I have blogged multiple times about what I call the Diabetes Conspiracy and the Avandia saga. Please check out this time line which shows that the Avandia scare is much more about politics then science, and really starts with both Vioxx and Elliot Spitzer (yes, that Elliot Spitzer).
Briefly, Avandia was known to increase fluid retention and may have the potential for congestive heart failure, so after the product was approved in the late 90's, as part of due diligence, GSK agreed to perform a large randomized trial looking at heart benefits and risks of Avandia called RECORD (pronounced like the noun, i.e. LP and not the verb) , as well as present the results of their planned studies, looking for any heart risks. The RECORD study was a large, randomized trial designed specifically to answer the question of cardiovascular safety and Avandia. The RECORD study was of particular interest because Avandia was also known to have potential heart benefits, including improving good cholesterol, lowering triglycerides, and reducing other markers of heart disease. The question was whether the potential heart benefits (reduced heart attacks) outweighed the potential heart risks (congestive heart failure). It is important to note that at no time did anyone predict that Avandia might cause heart attacks. If anything, it was thought that Avanida might reduce heart attacks.
In August of 2006, GSK submitted both a pooled analysis (meta-analysis) of 42 separate double-blinded, randomized, controlled clinical trials to assess the efficacy of rosiglitazone (Avandia) for treatment of type 2 diabetes as well as a population-based database of 33,363 patients looking for similar endpoints. The 42 studies showed a potential increase in myocardial ischemia in patients taking Avandia compared to placebo. There were no problems when Avandia was compared to other diabetes meds, no issues with death, heart attack, or other adverse events in the FDA's meta-analysis. In addition, the large data base showed no problems with ischemia.
The FDA also knew that two large, randomized trials (ADOPT and DREAM) were about to be published that could help shed some light on this surprising and potential safety signal. In addition, the RECORD study had be going on for almost 2 years, and even though the results would not be available until 2009, an independent group which periodically looks at the study to assess safety (commonly done in safety studies) had not reported any problems. Given that GSK's data was inconsistent and that more data would soon be published, the FDA did not publicly disclose the findings of either the pooled data or population based data. This decision was made by the head of the FDA.
Coincidentally, the issue of releasing this data was just discussed in the New York times as the FDA, under scrutiny, is reconsidering some of these policies. However, there might be a plus side to not releasing data before a scientific conclusion can be made. Due to an earlier settlement regarding Paxil (find out who was the prosecutor behind this case by reading the Diabetes Conspiracy ), all of GSK's studies were available online. Dr. Steve Nissen, cardiologist from the Cleveland Clinic and known champion of dangerous drugs like Vioxx went and got the data himself and published it in the New England Journal of Medicine. ( I suspect that he was tipped off from someone inside the FDA who did not want to wait for RECORD, ADOPT and DREAM to finish before letting the public know about a potential danger- read here for more info). The Nissen article created a media storm, and patients and doctors got scared about using Avandia, since Nissen claimed Avandia caused a 43% increase in heart attacks.
Because of fears that the ongoing RECORD study could be jeopardized as patients might pull out of the study if Avandia were dangerous, as well as a means to vindicate their medication, GSK felt obligated to publish the partial results of the RECORD study. Though only more than half way done, the study seemed to show no signs of heart attacks, thus refuting the Nissen publication. However, the study was accompanied by not one, but three editorials discrediting the study, and claiming that it proved nothing about the safety regarding Avandia.
However, the final results of the RECORD study will be at 4:15 on Friday June 5th, at the ADA meeting in New Orleans and we should hopefully know once and for all whether or not Avandia is safe or increases the risk of heart attacks.
What are the possible results?
1. Avandia does in fact cause an increase risk of heart attack. If this is the case, GSK will likely pull Avandia from the market and spend billions of dollars settling law suits. However, this is HIGHLY unlikely, mainly because 1) the interim results did not show this 2) a data safety monitoring board has been independently watching the study closely and would have pulled the plug (like the ACCORD study) at the first warning signs, and 3) every large major randomized trial such as ACCORD, ADOPT, DREAM, VADT. etc show no increase risk of heart attacks.
2. It is not clear whether or not Avandia causes an increase risk in heart attacks.
This is a possibility (probably the most likely) for a variety of reasons. First, the primary outcome of the study is hospitalization (for acute myocardial infarction, congestive heart failure, stroke, unstable angina pectoris, transient ischemic attack, unplanned cardiovascular revascularization, amputation of extremities, or any other definite cardiovascular reason) or death from cardiovascular causes (including heart failure, acute myocardial infarction, sudden death, and death caused by acute vascular events including stroke). Hospitalization and death from cardiovascular causes individually, as well as each of the individual components (heart attacks, heart failure, etc.) are secondary endpoints. Since we know that Avandia can increase the risk of heart failure (possibly hospitalization, much less likely death) it is possible that the primary endpoint will not meet the study's statistically significant number, even though death might actually be reduced on Avandia and/or hearts attacks are clearly no different. When the results come out, should some of these endpoints go in opposite direction, be aware that the people who staked their reputation on NEJM meta-analysis (Dr. Nissen, Dr. Graham, Dr. Nathan) will likely come out and say that the entire study shows nothing. However, this is just simply not the case. It would be reasonable to expect that the primary endpoint of the study did not clearly indicate safety from the combination of hospitalization and death, but if all the hospitalizations are due to heart failure ( a known complication), there are no more deaths (possibly fewer) and there are no more heart attacks (possibly fewer); these are very meaningful and positive results.
