Background TNF-a inhibitors have completely changed the prognosis of rheumatoid arthritis (RA). The number of molecules and the time of exposure have increased. However, few studies have compared i) the safety of the five now available TNF-a inhibitors between them, and ii) their medium term safety (after more than 6 months of use).
Objectives The aim of our study was to compare the data regarding medium-term safety obtained from randomized controlled trials for the five TNF-a inhibitors used in the treatment of RA. We particularly aimed to focus on two serious events: death and severe infections.
Methods A systematic review of articles published from January 2000 to January 2013 was performed using MEDLINE, EMBASE and COCHRANE databases. We included randomized, controlled, double blind and parallel-group trials, with a follow-up period of at least 6 months, comparing TNF-a inhibitors to placebo, with or without concomitant treatment with Methotrexate. The primary outcome measure was the occurrence of a serious adverse event, defined by the occurrence of death or of a serious infection (this latter defined in all studies by the requirement of hospitalization). Secondary outcome measures were the occurrence of death or serious infection. MedCalc software was used for the statistical analysis.
Results Among the 396 articles initially identified, 22 were finally selected. These articles included 6682 patients in the TNF-a inhibitor group (Adalimumab: 2507, Golimumab: 575, Certolizumab: 1512, Infliximab: 1087 and Etanercept: 1001) and 3607 in the placebo group. The duration of patient follow-up ranged from 24 to 78 weeks.
The analysis of our primary outcome showed no increased risk of serious adverse event (death or severe infection) in the TNF-a inhibitor group compared to the placebo group (Odds Ratio, OR: 1.14, 95% Confidence Interval, CI: 0.91-1.44). Sensitivity analysis with respect to each molecule did not change the main conclusion, except for Certolizumab, which showed an increased risk of serious adverse events (OR: 2.06, CI: 1.35-3.15).
Regarding secondary outcomes, no increased risk of death in the TNF-a inhibitor group compared to the placebo group was observed (OR: 1.27, CI: 0.71-2.25). However, increased risk of severe infections in the TNF-a inhibitor group was observed compared to the placebo group (OR: 1.66, CI: 1.25-2.23). Sensibility analysis with respect to each molecule did not change the main conclusion regarding the risk of death, but revealed that adalimumab had an increased risk of severe infection (OR: 2.10, CI: 1.28-3.46).
Conclusions This meta-analysis has been performed on a large number of patients and included the five TNF-a inhibitors currently available. It allowed an indirect comparison between the different molecules according to their medium-term safety. We did not find an increased risk of serious adverse events or deaths in patients treated with TNF-a inhibitors. However, we observed an increased risk of severe infections. Further studies aiming at the evaluation of the long-term safety are now needed to confirm these results.
Disclosure of Interest None declared