Sleep + Light Interventions that Shift Melatonin Rhythms Earlier Improve Peri- and Post-menopausal Depression: Preliminary Findings

Abstract

Objective:

Testing the hypothesis that a sleep+light intervention which phase-advances melatonin rhythms will improve peri-post menopausal (P-M; by FSH) depression.

Methods:

In at-home environments, we compared two contrasting interventions: (1) an active, phase-advance intervention (PAI): one night of advanced/restricted sleep from 9pm-1am, followed by 8 weeks of morning bright white light (BWL) for 60 min/day within 30 min of awakening, and (2) a control, phase-delay intervention (PDI): one night of delayed/restricted sleep (sleep 3–7am) followed by 8 weeks of evening BWL for 60 min/day within 90 min of bedtime. We tested 17 P-M participants, 9 normal control-NC and 8 depressed-DP (by DSM-5 criteria).

Clinicians assessed mood by structured interviews and subjective mood ratings. Participants wore actigraphs to measure sleep and activity, and collected overnight urine samples for the melatonin metabolite, 6-sulfatoxymelatonin (6-SMT), before, during and after interventions.

Results:

Baseline depressed mood correlated with delayed 6-SMT offset time (cessation of melatonin metabolite (6-SMT) secretion) (r = +.733, p=.038). After PAI vs PDI, 6-SMT Offset (start of melatonin and 6-SMT decrease) was significantly advanced in DPs (Mean ± SD = 2 h 15 min ± 12 min, p=.042); advance in 6-SMT acrophase (time of maximum melatonin and 6-SMT secretion) correlated positively with mood improvement (r = +.978, p=.001). Mood improved (+70%, p=.007) by both 2 and 8 weeks.

Conclusions:

These preliminary findings reveal significantly phase-delayed melatonin rhythms in DP vs NC P-M women. Phase-advancing melatonin rhythms improves mood in association with melatonin advance. Thus, sleep + light interventions may potentially offer safe, rapid, non-pharmaceutical, well-tolerated, affordable home treatments for P-M depression.

INTRODUCTION

Risk for new onset and recurrence of mood disorders increases during the marked ovarian hormonal change of the peri-menopause (P-M).^1,2–7 As estradiol advances and progesterone delays circadian rhythms (CR), enhancing consolidation and entrainment capacity in females,^8–12 we hypothesize that P-M hormonal fluctuation¹³ dysregulates CRs, precipitating mood, sleep and activity disorders. Subsyndromal or untreated affective lability may progress to a major depressive disorder, increasing insomnia, suicide, osteoporosis and cardiovascular risks.^14–24 Previously we documented CR disturbances in P-M depressed participants (DP) vs normal controls (NC): In plasma melatonin, the best available measure of human circadian rhythmicity, we observed delayed melatonin offset time and elevated morning melatonin levels²⁵ which correlated with poor subjective sleep. In pilot work, we corrected CR disturbances with critically-timed sleep and light interventions (SALI) using time-restricted sleep (wake therapy) and light interventions to improve mood within 1–2 weeks; improvement correlated with the amount that melatonin was temporally phase-advanced (shifted earlier).

We hypothesize that in P-M DP with phase-delayed melatonin CR, a phase-advance intervention (PAI) will improve mood and sleep more than a phase-delay intervention (PDI), and that the amount of symptom improvement will correlate with the amount of melatonin phase-advance. These hypotheses are based on models^26–29 positing that disturbed CR phase (i.e., timing) contributes to mood and sleep/activity dysfunction, and that correcting CR disturbances improves these functions.³⁰ We previously found that advancing sleep to 9pm (vs delaying it to 3 am)^31,32 and restricting sleep to 4 hours for a single night improved mood and sleep significantly; and morning (AM) bright white light (BWL) advanced, while evening (PM) BWL delayed melatonin CRs.³³ As evidenced by extensive studies and Cochrane reviews, sleep and light interventions improve function in mood disorders.^34–53 Combined SALI enhances circadian organization by shifting timing and amplitude of sleep and melatonin CRs. Using either intervention alone provides limited benefits: wake therapy benefits often are lost after subsequent sleep, but can be sustained by light treatment; and wake therapy hastens and potentiates light benefits that otherwise require weeks to show significant efficacy in non-seasonally depressed women. Substantial evidence confirms the value of combined SALI in mood disorders.^39–53 Our approach responds to a call for combined strategies to improve treatment responses. Specifically, we hypothesized that 1) the Active PAI: phase-advanced restrict sleep (sleep 9pm-1am) for a single night, followed by phase-advancing AM BWL for 2 weeks, would improve mood and sleep compared with the Control PDI: phase-delayed restricted sleep (sleep 3–7am) for a single night, followed by phase-delaying PM BWL for 2 weeks; and 2) symptom improvement magnitude would correlate with melatonin phase-advance magnitude. The control PDI is based on our data showing phase-delayed sleep alone (without light treatment) does not modify mood or phase-shift melatonin significantly, and PM vs AM BWL has less effect on melatonin CRs.^31,33 We extended our plasma melatonin and sleep polysomnography findings obtained in hospital environs to the primary urinary melatonin metabolite, 6-sulfatoxy melatonin (6-SMT), and to actigraphy measures obtained in more ecologically valid home environs.⁵⁵ We tested the following hypotheses (H):

