Effects of the Assessment of 4 Determinants of Structural Geometry on QCT- and DXA-Derived Hip Structural Analysis Measurements in Elderly Women

Previously we reported that the corresponding 2-dimensional (2D) structural geometry measures derived from quantitative computed tomography (QCT) and dual-energy X-ray absorptiometry (DXA) of femoral neck were different. Now, we test the hypothesis that the following 4 measures: areal bone mineral density, W, and 2 new measures, standard deviation (σ) of a normalized mineral mass projection profile distribution and the displacement between centre of mineral mass and geometric centre of mineral mass (δ) of the projection profile allow transformation from one measurement modality to the other with high precision. QCT and DXA scans and hip structural analysis (HSA) performed on 237 women were randomly allocated into cohorts of 118 (cohort A) and 119 (cohort B). Intercepts and gradients from linear regression of the 4 QCT- and DXA-derived measures were obtained from cohort A and used to convert cohort B QCT-derived structural geometry measurements into their DXA equivalent. Corresponding cohort B QCT- and DXA-derived structural geometrical measurements were compared using Bland-Altman plots and regression analysis. Apart from W, comparisons of the 7 nontransformed QCT- and DXA-derived variables were significantly different using paired t-tests. Cross-calibration with the set of 4 base measures resolved the differences in all original variables. These data provide a mechanism for cross-calibrating HSA outcomes acquired using QCT and DXA and demonstrate that a complex 2D digitized structure can be described by 4 variables.