Multiple Sclerosis at the Crossroads of Genetics and Metabolism: How GWAS, Polygenic Risk Scores, and Metabolites Are Reshaping Disease Biol

Why an integrated view of multiple sclerosis is scientifically necessary. Multiple sclerosis (MS) is no longer best understood as a disease explained by a single immunologic pathway or a handful of risk alleles. The modern literature increasingly frames MS as a multilayered disorder in which inherited susceptibility, immune-cell regulation, central nervous system resilience, and downstream metabolic changes all interact over time. Genomic studies have mapped a broad architecture of risk, while metabolomics has begun to capture the biochemical consequences of that architecture in blood, cerebrospinal fluid, and tissue. In practice, this means that GWAS identifies who may be predisposed, PRS estimates cumulative inherited burden, and metabolite profiling offers a dynamic readout of disease biology that genetics alone cannot provide.

GWAS has established MS as a strongly polygenic disease with a dominant but incomplete HLA signal. The landmark genomic map produced by the International Multiple Sclerosis Genetics Consortium analyzed 47,429 people with MS and 68,374 controls, identifying 32 susceptibility variants within the extended major histocompatibility complex, 200 autosomal variants outside the MHC, and one chromosome X variant. That work reinforced the central role of immune biology and microglial pathways, while also showing that the HLA region, although crucial, does not explain the full inherited component of disease. Subsequent articles continue to cite this architecture as the reference framework for MS genetics, with HLA-DRB1*15:01 remaining the best-known major risk allele. The scientific implication is important: MS susceptibility is distributed across many loci of modest effect, which is precisely the type of architecture that makes PRS analytically useful.

Polygenic risk scores convert dispersed genetic associations into a clinically interpretable measure of inherited risk. PRS aggregates the weighted contribution of many susceptibility variants into a single quantitative estimate of liability. In a large European-ancestry analysis summarized by UK Biobank, MS PRS showed meaningful discrimination for disease susceptibility, with an AUC of 0.73 in UK Biobank and 0.80 in Kaiser Permanente Northern California; people in the top 10% of PRS had more than a fivefold increased risk in UK Biobank and roughly fifteenfold increased risk in the KPNC cohort relative to the median decile. A separate Neurology study on lifetime risk further showed how strongly inherited burden stratifies long-term probability: among women, those in the lowest 30% of genetic risk had an estimated lifetime risk around 1 in 2,739, whereas those in the top 10% had an estimated risk around 1 in 92. These are not deterministic predictions, but they demonstrate that cumulative common-variant burden is biologically real and epidemiologically meaningful.

Metabolomics adds what static DNA cannot: a biochemical portrait of active disease biology. If GWAS and PRS speak to predisposition, metabolomics speaks to state. Reviews of omics in MS consistently emphasize that metabolomics can reveal inflammation, neuroaxonal injury, mitochondrial stress, lipid remodeling, and treatment response in ways that genotype alone cannot. Recent work has highlighted recurrent involvement of lipid species, amino-acid pathways, and energy-related metabolites across serum, plasma, CSF, and lesion tissue. This is why metabolomics is becoming attractive not only for biomarker discovery, but also for mechanistic interpretation: it may show how a genetically susceptible immune-neural system is behaving at a particular moment, and whether that behavior tracks relapse, progression, or therapeutic exposure.

Mendelian randomization is the bridge connecting GWAS-scale genetics to metabolite causality. One of the most important developments in this field is the use of metabolite GWAS instruments within Mendelian randomization frameworks. In the 2023 Journal of Neuroimmunology study by Ge and colleagues, investigators used genetic instruments for 571 circulating metabolites and tested their causal relationships with MS risk using two-sample MR. They reported 29 metabolites with suggestive evidence of causal association, including serine, lysine, acetone, acetoacetate, and several lipid measures. Genetically instrumented higher serine, lysine, acetone, and acetoacetate were associated with higher MS risk, while some lipid fractions showed directionally protective associations and others appeared risk-increasing depending on lipoprotein context. This is a pivotal conceptual shift: metabolites are no longer viewed only as downstream correlates, but as candidates that may sit on causal pathways between inherited variation and disease.

A major scientific advance is the recognition that susceptibility genetics and severity genetics are overlapping but not identical. Risk loci do not fully explain why some patients accumulate disability faster than others. In a Nature study of MS severity, a GWAS of age-related MS severity score in 12,584 cases, followed by replication in 9,805 more cases, identified rs10191329 in the DYSF–ZNF638 locus; homozygous carriers of the risk allele reached the need for a walking aid a median of 3.7 years earlier. The same study found heritability enrichment in central nervous system tissues and suggested, through Mendelian randomization, a potentially protective effect of higher educational attainment, leading the authors to emphasize CNS resilience rather than purely immune susceptibility as a determinant of progression. This distinction matters greatly when integrating metabolomics: the most informative metabolites for risk may not be the same metabolites that best reflect progression, repair failure, or neurodegeneration.

Multi-omic studies suggest that the most informative biology emerges when metabolites are interpreted alongside transcriptomic and genetic context. A 2024 iScience study reported that combined blood metabolomic and transcriptomic signatures stratified patient subgroups according to MS severity, underscoring that metabolic differences are more informative when embedded within broader regulatory networks. Rather than treating one altered metabolite as a standalone biomarker, this systems-level approach asks which pathways, cell states, and transcriptional programs travel with the metabolite pattern. That is more biologically persuasive, because metabolites are often many steps downstream of gene regulation and immune activation. The direction of the field is therefore clear: future clinically useful signatures will likely combine PRS, pathway-level genetic annotation, transcriptomic context, and metabolite panels instead of relying on any single data type in isolation.

