As academic researchers increasingly raised questions about Vioxx's heart safety, the company struck back hard. It even sued one Spanish pharmacologist, trying unsuccessfully to force a correction of an article he wrote. In another case, it warned that a Stanford University researcher would "flame out" unless he stopped giving "anti-Merck" lectures, according to a letter of complaint written to Merck by a Stanford professor. A company training document listed potential tough questions about Vioxx and said in capital letters, "DODGE!"

The revelations shed new light on the interplay between marketing and science at Merck as bad news piled up about a blockbuster drug used by some 20 million Americans. Amid growing danger signs, Merck fought a rearguard action for 4 years, clinging to a hope that somehow Vioxx's safety could be confirmed -- even though its research chief had already privately acknowledged its risks.

Some of the internal documents may also prove damaging to Merck in court, where it faces lawsuits by the families of those who suffered heart attacks after taking the drug. Such lawsuits had begun before Vioxx's withdrawal, and since the announcement the number of potential plaintiffs has multiplied.

TRIAL RESULTS A Merck trial compared people taking a high dose of Vioxx with those taking naproxen. VIOXX NAPROXEN Total number in trial 4,047 4,029 Cardiovascular adverse events 101 46 Digestive system adverse events 48 97 Source: FDA analysis of Vigor trial data

Merck said in a news release Friday that it "acted responsibly and appropriately as it developed and marketed Vioxx." It added, "When questions arose about the safety of Vioxx, Merck took steps to investigate and address these issues." The study that ultimately led Vioxx to be withdrawn was sponsored by Merck itself, the company has noted.

Ted Mayer, a lawyer representing Merck, said the internal e-mails and marketing materials were "taken out of context" and "do not accurately represent the conduct of Merck and its employees." People with access to a selection of internal documents that tend to reflect poorly on Merck permitted The Wall Street Journal to review them, but Merck didn't provide other documents to furnish context, citing ongoing litigation.

Merck declined to discuss in detail the internal documents or make their authors available for comment, citing ongoing litigation. The Friday news release said "the business practices of Merck may well be misrepresented in any reporting" because of the selective release of documents.

Mr. Mayer also said Merck "is committed to open and vigorous scientific debate" and "never has had a policy of retaliating against scientists" but "has a right to defend its medicines against false claims."

Older painkillers such as aspirin and Aleve, known generically as naproxen, block two enzymes -- Cox-1 and Cox-2 -- that are involved in inflammation and pain. Blocking Cox-1 can damage the stomach and intestines but it also may prevent blood clots. Vioxx and another drug, Pfizer Inc.'s Celebrex, were designed to block only Cox-2.

From early on, companies developing Cox-2 inhibitors faced a dilemma. The drugs seemed to offer clear benefits to arthritis and other pain sufferers who couldn't stand the stomach damage of aspirin, naproxen or ibuprofen. But that was a relatively small market. The real bonanza lay with the general mass of pain patients.

In the late 1990s Merck was facing the loss of patent protection on several top drugs and needed a big hit. However, it would be difficult to penetrate the mass market if doctors and patients believed that by choosing Vioxx, they were forgoing a potential heart benefit.

A Nov. 21, 1996, memo by a Merck official shows the company wrestling with this issue. It wanted to conduct a trial to prove Vioxx was gentler on the stomach than older painkillers. But to show the difference most clearly, the Vioxx patients couldn't take any aspirin. In such a trial, "there is a substantial chance that significantly higher rates" of cardiovascular problems would be seen in the Vioxx group, the memo said.

A similar view was expressed in a Feb. 25, 1997, e-mail by a Merck official, Briggs Morrison. He argued that unless patients in the Vioxx group also got aspirin, "you will get more thrombotic events" -- that is, blood clots -- "and kill [the] drug."

In response, Alise Reicin, now a Merck vice president for clinical research, said in an e-mail that the company was in a "no-win situation." Giving study subjects both Vioxx and aspirin, she wrote, could increase the "relative risk," apparently referring to gastrointestinal problems. But, she added, "the possibility of increased CV [cardiovascular] events is of great concern." From the context, it seems Dr. Reicin meant "increased" relative to older drugs.

She added in parentheses: "I just can't wait to be the one to present those results to senior management!" She proposed that people with high risk of cardiovascular problems be kept out of the study so the difference in the rate of cardiovascular problems between the Vioxx patients and the others "would not be evident."

Vigor Trial

It's not clear what happened to the proposed trial discussed in the 1996-97 documents. But in early 1999, Merck started an 8,000-person trial named Vigor -- for the Vioxx Gastrointestinal Outcomes Research study -- to prove the drug's gastrointestinal safety benefits. The trial compared people taking a high dose of Vioxx with those taking naproxen. It excluded patients who were at high risk of heart problems. No patients were allowed to take aspirin.

