators of the analytical validity of the genotyping methods, also influencing potential nondifferential misclassification of the exposure. In addition, only 31.2 percent of the studies (5/16) clearly reported that the analysts were unaware of the clinical status of the subjects when genotyping the samples; therefore, differential exposure misclassification may not be ruled out.

At the moment, the potential public health impact of this issue is limited, given the small association between gastric cancer and homozygosis TT for MTHFR 677. Additional studies on the possible additional risk of gastric cancer for subjects who are 677 TT homozygous and have low folate levels are urgently needed, however. If this preliminary result is confirmed, proper evaluation of the clinical utility of MTHFR C677T testing for identifying gastric cancer susceptibility among populations with folate deficiency, followed by the introduction of specific folate supplementation (vs. no folate supplementation), would be warranted. Currently, however, population testing for the MTHFR C677T polymorphism to prevent gastric cancer is not indicated.

MTHFR plays a central role in balancing DNA synthesis (which involves 5,10-methylentetrahydrofolate) and DNA methylation (which involves 5,10-methyltetrahydrofolate). Specifically, the 677T allele contributes to DNA hypomethylation, which in turn may lead to altered gene expression; at the same time, this polymorphism might exert a protective effect, as observed for colorectal cancer (24), by increasing the levels of the MTHFR substrate, essential for DNA synthesis. Therefore, exact interpretation of the MTHFR-cancer association is not straightforward, although the observed increased risk of gastric cancer associated with the MTHFR 677 homozygous variant suggests that dietary folate might be protective in gastric carcinogenesis mainly by limiting aberrant DNA methylation when folate status is impaired. In general, studying the association between sequence variants of folate-related genes and cancer has the advantage of being less prone to the confounding effect exerted by dietary or lifestyle factors (78). The observed increased risk of gastric cancer for MTHFR 677 TT individuals strengthens the hypothesis of a protective effect of folate in gastric carcinogenesis. If this hypothesis holds true, it would be interesting to explore whether the introduction of folate fortification in some common food items (79) in North America beginning in 1998 actually contributed to the decreasing rates of gastric cancer (80). However, in view of the lag time regarding an effect of folic acid and the lengthy induction time required for gastric cancer, this issue could probably be addressed in only the next decade.

The observation of a potential role of folate in gastric carcinogenesis is also strengthened by our results of an increased risk of gastric cancer for MTHFR 677–homozygous subjects with low folate levels. This observation suggests that concomitant inadequate folate intake and impaired MTHFR activity might be important susceptibility factors for gastric cancer.

Despite the limitations of this analysis in terms of comparable folate data, which requires confirmation from large prospective studies based on blood folate measurement, our results are in keeping with the model proposed by Friso et al. (21). With folate deficiency, a decrease in downstream MTHFR products results in a lower global DNA methylation status. Recently, aberrant methylation of proto-oncogenes has been explored as both a mechanism and a marker of carcinoma progression (81), with some studies reporting an altered methylation pattern particularly for diffuse gastric cancer (82). Additionally, it was recently reported that significant global DNA hypomethylation occurs in MTHFR 677 TT subjects when compared with those with the wild-type genotype (19, 20), especially when plasma folate level is reduced (21). Taken together, these results suggest that the increased risk of gastric cancer associated with the homozygous MTHFR 677 variant might be referable to the subsequent impaired folate levels affecting DNA methylation status. Therefore, the observed association between the homozygous variant MTHFR genotype and gastric cancer might be counterbalanced to some extent by adequate folate intake.

Other genes involved in folate metabolism should be considered for a more comprehensive understanding of the exact role of the folate pathway in gastric cancer susceptibility. Given the controversial evidence from nutritional studies on the effect of fruit and vegetables on gastric cancer, there is a need for large prospective cohort studies based on repeated serologic dosage of folate levels and/or detailed and repeated nutritional data that would further clarify the role of folate in gastric carcinogenesis. Such studies would lay the foundation for evaluating the possible benefits of preventive nutritional interventions for individuals at risk of gastric cancer.

This study was supported by a training fellowship Environmental Cancer Risk, Nutrition and Individual Susceptibility (ECNIS), an FP6 EC Network of Excellence (contract ETP/14/2007). The authors thank Barbara M. Stadterman, GSEC administrator, for her support (National Institutes of Health grant P30CA047904-19S1). M. Ebert is supported by the ProINNO II initiative of the BMWi/AIF, the Else Körner-Fresenius-Stiftung, and the Deutsche Krebshilfe, Germany.

The authors are indebted to Dr. Fan Liu for extracting data from the Chinese articles.