Combined models are beginning to show practical value in prediction, but they are not yet ready to function as standalone diagnostic tools. A 2024 Nature Communications study applied an MS genetic risk score model to people presenting with optic neuritis and found that adding genetic risk to demographic and clinical variables improved risk stratification for future MS diagnosis. The broader PRS literature similarly shows that genetic scores can improve discrimination beyond conventional factors, but performance remains moderate rather than definitive. This is the crucial translational lesson: PRS is informative, not dispositive. The same caution applies to metabolomics. Metabolite signatures may improve early classification, activity monitoring, or prognosis, but they remain vulnerable to treatment effects, diet, comorbidity, body composition, fasting state, and assay heterogeneity. The combined approach is therefore promising precisely because each layer compensates for the weaknesses of the others.

Ancestry remains one of the most important unresolved issues for both GWAS interpretation and PRS portability. Most MS genetic discovery has historically been driven by European-ancestry cohorts, and this creates a translational bottleneck. A 2023 Brain Communications study showed that PRS derived from European GWAS performs less accurately in a South Asian population, highlighting a familiar problem in complex-trait genetics: predictive models do not transfer cleanly across ancestries when training data are unbalanced. More recent multi-ancestry work has begun to address this. A 2024/2025 study described in PubMed and related sources identified 236 susceptibility variants in a multi-ancestry GWAS of 20,831 MS cases and 729,220 controls, linked 76 genes to risk-variant effects, and suggested that although T-cell enrichment remains strong, inhibitory neurons may also be relevant to susceptibility biology. This kind of expansion is essential if PRS and metabolite-informed stratification are to become scientifically fair and clinically generalizable.

The future of MS precision medicine will likely depend on integrating inherited liability with dynamic molecular readouts. The most compelling scientific trajectory is not a competition between genetics and metabolomics, but their convergence. GWAS and PRS define a person’s background architecture of risk; MR helps prioritize which metabolites may be mechanistically upstream rather than merely reactive; multi-omic profiling clarifies which biochemical networks track severity or progression; and emerging severity GWAS reminds us that susceptibility and disability are partly separable biological domains. The field now needs large longitudinal cohorts with repeated biospecimens, harmonized metabolite assays, multi-ancestry genetic representation, and clinically meaningful endpoints such as relapse recovery, radiographic progression, and disability accumulation. When that infrastructure matures, the combined use of genetic variation, GWAS, PRS, and metabolomics may move MS research from statistical association toward a genuinely mechanism-based precision neurology.

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., Chataway, J., Ramagopalan, S. V., & Dobson, R. (2023). Polygenic risk score prediction of multiple sclerosis based on European genome-wide association study results is less accurate in a South Asian population. Brain Communications, 5(2), fcad041.
Ge, A., Sun, Y., Kiker, T., Zhou, Y., & Ye, K. (2023). A metabolome-wide Mendelian randomization study prioritizes potential causal circulating metabolites for multiple sclerosis. Journal of Neuroimmunology, 379, 578105. https://doi.org/10.1016/j.jneuroim.2023.578105
Harroud, A., Stridh, P., Xu, Y., et al. (2023). Locus for severity implicates CNS resilience in progression of multiple sclerosis. Nature, 619, 323–331.
International Multiple Sclerosis Genetics Consortium. (2019). Multiple sclerosis genomic map implicates peripheral immune cells and microglia in susceptibility. Science, 365(6460), eaav7188. https://doi.org/10.1126/science.aav7188
Loginovics, P., Lechner-Scott, J., Čierny, D., et al. (2024). Applying a genetic risk score model to enhance prediction of future multiple sclerosis diagnosis at first presentation with optic neuritis. Nature Communications, 15, Article 785. https://doi.org/10.1038/s41467-024-44917-9
Loonstra, F. C., Álvarez Sirvent, D., Tesi, N., Holstege, H., Strijbis, E. M. M., Salazar, A. N., Hulsman, M., van der Lee, S. J., & Uitdehaag, B. (2024). Association of polygenic risk score with lifetime risk of developing multiple sclerosis in a population-based birth-year cohort. Neurology, 103(7), e209663. https://doi.org/10.1212/WNL.0000000000209663
Lorefice, L., Fenu, G., Frau, J., & Coghe, G. (2023). Omics approaches to understanding the efficacy and safety of disease-modifying treatments in multiple sclerosis. Frontiers in Genetics, 14, 1076421. https://doi.org/10.3389/fgene.2023.1076421
Oppong, A. E., Coelewij, L., Robertson, G., Martin-Gutierrez, L., Waddington, K. E., Dönnes, P., Nytrova, P., Farrell, R., Pineda-Torra, I., & Jury, E. C. (2024). Blood metabolomic and transcriptomic signatures stratify patient subgroups in multiple sclerosis according to disease severity. iScience, 27(3), 109225. https://doi.org/10.1016/j.isci.2024.109225
Shams, H., Shao, X., Santaniello, A., et al. (2022). Polygenic risk score association with multiple sclerosis susceptibility and phenotype in Europeans. Brain, 145(8), 2971–2986. https://doi.org/10.1093/brain/awac092
Zeng, L., Atlas, K., Lama, T., et al. (2025). GWAS highlights the neuronal contribution to multiple sclerosis susceptibility. Research Square [Preprint]. https://doi.org/10.21203/rs.3.rs-5644532/v1