Sunday, July 4, 2010

Avandia Smackdown!!

This is it. The final countdown.
The fate of Avandia will likely be decided in a matter of days.
The outcome will likely have long lasting impact regarding how the FDA manages potential safety issues with any current or future drugs on the market.
That is because on July 13th and July 14th, the FDA will hold an advisory committee to review the safety data on GSK's drug Avandia.
I have blogged on the Avandia topic extensively. (In particular see Avandia Vindicated and Diabetes Conspiracy).
Below is a re-cap of all the major events in the saga as well as possible outcomes and my prediction as to what will happen.

Prelude- Vioxx (2001 -2004)
It is important to note that the Avandia story begins with Vioxx. Prior to Vioxx, many doctors prescribed heavily marketed drugs with less concerns about safety, thinking that if a drug had been approved by the FDA, they were likely safe. One of the major studies that led to fall of Vioxx was published in JAMA on August 22nd, 2001 by Cleveland Clinic cardiologist Dr. Steve Nissen. It took nearly three years until September 30th, 2oo4, when Vioxx was finally pulled from the market. Dr. David Graham, an FDA insider who works on drug safety was clearly frustrated by the FDA's apparent lack of concern. He was called to congress in November of 2004 to testify regarding what other drugs the public might be concerned about. Though Avandia wasn't named as one of Dr. Graham's five drugs to look out for, his testimony to congress against his own institution made it clear that Graham intended to be a whistle blower and would place the FDA safety folks at odds with the other branches of the agency

August, 2006- FDA becomes aware of potential Avandia issue
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, which surprisingly 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. These results were surprising because though Avandia was known to cause fluid retention and a possible risk of congestive heart failure (CHF), due to some of Avandia's other properties (lowering triglycerides, increasing HDL, decreasing CRP), if anything, Avandia should have prevented heart attacks.

Rather than create a public hysteria and mass panic, the FDA decided to take a wait and see approach for several reasons. First, the FDA 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 was a large, randomized trial designed specifically to answer the question of cardiovascular safety and Avandia. Though the results would not be available until 2009, the study had be going on for almost 2 years and an independent group which periodically looks at the study to assess safety (commonly done in safety studies) had not reported any problems. And in fact, both DREAM trial and the ADOPT trial not only showed that Avandia prevented diabetes and sustained glycemic control better than the other available diabetes agents respectively, there were NO signs of heart attack.

Thus, the FDA felt confident in their decision not to needlessly scare the public, but rather wait until about June, 2009 when the RECORD trial was likely to report out. Given that it was a large, randomized trial specifically designed to look at Avandia and cardiovascular safety, the RECORD study would be able to definitely answer the question once and for all.

May 2007- Nissen Meta-Analysis published.
Due to a previous settlement regarding their drug Paxil by then attorney general Elliot Spitzer, all GSK's clinical trial data was publicly available. Dr. Steve Nissen was therefore able to perform his own meta-analysis of the GSK studies. Given that there had been no public mention of Avandia heart attack concerns, I suspect someone inside the FDA tipped off Dr. Nissen. His controversial and often criticized findings of a 43% increase in heart attacks with Avandia were rushed to press and published in the New England Journal of Medicine. The news created a media frenzy. Congressional hearings were called for by Senator Charles Grassley the very next day (almost as if Sen. Grassley knew about the results before they were published). In the following week's issue of the New England Journal, the interim results of RECORD were published ahead of schedule, mainly due to concerns that the RECORD study itself could be jeopardized as patients might pull out of the study if Avandia were dangerous. 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 the controversy persisted.

July 2007- First FDA Advisory Committee Convened
Given the publicity of the Nissen study, the FDA met to discuss the safety of rosiglitazone. After review of all of the FDA's (not Nissen's) data, they saw something that didn't make sense. Their own meta-analysis showed a signal of myocardial ischemia, but analysis of the ADOPT, DREAM and RECORD trials, as well as analysis of large claims databases showed nothing. The FDA voted almost unanimously to keep Avandia on the market. However, they added a boxed warning to Avandia. This boxed warning remains one of the most confusing warnings I have ever seen, stating that there is contradictory evidence so no conclusion can be made. Nevertheless, physicians who may have been on the fence about prescribing the drug, likely stopped writing prescriptions for Avandia. The class of medications, called the TZD's, decreased in general use, and Actos, the competing TZD which had 50% of the market share of TZD prescriptions became the heavily favored product.

June 2009- RECORD study results published
During the ADA's annual meeting, the long awaited study results of the RECORD trial were finally published. The study that should have put the issue to rest showed absolutely no difference in cardiovascular issues with Avandia. In addition, patients who used Avandia had improved diabetes control and less use of insulin. At the previous years ADA meeting, two other randomized studies (ACCORD and VADT), which both used Avandia extensively also showed no issues related to Avandia. Given RECORD, ACCORD, ADOPT, DREAM and VADT there have been randomized, controlled trials of 26,000 patients that have been studied over 4 to 5 years, of which more than 15,000 patients took Avandia and showed absolutely no difference in heart attacks. Yet, critics (mainly Nissen and Graham), still had their doubts. The FDA decided that they would meet in July 2010 to review all the newest data and re-look at the controversial issue of Avandia.