PAINFUL PROCESS February 1997: Internal Merck e-mail warns that a proposed trial may show Vioxx patients having more blood clots than those taking another medication and "kill [the] drug." May 1999: FDA approves Vioxx for arthritis and other types of pain. March 2000: Vigor trial results available inside Merck. They eventually show Vioxx group has five times the rate of heart attacks as naproxen group. March 9, 2000: Merck research chief says in internal e-mail that cardiovascular problems "are clearly there" in Vigor trial and appear to be "mechanism based." April 2000: Merck press release says Vigor trial results are "consistent with" clot-preventing effects of naproxen. October 2000: Merck official threatens that a Stanford researcher will "flame out" if he continues "anti-Vioxx" lectures. September 2001: FDA sends Merck a warning letter, accusing it of misleading public about Vioxx's cardiovascular safety. April 2002: FDA approves a label change for Vioxx, showing data about potential heart risk. September 2004: Merck announces world-wide withdrawal of Vioxx. Source: WSJ research

In March 2000, the results of Vigor came in. They showed that Vioxx patients suffered fewer stomach problems than the naproxen group, but significantly more blood-clot-related problems -- precisely the sort of result anticipated in the 1996-97 internal documents. The heart-attack rate in the Vioxx group appeared to be four times as high as the naproxen group. (Later analysis would show it to be five times as high.)

The difference was so wide that Dr. Scolnick, the Merck research chief, appeared to recognize it couldn't come solely from naproxen's protective effect but must involve some sort of risk inherent in Vioxx. In a March 9, 2000, e-mail with the subject line "Vigor," Dr. Scolnick said the results showed that the cardiovascular events "are clearly there." In an apparent acknowledgment that Vioxx's own mechanism was at least partially at fault for the heart data, he wrote: "it is a shame but it is a low incidence and it is mechanism based as we worried it was."

Dr. Scolnick wrote that he wanted other data available before the results were presented publicly, so "it is clear to the world that this" was an effect of the entire Cox-2 class, not just Vioxx. The research chief, by then nearing retirement after 15 years in his post, then recalled some of his greatest hits that also had side effects but were big sellers. In Vioxx, he wrote, "We have a great drug and like angioedema with vasotec and seizures with primaxin and myopathy with mevacor there is always a hazard. The class will do well and so will we." Dr. Scolnick didn't respond to phone messages seeking comment.

But in a news release that month, Merck offered no hint of Dr. Scolnick's suggestion that there was a "mechanism-based" problem with Vioxx or a "hazard" that went beyond Vioxx's failure to offer the protective benefits of other painkillers. Merck said the Vigor trial results were "consistent with" naproxen's favorable effects, implying that this could explain why Vioxx didn't do as well.

The next month Merck issued another news release headlined, "Merck confirms favorable cardiovascular safety profile of Vioxx." While acknowledging the Vigor results, it said other trials and data had shown "NO DIFFERENCE in the incidence of cardiovascular events" between Vioxx and a placebo or between Vioxx and older painkillers.

Mr. Mayer, the lawyer representing Merck, says such statements accurately reflected the state of scientific knowledge at the time. "The known antiplatelet properties of naproxen strongly suggested that a property of naproxen was responsible for the differential rates in the Vigor trial," he says. Mr. Mayer declined to comment on Dr. Scolnick's e-mail.

In November 2000, the Vigor results were published in the New England Journal of Medicine. The article, written by academics who received consulting contracts or research grants from Merck and by Merck employees, discussed Vioxx's benefits for the stomach and heart-attack rates. But it didn't include information that, in retrospect, was important. Among patients who weren't already at high risk for heart attacks, it said, Vioxx didn't show a significant rise in heart attacks. That implied it was all right for people with healthy hearts -- say, a jogger in his 30s with joint pain -- to take Vioxx. But the article didn't provide detailed information about other serious cardiovascular complications such as strokes or blood clots.

John Abramson, a family doctor and clinical instructor at Harvard Medical School, scrutinized detailed data on the Vigor trial provided by Merck to the FDA and posted on the FDA Web site. In a book published this summer, "Overdosed America: The Broken Promise of American Medicine," he concluded that even those without a history of heart trouble doubled their risk of developing a cardiovascular problem by taking Vioxx instead of naproxen.

Gregory Curfman, executive editor of the New England Journal, says the journal "didn't have all the details that the FDA had later on." Given the available data, he says editors "spent a great deal of time trying to make sure that these unexpected cardiovascular side effects were fairly and accurately represented" in the article.

