Analyzing cortical bone cross-sectional geometry by peripheral QCT: comparison with bone histomorphometry.

A distinct advantage of peripheral quantitative computed tomography (pQCT) is its ability to assess bone strength by measuring cross-sectional geometry and density of cortical bone. For accurate determination of cortical bone cross-sectional area (CoA), it is important to select the appropriate analysis mode and thresholds. No study has assessed which analysis protocol best represents tibial bone geometry--as determined by histomorphometry. We measured bone geometry from 16 human cadaver tibiae (mean age 74 [SD 6] yr) with pQCT (XCT 2000) at the 25% site, measured proximally from the distal tibia plafond. We conducted histomorphometry at the same site as the criterion standard. Scans were analyzed using modes and thresholds recommended by the manufacturer (Norland Stratec Medizintechnic GmbH, Pforzheim, Germany). We also investigated agreement of two additional thresholds (calculated by half-maximum height and inflection point methods) to define the endosteal border of cortical bone. Compared to the criterion, the smallest error (-1.0%, p=0.002) in total cross-sectional area (ToA) was obtained using Contour mode 3 with an outer threshold of 169 mg/cm(3). The smallest error (0.1%, NS) in CoA was obtained with Separation mode 4 (outer threshold 200mg/cm(3), inner threshold 670 mg/cm(3)). CoA was overestimated by 5-7% (p<0.001) from the criterion when an inner threshold of 480 mg/cm(3) was used in combination with any of the recommended outer thresholds. pQCT measurements of bone geometry in vitro vary to some extent between modes and thresholds selected. The effect of variation in bone geometry measurements on the predictive ability of bone strength indices derived from CoA needs to be assessed.