Night shift work and epigenetic modifications: cardiovascular-related miRNA expression within a European cohort of night shift workers

Abstract

Background

The major pathways contributing to cardiovascular disease among night shift workers remain poorly understood. MicroRNAs (miRNAs) have been identified as novel regulators of cardiovascular risk factors. Our objective was to examine whether night shift work was associated with altered expression of five cardiovascular-related miRNAs in night and day shift workers.

Methods

RNA was extracted from plasma of 106 night and 89 day shift healthcare workers, followed by deoxyribonuclease treatment, cDNA synthesis, and quantitative real-time PCR using target-specific assays for five miRNAs (miR-122-5p, miR-107, miR-155-5p, miR-21-5p, and miR-103a-2-5p). In adjusted linear regression models, we examined associations between night shift exposure and the relative expression of miRNAs. We predicted target genes of the miRNAs under study and preformed enrichment analyses to examine related functions and pathways.

Results

Participants were on average 43 years of age and mostly female (94%). Results were suggestive of higher miRNA expression levels among night shift workers, most notably for miR-155-5p (coefficient 0.46, 95%CI -0.03, 0.94) and miR-103a-2-5p (coefficient 0.45, 95%CI -0.06, 0.96), however results did not reach statistical significance. In dose-response analyses, for each additional year of prior night shift work duration history, the relative expression of several miRNAs was higher. Increased duration of night shift work was significantly associated with higher levels of miR-22-5p (coeff 0.03, 95%CI 0.00, 0.05), miR-107 (coeff 0.04, 95%CI 0.01, 0.06), miR-21-5p (coeff 0.04, 95%CI 0.01, 0.07) and miR-103a-2-5p (coeff 0.02, 95%CI 0.00, 0.05). The enrichment analyses identified several pathways and functions (e.g. lipid and atherosclerosis) related to cardiometabolic risk.

Conclusions

Night shift work appeared to be associated with an elevated expression of cardiovascular-related circulating miRNAs. The pathway analyses of downstream target genes provided evidence of their involvement in a variety of cardiometabolic disease mechanisms. This study presented new evidence linking night shift work exposure to cardiometabolic markers via epigenetic regulation.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12940-026-01308-5.

Introduction

Internal biological processes are regulated by circadian rhythms, which are synchronized primarily by the 24-hour light and dark cycle [1], with smaller influences from external factors such as meal and exercise timing. In healthy workers with normal sleep patterns and non-shift work schedules, the biological clock drives wakefulness during the daytime and a decrease of alertness and a corresponding drive for sleep in the nighttime [2]. However, night shift workers experience circadian misalignment because they remain awake during the biological night, which disrupts the timing of several key inputs to the circadian system [3]. Circadian misalignment can have deleterious impacts on human health, including negative cardiometabolic health outcomes [4].

Epidemiological studies have consistently shown an association between night shift work and the increased risk of cardiometabolic diseases including metabolic syndrome [5] and hypertension [6], among others [7]. However, information on the mechanisms linking night shift work to cardiometabolic disease is lacking, particularly how the environmental stressors of night shift work (including circadian disruption, changes in sleep, artificial light at night exposure and others) can result in stable, yet reversible, changes in cardiometabolic health which can promote disease. One likely mechanism is the epigenetic regulation of gene expression via microRNAs (miRNAs).

MiRNAs are small (18–25 nucleotides in length), single-stranded, non-coding RNA species that have key actions in post-transcriptional gene regulation [8], which can influence an array of physiological processes. MiRNAs have mainly been studied for their pivotal role in pathways including apoptosis, cell cycle function, and DNA damage repair [9, 10]. Recently, miRNAs have also been studied in relation to cardiovascular [11] and metabolic diseases [12].

Because circulating miRNAs are protected from ribonuclease degradation, their plasma levels can serve as biomarkers of cardiometabolic diseases [13]. However, few studies to date have examined miRNA expression within shift workers [14–17], all of which were very small in number (n ≤ 30).

