Prediction of Vertebral and Femoral Strength In Vitro by Bone Mineral Density Measured at Different Skeletal Sites

Authors

  • Xiao G. Cheng,

    1. Arthritis & Metabolic Bone Disease Research Unit, University Hospitals, Catholic University of Leuven, Belgium
    2. Department of Radiology, UCSF, San Francisco, California, U.S.A.
    Search for more papers by this author
  • Geert Lowet,

    1. Division of Biomechanics and Engineering Design, Catholic University of Leuven, Belgium
    Search for more papers by this author
  • Steven Boonen,

    1. Arthritis & Metabolic Bone Disease Research Unit, University Hospitals, Catholic University of Leuven, Belgium
    Search for more papers by this author
  • Patrick H. F. Nicholson,

    1. Division of Biomechanics and Engineering Design, Catholic University of Leuven, Belgium
    2. Orthopaedic Biomechanics Laboratory, Beth Israel Deaconess Medical Center and Harvard Medical School, Boston, Massachusetts, U.S.A.
    Search for more papers by this author
  • Georges Van Der Perre,

    1. Division of Biomechanics and Engineering Design, Catholic University of Leuven, Belgium
    Search for more papers by this author
  • Jan Dequeker

    Corresponding author
    1. Arthritis & Metabolic Bone Disease Research Unit, University Hospitals, Catholic University of Leuven, Belgium
    • Prof. Dr. J. Dequeker, Arthritis & Metabolic Bone Disease Research Unit, U.Z. Pellenberg, Division of Rheumatology, Weligerveld 1, B-3212 Pellenberg, Belgium
    Search for more papers by this author

  • Presented in part at the 19th Annual Meeting of the American Society for Bone and Mineral Research, Cincinnati, Ohio, U.S.A, September, 1997 (J Bone Miner Res 1997;12 [Suppl]:S130).

Abstract

The aim of the present study was to investigate the prediction of vertebral and femoral strength in vitro by bone mineral density (BMD) measured at different skeletal sites. The third lumbar vertebral body, the right proximal femur, and the right calcaneus were removed from 38 male and 32 female cadavers (mean age 69 years, range 23–92 years). Areal BMD of all bone specimens was determined by dual-energy X-ray absorptiometry (DXA). The failure load of the vertebral body and the femur was determined by mechanical testing. Vertebral and femoral strength were both greater in males than females (p < 0.01), as was BMD at all sites (p < 0.01). Vertebral strength correlated well with vertebral BMD (r2 = 0.64) but was only moderately correlated with BMD measured at the femur (r2 = 0.36) or the calcaneus (r2 = 0.18). Femoral strength showed the highest correlations with femoral BMD (r2 = 0.88) and somewhat weaker relationships with BMD at the vertebra (r2 = 0.50) and the calcaneus (r2 = 0.54). BMD values at the vertebra, femur, and calcaneus were only moderately interrelated (r2 = 0.31–0.65), and vertebral strength correlated only modestly with the strength of the femur (r2 = 0.36). These in vitro results support the concept that optimal prediction of vertebral or femoral strength by DXA requires site-specific assessments.

INTRODUCTION

MANY CROSS-SECTIONAL and prospective studies have documented that low bone mineral density (BMD) at different anatomic sites is significantly associated with the risk of hip, spine, and forearm fractures.1–5 In addition, these studies suggest that the risk of fracture is not only significantly associated with bone density measured at the anatomical site of the fracture but also with BMD measurements made at other skeletal sites.6 However, at least one study in elderly women has indicated that hip fractures are best predicted by BMD measurements specifically at the proximal femur.7

While BMD measurements at different sites are intercorrelated, the correlations are not strong enough to permit the prediction of BMD values for one site from measurements of another.8,9 Thus, for fracture prediction at a particular skeletal site, BMD measurements of that same site are preferred,9 since bone strength may correlate poorly with BMD measured at a different skeletal site. Previous studies have confirmed that proximal femoral strength in vitro is highly correlated with dual-energy X-ray absorptiometry (DXA)-determined BMD at the same site, with correlation coefficients ranging from r2 = 0.71–0.92 for femoral neck BMD and r2 = 0.73–0.88 for trochanteric BMD.10–12 For the spine, somewhat lower correlations (r2 = 0.59–0.67) have been reported between vertebral strength and BMD as measured by DXA in the same specimens.13–15

It is known that BMD measured by DXA at the spine can be influenced by the presence of osteophytes16–18 and calcification of aorta,19 particularly in the elderly. Thus, it is not unexpected that BMD measurements of the proximal femur are considered superior to vertebral BMD measurements for estimating femoral strength,18,20 since the proximal femur is less frequently involved in osteoarthritis. However, the prediction of vertebral strength by BMD measurements made at the proximal femur has not been addressed. A very recent study by Augat et al.21 has demonstrated that geometric properties of the distal radius assessed by peripheral quantitative computed tomography not only correlated well with bone strength at the same skeletal site (r2 = 0.83) but also with the strength of the vertebra (r2 = 0.56) and of the femur (r2 = 0.79). However, the prediction of vertebral or femoral strength by DXA-determined BMD at other skeletal sites has not been studied extensively.

