Independent Contributions of Nocturnal Hot Flashes and Sleep Disturbance to Depression in Estrogen-Deprived Women

Abstract

Context: Women are at increased risk for mood disturbance during the menopause transition. Hot flashes (HFs), sleep disruption, and fluctuating estradiol levels correlate with menopause-associated depression but co-occur, making cause and effect relationships difficult to disentangle.

Objective: Using a GnRH agonist (GnRHa) experimental model, we investigated whether depressive symptoms are associated with HFs and/or are explained by concomitant sleep fragmentation in the absence of estradiol fluctuation.

Design and intervention: Depressive symptoms, objective polysomnographic sleep parameters, subjective sleep quality, serum estradiol, and HFs were assessed before and 4 weeks after open-label depot GnRHa (leuprolide 3.75-mg) administration.

Setting: Academic medical center.

Participants: Twenty-nine healthy nondepressed premenopausal volunteers (mean age, 27.3 years).

Results: Serum estradiol was rapidly and uniformly suppressed. HFs developed in 69% of the subjects. On univariate analysis, worsening of mood was predicted by increases in time in light sleep (stage N1), number of transitions to wake, non-REM arousals, subjective sleep quality, and reductions in perceived sleep efficiency (all P < .045), as well as the number of nighttime (P = .006), but not daytime (P = .28), HFs reported. In adjusted models, the number of nighttime HFs reported, increases in non-REM arousals, time in stage N1, transitions to wake, and reduced sleep quality remained significant predictors of mood deterioration (P ≤ .05).

Conclusions: Depressive symptoms emerged after estradiol withdrawal in association with objectively and subjectively measured sleep disturbance and the number of nighttime, but not daytime, HFs reported. Results suggest that sleep disruption and perceived nighttime HFs both contribute to vulnerability to menopause-associated depressive symptoms in hypoestrogenic women.

Trial registration: ClinicalTrials.gov NCT01116401.

Figure 1.

Box plots showing the increase in depressive symptoms from before to after GnRHa treatment as measured by within-subject change in MADRS score in relation to the number of subjectively reported nighttime (A) and daytime (B) HFs. Box plots show the distribution of the data within each group from the 25th to 75th percentile, with the median line and the whiskers representing the winsorized extreme values. Data are presented by HF frequency groupings for visual purposes using a median split to define those with more or less frequent symptoms at night and those with none (< five HFs during the entire follow-up period). P values reflect results from univariate linear regression models.

Figure 2.

Box plots showing the increase in depressive symptoms from before to after GnRHa treatment as measured by within-subject change in MADRS score in relation to changes in the proportion of time spent in stage N1 (A), the number of non-REM arousals (B), the number of transitions to wake per hour (C), and sleep quality on the PSQI (D). P values reflect results from bivariate linear regression models, adjusting for nighttime HF frequency. Box plots show the distribution of the data within each group from the 25th to 75th percentile, with the median line and the whiskers representing the winsorized extreme values. Data are presented according to whether the PSG sleep parameter increased or did not increase after GnRHa therapy (A–C) and whether PSQI scores increased by ≥ 2 points or less after GnRHa therapy (D).