SEARCH

SEARCH BY CITATION

Keywords:

  • finite element analysis;
  • microarchitecture;
  • μCT;
  • osteoporosis;
  • wrist fracture

Abstract

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. REFERENCES

BMD, bone microarchitecture, and bone mechanical properties assessed in vivo by finite element analysis were associated with wrist fracture in postmenopausal women.

Introduction: Many fractures occur in individuals with normal BMD. Assessment of bone mechanical properties by finite element analysis (FEA) may improve identification of those at high risk for fracture.

Materials and Methods: We used HR-pQCT to assess volumetric bone density, microarchitecture, and μFE-derived bone mechanical properties at the radius in 33 postmenopausal women with a prior history of fragility wrist fracture and 33 age-matched controls from the OFELY cohort. Radius areal BMD (aBMD) was also measured by DXA. Associations between density, microarchitecture, mechanical parameters and fracture status were evaluated by univariate logistic regression analysis and expressed as ORs (with 95% CIs) per SD change. We also conducted a principal components (PCs) analysis (PCA) to reduce the number of parameters and study their association (OR) with wrist fracture.

Results: Areal and volumetric densities, cortical thickness, trabecular number, and mechanical parameters such as estimated failure load, stiffness, and the proportion of load carried by the trabecular bone at the distal and proximal sites were associated with wrist fracture (p < 0.05). The PCA revealed five independent components that jointly explained 86.2% of the total variability of bone characteristics. The first PC included FE-estimated failure load, areal and volumetric BMD, and cortical thickness, explaining 51% of the variance with an OR for wrist fracture = 2.49 (95% CI, 1.32–4.72). Remaining PCs did not include any density parameters. The second PC included trabecular architecture, explaining 12% of the variance, with an OR = 1.82 (95% CI, 0.94–3.52). The third PC included the proportion of the load carried by cortical versus trabecular bone, assessed by FEA, explaining 9% of the variance, and had an OR = 1.61 (95% CI, 0.94–2.77). Thus, the proportion of load carried by cortical versus trabecular bone seems to be associated with wrist fracture independently of BMD and microarchitecture (included in the first and second PC, respectively).

Conclusions: These results suggest that bone mechanical properties assessed by μFE may provide information about skeletal fragility and fracture risk not assessed by BMD or architecture measurements alone and are therefore likely to enhance the prediction of wrist fracture risk.


INTRODUCTION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. REFERENCES

Measurement of areal BMD (aBMD) by DXA is presently the standard method for diagnosis of osteoporosis and prediction of fracture risk. Low BMD is widely recognized as a major risk factor for fractures.(1–3) However, the mechanical competence or load-bearing capacity of bone, which is its ability to sustain a certain load without encountering fracture, depends on the amount of bone (bone mass), the bone morphology (geometry and microarchitecture), and the intrinsic properties of bone tissue.(4) Thus, the use of aBMD for fracture prediction may be enhanced by considering other independent risk factors, such as cortical and trabecular architecture.(5–7) Recent developments in high-resolution imaging enable quantification of bone microarchitecture in vivo at the distal radius and tibia, with an isotropic voxel size of 82 μm. We recently showed that alterations of cortical and trabecular structure measured by high-resolution pQCT (HR-pQCT) are associated with fragility fractures in postmenopausal women, partially independently of decreased aBMD measured by DXA.(6,7)

HR-pQCT also offers new possibilities for prediction of bone strength by using microfinite element (μFE) techniques.(8–11) Predicted failure loads calculated from μFE analysis agree well with those measured in experimental tests on cadaver forearms.(12,13) Moreover, Pistoia et al.(12) reported stronger correlations between μFE analyses and measured failure load of the distal radius than those provided by bone mass or bone structural parameters. However, it is unknown whether assessment of bone strength by finite element analysis (FEA) improves the prediction of fracture.

Therefore, in this study, we compared in vivo measurements of BMD, microarchitecture, and bone mechanical parameters assessed by μFE analysis in women with and without prior history of wrist fracture. Our first aim was to study whether these parameters were associated with wrist fracture. This approach leads to a large number of variables; thus, we analyzed the correlations between density, microarchitecture, and μFE-predicted parameters. Our second aim was to establish which parameters or combinations of parameters were most relevant, without redundancy, to best characterize distal radius bone using principal components (PCs) analysis (PCA). Our final aim was to determine whether these new variables (i.e., PCs) were associated with wrist fracture.

