Tuesday, June 16, 2009

The Avandia Scare: Why it Matters, Who's Responsible, and What to Do

We now know that Avandia is not associated with increased cardiovascular risk.

I have blogged quite a lot regarding Avandia. You might ask why would I so strongly support and blog about a drug/medication that, even if it doesn't seem to cause heart attacks, is expensive and is known to have other side effects, like congestive heart failure and bone fractures?

Here's why:

Why it matters?

1. Patients, doctors and the public were unnecessarily scared. You only need to recall the story of the boy who cried wolf to know why false alarms can be problematic. There were many patients who were unnecessarily worried that the medication their doctors prescribed might be killing them. Similarly, physicians were worried that a medication they thought they were giving their patients to help them, might have caused harm. The reaction can be seen in the precipitous drop in Avandia sales. However, it wasn't just that patients stopped taking Avandia and switched to its competitor Actos. Though there are now far many more prescriptions written for Actos than Avandia (was about 50/50 before the Nissen article), the entire class of TZD medications declined. Also, though there were increases in other drugs like metformin, many patients simply stopped taking their diabetes medications, and many of those patients did not tell there doctor about this. (We have documented this in our large, academic faculty practice and are in the process of publishing the results).

In addition, the Avandia scare was probably one of the key events that prompted the FDA to start warning the public about issues they were looking into BEFORE their analysis was complete. I blogged about this in March, 2008 in my post More FDA warnings should not be cause for worry. Thusfar, all of these "early warnings" have not panned out, and have undoubtedly scared more than a few concerned patients. Spiriva and Stroke? Doesn't' seem to be associated. Singulair and Suicide? Well the FDA just released their findings and though the label to Singulair was changed, the data does not seem to support a major concern, and the FDA's concerns were significanatly downgraded from the original warning. In other words, asthmatic and COPD patients on these drugs would have never needed to worry about these "warnings" if it wasn't for the Avandia scare.

This is not to say that we shouldn't take adverse effects of drugs seriously. Vioxx taught us that lesson. The Vioxx scare made everyone take a careful look at drug safety including the reporting and approval process. However, the Avandia scare added unnecessary fuel to the fire to the point of hysteria. Now, many patients will question the necessity of every medication a doctor prescribes. This itself is not inherently bad (an activated, inquistive patient is a plus), but it's the fear that causes this questioning that is problematic.

2. We need Avandia (and medicines like it)
Most diabetics are not able to be controlled on only one pill. In fact, of the two older, generic pills; we know that most patients on these medicines will eventually fail therapy. Analysis have shown that though both metformin and sufonlyurea lower blood sugar; about half of all patients taking either of these medications lost diabetes control by three years, with the vast majority losing control at 9 years.

Avandia, was the first medicine to show that it could keep patients' diabetes under control. The ADOPT trial showed that Avandia, a thiazolidinedione or TZD, substantially reduced loss of diabetes control compared to both metformin (Avandia 32% better) and sufonlyurea (Avandia 63% better). TZD's work by reducing insulin resistance, and also can improve beta cell dysfunction. Both effects were seen in ADOPT. The recent RECORD trial has now also showed this effect. Actos, another TZD, also seems to do this as well. Finally, some of the newer DPP4 medications like Januvia and (hopefully) soon to be available saxagliptin (Onglyza) have shown to preserve pancreas beta cell function in animals and human studies are only now just getting under way.

The point is that though the older medications are needed, they are not enough. Newer medications like Avandia have value, most importantly keeping blood sugar under control and preventing the need for insulin.

3. Patients were harmed.

The damage done to patients from the Nissen publication is hard to quantify, but in addition to unnecessarily worrying patients, patients were definitely harmed by this poorly done study and the subsequent media frenzy. As mentioned, many patients simply stopped taking their medications. It is not known what ill effects this might have caused. More importanly, those doctors who stopped using the TZD class altogether likely started prescribing more and more insulin. We now know from both RECORD and BARI 2D that not using TZD's like Avandia can double to triple the amount of patients needing insulin. Though the exact number from RECORD is not published, based on the what was presented at the ADA, I calculate that it takes at most 10 patients treated with Avandia for 5 years to prevent at least one patient from needing insulin. Put another way, for every 10 patients who stopped Avandia, one likely needed to go on insulin. Based on pharmacy data (via drugtopics.com, thank Mr. Medsaver) , there weree 11,331,000 prescriptions for Avandia and 11,329,000 for Actos in 2006. In 2008, that dropped to 3,103,000 and 12,518,000 respectively. Assuming that both Actos and Avandia prevent the need for insulin, and assuming that about every 6 scripts written represents one individual patient, over one million diabetics likely stopped taking Avandia or Actos, and thus the Avandia scare likely caused over 100,000 type 2 diabetic patients to need insulin, which could have been avoided.

