Well, well, well. We finally have solid clinical data from a large trial of a gamma-secretase inhibitor for Alzheimer's disease. And it doesn't work.
Background, for those outside the field: a hallmark of Alzheimer's is the appearance of plaques in the brain. These are insoluble clumps of a protein called amyloid-beta, surrounded by dead and dying neurons. This amyloid is split off (for some reason) from the middle of a larger precursor protein (APP), and there are two enzymes that make the cuts to release it: gamma-secretase and beta-secretase. Shutting down one or both of those has long been seen as the most direct route to keeping amyloid from accumulating, and compounds that do this have been sought for at least twenty years now.
Now this is interesting. The road to secretase inhibition data in the clinic has been a long one, to say the very least - I worked in this field myself in the early 1990s, when we were still guessing at the enzymes involved. I would not like to even guess about the man-hours that have been spent along the way. Gamma-secretase has been a beast of a target. One nasty surprise along the way was the discovery that it also processes Notch, which is a developmental signaling pathway that you'd really rather avoid, but people have persevered, and pushed compounds into the clinic.
(As an aside, I'd have to say that beta-secretase has been even harder. There are an awful lot of structures out there billed as beta-secretase (BACE) inhibitors - and so they are, in your choice of labware. Despite huge efforts, it's been extremely hard to make an inhibitor with a reasonable chance of getting into the brain and being a drug. The only one I know of is CTS21166, from CoMentis, about which news has been rather scarce recently).
Myriad had taken a sort-of kind-of gamma-secretase inhibitor (Flurizan) into the clinic, and failed dismally. But Eli Lilly's semagacestat (LY450139) has long been the most advanced pure gamma-secretase inhibitor. It inhibits the enzyme directly, and had shown dose-dependent lowering of amyloid formation in humans, which is all you can ask. There were side effects noted from Notch, mostly in the GI tract, but the profile was still good enough to go on into Phase III two years ago. And now we have the results.
Nothing. Worse than nothing - they saw real declines in cognitive function compared to the placebo group. It's not getting as much play in the news this morning, but it also appears - insult to injury - that the drug was associated with a greater risk of skin cancer. Update: a commenter points out that this risk was known). Lilly has halted any development, and told all the study centers to stop dosing immediately. All the patients who received it will be monitored to see how they do over the next few months.
This is about as bad a result as could possibly be obtained, and I think it really has to torpedo the idea of gamma secretase as a drug target. Unless someone comes up with a very compelling and intricate argument to explain these results, I don't see how anyone can risk going down this particular road again. What must they be thinking today over at Bristol-Myers Squibb, where they've been developing a direct competitor, BMS708163? And how about the other drug candidates behind them?
And what does this say about the amyloid hypothesis itself? Nothing good. This is the crucial period for the whole idea, with several different approaches finally yielding late-stage clinical data. And it's starting to look as if the whole idea may have been just a terrible diversion.