Print BCTV: Six Years of Avandia -- FDA's Woodcock, GSK's Stewart, Duke'sCaliff, Mahaffey dissect Avandia

Six Years of Avandia

Transcript of BioCentury This Week TV Episode 143




Janet Woodcock, Director, Center for Drug Evaluation and Research, FDA


Dr. Murray Stewart, Senior Vice President Metabolic Pathways and Cardiovascular Therapy Area Head GSK


Dr. Robert Califf, Director, Duke Translational Medicine Institute (DTMI) at the Duke University Medical Center


Dr. Kenneth Mahaffey, Professor of Medicine and Associate Director, Duke Clinical Research Institute






Pioglitazone (Actos)

RECORD study


United Kingdom Prospective Diabetes Study

Clinical Safety Research Consortium

Rosiglitazone Avandia

The Lancet

Phillip Home

The City Organization

Contract Research Organization



Steve Usdin, Senior Editor




STEVE USDIN: Is Avandia a poison pill? An FDA advisory committee says no. The nation's top drug regulator and Avandia's manufacturer respond. I'm Steve Usdin. Welcome to BioCentury This Week.


NARRATOR: Your trusted source for biotechnology information and analysis, BioCentury This Week.


STEVE USDIN: No medicine in modern history has caused as much sustained argument as the diabetes drug Avandia rosiglitazone. The drug, made by GlaxoSmithKline, came under attack six years ago from academics and members of FDA's own review staff. They charged Avandia increased the risk of heart attack and cardiovascular mortality. With Congress looking on, FDA put severe restrictions on the use of Avandia in 2010, and the drug was taken off the market in Europe.


But the debate didn't end. One GSK case study found Avandia poses little or no cardiovascular risk. Critics charged the pharma mishandled its data. Duke University researchers were asked to conduct an independent review. Last month, they confirmed the pharma's conclusions. Now, for the third time, an FDA advisory committee has poured over the data. After two days, the 26-member committee decisively recommended the restrictions on Avandia be relaxed.


The controversy has disrupted the lives of diabetics, raised hurdles for testing new drugs, and cost GlaxoSmithKline billions of dollars in fines, legal costs, and lost sales. The Avandia story is just the tip of the iceberg on a critical public health issue. What's the best way for FDA to make real-world risk-benefit decisions?


Today, we'll hear from GlaxoSmithKline, Avandia's manufacturer, and from the Duke University scientist responsible for reviewing the safety study. First, I'm pleased to be joined by Dr. Janet Woodcock. As director of FDA's Center for Drug Evaluation and Research, she made the decision in 2010 to restrict the use of Avandia. Now, she'll have the final word on changes made in response to the recent advisory committee meeting.


Dr. Woodcock, to start with, there've been members of Congress, there've been prominent physicians who have gone out publicly and said that this drug Avandia has killed large numbers of people. It's caused heart attacks. And that it should have been withdrawn a long time ago. What's your response to that?


JANET WOODCOCK: Well, I don't think we have a lot of evidence that shows a problem with mortality with Avandia. The real issue was there was a meta-analysis that showed there might be increase in heart attacks with this drug. Now, the trial that was just re-adjudicated the RECORD trial was probably the longest randomized trial looking at the cardiovascular safety of Avandia.


And there, the rate of death on people who took Avandia was lower -- slightly lower not statistically significant, but slightly lower -- than people who took standard diabetic regimens. And they were followed for quite a few years. So that's probably the best evidence we have right now that certainly doesn't look like there was an increase in mortality in that study in people who took the drug.


STEVE USDIN: So you mentioned there's this study -- so people get it clear -- the RECORD study. That's a phase three study, where they randomized people to receive either Avandia plus another drug, or other diabetes therapies. There's also two other sorts of data that are out there. One is what they call meta-analyses. Those are pooling or collections of data from a lot of different studies.


And the other is what's called observational data, which is really looking at electronic health records and seeing what actually happened with people in the world. How do you rank those three kinds of data? Because you get conflicting results in this case from the three different sorts of data.


