Background With more biological disease modifying antirheumatic drugs (bDMARDs) in rheumatoid arthritis (RA), switching between bDMARDs is becoming more frequent. Patients (Pts) discontinue bDMARDs for various reasons, including inadequate effectiveness (IE) and adverse events (AE). Observational data indicates that the reason for discontinuing a second TNF inhibitor is usually the same that led to the first TNFi discontinuation, suggesting that a change in bDMARD mode of action may be more favourable.
Objectives To investigate the impact of specific reasons for discontinuing the previous bDMARD on the clinical response to abatacept (ABA), a bDMARD with a different mode of action.
Methods This is a pooled observational database analysis of 10 prospective cohorts of RA pts treated with iv ABA. All pts with available information on the reason for discontinuation of the last bDMARD were included. Pts initiating ABA as a first bDMARD were excluded from the analysis. The predictor of interest was the reason for prior bDMARD discontinuation, categorized as IE, AE, or other. The primary endpoint was time to ABA discontinuation, defined as the time between drug initiation to last administration: 1) for any reason, and 2) specifically for AEs or IE. Cox regression was used to estimate hazard ratios (HRs) for drug discontinuation, adjusting for pts demographics, disease and treatment characteristics.
Results Of the 2001 RA pts included, 1272 discontinued ABA during the 3639 patient-years of follow-up. 499 pts (24.9%) had stopped their last bDMARD for AEs, 1290 pts (64.5%) for IE, and 212 pts (10.6%) for other reasons. Pts who discontinued their prior bDMARD for IE were younger, more frequently RF positive, had lower BMI, shorter disease durations, higher baseline DAS28 and had less comorbidities than pts who stopped their previous bDMARD for other reasons.
There was no association between reason for discontinuing prior bDMARDs (AE, IE, or other) and overall ABA retention (log-rank p=0.78). ABA discontinuation for AEs was significantly associated with prior discontinuation of bDMARDs for AEs (log-rank p<0.001), and ABA discontinuation for IE was significantly associated with prior bDMARD discontinuation for IE (log-rank, p=0.02). In adjusted models, discontinuation of the prior bDMARD due to AE doubled the risk of AE-related discontinuation with ABA (HR 2.15, 95% CI: 1.63 – 2.84), compared to pts who had not discontinued their last bDMARD for AEs. IE on the prior bDMARD also tended to increase the risk of discontinuing ABA for IE (HR 1.15, 95% CI: 0.98 – 1.35), compared to pts without IE on their last bDMARD.
Conclusions The same reason that led to discontinuing prior bDMARDs is likely to lead to the discontinuation of a new bDMARD with a different mode of action. Overall these results suggest that discontinuation is mostly explained by patient-specific characteristics (selection) and less by particular drug mechanism.
Acknowledgement Unrestricted research grant from BMS
Disclosure of Interest A. Finckh Grant/research support from: BMS, D. Courvoisier: None declared, J. Gottenberg: None declared, M. Hernandez: None declared, F. Iannone: None declared, E. Lie: None declared, H. Canhão: None declared, K. Pavelka: None declared, M. Hetland: None declared, C. Turesson: None declared, X. Mariette: None declared, D. Choquette: None declared