Lower Cortical Porosity and Higher Tissue Mineral Density in Chinese American Versus White Women

Authors


ABSTRACT

Asian women have lower rates of hip and forearm fractures compared to other racial groups despite lower areal bone mineral density (aBMD). We have demonstrated microarchitectural differences, including greater cortical thickness (Ct.Th) and cortical volumetric BMD (Ct.BMD), in Chinese American versus white women. Yet it is not known whether greater Ct.BMD in Chinese American women is a result of greater tissue mineral density (TMD) or reduced cortical porosity (Ct.Po). Using an advanced segmentation algorithm based on high-resolution peripheral quantitative computed tomography (HR-pQCT) images, we tested the hypothesis that Chinese American women have better cortical skeletal integrity owing to lower Ct.Po and higher Ct.TMD compared with white women. A total of 78 Chinese American women (49 premenopausal and 29 postmenopausal) and 114 white women (46 premenopausal and 68 postmenopausal) were studied. Premenopausal Chinese American versus white women had greater Ct.Th, Ct.BMD, and Ct.TMD at both the radius and tibia, and decreased Ct.Po (p < 0.05). A similar pattern was observed between postmenopausal Chinese American and white women. As expected, postmenopausal versus premenopausal women had lower Ct.BMD at the radius and tibia in both races (p < 0.001). Ct.Po largely increased between premenopausal and postmenopausal women, whereas Ct.TMD decreased by 3% to 8% (p < 0.001) in both races. Age-related differences in Ct.Po and Ct.TMD did not differ by race. In summary, both reduced Ct.Po and greater Ct.TMD explain higher Ct.BMD in Chinese American versus white women. Thicker and preserved cortical bone structure in Chinese American women may contribute to greater resistance to fracture compared to white women. © 2014 American Society for Bone and Mineral Research.

Introduction

The incidence of fragility fractures differs between races. Asian individuals have lower rates of cortical (hip and forearm) fractures compared to whites and other racial groups. These differences cannot be explained by areal BMD (aBMD) alone as Asian women have either lower or similar aBMD when bone size is taken into account. In contrast, a higher risk of trabecular (vertebral) fractures has been observed in Asian versus white women.[1-10] The underlying bone microstructural and material properties contributing to racial difference in site-specific fracture risk are largely undefined. With the recent development of new imaging techniques, such as micro–magnetic resonance imaging (µMRI) and high-resolution peripheral quantitative computed tomography (HR-pQCT), bone microarchitecture can now be assessed in vivo, which has revolutionized our ability to assess these characteristics. Using HR-pQCT, we have previously reported microarchitectural differences between Chinese American and white women at the distal radius and tibia.[11-14] We found that Chinese American women have smaller bone area at the radius and tibia, greater cortical volumetric bone density (vBMD) and thickness, and greater trabecular thickness than white premenopausal women. Wang and colleagues[15] reported similar results between premenopausal Chinese and white women residing in Australia.

The thicker and denser cortical bone structure in Chinese American is likely to confer greater resistance to fracture compared to white women. Indeed, cortical bone has recently been shown to play an important role in discriminating fracture from nonfracture subjects and carries most of the load even at the distal sites measured.[16-19] The mechanical competence of cortical bone is not only explained by its thickness and cross-sectional area but also by its microstructural and material properties, such as cortical porosity, mineralization, crystallinity, or the presence of microcracks.[20] Cortical bone stiffness is predominantly associated with mineral content and bone density whereas its toughness is associated with the collagen matrix quality.[21] Cortical porosity was found to account for about 70% of elastic modulus and 55% of yield stress.[22] From a mechanical perspective, age-related changes in porosity have been shown to account for a large part of the reduction in strength, while, in comparison, mineral content played only a minor role.[23] Yet the mineral phase has a strong impact on brittleness, which ultimately contributes to alterations in mechanical competence.

Cortical BMD is a function of both the amount of bone, which is inversely proportional to cortical porosity, and proportional to its degree of mineralization. It is not known whether greater cortical BMD in Chinese American women is a result of greater tissue mineral density (TMD) or reduced cortical porosity (Ct.Po) or which contributes more to racial differences in bone strength. Insight into racial differences in cortical porosity and TMD and which is more predictive of bone strength may further improve our understanding of the pathophysiology of race-specific risk for certain types of fractures (ie, hip/peripheral versus vertebral). Ultimately such data may be helpful in developing more effective programs of osteoporosis diagnosis, prevention, and tailored treatment. This is particularly important for minorities, in whom we currently have inadequate data to make evidence-based decisions. Until very recently it was not possible to assess cortical microstructure using HR-pQCT. Indeed, the standard approach to segment the cortical bone uses a constrained Gaussian smoothing operator to blur out thin trabeculae, followed by the application of a fixed-attenuation threshold to identify thick cortical structures and subsequent slice-wise removal of nonconnected residual trabecular islands.[24] This approach performs poorly for subjects with moderately thin or porous cortices.[25, 26] Therefore, an automated segmentation technique was developed to identify the periosteal and endosteal margins of the distal radius and tibia and detect cortical porosity and cortical tissue mineral density.[25, 27, 28] However, the pores captured by this technique are limited by the resolution of HR-pQCT; the cortical porosity measure obtained represents only large osteonal canals; additionally, cortical tissue mineral density is affected by and represents a surrogate of both matrix mineralization and microporosity beyond the resolution limit of the scanner.

Using this advanced cortical segmentation software we analyzed HR-pQCT images of the radius and tibia to test the hypothesis that Chinese American premenopausal and postmenopausal women have better cortical skeletal integrity owing to lower cortical porosity and higher cortical tissue mineral density compared with white women. Additionally, we assessed within-race age-related differences in cortical parameters and the independent contributions of cortical tissue mineral density versus porosity to whole bone stiffness.

