Diurnal Light Exposure and Rest-Activity Rhythms in Relation to MASLD: Insights From 2 Nationwide Cohort Studies

Abstract

Context

Circadian rhythms may influence metabolic dysfunction–associated steatotic liver disease (MASLD), but the effect of personal light exposure and rest-activity rhythms on MASLD risk remain unclear.

Objective

To examine the associations between 24-hour light exposure, rest-activity rhythm (24-h-RAR), and the risk of incident MASLD, as well as how these factors influence genetic risk and life expectancy in individuals with MASLD.

Methods

This study used accelerometry data from both the National Health and Nutrition Examination Survey (NHANES) and UK Biobank cohorts. 24 h-RAR was assessed using nonparametric metrics, including activity level during the most active 10 hours (M10), activity level during the least active 5 hours (L5), relative amplitude (RA), M10 onset, and L5 onset. Light exposure was categorized into daytime and nighttime periods, with exposure durations recorded separately at different threshold levels. The primary outcome was MASLD, with secondary outcomes including fibrosis and cirrhosis.

Results

In the UK Biobank prospective analysis, each 0.1-unit and 1-unit increase in RA and M10 were associated with a 30% and 2% reduction of MASLD risk, respectively. In contrast, each 1-unit increase in L5 and delayed L5 onset were linked to an 8% and 21% increase of MASLD risk. Moreover, each additional hour of daylight exposure above 6000 lux was associated with a 9% lower risk of MASLD, while each additional 30 minutes of nocturnal light exposure above 30 lux corresponded to a 22% higher risk of MASLD. Additionally, a favorable 24 h-RAR profile and adequate light exposure were associated with a lower risk of fibrosis and cirrhosis, as well as improved life expectancy among participants with MASLD. Similar associations were observed in the NHANES analysis.

Conclusion

Greater daytime light exposure, reduced nocturnal light exposure, and regulated RARs may protect against MASLD and prevent its progression to liver fibrosis and advanced liver disease.

As a leading cause of liver-related morbidity and mortality, metabolic dysfunction–associated fatty liver disease (MASLD) is the most prevalent liver disease globally, affecting approximately 32% of the population (1, 2). Given the limited number of approved therapies or specific pharmacological treatments for MASLD, prevention has become a central focus of its management, with a particular emphasis on identifying and addressing modifiable risk factors (3). A variety of hepatic functions, including glucose, triglycerides (TGs), and cholesterol metabolism, are tightly regulated by the circadian clock, and disruptions to this system contribute to metabolic disorders and exacerbate underlying pathological conditions (4). The results from animal studies indicate that genetic disruption of clock genes disrupts metabolic functions in enterocytes and the liver, ultimately leading to hypertriglyceridemia (5, 6). However, despite these links, research examining the relationship between objectively measured circadian rhythm disruptions and MASLD remains limited.

The circadian system enables humans to synchronize their behavior and physiology with the day-night cycle. The central clock in the suprachiasmatic nucleus is primarily synchronized by light, the most crucial environmental cue, which activates photosensitive retinal ganglion cells that transmit signals to the suprachiasmatic nucleus to coordinate endogenous circadian rhythms (7). However, the widespread use of electrical lighting has allowed humans to be more active indoors, where environments are dim during the day and bright at night compared to natural light-dark cycles, disrupting natural circadian physiology (8). Peripheral clocks, such as the liver clock, are not only regulated by the master clock but can also be entrained by a variety of stimuli, including behavioral cycles like rest-activity and feeding-fasting patterns (9, 10), which align with diurnal changes.

This prospective cohort study leveraged 7 days of actimetry and light monitoring from the US National Health and Nutrition Examination Survey (US NHANES) and UK Biobank to explore the associations between 24-hour light exposure, 24-hour rest-activity rhythm (24-h-RAR), and the risk of incident MASLD. We also assessed how light exposure and 24-h-RAR influenced genetic risk, as well as their effect on life expectancy in individuals with MASLD.

Materials and Methods

Study Design and Population

For both the US NHANES and UK Biobank study, we employed stringent inclusion and exclusion criteria to ensure the reliability of our analysis. Detailed information regarding the study designs and data collection can be found in prior publications (http://www.cdc.gov/nchs/nhanes) (11).

The NHANES study included adults aged 20 years or older who had reliable accelerometer data from the 2011 to 2012 and 2013 to 2014 cycles, along with data available to calculate the Fatty Liver Index (FLI) or the US Fatty Liver Index (USFLI). Participants were excluded for excessive alcohol consumption (>3 drinks/day for men, >2 drinks/day for women), or a history of hepatitis B (positive hepatitis B surface antigen) or hepatitis C infection (positive hepatitis C antibody or hepatitis C virus RNA). The present analysis of UK Biobank was restricted to participants who responded to an email invitation for the accelerometer substudy. We excluded those with unreliable accelerometry data and individuals with prevalent alcohol-related liver diseases, viral hepatitis, autoimmune hepatitis, cirrhosis, liver cancer, other liver diseases, and alcohol/drug use disorder at baseline. For the prospective analysis of incident MASLD risk among UK Biobank participants, only individuals without MASLD at baseline were included. In our life expectancy analysis, we excluded US NHANES participants with missing death information. The study workflow is illustrated in Supplementary Figs. S1 and S2 (12). Details on the accelerometer data can be found in the supplementary methods (12).

Ethics Approval and Consent to Participate

The present analysis plan was approved by the UK Biobank data committee (application ID No. 58082). The UK Biobank obtained ethics approval by the UK National Research Ethics Service (ref11/NW/0382; June 17, 2011). Written informed consent was obtained from each participant prior to enrollment.

