Rosiglitazone: one year later

Conclusions about rosiglitazone remain unclear a year after a controversial meta-analysis set off alarms.

After a meta-analysis of rosiglitazone was published in The New England Journal of Medicine in June 2007, the mainstream media offered extensive coverage of the reported adverse cardiovascular effects of this diabetes medication. The meta-analysis suggested an increased risk of myocardial infarction and death from cardiovascular disease, and with little delay, sales of the drug began to decline.

In the community of endocrinologists and clinical trial methodologists, however, a separate controversy brewed. Just what could be learned from that meta-analysis? How good were the data, and were the methods used sound? To date, further analyses have been performed, but with little new data to add, the controversy remains. Some argue that the drug’s benefits outweigh the possibility of cardiovascular risk, while others maintain the potential risk signal is enough to move away from prescribing rosiglitazone. To understand more firmly where the field sits requires following the rosiglitazone story from its beginning.

Rosiglitazone was the second entry in the class of thiazolidinediones, following the since-pulled troglitazone; the drug was granted FDA approval in 1999. The authors of last year’s meta-analysis, Steven Nissen, MD, and Kathy Wolski, MPH, wrote that the drug’s approval was based on studies that were “not adequately powered to determine the effects of this agent on microvascular or macrovascular complications of diabetes, including cardiovascular morbidity and mortality.” They undertook their meta-analysis because of that, and because they felt that hints of cardiovascular risk had been seen before. Their analysis included many smaller trials as well as the larger-scale DREAM and ADOPT trials, 42 trials in all.

The meta-analysis found that patients in the rosiglitazone group (n=15,565), when compared with the control group (n=12,282), had an odds ratio for MI of 1.43 (95% CI, 1.03-1.98). The odds ratio for death from cardiovascular causes was 1.64 (95% CI, 0.98-2.74).

“When you have the No. 1 diabetes drug in the world that increases risk of ischemic heart disease — the most common cause of death in diabetics — by 40%, you have a real problem,” Nissen told Endocrine Today.

Comments, editorials and re-analyses were published and presented almost immediately, with many pointing out that a meta-analysis does not carry the weight of a prospective, randomized controlled trial and that the confidence intervals made the result unconvincing. The NHLBI was prompted to ask the Data and Safety Monitoring Boards of the ACCORD and BARI 2D trials, two trials involving rosiglitazone, to conduct safety reviews of each. The NHLBI released a statement saying that “the Data and Safety Monitoring Boards of both ACCORD and BARI 2D found no evidence to require discontinuing the use of rosiglitazone in the trials or to revise the study protocols.”

Sanjay Kaul, MD MPH, director of the Vascular Physiology and Thrombosis Research Laboratory at the Burns and Allen Research Institute at Cedars Sinai Medical Center in Los Angeles, and colleagues conducted another analysis of the same data to illustrate some of the methodological problems with the initial analysis.

“There was sufficient heterogeneity in clinical trial design and in the patient populations and treatment protocols that they did not lend themselves to pooling,” Kaul said. “We believe that the quality of the data are not robust, as none of the trials were designed primarily to assess cardiovascular endpoints.”

He added that the short follow-up of most of the trials led to a sparse data set and that the analytical methodology used was incomplete and potentially biased against rosiglitazone.

“All of these issues led us to the conclusion that there is sufficient uncertainty that we can’t adjudicate the association between rosiglitazone and cardiovascular risk,” he said.

The FDA intervenes

Sanjay Kaul, MD, MPH, Director of Vascular Physiology and Thrombosis Research Laboratory at the Burns and Allen Research Institute at Cedars Sinai Medical Center. Photo by Bill Pollard Courtesy of Cedars-Sinai Medical Center

Others disagreed, however, and an FDA joint Endocrinologic and Metabolic Drugs/Drug Safety and Risk Management Advisory Committee meeting was convened in July to discuss the new information. Meta-analyses conducted by the FDA and by GlaxoSmithKline arrived at similar conclusions as the original meta-analysis, and the FDA committee members agreed, in a 20-3 vote, that the data indicated that rosiglitazone does increase cardiac ischemic risk. They also agreed, in a 22-1 vote, that the evidence was not conclusive enough to merit pulling the drug from the market. Warning labels were added, but the drug remains available.

“I thought that sounded odd,” said Bruce M. Psaty, MD, PhD, a professor of medicine and epidemiology at the University of Washington in Seattle who wrote an editorial on the topic in NEJM in July 2007. “It was not clear to me how you would conclude that it increases risk and yet not pull it from the market.” He pointed out that a GlaxoSmithKline analysis had been submitted to the FDA as far back as October 2005 that indicated the possibility of an increased risk, and yet nothing had been done until Nissen and Wolski’s paper was published.

“The main feature that you want to see in a well-functioning FDA is timely information about warnings and risks, and timely withdrawals. It did not seem to me that these things happened in a timely fashion,” he said.

