Dr Robert Califf

He said there was very little evidence to assess the actual balance of benefit (pain relief) and risk (cardiovascular/gastrointestinal events) with either the COX-2 inhibitors or the regular NSAIDs and questioned why this was the case. "Our system of drug development and postmarketing surveillance has failed to produce a solid estimate of the balance of risks and benefits for any patients taking chronic NSAIDs or coxibs. How can it be that, despite all the money spent on these drugs, we still can't tell the consumer what the balance of risks and benefits are for any of them?"

Califf pointed out that while cardiovascular disease may be the leading cause of death in the Western world, arthritis is the leading cause of disability. "Who among us does not have an aching joint? All of us have an interest in this." And NSAIDs will become more important as the population ages, he pointed out. "I can guarantee that almost everyone over 65 takes these drugs, and not just occasionally. But these drugs do cause GI bleeding, which is a real big problem."

How can it be that, despite all the money spent on these drugs, we still can't tell the consumer what the balance of risks and benefits are for any of them?

While the new COX-2-selective drugs were hoped to block the pain without causing bleeding and GI problems, they had another downside. "While preserving the gut, they appear to increase cardiovascular events—we are always dealing with a trade-off." Noting that all therapeutics represent a balance of risk and benefit, he suggested that the doctor's old maxim should be changed to from "Do no harm" to "On average, do more good than harm."

He said that all the panelists to whom he had spoken at the recent FDA meeting on the coxibs believed that cardiovascular adverse events were a class effect with the COX-2 agents and that the more COX-2-selective the agent, the greater the protective effect on the gut but also the greater the risk of cardiovascular events. "Cardiotoxicity is a class effect of the COX-2-selective drugs that may be dose dependent. It is likely that cardiotoxicity is proportional to COX-2 specificity. COX-2 inhibitors also cause GI toxicity, but this is probably inversely proportional to COX-2 specificity," Califf summarized.

Califf said that in general NSAIDs are probably worse than neutral for cardiovascular risk, but that naproxen is probably an exception, with a modest reduction in thrombotic events. Califf agrees with the recommendations made at the FDA last month that naproxen plus a proton pump inhibitor (PPI) should be the regimen of first choice now for patients needing an NSAID.

He said the National Institutes of Health (NIH) statement that it was terminating the trial on naproxen in Alzheimer's disease after discovering that the drug was associated with an increase in cardiovascular events vs placebo was "terribly misguided." "The results were preliminary, and only a few details of the trial have been released. The observation with naproxen was apparently statistically nonsignificant, and there are not enough data to make any recommendations," he commented.

Califf noted that there are an enormous number of people with acute or chronic pain, and as many of these people are elderly, "the same patients are at risk of GI ulceration from NSAIDs but may also have heart disease or are at risk of getting heart disease." So how to treat these people?

Califf made the following recommendations:

• Low-dose aspirin should be continued when indicated.
• Alternatives to NSAIDs should be considered, such as acetaminophen and topical therapies.
• If an NSAID is used, naproxen plus a proton pump inhibitor (to protect the gut) should be used first.
• COX-2-selective drugs should be avoided unless the above strategies fail.

Califf called for a large-scale study of COX-2-selective drugs that would have naproxen plus a proton pump inhibitor as a control arm and at least 10 000 patients in each arm, with a mix of aspirin takers and patients at high risk of heart disease enrolled. The primary end point should be one of net clinical benefit—taking into account pain relief, GI events, and cardiovascular events.

Other speakers at the COX-2 session were not sure whether a trial in patients at high risk of heart disease would be practical. In an interview with heartwire, Juni said he thought a large-scale trial was definitely necessary but asked: "Whom do we do the trial in? To get events we need patients who are likely to have a cardiovascular event, but these patients are going to be reluctant to enroll. I certainly would be."

We need patients who are likely to have a cardiovascular event, but these patients are going to be reluctant to enroll. I certainly would be.

Califf disagrees. "Since many patients with known coronary disease are already taking these drugs, I think we'd have no trouble with enrollment," he told heartwire.

But having recommended that an NSAID/PPI be the first-line choice and that coxibs should be reserved for patients who cannot take this combination or for whom it does not work, how can he recommend randomizing between an NSAID/PPI and a coxib in a clinical trial?

Califf commented: "Faced with uncertainty, we have to make recommendations using judgment and the data we have. The remarkable thing is that almost all experts recommended naproxen/PPI based on what we know now. But the uncertainty is so broad that we just don't know if the recommendation is right. So, if there is no randomized controlled trial to enter, go with the naproxen/PPI. If there is a randomized controlled trial, join it so we can resolve the uncertainty!"

Juni said he thought it was essential for patients enrolled in new trials of the coxibs to be carefully characterized. "It may always be a particular characteristic of a patient that is making them susceptible to these adverse effects. We must explore that. What we do not have is a large safety trial in the population of patients who actually take these drugs in reality. This is what we need."

Most of the speakers at the ACC session believed the coxibs should remain on the market. Califf said: "I think they should still be available for patients as long as they know the risks. My mother took Vioxx for her aching knees, as it was the only thing that worked. She ended up having a knee-replacement operation last week, which is not without risk, because Vioxx is no more."

Juni agrees. "I think the FDA will restrict COX-2 inhibitors to patients who can't take NSAIDs/PPI or in whom that combination is not effective and who don't have a cardiovascular risk." He added that while there is no scientific evidence to support the idea that COX-2 inhibitors are more effective than standard NSAIDs in reducing pain, there are many anecdotal reports from patients who say they are better. "I think it may be patient specific," he said.