Tuesday, September 18, 2007

Avandia and Actose: Are there lessons for diabetes care and drug safety?

In light of new developments in the thiazolidinedione (TZD) controversy now might be a good time to take stock of where we are in drug therapy of type 2 diabetes. First there’s the latest spin from the New York Times, an article worth reading principally as an example of popular media distortion.

It mentions two new studies in JAMA and notes a doubling of heart failure risk from Avandia without making a clear distinction between TZD associated heart failure and Avandia associated myocardial infarction (MI). TZD associated heart failure exacerbation was reviewed in the Cleveland Clinic Journal of Medicine. The review makes this statement:

The incidence of congestive heart failure reported in clinical trials is less than 1% and appears to be related to underlying dysfunction, with decompensation caused by sodium retention and fluid accumulation rather than a direct cardiac suppressive effect.

This important distinction from drugs which are truly cardiotoxic is known to physicians but not to most Times readers, who might think Avandia is a cardiac poison like adriamycin. The distortion gets worse. One of the JAMA authors is quoted as saying that Avandia causes blindness and doubles the rate of fractures in women. The Times doesn’t bother to elaborate or even cite primary sources. As the statement stands it’s a half truth.

The blindness comment refers to macular edema. The primary source is here in a report from CMAJ. According to that report macular edema due to Avandia is believed to be rare, may reverse with drug discontinuation, may be anticipated by the development of peripheral edema, and does not invariably lead to blindness. The New York Times article left out another detail---insulin therapy in patients with type 2 diabetes is an established risk factor for macular edema. [1] [2].

And the fractures? Because the article provides no elaboration or background, readers might naturally assume these are hip or spine fractures. But the primary source, a study in NEJM from last December, reveals no increased rate of hip or spine fractures. The problem was confined to arm and foot fractures. That finding, although worrisome and in need of better understanding, is less concerning than what would naturally be inferred from the Times story, which confuses more than it informs. Unfortunately this type of sound bite coverage will again hijack scientific debate on the findings of the new JAMA studies.

In refreshing contrast is a discussion between Steven Nissen, author of the controversial NEJM Avandia meta-analysis, and Robert Califf, a cardiologist and Professor of Medicine at Duke University. The video and transcript are linked at Medscape Diabetes and Endocrinology. The Medscape editor explains the importance of this resource in the introduction:

The May 2007 online publication of the meta-analysis "Effect of Rosiglitazone on the Risk of Myocardial Infarction and Death From Cardiovascular Causes," in The New England Journal of Medicine (NEJM) launched a controversy that has been difficult to navigate. In an attempt to clarify the debate, our colleagues at theheart.org organized a videotaped conversation between Steven E. Nissen, MD, MACC, Medical Director, Cleveland Clinic Cardiovascular Coordinating Center, Cleveland, Ohio -- and co-author of the NEJM article -- and Robert M. Califf, MD, Professor of Medicine, Vice-Chancellor for Clinical Research, and Director, Translational Medicine Institute, Duke University, Durham, North Carolina. Drs. Nissen and Cardiff review the meta-analysis and the US Food and Drug Administration (FDA) advisory panel hearing, as well as discuss the wide-ranging implications of the ongoing clinical and political debate. It is a thoughtful, collegial conversation that helps parse this controversy and what it means for clinical practice and public policy.

It’s must viewing for anyone who wants the fair and balanced version of this debate but so far it’s been drowned out by sensationalistic media coverage.

The New York Times distortion centered around two studies and an accompanying editorial published in the most recent issue of JAMA. One of the papers, another rosiglitazone (Avandia) meta-analysis, reached the same conclusion as Nissen’s NEJM study. The JAMA meta-analysis, which differed from the NEJM study in that it restricted papers for inclusion to those which reported long term follow up and had a pre-specified intention of looking at adverse cardiovascular outcomes, showed a relative risk for myocardial infarction of 1.42 (1.06-1.91) associated with rosiglitazone and no increased risk of mortality.