RECORD interim results

Also in July, investigators with the RECORD study published interim results of the trial to provide regulatory agencies, physicians and the public with all available information. The authors acknowledged the limited statistical power of the analysis due to a shorter follow-up than the trial will eventually have, but they concluded that there was no significantly increased risk indicated. The hazard ratio for death or hospitalization related to cardiovascular causes (adjudicated events only) was 1.08 (95% CI, 0.89-1.31; P=.43).

David M. Nathan, MD, of Harvard Medical School and the Diabetes Center at Massachusetts General Hospital in Boston, wrote an editorial discussing the RECORD study’s potential to answer the ongoing questions about rosiglitazone. He told Endocrine Today that the study “appears to be terribly under-powered. You can look at [the hazard ratios] and say that it is not statistically significant, and of course these were interim results, but the fact that it is underpowered and that the current results are trending in the wrong direction make me think that the study is not going to answer the question.”

Controversy continues

After decreases in rosiglitazone sales — from $617 million worldwide in the first quarter of 2007 to $327 million in the fourth quarter — the controversy sits roughly where it sat after the FDA advisory committee meeting last year.

“It would be great to have clinical trials that are definitive, but my guess is that they won’t be done,” Nathan said. “We will be left with this uncertainty until the drug just fades away or the FDA gets other data, but I don’t know what other data they can find other than good clinical trials to answer this question.”

One problem with the meta-analyses is a lack of quality data, according to Nancy Geller, MD, of the NHLBI. At the FDA meeting, Geller said that “the quality of the meta-analyses far exceeds the quality of the data that went into the meta-analyses. Time-to-event analyses were not included, which would have added statistical power and allowed for hazard ratio calculations.

Furthermore, many of the studies included in the analyses had few or even zero adverse events. Kaul noted the importance of the imbalance between zero event rates in the rosiglitazone arm vs. the control arm.

“Whenever you have imbalances in the zero event trials the risk estimates are biased toward the control and against the treatment arm,” she said. By correcting for that fact in his analysis, the suggestion of risk was diminished.

Nissen, however, has been adamant that whether or not the risk exists is not in question.

“You have three analyses: the company’s own showing a 31% increase in risk, the FDA’s showing approximately a 40% increase in risk, and our analysis showing approximately a 40% increase in risk,” he said in an interview. “In my view, its really not controversial whether the drug increases the risk of heart attack and related complications. What is controversial is what to do about it.”

The American Diabetes Association and the European Association for the Study of Diabetes released an update to the consensus algorithm for the management of hyperglycemia in January 2008 to include information on TZDs. The update concluded that although the available data are not conclusive and thus the drugs should not be removed from the algorithm, the possibility of increased risk of MI with rosiglitazone should engender increased caution in physicians’ prescribing practices. They also noted that all TZDs appear to be associated with increased risk for heart failure and edema, both of which are included in the FDA’s drug warning labels.

One reason that further large-scale studies of rosiglitazone may not be started is that another TZD option exists. Pioglitazone (Actos, Takeda Pharmaceuticals), approved in 1999, has shown to be effective at lowering HbA1c levels, and no CVD risk signal has been found at this point. One observational analysis of 29,911 patients found that pioglitazone resulted in a 22% lower rate of MI than rosiglitazone (adjusted HR, 0.78; 95% CI, 0.63-0.96) and a 15% lower rate of MI and coronary revascularization combined (adjusted HR, 0.85; 95% CI, 0.75-0.98). Furthermore, TZDs are not considered first-line therapy, and targets can often be met with lifestyle intervention, metformin, insulin and sulfonylureas.

Many physicians continue to prescribe rosiglitazone and argue that the drug’s potential benefits to a growing population of diabetic patients outweigh its risks.

“I continue to prescribe it. We don’t know what [the drug’s] effect on cardiovascular events is; the data that have been published do not provide a clear answer for that, neither for rosiglitazone nor for pioglitazone,” said Hertzel Gerstein, MD, a professor in the division of endocrinology and metabolism at McMaster University in Hamilton, Ontario.

In a muddying of the diabetes management picture, the recent interim results of the ACCORD study found that intensively lowering HbA1c levels by any means significantly increases the risk of MI and stroke. While the study remains ongoing, the intensive glycemic control arm has been discontinued. There was a difference of three deaths per 1,000 participants each year between the group with an HbA1c target of <6% and the standard therapy group with a target of 7% to 7.9%. The link was not associated specifically with rosiglitazone (or any specific therapy) but with the target in general.

“The ACCORD findings are important, but will not change therapy for most patients with type 2 diabetes,” Judith Fradkin, MD,director of the Division of Diabetes, Endocrinology and Metabolic Diseases at the National Institute of Diabetes and Digestive Kidney Diseases, said in a press release. “Few patients with high cardiovascular risk like those studied in ACCORD are treated to blood sugar levels as low as those tested in this study.”

