Dual-energy X-ray absorptiometry, peripheral quantitative computed tomography, and micro-computed tomography techniques are discordant for bone density and geometry measurements in the guinea pig

This study aims to examine agreement among bone mineral content (BMC) and density (BMD) estimates obtained using dual-energy X-ray absorptiometry (DXA), peripheral quantitative computed tomography (pQCT), and micro-computed tomography (μCT) against high-resolution μCT and bone ash of the guinea pig femur. Middle-aged (n = 40, 86 weeks) male guinea pigs underwent in vivo followed by ex vivo DXA (Hologic QDR 4500A) scanning for intact and excised femur BMC and areal density. To assess bone architecture and strength, excised femurs were scanned on pQCT (Stratec XCT 2000L) as well as on two μCT scanners (LaTheta LCT-200; Skyscan 1174), followed by three-point bending test. Reproducibility was determined using triplicate scans; and agreement assessed using Bland–Altman plots with reference methods being high-resolution μCT (Skyscan) for BMD and bone ashing for BMC. All techniques showed satisfactory ex vivo precision (CV 0.05–4.3 %). However, bias compared to the reference method was highest (207.5 %) in trabecular bone volume fraction (BV/TV) measured by LaTheta, and unacceptable in most total femur and cortical bone measurements. Volumetric BMD (vBMD) and BV/TV derived by LaTheta and pQCT at the distal metaphysis were biased from the Skyscan by an average of 49.3 and 207.5 %, respectively. Variability of vBMD, BV/TV and cross-sectional area at the diaphysis ranged from −5.5 to 30.8 %. LaTheta best quantified total femur BMC with an upper bias of 3.3 %. The observed differences among imaging techniques can be attributable to inherent dissimilarity in construction design, calibration, segmentation and scanning resolution used. These bone imaging tools are precise but are not comparable, at least when assessing guinea pig bones.