On 21 May 2007, the New England Journal of Medicine published a meta-analysis of 42 trials of rosiglitazone (Avandia, GlaxoSmithKline) for treating type 2 diabetes mellitus. It found that the drug was associated with an increased risk of myocardial infarction (odds ratio 1.43; 95% confdence interval 1.03 to 1.98; P=0.03) and death from cardiovascular causes (1.64; 0.98 to 2.74; P=0.06).1
Rosiglitazone, a thiazolidinedione, is an agonist at the peroxisome-proliferator activated receptors in cell nuclei. These receptors modulate the expression of a host of genes, and glycaemic control is achieved primarily through increased insulin sensitivity in peripheral tissues. Rosiglitazone was approved by the US Food and Drug Administration (FDA) in 1999 and by the centralised process of the European Medicines Agency (EMEA) in 2000. Its popularity has increased steadily, with more than one million prescriptions written in the one year period ending March 2006 in England alone—a 22% increase over the previous year.2 However, the recently published meta-analysis raises serious questions about the drug’s safety.
Meta-analyses have unique strengths and weaknesses and this one is no exception.3 Its singular strength is the statistical power generated by data on 15 560 patients from published and unpublished trials. However, it includes clinically heterogeneous trials and criteria used by individual trials to classify adverse events are somewhat unclear. Only summary data are available in the public domain—for example, whether or not a person had a myocardial infarction, not when it occurred—which makes time to event analyses impossible. Also, the total number of adverse events was small, so that misclassifi-cation of a few events could alter the conclusions.
In response to the concerns raised by this meta-analysis, an unplanned interim analysis of a large, manufacture sponsored, randomised, open label, non-inferiority trial specifcally designed to investigate the cardiovascular safety of rosiglitazone was recently released.4 Compared with patients taking metformin and a sulphonylurea, people taking a regimen that included rosiglitazone had no signifcant increase in the risk of myocardial infarction (hazard ratio 1.16, 0.75 to 1.81), although they had a signifcantly increased risk of heart failure (2.24, 1.27 to 3.97). When these new data are added to the trials in the previous meta-analysis, rosiglitazone is associated with an increased risk of myocardial infarction (odds ratio, 1.33; 1.02 to 1.72).5
To summarise, the meta-analyses show a signifcantly increased risk for myocardial infarction, whereas several individual prospective trials do not. More data would certainly help to clarify the matter, but the emerging safety concerns question the prudence of continuing ongoing trials. Notwithstanding the ethical concerns, it may be impossible to prevent an exodus of patients from these trials in light of the ongoing “trial by media” of the drug.
The broader question is how this refects on regulatory processes used to monitor drug safety. Postmarketing surveillance, or pharmacovigilance, remains the weakest link in the regulatory process on both sides of the Atlantic. The current approach—the FDA’s adverse event reporting system and the European EudraVigilance programme—relies heavily on passive surveillance, and it is based on reports of unusual adverse events from consumers, practitioners, manufacturers, and national regulatory authorities. At best, this creates a case series, one of the weakest forms of epidemiological evidence,6 that would be insensitive to an increase in common events like myocardial infarcts in diabetics.
Alternatively, the regulatory authorities may require systematic phase IV trials after market authorisation, but these are often not completed in a timely manner. In the United States, completion dropped from 62% in the 1970s to 24% in recent years,6 and the FDA is ill equipped to act against defaulters. As of September 2006, 930 (74%) of the 1259 postmarket studies were pending or delayed.7
This results in a fractured regulatory process, where the preapproval phase is marked by stringent requirements for safety and effcacy data, but performance in postmarketing surveillance falls short of the standards the agencies set for themselves. This is exemplifed by the case of rosiglitazone. Rosiglitazone comes from a family of drugs with well documented side effects,8 9 and itself is associated with increased heart failure, anaemia, and raised low density lipoprotein concentration. However, postmarketing safety data seven years after regulatory approval consist of a patchwork of heterogeneous manufacturer sponsored trials, many of which are unpublished. Of note, a similar meta-analysis submitted by the manufacturer to the EMEA and the FDA in August 2006 showed an increased risk in ischaemic events (hazard ratio, 1.31, 1.01 to 1.70).10 The EMEA updated the product label of the drug,11 but no specifc communication to healthcare professionals was issued. The FDA did neither.
