Responsiveness of magnetic resonance imaging-derived measures over 2.7 years

To compare the responsiveness of magnetic resonance imaging (MRI)-derived measures of knee osteoarthritis over 2.7 years.

There were 430 community-based participants (mean age 63.0 yrs, range 51-79 yrs; 51% female) measured at baseline and 2.7 years later. MRI of the right knee at both timepoints was performed to assess cartilage volume, cartilage defects, bone marrow lesions (BML), meniscal pathology, and tibial bone area. Global measurements were calculated as the sum of tibial and femoral measures. Standardized response mean (SRM) was calculated as the mean of change divided by the SD of change.

Global tibiofemoral cartilage volume and cartilage defects had the best SRM of -0.80 and 0.62, respectively. Site-specific measurements were lower (SRM range for cartilage volume -0.48 to -0.54 and cartilage defects 0.33 to 0.49). The SRM for BML was 0.12, meniscal pathology 0.39, and tibial bone area -0.09. Cartilage volume and/or defects tended to be more responsive in those with knee pain, those who were obese, those who were older, and those with radiographic osteoarthritis.

Global cartilage volume demonstrated the best sensitivity to change, suggesting that if we relied solely on SRM to optimize clinical trial design, then cartilage volume would be the best outcome measure. However, clinical trials have shown that cartilage volume may be less responsive to treatment compared to other measures that have lower SRM (such as BML). Therefore, although one can optimize trial efficiency by finding more responsive endpoints, both sensitivity to change and magnitude of benefit should be considered.