Abstract
Study Objectives:
To examine the association between sleep midpoint and inflammation in a population with a large proportion of individuals diagnosed with obstructive sleep apnea syndrome (OSAS), a group that is already prone to increased inflammation.
Methods:
Subjects from the Cleveland Family Study underwent overnight polysomnography and completed surveys on sleep habits. Morning and evening blood samples were collected and assayed for proinflammatory biomarkers interleukin (IL)-1, IL-6, and tumor necrosis factor α (TNF-α). Linear regression models were used, adjusting for potential confounders and sleep duration.
Results:
The study population included 587 adults (52.3% with OSAS). Mean ± standard deviation weekday sleep midpoint was 3.52 ± 2.09 (3:31 am) and weekend sleep midpoint was 4.46 ± 1.69 (4:28 am). The Mean difference between weekday and weekend sleep midpoint (social jetlag) was 0.94 ± 2.08 hours. After adjusting for OSA severity, greater social jetlag was associated with higher levels of the inflammatory cytokine IL-1 (beta: 0.435 pg/mL, 95% confidence interval [CI]: 0.091 to 0.779). Additionally, later timing of sleep during both the weekdays and the weekends was associated with increased levels of IL-6 (weekday beta: 0.182 pg/mL; 95% CI: 0.013 to 0.350; and weekend beta: 0.188 pg/mL; 95% CI: 0.004 to 0.373). No trends were observed with TNF-α and any sleep exposure.
Conclusions:
Later sleep timing was associated with elevated levels of IL-6 while increased social jetlag was associated with elevated levels of IL-1. Our results indicate that later sleep schedules and increased social jetlag may lead to higher inflammation, even after controlling for OSA severity.
Citation:
Girtman KL, Baylin A, O’Brien LM, Jansen EC. Later sleep timing and social jetlag are related to increased inflammation in a population with a high proportion of OSA: findings from the Cleveland Family Study. J Clin Sleep Med. 2022;18(9):2179–2187.
Keywords: social jetlag, sleep midpoint, inflammation, cytokines
BRIEF SUMMARY
Current Knowledge/Study Rationale: Late bedtimes and variability in bedtimes could adversely impact health, potentially via increased inflammation. We sought to examine how sleep timing was related to inflammation after accounting for other sleep characteristics.
Study Impact: We found that later sleep timing and higher social jetlag were related to higher levels of certain inflammatory markers. Findings suggest that, independently of obstructive sleep apnea syndrome status and sleep duration, sleep timing characteristics may be related to inflammatory processes.
INTRODUCTION
In modern society, late bedtimes and variability in bedtimes, especially from the weekends to the weekdays (known as “social jetlag”), are widespread and disrupt underlying circadian rhythms. Many epidemiologic studies suggest that later sleep timing and social jetlag have a negative impact on cardiometabolic function.1–4 For example, the social misalignment of sleep timing and duration between the workdays and rest days has been related to increased obesity and cardiovascular diseases.3–6 Beyond metabolic health, later sleep timing is associated with a wide range of negative health outcomes, including adverse cognitive function, poor bone health, and lower physical activity.7–10 The mechanisms linking later sleep timing and social jetlag with adverse health outcomes are unclear but may be related to inflammation. There are extensive data supporting the association of sleep duration with increased inflammation; studies indicate both long sleep duration and short sleep duration are linked to higher levels of proinflammatory biomarkers such as cytokines interleukin (IL)-1, IL-6, and tumor necrosis factor α (TNF-α).11–13 Inflammation has also been well documented in one of the most common sleep disorders, obstructive sleep apnea (OSA) syndrome (OSAS).
In the United States, it has been estimated that somewhere between 10% and 17% of adults between the ages of 30 and 70 years have moderate to severe sleep-disordered breathing,14 a condition that has a wide overlap with OSAS. OSAS is associated with increases in systemic inflammation, sympathetic nervous system activation, oxidative stress, and endothelial dysfunction. Many cardiometabolic conditions, including hypertension, coronary artery disease, diabetes, atherosclerosis, and stroke, have strong and well-reported associations with OSAS.15–17 Extensive prior research indicates that OSAS could also be characterized as a low-grade inflammatory condition.18 A case-control study by Ciftci et al19 reports that serum levels of cytokines IL-6 and TNF-α, independent of body mass index (BMI), are significantly higher in OSAS-diagnosed cases than in controls. Thus, within a population that is already at higher risk of inflammation, and that represents a high proportion of adults,20 it is essential to investigate how other sleep behaviors could impact inflammation and ultimately cardiometabolic complications. However, the relationship between sleep timing, social jetlag, and inflammation has yet to be researched, especially when adjusting for OSAS severity. Specifically, we aimed to investigate whether poor sleep timing and increased social jetlag promote an inflammatory state in a population where the majority of individuals are diagnosed with OSAS. The research was conducted using inflammatory biomarkers among individuals from the Cleveland Family Study, a family-based cohort that assessed the genetics of OSAS. We hypothesized that later sleep timing (midpoint of sleep) and greater social jetlag (difference between workday and rest day sleep midpoint) would be associated with higher serum levels of proinflammatory cytokines IL-1, IL-6, and TNF-α.
