Objective To assess long-term disease outcome of undifferentiated arthritis (UA) after initial treatment with methotrexate (MTX) or placebo.

Methods 110 patients with UA were randomised to receive MTX (n=55) or placebo (n=55) for 1 year. After 5 years the outcomes for diagnosis (rheumatoid arthritis, 1987 criteria (RA (1987)), UA or UA in remission) and radiographic progression were compared between treatment arms and anti-citrullinated protein antibody (ACPA)-positive and -negative patients. Outcomes were recalculated for patients who, with hindsight, might have been classified at baseline as having RA according to the 2010 criteria (RA (2010)).

Results 25 patients in the MTX group and 29 in the placebo group progressed to RA (1987) (p=0.45). MTX delayed progression from UA to RA (1987) but only in ACPA-positive patients. Drug-free remission was achieved in 35 patients, 20 of whom were initially treated with MTX, and 32 were ACPA-negative. ACPA-positive patients had more radiographic progression, regardless of treatment. Forty-three patients (39%) could be reclassified as having had RA (2010) at baseline, 6/24 (25%) of whom achieved remission after placebo treatment.

Conclusions After 5 years there is no lasting benefit of a 1 year initial course of MTX for patients with undifferentiated arthritis, compared with initial placebo. Progression to classifiable RA was not suppressed, drug-free remission not induced and the progression of radiological damage was similar in both groups. Reclassification at baseline with the 2010 criteria showed that 25% of patients with RA (2010) achieved spontaneous drug-free remission.