Vitamin D receptor genotype and breast cancer in Latinas (United States)

Cancer Causes and Control 11: 25±30, 2000. 25 Ó 2000 Kluwer Academic Publishers. Printed in the Netherlands. Vitamin D receptor genotype and breast cancer in Latinas (United States) Sue Ann Ingles1,*, Diana G. Garcia1, Wei Wang1, Alexandra Nieters1, Brian E. Henderson1, Laurence N. Kolonel3, Robert W. Haile1 & Gerhard A. Coetzee2 1 Department of Preventive Medicine, University of Southern California/Norris Comprehensive Cancer Center, 1441 Eastlake Ave, MS 44, Room 6419, Los Angeles, CA 90033. Ph: (323) 865-0498; Fax: (323) 865-0473; E-mail: ingles@hsc.usc.edu; 2Department of Urology, University of Southern California/Norris Comprehensive Cancer Center, Los Angeles, CA; 3Cancer Research Center, University of Hawaii, Honolulu, HI (*Author for correspondence) Received 5 November 1998; accepted in revised form 6 August 1999 Key words: breast neoplasms calcitriol, cohort studies, polymorphism (genetics), receptors. Abstract Objective: Polymorphism in the vitamin D receptor (VDR) gene has been associated with variation in bone mineral density and with prostate cancer risk. The purpose of this study was to determine whether polymorphism in the VDR gene may also in¯uence breast cancer risk. Methods: Polymorphisms in the 5¢ and 3¢ ends of the VDR gene were genotyped for 143 Latina women with breast cancer and 300 cohort controls. Results: Both the BsmI and poly-A polymorphisms in the 3¢ end of the VDR gene were associated with breast cancer risk, with a trend for increasing risk with increasing number of BsmI B alleles or short (S) poly-A alleles. Compared to subjects having two long poly-A alleles (genotype LL), odds ratios (and 95% con®dence intervals) were 1.5 (1.0± 2.3) and 3.2 (1.5±6.9) for subjects having genotypes SL and SS, respectively. Compared to BsmI genotype bb, odds ratios (and 95% con®dence intervals) were 1.6 (1.1±2.5) and 2.2 (1.0±4.7) for genotypes Bb and BB respectively. The start codon polymorphism, FokI, was not associated with breast cancer risk. Conclusion: These results suggest that polymorphic variation in or near the 3¢ end of the VDR gene in¯uences breast cancer risk in Latina women. Introduction Vitamin D plays an important role in modulating transcription of genes involved both in calcium and phosphorus homeostasis and in cellular di€erentiation and proliferation (reviewed in refs. 1±3). An obligatory mediator of these e€ects is the vitamin D receptor (VDR). Both the 5¢ and 3¢ ends of the VDR gene are polymorphic. Polymorphism in the ®rst of two possible translation start codons [4] produces receptor variants di€ering in size and activity [5]. Allelic variation in the 3¢ end of the VDR gene, although less clearly related to function, appears to have phenotypic consequences for calcium metabolism [6, 7], vitamin D metabolism [8, 9], bone density (reviewed in ref. 10) and prostate cancer risk [9, 11±13]. The observation that normal human breast epithelial cells [14] and most breast cancers [15, 16] express VDR raise the possibility that polymorphism in the VDR gene may also in¯uence breast cancer risk. To test this hypothesis we genotyped polymorphisms in the 5¢ and 3¢ ends of the VDR gene for 143 Latina women with breast cancer and 300 cohort controls. Materials and methods Subjects The Hawaii±Los Angeles Multiethnic Cohort Study is an ongoing epidemiological study of Japanese-Americans and whites residing in Hawaii, and AfricanAmericans and Latinos residing in California. For a study of vitamin D receptor genotypes and breast cancer we obtained DNA samples on female subjects from the 26 two Los Angeles-based (i.e., African-American and Latino) sub-cohorts. In this ®rst report we focus