MATERIALS AND METHODS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. REFERENCES

Subjects

The subjects and imaging procedures were described in detail in an earlier study showing the association of vertebral and nonvertebral fractures with alterations of cortical and trabecular architecture at the distal radius and tibia.(7) For this study, we focused on 34 women with wrist fracture and their age-matched controls. One woman was eliminated from the study because of obvious movement artefacts during data acquisition. The final cross-sectional analysis involved 33 postmenopausal women (mean age, 73.4 ± 6.1 yr) from the OFELY cohort who sustained a fragility fracture of the wrist during the 13 yr of follow-up of the study (beginning of the study: February 1992). These subjects underwent DXA and HR-pQCT measurements at the same visit (September 2004 to June 2006) to determine bone densities and microstructural properties in vivo at the distal radius. Each fracture case was randomly age-matched within 1 yr to a control from the same cohort that never had a fracture. All fragility fractures were confirmed by radiographs. The mean time interval between fracture and scan acquisition was 5.8 ± 3.6 yr. Only low-trauma fractures (i.e., those occurring after falls from standing height or less) were included. The use of treatments affecting bone metabolism (bisphosphonates, hormone replacement therapy, selective estrogen receptor modulator, tibolone, or aromatase inhibitors) was registered every year and taken into account in that study if the duration of treatment was 1 yr or more at the time (or within 6 mo before) of architectural evaluation. The protocol was approved by an independent Ethic Committee, and all patients gave written informed consent before participation.

Measurement of BMD and bone microarchitecture

aBMD (g/cm2) at the total hip was measured using DXA (QDR4500; Hologic, Waltham, MA, USA). At the radius, aBMD was measured using DXA with either a QDR 1000+ (9 women with fracture and 7 without) or a QDR 4500 (24 women with fracture and 26 without) (Hologic). Because the number of cases and controls were the same for each system, the different beam geometry (pencil or fan) had no influence on the difference between groups. We assessed aBMD at the ultradistal site because this region of interest is most similar to that analyzed with the HR-pQCT.

Volumetric BMD and microarchitecture were measured at the distal radius using an HR-pQCT system (XtremeCT; Scanco Medical, Bassersdorf, Switzerland) that acquires a stack of 110 parallel CT slices with an isotropic voxel size of 82 μm, as previously described.(14) Given the marked difference in bone morphology between the proximal and distal regions of the volume we examined (Fig. 1), we also examined whether either the proximal or distal morphology and/or mechanical properties were more strongly associated with fracture risk than measurements averaged over the whole region. To do this we decided to define two additional volumes of interest: the first 20 slices and the last 20 slices (1.6-mm length) representing, respectively, the more distal and more proximal part of the ultradistal radius, and compared these measurements to the entire 110 slices (9-mm length).

thumbnail image

Figure FIG. 1.. Boundary condition and distribution of Von Mises stresses in the bone tissue for a control (left) and a fracture case (right). The colors indicate the stress levels with red for high stresses and blue for low stresses. (A) Compression test in the axial direction with a 1000 N load applied to the first distal slice while the proximal surface was fully constrained. (B–D) Distribution of Von Mises stresses in the first distal (B) and last proximal (D) slices in a woman who never had a fracture. (C–E) Distribution of Von Mises stresses in the first distal (C) and first proximal (E) slices in a woman who had a wrist fracture.

Download figure to PowerPoint

The outcome variables used in our analyses included the following: DXA-derived aBMD (g/cm2) and HR-pQCT–derived volumetric BMD (g hydroxyapatite/cm3) for entire (Dtot), trabecular (Dtrab), and cortical (Dcort) regions; cortical thickness (CTh, μm); and trabecular number (TbN, mm−1), thickness (TbTh, μm), separation (TbSp, μm), and intra-individual distribution of separation (TbSp SD, μm). DXA and HR-pQCT measurements were performed at the nondominant wrist unless there was a history of fracture on that wrist, in which case, the nonfractured wrist was measured.