Exposure Assessment

We employed nonparametric analyses with the physical activity intensity data into 5-second epochs (UK Biobank field ID 90004) to derive 24-h-RAR parameters both for US NHANES and UK Biobank (13). Our analysis focused on five 24-h-RAR parameters: the average activity level during the least active 5-hour period (L5), the average activity level during the most active 10-hour period (M10), relative amplitude (RA), the onset time of L5 (L5 onset time), and the onset time of M10 (M10 onset time). RA reflects the difference between M10 and L5 over a 24-hour period, calculated as (M10 − L5)/(M10 + L5), with higher values indicating a greater distinction between activity levels during peak and low activity periods of the day (14, 15). RA, M10, L5, and M10 onset times were divided into tertiles. L5 onset time was examined in terms of deviation from the population mean. “Advanced” participants were defined as having an L5 onset of less than 0.5 SDs from the mean, and “delayed” participants were defined as having an L5 onset of more than 0.5 SD from the mean.

Nocturnal light and daylight exposures for each participant were derived from their 1-week light recordings by extracting 24-hour light profiles and applying factor analysis. The mean light intensity during day and night was calculated by averaging the recorded light exposure for each participant within the specified time ranges. In the UK Biobank, we calculated the daily duration of daylight at thresholds above 2000 lux (bright indoor environment), 4000 lux (shade outdoors), and 6000 lux (sunlight outdoors), categorized into (<1 hour/day, 1-2 hours/day, and ≥2 hours/day) (16). The duration of nighttime light exposure was calculated at thresholds of 5 lux (dim indoor lighting), 10 lux (recommended for ≥3 hours before bedtime), and 30 lux (the reference threshold for salivary dim light melatonin onset) (17). Exposure was categorized into 3 groups: less than 15 minutes/day, 15 to 30 minutes/day, and greater than or equal to 30 minutes/day. For US NHANES, since values of 2500 lux or greater were recorded as 2500 lux and the minimum detectable light level was 25 lux, daylight thresholds were set above 1500, 2000, and 2500 lux, and nighttime light thresholds above 25 and 30 lux, using the same categorization plan as the UK Biobank. Details in data processing, quality control and definition of light exposure are presented in the Supplementary Methods (12).

Outcome Ascertainment

In the US NHANES study, hepatic steatosis was defined by an FLI of 60 or greater or the USFLI of 30 or greater (18, 19). MASLD was diagnosed when hepatic steatosis was accompanied by one or more cardiometabolic criteria: (1) body mass index (BMI) 25 or greater or waist circumference greater than 94 cm (men)/80 cm (women); (2) fasting glucose of 100 mg/dL or greater, glycated hemoglobin A_1c (HbA_1c) of 5.7% or greater or a diagnosis of diabetes; (3) blood pressure of 130/85 mm Hg or greater or antihypertensive treatment; (4) TGs ≥150 mg/dL or lipid-lowering medication; or (5) high-density lipoprotein (HDL) cholesterol less than 40 mg/dL (men)/less than 50 mg/dL (women) or lipid-lowering medication (20). Mortality outcomes were obtained from the US National Death Index records. In the UK Biobank, MASLD cases were identified from inpatient hospital data based on the 9th and 10th revisions of the International Classification of Diseases (Supplementary Table S1) (12). Death records were sourced from death certificates held by the NHS Information Centre (England and Wales) and the NHS Central Register (Scotland) up to November 30, 2022. Time to event was calculated from the accelerometer assessment to the date of MASLD diagnosis, death, or censoring (October 31, 2022 for England; August 31, 2022 for Scotland; and May 31, 2022 for Wales), whichever came first. Details on the secondary outcome can be found in the supplementary methods (12).

Ascertainment of Covariates

Based on previous epidemiologic evidence, sociodemographic and lifestyle covariates were collected from participants using touchscreen questionnaires. These covariates included age at accelerometer assessment, sex, BMI, ethnicity (White/others), educational level (college or university degree and others), smoking status (never or ever), alcohol intake (never or ever), healthy diet score (UK Biobank only), sleep duration, Townsend deprivation index (UK Biobank only), the ratio of family income to poverty (US NHANES only), PM2.5 absorbance (UK Biobank only), shift work history (UK Biobank only), season of accelerometer wear (UK Biobank only), photoperiod (UK Biobank only), physical activity, history of diabetes, and history of hypertension. Detailed information on covariates and genotype information can be found in the supplementary methods (12). Variables were recorded at the time point closest to the accelerometry measurement to minimize temporal bias from fluctuating covariates (Supplementary Fig. S3) (12).

Statistical Analysis

Baseline characteristics are presented as median (interquartile range) for continuous variables and percentages for categorical variables. Due to the intricate sampling design of the US NHANES, our analyses incorporated sample weights, clustering, and stratification to fulfill the necessary criteria for analyzing US NHANES data. All analyses were adjusted for covariates.

For cross-sectional analyses, logistic regression models were used to estimate odds ratios (ORs) and 95% CIs for the association between light exposure, 24-h-RAR, and prevalent MASLD. Cox proportional-hazards regression models were used to estimate hazard ratios (HRs) and 95% CIs. The proportional-hazards assumption was tested using Schoenfeld residuals. Missing values were handled by assigning a missing indicator category for categorical variables and replacing continuous variables with their median values. Restricted cubic spline analyses with 3 knots (at the 5th, 50th, and 95th percentiles) were conducted to examine the dose-response association (21). Subgroup analyses were conducted to determine whether the relationship between light exposure, 24-h-RAR, and MASLD risk varied by polygenic risk, age, sex, BMI, physical activity, and late-night eating (US NHANES only). The multiplicative interaction was tested using adding product terms in the Cox models. We conducted proportional-hazards survival analyses to evaluate the effects of light exposure and 24-h-RAR on life expectancy among individuals with MASLD (using the stpm2 command in Stata), conditional on survival from age 45 to 100 years. We also employed a machine learning technique, Shapley Additive Explanations (SHAP), to rank the importance of 26 risk and protective factors in predicting MASLD. Detailed information can be found in the supplementary methods (12).

