Peter Nogueira Hadar ¹ Mike Westmeijer ^2,3 Haoqi Sun ³ Erik-Jan Meulenbrugge ^3* Jin Jing ³ Luis Paixao ^3,4 Ryan Tesh ³ Madalena Da Silva Cardoso ⁵ Pierrick Arnal ^6* Rhoda Au ⁷ Chol Shin ^8,9 Soriul Kim ⁸ Robert J Thomas ¹⁰ Sydney S Cash ¹ M. Brandon Westover ^3*

• ¹ Epilepsy Service, Department of Neurology, Massachusetts General Hospital, Boston, United States
• ² Utrecht University, Utrecht, Netherlands, Netherlands
• ³ Department of Neurology, Beth Israel Deaconess Medical Center, Harvard Medical School, Boston, Massachusetts, United States
• ⁴ Department of Neurology, University of Miami Health System, Miami, Florida, United States
• ⁵ Department of Radiology, NYU-Langone Medical Center, New York, United States
• ⁶ Dreem, Paris, France
• ⁷ Department of Epidemiology, Boston University School of Medicine, Boston, United States
• ⁸ Institute of Human Genomic Study, College of Medicine, Korea University, Seoul, Republic of Korea
• ⁹ Biomedical Research Center, Korea University Ansan Hospital, Ansan, Republic of Korea
• ¹⁰ Department of Medicine, Division of Pulmonary, Critical Care & Sleep, Beth Israel Deaconess Medical Center, Boston, United States

Although seizures are the cardinal problem, epilepsy is associated with other forms of brain dysfunction including impaired cognition, abnormal sleep, and increased risk of developing dementia. We hypothesized that, given widespread neurologic dysfunction in epilepsy, accelerated brain aging would be seen. We set out to measure the sleep-based Brain Age Index (BAI) in a diverse group of patients with epilepsy. BAI is a machine learning-based biomarker that measures how much the brain activity of a person during overnight sleep deviates from chronological age-based norms.This case-control study drew age-matched controls without epilepsy from home sleep monitoring volunteers and from non-epilepsy patients with sleep lab testing. Patients with epilepsy underwent in-patient monitoring and were classified by epilepsy type and seizure burden. The primary outcomes measured were BAI, processed from electroencephalograms, and epilepsy severity metrics (years with epilepsy, seizure frequency standardized by year, and seizure burden [number of seizures in life]). Subanalyses were conducted on a subset with NIH Toolbox cognitive testing for total, fluid, and crystallized composite cognition. Results: 138 patients with epilepsy (32 exclusively focal, 106 generalizable [focal seizures with secondary generalization]) underwent in-patient monitoring and age-matched, non-epilepsy controls were analyzed. Mean BAI was higher in epilepsy patients vs. controls and differed by epilepsy type: -0.05 years (controls) versus 5.02 years (all epilepsy, p<0.001), 5.53 years (generalizable, p<0.001), and 3.34 years (focal, p=0.03). Sleep architecture was disrupted in epilepsy, especially in generalizable epilepsy. Higher BAI was positively associated with increased lifetime seizure burden in focal and generalizable epilepsies, and associated with lower crystallized cognition. Lifetime seizure burden inversely correlated with fluid, crystallized, and composite cognition. Significance: Epilepsy is associated with accelerated brain aging. Higher brain age indices are associated with poorer cognition and more severe epilepsy, specifically generalizability and higher seizure burden. These findings strengthen the use of the sleep-derived, electroencephalography-based BAI as a biomarker for cognitive dysfunction in epilepsy.