Light exposure during sleep impairs cardiometabolic function

Ivy C Mason; Daniela Grimaldi; Kathryn J Reid; Chloe D Warlick; Roneil G Malkani; Sabra M Abbott; Phyllis C Zee

doi:10.1073/pnas.2113290119

. 2022 Mar 14;119(12):e2113290119. doi: 10.1073/pnas.2113290119

Light exposure during sleep impairs cardiometabolic function

Ivy C Mason ^a,^b,^c,¹, Daniela Grimaldi ^a,¹, Kathryn J Reid ^a, Chloe D Warlick ^a, Roneil G Malkani ^a, Sabra M Abbott ^a, Phyllis C Zee ^a,²

PMCID: PMC8944904 PMID: 35286195

Significance

Ambient nighttime light exposure is implicated as a risk factor for adverse health outcomes, including cardiometabolic disease. However, the effects of nighttime light exposure during sleep on cardiometabolic outcomes and the related mechanisms are unclear. This laboratory study shows that, in healthy adults, one night of moderate (100 lx) light exposure during sleep increases nighttime heart rate, decreases heart rate variability (higher sympathovagal balance), and increases next-morning insulin resistance when compared to sleep in a dimly lit (<3 lx) environment. Moreover, a positive relationship between higher sympathovagal balance and insulin levels suggests that sympathetic activation may play a role in the observed light-induced changes in insulin sensitivity.

Keywords: light, sleep, metabolism, sympathetic nervous system, insulin resistance

Abstract

This study tested the hypothesis that acute exposure to light during nighttime sleep adversely affects next-morning glucose homeostasis and whether this effect occurs via reduced sleep quality, melatonin suppression, or sympathetic nervous system (SNS) activation during sleep. A total of 20 young adults participated in this parallel-group study design. The room light condition (n = 10) included one night of sleep in dim light (<3 lx) followed by one night of sleep with overhead room lighting (100 lx). The dim light condition (n = 10) included two consecutive nights of sleep in dim light. Measures of insulin resistance (morning homeostatic model assessment of insulin resistance, 30-min insulin area under the curve [AUC] from a 2-h oral glucose tolerance test) were higher in the room light versus dim light condition. Melatonin levels were similar in both conditions. In the room light condition, participants spent proportionately more time in stage N2 and less in slow wave and rapid eye movement sleep. Heart rate was higher and heart rate variability lower (higher sympathovagal balance) during sleep in the room light versus the dim light condition. Importantly, the higher sympathovagal balance during sleep was associated with higher 30-min insulin AUC, consistent with increased insulin resistance the following morning. These results demonstrate that a single night of exposure to room light during sleep can impair glucose homeostasis, potentially via increased SNS activation. Attention to avoiding exposure to light at night during sleep may be beneficial for cardiometabolic health.

Exposure to artificial light during the night is widespread globally, particularly in industrialized countries (1–3). Given that light and dark exposure patterns play a key role in the timing of many behaviors and physiological functions (4), exposure to light in the evening and night has been posited to be deleterious for human health and well-being (1, 5–10). Impacts of light exposure during sleep are not as well studied as other kinds of nighttime light exposure. However, a recent cross-sectional observation study noted that, compared to no light exposure during sleep, any self-reported artificial light exposure in the bedroom during sleep (small nightlight in room, light from outside room, or television/light in room) was associated with obesity in women (11). Furthermore, the incidence of obesity was highest in those who reported sleeping with a television or light on in the bedroom (11). These findings suggest that light in the bedroom during nighttime sleep may negatively influence metabolic regulation.

Emerging evidence indicates that light exposure plays a role in human metabolic regulation, with evening light exposure having unfavorable effects on metabolic functions including decreased glucose tolerance and decreased insulin sensitivity (12, 13). In line with these data, we have previously shown that blue-enriched light exposure in the morning and evening alters glucose metabolism, with an increase in insulin resistance compared to dim light exposure (14). In addition, evidence indicates that nighttime indoor light exposure during the habitual sleep period while awake (15), and during sleep itself (16), likely has deleterious metabolic effects. A recent study prospectively measured light exposure in the bedroom during nighttime sleep and showed that higher levels of bedroom light exposure were associated with a higher incidence of type 2 diabetes in an elderly population (16). However, the exact mechanisms by which light exposure, particularly during nighttime sleep, impacts metabolic regulation are not well understood.

