Vitamin D Receptor Gene Polymorphisms and Prognosis of Breast Cancer among African-American and Hispanic Women

Background Vitamin D plays a role in cancer development and acts through the vitamin D receptor (VDR). Although African-Americans have the lowest levels of serum vitamin D, there is a dearth of information on VDR gene polymorphisms and breast cancer among African-Americans and Hispanics. This study examines whether VDR gene polymorphisms are associated with breast cancer in these cohorts. Methods Blood was collected from 232 breast cancer patients (Cases) and 349 non-cancer subjects (Controls). Genotyping for four polymorphic variants of VDR (FokI, BsmI, TaqI and ApaI) was performed using the PCR-RFLP method. Results An increased association of the VDR-Fok1 f allele with breast cancer was observed in African-Americans (OR = 1.9, p = 0.07). Furthermore, the FbTA, FbtA and fbtA haplotypes were associated with breast cancer among African-Americans (p<0.05). Latinas were more likely to have the VDR-ApaI alleles (Aa or aa) (p = 0.008). The VDR-ApaI aa genotype was significantly associated with poorly-differentiated breast tumors (p = 0.04) in combined Cases. Kaplan-Meier survival analysis showed decreased 5-year disease-free-survival (DFS) in breast cancer patients who had the VDR-Fok1 FF genotype (p<0.05). The Cox regression with multivariate analysis revealed the independent predictor value of the VDR-FokI polymorphism for DFS. The other three variants of VDR (BsmI, TaqI and ApaI) were not associated with disease outcome. Conclusions VDR haplotypes are associated with breast cancer in African-Americans, but not in Hispanic/Latinas. The VDR-FokI FF genotype is linked with poor prognosis in African-American women with breast cancer.


Introduction
Several studies have demonstrated a decrease in Vitamin D receptor (VDR) expression in breast cancer cells compared to normal breast cells [1]. Decreased levels of VDR in breast cancer cells could be due to VDR gene polymorphisms [2], and/or DNA methylation [3][4][5]. Alterations in VDR expression and activity could lead to deregulation of Vitamin D uptake, metabolism, and serum levels of biologically active Vitamin D.
VDR gene has multiple gene polymorphisms [6], and four important single nucleotide polymorphisms (SNPs) in exon 2 and 39UTR region. These are VDR-FokI (rs2228570), VDR-BsmI (rs1544410), VDR-TaqI (rs731236) and VDR-ApaI (rs7975232). Functional studies have identified that the VDR-FokI polymorphism leads to a shorter VDR protein by altering a translation initiation site [7,8]. Large meta-analyses have demonstrated a significant association of the VDR FokI polymorphisms with breast cancer [9,10], and other cancers across multiple ethnic cohorts [11,12]. The functional role of VDR-BSMI, VDR-TaqI, and VDR-ApaI, which are all located near the 39UTR region, are not as clear. However, studies do suggest that these polymorphisms may alter the polyadenylation of the VDR mRNA transcript, and thus affect mRNA stability [13]. Several studies show contradictory results on these polymorphisms and risk for breast cancer. Some studies found an association of breast cancer with BsmI, TaqI and ApaI [14][15][16][17][18] polymorphisms while others found no associations [19][20][21].
We have examined four VDR polymorphisms (FokI, TaqI, BsmI, ApaI) in African-American and Hispanic women from South Los Angeles and their association with breast cancer risk and survival.

