Vioxx Cardiovascular Risk and Merck's Disclosure (1999–2004)

Introduction

Vioxx (rofecoxib) was a COX-2 inhibitor painkiller marketed by Merck beginning in May 1999. It was heavily promoted as a safer alternative to traditional NSAIDs, with worldwide sales of $2.5 billion in 2003 and sales in more than 80 countries; in November 2004 Senate Finance Committee Chairman Charles Grassley said that 20 million Americans reportedly took it. On 30 September 2004, Merck announced a voluntary worldwide withdrawal of Vioxx after the APPROVe clinical trial, a placebo-controlled study of 25 mg daily, showed an increased relative risk of serious cardiovascular events, including heart attacks and strokes. The trial's first report (NEJM, 2005) put the relative risk at 1.92 and said it became apparent after 18 months of treatment.

What the record establishes is that Merck had a heart-attack signal from its own VIGOR trial by 2000, that the FDA's 2001 warning letter said Merck's promotion minimized it, that the New England Journal of Medicine later questioned the integrity of the VIGOR paper's cardiovascular data, and that the risk was confirmed by the placebo-controlled APPROVe trial in 2004. What it does not establish is that Merck intended to conceal or play down the risk: no plea or settlement contains an admission to that effect, and plaintiff verdicts against Merck were mixed, with some cut by damage caps or, according to Merck, reversed on appeal (see Litigation and Settlement). A Merck vice president told a House committee in May 2005 that the company "believed wholeheartedly in the safety of Vioxx".

The VIGOR Trial and the Cardiovascular Signal

Merck released the VIGOR results (Vioxx Gastrointestinal Outcomes Research) in March 2000, and the trial was published in the New England Journal of Medicine in November 2000. It randomised 8,076 patients with rheumatoid arthritis to 50 mg of Vioxx daily or 500 mg of naproxen twice daily and was designed to show Vioxx's gastrointestinal advantage. It did, with gastrointestinal events roughly halved, but heart attacks were more frequent on Vioxx: 0.4% versus 0.1% in the published paper, which the NEJM later set out as 17 versus 4 events, a relative risk of 4.25. With three further heart attacks that the journal said were left out of the paper, the figure is 20 versus 4, a relative risk of 5.0.

Merck's response was to advance the "naproxen hypothesis", the suggestion that naproxen was cardioprotective (like aspirin) rather than that Vioxx was harmful. Internal documents later reported by the Wall Street Journal showed that Merck staff had discussed the cardiovascular findings. In 2000 Merck's research chief, Edward Scolnick, wrote to colleagues that the cardiovascular events "are clearly there" and that the problem was "mechanism-based"; Merck's general counsel later said the comments were initial impressions written on the day the data became available. A separate 16-page sales-training document called Dodge Ball Vioxx presented sample physician questions about the safety concerns related to Vioxx and listed the correct responses as "DODGE!"; Merck said the title referred to a sales-training game in which representatives were encouraged to answer with FDA-approved language.

In September 2001 the FDA sent Merck a warning letter saying the company had engaged in "a promotional campaign for Vioxx that minimizes the potentially serious cardiovascular findings" observed in VIGOR and "misrepresents the safety profile for Vioxx". On the naproxen explanation the letter said: "That is a possible explanation, but you fail to disclose that your explanation is hypothetical, has not been demonstrated by substantial evidence and there is another reasonable explanation, that Vioxx may have pro-thrombotic properties." The FDA approved label changes reflecting the VIGOR findings in April 2002; the Associated Press reported that Merck added the warnings after two years of wrangling with the FDA.

APPROVe and Withdrawal

The APPROVe trial (Adenomatous Polyp Prevention on Vioxx), designed to test whether 25 mg of Vioxx daily prevented recurrence of colorectal polyps, was stopped in September 2004 after its Data Safety Monitoring Board decided to halt it, on data showing an increased relative risk of confirmed cardiovascular events beginning after 18 months of treatment. Merck contacted the FDA on 27 September 2004, told FDA officials on 28 September that it would withdraw Vioxx, and announced a worldwide withdrawal on 30 September 2004. The trial's full report (NEJM, 2005) found 46 confirmed thrombotic events on Vioxx versus 26 on placebo, a relative risk of 1.92.

