Vitamin D is a key player in cell proliferation, differentiation, and apoptosis in normal and malignant breast cells 12. Indeed, the active form of vitamin D is hypothesized to have antiproliferative effects in many types of cells, including breast cancer cells 3456. The observation of a reduced risk for breast cancer among women with high vitamin D status lends support to this hypothesis 34.

Vitamin D has also been established as a determinant of bone density and, furthermore, higher bone density has been shown to be associated with risk for developing breast cancer 789. Whether the association between bone density and breast cancer is directly related to vitamin D or is explained wholly or in part by other factors such as the relationship of bone density to endogenous hormones, particularly estrogen, remains unclear.

The autocrine/paracrine pathway of vitamin D biosynthesis has been implicated in breast cell carcinogenesis 1. In this pathway, when circulating 25-hydroxyvitamin D reaches the mammary tissue it is further converted to 1,25-dihydroxyvitamin D by endogenous 1-α-hydroxylase in the breast 12. The locally produced 1,25-dihydroxyvitamin D may bind to the vitamin D receptor (VDR) and regulate cell development or progression 2.

Both the 5' and 3' termini of the VDR gene exhibit sequence level variation in the population. Three restriction fragment length polymorphisms at the 3' end of the gene (BsmI, ApaI, and TaqI restriction sites) have been observed to be in linkage disequilibrium with one another and with the variable length Poly(A) sequence in the 3'-untranslated region 101112131415. Because of the key role that VDR plays in the vitamin D biosynthesis pathway, it has been hypothesized that polymorphisms within the VDR gene may modify the risk for breast cancer, either singularly or through gene-gene or gene-environment interactions.

The long sequence of Poly(A) has been associated with both increased and decreased risks for breast cancer 101617. Additionally, variation in length of the Poly(A) tract has been associated with prostate cancer risk in a subset of studies 121314. Studies aimed at elucidating the association between the BsmI polymorphism and breast cancer risk have reported positive 31016, inverse 17, and null findings 181920. More consistent associations have been observed between the BsmI polymorphism and bone density 2122.

The inconsistency in findings to date is not surprising because most studies have been limited by sample size, and there has been considerable heterogeneity in study designs. Moreover, prior studies have focused on primarily Caucasian populations; hence, little is known about the association of VDR polymorphisms and breast cancer risk in African-American women. The present study was undertaken to assess whether VDR polymorphisms BsmI and Poly(A) are associated with breast cancer risk in a large population-based case-control study of Caucasian and African-American women aged 35 to 64 years.

Table 1 shows the distribution of demographic and family history characteristics within the sampled study population with data for one or both of the VDR polymorphisms (n = 3,066). As expected by design, cases were similar to controls with regard to age and race, and the proportions with first-degree family history in our sampled group (37.4% cases and 23.6% controls) exceeded those in the underlying CARE study (17.0% cases and 9.7% controls). All subsequent analyses account for the sampled nature of the data, including the over-sampling of women with a first-degree family history of breast cancer, through sampling weight adjustment. For Poly(A), there was excellent observer agreement in the 6% of randomly selected duplicates of genotyping results that were included for quality control purposes (κ statistic of 0.99). Similar information was not collected for BsmI. Laboratory personnel were blinded to case-control status.

We did not observe deviation from the expected Hardy-Weinberg frequencies for the Poly(A) genotypes in the controls (P = 0.69, P = 0.47, and P = 0.48 for combined controls, Caucasian controls, and African-American controls, respectively). However, we did observe significant deviation for the BsmI genotype (P < 0.01, P < 0.01, and P = 0.03 for combined controls, Caucasian controls, and African-American controls, respectively). BsmI and Poly(A) genotypes were in linkage disequilibrium for the Caucasian but not the African-American population (Caucasian: r² = 0.84; African American: r² = 0.47).

Within the CARE study population, we observed a slight increase in risk for breast cancer among postmenopausal Caucasian women with the BsmI bb genotype. There was little evidence of a similar relationship in premenopausal or African-American women. In addition, there was no evidence of a relationship between the Poly(A) LL genotype and risk for breast cancer overall, or within any racial or menopausal status subgroup. We do note that genotype frequencies were different by menopausal status; however, there is no biologic explanation for this difference. Automated genotyping systems, to the degree that they are flawed, have a tendency to overcall homozygotes and undercall heterozygotes, but there is no reason to assume that bb or BB genotypes would be differentially miscalled according to any exposure, including menopausal status. This observation does not change the overall conclusions of the study, and the differing results for premenopausal and postmenopausal Caucasian women are probably accounted for by small numbers.

