Abstracts

Rationale: In focal epilepsy, various seizure features, such as seizure severity, seizure spread, and seizure duration, are known to change from one seizure to the next within the same patient. Recently, we demonstrated that the evolution of seizure spatiotemporal electroencephalographic (EEG) features also changes over time. Crucially, these changes were not random, but were instead consistent with fluctuations over circadian or longer timescales. Intriguingly, recent discoveries also suggest that circadian and multiday fluctuations, which can be extracted from interictal biomarkers such as spike rate, may dictate seizure likelihood. Thus, we hypothesise that interictal spike rate fluctuations over specific timescales may also reveal how seizure evolution is modulated within individual patients.

Methods: We compared variability in seizure evolutions in 10 patients with chronic intracranial EEG recordings from the NeuroVista seizure prediction study (185-767 days of recording time, 57-452 analysed seizures/patient). We described seizure evolutions as sequences of a finite number of functional network states, which were derived in a data-driven manner. These “seizure states” represent different spatiotemporal EEG features during the seizure. Using this framework, we can quantify (1) the presence or absence of each seizure state, and (2) the duration of a state if it is present in a seizure.

We additionally extracted fluctuations in interictal spike rate over different timescales (daily, multiday, multiweek, multimonth, or slower) using empirical mode decomposition. Almost every patient had a daily spike rate cycle, longer (multiday to multimonth) spike rate cycle(s), and a slower non-cyclical trend in spike rate. To uncover if these fluctuations were associated with changes in seizure evolutions, we compared (1) seizure state occurrences and (2) seizure state durations to each patient’s spike rate fluctuations.

Results: All patients had at least one seizure state that was associated with a spike rate fluctuation, either in terms of the state’s occurrence or duration. Across patients, (1) seizure state occurrence was associated with slow trends in spike rate more often than it was associated with spike rate cycles, and (2) seizure state durations were frequently associated with both slow trends in spike rate and spike rate cycles. A given seizure state’s occurrence and duration were usually not associated with the same timescale of fluctuation in spike rate.

Conclusions: Our results suggest that time-varying factors modulate intra-patient seizure evolutions over circadian and slower timescales. The different associated timescales of seizure state occurrence and seizure state duration indicate that these features are often modulated by separate time-varying factors. These findings provide new insight into the patterns and mechanisms of intra-patient seizure variability, with potential implications for forecasting and treating seizures.

Funding: Please list any funding that was received in support of this abstract.: This project received funding from the Australian Government National Health and Medical Research Council. Y.W. is supported by the Wellcome Trust (208940/Z/17/Z) and P.N.T. is supported by a UKRI Future Leaders Fellowship (MR/T04294X/1).