Subjects and Methods

Subjects

As described,[11-14, 29, 30] we conducted a cross-sectional study of premenopausal and postmenopausal white and Chinese American women. Inclusion criteria for white women required that all four grandparents were of white race/non-Hispanic ethnicity. For Chinese American women, all four grandparents had to be of Chinese descent (self-reported). Women of mixed racial ancestry were excluded. Women who met these criteria and who resided in the United States at the time of enrollment were included in the study. Country of birth was not a criterion for study entry.

Premenopausal status was evaluated by history as having regular menses with more than six cycles per year. We further limited inclusion to premenopausal women who were 29 to 40 years old in order to study women who had reached peak bone mass but in whom the perimenopausal transition or menopause had not yet influenced skeletal metabolism. Postmenopausal status was defined as the absence of menses for >1 year. Postmenopausal women who were 59 to 70 years old were included in order to study women who were past the perimenopausal transition but who had not yet reached an advanced age when comorbid conditions/medications would be more likely to affect bone metabolism. Participants were recruited by newspaper, Internet advertisements, flyers, and directly at primary care physician offices. Women were screened for medical conditions and medications thought to affect bone metabolism. Exclusion criteria included untreated hyperthyroidism, renal dysfunction (estimated glomerular filtration rate [GFR] ≤ 60 mL/min), liver dysfunction (aspartate transaminase or alanine transaminase two times the upper limit of normal), intestinal malabsorption owing to any cause, history of malignancy other than non-melanomatous skin cancer, metabolic bone diseases (primary or secondary hyperparathyroidism, hypoparathyroidism, Paget's disease), human immunodeficiency virus (HIV) disease, organ transplantation, and drug exposures affecting bone metabolism (ie, current or past use of glucocorticoids >3 months, tacrolimus, cyclosporine, methotrexate, antiepileptic medications, teriparatide, calcitonin, or aromatase inhibitors; current use of hormone-replacement therapy or raloxifene; and current or more than 1 year of use of bisphosphonates). Premenopausal who had amenorrhea ≥6 months, or were currently pregnant or lactating were also excluded from the study.

Ninety-five premenopausal women (46 white and 49 Chinese American) and 97 postmenopausal women (68 white and 29 Chinese American) were studied.[11-14] This study was approved by the Institutional Review Board of Columbia University Medical Center. All patients gave written, informed consent.

Clinical and biochemical evaluation

Information regarding past medical history and medications, as well as oral contraceptive use (OCP), calcium intake, physical activity (Baecke sport index), and smoking was collected as described.[11-14, 29, 30] Weight and height were measured by balance beam and a wall-mounted, calibrated Harpenden stadiometer (Holtain Ltd, Crymych, UK), respectively. Serum calcium was measured with colorimetric or spectrophotometric methods. Serum intact parathyroid hormone (PTH) was measured by chemiluminescence assay and 25-hydroxyvitamin D was measured by liquid chromatography tandem mass spectrometry (Quest Diagnostics, Nichols Institute, San Juan Capistrano, CA, USA).[31]

Assessment of bone density and microarchitecture

Volumetric BMD and microarchitecture were assessed at the nondominant distal radius and distal tibia using a single HR-pQCT device (XtremeCT, Scanco Medical AG, Brüttisellen, Switzerland). This system allows a noninvasive assessment of 110 CT slices, with a nominal voxel size of 82 µm. Settings were set at 60 kVp, 900 mA, with a matrix size of 1536 × 1536. The region of interest (ROI) at the radius consisted of an approximately 9-mm length of bone located from 9.5 to 18.5 mm proximal to the endplate. At the tibia, the ROI was located from 22 to 31 mm proximal to the endplate.[32] Previous studies reported microarchitecture and BMD assessment according to the default clinical evaluation protocol provided by the manufacturer, in which cortical and trabecular regions are automatically separated using a Gaussian filter and a threshold-based algorithm.[11-13]

Extended cortical parameters were measured with a newer algorithm incorporated in the manufacturer's analysis software (micro-computed tomography [µCT] Evaluation v6.0; Scanco Medical AG).[27, 28, 33] This automated cortical analysis was first based on a series of morphological operations (dilatation, erosion, etc.) applied to the binarized image (bone/background) that allows the segmentation of the dense cortical bone compartment with the generation of periosteal and endosteal contours that were visually validated. A second step consisted of segmentation of the cortical porosity. We have made the distinction between intracortical porosity (IntraCt.Po, %), consistent with Haversian canals, for which endosteal and periosteal voids, transcortical erosions, and objects with a volume less than 0.003 mm3 were discarded; and cortical porosity (Ct.Po, %) calculated as 1–bone volume/total volume (BV/TV) of the cortical bone compartment, which included the previous features. We included both of these parameters because each has been previously reported in the literature as cortical porosity.[27, 28, 33] Mean and distribution (SD) of intracortical pore diameter (IntraCt.Po.Dm and IntraCt.Po.Dm.SD, both measured as micrometers [µm]) were measured by direct three-dimensional [3D] method. Finally, mean cortical thickness (Ct.Th; µm) was measured by direct 3D method[34] disregarding the presence of pores. Cortical BMD (Ct.BMD, mg HA/cm3) was assessed as the average mineral density within the cortical bone compartment after exclusion of a 2-voxel-thick layer on the periosteal and endosteal boundaries to minimize partial volume effects. Cortical tissue mineral density (Ct.TMD, mg HA/cm3) was assessed after further exclusion of all pore space. Ct.TMD represents a surrogate of both matrix mineralization and microporosity beyond the resolution limit of the scanner.

