Receiver operating characteristic analysis of ultrasound joint inflammation in relation to structural damage and disease activity in rheumatoid arthritis

• Published in: Radiologia medica 2019-10, Vol.124 (10), p.1037-1042
• Main Authors: Li, HuiHua, Allen, John Carson, Thumboo, Julian, Tan, York Kiat
• Format: Article
• Language: English
• Subjects: Arthritis; Diagnostic Radiology; Health risk assessment; Imaging; Inflammation; Interventional Radiology; Joints (anatomy); Medicine; Medicine & Public Health; Neuroradiology; Radiology; Recesses; Rheumatoid arthritis; Sensitivity analysis; Ultrasonic imaging; Ultrasonography; Ultrasound
• ISSN: 0033-8362; 1826-6983
• DOI: 10.1007/s11547-019-01063-3

Objective To investigate whether ultrasound greyscale (GS) and power Doppler (PD) joint inflammation may be useful in identifying rheumatoid arthritis (RA) patients in different states of structural damage and disease activity. Methods In this cross-sectional study utilizing 36-joint ultrasonography, bone erosion was scored dichotomously (1 = yes/0 = no) while GS and PD joint inflammations were graded semi-quantitatively (0–3) at each joint recess. Sensitivity, specificity and receiver operating characteristic (ROC) curve analysis was applied to study ultrasound joint inflammation as a clinical marker for identifying patients with erosion score > 4.5 (median) and DAS28 > 2.6, > 3.2 and > 5.1, respectively. Results 1080 joints and 1800 joint recesses were scanned in 30 RA patients (mean disease duration, 70.3 months). Patients with GS score > 35.5 (median) had significantly higher ultrasound erosion scores when compared to those with GS score ≤ 35.5 (mean (95% CI) ultrasound erosion scores, 9.27 (6.12–12.4) versus 3.33 (2.31–4.36), respectively. p  = 0.0027). Patients with PD positivity had significantly higher DAS28 scores compared to those with PD negativity (mean (95% CI) DAS28, 3.84 (3.35, 4.34) versus 2.86 (2.18, 3.54), respectively. p  = 0.0457). Area under the ROC curve (AUC) based on cut-off GS scores ≥ 38 to identify patients with ultrasound erosion score >4.5 was 0.82 (sensitivity = 73.3%, specificity = 86.7%, accuracy = 80%). AUC based on cut-off PD scores ≥ 2.5 for identifying patients with DAS28 > 5.1 was 0.88 (sensitivity = 100%, specificity = 69.2%, accuracy = 73.3%). Conclusions Ultrasound GS and PD joint inflammation scores can be useful in identifying RA patients with high bone erosion burden (ultrasound erosion score > 4.5) and high disease activity (DAS28 > 5.1), respectively.