FEA

FE models were generated using Image Processing Language (IPL) software provided by Scanco Medical. In the FE models, different elastic properties were specified for cortical and trabecular bone tissue. To identify cortical and trabecular bone tissue, a special peeling algorithm, included with the IPL software, was used while specifying a minimum cortical thickness of 6 voxels. μFE models were created by converting each voxel in the model to an equally sized brick element,(15) resulting in μFE models with ∼2 million elements. Material properties were chosen isotropic and elastic. Elements representing cortical bone were assigned a Young's modulus of 20 GPa, and those representing trabecular bone tissue were assigned a Young's modulus of 17.5 GPa.(16) For all elements, a Poisson's ratio of 0.3 was specified. Boundary conditions were set to simulate a “high friction” compression test in the axial direction with a 1000-N load applied. To do so, displacements of nodes located at the proximal face were suppressed in all directions and those at the distal face were permitted in the axial direction and suppressed in the other directions.

The estimated failure load was computed based on a criterion developed by Pistoia et al.,(12) where fracture is assumed to occur when 2% of the bone tissue is strained beyond a critical limit of 7000 microstrain. However, because the elastic properties in our study were twice as high as those used by Pistoia et al. (who used an elastic modulus of 10 GPa), the critical strain limit in our study was set to 3500 microstrain to calculate comparable failure loads. Other FEA-derived variables used in our analyses included the following: stiffness (kN/mm), the percentage of load carried by the trabecular bone at the distal and proximal surface of the volume of interest (% load trab distal and % load trab proximal, respectively), and the average and SD values of the Von Mises stresses in the trabecular and cortical bone (Trab average stress and Cort average stress; Trab SD stress and Cort SD stress [MPa], respectively). All μFE analyses were performed using the FE-solver included in the IPL software.

Statistical analysis

Descriptive statistics were summarized by means and SDs. The differences in density, microarchitecture, and mechanical parameters among women with and without fracture were assessed by Wilcoxon signed rank test or by Student t-test, depending on the distribution of variables. The differences between women with and without fracture were expressed in percentage and in Z-scores calculated using means and SDs of the controls. Associations between density, microarchitecture, mechanical parameters, and fracture status were evaluated by univariate logistic regression analysis and expressed as ORs (with 95% CIs) per SD change. The relationships between estimated bone failure load, other mechanical parameters, densities, and microarchitecture were studied using Pearson or Spearman correlation, depending on the distribution of variables. Correlations among architectural parameters measured by HR-pQCT (|r| = 0.34–0.95) and between architectural parameters and aBMD measured by DXA at the radius (|r| = 0.54–0.85) have already been reported.(6,7)

We examined a relatively large number of parameters, many of which are intercorrelated. Thus, to avoid the consequences of multicollinearity, we conducted a PCA in the 66 women. All parameters of biomechanics, density, and microarchitecture (except TbTh and Dcort because of technical limitations), aBMD at the ultradistal radius, anthropometric parameters (height and weight), and age were introduced, after being standardized, in a PCA.(17) PCA is a statistical technique that identifies, from a set of correlated variables, a substantially smaller set of uncorrelated parameters: the PCs. The analysis searches for a few linear combinations of the original parameters that capture most of the information (variance) of the original parameters: the PCs. The first PC (PC1) accounts for as much of the variability in the data as possible, and each succeeding component accounts for as much of the remaining variability as possible. The varimax rotation was used to interpret the results more easily. This rotation maximizes the saturation of each parameter (estimated component weights) for a PC and minimizes it for the others PCs. Only PCs having eigenvalues > 1.0 were selected. Each PC can be interpreted based on the estimated component weights for each variable.

Associations between PCs and fracture status were evaluated by univariate logistic regression analysis and expressed as ORs (with 95% CIs) per SD change. Statistical analyses were performed using SPSS software (version 12.0) or SAS software (SAS V8; SAS Institute, Cary, NC, USA).