on Latinas, a population with strong linkage disequilibrium in the genomic region of interest, which allows BsmI and poly-A genotypes to be interpreted as markers of the VDR 3¢UTR allelotypes [17]. African-Americans have not been included in this report because characterization of allelotypes and validation of markers in the AfricanAmerican population is still in progress. The Latina sub-cohort includes more than 12,000 Latina women, age 45±75 at recruitment, residing in Los Angeles County, who were recruited by sampling Los Angeles County residents from drivers' license ®les as previously described [18, 19]. Newly diagnosed cases of breast cancer among Latinas were ascertained through linkage of the cohort to the Los Angeles County Surveillance, Epidemiology, and End Results (SEER) cancer registry, which is estimated to be 99% complete. Blood samples were obtained from incident breast cancer cases and from approximately a 1% random sample of the cohort members to serve as a control group. The participation rate for sample collection was in excess of 70% among both cases and controls. All subjects signed informed consents and the study was approved by the University of Southern California Institutional Review Board, which oversees studies involving human subjects. S.A. Ingles et al. were performed using a likelihood ratio test for significance of a genotype variable coded as 0, 1, or 2. Expected genotype frequencies under Hardy±Weinberg equilibrium and under the hypothesis of no linkage disequilibrium were calculated from observed genotype frequencies among controls, using standard methods [22]. Testing for departure from the hypothesis of no linkage disequilibrium was performed by comparing observed and expected joint genotypic distributions using a chi-square test. Results Of the 300 control women, 55 were included in our previous report describing linkage disequilibrium in the 3¢UTR of the VDR gene [17]. Among 55 female and 43 male Latino cohort members, we had observed 81% agreement between BsmI and poly-A genotypes. With 300 control women now available from the same cohort, agreement between the BsmI and poly-A genotypes is now estimated to be 88% (Table 1). Table 1. Joint distribution of 3¢VDR marker genotypes (BsmI and Poly-A) among Latina controls Poly-A Genotyping An 825 bp region of genomic DNA containing the BsmI polymorphic site in intron 8 was ampli®ed and analyzed as previously described [12]. The existence of the cut allele, b, is indicated by the formation of a 625 base-pair product. A region surrounding the poly-A polymorphism was ampli®ed as previously described [13]. Products were separated on polyacrylamide sequencing gels and autoradiographed. Alleles were sized and categorized as short (S) or long (L) as previously described [12]. A region of approximately 280 bp of genomic DNA containing the FokI polymorphic site was ampli®ed and analyzed as previously described [20]. The presence of the cut allele, f, is indicated by a band at approximately 200 bp. LL SL Total BsmI bb Bb BB 12 4 0 7 94 12 0 14 157 19 112 169 Total 16 113 171 300 Agreement = (12 + 94 + 157)/300 = 88%. Table 2. Age-adjusted odds ratios and 95% con®dence intervals for breast cancer by 3¢VDR genotype in Latinas Controls n (%) BsmI bb Bb BB Cases n (%) OR (95% con®dence interval) 169 (56.4) 112 (37.3) 19 (6.3) 61 (42.7) 68 (47.5) 14 (9.8) 1.0 1.6 (1.1±2.5) 2.2 (1.0±4.7) p-trend = 0.01 Poly-A LL 171 (57.0) SL 113 (37.7) SS 16 (5.3) 62 (43.4) 65 (45.4) 16 (11.2) 1.0 1.5 (1.0±2.3) 3.2 (1.5±6.9) p-trend < 0.01 Statistical methods FokI, BsmI and poly-A odds ratios were estimated by ®tting standard unconditional logistic regression models [21], using two indicator variables to encode the three