Circadian phase determined from melatonin profiles is reproducible after 1 wk in subjects who sleep later on weekends Introduction In humans, the circadian rhythm of melatonin is often used as a marker for the phase of the circadian clock [1], but melatonin must be sampled in dim light because light suppresses melatonin secretion [2]. A number of different points on the melatonin profile are used as phase markers of the clock, including the dim light melatonin onset (DLMO), the dim light melatonin offset (DLMOff) and the midpoint between the DLMO and DLMOff. Phase shifting studies, such as simulated and real night work and jet lag studies frequently involve sleep deprivation because the timing of sleep is shifted to daytime hours following night shifts or to night time in the new time zone. Bright light, dark and exogenous melatonin are often used to phase shift the clock, but they must be applied at appropriate times to produce the desired shift. Ideally, baseline phase should be determined immediately before any phase-shifting stimuli are applied, but if the phase assessment session involves subjects remaining awake overnight then this will result in the subjects being in a sleep-deprived state when the treatment begins. In addition to obvious burdens to the subjects, it is also possible that sleep deprivation itself could attenuate the magnitude of the phase shift as shown in rodents [e.g. 3]. A further disadvantage of measuring the baseline phase position on the day before treatment is that assays for melatonin take a few days so the baseline phase position can only be calculated retrospectively and thus the treatment cannot be precisely timed. Two strategies would allow baseline phase to be deter- mined from saliva samples whilst ensuring that the subjects begin treatment in a non-sleep deprived state. A shortened phase assessment, with samples only being taken in the early evening to determine the timing of the DLMO, would not require the subjects to stay awake overnight [e.g. 4]. The major disadvantage of this design is that if the whole melatonin profile is not collected then the DLMOff and midpoint cannot be determined. An alternative option is to carry out the baseline phase assessment several days before treatment and allow subjects to recover from any sleep deprivation. In addition, this would allow time for the samples to be assayed for melatonin and for baseline phase to be calculated. This method relies on the assumption that circadian phase does not change in between the baseline phase assessment and the first day of treatment, or if it does that it returns to baseline before the treatment begins. To test this alternative, we carried out two circadian phase assessments 1 wk apart. Subjects followed a pre- scribed sleep and outdoor morning light schedule before and in between the two sessions which were both on Tuesdays. Subjects were allowed to sleep 1 hr later and nap on weekends. We show that given these conditions circa- dian phase was very similar in the two sessions. Abstract: The aim of this study was to determine whether circadian phase from salivary melatonin profiles is the same when measured in phase assessments 1 wk apart. Eleven healthy young men and women maintained a fixed, home sleep–wake schedule, in bed, in the dark 23:00–07:00 hr on weekdays. On Friday and Saturday nights they were permitted to wake up and go to bed up to 1 hr later, and on Saturdays and Sundays they could nap between 13:30 and 16:30 hr. The study was run in the summer. Subjects wore sunglasses when outside during the day, and went outside for at least 15 min between 08:00 and 09:00 hr each morning. They maintained this schedule for 15 days before the first assessment and the 6 days in between the two assessments. During the assessments subjects remained awake overnight in <5 lux and gave saliva samples every 30 min. A recovery nap (13:00– 17:00 hr) followed the first session. The dim light melatonin onset (DLMO), offset (DLMOff) and midpoint were used as phase markers. There was minimal change in their timing between the two phase assessments. The average absolute change in midpoint (the change in phase regardless of direction) was 20 min. There was a small, 30 min delay in the DLMO. Thus, circadian phase can be measured a week in advance of any phase shifting intervention and, as long as the prescribed sleep and morning light schedule is maintained, the phase at the start of treatment can be confidently estimated. Victoria L. Revell 1 , Hyungsoo Kim 1 , Christine Y. Tseng 1 , Stephanie J. Crowley 2 and Charmane I. Eastman 1 1 Biological Rhythms Research Lab, Rush University Medical Center, Chicago, IL; 2 Department of Psychology, E.P. Bradley Hospital Sleep and Chronobiology Research Laboratory, Brown University, Providence, RI, USA Key words: melatonin, morning light, sleep Address reprint requests to Dr Victoria Revell, Biological Rhythms Research Lab, Rush University Medical Center, 1645 W Jackson Blvd, Suite 425, Chicago, IL 60612, USA. E-mail: victoria_l_revell@rush.edu Received December 21, 2004; accepted March 21, 2005. J. Pineal Res. 2005; 39:195–200 Doi:10.1111/j.1600-079X.2005.00236.x Copyright Ó Blackwell Munksgaard, 2005 Journal of Pineal Research 195