February 2010- Grassley releases investigation of FDA and GSK
As above, Sen. Grassley called for an investigation back in 2007, and it took three years to finally release his findings. The report is an indictment of both the FDA for not taking Avandia risk seriously and GSK for hiding data from the public. Interestingly, the report fails to mention any of the studies published since May, 2007 showing Avandia to be safe. It as if time stood still in the Grassley report. The other issue that the Grassely report raises is the ethics of the TIDE study. The TIDE study, which the FDA requested GSK perform is a head to head study comparing Actos and Avandia in regards to safety. Grassely contends that since Avandia risks are known, the TIDE study is unethical and should not be performed.

June 2010- Nissen and Graham take one more shot
In conjunction with the June ADA meeting and weeks before the July FDA advisory committee, both Dr. Nissen and Dr. Graham have separate publications once again questioning Avandia's safety. (Both claim publicly that the timing of the release of this data is coincidental). The Nissen study is an update of his 2007 study, which both addresses some previous criticism of the original study and adds new data, including the RECORD study. Since this is virtually the same study as he did in 2007, it is not surprising that his findings are the same. The Graham study analyzed Medicare claims databases comparing Actos and Avandia. Graham claimed that Avandia was far riskier than Actos when looking at stroke, heart failure and death. Retrospective reviews of databases are legitimate forms of research, but have even more limitations than meta-analaysis. More interesting (which no media seems to be reporting on) is that Graham's study showed no difference between Avandia and Actos when it came to heart attacks. Stroke had never been a question, and even Nissen's analysis showed Avandia didn't increase death; CHF is a risk for both with multiple studies suggesting the risk is similar. Nonetheless, both studies got major media attention, overshadowing a 3rd study presented at the ADA (analysis of BARI-2D) which showed Avandia prevented stroke, heart attacks and death!.

July 2010- FDA Advisory Commitee
The FDA will review all the data mentioned above and hopefully come up with some conclusion. The meeting was planned in advance of Sen. Grassley's report, and was timed such that RECORD, BARI-2D and Graham's review could be incorporated with existing data. Expect debate to be particularly heated.

What are the possible outcomes?

Avandia Gets Pulled from the Market ( Odds 30-1)

This would also be the end of the TIDE study. This is a relatively unlikely scenario given that, with the exception of Graham's retrospective reivew, the only new data since 2007 are RECORD and BARI-2D, which are large, randomized prospective trials specifically designed to look at cardiovascular safety, both showing no issues with Avandia. Another intersting aspect, (which I blogged about in Rough Times for New Diabetes Drugs: The Diabetes Conspiracy Part III) is that because of the Avandia scare, the FDA developed new criteria requiring that ANY new diabetes drug that gets approved by the FDA must first show that it does not cause heart attacks. Specifically, the FDA now requires randomized clinical trials that show a drug does not cause heart attacks with the upper limit of confidence of 1.3. In other words, there needs to be 95% certainty that a diabetes drug couldn't possibly increase the risk of heart attacks by as much as 30%. The confidence intervals for the RECORD trial for heart attack fall under 1.3. Thus, Avandia has already fulfilled the FDA's own criteria of being a safe drug in regards to heart attack. How could they possibly pull it?

That said, removal of Avandia is not impossible. Folks like Graham, Nissen and Grassely have a lot at stake in this game. Avandia staying on the market might make them look pretty bad. Rest assured that they will not go down without a fight.

Avandia is exhonorated (10-1)

Given the results of RECORD and BARI-2D, along with analysis of the VADT and ACCORD trials and multiple data bases, the FDA could be so overwhelmed with Avandia safety data that it proclaims Avandia to have no risk of heart attack and pulls any mention of this from its label. However, with all the surrounding controversy, this seems a little unlikely. The FDA has been harshly criticized for being soft on safety. That said, the FDA might want to take a stand against outsiders (Nissen, Grassley) as well as insiders (Graham) trying to do their job for them.

Nothing Happens (1-1)

This is the most likley scenario. There are no clear winners or losers, so everyone can save face. In this scenario, the TIDE study would likely proceed as planned, but not necessarily. The problem with this scenario is that it leaves physicians and patients in the dark. Is Avandia safe or not? The FDA owes the public an actual opinion, unlike its stance in 2007, which essentially was "we dont' know." In this scenario, the devil is in the details. The FDA will have to update Avandia's label with regards to the cardiovascular safety . The label could be harsher towards Avandia (i.e. more studies suggest concern), reassuring (i.e. despite one meta-analysis a multitude of randomized trials show safety) or neutral.

Actos vs. Avandia?

Finally, I have been asked multiple time about Actos vs. Avandia regarding safety. The conventional wisdom is as follows: even if there probably isn't any cardiovascular risk with Avandia, why would you not just simply switch to Actos just in case? Why do we even have Avandia on the market if Actos exists?

