CORRELATION BETWEEN CHANGES IN RADIOGRAPHIC JOINT SPACE WIDTH AND MRI CARTILAGE THICKNESS IN THE APPROACH COHORT

Abstract

In knee OA, JSW measured on weight-bearing radiographs is often used as a surrogate measure for cartilage thickness, although it represents multiple factors. Changes in JSW often show a weak correlation with changes in cartilage thickness as measured on non-weight-bearing MRI. A previous study suggested that the difference in weight-bearing between the two techniques (partly) causes this weak correlation. To 1) investigate the correlation between changes in radiographic JSW and MRI cartilage thickness using the APPROACH cohort, and 2) to explore the influence of different parameters that could contribute to weight-bearing related differences in JSW and cartilage thickness changes. 297 patients were included at five European centers. In the current study, only medial OA patients were included. Radiographs and 1.5T or 3T MRI scans, including 3D SPGR MRI, were acquired at baseline, 6 months, 1 year and 2 years. The medial JSW and minimum JSW were measured from radiographs. Medial FT cartilage thickness (MFTC) was measured from MRIs. Only patients with both measurements at ≥3 time points, including baseline and 2 years, were included. Linear regression was used to determine 2-year change. Pearson correlations were calculated between the changes in MFTC and JSW. The influence of 4 parameters on the correlations was investigated: cartilage quality (T2 mapping, 3 evenly sized groups; high, medium, low), meniscal extrusion (2 groups; MOAKS score 0 or ≥1), and potential pain-induced unloading (2 groups; >2 higher NRS pain in index knee than other knee or not) and increased BMI-induced loading (3 groups; healthy (BMI<25), overweight (25<BMI<30), obese (BMI≥30)). Modification of the slope of the correlation between changes in MFTC and JSW by the 4 parameters was investigated using interaction terms in linear regression models. These models (using standardized variables) were also used to compare Pearson correlations between JSW and MFTC changes. If there was a statistically significant difference between groups in correlation or slope (interaction term p<0.05), correlations were analyzed separately per group. The current analysis included 178 patients with medial knee OA. An average decrease in medial JSW (-0.06 mm; SD 0.49), minimum JSW (-0.07 mm; SD 0.56) and cartilage thickness (-0.09 mm; SD 0.14) was seen. Correlations between changes in cartilage thickness and medial (R=0.34; p<0.001) and minimum (R=0.22; p=0.003) JSW were weak. There was a difference in correlation of changes in MFTC with minimum JSW between T2 time categories (p<0.05), and in correlation and slope with medial JSW for meniscal extrusion (both p<0.05), but not for pain and BMI (all p>0.2). Only patients with the lowest T2 times and with meniscal extrusion showed significant moderate correlations. Potential pain-induced unloading and BMI-induced extra loading during weight-bearing radiograph acquisition do not seem to contribute to differences in the relation between JSW and MFTC. Patients with lower T2 times showed a stronger correlation, indicating that healthier cartilage is potentially less influenced by compression during weight-bearing. The relation between JSW and cartilage thickness changes was also influenced by meniscal extrusion. Future research should include a combination of factors to detect how weight-bearing JSW represents cartilage thickness, quality, meniscal extrusion, and other possible factors (e.g. radiographic positioning). Loaded MRI studies would be of value as well. EU/EFPIA Innovative Medicines Initiative Joint Undertaking (grant n° 115770). WW: Chondrometrics GmbH; FWR: Boston Imaging Core Lab. IMI-APPROACH (NCT03883568) participants and investigators. CORRESPONDENCE ADDRESS: [email protected].