Intrinsically, the separation between cortical and trabecular bone with the newer algorithm leads to different cortical and trabecular compartment sizes and BMD than with the standard analysis and therefore results may differ from prior published data on this population.[11-14, 29, 30]

Assessment of bone stiffness

Whole-bone stiffness was assessed by finite-element analyses (FEA) from HR-pQCT images of the radius and tibia, as described.[11, 12, 35, 36] Briefly, micro–finite element (µFE) models were created by converting each voxel in the thresholded HR-pQCT image to an equally sized brick element. Bone tissue was modeled as an isotropic linear elastic material with a Young's modulus of 15 GPa and a Poisson's ratio of 0.3. The axial stiffness was calculated as the reaction force divided by the uniaxial imposed displacement that equaled 1% of the bone segment height, applied perpendicularly to the distal surface of bone. All µFEAs were performed by using customized software[11, 12, 35, 36] with an element-by-element preconditioned conjugate gradient solver[37] and implemented on a Dell XPS PC workstation (Dell Inc., Round Rock, TX, USA).

Statistical analysis

Data are expressed as unadjusted mean ± SD. Comparisons of group characteristics between the Chinese American and the white groups were evaluated by independent two-sided t test or Wilcoxon signed rank test. Bone volumetric density and microarchitectural parameters were compared between the two racial groups for premenopausal and postmenopausal women and between menopausal status for each race, at both skeletal sites using independent two-sided t test or Wilcoxon signed rank test. Generalized linear models were used to compare geometrical parameters (areas, volumes, and cortical thickness) adjusted for height and weight. Additionally, we adjusted for age within menopausal status. Cortical porosity and TMD were only adjusted for age and not height or weight because both measures are already normalized to bone size. The differences and mean adjusted differences are expressed as percentage and number of SDs (based on the average SD calculated in premenopausal women for each variable). The age-related differences between races were assessed by the race–menopause interaction in generalized linear models adjusted for covariates. Associations between cortical microstructural parameters were assessed by Pearson's correlation coefficient. Stepwise linear regression models were used to assess the independent associations of cortical variables (Ct.Po and Ct.TMD) with mechanical properties (whole-bone stiffness).

Statistical analysis was performed using SPSS v18.0 (SPSS, Inc., Chicago, IL, USA), with significance set at the 5% level (two-tailed).

Results

As shown in Table 1, there was no significant differences in age between races within menopausal status. In both the premenopausal and postmenopausal groups Chinese American women weighed less and were shorter than white women. PTH level was slightly higher in premenopausal Chinese American than white women. Chinese American women had lower vitamin D levels and lower calcium intake than their white counterparts; though there was no difference in serum calcium level.

Table 1. Characteristics of Premenopausal and Postmenopausal Chinese American and White Women
 Premenopausal womenPostmenopausal women
 White (n = 46)Chinese (n = 49)White (n = 68)Chinese (n = 29)
  • Data are expressed as unadjusted mean ± SD.
  • BMI = body mass index; n/a = not applicable; OCP = oral contraceptive use; PTH = serum intact parathyroid hormone.
  • aOCP duration among past or current OCP users.
  • bData available for 174/192 women.
  • cData available for 152/192 women.
  • *p < 0.05 between race by menopausal status.
  • **p < 0.01 between race by menopausal status.
  • ***p < 0.001 between race by menopausal status.
Age (years)35 ± 436 ± 764 ± 361 ± 2
Height (cm)165 ± 7162 ± 6**162 ± 6157 ± 5***
Weight (kg)63 ± 1756 ± 10*66 ± 1258 ± 8***
BMI (kg/m2)23.1 ± 5.521.6 ± 3.525.3 ± 4.923.6 ± 2.6*
Years since menopause (years)n/an/a12 ± 510 ± 4
OCP duration (years)a7.4 ± 6.43.7 ± 3.3*n/an/a
Calcium intake (mg/d)1394 ± 1570885 ± 557*1557 ± 730901 ± 544***
Current smoking (%)4.1%6.5%1.5%0.0%
Baecke sport index1.6 ± 0.71.1 ± 0.6***1.2 ± 0.51.2 ± 0.7
Serum calcium (mg/dL)9.4 ± 0.39.4 ± 0.39.5 ± 0.49.5 ± 0.4
PTH (pg/mL)b31 ± 1337 ± 13*38 ± 1237 ± 11
25-hydroxyvitamin D (ng/mL)c36 ± 1425 ± 9***38 ± 1431 ± 10*
Figure 1.

Adjusted difference in SD, between Chinese American and white women by menopausal status, at the radius and tibia. *p < 0.05, **p < 0.01, ***p < 0.001 after adjustment for age; ° = adjustment for age, height, and weight.

Among premenopausal women, Chinese Americans had smaller bone at both the radius (mean ± SD; –10.8% ± –0.6%; p = 0.002) and tibia (–7.2% ± –0.5%; p = 0.015) along with higher total volumetric density (radius: 19.6% ± 1.0%; p < 0.001; tibia: 11.9% ± 0.6%; p = 0.002) compared with white women (Table 2). Trabecular area was also smaller (radius: –16.5% ± –0.8%; p < 0.001; tibia: –9.2% ± –0.5%; p = 0.010) and trabecular volumetric density was higher (radius: 13.7% ± 0.5%; p = 0.012; tibia: 9.9% ± 0.5%; p = 0.028) in Chinese American women. Their cortical area was higher at the radial site (8.2% ± 0.5%; p = 0.028) compared to white women.

