Abstract
OBJECTIVE: Magnetic resonance imaging (MRI) enables detection of early, multi-tissue changes in knee osteoarthritis (OA). A Delphi-derived MRI definition integrates findings across multiple joint tissues to classify tibiofemoral OA (TFOA), though it may identify OA in the absence of cartilage damage. We examined how often MRI-defined TFOA occurs without cartilage involvement in two large US cohorts. METHODS: We analyzed baseline data from participants without definite radiographic TFOA [Kellgren-Lawrence (KL) grade <2 in both knees] in the Osteoarthritis Initiative (OAI) and Multicenter Osteoarthritis Study (MOST) cohorts. OAI knees were scored using the MRI Osteoarthritis Knee Score (MOAKS) and MOST knees using the Whole-Organ MRI Score (WORMS). MRI-defined TFOA was determined per Delphi criteria, and these knees were assessed for absence of cartilage lesions, with a secondary analysis for osteophyte absence. RESULTS: Among participants with KL <2 in both knees, MRI-defined TFOA was observed in 283 of 1621 (17.5 %) in OAI and 206 of 641 (32.1 %) in MOST. Nearly all cases showed partial- or full-thickness cartilage lesions. Knees without cartilage involvement were rare: 1/283 (0.4 %) in OAI and 3/206 (1.5 %) in MOST. By contrast, knees without osteophytes were more common (13.4 % OAI, 5.8 % MOST). CONCLUSIONS: In two large cohorts at elevated risk for knee OA, MRI-detected cartilage damage was almost always present when knees met multi-tissue MRI-based TFOA criteria, suggesting cartilage involvement may be a consistent feature of disease characterization. These findings indicate that the Delphi definition rarely identifies OA in the absence of cartilage involvement.