Shoulder arthroscopy remains superior to direct MR arthrography for diagnosis of subtle rotator interval lesions.

Abstract

To assess the sensitivity, specificity and accuracy of MR arthrography, as opposed to shoulder arthroscopy, in diagnosing individual rotator interval (RI) structures lesions at different levels of severity. Seventy-five patients were enrolled in a prospective study. All the patients were diagnosed with full-thickness rotator cuff tendon tears on unenhanced MRI and had complimentary MR arthrography to search for obscure RI lesions. All the patients then underwent shoulder arthroscopy. The arthroscopist was blinded about the MR arthrography results. At arthroscopy, 42 patients (56 %) were found to have RI lesion(s) and represented the study group. The remaining 33 patients represented the control group. The sensitivity, specificity and accuracy of MR arthrography for detecting individual RI lesions varied widely depending on the location and severity of the lesions. MR arthrography showed intermediate sensitivity of 67-80 %, specificity of 83-89 % and accuracy of 89-92 % for diagnosing subtle RI lesions; and perfect (100 %) sensitivity, specificity and accuracy for diagnosing biceps long head tendon dislocation. For the rest of RI lesions, MR arthrography showed high sensitivity, specificity and accuracy. Inter-observer agreement was found to be almost perfect (K = 0.81-1.0). Shoulder arthroscopy remains the gold standard for diagnosing subtle RI lesions. Although MR arthrography has proved to be a valuable tool for diagnosing established RI lesions, it is of intermediate sensitivity for diagnosing subtle RI lesions resulting in early insufficiency of the biceps pulley system.