Routine musculoskeletal ultrasound findings impact diagnostic decisions maximally in autoantibody-seronegative early arthritis patients.

Abstract

The diagnostic value added by musculoskeletal ultrasound (MSUS) over standard clinical and laboratory parameters has proved difficult to quantify. The additive contribution to diagnostic classification of a pragmatic, 15 min MSUS protocol was appraised in a large, unselected cohort of early arthritis clinic attendees. Detailed baseline characteristics were recorded. Semi-quantitative MSUS scoring of the most symptomatic wrist, second/third MCPs and PIPs and second/fifth MTPs was recorded, along with the sonographer's scan impression (definitely inflammatory, possibly inflammatory or non-inflammatory). MSUS findings were available to rheumatologist diagnosticians during subsequent consultations. Persistent inflammatory arthritis (PIA) was classified only where patients were started on ≥1 DMARD. Multivariate and receiver operating characteristic (ROC) curve analyses were used to identify independent discriminators of PIA, and the added value of MSUS parameters. Eight hundred and thirty-one patients were enrolled, of whom 31.3% acquired a PIA diagnosis. Swollen joint count, CRP, age and ACPA status were non-redundant clinical/laboratory predictors of a PIA diagnosis by consulting rheumatologists, with good discriminatory utility (area under the ROC curve, AUROC, 0.88). While the additive contribution of summed parameters from the seven-joint MSUS protocol to this model was statistically significant (P = 0.004), it was numerically small (ΔAUROC 0.02). However, the additive contribution to diagnostic outcome of sonographer's scan impression over clinical parameters alone became substantial in the sub-cohort of ACPA-negative patients, increasing the AUROC by 9% from 0.81 to 0.90 (P < 0.0001). The clinical utility of a 15-min MSUS screen for diagnosing PIA requiring DMARDs is most evident among ACPA-negative patients attending an early arthritis clinic.