JANET WOODCOCK: Well, that's a good point. I don't rank them. But I think learned bodies have ranked these sorts of information. And there's what's called a hierarchy of evidence. And the gold standard is supposed to be a well-conducted, randomized controlled trial.


And why is that? Because the randomization is supposed to deal with sources of confounding or bias -- in other words, people got put into the wrong groups or whatever because the doctor made a decision rather than randomly assigned to each group. So that's probably the highest level of evidence, most definitive.


Meta-analyses, depending on how they're conducted and so forth, they also are from randomized data often. But usually the studies were conducted for other reasons. And you're pulling them together and trying to figure out some other problem that you're looking for.


And then observational studies, there's nothing wrong with them, but they are subject to confounding because they just look at how people were treated and then look what happened to them. And people may be channeled into different treatments based on their own characteristics. And that's called confounding and may cause problems.


STEVE USDIN: So one of the other criticisms that's come up here, including from some of FDA's own scientists, is the fact that the RECORD study, this phase three study, was paid for and conducted by GlaxoSmithKline, by a drug company. And then there was criticism of whether it did it properly. And that's why there was this independent analysis done by Duke. To start with, should patients, should the public trust trials that are conducted or paid for by drug companies?


JANET WOODCOCK: Well, I often hear that put forth by academics, by patient groups, by outside consumer groups. Well, the drug company did this trial, so there's a level of concern about the validity of the data. We take extraordinary steps at the FDA, and regulators around the world, to make sure that trials are conducted in a fair and unbiased manner and that the data are valid. And those include things called the good clinical practices, which is a whole range of procedures that make sure there isn't any kind of undue interference.


We also go and inspect the trials. We sample sites. For this trial, the RECORD trial, we inspected the center that was running the trial, which was run by the company, and so forth, to make sure that there hasn't been any influence.


And, in fact, many academic trials suffer much more from poor recording of the data, missing data, lack of precision, all sorts of problems, putting people in the trial who weren't correctly entered, and so forth. So there's six of one, half dozen of another. I wouldn't say that academic trials are the gold standard here.


STEVE USDIN: So one of the questions, I guess, about Avandia -- bottom line -- 14, 15 years after it was first approved, there's still a high level of uncertainty about what its safety is, isn't it?


JANET WOODCOCK: There's a level of uncertainty. High level depends on what you think about drugs in general and how much certainty we have. Our biggest problem in drug regulation is we don't have a lot of certainty about all the effects of a drug.


And why is that? Because of human variability. And the drug interacts differently with different people. And so even with a huge development program, it may cost hundreds of millions of dollars, maybe a billion dollars, and enroll tens of thousands of patients. We still don't know everything about any drug that we put on the market. And that is basically the bottom line.


STEVE USDIN: And in 2010, 12 FDA advisers voted that Avandia should be removed from the market. Here's how the vote changed at the June 6 advisory committee meeting.


NARRATOR: You're watching BioCentury This Week.




STEVE USDIN: We're discussing the controversial diabetes drug Avandia with FDA's Janet Woodcock. Dr. Woodcock, we just saw a slide showing the plummeting sales of Avandia, and, in part, that was a result of restrictions that FDA put on. Can you describe what the restrictions that were put on are? And what are the range of things that you might do as a result of this recent advisory committee?


DR. JANET WOODCOCK: Well the restrictions were to allow people who were taking the drug and were satisfied to stay on the drug. Because there were people who testified at our prior AC that the drug really worked for them. And because there's still uncertainty around the safety issues, it didn't seem reasonable to remove that from them. But there were severe restrictions put on placing new patients on this drug.


STEVE USDIN: And what are the range of things that FDA could do in response to the recommendations from the recent advisory committee meeting?


DR. JANET WOODCOCK: Well, we asked the advisers everything from taking the drug off the market to removing all the restrictions, and the votes were more toward the end of removing some of the restrictions. Keeping, allowing, access to this drug and probably removing the very tight restrictions.


STEVE USDIN: So I've spoken with endocrinologists, with physicians who treat patients who have diabetes, even with GlaxoSmithKline, the company that produces it. Everybody says that, really, whatever FDA does, Avandia is not coming back. We're never going to see large numbers of people take this drug again.