RESULTS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. REFERENCES

Women with wrist fractures did not differ from age-matched controls for height, weight, and age at menopause (Table 1). As expected, aBMD of the hip and the forearm was lower in women with fractures. The use of treatments affecting bone metabolism, such as bisphosphonates (n = 16), hormone replacement therapy (n = 9), and selective estrogen receptor modulators (n = 2), used 1 yr or more at the time (or within 6 mo before) of this study, was similar in women with and without fractures.

Table Table 1.. Characteristics and Mean Values for Women With Wrist Facture and Controls (Mean ± SD)
Thumbnail image of

FEA

Stiffness and estimated failure load were significantly lower (−15% to 16%), whereas average stresses and the SD of these stresses were higher (15–23%) in women with fractures compared with controls (Table 1). Specifically, the SD of trabecular stress was significantly higher in the fracture group and was about as high as the average value, indicating that the local stresses were not evenly distributed through the trabecular network. For cortical bone, both the average and SD of the stress were significantly higher in fracture cases than in controls. Figure 1 shows the distribution of Von Mises stresses in the radius of a control and a case.

FEA showed that, in both fracture cases and controls, most of the load was carried by cortical bone (Table 1). However, the proportion of load carried by cortical versus trabecular bone differed at the distal and proximal regions. As expected, because of the difference in morphology (Fig. 1), the percentage of load carried by trabecular bone was higher at the distal than at the proximal region in both fracture and control groups (37–40% versus 9–12%, respectively; Table 1). However, in the fracture group, the percentage of load carried by the cortical bone (63.4%) was higher than in the control group (59.6%) and thus the percentage of load carried by trabecular bone was lower in fracture patients (36.6%) than in controls (40.4%), resulting in a more heterogeneous distribution of load between trabecular and cortical bone in women with fractures.

In a univariate logistic regression model, each SD decrease of estimated failure load, stiffness, and proportion of load carried by the trabecular bone at the distal and proximal sites were associated with a significantly increased risk of wrist fracture, with ORs ranging from 1.75 to 3.09 (Table 2).

Table Table 2.. Association Between BMD/Microarchitecture/Mechanical Parameters and Fracture Status, Expressed as ORs (95% CI) per SD Change
Thumbnail image of

Microarchitecture

Examining the total volume of interest (∼9 mm), we found that women with fractures had lower total and trabecular density (−24% and −31%, respectively; p = 0.001), CTh (−22%; p = 0.003), TbN (−21%; p = 0.001), and TbTh (−12%; p = 0.003) and higher TbSp (40%; p = 0.001) and TbSp SD (73%; p = 0.001) than controls without fracture (Table 1).(7) To take the relatively high interindividual variations into account, Z-scores were calculated and ranged from −1.1 (for Dtrab and TbN) to 1.5 (for TbSp and TbSp SD). In a univariate logistic regression model, each SD decrease of areal and volumetric total and trabecular densities, cortical thickness, and trabecular number was associated with a significantly increased risk of wrist fracture, with ORs ranging from 2.54 to 5.38 (Table 2).

Examining the proximal and distal regions separately, we found that all microarchitectural parameters and volumetric bone densities were significantly different between women with and without fractures (Table 3). Moreover, most architecture variables and bone densities differed between the proximal and distal regions, regardless of fracture status. There was no clear advantage of using either the proximal or distal set of slices compared with the whole volume for discriminating wrist fracture cases from controls. Interestingly, however, the difference in cortical thickness between cases and controls was more pronounced at the distal region (−35%) versus the proximal region (−16%), whereas the difference in trabecular thickness between cases and controls was more pronounced at the proximal region (−20% versus −7.6% at the distal region).

Table Table 3.. HR-pQCT Measurements at the First (Distal) and Last (Proximal) 20 Slices in Women With Wrist Fracture and Controls (Mean ± SD)
Thumbnail image of

Correlations

μFE-estimated failure load was highly correlated with areal and volumetric total and trabecular BMD (r = 0.77–0.82), cortical thickness (r = 0.69), trabecular number, separation, and distribution (|r| = 0.68–0.71), SD of trabecular stress, and average and SD of cortical stress (|r| = 0.90–0.96) and stiffness (r = 0.99).

Correlations between parameters were high; thus, to avoid the consequences of multicollinearity, we conducted a PCA to establish which parameters or combinations of parameters were most relevant, without redundancy, to best characterize distal radius bone.

