Development and validation of a severity-focused score for the assessment of lateral epicondylitis using enhanced magnetic resonance imaging

Abstract

Background: Magnetic resonance imaging (MRI) does not sufficiently reflect the severity of lateral epicondylitis. Although some authors have reported meaningful findings on conventional MRI, it is not widely accepted that MRI is useful for evaluating lateral epicondylitis. The aim of this study is to evaluate the relationship between the lateral epicondylitis severity score (LESS) using contrast-enhanced T1-weighted MRI and patients' perceptive pain and clinical courses. Methods: This retrospective study included 80 patients diagnosed with lateral epicondylitis and who also went contrast-enhanced 3.0-T MRI. The study cohort was split into a development (n = 50) and a validation cohort (n = 30). The patient's pain was assessed using a visual analog scale (VAS), and the involvement of the common extension tendon (CET) origin, radial collateral ligament, lateral ulnar collateral ligament, and plica were evaluated using T2-weighted images. The enhancement of CET and subcutaneous tissue was scored using a novel classification in the contrast-enhanced T1-weighted MRI. LESS was calculated from MRI findings to determine the severity of lateral epicondylitis. Pearson's correlation analysis was performed to evaluate the relationship between patients' VAS scores and MRI findings. Receiver operating characteristic curve was constructed to determine cut-off value of LESS for the surgical treatment of lateral epicondylitis. Results: The average age of the development cohort (n = 50) and the validation cohort (n = 30) were 49.8 (31-66) years and 52.3 (37-67) years, and the average VAS score were 7.3 (4-9) and 7.2 (5-10) respectively. In T2-weighted image evaluation, the correlation of VAS and grade of CET, radial collateral ligament, lateral ulnar collateral ligament, and plica were −0.03 (P = .87), −0.03 (P = .87), −0.14 (P = .39), and 0.36 (P = .02). Discordant contrast enhancement in T1-weighted images compared to nonspecific T2-weighted images was observed in 46 of 50 cases (92%). A statistically significant correlation was observed between the patients' VAS and LESS scores (r = 0.67, P < .01). Area under curve was 0.76 (P < .01) and the optimal cut-off value of LESS for the surgical treatment of lateral epicondylitis was 5.5 (sensitivity 72.7%, specificity 70.6%). When the cut-off value was used to predict surgical treatment in the validation cohort, the sensitivity was 100% and the specificity was 75%. Conclusion: Contrast-enhanced T1-weighted images provided more detail than conventional T2-weighted images in evaluation of structural or unobserved abnormalities to assess the severity of the lateral epicondylitis, and had excellent inter- and intraobserver reliability. LESS showed significant positive correlation with pain severity. If the LESS is 5.5 or more, surgical treatment of lateral epicondylitis may need to be considered.