Perspective

Revisiting the Rosiglitazone Story — Lessons Learned

Clifford J. Rosen, M.D.

N Engl J Med 2010; 363:803-806August 26, 2010

Article

In July 2007, 24 members of the Endocrinologic and Metabolic Drug Advisory Committee (EMDAC) and the Drug Safety and Risk Management Advisory Committee of the Food and Drug Administration (FDA) held a joint meeting, which I chaired, and concluded that rosiglitazone, an insulin-sensitizing agent used in treating type 2 diabetes mellitus, was associated with an increased risk of myocardial ischemia.1 This conclusion was based on concordant evidence from three independent meta-analyses performed by the drug's sponsor, the FDA, and Nissen and Wolski.1,2 Nevertheless, the committee voted, as did I, to keep rosiglitazone on the market, although we recommended enhancing efforts to inform patients and providers about potential cardiovascular risk. The committee also recommended more studies to refine the risk estimate, including a complete analysis of the Rosiglitazone Evaluated for Cardiac Outcomes and Regulation of Glycemia in Diabetes (RECORD) trial and large observational cohort studies.

Subsequently, the FDA added a black box warning about myocardial ischemia to the drug's label. In July 2010, 33 members of the joint advisory committee met for 20 hours over 2 days to further advise the FDA about rosiglitazone. Ultimately, the committee agreed that the drug posed significant cardiovascular risk. Twelve participants voted that it should be removed from the market, 10 (including me) voted for much stricter control over prescriptions, 7 voted for further warnings, 3 believed no changes were necessary, and 1 abstained. That vote has been interpreted in two ways — as indicating that the majority wanted to keep rosiglitazone on the market or that the majority wanted it either removed from the market or severely restricted in its use. So what has changed in 3 years?

Three lines of evidence have emerged since 2007 suggesting that rosiglitazone's cardiovascular risk is significant and that major federal action is now required. First, two meta-analyses including 10 new randomized trials show that the risk of acute myocardial infarction is 30 to 80% higher among patients with diabetes taking rosiglitazone than among those receiving placebo.3,4 In comparisons of rosiglitazone with other antidiabetic agents, such as the sulfonylureas or metformin, the hazard ratio is lower, although it is still greater than 1. In comparison with the FDA's 2007 meta-analysis, the newer analyses showed higher odds ratios for major adverse cardiovascular events, myocardial infarction, and serious ischemic events in patients receiving rosiglitazone (see Table 1Table 1Odds Ratios for Cardiovascular Events According to 2007 Meta-Analysis of 42 Trials and 2010 Meta-Analysis of 52 Trials.).4 A similar level of risk for myocardial infarction was reported by Nissen and Wolski; these findings held constant regardless of whether the RECORD trial was included.3

Second, observational data, particularly from Medicare databases, provide further support for these risk estimates. Seven case–control and 14 cohort studies demonstrated that rosiglitazone use was associated with greater cardiovascular risk than pioglitazone.4

Third, several new data sets from both clinical trials and observational studies show that pioglitazone either conferred no increased cardiovascular risk or, in some circumstances, protected against myocardial infarction. In one case, FDA investigators performed a retrospective cohort study of 225,000 Medicare Part D beneficiaries who received either rosiglitazone or pioglitazone from 2006 to 2009; they found that the risk of a cardiovascular event or death was nearly 25% higher if the patient was taking rosiglitazone rather than pioglitazone (see Table 2Table 2Hazard Ratios for Death from Any Cause and Composite End Points in the Medicare Database, 2006–2009.).5

In the past few years, we have gained a better understanding of the pharmacology of rosiglitazone and pioglitazone — and of why they may have different cardiovascular-risk profiles although they belong to the same class of drugs, the thiazolidinediones. These agents work by binding to the nuclear receptor and transcription factor peroxisome proliferator–activated receptor gamma (PPARG). Stimulation of PPARG occurs when it is bound by exogenous ligands (e.g., thiazolidinediones) or endogenous ligands (e.g., fatty acids or prostaglandins) that initiate recruitment of co-activators or co-repressors to the transcriptional complex. This recruitment activates downstream target genes to enhance insulin sensitivity and promote adipocyte differentiation. However, many of PPARG's actions are ligand-specific and therefore dependent on the variable strength of co-activator or co-repressor recruitment, differences in histone deacetylation, and site-specific changes in phosphorylation of the receptor. Hence, it is not surprising that rosiglitazone and pioglitazone activate or repress unique gene sets that have distinct effects on lipids and the vascular system.

