Background/Purpose:

The objectives of this systematic review were to study and update the safety of anti-TNF agents. We examined whether etanercept (ETN), compared to the anti-TNF antibody therapies, infliximab (INF) and adalimumab (ADA), had an inverse impact on adult patients with rheumatoid arthritis (RA) in terms of malignancy, serious adverse events (SAEs), serious infection, and discontinuation due to adverse events (AEs).

Methods:

We conducted a systematic-literature review of randomized controlled trials (RCTs) that studied one of the three biologics used for rheumatoid arthritis and reported on our pre-specified adverse outcomes of malignancy, SAEs, serious infections, and discontinuation due to AEs.  We searched various databases including MEDLINE® via OVID and PubMed®, the Cochrane Library, Google Scholar, and ClinicalTrials.gov, and further mined the reference lists from systematic reviews and original publications to identify all English-language studies published from January 1, 1990 until September 30, 2011.  In addition, we searched the US Food and Drug Administration (FDA) database to review drug approval reports that could provide eligible trials.  The search strategy and data extraction processes were duplicated by independent reviewers.  For the meta-analysis, we performed random effect inverse variance, maximum-likelihood estimation (MLE), arcsine transformed, and Bayesian models. We abstracted the studies with 0 events in both arms and used software default correction coefficients for 0 events or missing data. Furthermore, we compared the results from randomized trials with published large nationally representative cohort studies and administrative databases.

Results:

Thirty-five trials met our inclusion criteria, including 5,524 patients who received anti-TNF biologic agent treatment and 3,257 patients who received MTX/placebo.  The risk of malignancy in patients treated with INF was significantly higher than that among those treated with MTX (risk difference=0.02 [95% CI, 0.00 to 0.05]) in the arcsine transformed model.  The risk of serious infections in patients treated with ADA or INF was significantly higher than that among those treated with MTX or placebo (OR=7.8 and 2.07; ARD=0.03 and 0.03), both in the Bayesian and arcsine transformed models (Table).  In contrast, patients treated with ETN tended to have non-significant lower risks of SAEs, serious infections, and discontinuation (ORs=0.84, 0.71, and 0.84, respectively) (Table).

Conclusion:

The findings of this meta-analysis suggest potential differences in adverse outcomes among ADA-, INF-, and ETA-treated RA patients.  The risk of malignancy may be increased among INF-treated RA patients, whereas the risk of serious infection may be increased ADA- or INF-treated patients.  These findings call for further post-marketing surveillance to clarify these risks with longer term exposure.