METHODS
Subjects
The study population comprises participants from the Cleveland Family Study Cohort, which was originally designed to investigate the genetic epidemiology of sleep apnea and included 2,284 individuals from 361 families. Data collection and recruitment details have been previously published.21,22 In short, families were identified through an original proband with a laboratory-confirmed diagnosis of OSAS (a respiratory disturbance index ≥ 20 or apnea “considered severe enough to warrant therapy”23) and at least 2 relatives were recruited as well as control families from the same neighborhood. These families were studied on 5 occasions for up to 16 years. Extensive phenotyping including polysomnography (PSG) was completed during the fifth visit on a subset of 735 individuals selected based on their genetic informativity; specifically, pedigrees where siblings had extremes (high or low) of the apnea-hypopnea index (AHI), a measure of OSAS severity. Further questionnaires and biochemical sampling were conducted to collect information on sleep behavior in these participants. Details on participant selection have been previously described.21–24 Of the 735 participants, those who reported steady nightshift work or rotating nightshift work (n = 146) were excluded from this analysis. Two additional participants were excluded as they did not provide any inflammatory biomarker data, leaving 587 participants available for analysis. All participants provided written informed consent and protocols were approved by the institutional review board of the University Hospitals of Cleveland.
Sleep variables
Sleep midpoint was operationalized as the median of self-reported normal bedtime and waketime, separately for the weekend and the weekday. The midpoint difference between weekend and weekday was calculated as weekend midpoint of sleep − weekday midpoint of sleep. Reported sleep time was converted to decimal time before analysis.
Inflammatory biomarkers
Participant blood collection was performed both prior to sleep and upon awakening on the day of the PSG test. The first sample was obtained at 10:30 pm (evening draw) and the second at 8:00 am (morning draw). Blood samples were centrifuged within 30 minutes of collection, aliquoted, and stored at −70°C until assayed for IL-1β (IL-1), IL-6, and TNF-α at the University of Vermont Clinical Biochemistry Laboratory. Evening readings of IL-1 were not assessed.
Covariates
Potential confounders were selected based on prior literature and included the following: sex, age, BMI, race, household income, Mean caffeine consumption, smoking status, AHI, and sleep duration. Height and weight measures taken the morning after PSG were used to calculate BMI (kg/m2). Height was measured to the nearest millimeter using a wall-mounted stadiometer, and weight was measured to the nearest 0.1 kg using a digital scale. A categorical smoking variable classified each participant as “never having smoked,” “current smoker,” and “quit smoking.” Responses to the question, “On average, how often do you drink caffeine,” included “none,” “less than 1 cup a day,” “approximately 1 cup a day,” “more than 1 cup, less than 3 cups a day,” and “more than 3 cups a day.” Self-reported sleep duration was collected as hours of sleep per night, separately for the weekends and the weekdays. A weighted sleep duration variable was included in the final analysis and is calculated as follows: weekday sleep duration × 5/7 + weekend sleep duration × 2/7. AHI was included in the analysis as a measure of the severity of sleep apnea. AHI, as measured by overnight PSG, is defined as the number of all apneas and hypopneas (with ≥3% oxygen desaturation or arousal) per hour of sleep.
Statistical analysis
All inflammatory biomarkers were log-transformed prior to running regression analysis to generate a normal distribution. P values in Table 1 were calculated using the Kruskal-Wallis nonparametric test to compare potential confounders with sleep measures. The comparison between confounders and inflammatory biomarkers was completed using the analysis of variance (ANOVA) parametric test in Table 2. The potential for each of the sleep timing variables to predict cytokine levels was assessed using linear regression analysis using beta coefficients, 95% confidence intervals (CIs), and P values. To evaluate potential nonlinear relationships, all midpoint exposures were modeled categorically in quartiles. Each of the confounders was associated with at least 1 of the 3 exposures. An unadjusted model, demographics model, and fully adjusted model were completed for each of these analyses. The unadjusted regression models only included the sleep midpoint exposure and inflammatory biomarker outcome. Demographics models included covariates sex, age, household income, and race, as well as the main exposure and outcome. Fully adjusted regression models included all previously mentioned potential confounders, exposure, and outcome, as well as the covariates smoking status, BMI, Mean caffeine intake, and AHI. In addition, weekend sleep midpoint exposure models included the variable weekend sleep duration, and weekday sleep midpoint exposure models included variable weekday sleep duration. The difference between weekend and weekday midpoint exposures included the weighted Mean of sleep duration. Complete case analyses were conducted such that missing data were removed from regression analyses across unadjusted, demographics, and unadjusted models. The final analysis sample size was n = 422 (IL-1 models), n = 494 (IL-6 am and IL-6 pm models), n = 495 (TNF-α am models), and n = 351 (TNF-α pm models). All tests and analyses were performed using R version 4.0.2 (R Foundation for Statistical Computing).
Table 1.
Sociodemographic characteristics by sleep midpoint exposures.
