P199 The reliability of using ultrasound for the detection of hypoechogenicity of enthesitis in patients with spondyloarthropathies

Abstract

Abstract Background/Aims Enthesitis, which is the inflammation of the entheses area where the tendon or ligaments are attached into the bone, is considered a hallmark and pathological feature for spondyloarthropathies (SpA). The lower limb entheses are more prone to being affected than the upper extremity enthesis, the most common one is Achilles tendon where heel enthesitis is considered the most frequent finding in SpA patients. Ultrasound is widely used by rheumatologists because it has higher sensitivity and specificity in detecting enthesitis compared with clinical assessment. The abnormal characteristics of enthesitis in greyscale ultrasound includes the hypoechoic area at the insertion or body of the tendon due to the loss of the normal fibrillar pattern, increase in the tendon thickness, bone erosion, calcification and increase in Dopplar signals. Although ultrasound has many advantages compared with other imaging modalities, it is operator-dependent and the diagnosis is exposed to the subjectivity of the observer, which leads to the variability in defining the abnormal features of enthesitis. Using static ultrasound images of Achilles tendon entheses of patients with SpA, we first determined the intra- and inter-observer reliability of grading ‘hypoechogenicity’ using a panel of readers and second, compared a computerised pixel counting method against an expert consensus score to determine the level of agreement between approaches. Methods Six participants (rheumatologists and sonographer) with experience in ultrasound scored the presence of hypoechogenicity in 100 static images of Achilles tendon entheses of patients with SpA. Two scoring systems were used- an OMERACT derived SQ system (0-3) of the whole enthesis and binary score system of the distal 2mm (0-1). The intra-class correlation coefficient (ICC) and Cohen's kappa was used to assess inter-observer reliability. The ImageJ software was used to measure the mean grey values (MGV) of the pixels within the enthesis. Results The inter-observer reliability was good for using the SQ score (ICC 0.780 (95% CI 0.691-0.849) and moderate for the binary score (ICC 0.632 (95% CI 0.490-0.745)). There was no match between the results from the quantitative score of the pixels MGV and the expert semi-quantitative score. Conclusion This study is novel as in that it has specifically evaluated the scoring of hypoechogenicity within the entheses of patients with SpA. It has demonstrated variation in scoring between observers and highlighted the challenges of image interpretation. A lack of correlation of expert scoring with the manual pixel MGV was limited by the image artefacts and different machine settings which may have implications as we work towards future AI systems. Disclosure A.S. Aldahes: None. R. Wakefield: None. K. Smith: None.