The Case of the Accelerated Clock: Tau as a Protein of Interest in Epilepsy

Abstract

In Vivo Tau in Epilepsy Reflects Clinical Severity and Immune- and Aging-Related Proteomic Changes

Hong SB, Shin YW, Moon J, Park KI, Jung KY, Cho K, Choi H, Chu K, Lee SK. Brain. 2026:awag145. doi: 10.1093/brain/awag145

Tau pathology plays a central role in a number of neurodegenerative diseases, but its presence and relevance in epilepsy remain incompletely understood. Emerging evidence suggests that epilepsy may promote tau accumulation, yet whether this occurs in vivo, independent of comorbid dementia or amyloid pathology, is unclear. In this study, we combined [18F]flortaucipir (FTP) positron emission tomography (PET) imaging with high-throughput plasma proteomics to characterize regional tau deposition and its clinical and molecular correlates in patients with nondemented epilepsy. We enrolled 75 epilepsy patients and 47 age- and sex-matched healthy controls, collecting detailed clinical data, electroencephalography (EEG) features, and plasma samples for SOMAscan proteomic profiling, and plasma p-tau217, total tau, and amyloid-β measurement. A subset underwent FTP PET and [18F]florbetaben (FBB) PET imaging. Regional standardised uptake value ratios (SUVRs) were quantified using the AAL3 brain atlas. Compared to controls, epilepsy patients exhibited globally elevated FTP uptake across cortical regions, particularly in the lateral and medial frontal, lateral parietal, and lateral occipital brain areas, while FBB SUVRs showed nonsignificant differences. Exploratory analyses highlighted EEG slowing, multifocal discharges, and continued seizure activity during adolescence as clinical features associated with higher FTP SUVRs. In lateralized epilepsy, asymmetry indices tended to favor the hemisphere with the seizure onset zone. Plasma proteomic analysis identified 473 differentially expressed proteins in epilepsy, enriched in pathways related to immune activation, metabolism, and cytoskeletal remodeling. Protein expression associated with regional tau SUVRs again emphasized immune pathways as well as mitochondrial dysfunction; and suggested distinct mechanisms of tau accumulation in a region-specific manner. Furthermore, using the OrganAge algorithm, we found accelerated biological ageing in epilepsy patients across several organs, including the brain, heart, and muscle. While brain age gaps showed the strongest positive correlation with tau, the heart, pancreas, and muscle age gaps also showed correlations with regional brain tau, suggesting a link between systemic ageing and brain tau accumulation. Together, these findings suggest that epilepsy is associated with widespread elevated tau tracer signal that relates to EEG abnormalities, clinical disease burden, and immune- and aging-related proteomic signatures. Our results raise the possibility that tau accumulation contributes to key aspects of epilepsy pathophysiology and may have relevance for biomarker development and future therapeutic targeting.

Commentary

The role of neurodegenerative proteins in the pathophysiology of epilepsy has garnered considerable attention in recent years. ¹ Animal models have shown that hyperphosphorylated tau, one of the core pathological features of Alzheimer's disease, increases with hyperexcitability, improves seizure outcomes when suppressed or knocked out, and is present in surgical specimens from individuals with epilepsy undergoing epilepsy surgery.^2,3 With the advent of tau positron emission tomography (PET) tracers, and importantly, FDA approval of these imaging tracers, it has become essential to ask: what role does p-tau build up play in human epilepsy? Is it a treatment target, or an epiphenomenon? This article tries to address this comprehensively by recruiting individuals with epilepsy without a history of dementia, and evaluating tau PET, epilepsy characteristics, proteomic markers, and organ aging clocks, incorporating SOMAscan proteomics to correlate elevated brain tau with distinct molecular pathways.

In this prospective study based in South Korea, ⁴ individuals admitted to the epilepsy monitoring unit were enrolled if they had no subjective memory complaints, an mini mental state examination score above 25, no diagnosis of dementia, and no history of major head trauma. Participants underwent clinically indicated 3T structural magnetic resonance imaging (MRI). Only lesion-negative patients were eligible for tau PET with [18F]flortaucipir (FTP), a subset also underwent amyloid PET with [18F]florbetaben. Of 75 enrolled epilepsy patients, 57 were lesion-negative and eligible for tau PET, but only 32 completed the scan, versus 24 of 47 controls; amyloid PET was completed in just 20 epilepsy patients and 7 controls. Scalp electroencephalography (EEG) was used to lateralize seizure onset and quantify interictal discharges. Standardized uptake value ratios (SUVRs) with the cerebellum as reference were extracted and aggregated into 12 composite cortical regions; amyloid PET was visually scored. Finally, the authors applied the OrganAge algorithm to estimate organ-specific biological age from proteomic data. Epilepsy patients, with a mean age of 46.9 and epilepsy duration ranging from recently diagnosed to multidecade disease, showed globally elevated FTP uptake relative to controls across nearly all cortical regions, most prominently frontal and lateral cortices, while amyloid burden did not differ between groups. This pattern was robust to adjustment for education and persisted after excluding late-onset epilepsy (age > 60). Plasma p-tau217 was not significantly elevated on average in epilepsy patients, but a significantly greater proportion had p-tau217 above the upper normal limit (24% vs 5%); plasma amyloid-β42/40 ratios did not differ between groups, and the large majority of amyloid PET scans in both groups were read as negative.