So what's the point of continuing to investigate it? Of having this advisory committee meeting, and taking votes, and making decisions about a drug that really isn't ever coming back?


DR. JANET WOODCOCK: Well I don't think it's just about the drug. First of all, I do think it's important to get to as much of the truth as we can, reduce our uncertainty as much as possible, understand the facts as we know them. However, this drug and the controversy around it caused a whole change in drug development in the world of diabetes, that also we are doing for the obesity drugs, and caused FDA to rethink its approach in many ways. And so I think we need to understand as much about this particular episode as possible.


STEVE USDIN: So what does it tell you about that approach if the decision was made to require a lot more data before a drug is approved about its cardiovascular effects? If you go back now and say, well the RECORD study suggests that actually it didn't have those acute cardiovascular effects does that suggest maybe that that threshold should be lowered for diabetes and obesity drugs going forward?


DR. JANET WOODCOCK: Well, I think the one big thing that the episode with Avandia tells us is we shouldn't persist with so much uncertainty about a serious adverse event so long, when many millions of patients in the United States are exposed to a drug. There should be a rapid attempt to resolve the issue and get to more clarity on the side effect.


And I think that was the problem here, is we didn't have that clarity about whether there was a big risk to people or not, at a time when millions of people, perhaps, were taking the drug for their diabetes.


STEVE USDIN: Very, quickly, we've only got a few seconds left. There's another controversy that's coming up in diabetes drugs. It's about pancreatitis with another class of diabetes drugs, the GLP-1s. What's FDA doing there?


DR. JANET WOODCOCK: Well, we have been aware of this, and certainly we're doing more investigations. Again, these are the types of things that need to be resolved as quickly as possible.


STEVE USDIN: Thank you, Dr. Woodcock. We'll get GlaxoSmithKline's perspective in a moment. And later, we'll sum up with an expert from Duke University.


NARRATOR: Now in its 21st year, visit for the most in-depth biotech news and analysis. And visit for exclusive free content.




STEVE USDIN: I'm pleased to be joined by Dr. Murray Stewart, who presented the Avandia data at the recent FDA advisory committee meeting. Dr. Stewart, I want to start, I mean, I think there are two sets of issues about Avandia and about GSK. One is about the disclosure of the data, and the other is about the safety of Avandia.


Today, we're going to be talking about the safety data. I think, up front, we have to mention also that, regarding the disclosure data, that the company paid civil fines and criminal penalties. About the safety data, I want to ask you, what's you're summary of where we are? What do we know about the cardiovascular safety of Avandia?


DR. MURRAY STEWART: So I think the key aspect to the RECORD study, which was a pivotal advisory committee, was that the RECORD study demonstrated for major cardiovascular events -- that's heart attacks, deaths, and stroke -- that it was no different from other commonly used agents, such as Metformin and Sulfonylureas.


STEVE USDIN: Well, and that's interesting point where you say that it was no different from them. Do we know what their cardiovascular risk is, long term?


DR. MURRAY STEWART: So there's been one study done with Metformin, the United Kingdom Prospective Diabetes study, that wasn't looking specifically at the same points, but was looking at overall complications of diabetes. And it suggested that Metformin may be beneficial, but it wasn't conclusive. And, when we look a lot of the diabetes studies, there's actually very little data that is conclusive with regards to cardiovascular outcome studies.


STEVE USDIN: And another thing that I think that's interesting and important, does GSK have any plans to try to bring Avandia back if FDA were to loosen the restrictions? Do you imagine that it's ever going to become a major diabetes drug that'll be widely used?


DR. MURRAY STEWART: So I don't see GSK actively promoting Avandia. I think GSK will welcome the opportunity for patients to have access to Avandia who really need the drug.


STEVE USDIN: I'm wondering about the criticisms of the RECORD study, there's two sets of criticisms. One of them gets back to what we were talking about. You said about the cardiovascular risk compared to Metformin and Sulfonylureas is that it wasn't a placebo controlled study. And another one is that it was an open label study. What would be your responses to those?


