Monday, October 31, 2005

The pioglitazone controversy continues

Pioglitazone (Actose) and rosiglitazone (Avandia) are members of the thiazoladinedione (TZD) class of oral medications for type 2 diabetes. These agents have been surrounded by controversy because their predecessor in the TZD class, troglitazone (Rezulin), was withdrawn from the market because of liver toxicity. Although their beneficial effects on metabolic risk factors for macrovascular disease have long been known, outcome based data regarding protection against such events have been lacking until very recently and the Public Citizen Health Research Group has placed TZD drugs on its “do not use” list.

Recently the PROactive Study (PROspective pioglitAzone Clinical Trial In macroVascular Events) demonstrated the efficacy of pioglitazone in preventing macrovascular events. I blogged it here following presentation of the results but shortly after the announcement and before final publication a BMJ opinion piece was harshly critical of the study. My response to the BMJ commentary is here.

Now PROactive has been published in Lancet October 8 along with a commentary by Hannele Yki-Järvinen. Unlike the BMJ editorial the Lancet commentary acknowledges the clinical benefit of pioglitazone in decreasing macrovascular events. The major controversy about the PROactive results concerned the lack of statistical significance for the primary outcome and reliance on the secondary outcome. But Yki-Järvinen points out that inclusion of procedure related endpoints in the primary outcome could have biased the results against pioglitazone and implies (as I said before) that the primary outcome would have reached statistical significance with a longer follow up period, as the curves were diverging at study’s end. The commentary poses questions about the clinical significance of the increase in heart failure and how it might counterbalance the improvement in vascular outcomes.

So is it time for Public Citizen to change its “do not use” recommendation? They defend the recommendation on the basis that the TZDs “may be less effective than other drugs for diabetes and cause liver damage, weight gain, anemia and heart failure.” PROactive and other evidence suggests that this statement may be unfounded. How can the question of effectiveness of TZDs compared to other agents be answered? Because different classes of medication for diabetes have mechanisms of action which are complementary to one another it may be simplistic to ask whether one class of agents is as effective as another. It now appears that pioglitazone can join metformin as another agent capable of improving macrovascular outcomes. As for liver damage, none was found in PROactive. Weight gain (4 kg more than placebo) was seen, but the clinical significance is unknown and anemia was not mentioned. The problem of heart failure remains troubling although no new heart failure concerns were raised by the study.


eauersperg said...

As many other people have pointed out, inclusion of procedures (eg amputation) DOES NOT introduce bias in the primary endpoint. How can it in a blinded randomized trial? This is an argument that was originally proposed by the study authors, and was a pathetic attempt to exclude data that they didn't like.

And don't forget how they defined their endpoints. Heart attack, which pioglitazone seems to reduce slightly (though no better than other diabetic drugs), includes any chst pain combined with the slightest blip of an increase in troponin. Therefore, many of the events they prevent were trivial in size. I myself would far rather have a so-calld heart attack than an above-the-ankle amputation, which pioglitazone significantly INCREASED.

Read my other comments (and those of others) regarding the numerous other flaws in the design and reporting of this trial in the BMJ online comments. Not surprisingly, the Lancet has refused to publish several letters criticizing the ProActive study.

Ed Auersperg

Robert W Donnell said...

Thanks for your comment. There are two things I'd like you to clarify, though.
1)You said pio doesn't reduce MI any better than other diabetes drugs, right? With the possible exeption of metformin I didn't know the others did decrease MI. Certainly not insulin or sulfonylureas, do they? Please elaborate.

2) You seem pretty well convinced that pio causes an increase in amputations. That would be world class breaking news! Can you propose a model as to how it does that? Thanks.

eauersperg said...

I appreciate the chance to discuss this with you. To answer your two questions:

1. If you check the UKPDS results (Lancet 2005; 352: 837-853) the table with the main results shows with insulin almost percisely the same results as were obtained with pio in the ProActive trial. ie some cardiovascular endpoints were reduced dramatically, some a little, some were not affected, some a tiny bit worse, but no individual endpoint reached statistical significance.

