Morning Circadian Misalignment Is Associated With Insulin Resistance in Girls With Obesity and Polycystic Ovarian Syndrome

Abstract

Context: To our knowledge, circadian rhythms have not been examined in girls with polycystic ovarian syndrome (PCOS), despite the typical delayed circadian timing of adolescence, which is an emerging link between circadian health and insulin sensitivity (SI), and decreased SI in PCOS.

Objective: To examine differences in the circadian melatonin rhythm between obese adolescent girls with PCOS and control subjects, and evaluate relationships between circadian variables and SI.

Design: Cross-sectional study.

Participants: Obese adolescent girls with PCOS (n = 59) or without PCOS (n = 33).

Outcome measures: Estimated sleep duration and timing from home actigraphy monitoring, in-laboratory hourly sampled dim-light, salivary-melatonin and fasting hormone analysis.

Results: All participants obtained insufficient sleep. Girls with PCOS had later clock-hour of melatonin offset, later melatonin offset relative to sleep timing, and longer duration of melatonin secretion than control subjects. A later melatonin offset after wake time (i.e., morning wakefulness occurring during the biological night) was associated with higher serum free testosterone levels and worse SI regardless of group. Analyses remained significant after controlling for daytime sleepiness and sleep-disordered breathing.

Conclusion: Circadian misalignment in girls with PCOS is characterized by later melatonin offset relative to clock time and sleep timing. Morning circadian misalignment was associated with metabolic dysregulation in girls with PCOS and obesity. Clinical care of girls with PCOS and obesity would benefit from assessment of sleep and circadian health. Additional research is needed to understand mechanisms underlying the relationship between morning circadian misalignment and SI in this population.

Trial registration: ClinicalTrials.gov NCT02157974 NCT02585830.

Copyright © 2019 Endocrine Society.

Figures

Figure 1.

Salivary melatonin concentrations by clock time for PCOS (open squares, dotted line) and control subjects (black circles, solid line). Rectangles represent actual sleep times for each group. Results from a piecewise linear mixed-effects model examining differences in melatonin values between groups over time show that (A) the overall change in trajectory of melatonin concentration was significantly different between groups; (B) the trajectory of melatonin increase from 7 pm to midnight was significantly higher for the PCOS group than the control group; and (C) the trajectory of melatonin concentration decrease from midnight to 12 pm was significantly higher for the PCOS group compared with the control group. Serial saliva sampling was conducted every 60 minutes during waking hours from 7 pm to 12 pm the following day; in addition, the CIRC cohort (i.e., obese control subjects only) underwent hourly sampling overnight. Thus, values from 1 am to 4 am are not present in the figure for the PCOS group. Because PCOS participants were not woken to provide samples, only one participant with PCOS contributed a saliva sample at midnight; hence, the lack of error bars at this time point. The number of participants with samples at each time point ranged from one to 25 in the PCOS group and 11 to 25 in the control group.

Figure 2.

PCOS status is associated with morning circadian misalignment. Analysis of covariance shows that the relationship between DLMOn and sleep timing measured by (A) bedtime and (B) wake time does not differ between PCOS (open squares, dotted line) and control subjects (black circles, solid line). By contrast, DLMOff is later relative to sleep timing measured by (C) bedtime and (D) waketime in PCOS compared with obese control subjects. P values indicate significance of differences between groups.

Figure 3.

Regression analyses for melatonin variables and free testosterone. (A) Later DLMOff, (B) longer duration of melatonin secretion, (C) wider PA between bedtime and DLMOff (indicating longer sleep duration during the biological night), and (D) waketime and DLMOff (indicating melatonin secretion continuing after wake time) were associated with higher concentrations of free testosterone.

Figure 4.

Regression analyses for melatonin and SI. Wider PA between waketime and DLMOff (indicating melatonin secretion continuing after wake time) was associated with worse SI as measured by higher HOMA-IR.