Am Fam Physician. 2008;77(1):16-17
Author disclosure: Nothing to disclose.
Each month, three presenters will review an interesting journal article in a conversational manner. These articles will involve “hot topics” that affect family physicians or will “bust” commonly held medical myths. The presenters will give their opinions about the clinical value of the studies discussed. The opinions reflect the views of the presenters, not those of AFP or the AAFP.
This Month's Article
Kahn SE, Haffner SM, Heise MA, et al., for the ADOPT Study Group. Glycemic durability of rosiglitazone, metformin, or glyburide monotherapy [published correction appears in N Engl J Med. 2007;356(13):1387–1388]. N Engl J Med. 2006;355(23):2427–2443.
Which controls type 2 diabetes best at five years: rosiglitazone, metformin, or glyburide?
Mark: For the estimated 20 million Americans with type 2 diabetes, this is a very important question. This article is an excellent demonstration of why reading just the abstract could lead you to the wrong conclusion.
What does this article say?
Mark: This study randomized 4,360 patients to monotherapy with rosiglitazone (Avandia), metformin (Glucophage), or glyburide (Micronase). The end point was fasting blood glucose over 180 mg per dL. The authors concluded that, at five years, only 15 percent of patients taking rosiglitazone failed mono-therapy versus 21 percent of those taking metformin and 34 percent taking glyburide. Although the abstract trumpets rosiglitazone as “the winner,” this is exactly the wrong conclusion to draw from this study.
Should we believe this study?
Mark: How did the authors come to the incorrect conclusion? First, fasting blood glucose was the wrong end point to measure. This is a surrogate marker of disease that does not correlate with clinical outcomes. Using such surrogate markers as end points is known as “DOE” or “disease-oriented evidence.” What we care about are patient out-comes, such as adverse cardiovascular events; these are known as “POEMs” or “patient-oriented evidence that matters.” Fasting blood glucose is a surrogate marker that is, perhaps, just as dependent on the half-life of the drug and when the drug was administered as anything else.
What about the more accepted gold standard A1C values (itself a DOE)? There was a negligible difference of 0.13 percent favoring the rosiglitazone group over the metformin group. This miniscule difference is clinically meaningless; it takes an increase of at least 1 percent in A1C level to barely increase the relative risk of a cardiovascular event to 1.18 (95% CI, 1.1 to 1.26).1
And what did the study show about the clinical end points that we are really interested in, such as cardiovascular events? Patients on rosiglitazone had more CHF and a greater number of cardiovascular events than those on metformin or glyburide. These are the end points that matter.
Bob: Also, there was an average weight gain of 10.6 lb (4.8 kg) with rosiglitazone and 3.5 lb (1.6 kg) with glyburide, which is not something we need in patients with type 2 diabetes. Metformin resulted in a net weight loss of 6.4 lb (2.9 kg) over five years.
Mark: What disturbs me most is that the conclusion in the abstract is different from the conclusion in the body of the paper. What the authors conclude in the paper is that the “findings confirm the value of metformin as an initial treatment for type 2 diabetes.” This is entirely different than what they concluded in the abstract.
Andrea: This disconnect between what is reported in the body of a paper and what is promoted in the abstract is not a new problem. In a review of first-tier journals, the data in the abstract differed from the data in the paper up to 68 percent of the time.2 This tells you that if you just read the abstract, you may be misled.
Additionally, the common use of surrogate outcomes is problematic. For example, we don't care about a blood pressure measurement—we care about cardiovascular events. You can reduce the blood pressure and still not reduce all cardiovascular events.3,4 So anytime you see a published article, make sure the outcome is something we care about: the patient.
What should the family physician do?
Mark: I would begin medical therapy with metformin and use sulfonylureas and insulin as second-line treatment options. If you use glitazones, watch for adverse events, including increased fluid retention, CHF, and cardiovascular events.
Bob: CHF has been seen with glitazone use in the past. In 2002, the FDA recognized the increased rates of CHF associated with glitazones and required the makers of pioglitazone (Actos) and Avandia to strengthen the warnings concerning adverse cardiovascular events.
Also, there have been numerous headlines and news reports on a meta-analysis published in the New England Journal of Medicine that documented a small but increased risk of death from cardiovascular causes in patients taking rosiglitazone.5 It has been said that the risk of death is small; however, even if the risk is small, why would you prescribe this drug at all? When prescribing drugs, you want a benefit, but with rosiglitazone, you are basically saying, “I have this drug that will cost you a lot of money, won't help your heart, oh, and by the way, there is a small chance it may make you worse.” Any takers?
Andrea: I agree. Start with metformin. This study should also serve as a warning that abstracts may not reflect the true results of a study, even in journals like the New England Journal of Medicine. You should evaluate the study yourself before you change your clinical practice (or find a group of medical nerds like us to do it for you).
Glitazones are associated with fluid retention, increased CHF rates, and, possibly, increased rates of cardiovascular events compared with sulfonylureas and metformin.
The FDA and the Canadian government have issued warnings about CHF and cardiovascular events with rosiglitazone.
Metformin should be the first-line drug for managing type 2 diabetes. Insulin and sulfonylureas should be second line, and glitazones should be reserved for third line.
Metformin is the only drug for type 2 diabetes that does not cause weight gain, which is an important advantage.
The data and conclusion in the abstract of an article may not be the same as the data in the paper. If you read only the abstract, you may be misled.
A lot of studies use surrogate markers as outcomes (e.g., fasting or postprandial blood glucose, FEV1). These are DOEs. What we care about are POEMs (e.g., stroke rates, MI rates, quality of life).
Something can be statistically significant but clinically meaningless (for example, A1C difference of 0.08 percent).