There are known black-white differences in bone density measured by DXA but less is known about bone architecture. We compared cross-sectional geometric properties of the proximal femur in U.S. black (n = 86) and white (n = 151) and South African black (n = 60) and white (n = 48) postmenopausal women. Results are consistent with greater bone strength in the black groups in both countries.
Introduction: There are well-known ethnic differences in bone density, but little is known about ethnic differences in bone architecture between U.S. and South African blacks and whites.
Materials and Methods: We compared bone density and cross-sectional geometric properties of the proximal femur in 237 U.S. black (n = 86) and white (n = 151) and 108 South African black (n = 60) and white (n = 48) postmenopausal women. The proximal femur (neck, intertrochanteric region, and proximal shaft regions of interest) was measured with DXA and further analyzed with a hip structural analysis program. For each region, BMD, cross-sectional area, outer diameter, section modulus, endosteal diameter, average cortical thickness, and the buckling ratio were estimated.
Results and Conclusions: In the femoral neck, in both countries, the blacks had narrower endosteal diameters (mean difference, 2.6% and 5.1% in U.S. and South African women, respectively), thicker cortices (9.3% and 11.0%), and a lower buckling ratio (11.6% and 15.2%) despite a similar outer diameter. In the intertrochanteric region, the whites had a greater outer diameter (2.2% and 3.0% in U.S. and South African women, respectively), lower cross-sectional area (4.8% and 7.2%), and a higher buckling ratio (7.6% and 3.6%). There are fewer differences in the shaft. Compared with South African whites, U.S. whites had wider (mean difference 2.9%) femoral necks and a greater section modulus (6.4%) in the shaft. U.S. whites also had greater cross-sectional area in both the neck and shaft (5.2% and 4.6%, respectively). The U.S. blacks had significantly greater outer diameters, cross-sectional areas, endosteal diameters, and section moduli in the neck region compared with South African blacks. Our observations are consistent with greater bone strength in the black groups in both countries, and they also suggest that there are fewer differences between the same ethnic groups in the two countries than there are between different ethnic groups within a country.
OSTEOPOROSIS IS DEFINED by abnormalities in both the amount of bone and the architectural arrangement of bone tissue, leading to decreased skeletal strength and increased fragility and fracture risk.(1) There are known differences in both bone mass and bone architecture in blacks compared with whites in the United States.(2-7) In numerous studies of bone density measured in several skeletal regions, U.S. blacks have a significantly higher bone mass on average than whites.(2-5) They also have a lower fracture rate.(8-11) In contrast, bone density data from black and white women in South Africa do not follow the general pattern observed among U.S. groups. Daniels et al.(12) reported that, in both pre- and postmenopausal black and white women in Johannesburg, there were no significant differences in bone mass in the spine or forearm, although blacks had a significantly higher hip bone density. The few data on hip fractures in South Africa indicate a dramatically lower rate of hip fracture in blacks compared with whites,(13) as is the case in the United States.
Less is known about bone architecture in these ethnic groups. Published data include histomorphometric analyses of iliac crest biopsy specimens(14-19) in whites and blacks in both the United States and South Africa, radiographic measurements of the hip,(7) and our previously published study of proximal femoral geometry in U.S. blacks and whites.(6) Using data obtained by DXA, our group reported significant differences in cross-sectional geometric properties of the proximal femur, such that black women have proportionately greater mechanical strength than white women.(6) An earlier histomorphometric study of black and white women in Detroit indicates that black women have thicker trabeculae and thicker cortices than whites.(14) Studies comparing South African blacks and whites also suggest thicker trabeculae in blacks, which, along with other histomorphometric features, may contribute to a lower fracture rate.(18, 19) Unlike U.S. blacks, turnover assessed by histomorphometry in South African blacks is increased compared with South African whites.
Given that bone mass and histomorphometric variables vary among ethnic groups and that patterns of ethnic differences also vary geographically, we directly compared bone density and geometric data from black and white women from Detroit in the United States and from Johannesburg in South Africa. We took advantage of existing data from DXA studies obtained independently in two separate investigations.(6, 12) Both studies included adequate numbers of postmenopausal women from both black and white ethnic groups and used similar DXA instruments. Both instruments were still available for cross-calibration, allowing more reliable comparison of DXA data. Our objective was to describe hip bone geometry and other bone variables in the two ethnic groups from Detroit and Johannesburg.
