The results from this work show that µMRI-based structural images from trabecular and cortical bone at the distal extremities acquired in patients are suited to derive finite-element models for estimating the mechanical consequences of estrogen loss following menopause and estrogen supplementation. This study is only one of a small number of in-vivo image-based studies compellingly showing not only the structural but also mechanical manifestations of intervention. A prior study involving a smaller number of subjects had provided evidence of the architectural(18
) and, subsequently, mechanical effects of testosterone replacement in men with hypogonadism.(28
) Unlike in that work, the effects expected here were smaller in consideration that the women studied were early postmenopausal (mean approximately 1 year from last menstrual period), in contrast to testosterone-deficient men who had been hypogonadal for very long times and therefore eliciting a very strong treatment response. Further, this work goes considerably beyond the prior pilot study in that we performed subregional µFE analysis of a small cuboid trabecular subregion (which, however, was larger than in ref 28)—10 × 10 × 5 mm3
versus 5 × 5 × 5 mm3
—but also examined the role of the cortex by performing full cross-sectional stiffness analysis similar to that described by MacNeil and Boyd.(40
) Further, we studied the hypothesized mechanical treatment effects at both the distal tibia and the radius, that is, a load-bearing and a non-load-bearing anatomic site. We further brought advanced registration methods to bear for mutual alignment of the images at the three time points and with the scanner coordinate frame.(33
In the tibia, the subregion analysis revealed significant reductions in all six moduli in the control group (ie, women who elected not to take estrogen) ranging from 3% to 5% from baseline after 24 months (p
< .05 to <.005), whereas only three of the moduli were significantly lower after 12 months. In contrast, in the estradiol group, all six mechanical constants remained unaltered at both follow-up time points. Between-group changes in the elastic moduli, in contrast to whole-section axial stiffness values, were not significant. These results are somewhat at variance with the structural changes reported for the same cohort of patients in ref. (20
). This discrepancy is likely due to the constraints imposed by the mechanical analysis in the choice of the subvolume in the present work (10 × 10 × 5 mm3
parallelepiped at the centroid versus larger irregularly shaped region for prior structural analysis in ref 20). It is interesting to note, however, that despite the highly significant decrements in the mechanical constants in the control subjects, BV/TV decreased only marginally at 24 months (−1.8%, p
= .03), whereas the reduction at 12 months did not reach significance (−1.0%, p
= .07). The data thus suggest that the predicted changes in elastic and shear moduli are largely the result of structural changes, notably the observed reduction in plate-to-rod ratio observed previously in the data.(20
) Similar topologic changes had been previously observed in paired biopsies following early menopause.(41
Of particular relevance are the effects of intervention on whole-section axial stiffness at both anatomic sites. In the tibia, women in the control group suffered a loss in estimated stiffness of about 4% within a year (p < .005) while gaining over 6% in response to estrogen supplementation during the same period (p < .05). In the radius, whole-section axial stiffness declined by about 4% during the initial 12 months (p < .005) and increased marginally in the estradiol group (by about 2.5%, p < .05). Of significance further are the large intergroup differences in change from baseline: 10% after 12 months in the tibia (p = .0001) and 6% in the radius (p < .0005).
The effect of estradiol is clearly greater at the tibia than at the radius, suggesting that the load-bearing site may be more sensitive to antiresorptive treatment. Possible synergistic effects of antiresorptive agents with mechanical stimulation have been studied widely (see, for example, ref (42
)). However, most work relies on studies in animals. For example, Westerlind and colleagues(43
) found that estrogen administration to ovariectomized rats reduced bone loss in the unloaded and prevented loss in the loaded limb following unilateral sciatic neurotomy. Tromp and colleagues(44
) also found an interaction between estrogen and mechanical loading on bone in the ovariectomized rat model.
The highly significant between-group changes from baseline in whole-section axial stiffness at both anatomic sites emphasize the role of the cortex as being affected by estrogen depletion and supplementation. The findings are counter to the notion that short-term remodeling changes after menopause (both loss owing to reduction in estrogen levels or accrual from estrogen supplementation) affect primarily trabecular bone, which remodels faster than cortical bone. Further, the within-group increases in trabecular bone elastic and shear moduli, as well as whole-section axial stiffness in the estradiol group suggest that estrogen supplementation not only preserves but actually improves bone mechanical competence, in contrast to prior structural studies indicating mere preservation of architecture.(45
) Nevertheless, the anabolic effects of estrogen replacement are well documented. Khastgir and colleagues found in older women large increases in trabecular bone volume and bone mineral density (BMD). as well as increased wall thickness, in response to estrogen supplementation.(46
) Similarly, based on paired bone biopsies obtained before and after 2 years of hormone replacement in postmenopausal women, Vedi and colleagues demonstrated significantly higher cortical width.(47
The lack of further significant gains from 12 to 24 months is somewhat puzzling but parallels the structural observations made previously.(20
) Nevertheless, in the tibia, four of the elastic moduli that had not changed significantly after 12 months were all lower after 24 months (). In contrast, there was no further decline in whole-section stiffness after 24 months and only an insignificant further increase in the estradiol group. There are several possible reasons for these observations. First, lifestyle changes in the control group could have mitigated further declines in structural and mechanical parameters. Second, in these very early postmenopausal women, the degradation in bone quality is likely to be largest early after the onset of menopause. Third, the number of subjects proceeding to the 24-month time point was lower than those who were examined at baseline.
This study has limitations because it ignores effects of treatment on the intrinsic bone tissue properties. Paschalis and colleagues(2
) found in early postmenopausal women that 2 years of estrogen/progestin supplementation increased mineral/matrix ratio, mineral crystallinity/maturity, and the relative ratio of collagen cross-links, clearly showing that antiresorptive treatment affects material properties. The same group recently has provided evidence that subjects with vertebral fractures differ in their matrix and mineral composition (ie, collagen maturity, mineral/matrix ratio, and crystallinity).(48
) However, the work did not provide data on structural group differences, and BMD was matched at the proximal femur rather than at the vertebrae.
As with all CT-based µFE work using in vivo structural images to create the FE model.(49
) our model analyzes the structural implications only of longitudinal changes on mechanical parameters. While CT can, in principle, make use of variations in mineralization density, such differences are not distinguishable from partial-volume averaging in the limited spatial resolution regime of in vivo imaging, where resolution is comparable with trabecular thickness, nor would CT be able to quantify changes in matrix chemistry. Also, with few exceptions,(36
) all prior in vivo µFE work was based on thresholded images in that a fixed tissue modulus was given to the voxels assigned to bone. We found in preliminary work grayscale-based µFE analyses to be more accurate than those relying on binarized images,(51
) which is what prompted us to avoid binarization, since it is well known to be fraught with error at the resolution achievable by in vivo structural imaging of trabecular bone. Finally, the study is limited to the linear regime in that ultimate strength was not evaluated, which could be obtained from the present data by invoking the Pistoia criterion,(52
) which suggests that failure will occur once 2% of the bone tissue is strained beyond 0.7%.(52
) Nevertheless, there is substantial evidence that yield strength and stiffness are highly correlated (r2
> 0.95; see, for example, ref. (36
In conclusion, MRI-based µFE analysis is able to predict the short-term mechanical implications of drug intervention of both trabecular and cortical bone. The data provide evidence of an anabolic effect of estrogen supplementation in terms of whole-section axial stiffness at both the distal tibia and radius, but the effects, in terms of both reduction in mechanical competence in the absence of estradiol administration and the observed increases in subjects receiving hormone supplementation, were greater at the tibia than at the radius.