Tibial subchondral bone mineral density: sources of variability and reproducibility

Summary Objectives It has been shown that subchondral bone mineral density (sBMD) measurement may be a relevant parameter of osteoarthritis (OA) progression. However, factors implicating the reproducibility and contributing to the variability of the measurement have not been fully described. Thus, the aim of this study was to explore the reproducibility of sBMD by Dual energy X-ray Absorptiometry (DXA) and to further examine its sources of variability. Methods In this study, short-term, intra and inter-observer reproducibility of sBMD was examined on knee images obtained on DXA scans. The influence of software (lumbar spine and forearm modes), knee positioning (flexion or extension), site and size of regions of interest (ROI) and use of rice, on both lateral and medial tibial sBMD, were assessed. Root mean square coefficient of variation (RMS CV) and least significant changes (LSC) were calculated. Results The short-term precision of sBMD ranged between 2.24% and 5.12% for RMS CV and between 0.053 and 0.135 g/cm 2 for LSC. Good intra-observer precision was found for knee flexion conditions whatever the software used (RMS CV ranging from 0.43 to 1.41%). The reproducibility was dependant from the ROI size (the ROI including joint space exhibiting better precision results than ROI including solely the subchondral plate). For a constant size of the ROI, the precision results were site-dependant. Inter-observer RMS CV results ranged from 0.59 to 5.01% according to ROI and software used. For the specific task of monitoring medial sBMD in the ROI including solely subchondral plate, forearm flexion condition produced the highest intra-observer and short-term precision (respectively RMS CV: 0.45% and 2.77%; LSC: 0.013 and 0.080 g/cm 2 ). Conclusion Taking account into the excellent precision of the sBMD measurements expressed as RMS CV with the protocol proposed in the present study, clinical application of these measurements might be envisaged.