AbstractBackground: Radiographic joint space width (JSW) and MRI-based cartilage thickness are key imaging biomarkers in knee osteoarthritis (OA), yet their weak correlation complicates cross-modality interpretation. This mismatch may potentially be explained by differences in patient posture and joint loading, as radiographs are typically obtained in a weight-bearing and flexed position, whereas MRI scans are acquired in a non-weight-bearing and extended position. Methods: Using a 0.25 T rotatable MRI scanner, 21 individuals with OA were scanned in four positions, extended and flexed in non-weight-bearing and weight-bearing conditions, to reflect typical clinical imaging conditions and functional joint motion and loading. Three-dimensional JSW and cartilage thickness maps of the tibia and femur were generated, registered to canonical surfaces, and analyzed vertex-wise using statistical parametric mapping. Spearman correlations between JSW and cartilage thickness were calculated for each position. Results: JSW narrowed under weight-bearing, particularly medially, and showed a posterior shift with flexion, especially laterally. Femoral cartilage responded more to positional changes than tibial cartilage, showing posterior thinning and anterior thickening under weight-bearing flexion. Correlations between JSW and cartilage thickness strengthened while weight-bearing, especially posteriorly, but weakened anteriorly and laterally during flexion. Overall, moving from non-weight-bearing extension (MRI positioning) to weight-bearing flexion (radiograph positioning) improved JSW-cartilage correlations in the tibia and posterior femur. Conclusions JSW and cartilage thickness are strongly influenced by joint loading and flexion. Weight-bearing flexion improves JSW–cartilage correspondence posteriorly but reduces it anteriorly. These findings primarily inform interpretation and standardization of imaging endpoints across modalities and imaging conditions; translation to routine clinical practice is currently limited by scanner availability and the modest magnitude of cartilage thickness differences.
Simonis et al. (Sun,) studied this question.