Circadian regulation, autonomic function, and Alzheimer’s disease
Study hypotheses. Credit © Li Lab
This page is intended to showcase what we have found from this project. While our primary audience is the research community, we are trying our best to have plain language summaries for each major finding. We hope that this practice will be of help for the broader community to benefit from our research. Note that this page will be updated on a regular basis until we officially close this project.
Cure for Alzheimer’s disease is still lacking. It is important to identify the risk factors for the disease and its multiple impacts on body functions in order to prevent or slow down the progression of the disease and treat related symptoms. Using novel non-invasive assessment of circadian regulation and autonomic function by wearable technology, this project is designed to determine whether changes in these two important physiological functions can predict the development and progression of Alzheimer’s disease and cognitive decline in the elderly people at early, preclinical stages.
Disrupted rest-activity rhythms as early phase manifestations of Alzheimer’s dementia
Circadian measures derived from daily motor activity predicted the risk of Alzheimer’s dementia and the transition from mild cognitive impairment stage to Alzheimer’s dementia, even at many years before the incident.
We analyzed >5,600 activity recordings from >1,000 older participants in the Rush Memory and Aging Project (MAP) who had been followed with motor activity assessment since 2015. To quantify the near 24-h rest-activity rhythms, we performed a series of quantitative analysis incluidng cosinor analysis, nonparametric analysis, as well as a novel data-driven analysis known as empirical mode decomposition. Statistical results showed that lower amplitude of the 24-h rest-activity rhythms (meaning reduced rhythm strength) and higher intradaily variability (meaning more fragmented rhythms) predicted elevated risk of developing Alzheimer’s dementia. Also, lower amplitude, higher intradaily variability, and lower interdaily stability (meaning less robust rhythms) predicted a faster transition to Alzheimer’s dementia in participants with mild cognitive impairment.
Predicted cumulative hazard for developing Alzheimer’s dementia. Shown are the cumulative hazard functions plotted against time since baseline of two representative subjects with their amplitude (left panel) or intradaily variability (right panel) at the 10th and 90th percentiles, respectively. Abbreviations: IV = intradaily variability.
We also modeled the longitudinal changes of the rest-activity rhythms as participants aged based on the annual follow-up data. We found that the amplitude, acrophase (representing the clock time of peak activity), and interdaily stability progressively decreased, and intradaily variability progressively increased over time. Alzheimer’s progression accelerated these aging effects by doubling or more than doubling the annual changes in these measures after the diagnosis of MCI, and further doubled them after the diagnosis of Alzheimer’s dementia.
Circadian disturbance interacts with Alzheimer’s disease progression. Shown are the predicted levels of (A) amplitude, (B) acrophase, (C) interdaily stability, and (D) intradaily variability based on mixed models for a hypothetical female (red) and separately a hypothetical male individual (blue) with mean age of the cohort (i.e., 81 years old) who developed mild cognitive impairment (MCI) and Alzheimer’s dementia (AD) 3·7 and 7·5 years, respectively, after baseline. Predicted confidence intervals are shown by filled polygons overlaid on the corresponding predicted means. Amplitude and interdaily stability progressively decreased over time, and their decreases were accelerated after the diagnosis of MCI, and were further accelerated after the diagnoses of AD. Acrophase progressively advanced, and the phase advance was further accelerated after the diagnosis of AD. Intradaily variability progressively increased overtime, the its increase was accelerated after MCI, and was further accelerated after AD. Sex did not affect the changes of any of these variables although there were significant sex differences in these variables at baseline.
- Li P, Gao L, Gaba A, Yu L, Cui L, Fan W, Lim ASP, Bennett DA, Buchman AS, Hu K. Circadian disturbances in Alzheimer’s disease progression: a prospective observational cohort study of community-based older adults. The Lancet Healthy Longevity. 2020 Dec;1(3):e96–e105. PMCID: PMC8232345
As a methodological improvement, we further developed the nonlinear data-adaptive approach (i.e., uniform phase masking empirical mode decomposition, or UPMEMD for short) for circadian rhythmicity analysis. We hope to provide a generic tool for circadian analysis to the related research community.
A data-adaptive approach for extracting ~24 h circadian daily rhythms from actigraphy. A (upper left) Actogram from a young adult who had stable sleep-wake timings and (lower left) actigraphy with 24-h cosinor component extracted from cosinor analysis and ~24 h component extracted from UPMEMD superimposed. B (upper right and lower right) Illustrated by the same means by with data from a young adult who had irregular sleep-wake timings.
This method allowed us to examine the variations in rest-activity cycle length and their relationships with aging and Alzheimer’s dementia. Based on data from the Rush Memory and Aging Project, we used the UPMEMD to extract the rhythm of ~24-h, including the length and peak timing, and the standard deviations (SD) of cycle lengths and peak timings. We examined their longitudinal changes using statistical models. We found that SD of cycle length increased by 0.04 hours/year when participants were cognitively intact (95%CI: 0.02-0.05, p < 0.001); the rate increased by an additional 0.04 hours/year after the onset of MCI (95%CI: 0.02-0.06, p < 0.001) and increased further by 0.18 hours/year (95%CI: 0.12-0.24, p < 0.001) after the onset of AD.
Trajectory of the variation of cycle lengths over time. Shows are model estimations for the trajectories of the variation of cycle lengths for a male adult and a female adult, respectively.
- Gaykova N, Gao C, Yang HW, Lo MT, Hu K, Li P. The uniform phase empirical mode decomposition method for analyzing circadian rhythms. Society for Research on Biological Rhythms 2022 Conference; Omni Amelia Island, FL
- Gao C, Gao L, Gaykova N, Yu L, Yang J, Bennett DA, Buchman AS, Hu K, Li P. Variations in rest-activity cycle lengths and peak timing during Alzheimer’s progression. Alzheimer’s Association International Conference 2022; San Diego, CA and Online
- Gao C, et al. Manuscript(s) in preperation