| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Special Features |
Endocrine Unit (J.S.F, R.M.N.), Department of Medicine, Massachusetts General Hospital, Boston, Massachusetts 02114; Department of Epidemiology (M.F.S.), University of Michigan, Ann Arbor, Michigan 48109; Channing Laboratory (M.-L.T.L.), Brigham and Women’s Hospital, Boston, Massachusetts 02115; Division of Research (B.E.), Kaiser Permanente Medical Care Program, Oakland, California 94611; Department of Epidemiology (J.A.C.), Graduate School of Public Health, University of Pittsburgh, Pittsburgh, Pennsylvania 15213; Department of Health Research and Policy (J.L.K.), School of Medicine, Stanford University School of Medicine, Stanford, California 94305; Division of Geriatrics (G.A.G., M.-H.H.), University of California Los Angeles School of Medicine, Los Angeles, California 90095
Address all correspondence and requests for reprints to: Joel S. Finkelstein, M.D., Endocrine Unit, Bulfinch 327, Massachusetts General Hospital, 55 Fruit Street, Boston, Massachusetts 02114. E-mail: . jfinkelstein{at}partners.org
Abstract
Bone mineral density (BMD) and fracture rates vary among women of differing ethnicities. Most reports suggest that BMD is highest in African-Americans, lowest in Asians, and intermediate in Caucasians, yet Asians have lower fracture rates than Caucasians. To assess the contributions of anthropometric and lifestyle characteristics to ethnic differences in BMD, we assessed lumbar spine and femoral neck BMD by dual-energy x-ray absorptiometry in 2277 (for the lumbar spine) and 2330 (for the femoral neck) premenopausal or early perimenopausal women (mean age, 46.2 yr) participating in the Study of Women’s Health Across the Nation. Forty-nine percent of the women were Caucasian, 28% were African-American, 12% were Japanese, and 11% were Chinese. BMDs were compared among ethnic groups before and after adjustment for covariates. Before adjustment, lumbar spine and femoral neck BMDs were highest in African-American women, next highest in Caucasian women, and lowest in Chinese and Japanese women. Unadjusted lumbar spine and femoral neck BMDs were 7–12% and 14–24% higher, respectively, in African-American women than in Caucasians, Japanese, or Chinese women. After adjustment, lumbar spine and femoral neck BMD remained highest in African-American women, and there were no significant differences between the remaining groups. When BMD was assessed in a subset of women weighing less than 70 kg and then adjusted for covariates, lumbar spine BMD became similar in African-American, Chinese, and Japanese women and was lowest in Caucasian women. Adjustment for bone size increased values for Chinese women to levels equal to or above those of Caucasian and Japanese women. Among women of comparable weights, there are no differences in lumbar spine BMD among African-American, Chinese, and Japanese women, all of whom have higher BMDs than Caucasians. Femoral neck BMD is highest in African-Americans and similar in Chinese, Japanese, and Caucasians. These findings may explain why Caucasian women have higher fracture rates than African-Americans and Asians.
BONE MINERAL DENSITY (BMD) and fracture incidence vary widely across racial and ethnic groups (1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11). African-Americans are reported to have higher BMD than Caucasians in the axial and appendicular skeleton, and these differences may account for their lower incidence of osteoporotic fractures (1, 3, 4, 5, 12). Asian women are reported to have lower BMD than Caucasian women (8, 13, 14, 15), yet they have lower hip fracture rates, an observation that remains poorly understood (6, 7). Although some ethnic differences in BMD may be due to ethnic differences in bone size (10, 11, 16) that create a scale artifact when BMD is measured by two-dimensional absorptiometry (17, 18), important racial and ethnic differences in BMD remain after adjustment for such artifacts (10, 11, 19, 20). Moreover, ethnic differences in vertebral BMD have been found when BMD is measured by quantitative computed tomography, a technique that measures true volumetric BMD and is not affected by bone size (5, 9, 21).
Although many studies have examined ethnic variation in BMD, these studies have important limitations. First, prior studies of ethnic variation in BMD have included modest numbers of subjects, limiting the ability to detect differences and assess the importance of covariates. Second, many prior studies have used convenience samples that may not reflect the true population. Third, most prior studies have been confined to comparisons between two ethnic groups. No study has simultaneously compared BMD among community-based samples of African-American, Asian, and Caucasian women. Finally, most studies have examined ethnic differences in BMD in postmenopausal women, in whom the results may be complicated by differences in rates of postmenopausal bone loss.
The Study of Women’s Health Across the Nation (SWAN) is a large, multiethnic, community-based study investigating a wide range of characteristics as women transition through the menopause (22). At the time of enrollment, all women were either premenopausal or early perimenopausal so their BMDs should be at or near their adult peak levels. In SWAN, BMD has been assessed at multiple skeletal sites in over 2000 women of either African-American, Caucasian, Chinese, or Japanese descent. Thus, the SWAN cohort provides a good opportunity to examine ethnic variation in BMD and factors that are related to those differences.
