Semi-quantitative weight-bearing assessment of knee osteoarthritis: COAKS (CT Osteoarthritis Knee Score) reliability.

Objective: Our aims were to 1) introduce the semi-quantitative CT Osteoarthritis Knee Score (COAKS); and 2) report intra- and inter-observer reproducibility. Design: Weight-bearing CT (WBCT) images of 106 participants were reviewed to develop the COAKS system and create a standardized atlas. Images of 10 knees were used to train musculoskeletal radiologists with the atlas. Once trained, two radiologists independently scored 35 knees on two occasions using reformatted images in orthogonal planes. Joint space narrowing (JSN), osteophytes, subchondral cysts and subchondral sclerosis were scored (0-3 scale) in the medial tibiofemoral, lateral tibiofemoral, patellofemoral, and proximal tibiofibular compartments. Weighted kappa statistics were calculated for intra- and inter-observer reliability. Compartment feature scores were plotted as heat maps for each knee to illustrate OA severity and location. Results: Scoring for nearly all features in all compartments had substantial to near-perfect reliability (0.61-1.00). Both inter- and intra-observer results combined across all compartments demonstrated near-perfect agreement for JSN (0.87 and 0.86) and subchondral cysts (0.84 for both) and substantial agreement for osteophytes (0.79 and 0.74) and subchondral sclerosis (0.66 and 0.67). Conclusions: COAKS is a feasible, multiplanar, semi-quantitative, compartment-by-compartment WBCT-based knee OA scoring system that demonstrates substantial to near-perfect intra- and inter-observer reliability. The capacity of COAKS to characterize the location and severity of OA in the weight-bearing knee could enable patient stratification, selection, and longitudinal monitoring of structural disease severity in clinical trials and cohort studies.