So what lessons were learned from the recent hearing, and how might they be useful to the FDA? First, although black box warnings sound impressive and were instituted after the 2007 meeting, rosiglitazone remains on many formularies, and its use remains relatively stable (see graphUse of Rosiglitazone and Pioglitazone in Medicare Part D Beneficiaries, 2006–2009.).4 Rosiglitazone use leveled off after the publication of the first meta-analysis by Nissen and Wolski in May 2007,2 yet surprisingly, there are still a million rosiglitazone prescriptions written in the United States each year. Although there is evidence that black box warnings and notifications from the FDA have an impact, results from the Medicare database and the totality of evidence from committee hearings would argue against limiting new FDA action to the addition of further warnings.

Another clear lesson is that postmarketing studies are fraught with hidden risks if the trial design is improper. RECORD, a noninferiority trial, was instituted 2 years after rosiglitazone was approved and was authorized by European regulatory agencies, not the FDA. RECORD's open-label, unblinded design led to many committee questions about the quality of the data. Committee discussions also reflected profound skepticism about the trial results because of the inherent bias of the sponsor and questions concerning data integrity. Clearly, postmarketing studies require more careful designs and safeguards to protect data integrity and objectivity.

During the recent hearing, very little was said about the balance of risks and benefits. Though the risks were apparent from the data presentations, there was virtually no discussion of any unique benefits of rosiglitazone. A third lesson, I believe, is that all committee hearings should be required to include a discussion of the risk–benefit balance.

Finally, the committee found itself in a difficult position because of a shift in end points over time. Ten years ago, evaluations of diabetes-drug studies were based on glycemic durability, a surrogate marker, not a patient-specific outcome. The FDA has worked hard to push sponsors toward designing preapproval trials that address hard outcomes rather than surrogate indexes. This rigorous path to approval, which generates invaluable data for assessing both risks and benefits, must continue despite sponsors' concerns about cost.

A remarkable change has occurred at the FDA that may mean that a final decision about rosiglitazone is near. Since 2007, the agency has become more open in its deliberations, held advisory committee meetings more frequently, and clarified its position on conflicts of interest. The recent EMDAC hearing exemplified some of these changes. The committee was larger by half than the 2007 committee, which allowed it to include broader representation from multiple disciplines. In addition, the FDA worked hard between 2007 and 2009 to produce guidance that included upper and lower boundaries of confidence intervals around cardiovascular risk for new diabetes drugs. These limits provided critical benchmarks for committee members to use in risk assessment.

Of course, the ultimate decision about rosiglitazone's fate lies with the FDA: advisory committees merely provide advice for the agency to use as it wishes. Despite what may have seemed an ambiguous vote, I would argue that the EMDAC clearly indicated that the FDA must do more than provide additional black box warnings. The vote may reflect a desire to balance patient protection with the availability of the broadest possible range of therapies for a complex, chronic disease, but the clear risk demonstrated by the aggregate data calls for serious measures. Thus, in my view, the FDA must consider only two choices: either stronger warnings plus the use of informed consent and a registry for compassionate use of rosiglitazone or removal of the drug from the market.

Disclosure forms provided by the author are available with the full text of this article at NEJM.org.

This article (10.1056/NEJMp1008233) was published on July 21, 2010, at NEJM.org.

Source Information

From the Maine Medical Center Research Institute, Scarborough.

References

References

1. 1

   Rosen CJ. The rosiglitazone story -- lessons from an FDA Advisory Committee meeting. N Engl J Med 2007;357:844-846
   Full Text | Web of Science | Medline

2. 2

   Nissen SE, Wolski K. Effect of rosiglitazone on the risk of myocardial infarction and death from cardiovascular causes. N Engl J Med 2007;356:2457-2471[Erratum, N Engl J Med 2007;357:100.]
   Full Text | Web of Science | Medline

3. 3

   Idem. Rosiglitazone revisited: an updated meta-analysis of risk for myocardial infarction and cardiovascular mortality. Arch Intern Med 2010 June 28 (Epub ahead of print).
   

4. 4

   Food and Drug Administration. Briefing document: July 13-14, 2010 meeting of the Endocrinologic and Metabolic Drugs Advisory Committee. (Accessed July 20, 2010, at http://www.fda.gov/AdvisoryCommittees/CommitteesMeetingMaterials/Drugs/EndocrinologicandMetabolicDrugsAdvisoryCommittee/ucm191113.htm.)
   

5. 5

   Graham DJ, Ouellet-Hellstrom R, Macurdy TE, et al. Risk of acute myocardial infarction, stroke, heart failure, and death in elderly Medicare patients treated with rosiglitazone or pioglitazone. JAMA 2010 June 28 (Epub ahead of print).
   

