Tears of the supraspinatus tendon: assessment with indirect magnetic resonance arthrography in 67 patients with arthroscopic correlation.

BACKGROUND: Magnetic resonance (MR) arthrography is generally regarded as the gold standard for shoulder imaging. As an alternative to direct MR arthrography, the less invasive indirect MR arthrography technique was proposed, offering logistic advantages because fluoroscopic or ultrasonographic guidance for joint injection is not required.

PURPOSE: To assess the diagnostic performance of indirect MR arthrography in the diagnosis of full- and partial-thickness supraspinatus tears in a symptomatic population.

MATERIAL AND METHODS: Two radiologists with different levels of experience independently and retrospectively interpreted indirect MR (1.5T) arthrograms of the shoulder obtained in 67 symptomatic patients who underwent subsequent arthroscopy. On MR, the supraspinatus tendon was evaluated for full- or partial-thickness tear. With arthroscopy as the standard of reference, sensitivity, specificity, and diagnostic accuracy of indirect MR arthrography in the detection of full- and partial-thickness tears of the supraspinatus tendon was calculated. Kappa (kappa) statistics were used for the assessment of the agreement between arthroscopic and imaging findings and for the assessment of interobserver agreement.

RESULTS: For full-thickness tears of the supraspinatus tendon, sensitivities, specificities, and accuracies exceeded 90% for both observers, with excellent interobserver agreement (kappa = 0.910). For partial-thickness tears, sensitivities (38-50%) and accuracies (76-78%) were poor for both reviewers, and interobserver agreement was moderate (kappa = 0.491). Discrepancies between MR diagnosis and arthroscopy were predominantly observed with small partial-thickness tears.

CONCLUSION: Indirect MR arthrography is highly accurate in the diagnosis of full-thickness rotator cuff tears. However, the diagnosis of partial-thickness tears with indirect MR arthrography remains faulty, because exact demarcation of degenerative change and partial rupture is difficult. On the basis of the above findings, we do not recommend indirect MR arthrography on patients for whom rotator cuff disease is suspected clinically.