Background/Purpose: Recently, the FDA has indicated that data supporting imaging biomarkers of relevance to clinically important outcomes, such as knee replacement (KR) may be considered as an endpoint for OA RCTs, including MRI-based cartilage evaluations. We therefore evaluated the relation of MRI-based cartilage lesions to knee replacement and knee pain severity, short-term change in cartilage lesions, which is of relevance for drug development programs so that insights can be gained regarding likelihood of clinically important outcomes after a shorter term RCT based on a surrogate.

Methods:

We used data from the Multicenter Osteoarthritis (MOST) Study, a NIH-funded longitudinal cohort of persons with or at risk of knee OA. All participants had knee MRIs and WOMAC pain obtained at baseline, 30-, 60-, and 84-months. Cartilage lesions were scored from 1.0 T knee MRIs using WORMS. To replicate eligibility for trials, we excluded knees with KL=4 or malalignment >5 degrees. We evaluated the relation of baseline and 2.5-year changes in cartilage lesions to subsequent risk of KR and WOMAC knee pain severity over 7 years. Baseline cartilage lesions were analyzed as: 1) maximal score across all subregions (0-6); 2) sum of scores across all subregions (0-84). Change over 2.5 years were analyzed as: 1) change in sum of scores; 2) proportion of subregions with worsening lesions. We evaluated the sample as a whole, as well as limited to those with JSN=2 (~ JSW=1.5-3.5mm), a group demonstrating more responsiveness in recent RCTs, and to those with WOMAC pain 8-18 (out of 20), a typical pain range for entry into RCTs. For the relation of cartilage lesions to KR, we used a mixed-effects Cox proportional hazards model with a random-intercept for subjects. For WOMAC pain, we used linear mixed-effects model with a random-intercept for subjects and a nested random-intercept for knees within subjects. For analyses of change in cartilage score from baseline to 2.5 years later, we evaluated outcomes occurring at time-points after 2.5 years.

Results: Overall, 1755 subjects (2554 knees) met inclusion criteria (mean age 62, 64% female, 86.5% White, mean WOMAC pain 2.7). 113 knees underwent KR overall (4.4%); of knees that were JSN=2 at baseline, 19.8% (41/258) underwent KR and of knees with WOMAC 8-18 at baseline, 10.2% (25/244) underwent KR.  Overall, both severity and extent of baseline cartilage lesions as well as change in cartilage lesions over 2.5 years were significantly associated with risk of KR (Table). In contrast, while cartilage lesions were significantly associated with WOMAC pain severity, the relationship was weak (Table).

Conclusion: These findings support the clinical relevance of cartilage lesions to the clinically important endpoint of KR. Relatively short-term changes in cartilage lesions over 2.5 years may be potentially useful as a prognostic marker for later clinical endpoints such as KR. However, the magnitude may not be sufficient for validation as an imaging biomarker. In contrast, cartilage lesions were not strongly associated with WOMAC pain, potentially partly reflecting the uncertainty in utility of WOMAC pain to detect long-term change in pain over several years. 

Relation of Baseline & 2.5-year Change in Cartilage Morphology to Risk of Knee Replacement and to Knee Pain Severity

« Back to ACR Convergence 2020

ACR Meeting Abstracts - https://acrabstracts.org/abstract/the-relation-of-mri-based-cartilage-lesions-to-knee-replacement-and-knee-pain-severity-in-osteoarthritis-the-multicenter-osteoarthritis-study/