Single-nucleotide polymorphisms in one-carbon metabolism genes, Mediterranean diet and the risk of pre- and postmenopausal breast cancer

Background: One-carbon metabolism pathway could interfere with the carcinogenesis of breast cancer (BC). Dietary pattern plays an important role in one-carbon metabolism linking the combination of dietary micronutrients. Methods: We assessed associations between single-nucleotide polymorphisms (SNPs) of one-carbon metabolism, adherence to Mediterranean dietary pattern (MDP) and BC risk from a case-control study (818 cases, 935 controls) among Chinese female population. The genotyping of 10 SNPs in seven one-carbon metabolism-related genes (MTHFD1, MTRR, MAT2B, CDO1, FOLR1, UNG2, ADA) were performed. Dietary intake measurements were assessed by a validated food-frequency questionnaire. Gene-diet interactions were analyzed. Results: No evidence demonstrated SNPs of one-carbon metabolism or their haplotypes were associated with BC risk. High adherence to the Mediterranean dietary pattern decreased the risk of breast cancer among post- but not premenopausal women and the association was inuenced by the SNPs genotype, with the increasing number of variant allele in MTHFD1 (rs11627387), MTHFD1 (rs2281603), MTRR (rs16879334), MTRR (rs2287780), MAT2B (rs4869087), FOLR1 (rs10501409), UNG2 (rs231622) and ADA (rs244072). The protective effect against BC risk from high adherence to the MDP was gradually weakened and disappeared. For MTHFD1 (rs8003567) and CDO1 (rs34869) genotypes, women with homozygous were less affected by adherence to the MDP than to women who with heterozygotes. No signicant gene-diet interactions were observed. Conclusions: SNPs of one-carbon metabolism genes modify the effect of high adherence to MDP against BC risk in Chinese women, as potential effect modiers. Genetic variants may inuence the association between diet and BC risk.


Introduction
The primary risk factors of breast cancer (BC) include overweight or obesity, physical inactivity, exogenous hormone intake (use of oral contraceptive and hormone replacement therapy), reproduction condition (late age at rst birth and low parity), family history of cancer and unhealthy diet (Bethea, et al. 2015, Brewer, et al. 2017, Deschasaux, et al. 2017, Ellingjord-Dale, et al. 2017, Neuhouser, et al. 2015, Trinh, et al. 2015. Besides these modi able risk factors, genetic predisposition also plays an important role in susceptibility to BC. Inherited mutations in BRCA1 and BRCA2 genes confer a high risk of BC to women of speci c families (Torres, et al. 2017), whereas, for a large percentage of cases, BC is more likely attributed to a combination of genetic variations and environmental or lifestyle factors. Elucidating the gene-environment interactions is therefore essential to unveil the molecular mechanisms involved in breast carcinogenesis.
Diet is one of the environmental factors and often related to individuals' health outcomes. A wellbalanced dietary pattern, such as the Mediterranean diet, is able to reduce the risks of a variety of diseases. The traditional Mediterranean diet is characterized by the liberal use of olive oil; the high consumption of vegetables, fruits, nuts, legumes and unprocessed cereals; the moderate amounts of sh and wine; the low intake of dairy products (with the exception of cheeses), red meat and meat products (Trichopoulou, et al. 2014). A number of epidemiological studies have suggested that this dietary pattern had substantial health bene ts and displayed inverse associations with multiple disease types, including cancer (Soltani, et al. 2019). Our study also showed that high adherence to the Mediterranean dietary pattern signi cantly reduced post-menopausal BC risk in Chinese women. Since cancer development is generally determined by the interplay between extrinsic and intrinsic factors, the roles of underlying molecular pathways in the association of the Mediterranean diet with BC risk require further investigation.
One-carbon metabolism is a complicated metabolic network composed of cascade reactions based on the transfer of one-carbon units, which is required by nucleotide synthesis, DNA replication, repair and methylation. Folate, a water-soluble vitamin-B found rich in leafy green vegetables and fruits, beans and peas, grains and cereals, is an essential component of this metabolism pathway as the functional donor of one-carbon units. The inadequate intake of nutrients in the daily diet could destroy the normal physiological process of one-carbon metabolism, associated with the disruption of DNA replication and repair as well as aberrant DNA methylation patterns, eventually leading to carcinogenesis (Miranti, et al. 2017). Some epidemiological studies provide some promising results, which focused on MTHFR (rs1801131, rs1801133) and MTR (rs1805087), intake of folate and vitamin B6/B12 and BC risk (Ma, et al. 2009, Platek, et al. 2009). Also, genetic polymorphisms in the one-carbon metabolism pathway genes encoding functional enzymes and coenzymes have been suggested to in uence individuals' susceptibility to cancer, such as Methylenetetrahydrofolate reductase (MTHFR) and methionine synthase (MTR) (Shao, et al. 2018, Wu, et al. 2016). However, in addition to these most common polymorphisms, the role of genetic variants in other one-carbon metabolism genes and their interactions with diet regarding BC risk remain largely unexplored, especially in the Asian female population.
In this study, we evaluated associations between genetic polymorphisms in one-carbon metabolismrelated genes and the risk of BC among Chinese women. We further investigated the interaction effects of selected genetic variants with high adherence to the Mediterranean dietary pattern on BC carcinogenesis.

Subjects
Subjects were from the Chinese Wuxi Exposure and Breast Cancer Study (2013-2014), a population-based case-control study of the role of biology, diet, lifestyle, and environmental factors in the etiology of BC in Asian women. The subjects were all women who lived in Wuxi city, Jiangsu Province, China, for more than ve years. Newly diagnosed breast cancer patients within one year were selected as the case group according to the local cancer registration system. All cases were identi ed according to the International Classi cation of Diseases for Oncology (ICD-10, code C50) and were con rmed by pathology, excluding patients with secondary or recurrent BC. For those with multiple incident cancers, only included those with BC as the rst original malignancy diagnosed. Controls were matched to the cases in a ratio of 1:1 by the same residence area and age (range of ±5 years), excluding individuals with any cancer history. The study protocol was approved by the Institutional Review Boards of Jiangsu CDC, and the informed consent were obtained from all subjects. Blood samples were collected from both cases and controls.

Data on diet
The usual diet was assessed by a validated, semi-quantitative food frequency questionnaire (FFQ), which included 149 items along with the recipes commonly used in China (Zhao, et al. 2002). Nutrient and energy intake were calculated through the Chinese Food Composition Database. Dietary intake assessment included whether the food was consumed, consumption frequency (times of per day/week/month/year) and the average amount of food consumption at each time. The 149 food items in the FFQ were classi ed into 18 prede ned food groups based on similarities in nutrient pro le and culinary usage.

Mediterranean diet scale
The Alternate Mediterranean Diet Score (aMED) established by Fung et al., (Fung, et al. 2005) includes nine dietary components and range from 0-9 scores (minimum to maximum conformity). One point is given to each subject when the food intake is equal to or above the median intake of controls for the following seven components regarded as healthy: grains, fruits, vegetables, legumes, nuts, sh, and monounsaturated fat-saturated fat ratio. One point is given when the intake amount of a subject is less than the median intake of unhealthy food such as red meat or processed meat, or alcohol consumption within 5-25 g/day for women as a speci ed range. The higher the score obtained from the questionnaire, the greater the adherence to the Mediterranean dietary pattern (MDP).
Lifestyle, anthropometric, medical history and reproductive history data Demographic, lifestyle characteristics, menstrual and reproductive events, dietary intake, disease history and physical activity-related data were collected from a structured questionnaire, through in-person interviews conducted by trained interviewers. Anthropometric measures were obtained by trained personnel following the protocol. Physical activity was measured by referencing the Global Physical Activity Questionnaire. Postmenopause was de ned as an absence of menstruation in the past 12 months.
Experimental design and SNP genotyping were carried out by CapitalBio Corporation (Beijing, China) using the Sequenom MassARRAY platform following the manufacturer's instructions. The call rate for each SNP among all the samples was >95%. In addition, ve percent of samples were randomly selected and then repeatedly genotyped with a concordant rate of 100%.

Statistical analysis
Chi-square tests were carried out to examine Hardy-Weinberg equilibrium (HWE) in the control group.
Linkage disequilibrium between SNPs was calculated as D' and R 2 values. The SNPs with strong linkage disequilibrium will be constructed as haplotypes for further analysis.
Unconditional logistic regression methods were used to assess: (1) associations between SNPs polymorphisms and BC risk (statistical signi cance should be after Benjamini & Yekutieli correction for multiple testing, p<0.05). (2) associations between adherence to the MDP and BC risk. (3) associations between adherence to the MDP and BC risk strati ed by SNPs genotype.
The potential confounders included are age at diagnosis for cases or enrollment for controls (by years), area (urban, rural), education (ordered as illiterate and primary, middle and high school, university and above), tobacco smoking (no, or yes: including smoking and second-hand smoking 3day/week), moderate physical activity (minutes/day), oral contraceptives use (no, or yes: current use or ever use), hormone replacement therapy (no, or yes: current use or ever use), family history of breast cancer (no, or yes: in a rst-degree relative), history of benign breast disease (no, or yes: including lactation mastitis, plasma cell mastitis, cyclomastopathy, broadenoma of breast, galactocele), age at menarche (by years), parity (0, 1, 2, 3), age at rst full-term delivery (by years), breastfeeding (no, or yes), body mass index (BMI; in kg/m 2 ).
A priori causal model of the causal relationship between the one-carbon metabolism pathway, Mediterranean dietary pattern (MDP) and breast cancer was constructed to determine which confounders need to be adjusted, presenting as a directed acyclic graph (DAG) (Greenland, et al. 1999) (Supplemental Figure 1). Characteristics and frequencies of one-carbon metabolism genes Genotype and allele frequencies of the 15 SNPs in eleven one-carbon metabolism pathway-related genes (MTHFD1, MTRR, MAT1A, MAT2B, CDO1, FOLR1, CBS, GLS, DNMT3B, UNG2, ADA) among cases and controls are shown in Table 1. Because there is no genetic variation (minor allele frequency less than 0.05 was found in the MTHFD1 T>C (rs2230491), MAT1A T>C (rs10887718), CBS T>C (rs11701048), GLS T>C (rs12185688) and DNMT3B G>A (rs13045669), they will not be introduced in the following analyses for the association of BC risk and the interaction of diet. The genotype frequencies of the SNPs included in the control group did not deviate from Hardy-Weinberg equilibrium (HWE).
Associations between single-nucleotide polymorphisms (SNPs) and breast cancer risk  Table  2.
For linkage disequilibrium analysis, two considerable degree of linkage disequilibrium were observed between the MTRR T>C (rs227780) and MTRR G>C (rs16879334) (D'=0.99, r 2 =0.99) and MTHFD1 G>A (rs8003567) and MTHFD1 G>A (rs11627387) (D'=0.96, r 2 =0.35), as shown in Figure 1. Haplotypes with a frequency greater than 0.03 were constructed in the case group and control group, but no signi cant difference was found in the distribution of haplotypes between cases and the controls, as shown in Supplemental Table 3.
Associations between adherence to the Mediterranean dietary pattern and breast cancer risk, strati ed by the SNPs genotypes A higher aMED score indicated a lower postmenopausal BC risk with a p-trend value<0.0000, the OR value is 0.54 (95% CI: 0.38-0.78) for comparing top with bottom quartiles, and OR value for per 1 point increase of the aMED is 0.80 (95% CI: 0.70-0.90). However, no signi cant association was found among premenopausal women (Table 2).
Among postmenopausal women, we found that the association between MDP and BC risk was affected by the SNPs genotype. Strati ed by the genotypes of a dominant model (e.g., AA+AB versus BB, A is the risk allele) and an additive model (e.g., AA versus AB versus BB, A is the risk allele), with the increasing number of variant allele, the protective effect against BC risk from high adherence to the MDP was gradually weakened and disappeared. While for MTHFD1 (rs8003567) and CDO1 (rs34869) genotypes, women with homozygous were less affected by high adherence to the MDP than to women who with heterozygotes. The results were shown in Table 3 and Figure 2, and the alternate Mediterranean Dietary Pattern score calculated as a continuous variable and a quartile variable, separately.
Interaction analyses between high adherence to the Mediterranean dietary pattern, One-carbon metabolism genes and breast cancer risk In the interaction analyses, the wild-type genotype of the SNPs at the lowest quartile values of aMED was used as the reference group. We found a nominal statistical signi cance of the relevant interaction between MTHFD1 G>A (rs8003567) polymorphisms with postmenopausal breast cancer risk based on the additive and dominant genotypic effects (P-interaction=0.0260 for additive genotypic effects and P-interaction=0.0465 for dominant genotypic effects, based on multiplication model), Table 3. However, this interaction is no longer signi cant under the Benjamini & Yekutieli correction for multiple comparisons.
The detail was shown in Supplemental Table 4 and Table 5.