FokI, BsmI or poly-A genotypes, and two indicator variables to adjust for tertiles of age. Tests for trend SS Total 300 (100) 143 (100) 27 VDR genotype and breast cancer Both BsmI and poly-A genotype frequencies were in Hardy±Weinberg equilibrium among the cohort controls, with observed genotype frequencies (Table 2) being nearly identical to the expected Hardy±Weinberg frequencies of 57%, 37%, and 6% for genotypes bb or LL, Bb or SL, and BB or SS, respectively. Cases, on the other hand, had a lower than expected frequency of bb or LL and higher than expected frequencies of Bb or SL and BB or SS genotypes. Thus 3¢VDR genotype, as measured by either the BsmI or the poly-A marker, was associated with breast cancer risk (Table 2). Compared to subjects having the BsmI bb genotype, risk was increased by approximately 60% in heterozygotes (Bb) and more than two-fold in BB homozygotes. Similar results were obtained using poly-A as a marker of 3¢VDR genotype. The start codon polymorphism, FokI, was not in linkage disequilibrium with the BsmI and poly-A polymorphisms in the 3¢ end of the gene. The observed joint distribution of FokI and BsmI genotypes did not di€er signi®cantly from the expected distribution under the hypothesis of no linkage disequilibrium (Table 3), indicating that the two ends of the VDR gene segregate independently, or nearly so in this population. Moreover, the FokI polymorphism was not associated with breast cancer risk (Table 4). Genotype frequencies were similar for cases and controls, and odds ratios comparing genotypes Ff and € to FF were not signi®cantly di€erent from the null hypothesis value of 1. Table 3. Joint distribution of 5¢VDR (FokI) and 3¢VDR (BsmI) genotypes among Latina controls: observed frequencies (and expected frequencies under the assumption of no linkage disequilibrium) BsmI FokI FF Ff € bb Bb BB 51 (61) 89 (81) 29 (27) 45 (40) 50 (54) 17 (18) 11 (7) 8 (9) 0 (3) Test for departure from hypothesis of no linkage disequilibrium: p = 0.21. Table 4. Age-adjusted odds ratios and 95% con®dence intervals for breast cancer by 5¢VDR genotype in Latinas Controls n (%) Cases n (%) FokI FF Ff € 107 (35.7) 147 (49.0) 46 (15.3) 53 (37.1) 65 (45.5) 25 (17.5) Total 300 (100) 143 (100) OR (95% con®dence interval) 1.0 0.9 (0.6±1.4) 1.1 (0.6±2.9) Of the 143 cases, 22 (15%) were diagnosed with in-situ disease, 86 (60%) with stage 1 disease, and 35 (25%) with higher-stage disease. Results did not appear to di€er by stage; however, power to detect heterogeneity was low. The average age was 65.5 (s.d. = 8.3) years for cases and 62.7 (s.d. = 7.9) years for controls. Discussion In this cohort of Latina women, the frequency of the BsmI b allele, at 75%, is higher than frequencies typically reported for non-Latino white populations (e.g., 59% among 591 control subjects in the Physician's Health Study [9]; 57% among 169 non-Latino white control subjects residing in Los Angeles County [17]). However, the BsmI b allele frequency among our Latina control subjects is similar to the frequency of 71% observed among 103 Mexican-American women in an observational study of fracture risk in Northern California [23]. Although the Northern California study was not population-based, any bias due to selection of women at increased fracture risk would be expected to produce downward bias in the frequency of the BsmI b allele, which is associated with lower fracture risk. The BsmI b allele frequency among Latinos, at 71±75%, is lower than frequencies typically reported in Asian populations (e.g., 88% among 488 healthy premenopausal Japanese women [24]; 95% among 96 members of a Singapore cohort [17]). The observation of BsmI allele frequencies in Latinos that are intermediate between