Table 2. Cortical Microstructure at the Distal Radius and Tibia in Premenopausal and Postmenopausal Chinese American and White Women
 Premenopausal womenPostmenopausal women
 WhiteChinese AmericanWhiteChinese American
  • Data are expressed as unadjusted mean ± SD.
  • Ar = area; Tt = total; BMD = bone mineral density; HA = hydroxyapatite; Tb = trabecular; Ct = cortical; Ct.Th = Ct thickness; Po = porosity; IntraCt = intracortical; IntraCt.Po.Dm = intracortical pore diameter; IntraCt.Po.Dm.SD = standard deviation of IntraCt.Po.Dm; TMD = tissue mineral density.
  • *p < 0.05 between race by menopausal status.
  • **p < 0.01 between race by menopausal status.
  • ***p < 0.001 between race by menopausal status.
  • p < 0.05 between race by menopausal status after adjustment for age.
  • ††p < 0.05 between race by menopausal status after adjustment for age, height, and weight.
Distal radius
Tt.Ar (mm2)233 ± 39208 ± 36**,††230 ± 42203 ± 34**
Tt.BMD (mg HA/cm3)330 ± 58395 ± 58***,298 ± 73341 ± 52**
Tb.Ar (mm2)182 ± 39152± 34***,††179 ± 43149 ± 33***
Tb.BMD (mg HA/cm3)152 ± 33173 ± 42*,132 ± 36124 ± 37
Ct.Ar (mm2)55 ± 959 ± 9*,††53 ± 857 ± 6**,††
Ct.Th (µm)916 ± 1631085 ± 158***,††886 ± 1881046 ± 137***,††
Ct.BMD (mg HA/cm3)1020 ± 311046 ± 26***,943 ± 74985 ± 50**
Ct.Po (%)4.8 ± 1.43.6 ± 1.1***,9.1 ± 3.96.9 ± 2.6**
IntraCt.Po (%)1.1 ± 0.60.9 ± 0.43.1 ± 2.23.0 ± 1.4
IntraCt.Po.Dm (µm)160 ± 17168 ± 23183 ± 32195 ± 27*
IntraCt.Po.Dm.SD (µm)65 ± 1473 ± 2380 ± 2385 ± 15
Ct.TMD (mg HA/cm3)1036 ± 261059 ± 22***,1000 ± 471029 ± 37**
Stiffness (kN/mm)83 ± 1594 ± 17***,††68 ± 1672 ± 12
Distal tibia
Tt.Ar (mm2)663 ± 83615 ± 100*685 ± 106618 ± 87**
Tt.BMD (mg HA/cm3)281 ± 46314 ± 55**,245 ± 44257 ± 43
Tb.Ar (mm2)556 ± 84505 ± 100**581 ± 108515 ± 89**
Tb.BMD (mg HA/cm3)160 ± 33176 ± 34*,152 ± 31142 ± 30
Ct.Ar (mm2)111 ± 18114 ± 18 ††108 ± 16110 ± 13
Ct.Th (µm)1162 ± 2071262 ± 232*,††1044 ± 1891172 ± 166**
Ct.BMD (mg HA/cm3)959 ± 41985 ± 36**,803 ± 87844 ± 69*
Ct.Po (%)7.6 ± 2.06.5 ± 1.5**,16.1 ± 5.513.7 ± 4.0*
IntraCt.Po (%)3.8 ± 1.53.3 ± 1.1*8.9 ± 3.58.9 ± 2.9
IntraCt.Po.Dm (µm)193 ± 20192 ± 27199 ± 25223 ± 34**,
IntraCt.Po.Dm.SD (µm)90 ± 1788 ± 2190 ± 19104 ± 26*,
Ct.TMD (mg HA/cm3)1012 ± 261030 ± 25***,930 ± 52956 ± 45*
Stiffness (kN/mm)231 ± 42249 ± 45*,††206 ± 39207 ± 32

At both skeletal sites, Chinese Americans had greater Ct.Th (radius: 18.5% ± 0.9%; tibia: 8.6% ± 0.4%; all p < 0.05), Ct.BMD (radius: 2.6% ± 0.8%; tibia: 2.8% ± 0.6%; all p < 0.01), and Ct.TMD (radius: 2.2% ± 0.8%; tibia: 1.8% ± 0.7%; all p ≤ 0.001) than their white counterparts. Cortical porosity was lower in Chinese American women at the radius (–23.5% ± –0.8%; p < 0.001) and tibia (–14.2% ± –0.6%; p = 0.007). Similarly, intracortical porosity was lower at the tibia (–15.3% ± –0.4%; p = 0.049), compared to white women. The size of the intracortical pores (IntraCt.Po.Dm) and its heterogeneity (IntraCt.Po.Dm.SD) did not differ between Chinese American and white women at either site. Whole-bone stiffness was significantly higher in Chinese American than white women (radius: 14.3% ± 0.7%; tibia: 8.5% ± 0.4; all p < 0.05).

After adjustment of geometrical parameters for age, height, and weight, racial differences in total bone area were attenuated. They remained significant at the radius but not at the tibia. Yet cortical area and thickness were significantly higher at both sites (radius: Ct.Ar 11.3% ± 0.7%, Ct.Th 18.3% ± 0.9%; tibia: Ct.Ar 6.6% ± 0.4%, Ct.Th 8.7% ± 0.5%; all p < 0.05) in Chinese American versus white women. Differences between races in Ct.BMD, Ct.Po, and Ct.TMD remained unchanged after adjustment for age at either site.

