A new meta-analysis of observational studies suggests that rosiglitazone, the diabetes drug suspended last year by European regulators and severely restricted by the U.S. Food and Drug Administration over concerns of elevated cardiac risk, is associated with modest but statistically significant increases in the risk of heart attack, heart failure, and death, compared with pioglitazone, another drug in the same class.
The findings, derived from 16 studies (12 retrospective cohort studies and 4 case control studies), were published March 18 in BMJ (2011;342:d1309) and seem to add weight to the case against rosiglitazone’s cardiovascular safety relative to pioglitazone.
Rosiglitazone, the meta-analysis found, was associated with an approximately 16% higher heart attack risk, 22% higher congestive heart failure risk, and 14% higher mortality risk than pioglitazone among 810,000 patients with type 2 diabetes (429,000 taking rosiglitazone and 381,000 on pioglitazone). All differences were significantly different, reported Dr. Yoon Kong Loke of the University of East Anglia, Norwich, England. These numbers translate into 170 excess myocardial infarctions, 649 excess cases of heart failure, and 431 excess deaths for every 100,000 patients receiving rosiglitazone instead of pioglitazone, they explained.
The investigators chose to evaluate data from observational studies because of a lack of randomized controlled trials directly comparing the drugs, and because "clinical trials have strict selection criteria that may exclude participants at high risk of adverse events," they wrote, making cardiovascular events rarer than they would be in clinical practice.
Though Dr. Loke and his colleagues acknowledged that data from observational trials are susceptible to bias and confounding, they also argued that such data more likely reflected the results to be expected in real-world settings.
One clinician who remained unconvinced that Dr. Loke and his colleagues had demonstrated a significant risk difference was Dr. Sanjay Kaul, director of the Vascular Physiology and Thrombosis Research Laboratory at Cedars-Sinai Medical Center in Los Angeles and lead author of a 2010 American Heart Association Scientific Advisory on the two medications (Circulation 2010;121;1868-77). In the advisory, Dr. Kaul and his colleagues, citing a number of studies including a 2010 FDA meta-analysis of clinical trial data, determined that there were insufficient data to support pioglitazone over rosiglitazone based on cardiac risk.
In an interview, Dr. Kaul criticized Dr. Loke and his colleagues’ findings sharply, noting that meta-analyses of observational data are generally discouraged because they have been shown to produce "very precise but equally spurious" results that are seldom replicated in randomized trials.
Further, Dr. Kaul continued, Dr. Loke and his colleagues justified pooling "because of lack of statistical heterogeneity. However there is substantial clinical heterogeneity that arguably precludes pooling" – including differences in study design (such as case-control vs. cohort), treatment exposure, type of comparator (single vs. combined therapies), end points, method of analysis, matching technique, and effect size measure.
Dr. Kaul noted that, during a July 2010 FDA Advisory Panel meeting on rosiglitazone and cardiovascular safety, the agency’s statistical experts and reviewers had cautioned that because of such heterogeneity, estimates of association from observational studies should not be pooled into a single meta-analytic estimate.
The hypothesis of increased myocardial infarction risk raised was not validated in the four largest and highest-quality studies included in Dr. Loke and his colleagues’ meta-analysis, Dr. Kaul pointed out. Moreover, "The risk estimates for death or composite end points were all less than an odds ratio of less than 1.5. Effect sizes of this magnitude in epidemiologic studies are of questionable relevance and credibility," he said, adding that these should be considered as hypothesis generating at best and that randomized clinical trials would be required to confirm them.
Dr. Loke, in an interview, said that while he did not disagree that the design and data quality of the studies in his team’s meta-analysis were less than consistent, "Let’s face it, there’s never going to be a randomized controlled trial comparing [rosiglitazone and pioglitazone]. But, if I were a patient deciding, I’d want the available evidence."
Two editorialists suggested that the study was further evidence of misconduct by drug companies and regulators in pushing a treatment to market with questionable benefit and substantial evidence of harm. Dr. Victor M. Montori, professor of medicine, and Dr. Nilay D Shah of the department of health services research, both at the Mayo Clinic in Rochester, Minn., said the "rosiglitazone story says much about how health care has become less about promoting patients’ interests, alleviating illness, promoting function and independence, and curing disease, and much more about promoting other interests, including those of the drug industry. Has the corruption of health care advanced so far that it is unreasonable, even naive, to expect responsible drug companies, enlightened regulators, and thoughtful prescribers?" (BMJ 2011;342:d1354).