Background/Purpose

TNF inhibitors and combinations of conventional disease-modifying antirheumatic drugs are commonly added to treat methotrexate non-responsive rheumatoid arthritis patients. In the 48-week double blind, RACAT trial (NEJM 2013;369:307-18), 353 methotrexate suboptimal-responders were randomized to two treatment strategies, either first adding sulfasalazine and hydroxychloroquine, triple therapy (T, N=178), or etanercept (E, N=175) to methotrexate.  Participants without significant improvement in DAS28 at 24 weeks switched treatment while maintaining the blind.  We examined differences between T and E in the most common adverse events reported in the trial, gastrointestinal (GI) toxicity and infections, taking into account possible confounding factors.  

Methods

All adverse events during the trial were recorded and coded by blinded reviewers.  Serious (SAE) or non-serious (NAE) infectious and GI events reported during the intervention period and for 4 weeks after completing the intervention were included in the analysis.  The trial design posed challenges for analysis for 13 switcher patients who had infection (N=8) or GI (N=5) events occur both before and after switching treatment (either T to E or E to T).  For these switchers we included in the analysis only the infections that occurred when patients were currently on T and GI events that occurred when patients were currently on E. This is a conservative approach because infections were the most common side effect for E and GI events were for T.  We calculated the rate of event, total number of events, and the mean duration in the treatment when events occurred.  Logistic and linear regression models were used for treatment comparison with and without controlling for participant characteristics (age, sex, race, BMI, smoking) and comorbidities. 

Results

Participants who were on E were more likely to have infection NAEs (OR= 1.68, p=0.02) and had a higher total number of events (mean of 0.7 vs. 0.4, p=0.004) than participants who were on T. Participants who were on T had a higher number of GI NAEs than those on E (mean 0.55 vs. 0.34, p=0.02). Furthermore, the amount of time on the treatment when the GI events occurred was shorter for T than E (mean duration 10.0 vs. 17.7 weeks, p=0.001). The same conclusions remain after using regression models to control for patient characteristics and comorbidities.  Serious infections and GI events were rare for both treatments. However, there were a greater number of SAE infections that occurred when receiving E than T (12 vs. 4). Pneumonia was the most common SAE infection for both treatments (6 E and 2 T).

Conclusion

In RACAT, infectious NAEs were significantly increased in participants receiving E compared with T even after controlling for participant demographics and comorbidities.  GI NAEs were significantly more frequent in T and occurred earlier in the course of therapy.  Though numbers of SAE were small, there was a trend toward greater number of infectious SAEs when receiving E, which were predominately pneumonias. Our study findings might help clinicians when prescribing medications by considering patients’ tolerance of each treatment’s more common adverse events.

---

« Back to 2014 ACR/ARHP Annual Meeting

ACR Meeting Abstracts - https://acrabstracts.org/abstract/infections-and-gastrointestinal-side-effects-in-a-comparison-of-rheumatoid-arthritis-therapies/