Basic Science and Pathogenesis.

Area of Science:

• Genetics
• Neuroscience
• Computational Biology

Background:

• Late-onset Alzheimer's Disease (LOAD) poses a significant global health challenge, with cases projected to triple by 2050.
• Genetic factors contribute over 60% to LOAD variation, yet traditional risk scores using common variants have limited predictive power across diverse populations.
• The predictive potential of rare damaging variants in a polygenic framework for LOAD has not been previously assessed.

Purpose of the Study:

• To develop the first rare variant polygenic risk score for LOAD (rvPRS-AD).
• To assess the trans-ancestral predictive power of rvPRS-AD.
• To evaluate rvPRS-AD's independence from common variant predictors and APOEε4 status.

Main Methods:

• Developed rvPRS-AD using a whole genome-adapted RV-EXCALIBER method for rare-variant gene burden testing.
• Utilized European LOAD cases and gnomAD control data to identify risk-conferring genes.
• Constructed rvPRS-AD in Hispanic participants by weighting the burden of rare damaging variants per gene.

Main Results:

• Identified ABCA7 as the most strongly associated gene with LOAD (OR = 1.40, p=1×10⁻⁴).
• rvPRS-AD showed significant trans-ancestral predictive power, increasing LOAD odds by 38% in Hispanics (1-SD increase).
• The score remained predictive independent of APOEε4 status and identified individuals at extreme risk (≥2-fold increased odds).

Conclusions:

• rvPRS-AD is the first genetic instrument leveraging rare variants in a polygenic framework for LOAD risk prediction.
• This approach offers improved prediction across ancestrally diverse populations.
• rvPRS-AD provides a novel tool for LOAD risk stratification independent of common variants.