Polygenic Risk Scores and the Future of Multiple Sclerosis Risk Prediction
The article by Shams and colleagues examines how polygenic risk scores can improve understanding of multiple sclerosis susceptibility and disease phenotype in individuals of European descent. Multiple sclerosis is a chronic inflammatory demyelinating disease of the central nervous system, and its genetic basis is distributed across many loci rather than determined by a single variant. The authors therefore use a polygenic risk score, or PRS, to aggregate small genetic effects across the genome into a single quantitative measure of inherited liability. Their central argument is that PRS can move multiple sclerosis genetics from association discovery toward risk stratification, mechanistic interpretation, and clinically relevant prediction.
Study Design and PRS Construction
The study derived its multiple sclerosis PRS from International Multiple Sclerosis Genetics Consortium genome-wide association data, using post-quality-control summary statistics from 14,802 cases and 26,703 controls. The authors applied LDPred2, a Bayesian method that adjusts variant effect estimates while accounting for linkage disequilibrium, and restricted development to high-quality HapMap3 variants shared across validation and test datasets. The workflow shown in the page 3 diagram illustrates the analytical pipeline: GWAS summary statistics were used to generate genetic models, models were optimized in UK Biobank Phase 1, and the best-performing model was then tested in UK Biobank Phase 2 and the Kaiser Permanente Northern California cohort.
Evidence for Strong Risk Discrimination
The principal finding is that MS-PRS substantially distinguishes individuals with multiple sclerosis from unaffected controls. In UK Biobank Phase 2, the PRS achieved an area under the receiver-operating curve of 0.73, while in the Kaiser Permanente Northern California cohort it reached 0.80, indicating stronger discrimination in the more clinically curated case-control setting. Individuals in the top 10% of the PRS distribution had more than a fivefold increased risk in UK Biobank and approximately a fifteenfold increased risk in the Kaiser cohort relative to the median decile. The figure on page 5 visually reinforces this result by showing higher PRS distributions among affected individuals and a progressive rise in disease prevalence across PRS percentiles.
Added Value Beyond Conventional Risk Factors
A major contribution of the article is its demonstration that genetic profiling improves risk models beyond conventional epidemiological and clinical variables. When PRS was added to models containing age, sex, mononucleosis history, smoking, body mass index, childhood overweight status, and family history, model performance improved consistently. In UK Biobank, the best-performing model including PRS, mononucleosis, smoking, sex, and age achieved an AUC of 0.78; in the Kaiser cohort, adding PRS to a model containing several conventional risk factors increased AUC to approximately 0.83. The authors also show that a genome-wide PRS outperformed a model based only on the HLA-DRB1*15:01 tagging variant, supporting the view that multiple sclerosis susceptibility is broadly polygenic rather than reducible to the major histocompatibility complex.
Biological Interpretation Through Pathway-Specific Scores
Beyond prediction, the article uses pathway-specific PRS to interpret the biological architecture of multiple sclerosis risk. Out of 85 replicated risk-associated pathway scores, many were linked to adaptive immune function, including IL-5 signalling, IL-12 signalling, T-cell receptor signalling, MHC class II antigen presentation, interferon-gamma signalling, and the complement cascade. The study also identifies viral and parasite infection response pathways, extracellular matrix organization, cell adhesion, protein glycosylation, VEGF signalling, and NOTCH-related pathways. These findings are biologically coherent because multiple sclerosis involves immune activation, antigen presentation, leukocyte trafficking into the central nervous system, and inflammatory tissue injury.
Familial Risk, Neuroimaging, and Disease Activity
The article extends PRS analysis into multicase families and neuroimaging phenotypes. In families with affected parent-child pairs and discordant siblings, affected individuals had higher familial PRS than controls, and the sibling-level AUC was approximately 0.65, suggesting that PRS may help identify elevated inherited risk within families. The figure on page 8 summarizes these familial patterns, including correlations between parental and sibling PRS. The study also links higher PRS to radiographic markers of disease progression, particularly thalamic atrophy over a 10-year UCSF-EPIC follow-up. On page 9, the thalamic atrophy panels show greater longitudinal volume loss with higher PRS, while additional analyses found an association between PRS and relapse activity but not clear EDSS worsening.
Clinical Relevance and Future Directions
The study positions PRS as a promising component of future multifactorial risk assessment for multiple sclerosis, particularly for individuals with suggestive symptoms, family history, or other established risk factors. However, the authors appropriately caution that PRS is not yet a stand-alone population screening tool because multiple sclerosis has a low baseline prevalence and genetic prediction alone remains insufficient for definitive diagnosis. A further limitation is ancestry: the analysis focuses primarily on European-descent cohorts, and the authors emphasize the need for larger, diverse GWAS resources to make polygenic profiling equitable and transferable across populations. Overall, the article represents an important step toward integrating statistical genetics, immunobiology, neuroimaging, and clinical risk modelling in multiple sclerosis research.
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.