H1:

The Active PAI will phase-advance 6-SMT timing (acrophase, onset, or offset) significantly more than the Control PDI;

H2:

PAI will improve mood significantly more than PDI;

H3:

Magnitude of 6-SMT phase-advance will correlate significantly with magnitude of symptom improvement.

METHODS

Participants

We identified women as peri-menopausal or post-menopausal based on the classification criteria developed by the Study of Women’s Health Across the Nation (SWAN),⁵⁶ the timing of their final menstrual period (FMP), and verified by Follicle Stimulating Hormone (FSH). DP met Diagnostic and Statistical Manual-5 (DSM-5) criteria for a depressive disorder.⁵⁷ Participants had their primary care physicians provide written consent for their participation; had a physical examination and laboratory tests for a chemistry panel, thyroid indices, complete blood count, urinalysis and a normal PAP smear and mammogram within the last year; were non-smokers without significant medical illness; were off medication that would interfere with the melatonin measures, and off hormone therapy for at least three months. Participants with bipolar illness were excluded. DP were off antidepressant medication ≥ two weeks (four weeks for fluoxetine) prior to entering the study. NC participants had no current history of psychiatric illness. DP and NC participants were without alcohol abuse within the last year.

All participants provided written informed consent after all the procedures, approved by the University of California San Diego (UCSD) Internal Review Board, had been explained fully.

Recruitment, Screening, Baseline Measures

We recruited participants through UCSD clinic referrals and newspaper advertisements. During a 2-week screening evaluation, we obtained informed consent, screening questionnaires, psychometric and diagnostic assessments (Structured Clinical Interview for DSM-5: SCID)⁵⁸ and a urine toxicology screen to rule out substance use (week 1); continued psychometric assessments, and provided participants forms and equipment necessary to collect baseline measures of actigraphy (for 7 consecutive 24-hr periods) and overnight urinary 6-SMT (for 18 hours)(week 2), starting at least 7 days prior to the sleep intervention night. Participant blind randomization, half assigned to the PAI and half to the PDI, occurred between screening weeks 1 and 2. After screening, a psychologist or trained mental health professional with inter-rater reliability ≥ .9 obtained weekly (at about the same time (10am-2pm), a Hamilton Rating Scale for Depression (HRSD)⁵⁹ as part of a Structured Interview Guide for the HRSD with an Atypical Depression Supplement (SIGH-ADS,⁶⁰ a validated objective, interview-based mood assessment⁶¹) and a Mania Rating Scale-MRS⁶² (for safety, to document intervention-induced manic symptoms). Participants completed the Beck Depression Inventory (BDI)⁶³ (validated subjective mood assessment) and Beck Anxiety Inventory-BAI⁶⁴ (validated subjective anxiety assessment). For study inclusion, DP had mean scores on the HRSD ≥ 14; BDI ≥ 10, for two weeks. NC participants had no clinically significant mood changes during the month, and for study inclusion, mean HRSD ratings < 8, and BDI ratings < 5. Subjects also completed Horne and Östberg⁶⁵ “morningness” vs “eveningness” (MEQ) ratings.

Importantly, we designed the study arms to create identical sleep restriction durations (4 hours), plus identical bright light exposures (60 min/day), but at different times of day, to effect maximal CR realignment. Thus, the treatments varied only in the times of day when sleep restriction and light exposure were instituted, with PAI designed to advance 6-SMT timing vs. PDI, designed to delay.

