TL;DR
This study reveals that specific EEG patterns, particularly those indicating temporal lobe hyperexcitability, are significantly associated with cognitive impairment in individuals with late-onset epilepsy (LOE). These findings suggest that scalp EEG could serve as a valuable, non-invasive tool for identifying LOE patients at higher risk of cognitive decline, potentially pointing towards neurodegenerative origins in some cases.
What Problem Does This Paper Address?
Late-onset epilepsy (LOE), defined as epilepsy starting at age 55 or later, is increasingly recognized as a distinct subtype with a strong link to cognitive impairment and dementia. This relationship appears to be bidirectional: LOE increases the risk of cognitive decline, and conditions like Alzheimer's disease can lead to seizures and epileptiform activity. While scalp EEG is a well-established tool for assessing brain function, its specific utility in understanding and predicting cognitive impairment in LOE has not been thoroughly explored. The challenge lies in identifying reliable, non-invasive biomarkers that can help stratify LOE patients based on their risk of cognitive decline, especially given the potential for early intervention.
Methods
The researchers conducted a cross-sectional study involving individuals who underwent 24-hour video-EEG (VEEG) monitoring. They included patients diagnosed with LOE and age- and gender-matched healthy controls. Cognitive impairment was assessed using standardized neuropsychological tests (MMSE, MoCA).
Key methodological aspects include:
- EEG Analysis: VEEGs were independently reviewed by two neurophysiologists. EEG features were extracted based on the standardized "background activity" and "interictal findings" categories. Specific attention was paid to focal slowing and interictal epileptiform discharges (IEDs), particularly their localization (temporal vs. extratemporal, unilateral vs. bilateral). Temporal intermittent rhythmic delta activity (TIRDA) was also analyzed.
- Statistical Modeling:
- Regression Analysis: Ordinal and binomial logistic regression models were used to identify clinical and EEG risk factors associated with cognitive impairment.
- LASSO Regression: The Least Absolute Shrinkage and Selection Operator (LASSO) regression was employed to select the most discriminative EEG features for predicting cognitive impairment, particularly in the LOE of unknown etiology (LOEU) subgroup. This method helps to identify a parsimonious set of predictors.
- Covariate Control: The analysis controlled for various demographic, clinical, and neuroimaging factors, including age, sex, years of education, epilepsy duration, seizure frequency, medication use, and cerebral small-vessel disease (CSVD) signs on MRI.
Datasets
The study utilized data from a single-center cohort at West China Hospital.
- Screened Individuals: 8318 individuals had their VEEGs reviewed.
- Included Participants:
- Late-Onset Epilepsy (LOE): 287 participants.
- Healthy Controls (HC): 132 participants.
- Subgroups:
- LOE of Unknown Etiology (LOEU): 177 participants.
- LOE with Definite Etiology (LOED): 110 participants.
- LOE with Cognitive Impairment (LOE-CI) and LOE with Normal Cognition (LOE-CN) were derived from the LOE cohort.
No new benchmarks were introduced; the study relied on existing diagnostic criteria and neuropsychological assessments.
Key Insights and Ideas
- EEG as a Cognitive Predictor: The study strongly suggests that specific EEG patterns are not just indicators of epilepsy but also valuable predictors of cognitive impairment in LOE. This moves beyond simply identifying seizures to understanding the underlying neural network disruptions.
- Temporal Lobe Hyperexcitability is Key: The findings consistently point to temporal lobe abnormalities, particularly anterior temporal epileptiform discharges and TIRDA, as being significantly associated with cognitive impairment. This highlights the critical role of the temporal lobe in both epilepsy and cognition.
- Bilateral Involvement Matters: Bilateral anterior temporal IEDs emerged as a particularly strong predictor, especially in LOE of unknown etiology. This suggests that widespread disruption in the temporal lobes may be more indicative of cognitive decline.
- Potential Neurodegenerative Link: The observation that LOE of unknown etiology, particularly with bilateral temporal IEDs, shows patterns reminiscent of Alzheimer's disease raises the intriguing possibility that some LOE cases might have a shared neurodegenerative origin.
- Scalp EEG's Practical Value: The use of long-term scalp EEG, a non-invasive and accessible tool, for identifying these risk factors makes this research highly relevant for clinical practice.
Key Results
- Prevalence of Impairment: 48% of LOE participants were cognitively impaired.
- EEG Abnormalities: 78% of LOE participants and 33% of controls showed abnormal EEGs.
- Slowing: Focal slowing was present in 56% of LOE participants, primarily in the temporal regions. Generalized slowing was found in 13%.
- Interictal Epileptiform Discharges (IEDs): 56% of LOE participants had IEDs, predominantly in the temporal lobe.
- TIRDA: Temporal intermittent rhythmic delta activity (TIRDA) was present in 28% of LOE participants.
- Discriminative Power:
- Bilateral anterior temporal IEDs were the most discriminative indicator of cognitive impairment in LOEU (OR = 53.280, 95% CI 11.359–249.917).
- A LASSO model combining age, overall abnormal EEG, generalized delta slowing, and anterior temporal IEDs achieved an Area Under the Receiver Operating Characteristic Curve (AUROC) of 0.869 for predicting cognitive impairment in LOEU.
- Specific Associations:
- In LOEU, anterior temporal IEDs (especially bilateral) and TIRDA were significantly associated with cognitive impairment.
- In LOED, TIRDA was associated with cognitive impairment.
Limitations and Future Work
The authors acknowledge several limitations:
- Underrepresentation of Structural LOE: The cohort had fewer participants with structural LOE, particularly cerebrovascular causes, which might limit generalizability.
- Single-Center Cohort: The study was conducted at a single center, potentially introducing selection bias.
- Cross-Sectional Design: The study design prevents establishing causality between EEG features and cognitive decline; it cannot determine the sequence of events.
- Qualitative EEG Analysis: While standardized, the EEG analysis was largely qualitative. Future work could benefit from quantitative EEG metrics.
Future Directions:
- Longitudinal Studies: The authors strongly recommend future longitudinal studies to explore the temporal relationship between EEG features, LOE, and dementia progression.
- Quantitative EEG: Further investigation using quantitative EEG analysis could provide more precise insights.
- Multimodal Data Integration: Combining EEG data with other modalities (e.g., neuroimaging, biomarkers) could lead to more accurate prediction and discrimination models.
- Expanded Sample Sizes: Larger, multi-center studies are needed to confirm these findings and improve generalizability.
Summary updated