Among postmenopausal women, Chinese Americans had smaller bone at both the radius (–11.4% ± –0.7%; p = 0.005) and tibia (–9.7% ± –0.7%; p = 0.004) along with smaller trabecular area (radius: –16.8% ± –0.8%; tibia: –11.4% ± –0.7%; all p < 0.01). Total volumetric density was higher at the radius (14.4% ± 0.6%; p = 0.006) in Chinese Americans and trabecular volumetric density was not different between groups (Table 2). Cortical area was higher at the radial site (7.6% ± 0.4%; p = 0.007) in Chinese American compared to white women.

At both skeletal sites, Chinese Americans had greater Ct.Th (radius: 18.0% ± 0.9%; tibia: 12.3% ± 0.6%; all p < 0.01), Ct.BMD (radius: 4.4% ± 1.3%; tibia: 5.1% ± 1.0%; all p < 0.03), and Ct.TMD (radius: 2.9% ± 1.1%; tibia: 2.8% ± 1.0%; all p < 0.02) than their white counterparts. Whole-bone stiffness did not differ between races at either site.

Chinese American women had lower cortical porosity at the radius (–23.7% ± –1.6%; p = 0.006) and tibia (–14.8% ± –1.3%; p = 0.036) compared to white women, yet intracortical porosity was not different between groups. In Chinese American women, intracortical pore size was higher at both the radius (6.5% ± 0.6%; p = 0.038) and tibia (11.6% ± 1.0%; p = 0.004) and was more heterogeneous at the tibia (15.7% ± 0.7%; p = 0.017) than in white women.

After adjustment of geometrical parameters for age, height, and weight, racial differences in bone total and trabecular area were attenuated and no longer significant at both the radius and tibia. Tibial cortical thickness was no longer different between groups. Yet cortical area and thickness remained significantly higher at the radius (Ct.Ar: 11.2% ± 0.6%; Ct.Th: 14.5% ± 0.7%; all p < 0.01) in Chinese American versus white women. After adjustment for age, differences in Ct.BMD, Ct.Po, and Ct.TMD were attenuated and no longer significant at either site. Intracortical pore size and its heterogeneity remained higher in Chinese American versus white women at the tibia (11.5% ± 1.0% and 14.5% ± 0.7%, respectively; both p < 0.05).

Figure 2.

Difference in SD, between premenopausal and postmenopausal women by race, at the radius and tibia. *p < 0.05, **p < 0.01, ***p < 0.001; ° = adjustment for height and weight.

At the radius, the total, trabecular and cortical bone size did not differ between premenopausal and postmenopausal women in both races. Moreover, cortical thickness did not differ between menopausal status in either race.

Figure 3.

3D representation of cortical porosity at the tibia in premenopausal and postmenopausal Chinese American versus white women.

As expected, postmenopausal women had lower Tt.BMD (Chinese American: –13.7% ± –0.8%; white: –9.7% ± –0.5%; all p < 0.02), Tb.BMD (Chinese American: –28.1% ± –1.2%; white: –13.5% ± –0.5%; all p < 0.01), Ct.BMD (Chinese American: –5.9% ± –2.0%; white: –7.5% ± –2.5%; all p < 0.001), and whole-bone stiffness (Chinese American: –23.9% ± –1.3%; white: –16.8% ± –0.8%; all p < 0.001) compared with premenopausal women (Table 2).

Ct.Po was markedly increased in postmenopausal women (Chinese American: 89.3% ± 2.4%; white: 89.8% ± 3.1%; all p < 0.001). Moreover, intracortical porosity almost tripled between premenopausal and postmenopausal women (Chinese American: 227% ± 3.9%; white: 167% ± 3.6%; all p < 0.001), whereas Ct.TMD decreased (Chinese American: –2.8% ± –1.1%; white: –3.5% ± –1.4%; all p < 0.001) compared with premenopausal women.

Intracortical pore size and its heterogeneity were higher in postmenopausal than in premenopausal women in both races (Ct.Po.Dm: Chinese American: 16.3% ± 1.3%; white: 14.4% ± 1.1%; Ct.Po.Dm.SD: Chinese American: 16.2% ± 0.6%; white: 23.6% ± 0.8%; all p < 0.01).

After adjustment of geometrical parameters for height and weight, we observed the same pattern of age-related differences in both races. Cortical thickness in Chinese American women was additionally lower in postmenopausal compared to premenopausal women (–8.3% ± –0.5%; p = 0.023) after adjustment.

The age-related differences did not differ by race as assessed by the race–menopause interaction in generalized linear models, except for Tb.BMD, which demonstrated a more marked decline with age in Chinese American than in white women (–29.7% versus –15.6%; p = 0.014).

Similar to the findings observed at the radius, at the tibia the total, trabecular, and cortical bone size did not differ between premenopausal and postmenopausal women in either race. Cortical thickness was lower in postmenopausal versus premenopausal white women (–10.1% ± –0.5%; p = 0.006) but the difference did not reach statistical significance in Chinese Americans (–7.1% ± –0.4%; p = 0.14).

Postmenopausal women had lower Tt.BMD (Chinese American: –18.3% ± –1.1%; white: –12.6% ± –0.7%; all p < 0.001), Ct.BMD (Chinese American: –14.3% ± –3.5%; white: –16.2% ± –3.8%; all p < 0.001), and whole-bone stiffness (Chinese American: –16.9% ± –0.9%; white: –10.2% ± –0.5%; all p < 0.01) compared with premenopausal women (Table 2). Tb.BMD was lower in postmenopausal Chinese American women (–19.2% ± –1.0%, p < 0.001) but was not significantly lower in white women (–5.5% ± –0.2%; p = 0.16) compared to premenopausal women.

