Paradoxically, Asians have lower areal bone mineral density (aBMD), but their rates of hip and wrist fractures are lower than whites. Therefore, we used high-resolution pQCT (HR-pQCT) to determine whether differences in bone macrostructure and microstructure, BMD, and bone strength at the distal radius were apparent in Asian (n = 91, 53 males, 38 females, [mean ± SD] 17.3 ± 1.5 years) and white (n = 89, 46 males, 43 females, 18.1 ± 1.8 years) adolescents and young adults. HR-pQCT outcomes included total BMD (Tt.BMD), trabecular bone volume fraction (BV/TV), and trabecular number (Tb.N), thickness (Tb.Th), and separation (Tb.Sp). We used an automated segmentation algorithm to determine total bone area (Tt.Ar), and cortical BMD (Ct.BMD), porosity (Ct.Po), and thickness (Ct.Th), and we applied finite element (FE) analysis to HR-pQCT scans to estimate bone strength. We fit sex-specific multivariable regression models to compare bone outcomes between Asians and whites, adjusting for age, age at menarche (girls), lean mass, ulnar length, dietary calcium intake, and physical activity. In males, after adjusting for covariates, Asians had 11% greater Tt.BMD, 8% greater Ct.BMD, and 25% lower Ct.Po than whites (p < 0.05). Also, Asians had 9% smaller Tt.Ar and 27% greater Ct.Th (p < 0.01). In females, Asians had smaller Tt.Ar than whites (16%, p < 0.001), but this difference was not significant after adjusting for covariates. Asian females had 5% greater Ct.BMD, 12% greater Ct.Th, and 11% lower Tb.Sp than whites after adjusting for covariates (p < 0.05). Estimated bone strength did not differ between Asian and white males or females. Our study supports the notion of compensatory elements of bone structure that sustain bone strength; smaller bones as observed between those of Asian origin compared with white origin have, on average, more dense, less porous, and thicker cortices. Longitudinal studies are needed to determine whether ethnic differences in bone structure exist in childhood, persist into old age, and whether they influence fracture risk.
Currently, an apparent paradox exists regarding ethnic differences in fracture rates. Although Asians have lower areal bone mineral density (aBMD, g/cm2) (by dual-energy X-ray absorptiometry [DXA]),[1-3] rates of hip and wrist fractures are lower among Asian compared with white adults,[1, 4, 5] and fracture incidence is also lower in Asian children and adolescents compared with their white peers. The ethnic difference in bone mass, already apparent during prepuberty and early puberty,[7, 8] has been attributed to the smaller bone and body size of Asians compared with whites.[1, 9, 10] However, the overall strength of bone is a function of many aspects of structural design (bone quality). Thus, Asians may differ in some attributes of bone quality (eg, bone macrostructure, bone microstructure, BMD) that influence bone strength; attributes that cannot be captured with planar DXA technology.
Recently, four studies used high-resolution pQCT (HR-pQCT) to investigate whether ethnic-specific variations in adult bone microstructure underpinned ethnic differences in fracture risk. These cross-sectional studies of premenopausal[11-13] and postmenopausal women reported smaller bone size at the distal radius and distal tibia in Asian compared with white women. However, Asian women had thicker, denser cortices and thicker trabeculae,[11, 12] which may compensate for their smaller bone size. There are no similar data (by HR-pQCT) reflecting bone quality in Asian and white men. However, at the proximal femur (assessed by QCT) Asian men had smaller bone area, thicker cortices, greater cortical bone volume, and greater total and trabecular BMD compared with white men. Importantly, whether bone macrostructure and microstructure differs between Asians and whites at adolescence and in young adulthood is not yet known.
We previously reported smaller cortical bone area and higher cortical BMD (by pQCT) at the tibial midshaft in prepubertal and early pubertal Asian girls compared with their white peers who lived in geographic proximity (Vancouver and Richmond, Canada). However, there were no differences in bone area or BMD between Asian and white boys. Thus, we conducted a cross-sectional study using a relatively novel instrument (HR-pQCT) to address the following two objectives: (1) to identify differences in bone macrostructure, microstructure, and BMD between Asian and white adolescents and young adults; and (2) to assess differences in estimates of bone compressive strength (finite element [FE] analysis using HR-pQCT images) between these same groups.