PCA

The PCA revealed five components that were above the 1.0-eigenvalue threshold. These five independent factors jointly explained 86.2% of the total variability of bone characteristics (Table 4). The weightings of these five components, representing the degree and direction of each of the measured variable's contribution to each component, are shown in Table 5. The highest weights (>0.6) are marked in bold.

Table Table 4.. Extraction of the PCs
Thumbnail image of
Table Table 5.. Component Loading Matrix After Varimax Rotation
Thumbnail image of

The first PC (PC1) incorporated 9 of the 19 traits in the model, including bone quantity (aBMD, volumetric BMD, cortical thickness) and stiffness, estimated bone failure load, and cortical and trabecular average and/or SD stresses. Thus, PC1 may be referred to as the component of bone strength and quantity, explaining 51% of the total variance. Within PC1, stiffness and estimated bone failure load had the highest weights. PC2 included trabecular number, separation, and distribution; thus, it may be referred to as the component of trabecular microarchitecture, explaining 12% of the variance. An additional 9% of the variance was explained by PC3, which included the variables defining the proportion of load carried in trabecular versus cortical bone at the distal and proximal sections and may thus be referred to as the component of load distribution. PC4 included the cross-sectional area of the radius (7% of the variance), and PC5 reflected subject height and weight (6% of the variance) and may, respectively, be referred to as the component of bone geometry and component of morphology.

We used univariate logistic regression to compute the association between PCs and fracture status. The OR for an SD decrease in PC1 was 2.49 (95% CI, 1.32–4.72; decreased BMD and estimated bone failure load and increased trabecular and cortical stresses; p = 0.005). The OR for an SD increase in PC2 was 1.82 (95% CI, 0.94–3.52; increased trabecular separation and distribution and decreased trabecular number; p = 0.075). The OR for an SD decrease in PC3 was 1.61 (95% CI, 0.94–2.77; decreased load carried by the trabecular bone at the distal and proximal sites; p = 0.086).

DISCUSSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. REFERENCES

Our study showed that, in addition to BMD, both microarchitecture and bone mechanical properties assessed by FEA based on in vivo HR-pQCT images were associated with fragility fracture of the wrist in postmenopausal women. Using PCA, we also found that distal radius bone was best characterized by a combination of FEA-derived bone strength and quantity parameters, which explained one half of the variability in bone characteristics. Moreover, characterization of distal radius properties was further improved by adding a combination of trabecular architecture, distribution of load between trabecular and cortical bone, bone geometry, and subject morphology.

As expected, differences in density, microarchitecture, and bone mechanical parameters were observed between women with and without fracture. We found a significant association between fracture status and ultradistal radius aBMD, volumetric total and trabecular BMD, trabecular number, cortical thickness, and estimated failure load. Similar observations were made by Melton et al.(9) in a recent case-control study of 18 women with wrist fracture.

Correlations among architectural parameters have been previously reported with HR-pQCT, showing that trabecular architectural measurements were strongly correlated to trabecular density and that total density was strongly correlated to trabecular density and cortical thickness at the radius.(6) Moreover, aBMD measured by DXA at the radius was significantly correlated with microarchitecture.(7) In this study, we also found that estimated mechanical parameters such as failure load were moderately to strongly correlated with aBMD and volumetric BMD, as well as microarchitecture. This is in agreement with earlier studies in which radius BMD measured by DXA or pQCT correlated strongly with experimentally measured bone failure load.(18) In addition, Pistoia et al.(12) found strong correlations between the microarchitecture of the distal radius measured by 3D-pQCT, and experimentally measured bone failure load (r = 0.62–0.73 for TbN, TbSp, and CTh).

