Before It's Too Late: Recognizing the Impact of Seizures and Risk Factors for Late-Onset Epilepsy in Dementia and Mild Cognitive Impairment

Abstract

Association of Seizure Control with Mortality, Cognition, and Function in People With Dementia

Zawar I, Quigg M, Ghosal S, Punia V, Calle-Lopez Y, Manning C, Kapur J. Ann Neurol. 2025;97(2):358–368. doi: 10.1002/ana.27125. Epub 2024 Oct 30. PMID: 39475168

Objectives: The effects of seizure control on outcomes in persons with dementia (PWD) remain unclear. Our study aimed to investigate the impact of seizure control on mortality, function, cognition, and mood among PWD. Methods: This longitudinal, multicenter study is based on 39 Alzheimer's disease centers (ADCs) in the United States from September 2005 to December 2021. PWD were grouped by seizure status into recurrent (seizures in the past year), remote (prior seizures but none in the past year), and no seizures (controls). The primary outcome was all-cause mortality among seizure groups. We used Weibull survival analysis to assess the mortality risks by seizure status after adjusting for age, sex, education, race, ethnicity, hypertension, diabetes, hyperlipidemia, degree of cognitive impairment, dominant Alzheimer's disease (AD) mutation, brain trauma, stroke, Parkinson's disease, alcohol abuse, and depression. Cognition (Clinical Dementia Rating), function (physical dependence and nursing home residence), day-to-day activities (Functional Assessment Scores), and mood (Geriatric Depression Scale) were compared among seizure groups after adjusting for dementia duration and age. Results: Among 26,501 participants, 374 (1.4%) had recurrent seizures and 510 (1.9%) had remote seizures. In multivariable survival analysis, recurrent seizures were associated with a higher mortality risk than remote and no seizures (adjusted hazard ratio [aHR], 95% confidence interval [95% CI]; recurrent aHR = 1.79, 95% CI = 1.51 to 2.12; remote aHR = 1.17, 95%CI = 0.98 to 1.38). Median time-to-death for recurrent, remote, and no seizures was 2.4, 4.0, and 4.7 years, respectively. People with recurrent seizures had worse cognition, day-to-day function, and physical dependence than those with remote seizures and controls. Interpretation: PWD with poorly controlled recurrent seizures have worse mortality, functional, and cognitive outcomes than PWD with remote and no seizures. These findings underscore the need for timely identification and management of ongoing seizures in PWD.

Risk Factors Associated With Late-Onset Epilepsy in Dementia and Mild Cognitive Impairment

Zawar I, Quigg M, Johnson EL, Ghosal S, Manning C, Kapur J. JAMA Neurol. 2025;82(6):605–613. doi: 10.1001/jamaneurol.2025.0552. PMID: 40227723.

Importance: The risk of developing epilepsy substantially increases after the age of 60 years (late-onset epilepsy [LOE]), particularly in people with cognitive decline (PWCD, ie, dementia and/or mild cognitive impairment). Epilepsy is associated with worse cognitive and mortality outcomes in PWCD. Identifying PWCD at risk for developing LOE can facilitate early screening and treatment of epilepsy. Objective: To investigate factors associated with LOE in PWCD. Design, Setting, and Participants: This longitudinal, multicenter study is based on participants from 39 US Alzheimer's Disease Research Centers from September 2005 through December 2021. Of 44 713 participants, 25 119 PWCD were identified. Of these, 14 685 were included who did not have epilepsy at enrollment, had 2 or more visits, and were 60 years or older at the most recent follow-up. Exposure: The association between various factors and LOE development in PWCD was investigated. Main Outcomes and Measures: The primary outcome was LOE, defined as seizures starting at or after 60 years of age. Those who did not develop LOE but were 60 years or older at follow-up served as controls. A multivariable Cox regression analysis assessed the association between various factors and LOE. Independent variables included age, sex, and socioeconomic factors (education, race, ethnicity), cardiovascular risks (hypertension, diabetes, hyperlipidemia), cerebrovascular disease (stroke or history of transient ischemic attack [TIA]), other neurologic comorbidities (Parkinson disease [PD], traumatic brain injury), cognition (age at dementia onset, dementia severity, type of dementia [Alzheimer disease (AD) versus non-AD]), genetics (apolipoprotein E4 [APOE4] status), lifestyle (alcohol misuse, smoking), and depression.

