Friday, November 23, 2007

Here We Go Again: Everything That's Wrong with Industry Sponsored Trials, Chapter 77

Alex Berenson wrote recently in The New York Times about the delay in reporting the results of the "Enhance" trial of the cholesterol-lowering drug, ezetimibe. His article contains a quick review course of most everything that is wrong with company sponsored clinical trials of drugs, in case you came in late.

Ezetimibe is currently sold by Merck and Schering-Plough, alone (as Zetia) and in combination with a statin drug (as Vytorin). The two drugs together sell about $4B annually. Ezetimibe's claim to fame is that it does not work like statins, so patients unable to tolerate statins can usually take it safely. It also is better at selectively lowering "bad" (LDL) cholesterol than statins alone. Proof that it lowered LDL was enough for the FDA to approve it. Yet there have been no trials demonstrating that ezetimibe, either alone or in combination, actually prevents serious disease or saves lives. (Lesson One) Three such trials are underway with results expected perhaps in 2010.

In the meantime, a lot has been riding on the Enhance trial, which is designed to measure the growth of atherosclerotic plaque ("hardening of the arteries") in patients taking only a statin vs. patients taking both the statin and ezetimibe. The thickness of plaque is still a "surrogate endpoint" rather than an actual disease outcome; but many believe that it's a reasonable predictor of whether the drug will actually work. Berenson reports that there is therefore considerable concern among cardiologists that the results of Enhance, initially expected to be presented in March 2007, will not be ready till March 2008.

The official word from the lead investigator, Dr. John Kastelein of Amsterdam, is that no one knows what the results are yet because no data have been released, and the data are still being analyzed. The delays were routine problems that arise when you have as huge a body of data as was generated by this trial. Kastelein strongly denied that there was anything fishy going on.

But besides this suggestion of normal delay, there were two worrisome features of Enhance reported by Berenson. First, it turns out that the companies have sole control of the data, and until they analyze it all and elect to release it, Kastelein himself has no way to access it. (Lesson Two.) Second, it turns out that the companies have decided to change the study's primary endpoint. Initially they had planned to focus attention on a combination of plaque thickness measuements at three different points in the carotid (neck) artery. They now intend to report as the primary endpoint the thickness at onely one point. (Lesson Three) This, they claim, will allow faster reporting of the results and will be just as reliable. They say that they will still report the three separate places in the carotid, but as a secondary and not as a primary endpoint.

Kastelein hinted in the interviews that he was personally not very happy with the change in endpoints. To determine that this was the right thing to do, the companies brought in an outside expert advisory panel. The panel concluded that the endpoint should be changed. The companies refuse to say who was on the panel. (Lesson Four)

Trial expert and industry critic Dr. Bruce Psaty is quoted by Berenson as highly critical of the endpoint shift. As a rule such a shift is very suspicious because it raises fears that you are deciding what counts as a successful trial result only after you know what some of the data show. It is analogous to hitting a bull's-eye with a bow and arrow by first shooting the arrow, then drawing the bull's-eye around the arrow wherever it happens to land.

So, in summary, here is the set of rules for drug companies as to how to make the scientific community--at least that part of it that has not been bought off with your consulting fees, speakers' fees, research funding, etc.-- most suspicious that you are pulling a fast one:

  1. Seek to get your drug on the market before we know whether it actually makes us live any longer or prevents real disease.
  2. Keep all the data under your strict control; don't even let the (so-called) lead investigator see it.
  3. Change the primary study endpoint in the middle of the trial.
  4. Keep all the processes as secret as possible at each step of the way.
Now, there may, in fact, be perfectly innocent explanations for all of this. (After all, Saddam Hussein turned out to have no WMDs.) Given the past track record of industry behavior in like circumstances, how many people are willing to bet in this case that the explanations are entirely innocent?

Berenson A. After a trial, silence. New York Times, Nov. 21, 2007: C1.

NOTE ADDED 11/26/07: I'm grateful for an e-mail from Matt Herper at Forbes who informs me that they got to the ENHANCE story first-- see:

1 comment:

Unknown said...

Can anyone tell us who the 18 or 19 sites are in the ENHANCE study? It doesn't seem to be available at, but this seems to be such a hot topic for the annual meetings this year and last that it can't be much of a secret.

I'm curious what kind of publication terms they agreed to with Merck/S-P in their Clinical Trial Agreements for ENHANCE. (Or to the extent that they cannot share that specific information, I'd like to know what they generally agree to for publication terms.)

We are now 2 years post-study conclusion and many academic institutions would have publication policies that would allow their investigators to at least publish or present their own findings with such a long delay on the publication of the entire study's multi-site data.