Secondly, the study is designed to show with more than 95% certainty that Avandia would not increase the risk of BOTH cardiovascular hospitalization AND death by more than 20%. If the results show that the combination increases the risk by 21%, the general findings will be considered inclusive even if some of the findings of the study show Avandia to be clearly beneficial. Again, look for the naysayers to focus on this should the primary endpoint fail to meet this criteria. However, what you shold focus on is whether or not there is a difference in what everyone was worried about in the first place: heart attacks.
Finally, the study may not be powered to make a conclusion. For the RECORD study, the patients not on Avandia need to have the 11% events per year ( 3% with deaths from cardiovascular causes and 8% with hospitalizations from cardiovascular causes). In the interim analysis of RECORD, the event rate was much, much lower at 3.1%. This is likely due to much better diabetes care occurring then when the study was originally designed, for example more diabetics being put on statin medications. This was a HUGE criticism when the interim analysis was published, and will likely be a huger criticism on Friday. Power calculations are used to estimate how many patients would be needed in a study to show a difference. A well powered study increases the confidence that the results are true if a difference is shown. That said, if there are differences (or in this case no difference is shown, the study's primary goal), even if the study is not adequately powered, it does not make the results untrue. All it does is decreases the level of confidence that these results are true. Power calculations for RECORD were based on a 99% certainty. If the results are 85% certain, then this is probably good enough.
3. Avandia does not cause an increase risk of heart attacks. This was the primary goal of the study was to show that Avandia does not increase hospitalizations or deaths from cardiovascular causes. If the study shows with statistical confidence that there is no difference in both this primary outcome AND there is no statistically different risk of heart attacks between Avandia containing regimens and non-Avandia containing regimens, then this should be the end of the Avanida controversy. Case closed. Again, given that the primary outcome includes heart failure and the issue of statistical power mentioned above, it may be hard to show this. That said, even if the primary outcome is inconclusive, if both death from cardiovascular cause and heart attack are no different, along with data from other studies, this should be enough evidence to put the issue to rest, though it will leave enough room for folks like Nissen, Graham, and Nathan to be critical.
4. Avandia actually DECREASES the risk of heart attacks. I might be the only one in the country (including the drug company that makes Avandia), but I am not going to be surprised if the data shows that Avandia prevents heart attacks. I realize I am going out on a limb here, and making a prediction that is not likely to come true, days before we will actually know the results. However, there are several reasons that this might occur. First based on all the data we have up to this point, it is clear that Avandia doesn't cause heart attacks. As I have blogged before, the Nissen study itself is riddled with problems. However, the way the study is designed, the patients in the group who are NOT taking Avandia will be more likely to take insulin.In my previous posts The Problem With Insulin and The Problem with Insulin- Part 2 , I explain that though insulin is life saving for type 1 diabetics, and can be life saving for some type 2 diabetics, it is not without complications. In the ACCORD study where diabetics who were actually treated more aggressively to get their sugars to normal had more heart attacks and death. They also had more hypoglycemia or low blood sugar, which could be a cause of heart attack or death as well. It may be that because Avandia has properties that might actually prevent heart attacks (improved good cholesterol or HDL for example) and because the patients not taking Avandia would be more likely to need insulin, that Avandia might be shown to actually reduce heart attacks!
Even if scenario #2 or #3 (3 obviously preferred) turns out to be the result, and their is a mix of finding BUT no difference in heart attacks, death, and only an increase in the risk for the known complications of CHF; AND in addition to this there turns out to be less use of insulin and less hypoglycemia in the Avandia group, isn't that reason enough to use Avandia (or Actos, the other TZD on the market)? For most of my diabetic patients, the thing they are most afraid of is needing insulin. How wonderful would it be to know that I can spare many patients the need for starting insulin by using pills (and not just the generic ones that although cheap, are known to fail over the long term)? Yet, current recommendations suggest that doctors should use JUST the old, generic diabetes medicine (there are only two) and if these fail, go right to insulin. This is why the RECORD study is so important. It is more than just about vindicating a maligned drug or making a cardiologist who likes the limelight look foolish. It's is about whether or not we should use insulin early in the game or as a last resort. Thus, the results of the study should have profound implications for the management of all patients with type 2 diabetes.