SCOT: Good Risk/Benefit Trade-Off With Celecoxib, Other NSAIDs in Patients Without CV Disease

September 02, 2015

LONDON, UK — Both CV-event rates and prevalence of serious gastrointestinal complications were surprisingly low in people without heart disease taking either celecoxib (Celebrex, Pfizer) or traditional nonsteroidal anti-inflammatory agents (NSAIDs) for arthritis, say researchers from a randomized trial reported here at the European Society of Cardiology (ESC) 2015 Congress[1].

They assert that the findings should reassure physicians and patients that celecoxib, a selective inhibitor of cyclooxygenase 2 (COX-2) long dogged by evidence that it raises the risk of MI, at least in patients with CVD, is as safe for chronic therapy as non–COX-selective (ns) NSAIDs.

But some observers say the trial wasn't set up statistically to provide reliably strong results, at least enough to have an effect on clinical practice guidelines. Also, an early pullout by the sponsor due to observed parity between the two types of anti-inflammatory agent may have complicated interpretation of the trial.

Prof Thomas Macdonald

The SCOT trialists further say their findings argue against concerns that even the older, mainstay ns-NSAIDs—predominantly diclofenac and ibuprofen in their trial—pose an untoward cardiovascular hazard if appropriately used in the right population.

"In patients with arthritis, without known cardiovascular disease, cardiovascular-event rates were in fact very low and serious ulcer-related complication rates were almost vanishingly low, and neither outcome differed significantly between [ns]-NSAIDs and celecoxib," Prof Thomas M McDonald (University of Dundee, Scotland) said in a presentation to the media at the ESC Congress.

"Striking" and "Curious"

Prof Jose Lopez Sendon

It's been more than a decade since rofecoxib (Vioxx, Merck), a drug in the same selective COX-2 inhibitor class as celecoxib, was pulled from the market after highly publicized reports of cardiovascular events in patients taking the drug. Celecoxib and other coxibs remained available, but hopes had been that both it and rofecoxib would be appreciated for posing less risk of gastrointestinal ulcers and GI symptoms than nonselective NSAIDs. But it has been unclear whether coxibs have even that advantage over the older, less expensive agents.

Prof Jose Lopez-Sendon (Hospital Universitario la Paz, Madrid, Spain), assigned discussant after MacDonald's formal presentation of SCOT at the ESC Congress, agreed that the trial confirmed its hypothesis that the coxib would be noninferior to the ns-NSAIDs, but that the low event rate was "striking" and "curious" and perhaps a sign that the patient population was at low CV risk.

"With no difference, really, between the two groups of treatment, I don't think clinical practice will change a lot. Probably the recommendations in the guidelines will follow the same as they are: use the lowest effective dose for the shortest period of time," Lopez-Sendon said. "And only in no-risk patients, because these drugs are still contraindicated in patients with cardiovascular disease."

Dr Scott Solomon

Dr Scott Solomon (Brigham and Women's Hospital, Boston, MA), a panelist discussing the trial following Lopez-Sendon, pointed out that "we did the initial cardiovascular safety review for celecoxib that established the risk against placebo. And we found that this risk was really mostly apparent in patients at high risk for cardiovascular events, not at low risk. So you might have seen something different had you gone to a higher-risk population."

"Not a blip of a signal"

The Standard Care vs Celecoxib Outcome Trial (SCOT), which McDonald presented to the broader congress later in the day, had randomized 7297 patients on ns-NSAIDS for osteoarthritis or rheumatoid arthritis to either continue on those agents or be switched to celecoxib by prescription from their primary-care physicians, who oversaw their management as usual. The cohort was followed for a mean of 3.2 years exclusively through review and adjudication of events documented in hospital records and other databases, as described by MacDonald.

Importantly, none of the entered patients could have "established cardiovascular disease"; otherwise, celecoxib would have been contraindicated. But many had risk factors. In both the celecoxib and ns-NSAID groups, the mean age was about 68 and body-mass index was about 29; almost half were hypertensive, a third had elevated cholesterol, and a fifth were on statins. About 38% of patients were on diclofenac and 31% were on ibuprofen for their arthritis, while 38% were on "ulcer-healing therapy."

MacDonald said many of the patients would be seen by cardiologists as high risk for CV disease. They might have been expected to have a 2% to 3% annual risk of cardiovascular events during the trial period, "but that was not the case." The patients went on and off their celecoxib or ns-NSAID regimens as any might in routine practice, so both intention-to-treat (ITT) and on-treatment analyses were performed, with lower-than-expected results in both cases, 1.1% and 0.9%, respectively.

Indeed, in noninferiority analysis, he said, there was "not a blip of a signal" for raised cardiovascular risk in the population studied among those getting celecoxib compared with ns-NSAIDs for the primary clinical-outcomes end point.

