Candidate EEG-derived biomarkers of epileptiform discharge occurrence in epilepsy

Objective. In epilepsy, neuronal excitability is a key factor underlying the occurrence of epileptiform discharges (EDs). EEG-derived biomarkers that reflect ED occurrence may therefore provide objective measures of cortical excitability. Here, we explore the relationship between three candidate EEG biomarkers and the occurrence of EDs to assess their potential use in epilepsy treatment monitoring.

Methods. We obtained ultra-long-term two-channel EEG recordings from patients with pharmacoresistant epilepsy participating in our ongoing PREDYct study using a subcutaneous electrode. 14–16 d of data per patient were visually assessed and EDs, including interictal EDs and ictal patterns, were annotated. We evaluated three approaches to estimate potential biomarkers for the ED occurrence: (1) parameters from a corticothalamic mean-field model involving four neuronal populations, where model parameters were individually fitted to each patient’s EEG power spectra; (2) the spectral slope of the aperiodic part of the EEG power spectrum as a proxy for the excitation-inhibition balance; (3) the wave area/spike amplitude (WA/SA) from annotated EDs. Candidate biomarkers were assessed for their association with ED occurrence using negative binomial regression, provided that at least 15 ED events were annotated. Analyses were conducted at both the group and individual levels.

Results. We included 10 patients. Eight had ⩾ 15 EDs, and seven were analyzable for spectral slope. Mean-field model gains G_ee, G_ese and G_esre were significantly related to ED occurrence at group level and in at least half of the individual patients. The spectral slope exhibited a negative association with ED occurrence at the group level and across all seven individual patients. WA/SA was not significantly related to ED occurrence.

Significance. We identified two candidate EEG-derived biomarkers that exhibited significant associations with the ED occurrence. These findings indicate their potential as biomarkers of epileptiform activity and their utility in guiding and monitoring epilepsy treatment.