Key points are not available for this paper at this time.
Abstract Introduction “Isolated” REM sleep behavior disorder (iRBD) is an early marker of Parkinson’s disease (PD) and dementia with Lewy bodies (DLB) that affects 1-2% of individuals after 60 years of age. We published on a classifier using nocturnal wrist actigraphy data which achieved 90% accuracy for iRBD detection. Here, we applied the same iRBD classifier in a population sample, the UK Biobank data to predict risk of incident PD or DLB. Methods 30,000 records were randomly selected from the United Kingdom Biobank (UKBB) cohort dataset and were included based on age60 years, no neurological diagnosis and 3-day valid nocturnal actigraphy data. Our published models were used to generate an iRBD score through automated detection of sleep periods and movement features from the accelerometer data. The iRBD score was generated for all the subjects and divided into: top 1%tile (predicted iRBD) and bottom 90%tile (predicted no-iRBD). The outcomes of interest were: 1) PD; 2) possible DLB. Odds ratio were then calculated using the iRBD scores (predictor) and the outcomes of interest (1 or 2). Results In the 10,087 records that met inclusion criteria, mean age was 63.8±2.8 years and subjects had on average 6.4±0.9 nights of valid actigraphy data. 109 subjects developed PD, and 67 DLB after 5.1±2.1 years. In the top 1%tile (n=101), 10 of 109 incident PD, and 18 of 67 DLB were accurately predicted by the calculated iRBD scores (sensitivity 9.2% and 26.8%, respectively). In the bottom 90%tile (n=9,078), 8,952 were accurately predicted to not develop PD or DLB (specificity 98.6%). Odds ratio were 9.1 4.0 – 20.1 for PD, 40.2 21.3 – 75.7 for DLB, and 19.8 11.9 – 32.76 for either PD or DLB. Conclusion These results provide proof of concept that the existing iRBD detection model can be used to predict neurodegenerative outcomes in the community setting using ≥3 nights of actigraphy data. Future studies may assess additional actigraphy features related to 24h rest-activity rhythms (RAR) and gait patterns. Support (if any) Michael J Fox Foundation
Wickramaratne et al. (Sat,) studied this question.