Merck pulled its arthritis painkiller Vioxx off the market in September 2004 on evidence that it could cause strokes or heart attacks, the company and its lawyers have stood by the premise that it was dangerous only to patients who took it for at least 18 months.
So it was news last week when prominent medical experts said that new data from Merck indicated that Vioxx’s risks started to emerge after only four months of use. The controversy is the latest illustration of how widely open to interpretation and potential corporate pressure the results of clinical trials can be even when reported in a leading medical journal.
Critics say it is now clear that the previous data analysis was done in a way that minimized the risks of the drug. Some also say that Merck and its academic collaborators should have known about that four-month threshold and made the earlier risks clearer in a medical journal article in March 2005.
It was the first scientific report of the clinical trial results that had prompted the company to withdraw the drug. That article, in The New England Journal of Medicine, concluded: “The increased relative risk became apparent after 18 months of treatment.”
The conclusion “makes the drug look a lot safer than it was,” Dr. Steven E. Nissen, the interim chairman of cardiovascular medicine at the Cleveland Clinic, said last week after reviewing the new data. “If you wanted to construct a legal defense that says nothing happens for 18 months, this is how you would cut the data.”
Merck said the new data, seeming to show an earlier risk from Vioxx, had not yet been compiled in early 2005. But the company argues that even now that the fuller data were available, the numbers do not show a danger from Vioxx before 18 months.
The outside academic scientists who were the authors on the 2005 paper with Merck scientists said they had not seen the new information until recently and would meet this week to evaluate it. But they, too, defend the original article, saying it accurately represents the trial’s findings at the time.
The new data were certain to be seized upon by lawyers representing former Vioxx users, who have filed more than 11,500 lawsuits against Merck. Doctors say the new data also suggest a somewhat greater risk for people who took Vioxx, although some said that by now, presumably more than a year and a half since anyone has taken the drug, the risks of future problems are small.
One thing that seems certain in the latest controversy is that the measurable risks posed by Vioxx depend on whose strokes and heart attacks were counted, and when.
The clinical trial at issue, called Approve, studied 2,600 people, half of whom took Vioxx and half a placebo for three years. The trial was designed to see if Vioxx could prevent precancerous colon polyps, but after concern rose about the heart risks of Vioxx, the trial was also used to study cardiac risk.
The results showed that Vioxx users had twice the risk of a heart attack, stroke, other blood clot or death from a cardiac cause than those who took a placebo. It was when Merck learned the preliminary results of that trial in September 2004 that it withdrew Vioxx from the market worldwide and the paper on the results was prepared. The New England Journal of Medicine published the results online in February 2005 and in print the next month.
The article, to illustrate its conclusion that risk did not arise before at least 18 months, used a type of graph known as a Kaplan-Meier curve to plot the cumulative number of heart attacks or other cardiac “events” that occur over time.
For 18 months the line for the Vioxx users and the one for the placebo users essentially coincide, like two runners neck and neck in a race. After 18 months, the Vioxx line rises at a much faster rate than the placebo line as the users of the drug began having heart attacks and strokes at a greater rate than the placebo group.
But as it turns out, the data used in that chart included heart attacks and other events that occurred only while a patient was taking the drug or for up to 14 days after they stopped. That point was noted in the paper, but only in passing.
The new data Merck submitted to the Food and Drug Administration earlier this month include what happened to the patients for a longer period after they stopped taking the drug. And in this presentation, the Kaplan-Meier chart the one critics have seized upon shows those lines starting to separate at around four months and remaining separate thereafter, although they do diverge more widely at the 18-month point.
The reason for the difference is that more patients taking Vioxx dropped out of the trial earlier than patients taking a placebo. And many of the patients who stopped taking Vioxx before the three-year study period was over did so because they had hypertension or other problems that could raise the risk of a heart attack. And so, critics say, the 14-day cutoff could miss heart attacks in these people.
“People may drop out because they had chest pain and then weeks later they had a heart attack,” said Dr. Alastair J. J. Wood, a drug safety expert at Vanderbilt University. Merck says it was simply adhering to the study’s original design, and that the more recent chart reflects data that were not available to analyze in early 2005. It said, and some outside experts agreed, that a 14-day cutoff is frequently used in clinical trials. Indeed, they said, measuring what happens to patients long after they have stopped using a drug could minimize the measured effects of the drug.
