Does Grey Matter? The Structural and Cognitive Footprint of Late-Onset Epilepsy

Abstract

Late-Onset Unexplained Seizures Are Associated With Cognitive Impairment and Lower Amygdala Volumes

Sarkis RA, Orozco J, Lemus HN, Hankerson A, Liu L, Lam AD, Johnson E, Stufflebeam S, Viswanathan A, Amariglio RE, Purandare M, Trouten P, Young GS, Locascio JJ, Pennell PB, Marshall GA. Brain Commun. 2025 Jan 31;7(1):fcaf050. doi:10.1093/braincomms/fcaf050. PMID: 39944741; PMCID: PMC11815171.

Late-onset epilepsy has been linked with accelerated cognitive decline and a higher risk of dementia. In this study, we sought to characterize the cognitive profile of participants with late-onset unexplained epilepsy and compare their MRI findings to healthy controls, to better understand underlying disease mechanisms. We recruited participants with at least one new-onset unexplained seizure at age 55 or later, without cortical lesions on MRI, within 5 years of the first seizure. We administered a neuropsychological battery to generate Preclinical Alzheimer Cognitive Composite and composite scores for delayed verbal recall, processing speed and executive function. We held a consensus meeting to determine whether the participants fulfilled criteria for mild cognitive impairment. An MRI volumetric analysis of hippocampal, amygdalae, and white matter hyperintensity volume was performed and compared to 353 healthy controls from the Harvard Aging Brain Study. On late-onset unexplained epilepsy participants, we also obtained 24-h EEG recording. Seventy participants were recruited, mean age 71.0 ± 7.0 years, 49% female, 15.6 ± 3.0 years of education. Impaired cognition (z-score ≤ −1.5) for late-onset unexplained epilepsy included the following: 15.9% for Preclinical Alzheimer Cognitive Composite −5, 23.2% for delayed verbal recall, 15.6% for processing speed, and 7.5% for executive function. Seventeen percent were found to have mild cognitive impairment. Late-onset unexplained epilepsy participants who were drug resistant were more likely to have cognitive impairment (50% vs. 9%). When controlling for age, sex, and race, late-onset unexplained epilepsy group had lower left AV (%; β = −0.003, P = 0.0016), right AV (%) (β = −0.003, P = 0.01), and log-transformed WMV (mm³; β = −0.21, P = 0.03) compared with Harvard Aging Brain Study (HABS); there were no differences in left or right HV between groups. EEG captured epileptiform abnormalities in 49% late-onset unexplained epilepsy participants, with a left temporal predominance (54%). In this single-site study of prospectively enrolled participants with late-onset unexplained epilepsy, we show that individuals with late-onset unexplained epilepsy exhibit cognitive impairments, mostly in verbal memory, and temporal dysfunction with left-sided predominance. Neuroimaging, when compared with healthy controls, shows lower amygdalae and white matter hyperintensity but not hippocampal volumes suggesting that the amygdalae is one of the earliest sites involved in the disease. The results also highlight the importance of seizure control given the association between mild cognitive impairment and drug-resistant epilepsy. Future studies extending these findings to Alzheimer's disease biomarkers and longitudinal follow-up will inform predictors of cognitive decline.

Association of Cognitive and Structural Correlates of Brain Aging and Incident Epilepsy. The Framingham Heart Study

Stefanidou M, Himali JJ, Bernal R, Satizabal C, Devinsky O, Romero JR, Beiser AS, Seshadri S, Friedman D. Epilepsia. 2025;66(1):160–169. doi:10.1111/epi.18160. Epub Nov 18, 2024. PMID: 39555677; PMCID: PMC11875459.

