Tuesday, July 22, 2008

Vytorin and SEAS: Here's An Alternative Point of View

There's been something of a feeding frenzy over reports of ! CANCER ! in the latest Vytorin trial. The study was acronymed 'SEAS' for the ‘Simvastatin and Ezetimibe in Aortic Stenosis Study,’ which randomly assigned over 1800 persons with ‘aortic stenosis’ to a placebo or to Vytorin. Different than coronary artery disease (in which the arteries feeding the heart muscle get blocked), aortic stenosis is scarring and thickening of a major heart valve. It usually evolves over decades and can lead to chest pain, shortness of breath and passing out. If the progression can be slowed with drugs, persons with mild stenosis may avoid severe stenosis and die of natural causes before the only known cure becomes necessary: major heart surgery.

The trial showed no difference in aortic valve outcomes. However, what the trial did show is an increase in cancer rates among subjects assigned to Vytorin:

In the subsidiary safety analyses, a total of 158 patients were recorded with a serious adverse event attributed to cancer. More of these events were observed among patients assigned the combination of simvastatin and ezetimibe than among those assigned placebo (93 (9.9%) versus 65 (7.0%); unadjusted p=0.03), and there were also slightly more cancer deaths (39 (4.1%) versus 23 (2.5%); unadjusted p=0.05). These apparent differences were not related to any particular type of cancer and did not become significantly larger with more prolonged treatment.

Shocking you say? While the mainstream print media’s alarmist framing (see here and here) can be forgiven (and let’s face it, they are also being distracted by the scourge of Mexican hot pepper Salmonella poisoning), some bloggers are really piling on: ‘this combination pill, still taken by millions of people, raises the risk of cancer,’ and it ‘unexpectedly suggests an increase in cancer…. disturbing finding…’ and cancer fears are putting stocks ‘in a tailspin.’

The Disease Management Care Blog advises folks to take a deep breath, calm down and look at the facts.

First off, fears that there are links between cholesterol lowering as well as cholesterol lowering drugs and cancer are not new. Reports as far back as the 1990s raised that possibility. Other studies have cast doubt on the association while other studies suggest lowering cholesterol may protect against cancer. When all the science is considered, the SEAS news is not surprising, not shocking and certainly not conclusive.

Secondly, there is the difference between causality and association. Just because two things happen at the same time doesn’t mean one directly causes the other. Between and surrounding Vytorin and cancer may be other events that drive the association. What’s more, even if there is a causal relationship, we don’t know the direction: it’s possible that developing cancer can cause a decrease in blood cholesterol levels, long before the cancer becomes apparent. The SEAS study may not have evenly randomized patients at risk for cancer.

Thirdly, if multiple studies are done, statistical significance – when none exists – is more likely to occur. The all important ‘p value’ of .05 means there is a 5% chance that the observation could have happened as result of random error. If multiple studies are done, the .05 for each study is additive. In other words, if there is a one in twenty chance that something will happen, by the time you get past ten chances (or you conduct ten studies), there is a reasonable expectation that that error (a spurious association) will happen. It was possible this was SEAS' bad luck.

Unfortunately, SEAS was unable to demonstrate that Vytorin can slow the progression of aortic stenosis. Based on prior studies, we still don't know about the cancer risk from cholesterol lowering medicines, but in general they suggest there isn't an association. The DMCB thinks the statistical/methodological issues above are more likely explanation for what happened.

No comments: