SEARCH

SEARCH BY CITATION

Keywords:

  • HIGH-RESOLUTION PQCT;
  • BONE MICROARCHITECTURE;
  • BONE STRENGTH;
  • AGING

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References

In this cross-sectional study, we aimed to predict age-related changes in bone microarchitecture and strength at the distal radius (DR) and distal tibia (DT) in 644 Canadian adults (n = 442 women and 202 men) aged 20 to 99 years. We performed a standard morphologic analysis of the DR and DT with high-resolution peripheral quantitative computed tomography (pQCT) and used finite-element analysis (FEA) to estimate bone strength (failure load) and the load distribution. We also calculated a DR load-to-strength ratio as an estimate of forearm fracture risk. Total bone area, which was 33% larger in young men at both sites, changed similarly with age in women and men at the DT but increased 17% more in men than in women at the DR (p < .001). Trabecular number and thickness (Tb.Th) were 7% to 20% higher in young men than in young women at both sites, and with the exception of Tb.Th at the DR, which declined more with age in men (−16%) than in women (−2%, p < .01), the age-related decline in these outcomes was similar in women and in men. In the cortex, porosity (Ct.Po) was 31% to 44% lower in young women than in young men but increased 92% to 176% more with age in women than in men (p < .001). The DR cortex carried 14% more load in young women than in young men, and the percentage of load carried by the DR cortex did not change with age in women but declined by 17% in men (p < .01). FEA-estimated bone strength was 34% to 47% greater in young men, but the predicted change with age was similar in both sexes. In contrast, the load-to-strength ratio increased 27% more in women than in men with age (p < .01). These results highlight important site- and sex-specific differences in patterns of age-related bone loss. In particular, the trends for less periosteal expansion, more porous cortices, and a greater percentage of load carried by the DR cortex in women may underpin sex differences in forearm fracture risk. © 2011 American Society for Bone and Mineral Research.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References

Currently, our understanding of the trajectory of age-related bone loss that contributes to osteoporosis and fragility fractures is based primarily on areal bone mineral density (aBMD) measurements in dual-energy X-ray absorptiometry (DXA) studies. Although aBMD (g/cm2) is a significant predictor of fracture risk,1 2D DXA technology is limited because it is unable to distinguish between cortical and trabecular bone and cannot quantify bone macro- and microstructural properties such as bone geometry and microarchitecture that influence whole-bone strength. This is illustrated by epidemiologic data that indicate that half of incident fractures occur in women whose aBMD values fall above the threshold for osteoporosis by World Health Organization (WHO) criteria.2, 3

The recent introduction of high-resolution peripheral quantitative computed tomography (HR-pQCT) allows for precise in vivo assessment of bone microarchitecture and volumetric BMD (vBMD). With a nominal isotropic resolution of 82 µm, HR-pQCT provides 3D macro- and microstructural data for both the cortical and trabecular compartments of the distal radius and distal tibia.4, 5 In cross-sectional studies, HR-pQCT detected significant differences in cortical6, 7 and trabecular8 structure between women with normal bone mass and those with low bone mass (by DXA) as well as between postmenopausal women with and without a history of fragility fractures.8–10 In addition to standard morphologic analyses, the application of finite-element analysis (FEA) to 3D HR-pQCT scans permits estimation of subject-specific bone strength,11 which was found recently to be associated with distal radius fractures in postmenopausal women independent of aBMD.12

To date, two studies have investigated age-related variation in HR-pQCT outcomes in women and men.13, 14 Khosla and colleagues13 performed the earliest population-based HR-pQCT study using a prototype HR-pQCT machine (XtremeCT, Scanco Medical, Brüttisellen, Switzerland) and reported results for the radius only. Subsequently, Dalzell and colleagues14 used HR-pQCT (XtremeCT, Scanco Medical) to assess both the distal radius and the distal tibia. Although Dalzell and colleagues14 provided a reference data set for both bone microarchitecture and FEA-derived estimates of bone strength, the sample size across a large age range was small (n = 132, 20 to 79 years) compared with that of Khosla and colleagues (n = 602, 21 to 97 years).13 Thus there is a need for a comprehensive HR-pQCT reference data set including outcomes for both the radius and tibia to establish the age-related variation in bone quality outcomes across adulthood in both women and men. The primary aim of this study was to describe age- and sex-related differences in bone macro- and microstructure and FEA-derived bone strength of both the distal radius and distal tibia in a population-based sample of women and men across a wide age range. This study will provide the first HR-pQCT reference data for Canadian adults with which we can compare bone quality and strength outcomes with those reported for adults in the United States13 and the United Kingdom.14

Materials and Methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References

Participants

We recruited participants from the Calgary, Alberta, cohort of the Canadian Multicentre Osteoporosis Study (CaMos). Briefly, CaMos is a 10-year prospective population-based study of more than 9000 women and men aged 25 years and older living within 50 km of nine Canadian cities.15 Participants were originally recruited between 1997 and 1998 using a stratified random-sampling technique that is described in detail elsewhere.15 At the time of the 10-year follow-up, we invited individuals from the Calgary cohort to participate in this HR-pQCT study. Of the 621 participants in the Calgary cohort at the 10-year follow-up (58% of the original Calgary cohort), 396 individuals (64%; 271 women, 125 men, 35 to 99 years of age) volunteered. We also recruited individuals from the Calgary CaMos youth cohort (16 to 25 years of age)16 within 1 year of their 2-year CaMos follow-up. Of 73 individuals, 31 (42%; 13 women, 15 men, 20 to 28 years of age) volunteered. To recruit more participants in the younger age strata (<60 years of age), we used three additional sampling strategies: (1) random sampling using the CaMos sampling technique,15 (2) snowball recruitment, whereby current volunteers were asked to provide names of friends and family in the target age range who might be interested in participating, and (3) advertising with posters. A total of 227 individuals (158 women, 62 men, 20 to 60 years of age) volunteered. Therefore, this sample included a total of 644 individuals (442 women, 202 men) aged 20 to 99 years. For this analysis we aimed to describe population-level variation in bone microarchitecture. Therefore, we did not exclude individuals based on self-reported clinical characteristics such as medication use (ie, glucocorticoids or bisphosphonates) or history of disease. We provide the number of participants with these clinical characteristics, as well as the results of secondary analyses from which these individuals were excluded. Approval for all procedures was obtained from the Conjoint Health Research Ethics Board at the University of Calgary, and all participants provided written informed consent.

