Background: Various protocols exist for magnetic resonance arthrogram (MRA) of the shoulder, including 3D isotropic scanning and positioning in neutral (2D neutral MRA), or abduction-external-rotation (ABER).
Hypothesis: MRA does not improve diagnostic accuracy for labral tears when compared to magnetic resonance imaging (MRI) performed using 3-Tesla (3T) magnets.
Method: Systematic review of the Cochrane, MEDLINE, and PubMed databases according to PRISMA guidelines. Included studies compared 3T MRI or 3T MRA (index tests) to arthroscopic findings (reference test). Methodological appraisal performed using QUADAS-2. Pooled sensitivity and specificity were calculated.
Results: Ten studies including 929 patients were included. Index test bias and applicability were a concern in the majority of studies. The use of arthroscopy as the reference test raised concern of verification bias in all studies. For anterior labral lesions, 3T MRI was less sensitive (0.83 vs. 0.87 p = 0.083) than 3T 2D neutral MRA. Compared to 3T 2D neutral MRA, both 3T 3D Isotropic MRA and 3T ABER MRA significantly improved sensitivity (0.87 vs. 0.95 vs. 0.94). For SLAP lesions, 3T 2D neutral MRA was of similar sensitivity to 3T MRI (0.84 vs. 0.83, p = 0.575), but less specific (0.99 vs. 0.92 p < 0.0001). For posterior labral lesions, 3T 2D neutral MRA had greater sensitivity than 3T 3D Isotropic MRA and 3T MRI (0.90 vs. 0.83 vs. 0.83).
Conclusions: At 3-T, MRA improved sensitivity for diagnosis of anterior and posterior labral lesions, but reduced specificity in diagnosis of SLAP tears. 3T MRA with ABER positioning further improved sensitivity in diagnosis of anterior labral tears.
Level of evidence: IV.
Keywords: 3-Tesla; Accuracy; Diagnosis; Labrum; MRA; MRI.