Do quantitative ultrasound measurements reflect structure independently of density in human vertebral cancellous bone?

Ultrasonic measurements were made in three orthogonal directions on 70 vertebral bone cubes. Apparent density (rho) was determined, and microcomputed tomography was used to derive a range of microstructural parameters. Qualitatively different ultrasonic behavior was observed in the craniocaudal (CC) axis, in which two distinct waves propagated. In this direction, only attenuation correlated strongly with rho (r2 = 80%), whereas, in the anteroposterior (AP) and mediolateral (ML) axes, there were significant correlations between all ultrasonic parameters and rho (r2 = 57%-79%). Microstructural parameters were, in general, correlated with ultrasonic properties, but when adjusted for rho, few significant relationships remained and the additional variance explained by individual microstructural parameters was relatively small (< 25% for CC axis, < 3% for AP, 0% for ML). In stepwise regression analysis including rho and all of the microstructural parameters, rho remained the primary determinant of ultrasonic properties in the transverse axes: Combinations of structural parameters explained, at most, an additional of 6% of the variability in ultrasonic properties in the AP axis, but failed to contribute significantly in the ML axis. In the CC axis, structural parameters played a greater role, but the pattern of associations was complex and the predictive power of the models was generally much less than that for the transverse axes. These data indicate that the ability of ultrasound to reflect aspects of trabecular structure is strongly dependent on the direction in which ultrasonic measurements are made, and provide only qualified support for the hypothesis that ultrasound reflects cancellous bone structure independently of bone density.