Questionnaires and clinical assessments

We obtained sociodemographic information and medical and fracture histories from the extensive interviewer-administered CaMos questionnaire. For the CaMos adult cohort, the questionnaire was administered at the time of the 10-year follow-up visit that occurred, on average, 8 ± 4 months prior to the HR-pQCT scan. For the CaMos youth cohort, the questionnaire was administered at the time of the 2-year follow-up visit that occurred, on average, 12 ± 6 months prior to the HR-pQCT scan. For the new recruits, the questionnaire was administered within 1 month of the HR-pQCT scan.

Standard CaMos clinical assessments included height, weight, and aBMD (g/cm2) measurements of the lumbar spine, total proximal femur, and femoral neck by DXA (Hologic QDR4500, Hologic, Bedford, MA). The DXA scans were acquired at the time of the 10- and 2-year follow-up visits for the CaMos adult and youth cohorts, respectively, and within 1 month of the HR-pQCT scans for the new recruits. Trained X-ray technicians acquired all scans, and quality assurance tests were performed daily. DXA scans of the CaMos adult and youth cohorts were analyzed by one technician at the CaMos coordinating center, and DXA scans of the new recruits were analyzed by three technicians (EFW Radiology, Calgary, Alberta, Canada).

High-resolution pQCT

We scanned the nondominant forearm (or the nonfractured forearm in the case of a previous wrist fracture) and the left tibia using HR-pQCT (XtremeCT, Scanco Medical) with the standard human in vivo scanning protocol (60 kVp, 1000 µA, 100-ms integration time). Prior to scan acquisition, the subject's arm or leg was immobilized in an anatomically formed brace provided by the manufacturer. The region of interest was defined using a 2D scout scan, on which the reference line was placed manually at the endplate of the radius or tibia (Fig. 1). The first of 110 slices was acquired 9.5 and 22.5 mm proximal to the reference line for the distal radius and tibia, respectively, with a nominal isotropic resolution of 82 µm. All scans were acquired by one of two trained technicians using standard positioning in order to minimize the influence of positioning errors on morphologic outcomes.17 The technicians also carefully examined each scan for motion artifacts, and in the event of significant motion artifacts (eg, blurring or discontinuities), the technician performed a second scan. Quality control was monitored daily using a daily calibration phantom provided by the manufacturer.

thumbnail image

Figure 1. Scout views representing the reference line (solid line) and scanned region (dashed lines) at the distal radius (A) and distal tibia (B).

Download figure to PowerPoint

All HR-pQCT images were scored for motion artifacts according to a motion scale of 0 (no motion) to 4 (significant blurring of the periosteal surface, discontinuties in the cortical shell, or streaking in the soft tissue).18 For this analysis, we excluded scans that were scored as 4 on our motion scale. The HR-pQCT images were analyzed by one of two trained technicians using the standard manufacturer's method5 that we19 and others8, 13 have described in detail. Briefly, a semiautomated, hand-drawn contouring scheme was used to segment the periosteal surface of the radius, and this was followed by a threshold-based algorithm that separated the cortical and trabecular bone compartments.5 From the standard morphologic analysis, we obtained measures of total BMD (BMDtotal mg HA/cm3) and trabecular BMD (BMDtrab). The ratio of trabecular bone volume to total bone volume (BV/TV) is derived by dividing BMDtrab by the assumed density of fully mineralized bone (1200 mg HA/cm3), and trabecular number (Tb.N, mm−1) is determined using ridge-extraction methods.20 Trabecular thickness (Tb.Th, mm) is derived as (BV/TV)/Tb.N, and trabecular separation (Tb.Sp, mm) is derived as (1 – BV/TV)/Tb.N according to standard morphologic relationships.21 These standard HR-pQCT morphologic outcomes were validated against “gold standard” micro–computed tomographic (µCT) imaging,19 and in our laboratory, differences in in vivo short-term reproducibility were less than 1% for density measures and less than 4.5% for morphologic measures.22

In addition to the standard morphologic analysis, we also used an automated segmentation algorithm to determine total and cortical bone cross-sectional areas (Tt.Ar, Ct.Ar, mm2), cortical porosity (Ct.Po, %), cortical thickness (Ct.Th, mm), and density (mg HA/cm3).6, 23 With this algorithm, we calculated cortical porosity as the number of void voxels in each thresholded cortex image divided by the total number of voxels in the cortex using Image Processing Language (IPL, Version 5.08b, Scanco Medical). We quantified cortical thickness after removing the intracortical pores from the thresholded cortex image using a distance transform,24 and we calculated cortical density as the average mineral density in the region defined by the autosegmentation cortical bone mask. The automated segmentation algorithm is robust for identification of cortical bone, particularly when the cortex is thin, and was found recently to have high reproducibility (coefficient of variation < 1.5%).25

