In rheumatoid arthritis (RA), the consensual 1st line conventional synthetic disease modifying antirheumatic drugs (csDMARD) of RA is methotrexate (MTX). In case of contra-indication or intolerance to MTX, leflunomide is an alternative. If the treatment target is not achieved with csDMARD strategy, addition of a biological DMARD (TNF inhibitors, anti-Interleukin 6 (anti-IL6)), abatacept, or rituximab) or a targeted synthetic (ts) DMARD (JAK inhibitors) is considered. Current practice is to start a bDMARD (biologic Disease Modifying Antirheumatic Drugs) and especially TNF inhibitors (etanercept or monoclonal anti-TNF antibodies) with the benefit of hindsight. However, abatacept and TNF inhibitors have demonstrated similar efficacy in patients with insufficient response to csDMARD (AMPLE trial). Although abatacept has shown a very good tolerance profile that might be superior to other bDMARDs rheumatologists might be reluctant to use it as a first line bDMARD as there is a belief of a slower efficacy compared to other bDMARDs or JAK inhibitors. Indeed, in real world study, compared to TNF inhibitors it seems that discontinuation of abatacept is more related to lack of effectiveness than safety issues. Investigators have hypothesized that first rapidly controlling the inflammation phase, using TNF inhibitors followed by abatacept to induce an immunological remission would optimize response and tolerance of ACPA positive patients with RA. To demonstrate our hypothesis, the investigaors propose a randomized controlled trial with one arm receiving an induction therapy for 12 weeks with a TNF inhibitor followed by a cell-targeted bDMARD (abatacept) and the other arm, receiving TNF inhibitors.