Our earlier reports have demonstrated that femoral strength is highly correlated with DXA-determined BMD at the trochanteric region (r2 = 0.88) and the femoral neck (r2 = 71).12 Similarly, vertebral BMD as measured by DXA accounts for 64% of the variance of vertebral strength.15 In this paper, we report the prediction of vertebral or femoral strength by DXA-determined BMD at the vertebra, the femur, and the calcaneus.

MATERIALS AND METHODS

The third lumbar vertebral body (L3), the right proximal femur (six femoral prostheses were excluded), and the right calcaneus were removed from 38 male and 32 female cadavers at autopsy within the framework of the EC BIOMED1 Concerted Action “Assessment of Bone Quality in Osteoporosis.”22 The mean age at death was 69 years, range 23–92 years. The characteristics of this population have been detailed elsewhere.12,15,23 In brief, the study group consisted of 70 unscreened subjects presenting at postmortem. The specimens were stored at –20°C. High resolution contact radiographs of the vertebrae, the femora, and the calcanei were taken with a Faxitron X-ray unit (Hewlett-Packard, McMinnville, OR, U.S.A.) to screen for pre-existing fractures and metastases, scoliosis, and other skeletal diseases.

A Hologic 1000W DXA scanner (Hologic Inc., Waltham, MA, U.S.A.) was used to measure the areal BMD (g/cm2) for all specimens. A 3.5-cm-thick Plexiglas sheet was placed beneath the bone to simulate the presence of soft tissue. The vertebral body was scanned in the posterior-anterior projection,15 the proximal femur was scanned in the simulated anteverted position,12 whereas for the calcaneus, BMD was measured in a 3-cm-wide rectangular region placed over the posterior part of the calcaneus corresponding to the site commonly used for clinical quantitative ultrasound measurements.23

All vertebral and proximal femoral specimens were tested to failure on a material testing machine. The configuration of the femoral test was designed to simulate a fall with impact to the greater trochanter,10,11 as has been described previously.12 The load was applied to the femoral head at a displacement rate of ∼14 mm/s until failure.12 Afterward, radiographs were taken to classify the fracture type as either cervical or trochanteric.12,23 The ultimate load of the third lumbar vertebra was determined by a uniaxial compressive test at a displacement rate of about 8 mm/s. The loading conditions for this test have been previously reported.15 Again, radiographs were taken after mechanical testing to verify the fracture mode. Five vertebrae with pre-existing fractures were excluded, and a further three vertebrae were excluded because of technical problems during testing.

Simple linear regression analysis was used to assess the relationship between BMD and fracture load, with the standard error of estimate (SEE) expressed in percentage form as 100 × SEE/mean. Differences between correlation coefficients were assessed using Fisher's Z transformation.24 Multiple regression analysis was used to test whether combinations of BMD at various sites improved the prediction of bone strength. Probabilities were considered significant at the 5% level.

RESULTS

Table 1 gives the summary of the descriptive statistics of the variables studied. Males had significantly higher bone strengths and BMD values than females.

Table TABLE 1. THE SUMMARY OF THE STATISTICS OF THE VARIABLES
original image

The correlations (r2) between vertebral or femoral strength and DXA-determined BMD at the vertebra, the femur, and the calcaneus are shown in Table 2. While vertebral BMD accounted for 64% of the variance in vertebral strength, its performance was poorer than that of trochanteric BMD in predicting femoral strength (r2 = 0.88; p < 0.05, Fisher's Z transformation). Femoral strength was more highly correlated with BMD measurements in the femoral trochanteric region than with BMD at the vertebra or calcaneus (p < 0.05, Fisher's Z transformation). Spinal BMD gave the highest correlation with vertebral strength, although the differences in predictive ability between the different BMD measurement sites did not, in this case, reach statistical significance (Fisher's Z transformation). Multiple regression analysis revealed that combining the vertebral, femoral, and calcaneal BMD did not improve the prediction of vertebral strength compared with vertebral BMD alone. Likewise, the correlation between femoral strength and femoral BMD was not improved by including vertebral and calcaneal BMD in the regression analysis.

Table TABLE 2. CORRELATIONS (R2) of Vertebral or Femoral Strength with BMD by DXA
original image

Table 3 shows the correlation matrix for BMD measurements at the vertebra, the femur, and the calcaneus. Figure 1 illustrates the relationship between vertebral and femoral strength (r2 = 0.36, p < 0.01).

Table TABLE 3. CORRELATIONS (R2) Matrix Between BMD by DXA Made at Different Sites
original image
Figure FIG. 1.