Finite element analysis

As in previous analyses,11 we generated linear, homogeneous finite-element meshes from the 3D HR-pQCT image data using the voxel-conversion approach.26, 27 The boundary conditions represented a uniaxial compression test with the nodes at the bottom platen surface fixed in the Z direction (ie, uniaxial testing direction). The nodes at the top and bottom surfaces were unconstrained in X and Y directions, whereas a displacement was applied in the Z direction that corresponded to 1% strain. A single homogeneous tissue modulus of 6829 MPa and a Poisson's ratio of 0.3 were applied to all elements.11

We solved the finite-element models using custom large-scale FEA software (FAIM, Version 4.0)28 on a desktop workstation (Mac, OS X, Version 10.5.6; 2 × 2.8 GHz Quad-Core Intel Xeon). The meshes generated for the radius and tibia scans resulted in models with approximately 1 and 2.5 million elements, respectively. Using custom software, the average solution time was 20 minutes for radius models and 60 minutes for tibia models. Outcomes included stiffness (N/mm), calculated as the reaction force (RFz) determined by the FEA model at 1% strain divided by the average bone cross-sectional area from the morphologic analysis. We used the stiffness value to estimate apparent bone strength (ultimate stress, MPa).11 In our laboratory, in vivo reproducibility for the stiffness measure is less than 3.5%.22 We also calculated failure load (N) using the Pistoia criterion29 and determined the percentage of the total load that is carried by the cortex at the most distal slice.11

Load-to-strength ratio

To estimate forearm fracture risk, we calculated the load-to-strength ratio (Φ)30, 31 as the ratio of the estimated load applied to the outstretched hand during a fall from standing height (subject-specific fall force, estimated with a forward dynamics simulation model that incorporates subject height and weight32) to the FEA-derived failure load. Theoretically, the biomechanical fracture threshold occurs at Φ = 1.31

Statistical analysis

To investigate the relationship between age and HR-pQCT outcomes in women and men, we used Pearson and Spearman correlations depending on the distribution of the HR-pQCT variables and linear regression. In each regression model we evaluated possible nonlinear associations with age using a quadratic age term (age2). Model assumptions were checked graphically (histograms, scatter plots, and quantile-quantile plots), and variables were transformed where necessary. We included a sex × age interaction term in the regression model in order to compare slopes between women and men, and we used age centered (age – mean age) to reduce the influence of collinearaity in the higher-order (quadratic) models. Similar to the analysis of Khosla and colleagues,13 we used the predicted values from each regression model to estimate change in each HR-pQCT outcome between 20 and 90 years in both sexes. Importantly, since this is a cross-sectional study, we were unable to determine true age-related changes in HR-pQCT outcomes, and we could not account for the possible influence of secular changes in body and bone size. However, in keeping with previous studies,13, 33 we aim to use our cross-sectional data to create a trajectory over age and predict age-related changes in bone outcomes. In addition to the regression analyses, we used Student's t tests to compare bone outcomes between young-adult women and men (20 to 29 years of age). To account for significant associations between bone size (total and cortical bone area) and height (data not shown), we performed the analyses using both unadjusted values of bone area and bone area adjusted for height. All analyses were performed with and without the inclusion of the participants who reported current use of glucocorticoids and/or antiresorptives (ie, bisphosphonates, selective estrogen receptor modulators, hormone-replacement therapy, etc.). We report the results of the subgroup analysis (without inclusion of medication users) in the text. All analyses were performed in Stata, Version 10.0 (StataCorp, College Station, TX, USA), and in light of the large number of regression models, we considered results statistically significant at p < .01.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References

We provide descriptive characteristics of the cohort in Table 1 and the age distribution in Table 2. After exclusion of two subjects (1 woman, 1 man) who had unacceptable HR-pQCT scans of both the radius and the tibia owing to motion artifacts, the final sample included 441 women and 201 men. Among the 642 participants, 614 (418 women, 196 men) had acceptable radius and tibia scans, 10 (7 women, 3 men) had acceptable radius scans only, and 18 (16 women, 2 men) had acceptable tibia scans only. Most of participants were white (94%), 3.5% were Asian, fewer than 1% were black, and 2.5% were of other ethnicities. Most of the women (60%) were postmenopausal, and of these women, 27% reported current use of bisphosphonates, fewer than 2% reported current use of a selective estrogen receptor modulator (SERM), and 9% reported current use of estrogen therapy. Fewer than 5% of the cohort reported current use of glucocorticoids.

Table 1. Descriptive Characteristics of the Cohort, Including Anthropometry, DXA Measurements, Menopause Status, and Self-Reported Use of Medications
 WomenMen
  • BMI = body mass index; aBMD = areal bone mineral density; PF = proximal femur; FN = femoral neck; LS = lumbar spine; SERM = selective estrogen receptor modulator; AR = antiresorptive; PTH = parathyroid hormone; HRT = hormone-replacement therapy; OCP = oral contraceptive pill.

  • a

    n = 431 women, 191 men.

  • b

    n = 426 women, 183 men.

  • c

    n = 421 women, 184 men.