A proposed pathway to explain the relationship between nighttime light exposure and altered metabolic function is via changes in sleep. Robust evidence from epidemiological and experimental studies indicates that nighttime light exposure, either from outdoor or indoor sources, has negative impacts on subjective and objective sleep quality as indicated by actigraphy or polysomnography (PSG) measures of reduced total sleep time (TST), sleep efficiency (SE), increased wake after sleep onset (WASO), reduced amount of slow wave sleep (SWS), or increased arousal index (AI) (17–20). Given the well-established contribution of sleep disruptions to metabolic dysfunction (21), it is plausible that nighttime light exposure alters glucose metabolism due to disturbances to sleep. However, nighttime light exposure also appears to have a direct effect on glucose regulation that is independent of sleep loss, as shown by a study that subjected healthy male individuals to sleep deprivation in the dark or to sleep deprivation with nighttime light exposure (22). This study showed that a full night of sleep deprivation with nighttime light exposure increased postprandial levels of insulin and glucagon-like peptide-1, increased insulin resistance, and reduced nighttime melatonin; these changes were not observed under conditions of sleep deprivation in darkness.

A second proposed mechanism to explain the impairment of glucose metabolism from nighttime light exposure is via light-induced changes to the endogenous circadian system, including suppression and phase shifting of the melatonin rhythm (23). It is well established that light exposure suppresses melatonin secretion (24, 25), and several studies have implicated suppression of nighttime melatonin with incidence of diabetes (26) and insulin resistance (27). The association between altered melatonin levels and changes in glucose regulation may be explained by evidence that melatonin plays a role in the secretion and action of insulin (28–30). In particular, lower melatonin levels resulting from light exposure during the nighttime sleep period, in a fasting condition, have been suggested to alter melatonin’s facilitation of pancreatic β-cell recovery (31). Moreover, evidence shows that light exposure, even of moderate intensity, during the nighttime sleep period can produce a phase shift of the internal circadian system (32, 33). Given the established role of the circadian system in the control of glucose metabolism, light exposure during the nighttime sleep period could facilitate the misalignment between the central clock and peripheral clocks in metabolic tissues, with consequent negative impact on glucose homeostasis (34).

A third potential mechanism is the effect of light exposure on autonomic nervous system (ANS) activity. Light exposure has an arousing effect on the sympathetic autonomic system as revealed by the increase in cortisol or heart rate (HR) associated with light exposure mainly during the morning and/or nighttime hours as compared to evening hours (35–37). Beyond the direct excitatory effect exerted by light exposure on sympathetic activity (35), alterations of the ANS characterized by a shift toward an increased sympathetic drive have also been suggested to mediate the negative effects of sleep disruption on many physiological systems such as glucose metabolism (38). Thus, it is plausible that light-induced autonomic activation, either directly and/or mediated by sleep disruption, significantly contributes to the observed relationship between nighttime light exposure and altered glucose metabolism. Notably, sympathetic overactivity has been shown to precede the development of insulin resistance and prediabetes and contribute to the development of obesity and metabolic syndrome (39–41).

Prior studies have reported that light exposure during sleep increases HR and decreases HR variability (HRV), consistent with increased sympathetic activation (42–44). These studies either examined bright light (1,000 lx) over the entire sleep period (42) or lower light levels (50 lx or dawn simulation) early or late in the sleep period (43, 44). However, the effect of a single night of moderate room light exposure across the entire nighttime sleep period on autonomic activation and its impact on metabolic function has never been fully investigated.

In the present study, we tested the hypothesis that room light exposure (100 lx) during habitual nighttime sleep is associated with increased insulin resistance as measured by the homeostatic model of insulin resistance (HOMA-IR), the Matsuda insulin sensitivity index, and impaired response to an oral glucose tolerance test (OGTT) the next morning. In addition, we hypothesized potential mechanisms of light-induced metabolic changes, such as reduced sleep quality, suppression of melatonin level, and elevated sympathetic activation (HR and HRV) during the sleep period.