Prior studies of VDR and breast cancer risk have been limited in number and yielded inconsistent results. Three hospital-based case-control studies of Caucasian women in the UK reported an increased risk for breast cancer in carriers of BsmI bb genotype 31016. However, these reports appear likely to involve overlapping populations. All three studies involved the same core group of authors and included patients from St. George's Hospital Medical School in London. Two used controls from the UK National Breast Screening Programme 1016, whereas the third used controls from a screening clinic 3. The largest of these studies 16 included 398 incident and prevalent breast cancer cases and 427 screened, disease-free controls. The investigators reported a 1.9-fold increased risk for breast cancer among carriers of the BsmI bb genotype, and a nearly twofold increase in risk for carriers of the Poly(A) LL genotype, when controlling for age, HT, and menopausal status. Participants were matched for age at sampling (controls) versus diagnosis (cases), but they differed significantly in terms of menopausal status and HT use. The second study 10 included 181 breast cancer cases and 241 controls from the same two institutions described above. Cases were not age-matched to controls. Controls were confirmed as having no detectable cancer at the time of sampling, although 89 had 'other' breast conditions such as calcifications, fibrocystic disease, or benign lumps. The authors reported a more than twofold increase in risk for breast cancer associated with both the BsmI bb genotype and Poly(A) LL genotype. The differences between this study and our own population-based study are significant, limiting our ability to compare results. The third study 3 was distinguished by the fact that the investigators also measured serum 25-hydroxyvitamin D levels. In this hospital-based, case-control study of 179 cases matched to 179 disease-free control volunteers by age, menopausal status (where possible), and time of year, a significant doubling in the risk for breast cancer was observed in women with BsmI bb compared with those with the BB genotype. The authors concluded that low circulating 25-hydroxyvitamin D levels, either alone or in combination with the BsmI Bb or bb genotype, may increase breast cancer risk.

In contrast, several prior studies did not report an increased risk for breast cancer associated with either the BsmI bb genotype or the Poly(A) LL genotype 1718192028. One of the largest, that conducted by Chen and colleagues 18, was conducted to investigate the role played by VDR polymorphisms and breast cancer risk in a nested case-control study of largely Caucasian women aged 43 to 69 years within the Nurses' Health study. The authors genotyped 1,180 cases and 1,547 controls for the BsmI and 1,234 cases and 1,676 controls for the Fok1 polymorphisms. They observed a significantly increased risk for breast cancer among carriers of the ff genotype of Fok1 (OR = 1.34, 95% CI = 1.06 to 1.69) as compared with those with the FF genotype, but they found no association between the BsmI polymorphism and breast cancer risk (OR = 0.93; 95% CI = 0.72 to 1.20) for BB versus bb. The conclusions did not change when the data were stratified by menopausal status. The Poly(A) variant was not tested.

To our knowledge, no prior studies of the VDR polymorphism and breast cancer risk have included a substantial number of African-American women. It is important to note that we did not observe a statistically significant increased risk for breast cancer associated with either the BsmI bb genotype or the Poly(A) LL genotype among the African-American women in our study. The (nonsignificant) reduced risk for breast cancer associated with carrying either the BsmI Bb or bb genotypes among never smokers, observed in African-American women, was probably due to small numbers in the referent category and not the presence of the allele or the effect of not smoking. Significantly larger studies of African-American women would be needed to address that question more completely.

Genetic factors such as the VDR polymorphism that influence disease risk may be subject to environmental influences. For example, smoking is an inconsistent risk factor for breast cancer 29, but the effects of polymorphisms in other genes on breast cancer risk have been observed to vary by smoking history 3031. Although not consistent in all subgroups, we did observe that the presence of any b allele elevated the risk for breast cancer among never smokers and that the relationship between the VDR polymorphisms and breast cancer risk varied among postmenopausal Caucasian women who reported ever or never smoking. The effect modification results should, however, be interpreted with caution; although we did observe an association between breast cancer and the BsmI bb genotype among postmenopausal Caucasian women, we did not observe an association between breast cancer and smoking in the same subgroup.

