Polysomnographic phenotypes: predictors of treatment response in obstructive sleep apnea with mandibular advancement devices

Purpose: This non-randomized clinical study aims to identify polysomnographic phenotypic characteristics that differentiate responders from non-responders to mandibular advancement devices (MAD) treatment for obstructive sleep apnea (OSA) and to establish a predictive model of treatment response for OSA using oral devices based on the set of anthropometric, demographic, and polysomnographic phenotypic characteristics. Methods: This study was registered under the identifier number: NCT02724865. It prospectively analyzed patients receiving MAD treatment for six years. The MADs used were two-piece adjustable appliances following a standardized protocol. Treatment response was defined according to the latest International Consensus Statement on OSA severity. The study analyzed polysomnographic phenotypes, categorizing them into positional phenotype, sleep-stage phenotype (REM/NREM-OSA), and airway collapsibility phenotype. A logistic regression model and a classification and regression tree were implemented. Results: A total of 112 patients completed the study (64 responders and 48 non-responders). Positional-OSA patients had higher response rates than non-positional (64.1% vs. 35.9; p 0.032). REM-OSA and apnea-predominant phenotype showed a lower response (p < 0.001). In these phenotypes, most patients were women, with higher body mass index, higher scores in the Epworth Sleepiness Scale, lower minSaO2 in REM-OSA phenotype, and higher T90% in apnea-predominant phenotype. Conclusion: This study underscores the importance of hypoxic burden in the severity of OSA. The parameters T90% and POSA formed predictive model. Additionally, MAD appears to be less effective in the REM-OSA phenotype. Moreover, although patients with an apnea-predominant phenotype responded less favorably, there was a conversion from apneas to hypopneas, reducing severity.