Accuracy and Reliability of Imaging Modalities for the Diagnosis and Quantification of Hill-Sachs Lesions: A Systematic Review

Purpose: To determine the reliability and accuracy of different imaging modalities in assessing Hill-Sachs lesions within the setting of anterior shoulder instability. Methods: A systematic review was performed according to the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-analyses) guidelines using the PubMed, Scopus, Embase, and Cochrane Library databases. The inclusion criteria were clinical trials or cadaveric studies that assessed the accuracy of humeral head bone loss imaging or reliability and English-language articles. The exclusion criteria were animal studies; imaging studies without measures of accuracy, reliability, or clinical predictive power; studies of shoulder injuries without humeral head bone loss; editorials; abstracts; reviews; case reports; and surveys. The search terms included "imaging" OR "radiographic" OR "CT" OR "MRI" AND "Hill-Sachs" OR "humeral head bone loss." Assessment of the methodologic quality of the included studies was performed using the original Quality Assessment of Diagnostic Accuracy Studies (QUADAS) tool. Results: Forty studies (2,560 shoulders) met the inclusion criteria and were assessed. For diagnosing the presence of Hill-Sachs lesions, computed tomography (CT) arthrography had the highest reported accuracy (median, 91%; range, 66%-100%). For the same assessment, CT arthrography also had the greatest reported sensitivity (median, 94%; range, 50%-100%). For the quantification of Hill-Sachs lesion parameters, reported intraobserver reliabilities were highest for 3-dimensional (3D) CT (intraclass correlation coefficient [ICC] range, 0.916-0.999), followed by 2-dimensional CT (ICC range, 0.858-0.861) and magnetic resonance imaging (MRI) (ICC range, 0.28-0.97). For the same quantification parameters, interobserver reliabilities were also reported for 3D CT (ICC range, 0.772-0.996), 2-dimensional CT (ICC range, 0.721-0.879), and MRI (k range, 0.444-0.700). Intraobserver reliabilities for determining glenoid tracking were only reported for 3D CT (k range, 0.730-1.00; ICC range, 0.803-0.901) and MRI (ICC range, 0.770-0.790). Conclusions: This study shows that the current literature supports a variety of different imaging modalities that provide clinically acceptable accuracy in diagnosing and quantifying Hill-Sachs lesions, as well as determining whether they will cause persistent anterior shoulder instability. Furthermore, this systematic review justifies that further research is needed to help develop a treatment algorithm on the proper imaging modalities needed to help treat patients with anterior shoulder instability that is both reliable and financially acceptable.