Validation of a Novel Magnetic Resonance Imaging Classification for Osteochondritis Dissecans of the Knee.

Abstract

The novel Kocher classification is a 3-group magnetic resonance imaging (MRI) classification system for osteochondritis dissecans (OCD) of the knee that was shown to have comparable reliability to that of the established 5-group Hefti classification. The purpose of this study was to evaluate the validity and clinical utility of this simplified system as an alternative to the Hefti classification. Demographic data and arthroscopic findings were retrospectively collected from medical and surgical records of 144 consecutive knees in children with arthroscopically diagnosed knee OCD. OCD lesions on preoperative MRIs and surgical reports (serving as the reference standard) were assessed by independent raters and assigned both a Kocher and Hefti classification. Agreement between MRI classification and arthroscopic findings for both systems was assessed using weighted kappa (kw) coefficients. Validation, accuracy, sensitivity, and specificity were measured by comparing a dichotomized Kocher classification for MRI and arthroscopy, and by estimating Cohen kappa (kc) coefficients. Agreement between arthroscopic findings and treatment type was measured using the Spearman correlation coefficient. Inter-rater reliability between the 2 MRI raters was substantial for the Kocher classification [ka=0.66; 95% confidence interval (CI)=0.56-0.75] and moderate for the Hefti classification (ka=0.57; 95% CI=0.47-0.67). There was no difference detected in the agreement statistics for Kocher versus Hefti classifications (P=0.89). Binary agreement using dichotomized Kocher classifications was worse than the 3-group category classification. When dichotomized, combining Kocher grades 1 and 2 demonstrated moderate agreement (kc=0.41; 95% CI=0.25-0.58), and combining grades 2 and 3 demonstrated fair agreement (kc=0.34; 95% CI=0.21-0.48). There was a strong correlation between arthroscopy-based finding and treatment category for both the Kocher classification (r=0.85; 95% CI=0.80-0.89) and the Hefti classification (r=0.82; 95% CI=0.75-0.86). The validity and clinical utility of the newer 3-group Kocher classification for knee OCD is comparable to that of the well-established 5-group Hefti classification. Both systems help determine lesion stability and characteristics on MRI, which correlate closely to arthroscopic findings. This simplified classification system, with less uncertainty, provides a foundation for further outcomes research to develop an evidence-based algorithm for effective surgical management of OCD lesions of the knee. Level II-diagnostic study.