Metabolic Pathways and Multiple Sclerosis: Insights from Mendelian Randomization
The study by Chen and colleagues, Assessing Causal Relationship Between Human Blood Metabolites and Five Neurodegenerative Diseases With GWAS Summary Statistics, investigates whether circulating metabolites contribute causally to neurological disease rather than merely changing as a consequence of illness. Although the analysis considers Alzheimer’s disease, amyotrophic lateral sclerosis, frontotemporal dementia, Parkinson’s disease, and multiple sclerosis (MS), its relevance to MS lies in its attempt to connect inherited variation in systemic metabolism with disease susceptibility. This distinction is scientifically important because conventional metabolomic studies cannot readily determine whether an altered metabolite is a causal risk factor, a marker of ongoing disease activity, a treatment-related effect, or a consequence of behavioral and physiological changes. By using genetic variants as proxies for lifelong differences in blood-metabolite concentrations, the investigators sought to move from descriptive association toward causal inference. The resulting MS findings are not definitive biomarkers or therapeutic targets, but they provide a structured set of hypotheses concerning carnitine-related metabolism and aromatic amino-acid pathways that can be tested in mechanistic and longitudinal studies.
Mendelian Randomization as a Framework for Causal Inference
The central methodology was two-sample Mendelian randomization, in which single-nucleotide polymorphisms associated with a metabolite were used as instrumental variables to estimate that metabolite’s effect on disease risk. In principle, inherited alleles are allocated before disease onset and are therefore less vulnerable than measured metabolite concentrations to reverse causation and many environmental confounders. The investigators first used inverse-variance-weighted Mendelian randomization and then applied weighted-median estimation, maximum-likelihood analysis, MR-Egger regression, MR-PRESSO, linkage-disequilibrium score regression, colocalization and reverse-direction analyses. These complementary procedures were intended to distinguish genuine causal effects from horizontal pleiotropy, shared genetic architecture and linkage disequilibrium with separate disease-causing variants. Mendelian randomization nevertheless depends on strong assumptions: genetic instruments must predict the metabolite, must not be associated with relevant confounders and must influence MS only through the metabolite under investigation. Sensitivity analyses can identify some violations of these assumptions, but they cannot prove that every instrument is biologically specific.
The Genetic and Metabolomic Data Architecture
The metabolomic component included 486 blood metabolites measured in up to 7,824 individuals of European ancestry, comprising 309 chemically identified metabolites and 177 compounds whose identities had not been conclusively established. The known molecules represented amino-acid, carbohydrate, vitamin, energy, lipid, nucleotide, peptide and xenobiotic metabolic classes. The MS genome-wide association dataset contained 68,379 participants, including 32,367 cases and 36,012 controls, with approximately 7.93 million genetic variants obtained through the International Multiple Sclerosis Genetics Consortium. The authors noted inflation in the MS association statistics, but linkage-disequilibrium score regression suggested that this pattern primarily reflected the highly polygenic architecture of MS rather than population stratification or cryptic relatedness. The genetic instruments had a minimum F statistic of 20.8, above the conventional threshold used to flag weak instruments, although the investigators adopted a relaxed metabolite-association threshold because the metabolomic GWAS was comparatively small. This decision increased analytical power but also reinforced the need for replication.
The Principal Metabolite Findings for MS
At a nominal probability threshold of 0.05, the analysis identified 18 known and 11 unknown metabolites associated with MS risk. However, none of the MS associations survived false-discovery-rate correction; the four experiment-wide significant results involved ALS or frontotemporal dementia rather than MS. This is the most important qualification when interpreting the MS results: they are suggestive, not statistically definitive after accounting for the hundreds of metabolite–disease comparisons. The strongest internally consistent MS signal was isovalerylcarnitine. Genetically predicted higher isovalerylcarnitine was associated with increased odds of MS in the inverse-variance-weighted analysis, with an odds ratio of 1.71, a 95% confidence interval of 1.20–2.45 and a nominal P value of 0.003. The association remained directionally consistent under MR-Egger, weighted-median and maximum-likelihood estimation, producing odds ratios of 3.29, 2.53 and 1.76, respectively. MR-Egger and MR-PRESSO testing did not provide evidence of horizontal pleiotropy for this relationship, which strengthens its plausibility but does not elevate it to a validated causal association.