Vitamin D itself is proposed to have anticancer properties 323334. Indeed, it has been shown that calcitriol (1,25-dihydroxyvitamin D₃), a hormonal derivative of vitamin D₃, has an antiproliferative and prodifferentiation effect on several cell types, specifically the squamous cells of the head and neck 35. In a study of prostate cancer, Ma and coworkers 13 reported that vitamin D intake modified the risk for prostate cancer that was associated with VDR polymorphism (BsmI). Specifically, they observed a 57% reduction in risk among men with a plasma 25-hydroxyvitamin D level below the median who had the BB versus bb genotype (relative risk = 0.43, 95% CI = 0.19 to 0.98). The risk reduction was strongest among older men (relative risk = 0.18, 95% CI = 0.05 to 0.68).

We did not have information on vitamin D levels, via either dietary intake or metabolism from routine sun exposure. However, several studies 789 support a relationship between an increase in breast cancer risk and bone density. Two prospective cohort studies have reported an association between greater bone mass and breast cancer risk 78, and one of these studies suggests that vitamin D intake can modify the risk of osteoporosis in women being treated for breast cancer 7. In addition, men undergoing androgen ablation therapy experience osteoporosis 3637, which also supports a relationship between sex hormones and bone modeling. It is possible that the increased bone density observed in women with an increased risk of breast cancer may simply be a marker for high levels of vitamin D uptake, reflecting specific vitamin D receptor genotypes, or perhaps a marker of higher levels of endogenous estrogens.

Our data support a role for the BsmI b allele in the development of breast cancer in women who never smoked, but not in those who did smoke. Why a subset of VDR genotypes would have a differential effect on smokers versus nonsmokers is unclear. One hypothesis is that there is some as yet undefined interaction between estrogen metabolism and smoking that is exacerbated by specific levels of vitamin D status.

The strengths of the present study include its large size, generalizability, and the ability to control for most of the established risk factors for breast cancer. To date, this is the largest study of VDR polymorphisms and breast cancer risk. The population-based design, wider age range, and inclusion of both African-American and Caucasian women allow a more comprehensive portrayal of the frequency of mutations in the general population than has been available to date. Despite the generous sample size, the number of women with referent genotypes was small, resulting in an inability to make precise estimates. In addition, the study was limited by the fact that the gene was interrogated at only two positions at the 3' end of the gene. Newer technologies now permit a more comprehensive approach to genotyping, and variation within the entire gene will be readily examined in subsequent studies. The departure from Hardy-Weinberg equilibrium that we observed for the BsmI polymorphism could be attributed to any number of factors, including undetected ethnic diversity, biologic explanations, or genotyping misclassifications. We did not include blinded duplicates in the genotyping dataset, and therefore we did not have the opportunity to check for systematic genotyping errors.

One difficulty in studies of single nucleotide polymorphisms, such as this, is the issue of whether the variant being tested is truly associated with the disease or whether there is another polymorphism in linkage disequilibrium with the polymorphism under consideration that could be responsible for the observed association. Only when the whole gene can be examined along with associated regulatory elements can we develop a clearer picture of the underlying biology that consistently links the VDR gene to cancer risk.

This study provides additional support for an increased risk for breast cancer associated with the BsmI polymorphism at the 3' end of the VDR gene in postmenopausal Caucasian women and sheds light on the differential risk observed in menopausal status and race. Contributions of vitamin D intake to risk for breast cancer and bone mineral density, as well as potential effect modification by history of smoking, should be further studied to elucidate the precise biologic mechanisms.

BMI = body mass index; bp = base pairs; CARE = Contraceptive and Reproductive Experiences; HT = hormone therapy; OC = oral contraceptive; OR = odds ratio; PCR = polymerase chain reaction.

The authors declare that they have no competing interests.

BT performed the statistical analysis and drafted the manuscript. KM participated in study design and coordination, hypothesis generation, and manuscript preparation and also oversaw data collection and statistical analyses. JD, LB, PM, and BS participated in study design, data collection, and manuscript review. DD participated in the design of the study, manuscript review, and performed statistical analyses related to sampling. GU participated in manuscript review. MH carried out molecular genetic analyses. EO oversaw molecular genetic analyses, participated in study design, hypothesis generation, and manuscript preparation. All authors read and approved the final manuscript.