ABSTRACT
A fundamental knowledge gap in epilepsy neuroscience concerns the varying propensity for seizures over the
24h circadian cycle. For example, it is not known why some patients with frontal lobe epilepsy may only seize
while asleep. Understanding the dynamics of epilepsy networks on circadian time scales is essential for
improving therapeutic prospects of the substantial fraction of epilepsy patients who fail all treatments. Current
network architectures of epilepsy are based on structural MRI or resting state (rs) fMRI. These modalities
reveal single experimental time points at fixed time scales and do not address the spatiotemporally dynamic
nature of seizure networks. Reports of seizure periodicity in chronic intracranial recordings do not sample the
whole epileptic network and only document seizure occurrence, not their causative network alterations. Our
long-term goal is to understand network dynamics in epilepsy to advance therapies. Our objective here, using
the intracerebral stereo-electroencephalographic (SEEG) signal, is to build a dynamic neurophysiological
SEEG-based connectome of the frontotemporal brain regions over the circadian cycle. Our central hypothesis
is that the topology of epileptic networks has specific circadian dependence, and that such dependence can
be modulated on longer time scales, including by anticonvulsant drugs. Our rationale for this project is that
knowledge of the network pathways, bandwidths and circadian state-dependence of epileptic networks will
inspire new neuromodulatory approaches to epilepsy (targeting brain regions in specific frequency bands and
their 24h cycles). Such insight may also drive new network-inspired ablative surgical approaches. We will
pursue two specific aims: (i) determine the SEEG-based connectomics of frontotemporal cortex across
circadian vigilance states; and (ii) Identify the infradian characteristics of epilepsy network dynamics in
frontotemporal cortex. Working with continuous multi-day SEEG recordings from patients at our clinical facility,
we will pursue these aims in parallel. We will organize the data by patient vigilance state, and using analytic
tools deployed in prior work, we will describe epileptiform frontotemporal cortical networks, and their
interaction at multiple time scales and with reference to the 24h and infradian cycles. We will identify key
network vulnerabilities locked to the circadian cycle and validate our results with comparisons with ictal onset
areas and the spatial distribution of metrics such as epileptogenicity index. Our proposal is innovative,
because we will move beyond the static nature of imaging-based connectomics to add the dimension of time
to descriptions of brain network architecture. Our contribution will be significant, by helping solve a scientific
riddle in epilepsy neuroscience while suggesting potential new treatments for refractory epilepsy. More
generally, our work will inform the ‘building brain maps’, ‘observing the brain in action’, and ‘advancing human
neuroscience’ priority areas of the NIH BRAIN initiative.