Introduction

Colorectal cancer (CRC) is one of the most common cancers, and the development of CRC is determined by a series of risk factors, such as genetic polymorphisms, environmental factors, lifestyles, and the complex interaction between these factors (de Kok and van Maanen 2000). Colorectal adenomas are well-established precursor lesions for colorectal cancer (Fearon and Vogelstein 1990; Peipins and Sandler 1994); prevention of colorectal adenomas may decrease the occurrence of colorectal cancer.

It is widely accepted that gene methylation is implicated in carcinogenesis (Toyota et al. 1999). 5,10-Methylenetetrahydrofolate reductase (MTHFR) is involved in DNA methylation; it manipulates the levels of folate, one of the methyl group donors (Kim et al. 1996). MTHFR catalyzes the irreversible conversion of 5,10-methylenetetrahydrofolate (required for purine and thymidine synthesis) to 5-methyltetrahydrofolate, which is necessary for methionine synthesis (Ueland et al. 2001). Insufficient thymidylate results in uracil misincorporation into DNA, leading to single-strand and double-strand breaks and increasing the incidence of DNA misrepair, thus increasing the risk of genetic instability (Blount et al. 1997).

MTHFR C677T (Ala222Val) (Frosst et al. 1995) and A1298C (Glu429Ala) (van der Put et al. 1998) are two common polymorphisms. Both of these polymorphisms decrease the activity of the enzyme; this effect is more pronounced in the homozygous than the heterozygous state (Ogino and Wilson 2003). The MTHFR C677T polymorphism exerts opposite effects in colorectal carcinogenesis, depending on the balance of thymidylate synthesis and DNA methylation, which affects the folate pool (Giovannucci et al. 2003). An association between MTHFR and genetic susceptibility to CRC and colorectal adenoma has been widely documented but with inconsistent results. A single study may be underpowered to detect a possible small effect from a real association of these polymorphisms with CRC and colorectal adenoma, especially when the sample size is relatively small. Different populations and sampling methods were applied in each study, which makes it complicated to interpret the data. Thus, a rigorous quantitative synthesis method is required in order to assess the discrepancy.

Here, we performed a meta-analysis of all eligible case-control studies in order to address the association of MTHFR C677T and A1298C polymorphisms with CRC and colorectal adenoma. Our results indicated that both the 677T and 1298C alleles might have a protective effect from CRC across a worldwide population. The 677T allele may exert its effect on CRC in a recessive genetic model in Asians, while the 1298C allele may decrease CRC risk in a recessive genetic model in Caucasians. No associations of C677T and A1298C polymorphisms were found to increase the risk of colorectal adenoma, which indicated that the two polymorphisms may protect from the adenoma becoming malignant cancer. More studies are needed to further confirm the protective roles of these polymorphisms in CRC development, and their effects in different populations.

Materials and methods

Identification and eligibility of relevant studies

To identify all articles that examined the association of MTHFR C677T and A1298C polymorphisms with CRC and colorectal adenoma, we conducted a literature search of the PubMed database (from January 1991 to August 2006) using the following keywords and subject terms: “MTHFR,” “polymorphism,” and “colorectal cancer” or “colorectal adenoma.” “Colon cancer (adenoma)” and “rectal cancer (adenoma)” were used to replace “colorectal cancer (adenoma)” in further searches of related studies. References in retrieved articles were screened. Abstracts, case reports, editorials, and review articles were excluded. Studies included in the current meta-analysis had to meet all of the following criteria: (a) an unrelated case-control design was used, (b) genotype frequency was available, and (c) the genotype distribution of the control population must be in Hardy–Weinberg equilibrium (HWE).

Data extraction

Data were collected on the genotypes of C677T or A1298C based on colorectal cancer and colorectal adenoma, respectively. First author, year of publication, ethnicity of study population, and characteristics of cases and controls were described.

Statistical analysis

The meta-analysis examined the overall association of the C677T allele T with the risk of CRC and colonrectal adenoma compared to that for allele C; homozygote TT was contrasted with CC, recessive (TT vs. CT + CC) and dominant (TT + CT vs. CC) models for allele T. The same contrasts were performed for allele C of the A1298C polymorphism.

