Search tips
Search criteria 


Logo of nihpaAbout Author manuscriptsSubmit a manuscriptHHS Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
Eur J Clin Nutr. Author manuscript; available in PMC 2014 January 27.
Published in final edited form as:
PMCID: PMC3902360

Single nucleotide polymorphisms in uracil-processing genes, intake of one-carbon nutrients and breast cancer risk



The misincorporation of uracil into DNA leads to genomic instability. In a prior study, some of us identified four common SNPs in uracil-processing genes (rs2029166 and rs7296239 in SMUG1, rs34259 in UNG, and rs4775748 in DUT) that were associated with significantly altered levels of uracil in human DNA. We investigated whether any of these SNPs are associated with an altered risk of developing breast cancer and if one-carbon nutrients intake can modify their effects.


We genotyped the four SNPs in 1,077 cases of incident breast cancer and 1,910 age and race-matched controls in the Western New York Exposures and Breast Cancer (WEB) Study and examined associations with breast cancer risk and interactions with intake of folate, vitamins B6 and B12.


After adjustment for known risk factors for breast cancer, there was increased risk of breast cancer among postmenopausal women who were heterozygous for either of the SMUG1 SNPs (OR 1.29, 95% CI 1.07–1.56) and (1.29, 1.07–1.55). Among premenopausal women, increased risk associated with the SMUG1 rs2029166 genotype was limited to those with low folate intake. There were no other interactions with vitamins B6 or B12 intake.


Our study suggests that the four selected SNPs are not robust determinants of breast cancer risk, but that the two SNPs in SMUG1 might modestly alter the risk of breast cancer. However, the increase in risk among heterozygotes in the two SNPs in SMUG1, which is thought to be the most active glycosylase in vivo, raises the possibility that subtle ‘heterosis’ effects on cancer risk might be produced by these SNPs.

Keywords: Uracil-Processing Genes, Single nucleotide polymorphisms, Breast cancer


Uracil misincorporation into DNA occurs spontaneously at low levels as a result of cytosine deamination or misincorporation of dUMP during DNA replication (Kavli et al., 2007; Sousa et al., 2007). Under normal conditions, such lesions are rapidly repaired by the base excision repair mechanism initiated by uracil-DNA glycosylase enzymes (Krokan et al., 2002; Visnes et al., 2009). In humans, four uracil-DNA glycosylases have been identified, encoded by the UNG, SMUG1, MBD4 and TDG genes (Mohrenweiser & Jones, 1998; Barnes & Lindahl, 2004; Hung et al., 2005). In addition, mammalian cells have enzymes (dUTPases) that prevent the incorporation of uracil into DNA by de-phosphorylating dUTP into dUMP. In humans the only known UTPase is encoded by the DUT gene (Barnes & Lindahl, 2004; Sousa et al., 2007).

Intact mechanisms for the removal of uracil from DNA and prevention of its incorporation are essential for maintaining DNA integrity. It has been shown that failure to remove misincorporated uracil can lead to mutations after DNA replication or to double strand DNA breaks leading to chromosomal aberrations, both hallmarks of the cancer phenotype (Fenech, 2001; Kapiszewska et al., 2005; Sousa et al., 2007; Mashiyama et al., 2008).

Thus, a major factor influencing DNA integrity is the deoxyribonucleotide pool homeostasis, an equilibrium that is directly dependent on the adequate availability of certain elements of the B-vitamin family of micronutrients. Folate deficiency leads to low 5,10-methylene-THF concentrations, the essential co-factor for the conversion of dUMP to dTMP and may thereby increase the level of uracil incorporation into DNA (Blount et al., 1997). The chromosomal fragility and breaks that accompany folate depletion under some circumstances are thought to be caused by increased uracil misincorporation and attempted repair followed by the development of double strand breaks (Ames, 1999; Jang et al., 2005). Similarly, vitamins B12 and B6 appear to be crucial factors in maintaining genomic stability of human cells as well, since each play integral roles in recycling folate into the co-enzymatic form used for thymidine synthesis. (Fenech, 2001; Ames & Wakimoto, 2002; Mashiyama et al., 2008). Epidemiologic studies of humans have shown increased risk, although not consistently, of cancers including breast cancer, associated with low intake or low blood concentrations of folate, vitamin B12 and B6 (Choi & Mason, 2000; Giovannucci, 2002; Zhang et al., 2003). It has therefore been postulated that the DNA damage caused by uracil misincorporation may explain the increased cancer risk associated with inadequate availability of these vitamins (Ames & Wakimoto, 2002).

