Abstract
Stroke is the leading cause of adult disability in the US, a top killer of Americans, and impacts Mexican
Americans (MAs) to a greater extent than non-Hispanic whites (NHWs). One opportunity to improve stroke
outcomes and reduce disparities may exist through identification and treatment of obstructive sleep apnea
(OSA) in individuals with stroke. OSA in the general population is heterogeneous with respect to
pathophysiology, expression of disease, response to therapies, and association with outcomes. Mechanistic
causes of airway collapse during sleep can be categorized from polysomnography (PSG) data as OSA
endotypes, including an anatomic cause (collapsibility), and 3 non-anatomic causes (pharyngeal muscle
compensation, chemoreflex feedback loop/loop gain, and arousal threshold). Unlike traditional PSG data that
reflect OSA severity and not underlying cause, these endotypes determine response to treatments, and thus
new PSG-based methods to determine endotypes create novel opportunities for personalized care. OSA is
overrepresented after stroke (~75%) and manifests differently compared to the general population.
Furthermore, OSA is more prevalent and severe among MA stroke patients, who on average have a higher
BMI than NHWs, and therefore likely more airway collapsibility. Reasons for the high prevalence and the
mechanistic causes of post-stroke OSA are unknown. Due to interruption of brain pathways, non-anatomical
causes of OSA may be more likely after stroke than in the general population. In contrast, stroke cases with
pre-existing OSA may have endotypes more similar to the general population, with greater contribution from
collapsibility. Leveraging the infrastructure of a longstanding population-based study (BASIC) and its ancillary
study for subject identification, baseline data collection, and baseline PSG, this prospective study with a stroke-
free comparison group, with longitudinal follow-up seeks to: 1) determine specific endotypes and endotypic
profiles that contribute to post-stroke OSA, and how these differ by ethnicity and from those without stroke, 2)
determine how specific endotypic profiles relate to improvement in OSA severity typically observed early after
stroke in order to inform which patients may need longer-term treatment and which may need repeat testing for
OSA, and 3) build a model to predict post-stroke OSA endotypic profiles based on clinical information including
phenotypic data, to assist in selection of most appropriate treatment options without the need for PSG. Newly
proposed facial morphometric measures and other phenotyping will complement the rich demographic and
clinical data for consideration as predictors of post-stroke OSA endotypic profiles. This study will expand our
understanding of the pathophysiology of post-stroke OSA and open the door to personalized medicine for
stroke patients, currently dominated by a one-size-fits-all approach.