Accuracy of standard MRI sequences for meniscal tears and grading of chondral lesions in the knee, relative to knee arthroscopy: A prospective study of 719 cases

Abstract

Introduction: Musculoskeletal magnetic resonance imaging (MRI) is commonly used for diagnosis and research, but its accuracy remains unproven and its clinical usefulness is questionable. The goal of this study was to compare the accuracy of knee MRI with clinical assessment for diagnosing meniscal tears, and to determine the accuracy of MRI for grading chondral lesions, relative to knee arthroscopy. Methods: Physically active patients presenting with acute / sub-acute onset of mechanical symptoms, in the absence of grade 2-3 ligament injury or true locking, that had been preventing them from taking part in their usual physical activity for at least 4 weeks, and with clinical findings that warranted a knee arthroscopy, had both a knee arthroscopy and an MRI performed. Patients were over the age of 18 years, fit for general anaesthetic, with no contraindications for arthroscopy or MRI, and prepared to undergo both. A standard knee arthroscopy was performed and the intra-operative findings were compared with those of the MRI, using the International Chondral Research Society (ICRS) grading for chondral damage, and the presence or absence of a meniscal tear. All MRIs were performed on either a 1.5T or 3.0T MRI machine, using standard sequences. Results: 719 patients were recruited over a period of 6.5 years. Their average age was 52 years (standard deviation, SD 5.2), the male:female ratio was 493:226. Kappa scores with standard errors (SE) for agreement between MRI and knee arthroscopy were 0.41 (SE 0.1) for medial meniscal tears, and 0.44 (SE 0.1) for lateral meniscal tears. For the grade of chondral damage, the Kappa scores with SE values were 0.09 (0.1), 0.17 (0.1), and 0.22 (0.07) for anterior, medial and lateral compartments respectively. Using areas under the receiver operating characteristic curves, we found clinical assessment was more accurate than MRI for diagnosis of lateral meniscal tears (P<0.001), and of similar accuracy for the diagnosis of medial meniscal tears (P=0.12). Discussion: MRI had relatively poor correlation with arthroscopic findings for grading chondral damage and was less accurate than clinical assessment for the diagnosis of lateral meniscal tears. Other research has suggested that MRI is overused in clinical practice and may increase the number of knee arthroscopies performed. Up to 40% of patients diagnosed with a meniscal tear on MRI had no meniscal tear at arthroscopy. Insufficient accuracy will undermine the validity of any research in which this imaging modality has a primary role in the study methodology. Conflict of interest declaration: My co-authors and I acknowledge that we have no conflict of interest of relevance to the submission of this abstract.