Patients don't like insulin. There is the cost of not only the medications (most doctors are using the newer, expensive insulins, not the older ones), but also the strips and the machines to check blood sugar. In addition, patients fear low blood sugar, or hypoglycemia, which can be fatal. I have blogged about The Problem With Insulin (as well as The Problem with Insulin- Part 2)

4. The entire approach to the way diabetes is managed has forever changed.

Dr. Nathan, an endocronologist from Harvard, is basically the current architect of the ADA's guidelines for the management of type 2 diabetes. In a Perspective piece in the New England Journal of medicine , he shows that he does not like new diabetes medicines stating that newer diabetes drugs like Avandia and Januvia , "are generally no more potent, and often less effective in lowering glycemia, than the three oldest classes (insulin, the sulfonylureas, and the biguanides), all of which are more than 50 years old. Moreover, the newer classes are uniformly more expensive and are associated with adverse effects."

Despite the results of ADOPT, in an early guideline, Nathan stated that TZD's were merely a secondary option. In the newer consensus statement, mostly because of the Avandia scare, he further pushes for the older medicines (metformin, sulfonylurea, and insulin) as the preferred medications to use, leaving Actos as only a second line alternative, and taking Avandia and Januvia off the map. Though not all endocrinologists necessarily agree with this approach, many primary care physicians follow what the ADA recommends. In addition to TZD benefits presented at the recent ADA meeting, there are other examples of gains for diabetic patients, specifically saxagliptin, which is now proven to show no cardiovascular side effects and may even prevent strokes, and once weekly exenatide (Byetta). This insulin before newer medications approach, which is promoted by Nathan and the ADA is, in my opinion, not good for patients.

5. It is now much harder for ANY new diabetes medications to be availble for patients.

The Avandia scare had another MAJOR impact. Because of this supposed risk of increased heart attacks, the FDA now requires ANY new diabetes medicine to prove that it doesn't cause heart attacks. I talk about this in more detail in Rough Times for New Diabetes Drugs: The Diabetes Conspiracy Part III. This finally came to fruition several months ago when allogliptin got it's approval delayed. (This is somewhat ironic since alloglipitin's maker is Takeda, who also makes Actos, Avandia's competitor. Takeda was banking on alloglipitin more than GSK was counting on Avandia for it's future, so even though many docs switched from Avandia to Actos, it is possible that in the long run, the Avandia scare will hurt Takeda far worse than GSK).

6. The false alarm was motivated by politcs, ego and greed. Not all false alarms are bad. However, there is a difference between someone yelling "fire" when they smell smoke, and the school boy pulling the fire alarm to get an earlier recess. Prior to the Nissen meta-analysis, no one thought that Avandia would cause heart attacks. In fact, if anything, the thought was that TZD's like Avandia might actually prevent heart attacks. In fact, in one of Nissen's own studies Actos (a competitor to Avandia) was shown to decrease the progression of plaque build up. (Yes, this might be a conflict of interest if Nissen is attacking a drug whose competitor funds his research). To further describe those who may have less than pure motivations, see below.

Who is to blame?

1. There is no questio n that cardiologist Dr. Steve Nissen is in part to blame. He is very smart and an excellent researcher. He knows that some of the techniques he used in his analysis were not up to par, such as excluding any Avandia study where there were no heart attacks from him analysiss. There are many flaws and critics of the this meta-analysis, as well as an alternative analysis that found completely different results.

