Polygenic scores derived from phenotype-specific genome-wide association studies of amyloid-β accumulation and age at onset do not significantly improve prediction of these traits beyond scores derived from Alzheimer's disease risk and resilience GWASs, likely due to limited sample size in discovery analyses.

Genetic variants that influence amyloid buildup and when symptoms start are too weak and scattered to be picked up accurately unless you have a huge number of people in the study. The broader Alzheimer’s risk scores already capture the strongest genetic signals that affect both amyloid and other disease processes, so adding more specific amyloid-focused genetic data doesn’t help unless you have way more people to find the subtle signals.