Very little is known about the functional effects of uracil-processing gene polymorphisms. It is thought that they can lead to altered enzyme activity, contributing to uracil concentrations, increased uracil misincorporation and therefore to human disease (Sousa et al., 2007). One study detected novel germline sequence variations in TDG, UNG and SMUG1 in colorectal cancer patients with familial aggregation, suggesting that these variants may play a role in disease susceptibility (Broderick et al., 2006). Other studies have described an association of a SNP in MBD4 with increased risk of lung (Shin et al., 2006; Miao et al., 2008) and esophageal cancer (Hao et al., 2004). Finally, in a previous study, some of us performed an extensive screening of SNPs in UNG, SMUG1, MBD4, TDG and DUT, and discovered that two SNPs in SMUG1 and one SNP in UNG were associated with increased blood uracil DNA concentration in carriers of the variant genotype, while one SNP in the DUT gene was associated with decreased uracil DNA concentration (Chanson et al., 2009).

The association of the four SNPs in SMUG1, UNG and DUT with altered uracil DNA levels and subsequent DNA damage led to the hypothesis that these SNPs may influence cancer risk and may affect DNA mutations in tumors. In order to test our hypothesis we sought out a cohort of subjects whose cancer was both a common one and one in which a reasonably extensive set of epidemiologic observations had shown it to be linked to the intake of 1-carbon nutrients. Breast cancer fulfilled both of these criteria. Moreover, we chose to study p53 mutations, since these mutations are among the most frequent genetic events in human cancer including 25–40% of breast cancers, and uracil misincorporation contributes to mutagenesis, and based on our previous observation that plasma uracil concentrations are correlated with strand breaks in the coding region of the human p53 gene (Chanson et al., 2009).

Therefore, we investigated the association of these four SNPs with breast cancer risk in a population based breast cancer case control study, the Western New York Exposures and Breast Cancer Study (WEB). Moreover, we investigated interactions with dietary folate, vitamin B6 and B12 and examined whether there was increased likelihood of mutations in the p53 gene in tumors associated with these SNPs.

Materials and methods


Subjects were participants in the Western New York Exposures and Breast Cancer (WEB) Study, a population based case-control study described in detail previously (Bonner et al., 2005; Tao et al., 2009). Briefly, 1,170 cases (35–79 years old) accrued between 1996–2001, residents of two counties in upstate New-York, were incident, primary, histologically confirmed breast cancer cases. Controls (n=2,115) were randomly selected from the New York State Department of Motor Vehicles driver’s license list age ≤ 65 years) and the Health Care Finance Administration rolls (age >65 years), and were frequency matched to cases on age (5-year intervals) and race. Both cases and controls had no previous history of cancer other than non-melanoma skin cancer. The study protocol was approved by the Institutional Review Boards of the University at Buffalo and all participating institutions, and written informed consent was obtained from all participants.

Data on participant demographics, medical history, reproductive history, lifetime alcohol consumption and other breast cancer risk factors were collected by extensive in-person interviews. A modified version of the self-administered Health Habits and History food frequency questionnaire (FFQ) was used for cases and controls to collect dietary intake data 12–24 months prior to interview, and the DietSys nutrient analysis software (version 3.7) was used to calculate nutrient intakes, adjusted for total energy (Block et al., 1986; Willett et al., 1997).

Laboratory methods

Blood or saliva samples were obtained from 1,099 (94%) cases and 1,945 (92%) controls, and genomic DNA was extracted from buffy coat fractions or buccal cells (17% of cases and 8% of controls). Genotyping for the four SNPs investigated was performed using an allelic discrimination method by real time PCR with TaqMan probes, in an ABI 7900HT real time PCR instrument (Appied Biosystems, Foster City, CA), according to the manufacturer’s instructions. As quality control measures, 10% duplicate samples were included in all measurements and all technical personnel was blinded to the case-control status of samples. The concordance rates between blinded duplicates were 98.1% for rs2029166, 98.7% for rs7296239, 97.4% for rs34259, and 97.4% for rs4775748. The call rates were 98.4% for rs2029166, 97.4% for rs7296239, 98.1% for rs34259, and 98.8% for rs4775748, hence in part the reason for the slightly different number of women analyzed for each SNP.

