I know that pyramid of study type strength is a standard thing, but I have an issue with putting meta-analyses at the apex. In order to accept the conclusions there, you have to look at every study they are including, understand how each study is rated for 'quality' (used to weight that study against the others included).

You can take a meta analysis as a good overview of work up to the date of writing - but without going into the included studies, it's an overview, no stronger than a similarly powered RCT. Best case is that it puts together multiple otherwise-perfect-but underpowered RCTs.

I also have issues with 'animal trials', but I'll let that go - there aren't going to be many people making decisions about their health based on mouse trials.

Off-topic: I'd suggest Peter Attia's interview with Francisco Gonzalez-Lima as a very good argument for late-onset Alzheimer's being a vascular and metabolic issue. I find the relative lack of cytochrome c oxidase in late onset Alzheimer's patients as compared to normal brains compelling, particularly as attempts to reduce amyloid and tau have been effective - at reducing amyloid and tau - but not reversing the effects of Alzheimer's.

Bringing this up as we've had a couple of interactions about this etiology. This makes four potential etiologies (amyloid/tau, inflammation, T3D, metabolic crisis). For late-onset, I'd expect it to be one of the last three - and there's not much reason they can't all be part of the same answer. A brilliant person might see the connection, but I'm not that person.