Parts of this work were presented at the 18th Annual Meeting of the American Society for Bone and Mineral Research in Seattle, WA, U.S.A., 1996 and at the 29th Annual Meeting of the Society for Epidemiologic Research in Boston, MA, U.S.A., 1996.
Bone mineral density (BMD) is under genetic control. Some studies in Caucasian and Asian women suggest that polymorphisms in the vitamin D receptor (VDR) gene are associated with BMD and the rate of postmenopausal bone loss. We determined if similar associations exist in 101 African-American women aged 65 years and older (71 ± 5 years, mean ± SD). We also examined the relation between VDR genotype and fractional45Ca absorption and markers of bone formation (osteocalcin) and resorption (N-telopeptides) in these women. BMD was measured at the proximal femur and whole body at baseline and after 1.9 ± 0.4 years (femur only) on a Hologic QDR-2000 densitometer using dual-energy X-ray absorptiometry. Calcaneal BMD was measured with single x-ray absorptiometry. VDR gene polymorphisms were defined by the endonucleases BsmI, ApaI, and TaqI. These polymorphisms were not associated with BMD at any skeletal site or with markers of bone turnover. There was a significant interaction between age and VDR genotype where the oldest women (>70 years) with the tt genotype experienced greater hip bone loss than women with the TT genotype (−2.1%/year vs. −0.4%/year, respectively), whereas heterozygous women experienced an intermediate rate of bone loss (−1.3%/year). Women homozygous for the B allele had 14% lower fractional45Ca absorption compared with women homozygous for the b allele, although this difference was not statistically significant (p = 0.08). We conclude that VDR gene polymorphisms are not associated with BMD or indices of bone turnover in this population of older African-American women. However, DNA sequence variation in the VDR gene or a nearby locus may influence intestinal calcium transport and the rate of postmenopausal bone loss in African-American women.
BONE MINERAL DENSITY (BMD) is under strong genetic control.1 As much as 60% of the variance in BMD has been attributed to genetic factors,1 and at least one twin study suggests a genetic influence on age-related bone loss.2 Some of the genetic influence on BMD may be mediated by DNA polymorphisms in the gene encoding the receptor for 1,25-dihydroxyvitamin D (VDR), an important regulator of intestinal calcium absorption and bone mineralization.3 In a study of Australian twins, Morrison et al.4 found lower lumbar spine BMD in women homozygous for the absence of a BsmI restriction endonuclease site in the VDR gene compared with women who were homozygous for the presence of this site. Subsequent studies have also linked population variance in premenopausal bone mass5–8 and the rate of postmenopausal bone loss9–11 to this VDR gene polymorphism. Although the mechanisms underlying this VDR gene effect remain largely undefined, some studies suggest that VDR genotype influences intestinal calcium transport12 and bone turnover.13,14 Nevertheless, attempts to replicate these findings have yielded conflicting results,15–20 and studies of VDR gene polymorphisms have almost exclusively been confined to Caucasian and Asian women.
Age-specific hip fracture rates in African-American women are about half those in Caucasian women.21 This difference in fracture risk may reflect in part a greater peak bone mass22 and slower age-related bone loss23 among African-Americans. To our knowledge, only one study has examined the skeletal effects of VDR gene polymorphisms in African-American women.5 That study confirmed an association between the BsmI VDR polymorphism and premenopausal bone mass,5 but did not examine the relation between VDR genotype and the rate of postmenopausal bone loss. Furthermore, two additional polymorphisms in the VDR gene defined by the endonucleases ApaI and TaqI24 were not assessed in that study.
In the present study, we examined the associations between the BsmI, ApaI, and TaqI VDR gene polymorphisms and bone mass and the rate of bone loss in a sample of community-dwelling African-American women aged 65 years and older. We also determined if VDR polymorphisms are related to fractional calcium absorption and biochemical markers of bone turnover in these women.
MATERIALS AND METHODS
A total of 156 African-American women were recruited between January, 1993 and August, 1994 for an ancillary study to the Study of Osteoporotic Fractures.25 Women were recruited by mailing information to population-based lists of voters, staff talks to community groups, paid advertisements, and word of mouth. Women who were at least 65 years of age or older, able to walk without the assistance of another person, community living, and who had not undergone a bilateral hip replacement were invited to participate.
