Rosiglitazone May Benefit CVD Patients
Published: June 28, 2010
* Explain that this study suggests a protective effect for rosiglitazone in patients with confirmed cardiovascular disease, but that finding was based on a post-hoc analysis that cannot prove causality.
* Note that this study was published as an abstract and presented at a conference. These data and conclusions should be considered to be preliminary until published in a peer-reviewed journal.
ORLANDO -- Rosiglitazone (Avandia) reduced major cardiovascular event risk in patients with co-existing type 2 diabetes and heart disease, according to a post-hoc analysis of the BARI 2D trial.
Those findings reported at the American Diabetes Association meeting presented a dramatically different picture than that which emerged from a 56-trial meta-analysis in the Archives of Internal Medicine and a FDA analysis of Medicare patients in the Journal of the American Medical Association released earlier in the day.
The BARI 2D trial -- sponsored by the National Institutes of Health, GlaxoSmithKline, maker of rosiglitazone, and other drug companies -- found a 29% reduced composite relative risk of death, MI, and stroke with the use of rosiglitazone versus no drug in the thiazolidinedione class (3.79 versus 5.81 per 100 patient-years, P=0.002).
This apparent protective effect of rosiglitazone, which was used at the physician's discretion as part of randomization to insulin sensitization versus insulin provision in the trial, remained significant after adjustment for baseline differences and use of other antidiabetic drugs (relative risk 0.72, P=0.01).
Lead author Richard G. Bach, MD, of Washington University in St. Louis, cautioned that the results were hypothesis-generating and could not prove that rosiglitazone was responsible for the effect.
If real, though, the results might be explained by the patient population studied since prior studies have included few patients with comorbid heart disease, he said in an interview.
Some clinicians here saw the BARI 2D data as a puzzling piece in the rosiglitazone controversy.
"This really confuses the atmosphere around the thiazolidinedione story, specifically is rosiglitazone a risk," said cardiologist Robert Eckel, MD, of the University of Colorado at Denver and a past president of the AHA.
He said he had begun to see it as a dying dog of a drug. But based on these data, "It may not be dead yet," Eckel said in an interview.
John Buse, MD, PhD, of the University of North Carolina in Chapel Hill, agreed, saying he wasn't sure what to make of this data. But he cautioned about the low level of evidence that this study represents.
"The study was a negative study," he said in an email. "This is now a post-hoc non-randomized comparison."
Prior meta-analyses and observational studies have dueled back and forth about the presence or lack of cardiovascular risk but to find cardiovascular outcome benefit is almost unheard of for an antidiabetic agent.
The new meta-analysis in the Archives by Steven E. Nissen, MD, and Kathy Wolski, MPH -- the same Cleveland Clinic team that initially blew the whistle on rosiglitazone in a 2007 meta-analysis -- indicated a consistent MI risk elevated by 28% to 39% across sensitivity analyses that was significant at the P=0.04 level in each case.
Cardiovascular death was not significantly linked to rosiglitazone in this meta-analysis.
Rosiglitazone also looked substantially worse than pioglitazone (Actos) in the JAMA paper led by Vioxx-whistleblower David J. Graham, MD, MPH, of the FDA Center for Drug Evaluation and Research.
In that analysis of Medicare data, rosiglitazone was associated with no greater acute MI risk than pioglitazone but significantly elevated risk of stroke (HR 1.27), heart failure (HR 1.25), death (HR 1.14), and the composite of all four endpoints (HR 1.18, 95% CI 1.12 to 1.23)
The risk attributable to rosiglitazone for the composite endpoint was 1.68 (95% CI 1.27 to 2.08) for excess events per 100 person-years of treatment compared with pioglitazone.
But in Bach's analysis of BARI 2D, the individual endpoints in the unadjusted analysis suggested a reduced risk of stroke in this group of diabetes patients with heart disease compared with no thiazolidinedione use (0.28 versus 0.77 per 100 patient-years, RR 0.36, P=0.008).
