Just as it appeared that the dust from the great COX-2 controversies had begun to settle, a jury in Texas awarded a plaintiff a judgment against Merck & Co for one quarter of a billion dollars – indeed, for a death that, arguably, was not Vioxx-related.  Many of us are wondering exactly how the situation has reached such an extraordinary moment. The events that followed the voluntary withdrawal of Vioxx (rofecoxib) by Merck & Co on September 30, 2004, have included a contentious three-day US Food and Drug Administration (FDA) advisory hearing,  the addition of sweeping black box warnings on selective and non-selective cyclooxygenase inhibitors and the withdrawal of Bextra (valdecoxib). Together the events have had a chilling effect on the prescription of NSAIDs and the remaining coxib, Celebrex (celecoxib) and, most importantly, the events have resulted in fewer available effective therapies for patients with osteoarthritis (OA) and other pain syndromes.

What is remarkable in contemplating the post-Vioxx withdrawal period is the rapid widespread and intense reaction, even though we, as physicians, have long appreciated that nonsteroidal anti-inflammatory drugs (NSAIDs)  can increase the risk of hypertension, heart failure and renal insufficiency – and even though, the data on NSAID-conferred cardiovascular (CV) risk, while compelling, are complex. Yet, the series of events that led to the withdrawal of Vioxx and the subsequent aftershocks had actually been stirring under the surface for years. So what made them erupt? Why did the reaction to the Adenomatous Polyp Prevention on Vioxx (APPROVe) study suddenly have dramatic, cascading effects that led to such a backlash against not only COX-2 inhibitors, but also against the pharmaceutical industry, the FDA and physicians as well?

While, at the end of the day, the FDA recommendations and the decreased utilization of anti-inflammatory drugs are understandable given the concern and uncertainty, the question that must be considered is what were the forces at work that led these changes to sweep across continents with brush-fire rapidity?

In thinking about the past year, one is reminded of the phenomenon designated the "tipping point" by Malcolm Gladwell in the book "The Tipping Point: How little Things Can Make a Big Difference." Gladwell describes how an idea or message – here a headline such as "Vioxx Causes Heart Attacks" – can spread in epidemic-like fashion.  The COX-2 controversy exhibited such a "tipping point" and the elements are worth considering as we try to learn the lessons of the past several years.

A Context of Distrust

Referring to the power of context, Gladwell notes that social epidemics reflect the larger circumstances of the times. In this case, a synchrony of events occurred during the past decade that undermined the public trust in the institutions responsible for healthcare and its oversight.  For those who chose to see a conspiracy in controversy, there were several candidate co-conspirators who were to be held accountable.  The pharmaceutical industry was already under fire because of the high cost of drugs to American consumers, and a perception by some that profit margins were excessive in the context of a healthcare system, which left select drugs unaffordable to the underinsured.  Direct-to-consumer (DTC) marketing increased the visibility of the pharmaceutical industry, and led to criticism regarding the overly aggressive marketing of drugs of marginal added value and understated risk. Of note, COX-2 drugs were marketed soon after the FDA reduced decades of restrictions on DTC advertising (triggered in part by DTC legislation enacted in 1997), and the COX-2 phenomenon was, in part, due to increased consumer demand for these prescription drugs. DTC advertising of Vioxx and Celebrex helped catapult worldwide sales of the drugs into the billions of dollars. In 2002, 61% of anti-inflammatory prescriptions were for a COX-2 inhibitor, up from 35% in 1999.

