Purpose: Compare the diagnostic performance of saline and gadolinium shoulder magnetic resonance arthrograms (MRA) in the detection of labral and rotator cuff injury compared to arthroscopy.
Materials and methods: Consecutive patients who underwent a gadolinium or saline MRA followed by arthroscopy were retrospectively reviewed. The reports were reviewed for injuries. A chi square or Fisher's exact test was performed to compare the MRA and surgery. Kappa values were calculated to correlate diagnosis of tear between MRA and arthroscopy.
Results: There were a total of 58 patients included, including 34 gadolinium arthrograms and 24 saline arthrograms. The accuracy of saline MRA was similar compared to gadolinium MRA in the diagnosis of tears of the supraspinatus (accuracy 0.88 vs 0.74, respectively) and infraspinatus (accuracy 0.88 vs 0.65, respectively) tendons and tears of the anterior/anterior inferior, posterior, and superior labrum, (accuracy 0.79 vs 0.76, 0.71 vs 0.62, and 0.58 vs 0.56), and saline vs gadolinium, respectively. Although there was a trend toward overall better saline MRA performance, a statistically significant difference in the accuracy to detect tears was only noted for the infraspinatus tendon. Interobserver agreement for rotator cuff tears was higher for saline than gadolinium MRA.
Conclusion: Saline MRA was accurate, with no significant differences compared gadolinium arthrograms in the diagnosis of labral and rotator cuff pathology. Given expense, and the potential additional information provided by fluid sensitive sequences over T1 fat-suppressed sequences, consideration should be given to using saline for shoulder MRAs.
Level of evidence: Level III, Retrospective Cohort Study.
Keywords: Diagnostic performance; Gadolinium; MR arthrography; Saline; Shoulder.