Abstract
This study was designed to characterize the reproducibility and sensitivity to change of magnetic resonance imaging-based cartilage morphology metrics and femoral regions of interest (ROIs), in order to provide preferable outcome measures in longitudinal studies of cartilage morphology. Test-retest acquisitions were obtained at 3 tesla (T) in 33 subjects with and without radiographic signs of osteoarthritis (OA) (reproducibility study) as well as baseline and 2-year follow-up acquisitions in 28 subjects with radiographic signs of advanced OA (sensitivity study). Cartilage was segmented in the tibia and two distinct anatomical femoral ROIs, a ‘long’ ROI extending 60% from the trochlear notch to the posterior end of the condyles, and a ‘short’ ROI extending to the intercondylar bone bridge. Coefficients of variation (reproducibility study) and standardized response means (SRMs, sensitivity study) were obtained for different morphology metrics and anatomical regions. The subchondral bone area of the long ROI was 20% greater and less variable than that of the short ROI; cartilage morphology metrics were generally more reproducible in the long ROI. Normalized cartilage volume (VCtAB) and mean cartilage thickness (over the entire subchondral bone area; ThCtAB.Me) tended to be more reproducible and more sensitive to change (SRM up to –0.62) than cartilage volume (SRM up to –0.44), cartilage thickness over the cartilaginous area (ThCcAB; SRM up to –0.48) or maximum cartilage thickness (ThCtAB; SRM up to –0.35). The long femoral cartilage ROI provided more reproducible measurements than the short one. VCtAB and ThCtAB.Me may be preferable metrics in longitudinal studies of articular cartilage adaptation or OA.