Citing Articles (19)

Citing Articles

1. 1

   M. Angelyn Bethel, Harald Sourij. (2012) Impact of FDA Guidance for Developing Diabetes Drugs on Trial Design: From Policy to Practice. Current Cardiology Reports 14:1, 59-69
   CrossRef

2. 2

   Sayuri N. Friedland, Aaron Leong, Kristian B. Filion, Jacques Genest, Iliana C. Lega, Salvatore Mottillo, Paul Poirier, Jennifer Reoch, Mark J. Eisenberg. (2012) The Cardiovascular Effects of Peroxisome Proliferator-activated Receptor Agonists. The American Journal of Medicine 125:2, 126-133
   CrossRef

3. 3

   Dawn M. Torres, Frances J. Jones, Janet C. Shaw, Christopher D. Williams, John A. Ward, Stephen A. Harrison. (2011) Rosiglitazone versus rosiglitazone and metformin versus rosiglitazone and losartan in the treatment of nonalcoholic steatohepatitis in humans: A 12-month randomized, prospective, open- label trial. Hepatology 54:5, 1631-1639
   CrossRef

4. 4

   Michael Hochman, Danny McCormick. (2011) Endpoint Selection and Relative (Versus Absolute) Risk Reporting in Published Medication Trials. Journal of General Internal Medicine 26:11, 1246-1252
   CrossRef

5. 5

   Kim Henriksen, Claus Christiansen, Morten Asser Karsdal. (2011) Serological biochemical markers of surrogate efficacy and safety as a novel approach to drug repositioning. Drug Discovery Today 16:21-22, 967-975
   CrossRef

6. 6

   Claudia Sedlinsky, María Silvina Molinuevo, Ana María Cortizo, María José Tolosa, Juan Ignacio Felice, María Laura Sbaraglini, Leon Schurman, Antonio Desmond McCarthy. (2011) Metformin prevents anti-osteogenic in vivo and ex vivo effects of rosiglitazone in rats. European Journal of Pharmacology 668:3, 477-485
   CrossRef

7. 7

   A. P. van de Woestijne, H. Monajemi, E. Kalkhoven, F. L. J. Visseren. (2011) Adipose tissue dysfunction and hypertriglyceridemia: mechanisms and management. Obesity Reviews 12:10, 829-840
   CrossRef

8. 8

   A. Chaudhuri, P. Dandona. (2011) Effects of insulin and other antihyperglycaemic agents on lipid profiles of patients with diabetes. Diabetes, Obesity and Metabolism 13:10, 869-879
   CrossRef

9. 9

   Jonathan P. Whitehead. (2011) Diabetes: New conductors for the peroxisome proliferator-activated receptor γ (PPARγ) orchestra. The International Journal of Biochemistry & Cell Biology 43:8, 1071-1074
   CrossRef

10. 10

    P. Burckhardt. (2011) Potential negative cardiovascular effects of calcium supplements. Osteoporosis International 22:6, 1645-1647
    CrossRef

11. 11

    Keith G Tolman. (2011) The safety of thiazolidinediones. Expert Opinion on Drug Safety 10:3, 419-428
    CrossRef

12. 12

    José L. García-Giménez, Fabián Sanchis-Gomar, Federico V. Pallardó. (2011) Could thiazolidinediones increase the risk of heart failure in Friedreich's ataxia patients?. Movement Disorders 26:5, 769-771
    CrossRef

13. 13

    Jean-Luc Vachiéry. (2011) Prostacyclins in pulmonary arterial hypertension: the need for earlier therapy. Advances in Therapy 28:4, 251-269
    CrossRef

14. 14

    Harold E. Lebovitz. (2011) Type 2 diabetes mellitus—current therapies and the emergence of surgical options. Nature Reviews Endocrinology 7:7, 408-419
    CrossRef

15. 15

    Robert Ratner, Jenny Han, Dawn Nicewarner, Irina Yushmanova, Byron J Hoogwerf, Larry Shen. (2011) Cardiovascular safety of exenatide BID: an integrated analysis from controlled clinical trials in participants with type 2 diabetes. Cardiovascular Diabetology 10:1, 22
    CrossRef

16. 16

    G. Schernthaner, R. J. Chilton. (2010) Cardiovascular risk and thiazolidinediones-what do meta-analyses really tell us?. Diabetes, Obesity and Metabolism 12:12, 1023-1035
    CrossRef

17. 17

    Taina K. Mattila, Anthonius de Boer. (2010) Influence of Intensive versus Conventional Glucose Control on Microvascular and Macrovascular Complications in Type 1 and 2 Diabetes Mellitus. Drugs 70:17, 2229-2245
    CrossRef

18. 18

    Yihong Wan. (2010) PPARγ in bone homeostasis. Trends in Endocrinology & Metabolism 21:12, 722-728
    CrossRef

19. 19

    Robert E Lasky, Wim van Drongelen. (2010) Is sucrose an effective analgesic for newborn babies?. The Lancet 376:9748, 1201-1203
    CrossRef