Subjects and Methods
This convenience sample included participants who were volunteers in the ongoing Healthy Bones Study III (HBSIII) and returned for measurement in Spring 2009. We describe the HBSIII cohort in detail elsewhere.[18, 19] Two hundred and twenty adolescent and young adult males and females between the ages of 14 and 22 years returned for data collection in Spring 2009. Of these participants, we included in the present analysis only those participants classified as being of “Asian” or “white” (“Caucasian”) ethnic origin based on a health history questionnaire that was previously completed by each participant's parents or guardians. Participants were classified as “Asian” (n = 97, 44%) if both parents or three of four grandparents were born in East Asian (eg, Hong Kong or China, Korea, Taiwan, Japan), Southeast Asian (eg, Philippines, Vietnam), or South Asian (eg, India) countries and as non-Hispanic “white” (n = 103, 47%) if both parents or three of four grandparents were born in North America or Europe. We also considered parental self-report of ethnicity to ensure we correctly classified each participant. We did not include participants of Hispanic (n = 2), Oceanic (n = 1), Black (n = 4), First Nations (n = 1), or of mixed ethnicity (n = 12) in this cross-sectional study.
We excluded 15 participants from the analysis: 9 participants did not have an HR-pQCT scan either due to concerns regarding previous radiation exposure from clinical scans (n = 2 white females) or lack of parental consent for radius scans (n = 7; 4 white males, 2 Asian males, 1 Asian female), 3 white males did not have a whole-body DXA scan, 2 participants (1 white male and 1 white female) presented with diseases known to affect bone metabolism (osteogenesis imperfecta, fetal alcohol syndrome), and 1 white female reported using oral corticosteroids for longer than 3 months prior to data collection. All other participants were included in the analysis. Thus, the present analysis includes 94 Asian (males = 53, females = 41) and 91 white (males = 48, females = 43) adolescents and young adults. All participants visited the Centre for Hip Health and Mobility, affiliated with the Vancouver Costal Health Research Institute, to complete their measurements. The UBC Clinical Research Ethics Board approved all procedures and we obtained written informed consent from all parents and participants.
Anthropometry and body composition
We measured standing height to the nearest 0.1 cm and weight to the nearest 0.1 kg using a wall-mounted digital stadiometer (Seca Model 242; Seca, Hanover, MD, USA) and an electronic scale (Seca Model 840; Seca), respectively. We measured ulnar length as the distance from the distal medial edge of the ulna to the proximal olecranon process with a standardized anthropometric tape. For each variable, we used the mean of two or the median of three measures for analyses. We obtained measures of lean mass (kg), fat mass (kg), and percent (%) body fat from a whole-body DXA scan (Hologic QDR 4500 W; Hologic, Inc., Waltham, MA, USA). Quality assurance scans were performed daily and qualified technicians conducted all measures following standard manufacturer's procedures (Hologic, Inc.). In our laboratory, short-term precision for body composition variables was determined based on three whole-body scans done with repositioning on 17 young adults. The coefficients of variation (CV%) for lean mass, fat mass, and percent body fat ranged from 0.3% to 1.9% (UBC Bone Health Research Group; unpublished data).
As in previous studies,[20, 21] we administered the Physical Activity Questionnaire for Adolescents (PAQ-A) to estimate daily physical activity (min/d) in the moderate to vigorous range over the previous 7 days. All participants completed a food frequency questionnaire to estimate dietary intake of calcium (mg/d). History of oral contraceptive pill use and current smoking status were assessed using a health history questionnaire. We assessed maturity as per the method of Tanner (self-reported pubic hair stage for males and breast stage for females). We also used a self-report questionnaire to determine age at menarche (years) for females, who were all postmenarcheal.
Bone microstructure and density measurements
As described elsewhere,[25, 26] we used HR-pQCT (XtremeCT; Scanco Medical AG, Brüttisellen, Switzerland) to assess bone macrostructure, microstructure, and BMD at the nondominant distal radius. In the case of a prior fracture, we scanned the contralateral limb (n = 8). We acquired an anteroposterior scout view to identify the region of interest (ROI) at the radius. The technologist placed a reference line at the medial edge of the distal radius and defined the ROI as the region proximal to the reference line and equivalent to 7% of the total ulnar length. We used 60 kVp effective energy, 900 µA, matrix size of 1536 × 1536, and 100 ms integration time to acquire 110 slices of the radius at an 82-µm nominal isotropic resolution. The first three and last three slices were excluded from the analysis as per the manufacturer's default. The effective dose equivalent for the radius scan is less than 3 µSv per measurement, with a measurement time of 2.8 minutes. Quality control was monitored daily for density fluctuation and weekly for geometry using a calibration phantom provided by the manufacturer.
