ABSTRACT
OBJECTIVES: To investigate, in anti-cyclic citrullinated peptide antibody positive individuals with musculoskeletal symptoms but no clinical synovitis (CCP+ at-risk), the additional value of ultrasound (US) for the prediction of inflammatory arthritis (IA). Furthermore, to define a concise US protocol for feasible risk prediction. METHODS: Demographic and clinical data were collected in 417 CCP+ at-risk (Leeds CCP cohort) with a baseline US scan assessing synovitis and bone erosions in 36 joints, and a follow-up duration ≥24 months. Multivariable binary regression models for IA development at 24 months evaluated routine clinical variables associated with IA alone ("clinical" model) and combined with a 36-joint US scanning protocol ("clinical-US extended" model). A "clinical-US short" model was developed. RESULTS: At 24 months, 92/417 (22.1%) CCP+ at-risk developed IA (median time: 7 months, IQR : 3-12). The "clinical-US extended" model performed better than the "clinical" model (AUC 0.788 vs AUC 0.731 respectively, p< 0.001) with an odds ratio for IA development of 3.18 (95% IC 1.80-5.63) for US synovitis and 2.54 (95% IC 1.21-5.37) for bone erosions. The "clinical-US short" model, which retained the wrists, knees and MTP5 joints, performed better (AUC 0.782) than the "clinical" model (p< 0.001) and similarly (difference in Akaike information criteria <2) to the "clinical-US extended" model. CONCLUSIONS: US provides valuable information for predicting progression to IA in CCP+ individuals both alone and in addition to clinical variables. US synovitis was associated with a threefold increase risk of IA development. A concise US protocol of 6 joints provides clinically feasible risk prediction in CCP+ at-risk.
