A retrospective case-control review was conducted on prospectively collected data on adult patients who underwent shoulder arthroscopy between 2019 and 2022 at a single institution. Inclusion criteria included a preoperative noncontrast MRI with a 1.5-Tesla magnet and subsequent arthroscopic evaluation. Patients younger than 18 years of age were excluded, as well as those with an MRI with motion artifact, history of ipsilateral shoulder surgery, proximal humerus or glenoid fracture, signs of multidirectional shoulder instability, or incomplete or inadequate operative documentation. Patients with an arthroscopically confirmed pulley lesion comprised the study group, whereas those with an arthroscopically intact biceps pulley were categorized as the control group. Arthroscopic findings were considered the gold standard. The protocol was approved by the local institutional review board before the study’s inception. The requirement for informed consent was waived.

All arthroscopic procedures were performed by a fellowship-trained shoulder surgeon (A.G.). At the time, the surgeon was aware of the MRI interpretation before surgery. Intraoperatively, patients were positioned in the lateral decubitus position, and conventional portals were used (i.e., posterior, anterosuperior and lateral). Subscapularis tendon integrity, the LHBT complex, and surrounding structures were evaluated with 30° and 70° arthroscopes inserted through the posterior portal. The presence of biceps medial pulley lesion ( Fig 1 ), lateral pulley lesion ( Fig 2 ), signs of tendinopathy of the LHBT, and the integrity of rotator cuff muscles adjacent to the rotator interval were documented. Therapeutic procedures performed at arthroscopy were documented.

Arthroscopic posterior portal view of a left shoulder showing the integrity of the medial bicipital pulley (∗) in relation to the LHBT and SSC tendon (A) and discontinuity of the medial pulley (∗) associated with an upper third SSC tear (B). (CP, chondral print; HH, humeral head; LHBT, long head biceps tendon; SSC, subscapularis tendon.)

Arthroscopic posterior portal view of a left shoulder showing the integrity of the lateral bicipital pulley (∗) in relation to the LHBT and SSP tendon (A) and discontinuity of the lateral pulley (∗) associated with an anterior SSP tear (B). (HH, humeral head; LHBT, long head biceps tendon; SSC, subscapularis tendon.)

Each patient underwent conventional shoulder MRI on a 1.5-Tesla whole-body scanner equipped with a 16-channel shoulder coil at varying institutions (i.e., outside MRIs). The shoulder was positioned neutrally in accordance with a standardized protocol. Imaging sequences included triplanar intermediate-weighted turbo-spin echo sequences (TSE) with spectral fat suppression, a sagittal T2-weighted TSE sequence, and a coronal T1-weighted TSE sequence. No intra-articular or intravenous contrast was administered for any examination.

Five independent assessors—comprising 3 fellowship-trained shoulder surgeons (assessors 1, 2, and 3) and 2 musculoskeletal radiologists with at least 10 years of experience in shoulder imaging (assessors 4 and 5)—evaluated the MRIs. All assessors were blinded to clinical and surgical information and analyzed the scans using predefined MRI assessment criteria. Specifically, they assessed 6 of the 7 MRI signs of LHBT instability described by Zappia et al. : chondral print ( Fig 3 ), humeral head subchondral bone edema at the chondral print, LHBT angle ( Fig 4 ), LHBT-groove distance ( Fig 5 ), LHBT subluxation or dislocation on the axial plane ( Fig 6 ), and displacement sign ( Fig 7 ). The detour sign was excluded because of its poor diagnostic performance in previous studies. Each assessor recorded the presence or absence of the individual signs and ultimately determined whether a biceps pulley lesion was present.

Chondral print sign (white arrowhead) and humeral head subchondral bone edema (black arrowheads) are shown in a sagittal view of a T1-weighted right shoulder MRI. (MRI, magnetic resonance imaging.)

LHBT angle sign is shown in a coronal view of a T1-weighted right shoulder MRI. The angle is measured between the intra-articular portion of the LHBT and the superior edge of the SSC tendon. An angle >35° is considered a positive sign. (LHBT, long head biceps tendon; MRI, magnetic resonance imaging; SSC, subscapularis tendon.)

LHBT-groove distance sign (dashed line) is shown in an axial view of a T1-weighted right shoulder MRI. The axillary nerve is indicated with a white arrowhead. The lines represent the axillary neurovascular bundle comprised of the axillary artery (red line), vein (blue line), and nerve (yellow line). (LHBT, long head biceps tendon; MRI, magnetic resonance imaging.)

LHBT subluxation sign is shown in an axial view of a T1-weighted right shoulder MRI. The white line marks the limit of the bicipital groove. The white curved arrow shows the direction of a subluxed or displaced LHBT from the bicipital groove. (LHBT, long head biceps tendon; MRI, magnetic resonance imaging.)

Displacement sign (white curved arrow) is shown in a sagittal view of a T1-weighted right shoulder MRI. (MRI, magnetic resonance imaging.)

With arthroscopy as the gold standard, the sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and accuracy of each sign, as reported by each assessor in the identification biceps instability, were calculated from merged interpretations from all assessors, and compared between surgeons (assessors 1, 2, and 3) and radiologists (assessors 4 and 5). If at least 1 of the 6 signs was present, the MRI was considered positive. Cohen’s kappa (k) was used to evaluate inter-rater agreement for each sign. Cohen’s kappa coefficient is interpreted as follows: a value of ≤0 indicates no agreement, 0.01 to 0.20 indicates slight agreement, 0.21 to 0.40 indicates fair agreement, 0.41 to 0.60 indicates moderate agreement, 0.61 to 0.80 indicates substantial agreement, and 0.81 to 1.00 indicates a nearly perfect agreement, as classified by Landis and Koch. To compare the diagnostic accuracy between assessors and radiologists versus surgeons, the area under the receiver operating curve was used. A threshold of 0.05 was used to denote statistical significance. The data were evaluated with the SPSS 29 program (IBM Corp., Armonk, NY).