Archived tumor blocks were obtained from 920 (78.60%) of cases, and clinical and pathological characteristics were abstracted from medical records. Estrogen receptor (ER) status was determined by immunohistochemistry and slides were scored by a certified pathologist as previously described (Tao et al., 2009). Screening for p53 mutation status was performed on all tumor blocks as previously described (Tennis et al., 2006). Briefly, DNA was extracted from the tumor tissue on slides after micro-dissection using a pathologically confirmed HE stained slide as a guidance. Screening for P53 mutations was performed using the Affymetrix p53 GeneChip system (Affymetrix, Santa Clara, CA), and mutation status was confirmed by by-directional sequencing.

Statistical analyses

All statistical analyses were performed using SAS v.9.1. (SAS Institute, Cary, NC). Hardy-Weinberg expectations were tested using the exact χ2 goodness-of-fit tests. Unconditional logistic regression models were used to calculate odds ratios (OR) and 95% confidence intervals (CI) for associations with breast cancer risk according to SNP genotype status. Due to the nature of self-administrated dietary information, recall bias is unavoidable; cases may differentially recall exposures of dietary folate, vitamin B6 and B12 intake because of their cancer diagnosis. However, this bias would not affect our comparisons, particularly in that the participants would not be aware of their genotypes in SMUG1, UNG and DUT. In order to minimize the influence of self-administrated FFQ, we focused our analyses on stratified analyses by median dietary intakes of folate, vitamin B6 and B12 to investigate the potential gene-nutrient interactions on breast cancer. Interactions between genotype and folate, vitamins B6 and B12 intake were investigated by the evaluation of a multiplicative term in the regression model. Unconditional logistic regression was used in case-case comparisons to estimate ORs and 95% CIs according to p53 mutations status. All models were forced to adjust for age and education, and extra adjusted total energy intake when we tested the associations of genotype with breast cancer risk stratified by dietary intake of folate, vitamin B6 and B12. The case-case comparisons were additionally adjusted for ER status. Other potential confounders including demographic factors and breast cancer risk factors (body mass index, age at first birth, age at menarche, age at menopause (for postmenopausal women), parity, family history of breast cancer, history of benign breast disease, alcohol intake and smoking status were examined as potential confounders; estimates of ORs were changed by less than 10% (data not shown).


Characteristics of the study participants

Descriptive characteristics of the WEB study cases and controls have been presented elsewhere (Bonner et al., 2005; Tao et al., 2009). Briefly, Table 1 presents the main characteristics of the subjects used for analysis in this study and the allelic distribution of the investigated SNPs.

Table 1
Allelic frequencies for the SMUG1, UNG and DUT SNPs and distribution of selected breast cancer risk factors among Caucasian cases and controls, Western New York Exposures and Breast Cancer (WEB) Study

Associations between uracil processing genes variants and breast cancer risk

The SMUG1 rs7296239 SNP was not in Hardy Weinberg equilibrium (HWE) for the whole group (p<0.05) while it was in HWE for Caucasians only. Therefore we limited further analyses for all SNPs to Caucasians, who constitute more than 90% of the study population, with 1,077 cases and 1,910 controls. Odds ratios and 95% confidence intervals for breast cancer risk by genotype for the four SNPs were investigated, overall and stratified according to menopausal status, the results of which are shown in Table 2.

Table 2
SMUG1, UNG and DUT SNP genotypes and Risk of Breast Cancer among Caucasians (WEB Study)

The genotypes for UNG rs34259 and DUT rs4775748 were not associated with breast cancer risk regardless of menopausal status for all of the genetic models tested (dominant, recessive, and codominant; results not shown for the different genetic models). Among pre-menopausal women, there was no association with risk for either of the SMUG1 genotypes. Among postmenopausal women, cancer risk was increased among those with the CT genotype of the rs7296239 SNP (OR 1.29, 95% CI 1.07–1.55) but not the CC (OR 1.03, 95%CI 0.74–1.43) genotype. Similarly there was an increased risk associated with the heterozygous CT genotype only for the rs2029166 genotype (OR 1.29, 95% CI 1.07–1.56).

Interactions between folate, vitamin B6, vitamin B12 intakes and SNP genotypes, and the effect on breast cancer risk

We stratified women in categories defined by the median dietary intake in controls for folate, vitamin B6 and vitamin B12 intake, and tested the association with genotype within these strata for each vitamin. As shown in Table 3, an association of the SMUG1 rs2029166 genotype with premenopausal breast cancer risk was observed, but limited to participants with lower intake of folate (p-interaction, 0.01). For the SMUG1 rs7296239 SNP, there was a similar association limited to those with low folate although the association was not statistically significant. Among postmenopausal women, there was no interaction with folate intake. There was no evidence on interaction of genotype by intake of either vitamins B6 or B12 (data not shown).