Ct.Po was increased in postmenopausal women (Chinese American: 110% ± 3.9%; white: 111% ± 4.6%; all p < 0.0001). Likewise, intracortical porosity more than doubled between premenopausal and postmenopausal women (Chinese American: 174% ± 4.1%; white: 130% ± 3.6%; all p < 0.0001), whereas Ct.TMD decreased (Chinese American: –7.1% ± –2.7%; white: –8.1% ± –3.0%; all p < 0.0001) compared with premenopausal women.

Intracortical pore size and its heterogeneity were higher in Chinese American postmenopausal women (Ct.Po.Dm: 15.9% ± 1.3%; Ct.Po.Dm.SD: 18.7% ± 0.9%; all p < 0.01) but did not differ in postmenopausal white women compared to premenopausal women.

After adjustment of geometrical parameters for height and weight, we observed a significant increase in Tt.Ar (Chinese American: 9.8% ± 0.6%; white: 6.6% ± 0.4%; all p < 0.005), Tb.Ar (Chinese American: 13.9% ± 0.7%; white: 8.9% ± 0.5%; all p < 0.002) and a decrease in Ct.Th (Chinese American: –12.9% ± –0.7%; white: –14.2% ± –0.7%; all p < 0.002) in postmenopausal women compared to premenopausal women in both races. Ct.Ar did not differ between premenopausal and postmenopausal women in both races.

The age-related differences did not differ by race as assessed by the race-menopause interaction in generalized linear models, except for Tb.BMD, intracortical pore size, and intracortical pore size heterogeneity. The age-related decline in Tb.BMD was more marked in Chinese American women (–19.2% versus –6.8%; p = 0.014), compared to white women. The age-related increase in intracortical pore size was also higher in Chinese American women (16.9% versus 3.8%; p = 0.002), as well as intracortical pore size heterogeneity (19.4% versus 0.7%; p = 0.01) compared to white women.

Associations between cortical parameters

Cortical BMD was moderately associated with cortical thickness (r = 0.64 and 0.53; p < 0.001) and highly correlated with cortical porosity (r = –0.93 and –0.97; p < 0.001), intracortical porosity (r = –0.80 and –0.91; p < 0.001) and cortical TMD (r = 0.94 and 0.95; p < 0.001) in the whole population at the radius and tibia, respectively. Associations between Ct.BMD and Ct.Po were lower in premenopausal women than in postmenopausal women at both the radius (r = –0.79 and –0.92, respectively; p < 0.001) and tibia (r = –0.89 and –0.94, respectively; p < 0.05). Moreover, Ct.Po and Ct.TMD were moderately correlated together at the radius and tibia (in premenopausal women: r = –0.65 and –0.65; both p < 0.001; and in postmenopausal women r = –0.73, and –0.74; both p < 0.001).

Prediction of bone stiffness by Ct.Po and Ct.TMD

At the radius, Ct.Ar, Ct.Th, and Ct.vBMD explained 46.2%, 28.6%, and 21.4%, respectively, of whole-bone stiffness; all p < 0.001. Ct.TMD explained 20.4% and Ct.Po explained 18.3%, whereas IntraCt.Po explained only 7.7% of whole-bone stiffness (all p < 0.001).

At the tibia, Ct.Ar, Ct.Th, and Ct.vBMD explained 31.0%, 22.3%, and 14.5%, respectively, of whole bone stiffness; all p < 0.001. Ct.TMD explained 17.3% and Ct.Po explained 10.2%, whereas IntraCt.Po explained only 5.6% of whole-bone stiffness (all p ≤ 0.001).

IntraCt.Po.Dm and IntraCt.Po.Dm.SD were not predictors of whole-bone stiffness at either site.

The explanatory effects of each cortical parameter on whole-bone stiffness did not differ by race or menopausal status as assessed by the race–parameter or menopause–parameter interactions in linear regression models.

When assessed in stepwise regression models (Ct.TMD and Ct.Po), only Ct.TMD remained a significant predictor of bone stiffness at both sites. When adding either Ct.Th or Ct.Ar to the previous candidate predictors, at the tibia, Ct.TMD, but not Ct.Po, remained independently associated with whole-bone stiffness. The combination of Ct.Ar and Ct.TMD explained 37.4% and the combination of Ct.Th and Ct.TMD explained 25.3% of whole-bone stiffness (all p < 0.01) at the tibia. However, at the radius, Ct.Po, but not Ct.TMD, remained independently associated with whole-bone stiffness. The combination of Ct.Ar and Ct.Po explained 50.1% and the combination of Ct.Th and Ct.Po explained 31.5% of whole-bone stiffness (all p < 0.01) at the radius.

Discussion

Using advanced cortical segmentation software, we confirmed that Chinese American women have thicker and denser cortices at distal appendicular sites compared with white women, as we and others have reported.[11-15, 42] This study extends this knowledge base by examining the underlying structural and bone material properties that contribute to greater cortical BMD in Chinese American versus white women. Additionally, our analysis assessed the independent contributions of cortical tissue mineral density versus porosity to whole bone stiffness and is the first study to do so. The higher cortical BMD observed in Chinese American versus white women was a result of both lower cortical porosity and higher tissue mineral density, which, owing to technical limitations, is a surrogate for both the degree of mineralization and microporosity beyond the resolution of the device. Although the absolute differences in cortical porosity and tissue mineralization density between races tended to be relatively small, we believe they are likely to be clinically relevant for several reasons. A number of studies indicate that the relationship between cortical porosity and bone strength is exponential, such that small increases in porosity lead to large decreases in bone strength.[38] In a recent report in white women, each SD increase in cortical porosity increased the odds of fracture by 22% to 55% depending upon the skeletal site.[39] Applying, these data, although the method of cortical porosity measurements differed between studies, the racial differences (up to 1.6 SD) in cortical porosity observed in our investigation could translate into a 35% increased risk of fracture in white versus Chinese American women.

