There has been considerable "buzz" about the latest study purporting to 
show that drugs can "prevent" type II diabetes.  That study just came out 
in the Lancet: 
http://www.thelancet.com/journals/lancet/article/PIIS0140673606694208/fulltext
(The DREAM Trial Investigators. Effect of rosiglitazone on the frequency of 
diabetes in patients with impaired glucose tolerance or impaired fasting 
glucose: a randomised controlled trial. Lancet 2006.)

For an example of the buzz, see this story in the Boston Globe, which 
opened  with "millions of adults at high risk of diabetes would ward off 
the disease by taking a drug commonly used to treat 
it...." 
http://www.boston.com/news/nation/articles/2006/09/16/drug_can_help_prevent_diabetes_study_says/

Also, a New York Times article quoted the study's first author that it 
showed "a very large effect, the kind of thing you don't expect in a 
clinical trial," and so he "would certainly consider using the drug in some 
patients."  On the other hand, the first author "took issue with the notion 
that the trial was a drug company study,"  although one of its funders was 
GlaxoSmithKline, which makes rosiglitazone (Avandia).  And the Times noted 
that GSK "is coordinating a publicity program around the study release."
http://www.nytimes.com/2006/09/17/world/17diabetes.html

But reading the study reveals two obvious problems:

1- It is not clear whether the drug really "prevents" diabetes, or just 
lowered the blood sugar of patients who became diabetic.  The criterion for 
preventing diabetes was having a normal fasting glucose or a normal fasting 
glucose tolerance test while on the drug.

2- It is not clear whether the drug provided any clinical benefit, much 
less one that outweighed its harms.  The article did not report 
measurements of many clinical outcomes, making it difficult to assess its 
clinical benefits beyond effects on laboratory test results.  Disturbingly, 
some of the outcomes it did measure may have been worse in the 
rosiglitazone group.  The rate of all cardiovascular events trended higher 
in that group (2.9% vs. 2.1% in placebo).  The rate of congestive heart 
failure was higher in that group, albeit still numerically small (0.5% vs 
0.1%).

Apparently, an editorial appeared with the article, which appears not to be 
available without a subscription.  According to the Times, it advocated 
considering intensive life-style changes rather than drugs to prevent 
diabetes, but may not have addressed the points above.

I wondered if anyone on this list had read this article, and if so, what 
they thought about the two problems noted above, and why they escaped much 
notice?

I should note that an article from Canada.com quoted Dr Jim Wright, of the 
British Columbia Therapeutics Initiative, about problem 1 above, saying the 
study "is giving the impression that it [rosiglitazone] prevents diabetes, 
when really what it is doing is delaying the diagnosis for one year."
http://www.canada.com/topics/bodyandhealth/story.html?id=e4b20bab-8e4a-4f22-b96d-0bb04c44bd04&k=40373

I should also note that several members of the writing committee had 
financial ties to GlaxoSmithKline, which could conceivably have affected, 
possibly unconsciously, how they wrote the article.

Finally, I should note that failure to distinguish "prevention" from 
suppressing laboratory manifestations of the disease also seemed to be a 
problem for an earlier study about using metformin to reduce "the incidence 
of type 2 diabetes."
(Diabetes Prevention Research Group. Reduction in the incidence of type 2 
diabetes with lifestyle intervention or metformin.  N Engl J Med 2002; 346: 
393-403).

A letter to the journal later pointed out the problem.
(Bo-Abbas YY. Diabetes prevention. N Engl J Med 2002; 346: 1829.)

The author of the study replied that "we have performed a drug-washout 
study among the participants who did not have diabetes at the end of the 
study, and the results will be reported separately."
(Knowler WC. Diabetes prevention. N Engl J Med 2002; 346: 1830.)

However, a PubMed and Web of Science search did not reveal such a publication.

Therefor, I wonder if any other members of the list have general comments 
on the use or lack thereof of washout periods in prevention trials to 
measure the effects of drugs which are known to affect the laboratory tests 
used to make the diagnosis of the disease the trials seek to prevent?


Roy M. Poses MD
Clinical Associate Professor
Brown University School of Medicine
<[log in to unmask]>