Conference Coverage

Ankylosing spondylitis progression slowed when NSAIDs added to TNFi


Key clinical point: Ankylosing spondylitis patients have less progression on a TNFi when they also receive an NSAID, especially celecoxib.

Major finding: At 4 years, the mSASSS score was 4.69 points (P less than .0001) lower on TNFi with celecoxib than on TNFi alone.

Study details: Retrospective study with causal interference modeling.

Disclosures: The study was not industry funded. Dr. Gensler reported financial relationships with Amgen, AbbVie, Janssen, Eli Lilly, Novartis, and UCB.

Source: Gensler L et al. Ann Rheum Dis. 2018;77(Suppl 2):148. Abstract OP0198.



– When combined with a tumor necrosis factor inhibitor, NSAIDs provide protection against long-term radiographic progression in patients with ankylosing spondylitis, according to an analysis of more than 500 patients presented at the European Congress of Rheumatology.

Dr. Lianne S. Gensler, director of the Ankylosing Spondylitis Clinic at the University of California, San Francisco

Dr. Lianne S. Gensler

“The greatest effect is really in those patients using celecoxib and tissue necrosis factor inhibitors [TNFi],” reported Lianne S. Gensler, MD, director of the Ankylosing Spondylitis Clinic at the University of California, San Francisco.

Relative to TNFi alone, the addition of NSAIDs of any type provided protection at 4 years against radiographic progression as measured with the modified Stoke Ankylosing Spondylitis Spine Score (mSASSS). However, the protection associated with adding celecoxib was significant at 2 years and greater than that of adding nonselective NSAIDs at 4 years.

These data were drawn from 519 patients participating in the Prospective Study of Ankylosing Spondylitis study. All patients in this analysis were followed for at least 4 years. Radiographs were obtained every 6 months.

Although the study was a retrospective analysis of prospectively collected data, Dr. Gensler explained that control of variables such as disease and symptom duration with a technique called causal interference modeling “allows simulation of a randomized, controlled trial with observational data.”

Whether measured at 2 or 4 years, the reductions in mSASSS score for TNFi use versus no TNFi use were modest and did not reach statistical significance. However, exposure to NSAIDs plus TNFi did reach significance at 4 years, and the effect was dose dependent when patients taking a low-dose NSAID, defined as less than 50% of the index dose, were compared with those taking a higher dose. In this study, 70% were on chronic NSAID therapy, and these patients were divided relatively evenly between those on a low-dose or high-dose regimen.

At 2 years, relative radiographic protection for TNFi plus NSAIDs was not significantly greater than with TNFi alone, but at 4 years the median mSASSS score was 1.24 points lower (P less than .001) in those receiving low-dose NSAIDs, and 3.31 points lower (P less than .001) in those receiving high-dose NSAIDs.

In the subgroup of patients taking high-dose NSAIDs, the protection from progression was greatest among those receiving the selective COX2-inhibitor celecoxib. In these, the median 3.98 points lower mSASSS score (P less than .001) was already significant at 2 years. At 4 years, the median mSASSS score in those receiving TNFi plus celecoxib was 4.69 points lower (P less than .001).

Further evaluation suggested that the benefit from celecoxib plus TNFi was not just additive but synergistic, according to Dr. Gensler. She reported that neither TNFi nor celecoxib alone provided radiographic protection at 2 or 4 years.

Despite the modeling strategy employed to reduce the effect of bias, Dr. Gensler acknowledged that residual confounding is still possible. But she contended that “a large effect [from a such a variable] would be required to negate the findings.”

One of the messages from this study is that “not all NSAIDs are alike,” Dr. Gensler said. “Despite this, when I sit with a patient across from me, I will still treat the patient based on symptoms and disease activity first, though perhaps choose to be more NSAID selective if this is warranted and feasible.”

Next Article: