Letter Urging FDA to Halt Rosiglitazone (Avandia) Trial

May 11, 2010

Margaret A. Hamburg, M.D., Commissioner
Food and Drug Administration
Department of Health and Human Services
WO 2200
10903 New Hampshire Avenue
Silver Spring, MD, 20993

Dear Dr. Hamburg,

We urge you to immediately order the cessation of the unethical Thiazolidinedione Intervention in Vitamin D Evaluation (TIDE) trial. TIDE is a randomized trial that the FDA, in 2007, ordered GlaxoSmithKline (GSK), the manufacturer of rosiglitazone to conduct. A major objective of this trial is “to test the cardiovascular effects of long-term treatment with rosiglitazone or pioglitazone when used as part of standard of care,” with the implicit intent of testing whether rosiglitazone is more dangerous than pioglitazone. This trial now involves 137 sites in 14 countries, including 19 sites in the U.S and 34 in Canada. Presently, 83 of these sites are recruiting subjects for the trial, which has an anticipated sample size of 16,000 subjects and a targeted completion date in 2015. In an apparent attempt to increase enrollment, 53 new recruitment sites were recently added between the previous posting on March 31, 2010 and the updated posting on April 23. It is worth noting that these new sites include developing countries such as Chile, Mexico, Colombia, Latvia, Pakistan and India.[1]

Our request that the trial be terminated is based on accumulating evidence of the increased dangers of rosiglitazone (Avandia) relative to pioglitazone (Actos) and in comparison to older, standard oral treatments for diabetes. In 2008, based upon even less unfavorable data on rosiglitazone than presently exists on this issue, the American Diabetes Association and the European Diabetes Association advised that “given that other options are now recommended, the consensus group members unanimously advised against using rosiglitazone.”[2]

We will briefly review the evidence of increased risk of rosiglitazone and conclude with four reasons why the TIDE trial must be terminated before more patients are harmed.


The thiazolidinediones (TZDs) rosiglitazone and pioglitazone are approved for type 2 diabetes. This approval was based solely upon their ability to modestly improve surrogate markers such as blood glucose in placebo controlled trials. This effect is mainly mediated by the binding of these drugs to their target - a nuclear receptor termed “PPARgamma” - in fat, muscle and the liver. However, this same receptor is widely distributed throughout other body tissues, and these drugs have been linked to a host of undesirable adverse effects, many of which are thought to reflect stimulation of PPARgamma receptors elsewhere in the body. Since their regulatory approval, both drugs have been linked to a broad spectrum of potentially serious adverse events including macular edema, anemia, edema, congestive heart failure, myocardial infarction and bony fractures.[3-9] Like their predecessor troglitazone (which was removed from the UK market in 1997 and the U.S. market in 2000), rosiglitazone and pioglitazone have also been associated with cases of acute liver injury, some of which have been fatal.

The majority of deaths among patients with type 2 diabetes result from cardiovascular disease. This is an important observation, because there is no convincing data from clinical trials that the TZDs lessen the risk of cardiovascular disease. Indeed, these drugs gained approval purely based upon their ability to lower blood glucose, and achieved “blockbuster” status in large measure because they were marketed as drugs with the ability to lessen insulin resistance - a common feature of type 2 diabetes. Regrettably, there is no convincing evidence from clinical trials that either drug lowers the risk of vascular complications of type 2 diabetes.

In contrast, strong evidence suggests that both drugs cause edema and congestive heart failure. The risk of CHF during pioglitazone therapy appears to be on the order of about 40 percent higher than placebo, as seen in the PROactive study (relative risk 1.4),[9] while the risk with rosiglitazone appears to be significantly higher than standard diabetes therapy (metformin and sulfonylurea) based on evidence from the RECORD study (relative risk 2.1).[3] Moreover, in the RECORD study, there was a significant increase in the risk of heart failure hospitalizations and cardiovascular death in patients who developed heart failure after being randomized to rosiglitazone.[3]