The widely quoted death-toll figure comes from FDA drug-safety scientist David Graham's Senate testimony, not from an official FDA estimate. On 18 November 2004 he said an FDA–Kaiser Permanente study report, posted on the FDA website, estimated nearly 28,000 excess cases of heart attack or sudden cardiac death caused by Vioxx, which he called "an extremely conservative estimate". FDA officials told the same hearing that agency scientists had raised questions about the design and statistical modelling of that study when reviewing his August 2004 conference poster. Separately, applying the risk levels seen in VIGOR and APPROVe, an extrapolation he said does not depend on the Kaiser study, he put the US total at 88,000 to 139,000 excess cases, of whom "30-40% probably died" (roughly 26,000 to 56,000 deaths by that arithmetic).

Litigation and Settlement

Multidistrict Litigation 1657 consolidated thousands of cases against Merck in federal court. On 9 November 2007, Merck announced that it would pay $4.85 billion to resolve Vioxx personal-injury claims in federal and state courts, covering lawsuits that represented about 47,000 plaintiffs. Claimants had to show a heart attack or stroke and at least 30 Vioxx pills. Merck stressed that it was not a class settlement and that it was not admitting fault. Trial outcomes had been mixed: the Associated Press counted Merck winning 10 of 15 verdicts by then. Plaintiffs won some trials, including a $253.4 million verdict in Texas in August 2005 (the first Vioxx trial; Texas damage caps later cut it to about $26 million), $13.9 million from a state jury in April 2006 and $32 million in Texas in April 2006 (later reduced to about $7.75 million under state damage caps). In March 2007 a New Jersey jury ruled against a widow's claim but found that Merck violated the state's consumer fraud law. According to Merck, a Texas appeals court reversed the $253.4 million verdict in 2008, finding that Vioxx did not cause the death, and a New Jersey court reversed awards of punitive damage and consumer fraud against the company.

In November 2011 the US Department of Justice announced a $950 million resolution with Merck Sharp & Dohme. The company pleaded guilty on 16 December 2011 to a one-count misdemeanour for introducing a misbranded drug into interstate commerce: it had promoted Vioxx for rheumatoid arthritis before the FDA approved that use, conduct the FDA had criticised in its September 2001 warning letter. Merck agreed to a criminal fine of $321,636,000. In a parallel civil settlement Merck agreed to pay $628,364,000 to resolve allegations, including that Merck representatives made inaccurate, unsupported or misleading statements about Vioxx's cardiovascular safety to increase sales. Merck said the civil settlement was not an admission of liability or wrongdoing and that it believed it had acted responsibly and in good faith. The criminal plea itself did not concern the cardiovascular risk. In January 2016 Merck agreed to pay $830 million to settle a shareholder class action over its statements about Vioxx; the settlement was not an admission of liability or wrongdoing.

Senate Hearings and Internal Documents

The Senate Finance Committee held a hearing on 18 November 2004, "FDA, Merck and Vioxx: Putting Patient Safety First?", at which documents provided by Merck and other parties to Vioxx litigation were discussed. In his opening statement Chairman Charles Grassley said the VIGOR findings went to the FDA in June 2000 and the label was not changed until 11 April 2002, and that "during these 22 months, Merck aggressively marketed Vioxx, knowing that consumers and doctors were largely unaware of the cardiovascular risks found in the VIGOR trial". That is a senator's characterisation, not a finding. FDA drug-safety scientist David Graham told the committee that "the FDA has let the American people down". In May 2005 a House Government Reform Committee hearing released Merck sales documents, including a February 2001 memo telling representatives "Do not initiate discussions" on a study that raised heart concerns. A Merck vice president told that hearing the company had promptly released details of the studies that first raised the concern and followed up by performing the study that ultimately doomed the drug, the Associated Press reported, and lawmakers said there was no indication the sales tactics were illegal.

