Can the Reparability of Meniscal Tears Be Predicted With Magnetic Resonance Imaging?

Background: Historically, magnetic resonance imaging (MRI) has been very useful in diagnosing meniscal tears but not as valuable in predicting whether a meniscal tear is reparable. Given that several recent studies suggested that MRI can be used to predict tear reparability, the topic has resurfaced as a controversy in the orthopaedic and radiology literatures. Hypothesis: Experienced musculoskeletal radiologists can use MRI to predict the reparability of meniscal tears with good to excellent accuracy using the same arthroscopic criteria used by surgeons intraoperatively. Study Design: Cohort study (diagnosis); Level of evidence, 3. Methods: Fifty-eight patients with meniscal tears treated with repair were matched by age and sex with 61 patients with tears treated with meniscectomy. Two senior musculoskeletal radiologists independently and blindly reviewed preoperative MRI of these 119 meniscal tears. Using established arthroscopic criteria, the radiologists were asked to grade each tear 0 to 4, with 1 point for each of the following: a tear larger than 10 mm, within 3 mm of the meniscosynovial junction, greater than 50% thickness, and with an intact inner meniscal fragment. Only a tear with a score of 4 would be predicted to be reparable. Results: The 2 radiologists' ability to correctly estimate reparability was poor, with 58.0% and 62.7% correct predictions (kappa = 0.155 and 0.250, respectively). Interrater reliability assessment showed that the raters agreed on a score of 4 (reparable) versus < 4 (not reparable) 73.7% of the time (kappa = 0.434) but came to identical scores only 38.1% of the time (kappa = 0.156). Determining the status of the inner fragment was the most predictive individual criterion and the only one to reach statistical significance (chi(2) = 14.9, P < .001). Conclusion: Magnetic resonance imaging is not an effective or efficient predictor of reparability of meniscal tears with the current arthroscopic criteria.