Objective: We sought to determine the diagnostic accuracy of unenhanced MRI and direct MR arthrography (MRA) for evaluation of the intra-articular long head of the biceps brachii tendon (LHBT) using arthroscopy as the gold standard.
Materials and methods: A retrospective review of patients who underwent shoulder MRI (n = 132) and MRA (n = 67) within 12 months prior to arthroscopy was performed. MR images were independently reviewed by two blinded musculoskeletal radiologists. Routinely recorded arthroscopic photos/videos were reviewed by an orthopedic surgeon. The LHBT was graded as normal, tendinosis, partial thickness tear less or greater than 50%, and complete tear. Sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and accuracy for tendinosis and tear detection were calculated.
Results: MRI correctly diagnosed fewer normal LHBTs compared to MRA (39-54% versus 74-84%, respectively; p < 0.005). MRI and MRA did not differ significantly in the diagnosis of tendinosis (18-36% and 15-38%, respectively; p > 0.05) and tears (75-83% and 64-73%, respectively; p > 0.05). For tendinosis, MRI versus MRA showed 18-36% and 15-38% sensitivity, 69-79% and 83-91% specificity, 22-28% and 18-50% PPV, 74-76% and 80-86% NPV, and 61-64% and 70-81% accuracy; respectively. For tears, MRI versus MRA showed 75-83% and 64-73% sensitivity, 73-75% and 82-91% specificity, 66-69% and 41-62% PPV, 82-87% and 92-94% NPV, and 74-78% and 79-88% accuracy; respectively.
Conclusions: No significant difference was found between unenhanced MRI and direct MRA for the detection of tendinosis and tears of LHBTs.