Results: Of the 14 685 participants (7355 female [50%] and 7330 male [50%]; mean [SD] age, 73.8 [8.5] years) who met the inclusion criteria, 221 participants (1.5%) developed LOE during follow-up. After adjusting for demographics, cardiovascular risks, neurologic comorbidities, genetics, cognitive factors, and depression, the following were associated with a higher risk of developing LOE: APOE4 allele (adjusted hazard ratio [aHR], 1.39; 95% CI, 1.04–1.86; P = .03), dementia onset before age 60 years (aHR, 2.46; 95% CI, 1.53–3.95; P < .001), worse cognition (aHR, 2.35; 95% CI, 1.97–2.79; P < .001), AD dementia subtype (aHR, 1.68; 95% CI, 1.13–2.49; P = .01), stroke/TIA (aHR, 2.03; 95% CI, 1.37–3.01; P < .001), and PD (aHR, 2.53; 95% CI, 1.08–5.95; P = .03). In sensitivity analysis, using an alternative LOE definition of epilepsy onset after age 65 years revealed the same factors associated with LOE. Conclusion and Relevance: This study showed that the APOE4 allele, dementia onset before age 60 years, AD dementia subtype, worse cognition, stroke/TIA, and PD are associated with LOE development in PWCD. PWCD with these risk factors may be considered for routine screening with an electroencephalogram for early identification of LOE.

Commentary

With advances in healthcare and a globally aging population, the prevalence of epilepsy in older adults is expected to rise substantially. ¹ Late-onset epilepsy (LOE)—typically defined as epilepsy developing between the ages of 55 and 65—is most commonly caused by neurodegenerative disease, cerebrovascular pathology, or brain tumors.^1,2 Unlike epilepsy in younger individuals, LOE is almost always focal and presents unique diagnostic and therapeutic challenges, including atypical semiology, greater comorbidity burden, increased frailty, and elevated risks of stroke and dementia. ² It also highlights the complex and bidirectional relationship between epilepsy and dementia—a relationship that remains incompletely understood. ³

Seizures are a frequent comorbidity in people with dementia (PWD), with estimated incidence ranging from 5% to 64%. ⁴ They occur 6 to 10 times more often in older adults with dementia than in cognitively intact peers and can further accelerate cognitive decline.^3,4 However, detecting seizures in PWD is particularly challenging due to symptom overlap and baseline cognitive impairment, often resulting in underdiagnosis, treatment delays, and poorer clinical outcomes. ² While awareness of epilepsy in patients with dementia and mild cognitive impairment (MCI) is increasing, major questions remain regarding its risk factors, clinical implications, and optimal management. Two recent studies by Zawar and colleagues provide timely and valuable insights into this evolving area.^5,6

In the first study, Zawar et al ⁵ conducted a large, multicenter analysis of 26,501 individuals with dementia and recurrent seizures, using data from 39 U.S. Alzheimer's Disease Centers (2005–2021). Participants underwent detailed clinical evaluations, and outcomes were compared between those with and without active seizures—defined as seizures within the past year or requiring ongoing treatment. The results were striking: individuals with active seizures had an approximately 1.8-fold higher mortality risk than those without. Median survival was 2.4 years for those with recurrent seizures, 4.0 years for those with remote seizures, and 4.7 years for those without seizures. Furthermore, PWD with active seizures experienced more severe cognitive and functional decline, greater physical dependence, and higher rates of mood disturbances than those with remote seizures or none.