Verdict

Partly true. The claim is that Merck knew Vioxx raised the risk of heart attacks and strokes and deliberately concealed or played down that risk. The risk itself is established: the placebo-controlled APPROVe trial found about double the risk of serious cardiovascular events, and Merck withdrew the drug. The FDA's 2001 warning letter said Merck's promotion minimized the VIGOR cardiovascular findings; that is a regulator's finding about promotion, not a court finding. The New England Journal of Medicine's 2005 expression of concern said three heart attacks, all on Vioxx, were left out of the published VIGOR paper; it concerned the paper's authors, who disputed it. What is not established is intent, whether to conceal or to play down the risk. The VIGOR imbalance was released in 2000 and given to the FDA, Merck stressed that the 2007 $4.85 billion settlement was not an admission of fault, plaintiff verdicts against Merck were mixed, and the 2011 guilty plea concerned off-label promotion for rheumatoid arthritis, not the cardiovascular risk. The death toll quoted online derives from FDA scientist David Graham's 2004 testimony, not from an official FDA figure.

The Pharmacology and Mechanism of Cardiovascular Risk

The leading mechanistic explanation for Vioxx's cardiovascular hazard is its selective inhibition of cyclooxygenase-2 (COX-2), an enzyme that produces prostaglandins; a 2001 JAMA analysis proposed that reduced prostacyclin production may increase prothrombotic activity. While traditional NSAIDs inhibit both COX-1 and COX-2, Vioxx's selectivity for COX-2 suppresses the production of prostacyclin—a vasodilator and platelet aggregation inhibitor—while preserving thromboxane A2 production, which promotes clotting. On this hypothesis the imbalance shifts the hemostatic equilibrium toward thrombosis (clot formation), which would increase the risk of myocardial infarction and stroke. In the epidemiological analysis Graham presented to the Senate in 2004, high-dose Vioxx was associated with a 3.7-fold increase in heart attack and sudden cardiac death and low-dose Vioxx with a 1.5-fold increase, compared with celecoxib.

Internal Decision-Making: VIGOR and the Naproxen Hypothesis

Merck's initial response to the VIGOR cardiovascular signal was the "naproxen hypothesis". The 2000 VIGOR study, published in the New England Journal of Medicine, reported fewer heart attacks on naproxen than on Vioxx (0.1% versus 0.4%); by the numbers the NEJM later set out, that is a 4.25-fold difference (17 versus 4 events), or five-fold (20 versus 4) once three omitted events are counted. Merck contended that naproxen protected the heart. The explanation was not absurd, and a Merck epidemiologist testified in 2005 that she believed naproxen has a cardioprotective effect. But in September 2001 the FDA said Merck had failed to disclose that the explanation was hypothetical and unproven, and the placebo-controlled APPROVe trial later showed increased cardiovascular risk on Vioxx without any naproxen comparison.

Internal Merck documents that emerged in litigation and congressional investigations show that staff discussed the cardiovascular signal. A 2004 Wall Street Journal investigation, as summarised by California Healthline, reported a 2000 email in which Merck's research chief, Edward Scolnick, wrote that the cardiovascular events "are clearly there" and that the problem was "mechanism-based", and a 16-page sales-training document called Dodge Ball Vioxx that presented sample physician questions about the safety concerns related to Vioxx and listed the correct responses as "DODGE!". Merck said the Scolnick comments were initial impressions written on the day the data arrived and that the document's title referred to a training game. A 2008 JAMA case study by Psaty and Kronmal, based on documents from the Vioxx litigation, reported that Merck's internal April 2001 analysis of pooled Alzheimer's-trial data showed significantly higher total mortality on rofecoxib (34 deaths among 1,069 rofecoxib patients versus 12 among 1,078 placebo patients; hazard ratio 2.99) and that these mortality analyses were neither provided to the FDA nor made public in a timely fashion. Merck's own September 2004 announcement said the VIGOR results, released in March 2000, indicated an increased risk of cardiovascular events versus naproxen. The period between VIGOR (2000) and APPROVe (2004) is the one in which critics, including the FDA's 2001 warning letter, say Merck's promotion understated the risk.

