SNOWMASS, COLO. – A recent, much-ballyhooed, large, observational U.S. study that revised downward the risk of serious infections posed by tumor necrosis factor–alpha inhibitor therapy may have rendered false comfort to prescribing physicians.
The findings of the U.S. multicenter Safety Assessment of Biologic Therapy study (SABER) were roundly celebrated by many rheumatologists, dermatologists, and gastroenterologists who prescribe biologic agents for autoimmune diseases. But the SABER results are strikingly out of step with those of many large, well-conducted patient registries.
Dr. Kevin L. Winthrop
Moreover, at least one SABER organizer is skeptical about the validity of the study’s key conclusion that the serious infection risk in patients who’ve started on anti-TNF therapy isn’t significantly different than the risk in those started on methotrexate or another nonbiologic disease-modifying antirheumatic drug (DMARD). He cautioned against taking that conclusion as a lesson from SABER.
"Actually, my belief is that most of our findings in SABER are explained by the fact that anti-TNF agents are prednisone-sparing drugs," Dr. Kevin L. Winthrop said at the symposium, which was sponsored by the American College of Rheumatology. "My belief is that anti-TNF drugs do increase the risk of serious bacterial infections. The fact that they allow people to decrease their prednisone may result in roughly a canceling out of that risk."
SABER was a retrospective cohort study combining data from four large U.S. automated health databases. Investigators were able to compare outcomes in 16,022 patients with rheumatoid arthritis (RA), psoriasis, psoriatic arthritis, ankylosing spondylitis, or inflammatory bowel disease (IBD) who were placed on an anti-TNF biologic vs. an equal number of matched patients who were started on a conventional DMARD.
The rate of serious infections requiring hospitalization during the first year after starting therapy – while impressively high – didn’t differ significantly between the two groups: roughly 8% per year in the RA patients regardless of whether they were on anti-TNF therapy or a conventional DMARD, 5.4% in those with psoriasis or spondyloarthropathies, and 10% in patients with IBD (JAMA 2011;306:2331-9).
But SABER had a major limitation: Data on prednisone use were collected for the year before the new therapy was introduced, but not after, noted Dr. Winthrop, an infectious disease specialist at Oregon Health and Science University, Portland.
"When you put someone on a biologic, you do a lot of things for that person differently than when you put someone on, say, Plaquenil [hydroxychloroquine]. You screen them for tuberculosis, you may counsel them, you give them vaccines, hopefully you decrease their prednisone, and you may even take off some methotrexate if they have good disease control. All of that probably happens to a much greater extent in the biologic arm than the nonbiologic arm," he explained.
Although infliximab was associated with a 23%-26% higher risk of serious infection compared with adalimumab or etanercept in SABER participants with RA, patients on infliximab were also more likely to concurrently be on methotrexate. "That might have been enough to explain this risk difference," according to Dr. Winthrop.
Unlike the SABER findings, a significantly elevated serious infection risk in association with RA patients receiving anti-TNF agents has been reported from the British Society for Rheumatology registry (Arthritis Rheum. 2006;54:2368-76), from a Japanese prospective registry (J. Rheumatol. 2011;38:1258-64), from the CORONNA (Consortium of Rheumatology Researchers of North American) registry (Ann. Rheum. Dis. 2010;69:380-6), and from the German RABBIT registry (Ann. Rheum. Dis. 2011;70:1914-20).
These studies generally showed a time-dependent decline in infection risk in patients treated with anti-TNF agents. But physicians shouldn’t assume that this apparent drop off in risk over time is a real phenomenon; rather, it is largely due to what epidemiologists call survivor bias. In other words, people who develop serious infections or other problems early on drop out of treatment, so the case mix in the registries changes, he explained.
Dr. Winthrop was particularly impressed with what he termed the "very elegant" 5,044-patient RABBIT study, which was designed to be able to answer the question SABER couldn’t – namely, what happens to prednisone dosing when patients go on a TNF inhibitor? The answer, as was well documented in RABBIT, is that the use of prednisone drastically decreases.
Moreover, the German investigators, using a multivariate analysis, identified a set of independent risk factors for serious infection. These included treatment with a TNF inhibitor, which was associated with a 1.8-fold increased risk, compared with conventional DMARDS; treatment with glucocorticoids at 7.5-14 mg/day during the study period, which had a 2.1-fold risk; and treatment with steroids at 15 mg/day or more, which conferred a 4.7-fold increased risk of infection. Other studies have also demonstrated that glucocorticoids convey a high and dose-dependent risk of serious infection.