Subjects and Methods
Study population
SWAN is a multisite, longitudinal cohort study being conducted in community-based groups of women. At baseline, 3302 women who belonged to one of five ethnic/racial groups were recruited: Caucasian (n = 1550), African-American (n = 935), Japanese (n = 281), Chinese (n = 250), and Hispanic (n = 286). Eligibility criteria for entry into the SWAN longitudinal cohort were: age, 42–52 yr; intact uterus and at least one ovary; no current use of estrogens or other medications known to affect ovarian function; at least one menstrual period in the 3 months before screening; and self-identification as a member of one of the five eligible ethnic groups. Cohort recruitment and enrollment have been described in detail (22). In brief, participants were enrolled at seven clinical sites in the following geographic areas: Boston, Massachusetts; Chicago, Illinois; Detroit, Michigan; Los Angeles, California; Hudson County, New Jersey; Oakland, California; and Pittsburgh, Pennsylvania. Recruitment techniques were designed to generate a representative sample of women at each of the seven sites. All seven sites enrolled Caucasians, and each site also enrolled women belonging to one prespecified minority ethnic group. African-American women were enrolled in Boston, Chicago, the Detroit area, and Pittsburgh, whereas Japanese, Chinese, and Hispanic women were enrolled in Los Angeles, Oakland, and Hudson County, respectively. The Chicago and Hudson County sites did not measure BMD, leaving a potential maximum of 2413 participants belonging to 4 ethnic groups for BMD analyses. Of these women, 82 did not have a baseline lumbar spine BMD measurement, and 54 women had technically unsatisfactory baseline spine BMD scans (31 with fewer than 4 usable vertebrae due to focal sclerosis or other anatomic problems; 11 with metal, calcium pills, or bone grafts overlying the region of interest; and 12 in which the computer could not identify the bone edges), leaving 2277 women with potentially usable lumbar spine BMD data. A baseline hip BMD measurement was lacking in 80 women, and 3 women had technically unsatisfactory baseline hip BMD scans (due to metal overlying the region of interest), leaving 2330 women with potentially usable hip BMD data.
We also excluded women with technically usable scans but self-reported histories of anorexia nervosa, bulemia, hypercalcemia, chronic liver disease, glucocorticoid use of at least 2 months duration, tamoxifen use, lithium use, anticonvulsant use, or Depo-Provera use because these conditions are known to affect BMD. A total of 146 women (6.4%) were excluded from the spine BMD analyses, leaving 2131 women in the spine cohort belonging to the following ethnic groups: African-American (n = 591), Caucasian (n = 1051), Japanese (n = 258), and Chinese (n = 232). A total of 153 women (6.6%) were excluded from the hip BMD analyses, leaving 2177 in the hip cohort women belonging to the following ethnic groups: African-American (n = 608), Caucasian (n = 1076), Japanese (n = 257), and Chinese (n = 235). Finally, 203 women in the spine cohort and 208 women in the hip cohort were missing at least one model covariate, leaving 1928 women in the spine sample and 1969 women in the hip sample for multivariable analyses.
Study protocol
All subjects were seen at a SWAN center for measurements of BMD of the lumbar spine and proximal femur; assessment of factors possibly related to BMD using questionnaires that were administered by a SWAN-certified interviewer; and measurement of height and weight. Standardized interviewer-administered or self-administered questionnaires were used to assess the following parameters: age (years), cigarette smoking (number of pack-years), alcohol intake (grams per day) (23), calcium intake (milligrams per day) (23), physical activity (a summary score of active living, home, and recreational physical activity estimated by the modified Baecke instrument) (24, 25), age at menarche (years), number of prior pregnancies, educational level, self-reported history of thyroid disease (overactive, underactive, or thyroidectomy), duration of prior amenorrhea (years), menopausal status (premenopausal or early perimenopausal), oral contraceptive pill use (years), diuretic use, and anticoagulant use. Women were classified as premenopausal if they reported monthly menstrual bleeding during each of the 3 months preceding study entry and they were classified as perimenopausal if they experienced menstrual bleeding during at least 1 of the 3 months preceding study entry but had some change in the regularity of their menstrual cycle. Weight was measured without shoes in light indoor clothing, using a calibrated digital or balance beam scale. Height was measured using a calibrated stadiometer. The protocol was approved by the Institutional Review Board at each center, and all women provided written informed consent.
Assessment of BMD
BMD of the posterior-anterior lumbar spine and femoral neck was measured by dual energy x-ray absorptiometry (DXA) using a Hologic QDR 2000 densitometer (Hologic, Inc., Waltham MA) in Pittsburgh and Oakland or a Hologic QDR 4500A densitometer in Boston, the Detroit area, and Los Angeles. Osteodyne positioners were used to position subjects for measurements of the proximal femur (26).
