No association between cSHMT genotypes and risk of breast cancer in the Nurses’ Health Study

Abstract

Objective

Increased breast cancer risk has been observed with both low folate status and a functional polymorphism in methylenetetrahydrofolate reductase (MTHFR 677C→T). Cytoplasmic serine hydroxymethyltransferase (cSHMT) affects the flow of one-carbon units through the folate metabolic network, but there is little research on a role for genetic variation in cSHMT in determining breast cancer risk.

Methods

A nested case-control study within the Nurses’ Health Study was used to investigate an association between cSHMT (1420C→T) and breast cancer risk.

Results

No evidence for an association of cSHMT genotype and breast cancer was 10 observed. There was also no evidence of a gene-gene interaction between cSHMT and MTHFR.

Conclusions

There was no evidence of an association between cSHMT genotype and breast cancer occurrence. Further research in populations with differing average folate intake may be needed to fully understand the interactions of folate nutrition, sequence variation in folate genes, and breast cancer risk.

Introduction

Cytoplasmic serine hydroxymethyltransferase (cSHMT) catalyzes the reversible conversion of serine and tetrahydrofolate to glycine and 5,10-methylenetetrahydrofolate, and mediates the competition for one-carbon units between methylenetetrahydrofolate reductase (MTHFR) and thymidylate synthase, preferentially partitioning 5,10-methylenetetrahydrofolate to synthesis of thymidine (Herbig et al., 2002). Given the importance of sufficient thymidine to prevent uracil misincorporation into DNA, a common nonsynonymous SNP in the cSHMT gene (1420C→T) has been investigated in relation to cancer risk. Associations of this SNP with decreased risk of leukemia and malignant lymphoma have been observed, while the only prior report of this SNP and breast cancer found no association (Lissowska et al., 2007). We investigated the relation of cSHMT genotype with incident breast cancer in the Nurses’ Health Study (NHS). We also considered a functional polymorphism in MTHFR (677C→T) for an interactive effect because MTHFR catalyzes a linked step in one-carbon metabolism, and a gene-gene interaction was observed between these genes in relation to cardiovascular disease risk (Lim et al., 2005).

Subjects and Methods

The study population consisted of 1,007 breast cancer cases diagnosed by June 1, 1998 and 1,441 matched controls. The case-control study, described elsewhere (Han et al., 2004), was nested within the sub-cohort of 32,826 women of the NHS who provided blood samples in 1989 and 1990, prior to FDA-mandated folate fortification in the United States.

Genotypes were determined by TaqMan (ABI Prism 7900 Sequence Detection System, Applied Biosystems, Foster City, CA) (further details: cSHMT (Skibola et al., 2002); MTHFR (Lim et al., 2005)).

Conditional logistic regression models were used to quantify the genotype—outcome association. Product terms were used to test for interactions between genotypes, folate quintile, and alcohol. Plasma means for homocysteine, folate, and vitamins B6 and B12 were compared within genotype combinations to assess genotype—biomarker associations. Statistical analyses were conducted using SAS 8.2 (SAS Institute, Cary, NC).

Results

Genotypes for both genes were available for 939 cases and 1226 matched controls (for descriptive data: (Han et al., 2004)). The concordance for quality control samples for genotyping assays was 100%. Both genes were in Hardy-Weinberg equilibrium. The allele frequency for cSHMT was similar to previous reports, while the allele frequency for MTHFR (40.9% among controls) was slightly higher than in other publications. Although there was no difference in the allele frequencies between cases and controls for either gene, the distribution of the MTHFR genotypes differed between the two groups (p < 0.04; Table 1).

Table 1.

Distribution of Genotypes for cSHMT 1420C→T and MTHFR 677C→T in Breast Cancer Cases and Controls in the Nurses Health Study.

	MTHFR CC		MTHFR CT		MTHFR TT		Total
	cases	controls	cases	controls	cases	controls	cases	controls
cSHMT CC	171	220	203	296	89	103	463* (49.3)	619 (50.5)
cSHMT CT	138	172	161	234	87	83	386 (41.1)	489 (39.9)
cSHMT TT	37	37	38	62	15	19	90 (9.6)	118 (9.6)
Total	346* (36.9)	429 (35.0)	402 (42.8)	592 (48.3)	191 (20.3)	205 (16.7)	939	1226

^*number (perce)

There was little or no association of the cSHMT polymorphism with breast cancer risk (Table 2). The addition of MTHFR genotype and dietary folate variables to the model did not appreciably change cSHMT effect estimates in either crude models or after adjustment for known risk factors. Little or no change in results was observed when cases were subdivided by estrogen/progesterone receptor status or tumor size. Regression coefficients for MTHFR genotype and folate status were virtually identical when cSHMT was excluded from the final model (data not shown). There was little or no evidence of a gene-gene interaction. Models including gene-gene interaction terms were not statistically different from models without these terms, and the addition of these terms had no effect on model coefficients. Similar results were seen with the other interaction terms tested (MTHFR-folate, cSHMT-folate, folate-alcohol), thus the additional consideration of these models did not change the findings.

