Associations between actigraphy-derived circadian rest-activity rhythms and cancer-related fatigue: Insights from the Harvest for Health Clinical Trial
openNCI - National Cancer Institute
PROJECT SUMMARY
Cancer-related fatigue affects nearly all people undergoing cancer treatment and sometimes persists for months
or years post-treatment. The etiology and pathophysiology of cancer-related fatigue are not well understood but
may be related to disruption in circadian rhythms. Circadian rhythms describe the daily oscillations of metabolic,
endocrine, and neuronal systems, all of which regulate energy levels. Behavioral interventions such as exercise
and diet can regulate (or dysregulate) circadian rhythms, though circadian mechanisms are rarely leveraged in
behavioral interventions. Ambulatory actigraphy data has become easy and inexpensive to collect, and there is
a plethora of methods to analyze rest-activity rhythms from actigraphy data. However, timing is rarely
incorporated into behavioral programs to address fatigue because the field has not yet defined unequivocal rest-
activity rhythm parameters from actigraphy datasets that are associated with fatigue. Thus, the objective of this
project is to quantify the associations between rest-activity rhythm parameters and persistent cancer-related
fatigue using a battery of analysis methods. This study leverages the Harvest for Health study, which was a
randomized controlled trial among older cancer survivors testing the effects of a 1-year gardening intervention
vs. waitlist control. The dataset has actigraphy and fatigue data for 279 participants over two years. Participants
wore an actigraph for 7 days and reported fatigue at baseline, 1 year, and 2 years. We will use traditional methods
to quantify rest-activity rhythms (i.e., cosinar analysis, non-parametric measures) as well as novel hidden Markov
modelling, based on a probabilistic framework. The hidden Markov models go beyond traditional methods in that
they provide visual Day Profiles of activity and can quantify where in the day circadian disruptions are occurring.
Aim 1 will assess the association between the rest-activity rhythm index parameters and fatigue from baseline
to 1 and 2 years. We predict that higher hidden Markov-derived rhythm indices and other parameters that reflect
stronger rest-activity rhythms will be associated with less fatigue. Aim 2 will assess the changes in rest-activity
rhythm parameters over time. We predict that rest-activity rhythms will get stronger over time, as people get
further from cancer treatment. Aim 3 will test the effects of the gardening intervention vs. control on Hidden
Markov model-derived rhythm indices. We hypothesize that the gardening intervention will result in higher
average rhythm index (stronger rest-activity rhythm) than the control at year 1. These findings will immediately
be useful to researchers in the optimization of behavioral interventions (e.g., exercise, nutrition, sleep hygiene)
to strengthen biological rhythms, address fatigue, and improve quality of life in cancer survivorship. The results
and the resultant code to easily run hidden Markov models will inform future research and clinical applications.
For example, these models will inform personalized recommendations for people struggling with fatigue based
on a person’s rest-activity rhythm—should we focus efforts to improve sleep, encourage morning activity,
minimize nutrient intake in the evening, or something else?—thereby accelerating people’s recovery from cancer.
Up to $411K
health research