Will the FDA finally get it about surrogate endpoints?

As you have likely heard, there is bad cholesterol (Low Density Lipoprotein, LDL) and "good" cholesterol (High Density Lipoprotein, HDL). Elevated LDL in the blood is associated with increased risk of heart disease and strokes. No doubt about it. Conversely, in observational studies, elevated HDL is associated with lower risk. Truth.

So, we know that statins, which lower LDL, are associated with reduced risk of heart attacks and stroke, probably cost-effectively in people who already have heart disease, maybe or maybe not worth it for people who just have risk factors. But there's definitely an effect.

So, obviously, a pill that raises HDL will also help. For years, doctors have prescribed time-release niacin (nicotinic acid) for this purpose. It has some unpleasant side effects, but it's worth it to raise your HDL and save your life, right?

Sadly, no. A randomized controlled trial has been stopped early because adding Niacin to a statin regimen actually seems to result in increased risk of stroke, even though, yup, it raises HDL.

We've seen this again and again -- drugs are approved because they have an effect on some biological indicator which is hypothesized to yield a health benefit. But then it turns out it doesn't after all. Meanwhile the drug companies have raked in billions and patients have experienced unpleasant or even deadly side effects, for no benefit.

It's easy to think of reasons why we can be misled in this way. For example, a lot of vigorous aerobic exercise tends to raise HDL. Maybe it was associated with reduced risk in observational studies because it is a marker of fitness. But raising it with a pill doesn't make you physically fit! It just makes you have elevated HDL, which appears to be beside the point.

The FDA must change its policy and require actual evidence of a health benefit before approving new drugs. And maybe there are a lot more old ones that should come off the market. Of course everybody is lining up to defend Niaspan, even though it doesn't work, because a) there's big bucks involved and b) they don't want to admit they were wrong. Grow up folks.