London, UK - In the five-plus years that it was on the market, rofecoxib (Vioxx, Merck & Co) may have been responsible for 88 000 to 140 00 excess cases of serious coronary heart disease (CHD) in the US alone, and an estimated 44% of these cases may have been fatal, according to FDA official Dr David Graham and colleagues. These estimates appear in a study based on Kaiser Permanente data, published online January 25, 2005 in the Lancet.
Preliminary data from this study were first reported by Graham last August and appeared on the FDA website last November, but since then several revisions have been made "to correctly apply enrollment criteria for cases and controls," so the figures are now slightly different. This is also the study at the center of a controversy surrounding the FDA, with allegations that Graham's superiors at the agency tried to bully and discredit him, as previously reported by heart wire .Analyzed records of 1.4 million patients on NSAIDs
Graham et al conducted a nested case-control study using data from Kaiser Permanente, a national integrated managed-care organization in California. They assembled a cohort of 1.4 million patients who had taken nonsteroidal anti-inflammatory drugs (NSAIDs) between January 1999 and December 2001 and identified those taking rofecoxib (n=26 748), celecoxib (Celebrex, Pfizer) (n=40 405) or a nonselective NSAID, including naproxen (around 435 492) and nearly a million users of ibuprofen (n=991 261).
They classified the patients as current users if the NSAID prescription overlapped with the index date, remote users if the drug supply had ended more than 60 days before the index date, and recent users if the prescription ended between one and 60 days before the index date.
The study outcome was incident serious CHD, defined as acute myocardial infarction (AMI) requiring hospital admission or sudden cardiac death. The researchers identified 8143 cases (of which 1508 were sudden cardiac death). The cases admitted with AMI had laboratory confirmation in 87% of cases, and 702 (11%) of these patients died. Each case was age- and sex-matched with four controls.
Overall, rofecoxib raised the risk of serious CHD by 1.34 (p=0.066). Celecoxib slightly decreased the risk (odds ratio 0.84). The researchers comment that a similar diminished risk with celecoxib has been reported in other studies, but they also note that an increased CV risk has recently been reported from the Adenoma Prevention with Celebrex (APC) trial.Risk of acute MI with current use of NSAIDs compared with remote use
|| Adjusted odds ratio
|| 95% CI
When all current users of rofecoxib were compared with current users of celecoxib, risk was increased by 1.59 (p=0.015). For high-dose rofecoxib (>25 mg/day), the odds ratio was 3.58 (p=0.016), while for standard-dose rofecoxib <25 mg/day), the odds ratio was 1.47 (p=0.054). The researchers comment that "relatively few patients" were exposed to high-dose rofecoxib (10 cases, eight controls), but this finding accords with two previously published epidemiological studies that have analyzed higher-dose rofecoxib.
The risk of serious CHD was also increased with recent use of any NSAID, current use of naproxen, and current use of other NSAIDs, the researchers note. The latter is attributable to the effects of diclofenac (odds ratio 1.60, p=0.09) and indomethacin (odds ratio 1.30, p=0.01). The finding that risk was also increased with naproxen contrasts with that of manybut not allprevious studies, which have suggested that naproxen has a cardioprotective effect. In this study, compared with celecoxib use, current use of naproxen had an odds ratio of 1.36 (p=0.016).
Graham et al conclude that "the data from this study provide further evidence that rofecoxib increases the risk of serious CHD." The mean duration of use before an event was 113 days (range 4-688) for standard-dose rofecoxib and 112 days (range 8-262) for high-dose use (p=0.96), consistent with the idea that risk begins early in treatment.
In an accompanying Lancet commentary, Dr Simon Maxwell and Prof David Webb (Clinical Pharmacology Unit, University of Edinburgh, Scotland) write: "We can conclude that rofecoxib does have adverse cardiovascular effects, that these are dose-related and manifest in the early months of use, and that this effect could have been predicted with confidence three years ago. It is also apparent that any effect of naproxen on MI is small and certainly could not have explained the findings of the VIGOR study. In addition, we have further reassurance that these concerns may not extend to celecoxib, although confirmation about its safety and that of the newer COX-2 inhibitors is needed from further studies." They add that the available evidence is not "entirely reassuring," citing the APC finding for celecoxib as well as findings from CABG patients with valdecoxib (Bextra, Pfizer) and TARGET findings for lumiracoxib (Prexige, Novartis).Potential public-health implications
Graham et al's estimate of excess CHD caused by rofecoxib appears in the last paragraph of the discussion and is based on an extrapolation of clinical-trial findings and estimates of background rates from NSAID risk studies.
The researchers used IMS data to estimate that 106.7 million prescriptions for rofecoxib were dispensed in the US (17.6% for the high dose) in between the drug's launch in 1999 and withdrawal in September 2004. They note that two Merck-sponsored randomized clinical trials (VIGOR and APPROVE) found a relative risk for AMI of 5 for the high dose and 2 for the standard dose. In studies of the cardiovascular risk in NSAID users, the background rate for AMI among control groups has varied from 7.9 per 1000 person-years (in CLASS) to 12.4 per 1000 person years (in TennCare). Using all of these data, they calculate that 88 000 to 140 000 excess cases of serious CHD probably occurred in the US during the market life of rofecoxib. And they cite a US national estimate of a 44% case-fatality to suggest that many of these cases were fatal.
"In the future, when trials such as VIGOR show that a new treatment confers a greater risk of a serious adverse effect than a standard treatment, we must be much more careful about allowing its unrestrained use," Graham et al comment.
The editorialists also comment on the "important lessons learned" from the rofecoxib story. They agree with previous commentators that Merck and the FDA both failed the public and say that more notice should have been taken of basic clinical pharmacology and mechanism-of-action studies.
In an interview with heart wire , Maxwell expressed surprised that a study published in the Archives of Internal Medicine January 24, 2005 found no increased cardiovascular risk with the COX-2 inhibitors compared with nonnaproxen NSAIDs. He queries the exclusion of naproxen from the analysis, pointing out that it's a standard and commonly used anti-inflammatory and says the number of events in this observational study (in 6000+ patients) is "minute" in comparison with Graham et al's data base. "I don't think this outweighs prior evidence on rofecoxib," he says.
The commentators say the rofecoxib saga shows how difficult it is to predict the effect of a drug on large numbers at the time of launch and so has emphasized the potential value of pharmacoepidemiological research like that of Graham et al and others as a component of pharmacovigilance.
"Finally, the demise of rofecoxib should make us reconsider our attitudes to new technologies," they write, noting that "the hype surrounding the arrival of coxibs persuaded many prescribers." Advances in therapeutics are rarely so important for the health of the public that they should not be based on caution and regular review of the accumulating evidence, they say. "In retrospect it is hard to understand why there should have been such a rapid escalation in prescribing a drug that provided only a modest improvement in gastrotoxicity, had continued questions about adverse effects, and incurred greatly increased costs for healthcare providers."
Heartwire from Medscape © 2005
Cite this: Excess heart disease and deaths caused by rofecoxib, study suggests - Medscape - Jan 25, 2005.