non-Latino whites and Asians is consistent with the known ethnic make-up (part European, part native-American) of this population. In this cohort of Latina women, breast cancer risk was associated with polymorphic variation in the 3¢ end of the VDR gene. The VDR, a transcription factor which regulates a number of genes involved in cell proliferation and di€erentiation (reviewed in refs 1±3), was ®rst observed in human breast cancer cells nearly 20 years ago [25]. Since that time, numerous studies have demonstrated that vitamin D and deltanoids (vitamin D analogues) inhibit proliferation of breast cancer cells both in vitro [25±33] and in vivo [15, 30±32, 34±36]. The anti-proliferative e€ect is con®ned to those cells possessing VDRs [28, 37], and is roughly proportional to VDR number [37]. The simplest hypothesis that might explain an association between breast cancer risk and 3¢VDR genotype is that two allelic variants encode receptors di€ering in steady-state expression or functional activity. Tests of this hypothesis, however, have yielded con¯icting results, with VDR gene expression having been reported 28 as: higher for BsmI B compared to b alleles [8, 38]; higher for b compared to B alleles [39]; and indistinguishable for B and b alleles [40, 41]. Contributing to these contradictory results may be the use of markers, such as BsmI, which do not lie in the 3¢UTR itself, to classify a relatively small number of cell lines in all of these studies. Although BsmI can be used as a marker of 3¢UTR polymorphisms in most populations [17], it cannot be presumed that the B allele, for example, is in cis with the functional allele of interest in any single cell line. Potentially functional polymorphisms in the 3¢UTR itself have not yet been tested in in vitro systems. Based on in vitro studies reported to date, the VDR 3¢UTR cannot be ruled out as a functional locus contributing to breast cancer risk. At the 5¢ end of the VDR gene, the start codon polymorphism, FokI, was not associated with breast cancer risk. Thus, the association between breast cancer and the 3¢VDR genotype is not due to linkage disequilibrium between the 3¢ end of the gene and the start codon polymorphism. Sequences upstream of the start codon are also unlikely candidates for the functional locus, since the start codon polymorphism lies outside the region of tight linkage disequilibrium surrounding the BsmI polymorphism. The region of disequilibrium extends at least 3 kb downstream from BsmI to the poly-A microsatellite approximately 1 kb from the end of the VDR 3¢UTR [17], and may extend further downstream to include other genes. Thus, we cannot rule out the possibility that BsmI and poly-A are markers for a nearby downstream gene. The association of 3¢VDR genotype with variation in traits that are clearly dependent on vitamin D status, such as calcium metabolism and bone mineral density, lends support to the hypothesis that polymorphism within the VDR gene itself is functionally signi®cant. The BsmI BB genotype, which we found to be associated with increased risk of breast cancer, has been associated with decreased bone density in most studies [42]. It can be hypothesized that variation in the VDR gene marked by the BsmI B allele, either by a€ecting VDR activity or VDR number, leads to decreased trans-activation of VDR target genes that in¯uence calcium metabolism or cellular growth or proliferation. Likely target genes include genes for insulin-like growth factor (IGF) binding proteins, which in addition to regulating bone formation, may mediate growth inhibitory e€ects of vitamin D in breast cancer cell lines [43, 44]. Although breast cancer and osteoporosis have been found to be inversely correlated at the