I use both Acots and Avandia, and do believe there is a role for Avandia. There are two important things to consider with regards to this issue:

1. The absence of evidence does not equal the evidence of absence.
In other words, just because no concerns have been raised with Actos does not make it necessarily a safer choice. The vast majority of large, randomized clinical trials (the gold standard of science) have been done with Avandia and not Actos. Avandia beats Actos in trials almost 2:1. It would be like saying 5/10 studies show drinking Coke causes obesity in teens, but 0/5 studies show that drinking Pepsi does not cause obesity; therefore, we should have all teens drink Pepsi instead of Coke to avoid obesity. Some have argued that the reason there has been no signal seen in Actos is because of differences regarding each drug's affect on lipids, with Actos having slightly more benefit than Avandia. However, there has been no true head to head trial of both (which is why the TIDE trial would be scientifically beneficial), and the lipid data varies from study to study (cholesterol actually increased in Actos' one large, randomized trial called Proactive). Furthermore, since there are many more large, randomized trials with Avandia, none of which show an increased cardiovascular risk, one could make the argument that Avandia is the "safer" choice. I am not suggesting that Actos is dangerous. However, stating that Actos is a safer choice is not scientifically valid. The American Heart Association and the American College of Cardiology came to similar conclusions, stating no safety difference between the two products in their recent review of all the data (which was released ahead of schedule in response to the scathing Grassley report).

2. In my experience, the most effective dose of metformin is 1000mg twice daily. The most effective dose of each TZD is the maximum dose. If used early in the disease process, the typical side effects (edema) have been minimal, even at the maximum dose of each TZD. Thus, in my opinion, the perfect dose of Avandamet ( a product which combines Avandia and metformin) is 4/1000mg twice daily and the perfect dose of ActoPlusMet would be 22.5/1000mg. Avandamet is available in this does, ActoPlusMet is not. To give this dose of ActoPlusMet, patients would need two separate prescriptions, two separate pills, and two separate co-pays.

Addendum 7/9/2010
The FDA just posted the preliminary documents for next week's Avandia "smackdown."
A few tidbits:

1. I mentioned three possibilities for Avandia's fate: removal, label revision or no change. There is actually a little more subtlety regarding the label change. They might either remove altogether any mention of heart attack, add more warnings (such as recommending Avandia only be used if other meds fail), or add more warning AND restrictions to prescribing. See the FDA's exact wording below:

A. Allow continued marketing and revise the current label to remove the boxed warning and other warnings regarding an increased risk of ischemic CV events, or
B. Allow continued marketing and make no changes to the current label, or
C. Allow continued marketing and revise the current label to add additional warnings (e.g., contraindications for certain patient populations, recommendation for second-line use in patients intolerant of or uncontrolled on other anti-diabetic agents); or
D. Allow continued marketing, revise the current label to add additional warnings, and add additional restrictions on use (such as restricting prescribing to certain physicians or requiring special physician and patient education)
E. Withdrawal from the U.S. market.

2. One of the presentations regarding the TIDE study, talks about the importance of large randomized controlled trials, citing all the bad things we would still be doing without the data. I love this....

If modern medicine relied on epidemiologic data, administrative databases, & meta-analyses of small trials to identify effective therapies, we would be …
routinely using HRT (WHI)
suppressing VPBs post-MI (CAST)
doing internal carotid artery bypasses (EC-IC study)
using alpha blockers to prevent stroke (ALLHAT)
giving estrogen to men with CAD (CDP)
giving high-dose GIK infusions in the CCU (CREATE)
using perioperative beta blockers (POISE)

2. There is going to be considerable time spent on poking holes in the RECORD study, which is really the pivotal evidence for Avandia safety. One of the main criticisms of RECORD is that the expected number of events (which determines how many patients they need in a study) was much lower than expected. Some has therefore stated that RECORD was "underpowered" to show Avandia is safe. In fact, the AHA/ACC consensus statement hinges its findings on this. However, the study was indeed powered to show Avandia is safe. The FDA presentation looking at this issue confirms this stating "Despite the initial over-estimate of events, the trial had substantial power to achieve its specified goal." This bodes well for Avandia.

3. Dr. Unger, a cardiologist from the FDA states that "the results of RECORD do not substantiate the findings from the Nissen/Wolski meta-analysis on myocardial infarction and cardiovascular death." In other words, from a cardiologist's perspective, RECORD proved Nissen wrong.

4. Nissen is not going down without a fight. Nissen main tactics seem to be 1) poke holes in the RECORD study and 2) present Actos (using non-head to head comparisons, meta-analyses, and claims databases) as a much safer alternative. Again, note that Takeda (makers of Actos) fund his research.

1 comment:

Anonymous said...

Also, when you look at the analysis from BARI 2D - which was done in patients with actual CVD - there was a net benefit of stroke/MI/Death with Rosi. Post hoc, yes, but it was a well done prospective study with a couple thousand patients over five years.

Full writeup on