New Debate Over Vytorin and Cancer
Data Are Reassuring, but Editor of Leading Medical Journal Says Drug Can't Be Cleared
WebMD News Archive
Assessing Cancer Risks continued...
But that, Peto says, is preposterous. When his team looked at the data from the IMPROVE-IT and SHARP trial, they found no increase in cancer over time -- which one would expect from a potent cancer-causing agent. Moreover, they found no increase in any specific kind of cancer.
"If this stuff was 100 times as dangerous as cigarette smoke, as the SEAS finding suggests, you would expect to see an increase over time in the hazard. And we don't," Peto tells WebMD. "And if you had no effect, you would expect the excess cases to be distributed evenly over time, and they are. So we do not get any increase over time in risk, either in cancer incidence or in death from cancer."
Yet Drazen and Steven Nissen, MD, chairman of the Cleveland Clinic's department of cardiovascular medicine, point to a disturbing statistical anomaly. If all the cancer deaths in the SEAS, IMPROVE-IT, and SHARP trials are added together, there were 134 cancer deaths in patients taking Vytorin and 92 deaths in patients taking inactive placebo or the cholesterol-lowering drug Zocor.
That suggests a 45% increase in cancer deaths with Vytorin -- with a statistical probability of 0.007, meaning the odds of it being a chance finding are seven in 1,000.
Peto, a professor of medical statistics, says this isn't a fair analysis. First of all, with all the thousands of clinical trials being done all over the world, every year there will be false findings that have only a one in 1,000 chance of happening.
More specifically, he says, the reason for looking at the IMPROVE-IT and SHARP results was to see whether the unexpected SEAS finding was true. Adding the SEAS data to the other data pollutes the analysis.
"Suppose I have a freak result. To test whether it is true, I will check out other data independently," Peto says. "What you cannot do is add up the independent test with the test that got you alarmed in the first place. If you include the test that got you freaky results in the first place, you will have a freaky result in the end."