Discussion
To our knowledge, this is the rst study performed in China that investigated the associations between genetic variants in one-carbon metabolism genes and BC risk as well as their interactions with MDP on BC risk. Among 10 SNPs involved in the present study, none of them were statistically signi cantly associated with the risk of BC despite the menopausal status. However, some meaningful results were found when analyzing the associations between high adherence to MDP and BC risk based on the strati cation of genotypes of the SNPs studied. The association between high adherence to MDP and risk reduction of BC was affected by the SNPs genotype associated with the one-carbon metabolism pathway. Our study suggests that the one carbon-metabolism genes may act as modi ers between diet and BC risk. The nutrients associated with one-carbon metabolism could not naturally be manufactured by the human body, which means they need to be obtained from foods. The disruption on the one-carbon metabolism pathway could interfere with DNA-repair, DNA-replication, and gene expression regulation, which could be highly carcinogenetic (Kim 2004, Locasale 2013). That has motivated lots of studies focusing on the potential link between the nutrients associated with one-carbon metabolism pathway and carcinogenesis (Milne, et al. 2017, Weinstein, et al. 2006. A previous study showed that SNPs of onecarbon metabolism gene have interactions with folate intake to affect the BC risk (Kakkoura, et al. 2017). However, most previous studies focus on the effect of individual nutrients rather than a combination of foods (Lissowska, et al. 2007), which have been limited in interpreting the high degree of intercorrelation among various nutrients. These associations were always weak because it is hard to attribute effects to single independent component foods (Kim 2004). In this context, we choose a 'posterior' dietary pattern instead of individual nutrients, which could better capture speci c diet characteristics and cumulative effects of nutrients.
The Mediterranean dietary pattern is characterized by high consumption of vegetables, fruit, legumes, and sh, rich in folate, choline, vitamins and methionine. Su cient levels of micronutrients play an important role in the one-carbon metabolism (Park, et al. 2012, Woodside, et al. 2005, because speci c enzymes and co-enzymes in one-carbon metabolism require ample quantities of dietary micronutrients (e.g., folate, methionine and other speci c amino acids and B 2 , B 6 and B 12 and other vitamins), as substrates to achieve their biological functions (Lucock 2000, Stevens, et al. 2007).
One-carbon metabolism is interconnected to the biological processes of DNA methylation and DNA synthesis (Xu and Chen 2009). Both processes are thought to play key roles in carcinogenesis (Lewis, et al. 2006, Maruti, et al. 2009). DNA methylation is an epigenetic mechanism by which cells regulate gene expression, which involves the addition of a methyl (-CH3) to the 5-carbocytosine residue, predominantly in the cytosine guanine dinucleotide (CPGs). Dietary micronutrients involved in one-carbon metabolism play an essential role in DNA methylation, such as folic acid, choline, betaine, ribo avin, vitamins B 6 and B 12 , and the amino acid methionine (Figure 3) (Crider, et al. 2012, Locasale 2013. Especially folic acid (or called vitamin B 9 ), whose role is crucial in the DNA methylation process, producing the methyl group donor, S-adenosylmethionine.
A nominally signi cant interaction result of MTHFD1 G>A (rs8003567) also implies that one-carbon metabolism genes may be related to diet and BC risk by affecting DNA methylation. The MTHFD1 gene product is a multifunctional enzyme possessing the activities of methylene-THF dehydrogenase, methenyl-THF cyclohydrolase and formyl-THF synthetase in one-carbon metabolism pathway (MacFarlane, et al. 2011). It usually catalyzes sequential and reversible reactions in multiple conversion of tetrahydrofolate (THF), the active form of folate, into 5,10 methylene-THF, which is essential for the de novo purine and thymidylate synthesis as well as the supply of one-carbon units for subsequent DNA methylation. The de ciency or dysregulation of the MTHFD1 enzyme may in uence cell division and global methylation pattern, eventually contributing to tumorigenesis (Ding, et al. 