By 2001, the Vigor data had clearly caused the debate to shift. The main question was no longer whether Vioxx lacked the benefits of older painkillers and if so whether that was significant. Now the issue was squarely Vioxx itself: Was the drug intrinsically risky?

In February 2001, the FDA presented its analysis of the Vigor data to an agency advisory committee. It showed that the number of people who had a digestive problem while taking naproxen was about double the figure for Vioxx takers -- but that difference was almost exactly the same as the additional number of Vioxx users who suffered a cardiovascular problem such as a stroke.

FDA officials wanted to highlight the cardiovascular risk prominently on Vioxx's label. Merck resisted, complaining that the agency was putting more weight on the negative findings than on the positive gastrointestinal aspects. In the end, the two sides compromised. The new Vioxx label, which went into effect in April 2002, listed the good news about fewer upset stomachs first. Then it added two tables with the bad news about more heart attacks and strokes.

The agency, meanwhile, had become increasingly concerned about Merck's marketing of the drug to doctors. It complained in a Sept. 17, 2001, warning letter about a Merck-sponsored presentation by a doctor in June 2000. The doctor had said the Vigor trial showed that naproxen was "a wonderful drug" for reducing the risk of heart problems -- not that there was anything wrong with Vioxx. Such statements, the FDA said, "minimized the potentially serious cardiovascular findings" of Vigor.

A Merck internal marketing document reviewed by The Wall Street Journal, addressed to "all field personnel with responsibility for Vioxx," provided an "obstacle handling guide." If a doctor said he was worried that Vioxx might raise the risk of a heart attack, he was to be told that the drug "would not be expected to demonstrate reductions" in heart attacks or other cardiovascular problems and that it was "not a substitute for aspirin." This wasn't a direct answer.

One training document is titled "Dodge Ball Vioxx" and consists of 16 pages. Each of the first 12 pages lists one "obstacle," apparently representing statements that might be made by a doctor. Among them are, "I am concerned about the cardiovascular effects of Vioxx" and "The competition has been in my office telling me that the incidence of heart attacks is greater with Vioxx than Celebrex." The final four pages each contain a single word in capital letters: "DODGE!"

Mr. Mayer, Merck's lawyer, declined to discuss the document specifically but said sales representatives were trained to discuss Vioxx in a manner "consistent with FDA-approved labeling" and "were not trained to avoid physicians' questions."

Merck also went on the offensive against academic researchers who began to question Vioxx's safety. Gurkirpal Singh of Stanford University, a prominent Cox-2 expert who was giving lectures sponsored by Merck and other companies, says he pressed Merck repeatedly for more cardiovascular safety data. When Merck refused, Dr. Singh added a slide to his presentations that showed a man -- representing the missing data -- hiding under a blanket. "This was the first time they didn't answer my questions," he says. "With Vigor, suddenly it was a clampdown."

'Flame Out'

Merck canceled several presentations by Dr. Singh that it had been scheduled to sponsor, and it didn't stop there. In October 2000, a Merck official, Louis Sherwood, called James Fries, a Stanford University Medical School professor, to complain that Dr. Singh's lectures were "irresponsibly anti-Merck and specifically anti-Vioxx," as Dr. Fries described the call in a January 2001 letter to Mr. Gilmartin, the Merck chief executive. The Merck official "suggested that if this continued, Dr. Singh would 'flame out' and there would be consequences for myself and for Stanford," Dr. Fries wrote.

Dr. Fries struck back. "There is a line that you can't go across. ... It had gone over that line," he says. He wrote to the Merck chief that researchers at several other top medical schools complained about "a consistent pattern of intimidation of investigators by Merck" on Vioxx.

Mr. Gilmartin responded that Merck had a "deep and abiding commitment to the highest ethical standards in all our dealings with physicians and other healthcare providers." Dr. Fries and other researchers mentioned in the letter say the company did try to repair relations subsequently. Dr. Singh, now an adjunct clinical professor at Stanford, says he stopped using the blanket slide after Merck gave him more data.

Lee Simon, a rheumatologist at Beth Israel Deaconess Medical Center in Boston, says he publicly mentioned data showing Vioxx might be associated with a risk of high blood pressure and swelling. While Dr. Simon was closely involved with research on the rival Cox-2 drug Celebrex, he had worked with Merck in another area. Merck's Dr. Sherwood called Dr. Simon and one of his superiors at the hospital to complain that Dr. Simon's lectures were slanted against Vioxx.

"I was shocked that there was a phone call made like that," Dr. Simon says. "The company was attempting to suppress a discussion about this data."

