Polygenic Risk Scoring for Multiple Sclerosis: Ancestry-Dependent Performance and the Need for Equitable Genomic Prediction
Multiple sclerosis is a chronic autoimmune disease of the central nervous system characterized by inflammatory demyelination and neurological disability. The article examines whether polygenic risk scoring can meaningfully stratify multiple sclerosis risk across genetically diverse populations. This question is important because multiple sclerosis has a substantial genetic component, yet most genome-wide association studies used to construct polygenic risk scores have been conducted in individuals of European ancestry, raising concerns about transferability to other populations. The study is a medRxiv preprint and explicitly notes that it has not been certified by peer review and should not guide clinical practice.
Study Objective and Dataset
The authors aimed to determine whether a multiple sclerosis polygenic risk score could effectively classify individuals of non-European ancestries into different risk strata. They used data from the All of Us Research Program, a large United States cohort designed to improve biomedical research diversity. The analysis included participants with both whole-genome sequencing data and electronic health record data collected between 2018 and 2023. This design allowed the investigators to combine genetic ancestry assignment, polygenic risk estimation, and clinical outcome ascertainment within a single large-scale research infrastructure.
Construction of the Polygenic Risk Score
The exposure of interest was a polygenic risk score based on 282 independent genetic variants associated with increased multiple sclerosis risk. For each participant, the score was calculated by summing the number of risk alleles across variants, weighted by each variant’s reported effect on disease susceptibility. The variants were selected according to standard criteria: minor allele frequency above 1%, biallelic structure, independence from each other, and genome-wide significant association with multiple sclerosis. The resulting score was normalized within each ancestry group and divided into quintiles representing very low, low, intermediate, high, and very high polygenic risk.
Cohort Composition and Case Distribution
The final analytic cohort contained 173,153 participants, with a mean age of 51.6 years and 60% female representation. The largest ancestry groups were European, African, and Latino/admixed American, and the authors randomly sampled European and African participants to match the Latino/admixed American group size of 32,428 individuals. Within these matched groups, multiple sclerosis cases represented 1.0% of the European ancestry sample, 0.56% of the African ancestry sample, and 0.46% of the Latino/admixed American sample. The table on page 11 further shows that multiple sclerosis cases were disproportionately female and more frequently of European ancestry compared with non-cases.
Main Findings Across Populations
The polygenic risk score performed as expected in participants of European ancestry: individuals in the highest quintile had substantially higher odds of multiple sclerosis than those in the lowest quintile. After adjustment for age, sex, and genetic principal components, the highest European PRS quintile was associated with an odds ratio of 2.41. The score also stratified risk in Latino/admixed American participants, where the highest quintile had an odds ratio of 2.56. In contrast, the same score did not significantly stratify risk among participants of African ancestry, with a weaker and statistically non-significant trend.
Interpretation of the Ancestry-Specific Differences
The article argues that reduced performance in the African ancestry group likely reflects the limited representation of African populations in existing genome-wide association studies. A polygenic risk score developed primarily from European datasets may fail to capture ancestry-specific linkage disequilibrium patterns, variant frequencies, causal alleles, and gene-environment interactions. The authors also emphasize that African populations possess greater genetic diversity, which can make direct transfer of European-derived risk models less accurate. These observations reinforce a central issue in statistical genetics: predictive models trained in one ancestry group may not generalize equitably across all populations.
Implications for Precision Medicine
This study provides a clear example of both the potential and the limitations of polygenic risk scoring in complex autoimmune disease. The results support the use of multiple sclerosis polygenic risk scores for risk stratification in European and Latino/admixed American populations, while cautioning against uncritical application in African ancestry populations. The authors conclude that more inclusive genome-wide association studies and ancestry-aware polygenic risk methodologies are necessary to ensure that genetic prediction benefits all groups rather than amplifying existing biomedical inequities. In practical terms, the findings argue for a future in which multiple sclerosis risk prediction integrates genetic, environmental, lifestyle, and clinical information rather than relying on polygenic risk alone.
Disclaimer: This blog post is based on the provided research article and is intended for informational purposes only. It is not intended to provide medical advice. Please consult with a healthcare professional for any health concerns.
References:
Rivier, C. A., Payabvash, S., Zhao, H., Hafler, D. A., Falcone, G. J., & Longbrake, E. E. (2024). Differential Results of Polygenic Risk Scoring for Multiple Sclerosis in European and African American Populations. medRxiv, 2024-06.