VDR polymorphisms
Genomic DNA was extracted from buffy coat samples using the Qiagen DNA extraction kit. Vitamin D receptor genotyping for all four single-nucleotide polymorphic sites (SNPs) was done by the PCR-RFLP method (polymerase chain reaction-restriction fragment length polymorphism analysis). We used the following flanking primers [16,21]. The BsmI polymorphism was detected using one primer originating in exon 7 (Forward: 59-CAACCAA-GACTACAAGTACCGCGTCAGTGA-39) and a second in intron 8 (Reverse: 59-AACCAGCGGAAGAGGTCAAGGG-39), amplifying an 823 base pair (bp) fragment spanning the BsmI site. The ApaI and TaqI polymorphisms were detected using one primer in intron 8 (Forward: 59-AGAGCATGGACAGGGAG-CAAG-39) and the other in exon 9 (Reverse: 59-GCAACTCCT-CATGGCTGAGGTCTCA-39), yielding a 745 bp fragment spanning the ApaI and TaqI site. VDR-FokI polymorphic site was amplified using following primers Forward -59GATGC-CAGCTGGCCCTGGCACTG39 and Reverse -59ATGGAAA-CACCTTGCTTCTTCTCCCTC 39. PCR conditions were as follows: denaturation at 94uC for 5 min, followed by 40 cycles of PCR at 94uC (30 sec), 60uC-63.5uC (30 sec), and 72uC (30 sec). Annealing temperature for BsmI, ApaI/TaqI and FokI PCR are 63.5uC, 61uC and 59.5uC, respectively. Following PCR, aliquots of the amplified PCR products was digested with BsmI, ApaI, TaqI and FokI in accordance with the manufacturer's specifications (New England Biolabs, Beverly, MA, USA). The presence (lowercase) or absence (uppercase) of the enzyme recognition site was identified by ethidium bromide staining of fragments separated in a 2% agarose gel. Genotypes were assigned as BB, Bb, bb for the BsmI polymorphisms, AA, Aa, aa for the ApaI polymorphisms, TT, Tt and tt for the TaqI polymorphisms and FF, Ff and ff for the VDR-FokI polymorphisms. For the BsmI cleaved PCR products, the single resultant fragment was 823 bp, whereas the two resultant fragments were 648 bp and 175 bp. The PCR products digested with TaqI enzyme resulted in fragments of 496 bp and 249 bp in the presence of the polymorphic cutting site. PCR products digested with ApaI enzyme result in fragments of 531 bp and 214 bp in the presence of the polymorphic cutting site. The PCR products of VDR-FokI polymorphism (273 bp) digested with FokI enzyme result in 198 bp and 75 bp products in the presence of restriction site.

Statistical Analysis
We used SPSS (ver 11.5) software to analyze our data. The Pearson Chi-square test was used to determine the statistically     Table 3. Cont.  Table 4. Linkage disequilibrium (LD) between VDR polymorphisms in African-American and Hispanic/Latina women.   [22]. Associations between each VDR polymorphism and breast cancer was estimated using the Logistic Regression test at 95% confidence interval, and controlled for age and BMI. For each polymorphic site, the wild type genotype (FF, BB, AA, TT) was considered as the reference based on published literature. We also performed analysis of linkage disequilibrium (LD) between polymorphic sites in a locus, to identify ''clusters'' of highly correlated sites based on LD statistics. We calculated the haplotype frequencies and LD in Cases and Controls using the Haplotype software package available online [23,24]. Table 1 shows the characteristics of subjects in the study. The mean age of African-American Cases and Controls is 52.9 and 50.7 years, respectively, and for Hispanic/Latinas Cases and Controls is 49.2 and 45.6 years, respectively. The mean BMI is .30 kg/m ' 2. Among Cases, the distribution of histopathological characteristics between African-American and Hispanic women with breast cancer was similar (Table 1), and approximately ,59.9% Cases had Stage I/II disease and 26.7% had Stage III/V disease. The majority of our patients had infiltrating ductal carcinoma (62.2%), moderately to poorly differentiated (.70%).

Results
The VDR gene polymorphisms were genotyped in 188 African-American and 393 Hispanics/Latina women. No deviation was observed from Hardy-Weinberg equilibrium (HWE) in the genotypic distribution of the VDR-FokI, VDR-BsmI and VDR-ApaI polymorphisms in the study subjects. However, the VDR-TaqI distribution was not within HWE. Further analysis identified that the deviation of VDR-TaqI was contributed from Cases (p allele = 0.75, q allele = 0.25, x 2 = 13.52), but not from Controls (p allele = 0.79, q allele = 0.21, x 2 = 2.34) -data not shown.
