Selecting patients for sleep surgery using novel physiological and surgical phenotyping

Up to 50 prospective candidates for multi-level sleep surgery will be recruited and all will complete the following two phases: Phase 1 - all participants. Duration: two individual overnight sleep studies + 14 nights of in-home use of mattress device. Two Sleep Studies: Prospective detailed physiological phenotyping investigations will be performed on each patient who consents by the sleep researchers: - Airway collapsibility: patients will be nasally intubated with an oesophageal balloon catheter as well as epiglottic and choanal catheters for detailed evaluations of breathing effort and airway collapsibility and function. Wake upper airway collapsibility will be determined during a ~15-minute wakefulness test (7) using 60 brief negative pressure pulses delivered through a continuous positive airway pressure (CPAP) mask with epiglottic and choanal pressure recording catheters in place. After this procedure the choanal catheter (only required for airway collapsibility assessments) will be removed. - Upper airway force / fatigue: Using a customised force transducer device to measure the maximal tongue strength (Pmax), participants will be asked to push their tongues against the force transducer as hard as possible. Fatigability will be measured by asking the participants to maintain ~50% of their maximum force for as long as possible. - Airflow analysis: Participants will wear a nasal mask fitted with a calibrated airflow sensor. Polysomnography (PSG) nasal cannula pressure, along with epiglottic and oesophageal pressure recordings analysis will be assessed to identify breathing effort and work of breathing, including effective versus ineffective snoring and motion components using our established analytical techniques. The shape of inspiratory and expiratory flow versus time curves of each breath throughout the PSG will also be evaluated to classify the most likely site of airway collapse including epiglottic, retro-lingual and palatal collapse of each breath exhibiting flow limitation - Acoustic analysis: During the overnight sleep studies, high quality respiratory motion (piezo-electric mattress sensor) and microphone-based acoustic recordings will be obtained time-synchronised to the PSG recording via a common timing signal channel on both systems. Acoustic data will be examined in relation to breathing phase (inspiration versus expiration), sleep stages, apnoeic and hypopnoeic events and as a function of airflow shape assessed site of airway collapse. Machine-learning algorithms will be developed to analyse clusters and identify acoustic features related to palate, lateral wall, retrolingual and epiglottic collapse as assessed from both airflow shape and in drug-induced sedation endoscopy (DISE). - Electromyography (EMG) recording (optional): Using small EMG needles placed into the genioglossus, the activity of the primary airway dilator muscle will also be assessed during PSG for classifying airway muscle responsiveness to airway collapse based on EMG versus epiglottic pressure - Obstructive sleep apnoea (OSA) severity: The piezo-electric sensor strip and acoustic recording system will be used during the overnight sleep study and for 14 days in the participants' homes to record multiple nights of detailed sleep, breathing and snoring data including sleep position, to improve the accuracy of sleep apnoea severity and its changes over time. Participants will be asked to complete a 14-day device use diary during this time. Phase 2 - All participants. One-off DISE, ~30-45 minute procedure. “Pheno-DISE”: Drug induced sedation endoscopy (DISE) will be performed utilising several advanced physiological assessments and additional monitoring to replicate PSG recordings. Extended DISE will allow for multiple assessments of airway collapsibility, function, sites and severity of airway collapse at a range of anaesthetic depths. Treating clinician and anaesthetist will conduct the DISE and administer the anaesthetic (intravenous propofol titrated to individual patient responses in terms of BIS (aiming for 40-60) until partial and/or complete airway collapse is observed nasendoscopically). Sleep researchers will manage the mattress device and collect data from the respiratory and acoustic measures: a) participants will wear a nasal mask fitted with a calibrated airflow sensor, and their mouth with be taped to ensure nasal breathing throughout the procedure. b) Electroencephalography (EEG) monitoring: EEG in addition to Bispectral Index Monitoring (BIS) will be used to classify sleep depth. c) Respiratory motion and acoustic analysis: Detailed acoustic feature analysis will be used to compare PSG to DISE audio recordings demonstrating similar airflow features (recognising a less-favourable acoustic environment in the operating theatre) towards DISE confirmation of PSG assessed sites of airway collapse. d) Airway collapsibility: Airway collapsibility will be determined using negative pressure pulses during DISE for direct comparison against awake assessments. In addition a Pcrit (critical pressure) machine will be used to briefly (3-6 breaths) drop airway pressures from a stable holding pressure that maintains normal airflow, to pressures that induce mild, moderate and severe airway obstruction. e) Airway collapse will be directly visualized via a nasendoscopy scope passed through a sealable cable gland port built into the nasal mask. Video recordings of the airway will also be obtained and time-synchronised with the physiology recording systems. f) OSA severity: The piezo-electric sensor strip and acoustic recording system will also be used during the DISE procedure (and the overnight sleep studies and by the participants at home during phases 1 and 2). Pheno-DISE classification: based on the observed findings, a new classification will be developed, encompassing the features of anatomical collapse (similar but more extensive than the existing VOTE (Velum oropharynx tongue base epiglottis)/NOHL (Nose oropharynx hypopharynx and larynx) classification systems) and phenotype (modified from the existing PALM (Pcrit (critical pressure), Arousal threshold, Loop gain, and Muscle responsiveness) classification system). Analysis of Pheno-DISE findings: machine learning will be applied to test if the simpler less-invasive phenotype markers can potentially replace some of the more invasive and time-consuming aspects of the pheno-DISE methodology.