Elevated tau signal correlated most robustly with multifocal interictal discharges, while spike frequency and monthly seizure frequency showed no significant association.

Among etiologies, postencephalitic epilepsy showed the highest FTP uptake, and tau signal trended higher in the epileptic hemisphere; regional tau showed no correlation with cortical atrophy pattern. Differentially expressed proteins in epilepsy, 473 in total and overwhelmingly skewed toward upregulation (462 up vs 11 down), were enriched in ErbB signaling, immune pathways, and ubiquitin-proteasome and mitochondrial components. Proteins positively correlated with regional tau were enriched in mitochondrial function and cellular stress response pathways, while negatively correlated proteins were enriched in innate immune pathways including interleukin (IL)-17 signaling. Finally, organ age gaps were significantly elevated in epilepsy for brain, kidney, muscle, and pancreas, and for overall organismal age, though heart age gap did not differ significantly between groups; despite this, brain and heart age gaps showed the most consistent positive correlation with regional tau SUVR, while muscle age gap correlated most strongly with the sensorimotor region. Epilepsy duration and antiseizure medication burden both correlated with organ age gap.

This is one of the first comprehensive studies to characterize in vivo tau in epilepsy and correlate it with clinical, EEG, and proteomic measures. Such studies are necessary to begin identifying what role, if any, tau plays in the pathophysiology of epilepsy and whether it carries genuine disease relevance rather than representing an incidental bystander. The findings indicate that tau is indeed present in at least a subset of epilepsy patients, that it correlates with multifocal discharges, and that it may be one more indicator of a broader pattern of brain and systemic aging, with the proteomic data pointing to accelerated aging in other organs as well.

It must be emphasized, however, that this study looks at epilepsy through an Alzheimer's lens, using AD-derived tools such as flortaucipir PET, which is most sensitive to mature 3R/4R paired-helical-filament tau aggregates and may miss other tau species previously identified in pathological specimens from epilepsy surgery. ⁵

Importantly, the distribution of tau observed here does not follow the classic AD cascade: burden is more attenuated than in AD, and the pattern instead highlights a frontal vulnerability, one also seen with amyloid imaging in a separate epilepsy cohort,^6,7 hinting at a shared regional susceptibility that may be disease rather than protein-specific. There are also technical limitations inherent to the tracer itself, including the possibility of artifactual signal from altered blood flow or off-target binding, and the sample supporting the imaging findings is itself a modest, nonrandomly selected subgroup of the full cohort. ⁸

The proteomic analysis is similarly compelling but cross-sectional, and relies on a cohort sampled during an epilepsy monitoring unit admission, a setting that carries substantial physiological stress, whether from the admission itself or the anticipation of a captured seizure, and this stress response could plausibly modify the proteomic signature attributed here to epilepsy. Curiously, 98% of the proteins were upregulated in epilepsy, which is an unusual pattern as one would often expect a mixture of both upregulated and downregulated proteins in disease states. The implicated mechanisms, mitochondrial dysfunction, ubiquitin-proteasome disruption, neuroinflammation, again point to processes that, as the field moves away from compartmentalized thinking, are increasingly understood to be interrelated; neuroinflammation in particular is taking center stage across a range of neurologic diseases, epilepsy included. ⁹

The authors extend this proteomic signal by applying an organ-aging algorithm, showing that proteins associated with brain, heart, and muscle differ in epilepsy and correlate with tau burden, potentially indicating a broader process of accelerated aging. Notably, the study did not find that seizure burden or interictal discharge frequency correlated with tau, which raises the question of whether the multifocal discharge finding reflects a genuine driver of tau pathology through diffuse network involvement, or is instead an incidental association. Epilepsy duration and antiseizure medication burden were both associated with accelerated organ aging, both modifiable factors, particularly antiseizure medication burden, which deserves closer scrutiny as a potential contributor rather than a passive marker.

Overall, it is valuable to apply tools from other disciplines to understand a relatively understudied disease process. But the data may also be telling us that epilepsy is its own entity: it likely overlaps mechanistically with Alzheimer's disease and other neurodegenerative conditions, but appears to involve distinct underlying processes rather than representing early or attenuated AD. Future work should aim to hone in on the modifiable factors, particularly medication burden and disease duration, that may be driving the accelerated aging signal identified here.