Genetic-Proteomic Integration Identifies Predictive Plasma Proteins for Multiple Sclerosis.

OBJECTIVE

Multiple sclerosis (MS) develops after a prolonged preclinical phase. Identifying circulating biomarkers that capture this early biology can improve risk stratification and guide intervention. We aimed to identify plasma proteins driving MS susceptibility using large-scale proteogenomic integration and to evaluate their prediagnostic predictive value and effects on severity.

METHODS

We used cis-acting protein quantitative trait loci (pQTL) for 2,545 plasma proteins (n = 80,824). Predicted protein levels were tested for association with MS risk in 14,802 cases and 26,703 controls using Mendelian randomization and colocalization. Splicing annotations were integrated to interpret platform-specific discrepancies. We validated candidates as predictors of incident MS in prediagnostic United Kingdom Biobank samples (124 cases, 52,515 controls; median 5.9 years prediagnosis) and assessed associations with disease severity (n = 12,584).

RESULTS

We identified 39 proteins associated with MS risk. Most formed a densely connected network enriched in immune regulatory pathways, B- and T-cell costimulation, cytokine signaling, and Epstein-Barr virus-related pathways. Transcriptomic enrichment was strongest in B-cell subsets. Splicing data suggested that discordances between proteomic platforms reflect distinct proteoforms. Among 28 genetically implicated proteins measured in prediagnostic samples, 8 were associated with time to MS diagnosis, demonstrating enrichment beyond chance expectation (pbinom = 4.92 × 10-5). DKKL1 showed concordant protective associations across risk, incidence, and severity. Integrating pQTLs improved fine-mapping resolution at colocalized loci by >10-fold and nominated 13 putative novel risk loci.

INTERPRETATION: This integrated genetic-proteomic framework identifies causal proteins, refines risk loci, and supports the development of early predictive biomarkers with translational potential. These findings support genetically anchored biomarkers for preclinical disease detection and intervention. ANN NEUROL 2026;100:268-281.