| Baseline Sociodemographic and Lifestyle Characteristics | n (%) | Weekday Sleep Midpoint, Mean ± SD | Weekend Sleep Midpoint, Mean ± SD | Difference Between Weekend and Weekday Midpoint, Mean ± SD, h/night |
|---|---|---|---|---|
| Overall | 587 | 3.53 ± 2.09 (3:31 am) | 4.47 ± 1.69 (4:28 am) | 0.94 ± 2.08 |
| Sex | ||||
| Male | 263 (44.8%) | 3.70 ± 2.43 | 4.57 ± 1.80 | 0.88 ± 2.40 |
| Female | 324 (55.2%) | 3.37 ± 1.77 | 4.37 ± 1.60 | 1.00 ± 1.789 |
| Pa | .0600 | .1700 | .4910 | |
| Age group (years) | ||||
| < 20 | 101 (17.2%) | 4.05 ± 2.14 | 5.74 ± 1.93 | 1.68 ± 1.41 |
| 20 to < 35 | 108 (18.4%) | 3.74 ± 1.31 | 5.15 ± 1.54 | 1.42 ± 1.22 |
| 35 to < 50 | 150 (25.6%) | 3.15 ± 1.35 | 4.24 ± 1.46 | 1.09 ± 0.97 |
| 50 to < 65 | 145 (24.7%) | 3.53 ± 3.18 | 3.84 ± 1.31 | 0.31 ± 3.55 |
| ≥ 65 | 83 (14.1%) | 3.28 ± 1.27 | 3.55 ± 1.33 | 0.27 ± 0.62 |
| P | .0090 | <.0001 | <.0001 | |
| BMI (kg/m2) | ||||
| < 18.5 | 15 (2.6%) | 3.87 ± 2.93 | 5.10 ± 2.83 | 1.23 ± 1.13 |
| 18.5 to < 25 | 101 (17.2%) | 3.44 ± 1.46 | 4.80 ± 1.72 | 1.36 ± 1.37 |
| 25 to < 30 | 138 (23.5%) | 3.44 ± 1.43 | 4.52 ± 1.65 | 1.09 ± 1.15 |
| ≥ 30 | 333 (56.7%) | 3.56 ± 2.42 | 4.30 ± 1.62 | 0.74 ± 2.52 |
| P | .8210 | .0261 | .0401 | |
| Race/ethnicity | ||||
| Black | 315 (53.7%) | 3.52 ± 1.92 | 4.45 ± 1.73 | 0.93 ± 1.77 |
| White | 257 (43.8%) | 3.44 ± 2.22 | 4.35 ± 1.53 | 0.92 ± 2.43 |
| Other | 15 (2.6%) | 4.92 ± 2.83 | 6.51 ± 2.29 | 1.59 ± 1.33 |
| P | .1200 | .0084 | .5670 | |
| Household income | ||||
| < $10,000 | 87 (16.5%) | 3.92 ± 1.59 | 5.10 ± 1.79 | 1.19 ± 1.11 |
| $10,000 to $29,999 | 173 (32.8%) | 3.66 ± 2.20 | 4.46 ± 1.68 | 0.80 ± 2.20 |
| $30,000 to $49,999 | 117 (22.2%) | 3.33 ± 2.35 | 4.16 ± 1.60 | 0.83 ± 2.53 |
| ≥ $50,000 | 150 (28.5%) | 3.19 ± 2.02 | 4.20 ± 1.38 | 1.01 ± 2.23 |
| P | .0387 | <.0001 | .5150 | |
| Mean caffeine consumption per day | ||||
| None | 72 (12.3%) | 3.64 ± 1.40 | 4.69 ± 1.43 | 1.05 ± 1.15 |
| < 1 cup | 132 (22.5%) | 3.23 ± 1.31 | 4.30 ± 1.65 | 1.07 ± 1.24 |
| 1 cup (8 ounces) | 118 (20.1%) | 3.39 ± 1.45 | 4.56 ± 1.60 | 1.17 ± 1.04 |
| > 1 to < 3 cups | 159 (27.1%) | 3.50 ± 1.62 | 4.55 ± 1.74 | 1.05 ± 1.10 |
| > 3 cups | 106 (18.1%) | 4.00 ± 3.79 | 4.29 ± 1.91 | 0.29 ± 4.20 |
| P | .0672 | .5340 | .088 | |
| Smoking status | ||||
| Never | 325 (55.4%) | 3.44 ± 1.83 | 4.51 ± 1.60 | 1.06 ± 1.78 |
| Current | 134 (22.8%) | 3.84 ± 2.46 | 4.91 ± 1.99 | 1.07 ± 2.51 |
| Quit | 128 (21.8%) | 3.41 ± 2.29 | 3.91 ± 1.44 | 0.50 ± 2.25 |
| P | .7550 | .0115 | .0175 | |
| AHI (events/h) | ||||
| Q1 (median = 0.82) | 147 (25%) | 3.52 ± 1.67 | 4.89 ± 1.79 | 1.380 ± 1.125 |
| Q2 (median = 3.34) | 146 (25%) | 3.69 ± 2.74 | 4.60 ± 1.75 | 0.908 ± 3.172 |
| Q3 (median = 9.55) | 146 (25%) | 3.65 ± 2.37 | 4.30 ± 1.62 | 0.658 ± 2.169 |
| Q4 (median = 32.19) | 146 (25%) | 3.22 ± 1.27 | 4.05 ± 1.50 | 0.832 ± 1.035 |
| P | .2170 | .0001 | .0231 | |
| Weekday sleep duration (h/night) | ||||
| Q1 (median = 5.75) | 139 (24.3%) | 3.73 ± 2.80 | 4.47 ± 1.77 | 0.74 ± 3.13 |
| Q2 (median = 6.75) | 147 (25.7%) | 3.16 ± 1.24 | 4.37 ± 1.57 | 1.21 ± 1.17 |
| Q3 (median = 7.75) | 144 (25.2%) | 3.44 ± 2.02 | 4.21 ± 1.32 | 0.77 ± 2.31 |
| Q4 (median = 9.17) | 142 (24.8%) | 3.88 ± 2.03 | 4.88 ± 1.96 | 1.00 ± 1.04 |
| P | .0163 | .0061 | .1790 | |
| Weekend sleep duration (h/night) | ||||
| Q1 (median = 5.57) | 141 (24.7%) | 3.70 ± 2.74 | 4.23 ± 1.64 | 0.53 ± 3.02 |
| Q2 (median = 7.25) | 138 (24.2%) | 3.32 ± 1.40 | 4.18 ± 1.46 | 0.85 ± 0.97 |
| Q3 (median = 8.08) | 139 (24.4%) | 3.70 ± 2.32 | 4.56 ± 1.84 | 0.86 ± 2.36 |
| Q4 (median = 9.54) | 152 (26.7%) | 3.47 ± 1.71 | 4.92 ± 1.68 | 1.45 ± 1.27 |
| P | .3590 | .0003 | .0018 |
aP values represent at least 1 statistically significant difference between categories, calculated from nonparametric Kruskal-Wallis tests. AHI = apnea-hypopnea index, BMI = body mass index, Q = quartile, SD = standard deviation.
