Diagnostic accuracy of magnetic resonance imaging for subscapularis tendon tears: a comparison with intraoperative arthroscopic findings

BACKGROUND: Subscapularis tendon (SSt) tears are frequently underdiagnosed by non-shoulder specialists. Delayed diagnosis may result in tendon retraction, muscle atrophy, and fatty degeneration, negatively affecting surgical outcomes. Magnetic resonance imaging (MRI) is widely used for preoperative evaluation, but its diagnostic accuracy for SSt tears remains controversial. This study aimed to evaluate the diagnostic performance of MRI in detecting SSt tears by comparing preoperative MRI findings with intraoperative arthroscopic findings. METHODS: A retrospective analysis was conducted on patients who underwent shoulder arthroscopy for rotator cuff pathology between January 2014 and December 2023. Preoperative MRI evaluations were performed by an experienced musculoskeletal radiologist, while intraoperative assessment was conducted by an orthopedic surgeon specialized in shoulder surgery and served as the reference standard. The sensitivity, specificity, positive predictive value, negative predictive value, overall accuracy, prevalence, positive likelihood ratio, and negative likelihood ratio of MRI for detecting SSt tears were calculated with 95% confidence intervals (CIs). RESULTS: A total of 128 patients were included, with a mean age of 65.4 years. Intraoperatively, SSt tears were identified and repaired in 77 patients, while 51 patients had an intact tendon. MRI reports identified SSt tears in 48 patients and reported an intact tendon in 80 patients. The sensitivity and specificity of MRI for detecting SSt tears were 45.5% (95% CI 0.346-0.567) and 74.5% (95% CI 0.610-0.848), respectively. The positive predictive value was 72.9% (95% CI 0.587-0.837), and the negative predictive value was 47.5% (95% CI 0.368-0.585). The overall accuracy was 57.0% (95% CI 0.483-0.654). The prevalence of SSt tears was 60.2% (95% CI 0.514-0.684). The positive likelihood ratio was 1.78 and the negative likelihood ratio was 0.73. Among 77 arthroscopically confirmed tears, 70 were partial thickness tears and 7 were full thickness tears. All 42 false negatives occurred in partial-thickness tears. CONCLUSIONS: MRI demonstrated limited sensitivity in detecting SSt tears despite moderate specificity. These findings indicate that MRI alone may be insufficient for establishing a reliable diagnosis. Clinical examination findings should be integrated with imaging results in the assessment of patients with shoulder pain.