Several sensitivity analyses were conducted to investigate potential sources of bias in our findings. First, events occurring within the first 2 years after accelerometer assessments were excluded to mitigate the risk of reverse causality. Second, we excluded participants with a history of shift work. Third, we paired models with and without BMI, as well as with and without physical activity. Fourth, the models were further adjusted for waist circumference, fasting glucose, HbA_1c, TGs, HDL cholesterol, sleep apnea, and the use of sedatives/antidepressants. Fifth, in the NHANES analysis, additional adjustments were made for the Healthy Eating Index 2015. Sixth, we further adjusted for M10 and L5 in the RA analysis. Finally, multiple imputation with chained equations (MICE package v3.17 in R) was used to address missing covariate data.

Statistical significance was defined as a 2-sided P less than .05. SAS version 9.4 (SAS Institute Inc), R software version 4.3.1 and Stata (version 16.0) were used for all statistical analyses.

Results

Baseline Characteristics

In our cross-sectional analyses of the US NHANES dataset, we included 7253 participants, among whom 3465 (47.8%) had MASLD. A total of 91 349 participants were enrolled in the UK Biobank study and observed for a median of 7.93 years, during which 776 MASLD cases were recorded. As shown in Table 1, participants with MASLD in both studies were older, more likely to be male, had a higher BMI, and faced poorer economic conditions. Moreover, they were more likely to be current smokers and engaged in lower levels of physical activity. The baseline characteristics of participants by 24-h-RAR and light exposure are shown in Supplementary Tables S4 and S5 (12).

Table 1.

Baseline characteristics of participants by metabolic dysfunction–associated steatotic liver disease^a

	US NHANES		UK Biobank
	No MASLD	Prevalent MASLD	No MASLD	Incident MASLD
No. of participants	3788	3465	90 573	776
Age, y	44.4 (30.7 to 58.9)	50.3 (37.9 to 62.0)	63.0 (55.8 to 68.1)	63.4 (56.2 to 68.4)
Sex, male, %	44.9	51.3	43.4	47.2
Race or ethnicity, White, %	40.1	42.5	96.6	94.9
Waist circumference, cm	88.5 (81.5 to 95.0)	109.3 (102.5 to 118.2)	87.2 (78.0 to 97.0)	98.0 (90.0 to 107.0)
BMI categories, %
Normal/Underweight, <25	53.8	2.39	39.6	12.1
Overweight, 25-30	39.0	25.5	41.1	38.4
Obese, ≥30	7.21	72.0	19.0	48.9
Missing	0.05	0.12	0.22	0.52
Townsend deprivation index	—	—	−2.46 (−3.83 to −0.22)	−1.67 (−3.42 to 1.09)
Missing, %	—	—	0.11	0.13
Poverty-income ratio	3.19 (1.46 to 4.99)	2.66 (1.33 to 4.75)	—	—
Missing, %	7.50	7.50	—	—
Education level, ≥college, %	32.0	20.1	43.1	31.4
Missing, %	0.06	0.03	0.99	1.68
Season of accelerometer wear, %
Spring	—	—	22.6	18.7
Summer	—	—	26.1	26.8
Autumn	—	—	29.9	30.3
Winter	—	—	21.4	24.2
Smoking status, %
Never smoker	61.9	54.7	57.3	47.3
Ever smoker	38.1	45.3	42.5	52.1
Missing, %	0.03	0.06	0.26	0.64
Alcohol intake
Never	28.9	27.0	5.50	9.54
Ever	67.3	67.7	94.4	90.3
Missing, %	6.81	5.31	0.08	0.13
Total MVPA volume, min/wk	298.3 (8.51 to 719.4)	179.8 (0.00 to 701.0)	115.3 (53.7 to 213.0)	55.0 (20.6 to 123.0)
Photoperiod	—	—	12.3 (9.38 to 15.4)	12.1 (9.16 to 15.4)
PM2.5 absorbance	—	—	9.85 (9.19 to 10.5)	10.1 (9.40 to 10.8)
Missing, %	—	—	8.17	2.92
HEI-2015	53.0 (43.5 to 62.9)	49.9 (41.6 to 58.9)	—	—
Healthy diet score	—	—	4.00 (3.00 to 4.00)	3.00 (2.00 to 4.00)
Sleep duration, %
<7 h/d	36.2	40.6	34.7	37.5
7-8 h/d	57.3	50.8	45.7	44.1
>8 h/d	6.52	8.66	19.9	18.4
Shift work history, yes, %	—	—	22.2	26.8
Diabetes, yes, %	6.68	20.0	2.82	8.76
Hypertension, yes, %	25.7	49.1	22.4	38.8
FBG, mg/dL	96.0 (90.0 to 103.0)	105.0 (97.0 to 121.0)	4.92 (4.59 to 5.28)	5.02 (4.65 to 5.5)
HbA1c, %	5.40 (5.20 to 5.70)	5.70 (5.40 to 6.20)	34.9 (32.5 to 37.3)	36.2 (33.4 to 39.5)
SBP, mm Hg	118.0 (108.0 to 130.0)	124.0 (114.0 to 136.0)	135.0 (123.5 to 148.0)	139.0 (128.5 to 151.0)
DBP, mm Hg	70.0 (62.0 to 76.0)	72.0 (64.0 to 80.0)	81.0 (74.5 to 88.0)	84.0 (77.5 to 90.5)
TGs, mg/dL	91.0 (64.0 to 129.0)	161.0 (109.0 to 242.0)	123.0 (88.6 to 179.5)	164.0 (116.9 to 240.7)
HDL cholesterol, mg/dL	55.0 (47.0 to 66.0)	45.0 (38.0 to 53.0)	56.1 (46.8 to 66.9)	47.9 (40.6 to 56.9)

Abbreviations: BMI, body mass index; DBP, diastolic blood pressure; FBG, fasting blood glucose; HbA_1c, glycated hemoglobin A_1c; HDL, high-density lipoprotein; HEI, Healthy Eating Index; MASLD, metabolic dysfunction–associated steatotic liver disease; MVPA, moderate to vigorous physical activity; NHANES, National Health and Nutrition Examination Survey; PA, physical activity; SBP, systolic blood pressure; TGs, triglycerides.