MATERIALS AND METHODS
To allow for a comparison of the South African groups with our data in U.S. women, we assessed cross-sectional hip geometry in the South African postmenopausal cohort studied by Daniels et al.(12) All subjects in both studies had a measurement of the proximal femur using a Hologic QDR bone densitometer, which theoretically allowed an appropriate comparative study of cross-sectional geometric properties derived from the Hologic data. Nonetheless, we first tested the calibration of the two instruments so that we could adjust the data from the two studies if necessary.
The two cohorts of women were recruited in metropolitan Detroit and Johannesburg for separate observational studies of bone density in healthy women. Thus, the study protocols were not the same, but all subjects were postmenopausal and not using hormone replacement therapy; they were in good general health and were ambulatory. The Johannesburg group was comprised of 60 black and 48 white nurses working in local hospitals.(12) The Detroit group of 235 white and 136 black women was recruited for an ancillary study of the Women's Health Initiative (WHI)(20) within 6 weeks of randomization to either the hormone replacement or dietary intervention trials. Over a period of 15 months, all subjects enrolled in these studies were invited to participate in our study; ∼95% of eligible women agreed to participate. The time period was proscribed by the initiation date of our ancillary study and the end date of the WHI.
The age ranges of the groups from the two countries differed, although there was significant overlap: the age range for the Detroit subjects was 50-79 years, whereas the Johannesburg subjects ranged from 45 to 64 years of age. Thus, we selected subjects in the Detroit study who were <65 years old to have more comparable groups based on age. Table 1 shows the final sample sizes of each of the four subgroups.
Table Table 1.. Group Characteristics, Mean (SD), and p Values for Statistical Significances Between U.S. White and U.S. Black (P1) and Between SA White and SA Black (P2) Groups
Data from DXA of the proximal femur in the two cohorts were used retrospectively for this comparison study. A Hip Structure Analysis (HSA) program developed by one of the authors (TJB) uses the distribution of mineral mass in a line of pixels across the bone axis to measure geometric properties of cross-sections in cut planes traversing the bone at that location. Current versions of the HSA program average measurements for five parallel mass profiles spaced ∼1 mm apart along the bone axis, effectively corresponding to a 5-mm section thickness. Analysis locations included the narrow-neck region across the narrowest point of the femoral neck, the intertrochanteric region traversing the bisector of the neck and shaft axes, and the shaft region located 1.5 times the neck width distal to the axes' intersection (Fig. 1). BMD, cross-sectional area of bone (CSA), outer diameter (subperiosteal width), and the cross-sectional moment of inertia (CSMI) were measured directly from mineral mass distributions using algorithms described previously.(21, 22) Section modulus (Z) was calculated as CSMI divided by the distance from the center of mass to the medial or lateral bone surface, whichever is greater.
Assumptions of cross-sectional shape are not required for the above measurements but are required for estimates of average cortical thickness and buckling ratio, an index of cortical instability.(23) Buckling ratios were calculated as the maximum distance from the center of mass to the outer cortical margin divided by the mean cortical thickness. As in previous studies, cortices of the narrow neck, intertrochanteric, and shaft regions were modeled as circular annuli with 60%, 70%, and 100% of the measured mass in the cortex, respectively.(21) The HSA program also reports neck shaft angle (NSA) and neck length. NSA is the obtuse angle between the neck and shaft axes, and neck length is the distance from the center of the femoral head to the intersection of the neck and shaft axes.
We assessed the precision of the hip structure measurements by scanning the proximal femur of three volunteers five times each, repositioning the subjects between scans. The error of the repeated measurements, expressed as the CV (%), ranged from 0.8% (intertrochanteric bone width) to 4.7% (neck cortical thickness), with an average of 2.2%.