All five centers employed a standard quality control program that involved training, certification, and recertification of DXA operators by Synarc, Inc., (Waltham, MA); daily measurement of a Hologic anthropomorphic spine phantom at each site; cross-site calibration with a single anthropomorphic spine phantom; and documented visual review of every scan image by a local site investigator experienced in bone densitometry. Measurements of the local spine phantoms and the circulating spine phantom were analyzed by Synarc and used by the study’s coordinating center to adjust DXA measurements for minor temporal or geographic variations in densitometer performance. Each of the five centers performed lumbar spine and proximal femur measurements in duplicate, with complete repositioning, on five women aged 42–52 yr (i.e. total of 25 duplicate measurements). Based on these data, our short-term in vivo measurement SD values are 0.014 g/cm2 (1.4%) and 0.016 g/cm2 (2.2%) for the lumbar spine and femoral neck. Scans were reviewed for prespecified quality control criteria to ensure that the region of interest was defined properly and that scans with problems that might affect BMD (e.g. areas of asymmetric sclerosis, fracture, severe scoliosis, overlying metal, overlying calcium pills, motion artifacts, poorly defined bone edges, etc.) were identified. All such scans were reanalyzed by Synarc who recommended whether the scan should be accepted as is, reanalyzed, or rejected. In addition, scans with outlier results and 5% of all scans (selected at random by the study coordinating center) were reviewed by Synarc.
Computation of bone mineral apparent density (BMAD)
Conventional BMD measurements by DXA are expressed in grams per square centimeter and are more properly referred to as areal BMD than true BMD, which would have units of grams per cubic centimeter. Areal BMD introduces a scale artifact that causes smaller bones to have lower areal BMD than larger bones. This scale artifact is reduced by expressing BMD as BMAD, a calculated variable that helps account for the size of the bones. For the lumbar spine, BMAD was computed as the BMC/(area)3/2 using the method of Carter et al. (17). For the femoral neck, BMAD was computed as the BMC/(area)2 using the method of Katzman et al. (18).
Data analysis
Preliminary analyses were conducted to determine whether the body thickness affected the relationship between ethnicity and BMD. With Hologic densitometers, body thickness is estimated using d0 values, and results may be unreliable if d0 falls below certain thresholds on any individual scan. Thus, all analyses were conducted by first including all women (after the exclusions noted above) and then after excluding women whose d0 value on the posterior-anterior spine scan was below the manufacturer’s recommended value (i.e. 100 for the QDR 2000 and 39 for the QDR 4500). Because inclusion of women with low d0 values did not affect the comparisons of bone density among ethnic groups and had negligible effects on the ß coefficients of the variables in the regression models, we report data for all eligible women. Preliminary analyses were also conducted to determine if there were differences in lumbar spine or femoral neck BMD between premenopausal and early perimenopausal women. Because the mean values were nearly identical, the results of premenopausal and early perimenopausal women were combined.
Spearman correlation coefficients of bone densities and each potential covariate were computed separately. Covariates with significant (P < 0.05) univariate associations with lumbar spine BMD were included in the multivariable models. These variables included body weight, age at menarche, years of oral contraceptive pill use, physical activity, number of prior pregnancies, educational level, and total calcium intake. In addition, cigarette smoking, alcohol intake, and age were included in the multivariable models because many prior studies have found that they are related to BMD. Body mass index and height were not included in the multivariable models because of their colinearity with body weight and because their univariate correlation with BMD was less strong than that of body weight with BMD.
Initial analyses suggested a significant interaction between weight and ethnicity for lumbar spine BMD and femoral neck BMAD. This interaction was due to five Asian women with extremely high weights who were not typical of the cohort. Regression diagnostics and formal outlier analysis demonstrated that the BMDs of these women were exerting extreme influence. Thus, the data from these women were excluded from the final multivariable analyses, and interaction terms were not used.
For each skeletal site, BMDs were first compared among the four ethnic groups by ANOVA without adjustment for covariates (Figs. 1–4
, top panels). To determine whether observed ethnic differences in BMD were due to ethnic variation in factors that affect BMD, ethnic differences in BMDs were then recompared after adjustment for the covariates listed above using multivariable regression analyses (PROC GLM procedure in SAS) (Figs. 1–4, bottom panels). The Tukey-Welsch adjustments (option REGWQ in the MEANS statement) were used for multiple comparisons among ethnic groups. Adjusted group means (least square means) were calculated after adjusting for the above covariates. Because the relationship between body weight and BMD was approximately linear (Figs. 1–4, middle panels), the adjusted group mean values were fit to straight lines (Figs. 1–4, bottom panels). Because body weight was a critical determinant of BMD (see Results) and there were very few Chinese or Japanese women with weights above 70 kg, all analyses were repeated on a subset of women with weights less than 70 kg. A cutoff point of 70 kg was chosen because there were large numbers of women from each of the four ethnic groups in this weight range. All data are expressed as the mean ± SD unless otherwise indicated. P values less than 0.05 are considered statistically significant, and all P values less than 0.1 are shown.