^a Continuous variables are expressed as median (interquartile range), and categorical variables are expressed as percentages.

Association of 24-Hour Rest-Activity Rhythm and Light Exposure With Metabolic Dysfunction–Associated Steatotic Liver Disease

Table 2 presents the association between 24-h-RAR and the risk of prevalent and incident MASLD. Individuals with each 0.1-unit increase in RA exhibited an 28% (UK Biobank) lower prevalence of MASLD. Prospective analyses in the UK Biobank further revealed that each 0.1-unit increase in RA was associated with a 30% reduction in the risk of incident MASLD. Additionally, a 1-unit increase in M10 and L5 was linked to a 37% reduction and a 59% increase in the risk of incident MASLD, respectively. Similar associations were observed in the cross-sectional analysis. No statistically significant association was found between M10 onset time and the risk of MASLD. However, in the UK Biobank analysis, a delayed L5 onset time was associated with a higher risk both of prevalent and incident MASLD, with an OR (95% CI) of 1.22 (1.05-1.41) and an HR of 1.21 (1.02-1.42), respectively. We found the same directional consistency in the US NHANES as observed in the UK Biobank, further supporting the robustness of the results across both datasets. The dose-response analysis revealed a U-shaped association with the risk of incident MASLD (Fig. 1).

Table 2.

Association of 24-hour rest-activity rhythm with risk of prevalent and incident metabolic dysfunction–associated steatotic liver disease

	US NHANES (baseline; N = 7253)a		UK Biobank (baseline; N = 91 559)a		UK Biobank (follow-up; N = 91 349)b
	Cases % (N)	Odds ratio (95% CI)	Cases % (N)	Odds ratio (95% CI)	Cases % (N)	Hazard ratio (95% CI)
RA
Continuous, per 0.1 unit	47.8 (7253)	0.89 (0.82-0.96)	1.07 (91 559)	0.72 (0.67-0.77)	0.85 (91 349)	0.70 (0.65-0.75)
Low	54.4 (2397)	1.00 (reference)	1.77 (30 519)	1.00 (reference)	1.39 (30 449)	1.00 (reference)
Intermediate	49.5 (2556)	0.85 (0.60-1.19)	0.94 (30 520)	0.70 (0.60-0.82)	0.75 (30 451)	0.70 (0.59-0.82)
High	38.9 (2300)	0.66 (0.51-0.84)	0.51 (30 520)	0.53 (0.44-0.65)	0.41 (30 449)	0.54 (0.43-0.67)
M10
Continuous, per 1 unit	47.8 (7253)	0.95 (0.93-0.97)	1.07 (91 559)	0.98 (0.97-0.98)	0.85 (91 349)	0.98 (0.97-0.98)
Low	57.9 (2417)	1.00 (reference)	1.74 (30 520)	1.00 (reference)	1.36 (30 449)	1.00 (reference)
Intermediate	46.0 (2422)	0.76 (0.62-0.93)	0.92 (30 519)	0.70 (0.60-0.81)	0.74 (30 450)	0.70 (0.59-0.83)
High	39.4 (2414)	0.61 (0.47-0.79)	0.56 (30 520)	0.60 (0.49-0.74)	0.46 (30 450)	0.63 (0.50-0.79)
L5
Continuous, per 1 unit	47.8 (7253)	1.06 (0.98-1.17)	1.07 (91 559)	1.08 (1.04-1.11)	0.85 (91 349)	1.08 (1.05-1.12)
Low	42.5 (2395)	1.00 (reference)	0.81 (30 519)	1.00 (reference)	0.63 (30 449)	1.00 (reference)
Intermediate	49.0 (2431)	1.12 (0.91-1.38)	0.97 (30 521)	1.16 (0.97-1.37)	0.77 (30 450)	1.19 (0.98-1.44)
High	51.8 (2427)	1.29 (1.03-1.62)	1.44 (30 519)	1.51 (1.27-1.79)	1.15 (30 450)	1.59 (1.32-1.92)
M10 onset time
Continuous, per 1 unit	47.8 (7253)	1.03 (0.98-1.08)	1.07 (91 559)	1.01 (0.97-1.06)	0.85 (91 349)	1.01 (0.96-1.06)
Advanced	49.9 (2424)	1.00 (reference)	1.05 (30 666)	1.00 (reference)	0.84 (30 600)	1.00 (reference)
Intermediate	49.5 (2412)	1.21 (0.94-1.56)	1.00 (30 483)	0.94 (0.80, 1.11)	0.77 (30 410)	0.91 (0.76-1.09)
Delayed	43.9 (2417)	1.12 (0.85-1.48)	1.17 (30 410)	1.09 (0.93, 1.27)	0.94 (30 339)	1.09 (0.92-1.29)
L5 onset time
Continuous (per 1 unit)	47.8 (7253)	1.02 (0.98-1.06)	1.07 (91 559)	1.01 (0.96-1.06)	0.85 (91 349)	1.00 (0.95-1.06)
Advanced	51.1 (2112)	0.92 (0.74-1.13)	1.05 (19 917)	1.12 (0.95-1.32)	0.84 (19 874)	1.13 (0.94-1.35)
Intermediate	45.9 (3384)	1.00 (reference)	0.96 (48 369)	1.00 (reference)	0.77 (48 271)	1.00 (reference)
Delayed	47.4 (1757)	1.08 (0.85-1.38)	1.32 (23 273)	1.22 (1.05-1.41)	1.03 (23 204)	1.21 (1.02-1.42)

Model was adjusted for age, sex, ethnicity, body mass index, smoking status, alcohol drinking status, educational level, healthy diet score (UK Biobank only), Townsend deprivation index (UK Biobank only), ratio of family income to poverty (US NHANES only), PM2.5 absorbance (UK Biobank only), shift work history (UK Biobank only), sleep duration, total MVPA volumes, season of accelerometer wear (UK Biobank only), history of diabetes, and history of hypertension.