We included “normal controls” to address two important questions: (1) What are the circadian rhythm abnormalities in menopausal DP vs NC at baseline? (2) If we “normalize,” i.e., correct the circadian rhythm abnormalities in menopausal DP with sleep + light interventions designed to challenge the circadian system with phase-shifts, so that they approach the values of menopausal NC participants, will that improve mood and sleep in DP? This approach allows us to determine if circadian rhythm abnormalities in menopausal depressed women are pathogenic, and can be targeted with sleep and light interventions to improve mood and sleep.

Study Flow

Sleep + Light Interventions:

After baseline collections, participants were randomly assigned to either (1) one night (4 hours) of Phase-Advance Sleep: Sleep 9pm-1am, followed by wake until the next habitual sleep time, or (2) one night (4 hours) of Phase-Delay Sleep: Wake until 3am, followed by sleep until 7am plus wake until the next habitual sleep time. While awake at night, they telephoned a study line every 30 min to ensure compliance with sleep protocols and to aid in staying awake.

After the sleep intervention night, participants received the light intervention initially for eight weeks at home: Phase advance sleep was followed by morning (AM) bright white light (BWL) initially 60 min/day starting within 30 min of wake; Phase delay sleep was followed by evening (PM) BWL initially 60 min/day ending 30 min before bedtime. Subsequently, after finding that 2 weeks of 30 min/day was as efficacious as 8 weeks of 60 min/day when preceded by wake therapy, we reduced the light intervention to 2 weeks of 30 min/day to reduce participant burden. Participants sat in front of a portable (5.5” x 6.25”) Litebook® light box (an array of 60 cool white light-emitting diodes behind a clear plastic screen with an intensity of 1,350 lux and an irradiance of 2.41x10^-9w/cm² at a distance 21 inches (in.), and spectral emission peaks at 464 nm and 564 nm)(The Litebook Company Ltd., Alberta Canada). They did not stare directly at the light source as it could cause discomfort to some, but is not harmful. The distance of the participant from the light source was calibrated individually for each light box by use of a Meterman LM631 Digital Light Meter (Meterman Test Tools, Everett, WA) to ensure an intensity of 1,350 lux at 21 in. A research assistant provided a measuring tape to ensure proper distance between light source and the participant. Ambient light intensity and spectra throughout the day were documented by the Actiwatch Spectrum. We obtained psychometric evaluations the day of the sleep intervention, the next 2 days and weekly during the light intervention. See Fig. 1.

Fig. 1.

We obtained measures of objective and subjective mood and sleep quality, 6 SMT onset, acrophase and offset) at baseline, for 8 DP and 9 NC, and during, and after interventions for the women who completed all requirements of the study protocol. Study Flow: Overnight urine samples for 6-sultoxy-melatonin (6-SMT) were collected in week 2 (baseline) and subsequently after 2 weeks of light intervention. Actigraphy began 7 days before the Phase-Advance (PAI) or Phase-Delay Intervention (PDI), during the interventions, and ended 7 days post-intervention.

Measures and Devices

Sleep/Activity/Illumination:

We obtained objective measures of activity (maximum and average activity counts per 60-min epoch), sleep (sleep onset time, sleep end time, total sleep time, sleep latency, daytime naps defined as at least 10 min of no activity) and ambient illumination (lux) at baseline starting 7 days prior to the sleep intervention, continued during interventions to ensure appropriate sleep/wake and light exposure times, and ending 7 days post-light intervention using the Actiwatch Spectrum Plus® (Philips Respironics), a small (48.5 x 36.7 x 13.8 mm) lightweight (29.8 g with band) wrist-actigraphy device containing a piezoelectric linear accelerometer (sensitive to 0.003 g and above), log-linear photometric transducer (sensitive from <0.01 lux to >100,000 lux), 3 photon flux and irradiance color sensors in red, green, and blue bands of visible light, a microprocessor with 2 Mbits RAM memory allowing for 36 days continuous recording of 1-minute epochs. The illumination measurements were roughly log-linear ranging in intensity from below moonlight to the brightest summer day at noon. Subjective sleep: We assessed subjective sleep quality by the Pittsburgh Sleep Quality Index (PSQI),⁶⁶ Sleep Quality Ratings (SQR: a Visual Analogue Scale for sleep, rest, alertness)⁶⁷ and logs on timing, duration, severity and frequency of hot flashes (that may affect sleep quality)^68–70 before, during and after interventions.