A single qualified technician scored all HR-pQCT images for motion artifacts using a grading scale proposed by the system manufacturer (1 = no motion, up to 5 = medium-large streaks/discontinuities). We excluded scans graded as 4 or 5 from our analysis. We segmented all images using the standard manufacturer's in vivo protocol. A threshold-based algorithm was used to separate cortical and trabecular bone compartments. The threshold used to differentiate cortical from trabecular bone was set to one-third of the apparent cortical bone density. We used the standard morphologic analysis to determine total BMD (Tt.BMD, mg hydroxyapatite [HA]/cm3) and trabecular BMD (Tb.BMD, mg HA/cm3). Trabecular bone volume fraction (BV/TV) is derived by dividing Tb.BMD by the assumed density of fully mineralized bone (1200 mg HA/cm3). Trabecular number (Tb.N, 1/mm) is defined as the inverse of the mean spacing of the three-dimensional (3D) ridges. Trabecular thickness (Tb.Th, mm) is derived as (BV/TV)/Tb.N and trabecular separation (Tb.Sp) is derived as (1–BV/TV)/Tb.N. In our laboratory, reproducibility is less than 3.8% for all parameters at the distal radius.
We also used an automated segmentation algorithm to determine total bone cross-sectional area (Tt.Ar, mm2) and cortical bone parameters.[32, 33] Outcomes included cortical porosity (Ct.Po, %) calculated as the number of void voxels within the cortex, cortical bone mineral density (Ct.BMD, mg HA/cm3) and cortical thickness (Ct.Th, mm) defined as the mean cortical volume divided by the outer bone surface.
We applied FE analysis to HR-pQCT images to estimate bone strength. As previously reported, we generated FE meshes from the 3D HR-pQCT images using the voxel conversion approach.[34, 35] We simulated uniaxial compression on each radius section up to 1% strain. A single homogenous tissue modulus of 6829 MPa and a Poisson's ratio of 0.3 were applied to all elements. We used a custom FE solver (FAIM, Version 4.0; Numerics88Solutions, Calgary, Canada) on a desktop workstation (Mac Pro, OSX, Version 10.5.6; Apple Inc., Cupertino, CA, USA; 2 × 2.8 GHz Quad-Core Intel Xenon) to solve the finite-element models. We estimated apparent bone strength (ultimate stress, MPa) using the stiffness value and failure load (N) using the Pistoia criterion. To estimate the risk of forearm fracture, we calculated the load-to-strength ratio[38, 39] based on a participant-specific fall force that incorporates standing height.
We performed separate analyses for male and female participants due to known sex differences in bone accrual and in the tempo and timing of growth and maturation. We used scatter plots and box plots to identify possible data errors and outliers prior to data analyses. We compared descriptive and anthropometric characteristics between Asian and white participants using Student's t tests for continuous variables and Pearson's chi-squared tests for categorical variables. We also used Student's t tests to compare unadjusted bone outcomes between Asians and whites by sex. We then fit multivariable regression models to compare bone macrostructure and microstructure, BMD, and bone strength between Asians and whites after adjusting for relevant covariates. As in previous studies,[16, 18] we chose covariates based on their known biological and biomechanical relationships with BMD, bone structure, and bone strength. Covariates for males and females were age, lean mass (an estimate of muscle force), ulnar length (an estimate of moment arm length), dietary calcium intake, and physical activity. In addition, we also included age at menarche in the regression models for females. Because the load-to-strength ratio is based on a participant-specific fall force that incorporates the participant's height, we did not include ulnar length (strongly correlated with height) as a covariate in the regression model for the load-to-strength ratio. We used histograms, inverse normal plots, and scatter plots to check model assumptions. We also identified influential data points using Cook's distance values. We performed all analyses using Stata, Version 10.1 (StataCorp, College Station, TX, USA) and we set the level of significance at p < 0.05.