Higher cortical BMD in Chinese individuals was found in several other studies using the same technique as well as other computed tomographic modalities. A subset of our population (83 premenopausal women: 47 white and 36 Chinese American and 50 postmenopausal women: 24 white and 26 Chinese American) underwent QCT scans of the hip.[14] Among premenopausal women, cortical BMD was 3% greater at the femoral neck (p = 0.05) and 3.6% greater at the total hip (p = 0.01) in Chinese American compared with white women. Cortical BMD was also 4% greater at the total hip (p = 0.02) and tended to be greater at the femoral neck (p = 0.058) in postmenopausal Chinese American versus white women. Using the same technique in the Osteoporotic Fractures in Men (MrOS) study, cortical BMD at the femoral midshaft was greater in Asian men than in Caucasian men older than 65 years but cortical BMD at the femoral neck was comparable between races.[40] At the tibial midshaft, Asian girls were found to have smaller cortical area and higher cortical BMD than Caucasians as measured by pQCT.[41] Wang and colleagues[15] reported that cortical BMD measured at the distal radius and tibia by HR-pQCT was not different between Chinese and Caucasians girls before puberty, but after menarche and in premenopausal women, Chinese individuals had greater cortical BMD. Consistent with our results, a recent study indicated that Canadian Asian adolescent males had greater cortical bone density and thickness and lower cortical porosity compared with white adolescent males.[42] Asian versus white female adolescents also had greater cortical vBMD and thickness, but the difference in cortical porosity was not significant. Our study extends these observations regarding cortical porosity to adults. Also congruent with our results, a recent study by Bjornerem and colleagues[39] suggests that taller (white) women have relatively thinner cortices and higher cortical porosity predisposing to an increased risk of fracture. These findings are consistent with our report as well, given the shorter height of Asian versus white women in our study.

Lower cortical porosity in Chinese American women is one explanation to account for their higher cortical BMD compared with white women. Another mechanism is the difference in cortical tissue mineral density, as observed at both skeletal sites. Indeed, our regression analysis revealed that cortical TMD explained more of the variance in whole-bone stiffness than cortical porosity. However, in our regression models containing cortical area and thickness, we found that both cortical porosity and TMD independently contributed to bone strength, although there were site-specific differences. This suggests that these extended cortical parameters have importance beyond the more standardly available measures, though the reasons for the site-specific differences remain unclear and require further study. Moreover, these results need to be interpreted with several limitations in mind. Current FEA methods address only the linear elastic properties of bone but not the nonlinear properties such as yield strength, which may be more relevant to fracture risk. Further, boundary and loading conditions used in the FEA models were simplified. Indeed, we applied a homogeneous tissue modulus for the mineralized phase that did not incorporate information about the distribution and degree of mineralization of bone in the cortex. Therefore, our models did not account for differences in tissue mineralization and strength. Interpretation of the relationship between tissue properties (TMD) and “microstructural” stiffness (from homogeneous FEA) is limited. In this context, models with material properties reflecting the distribution of mineralization would be more appropriate. Yet the relationship to apply for converting CT-attenuation to modulus depends on the image resolution and structure size. It has been approximated for HR-pQCT images,[43] but requires further testing and validation to establish reliable relations between the CT-attenuation measured at the radius and tibia and elastic modulus. Moreover, we simulated a uniaxial compression test on a planar parallel slice of the bone, which is not necessarily an accurate representation of the loading applied to this region during a fall. The affect of porosity on bone strength may have been underestimated by this loading condition and it may have been more pronounced with bending simulation. Other loading conditions incorporating both compression and bending would be of great interest to study; even though the actual model has been showed to highly correlate with failure load prediction based on a full 4-cm region of the radius to which more realistic boundary conditions were applied.[44]

Most studies have assessed tissue mineral density, also referred to as bone mineralization density distribution (BMDD), by quantitative backscattered electron imaging (qBEI),[45, 46] or degree of mineralization (DMB) by contact microradiography,[47] on either bone necropsies or biopsies. Available data suggest that BMDD does not vary with age, skeletal site or gender.[45-48] However, it is influenced by the level of bone remodeling activity.[49] Comparisons of racial differences in trabecular BMDD are limited, however, to studies of African Americans and whites.[50, 51] To our knowledge, no studies have included Asians. Roschger and colleagues[51] reported no difference in trabecular BMDD parameters assessed by the qBEI method between African American and white premenopausal women. However, BMDD in cortical bone can have remarkable variations even within a single skeletal site, which might be related to the highly heterogeneous osteonal structures found in cortical bone. Suitable cortical BMDD evaluation methods have been developed, but since then no study has assessed difference in cortical BMDD between races. In our study, whole cortical bone was analyzed, thus precluding a possible bias of osteonal selection. The higher cortical TMD observed in Chinese American women at both the radius and tibia could also be partly driven by their lower cortical porosity, resulting in a lower surface available for remodeling. Fewer remodeling units might lead to less undermineralized bone.

Although our previous reports with the standard analysis showed that Ct.Th was significantly higher in both premenopausal and postmenopausal Chinese American women compared with white women, differences were less pronounced with the advanced segmentation algorithm used in this study. The standard procedure used in our prior studies performs poorly on thin and/or porous cortices. Therefore, we expected to find greater differences between the two methods in postmenopausal compared to premenopausal women. Using the advanced segmentation algorithm in premenopausal women, Ct.Th was increased by 12% at the radius and 4% to 6% at the tibia in both races. Therefore, differences between races did not change between the two methods. In postmenopausal women, Ct.Th was 22% and 18% higher in Chinese American and 21% and 23% higher in white women than with the standard algorithm, at the radius and tibia, respectively. A greater difference was observed in white women at the tibia compared to Chinese American between the two methods, leading to a nonsignificant difference after adjustment at the tibia, where porosity is almost three times higher than at the radius.

