Diagnostic Accuracy of Preoperative 1.5-Tesla MRI for Three Categories of Meniscal Injuries in Anterior Cruciate Ligament Rupture: Correlation with the Arthroscopic Reference Standard

Background: Meniscal injuries are present in 40–90% of anterior cruciate ligament (ACL) ruptures, but the diagnostic accuracy of preoperative magnetic resonance imaging (MRI) varies considerably depending on the type of lesion. Ramp lesions of the medial meniscus are located in the peripheral vascularized red zone, with a healing potential of up to 95% when repaired concomitantly with ACL reconstruction, making accurate preoperative diagnosis clinically crucial. Methods: We aimed to determine the sensitivity, specificity, positive predictive value, negative predictive value, overall diagnostic accuracy, and Cohen’s kappa of preoperative 1.5-T MRI for three categories of meniscal injuries. These categories included ramp lesions, non-ramp medial meniscal lesions, and lateral meniscal lesions. Intraoperative arthroscopy was used as the reference standard. Results: A retrospective single-center diagnostic accuracy study (STARD 2015) was conducted in 40 consecutive patients who underwent primary arthroscopic ACL reconstruction. MRI examinations (1.5 T; proton density fat-saturated sequences) were interpreted by a subspecialty-trained musculoskeletal radiologist blinded to the surgical findings. Ninety-five-percent confidence intervals (Clopper–Pearson) and Cohen’s kappa (according to Landis and Koch) were calculated for each category. For non-ramp medial meniscal lesions, MRI demonstrated excellent diagnostic performance: sensitivity of 92.3% (95% CI 64.0–99.8%), specificity of 92.6% (75.7–99.1%), PPV of 85.7%, NPV of 96.2%, overall accuracy of 92.5%, and κ = 0.832 (almost perfect agreement). For ramp lesions, results were moderate and exploratory, being based on only three arthroscopically confirmed lesions: sensitivity of 66.7% (9.4–99.2%), specificity of 86.5%, PPV of 28.6% (3.7–71.0%), NPV of 97.0%, accuracy of 85.0%, and κ = 0.330 (moderate agreement). The corresponding likelihood ratios were LR+ 4.93 (95% CI 1.57–15.46) and LR− 0.39 (95% CI 0.08–1.92). For lateral meniscal lesions, MRI showed poor diagnostic performance: sensitivity of 50.0%, specificity of 66.7%, PPV of 33.3%, NPV of 80.0%, accuracy of 62.5%, and κ = 0.143 (weak agreement). Conclusion: The diagnostic accuracy of preoperative 1.5-T MRI for meniscal injuries in ACL-injured knees is heterogeneous and depends on the lesion type. Radiologists should apply differentiated levels of diagnostic confidence: high for non-ramp medial meniscal lesions (κ = 0.832), moderate with emphasis on NPV for ramp lesions, and cautious for lateral meniscal lesions due to near-chance agreement with arthroscopy (κ = 0.143). Surgical decision-making regarding the lateral meniscus should not rely solely on preoperative MRI findings.