|
|
|
|
Clinical demographic characteristics
Table 1 shows baseline characteristics of the spine BMD cohort as a whole and for each of the four ethnic groups. The mean age of the cohort was 46.3 ± 2.7 yr and there were small differences in age among groups. Fifty-five percent of women were classified as premenopausal, and 45% were classified as early perimenopausal. On average, African-American women were 9 kg heavier than Caucasian women and 25–27 kg heavier than Chinese and Japanese women. On average, African-American and Caucasian women were 6–7 cm taller than Chinese and Japanese women. Chinese and Japanese women were least likely to be perimenopausal. Mean calcium intake was highest in Caucasian women. African-American women had the most prior pregnancies and were more likely to be current smokers. Chinese women had the latest age at menarche and were the least likely to smoke cigarettes or drink alcohol.
|
). Many African-American women weighed more than 90 kg. In contrast, there were few Japanese or Chinese women who weighed more than 70 kg.
|
Mean lumbar spine areas were 57.2 ± 5.5, 58.7 ± 5.3, 56.3 ± 5.1, and 58.2 ± 4.4 cm2 in African-American, Caucasian, Chinese, and Japanese women, respectively (P < 0.001 for Chinese vs. Caucasian, Chinese vs. Japanese, and African-American vs. Caucasian; P = 0.016 for Chinese vs. African-American; P = 0.009 for African-American vs. Japanese; P = 0.214 for Japanese vs. Caucasian). Mean femoral neck areas were 4.7 ± 0.4, 4.9 ± 0.4, 4.5 ± 0.4, and 4.8 ± 0.4 cm2 in African-American, Caucasian, Chinese, and Japanese women (P < 0.001 for all comparisons except P = 0.385 for African-American vs. Japanese).
Lumbar spine BMD
Figure 1
shows the mean lumbar spine BMD of each of the four ethnic groups before and after adjustment for covariates. Before adjustment for covariates, lumbar spine BMD was highest in African-Americans, next highest in Caucasians, third highest in Chinese women, and lowest in Japanese women, and all groups differed significantly from one another (Fig. 1, top left). Unadjusted lumbar spine BMD was 7–12% higher in African-Americans than in the other groups and 2–5% higher in Caucasians than in Chinese or Japanese women (Fig. 1, top left). After adjustment for covariates, spine BMD was only 3–6% higher in the African-Americans than in the other groups (P < 0.001, P = 0.084, and P = 0.031 for comparisons with Caucasians, Chinese, and Japanese, respectively) and was 2–3% lower in Caucasians than in Chinese (P = 0.066) and/or Japanese women (Fig. 1, bottom left). Adjusted spine BMD did not differ between Japanese and Chinese women. In addition to ethnicity, body weight (P < 0.001) and physical activity (P = 0.010) were positively associated, and age at menarche (P < 0.001) was negatively associated with spine BMD in the multivariable model. Body weight explained 21% of the variability in spine BMD.
Visual inspection of the data suggested that spine BMD was similar in African-American, Chinese, and Japanese women and lowest in Caucasian women who weighed less than 70 kg (Fig. 1, middle left), a subset of the cohort in which there were substantial numbers of women of each ethnic group (Table 2). Therefore, the data were reanalyzed on women with weights less than 70 kg (Fig. 1, right panels). Within this subset, unadjusted spine BMD was highest in African-American women, and there were no significant differences among the other groups (Fig. 1, top right). Even in this subset, however, African-American and Caucasian women tended to predominate in the upper portion of the restricted weight range, and Chinese and Japanese women tended to predominate in the lower portion of the restricted weight range (Table 2). When spine BMD of this subset was adjusted for covariates, spine BMD was similar in African-American, Chinese, and Japanese women and lowest in Caucasian women (P = 0.003, P = 0.008, and P = 0.072 for comparisons of Caucasians with African-Americans, Chinese, and Japanese, respectively; Fig. 1, bottom right).
Lumbar spine BMAD
Figure 2 shows the mean lumbar spine BMAD for each of the four ethnic groups before and after adjustment for covariates. Before adjustment for covariates, lumbar spine BMAD was highest in African-American women, similar in Caucasian women and Chinese women, and lowest in Japanese women (Fig. 2, top left). After adjustment for covariates, lumbar spine BMAD was still highest in the African-American women (P < 0.001 vs. Caucasians and Japanese and P = 0.013 vs. Chinese) and was higher in Chinese women than in Caucasian women (P = 0.021; Fig. 2, bottom left). Lumbar spine BMAD was similar in Caucasian and Japanese women and in Chinese and Japanese women after correction for covariates. In addition to ethnicity, body weight (P < 0.001) was positively associated and age at menarche (P < 0.001) was negatively associated with lumbar spine BMAD in the multivariable model. Body weight explained 21% of the variability in lumbar spine BMAD.