Results

Participants.

A total of 20 healthy adults were randomized into the room light condition (n = 10, one night in dim light < 3 lx followed by one night with room light at 100 lx) and the dim light condition (n = 10, two consecutive nights in dim light < 3 lx), Fig. 1. The two groups were not significantly different for age, body mass index (BMI), sex, and race (Table 1). Actigraphy in the week preceding the laboratory stay (Table 1) indicated that participants randomized to the room light and dim light conditions had similar bedtime, sleep duration, and SE. Furthermore, a measure of daytime sleepiness (Epworth Sleepiness Scale [ESS]) was similar in both groups and was within normal range at screening (room light condition: 7 ± 2, dim light condition: 6 ± 3, P = 0.49).

Fig. 1. — Schematic of laboratory protocol. OGTT: oral glucose tolerance test.

Table 1.

Participant demographics and sleep variables from 1 wk of actigraphy in the week prior to the laboratory visit

Demographics	Room light condition	Dim light condition	p
Number	10	10
Age (years)	26.61 ± 4.34	26.78 ± 5.15	0.89^a
BMI (kg/m²)	23.25 ± 3.94	24.25 ± 3.71	0.39^a
Sex (females, n)	8	6	0.16^c
Race	6 White, 3 Asian, 1 African-American	4 White, 5 Asian, 1 African-American	0.24^c
Actigraphy measures
Bedtime (hh:mm)	23:05 ± 0:36	23:02 ± 0:37	0.96^a
Sleep duration (hh:mm)	7:13 ± 0:40	7:00 ± 0:27	0.46^a
SE (%)	88.07 ± 3.10	84.08 ± 6.21	0.09^c

Open in a new tab

SE = sleep efficiency. (a) two-sided Student’s t test; (b) nonparametric two-sided Wilcoxon rank-sum test; (c) χ² test.

HOMA-IR.

The change in HOMA-IR from Day 1 to Day 2 was significantly different (P = 0.018) between the room light and dim light conditions (Fig. 2) and was characterized by a ∼15% increase in the room light condition compared to a ∼4% decrease in the dim light condition.

OGTT.

Glucose and insulin profiles from the 2-h OGTT are shown in Fig. 3. Glucose values were not different between room light and dim light conditions (condition × day P = 0.18, condition × day × time P = 0.99; Fig. 3 A and B). The difference in insulin from Day 1 to Day 2 was significantly larger for the room light compared to the dim light condition (condition × day P = 0.034) and was observed across the 2-h OGTT (condition × day × time P = 0.95). However, post hoc analysis indicated that higher insulin levels in the room light condition on Day 2 were most pronounced at 20 and 30 min post ingestion of the glucose bolus (P = 0.01 and P = 0.03, respectively; Fig. 3 C and D).

The change in 2-h AUC of glucose and insulin from Day 1 to Day 2 did not differ between conditions (glucose: room light condition: 505 ± 1,398 mg / dL × min, dim light condition: −294 ± 192 mg/dL × min, P = 0.30; insulin: room light condition: 1,999 ± 1,740 µU/mL × min, dim light condition: −726 ± 1,226 µU/mL × min, P = 0.12). However, the change in 30-min AUC of insulin (a measure of early phase of insulin secretion) from Day 1 to Day 2 was significantly larger (P = 0.030) for the room light condition compared to the dim light condition (Fig. 4).

Fig. 4. — Early phase insulin response (30-min AUC) during the OGTT for room light (n = 10) and dim light (n = 10) conditions on Day 1 and Day 2. Insulin 30-min AUC and HOMA-IR were significantly higher on Day 2 compared to Day 1 in the room light condition versus the dim light condition. P values refer to the change from Day 1 to Day 2 between conditions (unpaired Student’s t test). The error bars represent SD.

Matsuda Insulin Sensitivity Index.

The change in Matsuda index from Day 1 to Day 2 was significantly different (P = 0.048) between the room light and dim light conditions (Fig. 5) and was characterized by a ∼16% decrease in the room light condition compared to a ∼3% increase in the dim light condition.

