ADA 2009: Avandia on the RECORD: No "Overall" CV Risk Increase, but Trial Remains Controversial

June 08, 2009

June 8, 2009 (New Orleans, Louisiana) — With something to offer both fans and critics of the thiazolidinedione in type 2 diabetes, a randomized trial with >4000 patients has concluded that rosiglitazone (Avandia, GlaxoSmithKline), added to standard glucose-lowering monotherapy, poses no more of an overall cardiovascular risk than a combination of metformin and sulfonylurea [1]. As found in previous studies, limb fractures and heart failure were more likely with rosiglitazone.

The trial, called Rosiglitazone Evaluated for Cardiac Outcomes and Regulation of Glycemia in Diabetes (RECORD), took direct aim at key issues that have dogged the highly effective antiglycemic agent for several years, including an onerous side-effect profile in some patients and, especially, published evidence that it might increase the risk of MI [2]. Some observers say the trial helps define rosiglitazone's optimal place in type 2 diabetes management, whereas others say it has methodologic limitations that keep it from proving much of anything.

Primary results of the trial, with lead author Prof Philip Home (Newcastle University, Newcastle upon Tyne, UK), were published online in the Lancet today to coincide with its live presentation here at the American Diabetes Association 2009 Scientific Sessions.

We find it difficult to see reasons for distinguishing rosiglitazone from other oral antiglycemic agents. They all have some problems.

In the trial's 4447 patients with type 2 diabetes, who were randomized to receive either rosiglitazone (on top of either metformin or sulfonylurea) or a metformin/sulfonylurea combination, the rate of CV hospitalization or death from cardiovascular causes was 14.5% in both groups over a mean follow-up of 5.5 years, Home et al report. The rosiglitazone-vs-active-control hazard ratio (HR) for that primary end point met RECORD's prospectively stated criteria for noninferiority.

There were no significant differences between the groups in risk of the individual end points of CV death, MI, or stroke, although the HR for MI was elevated by a nonsignificant 15%. The authors acknowledge that RECORD "had limited statistical power for individual components of the primary end point because it was never intended to answer these questions."

Speaking with heartwire , Home said that "the cardiovascular profile [of rosiglitazone] was no different overall from metformin/sulfonylurea. On that basis, we find it difficult to see reasons for distinguishing rosiglitazone from other oral antiglycemic agents. They all have some problems, and it can be very effective in some people. There seem to be no reasons to be especially cautious."

The authors do acknowledge many of the trial's limitations cited by critics; it was an open-label study, for example, and patients treated with rosiglitazone also received more loop diuretics and statins--the latter driven by rosiglitazone-associated increases in LDL cholesterol.

RECORD Outcomes, Mean 5.5-Year Follow-Up, Hazard Ratio (HR) for Rosiglitazone vs Metformin/Sulfonylurea

End points HR (95% CI)
CV hospitalization or deatha 0.99 (0.85–1.16)
CV death 0.84 (0.59–1.18)
MI 1.14 (0.80–1.63)
Stroke 0·72 (0·49–1·06)
Heart failure death or hospital admission 2.10 (1.35–3.27)b
a. Primary end point, meets prespecified criteria for noninferiority

b. p=0.001

Interpretations of the trial's outcomes, which are largely consistent with those of a nonprespecified interim analysis published in 2007 [3], range widely. RECORD's investigators and others say the trial shows the drug can be used cautiously and selectively. Others say it shows the drug is riskier than it's worth, at least in diabetics with cardiovascular disease, or blast the trial's methodology and discount the results.

Those views aren't new. As extensively covered by heartwire , the drug's credibility as a safe agent came under attack most prominently in 2007, when a published meta-analysis suggested a 43% increase in MI risk and a 64% rise in CV death risk among patients taking rosiglitazone [2]. The story deepened with developments that included document leaks, a congressional investigation, mea culpas, and an FDA slammed for lax regulation of the drug.

For their part, the RECORD researchers contend that, based on their study, "rosiglitazone is not recommended for people with a history of heart failure or with previous problems that might have led to myocardial dysfunction. Rosiglitazone should be used with caution in women at high risk of fractures."