Table 3
Risk of breast cancer associated with SMUG1, UNG and DUT SNP genotypes by folate intake, among Caucasians (WEB Study)

Case-only analysis for associations with p53 mutations, and interactions between dietary measures and genetic polymorphisms across p53 mutation status

The p53 mutation rate was 27% among cases, the majority of these mutations (94%) being found in exons 5–8 of the p53 gene. In case-only analysis (Table 4), there were no significant associations between the four genetic polymorphisms in uracil processing genes and p53 mutation status, although there were some trends towards increased risk of p53 mutations in premenopausal women with the CT/TT and the CT/CC genotypes for SMUG1 rs2029166 and rs7296239. Interestingly, in each of these two trends the heterozygotes demonstrated a more robust odds ratio compared to the homozygotes, although this mostly likely could be due to the fact that there are much less rare homozygotes than heterozygotes.

Table 4
Likelihood of p53 mutations in tumors by SMUG1, UNG and DUT SNP genotypes among Caucasians (WEB Study)


We examined the association with breast cancer risk of four single nucleotide polymorphisms in three uracil processing genes (two SNPs in SMUG1, one SNP in UNG and one SNP in DUT), which have been previously shown to be associated with altered blood uracil levels in humans (Chanson et al., 2009). Given their association with uracil levels, we hypothesized that these SNPs might be associated with p53 mutations, and that folate and related B vitamin intake would modify associations with breast cancer risk. To our knowledge this is the only study investigating uracil processing gene variants in relation to breast cancer risk.

We found associations between breast cancer risk and the SMUG1, but not the UNG or DUT, genotypes.

To date, only one study investigated sequence alterations in UNG and SMUG1 and their involvement in cancer risk by performing a mutation screening in these genes and others in colorectal cancer patients with familial aggregation (Broderick et al., 2006). They did not detect the same sequence variants that we investigated herein; however, they detected both novel and known sequence variants in these genes that were present only in the cancer patients and not in normal controls, although at very low frequency. Moreover, in silico analysis of the effect of these variants on protein structure suggested that they may have a deleterious effect. The low frequency of these variants in their population suggests that these variants may play a limited role as low penetrance genes in colorectal cancer susceptibility (Broderick et al., 2006).

In our study, carriers of the variant alleles for these SNPs had 15–30% increased risk of breast cancer, especially among postmenopausal women. Interestingly, these associations were limited to the heterozygous genotype carriers for both SNPs, even though we had previously found associations of increased blood uracil with the TT genotype for the rs2029166 SNP and with the CC genotype for the rs7296239 SNP (Chanson et al., 2009). Our results suggest a possible molecular heterosis effect, however we cannot exclude the possibility that the associations with the homozygote variant genotypes were not statistically significant due to less frequent subjects carrying this genotype in our population. Molecular heterosis takes place when heterozygous individuals for a specific locus show a significant increased or decreased effect for a quantitative or dichotomous trait compared to homozygous subjects, and is a rather controversial concept in population genetics, although well documented in plant, animal and human studies (Comings & MacMurray, 2000; Crow, 2008). Several well documented examples of molecular heterosis in human gene association studies are presented in a review on the subject by Comings and MacMurray 2000 (Comings & MacMurray, 2000). More recently, a meta-analysis of 19 case-control and 3 genome wide association studies of type 2 diabetes revealed that molecular heterosis still remains a plausible scenario explaining the effect of certain SNPs associated with disease (Salanti et al., 2009). We therefore contend that the present observations do not justify a definitive conclusion for a null effect of these SNPs.

We found that there was an interaction of genotype and folate intake for premenopausal breast cancer for the SMUG1 rs2029166 SNP. The association with genotype was limited to participants with low folate intake. For the SMUG1 rs7296239 SNP, results were similar but the interaction was not statistically significant. There was no evidence of interaction for postmenopausal breast cancer for these two genotypes. In addition, there was no interaction for the other SNPs examined or for vitamin B6 and B12. In our previous study of these SNPs in a Puerto Rican population including both men and women (Chanson et al., 2009), we did not see an effect of plasma folate and B vitamin levels on the association with altered uracil levels. The association of the SMUG1 rs2029166 genotype with breast cancer risk among premenopausal women with low intake of folate implies that a higher intake of these nutrients may be necessary in order to maintain genomic stability of human cells, as previously suggested (Fenech, 2001).

