To investigate whether baseline cartilage thickness and its longitudinal change are associated with incident widespread full-thickness cartilage loss (wsFTCL) in knee osteoarthritis, and whether there are optimal cut-off values for predicting wsFTCL.
Central medial tibial (cMT) and femoral (cMF) cartilage were assessed using quantitative magnetic resonance imaging data from the Osteoarthritis Initiative cohort (N = 600 knees). Cartilage thickness was measured at baseline and 12 months. wsFTCL was defined semi-quantitatively (scores 2 and 3 from the MRI Osteoarthritis Knee Score) and its incidence at 24 months recorded. Logistic regression was used to determine the odds of developing wsFTCL for baseline and for each 0.1 mm decrease in cartilage thickness. Cut-off values were investigated using the minimal-p method and area under the Receiver Operating Characteristic curves (AUC).
Incident wsFTCL was observed in 66 (12%) and 73 (14%) knees in cMT and cMF, respectively. Lower baseline cMT and cMF cartilage thickness values were associated with wsFTCL (OR = 1.20; 95% CI: 1.11, 1.28 and OR = 1.15; 95% CI: 1.06 to 1.24, respectively). Optimal cut-off AUCs for the tibia and femur were 0.64 (0.57-0.70) and 0.63 (0.57-0.69), respectively. Longitudinal decrease in femoral, but not tibial, cartilage thickness was associated with incident wsFTCL (OR = 1.77; 95% CI: 1.30 to 2.40); optimal cut-off AUC 0.65 (95% CI: 0.58-0.72).
Lower baseline cMT and baseline/change (decrease) over 12 months in cMF cartilage thickness were associated with incident, location-specific, wsFTCL at 24 months. Optimal cut-off values were relatively low and of uncertain utility for predicting incident wsFTCL.

Copyright © 2020 Osteoarthritis Research Society International. Published by Elsevier Ltd. All rights reserved.