Development of a Multiancestry Alzheimer’s Risk Score
Late-onset Alzheimer’s disease affects approximately 6.9 million people in the United States today. While memory loss and cognitive decline manifest over many years, researchers are increasingly aware that genetic factors influence who develops the condition and how quickly they decline. The APOE ε4 allele has long been recognized as the most significant genetic risk factor. However, genome-wide association studies have historically leaned heavily on data from European populations. This bias limits the accuracy of risk prediction for individuals of other ancestral backgrounds.
A new study published in Nature Genetics addresses this disparity by developing a polygenic risk score that performs across diverse populations. By aggregating data from African American, Caribbean Hispanic, East Asian, and European cohorts, researchers created a model that provides a more accurate measure of inherited susceptibility. This approach is critical because predictive models developed solely on European data often fail when applied to non-European groups.
Methodology for Genetic Prediction
The research team utilized summary statistics from multiple population ancestries to build their model. They integrated these data through weighted summation, allowing them to account for varied genetic architectures. Crucially, the researchers excluded variants within the APOE region to ensure the score reflects polygenic risk independent of that specific, dominant genetic marker. This separation allows clinicians to identify high-risk individuals even when they do not carry the high-risk APOE ε4 allele.
The resulting score was validated in several independent cohorts, including the All of Us Research Program and the Korean Brain Aging Study. In these datasets, the risk score consistently predicted clinical Alzheimer’s diagnosis and showed meaningful associations with biomarkers. By using this ancestry-aware framework, the team bypassed some of the biases inherent in previous risk assessments that overlooked the genetic nuances of non-European individuals.
Clinical Implications and Future Directions
Beyond predicting clinical diagnosis, this polygenic risk score tracks with physical markers of the disease. High scores correlated with reduced cerebrospinal fluid amyloid-β 42 levels and elevated tau protein concentrations. These findings mirror the biological progression of the disease from the preclinical phase to clinical onset. Researchers also found that the score aligns with structural brain changes, specifically hippocampal atrophy.
Cognitive assessments reveal that individuals with the highest risk scores show lower memory performance as early as age 60. This gap widens as patients age, suggesting the score captures a fundamental difference in cognitive reserve. While the score does not necessarily predict how fast someone will decline after an official diagnosis, it offers a window into who might face early-life cognitive strain.
This study underscores the necessity of broadening genetic datasets. As researchers refine these models, they move closer to personalized risk stratification. Future work will likely integrate these genetic scores with longitudinal neuroimaging and blood-based biomarkers to improve early detection efforts. The broader goal remains creating clinical tools that function fairly for every population, moving beyond the current limitations of genomics.