Of the 156 African-American women who participated in the baseline examination, 113 returned for a second clinic examination between March, 1995 and August, 1995 after an average of 1.9 years (range, 0.8–2.5 years). Peripheral blood leukocytes for DNA extraction and genotyping were available on 101 of these women for the present analyses. The protocol for this study was approved by the Institutional Review Board at the University of Pittsburgh, and written informed consent was obtained from each participant.
High molecular weight DNA was extracted from peripheral leukocytes by the salting-out procedure of Miller.26 Detection of the BsmI site in intron 8 was facilitated by polymerase chain reaction (PCR) amplification of a region spanning the BsmI site with primers originating in exon 7 (primer 1: 5′-CAACCAAGACTACAAGTACCGCGTCAGTGA-3′) and intron 8 (primer 2: 5′-AACCAGCGGGAAGAGGTCAAGGG-3′) producing an 825 basepair (bp) fragment.4 We were unable to amplify DNA adequately for BsmI genotyping on 14 samples. Detection of the ApaI and TaqI sites was facilitated using a single amplification with primers in intron 8 (5′-CAGAGCATGGACAGGGAGCAAG-3′) and exon 9 (5′-GCAACTCCTCATGGCTGAGGTCTCA-3′) producing a 740 bp fragment.14 Fragments were generated in a 60 μl reaction volume containing 100–200 ng of DNA, 0.46 μM of each primer, 185 μM of dNTPs, 50 mM KCl, 10 mM Tris-HCl (pH = 9.0), 1.5 mM MgCl2, 0.1% Triton X-100, and 0.8 U of Taq DNA polymerase.
A 10 μl aliquot of each PCR product was digested with 5 U of endonuclease BsmI (New England Biolabs, Beverly, MA, U.S.A.) at 65°C, ApaI at 37°C, or TaqI at 65°C for 1 h. Each digested PCR product was separated on a 2% agarose gel containing ethidium bromide for 3 h at 90 V. DNA fragments were visualized by ultraviolet illumination and fragment sizes estimated by comparison to a 1 kb ladder run on the same gel. The presence of the BsmI restriction site generates 175 bp and 650 bp fragments, whereas the absence of this site yields a 825 bp fragment. Homozygous absence of the TaqI site yields bands of 245 bp and 495 bp, whereas the homozygous presence of this site yields fragments of 205, 245, and 290 bp. Heterozygotes for the TaqI site exhibit fragments of 490, 290, 245, and 205 bp. Digestion of the 740 bp PCR product with ApaI yields fragments of 220 bp and 520 bp for the presence of the restriction site, whereas the absence of the restriction site leaves the fragment undigested. The presence of the restriction enzyme site is indicated by a lowercase letter and the absence of the site by an uppercase letter.
Measurement of bone mass and body composition
BMD (g/cm2) of the total hip, its subregions (femoral neck, trochanter, and intertrochanter), and whole-body BMD were measured on a Hologic QDR-2000 densitometer using dual-energy X-ray absorptiometry (DEXA) (Hologic, Inc., Waltham, MA, U.S.A.). Proximal femur BMD measurements were repeated at the second clinic examination. Paired baseline and follow-up scans were analyzed using the automated compare feature of the QDR 2000. Details of the DXA quality control procedures27 and longitudinal performance of the densitometers are published elsewhere.28 Calcaneal BMD was measured at the baseline clinic visit using single X-ray absorptiometry (SXA) (OsteoAnalyzer, Dove Medical Corp., Newburry Park, CA, U.S.A.). SXA quality assurance phantom scans were measured on a daily basis.
Broadband ultrasound attenuation (BUA) at the right calcaneus was measured on the same heel as calcaneal BMD using a Walker-Sonix UBA 575+ (Hologic, Inc.) using the scanning protocol provided by the manufacturer. Two measurements with repositioning were obtained on most participants. A third measurement was obtained if values differed by 10 db/MHz or more. The short-term reproducibility of BUA was assessed by obtaining three measurements with repositioning on 40 unselected volunteers. The mean coefficient of variation was 5.0%. Acoustic phantoms provided by the manufacturer were scanned weekly and showed no drift over time.
