In the previous post we looked at some possible mechanisms which could contribute to Avandia’s issues with cardiac events. Here I would like to look more closely at the actual data so that we can come to our own conclusions about Avandia.
The most recent is the re-analysis of the results of the RECORD (Rosiglitazone Evaluated for Cardiac Outcomes and Regulation of glycemia in Diabetes) trial which is available in pre-print form online through the European heart Journal. Rosiglitazone is the pharmacological (scientific) name for Avandia. Because the studies often use the pharmacological name for this drug and I will be quoting from them, I will also use rosiglitazone and Avandia interchangeably. The original RECORD trial was published as an interim analysis in 2007 in the New England Journal of Medicine and in final form in the Lancet in 2009. This study looked at 4458 people with type 2 diabetes in which glucose levels were not sufficiently controlled with either metformin or one of the sulfonylurea drugs. The metformin treated patients were then divided into half which got the addition of a sulfonylurea and half which got rosiglitazone (Avandia). Similarly the sulfonylurea group received either metformin or rosiglitazone. So we now have 3 groups: rosiglitazone plus a sulfonylurea, rosiglitazone plus metformin, and the sulfonylurea drug plus metformin. (There are a host of sulfonylurea drugs on the market and since they are pretty similar we are just going to refer to them by their pharmacological name). In both reports (2007 and 2009), the authors claimed no additional deaths were observed in the rosiglitazone (Avandia) groups as compared to the metformin/sulfonylurea group. In the abstract the final sentence stated “…rosaglitazone does not increase the risk of overall cardiovascular morbidity or mortality compared with standard glucose-lowering drugs.”
The trial was criticized on several levels in an editorial by Drs. Bruce Psaty and Curt Furberg, published in the New England Journal of Medicine in July, 2007. They felt that the definition of the primary outcome (all hospitalizations and deaths from cardiac causes) was too weak and promoted the likelihood of a null outcome. In other words, because life is complicated, there would probably be many reasons other than rosiglitazone that might bring the patient to the hospital and so we would lose the specific effect of rosiglitazone in the larger ebb and flow of events. Primary outcomes in clinical trials, also called endpoints, are the actual things that are measured and compared statistically between the various treatment groups. A trial may have multiple parameters which are measured but usually there is just one primary endpoint to establish if the trial is a “success” or a “failure”. Furthermore, continued Psaty and Furberg, the number of actual cardiac events seemed far too low for such an at risk population, suggesting problems with hospital reporting. As this was a multi-country study this was a valid concern.
This third analysis of the RECORD data, now in preprint form, addresses the criticisms of Psaty and Furberg and tells a different story. Rather than use all hospitalizations and deaths from cardiac causes as the endpoint, they established a Clinical Endpoint Committee which independently decided upon the appropriate parameters that should be measured. For example, they looked at time to first heart failure (HF) event and found that by this measure the rosiglitazone groups appeared to be double that of the metformin/sulfonylurea group. Similarly, fatal HF events were higher in the rosiglitazone group. From these and other measures the authors now conclude “These findings confirm the increased risk of HF events in people treated with rosiglitazone and support the recommendation that agent should not continue to be used in people developing symptomatic HF while using the medication.”
Another recent study was performed in Canada by a group headed by Dr. David Juurlink and published in the British Medical Journal in 2009. They did something called a “population based retrospective cohort study”. Since medical records in Canada are quite searchable, this group made use of the many thousands of people within their population who had been treated over the years. The fact that the patients had already been treated and the outcomes recorded is the retrospective part. The patients included in the study were the people who happened to live in Ontario, were 66 years or older, and were treated between the dates of April 1, 2002 and March 31, 2008. This is the population cohort part of the study. They examined 16,951 people who had been prescribed Actos and 22,785 people who had been prescribed Avandia. They then looked at death rates from any cause. After accounting for the variables within the population they found that indeed more Avandia patients had died than Actos patients. Their assessment was 1 extra death for every 93 patients. Interestingly, while cardiac deaths were increased, there was no difference in the rate of heart attack (myocardial infarction).
The Canadian study was preceded by a host of earlier studies. These were aggregated in something called a “meta-analysis” in which, through careful statistics, the results of many trials are compared. This is a tricky business in that each trial involves different investigative groups and designs. In some ways, it is an attempt to compare apples and oranges. The statisticians must come up with reasonable conversions to allow different populations and trial end points to be compared. One meta-analysis, published by Drs Nissen and Wolski in the New England Journal of Medicine looked at 42 separate clinical trials for a total of 12,282 patients. In some of these trials the rosiglitazone group had a few more cardiac events while in other trials the control group had a few more cardiac events. When everything was tallied up here is what they found: within the rosiglitazone groups from all of the trials considered, 86 patients had myocardial infarctions and 39 patients suffered death from cardiac causes. In the control groups from all of the trials considered, 72 patients had myocardial infarctions and 22 patients suffered death from cardiac causes. These and other measures do pretty much show that rosiglitazone does cause more cardiac events than other glucose lowering strategies.
Whether or not you choose to switch from Avandia to some other glucose lowering therapy is a decision you and your doctor should discuss. I do not necessarily think that Avandia should simply be abandoned. I keep thinking about this one measure that keeps cropping up in these studies: the all deaths or all hospitalizations measures. Given that heart disease is a major cause of death, why would the all deaths measure be so similar among the various groups in all of these clinical trials. Ruminations on this topic will be the subject of the next post.