Associations of actigraphic sleep and circadian rest/activity rhythms with cognition in the early phase of Alzheimer's disease

Abstract

Study objectives: To compare sleep and circadian rest/activity rhythms (RARs), quantified by standard and novel actigraphic metrics, between controls and participants with mild cognitive impairment (MCI), and to examine the cross-sectional relationships between these measures and cognition.

Methods: Actigraphy data were collected in 179 older individuals (mean age = 72.6 years) with normal cognition (n = 153) and MCI (n = 26). Sleep parameters (e.g. sleep efficiency), and standard nonparametric RARs (e.g. interdaily stability) were generated. Functional principal component analysis (fPCA) was used to generate three novel RAR metrics (fPC1, fPC2, and fPC3). Cognitive composite scores reflecting episodic memory and executive function were derived using factor analysis. Regression models compared sleep and RAR parameters between diagnostic groups and their association with cognitive performance.

Results: Compared to controls, the MCI group exhibited lower levels of the standard RAR parameter: relative amplitude and fPC3-a novel RAR whereby lower scores reflected a lower rhythm peak, as well as greater nighttime activity and less activity in the morning. Across groups, several standard RAR parameters (e.g. interdaily stability) and fPC3 were associated with better episodic memory and executive function performance. Additionally, several standard RAR measures (e.g. relative amplitude) and the novel RAR measure fPC1 (reflecting the total volume of activity and rhythm strength) were associated with better executive function performance.

Conclusions: Individuals with MCI have altered circadian RARs compared to controls, including the novel RAR metric fPC3, reflecting greater nighttime activity and less activity in the morning compared to mean values. Additionally, these measures are significantly associated with cognitive performance.

Keywords: Alzheimer’s disease; actigraphy; circadian rhythms; cognition; sleep.

Figure 1.

Visualization of Standard and Novel Circadian Rest/Activity Rhythms. Panel (a) represents the standard nonparametric rest/activity rhythm (RAR) parameter relative amplitude (RA). RA was dichotomized for visualization using a median split. The black line represents the mean rhythm of the overall sample, whereas the red line represents the mean rhythm among individuals with higher RA and the blue line reflects the mean rhythm among those with lower RA. Panels (b), (c), and (d) reflect the novel RAR parameters, functional principal components (fPC) 1, 2 and 3. In each panel, the black line represents the mean rhythm of the overall sample. The red line is two standard deviations above the mean for each respective fPC score, while the blue line is two standard deviations below the mean. Combined, fPC1, fPC2, and fPC3 account for approximately 69% of the total variation in RARs.

Figure 2.

Standard and Novel Circadian Rest/Activity Rhythms in the Normals vs. MCI. Panel (a) represents the standard nonparametric rest/activity rhythm (RAR) parameter relative amplitude (RA) in the Normals vs. MCI. The solid line reflects the mean rhythm of the Normals, while the dashed line shows the mean rhythm of the MCI group. Panel (b) demonstrates that RA is significantly different between the Normals and MCI subjects. Panel (c) demonstrates that functional principal component 3 (fPC3) is significantly different between the Normals and MCI subjects. For panels (b) and (c), y-axis units represent the standardized residuals after adjustment for age, gender, education, and APOE ε4 genotype.

Figure 3.

Associations of Standard and Novel Circadian Rest/Activity Rhythms with Cognitive Performance. Panel (a) demonstrates the associations of both the standard (open circles) and novel rest/activity rhythm (RAR) parameters (purple dots) with executive function performance. Panel (b) depicts the associations of both the standard and novel RAR parameters with episodic memory. Associations are depicted using partial regression plots of the standardized residuals after adjustment for age, gender, education, and APOE ε4 genotype.