Abbreviations: 24-h-RAR, 24-hour rest-activity rhythm; L5, least active 5-hour period; M10, most active 10-hour period; MASLD, metabolic dysfunction–associated steatotic liver disease; MVPA, moderate to vigorous physical activity; RA, relative amplitude; US NHANES, US National Health and Nutrition Examination Survey.

^a Obtained by using multivariable linear regression model.

^b Obtained by using multivariable Cox regression model.

Figure 1.

Restricted cubic spline models for the association between 24-hour rest-activity rhythm (24-h-RAR) and risk of metabolic dysfunction–associated steatotic liver disease (MASLD) in UK Biobank data. A to E illustrate the associations between relative amplitude (RA), most active 10 hours (M10), least active 5 hours (L5), M10S, L5S, and the incidence of MASLD in the UK Biobank datasets. Restricted cubic spline models fitted for Cox regression models with 3 knots. The 95% CIs of the adjusted hazard ratios (HRs) are represented by the shaded area. Restricted cubic spline model is adjusted the same as the model from Supplementary Table S3.

Table 3 and Fig. 2 present the association between daylight and nighttime light exposure with the risk of both prevalent and incident MASLD. In the US NHANES analysis, each additional hour of daylight exposure was associated with a 15%, 17%, and 20% reduction in the risk of prevalent MASLD when light intensity exceeded 1500, 2000, and 2500 lux, respectively. Conversely, for each 30-minute increase in nighttime light exposure, the risk of prevalent MASLD increased by 10% and 13% when light intensity exceeded 25 and 30 lux, respectively. Prospective analyses in the UK Biobank further showed that each additional 1 hour of daylight exposure was associated with a 4%, 7%, and 9% decrease in the risk of incident MASLD when daylight intensity surpassed 2000, 4000, and 6000 lux, respectively. Moreover, in the UK Biobank, each additional 30 minutes of brighter nighttime light exposure was associated with a 17%, 20%, and 22% increase in the risk of incident MASLD when nighttime light intensity surpassed 5, 10, and 30 lux, respectively.

Table 3.

Association of day and night light exposure with risk of prevalent and incident metabolic dysfunction–associated steatotic liver disease

	US NHANES (baseline; N = 4341)a			UK Biobank (baseline; N = 83 789)a		UK Biobank (follow-up; N = 83 595)b
	Cases % (N)	Odds ratio (95% CI)		Cases % (N)	Hazard ratio (95% CI)	Cases % (N)	Hazard ratio (95% CI)
Daylight duration			Daylight duration
>1500 lux			>2000 lux
Continuousc	48.4 (4341)	0.85 (0.73-1.00)	Continuousc	1.07 (83 789)	0.96 (0.93-1.00)	0.85 (83 595)	0.96 (0.92-1.00)
<1 h per d	48.0 (2784)	1.00 (reference)	<1 h per d	1.15 (22 915)	1.00 (reference)	0.90 (22 858)	1.00 (reference)
1-2 h per d	49.7 (1162)	0.91 (0.71-1.17)	1-2 h per d	1.11 (28 266)	1.00 (0.85-1.19)	0.90 (16 244)	1.02 (0.82-1.26)
≥2 h per d	47.9 (395)	0.73 (0.51-1.04)	≥2 h per d	0.99 (32 608)	0.90 (0.74-1.09)	0.80 (44 493)	0.90 (0.74-1.10)
>2000 lux			>4000 lux
Continuousc	48.4 (4341)	0.83 (0.69-0.99)	Continuousc	1.07 (83 789)	0.94 (0.89-0.98)	0.85 (83 595)	0.93 (0.88-0.98)
<1 h per d	47.9 (3197)	1.00 (reference)	<1 h per d	1.19 (37 118)	1.00 (reference)	0.93 (37 024)	1.00 (reference)
1-2 h per d	50.3 (921)	1.07 (0.81-1.42)	1-2 h per d	1.08 (17 003)	0.91 (0.76-1.09)	0.83 (16 960)	0.89 (0.72-1.09)
≥2 h per d	47.5 (223)	0.50 (0.32-0.79)	≥2 h per d	0.93 (29 668)	0.79 (0.66-0.94)	0.75 (29 611)	0.79 (0.64-0.96)
>2500 lux			>6000 lux
Continuousc	48.4 (4341)	0.80 (0.65-0.99)	Continuousc	1.07 (83 789)	0.92 (0.86-0.97)	0.85 (83 595)	0.91 (0.85-0.97)
<1 h per d	48.4 (3549)	1.00 (reference)	<1 h per d	1.19 (47 853)	1.00 (reference)	0.93 (47 729)	1.00 (reference)
1-2 h per d	48.8 (681)	0.99 (0.72-1.35)	1-2 h per d	1.02 (15 618)	0.84 (0.69-1.01)	0.83 (15 587)	0.87 (0.70-1.07)
≥2 h per d	47.8 (111)	0.46 (0.27-0.81)	≥2 h per d	0.84 (20 318)	0.70 (0.57-0.85)	0.66 (20 279)	0.68 (0.54-0.85)
Nocturnal light duration			Nocturnal light duration
>25 lux			>5 lux
Continuousd	48.4 (4341)	1.10 (1.01-1.21)	Continuousd	1.07 (83 789)	1.17 (1.06-1.28)	0.85 (83 595)	1.17 (1.06-1.29)
<15 min per d	51.7 (1303)	1.00 (reference)	<15 min per d	0.97 (60 289)	1.00 (reference)	0.76 (60 158)	1.00 (reference)
15-30 min per d	44.7 (1362)	1.00 (0.70-1.43)	15-30 min per d	1.07 (10 699)	1.01 (0.82-1.24)	0.90 (10 680)	1.10 (0.88-1.37)
≥30 min per d	48.8 (1676)	1.41 (1.03-1.93)	≥30 min per d	1.57 (12 801)	1.33 (1.12-1.57)	1.23 (12 757)	1.34 (1.11-1.62)
>30 lux			>10 lux
Continuousd	48.4 (4341)	1.13 (1.01-1.26)	Continuousd	1.07 (83 789)	1.21 (1.09-1.32)	0.85 (83 595)	1.20 (1.08-1.34)
<15 min per d	50.6 (1547)	1.00 (reference)	<15 min per d	0.97 (61 098)	1.00 (reference)	0.75 (60 966)	1.00 (reference)
15-30 min per d	45.6 (1419)	0.94 (0.68-1.29)	15-30 min per d	1.13 (10 532)	1.08 (0.88-1.31)	0.95 (10 512)	1.16 (0.93-1.45)
≥30 min per d	48.8 (1375)	1.41 (1.02-1.95)	≥30 min per d	1.56 (12 159)	1.33 (1.11-1.57)	1.22 (12 117)	1.33 (1.10-1.62)
			>30 lux
			Continuousd	1.07 (83 789)	1.22 (1.07-1.38)	0.85 (83 595)	1.22 (1.06-1.40)
			<15 min per d	0.97 (64 166)	1.00 (reference)	0.76 (64 027)	1.00 (reference)
			15-30 min per d	1.21 (9570)	1.16 (0.94-1.42)	1.01 (9549)	1.23 (0.99-1.54)
			≥30 min per d	1.58 (10 053)	1.33 (1.10-1.59)	1.25 (10 019)	1.34 (1.09-1.65)