Conflict of interest: none declared.

1. Goyette P, Sumner JS, Milos R, et al. Human methylenetetrahydrofolate reductase: isolation of cDNA mapping and mutation identification. Nat Genet (1994) 7:195–200.
2. Rozen R. Genetic predisposition to hyperhomocysteinemia: deficiency of methylenetetrahydrofolate reductase (MTHFR). Thromb Haemost (1997) 78:523–6.
3. Blount BC, Mack MM, Wehr CM, et al. Folate deficiency causes uracil misincorporation into human DNA and chromosome breakage: implications for cancer and neuronal damage. Proc Natl Acad Sci U S A (1997) 94:3290–5.
4. Stern LL, Mason JB, Selhub J, et al. Genomic DNA hypomethylation, a characteristic of most cancers, is present in peripheral leukocytes of individuals who are homozygous for the C677T polymorphism in the methylenetetrahydrofolate reductase gene. Cancer Epidemiol Biomarkers Prev (2000) 9:849–53.
5. Duthie SJ. Folic acid deficiency and cancer: mechanisms of DNA instability. Br Med Bull (1999) 55:578–92.
6. Sibani S, Leclerc D, Weisberg IS, et al. Characterization of mutations in severe methylenetetrahydrofolate reductase deficiency reveals an FAD-responsive mutation. Hum Mutat (2003) 21:509–20.
7. Frosst P, Blom HJ, Milos R, et al. A candidate genetic risk factor for vascular disease: a common mutation in methylenetetrahydrofolate reductase. Nat Genet (1995) 10:111–13.
8. van der Put NM, Gabreels F, Stevens EM, et al. A second common mutation in the methylenetetrahydrofolate reductase gene: an additional risk factor for neural-tube defects? Am J Hum Genet (1998) 62:1044–51.
9. Weisberg I, Tran P, Christensen B, et al. A second genetic polymorphism in methylenetetrahydrofolate reductase (MTHFR) associated with decreased enzyme activity. Mol Genet Metab (1998) 64:169–72.
10. Kim YI. Methylenetetrahydrofolate reductase polymorphisms, folate, and cancer risk: a paradigm of gene-nutrient interactions in carcinogenesis. Nutr Rev (2000) 58:205–9.
11. Trembath D, Sherbondy AL, Vandyke DC, et al. Analysis of select folate pathway genes, PAX3, and human T in a midwestern neural tube defect population. Teratology (1999) 59:331–41.
12. Rady PL, Szucs S, Grady J, et al. Genetic polymorphisms of methylenetetrahydrofolate reductase (MTHFR) and methionine synthase reductase (MTRR) in ethnic populations in Texas; a report of a novel MTHFR polymorphic site, G 1793A. Am J Med Genet (2002) 107:162–8.
13. Melo SS, Persuhn DC, Meirelles MS, et al. G 1793A polymorphisms in the methylene-tetrahydrofolate gene: effect of folic acid on homocysteine levels. Mol Nutr Food Res (2006) 50:769–74.
14. Parle-McDermott A, Mills JL, Molloy AM, et al. The MTHFR 1298CC and 677TT genotypes have opposite associations with red cell folate levels. Mol Genet Metab (2006) 88:290–4.
15. Friedman G, Goldschmidt N, Friedlander Y, et al. A common mutation A1298C in human methylenetetrahydrofolate reductase gene: association with plasma total homocysteine and folate concentrations. J Nutr (1999) 129:1656–61.
16. Klerk M, Verhoef P, Clarke R, et al. MTHFR 677C→T polymorphism and risk of coronary heart disease: a meta-analysis. JAMA (2002) 288:2023–31.
17. Lievers KJ, Boers GH, Verhoef P, et al. A second common variant in the methylenetetrahydrofolate reductase (MTHFR) gene and its relationship to MTHFR enzyme activity, homocysteine, and cardiovascular disease risk. J Mol Med (2001) 79:522–8.
18. Chango A, Boisson F, Barbe F, et al. The effect of 677C→T and 1298A→C mutations on plasma homocysteine and 5,10-methylenetetrahydrofolate reductase activity in healthy subjects. Br J Nutr (2000) 83:593–6.
19. Graziano F, Kawakami K, Ruzzo A, et al. Methylenetetrahydrofolate reductase 677C/T gene polymorphism, gastric cancer susceptibility and genomic DNA hypomethylation in an at-risk Italian population. Int J Cancer (2006) 118:628–32.
