Thought Leader: Scientific Expression

Back in December 2005, the New England Journal of Medicine (NEJM) issued an "Expression of Concern" over the VIGOR (Vioxx Gastrointestinal Outcomes Research) study published in 2000. The journal's editors stated that omissions regarding adverse cardiovascular events called the integrity of the study into question, and asked the VIGOR investigators to submit a correction.

Thomas Schnitzer, MD

But Merck wrote back (its letter was published in February), standing by the results of the study. Shortly thereafter, in the March 16, 2006 issue, NEJM ran another piece, reaffirming its expression of concern. But it also included a letter from the non-Merck investigators, which was equally vehement in affirming the validity of the study. From the looks of it, the back-and-forth is not over yet.

One of the authors of that letter, and a VIGOR investigator, was Thomas Schnitzer, MD, a professor of rheumatology and assistant dean for clinical research at Northwestern University Feinberg School of Medicine. Schnitzer has conducted clinical trials on all the major COX-2 drugs, and is currently the lead clinical trials investigator for the French biopharmaceutical company NicOx, which recently netted a technology deal with Merck. Here, Schnitzer reflects on this discourse with NEJM, and the problems with developing pain medications.

Pharm Exec: In their Expression of Concern, NEJM editors noted the omission of three myocardial infarctions in the published VIGOR study. Why were these events omitted?

Schnitzer: During the course of the study, there was an independent data safety and monitoring board (DSMB) that determined it needed to look at cardiovascular events. The DSMB determined a specific cut-off date for collection of this data. It also determined not to tell the investigators that, because investigators are supposed to be isolated from what the DSMB does—unless they want to stop the study.

What was unfortunate was the date for cutting off collection of gastrointestinal events was different than [the date for] cardiovascular events. The date picked was different by two weeks or so. As a result, what was reported was all the GI events up to the cut-off date for the GI events. And all the cardiovascular events were reported up to the date for the cardiovascular endpoint that was prespecified. But because the two dates were different, there were cardiovascular events that happened between the end of the GI and the end of the cardiovascular studies.

Did the investigators know about the additional adverse events?

The investigators, other than the people at Merck, didn't know about this for six months after the study was published. But one could argue that we didn't have to know about it because it's not a part of the predefined study.

The reality of the situation is that when you do clinical trials, you decide when you're going to stop them before you start working on things. You don't change the rules later. It's a cardinal rule of doing clinical research.

In your mind, do the additional cardiovascular events change the conclusion of your study?

It turns out that if you put those three events in, do the analyses all over again, the hazard ratio—the risk associated with rofecoxib—goes from four to five. The people who do this for a living will tell you that if you have four times the risk or five times the risk, it's essentially the same risk.

Why is this communiqué happening now?

The reality is that these extra events that occurred were in the public domain since early February 2001. The journal knew about them since that time. And if the journal had a concern for the last five years about these events and the implications they had, it had five years to do something about it.

The question is, why would they have an interest in doing something about it at the time that at least one high-profile trial is ongoing? But what probably started the ball rolling is when the editor at the New England Journal was subpoenaed by the plaintiffs.

What do you see is the heart of this issue?

I think there are two separate issues. There's one about not providing accurate data. And that's what NEJM's major concern is—they're saying three events were hidden. But it's rare that companies or people in academia try to hide or not provide real data. Quite honestly, pharmaceutical companies don't do research that way.

They said that by looking at "tracked changes" on an old disk sent to them that there were tables in the article which were deleted that included these cardiovascular deaths, and that they were specifically and purposefully removed so as not to provide that information. Actually, it was an empty table that had no data in it. So the table was deleted, but there were no data—or attempt to not include data.

Now, having said that, the other issue is the interpretation of data. And that's a whole other story. I mean, companies have tried in the past to take data and make them seem more positive than they really are. But the data themselves are rarely inaccurate. So I think one just needs to understand that probably for most cases, we're not dealing with the first, we're dealing with the second.

Whose responsibility is it to ensure that the interpretation is inclusive?

There are journals and then there are reviewers. And it's really the reviewers' responsibility. But the issue of everybody's credibility is at stake. What's the credibility of Merck, of the pharmaceutical industry, of the journals and their editors and staff? No one comes out a winner in this.

