From the 67th annual scientific sessions of the American Diabetes Association: Researchers debate the CV safety of rosiglitazone

Rosiglitazone did not demonstrate an increased risk of hospitalizations and death from cardiovascular causes (the composite end point) in an interim analysis of a prospective study.

Key Points

A special symposium at the 67th Annual Scientific Sessions of the American Diabetes Association that took place June 22 to June 26, 2007, in Chicago, Illinois, focused on the cardiac effects of rosiglitazone. The main speakers were the lead researchers of 2 studies assessing the drug's safety. Steven E. Nissen, MD, chairman of cardiovascular medicine at Cleveland Clinic, Ohio, conducted a meta-analysis (N Engl J Med. 2007;356:2457–2471) in which rosiglitazone was demonstrated to increase the risk of myocardial infarction (MI). Philip Home, DM (doctor of medicine), professor of diabetes medicine, University of Newcastle upon Tyne, UK, conducted the Rosiglitazone Evaluated for Cardiac Outcomes and Regulation of Glycaemia in Diabetes (RECORD) trial and recently published the results of an interim analysis of the prospective study (N Engl J Med. 2007;357:28–38) in which rosiglitazone did not demonstrate an increased risk of hospitalizations and death from cardiovascular causes (the composite end point).

Dr Nissen reviewed the chronology that led to the FDA review of the safety of rosiglitazone on July 30. He defended the results of his analysis of 42 randomized trials, most of them unpublished, that demonstrated a 43% relative excess hazard of MI with use of rosiglitazone compared with controls. According to Dr Nissen, these results are in line with the results of a meta-analysis conducted by GlaxoSmithKline (rosiglitazone's manufacturer), which were provided to FDA in 2006. Rosiglitazone was approved by FDA in 1999 for the treatment of type 2 diabetes. Nissen said "there was a signal from the very beginning" that rosiglitazone could represent a cardiovascular hazard. Studies presented to FDA in 1999 demonstrated an 18% increase in low-density lipoprotein cholesterol (LDL-C) among patients treated with rosiglitazone. Although a postmarket study was requested as a condition of approval, no trials assessing major cardiovascular outcomes were conducted.

In pooled registration trials and clinical trials that defined glycemic durability as an end point, hazard ratios for MI were consistently higher (33%–80%) with rosiglitazone versus the agent's comparators, according to Dr Nissen.

Subsequent to this meta-analysis, an observational study performed by GlaxoSmithKline demonstrated no excess cardiac risk with the use of rosiglitazone, but Dr Nissen pointed out several flaws in this study. He said it used an insurance claims database with an incomplete assessment of covariates, followed patients for only 1 year, and included no comparison to the other marketed thiazolidinedione, pioglitazone. "In my view, this study was very weak," Dr Nissen said.

Dr Home accused Dr Nissen of "data snooping on quite a big scale."

In reaction to Dr Nissen's meta-analysis, Dr Home released unplanned 3.75-year interim data from his RECORD study. In this analysis, use of add-on rosiglitazone was not demonstrated to be associated with an increased risk of hospitalizations or death from cardiovascular causes compared with a combination of metformin and a sulfonylurea.

Dr Home said that, based on these interim data, rosiglitazone should maintain a role in the glucose-lowering armamentarium and "RECORD should continue to its planned [6-year] end."

Dr Nissen disagreed: "Major errors in design have rendered this study futile." Dr Nissen noted numerous weaknesses of the RECORD trial, including the open-label design and the end point, which did not include MI.

Another weakness, according to Dr Nissen, is that the interim annual event rate in the RECORD trial was 3.1%, which was far lower than the 11% annual event rate that the authors postulated. Dr Nissen said this erroneous assumption renders the study underpowered to detect a difference between rosiglitazone and its comparators. "The study has less than 1% power to detect a difference in mortality," he said.

During the discussion, Dr Nissen was accused of undermining the RECORD trial, as some of the study participants withdrew after the results of his meta-analysis were published.

"The alternative [of] keeping the scientific [community] in the dark was unacceptable," Dr Nissen said. "We had an ethical and moral obligation to release the data. The scientific community has to decide for itself what to do with the data. When the signal [of excess cardiac risk] emerged, there were other routes by which this information could have been shared with the medical community-those routes failed."

Panel member David M. Nathan, MD, professor of medicine at Harvard Medical School, Boston, Massachusetts, said that drugs such as rosiglitazone are used to lower HbA1c levels in an effort to avoid or delay microvascular complications. When a member of the audience questioned him as to whether thiazolidinediones have been demonstrated to reduce the development of microvascular complications, Dr Nathan said that there is no evidence that they do.

Dr Nissen said each agonist of a peroxisome proliferator-activated receptor (PPAR) needs to be evaluated individually because each activates different genes. Pioglitazone, for example, has favorable effects on lipids, unlike rosiglitazone, and demonstrated a trend towards a reduction in a composite of death, MI, and stroke in the Prospective Pioglitazone Clinical Trial in Macrovascular Events (PROactive). According to Dr Nissen, the development pathways of >50 PPAR agonists have already been terminated due to toxicity; this toxicity has usually been direct cardiac toxicity.