Model was adjusted for age, sex, ethnicity, body mass index, smoking status, alcohol drinking status, educational level, healthy diet score (UK Biobank only), Townsend deprivation index (UK Biobank only), the ratio of family income to poverty (US NHANES only), PM2.5 absorbance (UK Biobank only), shift work history (UK Biobank only), sleep duration, total MVPA volumes, season of accelerometer wear (UK Biobank only), photoperiod (UK Biobank only), history of diabetes, and history of hypertension.

Abbreviations: MASLD, metabolic dysfunction–associated steatotic liver disease; MVPA, moderate to vigorous physical activity; US NHANES, US National Health and Nutrition Examination Survey.

^a Obtained by using multivariable linear regression model.

^b Obtained by using multivariable Cox regression model.

^c Per hour increment.

^d Per 30-minute increment.

Figure 2.

Restricted cubic spline models for the association between light exposure and risk of metabolic dysfunction–associated steatotic liver disease (MASLD) in UK Biobank data. A to F illustrate the associations between daylight or nightlight exposure and the prevalence or incidence of MASLD in the UK Biobank datasets. Restricted cubic spline models fitted for Cox regression models with 3 knots. The 95% CIs of the adjusted hazard ratios (HRs) are represented by the shaded area. Restricted cubic spline model is adjusted the same as the model from Supplementary Table S4.

Association of 24-Hour Rest-Activity Rhythm and Light Exposure With Secondary Outcomes

Consistently, individuals with each 0.1-unit increase in RA exhibited a lower prevalence of both incident and prevalent fibrosis and steatosis, with a 23% reduction in steatosis (OR; UK Biobank), 29% reduction in fibrosis (OR; UK Biobank), and a 25% reduction in cirrhosis (HR; UK Biobank) (Supplementary Table S6) (12). Similar associations were observed with other 24-h-RAR parameters as well, aligning with the findings for MASLD.

In the NHANES analysis, no statistically significant association was observed between light exposure and the risk of prevalent fibrosis. However, the UK Biobank results suggest that both daytime and nighttime light exposure may be also associated with the prevalence of fibrosis and cirrhosis. Notably, no association was found between nightlight exposure and the risk of incident cirrhosis (Supplementary Table S7) (12).

Supplementary and Sensitivity Analysis

In the subgroup analyses, individuals with varying genetic risk had similar associations with risk of MASLD (Supplementary Tables S8 and S9) (12), with no statistically significant interaction observed between genetic risk and 24 h-RAR or light exposure. This trend was consistent across different age, sex, BMI, physical activity, and late-night eating categories (Supplementary Tables S10 and S11; all P for interaction >.05) (12).

Results of our life expectancy analysis showed that individuals aged 45 years or older with MASLD had a shorter life expectancy compared to those without the disease (Supplementary Fig. S8) (12). Among individuals with MASLD, those with high RA gained an additional 6.29 years of life expectancy in the US NHANES and 8.17 years in the UK Biobank at age 45, compared to those with low RA (Fig. 3 and Supplementary Fig. S9) (12). Among the various day-night light exposure thresholds, MASLD patients gained up to 2.79 additional years of life expectancy in the US NHANES and 4.75 years in the UK Biobank at age 45, compared to those with unhealthy light exposure (Fig. 4, Supplementary Fig. S10, and Supplementary Tables S12 and S13) (12).

Figure 3.

Relationship between life expectancy gained and 24-hour rest-activity rhythm (24-h-RAR) among participants with metabolic dysfunction–associated steatotic liver disease (MASLD) in UK Biobank data. A to E illustrate the years of life gained among participant with different relative amplitude (RA), most active 10 hours (M10), least active 5 hours (L5), M10S, and L5S profiles in the UK Biobank datasets.