M. Thomas Stillman, a professor at the University of Minnesota, also discussed the data on high blood pressure and swelling in his lectures -- and also got a call from Dr. Sherwood. "We had a very direct conversation that I wouldn't call friendly," Dr. Stillman says. "It had a tone to me of, 'You better be careful of what you're saying.' ... I thought that was inappropriate." He had given Merck-sponsored lectures but that ended after the disagreement, he says.

In August 2001, researchers at the Cleveland Clinic published an analysis in the Journal of the American Medical Association that once again raised concerns about Vioxx's cardiovascular risks. Before it came out, Merck's Dr. Reicin and other officials met with the authors, arguing that "they didn't think there was a problem with the drug," says Steven Nissen, one of the Cleveland Clinic researchers. The company also asked the journal to run a Merck rebuttal but it refused, people with knowledge of the matter said at the time.

One of Merck's most aggressive moves came against Joan-Ramon Laporte of the Catalan Institute of Pharmacology in Barcelona, Spain. In the summer of 2002, a publication of the institute edited by Dr. Laporte repeated criticisms of Merck's handling of Vioxx that had been published in the British journal Lancet. Soon after, Dr. Laporte says, Merck officials sent him a "rectification" to publish, but he responded that there would be no correction. After Merck officials approached him twice more, the company filed suit in a Spanish court against Dr. Laporte and the institute, taking advantage of a Spanish law that allows plaintiffs to demand a public correction of inaccurate published information.

In January of this year, a judge ruled that Dr. Laporte's publication accurately reflected the medical debate about the cardiovascular safety of Vioxx, and ordered Merck to pay court costs.

This March, Dr. Laporte was a featured speaker at an annual update on the pharmaceutical world for about 1,000 Spanish family physicians. Merck had helped pay for the meeting for the previous eight years. It contacted the organizer, Ramon Morera i Castell, and told him that the company "preferred" if Dr. Laporte stayed off the program this year, says Dr. Morera. After Dr. Morera rejected the request, Merck withdrew its financing -- about $140,000. Though there wasn't any specific threat, "the message was clear," says Dr. Morera.

No one knows for sure why Vioxx might be tied to heart attacks and strokes. Some scientists point to a class effect of the Cox-2 inhibitors, but several studies suggest that Pfizer's Celebrex doesn't share Vioxx's risks. Pfizer says Celebrex might actually protect the heart.

Throughout 2002 and 2003, critics of Vioxx had one problem: The principal evidence against the drug came from a single source, the Vigor study. Conducting a new prospective trial, in which patients would be given Vioxx or a placebo and followed to see what happened, would be expensive for academic researchers to conduct on their own.

Several groups did conduct "retrospective" analyses, trawling through large databases of patient records for hints as to whether Vioxx was risky. Such studies are prone to confounding factors but can add to a body of evidence.

Merck itself sponsored one retrospective analysis by researchers at Harvard. It found Vioxx was "associated with an elevated relative risk" of heart attacks compared to use of Pfizer's Celebrex or no similar painkiller. Merck asked the researchers to delete or tone down the part of the statement about the no-painkiller group, but they refused, according to Daniel Solomon, a Harvard professor who was the lead author. "We made a decision that we should let the science rule the day," he says. Just before the paper was to be published in May 2004 in Circulation, the journal of the American Heart Association, Merck removed the name of an employee who had worked on the study from the paper's list of authors.

On Aug. 25, data presented at a medical conference by a researcher from the FDA's own drug-safety office showed that higher doses of Vioxx correlated with a tripled risk of a heart attack or sudden cardiac death compared to people who weren't taking any similar drug.

The evidence was now piling up, yet Merck stuck to the line it had kept since March 2000. A news release said Merck "strongly disagreed" with the FDA study's conclusion, noting that it was a retrospective analysis. The top of the release read: "Merck stands behind the efficacy, overall safety and cardiovascular safety of Vioxx."

The next month, company officials were informed that a safety monitoring board wanted to stop an ongoing study of Vioxx's ability to prevent colon polyps because people on the drug were having more heart attacks and strokes. The numbers were small. Among patients taking Vioxx for more than 18 months, there were 15 heart attacks or strokes for every 1,000 patients compared with 7.5 per 1,000 who were on placebo. For patients who took the drug 18 months or less, there was no increased risk, according to Merck.

This was a prospective study comparing Vioxx to a placebo. Merck's previous defenses -- criticizing retrospective studies or attributing results to the benefits of a pill used for comparison -- collapsed, and it withdrew the drug. Its stock price fell 27% and now stands about where it was in early 1996.

Write to Anna Wilde Mathews at anna.mathews@wsj.com ² and Barbara Martinez at barbara.martinez@wsj.com ³