Blood and urine samples were collected in the morning after a 5 h fast for measurement of biochemical markers of bone turnover. Serum and urine were immediately frozen at −70°C until analysis. Serum osteocalcin levels were determined by radioimmunoassay using highly specific rabbit antiserum raised against purified bovine osteocalcin, together with a bovine osteocalcin standard (Endocrine Sciences, Calabasas Hills, CA, U.S.A.). Urinary excretion of type I collagen cross-linked N-telopeptides (NTx) was measured as a marker of bone resorption in an enzyme-linked immunosorbent assay using a monoclonal antibody to NTx labeled with peroxidase enzyme (Endocrine Sciences). The urinary creatinine concentration was determined by standard colorimetric methods, and NTx values were corrected for urinary creatinine excretion. Intra- and interassay coefficients of variation for osteocalcin are 2.4% and 11.0%, respectively. Intra- and interassay coefficients of variation for NTx are 6.5% and 7.8%, respectively.
Calcium absorption testing was completed in the morning after a 5 h fast. Participants were instructed to avoid any alcohol-containing beverages for 24 h and any calcium supplements for 12 h prior to testing. Fractional45Ca absorption was estimated from the appearance of45Ca in blood after ingestion of 50 g of labeled apple juice (containing 63 mg of calcium) and 120 g of additional unlabeled Speas Farm apple juice (Sundor Brands, Inc., Mt. Dora, FL, U.S.A.). The total test load of calcium was 215 mg.45Ca was supplied by Dr. Robert P. Heaney, Osteoporosis Research Center, Creighton University, Omaha, NE, U.S.A. Actual dosing occurred midway during consumption of a standard light meal. Exactly 3 h after ingestion of the tracer, 15 ml of blood was drawn into a serum separator tube and allowed to clot at room temperature. Serum was separated within 2 h of collection and frozen at −70°C until analysis. Frozen serum samples were later shipped on dry ice by overnight delivery to Creighton University, Omaha, NE, U.S.A. for estimation of fractional calcium absorption using the single isotope method as described by Heaney et al.29,30
Body weight was measured without shoes or heavy outer clothing during the baseline and follow-up clinic examinations in a standardized fashion using a calibrated balance beam scale. Height was measured without shoes using a Harpenden stadiometer (Holtain Ltd., Dyved, U.K.). Height and weight were used to calculate body mass index (kg/m2).
Allele frequencies were estimated by gene counting. Hardy-Weinberg equilibrium was tested for each restriction fragment length polymorphism by a chi-square goodness of fit test. Rate of change in BMD at the hip was calculated from baseline and follow-up DEXA measurements and expressed as an annualized percentage (%/year) rate of change from initial BMD values. Analysis of variance (ANOVA) and covariance (ANCOVA) were used to test for differences across the VDR genotypes. We also determined if VDR gene polymorphisms modify age-related bone loss at the hip by dividing the group at their median age (70 years) and performing two-way ANCOVA with terms for age group, VDR genotype, and their interaction. We adjusted these models for baseline BMD at the respective site, baseline body weight, and change in body weight over time. This study had 80% power to detect a 0.8–0.9 SD BMD difference between homozygous genotypes at a two-sided α error of 5%. All statistical analyses were performed with SAS software (SAS Institute, Inc., Cary, NC, U.S.A.).
Gene frequencies for each restriction site were in Hardy-Weinberg equilibrium (B, 0.36; b, 0.64; A, 0.67; a, 0.33; T, 0.58; t, 0.42). Consistent with previous observations in Caucasian4 and Japanese31 women, the T and b alleles were closely associated (p < 0.001). There were genotype-related differences in age and body weight (Table 1). Thus, all subsequent analyses were adjusted for age and weight.
Table Table 1. MEAN (SD) CHARACTERISTICS OF OLDER AFRICAN-AMERICAN WOMEN BY VITAMIN D RECEPTOR GENOTYPE
The BsmI, TaqI, and ApaI genotypes were not significantly associated with BMD at any skeletal site or with calcaneal broadband ultrasound attenuation (Table 1). The annualized rate of change (%/year) in BMD at the total hip and subregions did not differ significantly by VDR genotype (data not shown). Further statistical adjustment for baseline BMD, baseline body weight, and change in body weight over time did not alter these results.