We provide descriptive characteristics for Asian and white participants in Table 1. We excluded 5 participants (white males: n = 2; Asian females: n = 3) from all analyses due to significant motion artifacts associated with HR-pQCT scans (motion grade of 4 or 5). The majority (n = 63, 69%) of Asian participants were of East Asian (Chinese) origin and a smaller percentage were either of Southeast Asian (n = 12, 13%) or South Asian descent (n = 13, 14%). We also had a small number of participants (n = 3, 3%) whose ethnic origin was a mix of East and Southeast Asian.
|Age (years)||17.2 (1.5)||17.7 (1.8)||0.167||17.4 (1.5)||18.5 (1.8)||0.003|
|Height (cm)||173.1 (6.1)||178.2 (8.0)||<0.001||158.5 (6.0)||166.5 (5.9)||<0.001|
|Ulnar length (mm)||284 (15)||290 (15)||0.045||251 (12)||262 (13)||<0.001|
|Weight (kg)||67.6 (16.5)||69.3 (13.2)||0.578||55.1 (8.7)||63.7 (12.8)||<0.001|
|Lean mass (kg)||51.6 (8.4)||54.2 (8.7)||0.139||35.7 (4.1)||41.6 (5.5)||<0.001|
|Fat mass (kg)||11.8 (9.2)||10.7 (6.3)||0.505||16.3 (5.3)||18.4 (7.8)||0.153|
|Percent body fat (%)||16.1 (7.3)||14.9 (5.9)||0.365||28.9 (5.3)||27.9 (6.4)||0.415|
|Physical activity (min/d)||72.8 (48.0)||84.4 (85.1)||0.414||53.7 (42.1)||60.8 (46.1)||0.478|
|Dietary calcium (mg/d)||854 (684)||1431 (848)||<0.001||735 (488)||1094 (655)||0.007|
|Current smoker (%)||3 (6%)||6 (13%)||0.096||4 (10%)||5 (13%)||0.725|
|OCP use (%)||—||—||—||5 (13%)||11 (30%)||0.161|
|Tanner 3/4/5 (n)c||0/29/24||0/8/37||<0.001||8/14/16||4/12/27||0.135|
|Age at menarche (years)||—||—||—||12.0 (1.3)||12.8 (1.2)||0.004|
Asian males were 5 cm shorter and had 6 mm shorter ulnar length than white males. Asian males were less mature as indicated by the smaller number of them in Tanner stage 5 (46%) compared with white males (82%). Asian males also reported significantly lower dietary calcium intake than white males. There were no significant differences in age, weight, lean mass, fat mass, % body fat, physical activity, or proportion of current smokers between Asian and white males.
Asian females were slightly younger, 8 cm shorter, 8.6 kg lighter, and had 5.9 kg less lean mass and 11 mm shorter ulnar length compared with white females. Dietary calcium intake was also significantly lower among Asian compared with white females. Asian females reported an earlier age at menarche and less oral contraceptive pill use than white females. Fat mass, % body fat, physical activity, proportion of current smokers, and Tanner stage were not significantly different between Asian and white females.
We present unadjusted and adjusted HR-pQCT outcomes for Asian and white males and females in Table 2 and Fig. 1. For males, Asians had 11% smaller Tt.Ar and 16% greater Ct.Th and these differences were maintained after adjusting for covariates (9% smaller Tt.Ar and 27% greater Ct.Th). Unadjusted values of Tt.BMD, Ct.BMD, and Ct.Po were not significantly different between Asians and whites. However, after adjusting for covariates, Asians had 11% greater Tt.BMD, 8% greater Ct.BMD, and 25% lower Ct.Po than whites. Trabecular microstructure outcomes (BV/TV, Tb.N, Tb.Th, and Tb.Sp) and FE variables (failure load, ultimate stress, and load-to-strength ratio) (Table 3) were not significantly different between Asian and white males before or after adjusting for covariates.