Figure 4.

Relationship between life expectancy gained and light exposure among participants with metabolic dysfunction–associated steatotic liver disease (MASLD) in UK Biobank data. A to F illustrate years of life gained among participant with different daylight or nightlight profiles in the UK Biobank datasets.

We compared the predictive ability of 24-h-RAR parameters and day-night light exposure durations with other risk factors for MASLD using the SHAP method. Among the 26 factors, lower RA emerged as the second-most significant factor for predicting MASLD, ranking just below MASLD genetic risk. Additionally, lower M10 and delayed L5 onset time were ranked as the 7th and 12th most significant factors, respectively, exceeding age 65 and older and higher PM2.5 abundances. Among the rankings for day-night light exposure, the factor of less than 1 hour per day daytime light (≥4000 lux) ranked the highest, placing 13th overall (Fig. 5B).

Figure 5.

Ranked feature importance by Shapley Additive Explanations (SHAP) values for 26 predictive features. A shows the average contribution of each variable in prediction models. B shows the information of how each variable influence the output of models for each participant. SHAP values greater than 0 indicate a positive effect on prediction while values less than 0 indicate a negative effect. Feature importance was calculated from the SHAP value for 26 predictive features for the XGBoost model.

In sensitivity analyses, the results remained generally consistent when excluding individuals who experienced outcomes within the first 2 years; excluding participants with a history of shift work, after adjusting for other covariates, further adjusted for M10 and L5 in the RA analysis, and with missing covariate data were addressed using multiple imputation (Supplementary Tables S15-S21) (12).

Discussion

In this large cross-sectional and prospective cohort study using data from the US NHANES and the UK Biobank, we found that participants with higher RA, higher M10, lower L5, and earlier L5 onset time were associated with a lower risk of MASLD. Prolonged exposure to daylight was linked to a lower risk of MASLD, with higher lighting levels offering a more pronounced protective effect. In contrast, nighttime light exposure showed the opposite effect. Similar associations were observed for the risk of fibrosis and cirrhosis. Furthermore, among participants with MASLD by age 45, those with optimal 24-h-RAR and light exposure had a greater life expectancy compared to individuals with poorer 24-h-RAR and light exposure. Importantly, comparative analyses revealed that lower RA has a greater predictive capability than traditional risk factors, such as alcohol consumption, obesity, and low physical activity.

Several studies have shown that shift work, a severe form of circadian disruption, increases the prevalence and incidence of MASLD (22-24). Building on this, our study examines the association between circadian rhythms and MASLD in the general population, considering both behavioral cycles and environmental exposures. To our knowledge, no prior study has directly investigated the relationship between light-dark exposure and MASLD risk. This study provides robust evidence from 2 large population-based cohorts indicating that daytime light exposure may reduce MASLD risk, with higher light intensity offering greater protective effects. This finding is important, as modern individuals spend nearly 90% of their time indoors under artificial lighting, typically providing only 300 to 500 lux during daytime hours—far below outdoor natural light intensity (25). A controlled crossover trial in overweight women found that a 3-week morning bright light treatment significantly reduced body mass, particularly fat mass (26). The reduction in appetite during the light session compared to the placebo session was evident, which may be attributed to light stimulating serotonin synthesis and turnover in the human brain (26). In this study, we found that the duration and intensity of daylight exposure are positively correlated with vitamin D levels, which were found to have a protective effect on MASLD development (27). Additionally, we observed a dose-response relationship between nighttime light exposure and an increased risk of MASLD, with no minimal exposure threshold; even levels above 5 lux may contribute to increased risk over time. This suggests that even short-term exposure to dim light at night can contribute to the risk of MASLD. For instance, a study of community-dwelling older adults found that nighttime light exposure was associated with metabolic dysfunction, such as a higher prevalence of obesity, diabetes, and hypertension (28). Animal studies also provide evidence that mice exposed to dim nighttime light gained more weight, had reduced glucose tolerance, and experienced disruptions in peripheral clocks, particularly in the liver and adipose tissue, despite similar caloric intake and activity levels (29-31). One possible mechanism underlying these effects involves the suppression of melatonin, a hormone produced by the pineal gland and released during darkness (32). Nighttime light exposure can suppress melatonin secretion and disrupt its release timing (33), potentially delaying its release, impairing pancreatic insulin secretion, and increasing hepatic glucose production (34). Additionally, altered light-dark exposure can affect the gut microbiome and its diurnal rhythms, leading to changes in gut microbiota composition, impaired gut barrier function, and contributing to the development of MASLD (35).

In addition to the effect of light exposure on the central circadian clock, which may elevate the risk of MASLD, our study also identified that disruptions in RARs are similarly associated with an increased risk of MASLD. Previous findings demonstrated a dose-dependent protective relationship between accelerometer-measured moderate-to-vigorous physical activity and the incidence of MASLD. In addition to these findings, our study highlights the importance of not only increasing overall physical activity levels but also of considering the amplitude and temporal distribution of RARs. The Osteoporotic Fractures in Men Study reported that a low amplitude of the RAR, as estimated by RA or cosinor amplitude, was associated with an increased risk of type 2 diabetes and higher BMI (36, 37). Animal studies indicate that exercising in the morning during the active phase has a more significant metabolic effect than nighttime activity during the rest phase, evidenced by increased utilization of carbohydrates and ketone bodies, as well as enhanced lipid and amino acid degradation (38). Recent research also highlights the role of hepatic Bmal1, a key regulator of liver circadian rhythms, in mitochondrial dynamics and metabolic fitness (39). Moreover, circadian misalignment by rest-activity cycles increased dysbiosis (40). The diurnal oscillations in the composition, function, and abundance of the gut microbiota affect metabolic homeostasis in the gastrointestinal tract (41), and its disruption is associated with obesity and metabolic disorders, influencing the host’s lipid metabolism. Disruptions in the circadian rhythm also intensify inflammatory responses and oxidative stress (42), which are key triggers in the development of metabolic disorders in the liver.