They studied two sulfonylureas, one of which looked bad, the other not as impressive as insulin.

2. I wouldn't say I'm "convinced" that pio increases amputations, but the effect was pretty dramatic, and shouldn't be ignored just becasue we don't like it. As for mechanism, I've been a doctor long enough to appreciate how much of medicine and biology simply doesn't make sense! As I get older, I am becoming less of a biologist, and more of an empiricist. For example, I used to understand perfectly how estrogen and folic acid and vitamin E each prevent heart disease....but once the megatrials were done, I can't explain why they don't! The PPAR Gamma receptor is involved in much more than just glucose and lipid metabolism and blood pressure. It's late at night, but I think I recall involvement in immunity, so perhaps it impairs the ability of people with diabetes to clear foot infections. As a calcium channel blocker of sorts, maybe pio creates a steal phenomenon. But hey, I don't think anyone knows how Tylenol works, so I don't feel obliged to explain how pio (perhaps) significanlty increases the risk of SERIOUS amputation.


eauersperg said...

I had a few minutes, so I searched "ppar", "gamma", and "infection". PPAR gamma is indeed involved in the inflammatory response, and TZDs have "anti-inflammatory" properties. Indeed, I came accross a trial that has been registered to study the effect of rosiglitazone on endometriosis!

So, as an "anti-inflammatory", pioglitazone could degrade the ability of the immune system to fight limb-threatening infections, and could at the same time mask symptoms which would otherwise trigger earlier antibiotic therapy. Combine that with worsening of edema, causing impairment of venous return and thus impairment of circulation, and I would say we have a biologically plausible explanation for a major increase in crippling complications of pio therapy.


Anonymous said...

So what would your suggest a newly diagnosed type 2 diabetic take that will not cause them to lose a leg? My doctor has had me on Avandamet 4/1000 along with a strict diet and exercise combination for about a year now and my blood sugar is excellent (6.0 A1C) but I do not want to take anything that is going to increase to risk of an amputation.

eauersperg said...

Obviously I can't recommend treatment to a person I've never met, but, in general terms, diet/exercise are clearly hugely important, as the American diabetes prevention trial showed so dramatically. As far as drugs are concerned, as a general rule, metformin is the hands down winner for the 80-90% of patients who don't get unacceptable GI side effects. Its' risk of causing lactic acidosis has been hugely overblown by those with an interest in peddling more expensive drugs.

The question is what should be your second choice if metformin is contraindicated or not tolerated or not sufficient. Acarbose is very popoular in Europe, and I think underused in North America. But the usual second choice these days is sulfonylurea vs TZD.

To answer that, you first have to answer a fundamental question. As I put it: are you a scientist or an empiricist? In other words, do you do things that "make scientific sense on paper" (eg estrogen lowers blood pressure and improves lipids so it probably prevents heart attack....we know now it's probably not useful for that), or do you wait for observational evidence that they work without caring too much about how (eg we don't know how Tylenol works to relieve pain, but we use it anyhow). As I get older and more jaded, I keep remembering that, on paper, the Titanic was unsinkable!

Thus, if you are a scient, you would say that pioglitazone reduces sugar, so it will likely reduce blindness and kidney disease and neuropathy, and it improves lipids and CRP and BP so it will likely prevent heart attack. On the other hand, if you are an empiricist, you would say we have proof (UKPDS) that sulfonylureas prevent eye/kidney/nerve disease, but we don't have proof that pioglitazone prevents anything.

As an empiricist, as a rule I prescribe sulfs before pio, but to be honest, I do prescribe pio fairly often. To the extent that we are convinced by the ProActive trial (and to be fair, I think it is safe to say that pio probably helps a bit to prevent vascular events), I would probably reach for pio before rosiglitazone (Avandia), since we have no endpoint studies at all for rosi.

I hope that's useful to you....if I get any more specific, your lawyer will probably sue me in the event our planet is destroyed by a meteor!

Anonymous said...

Refer to Diabetes for
useful information