Caffeine does not entrain the circadian clock but improves daytime alertness in blind patients with non-24-hour rhythms

Highlights • Most totally blind individuals have Non-24-Hour Sleep–Wake Disorder (N24HSWD).• Optimal treatment should reset the underlying non-entrained circadian pacemaker.• We tested daily caffeine treatment (150 mg, 10 a.m.) in three blind patients with N24HSWD.• Caffeine treatment improved daytime alertness at adverse circadian phases.• Caffeine treatment was unable to entrain the circadian clock.


Introduction
The circadian pacemaker in the suprachiasmatic nuclei (SCN) of the hypothalamus is synchronized daily to the 24-h light-dark cycle primarily via blue-light-sensitive melanopsin-containing retinal ganglion cells that project to the SCN via the retinohypothalamic tract (RHT) [1]. Most totally blind individuals have a nonfunctional RHT, and thus the circadian clock cannot be entrained by light and reverts to its intrinsic non-24-h period, leading to disruption of circadian rhythms, including melatonin, cortisol, sleepwake cycles, alertness, and performance [2][3][4]. The resultant clinical disorder, Non-24-Hour Sleep-Wake Disorder (N24HSWD), is characterized by cyclic episodes of poor nighttime sleep, an increased frequency and duration of daytime sleep, and disruptions in daytime alertness and performance [3,4]. While we and others have shown that daily melatonin administration can entrain the circadian clock in blind individuals [5,6], we aimed to investigate whether caffeine could perform this role. Although a direct phase-shifting effect of caffeine was not demonstrated in one study in mammals [7], several in vitro studies suggest that caffeine can phase-shift the clock [8][9][10][11] with a phase response curve similar to that of light [12,13]. We therefore hypothesized that morning caffeine administration (CT1-4) could induce the daily phase advance required to reset the circadian clock in the majority of totally blind people [14], and have direct stimulant benefits on daytime alertness and performance [15,16] while minimizing the negative impact on nighttime sleep.

