1002 Heart Rate Variability During Sleep and Future Risk of Parkinson’s Disease in Community-dwelling Older Men

Abstract Introduction Autonomic dysfunction is common in early stages of Parkinson’s disease (PD), providing an opportunity for early diagnosis and disease prediction. However, little is known about the longitudinal association between overnight—and sleep stage specific—heart rate variability (HRV), a marker of autonomic function, and incident PD. Methods Our sample consisted of 2470 older men (mean [SD] age = 76.1 [5.4] years) without PD who underwent polysomnography between 2003 and 2005 and were followed until 2016 for incident self-reported PD as part of the Osteoporotic Fractures in Men (MrOS) prospective study. HRV was calculated from the electrocardiogram in 5-minute epochs across the whole night and during rapid eye movement (REM) sleep and non-REM stages N2 and N3 (N2N3) alone. HRV metrics included measures of overall HRV (SDNN, standard deviation of NN intervals), short-term HRV/parasympathetic influence (RMSSD, root mean square of successive NN interval differences; HF, high frequency power), long-term HRV/sympathetic, parasympathetic, and baroreflex influence (LF, low frequency power), and long-term to short-term HRV balance (LF/HF; LFnu, normalized LF power, LF/[HF+LF]), which may represent sympatho-vagal balance. Associations between tertiles of HRV metrics and incident PD were assessed using multivariable logistic regression. Results During 11 years of follow-up, 60 incident PD cases were identified. After adjustment for covariates including age and comorbidities, men in the lowest tertile of LFnu measured across the whole night [OR (95% CI) = 2.13 (1.09,4.35)] and during REM alone [OR (95% CI) = 2.44 (1.23,5.55)] had approximately twice the odds of developing PD compared to those in the highest tertile. Results were similar for LF/HF, though the association was only significant during REM [OR (95% CI) = 2.33 (1.08,5.55)]. Further adjustment for apnea-hypopnea index did not attenuate these associations. HRV metrics assessed during N2N3 were not associated with incident PD, nor were SDNN, RMSSD, HF, and LF. Conclusion Individuals with decreased long-term to short-term HRV balance (possibly representing low sympatho-vagal balance) had higher risk of developing PD. REM may be a particularly important sleep stage for measurement of autonomic dysfunction in early PD and assessment of future PD risk. Support (if any) Support comes from the National Institute on Aging (R21AG085495).