DR. MURRAY STEWART: So I think we'll take the challenge against placebo. So this study was actually designed as a comparator study after discussion with the European authorities. They wanted to say, well, how does this compare to other commonly used agents? A lot of the discussions with the U.S. RECORD authorities, like to see a drug behave against placebo.


STEVE USDIN: Is that realistic in diabetes? Can you not treat people?


DR. MURRAY STEWART: So I think that's a real challenge. So what we tend to mean by placebo is against background care. But if you treated one group with a new drug that lowered glucose, and the other group you didn't treat a drug that didn't lower glucose, then are you really being fair to the patients?


STEVE USDIN: It doesn't seem like something that you could actually do in real life.


DR. MURRAY STEWART: So in real life, what I think you'd have to do in the placebo group, you would really need to adjust the normal standard of care, to ensure they got good control. But that would raise other challenges is what we call confounders, so the change in the background medication could influence the results. So you're really trading off, do you want to see a drug against background care or against an active comparator?


STEVE USDIN: Which, in some cases it might be the same. One of the outcomes from, or results, of the whole controversy over Avandia has been that FDA has raised the bar on the amount of data that companies need to provide in advance in order to get a diabetes drug approved. What do you think of where the bar is now? Is it at about the right place? Too high? Too low?


DR. MURRAY STEWART: I think it's about the right place, because I think it's important that when you get a drug approved and test that not only lowers glucose, but that it doesn't cause any complications. And we know that heart disease is an important complication for people with diabetes. And I think it's appropriate that we do assess the cardiovascular safety of all diabetes drugs.


STEVE USDIN: One of the arguments in favor of restricting Avandia's use was that there was another drug, Pioglitazone, also called Actos, that was available as an alternative. In fact, now there are concerns about bladder cancer for Actos, and patients are coming off of that. Overall, are these kinds of risks, how do they balance against the risks of not having someone's blood sugar under control?


DR. MURRAY STEWART: So I think, first thing, it's very important to control the blood sugar. And I think every patient is different, and they need to discuss all options with their doctors. In some case, the right drug may be Pioglitazone. In some case, the right drug may be Rosiglitazone Avandia.


STEVE USDIN: So, Dr. Stewart, what's your summary of what the advisory committee recommended to the FDA? Because the vote was a little bit confusing. And what do you think it is that FDA should do? What would you advise them to do?


DR. MURRAY STEWART: So I think the vote was mixed in terms of lifting the REMS, because I think a lot of the committee--


STEVE USDIN: I'm sorry, REMS is the Risk Evaluation Mitigation--


DR. MURRAY STEWART: --Mitigation Strategy. What I think the sentiment of the committee was that there should be easier access but maintaining adequate communication information. And I think they obviously want to discuss with the FDA our communication plans to ensure patients and physicians are aware of the risk, that they get access to the medicine.


STEVE USDIN: There's also, in the whole saga around Avandia, which has been going on since 2007, there's been a lot of criticism of GSK. There's been a lot of criticism of the pharmaceutical industry, in general, around how much data that it provides to independent investigators around possible bias in studies, things like that. What would be your response that? Has GSK changed its culture or its practices, as a result of all of the controversy around Avandia?


DR. MURRAY STEWART: I think GSK has certainly looked at itself and said, we want to maintain high standards. We want to allow access to our information, and we welcome the re-adjudication because it allowed us to give all the information we had to an independent group and see if they found the results that we did. And I'm glad to say that they did.


STEVE USDIN: And then, what about also the release of patient-level data? There have been a lot of calls for more transparency from drug companies about the data they generate in clinical trials.


DR. MURRAY STEWART: We welcome the opportunity for people to ask questions of GSK and approach. We don't encourage, what I would call fishing expeditions that people just come and just browse through it. But if people have appropriate scientific questions, we'll be setting up forums where people can come and ask questions and have access to the data.


STEVE USDIN: And another thing that's been brought up by the Avandia experience is kind of shining a spotlight on what are called meta analyses, this idea that you can pool the data from a lot of different trials and learn something that you wouldn't have from any of the individual trials. What do you see is the kind of strengths and weaknesses in that approach?


