Prediction of long-term drug-free outcomes in ACPA-positive and ACPA-negative rheumatoid arthritis by combined clinical and ultrasound assessment of residual disease: a 5-year prospective study.

Abstract

Objective: To delineate, within the framework of current clinical practice and criteria, the sustainability of first-line immuno-suppressive treatment discontinuation in rheumatoid arthritis (RA) and the impact of residual disease in remission on long-term drug-free (DF) outcomes.

Methods: RA patients, referring to the Pavia early arthritis clinic (EAC) between 2009 and 2021 and achieving remission after Disease Activity Score-driven methotrexate (MTX) monotherapy, were recruited. Eligible patients underwent DF follow-up at 3-month intervals over 5 years after MTX discontinuation. Pre-selected clinical, serological and ultrasound (US) exposure variables at MTX withdrawal were analysed using multivariable Cox regression to predict time-to-flare.

Results: Of 761 EAC patients with RA, 132 started DF follow-up (person-months: 3678). 62 experienced a flare after a median (range) of 9 (3-60) months, resulting in a progressive decline in flare-free survival throughout the observation period. Whole-cohort multivariate Cox regression identified anti-citrullinated protein antibody (ACPA) positivity (HR: 4.20, 95% CI 2.37 to 7.44) and hands' joints with grey scale (US-GS) alterations (GS>1; HR: 2.18, 95% CI 1.20 to 3.93) as independent predictors. ACPA-positive patients in Simplified Disease Activity Index (SDAI) remission displayed a flare-free survival estimate at 5 years of 6.4% (95% CI 1.2 to 35.7) versus 78.2% (95% CI 67.4 to 90.8) for ACPA-negative patients in SDAI remission without residual US-GS alteration in hands' joints (n=59); the latter group showing no evidence of radiographic progression and functional deterioration.

Conclusions: Long-term DF remission is attainable in a niche subset of ACPA-negative RA. Examining clinical and subclinical residual synovial abnormalities during remission allows for effective preemptive identification of this subset in real life.