Sleep.

Macrostructure.

The change in PSG parameters from Night 1 to Night 2 showed a greater percentage of TST spent in stage N2 (P = 0.004) and a lower percentage of TST spent in SWS (P = 0.017) and rapid eye movement sleep (REM, P = 0.033) in the room light condition compared to the dim light condition (Table 2). There was no difference between the two conditions in the change from Night 1 to Night 2 for PSG-derived measures of sleep fragmentation, number, and averaged duration of the wake episodes during the sleep period, and sleep–wake stage stability (SI Appendix, Table S1).

Table 2.

Polysomnographic features of sleep macrostructure

	Room light condition			Dim light condition
	n = 10			n = 10			p
	Night 1	Night 2	△ (Night 2) − (Night 1)	Night 1	Night 2	△ (Night 2) − (Night 1)
TST (min)	428.05 ± 47.79	440.25 ± 37.91	12.20 ± 59.33	427.20 ± 23.15	451.15 ± 13.85	23.95 ± 22.38	0.344^b
Stage N1 (%)	6.81 ± 4.38	5.07 ± 2.45	−1.73 ± 4.53	7.16 ± 3.62	3.88 ± 1.54	−3.28 ± 3.08	0.141^b
Stage N2 (%)	51.51 ± 7.18	51.44 ± 8.66	−0.08 ± 4.69	53.66 ± 5.73	46.28 ± 4.57	−7.39 ± 5.30	0.004^a
SWS (%)	20.64 ± 7.33	23.58 ± 7.84	2.94 ± 2.16	19.98 ± 5.62	26.31 ± 4.62	6.33 ± 3.46	0.017^a
REM sleep (%)	21.04 ± 5.13	19.91 ± 5.09	−1.13 ± 5.61	19.20 ± 2.65	23.53 ± 3.25	4.34 ± 4.95	0.033^a
SOL (min)	11.20 ± 16.80	6.15 ± 4.09	−5.05 ± 17.60	9.30 ± 6.46	7.75 ± 5.11	−1.55 ± 7.17	0.969^b
SE (%)	89.42 ± 10.07	91.90 ± 7.95	2.47 ± 12.44	89.04 ± 5.10	93.99 ± 2.83	4.94 ± 4.80	0.345^b
WASO (min)	39.50 ± 32.18	32.70 ± 36.12	−6.80 ± 47.56	43.35 ± 25.45	21.20 ± 12.13	−22.25 ± 21.44	0.361^a
AI	16.49 ± 5.35	13.82 ± 4.03	−2.67 ± 6.10	17.46 ± 6.37	12.13 ± 4.12	−5.32 ± 5.44	0.212^b

Open in a new tab

SWS = slow wave sleep; REM = rapid eye movement; SOL = sleep onset latency; SE = sleep efficiency; WASO = wake after sleep onset. P values refer to differences in △ (Night 2) − (Night 1) between room light and dim light conditions. (a) two-sided Student’s t test; (b) nonparametric two-sided Wilcoxon rank-sum test.

Relationship between the change in sleep macrostructure and metabolic measures.

There was no association between the change in sleep macrostructure (percentage of N2, SWS, or REM sleep) from Night 1 to Night 2 and the change from Day 1 to Day 2 in 30-min AUC of insulin, HOMA-IR, or Matsuda index (all P > 0.05 when analyzing participants from both groups together or by condition).

Microstructure.

There were no differences in the change from Night 1 to Night 2 between the room light and the dim light conditions for slow wave activity (SWA: 0.5 to 4 Hz) or slow oscillatory (SO) activity (0.5 to 1 Hz) across sleep cycles (SWA: condition × night P = 0.68, condition × night × cycle P = 0.86; SO activity: condition × night P = 0.96, condition × night × cycle P = 0.86; SI Appendix, Fig. S1).

Subjective sleepiness.

There were no differences in Karolinska Sleepiness Scale (KSS) from Day 1 to Day 2 between the room light and dim light conditions across the wake period (condition × day P = 0.61, condition × day × time P = 0.99).

Plasma Melatonin.