Avoiding the situation

Looking back on the entire rosiglitazone story, it seems pertinent to wonder if the process could have been handled better — by researchers, journals and the mainstream media — from the beginning. Kaul, whose analysis suggested the lack of good data undermined the risk signal, said that a dialogue could have been dealt with in less inflammatory fashion.

“Discussion needed to take place, because out of this discussion will come rational resolution. But unfortunately, the media ran with it, and that was very unfortunate. The discussion is best served in an academic environment rather than in the media. The media always loves the black and white headlines, but unfortunately the reality is always gray,” he said.

Gerstein said that while it may be understandable for the public to jump to conclusions about the risks of a drug, it is always important to wait for conclusive evidence one way or another.

“It’s natural for people to want to come to a final judgment about a lot of the things that we do,” he said. “When you look at all the information, however, it really is not sufficient to come to a clear conclusion. Obviously, if I thought this drug was clearly a harmful drug then I would not prescribe it to anybody. It has benefits and it has risks, which we discuss with patients and make a decision.”

Some experts think that the FDA might have been able to divert the media catastrophe by earlier handling of the information that appeared to be available before Nissen and Wolski’s paper was published.

David M. Nathan

“If they knew this, why didn’t they act earlier?” Nathan asked. “If there was a hint of risk, or real risk, or even just rumors, they are the ones who are supposed to be keeping an eye on this, and yet as far as I can tell they didn’t do a thing. It was going to be an inflammatory story no matter how it played out. The investigators don’t control the media.”

Many agree that whatever conclusions may ultimately be drawn from the rosiglitazone story, they have yet to become clear. It has, however, added another chapter into this country’s ongoing regulatory controversies. Some experts are calling for the FDA to require companies to do phase-4 safety testing following drug approval. In fact, the FDA did ask GlaxoSmithKline for such a study, but it ended up as the efficacy trial ADOPT.

While physicians await the final RECORD analysis and ponder the surprising ACCORD interim results, the only thing that is clear is that the landscape of diabetes management continues to shift.

“What we are all trying to do is find drugs that help our patients achieve a reasonable glucose goal safely and with as few side effects as possible,” Gerstein said. “There are a lot of things about this class of drugs that suggest some benefits. Unfortunately, the story is not completely clear yet.”– by Dave Levitan

For more information:

• Nissen SE, Wolski K. Effect of rosiglitazone on the risk of myocardial infarction and death from cardiovascular causes. N Eng J Med. 2007;356:2457-2471.
• Diamond GA, Bax L, Kaul S. Uncertain effects of rosiglitazone on the risk for myocardial infarction and cardiovascular death. Ann Intern Med. 2007;147:578-581.
• Summary minutes of the joint meeting of the Endocrinologic and Metabolic Drugs Advisory Committee and the Drug Safety and Risk Management Advisory Committee. July 30, 2007. www.fda.gov/ohrms/dockets/ac/07/minutes/2007-4308m1-final.pdf.
• Psaty BM, Furberg CD. The record on rosiglitazone and the risk of myocardial infarction. N Eng J Med. 2007;357:67-69.
• Division of Metabolism and Endocrine Products and Office of Surveillance and Epidemiology, FDA. Briefing document for the joint meeting of the Endocrinologic and Metabolic Drugs Advisory Committee and the Drug Safety and Risk Management Advisory Committee. July 30, 2007. www.fda.gov/ohrms/dockets/ac/07/briefing/2007-4308b1-02-fda-backgrounder.pdf.
• Gerrits CM, Bhattacharya M, Manthena S, et al. A comparison of pioglitazone and rosiglitazone for hospitalization for acute myocardial infarction in type 2 diabetes. Pharmacoepidemiol Drug Saf. 2007;16:1065-1071.
• GlaxoSmithKline quarterly results. www.gsk.com/investors/quarterly_results.htm.
• Home PD, Pocock SJ, Beck-Nielsen H, et al. Rosiglitazone evaluated for cardiovascular outcomes – an interim analysis. N Eng J Med. 2007;357:28-38.
• Nathan DM, Buse JB, Davidson MB, et al. Management of hyperglycemia in type 2 diabetes: a consensus algorithm for the initiation and adjustment of therapy: update regarding TZDs. Diabetes Care. 2008;31:173-175.
• Nathan DM, Buse JB, Davidson MB, et al. Management of hyperglycemia in type 2 diabetes: a consensus algorithm for the initiation and adjustment of therapy. Diabetes Care. 2006;29:1963-1972.
• Nathan DM. Rosiglitazone and cardiotoxicity – weighing the evidence. N Eng J Med. 2007;357:64-66.
• As quoted in: Wood, S. FDA advisory panel acknowledges signal of risk with rosiglitazone but stop short of recommending its withdrawal. Heartwire. July 30, 2007. www.theheart.org/article/804403.do