Notably, there were trends for reduced MI risk (2.16 versus 3.16 per 100 patient-years, RR 0.76, P=0.06) and death (1.88 versus 2.56 per 100 patient-years, RR 0.77, P=0.08) with rosiglitazone.
Adjustment left the stroke association (RR 0.36, P=0.02) but eliminated the links with MI (RR 0.77, P=0.15), heart failure (RR 1.17, P=0.35) and death (RR 0.83, P=0.29).
Asked to comment on the BARI-2D results, Nissen said the lack of a significant heart failure signal was troubling given that this is a well-documented effect for which both thiazolidinediones carry a boxed warning.
As expected from prior trials, though, fracture risk was high with rosiglitazone regardless of adjustment for other factors (RR 1.62, P=0.03).
The BARI 2D trial was primarily designed as a comparison of intensive medical therapy with revascularization either immediately or delayed until worsening stenosis.
As reported at the ADA meeting last year, the more aggressive revascularization strategy yielded no mortality advantage compared with intensive lifestyle, glycemic, and secondary prevention measures (88.3% versus 87.8% at five years, P=0.97).
The 2,368 patients with type 2 diabetes and stable ischemic heart disease were also randomized to insulin or insulin-sensitizers as part of medical therapy.
For insulin sensitizers, rosiglitazone was provided without charge and accounted for at least 90% of thiazolidinedione prescriptions.
Physicians were instructed to use metformin and rosiglitazone to target a glycosylated hemoglobin of 7.0% with additional agents used for those who couldn't reach at least 8.0%.
Not surprisingly, 992 of the patients received rosiglitazone at some point in the trial. And most was in combination with metformin. Of those in the insulin sensitization arm, 75% got metformin and 62% got a thiazolidinedione.
Bach pointed out that combination made it difficult to attribute the impact of treatment to rosiglitazone alone, calling this a major limitation aside from the trial design issues.
The fact that metformin was the combination agent may make interpretation even more difficult, noted David Nathan, MD, of Massachusetts General Hospital in Boston.
Supportive, though not conclusive, evidence has mounted to suggest that metformin may be good for heart disease, so the inability to separate out the effect of these two drugs is a critical, he said in an interview.
While metformin appeared to interact with rosiglitazone for the heart failure endpoint, it didn't appear to do so for the composite outcome, Bach noted, suggesting that it might not account for the protective effect seen.
Still, he acknowledged that the analyses couldn't rule out residual confounding.
The BARI trial began in 2001 but ran through 2009 after the rosiglitazone controversy arose, making some bias in prescribing likely during at least part of the trial.
Patients selected by physicians for rosiglitazone treatment did not appear to be a healthier group, with higher baseline hemoglobin A1c, longer duration of diabetes, and more albuminurea, but no significantly greater history of stroke or heart failure or prior revascularization.
In the end, these results may raise more questions than they answer.
Bach called the findings reassuring for high-risk patients currently receiving rosiglitazone in a manner similar to that in the trial but did not recommend initiating the drug on the basis of these results.
They are just one more additional source of information about the drug, he said at a press conference at which the results were released early ahead of presentation at the late breaking clinical trials session. This last minute change was the decision of the ADA in the face of the JAMA and Archives releases earlier in the day.
"The controversy goes beyond what any single trial can address," Bach said, but noted that the FDA has invited presentation of the data at its advisory committee meeting next month were the question of rosiglitazone's future is likely to be determined.
The BARI 2D study was supported by the NHLBI, NIDDK, GlaxoSmithKline, Lantheus Medical Imaging, Astellas Pharma US, Merck, Abbott Laboratories, Pfizer and others.
Bach reported recieving research grants from Bristol-Myers Squibb, Eli Lilly & Co., Merck & Co.
Buse reported having had relationships with GlaxoSmithKline and Takeda in the past, but all contracted through the University of North Carolina.
Eckel reported having a pending preclinical research grant with GlaxoSmithKline for an unrelated drug.
Primary source: American Diabetes Association
Bach RG, et al "Rosiglitazone and outcomes for patients with diabetes and cornary artery disease in the BARI 2D Trial" ADA 2010.