In women who weighed less than 70 kg, unadjusted spine BMAD was highest in the African-American women and was higher in Chinese women than in Caucasian or Japanese women (Fig. 2, top right). When spine BMAD of this subset was adjusted for covariates, spine BMAD was highest and similar in African-American and Chinese women, was higher in Chinese than in Caucasian women, and was lowest in Caucasian women (P < 0.001 for African-American vs. Caucasian, P = 0.024 for African American vs. Japanese, P = 0.008 for Chinese vs. Caucasian, other comparisons not significant; Fig. 2, bottom right).
Femoral neck BMD
Figure 3 shows the mean BMD of the femoral neck in each of the four ethnic groups before and after adjustment for covariates. Before adjustment for covariates, femoral neck BMD was highest in African-American women, next highest in Caucasian women, and lowest and similar in Japanese and Chinese women. Unadjusted femoral neck BMD was 14–24% higher in African-Americans than in the other groups and 7–9% higher in Caucasians than in Chinese or Japanese women (Fig. 3, top left). After adjustment for covariates, femoral neck BMD was only 6–9% higher in African-Americans than in Chinese, Japanese, or Caucasians (P < 0.001 for comparisons with all other groups) and was nearly identical in the other three groups (Fig. 3, bottom left). In addition to ethnicity, body weight (P < 0.001) and physical activity (P < 0.001) were positively associated, and age at menarche (P = 0.026) was negatively associated with femoral neck BMD in the multivariable model. Body weight explained 37% of the variability in femoral neck BMD.
In women who weighed less than 70 kg, unadjusted femoral neck BMD was highest in the African-American women and was higher in Caucasian women than in Chinese or Japanese women (Fig. 3, top right). The magnitude of the differences between groups was reduced, however. In this subset, femoral neck BMD was only 8–9% higher in African-Americans than in the other groups and only 2–3% higher in Caucasians than in Chinese or Japanese women (Fig. 3, top right). When femoral neck BMD of this subset was adjusted for covariates, femoral neck BMD remained highest in African-American women (P 
0.001 vs. Caucasians and Japanese; P = 0.071 vs. Chinese) and was nearly identical in the other three groups (Fig. 3, bottom right).
Femoral neck BMAD
Figure 4 shows the mean femoral neck BMAD of each of the four ethnic groups before and after adjustment for covariates. Like spine BMAD, femoral neck BMAD was highest in African-American women, next highest in Caucasian and Chinese women, and lowest in Japanese women before adjustment for covariates (Fig. 4, top left). After adjustment for covariates, femoral neck BMAD remained highest in African-American women (P 0.001 vs. other groups) and was now significantly higher in Chinese than in Caucasian women (P = 0.015; Fig. 4, bottom left). There were no other significant differences in BMAD between groups after adjustment for covariates. In addition to ethnicity, body weight (P < 0.001) and physical activity (P = 0.012) were positively associated, and age at menarche (P < 0.001) and educational level (P = 0.009) were negatively associated with spine BMAD in the multivariable model. Body weight explained 20% of the variability in femoral neck BMAD.
As with spine BMAD, unadjusted femoral neck BMAD remained highest in the African-American women but was now higher in Chinese women than in Caucasian or Japanese women who weighed less than 70 kg (Fig. 4, top right). When femoral neck BMAD of this subset was adjusted for covariates, femoral neck BMAD remained highest in African-American women (P < 0.001 vs. Caucasians and Japanese, P = 0.027 vs. Chinese), was borderline higher in Chinese than in Caucasian women (P = 0.075), and did not differ between Caucasian and Japanese or Chinese and Japanese women (Fig. 4, bottom right).
Discussion
These data demonstrate that there are significant differences in BMD of the lumbar spine and femoral neck among late premenopausal and early perimenopausal African-American, Caucasian, Chinese, and Japanese women. In unadjusted analyses, lumbar spine and femoral neck BMD are highest in African-American women, next highest in Caucasian women, and lowest in Chinese and Japanese women. Ethnic patterns of BMD change, however, when the effects of selected anthropometric and lifestyle variables (notably body weight) on BMD are considered. When analyses are restricted to women who weigh less than 70 kg, lumbar spine BMD is similar in African-American, Chinese, and Japanese women and lowest in Caucasians, a finding that differs from conventional dogma. At the femoral neck, adjusted BMD is highest in African-American women and similar in the other three groups. Differences in lumbar spine and femoral neck BMD between Caucasian and Chinese women are eliminated when values are corrected for the scale artifact inherent in areal BMD measurements. In fact, when BMD values are corrected for the scale artifact of DXA and then adjusted for differences in body weight and other covariates, Chinese women actually have higher BMD than Caucasians. This finding also differs from prior reports.
