Genetic Risk Prediction in Multiple Sclerosis: Why Ancestry Matters
Multiple sclerosis is a complex immune-mediated neurological disease in which genetic susceptibility arises from the combined influence of many common variants rather than from a single causative mutation. In the article “Polygenic risk score prediction of multiple sclerosis in individuals of South Asian ancestry,” Breedon and colleagues examine an important question in modern genomic medicine: can polygenic risk scores developed primarily from European ancestry genome-wide association studies accurately predict multiple sclerosis risk in people of South Asian ancestry? This question is scientifically and clinically significant because polygenic risk scores are increasingly discussed as tools for risk stratification, prevention trials, and early intervention, yet their validity depends heavily on the ancestry composition of the datasets used to construct them.
The Scientific Rationale for Polygenic Risk Scores
Polygenic risk scores aggregate the effects of many genetic variants across the genome to estimate an individual’s inherited susceptibility to a disease. In multiple sclerosis, the major histocompatibility complex, particularly human leukocyte antigen-related variation, has a strong influence on risk, while more than 200 additional non-MHC loci contribute smaller effects. The promise of polygenic risk scoring lies in its ability to identify individuals at elevated risk before disease onset, which could support preventive strategies such as Epstein–Barr virus vaccination studies, vitamin D supplementation trials, or monitoring during the prodromal phase of disease. However, this promise can only be realized equitably if these tools perform reliably across diverse populations.
Study Design and Population Resources
The authors used two major longitudinal genetic resources to test the transferability of European-derived multiple sclerosis polygenic risk scores. The primary analysis was conducted in the Genes & Health cohort, which includes British–Bangladeshi and British–Pakistani participants. After quality control, the study retained 40,532 individuals of South Asian ancestry, including 42 people with coded multiple sclerosis and 40,490 controls. For comparison, the authors used UK Biobank data from genetically European-ancestry participants, including 2,091 multiple sclerosis cases and 374,866 controls. This design allowed the investigators to contrast score performance between a South Asian-ancestry cohort and a much larger European-ancestry cohort while using broadly comparable analytical methods.
Construction and Evaluation of the Genetic Scores
The polygenic risk scores were generated using a clumping-and-thresholding framework implemented in PRSice-2, with variant effect sizes derived from the International Multiple Sclerosis Genetics Consortium genome-wide association study. The researchers created scores both including and excluding the MHC region, because this locus has an unusually large effect on multiple sclerosis susceptibility and can dominate prediction models. They also evaluated MHC-only scores to isolate the contribution of this region. Predictive performance was assessed using Nagelkerke’s pseudo-R² adjusted for disease prevalence, age, sex, and the first four genetic principal components, while discrimination was additionally examined using area under the curve statistics.
Principal Findings in the South Asian-Ancestry Cohort
The European-derived polygenic risk scores showed only modest predictive performance in the Genes & Health South Asian-ancestry cohort. The optimal score including the MHC region explained approximately 1.1% of multiple sclerosis liability, while the optimal score excluding the MHC explained approximately 1.5%. Notably, the MHC-only score was not significantly associated with disease status in this cohort, a finding that may reflect limited case numbers, ancestry-specific linkage disequilibrium patterns, or imperfect tagging of causal HLA alleles by variants identified in European studies. Although individuals in higher score quartiles tended to have greater disease risk, confidence intervals were wide, emphasizing the statistical uncertainty generated by the small number of multiple sclerosis cases.
Comparison with European-Ancestry Prediction Performance
When the same general approach was applied to European-ancestry UK Biobank participants, predictive performance was substantially stronger. In the full European-ancestry UK Biobank analysis, the score including the MHC explained approximately 4.4% of multiple sclerosis liability, while the non-MHC score explained approximately 2.3%. After subsampling UK Biobank to match the Genes & Health case-control structure, the MHC-containing score still performed better in European-ancestry participants than in the South Asian-ancestry cohort. These results support the broader observation that polygenic risk scores derived from European genome-wide association studies often lose accuracy when transferred to non-European populations, largely because allele frequencies and linkage disequilibrium structures differ across ancestral groups.
Implications for Genomic Equity and Future Research
The study provides a clear example of why genetic prediction tools must be developed and validated in ancestrally diverse populations before they are considered for clinical or public health use. Although the European-derived score retained some predictive signal in South Asian-ancestry individuals, its reduced accuracy raises concerns about inequitable implementation of genomic medicine. If polygenic risk scores are used to identify people for prevention trials, surveillance programs, or early intervention strategies, underperformance in underrepresented populations could worsen existing health disparities. The authors therefore argue for larger, more diverse multiple sclerosis genome-wide association studies and improved cross-ancestry statistical methods, so that genetic risk prediction can become scientifically robust, clinically meaningful, and equitable across populations.
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:
Breedon, J. R., Marshall, C. R., Giovannoni, G., van Heel, D. A., Dobson, R., & Jacobs, B. M. (2023). Polygenic risk score prediction of multiple sclerosis in individuals of South Asian ancestry. Brain Communications, 5(2), fcad041.
