Cognitive Decline Risk Stratification in People with Late-Onset Epilepsy of Unknown Etiology: an Electroencephalographic Connectivity and Graph Theory Pilot Study.

Background: Although people with late onset epilepsy of unknown etiology (LOEU) are at higher risk of cognitive decline compared to the general population, we still lack affordable tools to predict and stratify their risk of dementia. Objective: This pilot-study investigates the potential application of electroencephalography (EEG) network small-world (SW) properties in predicting cognitive decline among patients with LOEU. Methods: People diagnosed with LOEU and normal cognitive examination at the time of epilepsy diagnosis were included. Cerebrospinal fluid biomarkers, brain imaging, and neuropsychological assessment were performed at the time of epilepsy diagnosis. Baseline EEG was analyzed for SW properties. Patients were followed-up over time with neuropsychological testing to define the trajectory of cognitive decline. Results: Over 5.1 years of follow-up, among 24 patients diagnosed with LOEU, 62.5% were female, mean age was 65.3 years, thirteen developed mild cognitive impairment (MCI), and four developed dementia. Patients with LOEU developing MCI had lower values of SW coefficients in the delta (p = 0.03) band and higher SW values in the alpha frequency bands (p = 0.02) compared to patients having normal cognition at last follow-up. The two separate ANOVAs, for low and alpha bands, confirmed an interaction between SW and cognitive decline at follow-up. A similar gradient was confirmed for patients developing dementia compared to those with normal cognitive function as well as to those developing MCI. Conclusion: Baseline EEG analysis through SW is worth investigating as an affordable, widely available tool to stratify LOEU patients for their risk of cognitive decline.