Association of MTHFR C677T and A1298C polymorphisms with non-Hodgkin lymphoma susceptibility: Evidence from a meta-analysis

Methylenetetrahydrofolate reductase (MTHFR) is an important enzyme involved in folate metabolism and DNA synthesis. A number of studies have examined the association of MTHFR C677T and A1298C polymorphisms with non-Hodgkin lymphoma (NHL) susceptibility; however, the conclusions were contradictory. We searched available publications assessing the polymorphisms of MTHFR and NHL susceptibility from MEDLINE, EMBASE and CBM. Genotype-based mRNA expression analysis was performed using data from 270 individuals with three different ethnicities. Ultimately, a total of 7448 cases and 11146 controls from 25 studies were included for the C677T polymorphism, 6173 cases and 9725 controls from 19 studies for the A1298C polymorphism. Pooled results indicated that neither C677T nor A1298C polymorphism was associated with NHL susceptibility. However, C677T polymorphism showed a statistically significantly increased risk for Caucasians, but a decreased risk for Asians in the subgroup analysis by ethnicity. The same variants may confer increased susceptibility to develop follicular lymphoma (FL). Moreover, A1298C polymorphism was associated with increased NHL risk for Asians. This meta-analysis indicated that C677T polymorphism was associated with altered NHL susceptibility for Caucasians, Asians and FL. Increased NHL risk was also shown for A1298C among Asians. These findings warrant validation in large and well-designed prospective studies.

C ancer has been recognized as one of the most formidable public health challenges, with estimates of nearly 12.7 million cancer cases and 7.6 million cancer deaths having occurred in 2008. Non-Hodgkin lymphoma (NHL) is the eighth most frequently diagnosed tumor type among men and the tenth among women worldwide 1 . Noteworthily, according to GLOBOCAN 2008 estimates, 355900 new cases and 191400 deaths might have occurred in 2008. There are a variety of different subtypes of NHL. Generally NHL is categorized into two major groups: B cell lymphomas and T cell lymphomas, with B cell lymphomas making up majority of NHL cases (about 85%). Diffuse large B-cell lymphoma (DLBCL) and follicular lymphoma (FL) are the two major subtypes of B cell lymphomas 2,3 . North America, Australia/New Zealand, and Northern, Western, and Southern Europe with the highest incidence rates, and South-Central and Eastern Asia and the Caribbean with the lowest incidence rates 1 . Certain immunodefective conditions (e.g., immunosuppression, Epstein-Barr virus and human immunodeficiency virus infections) as well as occupational exposures to herbicides and chlorinated organic compounds are the main risk factors for NHL 1,4 . Moreover, deficiency of nutrients (e.g., folate) related to one-carbon metabolism is also a well-established risk factors for NHL 5 , which has been reported to likely lead to immune responses impaired 6 .
Folate is an important coenzyme in DNA synthesis. Methylenetetrahydrofolate reductase (MTHFR) is one of the most critical enzymes involved in folate metabolism and DNA synthesis, which catalyses the conversion of 5, 10-methylenetetrahydrofolate to 5-methyltetrahydrofolate irreversibly 7 . Reduced MTHFR activity may play an inhibitory role on the 5-methyltetrahydrofolate pathway, and may lead to the accumulation of 5-methylenetetrahydrofolate, and consequentially methylation of dUMP to dTMP is decreased. Uracil can misincorporate into DNA when the methylation of dUMP to dTMP is deficient, consequentially results in DNA double-strand breaks and other anomalies, if not repaired which may lead to carcinogenesis [8][9][10] .
The MTHFR gene is located at chromosome 1p36. 3. Among all the identified single nucleotide polymorphisms (SNPs) in this gene, C677T (Ala222Val, rs1801133) and A1298C (Glu429Ala, rs1801131), likely associated with reduced enzyme activity, have been widely investigated in a variety of diseases, such as psychiatric disorders 11 , congenital anomalies 12 , colorectal cancer 13 , and so on. Numerous studies have focused on the relationship between these two polymorphisms and NHL risk  , but the conclusions remain controversial. The discrepancies among studies may be ascribed to the relatively small sample size in each investigation as well as ethnicity difference. Therefore, we performed this meta-analysis using genotype data from all eligible investigations to provide a more precise evaluation of the association of MTHFR C677T and A1298C polymorphisms with NHL susceptibility.