2. The New England Journal of Medicine is also to blame. This is one of the most prestigious medical journals. To ensure the studies they publish are of the highest quality, they use a process of peer review. This means that when a study comes into a journal for publication, the editors first look at this to see if it is of interest to the journal and if so, sends it out to several experts in the field. These experts comment on the study, and based on these comments, the journal chooses whether or not to publish. Before publication, the author needs to address the comments by the peer reviewers and re-submit to the editor. This process generally takes several months. The New England Journal released the study to the public in just 3 weeks after Nissen submitted it. It seems impossible to me that The New England Journal followed it's own policies, especially with such an important, confusing and flawed study. The study was so confusing that one of the reviewers contacted GSK to make sense of the data (this was a major no-no, and the doctor got a lot of negative publicity about this move, but the point is that one of the experts could not make sense of a study that was published in just 3 weeks). Had the peer review process been followed strictly, the study may never have been published or the results would have been modified and likely gotten much less publicity.

3. An FDA insider. Until the Nissen publication, Avandia had known potential problems with heart failure, but no one was thinking about risk for heart attacks, and if anything possibly preventing heart attacks. Even if Nissen has combed through all of GSK's data on his own, now publicly available thanks to Eliot Spitzer, finding heart attacks with Avandia would be like finding a needle in a haystack. The only folks who knew about the problem were the FDA, and they decided not to release the data to the public, because other data was contradictory. The fact that 42 studies (not all the same) were used in both the Nissen and FDA meta-analysis seems more than just coincidental. I have no actual knowledge of any communication, but I strongly suspect that someone inside the FDA tipped off Nissen to the data that GSK submitted to the FDA, because they did not agree with the head of the FDA's decision not to make this data public.

4. The FDA in general.

The FDA made the right decision to hold off on any conclusions regarding GSK's surprising meta-analysis, since it knew DREAM and ADOPT would soon be published and there was no problems with RECORD study currently underway. When the Nissen paper was published in the NEJM , the FDA could have come out strongly stating it's reasoning for not acting on GSK's meta-analysis, and because by this time ADOPT and DREAM were published and showed no increased cardiovascular risk, the FDA could have clearly shown why their decision was right. In addition, another opportunity for the FDA to make a strong statement reassuring the public came the following week when the RECORD interim analysis also showed no increased cardiovascular risk. Instead, the FDA just bent over and took the criticism from the media, bloggers, Congress and others.

What should be done?

Because the Avandia scare caused unnecessary worry for doctors and patients, harmed patients, forever changed the way diabetes is managed, and made it more difficult for new diabetes medicine to be approved; because we need medicines like Avandia; and because the motivations behind the Avandia scare may not have been in the best interest of patients; something must be done. I believe a full investigation is in order, starting with the FDA. I would like to know whether Nissen communicated with anyone inside the FDA that leaked confidential information. I would also like an investigation into the New England Journal of Medicine. Though we do have freedom of the press, medical journals must be held to a higher standard, since doctors use information published in these journals to make life and death decisions. If the New England Journal has policies in place to ensure only high quality data is published and they did not follow these policies when publishing Nissen's study, they too should also be held accountable. I have sent emails to the Department of Health and Human services. I hope they respond.


james gaulte said...

I continue to be impressed with your through,thoughtful reviews and tireless effort in this regard.From the first I was skeptical that a single meta analysis should have stirred up so much trouble.My question is why did NEJM rush to judgment on this issue and publish so soon ?
I realize you have no special powers to discern motivations but I would be interested in your thoughts.

Dr. Matthew Mintz said...

thanks James, and excellent question. I am not exactly sure why the NEJM "fast tracked" the publication, but I suspect it's because:
1. They knew what a stir the Nissen study would create, and more attention for the NEJM means more business.
2. They probably knew that the FDA had it's own meta-analysis and if word got out that the NEJM was going to release a similar one, the FDA would try to get the data out first so as not to look bad (which is what happened). NEJM probably realized that they needed to act fast before the FDA got wind, and likely bypassed some of its peer review policies.
3.Another possibility (which is complete speculation) is that there was some political pressure. Congressional hearings were called for immediately after the Nissen study hit the press, almost as if the congressmen knew about the study beforehand. If that's the case, maybe they leaned on the NEJM to get the story out ASAP.

Anonymous said...

This is really an excellent analysis, and I think its pretty close to the truth. I hope someone from a significant media source picks up on this and makes it public.

I really dont think anyone will investigate this at all... the primary parties are all too invested in it.

GSK? They dont dare whistleblow on the FDA for fear of future reprisals. And all pharma has learned to never, ever cross Nissen.

The ADA? The American Diabetes Association would never turn on David Nathan.

