Limited value for ultrasonography in predicting flare in rheumatoid arthritis patients with low disease activity stopping TNF inhibitors

Objective. Ultrasonography (US) can be used for treatment decisions in RA patients. This study investigated the added value of US to clinical variables in predicting flare in RA patients with longstanding low disease activity when stopping TNF inhibitors (TNFi). Methods. Cox models with and without using US added to clinical variables were developed in the Potential Optimization of Expediency of TNFi-UltraSonography study. RA patients (n = 259), using > 1 year TNFi and csDMARD with DAS28 < 3.2 for 6 months prior to inclusion, were followed for 52 weeks after stopping TNFi. The added value of US was assessed in two ways: first, by the extent to which individual predictions for flare at 52 weeks with and without US differed; and second, by comparing how US information improved the prediction to classify patients at 52 weeks in the low risk (<33% flare), intermediate risk (33-50%) and high risk (50-100%) groups. Results. Although US was predictive of flare at group level (multivariate hazard ratio = 1.7; 95% CI: 1.1, 2.5), individual predictions for flare at 52 weeks with and without US differed little (median difference 3.7%; interquartile range: -7.8 to 6.5%). With US, 15.9% of patients were designated low risk; without US, 14.6%. In fact, 12.0% of patients were US-classified as low risk with/without knowing US. Conclusion. In RA patients with longstanding low disease activity, at time of stopping TNFi, US is a predictor for flare at group level, but at the patient level, US has limited added value when common clinical parameters are used already, though the predictive value of clinical predictors is modest as well.