Odds ratios (OR) corresponding to a 95% confidence interval (CI) were applied to assess the strength of association of C677T or A1298C with CRC and colorectal adenoma since case-control studies were used, and OR was calculated according to the method of Woolf (1955). A chi-square-based Q statistic test was performed to assess the between-study heterogeneity (Lau et al. 1997). Heterogeneity was considered significant for P<0.10 because of the low power of the statistic. A fixed-effect model using the Mantel–Haenszel method and a random-effects model using the DerSimonian and Laird method were used to pool the results (Petitti 1994). In the absence of between-study heterogeneity, the two methods provide similar results. Random effects are more appropriate when heterogeneity is present. The significance of the pooled OR was determined by the Z test, a P value of <0.05 was considered significant. For each genetic contrast, subgroup analysis according to ethnicity was only considered for Asian and Caucasian populations, in order to estimate ethnic-specific OR. The A1298C comparisons for colorectal adenoma were not stratified for subgroup analysis because of the limited studies included.

Publication bias was investigated by funnel plot, in which the standard error in log(OR) in each study was plotted against OR. An asymmetric plot suggested possible publication bias. Funnel plot asymmetry was assessed via Egger’s linear regression test, a linear regression approach to measuring funnel plot asymmetry on the natural logarithm scale of the OR (Egger et al. 1997). The significance of the intercept was determined by the t-test as suggested by Egger, and a P value of <0.05 was considered significant.

Hardy–Weinberg equilibrium was tested for by the chi-square test for goodness of fit using a web-based program (http://ihg.gsf.de/cgi-bin/hw/hwa1.pl). Analyses were performed using the software Stata version 7, ReviewManage 4.2. All P values were two-sided.

Results

Selection of studies

Twenty-eight articles were retrieved based on the search criteria for colorectal cancer, and 20 met our inclusion criteria. For various reasons, eight articles were not included. Genotype distributions in the control populations in five studies (Le Marchand et al. 2005; Jiang et al. 2005; Ma et al. 1997; Chen et al. 2002; Miao et al. 2005) significantly deviated from HWE. No detailed genotyping information was available from Marugame et al. (2003) and Kawakami et al. (2003). One article (Slattery et al. 1997) was replaced with its updated study (Slattery et al. 1999). All 20 eligible articles described C677T (Chen et al. 1996; Park et al. 1999; Slattery et al. 1999; Ryan et al. 2001; Keku et al. 2002; Le Marchand et al. 2002; Sachse et al. 2002; Shannon et al. 2002; Heijmans et al. 2003; Plaschke et al. 2003; Pufulete et al. 2003; Toffoli et al. 2003; Curtin et al. 2004; Jiang et al. 2004; Kim et al. 2004; Ulvik et al. 2004; Yin et al. 2004; Matsuo et al. 2005; Otani et al. 2005; Wang et al. 2006), and 11 of them also provide data on A1298C (Table 1). Among the 20 eligible articles included, 70% (14/20) stated that the age and sex status were matched between the case and control populations. PCR-RFLP was used to validate genotype in all but two articles. Ulvik et al. (2004) used real-time PCR and Otani et al. (2005) used Taqman SNP genotyping assay. All of the articles used blood samples for genotyping except Shannon et al. (2002), which used frozen tissue samples.

Table 1 Characteristics of eligible studies considered in the meta-analysis
Full size table

Thirteen articles were retrieved for colorectal adenoma. Crabtree et al. (2004) only examined the MTHFR polymorphism with respect to the severity of the adenoma. Ulvik et al. (2001) contained no detailed information on genotyping. These two studies were discarded. Among 11 eligible articles (Chen et al. 1998; Ulrich et al. 1999; Levine et al. 2000; Marugame et al. 2000; Giovannucci et al. 2003; Pufulete et al. 2003; Boyapati et al. 2004; Goode et al. 2004; van den Donk et al. 2005; Hirose et al. 2005; Mitrou et al. 2006), 27% (3/11) stated that the age and sex status were matched between the case and control populations. PCR-RFLP was used to validate the genotype in all of the studies. All articles included in the meta-analysis provided information on C677T, and three of them also genotyped A1298C (Table 1).

