Melatonin advances the circadian timing of EEG sleep and directly facilitates sleep without altering its duration in extended sleep opportunities in humans

Abstract

The rhythm of plasma melatonin originating from the pineal gland and driven by the circadian pacemaker located in the suprachiasmatic nucleus is closely associated with the circadian (approximately 24 h) variation in sleep propensity and sleep spindle activity in humans. We investigated the contribution of melatonin to variation in sleep propensity, structure, duration and EEG activity in a protocol in which sleep was scheduled to begin during the biological day, i.e. when endogenous melatonin concentrations are low. The two 14 day trials were conducted in an environmental scheduling facility. Each trial included two circadian phase assessments, baseline sleep and nine 16 h sleep opportunities (16.00-08.00 h) in near darkness. Eight healthy male volunteers (24.4 +/- 4.4 years) without sleep complaints were recruited, and melatonin (1.5 mg) or placebo was administered at the start of the first eight 16 h sleep opportunities. During melatonin treatment, sleep in the first 8 h of the 16 h sleep opportunities was increased by 2 h. Sleep per 16 h was not significantly different and approached asymptotic values of 8.7 h in both conditions. The percentage of rapid eye movement (REM) sleep was not affected by melatonin, but the percentage of stage 2 sleep and sleep spindle activity increased, and the percentage of stage 3 sleep decreased. During the washout night, the melatonin-induced advance in sleep timing persisted, but was smaller than on the preceding treatment night and was consistent with the advance in the endogenous melatonin rhythm. These data demonstrate robust, direct sleep-facilitating and circadian effects of melatonin without concomitant changes in sleep duration, and support the use of melatonin in the treatment of sleep disorders in which the circadian melatonin rhythm is delayed relative to desired sleep time.

Figure 1. Sleep profiles for one representative participant

Sleep opportunities are shown for the placebo (left side) and melatonin (right side) trials, as a function of clock time (h). Study days are presented beneath each other. For each day, sleep stages (REM, wake, stage 1, stage 2, stage 3, stage 4) are shown in the upper panel, and delta activity (power density in the 0.75–4.5 Hz band) is shown in the lower panel. On the constant routine days (D2 and D12), participants were instructed to remain awake, and therefore sleep profiles on these days are not presented. The onset of endogenous melatonin secretion (the time when plasma melatonin levels increased beyond 25% of the peak nocturnal level) is indicated as an open circle for each constant routine (redrawn with permission from Rajaratnam et al. 2003). Baseline sleep (D1) was similar in the two treatment trials. Sleep during the 16 h sleep opportunity following the first constant routine was well consolidated in both the placebo and the melatonin trials (D3). During the subsequent 16 h sleep opportunities (D4 to D10) clear differences between the two conditions emerged. In the placebo trial, the 16 h sleep opportunities were often characterized by relatively rapid onset to sleep, followed by long periods of wakefulness in the initial 8 h (i.e. 16.00 h to 00.00 h). This in turn was followed by consolidated sleep in the second 8 h (i.e. from 00.00 h to 08.00 h). In the melatonin trial, the longest consolidated bout of sleep was present in the first part of the sleep opportunity and more wakefulness was observed in the second part. It is noted that on several occasions, sleep was not initiated immediately after melatonin administration (e.g. D7 to D10). During the washout night on D11, when placebo was administered in both trials, sleep onset occurred earlier in the melatonin trial than in the placebo trial. Wakefulness was present in the second half of the sleep opportunity in the melatonin trial, but not in the placebo trial. Following the second constant routine, sleep was well consolidated in both trials and delta activity declined progressively in the course of the sleep episode, as it did in most other episodes of consolidated sleep.

Figure 2. Sleep efficiency profiles in the melatonin and placebo trials

Mean sleep efficiency profiles in the placebo (left side) and melatonin (right side) trials. Sleep efficiency is calculated across participants (n = 8) in 1 h intervals for each sleep opportunity. Data are represented as colour contour plots according to the legend presented in the lower panel, with lowest (0%) sleep efficiency indicated in dark purple and highest sleep efficiency (100%) indicated in dark orange. Consecutive study days are presented beneath each other. Clock time (h) and hours into the sleep opportunity are shown on the horizontal axis. For each day, important study events and treatments are indicated (baseline; CR1, constant routine 1; Rec1, recovery sleep 1; placebo or melatonin treatment administration; CR2, constant routine 2; Rec2, recovery sleep 2). In the placebo trial, sleep efficiency was generally low in the initial part of the 16 h sleep episode, i.e. 16.00 h to midnight, though D4 to D11. The major bout of sleep, as indicated by high sleep efficiency, occurred at a similar phase throughout the protocol, i.e. from about midnight to 08.00 h, with some day-to-day variability. In contrast, in the melatonin trial, the major sleep bout was substantially advanced in time compared to placebo, and occurred during the first half of the sleep opportunity. This change in distribution was found to persist on D11, when both groups received placebo, although the advance in the period of high sleep efficiency on D11 is less pronounced than when melatonin was administered on D10.

