Cutting to chase for those of you who want a quick answer: we don’t know, there aren’t enough people to study yet. There are hints, though.
Ezetimibe is commonly used in combination with a statin as Vytorin to reduce heart disease in patients. It doesn’t work all that well. I should say, it decreases cholesterol fine but whether that leads to less overall heart disease isn’t conclusive (latest study released today says maybe a little). Of course that study found an increase in cancer in those taking the drug. However, another analysis released today found that cancer deaths were increased but cancer cases were decreased (both were on the verge of stastical significance). Based on that reverse finding alone, it seems that not much is happening here, or if there is, not much. Animal studies show nothing that would indicate a problem.
This seems like a bit of ho hum but think about what this means for drug approval for a second. This drug was approved in 2002. It’s 2008 and we don’t really know if it decreases death an disease. This is the problem with approving on surrogate markers – that’s where the FDA says “well, you don’t know if it reduces heart disease but you know it decreases cholesterol and those are known to be related so good enough”. Sadly, this is very common with many types of drugs. The problems with this approach are four-fold (at least):
One, there may be a diluted association. For instance, if there is a certain amount that arterial plaque has to be lowered to effect risk and the drug lowers cholesterol a little bit which in turn effects plaque, by the time you look at an overall effect, it may be tiny or insignificant. Even if it exists, it may be so small as to not outweigh the risks from taking the drug.
Two, the surrogate may not be a causal factor, it may be a good marker. Take markers for cancer such as PSA for prostate cancer. Generally as the PSA level goes up that means the cancer is getting worse. However, the marker isn’t causing the cancer. Therefore, a drug that lowers PSA isn’t necessarily curing the cancer, only lowering the marker. This is like putting a picture of a empty highway infront of a traffic camera and concluding that there is no traffic today.
Three, the more assocaitions you make the more likely you are wrong. This is similar to one but isn’t talking about the magnitude of the effect but the likely you are plain old wrong. If each association has a uncertainty of 5% or so and you are using a couple of links, the possibility that the drug has nothing to do with the disease increases dramatically.
Four, if you don’t do a large study to figure out if it works, then you find out less about the toxicity; cancer for one among many.
On the other hand, if you do a big ol’ study, it takes longer to get the drug to market. I say FDA shouldn’t use surrogates unless there either aren’t any treatments available at all, or the new drug show a drastic improvement above the existing treatment. Of course, if the later is true, the study wouldn’t take as long so it could probably be limited to the former exemption.
Starting to wonder under what ‘good enough’ method the drugs you take got approved? So am I…