The significant effects of bone structure on inherent patient-specific DXA in vivo bone mineral density measurement inaccuracies

An extended analytic exposition is developed of the effects bone structure has on the form and extent of systematic inaccuracies in planar dual-energy x-ray absorptiometry (DXA) in vivo bone mineral density (BMD) measurements. Explicit expressions for absolute and percentage BMD inaccuracies are derived and criteria governing these BMD inaccuracies delineated. It is shown that the effect of bone structure is to introduce a scale factor which modulates the sizable and unavoidable DXA in vivo/in situ BMD inaccuracies that arise directly from patient-specific anthropometric and x-ray absorptiometric disparities among the several soft tissues present within the scan region of interest of any given bone site (i.e., lean muscle tissue, interposed and admixed fat, and red/ yellow marrow combinations). Different magnitudes and patterns of BMD inaccuracies are shown to pertain for bone structures that are (i) essentially wholly trabecular, (ii) wholly cortical, and (iii) those containing both cortical and trabecular bone. Over the range of soft tissue anthropometrics typical of adult patients, the overall percentage inaccuracies in DXA-measured BMD are shown to be quite sizable and to vary considerably for different bone structures. For a typical lumbar vertebral bone site, BMD inaccuracies are found to be as large as similar to25 % for normal patients, to exceed similar to35% for osteopenics, and to approach 50% for osteoporotic individuals. For bone sites with non-negligible cortical surrounds of trabecular structures (e.g., distal radius, some segments of proximal femur, etc.), it is shown that BMD percentage inaccuracies range up to similar to20% for normal, similar to25% osteopenic, and similar to35% for osteoporotic patients. The BMD % inaccuracies associated with wholly cortical bone (trabecular-free) sites (e.g., mid-shaft femur, mid-shaft radius, etc.) are comparatively small, being less than similar to2%. Depending on bone structure, bone size and shape, and patient-specific intra- and extra-osseous soft tissue particulars of any given adult patient, DXA in vivo BMD measurements can be grossly inaccurate, and can severely under- or over-estimate the true value of BMD and mask or exaggerate true changes in BMD in ways not previously elucidated. It is concluded that in vivo DXA-measured and actual BMD cannot be considered to be synonymous, and clinical reliance upon the two being the same may readily conduce to seriously flawed and misleading diagnostic, prognostic, and prospective results. (C) 2004 American Association of Physicists in Medicine.