Sleep AD: exploring the feasibility of a new approach to sleep research using at-home EEG remotely with people with prodromal and mild Alzheimer’s

Introduction: Modifying sleep quality is an untapped opportunity with potential to delay progression of Alzheimer’s disease (AD). Whilst in-laboratory polysomnography remains the gold-standard, the approach is expensive, burdensome and fails to capture a typical night’s sleep. The future of sleep trials is home-monitoring with commercially-available electroencephalography (EEG) devices. We aim to identify the optimal approach balancing reliability, comfort, efficiency and accuracy, for AD sleep intervention trials. We previously demonstrated that EEG headbands can be utilised remotely with this cohort, and now seek greater spatial resolution. Sleep AD assessed the feasibility, acceptability and sensitivity of using 32-channel EEG sleep caps remotely. We hypothesised that this would yield behaviourally-relevant sleep neurophysiology sufficient for clinical trials.

Materials and methods: We used a single-centre prospective observational feasibility study design. Participants with Mild Cognitive Impairment or mild AD completed baseline sleep and mood questionnaires, and provided a blood sample for AD biomarker analysis. Recordings were conducted over 3 consecutive nights; researchers attended evenings and mornings to set-up and test/reset equipment. Participants completed a Sleep Consensus Diary and feedback questionnaire. We used Brain Products actiCAP slim sleep caps with 32 active electrodes, and a matching wireless EEG LiveAmp amplifier and battery pack, worn in harness. Participants completed an overnight memory task on the final evening, and had the option to wear a wrist pulse oximeter for nights 2-3.

Results: Of 62 screened, 31 were invited, and 11 recruited. Recruitment stopped early (our original target was n=50) when interim analysis proved conclusive. Participants had a mean (SD) age of 71.3 (9.1), 4/11 were female, mean (SD) Montreal Cognitive Assessment score was 18.9 (6.0). One participant withdrew after completing baseline measures. Two completed <3 nights due to kit discomfort. Sleep data was recorded from 10 participants over a total of 28 nights. We achieved optimal recordings on 11/28 (39%) nights; sub-optimal full recordings on 5/28 (17.9%) nights; and partial/no data on 12/28 (42.9%) nights. 91% of optimal recordings were achieved on nights 2-3. On night 1, 90% of participants reported disrupted sleep, primarily due to discomfort. On nights 2-3, sleep ratings improved but mild discomfort/sleep disruption were frequently reported. Overnight memory testing appeared feasible and well-tolerated. Experiences of wearing a pulse oximeter were mixed. Participants reported satisfaction with taking part and liked the home-monitoring approach.

Conclusions: Our approach to use 32-channel sleep caps remotely yielded just 39% of recordings with data sufficiently sensitive for micro-architectural analysis. Data completeness was below expectations and insufficient for intervention trials. Despite iterative set-up improvements, ground/reference signals were frequently lost due to cap movement or adherence loss; real-time adjustments were not possible remotely. The 3 consecutive night design aimed to provide an adjustment night prior to more typical sleep on subsequent nights. Whilst partially successful, the scheduling was onerous for researchers. Despite withdrawals, feedback was unanimous that the concept of home-monitoring was appealing. We consider the approach of EEG home-monitoring with this cohort worth repeating, if we can develop a reliable and comfortable set-up with an alternative device.

Acknowledgments: Many thanks to our participants and funder Rosetrees Trust.