2018, MacFarlane, et al. 2011). Since rs8003567 located in the intronic region of the MTHFD1 gene and no disease-related studies on SNPs have been reported before, another possible explanation cannot be excluded that there are additional functional genetic variants in linkage disequilibrium with these two SNPs that modify BC risk in Chinese female population. However, the interpretation of a nominally signi cant interaction should be cautious, because the corrected P-values for multiple comparisons is no longer signi cant, the gene-diet interaction obtained may be a false positive result. Thus, replication of the ndings in other independent studies is needed before the rm conclusions can be drawn.
However, the relationship between nutrients and DNA methylation is complicated, and there is no uni ed conclusion now. A recent study in rural African women support that one-carbon nutrient may affect methylation levels, dietary intake of one-carbon metabolites and cofactors in diet uctuates with seasons. The concentration of biomarkers of maternal carbon metabolism nutrients during pregnancy was associated with the methylation of metastable epi-alleles in DNA from birth infant's lymphocytes and hair  (Turati, et al. 2018). The dietary pattern is a combination of food groups rather than isolated nutrients. A 'posterior' Mediterranean dietary pattern contains adequate levels of micronutrients associated with one-carbon metabolisms, such as Vitamins B 2 , B 6 , B 12 , folate and choline, which are all involved in DNA methylation and synthesis (Mas, et al. 2007).
Thus, it may be more predictive for vivo situation (Brennan, et al. 2010, Gerber 2003) and interpretative for disease risk and biological mechanism.
Several limitations should also be taken into account in our study. First, we did not obtain adequate information on various subtypes of breast cancer, such as hormone receptor status. The BC risk based on the strati cation analysis of breast cancer subtypes is not able to be further evaluated. Second, although we examined several SNPs in multiple key genes thought to be important in the one-carbon metabolism pathway, some other potential polymorphic sites may not be involved in the present study. Third, data were collected from a case-control study, which might be partially in uenced by the biases inherent in case-control designs, we analyzed the effect of selection bias in Supplemental Figure 3. Finally, since the number of cases and controls enrolled in this project is relatively small, the associations we founded requires replication in other larger sample independent studies. Further work should assess associations of BC risk and the concentrations of these nutrients in plasma associated with one-carbon metabolism and DNA methylation.

Conclusion
In conclusion, our results support that the SNPs in one-carbon metabolism genes modify the effect of high adherence to Mediterranean dietary pattern against BC risk in Chinese women. Genetic variants may as potential effect modi ers in uence the association between diet and BC risk.

Declarations Ethical approval
This study was approved by the ethical committee of the Jiangsu Center for Disease Control and Prevention (Jiangsu, China).

Informed consent
Informed consent was obtained from all individual participants included in the study.

Consent for publication
Not applicable

Availability of data and materials
The datasets generated and analyzed during the current study are not publicly available due to the racial genetic characteristics but are available from the corresponding author on reasonable request.

Competing interests
The authors declare that they have no competing interests All authors contributed to the preparation of the manuscript. PMW and MW (guarantors) had the original idea for the study and carried out the study design, formed the investigator group that obtained the funding, as well as oversaw the study implementation and data collection. JYZ, WL, YQD and YQ assisted data collection and interpreted the data. SC and ZZ carried out the analysis reported in this paper, under the supervision of PMW and MW, and also prepared the initial draft of the manuscript. All authors read and approved the nal manuscript.

Figure 2
Associations between high adherence to Mediterranean dietary pattern (MDP) and breast cancer risk, strati ed by SNPs genotypes.