Results
Study characteristics. A total of 38 articles were initially indentified from MEDLINE and EMBASE, five more articles were indentified from the reference of retrieved studies, and ultimately, additional two articles were indentified from the CBM database ( Figure 1). Of them, 14 articles were excluded after title and abstract assessment, while 31 articles met the crude inclusion criteria and were further evaluated. Among these remaining 31 publications, we further excluded two studies 15,26 that were covered by other included investigations 20,24 , three case-only designed studies [40][41][42] , and one more study in which genotype frequency data in the controls for both of the C677T and A1298C polymorphisms were deviated from HWE (P 5 0.009 for C677T and P 5 0.040 for A1298C) 38 . Overall, 25 articles were included in the final meta-analysis. There were 25 articles with 7448 cases and 11146 controls for the C677T polymorphism, and 19 studies with 6173 cases and 9725 controls for the A1298C polymorphism (Table 1). Intriguingly, in the remaining studies, 12 and nine studies pertaining to C677T polymorphism provided detailed genotype frequency data for the DLBCL and FL subtype, respectively, while nine and eight studies regarding A1298C polymorphism provided detailed genotype data for the these two subtypes, respectively (Supplemental Table 1). Sample sizes of case in the incorporated studies ranged from 28 to 1103 for C677T polymorphism and from 31 to 1124 for A1298C polymorphism.
For the C677T polymorphism, there were 19 studies conducted in Caucasians, three studies in Asians, and three studies in mixed ethnic group. Of these studies, 18 were population based (PB) and seven were hospital based (HB) designed, respectively. Furthermore, 11 studies were considered as low quality (quality socre # 9), and 14 (56%) were considered as high quality (quality score . 9). As to the A1298C polymorphism, there were 14 studies conducted in Caucasians, two studies in Asians, three studies in mixed ethnic group. While divided by the source of control, 15 studies were PB and four were HB. Among them, 6 were classified into low quality and 13 were classified into high quality. The mainly adopted genotyping methods were PCR-restriction fragment length polymorphism (12 and eight studies for C677T and A1298C, respectively) and Taqman (10 studies and eight studies for the C677T and A1298C, respectively).
Meta-analysis results. As shown in Table 2 and Figure 2, pooled analysis did not yield a significant association between MTHFR C677T polymorphism and overall NHL risk (homozygous: OR    The correlation between the mRNA expression and genotypes. The correlation between MTHFR mRNA expressions levels by the genotypes were explored for three ethnic groups (i.e., CEU, YRI and Asian) and the whole group (Table 3). No significant alteration in the mRNA expression levels was found for the C677T polymorphism under all the genetic models.
Interestingly, we found that C variant allele of MTHFR A1298C polymorphism significantly correlated with increased MTHFR  Heterogeneity and sensitivity analyses. As shown in Table 2, substantial heterogeneities were observed among all investigations for the C677T polymorphism and NHL risk (homozygous: P 5 0.092; heterozygous: P 5 0.027; dominant model: P 5 0.006 and allele comparing: P 5 0.006), except for the recessive model (P 5 0.465). We also observed considerable heterogeneities for the A1298C polymorphism (homozygous: P , 0.001; recessive model P , 0.001; dominant model: P 5 0.064 and allele comparing: P , 0.001), except for the heterozygous model (P 5 0.471). The metaregression analysis did not yield any significant difference between subgroup analysis. Thus, leave-one-out sensitivity analyses indicated that no single study could alter the pooled ORs obviously (data not shown).
Publication bias. The shape of the funnel plots seemed asymmetry for the C677T and A1298C polymorphisms (Figure 3 and

Discussion
To our knowledge, the current meta-analysis is the largest one to investigate the association between MTHFR gene polymorphisms and NHL risk. Pooled analysis for the C677T polymorphism contained 25 studies with a total of 7448 NHL patients and 11146 controls; meanwhile, pooled analysis for the A1298C polymorphism encompassed 19 studies with 6173 NHL patients and 9725 controls. The meta-analysis observed no significant association between MTHFR C677T and A1298C polymorphisms and overall NHL risk. However, stratified analyses by ethnicity revealed that the C677T polymorphism increased NHL risk for Caucasians but decreased risk for Asians. The study of the same SNP also observed a significantly increased risk of FL, but not DLBCL, when the analysis was stratified by subtype of NHL. Moreover, the A1298C polymorphism was associated with increased risk for Asians, while no effect was observed for other ethnic groups. Interestingly, we found the MTHFR mRNA expression levels was slightly increased in the Asians carrying 677T alleles (P 5 0.052), which was in accordance with our findings that C677T polymorphism was significantly associated with decreased NHL risk in the Asian group. Moreover, we also found the 1298C carriers showed significantly decreased MTHFR mRNA expression (P 5 0.028), which corresponded to the evidence of association of 1298C polymorphism with increased NHL risk. Therefore, this results suggested that our findings from association studies for Asians may be biological plausible. MTHFR gene variants play an important role in the outcome of NHL patients. Examination of polymorphisms in the folate pathway genes might facilitate to  reduce chemotherapy toxicity and improve survival by indicating when dose adjustments or alternative treatments are needed 40 .
