Alzheimer’s Risk Prediction: How Multi-Ancestry PRS Improves Accuracy

The Blind Spot in Alzheimer’s Genetic Risk Models

Accounting for roughly 6.9 million individuals in the United States, late-onset Alzheimer’s disease is a degenerative condition defined by memory deficits and declining cognitive abilities. The development of AD spans a lengthy timeframe, typically starting with imperceptible shifts in cognition before advancing to full-blown dementia. While the rate of functional loss differs from person to person, accumulating data indicates that inherited traits are pivotal in directing both when cognitive worsening begins and how it proceeds. While the APOE ε4 allele remains the strongest known genetic risk factor, genome-wide association studies (GWAS) have identified numerous additional common and rare variants contributing to susceptibility primarily in individuals of European ancestry (EA) and to a lesser extent in other populations. An individual’s inherited vulnerability to a specific disease can be quantified numerically through a polygenic risk score (PRS). However, previously derived PRS for AD perform inconsistently across diverse ancestries.

Researchers have tackled this disparity head-on. A newly published scientific paper outlines the creation and testing of an advanced multi-ancestry polygenic risk score for AD that vastly outperforms earlier models, particularly when applied to genetically mixed populations. Widening the scope of included demographic groups is vital for building sturdier and more universally applicable PRS tools, which in turn enhance disease likelihood forecasting across an ethnically diverse patient base.

“A critical challenge in the application of PRS lies in the underrepresentation of diverse genetic ancestries in AD GWAS datasets, which include predominantly white individuals of European ancestry. The availability of GWAS data from diverse populations, including African American, Hispanic and East Asian, provides an opportunity to enhance the transferability and accuracy of PRS across multiple ancestries,”

Methodology and Multi-Ancestry Validation Numbers

For the creation of a fairer predictive instrument, the investigators utilized aggregated genetic marker data spanning the entire genome, drawn from a diverse cohort encompassing upwards of 63,000 Alzheimer’s cases and 484,000 controls matched for age, as detailed in the publication. Evaluation of the PRS occurred within a separate, ethnically varied cohort consisting of 10,612 AD cases alongside 16,625 older control participants, followed by further verification in a mixed-ancestry population featuring 1,500 AD patients and 75,500 elderly controls.

In contrast to older generation tools built solely on European ancestry cohorts, the newly formulated multi-ancestry risk score demonstrated superior capability in forecasting clinical Alzheimer’s diagnoses across every evaluated demographic, with especially pronounced accuracy among East Asian, Hispanic (spanning both continental U.S. and Caribbean backgrounds), and African American individuals.

Connecting Genetic Risk to Biological Markers and Cognitive Decline

Beyond evaluating clinical diagnoses, researchers analyzed how the Alzheimer’s polygenic risk score correlates with a broad spectrum of clinical, neuroimaging, biochemical, and neuropathological features recorded among subjects enrolled in the Alzheimer’s Disease Sequencing Project, the Framingham Heart Study, the Alzheimer’s Disease Neuroimaging Initiative, and the Korean Brain Aging Study for the Early Diagnosis and Prediction of AD.

Measurable biological consequences emerged clearly from the data. Higher risk scores correlated significantly with lower scores in language, executive function, and memory tests; structural brain scans via MRI revealed diminished volume in the hippocampus, which is the brain structure targeted earliest by AD; and cerebrospinal fluid analyses detected abnormal concentrations of phosphorylated Tau (pTau)—showing an elevated pTau variance in women—alongside hallmark amyloid-beta proteins. Furthermore, tracking participants over time demonstrated that those carrying exceptionally high polygenic risk scores experienced the most rapid cognitive deterioration, most notably in the timeframe leading up to the onset of Alzheimer’s.

“the associations we observed with early biological and cognitive changes and potential sex-specific differences support the value of this ancestry-aware PRS for long-range risk prediction, selecting subjects for clinical trials, and personalized intervention and prevention strategies,”