Validation of 3D-DXA-derived proximal femur measurements against QCT across international clinical cohorts.

Abstract

Three-dimensional dual-energy X-ray absorptiometry (3D-DXA) reconstructs proximal femur models from standard scans to estimate cortical and trabecular bone parameters. The aim of this study was to evaluate 3D-DXA against quantitative computed tomography (QCT) across independent international cohorts. The study included 531 subjects from four cohorts: an adult population from Spain, a postmenopausal female population from the United States, an osteoarthritis population and a young population, both from Japan. Subjects underwent both 3D-DXA and QCT imaging. Accuracy was assessed using linear regression and Bland-Altman analysis to evaluate systematic and random errors. 3D-DXA parameters strongly correlated with QCT across all datasets, with correlation coefficients between 0.82 and 0.97. Random errors were consistent across cohorts and ranged between 16.55 and 19.91 mg/cm³ for integral volumetric bone mineral density (vBMD), between 13.52 and 18.47 mg/cm³ for trabecular vBMD, and between 9.13 and 11.37 mg/cm² for cortical surface bone mineral density (sBMD). Systematic errors ranged between -14.84 and 4.50 mg/cm³ for integral vBMD, between -8.31 and 14.41 mg/cm³ for trabecular vBMD, and between -5.58 and 3.21 mg/cm² for cortical sBMD. The variations in systematic errors were likely attributable to differences in QCT acquisition protocols. Overall, these results demonstrate consistent agreement between 3D-DXA and QCT across sex, age, ethnicity, geographic regions, and clinical profiles. Taken together, these findings support the use of 3D-DXA as an accurate, non-invasive, and clinically accessible technology for advanced assessment of the cortical and trabecular compartments of the proximal femur.