N441201
Age (SD)56.5 (19.5)53.5 (19.6)
Age range20.0–98.520.1–88.6
Height (cm)a161.2 (7.9)175.3 (8.0)
Weight (kg)a68.9 (14.5)84.4 (16.8)
BMI (kg/m2)a26.6 (6.0)27.5 (5.8)
PF aBMD (g/cm2)b0.883 (0.131)1.025 (0.145)
FN aBMD (g/cm2)b0.744 (0.121)0.831 (0.131)
LS aBMD (g/cm2)c0.948 (0.129)1.023 (0.168)
Underweight/normal weight/overweight/obese6/179/160/864/60/80/47
Pre/postmenopausal175/266
Previous hysterectomy and/or ovariectomy113 (26%)
Current use of bisphosphonates72 (16%)2 (1%)
Current use of oral glucocorticoids (>3 months)7 (2%)
Current use of SERMs or ARs (calcitonin, PTH)5 (1%)
Current use of testosterone1 (<1%)
Current use of HRT38 (9%)
Current use of OCPs83 (19%)
History of low-trauma fracture after age 2090 (20%)36 (18%)
Table 2. Summary of Participant Recruitment by Age Decade for Women and Men
Age (years)WomenMen
20–2960 (13.6%)30 (14.9%)
30–3949 (11.1%)31 (15.4%)
40–4955 (12.5%)32 (15.9%)
50–5952 (11.8%)27 (13.4%)
60–6987 (19.7%)30 (14.9%)
70–7995 (21.5%)32 (15.9%)
80+43 (9.8%)19 (9.5%)
Total441201

Valid proximal femur and lumbar spine DXA scans were available for 604 participants (421 women, 183 men). One man had a valid lumbar spine scan only, 5 women had valid femoral neck scans only, and 32 participants (15 women, 17 men) did not have valid proximal femur or lumbar spine scans.

HR-pQCT—radius

Since many of the sample outcome distributions were skewed, we describe the samples using median and interquartile range in Table 3. Results of the regression analyses are presented in Table 4. Among young adults, men had significantly greater (by 4% to 47%, p < .05) bone size (Tt.Ar, Ct.Ar), BV/TV, microarchitectural parameters of the trabecular region (Tb.N, Tb.Th), cortical thickness, estimated bone strength (failure load), and fall force than women (Table 4). Total BMD, Ct.Po, and ultimate stress also tended to be greater in young-adult men (p < .1), whereas Tb.Sp, Ct.BMD, the percentage of load carried by the cortex, and load-to-strength ratio were significantly greater in young-adult women than in young-adult men (by 6% to 41%, p < .05).

Table 3. Summary of High-Resolution pQCT (HR-pQCT) and Finite-Element Analysis (FEA) Outcomes at the Distal Radius and Distal Tibia for Women and Men
 WomenMen
RadiusTibiaRadiusTibia
Median (IQR)Median (IQR)Median (IQR)Median (IQR)
  • Tt.Ar = total bone area; BMDtot = total bone mineral density; BV/TV =bone volume to total volume ratio; Tb.N = trabecular number; Tb.Th = trabecular thickness; Tb.Sp = trabecular separation; Ct.Ar = cortical bone area; Ct.BMD = cortical bone mineral density; Ct.Po = cortical porosity; Ct.Th = cortical thickness; FEA = finite-element analysis.

  • a

    n = 431 women, 191 men.

  • b

    n = 415 women, 189 men.

HR-pQCTn = 425n = 434n = 199n = 198
Tt.Ar (mm2)255.3 (228.9, 284.0)667.8 (596.1, 741.2)375.0 (341.7, 416.4)857.3 (777.3, 984.2)
Tt.BMD (mg HA/cm3)301.7 (260.9, 343.9)278.4 (236.2, 317.5)327.5 (290.4, 368.5)313.5 (278.1, 344.9)
BV/TV0.121 (0.094, 0.142)0.131 (0.113, 0.151)0.156 (0.133, 0.180)0.161 (0.142, 0.178)
Tb.N (1/mm)1.85 (1.62, 2.04)1.73 (1.52, 1.95)2.09 (1.86, 2.24)2.02 (1.76, 2.20)
Tb.Th (mm)0.064 (0.057, 0.072)0.077 (0.067, 0.085)0.075 (0.067, 0.084)0.080 (0.071, 0.091)
Tb.Sp (mm)0.471 (0.425, 0.548)0.503 (0.436, 0.580)0.404 (0.372, 0.456)0.418 (0.377, 0.483)
Ct.Ar (mm2)57.6 (52.1, 65.0)123.3 (110.7, 138.2)86.9 (77.4, 97.6)176.7 (159.3, 199.7)
Ct.BMD (mg HA/cm3)823.0 (763.6, 877.5)793.2 (708.9, 855.8)797.2 (748.9, 846.1)777.2 (727.2, 825.4)
Ct.Po (%)7.7 (5.3, 10.7)12.2 (8.4, 17.2)9.2 (7.0, 12.5)13.8 (10.5, 18.2)
Ct.Th (mm)0.98 (0.86, 1.11)1.43 (1.23, 1.63)1.19 (1.05, 1.36)1.78 (1.58, 2.00)
FEA
 Ultimate stress (MPa)25.3 (19.4, 36.2)29.2 (22.8, 35.4)30.0 (23.0, 36.2)33.2 (27.6, 38.3)
 Predicted failure load (N)3289 (2705, 3856)9402 (8058, 10740)5447 (4548, 6208)13720 (11870, 15347)
 Percent load cortical distal43.0 (38.0, 50.0)40.0 (33.0, 47.0)35.0 (31.0, 41.0)39.0 (33.0, 44.0)
 Estimated fall load (Ns/m)a2671 (2627, 2704)2776 (2744, 2815)
 Load-to-strength ratiob0.80 (0.69, 0.97)0.51 (0.45, 0.60)
Table 4. HR-pQCT Outcomes for Young-Adult Women and Men and the Predicted Absolute and Percentage Change in Bone Outcomes Between 20 and 90 Years of Age in Women and Men
 WomenMenSex difference in young adultsSex × age interactionb
Predicted change between 20 and 90 yearsPredicted change between 20 and 90 years
Mean (SD) 20–29yrs (n = 58)Absolute changePercent changeAge correlationaMean (SD) 20 to 29 years (n = 28)Absolute changePercent changeAge correlation
  • Note: For Tt.Ar and Ct.Ar, both the unadjusted and height-adjusted values are provided.