6-SMT:

Participants recorded time and total volume of each urine voiding initially over 30+ hours from two overnight collections (as melatonin is secreted during the night, not the day), pre- and post-intervention. As pilot data confirmed high correlations between two adjacent nights of 6-SMT measures (Pearson r’s of p < .0001 for mesor (.926), amplitude (.925), acrophase (.974), onset (.951), offset (.814) and peak duration (.894)), we subsequently sampled overnight 6-SMT for only one night, starting at 6pm on one evening and ending at noon the next day, pre- and post-intervention, to decrease participant burden and enhance participant retention. Subjects collect two 2 mL aliquots from each voiding, which they froze at home for later transfer to two separate -70° C clinic freezers (located in separate buildings, each with independent backup power generators) where they were stored until assay. We trained participants to properly collect, accurately record time and volume, and store sample aliquots, instructing them to drink extra fluids (~200 mL every 2 h) to promote frequent voiding; provided an oversupply of duplicate sample vials with legible numbers on labels and vial tops to ensure correct sample vial identification with associated time and volume on the record sheet, and supplied 1000 mL plastic bottles for refrigerator storage of urine collected during night-time awakenings (so participants could postpone volume measures and transfer to freezer vials until morning). We have used these methods effectively for many years in home and laboratory settings.^71–74

Assays: 6-SMT^71–73 and FSH^75–78

6-SMT excretion was measured using Bühlmann 96-well ELISA kits (EK-M6S) obtained from ALPCO, Diagnostics, Ltd. (Windham, NH) as described previously.^71,72 At the usual dilution of 1:200, the analytical sensitivity of this EIA was 0.35 ng/mL and the functional least detectable dose for coefficient of variation (CV) < 20% was 1.3 ng/ml. When possible, all samples from an individual overnight collection were run at the same time on the same 96-well plate. Samples were assayed in duplicate or re-assayed at either increased or decreased dilution (1:25 to 1:3200) to obtain more accurate estimates, or to clarify irregular circadian patterns in excretion rate (ng/h) and outlying values. An advantage of urinary assays is the integration of pulsatile blood secretion, so that a much smaller number of assays are needed per 24 h. Also, we have known total volumes and voiding times with which to compute accurate excretion rates.

Calculations of 6-SMT timing parameters:

From the 6-SMT concentration, urine volumes and collection times, we computed 6-SMT excretion rate (ng/h) for each collection interval (between voidings). This value was associated with each 5-min epoch within the longer collection interval. We then computed circadian parameters from analyses of the full time series of 5-min intervals representing each 18 h overnight collection. The best-fitting 24-hour cosine curve was estimated with a least-squares technique, which yielded the times of onset (start of melatonin and 6-SMT increase), acrophase (peak melatonin and 6-SMT secretion time), offset (start of melatonin & 6-SMT decrease) amplitude (1/2 height) and mesor (mean excretion rate, ng/h) of the fitted curve. We used the actual ng/h curve to estimate, algebraically, nocturnal 6-SMT secretion onsets (from upward) and offsets (from downward) crossings of the cosine mesor; for visualization and statistical analyses, we transformed the data to a logarithmic scale⁷¹

FSH:

To document peri- or post-menopausal status (FSH > 25, 40 mIU/mL respectively) at baseline, participants collected a urine sample, repeated one week later, for qualitative assessment of urinary FSH via the Rapid Response™ FSH Menopause Test Cassette (BTNX Inc. Markham, ON, Canada). This test is a lateral flow immunoassay using a combination of antibodies plus a monoclonal anti-FSH antibody to selectively detect elevated FSH levels at concentrations of 25 mIU/mL or greater with 99.2% accuracy.

Statistical Analyses.

The between-subjects design was randomized, univariate (PAI vs PDI) and fixed effects; tests were two-tailed. The primary independent variable of interest was the critical timing of intervention (active PAI vs control PDI). The primary outcome measures were post-intervention differences from baseline in mood, sleep and 6-SMT timing (onset, offset or acrophase) in PAI vs PDI. For tests of H1 (whether PAI would advance 6-SMT timing more than PDI) and H2 (whether PAI would improve mood and sleep more than PDI) we used 2 X 2 Analyses of Variance and t-tests to compare the effects of independent variables PAI vs. PDI, and NC vs DP on mood, 6-SMT timing (onset, acrophase, offset) and sleep measures, before, during and after interventions. For H3 (whether the magnitude of 6-SMT advance was significantly correlated with magnitude of symptom improvement), we evaluated the relationships of mood and sleep changes to changes in 6-SMT timing using linear regression.