In any case, the company says, the difference between the two charts is not significant. In the old analysis, 22 people taking Vioxx and 20 taking a placebo had a heart attack or other event by the 18th month. The new data, which includes heart attacks after the 14-day cutoff, shows 26 events in the Vioxx group and 21 in the placebo group in the first 18 months.
“We don’t feel that these data change our assessment that over the first 18 months there is not a difference in risk between Vioxx and placebo,” Dr. Ned S. Braunstein, senior director of Merck Research Labs, said in an interview during the weekend.
The use of the 14-day cutoff was also criticized by some members of an advisory panel to the F.D.A., including Dr. Nissen of the Cleveland Clinic, at a meeting on Vioxx in February 2005. But the academic scientists defend the paper and say there was no undue influence.
“The product did not represent any pressure from Merck or skewed interpretation of the data,” said Dr. Robert S. Bresalier of the University of Texas M. D. Anderson Cancer Center, the lead author for the paper.
Dr. Gregory D. Curfman, the executive editor of The New England Journal of Medicine, said editors there would study the new information, but said he could not comment further until then. A spokeswoman for the journal, Karen Pedersen, said that could take several weeks.
The journal has already accused Merck of deliberately omitting information about three heart attacks from a paper in 2000 reporting the results of a previous Vioxx clinical trial. Omitting the information made the drug look less dangerous, they said.
Even before publication of the article in March 2005, there was significant debate over exactly how much emphasis should be placed on the difference in risk for patients who took the drug for less than 18 months and those who had taken it longer.
One of the investigators seems to have voiced some qualms, according to an e-mail message that first surfaced as part of the Vioxx litigation, and the reviewers of the original manuscript also voiced some unease over this finding and concern about the possibility that Merck had too much influence over the conclusions. Those reviewers were anonymous.
In an e-mail message sent to his co-author
s as the manuscript was being written, Dr. Marvin A. Konstam, a cardiologist at Tufts-New England Medical Center, wrote that the authors were “going out on a limb by emphasizing the 18-month issue,” saying they needed to make sure they properly underscored the main finding of the study the cardiovascular risk of the drug. Dr. Konstam, when contacted Friday, declined to comment.
Some of the outside reviewers of the article before it was published did raise concerns about the overall validity of the 18-month threshold as it was presented in the original draft.
“The ‘hand’ of the study sponsor seems too evident throughout the manuscript, which is written consistently in a fashion designed to support the company’s public positions,” said one reviewer. Another argued that the paper “aggressively promotes the safety of up to 18 months of use of rofecoxib. This goes beyond the data of the study.” Rofecoxib is Vioxx’s chemical name.
At the time, Merck was already the target of Vioxx-related plaintiffs’ lawsuits. In the cases that have gone to trial, the company has consistently said there was no danger from Vioxx at less than 18 months’ use.
So far, though, in the three of the five cases that Merck has lost in court with juries awarding verdicts of more than $10 million each time jurors have seemed to look more at the specifics of each patient’s case than whether the drug had been take for at least 18 months.
In The New England Journal of Medicine’s response to the original manuscript, the editor, Dr. Curfman, asked the authors to remove the assertion “that increased risk was observed only after 18 months.” Nevertheless, the 18-month finding was published as part of the article’s main results.
But Merck and the researchers continue to insist that the 18-month finding was a legitimate conclusion, based on the data that were analyzed. In commenting on this early draft, the reviewers were not aware of the statistical analysis that had been performed to reach the conclusion, said Dr. John A. Baron, a professor at the Dartmouth medical school who was one of the authors.
The researchers say they did not play down the risk of taking Vioxx, and they say they never made the claim that people taking the drug for a shorter period of time were not at any risk.
The reviewers’ comments were the result of the decision to submit a preliminary draft for review as well as the typical peer-review process, said Dr. Bresalier, the paper’s lead author.
While he acknowledged that Merck was involved in reviewing the manuscript, he said the paper’s authors responded to all the comments raised by the editor and the reviewers and that the final draft reflected changes made to address their concerns. “The end product was a good final product,” he said.
“If The New England Journal didn’t think this was an appropriate and satisfactory product, they didn’t have to publish it,” Dr. Bresalier said.
While Dr. Bresalier and the other outside authors reached insisted that they felt comfortable with how the risks were eventually portrayed, he said the 18-month finding was not seen as critical by the researchers, even if it had been “up-played” by Merck and its lawyers.
“Of course they want to minimize their liability,” he said.