Tailoring Human Sleep: selective alteration through Brainstem Arousal Circuit Stimulation

Abstract

Brainstem nuclei, such as the pedunculopontine nucleus, send activating projections to cortex, modulating states of sleep, wakefulness and arousal levels. Surgical modulation of subcortical activity using deep brain stimulation (DBS) is utilised in the management of pain and movement disorders. DBS of brainstem arousal circuits in a state-dependent manner could offer an attractive alternative in severe pharmacoresistant cases of hypersomnia and in disorders of consciousness, where behavioural activation is desired. We wanted to investigate if we can selectively induce wakefulness and/or alter sleep state through DBS of the PPN region (PPNR). To this end, we used the opportunity of implanted PPNR electrodes in stimulation-naive patients with multiple systems atrophy. PPNR activity was recorded during both slow wave sleep (SWS) and quiet wakefulness with simultaneous cortical EEG, in order to identify differences in brainstem oscillatory patterns during different states of excitability. PPNR DBS in two gamma frequency protocols (40Hz and 100Hz) was delivered during SWS of the same sleep stage and with comparable pre-trial levels of slow wave activity. Additionally, SHAM trials were used as a control where no stimulation was applied. We examined changes in cortical oscillatory power, changes in functional connectivity (coherence and causality) from pre- to post- stimulation and phase-locking of cortical oscillations with DBS frequencies and their sub-harmonics during stimulation. We also evaluated connectivity changes induced by DBS and corresponding differences in circuit dynamics between SWS and wakefulness. Beta and gamma PPNR oscillatory power increased when wake was compared to sleep. We saw clear PPNR power modulation by the phase of EEG slow wave, with significant increase in gamma compared to beta power during the excitable part of the slow-wave cycle. Gamma PPNR DBS induced transitions to wakefulness and REM, while it truncated sleep time compared to baseline. The 40Hz stimulation protocol was more efficient in reducing slow wave activity and increasing cortical beta power, compared to PPNR DBS at 100Hz. Furthermore, intrinsic cortical rhythms phase-locked with 40 Hz PPNR DBS to a significantly higher degree compared to the 100Hz protocol, with regional differences in phase-locking, suggesting a complex biological phenomenon. Finally, functional connectivity changes induced by PPNR DBS were consistent with differences in circuit dynamics between SWS and wakefulness. Overall, these results highlight the possibility of using DBS of brainstem arousal circuits to promote arousal and wakefulness, which opens new perspectives for using closed-loop approaches to modulate vigilance states in humans for therapeutic benefit.

Bioelectronic Zeitgebers: targeted neuromodulation to re-establish circadian rhythms

“…Preferentially active during wakefulness and REM sleep, its reported firing rates lie within the beta/gamma range (20-60Hz) with current clinical stimulation settings at 40 Hz often said to reflect a physiologic wake frequency [17], [18]. In keeping with its important role in sleep/wake activity, findings from our group have also shown that slow-wave sleep can be affected by PPN DBS timing [19]. The PPN additionally has long-range connections with areas important for gait and locomotion (notably basal ganglia) [20].…”

“…Along the lines of our described bioelectronic Zeitgeber strategy, we have preliminary proof of broad cortical circuit engagement during daytime PPN gamma stimulation from both patients. Acute effects of daytime parameters on the power spectra involve non-linear acceleration of peak beta frequency (13)(14)(15)(16)(17)(18)(19)(20)(21)(22)(23)(24)(25)(26)(27)(28)(29)(30) in response to stimulation, when compared to an off-stimulation state within each testing session in both patients (Fig. 3A).…”

The Design of Brainstem Interfaces: Characterisation of Physiological Artefacts and Implications for Closed-loop Algorithms

“…The introduction of closed-loop control policies presents significant potential for the refinement of neuromodulation; however, it relies on the accurate sensing of µV-scale physiological signals [21]. While the brainstem appears to be an attractive therapeutic target for disorders of consciousness [4], the presence of physiological artefacts poses a challenge to closed-loop algorithm development. Although the cardiac artefacts noted in this study could be mitigated through either software (e.g.…”

“…LFPs were further band-passed to artefact FOI, with power spectra visualised after a short-time Fourier transform (STFT, with a 5 s time window and 3 s overlap), then averaged across the FOI spectrum and compared across different lead contact dipoles. For constant cardiac artefacts, signal power within the artefact FOI was compared to power within the beta (15-30 Hz) and gamma (35-50 Hz) bands, as based on prior reports of human recordings these correspond to physiological PPN activity [4], [27]. Power within these activity bands was compared to pooled trials from both patients, for motion artefact-free periods averaged for the same dipole, using both pairwise t-tests with Bonferroni correction as well as a two-way ANOVA (with FOI and patient code as subgroup analyses) for pooled data.…”

“…Closed-loop designs are susceptible to artefact sensitivity however, which can be particularly prominent in areas where there is a physiological tendency for increased mobility. One such area with increased mobility due to its anatomical characteristics is the brainstem, a brain region with promising DBS targets for treatment of movement disorders, chronic pain, disorders of consciousness and sleep [3], [4].…”