Bone quality parameters of the distal radius as assessed by pQCT in normal and fractured women

The aim of this study was to test the ability of some indicators of different aspects of bone quality (assessed by peripheral quantitative computed tomography in the distal radius) to discriminate between fractured and nonfractured individuals. The study compared 214 women aged 45-85 years, free of any bone-affecting treatment, of whom 107 had suffered a Colles' fracture in the previous 6 months and 107 did not. The determinations included bone tissue or mineral 'mass' indicators (trabecular, cortical and total volumetric mineral content, cortical bone area); bone 'density' estimates (trabecular, cortical and total volumetric mineral density), and the Cartesian (rectangular) and polar moments of inertia as influences of cross-sectional architecture on resistance to bending and torsional loads, respectively. The influences of body height, weight and age on the tomographic indicators were minimized by adjusting the data according to the partial coefficients of multiple stepwise regressions. The adjusted values of all the indicators were lower in fractured than in nonfractured groups. The prevalence of fractures was directly related to the actual values of the indicators, rather than the age or body habitus of the individuals. The significance of these differences between the assessed indicators decreased in the following order: trabecular 'mass' > trabecular 'density' > cortical or total 'mass' > cortical architecture > total or cortical 'density' indicators. Within the same type of bone, the tissue or mineral 'mass' indicators performed better than the 'density' indicators. The cortical bone density did not give useful information, probably because of technical difficulties. Odds-ratios and receiver-operating characteristic (ROC) analyses confirmed those features. The selected 'cut-off' values of the indicators as determined by the ROC curves (very close to those determined by the inflexion points of the logistic reression curves) may indicate reference limits to detect persons at risk of fracture according to the type of information provided by each variable. These results show that these tomographic indicators discriminate well between fractured and nonfractured individuals, and should be suitable to assess how total, cortical and trabecular bone strength in the distal radius could affect different kinds of strength regardless of the age or body habitus of the individual. Their ability to estimate fracture risk from different biomechanical points of view should be assessed by adequately designed, prospective studies.