The CV risk with coxibs: is it really proven, how big is it, and are traditional NSAIDs any safer in this regard?
The increased cardiovascular risk with rofecoxib (Vioxx, Merck & Co) that led to this drug’s withdrawal now appears to be a class effect of the COX-2 inhibitors, but as the data are analyzed, they are throwing up more questions than answers. Some question whether the CV risk has really been proven, others how large this risk really is, while others still point out that some of the traditional nonsteroidal anti-inflammatory drugs (NSAIDs) also have a CV signal and so may also carry an increased CV risk.
One consequence of the growing belief that the CV risk is a class effect of the coxibs is Merck considering the return of rofecoxib to the market. This proposal is “astonishing” and “a mockery to the international medical community,” says Dr Osvaldo Messina (Dr Cosme Argerich Hospital, Buenos Aires, Argentina), commenting in the email forum of www.jointandbone. All of these COX-2 inhibitors share the same mechanism of action and the hence same increased cardiovascular risk, and so “we should be thinking about taking them all off the market,” he argues.The COX-2-selective inhibitors allow the effects of COX-1 to predominate, creating clinical conditions that lead to a trend to arterial thrombosis (vasoconstriction, platelet aggregation, and endothelial proliferation), Messina explains. “This is the biochemical reality,” he tells rheumawire. There are strong interests other than medical at work here, he says, and proposing that rofecoxib should return is saying that “1 more member of a questionable pharmaceutical class should be marketed because there are others already being sold.”But has the increased CV risk of rofecoxib really been proven? Dr Jeffrey Mann, a retired physician from Salt Lake City, UT, suggests that it hasn’t. A former emergency-medicine doctor, he has a particular interest in clinical trials and runs a medical education website on which he publishes detailed critiques of clinical trials, including a recent 1 about the Adenomatous Polyp Prevention on Vioxx (APPROVE) study. This showed nearly a doubling of CV risk for rofecoxib compared with placebo. However, Mann points out that the increase in CV risk was seen only 18 months into the trial and “primarily reflects the fact that the placebo group had an unexpectedly low control event rate during the last 18 months of the trial.” The Kaplan-Meier curves for cumulative adverse CV events in the placebo group show a dramatic flattening in the second half of the study, he points out; if it had remained similar to what it had been in the first half of the study, the difference between rofecoxib and placebo would probably have been deemed not to be clinically significant. “I cannot think of any intrinsic reason why the rate of adverse events should vary markedly from month to month other than due to the play of chance.”A problem with this trial, and many other clinical studies, is the heterogeneity of patients when the control event rate is low. “Many clinical researchers believe that randomization will ensure that there is a balance of patients in the control and treatment group, but common sense tells us this is not necessarily true,” Mann commented in an interview with rheumawire. Normally, a slight difference between the 2 groups in pattern distribution of baseline prognostic variables should not affect a study’s results, but when the control event rate is low, it can have a significant confounding effect, he explains. However, in the case of the APPROVE study, the most likely explanation is unrelated to any pattern-distribution imbalance: the placebo patients had a low risk of an adverse event (around 2%) over the 36-month study time period, and it was “pure chance” that very few of those events occurred during the second half of the study, he says. If the patient in question has a 1% chance of a heart attack, and rofecoxib raises this to 2%, well, many patients would be prepared to accept this in exchange for the benefits they derive from the drug. Hence, Mann suggests that the conclusion that rofecoxib increases CV risk is “scientifically invalid” and says he was “quite amazed” that this 1 finding led to the drug’s withdrawal. In his opinion, it is “foolish to definitively state that rofecoxib increases the risk of adverse CV events based on a single trial that has a low signal/noise ratio.” Mann expresses similar disquiet about the conclusions of the Adenoma Prevention with Celecoxib (APC) trialthe only study so far to show an increased CV risk with celecoxib (Celebrex, Pfizer). The APC trial also has a low control event rate, and it is difficult for randomized trials that recruit low-risk individuals to generate a strong signal, he comments. Science demands repeatable results, and the fact that the findings of both the APPROVE and APC studies differ from that of many other trials is in itself a cause for suspicion, he adds.Allow patients to choose what risks to take Also, the increased CV risk should be placed into context, says Mann. If rofecoxib does double the risk of a CV eventwhich Mann emphasizes isn’t proventhen whether or not this is clinically significant depends on “one’s mind-set. . . . If the patient in question has a 1% chance of a heart attack, and rofecoxib raises this to 2%, well, many patients would be prepared to accept this in exchange for the benefits they derive from the drug, especially if it’s put to them that they have a 98% chance of not having a heart attack, as opposed to a 99% chance.” And there are many other factors that need to be taken into account, he adds; for example, smoking is believed to double CV risk. “My patients have the right to accept whatever risks they choose in light of the individual benefits and their own informed consent,” says Dr Kenneth Hardy (University of Mississippi Medical Center, Jackson). He argues strongly in favor of bringing back Vioxx in the email forum of www.jointandbone.org and describes patients now crippled with arthritis who would be willing to take the risk to gain the benefit of this drug. Hardy objects to what he sees as the hijacking of this issue by an out-of-control media and legal system and a result as well of paranoia and loathing of industry wealth and power, all of which “have clouded our otherwise good clinical judgment.” He says: “Rheumatologists are among the most thorough of all internists” and should