Table 2.
Association between categorical measures of sleep midpoint and cytokine IL-1 levels.
| Difference in log IL-1 (pg/mL) (n = 422) | Unadjusted Model | Adjusted Model* | ||
|---|---|---|---|---|
| Beta Coefficient (95% CI) | P | Beta Coefficient (95% CI) | P | |
| Weekday sleep midpoint (Decimal time) | ||||
| Q1 (median = 2.03) | Reference | Reference | ||
| Q2 (median = 2.78) | 0.11 (−0.20, 0.42) | .4840 | 0.12 (−0.19, 0.43) | .4506 |
| Q3 (median = 3.53) | −0.13 (−0.44, 0.18) | .4110 | −0.12 (−0.43, 0.21) | .4986 |
| Q4 (median = 5.18) | −0.04 (−0.36, 0.27) | .7970 | −0.10 (−0.43, 0.22) | .5316 |
| Weekend sleep midpoint (Decimal time) | ||||
| Q1 (median = 2.78) | Reference | Reference | ||
| Q2 (median = 3.78) | 0.20 (−0.11 0.52) | .2106 | 0.17 (−0.15, 0.49) | .3010 |
| Q3 (median = 4.78) | 0.29 (−0.01, 0.60) | .0624 | 0.20 (−0.12, 0.51) | .2306 |
| Q4 (median = 6.50) | 0.26 (−0.05, 0.56) | .0953 | 0.06 (−0.29, 0.42) | .7326 |
| Difference between weekday and weekend midpoint (hours) | ||||
| Q1 (median = 0.00) | Reference | Reference | ||
| Q2 (median = 0.50) | 0.30 (−0.02, 0.62) | .0684 | 0.28 (−0.05, 0.61) | .0959 |
| Q3 (median = 1.25) | 0.21 (−0.10, 0.51) | .1811 | 0.13 (−0.20, 0.45) | .4415 |
| Q4 (median = 2.25) | 0.58 (0.27, 0.88) | .0003 | 0.44 (0.09, 0.78) | .0135 |
*Variables apnea-hypopnea index, sex, age, smoking status, race, household income, body mass index, sleep duration, and caffeine intake are included in these models. CI = confidence interval, IL = interleukin, Q = quartile.
RESULTS
Among the 587 participants (55.2% female), 52.3% had an AHI greater than or equal to 5 events/h. Only 12% of the sample reported to use continuous positive airway pressure therapy regularly or during the sleep study. Other demographic characteristics of the study sample are displayed in Table 1. Mean ± standard deviation weekday sleep midpoint was 3.52 ± 2.09 hours (3:31 am) and weekend sleep midpoint was 4.46 ± 1.69 hours (4:28 am). Younger age, longer sleep duration, male sex, higher BMI, Black race/ethnicity, higher apnea-hypopnea severity, smoking, and a lower household income were associated with later midpoint of sleep, either during the week or the weekend (or both). The Mean difference between weekday and weekend sleep midpoint was 0.94 ± 2.08 hours, although more than one-quarter of the sample had < 15 minutes of difference from weekdays to weekends (29%). At the other extreme, 16% of the sample had > 2 hours of difference from weekdays to weekend. Longer sleep duration (weekend only), increased BMI, increased caffeine consumption, increased apnea-hypopnea severity, and being a smoker were associated with a larger difference between weekday and weekend midpoint. In general, there were positive associations between age, BMI, caffeine intake, smoking, AHI severity, and cytokine levels ( Table S1 (391.4KB, pdf) in the supplemental material).
Associations between sleep midpoint exposures and inflammatory biomarker outcomes
Unadjusted and adjusted analysis of midpoint exposures in relation to IL-1, IL-6, and TNF-α outcomes are displayed in Table 2, Table 3, and Table 4. Increased social jetlag was associated with elevated levels of IL-1, as those in the highest quartile of sleep difference had an 0.435-pg/mL inflammation increase (95% CI: 0.091, 0.779; P = .0135). Later weekday and weekend midpoints were associated with higher IL-6 am levels. To illustrate, those with the latest sleep midpoints had 0.18 pg/mL (95% CI: 0.01, 0.35) and 0.19 pg/mL (95% CI: 0.004, 0.37) higher IL-6 am levels compared with those with the earliest sleep midpoints. These associations persisted in models that additionally adjusted for weekday–weekend midpoint difference. In addition, there was a nonlinear positive relationship between weekday–weekend difference and IL-6 pm levels, such that those in quartile 3 of weekday–weekend difference had a 0.15-pg/mL (95% CI: −0.001, 0.31) higher IL-6 pm level. There was no significant relationship observed between TNF-α (am and pm) and any midpoint exposure. In addition, IL-1 was not associated with weekend or weekday sleep timing. Sensitivity analyses adjusting for comorbid cardiometabolic conditions (diabetes, cardiovascular disease, and hypertension) or for regular use of continuous positive airway pressure did not substantially alter effect estimates (data not shown). Finally, sensitivity analysis that included family membership as a random intercept in a linear mixed model did not alter findings other than that CIs became slightly narrower.
Table 3.