Keku et al. (2002) provided data on two ethnicities (African–Americans and Caucasians), while Le Marchand et al. (2002) reported on three separate populations: Japanese, Caucasians, Hawaiians. Thus, each subpopulation in these two articles was treated as a separate study in the meta-analysis. Studies that provide genotyping data for mixed populations are indicated as having “unknown” ethnicity in Table 1.

Summary statistics

A total of 10,131 CRC patients and 15,362 controls (C677T), 4,764 CRC patients and 6,592 controls (A1298C), 4,616 adenoma patients and 6,338 controls (C677T), and 1,272 adenoma patients and 1,684 controls (A1298C) were investigated. Electronic Supplementary Material Table 1 shows the genotype and allele frequencies of both CRC and colorectal adenoma patients as well as the controls in the eligible studies. The allele frequencies were calculated for controls from the corresponding genotype distributions. The variant allele frequencies of 677T and 1298C in the individual studies are also shown in Table 1.

The C677 allele was a little less common among controls of Asian descent (62.4%, 95% CI 54.0–70.8) than in controls of European descent (67.8%, 95% CI 65.1–70.5). The A1298 allele was slightly more common in controls of Asian descent (77.7%, 95% CI 68.0–87.3) than in controls of European descent (71.0%, 95% CI 65.8–76.2). Overall, the prevalence of 677TT homozygosity was 15.6% and 16.8% in control subjects of Asian and European descent, respectively. The prevalence of 1298CC homozygosity was 5.3 and 9.0% in control subjects of Asian and European descent, respectively.

Genetic contrasts

No significant heterogeneity existed between the 23 studies upon comparing the C677T C to the T allele in CRC (P=0.22). First, a fixed-effect model was used to pool the results (Fig. 1A). There was evidence that the T allele resulted in decreased susceptibility to CRC in a worldwide population. The overall OR was 0.93, 95% CI (0.89–0.98) by fixed effects (P=0.004). No significant between-study heterogeneity was found within each subgroup. The decreased effect of the T allele on the susceptibility was seen in both subgroups, with eight comparisons in populations of Asian descent (P=0.001, OR=0.87, 95% CI 0.81–0.95) and nine comparisons in those of Caucasian descent (P=0.03, OR=0.93, 95% CI 0.88–0.99). Meta-analyses for other genetic contrasts further suggested that the 677T allele was more likely to affect the CRC risk in a recessive genetic model worldwide (P=0.009, OR=0.86, 95% CI 0.76–0.96, random effects model), with between-study heterogeneity present (P=0.06). This result implies that a dose effect may exist for this polymorphism. Interestingly, this effect is only significant in Asians (P=0.0005, OR=0.75, 95% CI 0.64–0.88, P=0.71 for heterogeneity), not in Caucasians (P=0.16, OR=1.02, 95% CI 0.79–1.32, P=0.006 for heterogeneity). The TT versus CC contrast also showed that the association between C677T polymorphism and CRC risk was maintained in Asians but not Caucasians (Table 2).

Fig. 1A–B
figure 1

Overall meta-analysis for C677T polymorphism (T vs. C allele) in CRC and colorectal adenoma. Point estimates of the OR for each study and the accompanying 95% CI values obtained with a fixed effects model are shown. A Analysis of the comparison with CRC, and B analysis of the comparison with colorectal adenoma. n indicates the total number of T or C alleles, N indicates the total number of T plus C alleles

Full size image
Table 2 Summary of ORs for various genetic contrasts performed when investigating the association of C677T and A1298C polymorphisms with CRC and colorectal adenoma
Full size table

Eleven studies investigated the association between C677T polymorphism and colorectal adenoma. No heterogeneity was found among the studies (P=0.87) when comparing the C to the T allele. No association was found with colorectal adenoma when a fixed model was used to perform a worldwide allele comparison (Fig. 1B, P=0.52, OR=0.98, 95% CI 0.93–1.04), as well as comparisons for Asian (P=0.78, OR=0.98, 95% CI 0.85–1.13) and Caucasian (P=0.35, OR=0.96, 95% CI 0.88–1.05) populations. No evidence of association between C677T and colorectal adenoma was discerned upon comparing with other genetic models.

