Improving the Assessment of Osteoporosis and Fracture Risk using DXA of the Calcaneus

Tishya A. L. Wren, Scott A. Yerby, Gary S. Beaupré, and Dennis R. Carter

VA Palo Alto Rehabilitation R&D Center and Stanford University Biomechanical Engineering Division

Objectives: Ten million Americans have osteoporosis, and 1.5 million osteoporotic fractures occur each year in the United States. These fractures are associated with significant morbidity and mortality, especially in elderly patients. To prevent osteoporotic fractures, clinicians must be able to identify individuals with low bone density who are at risk for osteoporotic fracture. Dual-energy X-ray absorptiometry (DXA) of the calcaneus is one of the best procedures available for assessing bone density and fracture risk at many commonly fractured sites including the hip and spine. However, DXA normally gives an areal bone mineral density (BMD) that depends on both volumetric bone density and bone size. We propose that the assessment of osteoporosis and fracture risk can be improved if volumetric densities are used to remove the confounding effects of bone size. The purpose of this study was to develop a simple method for determining such volumetric densities of the calcaneus.

Methods: We measured the volume (V_disp) and average medial-lateral thickness (t_calip) of 23 excised human calcanei. We then performed DXA scans with the specimens immersed in 15 cm of water and the X-ray beam oriented in the medial-lateral direction. Bone mineral content (BMC), projected area (A_DXA), and areal bone mineral density (BMD=BMC / A_DXA) were recorded. The volumetric bone apparent density was calculated as=BMC / V_disp. Because direct measurements of calcaneus thickness are not available for in vivo studies, an alternative method for computing the calcaneus thickness is needed. Assuming geometric similarity, the calcaneus thickness should be proportional to (subject weight)^1/3, (subject height)¹, and (calcaneus area)^1/2. Linear regressions of t_calip versus these variables were performed to identify the best estimator of the average calcaneus thickness (t^*). Bone mineral apparent density was calculated as BMAD=BMC / (A_DXA t^*). Regressions of BMD and BMAD versus V_disp were performed to investigate the dependence of BMD and BMAD on calcaneus size. Linear regression was also used to evaluate the relationship Þ=k BMAD and to determine the value of the proportionality constant k.

Results: The variable (calcaneus area)^1/2 was the best predictor of calcaneus thickness, followed by (subject height)¹. Assuming that t^*=A_DXA^1/2, BMAD=BMC / A_DXA^3/2. Linear regressions of BMD and BMAD versus V_disp showed that BMAD is independent of bone volume (slope=0.000; p=.786) while BMD increases with increasing bone volume (slope=0.003; p<.05). Linear regression of versus BMAD indicated a significant linear relationship (p < .0001) with the constant of proportionality k=1.822 +/- .026 SE.

Conclusions: We have developed a simple method for determining volumetric densities of the calcaneus from standard D)(A measurements. By using BMAD and Þ, we can improve the utility of DXA measurements in clinical applications. Osteoporosis, for example, can be better diagnosed using volumetric densities that do not depend on bone size. Fracture risk can also be better assessed using volumetric densities since low BMD reflects increased fracture risk when it is due to low volumetric bone density but not when it is due to small bone size. Thus, more useful clinical information can be extracted from DXA exams by calculating volumetric densities as described in this study.

Acknowledgments: Support provided in part by VA project A2058-2RA.