As they reported, the risk of any bone fracture was increased with rosiglitazone by 57% (p<0.0001) overall and by 82% in women and 23% in men. Fractures of the upper limbs or distal lower limbs predominated.

"Although our evidence is insufficient to rule out a small increased risk of myocardial infarction caused by rosiglitazone when compared with other glucose-lowering agents, rosiglitazone does not increase overall cardiovascular morbidity or mortality."

It is impossible to assess the safety of a drug when patients are not actually taking it.

An accompanying editorial describes a cautious, measured approach to using the drug [4]. "It is generally accepted that the use of half-maximal doses could provide more than half-maximal therapeutic effects while limiting the magnitude of unwanted side effects," write Drs Ravi Retnakaran and Bernard Zinman (University of Toronto, ON). "Although no studies have tested this strategy, we suggest that low-dose thiazolidinediones, in combination with metformin, might provide glycemic durability with lower risk. Indeed, this combination therapy is currently being assessed for the prevention of diabetes in individuals with impaired glucose tolerance [a reference to the ongoing Canadian Normoglycaemia Outcomes Evaluation (CANOE) trial]. If the efficacy of this strategy is confirmed, we might be able to find the optimal way to use this class of medications in the treatment of type 2 diabetes."

However, Dr Steven E Nissen (Cleveland Clinic Foundation, OH), who coauthored the 2007 meta-analysis that claimed an untoward risk of MI from rosiglitazone [2], told heartwire today that "the RECORD trial is seriously flawed." For starters, he said, its interim analysis, which was published after his meta-analysis, notes that 30.2% of its patients were no longer on the treatment to which they were randomized. (The published rates by treatment group were 26.8% for those in the rosiglitazone group and 33.7% for those in the active-control group.)

The current RECORD data in the Lancet, Nissen observed, don't include the corresponding numbers at the trial's end. But he said that GlaxoSmithKline has publicly acknowledged that they are 40% and 50% for those in the rosiglitazone and control groups, respectively.

"Obviously, it is impossible to assess the safety of a drug when patients are not actually taking it," he said.

Home said to heartwire that such numbers are relevant in an efficacy trial, but "because this is a safety study, what matters is the percentage of time that people within the study are exposed to their randomized drug. And for a five-and-a-half year study, it was quite good, actually; it was between 80% and 90% for both groups."

Nissen also noted that "significantly more patients in the Avandia-treatment group received statins, a class of drugs known to reduce risk of coronary events. Essentially, the study sought to neutralize the unfavorable lipid effects of the drug by administering statins more often to patients in the Avandia group. This imbalance biases the study in favor of Avandia and distorts the results."

As Home et al report, the rate of statin usage increased from 18% at baseline to 55% at five years among patients in the rosiglitazone group and from 19% to 46% among those in the control group. Loop-diuretic use also went up more sharply with rosiglitazone, from 3.1% to 13%, compared with 3.1% to 8.1% among controls.

Despite these issues, "notably, in patients with preexisting heart disease, there was a 26% higher risk of heart attack that fell short of statistical significance because of the small size of the study," according to Nissen, referring to the HR for the primary end point of 1.26 (95% CI 0.95–1.68; p=0.055) among patients with previous ischemic heart disease in a prespecified subgroup analysis.

"I agree with the authors and editorial writers that the results of RECORD are inconclusive with respect to the effects of the drug on the risk of heart attack."

RECORD was sponsored by GlaxoSmithKline. Home and six other coauthors "or the institutions with which they are involved receive funding for research, educational, and/advisory activities from pharmaceutical companies, including GlaxoSmithKline, and in some cases from the manufacturers of sulfonylureas and metformin preparations and other competing products." Two of the report's coauthors are noted as employees of and stockholders in GlaxoSmithKline. Zinman discloses research support and scientific advisory board honoraria from GlaxoSmithKline and scientific advisory board and speaking honoraria from Eli Lilly. Retnakaran had no disclosures. Nissen has disclosed receiving research support through his institution from Pfizer, AstraZeneca, Novartis, Roche, Daiichi-Sankyo, Takeda, Sanofi-Aventis, Resverlogix, and Eli Lilly and reports that he consults for many pharmaceutical companies and requires them to donate all compensation directly to charity.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.