Folate is a critical nutrient and coenzyme involved in DNA synthesis and methylation, and folate deficiency has been reported associated with numerous malignancies 43 . The product of MTHFR gene plays an important role in the methylation of homocysteine into methionine, sequentially leading to DNA methylation 44 . MTHFR C677T and A1298C polymorphisms were widely investigated in varieties of cancers. The former genetic variation is located in exon 4, and can lead to amino acid change from alanine to valine, which was first reported in 1995. This variation was reported to associate with reduced enzyme activity from a total of 40 subjects, with the CT and TT genotypes having ,60% and ,30% of the wild-type enzyme activity, respectively 45 . The A1298C polymorphism is located in exon 7. It has also been reported the CC genotype carriers having ,60% of wild-type enzyme activity 46 . Some of the previous studies failed to found this association 21,47 . Based on the largest meta-analysis to date, none of these two polymorphisms of interest was associated with overall NHL risk. Nonetheless, stratification analyses by ethnicity detected that analyzed SNPs significantly altered the risk of developing NHL in different ethnic groups. SNPexp online tool allows us to evaluate genotypes of MTHFR C677T and A1298C polymorphisms and their respective MTHFR transcript expression levels. With this in mind, we further investigated whether the biological results are in accordance with the observed association. We performed genotype-based mRNA expression analysis using the data from 270 individual with three ethnicities. We did not find a similar trend in the mRNA expression for the Caucasians but for the Asians, which may be due to the fact that the genotype counts for the homozygous variants is relatively small.
As so far, only two meta-analyses, which were nested in casecontrol studies, have investigated the association of MTHFR C677T and/or A1298C polymorphisms and NHL susceptibility. The study carried out by Lee et al. 28 only studied the C677T poly-   37 . Some of the significant findings described above were not validated in our meta-analysis. For instance, the finding that the MTHFR C677T polymorphism was associated with elevated risk for DLBCL in homozygous model 28 could not be duplicated in our study. Such associations were no longer significant in the current meta-analysis upon the inclusion of seven more studies. We also found some significant associations that were not observed in the previous studies, one example of which was that we found the C677T polymorphism decreased NHL risk for Asians, whereas the A1298C polymorphism conferred an increased risk to them. These new findings may be ascribed to the inclusion of more investigations with much large sample size in the current meta-analysis. It was noteworthy that we found the C677T polymorphism was associated with increased NHL risk for Caucasians while with decreased NHL susceptibility for Asians. The opposite findings in different ethnic groups may be resulted from ethnicity difference as well as the number of investigations. Earlier studies indicated that diet of Western contains high heterocyclic amines and polycyclic aromatic hydrocarbons 48,49 , which is relatively low in the diets of other ethnicities, which may contribute to the different effects of C677T polymorphism on cancer risk. Though we have included the latest investigations as well as publications written in Chinese, the current meta-analysis still has several limitations to be addressed. First, the sample size of cases from most eligible studies is relatively limited (,500), except for five studies 21,23,28,29,32 . Among the five studies, only one study has case more than one thousand 29 . The relative limited cases may have compromised statistical power. Second, we only included the studies written in English and Chinese, and might miss some investigations written in other languages. Third, heterogeneity was observed under some genetic models, so the results should be interpreted cautiously. Finally, our results were derived from unadjusted estimates due to lack of the original data, such as age, gender, smoking, drinking and dietary intaking habits, occupational exposures, as well as virus infections.
In summary, this study indicated that C677T and A1298C polymorphisms in the MTHFR gene may be associated with NHL susceptibility, especially for Caucasians, Asians and FL. Well-designed prospective studies with large sample size should be conducted to validate our findings.

Methods
Literature search strategy. We identified publications examined the association between MTHFR gene polymorphisms and NHL from MEDLINE and EMBASE using the following search items: ''MTHFR or methylenetetrahydrofolate reductase'', ''polymorphism or variant or variation'' and ''non-Hodgkin lymphoma or non-Hodgkin's lymphoma or NHL'' (the last search updated was on January 8, 2014). We also identified related publications written in Chinese from Chinese Biomedical (CBM) database using the combinations terms of ''MTHFR'', ''NHL'' and ''polymorphism'' in Chinese. Besides, we identified additional studies by searching of the references from retrieved studies manually. We only included the latest or the largest sample size studies in our final meta-analysis, if there exists more than one article published using the same subjects or overlapping data.
