Have the DXA-Based Exercise Studies Seriously Underestimated the Effects of Mechanical Loading on Bone?

To the Editor:

In a recent issue of the Journal, Adami et al.¹ reported interesting results of a controlled 6-month exercise intervention trial of 250 postmenopausal women. The exercise group took part in a training program designed to load primarily the wrist region. At baseline and after the training, dual-energy X-ray absorptiometry (DXA) was performed at the lumbar spine, proximal femur, and radius, and peripheral quantitative computed tomography (pQCT) at the proximal and ultradistal radius. DXA showed no significant training effects in the areal bone mineral density (BMD) at any measured site. In contrast, pQCT measurement of the ultradistal radius revealed a significant training effect in the cross-sectional area of the cortical bone (+3%) and thus also in the cortical bone mineral content (BMC) (+3%), and, of note, a decrease in the trabecular BMC (−4%). At the cortical region of the proximal radius, there was no training effect. The authors concluded that despite no effect on the total BMC, the stressed ultradistal radial bone segment adapted to incident loading by increasing both the cross-sectional area and density of the cortical component at apparent expense of the trabecular component of the same segment. They stated further that the observed structural changes were theoretically associated with improved bending strength of the bone and expressed the need for additional studies to confirm this hypothesis.

We have already addressed this issue in our recent exercise trial of rats, a study in which the effects of sudden impact loading regimen on mineral and mechanical properties of rat femora were evaluated.² In full accordance with the results of Adami et al.,¹ we neither observed differences in the DXA-derived BMC nor the external dimensions of the femoral necks between the intervention and control groups. However, the failure load of the femoral neck was significantly higher (+14%) in the impact-loaded animals, the finding suggesting redistribution of bone mineral and thus “stronger” cortex of the femoral neck. Subsequently, we speculated that if similar changes in the cross-sectional and mechanical characteristics of the bones (without simultaneous gain in BMC) are true in humans, too, a serious reevaluation of the results of the numerous longitudinal exercise studies,³ which have shown no or only mild-to-moderate exercise-induced gains in BMC and BMD, must be made; the true increases in bone strength may have been much better than indicated by these BMC–BMD changes alone.

Adami et al.¹ proposed two different strategies by which bones can cope with the mechanical demands caused by increased loading: (1) reshaping (periosteal expansion) of the bone cross-section, and (2) redistribution of bone mineral from the trabecular component to the cortical one. Both of these adaptive responses can deposit the bone mineral further from the centroid of the bone, and, according to common engineering principles, increase the bending and torsional strength of the bone. However, any actual bone strength index data (e.g., density-weighed polar section modulus, BSI) were not presented by Adami et al.¹ The BSI takes the bone size and shape, as well as the density distribution in the measured bone section, into account and would thus have provided a reasonable measure of bone strength in the study of Adami et al.¹

As regards the loading-induced potential reshaping of bone sections, adult joints and physeal regions are not likely to substantially grow in size,⁴ a fact that apparently compels the changes in the apparent density (or mineral content) of the trabecular component and the cortical area and density to be virtually the only adaptive elements in the long bone ends. In light of this, the argued ability of the epiphyseal bone (i.e., ultradistal radius) to increase its total size in response to altered loading, as suggested by Adami et al.¹ is somewhat questionable, whereas the observed redistribution of bone mineral manifesting as a larger cortical area is likely. The reported marginal and insignificant increase (+0.4 ± 8.8%) in the pQCT-derived total area of the ultradistal radius hardly warrants the conclusion that periosteal apposition of bone really occurred in the study in question.

To our knowledge, Adami et al.¹ are the first to provide rather good direct clinical evidence that bones can adapt themselves to increased loading by transferring bone mineral from the trabecular bone compartment to the cortical shell. The study of Adami et al.¹ shows that bones tend to primarily adjust their apparent strength, not necessarily BMC nor BMD, to the changes that occur in their functional environment. For this reason, there is an urgent need to reevaluate the appropriateness of the current bone measurements for evaluation of the skeletal response to mechanical loading or other treatments.

DXA undoubtedly provides a reasonable overall picture of the bone status, but due to the inherent planar nature of the DXA data, it is bound to overlook such structural bone alterations (interior and periosteal) that can largely and independently influence the bone strength, the bottom line! Therefore, we recommend that in every future bone study in which structural alterations may occur, the DXA should be supplemented with measurements, such as pQCT, that provide mechanically more relevant and detailed information.^{1, 5}