Logistic regression analysis was also performed (adjusted for age and BMI) and showed no significant association of breast cancer with any of the VDR polymorphisms when all of the Cases were  pooled together (Table 2). However, when the analysis was performed based on ethnic stratification of the African-American and Hispanic/Latina subjects, the VDR-FokI Ff genotype (OR = 2.2, 95% CI = 0.95-5.1, p = 0.06) and the f allele (OR = 1.9, 95% CI = 0.9-3.7, p = 0.07) showed an increased association with breast cancer among African-American subjects ( Table 2) compared with the Hispanic/Latina subjects. In the Hispanic/Latina population, we found no significant difference in the frequency distribution of the VDR genotypes between Cases and Controls, and no association with any of the genotypes with breast cancer based on logistic regression analysis (Table 2).
Haplotype analysis of the polymorphic variants produced 16 different haplotypes ( Table 3). The FBTA (FokI-F, BsmI-B, TaqI-T and ApaI-A) was the most common haplotype in Controls and Hispanic/Latina Cases ( Table 3). The most common haplotype in African-American Cases was the FbTA haplotype (FokI-F, BsmI-b, TaqI-T and ApaI-A). Several VDR-haplotypes were significantly associated with breast cancer in the African-American cohort, however, no association was observed between breast cancer and VDR-haplotypes in the Hispanic/Latina cohort. Logistic regression analysis, using wild-type alleles (FBTA) as the reference category, revealed that FbTA was 1.9 times higher (p = 0.02), FbtA was 1.6 times higher (p = 0.03) and fbtA was 1.4 times higher (p = 0.05) in Cases than in Controls in the African-American cohort (Table 3). Logistic regression analysis performed in Cases showed that African-American women with breast cancer were more likely to have the FbTA (OR = 2.4, p = 0.005) and FbtA (OR = 1.7, p = 0.04) haplotypes compared with Hispanic/Latina women with breast cancer. In contrast to Cases, African-American Controls with the fbTa (OR = 0.8, p,0.001), fbTA (OR = 0.8, p = 0.03), fBTA (OR = 0.8, p = 0.04) and fbtA (OR = 0.7, p = 0.02) haplotypes were significantly less frequently observed than among Hispanic/Latina Controls. Table 4 demonstrates the results from linkage disequilibrium (LD) analysis performed on the four VDR variants. The strongest LD is observed between VDR-TaqI and VDR-ApaI among Hispanic/Latina subjects (D9 = 0.66). Weaker LDs were observed between VDR-FokI and the other 3 variants (VDR-BsmI, VDR-ApaI and, VDR-TaqI) in both African-American and Hispanic/Latinas.
The associations between the VDR genotypes and breast tumor histopathology, tumor stage, and metastatic disease progression were assessed in Table 5. The VDR-FokI FF genotype was significantly more common in African-American patients than in Hispanic/Latina patients (46.1% vs. 35.9%, respectively, p = 0.02), while the VDR-FokI ff genotype was more common in Hispanic/Latina patients as compared to African-American patients (17.1% vs. 6.1%, respectively). African-American patients were more likely to have VDR-ApaI AA genotype and Hispanic/ Latina patients were likely to have VDR-ApaI Aa or aa genotypes (p = 0.008). There was also a significant association identified between VDR-ApaI polymorphisms and tumor differentiation (p = 0.04). Poorly differentiated tumors were more likely to express a polymorphic allele of VDR-ApaI (Aa or aa). There were no other significant associations found between any of the polymorphic VDR variants and tumor stage, tumor differentiation, and ER/ PR/HER2 receptor status.
