What will happen, or is happening, with a clinical trial is that there will probably be some people in the clinical trial who don't actually have the disease, and so they are unlikely to respond to the drug.

But, if 80% of the people in the clinical trial DO have it, and maybe half of them respond, that would still be a 40% efficacy rate, which would be enough to get it approved.

All figures pulled out of thin air, of course.

We really need a blood test for Alzheimer's but I haven't heard about anyone getting to the point of actually filing for approval yet.