Although not mentioned as a potential conflict of interest, the paper could be interpreted as more ammunition against the FDA. One of its authors, Curt Furberg, has been critical of the FDA’s approval and safety monitoring procedures and has had conflict with the agency. When removed form an FDA advisory panel meeting on arthritis drugs in 2004 Furberg expressed concern that agency officials were trying to silence him. The same Washington Post article quotes from an FDA official:

Sandra Kweder, deputy director of the FDA Office of New Drugs, said it was not unusual for advisory panel members to be kept from participating in a meeting if they have clear financial interests or intellectual positions that could keep them from being objective.

The importance of the potential conflict is underscored in this analysis by the fact that only 4 studies out of 140 were selected. Despite this caution the analysis must be taken seriously. It agrees with other research findings and when combined with other reports it strengthens the “signal” for macro vascular harm associated with Avandia.

The other paper, though conflicted due to co-author Steven Nissen’s ties with Takeda pharmaceuticals, the makers of the study drug, also confirms previously known findings. Pioglitazone (Actose), Avandia’s competitor and companion TZD, was found in a meta-analysis to be associated with a reduction in the primary composite outcome of death, myocardial infarction and stroke. This is in line with the controversial PROactive study. Thus it would appear, surprisingly enough, that the two approved members of the TZD class have divergent effects on macro vascular health, once again challenging simplistic assumptions about class effect. (Mea culpa!).

The accompanying editorial gives a helpful clinical perspective and makes suggestions for change at the FDA. Unfortunately label revisions and black box warnings are not enough. As the editorial points out and as I wrote in a previous Medscape Roundtable it has been demonstrated again and again that doctors ignore such warnings. This has not only led to untold instances of direct patient harm but has also deprived patients of beneficial therapy due to the withdrawal of effective drugs form he market. It has unfairly tarnished the reputations of drug companies and the FDA.

Unfortunately this debate has been fueled by a fundamental and pervasive misunderstanding of diabetes treatment. The problem is an inappropriate expectation that anti-diabetes drugs should improve macro vascular outcomes. Myocardial infarction and stroke are examples of macro vascular outcomes. While it’s well established that glycemic control improves micro vascular outcomes (e.g. retinopathy, neuropathy and nephropathy) clinical epidemiologists have been telling us for years that not even the strictest glycemic targets produce macro vascular benefits. In fact it has been known long before the advent of TZDs that drug treatment for type 2 diabetes is associated with macro vascular harm. This fact was recognized decades ago when the University Group Diabetes Program (UGDP) reported increased cardiovascular mortality in association with first generation sulfonylureas.

Although the UGDP study was criticized its findings have been supported in subsequent reports such as this recent analysis of a large Canadian health care database. Not only first generation but also second generation sulfonylureas were associated with cardiovascular death, and the association was dose related. These alarming findings, published last year, went largely unnoticed in the arena of public debate. Where was the outrage? Where was the New York Times? There can only be one explanation. The fact is, these are old, off patent drugs from decades ago. That makes it difficult to bash drug companies and the FDA. While the findings are much more important (because patient deaths are involved) than the Avandia results there’s low potential for hype. Clearly the popular media hysteria is more about finger pointing than patient care.

In order to improve macro vascular health in type 2 diabetes you can’t just lower blood glucose. You have to do something else. What is that “something else”? It is, at the risk of oversimplification, treatment of the metabolic syndrome, the principal dyslipidemia of type 2 diabetes. Such treatment is generally multimodal, based on hygienic measures (diet and exercise) often in conjunction with lipid regulating drugs. Glycemic control doesn’t accomplish this goal, and glucose controlling drugs are ineffective unless they exert additional direct effects on one or more adverse components of the metabolic syndrome. For a glucose controlling drug this is not the norm. When it occurs it’s a bonus. Thus far only two drugs in the vast array of diabetes medications appear promising in this regard: pioglitazone and metformin. Realistically the best we can expect for most diabetes drugs is a neutral effect on macro vascular health.

Background: Here’s the FDA warning for a popular sulfonylurea (Amaryl).

1 comment:

Anonymous said...

It would be very helpful for those of us in real world practice,to have some respected and disinteresested statistian publish the NNT and NNH for these drugs. I think these numbers should be required in any published study.

I'm tired of these relative incidence numbers which often turn out to represent clinicaly insignificant differences.It might also be helpful to recalculate the results of the meta analysis including the studies originally dropped. It appears that including them would have resulted in a totally insignificant difference in cardiac events.