|Tt.Ar (mm2)||269.7 (46.6)||300.6 (46.5)||0.001||204.3 (21.1)||237.0 (34.0)||<0.001|
|Tt.BMD (mg HA/cm3)||356.3 (70.8)||334.2 (72.3)||0.128||368.8 (64.3)||346.9 (54.7)||0.206|
|Ct.BMD (mg HA/cm3)||764.1 (78.0)||730.9 (112.2)||0.088||832.5 (58.5)||812.6 (64.9)||0.153|
|BV/TV||0.161 (0.033)||0.167 (0.027)||0.327||0.136 (0.034)||0.144 (0.028)||0.208|
|Tb.N (1/mm)||1.91 (0.27)||1.98 (0.21)||0.196||1.80 (0.29)||1.94 (0.22)||0.017|
|Tb.Th (mm)||0.084 (0.014)||0.085 (0.014)||0.783||0.075 (0.014)||0.074 (0.010)||0.737|
|Tb.Sp (mm)||0.449 (0.079)||0.427 (0.058)||0.123||0.494 (0.093)||0.449 (0.064)||0.011|
|Ct.Th (mm)||0.96 (0.25)||0.83 (0.34)||0.033||0.94 (0.17)||0.88 (0.16)||0.146|
|Ct.Po (%)||3.3 (1.8)||3.8 (2.4)||0.222||1.2 (0.6)||1.3 (1.1)||0.354|
|Tt.Ar (mm2)||272.4 (260.3–284.5)||297.5 (284.4–310.6)||0.009||215.2 (205.7–224.7)||227.4 (218.6–236.2)||0.098|
|Tt.BMD (mg HA/cm3)||363.0 (346.7–379.5)||326.4 (308.7–344.1)||0.005||363.0 (342.5–383.5)||347.6 (328.6–366.6)||0.331|
|Ct.BMD (mg HA/cm3)||775.0 (756.6–793.4)||718.3 (698.5–738.2)||<0.001||843.3 (824.5–862.1)||803.0 (785.6–820.4)||0.006|
|BV/TV||0.162 (0.154–0.169)||0.166 (0.158–0.175)||0.443||0.132 (0.121–0.143)||0.147 (0.137–0.158)||0.076|
|Tb.N (1/mm)||1.92 (1.85–1.98)||1.97 (1.90–2.04)||0.321||1.80 (1.71–1.89)||1.94 (1.85–2.02)||0.056|
|Tb.Th (mm)||0.085 (0.081–0.088)||0.085 (0.081–0.089)||0.823||0.073 (0.069–0.077)||0.076 (0.072–0.080)||0.375|
|Tb.Sp (mm)||0.448 (0.430–0.466)||0.429 (0.409–0.449)||0.183||0.496 (0.467–0.525)||0.447 (0.421–0.474)||0.030|
|Ct.Th (mm)||1.00 (0.93–1.06)||0.79 (0.72–0.86)||<0.001||0.96 (0.91–1.02)||0.86 (0.81–0.91)||0.017|
|Ct.Po (%)||3.2 (2.7–3.6)||4.0 (3.5–4.5)||0.029||1.2 (0.8–1.4)||1.4 (1.1–1.6)||0.346|
|Predicted failure load (N)||2580 (621)||2713 (657)||0.305||1898 (377)||2048 (300)||0.050|
|Ultimate stress (MPa)||39.9 (11.3)||37.4 (11.2)||0.264||37.8 (10.1)||35.4 (8.3)||0.246|
|Estimated fall load (N)||2761 (48)||2801 (63)||<0.001||2642 (51)||2709 (48)||<0.001|
|Load-to-strength ratio||1.14 (0.30)||1.09 (0.26)||0.420||1.45 (0.29)||1.35 (0.20)||0.083|
|Predicted failure load (N)||2653 (2520–2787)||2628 (2484–2773)||0.808||1961 (1842–2080)||1992 (1881–2102)||0.739|
|Ultimate stress (MPa)||40.7 (37.9–43.6)||36.4 (33.3–39.4)||0.050||37.0 (33.7–40.3)||36.0 (33.0–39.1)||0.702|
|Load-to-strength ratio||1.10 (1.03–1.16)||1.13 (1.06–1.20)||0.477||1.41 (1.32–1.50)||1.38 (1.30–1.46)||0.654|
For females, Asians had 16% smaller Tt.Ar than whites before adjusting for covariates. However, after adjusting for covariates, Tt.Ar was not significantly different between groups. Unadjusted values of Tt.BMD, Ct.BMD, and Ct.Th were not significantly different between Asian and white females; however, after adjusting for covariates, Ct.BMD and Ct.Th were 5% and 12% greater in Asian than white females, respectively. We did not observe any significant differences in BV/TV or Tb.Th between Asian and white females. However, unadjusted values of Tb.N and Tb.Sp were 8% to 10% lower in Asian females, and the difference in Tb.Sp was maintained after adjusting for covariates. There were no significant differences in any FE outcomes between Asian and white females (Table 3).