Several prospective observational studies have reported that BMD is a major predictor of fracture risk in postmenopausal women (27). In general, for every SD that BMD is reduced, the risk of fragility fractures increases by 50–100% (27). In our subjects, unadjusted lumbar spine and femoral neck BMDs were 1–2 SD values higher in African-American women than in Caucasian, Japanese, or Chinese women. These differences in BMD are sufficiently large to account for the well known lower fracture risk of African-American women although other factors, such as their higher body mass index, may also contribute. Most of these differences are due to differences in body weight rather than to the effects of ethnicity per se.
Previous studies have reported that BMD at multiple skeletal sites is higher in pre-, peri-, and postmenopausal African-American women than in Caucasian women (3, 4, 5, 19, 28, 29, 30). Most of these studies, however, did not account for possible effects of ethnic differences in bone size. Although African-American women are heavier than Caucasian women, their lumbar spine and femoral neck bone areas are actually smaller. This difference may be due, in part, to magnification effects related to differences in subject thickness. Thus, it is not surprising that correcting for differences in bone size fails to eliminate BMD differences between African-American and Caucasian women. Previous studies have also reported that differences in anthropometric and lifestyle variables, particularly body weight or body mass index, account for part of the difference in BMD between African-American and Caucasian women (5, 19, 28). For example, Ettinger et al. (19) found that adjustment for a series of anthropometric, lifestyle, and biochemical variables reduced the difference in BMD between black and white women by 34%. Cauley et al. (30) found that BMD remained higher in postmenopausal black women even after adjusting for age, obesity, serum estrogen levels, physical activity, smoking, alcohol consumption, and medication use. Our data agree that adjustment for anthropometric and lifestyle factors, particularly body weight, reduces the magnitude of the difference in BMD between African-Americans and Caucasians. Moreover, our data demonstrate that when weight is made comparable by restriction, spine BMD is similar in African-American and Chinese and Japanese women, although caution is needed when interpreting analyses based on subsets of subjects. This finding contrasts with existing data that suggest that lumbar spine and femoral neck BMD are higher in African-American than in Asian women (31, 32). Differences in BMD between African-American women and other ethnic groups that remain after correction for anthropometric and lifestyle factors may reflect true genetic differences in BMD.
Several studies have compared BMD of adult Chinese or Japanese women with women of other ethnic groups. If BMD is not corrected for bone size or body size, BMD at multiple skeletal sites is clearly higher in Caucasian women than in Chinese or Japanese women (8, 9, 10, 13, 14, 31, 33). Our data confirm these findings. Nonetheless, although some reports suggest that fracture rates in Asian women from urban areas are approaching those reported in Caucasians from the United States (34, 35), most studies report that hip fracture rates are lower in Asians than in Caucasians (6, 7), an observation that seems inconsistent with their lower BMD. To some extent, this paradox may be due to differences in hip geometry between Asian and Caucasian women (33), although no differences in hip axis length between Japanese and Caucasian were detected in the SWAN cohort (36). Ethnic differences in body mass could also contribute to differences in fracture rates, although the lower body mass index of Asian women should increase their risk of hip fracture compared with Caucasian women rather than decrease it. It is also possible that hip fracture rates are lower in Asians than in Caucasians because Asian women fall less frequently (37). A more plausible explanation for the apparent paradoxical relationship between BMD and fracture rates in Asian and Caucasian women is that BMD values have been misinterpreted. When corrections are made for body size, most authors report that total body, femoral neck, and posterior-anterior lumbar spine BMD (or bone mineral content) are similar in Caucasian and Asian women (8, 10, 11, 13, 14, 33). Our data confirm and extend these findings. When lumbar spine BMD values of the entire cohort were adjusted for covariates, BMD was similar in Chinese, Japanese, and Caucasian women. When similar adjustments were performed in women who weighed less than 70 kg, the portion of the cohort that included most of the Asian women, lumbar spine BMD was actually greater in Chinese and Japanese women than in Caucasian women. Moreover, because Chinese women have smaller bones than Caucasians, differences in lumbar spine and femoral neck BMD between Caucasian and Chinese women were eliminated when BMD was expressed as BMAD. When the spine and femoral neck BMAD values were also adjusted for differences in weight, Chinese women again had significantly higher values than Caucasians. Thus, Asians may have lower fracture rates than Caucasians simply because they have higher BMD when one adjusts for the effects of bone size and body weight.
The rank order of spine BMD among ethnic groups changed when the data were adjusted for covariates and changed further when the cohort was restricted to a weight range in which there was considerable overlap of the four ethnic groups. Whereas BMD of Caucasians ranked second at all skeletal sites in unadjusted analyses, Caucasians consistently had the lowest adjusted BMDs. Spine BMD was no longer higher in African-Americans than in Asians when values were adjusted within the weight-restricted subset of women. The changes in rank ordering of BMD when weight is considered in these ways highlight the importance of having a cohort of sufficient size to account for the complex effects of weight or other covariates on BMD in women of multiple ethnicities.