But the pharmaceutical industry was not the only perceived co-conspirator that emerged during this time.  There was a perception building, after years of being too restrictive and denying access to drugs available abroad, that the FDA had become overly permissive with drug approval and oversight, and was slow to act upon available drug adverse event data. In this context, in February 2001, the FDA Arthritis Advisory Committee, met to review data on the gastrointestinal (GI) and CV safety of Vioxx, with a particular focus on the Vioxx Gastrointestinal Outcome Research (VIGOR) trial.¹ The committee recommended that the Vioxx label be modified to include the information that Vioxx could increase the risk of CV events.  Unfortunately, it required 14 months of discussions between the FDA and Merck before a label change was made in April 2002. During that time of negotiation, and indeed until the withdrawal of Vioxx in September of 2004, marketing efforts by Merck promoting Vioxx continued. Indeed, what is striking in reviewing the dynamics of this period is the extent of imbalance which can exist between the messages of caution delivered by the FDA to consumers versus the messages of benefit delivered by DTC marketing.

Finally, in addition to the pharmaceutical industry and the FDA, physicians were increasingly viewed by many as biased in favor of the drug industry. Participation in sponsored symposia, the development of endurable educational materials and participation as consultants to industry were seen as compromising the impartiality of members of the medical profession.  As a result, concern grew that doctors--and their prescribing habits--were heavily influenced by industry-sponsored support and they could not be trusted as honest brokers in protecting patient welfare. The line between consultancy and advocacy on the part of the medical profession had become increasingly blurred.

            What is important to appreciate, therefore, is that the September 2004 COX-2 "tipping point" came in the context of increasing distrust of the motivations of the professions and organizations upon which the public relied for its protection: the drug industry, the FDA and physicians.  The distrust was much wider than the coxib controversy, per se (editorials by Marcia Angell, MD, former Editor-in-Chief of The New England Journal of Medicine, and the British Medical Journal, were broadly based) but the series of disclosures regarding coxib-related CV adverse events, interpreted by some as having been understated for economic gain, were made to a highly sensitized public, and were eagerly amplified by the media.

Why the Epidemic-like Reaction?

The above describes the "power of context" in which the COX-2 controversy emerged, which allowed the very rapid uptake and acceptance of the CV toxicity data.  What was historic, however, was the epidemic-like fashion by which the repercussions spread, involving intense media coverage, plummeting stock prices, the resignation of a CEO, Senate hearings, the withdrawal of two coxibs, marked change in prescribing habits amid significant confusion, and an extraordinary judgment in the first of over 4,000 lawsuits against Merck.

Did the CV data merit the magnitude of the response and, if so, why wasn't something done sooner? Part of the dilemma for the FDA, the companies and for physicians who closely followed the emerging data is that, in any individual study, the numbers of adverse events were small. The debate over VIGOR, which contributed to the delay in label change, also contributed to the quandary. Was there increased CV risk attributable to rofecoxib, or protection from naproxen? The accumulated data now suggest that the answer is most probably "yes' and "yes". Rofecoxib does confer increased CV risk and high-dose naproxen may indeed be cardioprotective by exerting "aspirin-like" sustained inhibition of platelet COX-1. Therefore, at the moment of the evaluation of each clinical data set, there was room for disagreement. Accumulation of data following VIGOR corroborated an increased CV risk for Vioxx, but not in all studies and usually at the >25 mg recommended dose. Moreover, in such population studies through 2004 there did not appear to be significant risk conferred by other coxibs or NSAIDs.  At the time, the question that plagued those, including the FDA, who tried conscientiously to evaluate the data, was the determination of when does a small number in a randomized trial become a clinically meaningful number? How many trials need to confirm a finding of a single study? How do we compare RCT data with population-based data, which could be influenced by channeling bias? There is little doubt that both the FDA's and the academic community's historic reluctance to accept the findings of individual studies (clinical or lab-based) contributed to the lag period between the initial observations and the ultimate acceptance of CV risk. Moreover, two important pharmacoepidemiologic studies - -- including a study by Wayne Ray, PhD, of Vanderbilt University Medical Center in Nashville, Tennessee, and one by Muhammad Mamdani, MD, of Institute for Clinical Evaluative Studies in Toronto, Ontario, Canada,  yielded conflicting data on CV risk and left uncertainty about causality. Ray et al reported that rofecoxib doses greater than 25 mg/day increased risk of CV events. Naproxen was not the comparator in this study. By contrast, Mamdami et al reported no increase in CV adverse events with rofecoxib and no decrease with naproxen.