The FDA's Limited Oversight and Regulatory Missteps

The FDA approved Vioxx in May 1999; its testimony to the Senate said the original safety database of about 5,000 patients on Vioxx "did not show an increased risk of heart attack or stroke". After VIGOR, the FDA's Arthritis Advisory Committee discussed the findings in February 2001, the FDA sent its warning letter in September 2001, and in April 2002 it approved label changes reflecting the VIGOR findings, including detailed information on the increased risk of cardiovascular events relative to naproxen. Graham told the Senate that the change came about 18 months after the VIGOR results were published and was not placed in the "Warnings" section. The Associated Press reported in May 2005 that the FDA's drug chief acknowledged it may have taken too long to put a new warning label on the drug. This regulatory delay did not itself amount to concealment by Merck, but it meant the label did not carry the findings until about 22 months after Merck gave them to the FDA.

APPROVe and the Unambiguous Confirmation

The APPROVe trial began enrolling in 2000 to test 25 mg of Vioxx daily as a way of preventing recurrence of colorectal polyps, and was halted in September 2004. Its first report (NEJM, 2005) found an increased relative risk of confirmed thrombotic cardiovascular events (46 on Vioxx versus 26 on placebo, relative risk 1.92) that became apparent after 18 months of treatment, with similar event rates in the first 18 months. The final analysis, published in The Lancet in 2008, found that the hazard ratio "did not substantially change over time" and that the data were compatible with an early increase in risk. Because APPROVe tested a single dose, it cannot show how risk varies with dose.

Because APPROVe compared Vioxx with placebo, the result could not be explained by a protective effect of a comparator drug such as naproxen. Merck withdrew Vioxx within days of the Data Safety Monitoring Board's decision. Merck's announcement said it believed it would have been possible to continue marketing Vioxx with labeling that incorporated the new data, but that a voluntary withdrawal was the responsible course given the alternative therapies and the questions the data raised. The withdrawal came about four and a half years after the VIGOR results were released in March 2000.

Counter-Arguments and the Limits of Individual Variation

One counter-argument is that Vioxx's cardiovascular risk depends on the patient, the dose and the duration of use, so that careful prescribing could have limited harm. Merck's withdrawal announcement said the first 18 months of APPROVe did not show an increased risk, and the FDA's April 2002 label noted that Vioxx 50 mg was not recommended for chronic use. Proponents of this view contend that Vioxx, rather than being unsuitable for everyone, needed clearer prescribing limits.

However, variation between patients does not answer the criticism in the FDA's 2001 warning letter that Merck's promotion minimized the VIGOR findings and so left physicians without a balanced picture on which to judge individual patients. Moreover, an FDA memo of 6 April 2005 concluded that the available data were best interpreted as consistent with a class effect of an increased risk of serious cardiovascular events for COX-2 selective and non-selective NSAIDs alike, and recommended a boxed warning on all prescription NSAIDs.

Public Health Impact and Regulatory Aftermath

The best-known estimates of the harm are those of one FDA scientist, David Graham, rather than an official FDA count. He told the Senate in November 2004 that, extrapolating from VIGOR and APPROVe, Vioxx had caused 88,000 to 139,000 excess heart attacks and sudden cardiac deaths in the United States, of whom 30-40% probably died, and said that Eric Topol had estimated up to 160,000 cases of heart attacks and strokes. A Kaiser Permanente–FDA analysis he led gave a lower figure of nearly 28,000 excess cases, which he called extremely conservative, and FDA officials questioned aspects of that study's design. Vioxx was sold in more than 80 countries and, according to the Senate hearing, about 20 million Americans reportedly took it. Estimates therefore range widely.

The settlements and the 2011 plea are described under Litigation and Settlement above. None of them includes an admission that Merck intended to conceal or play down the cardiovascular risk.

The Vioxx case prompted a broader FDA reassessment of COX-2 inhibitors and other NSAIDs. In February 2005 FDA advisers narrowly voted that Vioxx might be able to come back onto the market under certain conditions, according to the Associated Press, and in May 2005 a Merck vice president told Congress the company was in preliminary discussions with the FDA about what information it would need. An FDA memo of 6 April 2005 concluded that the available data were best interpreted as consistent with a class effect of increased cardiovascular risk for COX-2 selective and non-selective NSAIDs, and recommended boxed warnings on all prescription NSAIDs.