053 REM Sleep Apnea Severity Contributes to Poor Verbal Memory in Cognitively Asymptomatic Older Adults

Abstract

Abstract Introduction The prevalence of obstructive sleep apnea (OSA) rises with age, leading to increased dementia risk and memory decline. However, it remains unclear which OSA features drive this relationship. Here, we examine associations between verbal memory and multiple OSA features in healthy older adults. Methods 58 cognitively asymptomatic adults (61.4±6.3 years; 38 females) underwent polysomnography (PSG) and the Rey Auditory Verbal Learning Test (RAVLT; 0.2±0.5 years between assessments). OSA measures included apnea-hypopnea index (AHI), respiratory disturbance index (RDI), and desaturation index in both NREM and REM sleep. RAVLT measures included total learning (sum of trials 1–5), short-delayed recall, and long-delayed recall. Pearson correlations with FDR correction were calculated between OSA-related measures and RAVLT. Multiple regression was then used to adjust for OSA features in other sleep stages (i.e., REM versus NREM), age, sex, time between PSG and neuropsychological assessments, years of education, and APOE4 status. We also explored whether age moderated these relationships. Results REM RDI is negatively associated with RAVLT total learning (r=-0.31, p<0.01 with age moderating the effect at 1SD above mean: B=-0.22, t(49)=-2.88, p=0.01) and RAVLT long-delayed recall (r=-0.36, p<0.001; with age moderating the effect at mean: B=-0.05, t(49)=-2.38, p=0.02; and moderating at 1SD above mean: B=-0.09, t(49)=-3.91, p<0.01). REM desaturation index was also associated with RAVLT total learning (r=-0.21, p<0.01) and RAVLT long-delayed recall (r=-0.34, p<0.01). REM AHI was negatively correlated to RAVLT long-delayed recall (r=-0.34, p<0.01; with age significantly moderating the effect at the mean: B=-0.06, t(49)=-2.87, p=0.01; and moderating at 1SD above the mean: B=-0.11, t(49)=-4.23, p<0.01). Of note, NREM OSA features were not significantly correlated to RAVLT measures when REM OSA features were included in the model. Conclusion These findings demonstrate that REM OSA features in particular contribute to poor verbal memory encoding and retrieval, especially at older ages. Verbal memory decline has been predictive of conversion to Alzheimer’s disease (AD). Future studies including brain imaging, AD biomarkers, REM sleep oscillations, and comprehensive neuropsychological testing may elucidate the underlying mechanisms linking REM OSA features to memory decline and dementia risk. Support (if any) This research was supported by R56 AG052698 and P50AG033514.