Identification and prediction of isolated lateral meniscal lesions in the Chinese population using MRI-based anatomical parameters.

backgroundPrevious studies have demonstrated a close association between knee joint anatomical features and lesions of the anterior cruciate ligament (ACL) and meniscus. However, the anatomical risk factors specifically related to isolated lateral meniscus lesions remain inadequately investigated. The aims of this study were to identify anatomical risk factors for isolated lateral meniscus lesions and present a predictive model.

methodsA retrospective analysis was conducted using data from 142 Han Chinese patients, including 71 patients with isolated lateral meniscus lesions and 71 matched controls. Magnetic resonance imaging (MRI) was used to measure the posterior tibial slope (PTS), posterior slope of the meniscus (PSM), medial proximal tibial angle (MPTA), medial tibial depth (MTD), femoral condylar width (FCW), intercondylar notch width (NW), and the lateral angle at the distal end of the popliteus tendon groove (LATG). Demographic and anatomical characteristics were compared between the groups. Parameters showing significant differences were subjected to binary logistic regression to identify independent risk factors. Receiver operating characteristic (ROC) curves were constructed to evaluate the predictive performance and optimal cutoff values for each risk factor, and a nomogram was developed for model visualization. An additional cohort of 35 patients (17 with injury, and 18 controls) was used for external validation. The model performance was further assessed using ROC and calibration curves.

resultsDiscoid lateral meniscus (DLM), medial tibial slope (MTS), and LATG were independent risk factors for isolated lateral meniscus lesions. ROC curve analysis revealed that LATG had the highest area under the curve (AUC) at 0.766 (95% CI 0.688-0.844), followed by MTS at 0.691 (95% CI 0.604-0.777) and DLM at 0.662 (95% CI 0.572-0.752). The combined model incorporating all three variables yielded an AUC of 0.860 (95% CI 0.799-0.922). Tenfold cross-validation revealed an average AUC of 0.860 (95% CI 0.798-0.922) and a mean absolute error (MAE) of 0.014 in the training set. External validation produced an AUC of 0.794 (95% CI 0.634-0.954) and an MAE of 0.015.

conclusionDLM, MTS, and LATG are independent anatomical risk factors for isolated lateral meniscus lesions. The combination of the DLM, MTS, and LATG can effectively predict the risk of lesions. However, it should be noted that the ROC curve relies on the current retrospective cohort data, and caution is needed when the model is extrapolated to other populations for validation.