There was a significant interaction between age and the TaqI VDR polymorphism where the oldest women with the tt genotype had about five times higher bone loss than women with the TT genotype, whereas heterozygous women had intermediate rates of bone loss (Fig. 1). A significant interaction between age and the BsmI polymorphism was also present, where the oldest heterozygous women experienced the greatest bone loss (Fig. 1). Similar results were observed at the femoral neck, greater trochanter, and intertrochanteric regions of the hip (data not shown). In contrast, a significant interaction between age and the ApaI polymorphism was not observed.
Biochemical indices of bone formation (osteocalcin) and bone resorption (N-telopeptides) were not related to the VDR gene polymorphisms (Table 1). Fractional45Ca absorption was 14% lower in women with the BB genotype compared with those with the bb genotype, although this difference did not achieve statistical significance (p = 0.08). Fractional45Ca absorption was not related to the TaqI or ApaI VDR polymorphisms. There was also no significant interaction between age and the VDR genotypes for fractional calcium absorption, osteocalcin, or N-telopeptides in this sample of older African-American women (data not shown).
The present study examined the association between DNA polymorphisms in the VDR gene defined by the BsmI, ApaI, and TaqI restriction endonucleases and BMD at the proximal femur, calcaneus, and whole-body and calcaneal ultrasound in older community-dwelling African-American women. None of the individual restriction fragment length polymorphisms were associated with BMD or calcaneal ultrasound. There was also no difference in biochemical markers of bone formation (osteocalcin) or bone resorption (N-telopeptides) across the VDR genotypes. These results indicate that the BsmI, ApaI, and TaqI polyorphisms do not have a major influence on bone mass or bone turnover in our population of older African-Americans.
Recent studies on the association between VDR gene polymorphisms and BMD have yielded conflicting results. Our findings contrast with studies in older Caucasian4,6,7 and Asian women11,31 showing that the B and t32 alleles are associated with lower BMD, but are consistent with other reports in Caucasian women that were unable to find differences in BMD between VDR genotypes.15,17,19,20,33,34 Our results also conflict with those of Fleet et al.5 who documented lower femoral neck BMD in premenopausal African-Americans with the B allele. These conflicting results may be due in part to the doubtful functional significance and low informativeness of the diallelic VDR polymorphisms. For example, the BsmI and ApaI restriction sites lie in an untranslated region of the VDR gene and probably do not confer any functional diversity per se.4 Similarly, the silent nucleotide substitution in exon 9 that creates the TaqI polymorphism does not affect the amino acid composition of the VDR protein.4 These neutral polymorphisms should therefore be considered as possible markers, in linkage disequilibrium with functionally relevant genetic variants affecting the structure or expression of the vitamin D receptor or with a nearby bone metabolism gene. Since linkage disequilibrium may exist in one population but not in others, associations due to linkage disequilibrium often yield contradictory results in studies of complex traits.35 Interactions between VDR alleles and lifestyle factors36 may further exacerbate the likelihood of conflicting results within and between populations.
Since conclusions drawn from observations based on linkage disequilibrium can only be applied over relatively small genetic distances, our negative findings for the individual VDR polymorphisms cannot completely exclude the whole VDR locus from influencing BMD. Furthermore, the present study may have had limited statistical power to detect small differences among the VDR genotypes. In the future, large population-based studies employing higher-resolution haplotypes may yield greater insight into the possible relationship between DNA sequence variation in the VDR gene and BMD, including the identification of the probable functional variant. Indeed, one recent study employing VDR haplotype analysis in 1782 elderly men and women in the Netherlands documented lower BMD in individuals homozygous for the bAT VDR haplotype.37 Too few subjects were available in the present study to perform haplotype analysis.