20. Castro R, Rivera I, Ravasco P, et al. 5,10-methylenetetrahydrofolate reductase (MTHFR) 677C→T and 1298A→C mutations are associated with DNA hypomethylation. J Med Genet (2004) 41:454–8.
21. Friso S, Choi SW, Girelli D, et al. A common mutation in the 5,10-methylenetetrahydrofolate reductase gene affects genomic DNA methylation through an interaction with folate status. Proc Natl Acad Sci U S A (2002) 99:5606–11.
22. Chen J, Ma J, Stampfer MJ, et al. Linkage disequilibrium between the 677C>T and 1298A>C polymorphisms in human methylenetetrahydrofolate reductase gene and their contributions to risk of colorectal cancer. Pharmacogenetics (2002) 12:339–42.
23. Lievers KJ, Boers GH, Verhoef P, et al. A second common variant in the methylenetetrahydrofolate reductase (MTHFR) gene and its relationship to MTHFR enzyme activity, homocysteine, and cardiovascular disease risk. J Mol Med (2001) 79:522–8.
24. Botto LD, Yang Q. 5,10-Methylenetetrahydrofolate reductase gene variants and congenital anomalies: a HuGE review. Am J Epidemiol (2000) 151:862–77.
25. Sharp L, Little J. Polymorphisms in genes involved in folate metabolism and colorectal neoplasia: a HuGE review. Am J Epidemiol (2004) 159:423–43.
26. Wilcken B, Bamforth F, Li Z, et al. Geographical and ethnic variation of the 677C>T allele of 5,10 methylenetetrahydrofolate reductase (MTHFR): findings from over 7000 newborns from 16 areas world wide. J Med Genet (2003) 40:619–25.
27. Gueant-Rodriguez RM, Gueant JL, Debard R, et al. Prevalence of methylenetetrahydrofolate reductase 677T and 1298C alleles and folate status: a comparative study in Mexican, West African, and European populations. Am J Clin Nutr (2006) 83:701–7.
28. Zintzaras E. Association of methylenetetrahydrofolate reductase (MTHFR) polymorphisms with genetic susceptibility to gastric cancer: a meta-analysis. J Hum Genet (2006) 51:618–24.
29. Parkin DM. International variation. Oncogene (2004) 23:6329–40.
30. Verdecchia A, Mariotto A, Gatta G, et al. Comparison of stomach cancer incidence and survival in four continents. Eur J Cancer (2003) 39:1603–9.
31. IARC Working Group on the Evaluation of Carcinogenic Risks to Humans. Schistosomes, liver flukes and Helicobacter pylori. IARC Monogr Eval Carcinog Risks Hum (1994) 61:1–241.
32. Correa P, Schneider BG. Etiology of gastric cancer: what is new? Cancer Epidemiol Biomarkers Prev (2005) 14:1865–8.
33. Aoki M, Yamamoto K, Noshiro H, et al. A full genome scan for gastric cancer. J Med Genet (2005) 42:83–7.
34. Gonzalez CA, Sala N, Capella G. Genetic susceptibility and gastric cancer risk. Int J Cancer (2002) 100:249–60.
35. Roberts-Thomson IC, Butler WJ. Polymorphism and gastric cancer. J Gastroenterol Hepatol (2005) 20:793–4.
36. Bailey LB. Folate status assessment. J Nutr (1990) 120(suppl):1508S–11S.
37. Bailey LB. Folate, methyl-related nutrients, alcohol, and the MTHFR 677C→T polymorphism affect cancer risk: intake recommendations. J Nutr (2003) 133(suppl):3748S–53S.
38. Mason JB, Choi SW. Effects of alcohol on folate metabolism: implications for carcinogenesis. Alcohol (2005) 35:235–41.
39. Barak AJ, Beckenhauer HC, Tuma DJ, et al. Effects of prolonged ethanol feeding on methionine metabolism in rat liver. Biochem Cell Biol (1987) 65:230–3.
40. Chiuve SE, Giovannucci EL, Hankinson SE, et al. Alcohol intake and methylenetetrahydrofolate reductase polymorphism modify the relation of folate intake to plasma homocysteine. Am J Clin Nutr (2005) 82:155–62.
41. Tungtrongchitr R, Pongpaew P, Soonthornruengyot M, et al. Relationship of tobacco smoking with serum vitamin B12, folic acid and haematological indices in healthy adults. Public Health Nutr (2003) 6:675–81.
42. Stark KD, Pawlosky RJ, Beblo S, et al. Status of plasma folate after folic acid fortification of the food supply in pregnant African American women and the influences of diet, smoking, and alcohol consumption. Am J Clin Nutr (2005) 81:669–77.