There was heterogeneity among the 14 studies comparing the C versus the A allele (P=0.09) for A1298C polymorphism. Thus, the random effects model is more appropriate than the fixed model. The C allele tends to decrease the risk of CRC, but no significant association of the C allele with CRC susceptibility was found in a worldwide population (Fig. 2A, P=0.08, OR=0.93, 95% CI 0.85–1.01) and in an Asian population (P=0.2, OR=0.89, 95% CI 0.74–1.06, P=0.03 for heterogeneity). No heterogeneity (P=0.22) existed among the studies of Caucasian populations, and the fixed model did not reveal an association between 1298C and CRC risk (P=0.08, OR=0.90, 95% CI 0.79–1.01). Evidence for an association with CRC was discerned upon comparing the 1298CC genotype with the 1298AA genotype in a worldwide population (P=0.005, OR=0.80, 95% CI 0.69–0.94, P=0.18 for heterogeneity) and a Caucasian subgroup (P=0.04, OR=0.74, 95% CI 0.55–0.98, P=0.23 for heterogeneity). No association with 1298CC that decreases CRC risk in Asians was found using a random-effects model (P=0.35, OR=0.78, 95% CI 0.46–1.33, P=0.05 for heterogeneity) existed in the homozygote contrast. Importantly, the 1298C allele also tends to decrease CRC risk in a recessive genetic model worldwide (P=0.005, OR=0.81, 95% CI 0.70–0.94, P=0.40 for heterogeneity). This effect is only significant in the Caucasian subgroup (P=0.04, OR=0.75, 95% CI 0.57–0.99, P=0.35 for heterogeneity), not in Asians (P=0.17, OR=0.81, 95% CI 0.60–1.09, P=0.11 for heterogeneity) using a fixed effects model.

Fig. 2A–B
figure 2

Overall meta-analysis for A1298C polymorphism (C vs. A allele) in CRC and colorectal adenoma. Point estimates of the OR values for the studies are shown, as well as the accompanying 95% CI values obtained using a random effects model in (A; analysis of the comparison with CRC) and using a fixed effects model in (B; analysis of the comparison with colorectal adenoma). n indicates the total number of C or A alleles, N indicates the total number of C plus A alleles

Full size image

Only three studies reported A1298C polymorphism in colorectal adenoma, one for an unknown population and the other two for populations of Caucasian descent. Ethnic subgroup analysis was not performed due to limited studies. No association was found between A1298C and colorectal adenoma in a worldwide population when comparing the A and C alleles (Fig. 2B, P=0.15, OR=1.09, 95% CI 0.97–1.22, P=0.24 for heterogeneity), as well as for other genetic contrasts (Table 2).

Sensitivity analysis

Sensitivity analysis was performed by sequential omission of individual studies from various contrasts performed on a worldwide population and on Caucasian or Asian subgroups. The pooled ORs (including 95% CI) were consistently below 1.0 in all of the comparisons for the three groups (worldwide population, Caucasian, Asian), with either fixed or random effect models applied, except for the dominant genetic model of C677T. The results did not alter the protecting role exhibited by the 677T allele against the CRC risk when using a recessive genetic model, indicating that the significance of the pooled ORs was not excessively influenced by any single study (data not shown). When individual studies were sequentially omitted for A1298C under homozygote and recessive contrasts in the worldwide population and for Caucasians, the pooled ORs were consistently below 1.0, suggesting that the protecting role exhibited by 1298C against the CRC risk when using a recessive genetic model is not affected by a single study. The associations of the C677T and A1298C polymorphisms with colorectal adenoma did not change during the sensitivity analysis.