SCOT: Hazard Ratio (95% CI) for Primary and Secondary Outcomes, Celecoxib vs Nonselective NSAIDs

Analysis HR (95% CI) P
Primary outcome on treatment 1.12 (0.81–1.55) 0.50
Primary outcome ITT 1.04 (0.81-1.33) 0.75
All-cause mortality on treatment 1.20 (0.76-1.88) 0.43
All-cause mortality ITT 0.92 (0.70-1.21) 0.56
Primary end point=hospitalization for nonfatal MI, nonfatal stroke, or cardiovascular death (adjudicated)
ITT=intention to treat

"There were only 12 [on-treatment] ulcer-related upper gastrointestinal complications in the whole study. And by intent to treat only 15. We were as astonished as anyone else. This is a reflection of what is happening in the real world," MacDonald said.

"In this study population, celecoxib in my view appeared safe, and [nonselective] NSAIDs appeared safe. And in patients who get significant symptomatic relief—that is, they get benefit—in my view, from these medicines, the benefit/risk balance appears positive."

A Margin Call

SCOT has some limitations, some arguably significant, that call for caution in interpreting its outcome. Randomization wasn't double-blind, of course, and even through his optimism about the results, MacDonald acknowledged that "the very low event rates made this trial somewhat difficult to complete."

At his formal presentation on SCOT to the ESC Congress, MacDonald said, "It would have been good if this trial had had more patients and follow-up for longer. Unfortunately, Pfizer decided that this trial was futile and persuaded the European Medicines Agency to release it from its regulatory commitment. The investigators didn't agree with that, and we had to continue the trial with rather parsimonious funding."

And the protocol powered the analysis for noninferiority to exclude a 40% increase in cardiovascular events in patients in either group, which required 277 on-treatment events to fulfill, he noted.

Dr Paul Armstrong

Dr Paul Armstrong (University of Alberta, Edmonton) observed that "out in clinical practice and with registries we get information that we don't get in trials. In the real world, patients have comorbidities." So are the SCOT findings as reassuring as the trialists assert, regarding the relative and absolute CV safety of celecoxib and ns-NSAIDs? Armstrong told heartwire from Medscape that the issues explored in the SCOT trial are "worth looking at," but that "a 40% margin of noninferiority does not provide me much reassurance."

But Dr Lars Ryden (Karolinska Institute, Stockholm, Sweden) said he thinks SCOT is indeed "reassuring." Speaking with heartwire , Ryden said he thinks the trial was carefully conducted, and though one can challenge its conclusions, "Perhaps these drugs are not as dangerous as we believe."

Events vs Reactions

MacDonald reported that the rates of "serious adverse events"—that is, safety end points requiring hospitalization that were documented without regard to whether they were treatment related—were essentially equal, at 31.7% for the celecoxib group and 32.4% for the ns-NSAID group. He and his colleagues also looked at more amorphous end points they called "adverse reactions," which they defined as events or signs considered related to the drug treatment. They included:

  • "Adverse reactions" in 22 on celecoxib and 16.1% on ns-NSAIDs (P<0.001).

  • "Serious adverse reactions" in 5.2% and 5.8%, respectively.

There were very few "serious GI adverse reactions," according to MacDonald; still, they were significantly different, at 38 events from the coxib and 66 events from the ns-NSAIDs (P<0.007).

Solomon, from the panel, was interested in this comparison, "because as far as I can tell, that is the only reason that I would choose celecoxib over a standard, cheaper nonsteroidal for these kinds of patients."

A total of 50.9% of those assigned celecoxib and 30.2% in the ns-NSAID group withdrew from their assigned treatment (P<0.001). The most common reasons were perceived lack of efficacy (11.2% for celecoxib group and 3.0% for the ns-NSAID group, respectively); "switch or stopped" the drug (6.6% and 8.1%); adverse events (8.3% and 4.4%); doctor recommendation but not due to adverse events (4.7% and 6.0%); and "patient requested" (6.0% and 2.3%).

About 4% of celecoxib patients withdrew because the agent was "not tolerated," more than triple the rate for ns-NSAIDs.

MacDonald said the high rate of withdrawal from celecoxib was an aspect of the real-world setting. "It was done when there was a lot of negative publicity [about coxibs]. That's the first thing. The second thing is it was an open trial. Patients knew they were getting it, and certainly patients who have already been on chronic NSAID therapy and then were switched might be less happy to keep on the switched drug." And that, he acknowledged, "resulted in some lack of power in the on-treatment analysis."

According to the ESC, MacDonald had part of his university salary reimbursed to his institution by Pfizer, which initially sponsored the trial, and discloses receiving honoraria from and consulting for NiCox, Novartis, and AstraZeneca for research related to NSAIDs. Dundee University reports holding research grants from Pfizer, Menarini, Novartis, and Amgen. Armstrong has previously disclosed receiving research grants from Boehringer Ingelheim, Sanofi, Merck, AstraZeneca, Eli Lilly, GlaxoSmithKline, and Regado Biosciences; consulting for Regado Biosciences; and being on safety and data monitoring boards for Roche, Orexigen, and Eli Lilly. Solomon has served as an advisor to Bayer HealthCare.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.