Hearing in Our Sleep—Can We Drown Out Epileptic Spikes and Boost Cognition?

Commentary

Sleep is intertwined with outcomes in epilepsy. Too little sleep can lower threshold for seizure occurrence, and sleep that is disordered due to a high burden of interictal epileptic activity can disrupt cognition and behavior. ¹ Sleep disorders are also common epilepsy comorbidities, leading to increased risk of other adverse effects on health and well-being. ² Thus, approaches that improve sleep quality are particularly relevant to patients with epilepsy. Wong et al tackle this challenge by developing and testing a closed-loop protocol aimed at modulating slow-wave sleep in children with refractory epilepsy. ³

They leverage our increasing understanding of the fundamental corticothalamic oscillatory components of nonrapid eye movement (NREM) sleep that contribute to its function. One key oscillation is the slow oscillation (SO, 0.5-4 Hz), the phase of which either facilitates the occurrence of faster oscillatory frequencies and neural spiking (the “UP” state) or inhibits them (the “DOWN” state). ⁴ Though cellular mechanisms remain elusive, evidence suggests that bolstering the SO could improve sleep quality, leading to downstream positive effects on health and cognition. ⁵ The authors artificially tapped into the SO using an age-old EEG observation that nonarousing auditory stimuli can induce a K-complex that is thought to be a variant of the spontaneous SO (sharing at least some thalamocortical mechanisms) and potentially functioning to deepen sleep. ⁶ They monitored the SO in real-time from EEG electrodes and responsively delivered auditory stimulation to specific phases of the SO, a method known as closed-loop phase-targeted auditory stimulation (PTAS). Their goal was to characterize the response of epileptic networks to PTAS and evaluate for the effects on cognitive function.

Critically, they adapted PTAS to be robustly performed in a challenging clinical scenario: children with refractory epilepsy implanted with iEEG electrodes in the epilepsy monitoring unit. This experimental set-up can confer major advantages, including access to concomitant continuous iEEG monitoring and capacity for conducting a fully supervised, multiday protocol. It also comes with potential difficulties, as sleep is arguably perturbed in an intensive hospital environment after a neurosurgical procedure and patients are experiencing variable anticonvulsant medication doses and seizure occurrence that could alter responses to stimulation and cognitive performance. The authors took these factors into account, employing a randomized, counter-balanced block design to try to minimize heterogeneity of clinical variables over the course of the study. This patient population also has a relatively high burden of interictal epileptiform discharges (IEDs), necessitating a sophisticated computational method to prevent contamination of SO detection by the high broadband power associated with IEDs. ⁷

They first applied randomly timed PTAS to a Discovery cohort of children, exploring the network effects of the auditory stimulation. Using the high spatiotemporal resolution of iEEG compared to scalp EEG, they were able to track responses to auditory stimulation from auditory cortex to prefrontal and cingulate cortices, additionally discovering involvement of the central thalamus. They found that network response to the auditory stimulation was highly dependent on the phase of the spontaneous SO at which it was delivered. PTAS administered during the cortical “UP” state (UP-PTAS) resulted in higher resultant SO power across a broad cortical network spanning frontal, parietal, and temporal regions, with SOs closely resembling spontaneously occurring SOs. In contrast, PTAS administered during the cortical “DOWN” state decreased SO power and disrupted typical SO appearance. These results highlight that the outcome of perturbations applied to neural networks can be highly sensitive to ongoing activity patterns, such that the same stimulation applied at a marginally different time could have a different functional outcome. They further support the emerging notion that closed-loop intervention approaches armed with appropriate biomarkers could improve the efficacy and consistency of neuromodulation compared to open-loop approaches that are agnostic to brain state. ⁸ However, in-depth understanding of such biomarkers is important. For instance, some evidence suggests that there could actually be 2 types of slow waves that mediate different sleep functions. ⁹ These oscillatory patterns cannot be differentiated with the authors’ current detection approach, raising opportunities for even more precisely targeted interventions as mechanistic knowledge improves.

Next, the authors used a randomized cross-over study design to investigate the effects of UP-PTAS compared to no stimulation in a separate Experimental cohort of children. They determined that blocks of UP-PTAS were associated with increased physiologic-appearing SO amplitude, significantly decreased rate of IEDs, and improved performance on a sleep-related response inhibition task the following day. This result is noteworthy because it provides support for the utility of a noninvasive intervention during sleep to enhance physiologic oscillations and suppress epileptic activity, increasing the brain's cognitive capacity during subsequent wakefulness.

The specific mechanisms underlying these outcomes, however, remain unclear. The authors posit that UP-PTAS may engage physiologic thalamocortical activity patterns that are broadly inhibitory and/or protective against IED generation. PTAS can evoke thalamocortical sleep spindles, another oscillation implicated in mediating essential functions of NREM sleep. ¹⁰ Because IEDs exhibit a complex relationship with both the SO and spindles,^11,12 more detailed investigation of the interplay between these phenomena could shed light on how UP-PTAS suppresses IEDs. Which neurophysiologic parameter(s) are critical for driving the cognitive outcome are similarly uncertain. High rates of IEDs during sleep in children are correlated with impaired cognitive function and risk of abnormal neurodevelopment, suggesting that the UP-PTAS-mediated decrease in IEDs could benefit cognition. But UP-PTAS also stabilizes physiologic SO patterns, providing an alternative pathway for functional improvement. A key related question is how generalizable the cognitive benefits are; would tasks requiring memory or other executive functions be facilitated as well?

Cognitive comorbidities of epilepsy are prominent in children and poorly addressed by current treatments. This study establishes PTAS as a potential novel therapeutic avenue to boost cognitive function in children with refractory epilepsy. The protocol could be accomplished in a noninvasive manner, enhancing translational potential, though implementation in a home environment for long-term treatment could pose technical hurdles. Advances in neuromodulation devices that are self-administered, such as noninvasive vagal nerve stimulation, provide evidence that such technical issues could be overcome. Additionally, any form of neuromodulation raises the possibility of chronic, neuroplasticity-related effects on neural networks that may diverge from acute effects, especially in the developing brain. Whether PTAS can harmonize with physiologic sleep and quiet the epileptic network in the long-term remains to be seen (or heard).