The circadian system modulates the cortisol awakening response in humans

Abstract

Background: In humans, circulating cortisol usually peaks 30-60 min after awakening from nocturnal sleep, this is commonly referred to as the cortisol awakening response (CAR). We examined the extent to which the CAR is influenced by the circadian system, independent of behaviors including sleep.

Materials and methods: We examined the CAR in 34 adults (20 female) using two complementary multiday in-laboratory circadian protocols performed in dim light, throughout which behavioral factors were uniformly distributed across the 24-hour circadian cycle. Protocol 1 consisted of 10 identical consecutive 5-hour 20-minute sleep/wake cycles, and protocol 2 consisted of 5 identical consecutive 18-hour sleep/wake cycles. Salivary melatonin was used as the circadian phase marker (0° = dim light melatonin onset). During each sleep/wake cycle, salivary cortisol was measured upon scheduled awakening and 50-minutes later, with the change in cortisol defined as the CAR. Cosinor analyses were used to detect any significant circadian rhythmicity in the CAR. In secondary analyses, we adjusted the models for time awake before lights on, total sleep time, percent of rapid eye movement (REM) sleep, and percent of non-rapid eye movement (NREM) sleep.

Results: Both protocols revealed a similar circadian rhythm in the CAR, with peaks occurring at a circadian phase corresponding to 3:40-3:45 a.m., with no detectable CAR during the circadian phases corresponding to the afternoon. In addition to the sinusoidal component of the circadian rhythm, total sleep time was also associated with the CAR for protocol 1. The percent of sleep spent in REM or NREM sleep were not associated with the CAR in either protocol.

Conclusion: Our results show that the CAR exhibits a robust circadian rhythm that persists even after adjusting for prior sleep. Presuming that the CAR optimizes physiological responses to the anticipated stressors related to awakening, these findings may have implications for shift workers who wake up at unusual circadian phases. A blunted CAR in shift workers upon awakening in the evening may result in diminished responses to stressors.

Keywords: ACTH; HPA-axis; awakening; chronic stress; circadian; cortisol; sleep; time of day.

FIGURE 1

Schematic representation of study protocols for an individual with a habitual sleep time from midnight to 8:00 a.m. (clock time is provided on the x-axis). For protocol 1, (A) a baseline night of sleep at this habitual time is followed by a baseline day of wakefulness (11-hours). After baseline assessments, the circadian protocol is introduced which evenly distributes behaviors across all circadian phases with 10 identical consecutively recurring 5-hour 20-minute cycles (2-hour and 40-minute of scheduled wakefulness and 2-hour and 40-minute of a scheduled sleep opportunity) across 5 days. For protocol 2, (B) two nights of sleep at the participant’s habitual sleep time with 16-hour scheduled wakefulness in between, is followed by a circadian protocol with reoccurring 18-hour “days” (12-hour wake and 6-hour sleep cycles) across 7 days. Black boxes indicate scheduled sleep opportunities in darkness (<0.1 lux), gray bars indicate scheduled wakefulness in dim light (∼3 lux), and white boxes represent regular room light used when participants are admitted or discharged from the study.

FIGURE 2

Circadian rhythm in the cortisol awakening response (CAR). The cortisol concentrations upon-awakening and 50 min post-awakening as a function of circadian time are plotted on the left column (A,C) and the CAR is plotted on the right (B,D). Data are expressed as absolute values (left y-axes) and as a percentage of the mean of each participant’s upon-awake values (right y-axes). Participants’ binned data (60°, 4-hour intervals, means±SEM, left y-axes) are depicted as black circles (upon-awake) and red triangles (50 min post-awakening). Binned data were averaged within an individual first, as an individual could contribute more than one point to each bin. The solid lines represent the cosinor model fits. The blue arrows in (A,C) highlight the peaks of CAR in the two protocols. The corresponding clock times (determined from the average times of the DLMO for these participants) are shown on the top x-axes. Gray bars also on the top x-axes indicate the participants’ average habitual sleep times (determined from the at home assessment).

FIGURE 3

Associations between measured sleep parameters during the 10 forced desynchrony sleep opportunities of protocol 1 and the cortisol awakening response (CAR). Participants’ binned data (binned into 5–6 intervals for a given sleep parameter, means±SEM) are depicted as black squares. Binned data were averaged within an individual first, as an individual could contribute more than one point to each bin. The solid lines represent significant linear regressions as tested using mixed models (see Table 2). A quadratic term was also estimated but not significant.

FIGURE 4

Associations between measured sleep parameters during the five forced desynchrony sleep opportunities of protocol 2 and the cortisol awakening response (CAR). Participants’ binned data (binned into 4–5 intervals for a given sleep parameter, means±SEM) are depicted as black circles. Binned data were averaged within an individual first, as an individual could contribute more than one point to each bin. Linear (see Table 2) and quadratic models were not significant.