Novel Magnetic Resonance Imaging Classification of Osteochondritis Dissecans of the Knee: A Reliability Study.

Background: Current advanced imaging classification systems for osteochondritis dissecans (OCD) of the knee grade severity of disease by identifying certain lesion characteristics. The most widely used are the Hefti and Nelson systems. A novel classification presents a simpler 3-group approach to diagnose knee OCD by magnetic resonance imaging (MRI), compared with the Hefti (5-group) and Nelson (4-group) classifications. The purpose of this study was to compare the reliability of this novel classification with that of the more complex, established systems-an initial step in establishing validity and clinical utility.

Methods: In total, 120 standardized knee MRIs of patients with established knee OCD were preselected to capture the spectrum of lesion types, with regard to both progression and location of the lesion. Each of the MRIs were independently classified by 2 readers into the novel, Hefti, and Nelson classification systems. A random sample was rereviewed by 1 rater 6 weeks after initial review. The inter-rater and intrarater agreements were evaluated by estimating Krippendorff α.

Results: In total, 106 knees were classified by the novel, Hefti, and Nelson classification systems, as 14 of the knees lacked the necessary MRI sequences. There were no differences in inter-rater and intrarater agreement across classification systems. Krippendorff α for inter-rater agreement was 0.51 (95% confidence interval, 0.33-0.66) for the Hefti classification, 0.50 (0.34-0.64) for the Nelson classification, and 0.49 (0.32-0.65) for the novel classification. The intrarater agreement was 0.88 (0.75-0.97) for the Hefti classification, 0.94 (0.86-0.99) for the Nelson classification, and 0.98 (0.94-1.00) for the novel classification system.

Conclusions: The novel classification for knee OCD demonstrated near-perfect intrarater agreement and moderate inter-rater agreement, consistent with the current, well-established classification systems. Pending a subsequent study on validity and clinical utility, this simpler classification system may offer an alternative, noninvasive diagnostic method to guide clinical treatment.

Level Of Evidence: Level IV.