Background Early identification of patients at risk of persistent as opposed to spontaneously resolving arthritis is essential to allow the prompt institution of therapy and to provide vital information from the patient’s perspective. Currently available prediction strategies for persistent disease include that developed by Visser et al1. The identification of biomarkers that can improve such prediction strategies is an important goal. We have previously shown that data from ultrasound joint assessment adds to the accuracy of the Leiden Rule for prediction of RA. Here we present data on the utility of ultrasound to predict persistence in a very early synovitis cohort.
Objectives To evaluate musculoskeletal ultrasound (MSUS) detection of synovitis and erosions as a predictor of persistent disease in patients with very early synovitis.
Methods 91 patients with clinically apparent synovitis of at least one joint and duration ≤3 months of any symptom attributable to inflammatory arthritis were followed prospectively for 18 months and underwent clinical, laboratory, radiographic and 38 joint ultrasound (MSUS) assessments at baseline. MSUS variables included power Doppler (PD) and greyscale (GS) defined on 0-3 categorical scales. Sensitivity and specificity for persistent disease (at 18 months of follow-up) were determined for MSUS variables. Logistic regression models were used to evaluate the predictive ability of MSUS over and above the Visser score.
Results 35 (38%) patients had resolving disease, whilst 56 (62%) developed persistent disease (including both RA and non-RA phenotypes). Ultrasound demonstrated subclinical joint involvement in both patient groups for all joint regions assessed. However power Doppler (PD) and Greyscale (GS) median joint counts were all significantly higher in the patients with persistent disease compared with those with resolving disease (PD: 10 vs 3 joints, p<0.001 and GS: 14 vs 4 joints, p<0.001, respectively). 18 patients (32.1%) with persistent disease had erosions of hands or feet detectable at baseline by ultrasound, not present in any patients with resolving disease (p<0.001). Global ultrasound joint counts and several individual MSUS variables, particularly at the MCP and PIP joints, were associated with persistence independently of the Visser score. Logistic regression analysis showed that compared to the Visser score alone, the Visser score with added MSUS variables GS≥2 at the MCP joints and PD≥2 at the PIP joints improves area under the curve values significantly (0.816 and 0.912 respectively, p<0.01).
Conclusions Subclinical disease detected by ultrasound is more common in persistent than resolving disease. We have identified a number of ultrasound variables that provide important prognostic information on the development of persistent disease and have shown that useful data can be gathered from the assessment of a limited number of joints. Larger studies are required to fully evaluate the optimal weightings of these variables for use in a persistence algorithm.
 Visser H, le Cessie S, Vos K, Breedveld FC, Hazes JM. How to diagnose rheumatoid arthritis early: a prediction model for persistent (erosive) arthritis. Arthritis Rheum 46(2):357-65.
Disclosure of Interest A. Filer Grant/Research support from: Cellzome and Pfizer, M. Cader: None Declared, A. Abhishek: None Declared, G. Allen: None Declared, C. Buckley Grant/Research support from: Wyeth, Cellzome, UCB and Pfizer, P. de Pablo: None Declared, K. Raza Grant/Research support from: Wyeth, Cellzome, UCB and Pfizer