Sulfasalazine for the Treatment of Ankylosing Spondylitis: Relic or Niche Medication?

The primary goals of treatment of ankylosing spondylitis (AS) are to reduce axial musculoskeletal pain and stiffness, control enthesitis, improve fatigue, and preserve flexibility and mobility. Inflammatory arthritis in peripheral joints, typically affecting five or fewer joints and occurring in up to 40% of patients, also requires treatment. Before the introduction of tumor necrosis factor-alpha (TNF) inhibitors, sulfasalazine was the slow-acting medication most often used to treat patients with AS whose symptoms were not adequately controlled with nonsteroidal anti-inflammatory medications, exercise, and rehabilitative therapies. Sulfasalazine was first used in AS in a 16-week open trial of eight patients with peripheral arthritis (1). Six of the eight patients responded, with improvement in symptoms and reductions in tender joint counts and acute phase reactants.

The efficacy of sulfasalazine in doses of 2 to 3 grams daily in improving the symptoms of AS was subsequently evaluated in several controlled trials. Pooled results of these trials indicated that sulfasalazine improved morning stiffness and decreased levels of acute phase reactants more than placebo, but there were no significant effects on other measures, including back pain, joint counts, global assessments, and physical functioning (2). However, its efficacy may be greater in patients with early AS and those with peripheral arthritis. For example, in a trial of patients with a mean duration of AS of 5.4 years, 68% of whom had peripheral arthritis, sulfasalazine was significantly better than placebo in improving pain, patient global assessments, morning stiffness, chest expansion, and the erythrocyte sedimentation rate (3). Other studies, however, indicated that the benefit of sulfasalazine was limited to improvements in peripheral arthritis itself, and not improvement in the axial symptoms of those who had peripheral arthritis (4). Other conventional disease-modifying medications, such as methotrexate, have not been tested in large clinical trials, but the lack of adoption of these medications in clinical practice suggests that an important benefit in AS has not been overlooked (5). Some evidence suggests sulfasalazine may reduce recurrences of uveitis among patients with AS, but there is no evidence that it prevents or slows the progression of spinal fusion or structural damage (6).

Perhaps because of its limited clinical effect on axial symptoms, selective targeting to the smaller subset of patients with peripheral arthritis, or its sometimes troublesome side effects and requirements for laboratory monitoring, sulfasalazine has not been widely used in the treatment of AS. Surveys in the U.S. in the 1990’s indicated that sulfasalazine was used by fewer than 20% of patients, even among those with moderately active or very active AS (5,7). In the Prospective Study of Outcomes in AS (PSOAS), a large observational study of patients enrolled at five U.S. rheumatology centers, 30% of 888 patients reported ever having been treated with sulfasalazine (unpublished observations). The proportion of patients actively treated with sulfasalazine decreased from 14% in 2004 to 3% in 2010. Use of sulfasalazine was similarly low in the Outcomes in AS International Study (OASIS) cohort from western Europe (8). This contrasts with the established efficacy and rapid adoption of TNF inhibitors to treat active AS. Interestingly, the initial studies of TNF inhibitors in AS also enrolled a high proportion of patients with peripheral arthritis, and those with peripheral arthritis may respond better to treatment with TNF inhibitors than patients with purely axial AS (9,10).

Against this backdrop, and as reported in this issue of Arthritis and Rheumatism, Braun and colleagues conducted the ASCEND trial, the first head-to-head comparison of sulfasalazine and a TNF inhibitor in the treatment of AS (11). The purpose of this randomized trial was to compare the short-term efficacy of sulfasalazine and etanercept in improving overall AS activity in patients with active AS. Noting that some professional organizations recommend that a trial of sulfasalazine be considered for patients with active peripheral arthritis before a TNF inhibitor is prescribed, patients with peripheral arthritis were targeted for enrollment, but this was not an inclusion criterion. Subjects were required to have a Bath AS Disease Activity Index of 30 or higher (on a 0 – 100 scale, and slightly more permissive than the conventional criterion of 40 or higher) despite treatment with nonsteroidal anti-inflammatory drugs, similarly-rated levels of morning stiffness, global assessment, back pain or functional limitations, and judged to be a suitable candidate for treatment with either medication. The trial compared etanercept 50 milligrams weekly with sulfasalazine 3 grams daily, using the double-dummy method to aid in blinding. The primary outcome was the proportion of subjects in each group that had an ASAS20 response, which assesses spinal pain, morning stiffness, functioning, and patient global assessment, after 16 weeks of treatment.

