Objective: In this longitudinal pilot study, we aimed to investigate the intra-, interobserver, and scan-rescan reproducibility of marrow fat fraction (FF) measurements using three-point Dixon imaging in osteoporotic rabbits: comparison with histopathology.
Methods: Twenty female rabbits were randomly assigned to sham-operation and ovariectomy in combination with daily methylprednisolone hemisuccinate groups (n = 10 per group). Marrow FF by three-point Dixon technique and bone density by dual-energy x-ray absorptiometry were assessed at baseline, 6 and 12 weeks after operation. Intra-, inter-reader, and scan-rescan reliability of FF measurements were evaluated using intraclass correlation coefficient (ICC) and Bland-Altman 95% limit of agreement. Histomorphometry was performed to quantify marrow adipocyte parameters.
Results: Intra- and inter-reader reproducibility of FF measurements was "substantial" (ICC = 0.984 and 0.978, respectively). Although the ICC for scan-rescan reliability was excellent (ICC = 0.962), increased measurement variability was observed using Bland-Altman plot. Relative to the sham-operated rabbits, the adipocytes mean diameter, density, and percent adipocytes area in the osteoporotic rabbits increased by 23.4%, 68.9%, and 117.0%, respectively. Marrow FF was positively correlated with the quantitative parameters of adipocytes, particularly with percent adipocyte area, but inversely associated with bone density. At the relatively early stage, the percentage of bone loss was similar to that of elevated fatty marrow in the osteoporotic rabbits; at the later stage, the change for the latter outweighed that of the former.
Conclusions: Results of three-point Dixon technique demonstrated a very reproducible manner within and between observers and acceptable scan-rescan performance in the assessment of marrow fat in rabbits.