Using the UK Biobank data, we performed a 17-year longitudinal survival analysis to examine the link between 2,920 plasma proteins and the risk of epilepsy onset.

Epilepsy, a prevalent spectrum disorder, is characterized by the absence of a definitive diagnostic gold standard. The emergence of proteomics as a powerful technology has significantly enhanced the ability to identify disease-related protein markers and uncover potential drug targets.

The multiple Cox proportional hazard regression model explored the association between 2920 plasma protein levels and incident epilepsy risk. Then, we selected epilepsy-related proteins and showed the trajectory changes of these proteins over the 15 years before epilepsy diagnosis. The linear regression model was applied to investigate the associations of selected epilepsy-associated proteins with neuroimaging data, polygenic risk scores (PRS) for epilepsy, and environmental variables. Biological analyses such as functional enrichment analysis and PPI networks are utilized to find potential biological pathways and drug targets. Finally, we investigated the potential of plasma proteins to predict epilepsy incidence through a two-stage machine-learning approach.

We identified 103 proteins with significant associations with epilepsy, with NEFL (HR [95% CI]: 2.13 [1.85-2.46], P = 3.36×10^-25) and GDF15 (1.82 [1.60-2.07], P = 5.93×10^-20) exhibiting the strongest correlations. Then we mapped the trajectory of changes in abnormal plasma protein levels over 15 years preceding epilepsy diagnosis. Biological analyses uncovered the pivotal role of the immune response and pinpointed four central hubs (TNFRSF1A, HAVCR2, CD274, and TIMP1). Furthermore, 103 screened proteins were significantly associated with several brain regions implicated in epileptogenesis, including the hippocampus and thalamus. These proteins showed a significantly stronger correlation with stress-related events than with genetic predisposition to epilepsy.

These findings are instrumental for the identification of early predictive biomarkers for epilepsy and the optimization of therapeutic strategies.