Two-year change in cartilage thickness on quantitative magnetic resonance imaging (qMRI) and joint space width (JSW) on X-ray demonstrate similar predictive performance for knee replacement in OA subjects eligible for clinical trials

Purpose: Narrowing of joint space width (JSW), as measured on conventional X-ray, is the only primary structural endpoint currently accepted by FDA and EMA to evaluate efficacy in trials of Disease Modifying Osteoarthritis Drugs (DMOADs). EMA guidance to industry states that quantitative MRI (qMRI)-based measures of cartilage thickness might be considered as a surrogate endpoint if evidence of clinical relevance is provided with regard to disease progression. The objective was to assess, among a cohort of individuals eligible for a DMOAD clinical trial, the clinical relevance of 2-year changes in qMRI cartilage thickness and X-ray JSW by evaluating predictive performance of these measures for knee replacement (KR) as a clinical outcome. Methods: The sample included Osteoarthritis Initiative (OAI) participants who met typical eligibility criteria for a DMOAD trial at either the OAI baseline or at a later visit, and who had both JSW and qMRI readings available at the time of eligibility and 2-years later. Eligibility criteria included: Kellgren & Lawrence grade 2 or 3 in the target knee, medial minimum JSW (mJSW) ≥ 2.5 mm, a history of knee pain with a severity of 4–9 (on a 0–10 numerical rating scale) in the past 30 days, or a severity of 0–9 if the participant reported taking medication for joint pain on most of the past 30 days. X-ray measures included medial compartment minimal JSW (mJSW) and fixed JSW (fJSW: x=0.225mm for medial and x=0.775mm for lateral compartments). qMRI measures included cartilage thickness in the medial femorotibial compartment [MFTC], lateral femorotibial compartment [LFTC] and total femorotibial joint [FTJ=MFTC+LFTC]). Among other measures of predictive performance, area under the receiver operating characteristic curves (AUCs) were used to evaluate 2-year changes in imaging measures to discriminate between participants who later underwent KR versus those who did not. Cox proportional hazard models were used to estimate the association between 2-year change in X-ray and qMRI-based measures and time to KR. Results: The sample included 597 participants with median follow-up of 4.9 years after 2-year imaging, of whom 79 underwent KR. The 2-year changes in mean MRI MFTC and FTJ cartilage thickness, and X-ray medial fJSW measures, had similar discrimination for future KR, with AUCs significantly different from chance (Table 1 & Figure 1). The comparison of X-ray medial mJSW AUC to medial fJSW AUC revealed the only significant difference in AUCs (p=0.027) for the structural measures. There was no evidence of AUC greater than chance for lateral compartment MRI and X-ray measures. All MRI and X-ray measures were significantly associated with time-to-KR (Table 2). Conclusions: In a sample of persons with knee OA that could be enrolled in a DMOAD trial, 2-year changes in qMRI and X-ray measures had similar predictive performance with respect to future KR, and were similarly associated with time to KR. Given that qMRI measures and fJSW demonstrated similar clinical relevance, health agencies may wish to consider qMRI measures as a structural endpoints in DMOAD clinical trials.Tabled 1Table 1. AUC for ROC by qMRI and X-ray JSW measures relative to KR.MeasuresAUC95% CIP-valueMFTC0.600.52, 0.670.0175LFTC0.540.47, 0.620.2785FTJ0.620.55, 0.700.0013Medial mJSW0.590.52, 0.670.0162Medial fJSW0.640.57, 0.710.0001Lateral fJSW0.540.46, 0.610.3775AUC = area under curve, ROC = receiver operating characteristic. MRI: MFTC = medial femorotibial cartilage thickness, LFTC = lateral femoro-tibial cartilage thickness; FTJ =MFTC+LFTC, X-ray: mJSW = minimal joint space width, fJSW = fixed joint space width (x=0.225 mm for medial; x=0.775 mm for lateral) Open table in a new tab Tabled 1Table 2. Standardized hazard ratios for 2-year changes in qMRI and JSW measures with Time-to-KR.MeasuresStandardized hazard ratio (HR)95% CIP-valueMFTC1.71.4, 2.0