The second study ⁶ explored predictors of LOE in people with cognitive decline (PWCD), including both dementia and MCI. Using the same multicenter database, the authors followed 14,685 individuals without epilepsy at baseline, 221 of whom developed LOE during the study period. Several important risk factors were identified: early-onset cognitive decline (before age 60) conferred a 2.46-fold higher risk; Alzheimer's disease subtype was associated with a 1.68-fold increase; apolipoprotein E4 (APOE4) allele(s) with a 1.39-fold increase; history of stroke or transient ischemic attack (TIA) with a 2.03-fold increase; and Parkinson's disease with a 2.53-fold increase. These associations remained significant even when LOE was defined more stringently as onset after age 65, adding to the robustness of the findings.

Together, these studies underscore the critical association between seizures, cognitive trajectories, and mortality in individuals with cognitive decline. The first study builds upon prior literature^3,4 by showing that recurrent—but not remote—seizures are linked to significantly higher mortality in PWD. Recurrent seizures were also associated with greater morbidity across multiple health domains. These findings reinforce the importance of early seizure recognition, given its strong correlation with adverse outcomes. Although most seizures are pharmaco-responsive and fewer than 10% develop drug-resistant epilepsy, optimal care in dementia must go beyond seizure suppression. ⁷ Untreated seizures carry increased risks of stroke, faster cognitive deterioration, and death, emphasizing the need for proactive, multidisciplinary care.^1–3

The second study highlights the importance of monitoring high-risk populations—particularly those with early-onset cognitive decline, APOE4 positivity, or a history of stroke or TIA, or Parkinson's disease—for the timely detection and management of epilepsy. Clinicians should remain vigilant for seizures in these individuals, as early intervention may improve outcomes. Yet, seizure presentations in the context of dementia are often subtle—manifesting as unresponsiveness, confusion, or prolonged postictal states—contributing to under-recognition. ⁸ Additionally, routine EEG has a lower diagnostic yield in older adults. ⁹ As such, extended recordings of at least 60 min, or preferably 24 h (or longer) ambulatory EEGs, should be considered to improve detection of epileptiform abnormalities. ⁹

Effective seizure management in high-risk PWD may help slow cognitive decline and reduce mortality—hypotheses that require prospective validation. Evidence suggests that individuals with poorly controlled seizures experience faster cognitive decline, while better seizure control may positively influence cognition.^1–4 However, managing seizures in this population is complex. Clinicians must weigh seizure control against the potential cognitive side effects of antiseizure medications (ASMs), particularly in individuals with multiple comorbidities. Polypharmacy, drug interactions, and frailty further complicate treatment decisions. ¹⁰ The interplay between epilepsy, ASMs, and neurodegeneration remains poorly understood, and whether ASMs modify the progression of dementia is a critical, unanswered question.

While both studies leverage large, multicenter, longitudinal datasets, several limitations must be acknowledged. Their observational design limits the ability to establish causality. Seizure diagnoses were based on clinical assessments rather than EEG confirmation, raising the potential for misclassification. Notably, clinically silent or nocturnal-only seizures—often undetected without targeted investigation—were not addressed.^8,11 Additionally, important clinical variables such as seizure type, frequency, duration, and ASM use were not fully explored, leaving treatment effects uncertain. The definition of “active seizures” in the first study was broad, encompassing both individuals with ongoing seizures and those on ASM therapy, potentially conflating controlled and uncontrolled cases. Finally, incomplete documentation in the mortality analysis limited the ability to disentangle whether death was due to seizures, underlying disease, comorbidities, or treatment effects. Future research should aim to overcome these limitations through prospective study designs, objective diagnostics (eg, EEG), and detailed analyses of treatment responses and outcomes.

These limitations withstanding, the twin studies by Zawar and colleagues emphasize 2 key findings: (1) recurrent seizures are associated with earlier mortality and greater morbidity, and (2) several identifiable risk factors significantly increase the likelihood of LOE in PWCD. These insights should guide clinicians toward implementing targeted screening strategies, reinforcing the importance of timely seizure recognition, and supporting the development of prospective and mechanistic studies. Integrating seizure assessment into the routine care of patients with cognitive decline is a crucial step—before it is too late—with the potential to reduce the global burden of dementia.