The fact that all positive associations were limited to those with the heterozygous genotype possibly implies the presence of molecular heterosis in our study. However, caution should be used when interpreting these results due to the low magnitude of the associations. We had previously found associations with plasma uracil that were limited to homozygous genotypes for both SNPs. It may be that multiple factors are involved in the complex association between genes involved in uracil and one carbon metabolism and breast cancer risk, also explaining the inconsistency in epidemiology studies on the matter (Xu & Chen, 2009).

In this study, there was a suggestion of an increased number of p53 mutations among premenopausal breast cancers from women with the heterozygous genotype for both SMUG1 SNPs. These findings, while not statistically significant, are consistent with uracil misincorporation contributing to mutagenesis and with our previous observation that plasma uracil concentrations are correlated with strand breaks in the coding region of the human p53 gene (Chanson et al., 2009).

The strengths of this study include the population-based study design and the relatively large number of subjects. The detailed exposure information enabled an investigation of gene-environment interaction. However, the statistical power in subgroups of our study remained limited due to low frequencies of the variant alleles, which limited our ability to identify weak associations. We studied four polymorphisms that had been shown in another population to be associated with blood uracil concentrations but with unknown functional effects on the enzymatic activity of the encoded proteins, so we cannot rule out the possibility that other unidentified SNPs tightly linked to the ones described herein might instead be the functional SNPs that are associated with breast cancer risk.

In conclusion, our study suggests that the four selected SNPs are not robust determinants of breast cancer risk, but that the two SNPs in SMUG1 might modestly alter the risk of breast cancer, especially in postmenopausal women and in premenopausal women with lower intakes of folate and that these variants may contribute to p53 mutations. Future studies should be directed towards confirming these associations and probing for the presence of molecular heterosis.


Conflict of interest

The authors declare no conflict of interest.