Association between categorical measures of sleep midpoint and cytokine IL-6 levels.
| Unadjusted Model | Adjusted Model* | |||
|---|---|---|---|---|
| Beta Coefficient (95% CI) | P | Beta Coefficient (95% CI) | P | |
| Difference in log IL-6 am (pg/mL) (n = 494) | ||||
| Weekday sleep midpoint (Decimal time) | ||||
| Q1 (median = 2.03) | Reference | Reference | ||
| Q2 (median = 2.78) | 0.13 (−0.05, 0.31) | .1485 | 0.18 (0.01, 0.37) | .0333 |
| Q3 (median = 3.53) | 0.08 (−0.11, 0.26) | .4260 | 0.07 (−0.10, 0.24) | .4122 |
| Q4 (median = 5.18) | 0.17 (−0.02, 0.35) | .0843 | 0.18 (0.01, 0.35) | .0349 |
| Weekend sleep midpoint (Decimal time) | ||||
| Q1 (median = 2.78) | Reference | Reference | ||
| Q2 (median = 3.783) | 0.07 (−0.11, 0.26) | .4370 | 0.11 (−0.06, 0.27) | .2099 |
| Q3 (median = 4.78) | 0.15 (−0.03, 0.33) | .0957 | 0.27 (0.10, 0.43) | .0014 |
| Q4 (median = 6.50) | −0.05 (−0.23, 0.13) | .5926 | 0.19 (0.01, 0.37) | .0459 |
| Difference between weekday and weekend midpoint (hours) | ||||
| Q1 (median = 0.00) | Reference | Reference | ||
| Q2 (median = 0.50) | 0.13 (−0.06, 0.32) | .1810 | 0.13 (−0.04, 0.30) | .1313 |
| Q3 (median = 1.25) | −0.05 (−0.23, 0.13) | .6010 | 0.08 (−0.09, 0.25) | .3623 |
| Q4 (median = 2.25) | −0.13 (−0.31, 0.06) | .1810 | 0.11 (−0.06, 0.29) | .2114 |
| Difference in log IL-6 pm (pg/mL) (n = 494) | ||||
| Weekday sleep midpoint (Decimal time) | ||||
| Q1 (median = 2.03) | Reference | Reference | ||
| Q2 (median = 2.78) | 0.05 (−0.13, 0.23) | .5990 | 0.12 (−0.04, 0.27) | .1350 |
| Q3 (median = 3.53) | −0.07 (−0.26, 0.12) | .4520 | −0.05 (−0.21, 0.11) | .5341 |
| Q4 (median = 5.18) | −0.09 (−0.28, 0.10) | .3660 | −0.04 (−0.20, 0.12) | .5861 |
| Weekend sleep midpoint (Decimal time) | ||||
| Q1 (median = 2.78) | Reference | Reference | ||
| Q2 (median = 3.78) | −0.06 (−0.25, 0.13) | .5203 | 0.01 (−0.15, 0.16) | .9538 |
| Q3 (median = 4.78) | −0.01 (−0.19, 0.18) | .9417 | 0.12 (−0.04, 0.28) | .1323 |
| Q4 (median = 6.50) | −0.32 (−0.50, −0.13) | .0008 | −0.03 (−0.21, 0.14) | .7256 |
| Difference between weekday and weekend midpoint (hours) | ||||
| Q1 (median = 0.00) | Reference | Reference | ||
| Q2 (median = 0.50 | 0.08 (−0.12, 0.27) | .4479 | 0.07 (−0.09, 0.23) | .3958 |
| Q3 (median = 1.25) | 0.001 (−0.18, 0.18) | .9962 | 0.15 (−0.01, 0.31) | .0611 |
| Q4 (median = 2.25) | −0.17 (−0.35, 0.02) | .0775 | 0.12 (−0.05, 0.28) | .1692 |
*Variables apnea-hypopnea index, sex, age, smoking status, race, household income, body mass index, sleep duration, and caffeine intake are included in these models. CI = confidence interval, IL = interleukin, Q = quartile.
Table 4.
Association between categorical measures of sleep midpoint and cytokine TNF-α levels.
| Unadjusted Model | Adjusted Model* | |||
|---|---|---|---|---|
| Beta Coefficient (95% CI) | P | Beta Coefficient (95% CI) | P | |
| Difference in log TNF-α am (pg/mL) (n = 495) | ||||
| Weekday sleep midpoint (Decimal time) | ||||
| Q1 (median = 2.03) | Reference | Reference | ||
| Q2 (median = 2.78) | 0.06 (−0.13, 0.25) | .5670 | 0.02 (−0.16, 0.21) | .8059 |
| Q3 (median = 3.53) | 0.14 (−0.05, 0.34) | .1500 | 0.13 (−0.06, 0.33) | .1825 |
| Q4 (median = 5.18) | 0.14 (−0.06, 0.33) | .1760 | 0.12 (−0.08, 0.31) | .2535 |
| Weekend sleep midpoint (Decimal time) | ||||
| Q1 (median = 2.78) | Reference | Reference | ||
| Q2 (median = 3.78) | 0.01 (−0.19, 0.20) | .9400 | −0.03 (−0.23, 0.16) | .7447 |
| Q3 (median = 4.78) | 0.12 (−0.07, 0.31) | .2270 | 0.14 (−0.05, 0.33) | .1590 |
| Q4 (median = 6.50) | −0.04 (−0.23, 0.15) | .6820 | −0.01 (−0.22, 0.21) | .9565 |
| Difference between weekday and weekend midpoint (hours) | ||||
| Q1 (median = 0.00) | Reference | Reference | ||
| Q2 (median = 0.50) | −0.09 (−0.28, 0.11) | .3954 | −0.09 (−0.29, 0.11) | .3602 |
| Q3 (median = 1.25) | −0.11 (−0.30, 0.09) | .2772 | −0.05 (−0.25, 0.15) | .6055 |
| Q4 (median = 2.25) | −0.17 (−0.36, 0.02) | .0859 | −0.11 (−0.32, 0.10) | .3169 |
| Difference in log TNF-α pm (pg/mL) (n = 351) | ||||
| Weekday sleep midpoint (Decimal time) | ||||
| Q1 (median = 2.03) | Reference | Reference | ||
| Q2 (median = 2.78) | 0.06 (−0.09, 0.21) | .4310 | 0.08 (−0.07, 0.22) | .3268 |
| Q3 (median = 3.53) | 0.07 (−0.09, 0.23) | .3860 | 0.04 (−0.12, 0.20) | .6180 |
| Q4 (median = 5.18) | 0.10 (−0.05, 0.26) | .1940 | 0.08 (−0.08, 0.24) | .3366 |
| Weekend sleep midpoint (Decimal time) | ||||
| Q1 (median = 2.78) | Reference | Reference | ||
| Q2 (median = 3.78) | 0.10 (−0.06, 0.25) | .2147 | 0.06 (−0.10, 0.21) | .4900 |
| Q3 (median = 4.78) | 0.05 (−0.11, 0.20) | .5589 | 0.020 (−0.14, 0.17) | .8057 |
| Q4 (median = 6.50) | 0.14 (−0.02, 0.29) | .0785 | 0.10 (−0.08, 0.27) | .2820 |
| Difference between weekday and weekend midpoint (hours) | ||||
| Q1 (median = 0.00) | Reference | Reference | ||
| Q2 (median = 0.50) | 0.08 (−0.08, 0.23) | .3340 | 0.07 (−0.09, 0.22) | .3995 |