How do you think the issues should be addressed?

There needs to be a discussion. Is it best to do it in the journal? I think it's best to do it in a neutral third party, because the journal clearly controls the media. And I don't think it's a disinterested client.

The reality is that there needs to somehow be a better means for dealing with the media. Scientific medical journals are accountable in a way, because they are the ones clearly controlling the expression. Some of the things they do are commendable. But it's clear that it would benefit everyone if these vehicles for expression stay relatively neutral, as opposed to trying to take sides on issues.

Why would you say it's not a disinterested client?

Probably most telling is if you look at what the guidelines are for publishing an Expression of Concern, there are very strict rules about how you do this. It was interesting because the New England Journal subsequently did the right thing when it published an Expression of Concern. [NEJM is] supposed to approach the institution at which this was done, contact the dean or the chief officer of that institution with the expression, have them do an investigation, and then [the journal is] not supposed to publish anything until it gets a report back about what happened.

But they called the lead investigator [before the first Expression of Concern was published in December] on a Wednesday and basically said, "We're going to post it tomorrow." [The investigator] said that it wasn't really fair—that we haven't had a chance to respond or see it or anything. [NEJM] said, "We feel we had to do this." And that's all they said.

Despite the issues, the withdrawal of Vioxx means one less option for pain sufferers. Why has it been so difficult for companies to develop new pain medications?

The last general pain medicine was aspirin. Non-steroidals came on in the middle of the last century, and were advances because they caused less GI problems and were tolerated better. And then the COX2s came on, and they were tolerated even better. But they're not fundamentally different from aspirin.

I don't think we have predictive animal models. It's hard to get mice to tell you how much pain they have. What we're left with is an approach like the one NicOx is taking—basically, they're improving on the drugs we already have in some way, rather than coming up with a brand new drug.

Merck recently inked a deal with NicOx. What's NicOx's approach?

NicOx takes the view that they're going to improve on the products by adding nitric oxide [NO]. It turns out that NO also protects your stomach, or prevents the type of damage you get with nonsteroidals in your stomach. The goal here, initially, was to look at that as an endpoint. In the process of the development program, it also became clear that it mitigated the blood pressure response that you see with non-steroidals. So the goal of the program now is really to demonstrate, ideally, a difference in the blood pressure response.

The Editors Respond

We called the editors at the New England Journal of Medicine. Though they weren't taking interviews, the publication answered a few questions by e-mail.

Pharm Exec: Where should have the VIGOR investigators included the cardiovascular information?

NEJM: The VIGOR article should have covered both the gastrointestinal and cardiovascular adverse events in roughly equal proportions. The cardiovascular information should have been presented and discussed in all sections of the article. Supportive data in the form of one or more tables and a figure should have been included. The tables and figure that would have been most relevant have been reproduced on our Web site as a supplement to our Expression of Concern Reaffirmed.

How would you counsel article authors to include information about risk—particularly when it's there but not represented in the study?

In a study on the safety of a drug, it is mandatory to present all relevant data on risk. In the case of the VIGOR article, the presentation on risk was unbalanced, and very little of the cardiovascular risk data were included. Authors are not permitted to include only selected safety data. They are obligated to tell the whole story. Even unanticipated adverse events, if they are clinically relevant, must be reported in the article.

When authors do not follow the principle of complete reporting of data, physicians get incomplete information and their patients can be put at risk. Our expression of concern has reminded the research community about this important principle.

Did you follow the appropriate guidelines for running the first Expression of Concern?

We followed the guidelines for issuing an expression of concern set forth by the International Committee of Medical Journal Editors (www.icmje.org). Whenever we have issued an expression of concern, we have notified our readers of that concern at the same time we notified the authors, and we have then followed up with a response from the authors or their institution.

Thomas Schnitzer, MD, has conducted clinical trials on drugs such as Vioxx (rofecoxib), Celebrex, and Prexige, and has been published in many top journals. Currently, Dr. Schnitzer is the principal investigator for NicOx Pharmaceuticals' lead pain-control compound AZD3582, currently in Phase II. He also is a professor of rheumatology and assistant dean for clinical research at Northwestern University Feinberg School of Medicine. He graduated from Harvard University in 1971.