The estimated failure load prediction was based on a criterion reported by Pistoia et al.(12) It should be noted that, in that earlier study, the images were acquired using a larger voxel size (165 μm) than provided by the newer equipment used in this study (voxel size = 82 μm). However, our estimated bone failure loads were comparable with those reported in experimental studies of human cadaver forearms.(18–20) Hayes et al.(21) introduced the concept of factor of risk defined as the ratio of the applied load to the failure load. In this study, the load applied to the wrist was considered as the load applied to the outstretched hand during a fall from standing height.(22) Higher values of factor of risk indicate increasing risk of fracture. Here, women with wrist fractures did not differ from controls for height; thus, the mean subject-specific load was the same in women with fractures and controls (2631 ± 41 and 2634 ± 38 N, respectively; p = 0.75). Therefore, as estimated failure load was ∼15–16% lower in women with fracture; the factor of risk for wrist fracture was 16% higher in women with previous wrist fracture (1.08 ± 0.16) compared with controls (0.93 ± 0.19; p < 0.001). Trabecular and cortical average and SD stresses were also consistent with results found in different scenarios for simulated bone atrophy in the distal radius.(23)

The strong intercorrelation among the variables of interest led us to conduct a PCA to highlight differences and similarities in those structural and mechanical parameters and thus to reduce the number of variables in future analyses. We excluded two parameters from these analyses: (1) trabecular thickness, which is calculated from trabecular density and number, assuming fully mineralized bone to have a mineral density of 1.2 g HA/cm3, whereas changes in bone remodeling rate are known to influence this degree of bone mineralization(24); and (2) cortical density, which should be considered with caution because of partial volume effects that may limit the reliability of this measurement. We found that 86% of the variability of distal radius characteristics was explained with the five first noncorrelated PCs. Interestingly, stiffness and estimated bone failure load had the highest weights of all parameters within the first PC. PC1 also included areal and volumetric density, as well as cortical thickness, with a high correlation coefficient.

The PC2, which reflected trabecular microarchitecture, also tended to be associated with fracture status. It should be noted that our population is part of a larger case-control study including 101 women with fracture and 101 controls.(7) In that study, focusing on wrist fracture, we showed that, after adjusting for ultradistal aBMD, differences between women with fracture and controls remained significant for Dtot, Dtrab, and TbN and that there was a trend for TbSp and TbSpSD. Moreover, on the entire population, a trend was observed for an increased risk of fracture (all fragility fracture) for each SD change of TbN, TbSp, and TbSp SD after adjusting for aBMD at the distal radius. This finding is also consistent with previous studies reporting an association between fracture status and trabecular architecture partially independent of BMD.(6,9)

The PC3, which reflected the distribution of load between trabecular and cortical bone at the distal and proximal regions, also tended to be associated with fracture status. This means that when trabecular bone carried less load, or inversely when cortical bone carried more load, at the distal and proximal sites, the risk of fracture was increased. In the control group, the percentage of load carried by the cortical and trabecular bone was such that there was a better distribution of load between trabecular and cortical bone than in the fracture group. However, in both groups, the load was mostly applied to the cortical bone, ranging from 60% at the distal region to 91% at the proximal region. Ex vivo studies confirm that cortical bone is an important determinant of forearm bone strength.(18,20) In a recent in vivo study, Melton et al.(9) reported that the percentage of load borne by the cortex at the ultradistal radius was 63%, but they did not find a difference between fractures and controls, which may likely be explained by their limited sample size. Moreover, in an analytical model for simulated bone atrophy in the distal radius, failure load was more profoundly affected by a decrease in cortical thickness than by a reduced trabecular thickness or number.(23) The authors also suggested that the load distribution between cortical and trabecular bone plays an important role in bone strength, something that was confirmed in this study. This distribution was closely linked with the evolution of cross-sectional area, trabecular architecture, and cortical thickness, along the axial direction from the distal joint line to proximal radial shaft. Some studies reported that the progression of osteoporosis is not necessarily uniform in axial and perpendicular parts of bones and that trabeculae may be lost at a faster rate or earlier at the proximal region than at the denser trabecular skeleton located more distally.(25–29) In our study, we did not find a difference in the evolution of densities and microarchitecture along the axial direction in women with and without fracture. However, it should be noted that our entire region of interest is only a 9-mm slice of bone; thus, it may not have been large enough to observe such differences.