Mood, Sleep, 6-SMT:

We evaluated changes in (1) Mood, (2) Sleep, and (3) 6-SMT as a function of PAI vs PDI using MANCOVA. Further, we examined the relationships of 6-SMT changes to mood and to sleep changes using linear regression. We assessed change in subjective sleep quality (PSQI), and other outcome measures as a function of 6-SMT quantitative measures. Variables that were correlated with melatonin and/or sleep indices at the p< .10 level were included as covariates in analyses.

RESULTS

As not all enrolled participants completed all aspects of the protocol, we report the main findings from the available data, below.

In NC+DP combined, baseline SIGH-ADS depressed mood score was correlated significantly with 6-SMT Offset (r = +.575, p = .041); i.e., greater baseline depressed mood was associated with greater phase-delay in 6-SMT offset. In the 4 DP vs. 7 NC women for whom complete data were obtained, acrophase mean time was somewhat delayed in DP vs NC (4:45 + 1:09 vs 3:26 ± 0:47 hh:mm, p=.052; see Fig. 2); onset (22:54 ± 0:51 vs 22:48 ± 1:34) and offset (10:00 ± 1:31 vs 9:03 ± 1:04) of 6-SMT mean times were not significantly delayed (p>.05) in DP vs NC. Furthermore, in the women for whom data were available, 6-SMT measures in DP vs NC were comparable, statistically (all p > .05) for log mesor (2.70± .55 vs. 2.40 ± 0.43); log amplitude (2.51 ± .47 vs 2.37 ± 0.42); and duration (11 h 06 min ± 1h 22 min vs 10 h 15 min ± 1 h 38 min).

Fig. 2.

Schematic depicting relative advance from Baseline in 6-SMT Onset, Acrophase and Offset times and ng/hr values in High Symptom Severity (HSS/Depressed) vs Low Symptom Severity (LSS/Normal Control) perimenopausal women.

Intervention Effects on Outcome Measures

6-SMT:

A. The 4 DP women who completed the PAI active condition had a significant phase-advance in 6-SMT offset (mean ± SD = 2 h 15 min ± 12 min, p=.042) and a non-significant mean phase advance in 6-SMT onset (mean ± SD = 1 h 57 min ± 3 h 21 min, p=.081). B. The 4 DP women receiving the control condition (PDI) showed non-significant phase-delays in onset (mean = 1h 27 min ± 1 h 46 min, p>.05) and offset (mean = 51 min ± 38 min, p>.05). See Fig. 3:

Fig. 3.

Schematic depicting mean Pre- vs Post-Intervention 6-SMT onset, acrophase and offset times in Menopausal Depressed Participants. A. The active Phase-Advance Intervention (PAI) advanced the mean 6-SMT Acrophase from Pre- to Post-PAI (p = .042) B. The control Phase-Delay Intervention (PDI) did not significantly advance 6-SMT Acrophase (p > .05).

Baseline Measures

Demographics:

At baseline, of 17 participants (8 DP + 9 NC) who completed the study, the DPs all reported personal or family histories of depression, were significantly younger than NC (54.1 + 5.6 vs 60.5 + 6.4 y, p = .046); had somewhat fewer years past FMP Final Menstrual Period (3.6 ± 4.3 vs 7.3 ± 6.5, p >.05); and numerically lower mean MEQ scores (52.3 ± 9.4 vs 57.4 ± 13.7, p >.05) indicating they were somewhat more “evening” types than NC (lower MEQ score = more evening type). As expected, DP vs NC were significantly more depressed on the SIGH-ADS (25.5 ± 10.8 vs 6.0 ± 7.3, p=.001) and the BDI (19.0 ± 7.4 vs 2.7 ± 2.5, p=.001); had poorer global sleep quality scores (Psychological General Well-Being Index-PGWBI)79 (63.7 ± 11.9 vs 87.3 ± 12.7, p=.002); but were not significantly different in mean parity (2.0 ± 1.3 vs 2.3 ± 1.5 children, p>.05) or BMI (28.4 ± 5.0 vs 27.3 ± 4.9, p>.05), respectively. We collected participant data across all months in the year; 8 of 17 (4 NC, 4 DP) were tested in Spring/Summer; 9 of 17 (5 NC, 4 DP) were tested in Fall/Winter.