population level [45, 46], this may be explained by the role of estrogen both in maintaining bone mineral density and in driving cellular proliferation in the breast. The relative contributions of estrogen and vitamin D status to risk of S.A. Ingles et al. breast cancer and osteoporosis, as well as possible interactions among estrogen, vitamin D, and the IGF system, need to be further studied. Finally, we note that the BsmI b and poly-A L alleles that were associated with protection against breast cancer were previously found to be associated with increased risk of prostate cancer [9, 12]. While the reason for this ®nding is not yet clear, it is not surprising that a steroid hormone such as 1,25(OH)2D3 may have di€erent e€ects in di€erent tissues. The VDR can act both as an activator and as a repressor of transcription, depending on the nature of the target gene promoter and on tissue-speci®c VDR interacting proteins [47]. The speci®c target genes and regulatory factors involved in breast- and prostate-speci®c VDR responses have not been identi®ed; nevertheless, even in the absence of mechanistic explanations, the ®nding of an association between VDR genotype and breast cancer risk supports the hypothesis that vitamin D may in¯uence breast cancer etiology. However, because the sequence variants de®ning the functional 3¢VDR genotype have not yet been identi®ed, it is especially important that this epidemiologic ®nding is replicated in other populations and in other ethnic groups. If con®rmed, these ®ndings suggest that vitamin D and/or vitamin D analogues may be useful for breast cancer prevention and/or treatment, and that assessment of VDR polymorphisms might someday be useful to identify individuals most at risk and/or most responsive to intervention. Acknowledgements We thank Wu Zhang for laboratory assistance and Hank Hwang for programming assistance. This study was supported by funds from the California Breast Cancer Research Program of the University of California, Grants number 1IB-0353 and 3IB-0089 and by NIH/NCI grant R01 CA54281. Dr Ingles was supported by the STOP Cancer Foundation. References 1. Minghetti PP, Norman AW (1988) 1,25(OH)2-vitamin D3 receptors: gene regulation and genetic circuitry. FASEB J 2: 3043±3053. 2. Darwish H, DeLuca HF (1993) Vitamin D-regulated gene expression. Crit Rev Eukaryot Gene Expr 3: 89±116. 3. Hannah SS, Norman AW (1994) 1a,25(OH)2 vitamin D3-regulated expression of the eukaryotic genome. Nutr Rev 52: 376±382. 4. Saijo T, Ito M, Takeda E, et al. (1991) A unique mutation in the vitamin D receptor gene in three Japanese patients with vitamin Ddependent rickets type II: Utility of single-strand conformation polymorphism analysis for heterozygous carrier detection. Am J Hum Genet 49: 668±673. VDR genotype and breast cancer 5. Miyamoto K, Kesterson RA, Yamamoto H, et al. (1997) Structural organization of the human vitamin D receptor chromosomal gene and its promoter. Mol Endocrinol 11: 1165±1179. 6. Dawson-Hughes B, Harris SS, Finneran S (1995) Calcium absorption on high and low calcium intakes in relation to vitamin D receptor genotype. J Clin Endocrinol Metab 80: 3657±3661. 7. Wishart JM, Horowitz M, Need AG, et al. (1997) Relations between calcium intake, calcitriol, polymorphisms of the vitamin D receptor gene, and calcium absorption in premenopausal women. Am J Clin Nutr 65: 798±802. 8. Morrison NA, Qi JC, Tokita A, et al. (1994) Prediction of bone density from vitamin D receptor alleles. Nature 367: 284±287. 9. Ma J, Stampfer MJ, Gann PH, et al. (1998) Vitamin D receptor polymorphisms, circulating vitamin D metabolites, and risk of prostate cancer in United States physicians. Cancer Epidemiol Biomarkers Prev 7: 385±390. 