ABSTRACT Obstructive Sleep Apnea (OSA) is a common disorder with serious health consequences that often remains undertreated due to few therapeutic options beyond continuous positive airway pressure (CPAP). Alternative treatments are available, such as mandibular advancement devices (MADs) and hypoglossal nerve stimulation (HGNS). However, these treatments are not always effective, and the predictors of success are poorly understood or difficult to obtain. Furthermore, patients who fail these therapies are often left untreated and therefore susceptible to the clinical consequences of OSA. In the previous grant period, we developed methods for estimating the pathophysiological endotypes of OSA (pharyngeal collapsibility, loop gain, pharyngeal muscle compensation, and arousal threshold), as well as the primary site of airway collapse (palate, tongue, lateral walls, epiglottis), from the clinical polysomnogram (PSG). The objective of this grant is to apply these methods to 1) find predictors of MAD and HGNS response and 2) test endotype-specific pharmacotherapies in MAD and HGNS non-responders. Our hypothesis is that these endotypes/sites of collapse, as determined from the PSG, are important predictors of MAD and HGNS response. We also hypothesize that the addition of a drug targeting an abnormal endotype, e.g., high loop gain, can be used to more effectively treat MAD and HGNS non-responders. In the first two aims, the physiological endotypes and sites of collapse will be determined from the clinical PSG using the methods developed in the previous grant cycle. Patients will then undergo MAD (Aim 1) or HGNS (Aim 2) therapy. A follow-up PSG will be performed to evaluate the success of treatment. Using multivariable logistic regression, the significant physiological predictors of success will be determined. In Aim 3, the patients who fail MAD or HGNS in the previous aims, approximately one-third of individuals, will be treated with a drug targeting the most abnormal endotype (acetazolamide for high loop gain, atomoxetine-plus-oxybutynin for poor pharyngeal muscle compensation, or trazodone for low arousal threshold). Recent evidence suggests that these drugs can manipulate the endotypes. The drug will be administered concurrently with MAD or HGNS treatment (combination therapy). Aims 1 and 2 are expected to find the important physiological predictors of MAD and HGNS response, respectively, using novel metrics derived from the clinical PSG. Aim 3 is expected to provide proof-of-principle for a pharmacologic approach to treating MAD and HGNS failures, patients who otherwise have limited treatment options. Ultimately, these studies have the potential to improve patient selection for non-CPAP alternatives and broaden the treatment options for OSA.