There was no difference in the change of 24-h melatonin levels from Day 1 to Day 2 between the room light and dim light conditions (condition × day P = 0.96, condition × day × time P = 0.99, SI Appendix, Fig. S2A). In addition, there was no difference in the change of timing of melatonin onset from Night 1 to Night 2 between the room light and dim light conditions (room light condition: 0 ± 1.1 h and dim light condition: 0 ± 0.9 h, P = 0.79). The change in the timing of melatonin offset from Night 1 to Night 2 showed a trend toward an advance in the room light versus the dim light conditions (room light condition: −0.4 ± 1.1 h and dim light condition: 0.9 ± 1.3 h, P = 0.07).

There was no difference in the AUC of melatonin levels during the sleep opportunity from Night 1 to Night 2 between the room light and dim light conditions (Δ in the room light condition: −155 ± 177 pg/mL × min and Δ in the dim light condition: −156 ± 167 pg/mL × min, P = 0.49, SI Appendix, Fig. S2B).

HR, HRV, and Blood Pressure (BP).

HR and HRV during the sleep period.

There was a significant increase in HR during the sleep period from Night 1 to Night 2 (condition × night P < 0.0001) in the room light compared to the dim light condition, which was maintained across the entire night (condition × night × time P = 0.98) (Fig. 6 A and B).

Fig. 6. — (*Top*, A and B): HR during the sleep period for room light (n = 10) and dim light (n = 10) conditions on Night 1 and Night 2. (*Lower*, C and D): LF/HF derived from HRV analysis during the sleep period for room light (n = 10) and dim light (n = 10) conditions on Night 1 and Night 2. Beat-to-beat HR during the sleep period was averaged every 10 min across the sleep period, starting from the time at lights off. Change in HR from Night 1 to Night 2 was significantly larger in participants randomized to the room light condition (A) compared to those randomized to the dim light condition (B) (General Linear Model: condition × night P < 0.001). LF/HF was calculated on time windows with a stable signal of at least 5 min for the entire duration of the sleep period, starting from lights off. Change in LF/HF from Night 1 to Night 2 was significantly larger in participants randomized to the room light condition (C) compared to those randomized to the dim light condition (D) (General Linear Model: condition × night P < 0.019). The error bars represent SD.

HRV analysis showed that the change in HF relative power from Night 1 to Night 2 was not different between the room light and dim light conditions (condition × night P = 0.86, condition × night × time P = 0.48, SI Appendix, Fig. S3 A and B), while the change in LF relative power from Night 1 to Night 2 was significantly higher in the room light compared to the dim light condition across the 8-h sleep period (condition × night P < 0.0001, condition × night × time P = 0.49, SI Appendix, Fig. S3 C and D). In addition, the change from Night 1 to Night 2 of low-frequency to high-frequency ratio (LF/HF) was significantly higher in the room light compared to the dim light condition across the 8-h sleep period (condition × night P = 0.019, condition × night × time P = 0.11, Fig. 6 C and D).

Relationship between the change in HR and HRV during the sleep period and metabolic measures.

When considering participants from both conditions together, the increase in the average HR and LF/HF during the sleep period from Night 1 to Night 2 were positively correlated with the increase in the initial 30-min AUC of insulin from the OGTT (HR: R = 0.47, P = 0.037; LF/HF: R = 0.65, P = 0.0019). When analyzing the two conditions separately, the positive association between the change in HR during the sleep period and the change in the early phase of insulin response on the following morning was not significant in both conditions, although it appeared to be stronger in participants randomized to the room light condition (R = 0.45, P = 0.19) compared to those randomized to the dim light condition (R = 0.16, P = 0.64). However, when analyzing the two conditions separately, the positive association between the change in LF/HF during the sleep period and the change in the early phase of insulin response was significant in the room light condition (R = 0.74, P = 0.013) but not in the dim light condition (R = 0.56, P = 0.10), Fig. 7.

Fig. 7. — Pairwise correlations between the change in LF/HF obtained from HRV and averaged across the sleep period (Night 1 to Night 2) and the change in the AUC of insulin at 30 min from OGTT (Day 1 to Day 2). A significant positive correlation between LF/HF change and the change in the AUC of insulin at 30 min was observed in participants randomized to the room light condition (A, n = 10), and not in those randomized to the dim light condition (B, n = 10). The dotted lines represent 95% CIs.

