Which MRI signs truly predict subscapularis tears? A multivariate analysis of 278 arthroscopic cases

To determine the diagnostic performance of magnetic resonance imaging (MRI) for subscapularis (SSC) tendon tears, identify the strongest independent MRI predictors, and evaluate the accuracy of MRI in predicting tear size.

A retrospective cohort study of patients who underwent arthroscopic rotator cuff repair was conducted. From a total of 376 surgeries, 278 patients met the inclusion criteria. All patients had a preoperative 1.5-T MRI. A blinded shoulder surgeon and a musculoskeletal radiologist independently evaluated six MRI signs for SSC rupture. Arthroscopic findings served as the gold standard. Diagnostic performance metrics (accuracy and predictive value) were calculated, and a multivariate logistic regression was performed to identify independent predictors.

Arthroscopy confirmed an SSC tear in 99 of 278 patients (35.6%). For the radiologist, overall accuracy was 73%; for the surgeon, it was 70%. Performance was substantially higher for complete tears than for partial tears, with a positive predictive value for partial tears of only 24%. Multivariate logistic regression identified two independent predictors of an SSC tear: tendon hyperintensity on axial images (OR 3.97, p < 0.001) and a "naked" lesser tuberosity on sagittal images (OR 4.83, p = 0.029). A simplified diagnostic model based on these two signs achieved an accuracy of 75.5%, with 74.7% sensitivity and 76.0% specificity.

MRI performance is poor for diagnosing partial SSC tears but good for complete tears. Tendon hyperintensity (axial) and a naked lesser tuberosity (sagittal) are the strongest independent predictors of an SSC tear. A simplified diagnostic model focusing on these two signs may offer a more efficient and accurate approach to preoperative assessment.