| Q3 (median = 1.25) | 0.03 (−0.12, 0.18) | .7210 | 0.01 (−0.15, 0.17) | .8777 |
| Q4 (median = 2.25) | −0.06 (−0.21, 0.10) | .4600 | −0.08 (−0.25, 0.08) | .3315 |
*Variables apnea-hypopnea index, sex, age, smoking status, race, household income, body mass index, sleep duration, and caffeine intake are included in these models. CI = confidence interval, Q = quartile, TNF-α, tumor necrosis factor α.
DISCUSSION
Within this cohort of adults (a majority of whom had OSAS), linear associations were observed between increased social jetlag and levels of the inflammatory cytokine IL-1. Additionally, later timing of sleep during both the weekdays and the weekends, and higher social jetlag, were associated with increased levels of IL-6. The magnitude of these associations (∼0.2-pg/mL difference between the latest midpoints and earliest midpoints) are clinically relevant and on par with a 20-unit difference in AHI (ie, our data showed that each unit of AHI was associated with a 0.01-pg/mL higher am IL-6; 95% CI: 0.006, 0.01). No trends were observed with TNF-α. These results were independent of age, sleep duration, and AHI severity. Altogether, findings indicate that sleep timing behavior may influence cytokine levels.
Our findings on social jetlag and later sleep timing align with those from studies on shift-work populations. The importance of sleep timing was investigated in a trial of shift workers, where inflammatory biomarkers C-reactive protein (CRP), IL-6, and TNF-α were assessed in 14 healthy adult participants during a randomized crossover study.25 Associations were observed between short-term circadian misalignment increases and elevated levels of CRP, IL-6, and TNF-α. Similarly, a study of long-term circadian rhythm misalignment measured inflammatory biomarkers CRP, IL-10, and TNF-α in 17 healthy adult participants over several weeks of misalignment.26 Their research found associations between misalignment and elevated levels of CRP, IL-10, and TNF-α.
The literature further indicates that social jetlag and sleep timing have associations with other inflammatory-related cardiometabolic diseases, such as coronary artery disease, obesity, and type 2 diabetes.2,27 Another study by Parsons et al28 was completed on 815 non–shift workers in New Zealand. Researchers described associations between increased social jetlag and measures of metabolic dysfunction and obesity. Although this study indicated a trend between social jetlag and elevated inflammatory biomarker levels, no significant associations were observed. Our findings on sleep timing and inflammation are also consistent with research from Patel et al,12 which suggested that extremes of sleep duration are associated with elevations in inflammatory biomarkers CRP, IL-6, and TNF-α. Although our associations were independent of sleep duration, extremes in sleep duration, especially short sleep duration, are typically positively correlated with later sleep timing and social jetlag.
In contrast to prior studies, our results indicated no associations between sleep and cytokine TNF-α. Previous studies have shown that increases in cytokine IL-1 levels are usually parallel with increases in TNF-α,29 so it is interesting that we did not observe trends between social jetlag and TNF-α. Similarly, differences between weekday and weekend midpoint were associated with IL-1 in the unadjusted, demographic, and adjusted analyses, but no such trends were observed with TNF-α. Reasons for this discrepancy are unclear, although differences in study populations and protocols (eg, 1 week of observation vs longer-term habits) could play a role.
Cytokine rhythms have shown remarkably consistent patterns in several studies; proinflammatory cytokine production such as IL-6 and TNF-α have been reported to be maximal during nocturnal sleep.30–32 Of note, TNF-α rhythms appear to be completely dependent on sleep, while IL-6 production is not.30–32 It is likely that chronic changes in circadian misalignment represent a different challenge to more acute changes in sleep, and how this impacts the inflammatory cascade is not fully understood. Since the current study used cross-sectional data, we were unable to determine chronic vs acute alterations. Furthermore, it has been postulated that individuals with pre-existing conditions such as OSAS could be primed for an inflammatory response26 and that growth hormone, prolactin cortisol, and catecholamine levels may mediate the influence of sleep on the inflammatory cascade.30,33,34 Since approximately half of our population had objectively defined obstructive sleep apnea (OSA) and hormone levels were not measured, it is plausible that our study population could have an altered inflammatory response to circadian misalignment compared with otherwise healthy individuals.