10. Morrison N (1998) Vitamin D receptor gene variants and osteoporosis: a contributor to the polygenic control of bone density. In: Feldman D, Glorieux FH, Pike JW, eds. Vitamin D. San Diego, CA: Academic Press, 713±732. 11. Taylor JA, Hirvonen A, Watson M, Pittman G, Mohler JL, Bell DA (1996) Association of prostate cancer with vitamin D receptor gene polymorphism. Cancer Res 56: 4108±4110. 12. Ingles SA, Ross RK, Yu MC, et al. (1997) Association of prostate cancer risk with genetic polymorphisms in vitamin D receptor and androgen receptor. J Natl Cancer Inst 89: 166±170. 13. Ingles SA, Coetzee GA, Ross RK, et al. (1998) Association of prostate cancer with vitamin D receptor haplotypes in AfricanAmericans. Cancer Res 58: 1620±1623. 14. Berger U, Wilson P, McClelland RA, et al. (1988) Immunocytochemical detection of 1,25-dihydroxyvitamin D receptors in normal human tissues. J Clin Endocrinol Metab 67: 607±613. 15. Colston KW, Berger R, Coombes RC (1989) Possible role for vitamin D in controlling breast cancer cell proliferation. Lancet 1: 188±191. 16. Berger U, McClelland RA, Wilson P, et al. (1991) Immunocytochemical determination of estrogen receptor, progesterone receptor, and 1,25-dihydroxyvitamin D3 receptor in breast cancer and relationship to prognosis. Cancer Res 51: 239±244. 17. Ingles SA, Haile RW, Henderson BE, et al. (1997) Strength of linkage disequilibrium between two vitamin D receptor markers in ®ve ethnic groups: implications for association studies. Cancer Epidemiol Biomarkers Prev 6: 93±98. 18. Reichardt JKV, Makridakis N, Henderson BE, Yu MC, Pike MC, Ross RK (1995) Genetic variability of the human SRD5A2 gene: implications for prostate cancer risk. Cancer Res 55: 3973±3975. 19. Monroe KR, Yu MC, Kolonel LN, et al. (1995) Evidence of an X-linked or recessive genetic component to prostate cancer risk. Nat Med I: 827±829. 20. Gross C, Eccleshall TR, Malloy PJ, Villa ML, Marcus R, Feldman D (1996) The presence of a polymorphism at the translation initiation site of the vitamin D receptor gene is associated with low bone mineral density in postmenopausal Mexican-American women. J Bone Miner Res 11: 1850±1855. 21. Breslow NE, Day NE (1980) Statistical Methods in Cancer Research, vol. 1: The Analysis of Case±Control Studies. Lyon: IARC Scienti®c Publications. 22. Weir BS (1990) Genetic Data Analysis. Sunderland, MA: Sinauer Associates Inc. 23. McClure L, Eccleshall TR, Gross C, et al. (1997) Vitamin D receptor polymorphisms, bone mineral density, and bone metabolism in postmenopausal Mexican-American women. J Bone Miner Res 12: 234±240. 29 24. Tokita A, Matsumoto H, Morrison NA, et al. (1996) Vitamin D receptor alleles, bone mineral density and turnover in premenopausal Japanese women. J Bone Miner Res 11: 1003±1009. 25. Eisman JA, Martin TJ, MacIntyre I, Moseley JM (1979) 1,25dihydroxyvitamin-D receptor in breast cancer cells. Lancet 2: 1335±1336. 26. Frampton RJ, Omond SA, Eisman JA (1983) Inhibition of human cancer cell growth by 1,25-dihydroxyvitamin D3 metabolites. Cancer Res 43: 4443±4447. 27. Chouvet C, Vicard E, Devonecc M, Saez S (1986) 1,25-dihydroxyvitamin D3 inhibitory e€ect on the growth of two human breast cancer cell lines (MCF-7, BT-20). J Steroid Biochem 24: 373±376. 28. Haussler CA, Marion SL, Pike JW, Haussler MR (1986) 1,25dihydroxyvitamin D3 inhibits the clonogenic growth of transformed cells via its receptor. Biochem Biophys Res Commun 139: 136±143. 29. Frappart L, Falette N, Lefebvre MF, Bremond A, Vauzelle JL, Saez S (1989) In vitro study of e€ects of 1,25 dihydroxyvitamin D3 on the morphology of human breast cancer cell line BT.20. Di€erentiation 40: 63±69. 