Chronic Circadian Disruption and Sleep Restriction Influence Subjective Hunger, Appetite, and Food Preference

Abstract

Chronic circadian disruption (CCD), such as occurs during rotating shiftwork, and insufficient sleep are each independently associated with poor health outcomes, including obesity and glucose intolerance. A potential mechanism for poor health is increased energy intake (i.e., eating), particularly during the circadian night, when the physiological response to energy intake is altered. However, the contributions of CCD and insufficient sleep to subjective hunger, appetite, food preference, and appetitive hormones are not clear. To disentangle the influences of these factors, we studied seventeen healthy young adults in a 32-day in-laboratory study designed to distribute sleep, wakefulness, and energy intake equally across all phases of the circadian cycle, thereby imposing CCD. Participants were randomized to the Control (1:2 sleep:wake ratio, n = 8) or chronic sleep restriction (CSR, 1:3.3 sleep:wake ratio, n = 9) conditions. Throughout each waking episode the participants completed visual analog scales pertaining to hunger, appetite, and food preference. A fasting blood sample was collected to assess appetitive hormones. CCD was associated with a significant decrease in hunger and appetite in a multitude of domains in both the Control and CSR groups. This change in hunger was significantly correlated with changes in the ghrelin/leptin ratio. These findings further our understanding of the contributions of CCD and insufficient sleep on subjective hunger and appetite as well as of their possible contributions to adverse health behaviors.

Keywords: circadian misalignment; ghrelin; insufficient sleep; leptin; shiftwork.

Figure 1

Acute and chronic impact of circadian disruption and sleep restriction on subjective hunger and appetite. Data from the Control (n = 8, 1:2 sleep:wake ratio) participants are denoted by open circles and those from the Chronic Sleep Restriction (CSR, n = 9, 1:3.3 sleep:wake ratio) participants are denoted by closed circles. Higher scores indicate higher subjective feelings of each aspect of hunger and appetite; the possible range was 0–100. Beat cycles (i.e., time to complete a cycle of circadian and sleep:wake schedule combinations, which was ~6 protocol days in this forced desynchrony design) of the protocol are shown on the x-axis. Error bars represent SEM. Note, the y-axis scale of Nausea is one-half the range of the others. p values are derived from linear mixed models with beat cycle, condition, and their interactions as fixed effects and participant as a random factor.

Figure 2

Chronic impact of circadian disruption on subjective hunger and appetite. Data from the Control (n = 8, 1:2 sleep:wake ratio) participants are denoted by open circles and those from the Chronic Sleep Restriction (CSR, n = 9, 1:3.3 sleep:wake ratio) participants are denoted by closed circles. Higher scores indicate higher subjective feelings of each aspect of hunger and appetite. Beat cycles (i.e., time to complete a cycle of circadian and sleep:wake schedule combinations, which was ~6 protocol days in this forced desynchrony design) of the protocol are shown on the x-axis. Group means are each beat cycle are denoted by the open column and error bars represent SEM. Note, the y-axis of Nausea is approximately one-half the range of the others. p values are derived from independent t-tests used to test differences between Beat cycle 1 and Beat cycle 4.

Figure 3

Acute and chronic impact of circadian disruption and sleep restriction on subjective food preference. Data from the Control (n = 8, 1:2 sleep:wake ratio) participants are denoted by open circles and those from the Chronic Sleep Restriction (CSR, n = 9, 1:3.3 sleep:wake ratio) participants are denoted by closed circles. Higher scores indicate higher subjective feelings of each aspect of food preference; the possible range was 0–100. Beat cycles (i.e., time to complete a cycle of circadian and sleep:wake schedule combinations, which was ~6 protocol days in this forced desynchrony design) of the protocol are shown on the x-axis. Error bars represent SEM. p values are derived from linear mixed models with beat cycle, condition, and their interactions as fixed effects and participant as a random factor.

Figure 4

Acute and chronic impact of circadian disruption on food type preference. Data from the Control (n = 8, 1:2 sleep:wake ratio) participants are denoted by open circles and those from the Chronic Sleep Restriction (CSR, n = 9, 1:3.3 sleep:wake ratio) participants are denoted by closed circles. Higher scores indicate higher subjective feelings of each aspect of food preference. Beat cycles (i.e., time to complete a cycle of circadian and sleep:wake schedule combinations, which was ~6 protocol days in this forced desynchrony design) of the protocol are shown on the x-axis. Group means of each beat cycle are denoted by the open column and error bars represent SEM. p values are derived from independent t-tests used to test differences between Beat cycle 1 and Beat cycle 4.

Figure 5

Impact of circadian disruption on ghrelin/leptin ratio and the relationship between the change in subjective hunger on a visual analog scale and the change in fasted ghrelin/leptin ratio across chronic circadian disruption (Beat cycle 1 to Beat cycle 4). Data from the Control (n = 8, 1:2 sleep:wake ratio) participants are denoted by open circles and those from the Chronic Sleep Restriction (CSR, n = 9, 1:3.3 sleep:wake ratio) participants are denoted by closed circles in the (right) panel. Beat cycles (i.e., time to complete a cycle of circadian and sleep:wake schedule combinations, which was ~6 protocol days in this forced desynchrony design) of the protocol are shown on the x-axis. Group means for each beat cycle are denoted by the open column and error bars represent SEM. p values are derived from independent t-tests used to test differences between Beat cycle 1 and Beat cycle 4. In the (left) panel, data points represent individual participants and the solid line represents the linear fit of the data. p values derived from Pearson correlation.