Materials and methods
Three blind males (63.0 ± 7.5 years) who were habitual caffeine users (6-15 cups of tea/coffee per day) and had no light perception according to self-report (n = 2; S84 and S85) or no eyes (n = 1; S33) were studied at home. All participants were healthy and drug-free (Supplementary Table S1). Sequential urine samples were collected every~4 h (~8 h overnight) for 48 h every 1-2 weeks and assayed for 6-sulphatoxymelatonin (aMT6s), the major urinary metabolite of melatonin and a reliable marker of the circadian clock, by radioimmunoassay [17] (Stockgrand Ltd., University of Surrey, Guildford, UK). Each participant was studied for at least two circadian beat cycles based on circadian period (τ) estimates over four weeks of screening (Supplementary Materials). The period of the aMT6s rhythm was assessed using a regression analysis [2] (Origin 8.5 Pro, OriginLab Corporation, Northampton, MA, USA) weighted by the inverse of the squared standard error of the cosinor-derived acrophase (peak) times. Participants completed daily sleep-wake logs (Supplementary Materials) [3,18]. Alertness and mood were assessed using four nine-point scales (alert-sleepy, cheerful-miserable, calm-tense, and depressed-elated) every~2-4 h while awake on urine sampling days [4].
Caffeine (150 mg fast-release preparation; Martindale Pharmaceuticals, UK) was administered daily at 10 a.m. uninterrupted for approximately one circadian beat cycle in a single-masked design. Caffeine treatment was scheduled to be initiated at CT1-4 at or just after each participant reached a normal circadian phase (ie, aMT6s peak = 04:30 a.m.). Placebo was also administered in S33 and S85 for approximately one circadian beat cycle split before and after caffeine treatment (see Fig. 1); S84 received caffeine only. The study was approved by the University of Surrey Ethics Committee (EC/ 2003/144/SBMS). Written informed consent was obtained prior to the study and participants were informed that they were free to withdraw at any time.
The mean and standard deviation of each sleep-wake and mood parameter were calculated for each condition for each participant (PROC MEANS, SAS v9.2; SAS Institute, Cary, NC, USA). Statistically significant differences between conditions were calculated using the Wilcoxon Rank-Sum Test (PROC NPAR1WAY, SAS v9.2). The Students' t-test was used to assess whether τ differed significantly from 24 h, whether participants were entrained to the 24-h day, and to compare τ between conditions. A general linear model (PROC GLM in SAS 9.2) was used to compute statistical differences between treatment conditions (placebo/no treatment vs. caffeine) across all individuals with respect to circadian phase [calculated as the difference between the midpoint of the sleep episode (for sleepwake parameters) or the time of the mood assessment (for alertness/ mood) and the aMT6s acrophase divided by the estimated circadian period and multiplied by 360 (degrees), in 45°bins] (α < 0.05 after Bonferroni correction). Post-hoc comparisons for significant treatment-by-phase interactions were conducted using PROC MULTTEST (SAS v9.2).
The direct effects of caffeine on sleep were variable across participants, with shorter sleep latencies in two participants compared with placebo/no treatment (p = 0.007 and 0.02 for S33 and S85, respectively), but longer latencies during caffeine in S84 (p = 0.01) ( Table 1). Nighttime and 24-h sleep duration were significantly different across all three conditions for S33 (p = 0.02 for both). Post-hoc comparisons showed significant differences between the no treatment and placebo conditions (p = 0.03 and 0.01, respectively): total sleep amounts were lower during no treatment, which coincided with an adverse circadian phase (Fig. 1A). Although sleep offset was significantly different across conditions for S33 (p = 0.03), post-hoc comparisons revealed no pairwise differences ( Table 1).
As expected, significant circadian rhythms were observed in nighttime awakening duration, nighttime sleep duration (Fig. 1D), sleep offset (Fig. 1E), number of naps, and duration of naps ( Fig. 1F) (both placebo/no treatment and caffeine, all p < 0.03) and sleep onset and sleep quality (caffeine only, p = 0.0006 and 0.0005, respectively), but no significant main effect of treatment or treatment-by-phase interaction was observed for any sleep-wake parameters.
Alertness and mood changes were not consistent across participants (Table 1); caffeine improved alertness and cheerfulness in S33, and S84 and S85 rated themselves calmer during caffeine treatment (Table 1). A significant circadian rhythm was observed in alertsleepy scores for both the placebo/no treatment (p = 0.04) and caffeine (p = 0.0004) conditions when averaged across participants, with peak sleepiness occurring during the biological night at the aMT6s acrophase (Fig. 1G). There was a significant groupalerting effect of caffeine (p = 0.0005), with interaction effects at 135°( p = 0.03) and 225°(p = 0.002), equivalent to~1.30 p.m. and~7.30 p.m., respectively, under normal entrainment (ie, sleep at night, awake during the day); caffeine significantly increased alertness when participants were awake at these adverse circadian phases. A similar trend was observed for the circadian rhythm in cheerfulmiserable assessments (p = 0.07); participants rated themselves more cheerful during caffeine treatment. No significant circadian rhythms were observed in the calm-tense or depressed-elated scores.