The strength of our study lies in the noninvasive assessment of bone microstructure and material properties but technical limitations related to these in vivo measurements should also be considered. Indeed, assessment of cortical TMD should be considered with caution, because it might be impacted by thicker cortices, resulting in more beam hardening artifact, which can cause confounding error in density measurements.[52] Moreover cortical TMD depends on accurate depiction of cortical porosity, which is limited by the resolution of the device. Cortical TMD is therefore a surrogate for both matrix mineralization and microporosity, ie, porosity below the limit of the HR-pQCT resolution, and was moderately correlated with cortical porosity. Mean pore diameter was lower in premenopausal than postmenopausal women in both races and at both sites, with exception of white women at the tibia. The higher Ct.TMD in Chinese American versus white women may reflect, at least in part, racial differences in microporosity rather than true differences in tissue mineral density. Future biopsy studies would be helpful to determine whether cortical tissue mineral density is truly increased in Chinese American women or if differences in Ct.TMD assessed by HR-pQCT are mainly driven by their lower microporosity compared to white women.

Movement artifacts and partial volume effects are likely to impact the accuracy of Ct.Po and Ct.TMD. Therefore, the quality of HR-pQCT scans was reviewed by an experienced operator, and graded into five classes from 1 for perfect scans to 5 for the poorest scans, as described.[53, 54] Scans with poor quality; ie, grade 5, were excluded from the analyses (n = 12 at the radius and n = 5 at the tibia). When validated on 10 cadaver forearms against µCT at a resolution of 19 µm, the number of pores was underestimated with HR-pQCT and the correlation between the two modalities was r = 0.52. Yet the correlation for cortical porosity was r = 0.89 with a mean overestimation of 6.3% when measured from HR-pQCT images.[28] Accuracy of fine structure measurements usually depends on the scale used, and in vivo assessments these measurement are challenging whatever the technique used. Even if absolute values of cortical TMD and cortical porosity are partly inaccurate, errors are systematic; therefore, differences between groups should not be affected.

Limb length difference between races may affect the measurement site and therefore the results. Indeed, cortical thickness is three times thicker on the 20 most proximal slices compared to the 20 most distal slices.[16] Because we did not measure limb length in our study, we cannot exclude the possibility that this may partly account for the differences observed between racial groups. Yet Wang and colleagues[15] reported that they found similar differences between Chinese and white postpubertal girls when measured with the standard procedure, ie, the first CT slice fixed at 9.5 mm from the reference line at the radius, and when the ROI was chosen based on 4% of radius length. Average forearm length reported in women aged 18 years was 23.7 ± 1.5 cm[55]; therefore minimum and maximum lengths as evaluated by mean ± 2 SD were 20.7 and 26.7 cm, respectively. A relative positioning of 4% would differ by up to 0.24 cm but most differences would range between –1 SD and +1 SD; therefore by, ∼15 HR-pQCT slices (13% of the scan length).

Other cross-sectional studies have assessed age-related differences in cortical porosity and found increases with age ranging from 3.2% to 12.9% in approximately 30 year-old individuals[28] to a more dramatic increase[33, 56] of threefold in women aged 20 to 60 years,[33] which was similar to our finding. These studies are limited to white individuals and our investigation is the first to assess whether age-related differences in cortical porosity vary by race. Differences in Ct.Po and Ct.TMD between Chinese American and white women were more marked in premenopausal compared to postmenopausal women, yet their age-related differences did not differ by race. However, at the tibia, Chinese American women had a higher age-related increase in intracortical pore size and its heterogeneity compared to white women. A possible explanation for these differences could be that women in different racial groups respond to menopause/estrogen deficiency in different ways. Postmenopausal white women may create cortical pores at a faster rate than postmenopausal Chinese American women, keeping the mean pore size relatively constant while cortical pores may preferentially enlarge in Chinese American postmenopausal women. Future longitudinal studies are necessary to confirm our cross-sectional findings and hypothesis.

The greater cortical BMD in Chinese American women can be explained both by reduced cortical porosity and greater tissue mineral density as assessed by HR-pQCT. Ct.TMD is likely a surrogate for both matrix mineralization and microporosity, which cannot be discriminated with the current imaging technology. Therefore, studies using more dedicated techniques are needed to determine whether higher Ct.TMD is owing to a true increase in the degree of mineralization of bone and/or to a lower microporosity in Chinese American compared to white women. These results, however, are consistent with the lower bone remodeling observed in Chinese American compared to white women. Given the relationship between cortical porosity, cortical tissue mineral density, and bone strength, our findings suggest that the thicker and more preserved cortical bone structure in Chinese American women may contribute to greater resistance to fracture compared to white women.

Disclosures

All authors state that they have no conflicts of interest.

Acknowledgments

This research was supported by grants from the National Institutes of Health grants K23 AR053507 and R01 AR051376, a National Osteoporosis Foundation grant, Thomas L. Kempner and Katheryn C. Patterson Foundation, and the Mary and David Hoar Fellowship Program of the New York Community Trust and the New York Academy of Medicine. In addition, this work was supported by a fellowship grant from Laboratoire Servier (to SB). We thank Dr. Clyde Wu for his vision and support of this study.

Authors' roles: All authors participated in the design of the study, and/or the analysis of the study data. All authors contributed to the manuscript and approved the final version for submission.

Ancillary