Table 2.

The Relation of cSHMT 1420C→T and Breast Cancer Risk in the Nurses’ Health Study.

	Odds Ratio (95% Confidence Interval)*	Odds Ratio (95% Confidence Interval)*
	Unadjusted Models	Adjusted Models†
1. cSHMT alone
cSHMT (referent genotype: CC)
CT	1.06 (0.88 – 1.28)	1.13 (0.88 – 1.45)
TT	0.98 (0.72 – 1.34)	1.13 (0.75 – 1.72)
P for trend:	0.81	0.36
2. cSHMT adjusted for MTHFR
cSHMT (referent: CC)
CT	1.05 (0.87 – 1.27)	1.09 (0.84 – 1.40)
TT	0.98 (0.72 – 1.33)	1.14 (0.75 – 1.75)
P for trend:	0.88	0.43
3. cSHMT adjusted for MTHFR, & Plasma Folate Quintiles
cSHMT (referent: CC)
CT	1.04 (0.82 – 1.32)	1.05 (0.77 – 1.43)
TT	0.94 (0.63 – 1.39)	0.91 (0.54 – 1.52)
P for trend:	0.92	0.91

^*Odds ratio and 95% confidence intervals are calculated from conditional logistic regression model coefficients and standard errors

^†Adjusted multivariate conditional logistic regression model accounted for matching factors and adjusted for height, BMI, calorie intake, alcohol intake (≥15 g/day or <15 g/day), age at menarche, age at first birth, parity, age at menopause, post-menopausal hormone use (current user or not current user), family history of breast cancer in a first degree relative (yes or no), and history of benign breast disease (yes or no).

Among controls, there were no statistically significant associations of cSHMT genotype, either alone or in interaction with the MTHFR genotype, with plasma levels of homocysteine, folate, or vitamins B6 and B12.

Discussion

One-carbon metabolism is hypothesized to play a role in breast cancer risk (Zhang et al., 2003), but only one prior study (Lissowska et al., 2007) considered a common polymorphism (1420C→T) in cSHMT in relation to breast cancer. In agreement with the prior study, cSHMT genotype was not associated with the risk of breast cancer, whether considered alone or as a modifier of the effect of the MTHFR genotype or folate nutrition.

Strengths of this study include the large number of incident cases and matched controls that were evaluated in a nested case-control design, an optimal design for the analysis of gene-environment interactions. The collection of plasma to measure nutrient cofactors before diagnosis limited the possibility of alteration of nutrient status by disease. Although only 27% of the full cohort gave blood and were therefore eligible for the study, selection bias is unlikely given that the distribution of known risk factors for breast cancer were similar between women who did and did not provide blood samples. Population stratification may have confounded an association between cSHMT 1420C→T and breast cancer, but this bias seems improbable as approximately 93% of the population were Caucasian. The perfect concordance of genotyping quality control samples rules out measurement error as a cause of bias.

The background level of folate nutrition in this population may be important in evaluating the results. Mean dietary folate intake (without supplements) in the control group was 283 μg/day, and mean intake including supplements was 409 αg/day. Although the average total folate intake in controls is not high, it may be that a comparatively lower level of folate nutrition is necessary to observe an association of cSHMT genotype with breast cancer risk. Prior findings in the same cohort show a protective association of folate with breast cancer risk, which is limited to women with a high intake of alcohol (a folate antagonist) (Zhang et al., 2003). The effect of MTHFR genotype on cardiovascular risk has been shown to differ across populations by level of folate (Lewis et al., 2005) and reported associations of genotype and breast cancer are often strongest in women with the lowest levels of dietary folate (Chen et al., 2005; Shrubsole et al., 2004). As folate status modifies the relation between MTHFR and disease risk, a gene-gene interaction between MTHFR and cSHMT on disease may also depend on folate nutrition. Importantly, there was no association between MTHFR TT and breast cancer risk in this cohort; effect modification by cSHMT may only be apparent in populations in which this association is observed. Further research in populations with varied folate nutrition may be needed to fully explore the complex interactions of folate, genetic variation, and disease risk. In conclusion, the results from a large, nested case-control study do not support an association of cSHMT 1420C→T genotype and breast cancer risk, and provide no evidence of an interactive effect of this polymorphism with MTHFR 677C→T.