Alterations in cytokines might also be linked to BMI and metabolic function, as later sleep timing has been associated with obesity.1 Other possible sleep-inflammation pathways may involve behavioral and stress-related mechanisms. Circadian misalignment, such as that observed in shift workers, has been related to increased use of alcohol, increased smoking, and engaging in less physical activity.27 Diet could also be involved. A study of 2,433 adults in the United Kingdom reported that increasing social jetlag was inversely associated with healthy dietary patterns.35 Altering sleep patterns between the workdays and rest days may influence unhealthy eating habits, such as larger food portions, infrequent meals, and increased intake of snack foods.
It should be noted that this analysis has both strengths and limitations to consider. Because of the cross-sectional exploration of this study, a causal direction between later sleep timing and increased cytokine levels cannot conclusively be established. Due to the dependence on self-reported wake times and rest times, overall sleep exposure may be imprecise. The nature in which this study was conducted may have impacted the differing levels of inflammation, in that self-reported sleep exposure is a long-term measure and inflammatory cytokines were determined at 1 or 2 points in time. Further, blood was drawn at the same clock time for each participant rather than equivalent circadian times (ie, taken at the same time relative to each participant’s dim-light melatonin onset). Sample size represents a strength in this investigation, as all models were assessed with a reasonably large sample size.
In summary, our findings suggest that, independent of important confounders including age, sleep duration, and AHI severity, later sleep timing is associated with increased IL-6 levels and increased social jetlag is associated with increased IL-6 and IL-1 levels. Our results suggest that later sleep schedules and increased social jetlag could exacerbate systemic inflammation, irrespective of OSA status. Future studies that examine potential interactions between sleep timing and OSA on inflammation could provide meaningful insights for patients with both OSA and social jetlag.
DISCLOSURE STATEMENT
All authors have seen and approved the final version of the manuscript. Work for this study was performed at the University of Michigan. The Cleveland Family Study was supported by grants from the National Institutes of Health (HL46380, M01 RR00080-39, T32-HL07567, RO1-46380). The National Sleep Research Resource was supported by the National Heart, Lung, and Blood Institute (R24 HL114473, 75N92019R002). Dr. Jansen was supported by the National Heart, Lung, and Blood Institute (K01HL151673) and the National Institute of Environmental Health Sciences (R01ES032330). The authors report no conflicts of interest.
ABBREVIATIONS
- AHI
apnea-hypopnea index
- BMI
body mass index
- CI
confidence interval
- CRP
C-reactive protein
- IL
interleukin
- OSA
obstructive sleep apnea
- OSAS
obstructive sleep apnea syndrome
- PSG
polysomnography
- Q
quartile
- SD
standard deviation
- TNF-α
tumor necrosis factor α
REFERENCES
- 1. Thomas EA , Zaman A , Cornier M-A , et al . Later meal and sleep timing predicts higher percent body fat . Nutrients. 2020. ; 13 ( 1 ): 73 . [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. Depner CM , Stothard ER , Wright KP Jr . Metabolic consequences of sleep and circadian disorders . Curr Diab Rep. 2014. ; 14 ( 7 ): 507 . [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Merikanto I , Lahti T , Puolijoki H , et al . Associations of chronotype and sleep with cardiovascular diseases and type 2 diabetes . Chronobiol Int. 2013. ; 30 ( 4 ): 470 – 477 . [DOI] [PubMed] [Google Scholar]
- 4. Wong PM , Hasler BP , Kamarck TW , Muldoon MF , Manuck SB . Social jetlag, chronotype, and cardiometabolic risk . J Clin Endocrinol Metab. 2015. ; 100 ( 12 ): 4612 – 4620 . [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Wittmann M , Dinich J , Merrow M , Roenneberg T . Social jetlag: misalignment of biological and social time . Chronobiol Int. 2006. ; 23 ( 1-2 ): 497 – 509 . [DOI] [PubMed] [Google Scholar]
- 6. Mota MC , Silva CM , Balieiro LCT , Fahmy WM , Crispim CA . Social jetlag and metabolic control in non-communicable chronic diseases: a study addressing different obesity statuses . Sci Rep. 2017. ; 7 ( 1 ): 6358 . [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Chaput JP , Dutil C , Featherstone R , et al . Sleep timing, sleep consistency, and health in adults: a systematic review . Appl Physiol Nutr Metab. 2020. 45 , 10 (Suppl. 2) : S232 – S247 . [DOI] [PubMed] [Google Scholar]