Our study of adolescents and young adults builds upon the few previous studies that evaluated differences in bone macrostructure, microstructure, BMD, and estimated bone strength (by HR-pQCT) in Asian and white adult women.[11-14] We extend this literature further by evaluating cortical porosity using a customized measure and estimating bone strength using FE analysis in both males and females. Thus, we addressed the challenge of assessing key structural elements to better understand their contribution to bone strength.
Before discussing our findings in detail, we wish to acknowledge the potential limitations associated with our ethnicity classification. Although we used this method in previous studies,[8, 18, 19] we recognize that our broad groupings of Asian and white may mask important variability in bone outcomes associated with geographic, biological, environmental, and cultural factors.[41, 42] Previous studies of ethnic differences in bone microstructure focused on individuals of one ethnic origin such as Chinese,[11, 13] and thus targeted recruitment at this ethnic group (along with whites). In the present study, we invited all students to participate regardless of ethnic origin, and thus, our sample reflects the ethnic diversity of Metro Vancouver, where visible minority groups represent 41.7% of the population (compared with 16.2% in the rest of Canada). Although the majority (69%) of our Asian participants were of East Asian (Chinese) descent, we also included participants of Southeast Asian and South Asian origin in our analysis.
To our knowledge, this is the first study to compare BMD, bone macrostructure and microstructure and estimated bone strength between Asian and white males. However, the ethnic differences in bone size and cortical thickness we observed in males are similar, in part, to QCT results from the proximal femur in older men (≥65 years) in the Osteoporotic Fractures in Men (MrOS) study. Although femoral neck cortical thickness remained 5% greater in Asian men compared with whites after adjusting for age and body size, size-adjusted femoral neck cross-sectional area was not significantly different between Asian and white older men. In addition, femoral neck cortical BMD did not differ between Asian and white men. In our previous pQCT study, white boys aged 9 to 11 years had slightly greater cortical area (3%) at the tibial midshaft (50% site) compared with Asian boys even after adjusting for limb length and muscle cross-sectional area. We also observed a nonsignificant trend for higher cortical BMD (∼1%) in Asian compared with white boys. Despite the variability across studies in terms of skeletal site assessed, bone imaging technology, and study population, these findings suggest that there may be significant differences in bone size and cortical bone characteristics between Asian and white males. Further study is needed to confirm these trends, and to uncover potential mechanisms underpinning these ethnic differences.
As mentioned, we extended traditional measures of cortical BMD and thickness by assessing cortical porosity, a significant predictor of bone strength. Asian males had less porous cortices than white males; however, this difference was attenuated after adjusting for age and other covariates. The higher porosity in white males may be related to the slightly greater amount of exercise (∼14 min/d) that white males reported participating in compared with Asian males. Although the difference in physical activity between ethnicities did not achieve statistical significance, we showed in a younger age group that even small bouts of high-impact exercise (only a few minutes per day) can substantially enhance bone mass in boys and girls and bone strength in boys. White males' significantly greater intake of dietary calcium may have served as a modulating factor as per the functional model of bone development.
Our findings in females align with recent HR-pQCT findings in premenopausal[11, 12] and postmenopausal women at the distal radius and also with our earlier pQCT findings at the tibial shaft in prepubertal and early pubertal girls. That is, the pattern whereby Asian females had smaller bone area but thicker more dense cortices compared with whites appears to persist across the lifespan, and may also be present at multiple skeletal sites. Despite ethnic differences in cortical BMD and bone structure (Ct.Th) having been described,[11, 12, 14, 16] the mechanisms that underpin them are not well understood. Although we did not assess these factors, it is likely that both genetics and the endocrine environment play a role. If estrogen levels were higher in Asian compared with Caucasian females as has been reported previously for premenopausal women, this could partially explain the differences we observed. That is, this hormonal difference might theoretically stimulate endocortical apposition, decrease intracortical remodeling, and result in thicker, more dense, and less porous cortices. The actual influence of estrogen and other factors on cortical bone in Asian compared with white youth and young adults requires direct measure of these factors in future studies.