There are differences in BMD between ethnic groups that remain after adjustment for bone size and various anthropometric and lifestyle variables. There are many plausible factors that may contribute to these differences. African-Americans have lower 25-hydroxyvitamin D concentrations and equivalent or higher PTH levels than Caucasians (5, 19, 38, 39, 40, 41), although these differences do not appear to account for differences in BMD (5, 19, 38, 39, 40, 41). Two studies suggest that the skeletons of African-Americans are less sensitive to the resorptive effects of PTH (39, 40), a finding that may account for their lower rates of bone turnover (5, 42) and higher BMD. Finally, racial/ethnic variation in genes related either to the development of peak bone mass or to the timing or rate of adult bone loss is a plausible cause of racial/ethnic differences in bone mass. Few studies have examined this issue, however. The distribution of polymorphisms in the vitamin D receptor gene identified by the presence or absence of a BsmI restriction enzyme site is similar in African-American and Caucasian women as is the effect of specific vitamin D receptor genotypes on BMD (43). Further studies, particularly of genes that are clearly related to BMD or body size, are needed to determine the biological basis for the racial/ethnic variation in BMD that is not accounted for by variables in our analysis.
The mean femoral neck BMD of our Caucasian women is about 4% higher (0.826 g/cm2 vs. 0.796 g/cm2) than that of Caucasian women aged 40 to 49 who had similar measurements as part of the Third National Health and Nutrition Examination Survey (NHANES III) (44). This difference may be due to the fact that all SWAN participants were either premenopausal or early perimenopausal, whereas women of any menopausal status, including postmenopausal women, were included in NHANES III. By comparison with 20- to 29-yr-old Caucasian women surveyed in NHANES III (45), only 3 of 1076 Caucasian women, 2 of 235 Chinese women, 1 of 258 Japanese women, and 0 of 608 African-American women in our cohort would be classified as osteopenic, and no women would be classified as osteoporotic at the femoral neck using the nomenclature adopted by the World Health Organization (46).
Certain limitations of our study deserve mention. First, although our sample is community-based, it may not be representative of the entire U.S. population (22). Moreover, we have no data on Hispanic women or women of other minority groups. Second, our analysis does not include all factors that may affect BMD and may vary by ethnicity. For example, our only dietary variables are calcium and alcohol intake, and we have not included hormonal measures such as vitamin D, PTH, gonadal steroids, or IGF-I in these analyses. We feel that this approach is reasonable, however, because our goal was to determine whether ethnic differences in BMD truly exist among premenopausal and early perimenopausal women of various racial and ethnic groups and the extent to which lifestyle and anthropometric variables contribute to these differences.
In summary, we found significant differences in lumbar spine and femoral neck BMD in premenopausal and early perimenopausal women among women of various racial/ethnic groups. These data provide new insights into ethnic variation in BMD. Contrary to conventional wisdom, ethnicity per se is not associated with differences in lumbar spine BMD among African-American, Chinese, and Japanese women, and all three groups have higher BMDs than Caucasians of comparable weights. African-American ethnicity is associated with higher femoral neck BMD than Caucasians, Chinese, or Japanese, all of whom have similar femoral neck BMD. These findings provide a plausible biological basis for the known ethnic differences in fracture rates. Ethnic differences in BMD that persist after adjustment for bone size and other factors may be due to ethnic variation in genes related to BMD.
Acknowledgments
We acknowledge Dr. Christopher Gallagher for his leadership in the development of the study; Dr. Gordon Fitzgerald and Beth Willis for helping with data management and data analysis; Dr. Sybil Crawford for her advice on data analysis; Sarah Zhang, Robbin Cleary, Priscilla Marbury, Jean Young, Anna McDivit, Martha Kisshauer, and Cindy Granny for performing the bone density tests; the project directors and research assistants at the participating centers for their dedicated work on the protocol; and all of the SWAN participants for the generous donation of their time. The project officers were Taylor Harden, Carole Hudgings, Marcia Ory, and Sherry Sherman, and the steering committee chair was Jennifer L. Kelsey. This manuscript was reviewed by the Publications and Presentations Committee of SWAN and has its endorsement.
Footnotes
SWAN was funded by the National Institute on Aging, the National Institute of Nursing Research, and the Office of Research on Women’s Health of the National Institutes of Health. Supplemental funding from the National Institute of Mental Health, the National Institute on Child Health and Human Development, the National Center on Complementary and Alternative Medicine, the Office of Minority Health, and the Office of AIDS Research is also gratefully acknowledged.
