Main

What we know about the safety of selective cyclooxygenase 2 (COX2) inhibitors is dwarfed by what we don't know — this was the message that emerged from the FDA's joint Arthritis Drugs/Drug Safety & Risk Management Advisory Committee meeting held in Gaithersburg, Maryland, USA, last month.

After 3 days of testimonies, deliberations and often heated exchanges the committees voted that the selective COX2 inhibitors celecoxib (Celebrex; Pfizer), rofecoxib (Vioxx; Merck) and valdecoxib (Bextra; Pfizer) show an increased risk of cardiovascular events and should have a black-box warning on the label —the strongest warning the FDA can give to a drug.

There was agreement that the increase in cardiovascular events was greatest in Vioxx and Bextra, and was only apparent in the highest doses of Celebrex. But assessing the true risk of these drugs is difficult. Trials have used different populations with different comparator drugs, and often lacked a placebo arm, said Robert Temple, Associate Director for Medical Policy at the FDA, in his presentation to the committees.

Temple raised the question of running a large, long-term trial, similar to the National Institutes of Health-funded ALLHAT study (Anti-hypertensive and Lipid-Lowering Treatment to Prevent Heart Attack Trial), which compared commonly prescribed classes of antihypertensives without a placebo arm.

The proposed trial would compare the cardiovascular and gastrointestinal risk in Celebrex with the traditional non-steroidal anti-inflammatory drugs (NSAIDs) ibuprofen, naproxen and diclofenac, which differ with respect to their degree of COX2 selectivity. Ideally, said Temple, aspirin plus a proton-pump inhibitor (PPI) would be included too.

Thomas Fleming, Professor of Biostatistics at the University of Washington and an FDA consultant, believes that this trial could and should be done. “This trial should be carried out in an osteoarthritis or rheumatoid arthritis setting and should enrol 10,000 patients per arm with 2–3 years follow-up,” he says.

With a placebo arm being unethical, as it would leave some patients without any pain medication, comparator consistency in these trials is crucial, says Steven Nissen, Vice-Chairman of the Department of Cardiovascular Medicine at the Cleveland Clinic.

Nissen's trial proposal would compare celecoxib with naproxen, and have a third arm of another traditional NSAID, such as diclofenac. “If it is true, as Garret FitzGerald believes, that diclofenac and celecoxib are very similar with respect to their cardiovascular profile, as they are with respect to their COX2 selectivity, then diclofenac is not a neutral comparator,” says Nissen, who was also on the Advisory Committees panel.

Alastair Wood, Professor of Medicine at Vanderbilt University, and Chair of the joint committee meeting says that more needs to be known about naproxen for naproxen-plus-PPI to become the standard comparator for trials. “First, we need a naproxen-plus-PPI versus placebo comparison to establish whether naproxen is or is not [cardio]protective,” says Wood.

But FitzGerald, Professor of Medicine and Pharmacology at the University of Pennsylvania, is against performing a large study of all NSAIDs at the taxpayer's expense. A black-box label for COX2 inhibitors will provide a motive to their manufacturers to perform safety studies, but the challenges are different with NSAIDs, he says.

“In the case of diclofenac, I believe the evidence exists to word a non-black-box precaution in a way that suggested that the precautions of selective COX2 inhibitors should probably apply,” says FitzGerald.

Comparative data indicate that either naproxen is cardioprotective and COX2 agents are neutral, or naproxen is neutral and selective COX2 agents have a higher cardiovascular risk. “Either way, naproxen looks better than a selective COX2 inhibitor,” says FitzGerald.

With efficacy and safety variations within the NSAIDs needing to be clearly defined, the whole NSAID class should have a warning to indicate a possible cardiovascular risk, says John Cush, Chief of Rheumatology and Clinical Immunology at the Presbyterian Hospital of Dallas, and member of the Arthritis Advisory committee. “Performing a trial against naproxen as comparator and demonstrating equal or better risk profile would allow the warning to be removed,” he says.

Although Cush supports the idea of the ALLHAT-like trial, he does have some concerns. “In my experience, patients frequently switch NSAID therapies in the first year, which would confound long-term analysis,” he says. “The trial design will be difficult and must include input from rheumatologists who specialize in such trials.”

figure z

Robert Temple at the FDA has proposed an ALLHAT-like trial to assess the risk of COX2 inhibitors. US Food and Drug Administration

figure y

Garret FitzGerald is against running a large, long-term trial to examine all NSAIDs. University of Pennsylvania