To ensure the comparability of the two DXA data sets, we assessed the calibration difference between the DXA instruments in Detroit (Hologic QDR 1000 Plus) and Johannesburg (Hologic QDR 1000). A three-step calibration phantom obtained from Hologic (Bedford, MA, USA) was scanned on both instruments, 10 times each. Pixel values were generated for the block phantom and the mean value of mass was calculated for each step of the phantom. The South Africa values were regressed on the United States values to generate a linear calibration function. The regression formula predicting the Hologic 1000 Plus (United States) values from the Hologic 1000 (South Africa) readings indicate a strong association with a slope of nearly 1: y = 1.055x + 0.0196.
This function was used to adjust the BMD, CSA, and CSMI. Section modulus and cortical thickness estimates were computed from the corrected CSMI and CSA values, respectively, together with the measured bone widths.
Means and SDs were used as descriptive statistics. The ethnic differences in bone variables were examined by an analysis of covariance using ethnicity and country as the factor variables. Because of significant demographic differences (Table 1), age, height, and body weight were used as covariates. The mean relative differences between the study groups and 95% CIs were determined through log-transformation of the variables. When the 95% CI did not include zero, the difference was regarded statistically significant at α = 0.05. Statistical analyses were done with SPSS for Windows version 10.0 (SPSS, Chicago, IL, USA).
Descriptive statistics of the demographic characteristics of the four groups are provided in Table 1. The U.S. women were on average 4-5 years older than the South African women. The South African black women were significantly shorter than the other three groups, and the South African white women had a significantly lower mean weight than the other groups. These differences are reflected in the body mass index (BMI) for the groups, with South African blacks having the highest BMI (34.1 kg/m2) and South African whites having the lowest (25.9 kg/m2).
2D values derived from the DXA data are also shown in Table 1. The NSA is widest in the South African white group and similar in the other three groups. The femoral neck length is shortest in the South African black group, differing significantly from the other three groups. However, when femoral neck length is adjusted for height, the difference is no longer significant (p = 0.42, data not shown).
Table 2 shows the age- and body-size adjusted results for the other variables derived from the HSA, including the BMD derived from the neck, intertrochanteric, and shaft regions (and the conventional region of the femoral neck defined by the scanner manufacturer). There are significant differences between ethnic groups within the same country, as well as between the two countries. However, there was no interaction between ethnicity and country, and therefore, the between-group differences are described separately for ethnicity and country in Figs. 2A-2C.
Table Table 2.. Age, Height, and Weight-Adjusted Values (SE), and p Values for Statistical Significances Between U.S. White and U.S. Black (P1) and Between SA White and SA Black (P2) Groups
Ethnic differences within countries
There were some, albeit small, ethnic differences between the two countries (Table 2). In the neck region, blacks had a higher BMD in both countries: the mean difference (95% CI) was 8.6% (5.1-12.0%) between the U.S. groups and 9.9% (4.3-15.2%) between the South African groups. In the intertrochanteric region, the respective mean differences were 6.9% (3.0-10.6%) and 9.9% (3.8-15.6%). In the femoral shaft region, the ethnic differences in BMD were smaller and statistically significant only between the U.S. groups.
Bone CSA is greater in blacks from both countries in the neck and intertrochanteric regions, but is greater in the shaft region only in U.S. blacks. Bone outer diameter is statistically significantly different in the intertrochanteric region only, where whites have wider diameters in both countries. Section moduli in the neck and shaft regions are greater in U.S. blacks compared with whites, but there were no significant ethnic differences in the South African data. The endosteal diameter is significantly smaller, and the average cortical thickness significantly greater in blacks in both countries in the femoral neck and intertrochanteric regions, whereas the average cortical thickness is greater in the shaft region only among the U.S. blacks. The buckling ratio is significantly higher in whites in both countries in the neck and intertrochanteric regions but not in the shaft.
Ethnic differences between countries
Comparisons between the two countries show fewer differences compared with within-country ethnic differences (Figs. 2A-2C). Generally, the South African groups had lower CSA and section moduli and narrower outer diameters than Americans in the neck and shaft, with no differences in the intertrochanteric region. For example, there were no statistically significant BMD differences between the whites in the two countries or the black groups in the two countries. However, both whites and blacks in the two countries differed with respect to outer diameter of the femoral neck and the section modulus in the femoral shaft (greater in the United States). Among the black groups, in the femoral neck, the U.S. blacks had significantly greater bone CSAs, bone outer and endosteal diameters, and section moduli. These differences range from 4.5% (1.8-7.0%) in outer diameter to 7.5% (2.1-12.6%) in the section modulus. In the femoral shaft, the only significant difference between the two white groups is a higher section modulus in the U.S. whites (6.4%; 95% CI, 1.6-11.3%). The U.S. blacks also had significantly greater section moduli than South African blacks, as well as greater cross-sectional areas and outer diameters, ranging from 3.2% (0.9-5.3%) in outer diameter to 10.6% (6.1-14.9%) for the section modulus. There were no significant differences in the intertrochanteric region.