  • SD = standard deviation; Tt.Ar = total bone area; Tt.BMD = total bone mineral density; BV/TV = bone volume to total volume ratio; Tb.N = trabecular number; Tb.Th = trabecular thickness; Tb.Sp = trabecular separation; Ct.Ar = cortical bone area; Ct.BMD = cortical bone mineral density; Ct.Po = cortical porosity; Ct.Th = cortical thickness.

  • a

    Pearson correlation coefficient for normally distributed variables and Spearman rank correlation for non–normally distributed variables.

  • b

    Comparison of slopes.

  • c

    Indicates that regression model is nonlinear.

  • d

    Variable required transformation.

Radius
 Tt.Ar (mm2)262.8 (42.7)3.920.04349.0 (56.4)74.1220.32<0.001<0.001
 Tt.Ar (mm2/ht)159.0 (2.9)17.111c0.16194.0 (5.5)55.528c0.41<0.001<0.001
 Tt.BMD (mg HA/cm3)319.7 (60.8)−110.6−36c−0.35350.2 (11.3)−107.3−32c−0.400.030.97
 BV/TV0.126 (0.028)−0.028−23c−0.210.169 (0.030)−0.043−26c−0.36<0.0010.13
 Tb.N (1/mm)d1.95 (0.21)−0.41−21c−0.312.20 (0.25)−0.29−13c−0.27<0.0010.41
 Tb.Th (mm)d0.064 (0.011)−0.001−2c0.020.077 (0.015)−0.012−16c−0.22<0.0010.004
 Tb.Sp (mm)d0.454 (0.065)0.15234c0.290.382 (0.049)0.08422c0.30<0.0010.38
 Ct.Ar (mm2)d62.8 (10.3)−12.0−20c−0.3586.6 (14.6)−7.5−9c−0.12<0.0010.008
 Ct.Ar (mm2/ht)38.1 (0.8)−5.5−15c−0.2148.3 (1.6)−2.3−5c−0.03<0.0010.13
 Ct.BMD (mg HA/cm3)d835.6 (56.0)−196.0−24c−0.54785.6 (62.8)−114.9−15c−0.32<0.001<0.001
 Ct.Po (%)d6.2 (3.1)10.2176c0.568.1 (4.3)7.684c0.430.020.001
 Ct.Th (mm)d1.06 (0.19)−0.24−31c−0.331.25 (0.25)−0.25−21c−0.29<0.0010.51
 Ultimate stress (MPa)d31.0 (9.0)−16.8−57c−0.4735.2 (9.1)−17.7−54c−0.500.050.88
 Failure load (N)2033 (352)−908−46c−0.572993 (574)−839−28c−0.36<0.0010.60
 % load corticald41.8 (8.8)−0.2−0.5c0.0137.0 (7.9)−6.0−17c−0.200.020.007
 Fall force2694 (58)−70−3−0.372812 (61)−78−3−0.26<0.0010.71
 Load-to-strength ratiod0.72 (0.15)0.5574c0.510.51 (0.10)0.2447c0.33<0.0010.009
Tibia
 Tt.Ar (mm2)648.1 (11.1)53.380.13856.1 (26.5)53.360.11<0.0010.83
 Tt.Ar (mm2/ht)392.2 (5.5)61.8160.28475.0 (11.8)61.8130.23<0.0010.91
 Tt.BMD (mg HA/cm3)319.7 (49.8)−118.3−37c−0.52331.2 (44.3)−71.1−22c−0.330.300.002
 BV/TV0.148 (0.029)−0.026−18−0.220.178 (0.025)−0.026−15−0.27<0.0010.56
 Tb.N (1/mm)d1.82 (0.26)−0.18−10−0.162.06 (0.28)−0.16−8−0.15<0.0010.77
 Tb.Th (mm)d0.082 (0.014)−0.005−6−0.070.088 (0.014)−0.005−6−0.160.070.36
 Tb.Sp (mm)d0.480 (0.087)0.073160.170.407 (0.060)0.052130.18<0.0010.99
 Ct.Ar (mm2)d135.5 (24.3)−29.5−23c−0.35180.8 (44.5)−15.4−9c−0.10<0.0010.002
 Ct.Ar (mm2/ht)82.2 (2.0)−14.1−18c−0.25100.7 (4.7)−3.6−4c−0.01<0.0010.04
 Ct.BMD (mg HA/cm3)855.7 (43.6)−276.9−32c−0.73784.9 (52.5)−115.2−15c−0.36<0.001<0.001
 Ct.Po (%)d8.6 (3.5)19.4259c0.7312.4 (4.6)10.483c0.41<0.001<0.001
 Ct.Th (mm)d1.59 (0.34)−0.36−24c−0.321.89 (0.61)−0.23−13c−0.140.0040.03
 Ultimate stress (MPa)36.6 (7.8)−17.8−49c−0.5337.7 (7.5)−12.2−33c−0.380.530.02
 Failure load (N)5655 (890)−1940−35c−0.517579 (1165)−1533−21c−0.28<0.0010.21
 % load cortical43.4 (7.5)−14.3−34c−0.3838.1 (7.5)−7.1−19c−0.170.0030.004