Intervention Effects on Outcome Measures

6-SMT:

A. The 4 DP women who completed the PAI active condition had a significant mean phase-advance in 6-SMT offset (mean ± SD = 2 h 15 min ± 12 min, p=.042) and a non-significant mean phase advance in 6-SMT onset (mean ± SD = 1 h 57 min ± 3 h 21 min, p=.081). B. The 4 DP women receiving the control condition (PDI) showed non-significant phase-delays in onset (mean = 1h 27 min ± 1 h 46 min, p>.05) and offset (mean = 51 min ± 38 min, p>.05). See Fig. 3:

Mood:

Importantly, in 5 DP women receiving active PAI, mood improvement correlated positively with advance in 6-SMT acrophase (r = +.978, p=.001; See Fig. 4), as hypothesized. Furthermore, in DP receiving the active PAI, mood improved significantly (+70%, p=.007) by both 2 weeks and 8 weeks; i.e., PAI was equally efficacious in improving mood after 2 vs 8 weeks (delta SIGH-ADS scores = 11.8+5.0 vs 8.7+7.4 after 2 vs 8 weeks, p=.28). In contrast, in 2 DP receiving the control PDI, mood was not altered significantly after both 2 and 8 weeks (p>.05); thus, PDI produced no significant change in acrophase, and no evidence of clinically significant worsening of mood.

Fig. 4.

Relationship of change in peak urinary melatonin metabolite (6-SMT Acrophase) to post-intervention objective improvement (SIGHADS score). Of N =7 DP, 5 women received PAI and 2 women received PDI. Closed circles denote changes after the active, Phase Advance Intervention (PAI: LWT+AM Bright White Light); open squares denote changes after the control, Phase Delay Intervention (PDI: EWT+PM Bright White Light).

Sleep Quality

Pittsburgh Sleep Quality Index (PSQI):

At baseline, DP vs NC had a higher mean global score (9.0 + 3.3 vs 4.7 + 0.9, respectively, p = .012); i.e., as expected, DPs reported poorer subjective sleep than NCs. After PAI, DP vs NC reported only a non-significant mean improvement in PSQI; after PDI, however, improvement in global sleep quality was significantly greater in DP vs NC (p < .05). These findings may indicate that both circadian rhythms, as targeted primarily by the PAI, as well as sleep, as targeted primarily by PDI, are needed to benefit menopausal DP.

Sleep logs:

DP vs NC reported significantly more frequent awakenings (1.43 ± 1.73 vs 0.60 ± 1.06, p < .05), but not longer mean durations of awakenings (1:02 ± 1:1 vs 01:00 ± 1:06 h, p >.05), respectively. Increased awakenings, with potential eye-opening, could mean functionally increased exposure to dim light at night, which may alter circadian clock function and impair sleep. The control PDI did not worsen sleep or mood in NC (p > .05).

Hot Flashes:

Using skin conductance measures, we previously found only a non-significantly higher frequency of hot flashes during wake times, stages 2 – 3, and REM sleep in NC vs DP. By sleep logs, hot flash severity and number were not significantly different between DP vs NC. Thus, our data do not support a contribution of hot flashes to DP reporting more awakenings than NC women.

DISCUSSION

Our findings supported our initial three hypotheses (H): H1: The Active PAI phase-advanced 6-SMT timing (offset) significantly more than the Control PDI; H2: PAI improved mood significantly more than PDI; H3: Magnitude of 6-SMT phase-advance (acrophase) correlated significantly with magnitude of symptom improvement.

In this pilot study, despite the relatively small sample, we found the following important results: 1) At baseline, greater depression ratings by interview-based assessments (SIGH-ADS) correlated with later 6-SMT offset time; 2) After the active PAI, DP had a significant phase-advance in 6-SMT offset, whereas the control PDI had no significant effect; 3) After the active PAI, mood improvement correlated significantly (r=0.978, p= .001) with phase-advance in 6-SMT acrophase; 4) The active PAI was equally efficacious after 2 and 8 weeks, whereas the control PDI was not. We discuss the implications of these findings below.

