Avandia Vindicated (Again): The Diabetes Conspiracy Part II

Late this morning, results from the APPROACH study were presented at the American Heart Association. Unlike the JUPITER study which I blogged about the other day, this study seems to have generated little press (at the time of this posting only AP press and MSNBC). However, already the press makes it seem like Avandia remains a bad drug, with headlines stating:

"Controversial diabetes med doesn't slow plaque"

I recently posted (The Diabetes Conspiracy (Part I) ) that Avandia and Januvia were taking the heat for increasing the cost of diabetes care, yet no one seemed to comment on the newer insulins which in addition to contributing about equally to the increase in cost of diabetes care, were being prescribed at a much faster pace then either of the two newer diabetes pills.

I have previously blogged about the Avandia scare (for details see Avandia Vindicated! ). Briefly, in May 2007, Dr. Steve Nissen published a highly publicized, as well as highly criticized meta-analysis of 42 studies stating that Avandia caused heart attacks. The study was published ahead of press in unprecedented fashion, and congressional calls for investigation were sounded immediately (almost as if the politicians knew in advance). The FDA found something different. Though their similar analysis did find an association with myocardial ischemia (not heart attack) with Avandia, it was only when Avandia was compared to placebo and not other diabetes drugs. In addition, their analysis of three large, randomized prospective studies (RECORD, ADOPT, DREAM) and several large databases of patients showed no evidence of a problem. Since the FDA's analysis, several other large diabetes studies in which patients took Avandia have been published, all showing no risk of heart attack with Avandia. We now have randomized studies where over 26,000 patients have been studied for over 3.5 years, of which more than 15,000 patients took Avandia and showed absolutely no difference in heart attacks or myocardial ischemia.

The APPROACH study was a prospective, randomized, double-blind study which looked at 672 diabetic patients who needed to have an angiogram or angioplasty (thus, high risk for heart attack). They looked at one of the coronary vessels that was not going to be intervened on and measured the amount of plaque build up. Half the patients got Avandia and the other half got glipizide, an older, generic diabetes medication. Patients took either drug for 18 months, and by the end of the study, compared to glipizide, patients on Avandia had an increase in their high-density lipoprotein (HDL) (or good cholesterol), lowering of their blood pressure, lowering of their triglycerides and lowering of their CRP (which thanks to JUPITER seems even more important). There was a small increase in low-density lipoprotein (LDL, “bad” cholesterol) in patients taking Avandia. The investigators used intravascular ultrasound (IVUS) to measure the buid up of plaque. They looked at change in % atheroma volume (primary endpoint) as well as change in total volume, and change in the worse segment. The patients taking the older diabetes medicine had an increase in there % atheroma volume (0.42%) (more plaque= bad) where the patients taking Avandia had a decrease in their % atheroma volume (-0.22%), but this was not statistically significant (p=0.12). Though there was similar non-significant reductions in the worst segment, there was a decrease in the total atheroma volume which as statistically significant.

Technically, this is a "negative" study because the primary endpoint was not found to be statistically significant. However, this is not "bad news" for Avandia, nor does it mean it is a bad drug.
1. Patients on the older generic drugs had increase in plaque build up, Avandia patients did not. The p value of 0.12 means there is an 88% chance these findings are true. By convention, we only accept the results if there is more than 95% chance. However, given the statistically significant decrease in atheroma volume as well as positive changes in in BP, CRP, etc. if anything it would appear that Avandia leans toward reducing risks for heart attacks. Though this study was small, these patient were all at high risk for heart attacks (they were having angioplasties!). Yet, there was no difference in heart attacks in patients taking Avandia. Thus, this study provides even more evidence that Dr. Nissen's flawed meta-analysis was completely wrong!

2. The patients on the older diabetes drugs had more hypoglycemia. I will blog on this in more detail, but hypoglycemia is not trivial. It can be life threatening. Avandia does not cause hypoglyccemia.

3. A virtually identical study was done with the other TZD on the market, Actos. The PERISCOPE study had almost identical findings. In that study, the patients on Actos had a 0.16% decrease in their % atheroma volume (numerically not as good as Avandia, but probably the same). However, the patients taking glimepiride (a similar, older, generic diabetes medicine) has a 0.73% increase in % atheroma volume (the patients on the generic diabetes medicine in PERISCOPE had more plaque build up the the patients taking the generic medications in APPROACH). However, this difference was statistically significant. It is possible that APPROACH was a "negative" study and PERISCOPE was "positive" because the patients on the older medicines did worse in PERISCOPE. Drug companies and diabetes experts will likely debate this for years. The PROACTIVE study showed that Actos actually prevented heart attacks in strokes in patients with high risks for diabetes. My point is not to say whether Actos is or is not different from Avandia. The point is that the study designs and findings of both studies were similar, suggesting that if anything both Actos and Avandia would protect you from having a heart attack, and not cause one.

Already, the media is quoting our old friend Dr. Steven Nissen on the findings of APPROACH. He claims that these results "give one more reason" to use Avandia with caution. Oh, by the way, do you know who the lead investigator for the PERISCOPE study was? Yep, Dr. Nissen. Guess who funded his research? (Hint: Takeda makes Actos). Nissen did another meta-analysis on Actos published in JAMA that unlike his NEJM Avandia meta-analysis, found no problem with Actos. Many of those patients in that analysis came from the PROACTIVE study. Please note that Takeda gave Dr. Nissen and friends $25,000 to help analyze the results.

Bottom Line: It's too early as I blog to judge media reaction. They may give this study very little attention. What attention this study does get will make it sound like this is yet another problem for Avandia, which will be reinforced by comments by Dr. Nissen. The reality is that though the primary endpoint of APPROACH was not statistically significant, both Avandia and Actos likely prevent, not cause heart attacks. One should be very skeptical of Dr. Nissen's determination to take down Avandia. He is just one of several players in the Diabetes Conspiracy.