With the exception of Tt.Ar, all bone outcomes at the distal radius showed significant quadratic associations with age (Table 4). Between 20 and 90 years of age, the predicted decrease in BMD, BV/TV, ultimate stress, and fall force was similar in women and in men. Total bone area was positively associated with age in men, with a predicted increase of 22% between 20 and 90 years of age (p < .001), whereas bone area remained unchanged in women. When adjusted for height, Tt.Ar increased with age to a larger extent in men than in women (+28% versus +11%, p < .001). In the trabecular compartment, Tb.Th was negatively associated with age in men, with a predicted decline of 16% (p < .001), whereas Tb.Th remained unchanged in women (Fig. 2). Although not statistically significant, the predicted decline in Tb.N tended to be greater in women than in men (by 8%, p = .41). Similarly, the predicted increase in Tb.Sp tended to be greater in women than in men (by 12%, p = .38). In the cortical compartment, the predicted declines in Ct.Ar and Ct.BMD were larger in women than in men (by 11% and 9%, respectively, p < .01); however, after adjusting for height, the age-related decline in Ct.Ar was not significantly different between the sexes. In addition, the predicted increase in Ct.Po was more than twofold greater in women than in men (by 92%, p = .001, Fig. 2), and Ct.Th tended to show a larger decline with age in women than in men (by 10%, not significant). Although the relationship between age and ultimate stress was similar in women and men, the predicted decline in failure load was larger in women than in men (by 18%; Fig. 3), and the percentage of load carried by the cortex remained unchanged with age in women, whereas it declined by 17% in men. Finally, the predicted increase in the load-to-strength ratio, an estimate of forearm fracture risk, was greater in women than in men (by 27%, p = .009; Fig. 3).

thumbnail image

Figure 2. Predicted age-related changes in bone volume to total volume ratio (BV/TV) (A), trabecular number (Tb.N) (B), trabecular thickness (Tb.Th) (C), and cortical porosity (Ct.Po) (D) at the distal radius in women (circles) and men (triangles). The solid line represents the fitted mean from the regression model, and the dashed lines represent the 95% confidence interval of the prediction.

Download figure to PowerPoint

thumbnail image

Figure 3. Predicted age-related changes in FEA-estimated failure load (A) and the load-to-strength ratio (B) at the distal radius in women (circles) and men (triangles). The solid line represents the fitted mean from the regression model, and the dashed lines represent the 95% confidence interval of the prediction.

Download figure to PowerPoint

When the subjects who reported current use of medications were excluded, we observed linear associations with age for Tb.N (negative) and Tb.Sp (positive) and a quadratic association with age for Tt.Ar (data not shown). The predicted age-related changes were similar in magnitude to those observed in the initial analysis. All other relationships with age remained the same.

HR-pQCT—tibia

In young adults, Tt.BMD, Tb.Th, and ultimate stress were not significantly different between women and men. However, bone size (Tt.Ar, Ct.Ar), BV/TV, microarchitectural parameters of the trabecular (Tb.N) and cortical (Ct.Po, Ct.Th) compartments, and failure load were significantly greater in men than in women (by 13% to 44%). Young-adult women demonstrated a greater degree of trabecular separation (by 18%), higher Ct.BMD (by 9%), and percentage of load carried by the cortex (by 14%) than young-adult men.

In contrast to results for the distal radius, parameters of trabecular microarchitecture at the distal tibia showed linear associations with age (Table 4). We observed significant quadratic associations with age for all cortical bone outcomes and strength estimates. Between 20 and 90 years of age, the predicted increase in Tt.Ar was similar in women and in men, even after adjusting for height; however, women demonstrated a greater age-related decline in Tt.BMD than men (by 15%, p = .002). In the trabecular compartment, the predicted declines in BV/TV, Tb.N, and Tb.Th were similar in women and men (by 6% to 18%; Fig. 4), and the predicted increase in Tb.Sp also was similar in women and men. In the cortical compartment, Ct.Ar and Ct.BMD declined more with age in women than in men (by 14% and 17%, respectively), and the predicted increase in Ct.Po was more than threefold greater in women than in men (by 176%, p < .01 for all; Fig. 4). Cortical thickness also tended to decline more with age in women compared with men (by 11%, p = .03). The predicted decline in ultimate stress and in the percentage of load carried by the cortex was greater in women than in men (by 16% and 15%, respectively; p < .01); however, the predicted decline in failure load was not significantly different in women and men.

thumbnail image

Figure 4. Predicted age-related changes in bone volume to total volume ratio (BV/TV) (A), trabecular number (Tb.N) (B), trabecular thickness (Tb.Th) (C), and cortical porosity (Ct.Po) (D) at the distal tibia in women (circles) and men (triangles). The solid line represents the fitted mean from the regression model, and the dashed lines represent the 95% confidence interval of the prediction.

Download figure to PowerPoint

These relationships with age were similar when we excluded subjects who reported current use of medications, with the exception of failure load, which became linear. The magnitude of the predicted age-related changes was similar to the initial analysis (data not shown).

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References

This population-based study provides the first data describing age- and sex-related differences in cortical and trabecular bone microstructure and estimated bone strength in the peripheral skeleton in Canadian adults. In addition to standard HR-pQCT morphologic outcomes, this study is strengthened by the assessment of both cortical porosity and bone strength estimated from finite-element analysis. Consistent with previous population-based studies of adults in the United States13 and the United Kingdom,14 our results suggest that patterns of age-related bone loss differ between women and men and also between the cortical and trabecular compartments at the distal radius and distal tibia. The predicted macro- and microstructural changes with aging in our cohort may help to explain the significantly greater incidence of forearm fractures in women than in men,34 as well as how age-related structural changes differ between weight-bearing and non-weight-bearing sites.