That peri- and post-menopausal depressed women have delayed urinary melatonin rhythms manifested more clearly in offset time than in onset time -- replicates our previously reported findings from obtained from frequently-sampled plasma levels.²⁵ As also previously observed in that study and in another report,⁸⁰ delayed offset time (and longer melatonin duration) was associated with increased body mass index (BMI), more Eveningness, and delayed sleep end time, all of which predicted greater depression ratings (HRSD). Such delayed circadian rhythms, manifested in delayed sleep end time, may prevent peri- and post-menopausal women from experiencing morning bright white light that is critical in diurnal humans for synchronizing internal circadian rhythms with the external environment. Such resulting desynchronization may contribute to the development of mood disturbances.²⁶ We also reported phase-delayed melatonin circadian rhythms in the symptomatic luteal compared with the asymptomatic follicular phase in Premenstrual Dysphoric Disorder ³¹ and in Postpartum Depression.⁸¹ Thus phase-delayed circadian rhythms may reflect a pathophysiology that contributes to depressed mood, increased BMI and altered sleep timing, all of which have untoward health consequences, but may be treatable by correcting the circadian rhythm disturbance as described below.

The active PAI, but not the control PDI, corrected the baseline phase-delay disturbance in 6-SMT by significantly advancing it. Advancing and restricting sleep alone for a single night (without the sustaining effect of light treatment) in and of itself has the capacity to improve mood³² and phase-advance melatonin rhythms.³¹ It also serves to hasten and potentiate the effects of light treatment, which in non-seasonal depression can take five to ten weeks to exert significant antidepressant effects.^82–86 Thus a relatively simple, rapid-acting sleep and light intervention that can be administered at home can correct the underlying circadian rhythm disturbance in peri- and post-menopausal depression, which if left untreated, can lead to such adverse consequences.

Furthermore, after the active PAI, mood improvement was highly correlated with the phase-advance in 6-SMT acrophase. The fact that this correlation was so robust (r = .978, p =.001) despite a relatively small N, suggests that the active PAI intervention is effectively targeting the underlying circadian rhythm pathophysiology. That a phase-advancing sleep and light intervention also improves mood in association with correcting phase-delayed rhythms in other women’s reproductively-related depressive disorders such as Premenstrual Dysphoric Disorder and Postpartum Depression⁸⁷ underscores that the sleep and light intervention must be critically-timed to address the specific circadian pathology. In contrast, for example, in pregnancy depression characterized by phase-advanced melatonin rhythms⁸¹ it is phase-delayed sleep and light interventions that improve mood.^32,87 Note also that in peri- and post-menopausal depression in the current study, the control PDI did not significantly alter either mood, sleep or 6-SMT rhythms. These findings, that peri- and post-menopausal DP had phase-delayed circadian rhythms at baseline, corrected by a PAI, but not a PDI, and that the magnitude of the phase-advance in 6-SMT timing measures was associated with the magnitude of mood improvement, are consistent with our conceptual model as illustrated in Figure 5.

Fig. 5.

Hypothetical conceptual model illustrating relationship of 6-SMT Offset timing to expected (A) baseline symptom severity, and (B) post-intervention symptom improvement after active (PAI; closed circles) and inactive control (PDI; open squares).

That mood improved significantly (+ 70%) after the PAI, but not the PDI, and equally after 2 and 8 weeks of treatment suggests that this chronotherapeutic regimen is rapid-acting, conferring benefits over antidepressant medication or psychotherapeutic effects that may take 12 weeks to exert efficacy, and may be limited by clinician availability. As Wirz-Justice et al. describe in Science, “Given the psychological suffering that depression inflicts, including the danger of suicide…, it is surprising how little notice is taken of these chronobiological interventions (e.g., sleep and light therapies).”⁸⁸ Treating sleep and circadian problems promotes mental health.⁸⁹

LIMITATIONS

The major limitation of this study is in the preliminary nature of the findings, due to the small sample of participants who completed the study, at a single test site. As with most preliminary findings, replication of these results in more rigorous studies with a larger N is essential. Future studies also need to include follow-up beyond the initial testing weeks to determine the durability of the intervention effects.

CONCLUSIONS

Despite the limitations of this pilot study, the findings point to the potential importance of targeting the chronobiologic basis of peri- and post-menopausal depression, using critically timed sleep and light interventions to correct circadian pathology, and thereby improve mood. With further study, these chronotherapeutic interventions may offer safe rapid-acting, non-pharmaceutical, well-tolerated, affordable, home treatments that can be applied readily by paraprofessionals across cultures, helping reduce health disparities in underserved populations.