The system needs to be fixed. The Institute of Medicine recommends a life cycle approach to drug
evaluation.12 This would involve a systematic effort to monitor the safety and effcacy of a drug before and after approval using data from well designed clinical trials to inform ongoing risk-benefit analyses. This process could be made more systematic by requiring regulatory authorities to periodically and independently re-evaluate all data gathered after approval for all new molecular entities—particularly drugs with high sales.
In addition, the lack of transparency in the current system needs to be dealt with. There should be a legal requirement for all phase II-IV trials to be registered in a centralised database, such as the National Library of Medicine’s clinicaltrials.gov or an equivalent. Complete datasets from these trials, systematic analyses of the results, and reports of periodic evaluations by the regulatory agencies must be publicly available.
A radical change is needed in the culture of existing regulatory institutions that regard postmarketing surveillance as their secondary mandate. This will require systematic rethinking of the existing regulatory and funding processes, and expediting changes currently in the pipeline.13 Progress will entail empowering the regulatory agencies with additional authority and resources.
The manufacturer and the FDA will share the spotlight as congressional investigation into the matter starts. In the meantime, what are the implications for patients currently on rosiglitazone? Doctors will need to revisit the indication for the drug on a case by case basis, bearing in mind that several alternatives are cheaper, supported by robust evidence, and now perhaps safer.14 The decision to switch drugs must be tempered by the fragility of the available evidence and the risks associated with altering patients’ medical regimens. Needless to say, the ongoing use of rosiglitazone merits careful deliberation.


1 Nissen SE, Wolski K. Effect of rosiglitazone on the risk of myocardial infarction and death from cardiovascular causes. N Engl J Med Published online 21 May 2007.
2 National Institute for Health and Clinical Excellence. NICE implementation uptake report: glitazones (rosiglitazone and pioglitazone). 2007. www.nice. org.uk/page. aspx?o=423524.
3 EysenckHJ. Meta-analysis and its problems. BMJ 1994;309:789-92.
4 Home PD, PocockSJ, Beck-Nielsen H, Gomis R, Hanefeld M, Jones N, et al. Rosiglitazone evaluated for cardiovascular outcomes—an interim analysis. N Engl J Med Published online 5 June 2007.
5 Psaty BM, Furberg CD. The record of rosiglitazone and the risk of myocardial infarction. N Engl J Med Published online 5 June 2007.
6 US Senate. Testimony of Bruce M Psaty, before the House Committee on Energy and Commerce subcommittee on oversight and investigations. 2007. http://energycommerce.house.gov/cmte_mtgs/110-oi-hrg.032207.Psaty-testimony.pdf.
7 Food and Drug Administration. Report on the performance of drug and biologic firms conducting postmarketing commitment studies; availability. Federal Register 2007;72:5069-70.
8 Gale EA. Lessons from the glitazones: a story of drug development. Lancet 2001;357:1870-5.
9 Nissen SE, Wolski K, Topol EJ. Effect of muraglitazar on death and major adverse cardiovascular events in patients with type 2 diabetes mellitus. JAMA 2005;294:2581-6.
10 GlaxoSmithKline. Study no. ZM2005/00181/01. Avandia cardiovascular event modeling project. http://ctr.gsk.co.uk/ Summary/Rosiglitazone/II I_CVmodeling.pdf.
11 European Medicines Agency. European public assessment report for authorised medicinal products for human use. Avandia. 2007.www. emea.europa.eu/humandocs/Humans/EPAR/avandia/avandia. html.
12 Psaty BM, Burke SP. Protectingthe health of the public—Institute of Medicine recommendations on drugsafety. N Engl J Med 2006;355:1753-5.
13 European MedicinesAgency. Work programme for European Medicines Agency. London: EMEA, 2007:20-2.
14 Nathan DM. Rosiglitazone and cardiotoxicity—weighingthe evidence. N Engl J Med Published online 5 June 2007.