Before discussing our findings in detail, we acknowledge that in this cross-sectional study we cannot determine true age-related changes in HR-pQCT outcomes in women and men. Further, the predicted changes and sex differences we report based on our regression analyses may be influenced by secular trends in both body and bone size.35, 36 Although we adjusted for body size in our analyses of bone area (total and cortical), this may not necessarily reduce the influence of secular trends because they may vary between sexes and across limbs.37, 38 Despite these limitations, previous studies reported good agreement between the rate of bone loss predicted from cross-sectional data and the true rate of bone loss determined from 16 years of longitudinal data.39 Further, our findings are in agreement with paired biopsy40 and longitudinal pQCT41, 42 studies. In discussing our results, we recognize the limitations associated with predicting change in bone outcomes from cross-sectional data and highlight the need for prospective HR-pQCT trials to more accurately determine age-related changes in bone structure and strength in women and men.

In young adults, we found that men have larger bone size and higher-quality indices of trabecular morphology (ie, a larger number of thicker trabeculae) and cortical bone structure (ie, thicker cortices) at both peripheral sites than women. Together these characteristics may contribute to a bone structure that is more resistant to fracture. As men and women age, they appear to lose a similar volume of trabecular bone at the distal radius. However, the microstructural changes underpinning this bone loss differed between sexes; women tended to lose a larger number of trabeculae, whereas men appeared to experience more trabecular thinning together with loss of trabecular elements. These findings are in agreement with the HR-pQCT study of Khosla and colleagues13 and earlier histomorphometric results of iliac crest biopsies.21, 43 Trabecular thinning is believed to be a more common mechanism of bone loss in men owing to decreased rates of bone formation,21, 44 whereas loss of trabecular elements is thought to be more common in women owing to menopause-associated increases in osteoclastic activity that lead to a more discontinuous trabecular network.21 Importantly, in this study, both trabecular thinning and loss of trabecular elements contributed to loss of trabecular bone volume at the distal tibia in both women and men. Our findings at the tibia are consistent with results of a recent µCT study of human femoral neck cadavers that indicated the relationship with age was similar in men and women for Tb.N, Tb.Th, and Tb.Sp.45 Thus patterns of trabecular bone loss may vary with skeletal site.

Our regression analyses suggest that BV/TV remains relatively constant at the distal radius until midlife and declines thereafter (Fig. 2), whereas at the distal tibia, our data suggest that loss of trabecular bone volume may begin in young adulthood (Fig. 4). The latter finding is in agreement with previous HR-pQCT13 and pQCT41, 42 results showing a relatively continuous age-related decrease in BV/TV and trabecular BMD at peripheral sites in men and women. Underpinning the decline in BV/TV, Khosla and colleagues reported a 14% increase in trabecular number in men between 20 and 49 years of age and a 26% decrease in Tb.Th at the distal radius.13 We did not observe similar findings at either site in this study. Instead, Tb.Th at the distal radius appeared to increase slightly prior to age 50 in women and men, whereas Tb.N decreased slightly before midlife and showed accelerated loss thereafter. At the distal tibia, the linear relationship between age and Tb.N and Tb.Th suggests a steadier decline across adulthood. Longitudinal HR-pQCT studies are needed to confirm the age of onset of microarchitectural deterioration in women and men at these two peripheral sites and to elucidate the role of sex hormones in these age-related changes.

Interestingly, in their recent population-based HR-pQCT study, Dalzell and colleagues found that age was not significantly associated with any of the trabecular indices at either the distal radius or the tibia in men or women.14 The difference between our findings and those of the two earlier HR-pQCT studies may be related to scanner type or sample size. Khosla and colleagues13 used a prototype model of the XtremeCT that employs a larger voxel size (89 µm) than the XtremeCT used in this study (82 µm). The larger voxel size may have led to differences in their assessment of trabecular structure. However, both the study by Khosla and colleagues13 and this study benefited from a larger sample. It is possible that owing to a small sample (n = 132) with particularly low numbers in the younger (<30 years) and older (>70 years) age groups, Dalzell and colleagues14 were not sufficiently powered to detect age-related changes in trabecular bone microarchitecture.

In addition to trabecular bone loss, declines in cortical BMD associated with increased porosity also contribute to bone fragility with aging.46 In our cohort, cortical BMD was greater in young-adult women owing to a less porous cortex than young-adult men. However, with advancing age, loss of cortical BMD was more pronounced in women than men at both peripheral sites. Cortical BMD reflects both the volume of cortical bone and the degree of mineralization.47 Since mineralization of cortical bone changes minimally with age,46 the predicted age-related changes in cortical BMD likely reflect changes in cortical bone volume, which, in turn, is highly associated with cortical porosity.46, 47 Although the specific age at which this change in cortical bone begins is unclear, it appears that both cortical BMD and porosity of the distal radius remain relatively stable in women until approximately the time of menopause, after which porosity increases exponentially. This is consistent with earlier pQCT41 and HR-pQCT13, 14 reports and supports the current belief that estrogen deficiency plays a key role in cortical bone loss.48 Importantly, we also observed that the proportion of load carried by the cortex at the distal radius decreased with age in men, whereas it remained unchanged in women. Thus the apparent changes in cortical microstructure at the distal radius in women, combined with a higher proportion of load carried by the cortex, may contribute to higher fracture risk at this site in postmenopausal women.

