DUE 9/23 10 A.M EST BE ON TIMEQUALITY WORK, NO PLAGARISM1 PAGE NOT INCLUDING MIN 3 REF APAONCE AGAIN PUBLIC HEALTH LEADERSHIPAs one of the most widely-used prescriptions withdrawn from the market, R
n engl j med 351;17 www.nejm.org october 21, 2004 1707 PERSPECTIVE On May 21, 1999, Merck was granted approval by the Food and Drug Administration (FDA) to market rofecoxib (Vioxx). On September 30, 2004, after more than 80 million patients had taken this med- icine and annual sales had topped $2.5 billion, the company withdrew the drug because of an excess risk of myocardial infarctions and strokes. This rep- resents the largest prescription-drug withdrawal in history, but had the many warning signs along the way been heeded, such a debacle could have been prevented.
Neither of the two major forces in this five- and-a-half-year affair — neither Merck nor the FDA — fulfilled its responsibilities to the public. The pivotal trial for rofecoxib involved 8076 patients with rheumatoid arthritis and demonstrated that this coxib had lower gastrointestinal toxicity than naproxen. 1 Even though the drug was approved in 1999 on the basis of data submitted to the FDA, the data were not submitted to a peer-reviewed jour- nal until the following year and did not appear in print until November 23, 2000, one and a half years after commercial approval had been granted. The cardiovascular data reported in that article were incomplete, in part because of incomplete ascer- tainment: the design and execution of the trial had not anticipated that untoward cardiovascular events might occur. 1 It was not until February 8, 2001, that the FDA Arthritis Advisory Committee met to discuss con- cern about the potential cardiovascular risks asso- ciated with rofecoxib. It remains unclear why the FDA waited two years after its review and approval of rofecoxib to conduct this meeting. My colleagues and I reviewed the data from the meeting that were made publicly accessible and published an analy- sis of all the available data on rofecoxib and cele- coxib on August 22, 2001. 2 Our primary conclusion, based on the clear-cut excess number of myocar-dial infarctions associated with rofecoxib and the numerical, albeit not statistically significant, ex- cess associated with celecoxib, was that “it is man- datory to conduct a trial specifically assessing car- diovascular risk and benefit of these agents.” 2 Such a trial needed to be conducted in patients with es- tablished coronary artery disease, who frequently have coexisting osteoarthritis requiring medication and have the highest risk of further cardiovascular events. Given the very high coincidence of coro- nary disease and arthritis, this group may represent the largest segment of the population for whom rofecoxib was prescribed. In light of the insight that arterial inflammation is the basis for myocar- dial infarction and stroke and the knowledge that coxibs reduce the production of biomarkers of in- flammation such as C-reactive protein and improve endothelial function, such a trial would also have been quite attractive from the standpoint of po- tential benefit. The trial would have prospectively determined the incidence of cardiovascular events, whose possible association with coxib treatment had not been anticipated in the early and pivotal trials of these drugs.
Unfortunately, such a trial was never done. The FDA has the authority to mandate that a trial be conducted, but it never took the initiative. Instead of conducting such a trial at any point — and espe- cially after the FDA advisory committee meeting in 2001 — Merck issued a relentless series of publi- cations, beginning with a press release on May 22, 2001, entitled “Merck Reconfirms Favorable Car- diovascular Safety of Vioxx” and complemented by numerous papers in peer-reviewed medical litera- ture by Merck employees and their consultants. The company sponsored countless continuing medi- cal “education” symposiums at national meetings in an effort to debunk the concern about adverse car- diovascular effects. The message that was duly re- Failing the Public Health — Rofecoxib, Merck, and the FDA Eric J. Topol, M.D. n engl j med 351;17 www.nejm.org october 21, 2004 1708 PERSPECTIVE inforced was that rofecoxib had no cardiovascu- lar toxicity: rather, naproxen was cardioprotective.