The 5-year disease-free-survival (DFS) from breast cancer in association with VDR polymorphisms is shown in Figure 1. The 5year DFS in patients carrying VDR-FokI FF genotype is reduced significantly compared to VDR-FokI Ff and ff genotypes (p = 0.05). The VDR-BsmI, VDR-ApaI and VDR-TaqI polymorphisms did not influence DFS significantly. Results of multivariate analyses, adjusted for the tumor characteristics, ethnicity, and the four VDR polymorphic variants, confirm the independent predictor value of VDR-FokI in DFS as shown in Table 6. The relative risk of tumor recurrence or metastases was decreased by 43% in patients who had the polymorphic VDR-FokI alleles, Ff or ff, compared to subjects who had VDR-FokI FF genotype (Table 6).

Discussion
The results from our study reveal an increased association between one of the VDR polymorphic variants, VDR-FokI, and breast cancer in the African-American cohort. The odds ratio (OR) for breast cancer in association with the VDR-FokI f allele was 1.9 (95% CI = 0.9-3.7; p = 0.07). This is consistent with other studies [25] which have identified a positive association between the VDR-FokI ff genotype and/or f allele and breast cancer ( Table 7). Our study further identifies that this trend is present among African-American women, but not Hispanic/Latina women. The other most commonly studied VDR polymorphism, VDR-BsmI, was not associated with breast cancer in our study, similar to most of the other studies conducted on this polymorphism (Table 7). While Ingles et al [14] identified a significant association of the B allele with breast cancer among Hispanics, recent large studies and meta-analyses on various ethnic groups, including Hispanics, have suggested otherwise. For example, the large analyses conducted by Rollison et al [19] and Tang et al [10] have confirmed that there is no significant association between VDR-BsmI and breast cancer. Furthermore, data from our current study shows no associations between the two other VDR variants (TaqI and ApaI) and breast cancer, consistent with most studies to date on Caucasian populations ( Table 7). The VDR-TaqI (tt genotype) and VDR-ApaI (a allele) were significantly associated with breast cancer in only two hospital-based studies [15,16]. However, these polymorphisms were found to be nonsignificant in recent large population meta-analyses [10]. Linkagedisequilibrium analysis of the polymorphisms in our study is also consistent with findings from previous studies, confirming linkage between VDR-TaqI and VDR-ApaI in Hispanic/Latina women, and absence of LD between the VDR-FokI and other SNPs [6].
Additional studies on VDR SNPs [26][27][28][29][30][31][32][33][34][35] are presented in Table 7. Abbas et al [26], suggested that VDR polymorphisms may affect postmenopausal breast cancer risk, and be associated with receptor status. Our study shows no receptor-specific association (ER/PR/HER +/2) with any of the VDR polymorphic variants; however, the VDR-ApaI aa genotype is significantly associated with poorly differentiated tumors (p = 0.04). Since the VDR-ApaI a allele is most common in Hispanic/Latinas, and the Latinas in the present study were slightly younger than the African-American subjects, these findings may potentially implicate VDR-ApaI in the progression of cancer among younger breast cancer patients. Further investigation among larger numbers of younger women with breast cancer is necessary in order to discern a definite association.
Lastly, recent reviews have concluded that the scientific evidence between vitamin D status and breast cancer risk and outcome are fairly strong, and improving vitamin D levels could improve health status [36]. Unfortunately, studies to date which have utilized vitamin D supplementation to overcome vitamin D insufficiency have not yielded conclusive improvements in cancer health risks and outcomes [37,38]. Hence examination of the Vitamin D receptor (VDR), and genetic changes such as polymorphisms potentially impacting VDR protein function may be central in identifying confounding factors that play a role in Vitamin D activity and ultimately patient outcome. Our study identifies that the 5-year disease-free-survival from breast cancer is significantly reduced among breast cancer patients with the VDR-FokI FF genotype (Log rank = 6.06 and p = 0.05). Cox regression with multivariate analysis demonstrates the independently predictive value of VDR-FokI on DFS. Hence, findings from our study suggest that while the VDR-FokI f allele may play a role in early association with breast cancer development, the F allele may play a role on tumor progression and patient outcome. The findings on these VDR variants, together with significant differences in haplotypes between Cases and Controls suggest the potential need to screen for VDR polymorphisms in the context of breast cancer, particularly before considering Vitamin D supplementation.