Someone complain about Nissen? No drug company would complain about him. Look how many torpedoes he has fired and how many drugs he's sank. And Takeda loves him. Just watch his positive studies that will be done on Alogliptin and Actos. Heck, they even are naming their new drug after him...

Roy M. Poses MD said...

I don't believe that the results of RECORD support your contention that the concerns about cardiovascular risks of rosiglitazone were a "false alarm," or crying wolf.

In fact, RECORD had several problems that prevented it from proving that rosiglitazone does not elevate the risk of myocardial infarction (MI, or heart attack). These were:

1. Possibly because RECORD was an "open-label," rather than a double-blind trial, patients in the rosiglitazone group received different co-interventions than patients in the control group. In particular, about 10% more patients in the rosiglitazone group (55.2% vs 46.0%) were receiving statins at the end of the trial than were patients in the control group. There is pretty good evidence that statins reduce the risk of coronary artery disease, and hence, of MIs. Thus, thiis version of co-intervention bias may have decreased the apparent rate of MIs in the rosiglitazone group.

2. The RECORD trial suffered from a significant loss to follow-up. For 8.9% of patients, ascertainment of cardiovascular outcomes was incomplete. Since the rate of MIs in the whole group was less than 3% over 6 years of observation, it is possible that had all these cardiovascular outcomes been ascertained, the ratio of MIs in the rosiglitazone and control groups would have turned out quite differently, either indicating less, or more risk due to rosiglitazone.

3. The RECORD trial was too small to prove that rosiglitazone did not raise MI risk, possibly substantially. The 95% confidence interval for the risk of MI due to rosiglitazone was (0.80, 1.63). A somewhat crude way to interpret this is that the chances of rosiglitazone causing a risk of MI higher than 1.63 was less than 5%. On the other hand, the trial could not disprove that rosiglitazone raised the risk of MI by as much as 1.63 times. (Note that the point estimate of risk of MI was 1.14, meaning that the best guess of the effect of rosiglitazone on MI risk was that it raised it by 1.14 times, or 14%)

Moreover, I do not see how you can justify your point that patients who discontinued Avandia were harmed. Can you show any evidence that the drug provides clinical benefits (that is, to do with outcomes that actually affect patients, not laboratory values they cannot feel) that outweigh its harms? In the absence of such evidence, how could a patient be harmed by withdrawing the drug (and perhaps substituting something else)?

There may be some evidence that the drug improves diabetic control when added to other, older drugs. But it is not the only drug that can do this. Furthermore, I do not believe there is clear evidence that tight glucose control in Type 2 diabetes produces clinical benefits that clearly outweighs the risks of hypoglycemia. For example, VADT did not show a statistically significant decrease in the primary outcome, but did show an increase in hypoglycemic events in the intensive-treatment group. ADVANCE showed that intensive control reduced nephropathy (absolute decrease 1.1%) but increased severe hypoglemia (absolute increase 1.2%)

Finally, shouldn't you disclose that you have served on an advisory board and the speakers' bureau of GlaxoSmithKline, the manufacturers of Avandia?
See: http://cme.medscape.com/viewprogram/6294 and

Anonymous said...

The same thing--almost--happened in 2000, when "consumer activists" got Rezulin pulled from the market, to be replaced by Avandia (and Actos). We will never know the real benefit-risk profile of Rezulin, but at least we have more data now on Avandia.

See my 2000 op-ed in the Wall Street Journal, "Rezulin Proves the System Works."
Gilbert Ross MD www.acsh.org

Anonymous said...

Given that Dr. Mintz is quick to insinuate that conflicts of interest are partly to blame for the Avandia scare, I think it is only fair that he disclose all potential conflicts that he may have.

The Nissen meta-analysis may be flawed, but the RECORD trial is not conclusive either. Furthermore, the impact of lower Avandia usage on patient outcomes is merely speculation on Dr. Mintz's part, unless he has actual data (and not just extrapolations and assumptions) to support his viewpoint.

Rhonda Miner said...

Ironic, isn't it? I have had Type 1 diabetes for over 40 years and the simple logistics are, most diabetics will die from 1) diabetes itself, 2) heart disease, and/or 3) kidney failure. You already have a fractured equation when you consider that 33 to 40% of diabetics will have a heart attack anyway.