Predicting amyloid-beta pathology in the general population

Area of Science:

• Neurology
• Biomarker Discovery
• Gerontology

Background:

• Accurate prediction models for amyloid beta (Aβ) positivity in the aging population are needed for early Alzheimer's disease (AD) risk identification.
• Such models could serve as cost-effective tools for identifying individuals at risk.

Purpose of the Study:

• To develop and validate models for predicting Aβ positivity using easily obtainable predictors.
• To assess the generalizability of these models in a population-based sample.

Main Methods:

• Developed Aβ prediction models in the Anti-Amyloid Treatment in Asymptomatic Alzheimer's (A4) Study (n=4,119).
• Included predictors such as demographics, cognition, daily functioning, health, and lifestyle factors.
• Validated models in the population-based Rotterdam Study (n=500).

Main Results:

• The best model in the A4 Study (AUC=0.73) included age, APOE ε4, family history, cognition, walking, and sleep.
• This model achieved higher accuracy in the Rotterdam Study (AUC=0.85).
• The improvement over a model with only age and APOE ε4 was marginal.

Conclusions:

• Developed Aβ prediction models using inexpensive, non-invasive measures.
• Successfully applied models to a general population sample representative of older non-demented adults.
• These models offer a viable approach for risk stratification in the general aging population.