Only by happenstance, in a trial involving 2600 pa- tients with colon polyps who could not have been enrolled if they had had any cardiovascular disease, was it discovered that 3.5 percent of the patients assigned to rofecoxib had myocardial infarction or stroke, as compared with 1.9 percent of the patients assigned to placebo (P<0.001), necessitating pre- mature cessation of the trial and the decision to discontinue treatment with rofecoxib.
Over the course of the five-and-a-half-year saga, many epidemiologic studies confirmed and am- plified the concern about the risk of myocardial in- farction and serious cardiovascular events associ- ated with rofecoxib. 3 These studies considered large populations, up to 1.4 million patients, track- ing the use of various nonsteroidal antiinflamma- tory medications or coxibs to determine the risk of adverse events. Each time a study was presented or published, there was a predictable and repetitive re- sponse from Merck, which claimed that the study was flawed and that only randomized, controlled tri- als were suitable for determining whether there was any risk. But if Merck would not initiate an ap- propriate trial and the FDA did not ask them to do so, how would the truth ever be known?
Meanwhile, Merck was spending more than $100 million per year in direct-to-consumer ad-vertising — another activity regulated by the FDA and a critical mechanism in building the “block- buster” status of a drug with annual sales of more than $1 billion. For the past few years, every month has seen more than 10 million prescriptions for ro- fecoxib written in the United States alone. At any point, the FDA could have stopped Merck from us- ing direct-to-consumer advertising, especially giv- en the background concern that the cardiovascular toxicity was real and was receiving considerable confirmation in multiple studies conducted by in- vestigators who were independent of Merck. The only significant action taken by the FDA occurred on April 11, 2002, when the agency instructed Merck to include certain precautions about cardiovascu- lar risks in its package insert. The FDA also spon- sored one of the large epidemiologic studies per- formed in a cohort of Kaiser Permanente patients.
Considering the tens of millions of patients who were taking rofecoxib, we are dealing with an enormous public health issue. Even a fraction of a percent excess in the rate of serious cardiovascu- lar events would translate into thousands of af- fected people. Given the finding in the colon-polyp trial in low-risk patients without known cardio- vascular disease — an excess of 16 myocardial in- farctions or strokes per 1000 patients — there may be tens of thousands of patients who have had major adverse events attributable to rofecoxib (see Figure).
I believe that there should be a full Congres- sional review of this case. The senior executives at Merck and the leadership at the FDA share respon- sibility for not having taken appropriate action and not recognizing that they are accountable for the public health. Sadly, it is clear to me that Merck’s commercial interest in rofecoxib sales exceeded its concern about the drug’s potential cardiovascular toxicity. Had the company not valued sales over safety, a suitable trial could have been initiated rap- idly at a fraction of the cost of Merck’s direct-to- consumer advertising campaign. Despite the best efforts of many investigators to conduct and pub- lish meaningful independent research concerning the cardiovascular toxicity of rofecoxib, only the FDA is given the authority to act. In my view, the FDA’s passive position of waiting for data to accrue is not acceptable, given the strong signals that there was a problem and the vast number of patients who were being exposed. Furthermore, the tradeoff here involved a drug for symptoms of arthritis, for which many alternative medications are available, in the Failing the Public Health — Rofecoxib, Merck, and the FDA Risk of Myocardial Infarction (MI) or Stroke Associated with Rofecoxib Use. Data are from Mukherjee et al. 2 and the Adenomatous Polyp Prevention on Vioxx (APPROVe) study. No. of Patients (log scale) 100,000 10,000,000 1,000,000 10,000 1,000 10 100 1 Risk in Colon Polyp Trial and VIGOR TrialPotential Risk in the Population 100,000,000 161,000160,00010,000,000 No. of patients with MI or strokeTotal patients exposed n engl j med 351;17 www.nejm.org october 21, 2004 1709 PERSPECTIVE context of serious, life-threatening cardiovascular complications. Certainly there are many facts that we are not privy to, such as the direct communica- tion between the FDA and Merck, but all the facts can and should be scrutinized closely in a Congres- sional review in order to avert such a catastrophe in the future. From the Cleveland Clinic Foundation, Cleveland. 1. Bombardier C, Laine L, Reicin A, et al. Comparison of upper gas- trointestinal toxicity of rofecoxib and naproxen in patients with rheumatoid arthritis. N Engl J Med 2000;343:1520-8. 2. Mukherjee DM, Nissen SE, Topol EJ. Risk of cardiovascular events associated with selective COX-2 inhibitors. JAMA 2001; 286:954-9. 3. Topol EJ, Falk GW. A coxib a day won’t keep the doctor away.