43. Mannino DM, Mulinare J, Ford ES, et al. Tobacco smoke exposure and decreased serum and red blood cell folate levels: data from the Third National Health and Nutrition Examination Survey. Nicotine Tob Res (2003) 5:357–62.
44. Heimburger DC. Localized deficiencies of folic acid in aerodigestive tissues. Ann N Y Acad Sci (1992) 669:87–96.
45. Piyathilake CJ, Hine RJ, Dasanayake AP, et al. Effect of smoking on folate levels in buccal mucosal cells. Int J Cancer (1992) 52:566–9.
46. Stolzenberg-Solomon RZ, Qiao YL, Abnet CC, et al. Esophageal and gastric cardia cancer risk and folate- and vitamin B(12)-related polymorphisms in Linxian, China. Cancer Epidemiol Biomarkers Prev (2003) 12:1222–6.
47. Lacasana-Navarro M, Galvan-Portillo M, Chen J, et al. Methylenetetrahydrofolate reductase 677C>T polymorphism and gastric cancer susceptibility in Mexico. Eur J Cancer (2006) 42:528–33.
48. Zhang FF, Terry MB, Hou L, et al. Genetic polymorphisms in folate metabolism and the risk of stomach cancer. Cancer Epidemiol Biomarkers Prev (2007) 16:115–21.
49. Gao C, Wu J, Ding J, et al. Polymorphisms of methylenetetrahydrofolate reductase C677T and the risk of stomach cancer. (In Chinese). Zhonghua Liu Xing Bing Xue Za Zhi (2002) 23:289–92.
50. Miao X, Xing D, Tan W, et al. Susceptibility to gastric cardia adenocarcinoma and genetic polymorphisms in methylenetetrahydrofolate reductase in an at-risk Chinese population. Cancer Epidemiol Biomarkers Prev (2002) 11:1454–8.
51. Boccia S, Gianfagna F, Persiani R, et al. Methylenetetrahydrofolate reductase C677T and A1298C polymorphisms and susceptibility to gastric adenocarcinoma in an Italian population. Biomarkers (2007) 12:635–44.
52. Lunet N, Lacerda-Vieira A, Barros H. Fruit and vegetables consumption and gastric cancer: a systematic review and meta-analysis of cohort studies. Nutr Cancer (2005) 53:1–10.
53. Gonzalez CA, Pera G, Agudo A, et al. Fruit and vegetable intake and the risk of stomach and oesophagus adenocarcinoma in the European Prospective Investigation into Cancer and Nutrition (EPIC-EURGAST). Int J Cancer (2006) 118:2559–66.
54. Larsson SC, Giovannucci E, Wolk A. Folate intake, MTHFR polymorphisms, and risk of esophageal, gastric, and pancreatic cancer: a meta-analysis. Gastroenterology (2006) 131:1271–83.
55. Miao X, Xing D, Tan W, et al. Single nucleotide polymorphisms in methylenetetrahydrofolate reductase gene and susceptibility to cancer of the gastric cardia in Chinese population. (In Chinese). Zhonghua Yi Xue Za Zhi (2002) 82:669–72.
56. Shen H, Xu Y, Zheng Y, et al. Polymorphisms of 5,10-methylenetetrahydrofolate reductase and risk of gastric cancer in a Chinese population: a case-control study. Int J Cancer (2001) 95:332–6.
57. Gao CM, Lu JW, Toshiro T, et al. Polymorphism of methylenetetrahydrofolate reductase and sensitivity of stomach cancer to fluoropyrimidine-based chemotherapy. (In Chinese). Zhonghua Liu Xing Bing Xue Za Zhi (2004) 25:1054–8.
58. Gao CM, Takezaki T, Wu JZ, et al. Polymorphisms in thymidylate synthase and methylenetetrahydrofolate reductase genes and the susceptibility to esophageal and stomach cancer with smoking. Asian Pac J Cancer Prev (2004) 5:133–8.
59. Gao CM, Wu JZ, Liu YT, et al. Interactions between lifestyle, methylenetetrahydrofolate reductase gene and polymorphisms in thymidylate synthase gene with risk of stomach cancer. (In Chinese). Zhonghua Liu Xing Bing Xue Za Zhi (2003) 24:599–603.
60. Shen H, Newmann AS, Hu Z, et al. Methylenetetrahydrofolate reductase polymorphisms/haplotypes and risk of gastric cancer: a case-control analysis in China. Oncol Rep (2005) 13:355–60.
61. Si PR, Fang DC, Zhang H, et al. The relationship between methylenetetrahydrofolate reductase gene polymorphism and microsatellite instability in gastric cancer. (In Chinese). Zhonghua Liu Xing Bing Xue Za Zhi (2005) 26:794–9.