Our study had several limitations. First, the sample size was limited, which may explain the borderline significance for the association of PC2 and PC3 with wrist fracture. Second, boundary and loading conditions used in the FEAs were simplified. We simulated a compression test on a planar parallel slice of the radius, which is not necessarily an accurate representation of the loading applied to this region during a fall on the outstretched hand. However, Pistoia et al.(13) showed that failure load predictions based on a 1-cm cross-sectional region at the distal radius were highly correlated with those based on a full 4-cm region to which more realistic boundary conditions were applied. Material properties were fixed and did not take into account the bone tissue properties, whereas it has been recently shown that applying material properties proportional of BMD may improve the results of FEAs.(30) Nevertheless, to make this technique available in epidemiologic and clinical studies, simplifications of the model and the boundary conditions are currently required to reduce computing time. Finally, our study was cross-sectional, with HR-pQCT images available only after the fractures had occurred, and therefore, the measurements may not necessarily reflect the architecture at the time of the fracture. However, examining these associations as a function of time interval between the fracture date and the measurements procedure did not alter results (data not shown).

These limitations notwithstanding, the results indicate that bone mechanical properties derived from FEA of in vivo HR-pQCT images are associated with wrist fracture. Furthermore, PCA suggested that some of these FEA-derived mechanical properties provide information about skeletal fragility and fracture risk that is not captured by BMD or architecture measurements alone. In particular, the distribution of load between cortical and trabecular bone, assessed by FEA, seems promising for improving wrist fracture prediction. Altogether, these data provide strong rationale for additional prospective studies testing the ability of FEA based on HR-pQCT to predict fracture risk.

Acknowledgements

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. REFERENCES

The authors thank A Bourgeaud-Lignot, B Vey-Marty, and W Wirane for valuable technical assistance. This study was supported in part by research grants from Eli Lilly and from the Société Française de Rhumatologie to INSERM.