  • Ames BN. Micronutrient deficiencies. A major cause of DNA damage. Ann N Y Acad Sci. 1999;889:87–106. [PubMed]
  • Ames BN, Wakimoto P. Are vitamin and mineral deficiencies a major cancer risk? Nat Rev Cancer. 2002;2:694–704. [PubMed]
  • Barnes DE, Lindahl T. Repair and genetic consequences of endogenous DNA base damage in mammalian cells. Annu Rev Genet. 2004;38:445–476. [PubMed]
  • Block G, Hartman AM, Dresser CM, Carroll MD, Gannon J, Gardner L. A data-based approach to diet questionnaire design and testing. Am J Epidemiol. 1986;124:453–469. [PubMed]
  • Blount BC, Mack MM, Wehr CM, MacGregor JT, Hiatt RA, Wang G, Wickramasinghe SN, Everson RB, Ames BN. 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;94:3290–3295. [PubMed]
  • Bonner MR, Han D, Nie J, Rogerson P, Vena JE, Muti P, Trevisan M, Edge SB, Freudenheim JL. Breast cancer risk and exposure in early life to polycyclic aromatic hydrocarbons using total suspended particulates as a proxy measure. Cancer Epidemiol Biomarkers Prev. 2005;14:53–60. [PubMed]
  • Broderick P, Bagratuni T, Vijayakrishnan J, Lubbe S, Chandler I, Houlston RS. Evaluation of NTHL1, NEIL1, NEIL2, MPG, TDG, UNG and SMUG1 genes in familial colorectal cancer predisposition. BMC Cancer. 2006;6:243. [PMC free article] [PubMed]
  • Chanson A, Parnell LD, Ciappio ED, Liu Z, Crott JW, Tucker KL, Mason JB. Polymorphisms in uracil-processing genes, but not one-carbon nutrients, are associated with altered DNA uracil concentrations in an urban Puerto Rican population. Am J Clin Nutr. 2009;89:1927–1936. [PubMed]
  • Choi SW, Mason JB. Folate and carcinogenesis: an integrated scheme. J Nutr. 2000;130:129–132. [PubMed]
  • Comings DE, MacMurray JP. Molecular heterosis: a review. Mol Genet Metab. 2000;71:19–31. [PubMed]
  • Crow JF. Mid-century controversies in population genetics. Annu Rev Genet. 2008;42:1–16. [PubMed]
  • Fenech M. The role of folic acid and Vitamin B12 in genomic stability of human cells. Mutat Res. 2001;475:57–67. [PubMed]
  • Giovannucci E. Epidemiologic studies of folate and colorectal neoplasia: a review. J Nutr. 2002;132:2350S–2355S. [PubMed]
  • Hao B, Wang H, Zhou K, et al. Identification of genetic variants in base excision repair pathway and their associations with risk of esophageal squamous cell carcinoma. Cancer Res. 2004;64:4378–4384. [PubMed]
  • Hung RJ, Hall J, Brennan P, Boffetta P. Genetic polymorphisms in the base excision repair pathway and cancer risk: a HuGE review. Am J Epidemiol. 2005;162:925–942. [PubMed]
  • Jang H, Mason JB, Choi SW. Genetic and epigenetic interactions between folate and aging in carcinogenesis. J Nutr. 2005;135:2967S–2971S. [PubMed]
  • Kapiszewska M, Kalemba M, Wojciech U, Milewicz T. Uracil misincorporation into DNA of leukocytes of young women with positive folate balance depends on plasma vitamin B12 concentrations and methylenetetrahydrofolate reductase polymorphisms. A pilot study. J Nutr Biochem. 2005;16:467–478. [PubMed]
  • Kavli B, Otterlei M, Slupphaug G, Krokan HE. Uracil in DNA--general mutagen, but normal intermediate in acquired immunity. DNA Repair (Amst) 2007;6:505–516. [PubMed]
  • Krokan HE, Drablos F, Slupphaug G. Uracil in DNA--occurrence, consequences and repair. Oncogene. 2002;21:8935–8948. [PubMed]
  • Mashiyama ST, Hansen CM, Roitman E, Sarmiento S, Leklem JE, Shultz TD, Ames BN. An assay for uracil in human DNA at baseline: effect of marginal vitamin B6 deficiency. Anal Biochem. 2008;372:21–31. [PMC free article] [PubMed]
  • Miao R, Gu H, Liu H, et al. Tagging single nucleotide polymorphisms in MBD4 are associated with risk of lung cancer in a Chinese population. Lung Cancer. 2008;62:281–286. [PubMed]
  • Mohrenweiser HW, Jones IM. Variation in DNA repair is a factor in cancer susceptibility: a paradigm for the promises and perils of individual and population risk estimation? Mutat Res. 1998;400:15–24. [PubMed]
  • Salanti G, Southam L, Altshuler D, et al. Underlying genetic models of inheritance in established type 2 diabetes associations. Am J Epidemiol. 2009;170:537–545. [PMC free article] [PubMed]
  • Shin MC, Lee SJ, Choi JE, Cha SI, Kim CH, Lee WK, Kam S, Kang YM, Jung TH, Park JY. Glu346Lys polymorphism in the methyl-CpG binding domain 4 gene and the risk of primary lung cancer. Jpn J Clin Oncol. 2006;36:483–488. [PubMed]
  • Sousa MM, Krokan HE, Slupphaug G. DNA-uracil and human pathology. Mol Aspects Med. 2007;28:276–306. [PubMed]
  • Tao MH, Shields PG, Nie J, et al. DNA hypermethylation and clinicopathological features in breast cancer: the Western New York Exposures and Breast Cancer (WEB) Study. Breast Cancer Res Treat. 2009;114:559–568. [PubMed]
  • Tennis M, Krishnan S, Bonner M, et al. p53 Mutation analysis in breast tumors by a DNA microarray method. Cancer Epidemiol Biomarkers Prev. 2006;15:80–85. [PubMed]
  • Visnes T, Doseth B, Pettersen HS, Hagen L, Sousa MM, Akbari M, Otterlei M, Kavli B, Slupphaug G, Krokan HE. Uracil in DNA and its processing by different DNA glycosylases. Philos Trans R Soc Lond B Biol Sci. 2009;364:563–568. [PMC free article] [PubMed]
  • Willett WC, Howe GR, Kushi LH. Adjustment for total energy intake in epidemiologic studies. Am J Clin Nutr. 1997;65:1220S–1228S. [PubMed]
  • Xu X, Chen J. One-carbon metabolism and breast cancer: an epidemiological perspective. J Genet Genomics. 2009;36:203–214. [PMC free article] [PubMed]
  • Zhang SM, Willett WC, Selhub J, Hunter DJ, Giovannucci EL, Holmes MD, Colditz GA, Hankinson SE. Plasma folate, vitamin B6, vitamin B12, homocysteine, and risk of breast cancer. J Natl Cancer Inst. 2003;95:373–380. [PubMed]