- 8. Lucassen EA , de Mutsert R , le Cessie S , et al. ; NEO Study Group . Poor sleep quality and later sleep timing are risk factors for osteopenia and sarcopenia in middle-aged men and women: the NEO study . PLoS One. 2017. ; 12 ( 5 ): e0176685 . [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Shechter A , St-Onge MP . Delayed sleep timing is associated with low levels of free-living physical activity in normal sleeping adults . Sleep Med. 2014. ; 15 ( 12 ): 1586 – 1589 . [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Facco FL , Parker CB , Hunter S , et al. ; NICHD NuMoM2b and NHLBI NuMoM2b Heart Health Study Networks . Later sleep timing is associated with an increased risk of preterm birth in nulliparous women . Am J Obstet Gynecol MFM. 2019. ; 1 ( 4 ): 100040 . [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Irwin MR , Olmstead R , Carroll JE . Sleep disturbance, sleep duration, and inflammation: a systematic review and meta-analysis of cohort studies and experimental sleep deprivation . Biol Psychiatry. 2016. ; 80 ( 1 ): 40 – 52 . [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Patel SR , Zhu X , Storfer-Isser A , et al . Sleep duration and biomarkers of inflammation . Sleep. 2009. ; 32 ( 2 ): 200 – 204 . [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Almeida CM , Malheiro A . Sleep, immunity and shift workers: a review . Sleep Sci. 2016. ; 9 ( 3 ): 164 – 168 . [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Peppard PE , Young T , Barnet JH , Palta M , Hagen EW , Hla KM . Increased prevalence of sleep-disordered breathing in adults . Am J Epidemiol 2013. ; 177 ( 9 ): 1006 – 1014 . [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Redline S , Yenokyan G , Gottlieb DJ , et al . Obstructive sleep apnea-hypopnea and incident stroke: the Sleep Heart Health Study . Am J Respir Crit Care Med. 2010. ; 182 ( 2 ): 269 – 277 . [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. Peker Y , Carlson J , Hedner J . Increased incidence of coronary artery disease in sleep apnoea: a long-term follow-up . Eur Respir J. 2006. ; 28 ( 3 ): 596 – 602 . [DOI] [PubMed] [Google Scholar]
- 17. Hou H , Zhao Y , Yu W , et al . Association of obstructive sleep apnea with hypertension: a systematic review and meta-analysis . J Glob Health. 2018. ; 8 ( 1 ): 010405 . [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Kheirandish-Gozal L , Gozal D . Obstructive sleep apnea and inflammation: proof of concept based on two illustrative cytokines . Int J Mol Sci. 2019. ; 20 ( 3 ): 459 . [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Ciftci TU , Kokturk O , Bukan N , Bilgihan A . The relationship between serum cytokine levels with obesity and obstructive sleep apnea syndrome . Cytokine. 2004. ; 28 ( 2 ): 87 – 91 . [DOI] [PubMed] [Google Scholar]
- 20. Senaratna CV , Perret JL , Lodge CJ , et al . Prevalence of obstructive sleep apnea in the general population: a systematic review . Sleep Med Rev. 2017. ; 34 : 70 – 81 . [DOI] [PubMed] [Google Scholar]
- 21. Redline S , Tishler PV , Tosteson TD , et al . The familial aggregation of obstructive sleep apnea . Am J Respir Crit Care Med. 1995. ; 151 ( 3 Pt 1 ): 682 – 687 . [DOI] [PubMed] [Google Scholar]
- 22. Zhang GQ , Cui L , Mueller R , et al . The National Sleep Research Resource: towards a sleep data commons . J Am Med Inform Assoc. 2018. ; 25 ( 10 ): 1351 – 1358 . [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. Patel SR , Palmer LJ , Larkin EK , Jenny NS , White DP , Redline S . Relationship between obstructive sleep apnea and diurnal leptin rhythms . Sleep. 2004. ; 27 ( 2 ): 235 – 239 . [DOI] [PubMed] [Google Scholar]
- 24. Palmer LJ , Buxbaum SG , Larkin EK , et al . Whole genome scan for obstructive sleep apnea and obesity in African-American families . Am J Respir Crit Care Med. 2004. ; 169 ( 12 ): 1314 – 1321 . [DOI] [PubMed] [Google Scholar]
- 25. Morris CJ , Purvis TE , Hu K , Scheer FA . Circadian misalignment increases cardiovascular disease risk factors in humans . Proc Natl Acad Sci USA. 2016. ; 113 ( 10 ): E1402 – E1411 . [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26. Wright KP Jr , Drake AL , Frey DJ , et al . Influence of sleep deprivation and circadian misalignment on cortisol, inflammatory markers, and cytokine balance . Brain Behav Immun. 2015. ; 47 : 24 – 34 . [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27. Kecklund G , Axelsson J . Health consequences of shift work and insufficient sleep . BMJ. 2016. ; 355 : i5210 . [DOI] [PubMed] [Google Scholar]
- 28. Parsons MJ , Moffitt TE , Gregory AM , et al . Social jetlag, obesity and metabolic disorder: investigation in a cohort study . Int J Obesity (Lond). 2015. ; 39 ( 5 ): 842 – 848 . [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29. Clark IA , Vissel B . Inflammation-sleep interface in brain disease: TNF, insulin, orexin . J Neuroinflammation. 2014. ; 11 ( 1 ): 51 . [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30. Petrovsky N . Towards a unified model of neuroendocrine-immune interaction . Immunol Cell Biol. 2001. ; 79 ( 4 ): 350 – 357 . [DOI] [PubMed] [Google Scholar]
- 31. Haus E , Smolensky MH . Biologic rhythms in the immune system . Chronobiol Int. 1999. ; 16 ( 5 ): 581 – 622 . [DOI] [PubMed] [Google Scholar]
- 32. Lange T , Dimitrov S , Born J . Effects of sleep and circadian rhythm on the human immune system . Ann N Y Acad Sci. 2010. ; 1193 ( 1 ): 48 – 59 . [DOI] [PubMed] [Google Scholar]
- 33. Besedovsky HO , del Rey A . Immune-neuro-endocrine interactions: facts and hypotheses . Endocr Rev. 1996. ; 17 ( 1 ): 64 – 102 . [DOI] [PubMed] [Google Scholar]
- 34. Dimitrov S , Lange T , Fehm HL , Born J . A regulatory role of prolactin, growth hormone, and corticosteroids for human T-cell production of cytokines . Brain Behav Immun. 2004. ; 18 ( 4 ): 368 – 374 . [DOI] [PubMed] [Google Scholar]
- 35. Almoosawi S , Palla L , Walshe I , Vingeliene S , Ellis JG . Long sleep duration and social jetlag are associated inversely with a healthy dietary pattern in adults: results from the UK National Diet and Nutrition Survey Rolling Programme Y1-4 . Nutrients. 2018. ; 10 ( 9 ): 1131 . [DOI] [PMC free article] [PubMed] [Google Scholar]