AB1167 RELIABILITY AND VALIDITY OF ULTRASOUND PATHOLOGIES IN KNEE OSTEOARTHRITIS FOR SEMI-QUANTITATIVE AND QUANTITATIVE METHODS WITH MRI AS A REFERENCE

Abstract

Background Ultrasound can visualize both structural and inflammatory changes of osteoarthritis (OA) with greater accessibility. Therefore, the Outcome Measures in Rheumatology (OMERACT) Ultrasound Task Force recently proposed a semi-quantitative knee ultrasound scoring system1 which requires further validation. Objectives To investigate inter-rater reliability (IRR) of ultrasound pathologies between three raters with varying experiences and examine construct validity of these features against magnetic resonance imaging (MRI) using OMERACT semi-quantitative ultrasound image atlas and quantitative ultrasound measures. Methods According to standardized OMERACT scanning protocol, 20 participants with symptomatic and radiographic knee OA were scanned dynamically with a multi-frequency linear transducer (6-18MHz) of Aplio Platinum 500 machine, Toshiba, Japan. Following an initial calibration of ultrasound scores using the image atlas, three raters with varying experience (a physician operator certified with RhMSUS, a musculoskeletal ultrasonographer and a medical student) independently obtained semi-quantitative scores (0-3) for synovitis and power Doppler activity in suprapatellar recess, medial osteophytes and medial meniscal extrusion, as well as quantitative measurements (mm) of synovitis, effusion, synovial hypertrophy, medial osteophyte and meniscal extrusion (maximal measures). Semi-quantitative MRI Osteoarthritis Knee Score (MOAKS) scores of effusion-synovitis, osteophyte and meniscal extrusion were independently calculated by an experienced researcher. 3-T Sagittal proton-density (PD) weighted fat-suppressed turbo spin-echo (TSE) non-contrast MRI sequences were used for quantitative effusion-synovitis, and coronal plane for osteophyte and meniscal extrusion. Weighted kappa coefficient (Kw) or intra-class correlation coefficient (ICC) were calculated for IRR, Spearman’s rank (rs) and Pearson (rp) correlation coefficients for construct validity and Bland-Altman plots for extent of agreement between ultrasound and MRI. Results Demographics were 60.2±8.3 years of age, 65% female, BMI of 28.6±4.5 of kg/m2; 60% with Kellgren and Lawrence (KL) grade 2 and 40% with a KL grade 3. Semi-quantitative OMERACT ultrasound scores (Table.1) revealed good to excellent IRR (Kw=0.73 -0.88) for osteophyte, and moderate to good IRR (Kw=0.42-0.66) for synovitis. Conversely, quantitative measures of ultrasound pathologies had excellent IRR (ICC= 0.84-0.95) except for synovial hypertrophy (ICC=0.67-0.72). A significant association was found between semi-quantitative ultrasound synovitis and MRI effusion-synovitis (rs= 0.48, p=0.03). All three quantitative ultrasound measurements were significantly associated with quantitative scores of MRI (Table. 2). In Bland–Altman plots for quantitative measurements, there were systemic offsets of 0.6 mm, 1.9 mm and 9.8 mm for osteophyte and meniscal extrusion and synovitis respectively (Fig. 1). Conclusion These is moderate to good IRR between operators with varying experience using the OMERACT knee scoring image atlas for osteophyte and synovitis. While quantitative ultrasound measurements showed excellent IRR and significant association with MRI quantitative outcomes, the absolute feature-specific agreement is called into question.