DR. MURRAY STEWART: So I think, sometimes, meta analyses is answering a different question. But if you wanted to know clinical evidence, I think the hierarchy still suggests that a well conducted, randomized, control trial is the highest source of evidence.


STEVE USDIN: Great. Well, thank you. Thank you very much, Dr. Stewart. Next, we'll speak with Dr. Rob Califf of Duke University.


NARRATOR: BioCentury, named the 2012 Commentator of the Year by the European Mediscience Awards for excellence in communications and clear, concise commentary.




STEVE USDIN: Now Dr. Robert Califf joins us from the campus of Duke University where he's Vice Chancellor for Clinical Research. Dr. Califf, 14 years after Avandia was first approved, do we know now, do you have confidence that you know whether it causes heart attacks?


ROBERT CALIFF: You asked the question, does Avandia cause heart attacks, which to me is really the wrong question. The question is do we know the balance of risk and benefit overall of Avandia? And the second question would be, do we know anything more about, or less about Avandia than any of the other diabetes medicines that are currently being prescribed? But let's face it the only one we really have assurance about at this point is Metformin. It's been around longer than 14 years. So if you ask me specifically about does it cause MI, I don't know. I'm not sure. But if you ask me, what does it do relative to the other diabetes medicines, I'd say, I don't know, because they haven't been studied either.


STEVE USDIN: So how are we in the situation where potentially, very, very large numbers of people are being treated with these drugs for diabetes, they need to get their blood sugar under control, and we don't know what the cardiovascular side effects or profiles of these drugs are, and how do we get out of that?


ROBERT CALIFF: Well, again, I'd like to generalize this. I don't think it's just a matter of cardiovascular risk. We're doing some work right now on the issue of long term opioid therapy. Billions of dollars being sold. We don't even know if opioids improve pain relief beyond the first two months. So this is not a unique issue in diabetes, but I think diabetes is a great paradigm to talk about.


Because so many people are affected by diabetes, and there's so many interesting and new potential therapies on the way. You know, this is really an artifact of the misconceptions that existed about drug development going back 20 years, that you could use surrogate endpoints to really understand the balance of risk and benefits. You know, biomarkers, measurements like hemoglobin A1C are extraordinarily useful.


And development of drugs and following therapy, but if we take a pill chronically, what we really need to know is do the benefits overall exceed the risk? And very often the benefit-risk profile is determined by things that are not specifically in the main pipeline of the disease of interest. It's due to off-target effects, due to the fact that most medicines affect biology in a lot of different places.


The example where this has worked out really well would be statins, for example, developed to lower LDL cholesterol. But it turns out for a variety of problems, statins are really net beneficial. But we didn't know that until we did long term studies.


STEVE USDIN: So how long will it take, or would it take to be able to get the long-term profiles of drugs for chronic disease like diabetes?


ROBERT CALIFF: Well, the old way we did clinical trials, it would take a real long time because we would design trials in narrow populations, and be very careful about inclusion, exclusion criteria and end up with small studies that took almost forever to do. Which, you know, if you ask the question, how many people on the face of the earth have diabetes, it's in the tens of millions of people.


So with modern technology, we really have the capability of doing trials much quicker. But even though we can do trials quickly, we do want to know about long-term effects. So I think the new rules that the FDA and the European regulators have put into place are actually quite good. And the rules basically say to get a drug on the market, you've got to show that you don't increase risk above a threshold of an odds ratio of about 1.8.


STEVE USDIN: Which means about an 80% chance of an 80% maximum 80% increase that causes some kind of adverse cardiovascular issue.


ROBERT CALIFF: That's right, so an odds ratio of 1.8, I think a reasonable way to think of it is that way. But it's also, I think it's critical to point out to people that when you look at a point estimate in a clinical trial, it's essentially a bell-shaped curve there.


That is, those outer confidence limits like, let's say it was 1.7, there's a low probability that that's the truth. The central tendency to get a 1.7 in the kind of trials that are being done would need to be pretty close to 1.