There was no significant association between the change in HR and LF/HF during the sleep period from Night 1 to Night 2 with the change in HOMA-IR or Matsuda index from Day 1 to Day 2 (P > 0.05 when analyzing participants from both groups together or by condition).

Daytime BP and HR.

There were no differences in HR and systolic or diastolic BP across the wake period from Day 1 to Day 2 between the room light and dim light conditions (HR: condition × day P = 0.91, condition × day × time P = 0.86; systolic BP: condition × day P = 0.26, condition × day × time P = 0.56; diastolic BP: condition × day P = 0.90, condition × day × time P = 0.71).

Visual Analog Scale (VAS).

There were no differences in the change from Day 1 to Day 2 of subjective hunger (VAS-H) or vigor and affect (VAS-GVA) across the wake period between the room light and dim light conditions (VAS-H: condition × day P = 0.82, condition × day × time P = 0.32; VAS-GVA: condition × day P = 0.19, condition × day × time P = 0.77).

Discussion

This study provides insight into the physiological mechanisms underlying the relationship between nighttime light exposure, specifically during sleep, with cardiometabolic function. The primary finding of this study is that exposure to a single night of room light (100 lx) during sleep increased measures of insulin resistance the next morning. Interestingly, the effect of nighttime light exposure on metabolic function was correlated with an increase in sympathovagal balance during sleep.

As hypothesized, participants randomized to the room light condition showed increased insulin resistance in the morning (i.e., higher fasting HOMA-IR and lower Matsuda index from the OGTT) when compared to participants in the dim light condition. As expected for these healthy, normal weight participants, similar glucose levels were maintained in both conditions during the OGTT. However, in the room light condition, there were higher insulin levels during the OGTT in the morning after sleeping with overnight room light exposure when compared to the dim light condition, which is indicative of a compensatory insulin response to maintain euglycemia under a condition of increased insulin resistance (45). Interestingly, while there were no differences in the overall AUC of insulin during the OGTT, there were higher insulin levels in the early phase of insulin secretion (i.e., AUC of insulin response within the first 30 min of the OGTT) following the night of room light exposure for participants in the room light condition when compared to those in the dim light condition. This early phase of insulin secretion, or acute insulin response, is a measure of pancreatic β-cell function and plays a physiological role in the maintenance of postprandial glucose homeostasis (46). Higher acute insulin response has been shown to be a predictor of the onset of type 2 diabetes in individuals with both normal fasting and 2-h OGTT plasma glucose concentrations (47). Increased insulin resistance and altered β-cell function play a key role in the pathogenesis of diabetes (48). As such, our findings might have implications for individuals who are frequently exposed to nighttime light during sleep and who are at increased risk for type 2 diabetes. The long-term (10 wk) negative effect of constant light exposure on glucose homeostasis has been shown in a rodent model (48); long-term effects in humans of evening and nighttime light exposure on obesity and metabolic disorders remain unknown.

Our initial hypothesis was that room light exposure would impact sleep and that the changes in sleep could be a primary mechanism to explain altered glucose regulation. In our study, nighttime light exposure was accompanied by changes in sleep macrostructure but not in sleep microstructure (i.e., SWA and SO activity across sleep cycles) nor subjective sleep quality. In line with existing literature (20), there was a greater percentage of stage N2 and a lower percentage of SWS and REM sleep during the night with room light exposure compared to during the dim light condition. Sleep did not appear to be more fragmented in response to the room light exposure as there were no significant changes in PSG-derived measures of cortical arousal, sleep fragmentation, or wake–sleep stage stability between the room light and dim light conditions. It has been postulated that the association between sleep disruption with metabolic function is via ANS activation, which has been reported in association with measures of sleep fragmentation (49). The lack of increased sleep fragmentation or arousals with room light exposure during sleep might explain why we did not observe an association between changes in sleep macrostructure with changes in glucose metabolism.