Most subjects had relatively early and active AS. Although 73% had peripheral joint symptoms, only 31% had peripheral synovitis at study entry. Seventy-six percent of subjects treated with etanercept had an ASAS20 response at 16 weeks, compared to 53% of those treated with sulfasalazine. A number of secondary endpoints, including decreases in the number of tender and swollen joints, physical functional limitations, and serum C-reactive protein levels, also favored etanercept. Few patients withdrew from the trial, and both medications were well tolerated.

This large, well-executed trial suggests that etanercept is more efficacious than sulfasalazine in the treatment of overall AS symptoms. The notable aspect of this study is that the trial enrolled patients with AS who were the most likely candidates to respond to sulfasalazine, thereby providing a more rigorous challenge to etanercept. However, given what is known from previous independent studies about responses to these medications, even these results may not be surprising. And although the presence of peripheral arthritis may predict better responses to sulfasalazine, it also predicts better responses to TNF inhibitors (10).

The strength of the conclusions that can be drawn from the ASCEND trial is tempered by certain aspects of the trial design. Sulfasalazine was started at a dose of 500 milligrams daily, and appropriately, escalated slowly so that the full dose of 3 grams daily was not reached until week 6. Given that the study endpoints were assessed 10 weeks later, one can ask if the duration of treatment with sulfasalazine at full dose was sufficiently long to judge its potential effect. Many would consider 4 to 6 months of treatment with sulfasalazine at 3 grams daily to be an adequate trial (12). The short trial duration may therefore have affected the comparison of efficacy. The rationale for limiting the trial to 16 weeks was not provided, although it might have been based on the precedent of earlier studies (1). The dose escalation schedule of sulfasalazine should also be considered when evaluating comparisons of study endpoints early in the trial.

Eighteen percent of subjects had previously been treated with sulfasalazine. Although prior ineffectiveness of sulfasalazine was an exclusion criterion, the comparisons would have been cleaner if subjects had been required to be naïve to both active treatments. Eleven percent of etanercept-treated subjects had injection site reactions that may have led to unblinding. We do not know if unblinding to sulfasalazine also occurred, due to its common gastrointerstinal and neurological side effects. ASAS20 responses among subjects in the etanercept arm were higher than projected, and higher than those seen in placebo-controlled trials of etanercept in AS, but comparable to those in other active comparator trials that studied etanercept. Had a placebo arm been included, the difference in response between etanercept and sulfasalazine might have narrowed, and the trial would have been able to provide additional information about the efficacy of sulfasalazine. However, even attending to these issues, it is not clear that equivalence would have been observed.

A much larger question relates to how we are to apply these results in clinical practice. The ASCEND trial was cast as an examination of treatment options for patients with AS and peripheral joint involvement despite nonsteroidal anti-inflammatory drug treatment, a group for which there is some evidence supporting the use of sulfasalazine. However, not all subjects had peripheral joint involvement, and the primary endpoint assessed axial symptoms, not peripheral joint manifestations. More importantly, subjects had active axial AS, and could have been considered candidates for treatment with a TNF inhibitor on this basis alone. While the ASCEND trial suggests greater short-term clinical benefit with etanercept, even in this group of patients that may be more sulfasalazine-responsive, this trial does not address two more clinically relevant questions about whether sulfasalazine fills a niche in strategies of AS treatment: 1) among patients with AS who have active peripheral joint manifestations but mild axial symptoms (that would not warrant treatment with TNF inhibitors), is sulfasalazine or a TNF inhibitor a better treatment? and 2) among patients with both active axial and peripheral joint manifestations of AS, is a strategy of sulfasalazine first, followed by a TNF inhibitor in sulfasalazine-failures, better than a strategy of treating with a TNF inhibitor from the start? Given the chronic nature of AS, outcomes in such a study would need to include not only symptom control but also toxicities and costs, likely examined over several years. These questions separate the evaluation of sulfasalazine effects on peripheral arthritis from its effects on overall AS in patients who also have peripheral manifestations, a distinction not always clearly made in the literature. The former question examines peripheral arthritis as the outcome, while the latter question examines peripheral arthritis as a predictor of response. These questions about the role of sulfasalazine in the treatment of patients with AS remain open.