Late-onset epilepsy (LOE) may represent a distinctive neurodegenerative-linked epilepsy subtype, with new data showing that individuals who develop epilepsy after midlife have substantially greater Alzheimer-type pathology at autopsy, a higher burden of medical comorbidities, and a large proportion of cases with no identifiable cause.

Epilepsy is most common in older adults, and LOE —typically defined as recurrent seizures beginning after the age 65 — is typically attributed to an acquired brain insult.

Preliminary clinical data suggest that LOE carries its own signature. Patients are more frequently female, have a heavier comorbidity load but lower disability, and rarely develop drug-resistant epilepsy (DRE) than those with early-onset epilepsy (EOE). Separate autopsy findings extend the picture, revealing that those who develop seizures later in life harbor more advanced Alzheimer-type and related neurodegenerative pathology, supporting the idea that LOE may arise from different underlying mechanisms.

“Late-onset and early-onset epilepsy should be recognized as distinct profiles and not just lumped together under the banner head of ‘epilepsy,’” study investigator Syeda Amrah Hashmi, MD, Research Fellow, Department of Neurology, University of Virginia (UVA), Charlottesville, Virginia, told Medscape Medical News.

She added that recognizing these distinctions could help refine management and screening strategies and ultimately improve patient outcomes.

The findings were presented on December 6 at the American Epilepsy Society (AES) 79th Annual Meeting 2025.

Profiling LOE

Although stroke and neurodegenerative disorders like Alzheimer’s disease (AD) are two of the most common causes of LOE, tumors and autoimmune disorders may also contribute, said study co-investigator, Ifrah Zawar, MD, assistant professor of neurology, UVA Health.

In the retrospective single-center study, 74 patients with clinically diagnosed epilepsy were evaluated. Demographic and comorbidity data were collected, along with seizure-related variables such as age at onset, seizure type, and prior status epilepticus, using information from electronic medical records.

Based on age at onset, researchers classified half of the cohort as having LOE and half as having EOE. The mean age at onset was 72.8 years in the LOE group and 57.7 years in the EOE group. Hashmi noted that a larger ongoing dataset includes patients with EOE with younger onset ages.

Women were significantly more likely to have LOE than EOE (62.2% vs 37.8%; P = .0377). Patients with LOE were also more likely to be White than non-White (89.2% vs 70.3%; P = .0496).

Patients with LOE, who were older, had significantly higher Charlson Comorbidity Index scores than those with EOE (5.11 vs 3.84; P = .0082). Still, disability classifications were more common in EOE (16.2% vs 0%; P = .010), a difference Zawar attributed to work-related disability categories rather than neurologic impairment.

“Individuals whose epilepsy began before retirement age may experience employment interruption, driving limitations, and socioeconomic burden over time, whereas late-onset patients are generally beyond working age and therefore not categorized as disabled despite potentially having significant medical comorbidity,” said Zawar.

DRE was identified in 10.8% of patients with EOE but in none of those with LOE. This pattern is consistent with clinical experience, where LOE is generally more treatment-responsive than epilepsy that begins earlier in life and may have been evolving for years.

Exactly why LOE tends to be easier to treat remains unclear, Zawar noted, but it is likely related to LOE being etiologically distinct from EOE.

Routine EEGs were performed in 67.6% of patients with EOE and 73.0% of those with LOE, with epileptiform discharges or recorded seizures detected at similar rates (28.0% vs 25.9%). MRI abnormalities were common, present in 92.5% of patients overall. Routine EEG was performed in 67.6% of patients with EOE and 73.0% of patients with LOE. 

Rates of epileptiform discharges or captured seizures were comparable between patients with EOE and LOE (28.0% vs 25.9%), and MRI abnormalities were seen in 92.5% of participants.

“The details of neuroimaging results are still under investigation and will be better described in our future studies,” said Zawar.

Hashmi considers this pilot analysis a first step, with a larger dataset expected to clarify how LOE might inform more tailored treatment approaches.

Neuropathologic Differences

Another poster presented at the AES meeting examined neuropathologic features of LOE. In a large autopsy cohort, individuals with LOE died at a younger age (73 vs 81 years) and showed substantially greater neurodegenerative pathology — including tau and amyloid deposition — than those without epilepsy.

“These findings suggest that LOE may represent an early or parallel manifestation of underlying neurodegenerative disease. Identifying LOE as a potential marker of accelerated neurodegeneration has critical implications for early identification, risk stratification, and targeted intervention in aging populations,” said Zawar.

The broader picture is one of considerable etiologic uncertainty: 25%-50% of new-onset epilepsy in older adults has no identifiable cause. Reflecting this, Zawar’s group also presented a systematic review of 94 studies examining late-onset unexplained epilepsy (LOUE).

This analysis showed focal seizures were the most common seizure type, and vascular risk factors, particularly hypertension, were a common comorbidity. Participants with LOUE often showed interictal epileptiform discharges on EEG.

About 84% of individuals with LOUE achieved seizure freedom at 1-3 years of follow-up. Across studies that reported outcomes, an average of 76% of patients with LOUE were on monotherapy.

“Our systematic review highlights the many gaps that remain in understanding the epidemiology, clinical presentation, EEG, etiology, and treatment options for LOUE,” said Zawar.

The NEUROPATHOLOGY study received funding from the National Institutes of Health, the American Epilepsy Society, the Alzheimer’s Association, and the BAND Foundation. The investigators reported having no relevant financial disclosures.