REFERENCES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. REFERENCES
  • 1
    Cummings SR, Bates D, Black DM 2002 Clinical use of bone densitometry: Scientific review. JAMA 288: 18891897.
  • 2
    Kelsey JL, Browner WS, Seeley DG, Nevitt MC, Cummings SR 1992 Risk factors for fractures of the distal forearm and proximal humerus. The Study of Osteoporotic Fractures Research Group. Am J Epidemiol 135: 477489.
  • 3
    Nguyen TV, Center JR, Sambrook PN, Eisman JA 2001 Risk factors for proximal humerus, forearm, and wrist fractures in elderly men and women: The Dubbo Osteoporosis Epidemiology Study. Am J Epidemiol 153: 587595.
  • 4
    Seeman E, Delmas PD 2006 Bone quality–the material and structural basis of bone strength and fragility. N Engl J Med 354: 22502261.
  • 5
    Laib A, Newitt DC, Lu Y, Majumdar S 2002 New model-independent measures of trabecular bone structure applied to in vivo high-resolution MR images. Osteoporos Int 13: 130136.
  • 6
    Boutroy S, Bouxsein ML, Munoz F, Delmas PD 2005 In vivo assessment of trabecular bone microarchitecture by high-resolution peripheral quantitative computed tomography. J Clin Endocrinol Metab 90: 65086515.
  • 7
    Sornay-Rendu E, Boutroy S, Munoz F, Delmas PD 2007 Alterations of cortical and trabecular architecture are associated with fractures in postmenopausal women, partially independent of decreased BMD measured by DXA: The OFELY study. J Bone Miner Res 22: 425433.
  • 8
    MacNeil JA, Boyd SK 2007 Load distribution and the predictive power of morphological indices in the distal radius and tibia by high resolution peripheral quantitative computed tomography. Bone 41: 129137.
  • 9
    Melton LJ III, Riggs BL, van Lenthe GH, Achenbach SJ, Muller R, Bouxsein ML, Amin S, Atkinson EJ, Khosla S 2007 Contribution of in vivo structural measurements and load/strength ratios to the determination of forearm fracture risk in postmenopausal women. J Bone Miner Res 22: 14421448.
  • 10
    Ulrich D, van Rietbergen B, Laib A, Ruegsegger P 1999 Load transfer analysis of the distal radius from in-vivo high-resolution CT-imaging. J Biomech 32: 821828.
  • 11
    Van Rietbergen B, Muller R, Ulrich D, Ruegsegger P, Huiskes R 1999 Tissue stresses and strain in trabeculae of a canine proximal femur can be quantified from computer reconstructions. J Biomech 32: 165173.
  • 12
    Pistoia W, van Rietbergen B, Lochmuller EM, Lill CA, Eckstein F, Ruegsegger P 2002 Estimation of distal radius failure load with micro-finite element analysis models based on three-dimensional peripheral quantitative computed tomography images. Bone 30: 842848.
  • 13
    Pistoia W, van Rietbergen B, Lochmuller EM, Lill CA, Eckstein F, Ruegsegger P 2004 Image-based micro-finite-element modeling for improved distal radius strength diagnosis: Moving from bench to bedside. J Clin Densitom 7: 153160.
  • 14
    Laib A, Hauselmann HJ, Ruegsegger P 1998 In vivo high resolution 3D-QCT of the human forearm. Technol Health Care 6: 329337.
  • 15
    van Rietbergen B, Weinans H, Huiskes R, Odgaard A 1995 A new method to determine trabecular bone elastic properties and loading using micromechanical finite-element models. J Biomech 28: 6981.
  • 16
    Turner CH, Rho J, Takano Y, Tsui TY, Pharr GM 1999 The elastic properties of trabecular and cortical bone tissues are similar: Results from two microscopic measurement techniques. J Biomech 32: 437441.
  • 17
    Jolliffe IT 2002 Statistics, 2nd ed. Springer, New York, NY, USA.
  • 18
    Muller ME, Webber CE, Bouxsein ML 2003 Predicting the failure load of the distal radius. Osteoporos Int 14: 345352.
  • 19
    Wigderowitz CA, Paterson CR, Dashti H, McGurty D, Rowley DI 2000 Prediction of bone strength from cancellous structure of the distal radius: Can we improve on DXA? Osteoporos Int 11: 840846.
  • 20
    Ashe MC, Khan KM, Kontulainen SA, Guy P, Liu D, Beck TJ, McKay HA 2006 Accuracy of pQCT for evaluating the aged human radius: An ashing, histomorphometry and failure load investigation. Osteoporos Int 17: 12411251.
  • 21
    Hayes WC, Piazza SJ, Zysset PK 1991 Biomechanics of fracture risk prediction of the hip and spine by quantitative computed tomography. Radiol Clin North Am 29: 118.
  • 22
    Chiu J, Robinovitch SN 1998 Prediction of upper extremity impact forces during falls on the outstretched hand. J Biomech 31: 11691176.
  • 23
    Pistoia W, van Rietbergen B, Ruegsegger P 2003 Mechanical consequences of different scenarios for simulated bone atrophy and recovery in the distal radius. Bone 33: 937945.
  • 24
    Boivin G, Meunier PJ 2002 Changes in bone remodeling rate influence the degree of mineralization of bone. Connect Tissue Res 43: 535537.
  • 25
    Genant HK, Block JE, Steiger P, Glueer CC, Ettinger B, Harris ST 1989 Appropriate use of bone densitometry. Radiology 170: 817822.
  • 26
    Hannan MT, Felson DT, Anderson JJ 1992 Bone mineral density in elderly men and women: Results from the Framingham osteoporosis study. J Bone Miner Res 7: 547553.
  • 27
    Davis JW, Ross PD, Wasnich RD 1994 Evidence for both generalized and regional low bone mass among elderly women. J Bone Miner Res 9: 305309.
  • 28
    Majumdar S, Link TM, Augat P, Lin JC, Newitt D, Lane NE, Genant HK 1999 Trabecular bone architecture in the distal radius using magnetic resonance imaging in subjects with fractures of the proximal femur. Magnetic Resonance Science Center and Osteoporosis and Arthritis Research Group. Osteoporos Int 10: 231239.
  • 29
    Steiger P, Cummings SR, Black DM, Spencer NE, Genant HK 1992 Age-related decrements in bone mineral density in women over 65. J Bone Miner Res 7: 625632.
  • 30
    Cole J, Myers ER, Van der Meulen MC 2007 Finite element models more accurately predict structural behavior of human cancellous bone when using specimen-specific tissue properties. Presented at the 29th Annual Meeting of the American Society for Bone and Mineral Research, Honolulu, HI, USA, September 16–19, 2007.