Two papers in this week's NEJM prompt some thoughts.
The first is an editorial about the fate of a new drug for Alzheimer's disease. The story starts with the well known fact that people with Alzheimer's have extensive amyloid plaques in their brain. So perhaps if we get rid of the plaques we get rid of the Alzheimer's. That's a pretty weak molecular basis for thinking the treatment will work, but it's a huge problem so anything is worth a try. Eli Lilly had a go at it with their monoclonal antibody solenazumab. Overall the trial was a flop but a post hoc analysis in a small subgroup suggested benefit. The FDA were not convinced and asked for a full trial in that specific sub group; several (likely hundreds) of millions of dollars later, the trial showed no benefit and Eli Lilly pulled out. This cost will show up in the price of everything else they sell.
The editorialist reckons that by insisting on the full trial the FDA prevented widespread adoption of the drug on weak evidence and saved Medicare billions of dollars. Credit goes to the FDA partly because there is no tradition of economic analysis in the USA so once licensed a drug has no further barriers to clear; in the UK the drug would certainly have failed at NICE.
The second paper is an absolute megatrial with 27 567 patients randomised. That sort of trial is way beyond the reach of any public sector funder, so credit to Amgen (though again the cost of the trial must end up in higher costs somewhere). This time the effort was needed to show that a proxy marker (lower cholesterol) translated into a true benefit (fewer cardiovascular events). I guess that shows why we need proxy markers - launching a trial to show the biological effect is often simply not feasible because of the sample size required. This publication drew a blizzard of critique in the Journal's online comments pointing out that although cardiovascular events are down, all cause mortality isn't. Perhaps the drug has some subtle and unremarked adverse effect?
And of course why do we need a drug at all for high cholesterol? FIX YOUR DIET people!
It's unfair to expect to detect a drop in all cause mortality for something rare (and for this purpose even breast cancer counts as rare) but cardiovascular disease is a large cause of mortality. Petr Skrabanek used to enjoy irritating people by claiming that all sorts of things (such as breast cancer screening) were useless because they didn't decrease all cause mortality.