Basic Science and Pathogenesis.

Area of Science:

• Neuropathology
• Neuroimaging
• Machine Learning

Background:

• Limited autopsy studies exist for Hispanic individuals with Alzheimer Disease (AD).
• Machine learning (ML) models offer scalable quantification of neuropathology.
• Diverse cohorts are crucial for understanding AD heterogeneity.

Purpose of the Study:

• To evaluate amyloid beta (Aβ) deposit densities in Hispanic decedents (HD) versus non-Hispanic White decedents (NHWD).
• To compare Aβ plaque densities across three brain regions using ML.
• To identify demographic and pathological correlations with Aβ deposition.

Main Methods:

• Compared Aβ deposit densities (CAA, cored, diffuse plaques) in frontal, temporal, and parietal cortices of HD (n=92) and NHWD (n=185).
• Utilized a validated ML pipeline on digitized whole slide images (WSI).
• Employed a 2:1 matching ratio for age and sex between groups, adjusting for study center.

Main Results:

• Hispanic decedents showed 1.71-fold higher cerebral amyloid angiopathy (CAA) density in temporal lobe grey matter.
• Hispanic decedents had lower cored plaque densities in frontal and parietal white matter compared to NHWD.
• Significant differences in Aβ densities were observed across study centers.

Conclusions:

• The study validates an ML pipeline for Aβ quantification.
• Reveals both similarities and differences in Aβ deposition between Hispanic and non-Hispanic White individuals.
• Emphasizes the need for diverse cohorts and generalizable findings for AD precision medicine.