Introduction:​ Quantifying the determinants of bone strength is essential to understanding if or how the structure will fail under load. Determining failure requires knowledge of material and geometric properties. However, characterizing the relative contributions of geometric parameters of bone to overall bone strength has been difficult to date because of limitations in imaging technology. Peripheral quantitative computed tomography (pQCT) uses digital images to derive estimates of bone strength in the peripheral skeleton and is a relatively safe technique to differentiate cortical from trabecular bone and assess bone geometry and density. However, in a compromised osteoporotic bone, thin cortices and low scan resolution can limit accurate analysis.

Methods:​ Therefore, in this two-part investigation we scanned ten pairs (n=20) of fresh-frozen radial specimens [female, mean (SD) age 79(6) years] using pQCT (XCT 2000) at the 4 and 30% sites of the distal radius. We investigated the accuracy of four different acquisition resolutions (200, 300, 400, 500 μm) and several analysis modes and thresholds. We evaluated (1) the accuracy of the Norland/Stratec XCT 2000 pQCT in assessing low-density bones by comparing pQCT outcomes to ashing and histomorphometry and (2) the association of geometric parameters by pQCT and areal bone mineral density (aBMD) by dual-energy X-ray absorptiometry (DXA) to failure load at the distal radius.

Results:​ Using histomorphometry and ashing as reference standards, we found that pQCT scans varied systematically and underestimated or overestimated total area and mineral content at the radial midshaft depending on the analysis algorithm and selected threshold. Overall, most pQCT analysis modes were accurate. In the mechanical testing studies, bone mineral content and cortical bone content at the midshaft were strongly associated with failure load. The pQCT parameters that best accounted for failure load were total content at the 4% site and cortical thickness at the 30% site and they accounted for up to 81% of the variance. The best DXA predictor of failure load was total density at the distal third site and it explained 75% of the variance.

Conclusions:​ In summary, analysis mode, resolution and thresholding affected pQCT outputs at the radial midshaft. This study extends our understanding of pQCT analysis and provides important data regarding determinants of bone strength at the distal radius.