The comparisons between the ethnic groups within each country show remarkable similarities. In the femoral neck, in both countries, the cross-sectional variables in blacks suggest a narrower marrow cavity, thicker cortices, and a lower buckling ratio despite nonsignificant differences in bone outer diameter. That is, the higher femoral neck BMD in the black groups is mainly caused by a greater amount of bone within the periosteal envelope. In the intertrochanteric region, the bone width is greater in whites, but blacks in both countries still have more bone in their cross-sections and thus lower buckling ratios. These characteristics are consistent with greater bone strength in the black groups in both countries in these skeletal regions. These observations of structural differences that would impart greater mechanical strength in the black groups are consistent with lower reported hip fracture rates in blacks compared with whites in both the United States and South Africa.(24-27) In the shaft, there are no ethnic differences in the South African groups, but those in the U.S. groups are consistent with the measurements in the neck.
It is not clear why there are fewer differences in the shaft region. It is important to note that DXA measurements integrate cortical and trabecular bone, and it is possible that differences in proportions of trabecular bone in the neck and intertrochanteric regions are important in the group comparisons. This could help explain why there are fewer differences in the shaft region. However, the differences in the NSA noted above may contribute to mechanical differences in the more proximal regions. It is also possible that differences in physical activity patterns among the four groups, which are beyond the scope of this paper, could be affecting the cross-sectional geometry of the shaft. For example, it is likely that the NSA affects the manner in which loads are transmitted to the femoral shaft. Another consideration is the uncertainty surrounding the extent to which genetic influences, including differing degrees of admixture, affect these regions of the proximal femur in the four groups.
There are fewer differences between the same ethnic groups in the two countries than there are between ethnic groups within a country. U.S. whites had a wider femoral neck and greater section modulus in the shaft compared with South African whites. Among blacks, there are more differences: in the neck region, U.S. blacks have significantly greater bone widths, cross-sectional areas, endosteal diameters, and section moduli. The pattern is similar in the shaft region. It is important to note that, in the South African groups, the findings of ethnic differences in hip bone density differences are not echoed in the lumbar spine or radius, where no differences in DXA-measured BMD were noted between South African blacks and whites.(12) This contrasts with studies of U.S. blacks and whites that differ significantly at virtually all measurement sites.(2-5)
It is well established that bone strength has both material and structural aspects, although our study only evaluated structural variables. Material strength is influenced by tissue mineralization, porosity, and collagen properties,(28) and it is possible that some apparent ethnic differences in fracture rates may be because of material differences that were not evaluated here. Unfortunately, neither DXA nor any other currently available noninvasive methods can reliably measure bone material properties.
Given the recent evolutionary histories of whites and blacks in the two countries, it is not unexpected that there are more differences between the two black groups than among the two white groups. In both countries, the white populations are largely northern and/or eastern European in origin. The black populations come from different geographic regions (West Africa and South Africa), although the South African group probably represents any number of the Bantu-speaking ethnic groups that originally migrated from West Africa.(29, 30) The U.S. black population has a relatively large degree of admixture with Europeans and with Native Americans.(31) Genetic influences almost certainly contribute to variations in the size, shape, and strength of the proximal femur (and other skeletal regions). Characterizing these populations from a genetic standpoint may be possible in the future, but for now, the relative role of genetics and environmental factors remains speculative. There are also many sociopolitical factors that could affect physical activity patterns and general health and nutrition, which in turn affect skeletal health. These factors deserve further investigation. The results of this study underscore the need to recognize ethnic differences among geographically and culturally distinct populations in biomedical research.
This study was supported by National Institutes of Health Grants AG11541 and AR44655. We thank Dr Elvis Daniels for his work with the South African nurses cohort.