It was however, the placebo-controlled Adenomatous Polyp Prevention on Vioxx (APPROVe) study, with its confirmed thrombotic endpoint that resulted in Merck & Co voluntarily pulling rofecoxib from the market. This study found that after 18 months of continuous treatment, there was an increased risk of confirmed CV events including MI and stroke in the rofecoxib-treated arm. Celecoxib and valdecoxib soon faced intense scrutiny, and each was found to have evidence of increased cardiovascular risk in placebo-controlled trials in non-arthritic patients, using higher doses than are recommended for osteoarthritis. As conventional NSAIDs also act by inhibiting COX-2 to achieve an anti-inflammatory effect, questions also arose about the safety of these drugs.  The immediate concern – and perhaps a pivotal tipping moment–however, centered on the coxibs and their capacity to promote mechanism-based thromboses. Indeed, the elegant scientific studies of Fitzgerald over this period, which demonstrated a basis for COX-2 inhibitor dependent thromboses, were critical in providing the theoretical construct, or context, around which one could interpret the clinical data.

Thus, during the final quarter of 2004, beginning with the withdrawal of Vioxx and capped by the report of increased CV events associated with Celebrex in the Adenoma Prevention with Celecoxib (APC) trial the media frenzy about CV risk had become intense. The spread of the message was fueled by the story told by Dr. David Graham, of the FDA. Graham accused the FDA of being excessively slow to respond to clear signals of CV risk. As Gladwell might describe it, Dr. Graham was the Paul Revere of the COX-2 counter-revolution, with his memorable testimony before Congress repeatedly shown on 60 Minutes and cable TV, providing a provocative analogy that the risk of Vioxx administration was comparable to 900 commercial jet-liners falling from the sky each year, a graphic and disturbing vision in the post-9/11 era. Interestingly, the Kaiser Permanente data set he used was of interest in that it again showed that while Vioxx was associated with CV risk, Celebrex was not.  At the FDA hearings in February 2005, Dr. Graham reported a new set of data from the Medi-Cal database, which was subsequently reported by Singh et al at the recent at the Annual European Congress of Rheumatology of the European League Against Rheumatism (EULAR) meeting. These data showed that each of the NSAIDs -- not just the coxibs – appeared to increased risk of CV events in a dose-dependent fashion.

The media were quite receptive to the coxib story, which included allegations of cover-up by the pharmaceutical industry and suppression of information by the FDA. The complexity of the studies, the increased risk attributable to the class of NSAIDs, the role of aspirin and the benefit of anti-inflammatory agents were generally given short shrift in the analyses of the data. The message of fear was also fed to the media by personal injury lawyers, who placed newspaper and TV ads trolling for individuals who suffered MIs or strokes while taking coxibs. The fact that a simple Google search of the words Vioxx and lawyer yields 675,000 hits is a remarkable sign of how an epidemic can spread in a connected world.

So, what is the result of this tipping point in the history of the cyclooxygenase inhibitors? Sweeping changes in the way that these medications are labeled, a movement to review and restrict DTC advertising, financial implications to drug sales and stock values, the request by the FDA that Pfizer withdraw Bextra citing additional risks namely Steven Johnson's syndrome, but importantly, no evidence for greater CV risk than other available NSAIDs.  And, looming on the horizon are the numerous lawsuits, where direct causal relationships of coxibs to CV events will be argued and, as in Texas, often won – whether or not in that particular patient COX-2 inhibition had in fact a causal effect.

How do we interpret these events - are they fair and justifiable? Certain actions are clearly appropriate. The action of the FDA to label the class of NSAIDs is justified by the available data, although placebo-controlled studies of the traditional NSAIDs are lacking. 

