Using this large family-based case-control study of colorectal cancer, genetic variation in genes involved in folate cellular uptake and distribution, FOLR1, FPGS, GGH and SLC19A1, were unassociated with colorectal cancer risk. We observed no evidence that associations, should they exist, were modified by multivitamin use, folic acid use, or dietary/total intake of folate. Furthermore, we found no evidence of heterogeneity in the SNP risk estimates by family history of colorectal cancer, tumor subsite, or MSI status.
The solute carrier family 19 (folate transporter) member 1 (SLC19A1
, RFC1) transports folate compounds into cells and plays a role in maintaining intracellular concentrations of folate. Mean expression levels of the RFC1 protein have been shown to be higher in tumor tissues compared with normal colonic mucosa [29
] and higher expression of folate receptors has been associated with the resistance to folate antagonist drugs [30
]. The SLC19A1
-G80A polymorphism has been associated with alterations in folate and homocysteine (Hcy) metabolism in healthy individuals [31
]; that study further suggested that the variant SLC19A1
-80A allele was associated with higher levels of serum folate [31
]. Other studies have reported no association between this polymorphism and plasma folate or homocysteine [32
]. To date, studies have suggested no association for this polymorphism with cancers of the breast [35
] and colon [36
]; other studies have suggested a potentially increased risk of bladder (borderline) [37
] and esophageal cancers [38
] for the AA versus G/A or G/G genotypes. The homozygote variant genotype has also been reported as not associated with CIMP+ or CIMP− colon cancers [39
Folate-binding proteins also transports folates in cells. These exist in three isoforms (FRα, FRβ and FRγ) that are differentially expressed in various tissues. The FRα isoform, known as FOLR1, is the most widely studied and is over-expressed in colon tumors [29
]. Expression of FOLR1 has been shown to be an important prognostic marker in some studies [29
]. Ma and colleagues reported that FOLR1
gene inactivation in mice increased sensitivity to colon carcinogenesis [42
may confer a growth advantage to the tumor by modulating folate uptake or generating regulatory signals [43
]. This gene is highly polymorphic and a selected set of variants may be associated with homocysteine and folate levels [44
], although further study is needed. To our knowledge, no study has reported on the potential role of common genetic variants in this gene and cancer risk. We found no evidence to support the hypothesis that polymorphisms in this gene are associated with risk of colorectal cancer. FOLR1 has very high affinity for folic acid [17
], but we found no evidence that the association between any FOLR1
polymorphism and risk differed between individuals taking folate-containing supplements and those that did not.
Folates derived from dietary sources exist mainly as polyglutamated forms. Gamma-glutamyl hydrolase (GGH) removes glutamate residues from folylpolyglutamates, thereby permitting movement into or out of cells [13
]. When the expression of GGH increases, more rapid hydrolysis of cellular folylpolyglutamates results in the depletion of intracellular folates. Studies have suggested that selected polymorphisms in GGH
, (−401C>T, rs3758149 and −124T>G, rs11545076) may increase promoter activity when introduced into both hepatocellular liver carcinoma (HepG2) and breast cancer (MCF-7) cell lines [46
]. A recent study suggested that the G allele of the GGH
-124T>G polymorphism was associated with a stepwise increase in DNA uracil content, but not plasma total homocysteine levels [47
]. We found no association between either polymorphism and risk of colorectal cancer.
FPGS catalyzes an essential polyglutamation step in FOCM, the addition of multiple glutamates to compounds with the basic pteroylglutamate structure such as tetrahydrofolate and many other folate analogues [18
]. Polyglutamation of endogenous reduced folates allows for retention and accumulation of these essential cofactors within the cell. Low expression of FPGS in normal-appearing mucosa in the colorectum in individuals with colorectal cancer has been associated with poor survival [29
]. In a study that resequenced the FPGS
gene in four ethnic populations, five SNPs were shown to alter an amino acid and two of these non-synonymous SNPs, -R424C and -S457F, affected protein expression, in vitro
substrate enzyme kinetics, and efficacy of anti-folate therapy [48
]. Few studies have been conducted investigating the role of FPGS
polymorphisms in cancer risk [49
]. We found no evidence that any of the selected tagSNPs was associated with colorectal cancer risk. We did not include some known non-synonymous variants in FPGS
, and therefore further study may be warranted.
Polymorphisms in genes involved in the provision of methyl groups may be more important for the development of MSI-H colorectal cancers than for those with the MSI-L or MSS phenotype. The majority of sporadic MSI-H colorectal tumors show hypermethylation of the MLH1 gene promoter and CpG island methylator phenotype (CIMP) [51
]; therefore, because folates play a key role in methylation, genetic variants that influence folate levels may contribute to the risk of MSI-H tumors. This hypothesis has received little attention. One publication by Curtin et al. found little evidence that a selected set of functional variants in folate genes were associated with CIMP+ or CIMP− cancers except for MTHFR
]. We found limited evidence that variation in genes involved in the uptake and distribution of folates differential influence colorectal cancer risk by MSI-status.
This study has several strengths and limitations. The case-unaffected sibling design controls for any potential confounding by ethnicity and is more powerful for detecting gene-environment interactions than studies with population-based unrelated controls. However, a limitation to this design is that it may have lower power or detecting main effects [20
]. We used a validated semi-quantitative food frequency questionnaire, but these are subject to measurement error, which may introduce substantial biases, generally conservative [52
]. Detailed data using a food frequency questionnaire were collected for only a subset of the participants in this study, so we had limited power to detect heterogeneity by dietary and total folate intake. Strengths include the large sample size, comprehensive evaluation of genes, and the availability of systematically collected data on lifestyle and tumor characteristics.
In summary, we found no evidence that 29 common genetic variants in FOLR1, GGH, FPGS and SLC19A1 are associated with risk of colorectal cancer. Nonetheless given the limited data on these genes in cancer risk, further confirmation by other studies is needed.