be left “the hell alone to deal with the medicines and the patients we know best.”Hysteria is unfair to physicians and patients We have scraped the bottom of the statistical barrel looking for meaningful information and received instead controversial, variable, and highly sensationalized data of minimal practical value to anyone other than trial attorneys and the media, “The reason for all this coxib/NSAID/CV-risk hysteria is that we are all unwittingly wallowing in confusion at the bottom of a statistical barrel,” says Hardy. “The CV risks, while admittedly real, are so small that earlier studies simply couldn’t statistically and/or reliably detect them at all. In fact, everyone currently agrees that large, prospective trials with multiple cohorts over many years will be needed before we can truly understand whatif anydifferences in CV risk exist between specific (rofecoxib, valdecoxib), selective (meloxicam, celecoxib, etodolac, etc), and nonspecific (ie, traditional) NSAIDs. The risks and differences are simply too small to be reliably interpreted from our available clinical trials.” Hardy makes the point that naproxen has come out showing different effects in different studiessome show it to have a decreased CV risk, but some show it to increase the CV risk, and some show no effect at all. “We have scraped the bottom of the statistical barrel looking for meaningful information and received instead controversial, variable, and highly sensationalized data of minimal practical value to anyone other than trial attorneys and the media,” says Hardy. This is “grossly inappropriate and unfair to physicians and their patients,” he argues. “Huge numbers of patients have benefited enormously, and safely, especially from the coxibs. . . . Armed with practical information, patients and their physicians should be the ones to discuss risk/benefit on an individualized basis.” Does problem extend to all NSAIDs? One of the fears of practicing clinicians is that the publicity about the “dangers” of the increased CV risk with the coxibs will lead doctors and patients to older drugs that do not have any better safety records and that in fact may be worse. This is already being advocated in some areasas reported by rheumawire, New Zealand recently ruled that the risks of COX-2 inhibitors outweigh the benefits in the general population and advised using “alternative therapeutic options” while listing available traditional NSAIDs (including diclofenac, ibuprofen, sulindac, tiaprofenic acid, ketoprofen, naproxen, tenoxicam, and piroxicam).But the safety of many of these older drugs hasn’t been thoroughly documented, a point that came out strongly at the recent US FDA hearing on the safety of NSAIDs. “We are worried about some of the other NSAIDs, and I hope that message came through,” says 1 of the FDA advisory panel members, Dr Steven Nissen (Cleveland Clinic Cardiovascular Coordinating Center, OH). Some of the older drugs are also relatively COX-2 selective, he points out, and “we aren’t so sure that agents like diclofenac and meloxicam don’t have the same cardiovascular risks. And we’re not so sure about ibuprofen, either.” His comments appear in a report of the FDA hearing on AMednews.com, and he adds: “My concern is that if we are going to push people to a different pattern of use, that pattern ought to reduce risk, but we don’t really know if these drugs are going to reduce the risk of cardiovascular complications.” The same point is highlighted in the latest hotline on the safety of COX-2 inhibitors from the American College of Rheumatology (ACR), which summarizes deliberations at the FDA hearing. As already reported by rheumawire, the ACR comments that the data on CV risks of traditional NSAIDs are “incomplete, but concerning.” It notes that trials of celecoxib using diclofenac and ibuprofen as comparators fielded comparable rates of CV events but adds that this “does not confirm the safety of the coxib, as these nonselective NSAIDs may also share an increased CV risk.” Unpublished observational studies reported at the meeting by Dr David Graham using 2 large US databases suggest there is an increased CV risk for many NSAIDs (indomethacin, meloxicam, sulindac), and preliminary data from a Norwegian study (Sudbo et al) showed increased CV risks with traditional NSAIDs, with CV hazard ratios ranging from 1.70 for naproxen to 2.86 for ibuprofen.The mechanism behind the increased CV risk is not entirely clear, comments Dr Andrew Laster, a rheumatologist in private practice in Charlotte, NC and an editorial consultant to www.jointandbone.org. “We have assumed until now that COX-2 drugs are implicated because they block prostacyclin and allow thromboxane to work unopposed,” he comments. However, aspirin is a selective COX-1 inhibitor, and yet a subset analysis of the coxib trials that have shown a CV signal indicates that the increased CV risk occurred both in patients who were taking aspirin and in those who weren’t. “So if aspirin was unable to block this response, does this mean that an alternate mechanism is in effect? Is this why some NSAIDs that are nonselective may also be associated with increased CV risk?”So if aspirin was unable to block this response, does this mean that an alternate mechanism is in effect? Dr David Pisetsky (Duke University, Durham, NC), an editorial consultant to www.jointandbone.org, says that there are several important issues that need to be considered. He points out that the trials showing an increased CV risk with the coxibs were long-duration cancer-prevention studies where patients were taking the drug every day and asks whether it is reasonable to extrapolate from them to the arthritis population and the more common situation where people take NSAIDs periodically1 estimate suggests for 60 days per year. “I agree that all drugs have side effects and that patients need to evaluate for themselves the relative risks and benefits,” he says. “And like many physicians, I have patients who claim unique benefits from 1 or another NSAID or coxib.” But while alternatives should be available for these individuals, this need should not equate with widespread use of a class of drugs that potentially has more side effects, he adds. However, he also wonders whether by now the media and direct-to-consumer advertising have clouded the discussion to such an extent that a reasonable middle ground will be hard to achieve. (Source: Joint & Bone: March 2005.)