Lancet 2004;364:639-40. The coxibs are a subclass of nonsteroidal anti- inflammatory drugs (NSAIDs) designed to inhib- it selectively cyclooxygenase-2 (COX-2).
1 Their development was based on the hypothesis that COX-2 was the source of prostaglandins E 2 and I 2 , which mediate inflammation, and that cyclooxy- genase-1 (COX-1) was the source of the same pros- taglandins in gastric epithelium, where they afford cytoprotection. Three coxibs — celecoxib, rofecox- ib, and valdecoxib — have been approved for use by the Food and Drug Administration (FDA); a fourth, etoricoxib, has been approved by the Euro- pean regulatory authority, and it and a fifth, lumira- coxib, are currently under consideration for FDA approval.
Coxibs have been aggressively marketed directly to consumers in the United States and have rapid- ly dominated the prescription-drug market for NSAIDs, accounting for worldwide sales of roughly $10 billion. Rofecoxib has now been withdrawn from the market by Merck, following the prema- ture cessation, by the data and safety monitoring board, of the Adenomatous Polyp Prevention on Vioxx (APPROVe) study, which was designed to de- termine the drug’s effect on benign sporadic co- lonic adenomas. This action was taken because of a significant increase by a factor of 3.9 in the inci- dence of serious thromboembolic adverse events in the group receiving 25 mg of rofecoxib per day as compared with the placebo group. Blood pres- sure was elevated in patients in the rofecoxib group early in the course of the study, but the incidence of myocardial infarction and thrombotic stroke in the two groups began to diverge progressively after a year or more of treatment.
Coincident with the approval of rofecoxib andcelecoxib in 1999, my colleagues and I reported that both drugs suppressed the formation of pros- taglandin I 2 in healthy volunteers. 2 Prostaglandin I 2 had previously been shown to be the predomi- nant cyclooxygenase product in endothelium, in- hibiting platelet aggregation, causing vasodilata- tion, and preventing the proliferation of vascular smooth-muscle cells in vitro. However, it was as- sumed that prostaglandin I 2 was derived mainly from COX-1, the only cyclooxygenase species ex- pressed constitutively in endothelial cells. This as- sumption later proved incorrect, since studies in mice and humans showed that COX-2 was the dominant source. The individual cardiovascular effects of prostaglandin I 2 in vitro contrast with those of thromboxane A 2 , the major COX-1 prod- uct of platelets, which causes platelet aggregation, vasoconstriction, and vascular proliferation.
Whereas aspirin and traditional NSAIDs in- hibit both thromboxane A 2 and prostaglandin I 2 , the coxibs leave thromboxane A 2 generation unaf- fected, reflecting the absence of COX-2 in platelets.
Increasing laminar shear stress in vitro increases the expression of the gene for COX-2, leading our group to suggest that COX-2 might be hemody- namically induced in endothelial cells in vivo. If so, suppression of the COX-2–dependent formation of prostaglandin I 2 by the coxibs might predispose patients to myocardial infarction or thrombotic stroke.
Thus, a single mechanism, depression of pros- taglandin I 2 formation, might be expected to ele- vate blood pressure, accelerate atherogenesis, and predispose patients receiving coxibs to an exagger- ated thrombotic response to the rupture of an ath- erosclerotic plaque. The higher a patient’s intrin- Coxibs and Cardiovascular Disease Garret A. FitzGerald, M.D. Failing the Public Health — Rofecoxib, Merck, and the FDA