We know much less about ethnic differences in trabecular microstructure and estimated bone strength because few studies have reported these outcomes, and none compared these differences between Asian and white youth and young adults. In male participants, there were no apparent ethnic differences for any measure of trabecular microstructure. However, when we compared trabecular microstructure outcomes between males of East Asian descent (the majority of our Asian cohort) and white males, we noted that East Asian males had lower BV/TV, lower Tb.N, and greater Tb.Sp (data not shown). These differences in trabecular microstructure are similar to what we observed in females, because Asian females had a smaller number of more widely spaced trabeculae compared with white females. However, we did not observe any ethnic differences in trabecular thickness as was reported in recent HR-pQCT studies. For example, Wang and colleagues assessed the distal radius in Chinese and white premenopausal women aged 18 to 45 years, living in Australia. Chinese women had fewer trabeculae of greater thickness, on average, compared with white women. In contrast, Walker and colleagues reported no differences in trabecular microstructure at the distal radius in Asian American compared with white American premenopausal women aged 29 to 40 years. Discrepancies across studies likely reflect differences in the age, ethnic origin, and geographic location of study participants.
Despite the observed ethnic differences in bone macrostructure and microstructure, and BMD, estimates of bone strength and fracture risk (load-to-strength ratio) did not differ between Asians and whites with the exception of ultimate stress in males, which trended higher in Asians (12%, p = 0.05). Overall, our bone strength findings do not align with the recent results of Wren and colleagues who reported significantly lower fracture risk among Asian children and youth aged 6 to 17 years compared with their white peers. There are several possible explanations for this result. First, resistance to compressive loads at the distal radius (estimated with our FE analysis) may be similar in a larger bone with a thinner, less dense cortex compared with a smaller bone with a thicker, more dense cortex because both bone size and geometric properties of the cortex are predictors of distal radius failure load. Second, limitations associated with our FE analysis may have masked any potential ethnic differences in bone strength. For example, the FE model estimates bone strength under uniaxial compression. Thus, it is possible that ethnic differences in bone strength may be more apparent under bending or torsional loads, or a combination of several loading conditions. In addition, we used a homogenous tissue modulus in our FE models. Had we used a scaled model in which the CT attenuation values were converted to tissue moduli according to previously established relationships, we may have been able to distinguish differences in bone strength between Asians and whites. However, scaled FE models require a significantly longer amount of time to solve and were thus, not feasible for this study. Finally, we may have been underpowered to detect ethnic differences in FE outcomes due to our relatively small sample size combined with the higher variability in failure load and ultimate stress (15% to 30%) compared with the variability in Tt.Ar and Ct.BMD (7% to 18%).
We acknowledge several additional limitations or our study. First, the cross-sectional study design did not allow us to examine antecedents of ethnic differences in bone structure. Longitudinal data would clarify differences in bone accrual and growth velocities between Asian and whites as reported in other studies.[10, 18, 51] Second, we acknowledge the recruitment bias introduced by accessing this convenience sample of children and the limitations related to generalizing our results to a wider population of children, adolescents, and young adults. Finally, the customized algorithm we used to quantify Ct.Po cannot detect pores smaller than 82 µm. As smaller pore size may be more prevalent in younger people, we may have underestimated Ct.Po in the youngest members of our cohort.
In conclusion, our study supports the notion of compensatory elements of bone structure that sustain bone strength; whereby smaller bones as observed between those of Asian compared with white origin, have, on average, more dense, less porous and thicker cortices. Conducting longitudinal studies across a broader age range would be important to determine whether ethnic differences in bone macrostructure and microstructure exist in childhood, persist into old age and whether they influence fracture risk.
All authors state that they have no conflicts of interest.
This work was supported by a grant from Canadian Institutes of Health Research (MOP-84575). We gratefully acknowledge the students, staff, and parents for their participation in the Healthy Bones III Study; without their dedication, this study would not have been possible. We also thank Drs. Steven Boyd and Kyle Nishiyama at the University of Calgary for their assistance with the autosegmentation and finite element analyses. Finally, we thank all staff at the Centre for Hip Health and Mobility for their ongoing support.
Authors' roles: Study design: SK, HMM, LN, and HAM. Study conduct: HAM. Data collection: HAM. Data analysis: SK, HMM, and LN. Data interpretation: SK, HMM, LN, and HAM. Drafting manuscript: SK and HMM. Revising manuscript content: SK, HMM, LN, and HAM. Approving final version of manuscript: SK, HMM, LN, and HAM. HAM takes responsibility for the integrity of the data analysis.