The lack of difference in melatonin between the room light and dim light conditions is in accordance with previous studies reporting mixed findings on the effect of light exposure during nighttime sleep on the level and timing of melatonin (19, 43, 50,). We chose 100 lx for the room light condition as this level is within the range of illuminance that has been employed in previous experimental studies as typical indoor illumination (23, 51–53). Although there are some data showing that exposure to light levels >65 lx can suppress melatonin secretion and delay biological rhythms (23, 54), only 5 to 9% of light is estimated to be transmitted through the closed eyelids to the eyes and subsequently to the retinohypothalamic pathway that mediates melatonin suppression (55, 56). In addition, the known interindividual variability in melatonin suppression to light exposure might account for the negative findings in this study (57). However, due to our relatively small sample size and since melatonin samples during wake were not collected under dim light conditions, we are unable to rule out the possible effect of light-induced changes of melatonin level or phase on metabolic parameters.

A finding from this study is the sustained increase in HR during the sleep period in the room light condition. In addition, and consistent with the observed increase in HR, increase in LF/HF derived from HRV during the sleep period was seen in the room light condition, indicative of an increased sympathovagal balance. The increase in LF/HF in response to light exposure during the nighttime sleep period has been reported by a previous study that exposed participants to much brighter light levels (1,000 lx) (42) than used in the present study. Other studies utilizing 50 lx or dawn simulation early or late in the sleep period have also reported increases in HR and decreases in HRV, consistent with increased sympathetic activation (43, 44). It is unknown whether light exposure levels similar to those used in the current study would have any or similar impact in awake individuals during the biological night.

Moreover, the significant positive association between the change (Night 1 to Night 2) in LF/HF during the sleep period and the change (Day 1 to Day 2) in the early phase of insulin secretion, which was observed in participants in the room light condition but not in those in the dim light condition, provides a potential mechanism underlying the metabolic changes seen following nighttime light exposure during sleep. While we are unable to definitively conclude that nighttime light exposure caused sympathetic activation, these results provide further support for this possibility.

There are multiple pathways through which light can impact the ANS. Increases in HR and LF/HF in response to room light exposure during sleep are consistent with existing data that indicate that light exposure can stimulate cardiac activity by eliciting a sympathoexcitatory response (36, 58). An important pathway proposed to mediate the effect of light on HR is via the retinohypothalamic tract to the suprachiasmatic nucleus (SCN) and then to the paraventricular nucleus, brainstem, and spinal cord to the heart (36, 59, 60). Other than via the SCN, evidence from animal models indicates that light can also directly affect brain regions with sympathoexcitatory effects (e.g., orexinergic system) (61), which might, in part, explain the differences we observed between the changes in melatonin and HR in response to light. Finally, given the state-dependent regulation of the ANS, it is possible that the observed changes in sleep macrostructure associated with nighttime light exposure may have also contributed to the increase in sympathetic activity. While there may be multiple pathways by which light may increase sympathetic activity, the relationship between sympathetic activation and glucose homeostasis are well established. Specifically, in response to increased sympathetic outflow to skeletal muscle, postprandial glucose uptake from skeletal muscles is reduced, resulting in an increase of plasma glucose levels which, in turn, stimulate additional insulin production by the pancreas and promote insulin resistance (40).

The effects of light exposure at night, particularly during sleep, on cardiometabolic function could have implications for those living in modern societies where indoor and outdoor nighttime light exposure is increasingly widespread and where concerns regarding cardiometabolic health are also on the rise. Thus, it is plausible that decreasing exposure to indoor nighttime light during sleep could have beneficial effects on cardiometabolic health. Future studies using a larger sample size and a randomized cross-over design to study the effects of varying light wavelengths, duration, and intensities are needed to confirm our findings and potential ecological translatability. Follow-up studies are warranted in sectors of our society who are more likely to be affected by nighttime light exposure during the sleep period, including those living in long-term care facilities. Larger studies powered to assess the role of sex are also warranted given the known effect of sex on ANS activity. Moreover, research is needed to determine how nighttime light exposure during sleep may interact with daytime light exposure history during wake and to establish whether chronic nighttime light exposure has long-term effects on cardiometabolic function.