The rate of bone loss at the hip increases steadily with advancing age in older women.28 The present results indicate that the TaqI VDR alleles may be codominantly associated with hip bone loss in our oldest African-Americans. Our oldest African-American women with the tt genotype had about five times greater bone loss at the hip than women with the TT genotype, whereas heterozygous women experienced an intermediate rate of bone loss. The annualized rate of bone loss at the hip in our oldest African-American women with the TT genotype (−0.4%/year) is comparable to the overall rate of hip bone loss reported in Caucasian women aged 65 years and older (−0.6%/year),28 whereas the rate of loss in our oldest African-Americans with the tt genotype (−2.0%/year) is approximately three times greater than that observed in Caucasians (−0.6%/year). Thus, while the overall rate of bone loss at the hip may be less in African-Americans,23 our results suggest that bone loss may be greater in older African-American women who are homozygous for the t allele. A significant interaction between age and the BsmI polymorphism was also present, but there was no evidence of an allele dose effect for this particular polymorphism. Nevertheless, subdivision of our women into the two age categories resulted in smaller sample sizes in the VDR genotype classes, and our findings for the TaqI polymorphism will require confirmation in a larger cohort of African-American women.
Only a few other prospective studies have determined if VDR genotype is associated with the rate of postmenopausal bone loss. Keen et al.38 also observed greater hip bone loss in early postmenopausal women with the tt genotype, although this difference was not statistically significant. Krall et al.9 found significantly greater bone loss at the femoral neck, radius and lumbar spine in postmenopausal Caucasian women with the BB VDR genotype, but were unable to document a significant interaction between years since menopause and this particular VDR polymorphism. Two other studies also noted greater bone loss in subjects with the BB genotype,10,11 whereas this observation was not confirmed in either early or late postmenopausal women.16,18
The active metabolite of vitamin D, 1,25-dihydroxvitamin D (1,25(OH)2D), is known to stimulate intestinal calcium absorption, and even subtle qualitative or quantitative differences in the VDR may alter intestinal sensitivity to vitamin D metabolites. We found 14% lower fractional45Ca absorption in African-American women with the BB VDR genotype compared with the bb genotype, although this difference did not achieve statistical significance (p = 0.08). This observation is consistent with the findings of Dawson-Hughes et al.12 who found reduced calcium absorption in postmenopausal Caucasian women with the BB VDR genotype on a low dietary calcium intake. We are unaware of other studies relating calcium absorption to polymorphisms in the VDR gene, but these results raise the possibility of a variant in the intestinal VDR that may alter 1,25(OH)2D-mediated calcium transport.
The results of at least one study suggest that there may also be differing sensitivities between VDR genotypes to 1,25(OH)2D-mediated bone remodeling.13 However, we were unable to demonstrate differences in bone formation or resorption markers between VDR genotypes in either our total sample or in our younger and older subgroups. Our inability to link bone turnover markers to VDR genotype is consistent with the results of some studies in Caucasians,15,16 but contrasts with other reports demonstrating a relation between VDR genotype and the bone formation markers, osteocalcin,14,31 and bone-specific alkaline phosphatase.31 The present study may have had limited statistical power to detect a small difference in osteocalcin and N-telopeptides across VDR genotypes given the wide variation in these markers. It is also possible that the use of other indices of bone turnover or measurement of the change in bone turnover markers during the follow-up would have revealed a stronger association with VDR genotype.
The frequency of the BB genotype in our population of African-Americans in Pittsburgh (13%) was higher than the frequencies in African-Americans reported by Fleet et al. (8%)5 and Hustmyer et al. (4%).24 These genotype frequencies are generally lower, however, than those in Caucasian populations in the U.S. (16%),7 U.K. (21%),39 France (17%),15 Australia (17%),14,31 Greece and Italy (33%),34 and the Netherlands (18%).37 The BB genotype frequencies in African-Americans also differ from those in the Japanese population (1.4%).31 Thus, there appears to be marked racial and ethnic differences in BsmI VDR genotype frequencies.
In conclusion, polymorphisms in the VDR gene defined by the BsmI, ApaI, and TaqI endonucleases are not associated with BMD or bone turnover in our population of African-American women. However, variation in the VDR gene or a nearby locus may influence intestinal calcium transport and the rate of postmenopausal bone loss in older African-American women.
This research was supported in part by the United States Public Health Science grant AR35582 and a grant from the Office of Research on Women's Health.