Discussion
Daily administration of 150 mg of caffeine was unable to entrain the non-24-h rhythms in any of the three totally blind individuals studied. These results indicate that a daily 150 mg dose of caffeine at 10.00 a.m. is not effective as a circadian entraining agent. Our results do show, however, that the morning administration of caffeine directly mitigates some of the negative impact of nonentrained rhythms on daytime alertness and mood, without addressing the underlying circadian disorder.
The strength of this study is that the circadian effects of caffeine were studied in non-entrained individuals and in the absence of light, considered a gold-standard approach for assessing circadian rhythm entrainment. Although we cannot exclude the possibility that caffeine had a small resetting effect at some phases of administration that exceeded the limit of detectability in this study, we can state definitively that caffeine at this dose and duration of administration did not entrain the circadian clock, which was the primary aim of the experiment, as evidenced by a failure to observe a change in the intrinsic circadian period during the caffeine phase of the protocol. While the study was limited to a small number of cases, the distinct non-entrained phenotype would have permitted clear evidence of entrainment, if present, at this single dose and preparation even with this limited number of participants. The response to other doses may vary, however, and we did not screen for interindividual differences in caffeine sensitivity (eg, ADORA2A gene polymorphisms [19,20]). Despite exposure to multiple additional non-photic time cues (eg, activity, meal timing, exercise, and alcohol) [21,22], there was also no evidence of entrainment to the 24-h day during no treatment and placebo conditions, consistent with our previous work [2,5,23].
Although we attempted to initiate caffeine treatment at the same circadian phase across participants, post-hoc analysis of the initial treatment phase showed that it was not consistent (Fig. 1). Administering caffeine for most or all of a circadian beat cycle ensured that caffeine was administered during both the advance and delay phases of any theoretical phase response curve, however, thereby mitigating this confound. Even among participants who did not receive treatment for a full beat cycle (S33 and S85), all participants received treatment for almost all the phase advance portion of the phase response curve. If the caffeine phase response curve is similar to that of light as we assumed [12,13], then we would expect to see evidence of phase advances, such as a shortened circadian period, even in the absence of full entrainment, which we did not observe [23]. In addition, although we used sleep-wake logs rather than actigraphy recordings to measure nocturnal sleep and daytime naps, prior studies in blind individuals [3,18] have reported a good correlation between the two methods to measure the timing and duration of the sleep episode (r = 0.48-0.88) [18].
In conclusion, daily administration of 150 mg caffeine failed to entrain the circadian pacemaker in totally blind patients. Caffeine was able to increase alertness and mood directly, however, and may therefore be a useful adjunct therapy to provide temporary relief for the sleepiness symptoms of N24HSWD in the absence of appropriate treatment with a circadian regulator.

Conflict of interest
Neither author has any conflicts of interest directly associated with the current study. Dr. Lockley was the principal investigator of two recently completed and one ongoing sponsored clinical trials of a melatonin agonist for the treatment of non-24-hour sleepwake disorder in the blind, sponsored by Vanda Pharmaceuticals. Inc., and has received an investigator-initiated research grant and two service agreements from Vanda Pharmaceuticals, Inc., related to non-24-hour rhythms in the blind. He has also received minor consulting fees from 14 financial companies related to non-24hour sleep-wake disorder in the blind and the publicly available clinical trial results. He has also received honoraria from MediCom Worldwide, Inc., for teaching on a CME course sponsored by Vanda Pharmaceuticals, Inc.; for contributing text about non-24-hour sleepwake disorder for the National Sleep Foundation and textbook chapters published by Elsevier; and in 2007 received an authorship fees from Servier Inc., for writing a review of circadian rhythm disorders in the blind.
Dr. Lockley also reports receiving consulting fees in the past from American Family, Apollo Lighting, Brigham and Women's Hospital, Naturebright, Thomas Jefferson University, Warwick Medical School, and Wyle Integrated Science and Engineering and Wyvern Funds; and holds current consulting contracts with Headwaters, PlanLED and Wyle Integrated Science and Engineering; unrestricted equipment gifts from Bioilluminations LLC, Bionetics Corporation, and ResMed Inc.; an unrestricted monetary gift to support research from Swinburne University of Technology, Australia; a fellowship gift from Optalert, Pty, Melbourne, Australia; equity in iSLEEP, Pty, Melbourne, Australia; advance author payment and royalties from Oxford University Press; honoraria for written articles by AMO Inc., and the Wall Street Journal; honoraria plus travel, accommodation or meals for invited seminars, conference presentations or teaching from 2nd International Symposium on the Design of Artificial Environments; American Society for Photobiology; Bassett Research Institute; Brookline Adult Education; Brown Table 1 Average sleep and mood parameters by participant across placebo and caffeine conditions.