Clearly, from the scientific perspective, what is needed are large population, longitudinal clinical trials for safety that compare individual coxibs, traditional NSAIDs and naproxen. Ideally, such large studies could be funded collaboratively by industry with an interest in a particular drug and sponsored by the National Institutes of Health (NIH). Scientific studies could be designed – perhaps these funded via NIH peer review – to examine biomarkers and predictors of adverse events in an effort to sort through potential differences among the drugs. In addition, there should be some attempt to validate the large clinical database, which can be utilized to track adverse events.  Pharmacovigilance through such databases is increasingly important–and indeed the studies of Ray, Solomon, Graham, Singh, and others were instrumental in establishing a recurring signal of NSAID-related CV risk. At the same time, population-based studies are subject to criticism due to channeling bias, and indeed the various NSAID/coxib data vary among the published studies.

Beyond the scientific questions that remain, many of the resultant changes have had a positive impact on the pharmaceutical landscape, including closer scrutiny of DTC advertising, the ability of the FDA to act more swiftly, and the need for transparency in physician-industry relationships. 

What is important is to have the next phase be rational and driven by data. While some would argue that the pendulum has swung too far – and certainly the ongoing litigation will test the limits of evidence-based–and lack of evidence–judgments, it is certainly appropriate to assess many of the events and determine where behaviors should be modified. In general, the good intentions of each of the participants in this story, and the efforts to understand the data should be emphasized.  This message is critical in order to restore the trust of the public in the medical profession, the FDA and the pharmaceutical industry.

Steven B. Abramson, MD, is Professor of Medicine and Pathology at the NYU School of Medicine. He is also Director of Rheumatology at NYU and the NYU Hospital for Joint Diseases.  He served as Chairman of the Arthritis Advisory Committee, Center for Drug Evaluation and Research of the Food and Drug Administration (FDA) and continues as an active consultant to the Committee.

References

1.      Bombardier C, Laine L, Reicin A, et al. Comparison of upper gastrointestinal toxicity of rofecoxib and naproxen in patients with rheumatoid arthritis. VIGOR Study Group. N Engl J Med. 2000; 23;343:1520-1530.

2.      Ray WA, Stein CM, Daugherty JR et al. COX-2 selective non-steroidal anti-inflammatory drugs and risk of serious coronary heart disease. Lancet 2002;360:1071-1073.

3.      Mamdani M, Rochon P, Juurlink DN, et al. Effect of selective cyclooxygenase 2 inhibitors and naproxen on short-term risk of acute myocardial infarction in the elderly. Arch Intern Med. 2003;63:481-486.

4.      Bresalier, RS. Sandler RS. Quan, H, et al. Cardiovascular events associated with rofecoxib in a colorectal adenoma chemoprevention trial. N Engl J Med. 2005;352:1092-1102.

5.      Wong D, Wang M, Cheng Y, Fitzgerald GA. Cardiovascular hazard and non-steroidal anti-inflammatory drugs. Curr Opin Pharmacol. 2005;5:204-210.

6.      Solomon SD, McMurray JJV, Pfeffer MA, et al. Cardiovascular risk associated with celecoxib in a clinical trial for colorectal adenoma prevention. N Engl J Med. 2005;352:1071-1080.

7.      Graham, DJ, Campen D, Hui R, et al. Risk of acute myocardial infarction and sudden cardiac death in patients treated with cyclo-oxygenase 2 selective and non-selective non-steroidal anti-inflammatory drugs: nested case-control study. Lancet. 2005;365:475-481.

8.      Singh G, Mithal A, Triadafilopoulos G. Both selective COX-2 inhibitors and non-selective NSAIDs increase the risk of acute myocardial infarction in patients with arthritis: selectivity is with the patient, not the drug class. Presented at: Annual European Congress of Rheumatology of EULAR; June 8-11, 2005; Vienna, Austria. Abstract OP0091.