The coordinating center for this study was New England Research Institutes, Watertown, Massachusetts [Grant U01 AG12553, Sonja McKinlay, Principal Investigator (PI)]. Clinical centers included: University of Michigan, Ann Arbor, Michigan (Grant U01 NR04061, MaryFran Sowers, PI); Massachusetts General Hospital, Boston, Massachusetts (Grant UO1 AG12531, Robert M. Neer, PI 1995–1999, and Joel S. Finkelstein, PI 1999 to present); University of California/Kaiser, Davis, California (Grant U01 AG12554, Ellen Gold, PI); University of California, Los Angeles, California (Grant U01 AG12539, Gail Greendale, PI); and the University of Pittsburgh, Pittsburgh, Pennsylvania (Grant U01 AG12546, Karen Matthews, PI). This work was also supported by NIH Grant K24 DK02759 to Dr. Finkelstein, the Iris Cantor–UCLA Women’s Health Center, a UCLA Center of Excellence in Women’s Health grant to Dr. Greendale (RFP 282-97-0025), and NIH Grant RR-1066.
Abbreviations: BMAD, Bone mineral apparent density; BMD, bone mineral density; DXA, dual-energy x-ray absorptiometry; NHANES III, Third National Health and Nutrition Examination Survey; SWAN, Study of Women’s Health Across the Nation.
Received October 29, 2001.
Accepted March 8, 2002.
References
This article has been cited by other articles:
 |
|
 |
|
  L.-J. W. Lu, F. Nayeem, K. E. Anderson, J. J. Grady, and M. Nagamani Lean Body Mass, Not Estrogen or Progesterone, Predicts Peak Bone Mineral Density in Premenopausal Women J. Nutr., February 1, 2009; 139(2): 250 - 256. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. P Reis, E. D Michos, D. von Muhlen, and E. R Miller III Differences in vitamin D status as a possible contributor to the racial disparity in peripheral arterial disease Am. J. Clinical Nutrition, December 1, 2008; 88(6): 1469 - 1477. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. D. Walker, R. Novotny, J. P. Bilezikian, and C. M. Weaver Race and Diet Interactions in the Acquisition, Maintenance, and Loss of Bone J. Nutr., June 1, 2008; 138(6): 1256S - 1260S. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. S. Finkelstein, S. E. Brockwell, V. Mehta, G. A. Greendale, M. R. Sowers, B. Ettinger, J. C. Lo, J. M. Johnston, J. A. Cauley, M. E. Danielson, et al. Bone Mineral Density Changes during the Menopause Transition in a Multiethnic Cohort of Women J. Clin. Endocrinol. Metab., March 1, 2008; 93(3): 861 - 868. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 E. D. Taylor, K. R. Theim, M. C. Mirch, S. Ghorbani, M. Tanofsky-Kraff, D. C. Adler-Wailes, S. Brady, J. C. Reynolds, K. A. Calis, and J. A. Yanovski Orthopedic complications of overweight in children and adolescents. Pediatrics, June 1, 2006; 117(6): 2167 - 2174. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. R. Sowers, M. Jannausch, D. McConnell, R. Little, G. A. Greendale, J. S. Finkelstein, R. M. Neer, J. Johnston, and B. Ettinger Hormone Predictors of Bone Mineral Density Changes during the Menopausal Transition J. Clin. Endocrinol. Metab., April 1, 2006; 91(4): 1261 - 1267. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. F. Aloia, S. A. Talwar, S. Pollack, and J. Yeh A Randomized Controlled Trial of Vitamin D3 Supplementation in African American Women Arch Intern Med, July 25, 2005; 165(14): 1618 - 1623. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 D. Nash, L. S. Magder, R. Sherwin, R. J. Rubin, and E. K. Silbergeld Bone Density-related Predictors of Blood Lead Level among Peri- and Postmenopausal Women in the United States: The Third National Health and Nutrition Examination Survey, 1988-1994 Am. J. Epidemiol., November 1, 2004; 160(9): 901 - 911. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. J. De Souza and N. I. Williams Physiological aspects and clinical sequelae of energy deficiency and hypoestrogenism in exercising women Hum. Reprod. Update, September 1, 2004; 10(5): 433 - 448. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. C. Looker The Skeleton, Race, and Ethnicity J. Clin. Endocrinol. Metab., July 1, 2002; 87(7): 3047 - 3050. [Full Text] [PDF]
|
|
|
|
|
|
J. S. Finkelstein, M. Sowers, G. A. Greendale, M.-L. T. Lee, R. M. Neer, J. A. Cauley, and B. Ettinger Ethnic Variation in Bone Turnover in Pre- and Early Perimenopausal Women: Effects of Anthropometric and Lifestyle Factors J. Clin. Endocrinol. Metab., July 1, 2002; 87(7): 3051 - 3056. [Abstract] [Full Text] [PDF]
|
|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Endocrinology | Endocrine Reviews | J. Clin. End. & Metab. |
| Molecular Endocrinology | Recent Prog. Horm. Res. | All Endocrine Journals |
