How is sleep-disordered breathing linked with biomarkers of Alzheimer disease?

Sleep-disordered breathing (SDB) is prevalent in Alzheimers disease (AD). We assessed whether and how SDB affects neuroimaging biomarkers of AD, including amyloid-beta plaque burden (A{beta}), regional uptake of fluorodeoxyglucose using positron emission tomography (rFDG-PET), grey matter volume (GMV), as well as cognitive scores and cerebrospinal fluid (CSF) biomarkers. MethodsWe selected 757 subjects from the Alzheimers Disease Neuroimaging Initiative (ADNI) database based on cognitive status (AD, mild cognitive impairment (MCI), cognitively unimpaired (CU)), and SDB condition (with/without SDB). To ensure the reliability of our findings and considering imbalanced sample size across groups, we used a stratified subsampling approach generating 10,000 subsamples (n=10/group). We then selected 512 subsamples with matched covariates. The effect size of the cognitive status-SDB interaction was computed for each biomarker and cognitive score. For reference, we computed 1000 null models by shuffling group labels randomly. The average value of effect sizes for each biomarker in each region was estimated through bootstrapping with 10,000 iterations for both the main and null models and compared with the null models distribution. Linear regression models were next implemented to identify associations between the effect size on A{beta}, rFDG, and GMV with the effect size on cognitive scores and CSF biomarkers across all subsamples. ResultsThe cognitive status-SDB interaction had a medium-sized effect on A{beta}, rFDG and GMV biomarkers in several brain areas. The effect sizes of the mentioned interactions on A{beta} plaque burden in the right precuneus, left middle temporal gyrus, and left occipital fusiform gyrus were associated with the effect sizes of the interactions on cognitive scores. Further, the interaction effect sizes on CSF A{beta}42 were related to the interaction effect sizes on A{beta} in the right precuneus and posterior cingulate cortex, as well as rFDG in the left precuneus cortex and GMV in bilateral angular gyrus and right occipital fusiform gyrus. Effect sizes on CSF p-tau were also correlated with the effect sizes on A{beta} in the left lateral occipital cortex and GMV in the left middle temporal gyrus. ConclusionWe observed that SDB interacts with neuroimaging and CSF biomarkers of AD. Specifically, SDB has a robust association with markers of A{beta} pathology in PET and CSF relative to rFDG and GMV in the AD group. The cognitive status-SDB interaction on A{beta} is associated with cognitive decline. This study further supports the hypothesis that SDB may precipitate AD pathology.