Inclusion and exclusion criteria. Studies included in the final meta-analysis had to meet the following criteria: (1) evaluate MTHFR C677T and/or A1298C polymorphisms and NHL risk; (2) be a case-control study, nested case-control study or a cohort study; (3) written in English or Chinese; (4) contain SNP genotype data; (5) independent from other studies; (6) provide sufficient data to calculate odds ratios (ORs) and their corresponding 95% confidence intervals (CIs).
The studies were excluded if genotype frequency data in the controls for MTHFR C677T and A1298C polymorphisms demonstrated departure from Hardy-Weinberg equilibrium (HWE) without further evidence showing that genotype distribution of other SNPs in controls followed HWE. In addition, case-only studies, case reports, conference abstract, reviews, meta-analyses and studies without detailed data were excluded.
Data extraction. Two investigators (JH and XL) independently extracted the following information from all eligible studies according to the inclusion and exclusion criteria: the first author's surname, year of publication, country of origin, ethnicity, cancer type and subtype (FL and DLBCL), control source (population based or hospital based), the total number of cases and controls, genotyping methods, minor allele frequency (MAF) for controls, P values for HWE for the control subjects and numbers of cases and controls with the CC, CT and TT genotypes for the C677T polymorphism and AA, AC and CC genotypes for the A1298C polymorphism. Any disagreement was resolved by discussion within our team numbers until consensus was reached.
Genotype based mRNA expression analysis. The genotypes data for MTHFR C677T and A1298C polymorphisms were available from HapMap (http://hapmap.ncbi.nlm. nih.gov/) for 270 subjects with three different ethnicities and their corresponding mRNA expression levels data were available from SNPexp (http://app3.titan.uio.no/ biotools/tool.php?app5snpexp) as described previously [50][51][52][53] . Quality assessment. Two investigators assessed the quality of each investigation using the quality assessment criteria (Supplemental Table 2), which was derived from previously published meta-analysis of molecular association studies 54 . Quality scores of studies ranged from 0 (lowest) to 15 (highest). Studies with scores # 9 were categorized into low quality, while those with scores . 9 were considered as high quality. A third investigator (JZ) would be involved if there existed any disagreement.
Statistical methods. Crude ORs and their corresponding 95% CIs were used to evaluate the strength of associations between MTHFR gene polymorphisms and NHL risk. The pooled ORs were estimated for C677T polymorphism under the homozygous model (TT vs. CC), heterozygous model (CT vs. CC), recessive model (TT vs. CT 1 CC), dominant model (CT 1 TT vs. CC) and allele comparison (T vs. C). The same genetic models were also adopted for A1298C polymorphism as followed: homozygous model (CC vs. AA), heterozygous model (AC vs. AA), recessive model (CC vs. AC 1 AA), dominant model (AC 1 CC vs. AA) and allele comparison (C vs. A). Goodness-of-fit chi-square test was used to test deviation from HWE for the genotypes of controls. P , 0.05 was considered significant. The Chi-square based Q-test was performed to evaluate the heterogeneity across the studies. The random-effects model 55 was chosen when significant heterogeneous exist (P heterogeneity , 0.10); otherwise, fixed-effects model (the Mantel-Haenszel method) 56 would be adopted. Stratification and metaregression analyses were conducted to explore the potential source of heterogeneity across studies. Furthermore, stratification analyses were conducted by ethnicity (i.e., Asians, Caucasians, and Mixed that contained more than one ethnic group), control source (hospital-based and population-based), quality score of studies (low and high), and tumor subtype (FL and DLBCL). Sensitivity analysis was performed to assess the stability of the results by sequentially excluding one study at a time and recalculating the pooled ORs and their corresponding 95% CIs. Furthermore, both the Begg's funnel plot 57 and the Egger's linear regression test 58 were performed to assess the potential publication bias. The differences in mRNA expression levels for different genotypes were evaluated by one-way ANOVA, and the mRNA expression levels trend were evaluated by General linear model. All statistical tests were performed with STATA software (version 11.0; Stata Corporation, College Station, TX) and SAS software (version 9.1; SAS Institute, Cary, NC). All the statistics were two-sided, and P , 0.05 was considered as significant findings.