Needle Diagnostic Arthroscopy and Magnetic Resonance Imaging of the Shoulder Have Comparable Accuracy With Surgical Arthroscopy: A Prospective Clinical Trial

摘要

Purpose: To examine the accuracy, sensitivity, and specificity of a minimally invasive needle arthroscopy device and magnetic resonance imaging (MRI) compared with diagnostic arthroscopy, the gold standard in diagnosing intra-articular shoulder pathologies. Methods: This was a prospective, blinded clinical trial over 6 months on 50 patients with shoulder pathology requiring arthroscopy. Patients were eligible if they had an MRI and consented for surgical arthroscopy. Patients were excluded if they didn't consent. Each underwent a clinical evaluation, MRI, needle arthroscopy, and surgical arthroscopy. Videos and images were blindly reviewed postoperatively. Analysis included sensitivity, specificity, positive predictive value (PPV), negative predictive value, Cohen's kappa agreement coefficient, and the McNemar test. Results: Needle arthroscopy had similar accuracy to MRI in diagnosing intra-articular shoulder pathologies when both were compared with the gold standard of diagnostic arthroscopy. It had high specificities and PPV for certain rotator cuff tears, biceps pathology, and anterior labral tears. When compared with the gold standard, specificity of needle arthroscopy for diagnosing rotator cuff tear and cartilage lesions was 1.00 and 0.97 and 0.72 and 0.86 for MRIs, respectively. Sensitivity of needle arthroscopy for rotator cuff and cartilage lesions was 0.89 and 0.74, respectively, lower than MRI. For most intra-articular pathologies, needle arthroscopy was at least equally accurate to MRI at diagnosing intra-articular shoulder pathologies, with similar or high kappa statistics when correlated with surgical arthroscopic findings. Conclusions: Needle arthroscopy is a promising diagnostic modality for intra-articular shoulder pathologies. It had comparable accuracy with MRI for diagnosing articular cartilage, labrum, rotator cuff, and biceps pathology. Across all pathologies, needle arthroscopy had better ability to "rule in" a diagnosis (high specificities and PPV), but slightly worse ability to "rule out" a diagnosis (lower sensitivities and negative predictive value) compared with MRI.