Figure 3. Wakefulness, delta activity and sigma activity

Time course of mean wakefulness (% per 2 h; A), mean delta activity in NREM sleep (power density in the 0.75–4.5 Hz band) expressed as a percentage of baseline (D1; B), and mean sigma activity in NREM sleep (power density in the 12.0–13.59 Hz band) expressed as a percentage of baseline (D1; C) (n = 8). Standard error of the mean (s.e.m.) is shown. Data points have been calculated across the treatment administration period (D4 to D10). D3 was excluded due to the influence of the extended period of wakefulness associated with the initial constant routine. During the placebo trial, wakefulness was greater during the first half of the sleep opportunity compared to the second half. In contrast, the opposite pattern is observed for wakefulness during the melatonin trial. The time course of delta activity is similar for the melatonin and placebo trials, showing a decrease as a function of the number of hours into the sleep opportunity. Sigma activity is increased in the first half of the sleep opportunity after melatonin treatment, but is not different to placebo in the second half of the sleep opportunity.

Figure 4. Total sleep time, REM sleep and NREM sleep

Time course of total sleep time (TST; h), REM sleep (h) and NREM sleep (h) during the melatonin and placebo trials quantified with an exponential decaying function fitted to mean data (n = 8) using a non-linear regression procedure. For TST, the function used was TST_day = TST₀× e^−day/t+ TST_∞. TST_∞ represents the value of TST when ‘day’ approaches ∞; TST₀ is the hypothetical intercept with the ordinate if TST₀ were 0; t is the time constant of the exponential decaying function. The parts of the study involving sleep deprivation (i.e. the constant routines, CR) are indicated with light grey shading. The horizontal asymptote is shown with a horizontal line, and the asymptote value is indicated for each function. Asymptotic values were estimated to be 8.7 h for TST (95% CI: melatonin 7.7–9.7, placebo 8.3–9.1), 2.0 h for REM sleep (95% CI: melatonin 1.8–2.2, placebo 1.9–2.2) and 6.6 h for NREM sleep (95% CI: melatonin 5.5–7.8, placebo 6.4–6.9). These values were identical for the melatonin and placebo trials.

Figure 5. Separation of direct and circadian effects of melatonin

Mean sleep efficiency levels (% per hour; A and B, n = 8) and mean plasma melatonin levels (C and D n = 8). Standard error of the mean (s.e.m.) is shown for sleep efficiency and plasma melatonin data. The direct, sleep-facilitating effect of melatonin (A) is illustrated by a comparison between sleep efficiency profiles on the last day of melatonin treatment (D10, •) and sleep efficiency on the washout day (D11, ○). Increased sleep efficiency is observed for the first 2–3 h during melatonin treatment. The circadian effect of melatonin on sleep (B) is illustrated by a comparison between sleep efficiency levels on the washout day (D11), which was the day after melatonin (•) or placebo (○). On D11, placebo was administered to all participants. A shift in the distribution of sleep can be observed after melatonin treatment, with the major bout of sleep occurring earlier in the sleep opportunity. On the corresponding day after placebo, the major bout of sleep occurred later in the sleep opportunity, although an initial rise in sleep efficiency is noted at around the commencement of the sleep opportunity. Mean plasma melatonin profiles are presented (C) to illustrate the difference in circulating melatonin levels during melatonin treatment (•) compared to the levels after treatment had stopped (○) (reproduced and re-analysed with permission from Rajaratnam et al. 2003). Note that during melatonin treatment, plasma melatonin levels remained elevated for the duration of the 16 h sleep opportunity. Plasma melatonin levels during the second constant routine (i.e. after the treatment period) are presented (D) to illustrate the melatonin-induced advance in timing of the endogenous melatonin profile (•) compared to placebo (○) (reproduced with permission from Rajaratnam et al. 2003).

Figure 6. Relationship between the sleep efficiency profile and the plasma melatonin rhythm

Mean cross-correlation coefficients, where the observations from one data series are correlated with another series at various lags and leads. Sleep efficiency data from the washout night are compared to plasma melatonin levels during the second constant routine, for the melatonin trial (upper panel) and the placebo trial (lower panel). Plasma melatonin data were taken with permission from Rajaratnam et al. (2003). The horizontal dashed line represents a correlation coefficient (r) of 0. Time lag (h) is shown on the horizontal axis. Data points represented by open circles indicate time lags with significant correlation coefficients (P < 0.05). Maximal correlation coefficients obtained for both trials were at 0 h lag.