Objectives: Late-onset epilepsy has the highest incidence among all age groups affected by epilepsy and often occurs in the absence of known clinical risk factors such as stroke and dementia. There is increasing evidence that brain changes contributing to epileptogenesis likely start years before disease onset, and we aim to relate cognitive and imaging correlates of subclinical brain injury to incident late-onset epilepsy in a large, community-based cohort. Methods: We studied Offspring Cohort of the Framingham Heart Study participants 45 years or older, who were free of prevalent stroke, dementia, or epilepsy, and had neuropsychological (NP) evaluation and brain magnetic resonance imaging (MRI). Cognitive measures included Visual Reproduction Delayed Recall, Logical Memory Delayed Recall, Similarities, Trail Making Test B minus A (TrTB-TrTA; attention and executive function), and a global measure of cognition derived from principal component analysis. MRI measures included total cerebral brain volume, cortical gray matter volume (CGMV), white matter hyperintensity volume (WMHV), and hippocampal volume. Incident epilepsy was identified through a review of administrative data and medical records. Cox proportional hazards regression models were used for the analyses. All analyses were adjusted for age, sex, and educational level (cognition only). Results: Among participants who underwent NP testing (n = 2349, 45.81% male), 31 incident epilepsy cases were identified during follow-up. Better performance on the TrTB-TrTA was associated with a lower risk of developing epilepsy (hazard ratio [HR] .25, 95% confidence interval [CI] .08–.73; p = .011). In the subgroup of participants with MRI (n = 2056, 46.01% male), 27 developed epilepsy. Higher WMHV was associated with higher epilepsy risk (HR 1.5, 95% CI 1.01–2.20; p = .042), but higher CGMV (HR .73, 95% CI .57–.93; p = .001) was associated with lower incidence of epilepsy. Significance: Better performance on the (TrTB-TrTA), a measure of executive function and attention, and higher cortical volumes are associated with lower risk of developing epilepsy. Conversely, higher WMHV, a measure of occult vascular injury, increases the risk. Our study shows that non-invasive tests performed in mid-life may help identify people at risk for developing epilepsy later in life.

Commentary

Late-onset epilepsy (LOE) is increasingly recognized amid the rising tide of the “silver tsunami”—the unprecedented surge of the aging population worldwide. As life expectancy climbs, the burden of age-related neurological disorders, including LOE, is rising in tandem. Notably, epilepsy incidence peaks in older adults. ¹ LOE is typically defined as new-onset epilepsy after age 55 to 65, ¹ with neurodegeneration, cerebrovascular disease, and neoplasms cited as its most frequent culprits. Yet in over half of the cases, no discernible etiology is identified, leading to a diagnosis of late-onset unexplained epilepsy (LOUE). ² LOE carries dire consequences, including significantly heightened risks of stroke and dementia.^1,3 Just as coastal communities brace for the rising seas, the epilepsy field must prepare for this demographic shift by refining strategies for timely recognition, diagnosis, and management of LOE.

In this context, two recent studies offer valuable insights into LOE's structural and cognitive underpinnings. Stefanidou et al. ⁴ leveraged the longitudinal population-based Framingham Heart Study (FHS) to identify midlife neuroimaging and cognitive predictors of LOE later in life. In parallel, Sarkis et al. ¹ offered a complementary lens by adopting a prospective design to delineate the clinical and neuroimaging profile of LOUE at or near seizure onset in individuals aged 55 and older.

A central focus of both studies was the role of white matter hyperintensities (WMHs), which serve as a surrogate marker for occult cerebrovascular injury. In FHS, an elevated midlife WMH burden predicted an increased risk of LOE ⁴ —echoing prior findings from the Atherosclerosis Risk in Communities (ARIC) cohort, which demonstrated that midlife WMHs, particularly in periventricular and deep white matter, conferred a higher risk of LOE onset after age 60. ⁵ These findings collectively support a model in which cumulative, subclinical vascular damage facilitates epileptogenesis. ⁴

In an interesting contrast to FHS and ARIC, Sarkis et al. ¹ found an inverse association: individuals with LOUE exhibited lower WMH burden than matched healthy controls. This discrepancy may reflect methodological differences, including prospective versus retrospective designs, sample size differences, or selection bias. Notably, Sarkis et al.'s ¹ LOUE cohort was rigorously screened to exclude structural lesions and overt cerebrovascular disease, potentially influencing WMH prevalence. These conflicting findings underscore the complexity of LOE and highlight the need for more harmonized, longitudinal, multimodal imaging studies to clarify the temporal and causal relationship between WMHs and seizure onset.

These observations naturally raise an important question: Does grey matter (GM) also matter? Cortical volume loss has been implicated in both chronic and new-onset epilepsy. ⁶ The ARIC study reported reduced cortical volumes among individuals with LOE, ⁵ a finding corroborated by Sarkis et al., ¹ who noted smaller cortical volumes in their LOUE cohort. Stefanidou et al. ⁴ extended this further, showing that GM atrophy may emerge as early as midlife—well before clinical seizures—suggesting its potential role as a preclinical predictor of vulnerability to LOE. The progressive nature of cortical atrophy and GM loss in epilepsy ⁷ may also underpin the cognitive deficits commonly observed in this population.

Intriguingly, despite their methodological and cohort differences, both studies strikingly converge on an unexpected finding: hippocampal volumes did not significantly differ between individuals with epilepsy and healthy controls.^1,4 This absence of a detectable difference may reflect the subtlety of hippocampal atrophy in this population or the limitations of conventional volumetric analyses. In contrast, morphological shape analysis, capable of detecting more nuanced neuroanatomical changes, may prove more informative in future studies. ⁸

Strikingly, Sarkis et al. ¹ reported significantly smaller bilateral amygdala volumes in their LOUE cohort compared to healthy controls, representing a novel and potentially pivotal finding. Neuronal loss and microstructural abnormalities in the amygdala have been described in epilepsy, reinforcing the plausibility of this region's role in the emergence of late-onset seizures. ⁹ Prior studies have also demonstrated amygdalar susceptibility to tau and amyloid-β deposition in aging and Alzheimer's disease, which may account for the pronounced atrophy observed in LOUE, which has intimate ties with neurodegeneration.^3,10 Future studies employing high-resolution, longitudinal imaging are warranted to clarify whether amygdalar atrophy precedes seizure onset and/or serves as an early biomarker of LOUE risk.

Beyond structural correlates, both studies underscore cognitive impairment as a salient feature of LOE. In FHS, better midlife performance on executive function and attention tests was associated with a significantly lower risk of developing LOE. ⁴ These findings point to the predictive value of cognitive assessments conducted well before seizure onset. Meanwhile, Sarkis et al. ¹ identified cognitive impairment at or near LOUE diagnosis, with deficits most prominent in delayed verbal memory and, to a lesser extent, executive function. While both studies employed comprehensive neuropsychological batteries, Sarkis et al. ¹ detected a stronger signal for memory impairment, particularly delayed verbal recall, possibly reflecting a temporal lobe predilection in LOUE.

More importantly, 17% of participants in Sarkis et al. ¹ met criteria for mild cognitive impairment (MCI), aligning with prior observations that cognitive dysfunction is prevalent in older adults with epilepsy—especially in those with vascular or unknown etiology. Furthermore, their findings link seizure control with cognitive outcomes: individuals with drug-resistant epilepsy were significantly more likely to exhibit MCI. These results are consistent with recent work showing an elevated risk of progression from normal cognition to MCI and dementia among those with poorly controlled seizures.^3,11 This association carries immediate clinical implications, suggesting that early and aggressive seizure management in LOUE may mitigate cognitive decline. Future studies integrating longitudinal cognitive trajectories and multimodal imaging will be essential in delineating those with LOE who are on a trajectory toward dementia, and those who are not.

Despite their differing designs and some contradictory findings, both studies converge on a central theme: structural brain changes and cognitive deficits may not merely accompany LOE but precede and perhaps even predispose to its emergence. The strength of both studies lies in their robust neuropsychological and imaging datasets, meticulous analytic approaches, and efforts to account for vascular and demographic confounders.^1,4 Nevertheless, certain limitations should be acknowledged. Participants in both cohorts were predominantly white, which may restrict generalizability. EEG findings and seizure types were unavailable in FHS, and Sarkis et al. ¹ included individuals with single unprovoked seizures, potentially diluting epilepsy-specific inferences. ⁴ Moreover, the age threshold of 55 for LOUE in Sarkis et al. ¹ remains a point of contention, highlighting the need for a more uniformly accepted definition of LOUE.

Taken together, the works of Stefanidou et al. ⁴ and Sarkis et al. ¹ illuminate the multifactorial origins of LOE and delineate promising cognitive and imaging biomarkers of disease risk and trajectory. While Stefanidou et al. ⁴ highlight early predictive indicators detectable in midlife, Sarkis et al. ¹ capture the neurocognitive landscape surrounding the onset of unexplained seizures. Their complementary findings suggest a potential role for precision neurology in identifying at-risk individuals—before seizures occur—through midlife assessments of cognitive function, as well as volumetric and morphological imaging of the brain. These findings offer critical footholds in the quest to prevent, delay, or modify epilepsy and its downstream cognitive consequences in older adults.