62. Mu LN, Ding BG, Chen CW, et al. A case-control study on the relationship between methyl-tetra-hydrofolic acid reductase 677 gene polymorphism and the risk of stomach cancer. (In Chinese). Zhonghua Liu Xing Bing Xue Za Zhi (2004) 25:495–8.
63. Weng YR, Sun DF, Fang JY, et al. Folate levels in mucosal tissue but not methylenetetrahydrofolate reductase polymorphisms are associated with gastric carcinogenesis. World J Gastroenterol (2006) 12:7591–7.
64. Kim JK, Kim S, Han JH, et al. Polymorphisms of 5,10-methylenetetrahydrofolate reductase and risk of stomach cancer in a Korean population. Anticancer Res (2005) 25:2249–52.
65. Wang Y, Guo W, He Y, et al. Association of MTHFR C677T and SHMT(1) C1420T with susceptibility to ESCC and GCA in a high incident region of Northern China. Cancer Causes Control (2007) 18:143–52.
66. Gotze T, Rocken C, Rohl FW, et al. Gene polymorphisms of folate metabolizing enzymes and the risk of gastric cancer. Cancer Lett (2007) 251:228–36.
67. Zeybek U, Yaylim I, Yilmaz H, et al. Methylenetetrahydrofolate reductase C677T polymorphism in patients with gastric and colorectal cancer. Cell Biochem Funct (2006) 25:419–22.
68. Wang LD, Guo RF, Fan ZM, et al. Association of methylenetetrahydrofolate reductase and thymidylate synthase promoter polymorphisms with genetic susceptibility to esophageal and cardia cancer in a Chinese high-risk population. Dis Esophagus (2005) 18:177–84.
69. Sarbia M, Geddert H, Kiel S, et al. Methylenetetrahydrofolate reductase C677T polymorphism and risk of adenocarcinoma of the upper gastrointestinal tract. Scand J Gastroenterol (2005) 40:109–11.
70. Deeks JJ, Altman DG, Bradburn MJ. Statistical methods for examining heterogeneity and combining results from several studies in meta-analysis. In: Systematic reviews in health care—meta-analysis in context—Egger M, Davey Smith G, Altman DG, eds. (2005) London, United Kingdom: BMJ Books. 285–312.
71. DerSimonian R, Laird N. Meta-analysis in clinical trials. Control Clin Trials (1986) 7:177–88.
72. Roff DA, Bentzen P. The statistical analysis of mitochondrial DNA polymorphisms: chi 2 and the problem of small samples. Mol Biol Evol (1989) 6:539–45.
73. Ioannidis JP, Trikalinos TA. The appropriateness of asymmetry tests for publication bias in meta-analyses: a large survey. CMAJ (2007) 176:1091–6.
74. Taioli E. International collaborative study on genetic susceptibility to environmental carcinogens. Cancer Epidemiol Biomarkers Prev (1999) 8:727–8.
75. Egger M, Davey Smith G, Schneider M, et al. Bias in meta-analysis detected by a simple, graphical test. BMJ (1997) 315:629–34.
76. Irwig L, Macaskill P, Berry G, et al. Bias in meta-analysis detected by a simple, graphical test. Graphical test is itself biased. BMJ (1998) 316:470–1.
77. Blettner M, Sauerbrei W, Schlehofer B, et al. Traditional reviews, meta-analyses and pooled analyses in epidemiology. Int J Epidemiol (1999) 28:1–9.
78. Smith GD, Ebrahim S. ‘Mendelian randomization’: can genetic epidemiology contribute to understanding environmental determinants of disease? Int J Epidemiol (2003) 32:1–22.
79. Malinow MR, Duell PB, Hess DL, et al. Reduction of plasma homocyst(e)ine levels by breakfast cereal fortified with folic acid in patients with coronary heart disease. N Engl J Med (1998) 338:1009–15.
80. American Cancer Society. Cancer facts and figures 2007. (2007) Atlanta, GA: American Cancer Society.
81. Dunn BK. Hypomethylation: one side of a larger picture. Ann N Y Acad Sci (2003) 983:28–42.
82. Yamashita K, Park HL, Kim MS, et al. PGP9.5 methylation in diffuse-type gastric cancer. Cancer Res (2006) 66:3921–7.
83. Hung RJ, Hashibe M, McKay J, et al. Folate related genes and the risk of tobacco-related cancers in Central Europe. Carcinogenesis (2007) 28:1334–40.