Objective To examine the relationship of baseline clinical, radiographic, molecular and MRI measures with structural progression (subregional MRI-based femorotibial cartilage loss) in knee osteoarthritis (OA).

Methods Single knees of 75 female participants with radiographic knee OA (and 77 healthy control participants) were examined over 24 months using MRI. Subregional femorotibial cartilage thickness was determined at baseline and follow-up. Baseline clinical, radiographic, molecular (n=16) and quantitative MRI-based measures of the meniscus and cartilage, including delayed gadolinium-enhanced MRI (dGEMRIC) and T2, were obtained. Differences in these baseline measures between radiographic osteoarthritic knees with longitudinal cartilage thinning (or thickening) and those with no significant change were evaluated by receiver operator characteristic analyses and Wilcoxon rank sum tests.

Results The relatively strongest predictors of longitudinal cartilage thinning were reduced baseline cartilage thickness in the medial femur (area under the curve (AUC)=0.81), varus malalignment (AUC=0.77), reduced minimum joint space width and a greater radiographic joint space narrowing (JSN) score (both AUC=0.74). These remained significant after adjusting for multiple comparisons using false discovery rates. Reduced bone resorption (C-terminal telopeptide of type I collagen; AUC=0.65) and a low dGEMRIC index (reflecting low proteoglycan content) in the medial tibia (AUC=0.68) were associated with longitudinal cartilage thinning, but failed to reach statistical significance after correction for multiple testing in this (small) sample.

Conclusions This exploratory study indicates that baseline molecular or MRI cartilage compositional markers may not provide better discrimination between knees with cartilage thinning and those without longitudinal change than simple radiographic measures, such as greater JSN score.