Publication bias

A funnel plot for the comparison of the 677C allele with the 677T allele in the OR analysis and Egger’s test provided no evidence for funnel plot symmetry for both colorectal adenoma (t=0.42, P=0.69) and CRC (t=1.50, P=0.15). Similarly, no publication bias was detected for the A versus C allele contrast of A1298C polymorphism for both colorectal adenoma (t=−0.36, P=0.72) and CRC (t=0.87, P=0.54).

Discussion

This meta-analysis examined the association of MTHFR C677T and A1298C polymorphisms with susceptibility to CRC and colorectal adenoma. Little et al. (2003) found a protective effect of MTHFR C677T against CRC risk using a meta-analysis on 14 studies in 2003, including Ma et al. (1997), which deviated from HWE. A protective effect of MTHFR A1298C against colorectal cancer risk was also found using six studies. No association of MTHFR C677T with colorectal adenoma was found in their meta-analysis using six studies. Our meta-analysis results are consistent with those of Little et al. (2003), which were obtained by recruiting more up-to-date studies. We also investigated the association of MTHFR A1298C with colorectal adenoma risk. Importantly, we analyzed the effect of C677T and A1298C on CRC and colorectal adenoma risk by stratifying the ethnic populations studied.

A significant association between the MTHFR C677T polymorphism and CRC in a worldwide population for the overall effect of allele T versus C was found. Subgroup analysis suggested 677C polymorphism was a protecting factor against susceptibility to CRC for Asians when either homozygous (TT vs. CC) or recessive genetic model contrasts were performed. Significant associations were found between MTHFR A1298C polymorphism and CRC for the homozygote contrast and recessive genetic models. However, the Asian population does not show an association in the homozygote and recessive genetic analysis, in contrast to Caucasians, which retain the protecting effect of the 1298C against CRC risk. Thus, both the 677T allele and the 1298C allele had a decreased effect on CRC risk under the recessive genetic model, but the effect may be more pronounced in Asians and Caucasians, respectively. We investigated five studies which provided case and control numbers of the nine combined genotypes of C677T and A1298C (Curtin et al. 2004; Jiang et al. 2004; Keku et al. 2002; Plaschke et al. 2003; Yin et al. 2004). The probabilities of 677TC + 1298CC, 677TT + 1298AC, and 677TT + 1298CC genotypes were very rare from 3,243 cases and 4,198 controls of those studies. Ogino and Wilson (2003) calculated the haplotype frequencies of the C677T and A1298C polymorphisms in pooled general populations derived from data published in 16 articles. They found that most 677T and 1298C alleles were associated with the 1298A and 677C alleles, respectively. There may be an increased frequency of the very rare cis 677T/1298C haplotype in some parts of the UK and Canada, possibly due to a founder effect. Chen et al. (2002) reported the linkage disequilibrium between C677T and A1298C polymorphisms in MTHFR. Therefore, the effects on the CRC risk are independent for the two polymorphisms; both the 677T and 1298C alleles may be causal variants.

The 677T allele and the 1298C allele may decrease the risk of CRC, but they show no association with the risk of colorectal adenoma in a worldwide population and in the subgroups in our meta-analysis. The results suggest that the two polymorphisms may have a protecting effect, preventing colorectal adenoma from transforming into malignant CRC, since colorectal adenomas are precursor lesions of colorectal cancer (Fearon and Vogelstein 1990; Peipins and Sandler 1994). It is possible that MTHFR polymorphisms could modify the association between diet and colorectal adenomas. However, various results have been reported and no conclusions are available as yet. The MTHFR TT genotype in combination with a low folate status may be a risk factor for colorectal adenomas (Ulrich et al. 1999; Levine et al. 2000), or there may be no interaction (van den Donk et al. 2005).

