In this study we used QCT-based FEA of lumbar and thoracic vertebrae of young and old men and women to determine age-related changes in mechanical strength, bone mass, and bone density of cortical and trabecular bone compartments. As expected, vertebral strength decreased with age for both men and women, but it decreased more dramatically in women than in men because of a greater decline in bone mass in both trabecular and peripheral bone compartments. Notably, in men there was little age-related decline in peripheral bone strength. These results provide evidence of a different compartment-specific pattern of age-related decline in vertebral bone mass and strength in women vs. men that may contribute to the higher incidence of vertebral fractures among women.
As expected, compressive strength predicted by finite element analysis was higher in men than women, and higher in L3 than T10, both of which can largely be explained by differences in bone size. It has previously been shown that vertebral compressive failure loads are lower in women, but estimated failure stresses are similar in both sexes,(26
) suggesting that vertebral size explains much of the difference in compressive failure loads between men and women. Our data support this, as both compressive strength and average vertebral cross-sectional area are larger in men than women, but no sex-related difference in estimated failure stress (vertebral body strength/average CSA) was observed. Similarly, previous studies(26
) have reported variation in compressive strength of human cadaveric vertebrae along the thoracic and lumbar spine, with an increase in vertebral compressive failure load and a decrease in estimated failure stress (failure load / average vertebral cross-sectional area) from the thoracic to lumbar spine.(26
) We observed a similar pattern, because T10 failure stress was higher than L3 failure stress for all groups. For these calculations we used the average CSA of the vertebral bodies. It is possible that minimum CSA instead of average CSA would yield different results for estimates of failure stress. Unfortunately, we are unable to calculate minimum CSA using our current software. However, we predict that differences observed between young and old and between thoracic and lumbar vertebrae will be maintained whether we normalize by average CSA or minimum CSA. In the absence of minimum CSA measures, geometric strength can provide similar information, because it is a strength measure that is wholly dependent on geometry (and presumably minimum CSA).
In contrast to sex-specific differences, age-related differences in compressive strength cannot be explained by changes in bone size but rather are due primarily to changes in bone mass and density. Geometric strength, a measure of the isolated contribution of bone geometry to compressive strength, was higher in old subjects than in young subjects, indicating that considering only bone size/geometry
, older subjects have stronger vertebrae than young subjects, but this age-related increase in geometric strength was generally small and did not offset age-related declines in overall vertebral strength. This finding is supported by previous studies that have shown an increase in cross-sectional area of vertebral bodies with age.(5
) However, it is well established that volumetric bone density (vBMD) declines in both men and women with age, resulting in an overall loss of vertebral body strength. Previous studies have shown that vBMD is similar in young men and women,(35
) and may even be slightly higher in women,(36
) but that women clearly exhibit a greater age-related decline in vBMD and compressive strength at the lumbar spine than men.(36
) Our study confirms these prior observations, because young men and women had similar vBMD values, yet the women exhibited significantly greater age-related declines in bone mass, density, and strength than men.
Age-related decline in vertebral compressive strength in men can be attributed almost exclusively to a decline in trabecular strength, because peripheral strength and density were largely maintained. In women, changes in both the trabecular and the peripheral compartment contribute to the age-related loss of strength, with a relatively larger loss in the trabecular compartment. As a result, the percentage of total bone strength attributable to the peripheral compartment increases with age in both men and women. Altogether, these data suggest that sex-specific differences in the age-related changes in cortical bone contribute to the lower incidence of vertebral fractures in men than in women. The finding that bending stiffness significantly decreased with age in women (−23% at T10, −34% at L3) but not in men (+3% at T10, −6% at L3) is likely due to differences in bone loss from the peripheral compartment. These differences in the peripheral compartment may result in a decreased resistance to loads induced by forward flexion for women relative to men, making women more vulnerable to sustaining wedge fractures.
The role of vertebral osteophytes must also be considered when interpreting the bone mass and strength changes in the peripheral compartment. Osteophytes are not specifically removed by the image processing used in this study and are therefore included in the peripheral bone measurements. Inclusion of osteophytes in the peripheral compartment may mask underlying age-related declines in bone mass and strength. In addition, because the peripheral compartment is defined as the outer 2 mm of bone in this study, areas with large osteophytes may cause the trabecular compartment to include some regions of cortical bone. One study reported a slightly higher prevalence of vertebral osteophytosis in men than in women older than age 50 (84% vs. 74%, respectively), although the distribution of osteophytes along the spine was similar in both sexes.(37
) In contrast, another study reported a higher prevalence of vertebral osteophytosis in women than in men.(38
) Altogether, these epidemiologic studies do not indicate a marked difference in the prevalence of osteophytes by sex, thus limiting the confounding role of osteophytes on sex-specific differences observed in the current study. Nonetheless, further studies may be needed to delineate compartment-specific changes in bone mass and bone strength without the possible confounding contribution of osteophytes.
