They could also develop it at the exact same rate and mask it effectively. Had a very sharp family member who was completely lucid through all of dementia, it was like watching a tape rewind. Got to meet each decade's version of them over a couple of years. Towards the end they were completely normal

Plausible, but then you'd expect to find the same effect in other people who passed similarly demanding tests once. For example in people who studied something geometry-heavy.

Also, The Knowledge Test is, IIRC, passed only once and then you do not have to retake it for the rest of your career.

If a test passed in your 20s or 30s predicted accurately a much lower incidence of Alzheimer's in your 60s, that would be a remarkably strong early signal. Not impossible, but worth extra scrutiny. For the vast majority of old age conditions, we cannot predict them so long in advance.