3-T MRI of the Shoulder: Is MR Arthrography Necessary?

The purpose of this study is to report the diagnostic sensitivity of 3-T conventional MRI versus MR arthrography of the shoulder in the same patient population. One hundred fifty consecutive conventional shoulder MRI and MR arthrography examinations performed on patients 50 years or younger who subsequently underwent arthroscopy were reviewed retrospectively by consensus reading by two musculoskeletal radiologists. All patients selected for arthroscopy had abnormal findings on clinical examination and MRI or MR arthrography. All 150 patients were referred from one orthopedic group. All patients consented to undergo both MRI and MR arthrography. None had undergone prior shoulder surgery. Full- or partial-thickness supraspinatus tendon tears, superior labral anterior-to-posterior (SLAP) tears, and anterior or posterior labral tears were assessed. Three full-thickness and nine partial-thickness supraspinatus tendon tears, seven SLAP tears, six anterior labral tears, and two posterior labral tears were seen on MR arthrography but not on conventional MRI. All additional MR arthrography findings were confirmed at arthroscopy. On conventional MRI, sensitivities and specificities compared with arthroscopy were as follows: anterior labral tear, 83% sensitivity and 100% specificity; posterior labral tear, 84% and 100%; SLAP tear, 83% and 99%; supraspinatus tendon tear, 92% and 100%; partial-thickness articular surface tear, 68% and 100%; and partial-thickness bursal surface tear, 84% and 100%. On MR arthrography, sensitivities and specificities compared with arthroscopy were as follows: anterior labral tear, 98% sensitivity and 100% specificity; posterior labral tear, 95% and 100%; SLAP tear, 98% and 99%; supraspinatus tendon tear, 100% and 100%; partial-thickness articular surface tear, 97% and 100%; and partial-thickness bursal surface tear, 84% and 100%. MR arthrography showed a statistical improvement in sensitivity (p