Conventional assessment of spine BMD typically analyzes only vertebrae of the lumbar region (typically L2–L4 or L1–L4), yet many fractures occur in the thoracic spine. To estimate the error in predicting thoracic vertebral strength measurements from lumbar analyses, we determined the association between FE-determined lumbar and thoracic vertebral strength. We found a strong correlation between compressive strength estimates for T10 and L3 (r2
= 0.77 for all subjects), but when each age and sex group was considered individually, we found that the association was weaker in old vs. young subjects and also weaker in women vs. men (ie, r2
= 0.50 for old women). Similarly, Bürklein et al.(31
) compared the compressive strength of T6, T10, and L3 vertebrae in 119 cadavers and reported only modest correlations between the different levels (eg, r2
= 0.46 for T10 vs. L3). These results indicate that there is heterogeneity of vertebral strength along the spine. It remains to be determined whether clinical fracture risk assessment can be improved by assessing vertebral levels in both the thoracic and lumbar spine.
This study had several strengths that are novel contributions to study of vertebral fractures. First, we analyzed an age-stratified set of subjects from a community-based population. Therefore, the observed trends should reflect typical changes that occur in the population in general, although the racial representation for this study was primarily white people. Second, the use of finite element analysis and the controlled parameter studies enabled us to simulate different loading conditions and isolate the contributions of the trabecular and peripheral compartments to the strength of the whole bone. This provided unique insight into the role of the trabecular and peripheral compartments, which would be difficult to achieve using simpler structural models based on beam-and-column theories.
This study also had several limitations. First, the study was cross-sectional, and therefore age-related “changes” reported for bone strength or other contributing factors were inferred based on cross-sectional differences between young and old subjects. Second, the peripheral density and strength measurements included the outside 2 mm of bone, which contained trabecular as well as cortical bone, including osteophytes. In young subjects, the cortical shell of the vertebral bodies is approximately 400–500 µm thick and decreases to only 200–300 µm in elderly individuals.(39
) Ideally, to observe differences between cortical and trabecular bone, the peripheral shell would contain only cortical bone, but the spatial resolution of these clinical scans precludes accurate segmentation of this thin cortex. However, our previous micro-CT-based finite element analysis of T10 vertebrae, which captured the cortical shell at high resolution, have shown that the cortical shell supports approximately 40–50% of the compressive load.(41
) This is consistent with the load-sharing estimates of the peripheral bone in the current study, providing a degree of validation to these model predictions. A third limitation is that our sample size was modest (n
= 30 subjects/group), although it was adequately large for us to find significant differences for all variables examined. Fourth, the finite element models were loaded via PMMA plates at the top and bottom of the vertebral body, as is commonly done in cadaver studies. Again, our prior studies using micro-CT-based FEA have shown that overall load-sharing trends are relatively insensitive to the presence of a disc(41
) and thus we would not expect our reported trends to differ notably with an intervertebral disc instead of PMMA at the endplates. Finally, the method used to assess “peripheral” properties (eg, total strength – trabecular strength) ignores load sharing between the two compartments. A thorough analysis of load sharing between trabecular and cortical bone would require a high-resolution micro-CT-based analysis.(41
) Unfortunately, because of the resolution used to obtain the CT scans in this study, this type of analysis was not possible. However, the contributions of the individual compartments that we calculate with continuum models in the current study is consistent with what Eswaran reported with the micro-CT-based models, which suggests that by taking off the 2 mm of bone, we are effectively removing the cortical shell (about 0.4 mm thick) and adjacent trabeculae that would be unloaded upon removal of the cortical shell. Therefore, we conclude that removal of the outer 2 mm in the continuum models provides a good estimate of the results that would be obtained with removal of just the real cortical shell—because for the latter, the adjacent trabeculae become unloaded since there is no cortical shell to transmit load in the vertical direction to and from these trabeculae.(13