30. Abe J, Nakano T, Nishi Y, Matsumoto T, Ogaata E, Ikeda K (1991) A novel vitamin D3 analog, 22-oxa-1,25-dihydroxyvitamin D3, inhibits the growth of human breast cancer in vitro and in vivo without causing hypercalcemia. Endocrinology 129: 832±837. 31. Colston KW, Chander SK, MacKay AG, Coombes RC (1992) E€ects of synthetic vitamin D analogues on breast cancer cell proliferation in vivo and in vitro. Biochem Pharmacol 44: 693±702. 32. Colston KW, Mackay AG, James SY, Binderup L, Chander S, Coommbes RC (1992) EB1089: a new vitamin D analogue that inhibits the growth of breast cancer cells in vivo and in vitro. Biochem Pharmacol 44: 2273±2280. 33. Mathiasen IS, Colston KW, Binderup L (1993) EB 1089, a novel vitamin D analogue, has strong antiproliferative and di€erentiation inducing e€ects on cancer cells. J Steroid Biochem Mol Biol 46: 365±371. 34. Oikawa T, Yoshida Y, Simamra M, et al. (1991) Antitumour e€ect of 22-oxa-1a-25-dihydroxyvitamin D3, a potent angiogenesis inhibitor of rat mammary tumours induced by 7,12-dimethybenz[a]anthracene. Anticancer Drugs 2: 475±481. 35. Abe-Hashimoto J, Kikuchi T, Matsumoto T, Nishii Y, Ogata E, Ikeda K (1993) Antitumor e€ect of 22-oxa-calcitriol, a noncalcemic analogue of calcitriol, in athymic mice implanted with human breast carcinoma and its synergism with tamoxifen. Cancer Res 53: 2534±2537. 36. Anzano MA, Smith JM, Uskokovic MR, et al. (1994) 1alpha,25dihydroxy-16-ene-23-yne-26,27-hexa¯uorocholecalciferol (Ro245531), a new deltanoid (Vitamin D analogue) for prevention of breast cancer in the rat. Cancer Res 54: 1653±1656. 37. Buras RR, Schumaker LM, Davoodi F, et al. (1994) Vitamin D receptors in breast cancer cells. Breast Cancer Res Treat 31: 191± 202. 38. Carling T, Rastad J, AÊkerstroÈm G, Westin G (1998) Vitamin D receptor (VDR) and parathyroid hormone messenger ribonucleic acid levels correspond to polymorphic VDR alleles in human parathyroid tumors. J Clin Endocrinol Metab 83: 2255±2259. 39. Verbeek W, Gombart AF, Shiohara M, Campbell M, Koe‚er HP (1997) Vitamin D receptor: no evidence for allele-speci®c mRNA stability in cells which are heterozygous for the TaqI restriction enzyme polymorphism. Biochem Biophys Res Commun 238: 77±80. 40. Mocharla H, Butch AW, Pappas AA, et al. (1997) Quanti®cation of vitamin D receptor mRNA by competitive polymerase chain reaction in PBMC: lack of correspondence with common allelic variants. J Bone Miner Res 12: 726±733. 30 41. Gross C, Musiol IM, Eccleshall TR, Malloy PJ, Feldman D (1998) Vitamin D receptor gene polymorphisms: analysis of ligand binding and hormone responsiveness in cultured skin ®broblasts. Biochem Biophys Res Commun 242: 467±473. 42. Cooper GS, Umbach DM (1996) Are vitamin D receptor polymorphisms associated with bone mineral density? A meta-analysis. J Bone Miner Res 11: 1841±1849. 43. Rozen F, Yang X-F, Huynh H, Pollak M (1997) Antiproliferative action of vitamin D-related compounds and insulin-like growth factor-binding protesin 5 accumulation. J Natl Cancer Inst 9: 652± 656. 44. Colston KW, Perks CM, Xie SP, Holly JMP (1998) Growth inhibition of both MCF-7 and Hs578T human breast cancer cell S.A. Ingles et al. lines by vitamin D analogues is associated with increased expression of insulin-like growth factor binding protein-3. J Molec Endocrinol 20: 157±162. 45. Cauley JA, Lucas FL, Kuller LH, Vogt MT, Browner WS, Cummings SR (1996) Bone mineral density and risk of breast cancer in older women. The study of osteoporotic fractures. JAMA 276: 1404±1408. 46. Zhang Y, Kiel DP, Kreger BE, et al. (1997) Bone mass and the risk of breast cancer among postmenopausal women. N Engl J Med 336: 611±617. 47. Haussler MR, Whit®eld GK, Haussler CA, et al. (1998) The nuclear vitamin D receptor: biological and molecular regulatory properties revealed. J Bone Miner Res 13: 325±349.