The notion that rosiglitazone is more likely than pioglitazone to cause heart failure is supported by strong biologic plausibility and also by several other meta-analyses of randomized controlled trials and observational studies. FDA scientists estimate that with the risk of heart failure with rosiglitazone is roughly 50 percent higher than with pioglitazone. Importantly, rosiglitazone-induced heart failure is also accompanied by an excess risk of death relative to pioglitazone. This is evident in a large population-based cohort study led by one of us (Dr. Juurlink), comparing outcomes in 39,000 people using either rosiglitazone or pioglitazone.[7] The authors found major differences in the risk of congestive heart failure and death from any cause in patients taking rosiglitazone as compared to those taking pioglitazone. The authors estimated that one additional hospitalization for heart failure would occur annually for every 120 patients prescribed rosiglitazone rather than pioglitazone, and that one additional death would occur each year for every 269 patients treated with rosiglitazone rather than pioglitazone.[7] At the population level, this translates into many thousands of additional adverse outcomes resulting from the use of rosiglitazone rather than pioglitazone. Other well-conducted observational studies examining the relative safety of the TZDs have also found a higher mortality with rosiglitazone.[10,11] 

Rosiglitazone and pioglitazone also appear to differ in their ability to provoke myocardial ischemic events, with several studies suggesting that rosiglitazone carries a higher relative risk of myocardial infarction than pioglitazone. Several meta-analyses of clinical trials, including those conducted by the manufacturer of rosiglitazone, the FDA and several independent scientists have confirmed that rosiglitazone increases the risk of myocardial infarction, whereas pioglitazone is at best neutral or perhaps somewhat protective.

In July of 2007, an FDA advisory committee agreed that evidence suggested that rosiglitazone increased the risk of angina and myocardial infarction, but the committee did not believe the drug should be removed from the market. Subsequently, the FDA ordered a black box warning concerning these risks for rosiglitazone, but not for pioglitazone. In 2010, even more data are available.[7] Although both drugs share similar safety concerns, a large and growing body of evidence suggests that rosiglitazone imparts greater cardiovascular risks than pioglitazone, and some have speculated that this reflects the fact that rosiglitazone is a more potent PPARgamma agonist. Not a single study, however, has suggested that pioglitazone might carry more risk. In the context of the TIDE trial, it is difficult to overstate the importance of this observation

In light of the growing evidence that rosiglitazone imparts greater cardiac risk than pioglitazone yet offers no particular advantage, the Saudi Arabian drug regulatory agency has removed rosiglitazone from the market. In the United States, the VA formulary and other insurers have placed restrictions on the use of rosiglitazone. Many physicians have already switched from rosiglitazone to pioglitazone as these warnings have slowly percolated into practice, with four times as many prescriptions written for pioglitazone in 2008 than for rosiglitazone.[12] As noted earlier, the American Diabetes Association and the European Association for the Study of Diabetes have effectively indicated that rosiglitazone has no role in the present-day management of type 2 diabetes, uniformly recommending against its use.

The TIDE trial

We believe the TIDE trial is unethical for several reasons:

1. Misplaced scientific objectives

A primary purpose of the TIDE trial is to establish with certainty whether or not rosiglitazone is indeed more dangerous than pioglitazone. However, there are now well-documented differences in cardiovascular risks between rosiglitazone and pioglitazone, demonstrated in several studies conducted in the United States, Canada and the UK. The TIDE trial defies a basic tenet of clinical trial design - that trials should be conducted to determine the balance of risk and benefit and not simply to provide absolute proof of harm. Because rosiglitazone has no safety or efficacy advantage - not even a theoretical one - over pioglitazone, and because a wealth of data now suggests rosiglitazone carries greater risks than pioglitazone, it is not possible to advance a cogent argument that this trial is ethical given the present state of evidence.

2. Absence of clinical equipoise

Clinical equipoise requires that no subject receive an intervention known to be inferior to current standards of care.[13,14] RCTs are justified in cases in which the expert scientific community is unsure about the comparative merits of interventions, and there should be equivalent evidence for the two interventions.[14] This is clearly not the case in the context of rosiglitazone and pioglitazone. Several guidelines and systematic reviews (the highest level of evidence) have all demonstrated that rosiglitazone has an inferior safety profile relative to pioglitazone, and the positions of the ADA and EASD support this.