Reasons for the conflicting results where MTHFR C677T and A1298C play different roles in different studies may be genetic heterogeneity in different populations and clinical heterogeneity in different studies. Potentially, differences in patient populations (e.g., in terms of age and years from onset, gender difference, and lifestyle) might cause different results. Shannon et al. (2002) showed that the MTHFR 677TT genotype conferred an increased risk of colorectal cancer on subjects older but not younger than 70 years. The risk effect may be dependent on gene methylation, which is affected by folate intake and its interaction with other risk factors. Chen et al. (1996) suggested a possible interaction between the MTHFR C677T polymorphism and folate status in CRC. TT genotype was protective in folate-replete subjects, whereas the combination of TT and low folate status conferred no protection, and even increased risk. Otani et al. (2005) reported no association between MTHFR C677T and A1298C–nutrient interaction in CRC, including vitamin B6, vitamin B12 and folate. Moreover, the association of alcohol consumption with colorectal cancer risk has been related to its anti-folate effects and subsequent effects on DNA methylation (Freudenheim et al. 1991; Giovannucci et al. 1993). Thus, diet, particularly low folic acid intake, can modify the effects of the MTHFR polymorphisms. However, no definite conclusion about the relation between folate intake and CRC risk has been possible so far. More careful stratification analysis that takes into account the clinical character and diet, smoking status and alcohol consumption are needed.

The 677T and 1298C SNPs in the MTHFR gene both decrease the activity of the enzyme, leading to hyperhomocysteinemia, particularly in folate-deficient states (Frosst et al. 1995; van der Put et al. 1998). Elevated plasma homocysteine concentration had been suggested to be a risk factor for schizophrenia (Muntjewerff et al. 2006), coronary heart disease (Klerk et al. 2002), and stroke (Casas et al. 2005), particularly in low folate status. 677TT homozygotes show significantly greater mean homocysteine and risks of those three diseases than people who are 677CC homozygotes. However, Lewis et al. (2005) found no strong association of the MTHFR 677T polymorphism with coronary heart disease in Europe, North America, or Australia using a meta-analysis, which might due to the higher folate intake in North America and Europe. Here, folate provides one-carbon groups for the methylation of homocysteine to form methionine, which then decreases the concentration of homocysteine (Hankey and Eikelboom 2005). Zintzaras recently used meta-analysis to examine the association of MTHFR C677T and A1298C polymorphisms with breast cancer and gastric cancer, respectively (Zintzaras 2006a, 2006b). No association was found with breast cancer, but an increased risk from MTHFR 677T and 1298C was indicated for gastric cancer, in contrast to our meta-analysis results which indicate that MTHFR 677T and 1298C might be protective against CRC. Decreased levels of methyltetrahydrofolate may adversely affect the methylation of oncogenes and tumor suppressor genes, contributing to carcinogenesis. Also, depletion of methylenetetrahydrofolate interferes with thymidylate biosynthesis, which may lead to an accumulation of deoxyuridylate DNA, and subsequent removal of this abnormal base may destroy the integrity of DNA (Houlston and Tomlinson 2001). Thus, the balance between DNA synthesis and DNA methylation, which is determined by the MTHFR polymorphisms, may influence the cancer risk, which is also affected by folate intake.

Only two of the studies involved in our meta-analysis investigated gene–gene interactions. Ulvik et al. (2004) found no association of MTHFR C677T and MTR (methionine synthase) A2756G with CRC risk, and Hirose et al. (2005) revealed a decreased risk of CRC from ALDH2*2 and MTHFR 677T. More research on gene–gene interactions such as those involving MTR, thymidylate synthase, and ALDH2 will provide a more comprehensive insight into the associations studied here.

In conclusion, our meta-analysis (along with other studies) suggests that the MTHFR 677T and 1298C alleles have no association with colorectal adenoma risk, and they may both provide protecting effects against CRC risk in a recessive genetic model; however, the protecting effect of 677T was only significant in Asians, while 1298C was only significant in Caucasians. Whether other MTHFR polymorphisms provide protective effects against CRC risk in different populations in a recessive genetic model requires further investigation. More studies or large case-control studies should be performed to clarify the possible roles of C677T and A1298C in CRC.