In the present study, we also found that 24-h-RAR and light exposure are associated with different stages of liver disease, including steatosis, fibrosis, and cirrhosis. Additionally, our findings suggest that individuals with MASLD have a reduced life expectancy compared to those without this condition, consistent with previous evidence indicating that MASLD patients have, on average, 2.8 years’ shorter life expectancy than controls (43). Moreover, maintaining a favorable 24-h-RAR profile and adequate light exposure among individuals with MASLD was associated with decreased mortality risk and fewer years of life lost. Therefore, monitoring 24-h-RAR and light exposure both in MASLD and non-MASLD individuals may help mitigate the onset and progression of MASLD and reduce mortality rates. These findings highlight the importance of integrating 24-h-RAR and light exposure assessments into clinical practices and public health strategies to enhance the management and prevention of MASLD.

The primary strengths of this study include the objective measurement of 24-h-RAR and light exposure, which minimizes recall bias associated with self-reported data. Additionally, the large sample sizes and extended survey periods of the US NHANES and the UK Biobank enhance the study's robustness. However, several limitations must be acknowledged. First, baseline 24-h-RAR and light exposure were assessed over a 7-day period, which typically achieves high intraclass correlations while capturing behaviors on weekdays and weekends (44). Nevertheless, it remains uncertain whether a 7-day accelerometer measurement adequately reflects longer-term behavioral patterns. Second, light monitoring did not measure light exposure at the ocular level, and the device cannot distinguish between true darkness and instances when it is covered. Consequently, this provides only a rough estimate of light's actual effects on MASLD. Third, we lack information on the date during which data collection occurred, as well as details on shift work in the US NHANES analysis. These factors may influence RARs, light-dark exposure patterns, and metabolic health. Fourth, in the UK Biobank study, lifestyle-related covariates were not collected during the baseline accelerometer mail-out; instead, they were gathered during physical visits to the assessment centers. However, these covariates are generally stable over time (45). Fifth, feeding rhythms may independently influence the liver's circadian clock, apart from light exposure (46, 47). However, we analyzed late-night eating behaviors only in the US NHANES population, which may not fully represent broader populations. Sixth, the small sample size in the life-expectancy analysis limited our ability to adjust for important variables like cardiovascular disease, kidney disease, and cancer, which are linked to premature mortality. This may affect the accuracy of the life-expectancy estimates. Lastly, causal inferences cannot be drawn from our observational study. However, we mitigated this limitation by adjusting for potential confounders and excluding participants diagnosed with MASLD in the first 2 years, ensuring our findings remained consistent.

Conclusion

Our findings indicate that abnormalities in RARs are associated with an increased risk of MASLD. Additionally, higher-intensity and longer-duration light exposure during the day, combined with maintaining a dark environment at night, may help mitigate the risk of developing MASLD and its progression to liver fibrosis and advanced liver disease. This emphasizes the potential of managing circadian rhythms early as a promising strategy for preventing chronic liver diseases.

Abbreviations

BMI

body mass index

DBP

diastolic blood pressure

FBG

fasting blood glucose

FLI

Fatty Liver Index

HbA_1c

glycated hemoglobin A_1c

HDL

high-density lipoprotein

HR

hazard ratio

L5

least active 5 hours

M10

most active 10 hours

MASLD

metabolic dysfunction–associated steatotic liver disease

MVPA

moderate to vigorous physical activity

OR

odds ratio

PA

physical activity

RA

relative amplitude

RAR

rest-activity rhythm

SBP

systolic blood pressure

SHAP

Shapley Additive Explanations

TGs

triglycerides

US NHANES

US National Health and Nutrition Examination Survey

Funding

X.T. was supported by the National Natural Science Foundation of China (82570128), Pioneer R&D Program of Zhejiang (2025C01119), and the Rut and Arvid Wolf Memorial Foundation (2023-02467). H.F. was supported by the National Natural Science Foundation of China (82571696), the National Science and Technology Innovation 2030 of China-Major Projects (2022ZD0214100), Guangzhou Municipal School (College)–Enterprise Joint Funding Project (2024A03J0214), Guangzhou Science and Technology Plan Project (2025A03J3929), and Guangzhou Key Clinical Specialty (Clinical Medical Research Institute). H.W. was supported by the Young Elite Scientists Sponsorship Program by CAST—Doctoral Student Special Plan (156-O-330-0000125-8). The funders had no role in the conduct of the study; collection, management, analysis, or interpretation of the data; preparation, review, or approval of the manuscript; or the decision to submit the manuscript for publication.

Author Contributions

H.W.: conceptualization, methodology, software, investigation, writing—original draft, writing—review and editing, visualization; W.W.: methodology, software, investigation, writing—review and editing; B.W.: methodology, software, investigation, writing—review and editing; J.W.: investigation, writing—review and editing; N.W.: investigation, writing—review and editing; J.Z.: investigation, writing—review and editing; H.F.: methodology, software, investigation, writing—original draft, writing—review and editing; X.L.: methodology, software, investigation, writing—original draft, writing—review and editing; X.T.: conceptualization, software, formal analysis, data curation, investigation, writing—original draft, writing—review and editing, supervision, project administration, funding acquisition.

Disclosures

None of the authors has any potential conflict of interest.

Data Availability

The raw UK Biobank data are protected and are not available due to data privacy laws. Researchers can apply to use the UK Biobank resource for health-related research and public interest via the UK Biobank Access Management System (https://ams.ukbiobank.ac.uk/ams/).