3. Unfavorable balance of risks and benefits of rosiglitazone

The balance of risks and benefits on rosiglitazone is clearly unfavorable. A priori the risks of harm with rosiglitazone are substantial, and it is highly unlikely and statistically improbable that patients in the rosiglitazone arm will derive any additional clinical benefit beyond that provided by pioglitazone. There are several different classes of drugs available to treat patients with type 2 diabetes that do not carry these risks.

4. Mischaracterization of risk in the informed consent form

The sine qua non of an ethical trial is that intervention-specific risks are made available to participants so that they may make an informed decision regarding participation. The consent form of the TIDE trial addresses the safety concerns regarding TZDs, but does not effectively differentiate between the two drugs. Indeed, the informed consent form co-mingles the drugs, obscuring the fact that a wealth of data now suggests that one treatment arm is thought to have an inferior safety profile. Surely no patient would willingly participate in a trial in which they have a substantial likelihood of being randomized to a drug that, in the opinion of a large group of experts, has no role in present day therapeutics because of its inferior safety profile. Unfortunately, prospective study participants are denied the ability to make an informed decision because the TIDE consent form does not present an accurate portrayal of the existing safety data.

Moreover, it remains unclear if the site-specific Institutional Review Boards (IRBs) responsible for authorizing the TIDE had access to complete information at the time of the study’s initial approval. We anticipate that most IRBs would not permit the continuation of the trial if they were fully apprised of the ongoing safety concerns.


The TIDE trial continues to recruit patients despite a lack of clinical equipoise, exposing thousands of high-risk patients with diabetes to a drug with an unfavorable safety profile and no clinical advantage over its comparator. Prospective study subjects are deprived of the opportunity to make a fully informed decision because the consent form does not present an accurate portrayal of existing safety concerns. It is difficult to imagine that a patient would willingly participate in a trial involving a drug that, according to the American Diabetes Association and its European equivalent, has safety concerns that leave it with no present-day role in the management of type 2 diabetes. The TIDE trial can only continue with the misplaced objective of providing definitive proof of what many studies have already suggested: that rosiglitazone is indeed more dangerous than pioglitazone. The price of such definitive proof will almost certainly be measured in the lives of study subjects who have been incompletely informed about the risks and benefits of participation.

We respectfully request a timely response to this letter, and we hope that you agree that simply waiting to act until the FDA advisory committee meeting scheduled for July guarantees that more patients - now including those in developing countries - will be exposed to and very possibly injured by a drug that, for the reasons we have outlined, should not be in use.


David N. Juurlink M.D., Ph.D., FRCPC, FACMT, FAACT
Head, Division of Clinical Pharmacology and Toxicology
Sunnybrook Health Sciences Centre
Scientist, Institute for Clinical Evaluative Sciences
Medical Toxicologist, Ontario Poison Centre
Toronto, Ontario

Sidney M. Wolfe, M.D.
Director, Public Citizen’s Health Research Group
Washington, DC

cc: Congresswoman Rosa DeLauro
 Senator Charles Grassley


1. Clinicaltrials.gov, accessed 4/26/10 (4/23 posting).
2. Diabetologia. 2009; 52(1):17-30.
3. Eur Heart J. 2010;31(7):824-31
4. N Engl J Med. 2007;356(24):2457-71
5. JAMA. 2007;298(10):1189-95
6. CMAJ. 2009;180(1):32-9.
7. BMJ. 2009;339:b2942
8. Am J Ophthalmol. 2009;147(4):583-586.
9. Lancet. 2005; 366(9493):1279-89.1.
10. JAMA 2007;298(22):2634-43.
11. Arch Intern Med 2008;168(21):2368-75
12. Drug Topics data, accessed 4/26/10
13. Kennedy Institute of Ethics Journal 17.4 (2007): 321-350.
14. NEJM 1987;317(3):141-5.