Aromatic Amino-Acid Metabolism as an MS-Related Pathway
The pathway-enrichment analysis provided a broader biological interpretation by associating MS with “phenylalanine, tyrosine and tryptophan biosynthesis,” represented specifically by L-tyrosine and yielding a nominal P value of 0.0232. The same pathway was associated with ALS through L-phenylalanine, suggesting that distinct neurological conditions may converge on related metabolic systems without necessarily sharing the same causal metabolite or clinical course. Phenylalanine is metabolically coupled to tyrosine, and the article notes that this conversion is linked to the biosynthesis of dopamine, norepinephrine and epinephrine. It also cites previous observations of altered tryptophan and other neutral amino acids in neurodegenerative disease. For MS, the result should therefore be interpreted as evidence that genetically influenced variation within an interconnected aromatic amino-acid pathway may relate to susceptibility. It does not establish that dietary supplementation, restriction or direct manipulation of phenylalanine, tyrosine or tryptophan would prevent or treat MS. Pathway enrichment is especially vulnerable to correlated metabolites and incomplete pathway annotation, and its significance in this study was nominal rather than corrected across all tested pathways.
Biological and Translational Significance
Together, the isovalerylcarnitine and aromatic amino-acid findings suggest that the metabolic component of MS risk may not be confined to a single molecule. Instead, systemic networks involving substrate utilization, metabolite transport and amino-acid processing may interact with genetically determined disease susceptibility. A second suggestive result illustrates this complexity: the peptide designated ADpSGEGDFXAEGGGVR was associated with lower MS risk, with an odds ratio of 0.58, while showing an association in the opposite direction for ALS. Such disease-specific effect directions caution against treating “neurodegeneration” as one metabolically homogeneous process. From a translational standpoint, the study is more valuable for prioritizing experiments than for defining clinical tests. Isovalerylcarnitine could be evaluated in independent cohorts alongside disease stage, treatment status, inflammatory activity and longitudinal outcomes. The aromatic amino-acid pathway could similarly be examined through targeted metabolomics, isotope-tracing experiments and integrated genomic analyses. Any useful MS biomarker would need to demonstrate reproducibility, disease specificity and incremental predictive value beyond established clinical, imaging and laboratory measures.
Limitations and the Research Agenda for Multiple Sclerosis
Several limitations restrict the immediate application of these findings. The metabolite GWAS was modest in size, reducing power and increasing uncertainty around instrumental effects. Both the metabolite and neurological-disease datasets were restricted to participants of European ancestry, limiting generalizability to populations with different genetic backgrounds, environmental exposures and metabolic profiles. Most importantly, the study measured metabolites in blood rather than in brain, cerebrospinal fluid or specific immune-cell populations; the authors explicitly acknowledge that metabolite concentrations differ among tissues and that circulating measurements may not represent the most biologically relevant compartments. The MS associations also remained below experiment-wide significance after multiple-testing correction. Consequently, the article does not establish isovalerylcarnitine or aromatic amino-acid metabolism as clinical targets. Its principal contribution is conceptual: it demonstrates how genomics and metabolomics can be combined to separate potential metabolic causes from secondary disease effects. Future MS research should replicate these signals in larger, ancestrally diverse metabolomic GWAS datasets, identify the responsible tissues and cell types, resolve temporal relationships with disease activity and test whether the implicated metabolites participate directly in pathogenesis or merely index broader biochemical networks.
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:
Chen, H., Qiao, J., Wang, T., Shao, Z., Huang, S., & Zeng, P. (2021). Assessing causal relationship between human blood metabolites and five neurodegenerative diseases with GWAS summary statistics. Frontiers in Neuroscience, 15, 680104.