In contrast to the radius, our results indicate loss of cortical BMD at the distal tibia may begin earlier in women, during pre- or perimenopause. These findings are in agreement with the recent results of Burghardt and colleagues,7 who found that the largest differences in cortical porosity at the radius occurred between women in their 50s and 60s, whereas at the tibia, the largest difference in porosity was observed in the 40s and 50s. Porosity may increase earlier at the distal radius; however, the resolution of HR-pQCT may be insufficient to detect the presence of smaller pores. In men, cortical porosity was higher than in women across the entire age range, with the exception of the eighth decade (data not shown), and the onset of cortical bone loss was not apparent until later in life. Further study is required to determine the role of sex hormone deficiencies in these age-related changes.

In addition to higher-quality trabecular and cortical microstructure, the bone-strength advantage in men across the lifespan is also a function of bone size. This biomechanical advantage is established during growth and is maintained with aging.48 At both sites, the predicted age-related increase in total bone area (height-adjusted) in men and women suggests continued periosteal apposition. However, at the distal radius, bone size appeared to increase more with age in men than in women. Although we cannot rule out secular trends as a possible explanation for this finding, this finding is in line with the results of a recent longitudinal pQCT study of the tibial shaft.42 Periosteal apposition is thought to offset bone resorption on the inner endocortical surface that leads to cortical thinning and reduced bone strength, but in women, the rate of periosteal apposition may not be sufficient to compensate for endocortical bone loss.42, 48 This apparent sex difference in macrostructural changes with aging may partially explain the higher incidence of Colles' fractures in women than in men.34

Despite a larger bone size, thicker cortices, and greater volume of trabecular bone in young-adult men, our estimate of size-adjusted bone strength (ultimate stress) was not significantly different between young-adult men and women. Also, the predicted decline in size-adjusted bone strength was similar in magnitude in both sexes despite sex differences in the pattern of age-related bone loss in the trabecular and cortical compartments. However, the load-to-strength ratio, which provides an estimate of distal radius fracture risk,31 increased with age to a greater extent in women than in men. This is similar to previous pQCT34 and HR-pQCT14 results and suggests that even though fall force is less in women at all ages (owing to women being shorter and lighter), the deterioration in trabecular and cortical microarchitecture with age is a more significant factor related to increased forearm fracture risk in postmenopausal women.

Although it is unclear which macro- and microarchitectural parameters were the largest contributors to the age-related declines in bone strength in women and men, bone strength is a function of both trabecular and cortical bone at common fracture sites.49 Regarding microarchitectural changes, results from mechanical testing studies indicate that reductions in Tb.N have a 2.5-fold greater effect on bone strength than reductions in Tb.Th.50 However, these findings are based on studies of vertebral trabecular bone in isolation rather than the entire vertebral body with both cortical and trabecular bone. In contrast, a parametric study of changes in bone microarchitecture at the distal radius found that the cortex carried the majority of the load at the distal radius, and as a result, reductions in cortical thickness had the greatest impact on bone strength.51 Scenarios in which the trabecular structure was atrophied (eg, thinning of trabeculae or removal of trabeculae) affected bone strength to a lesser degree. Thus it is clear that the cortex is also an important contributor to bone strength, but the relative importance of cortical versus trabecular changes depends, in part, on loading type (ie, axial compression versus bending).52

In addition to limitations associated with the cross-sectional study design discussed previously, we note several other limitations of our study. First, we recruited many of our young subjects (40 years of age and younger) from the general Calgary population using non-random-sampling techniques. Nevertheless, compared with subjects younger than 40 years of age, who were recruited using random sampling techniques, the nonrandom recruits were similar in age, height, and weight and were comparable with respect to demographic variables such as race. Second, the majority of participants were white (94%); this is higher than the proportion of whites in the general Calgary population (78%) and in the Canadian population as a whole (84%), as reported in the 2006 Census.53, 54 Thus additional investigations are warranted in other ethnic groups. Third, we recognize limitations associated with quantification of cortical porosity using HR-pQCT because a resolution of at least 10 µm is recommended for direct assessment of individual cortical pores in humans.55 However, results from our validation study suggest that HR-pQCT accurately measures total porosity (ie, pore volume).6 Finally, we used axial compression as the loading condition in our finite element analyses. Since Colles' fractures typically represent a combination of loads, possibly including compression, bending, and torsion, our FEA may not provide an accurate representation of the forces sustained during a fall on an outstretched hand. Thus further developments in FEA should consider how best to incorporate other loading conditions.

In summary, this study provides the first HR-pQCT data for a large, population-based sample of Canadian adults. Our findings highlight important sex differences in age-related macro- and microarchitectural changes in the trabecular and cortical bone compartments that may underpin sex differences in fracture risk. In particular, the observed trends for less periosteal expansion, more porous cortices, and a greater percentage of load carried by the cortex at the distal radius may contribute to the 600% higher incidence of forearm fractures in women.34 Importantly, these findings must be confirmed in longitudinal studies that are able to accurately determine changes in trabecular and cortical microarchitecture with age in women and men.

Disclosures

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References

All the authors state that they have no conflicts of interest.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References

We gratefully acknowledge the participants who volunteered their time to take part in this study, the assistance of Ms Jane Allan and Ms Bernice Love in subject recruitment, the HR-pQCT scanning expertise of Ms Irene Hanley and Ms Shannon Boucousis, and the technical assistance of Dr Helen Buie in scan analysis. In addition, we appreciate support from Dr David Goltzman (CaMos principal investigator) and the CaMos Research Group. Funding from the Canadian Institutes for Health Research (CIHR) supported this research. SKB is an Alberta Innovates—Health Solutions (AIHS) Senior Scholar, and HMM is supported through CIHR and AIHS postdoctoral fellowships.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References