STEVE USDIN: So I'm sorry, Dr. Califf, that's all the time that we've got today. But I think that sums it up basically that it's a highly complex issue, and it's going to take a long time to resolve these questions. That's this week's show. To find out more about Duke's review of the Avandia data, watch our exclusive interview with Dr. Kenneth Mahaffey at


I'd like to thank Janet Woodcock, Murray Stewart and Rob Califf. Remember to share your thoughts about today's show on Twitter. Join the conversation by using the hashtag I'm Steve Usdin, thanks for watching.




NARRATOR: Now, a BioCentury This Week Web Exclusive.


STEVE USDIN: Dr. Mahaffey, I wanted to ask you about the re-adjudication of the RECORD results. You were in charge of that process. To start with, can you tell us why you did it in the first place, and what were the results? And how you did it, and what were the results?


KENNETH MAHAFFEY: Certainly. So following an Advisory Committee meeting in July of 2010, the FDA required the sponsor GSK to commission a re-evaluation of the endpoints in the RECORD trial. GSK asked us to provide a bid. We did so, and we were awarded the project. The project included a comprehensive review of all of the RECORD data.


We received electronic data sets from GSK. We also received all of the case report forms that the site investigators completed when they were following the patients in the RECORD trial. We systematically reviewed all of those sources of information to identify possible myocardial infarctions, possible strokes, and all the deaths.


And then we sent them through a review process, or adjudication process, where these events were reviewed by physicians to make a determination of whether or not the event did or did not occur based on pre-specified definitions for those endpoints.


STEVE USDIN: And in the end, what did you determine? What did you decide, both compared to the results that GSK originally announced, and just in absolute terms, what was the conclusion of the study?


KENNETH MAHAFFEY: So in the re-evaluation effort, when we looked at the composite endpoint that was the primary endpoint of cardiovascular death, MI, or stroke, we found a modest number of additional MIs and strokes, and a few additional deaths compared with the original RECORD adjudication by their Clinical-Events committee.


When we performed the statistical analyses to look at the comparisons between Rosiglitazone and the comparator arm, our results were very similar to what was previously published from the RECORD database by Home and colleagues in The Lancet.


STEVE USDIN: And so there's been criticism of the study, and there's been criticism of Duke. The criticism of the study saying that you can't trust a study that was sponsored by a drug company, and criticism of Duke in saying that you can't trust an analysis that similarly was sponsored by a drug company, and that they could have somehow also manipulated the data that you received. What would be your responses to those criticisms?


KENNETH MAHAFFEY: Well, I think that the Advisory Committee had a lengthy discussion about that, and I think summarized nicely the issues. The first is that GSK says that they sent us all the data. That they're legally required to send all the data to the FDA and therefore to us, and there's no indication that they didn't do that.


And then for the second thing, where you're questioning research being done that's sponsored by industry, and I'll remind you that the majority of research around the world where there's tens of billions of dollars spent on clinical research, the majority of that research is funded by the pharmaceutical industry. And with shrinking NIH dollars, more work is likely to be done by industry, or a higher proportion of the research will be funded by industry.


And therefore I think it's critical that we continue to have academic groups who don't have a vested interest in the outcome of a specific drug, because of the need to, in publicly traded companies to address shareholders, discuss concerns, and increase profits, that I think the best way to do research in the future will be to bring the key stakeholders together -- academia, industry, the regulatory agencies and other important bodies -- to run research in the best way possible.


STEVE USDIN: So just briefly, we've got a few moments left. You think going forward, then, that a group like yours should or could have been involved in the original adjudication of the study, and that that might have prevented the kind of disputes that have been going on for the last more than two years about the RECORD study?


KENNETH MAHAFFEY: Well, there was an academic group that was the Events Committee for the RECORD trial. It was supported by Contract Research Organization, and as well as the sponsor. I think going forward, that again, if we bring together key groups in academics, industry, government agencies, regulatory bodies, and there's a number of great examples where there's been terrific successes. The City Organization, the Clinical Safety Research Consortium, where we're bringing together stakeholders to try and evaluate not only how research is being done now and how to improve on it, but where we need to get to in order to be able to do research more efficiently, and answer important questions in the future.


STEVE USDIN: Great, well thank you very much, Dr. Mahaffey.