Relationship between L3 and femoral strength (r2 = 0.36, p < 0.01). Open circles indicate males and filled circles indicate females.

DISCUSSION

In the clinical setting, BMD can be measured by DXA at various sites, but the effectiveness of combining BMD measurements from different sites has not been conclusively demonstrated. Wasnich25 and Davis et al.26 have suggested that the combination of BMD measurements from multiple anatomical sites might be helpful in predicting the risk of spine fractures for the individual patient. However, Black and colleagues27 investigated the utility of combining hip and spine BMD to identify a high-risk group for hip fracture in a large prospective study (the Study of Osteoporotic Fractures). They found that after adjusting for age and femoral neck BMD, spinal BMD was no longer significantly associated with the risk of fracture. In the present cadaver study, the principal finding was that trochanteric BMD alone explained 88% of the variance in femoral strength, and combinations of femoral, vertebral, and calcaneal BMD did not improve the predictive ability. Similarly, femoral and calcaneal BMD did not contribute significantly to the prediction of vertebral strength compared with vertebral BMD alone. This suggests that additional scans at other sites are of little value in predicting bone strength if BMD measurements are available at that specific site. However, spinal BMD may be an exception. Spinal degenerative joint disease,16,17 intervertebral disk narrowing, scoliosis, or aortic calcification19 can affect vertebral BMD measurements. In the light of this, BMD measurements at other sites, such as the radius, femur, and calcaneus, may be considered potentially valuable.

In the present study, we found BMD measured in vitro at the vertebra, the femur, and the calcaneus to be moderately intercorrelated (r2 = 0.31–0.65), consistent with previous reports.8,9,25,26 Our data also revealed a rather poor relationship between bone strength at the femur and the vertebra (r2 = 0.36), which may reflect, in part, the different geometry and composition of the two bones and the differences in the type of mechanical test performed. However, these observations do indicate significant heterogeneity in BMD and bone strength among different sites in the same individual and that these properties may therefore be determined to a large extent by local factors. This again suggests that the optimal prediction of bone strength or fracture risk assessment requires site-specific BMD measurements.

While vertebral BMD accounted for 64% of variance in vertebral strength, its performance was significantly poorer than that of trochanteric BMD in predicting femoral strength (r2 = 0.88). Several factors might account for this observation. Spinal BMD measurements will be affected by the intrinsic integration of the mineral content of the posterior elements, even in the absence of osteophytes. The inclusion of the posterior elements could be particularly important in osteoporosis since bone loss occurs preferentially in the vertebral body. In addition, vertebral and femoral specimens were tested under distinct loading conditions and the failure mechanisms are likely to have been very different. The poorer correlation between vertebral strength and BMD suggests that strength at this site is more profoundly affected by factors that are not quantitated by DXA. For instance, the geometrical and structural heterogeneity at the vertebra may be significantly greater than at the femur. Also, since the vertebral bodies consist primarily of cancellous bone, trabecular microstructure might play a more significant role in determining vertebral strength than at the highly cortical proximal femur.

Significant gender-related differences in both bone strength and BMD were observed. These may reflect differences in bone size between the genders. In an earlier report,15 using the same specimens as the present work, we found that correcting for cross-sectional area (i.e., in effect expressing strength as ultimate stress rather than ultimate load) negated the gender-related difference in femoral strength. However, in the vertebra we found that a gender-related difference remained even after such a normalization.15 This appears consistent with the hypothesis that stuctural or geometrical factors play a relatively larger role in the vertebra compared with the femur.

This study adds to the growing body of evidence that useful correlations exist between bone strength and DXA-determined BMD. However, while the predictive ability of BMD is well established, these correlations do not tell us why, from a biomechanical viewpoint, such relationships exist. From engineering principles it is known that bone strength is a function of the local geometric properties and the intrinsic mechanical properties of the bone tissue. DXA BMD measurements represent an integrated measure of the bone mineral within a given projected area, and as such reflect both true BMD and bone geometry. Given the complexity of the situation, the present study addresses only limited aspects of this problem. More sophisticated studies are needed to establish the underlying causative relationships between geometrical, structural, and material properties and bone strength.

In summary, this cadaver study indicates that the optimal prediction of vertebral or femoral strength by DXA requires site-specific BMD measurements. Furthermore, the predictive ability of a site-specific BMD for bone strength is not improved by combining additional BMD measurements from other sites.

Acknowledgements

This work was supported by the EC BIOMED1 Concerted Action “Assessment of Quality of Bone in Osteoporosis” contract number BMH1-CT92–0296. We are grateful to Dr. K.G. Faulkner of Oregon Osteoporosis Center, Portland, OR, U.S.A. for his helpful remarks. We wish to thank Prof. E.K. Verbeken for providing the bone specimens, Mr. H. Borghs for the DXA measurements, and Mrs. J. Cartois and Ms. A. Vandereijcken for expert secretarial help.

Ancillary