86. A Case Study of Avandia

GlaxoSmithKline (GSK) was recently accused of offering substantial sums as bribes in order to facilitate sales of their products in China. This is not the first time the drug company has encountered problems because of behaviour which fails to match up to the highest standards of corporate social responsibility.

In 2007 Steven Nissen and Kathy Wolski of the Department of Cardiovascular Medicine in the Cleveland Clinic published a paper in the New England Journal of Medicine (NEJM) which raised doubts about the safety of the drug rosiglitazone (1). This was produced and marketed by GSK under the brand name Avandia. The drug had been approved by various regulatory authorities, including the FDA in the USA on the grounds that it reduced the blood glucose levels in diabetics. Therefore it was assumed that the damaging effects associated with diabetes would be somewhat alleviated. One of the main concerns with diabetes is an increased risk of developing heart disease. Using data obtained in 42 studies involving almost 28,000 patients, the researchers concluded that those who were prescribed Avandia were 43% more likely to suffer from a heart attack and 64% more likely to die of a heart related condition.

So it was no surprise that when this report was published there were ructions and eventually the US Senate Finance Committee decided to investigate the matter. A report prepared by the Committee staff was extremely comprehensive and illuminating because it was able to demand access to documents not normally available (2).Here are some of the points which were revealed.

  • Before accepting papers for publication the NEJM editors arrange for manuscripts to peer-reviewed by independent experts in the field. One of the conditions is that this is done on a confidential basis and the material must not be shared. However one of the reviewers, Dr Steve Haffner forwarded the Nissen/Wolski paper to GSK executives. Some GSK staff examined the case and performed their own independent analysis of the data. This exercise produced essentially the same values for the increased risk of heart attacks, which was also in agreement with the results of an analysis by the FDA
  • At this point, the plot begins to thicken because when the Nissen/Wolski paper became public knowledge. GSK released a statement which said:

GSK strongly disagrees with the conclusions reached in the NEJM article, which are based on incomplete evidence and a methodology that the author admits has significant limitations.’’ 

The justification cited in support of the above was GSK trial entitled RECORD which still had some years before completion. In view of developments GSK prepared a manuscript on the information available to date and submitted it the NEJM, which then passed it on to various reviews for their opinion. In a letter to the authors, the editor commented that:

In the opinion of all the readers, the data that you present are completely

compatible with the results of the meta-analysis by Nissen…”

‘‘The editors feel strongly that your data do not support the statement that the RECORD results for MI contradict the Nissen meta-analysis; this statement must be removed or modified.’’

In spite of these adverse comments, the NEJM published the paper, which had the following conclusion.

“Our interim findings from this ongoing study were inconclusive regarding the effect of rosiglitazone on the overall risk of hospitalization or death from cardiovascular causes. There was no evidence of any increase in death from either cardiovascular causes or all causes. Rosiglitazone was associated with an increased risk of heart failure. The data were insufficient to determine whether the drug was associated with an increase in the risk of myocardial infarction” (3).

It is obvious that this paper was an attempt by GSK to undermine the credibility of the earlier work showing an increased risk of heart disease. Nevertheless there was no way that the authors could avoid the increased risk of heart failure.

It is possible that the NEJM felt pressurised into accepting the paper for publication and the editors took the unusual step of including 3 different commentaries in the same issue which were:

  • It is highly unlikely that the study will ever establish a cardiovascular benefit for Avandia. The interim results in the RECORD trial do not provide any assurance that treatment with Avandia is safe(4)
  • The findings of the RECORD trial are not inconsistent with the results of the Nissen Wolski analysis. There is continued uncertainty about the cardiovascular safety of Avandia (5)
  • The primary weakness that the incidence of events such as heart attacks is very low…only about 40% of what would be expected in diabetics aged 55-60 years old. This is probably because of a high loss to follow-up and the high proportion of patients from Eastern Europe, where the medical care and criteria for admission to hospital are different from those in the West(6).

The position was summarised neatly by Clifford Rosen who chaired an FDA committee which considered this issue when he wrote that:

The basic plot of the rosiglitazone story quickly became obvious to the advisory committee: a new “wonder drug,” approved prematurely and for the wrong reasons by a weakened and underfunded government agency subjected to pressure from industry, had caused undue harm to patients”(7).

Even if we disregard the murkier aspects of this particular case, it does highlight the fundamental weakness of using a surrogate (in this instance blood glucose) rather than a more realistic one such as years of good health or life expectancy.



  1. S E Nissen and K Wolski (2007) New England Journal of Medicine 356 (24) pp2457-2471
  2. Available at http://www.finance.senate.gov/newsroom/chairman/release/?id=bc56b552-efc5-4706-968d-f7032d5cd2e4
  3. P D Home et al (2007) New England Journal of Medicine 357 (1) pp28-38
  4. D M Nathan (2007) New England Journal of Medicine 357 (1) pp64-665
  5.  M Drazen et al (2007) New England Journal of Medicine 357 (1) pp63-64
  6. B H Psaty and C D Furberg (2007) New England Journal of Medicine 357 (1) pp67-69
  7. C J Rosen (2007) New England Journal of Medicine 357 (9) pp844-846.
Scroll to Top