In this study, our aim was to we examine whether night shift work was associated with altered expression of 5 miRNAs (hsa-miR-122-5p, hsa-miR-107, hsa-miR-155-5p, hsa-miR-21-5p, hsa-miR-103a-2-5p) related to cardiovascular outcomes and circadian rhythms (Supplemental Table 1) [18–25]. These epigenetic changes may indicate key pathways related to cardiometabolic comorbidity among this population.

Methods

Study population

Data from this study came from the Exposome Project for Health and Occupational Research Night Shift Study (EPHOR- NIGHT), a study which applied an exposome approach to collect data from night shift workers [26]. The EPHOR-NIGHT study enrolled male and female night shift workers (exposed) and day shift workers (unexposed) from four EU countries (N = 937). Information on this study is available elsewhere. Briefly, each participant completed data collection over 5 days. Relevant to the present analyses, blood was collected from each participant after 2 days of their typical shift schedule at 07:00 h (+/-1 h allowed) to ensure homogeneity for all shifts, and participants responded to a baseline questionnaire and wore sensors (light, actigraphy) over the course of 5 days. The present study incorporated data from a subset of 189 participants from 2 of the 4 centers, including 100 from the Swedish center and 89 from the Spanish center who had valid blood samples analyzed for miRNA expression and had complete information on the study covariates. The study received ethical approval from CEIm-Parc de Salut Mar approval #2021/9845 (Spain) and Stockholm Ethics Review Authority approval #2021–02386 (Sweden) and all participants gave written informed consent.

Exposure

Our primary exposure was night shift work, which was a binary variable. At the time of sampling, to be eligible for the study, night-shift workers were required to be working a minimum of two consecutive night shifts per week, and to have worked night shifts for 3 + years prior to enrollment. A night shift was defined as a working schedule that involved working 4 + hours of the shift between 00:00 h and 06:00 h. For day shift workers, at the time of sampling, workers were not allowed to be currently working night shifts, nor have worked night shifts for the prior 3 years at the time of enrollment. Incidental night shift work was allowed (working less than one night shift per month, however, ≥ 1 week must have elapsed after an incidental night shift before the day shift worker was enrolled into the study). A day shift was defined as a working schedule that involved working 4 + hours of the shift between 07:00 h and 18:00 h (shift must start at 6:00am or afterward). In addition to examining associations with night shift work, we further examined how night shift characteristics including (a) # consecutive night shifts/week, (b) intensity of night shifts/week and (c) duration of prior night shift work history was related relative expression levels of the miRNAs of interest.

Outcome

Plasma collection and RNA extraction

Venous blood samples were collected from the forearm in BD Vacutainer EDTA Tubes. Plasma was separated by centrifugation, and the plasma supernatant was transferred to cryotubes and kept at -20 ℃ until transportation to the laboratory at the Division of Occupational and Environmental Medicine, Lund University. They were then stored at -80 °C until analysis.

The plasma samples were thawed in a 37 °C water bath, and 200 µL of each plasma sample were used to isolate total RNA using the miRNeasy Serum/Plasma Advanced Kit (Qiagen, Hilden, Germany), according to the manufacturer’s protocol. An additional step of deoxyribonuclease treatment with the RNase-Free DNase Set (Qiagen) was added in order to digest DNA during the RNA purification process. Plasma total RNA was eluted in a final volume of 20 µL of RNAse-free water. RNA samples were put on ice and then reverse transcribed to form complementary DNA (cDNA).

cDNA synthesis

The cDNA conversion was completed the same day as the RNA extraction. To minimize variability in input material, a fixed volume of starting biofluid (200 µL plasma) was used for all samples, and all samples were processed using the same extraction and RT-qPCR workflow. A total of 10 µL of the RNA sample was added to 15 µL of RNAse free water, and then 2 µL of diluted RNA was reverse transcribed using the miRCURY LNA RT Kit (Qiagen), according to the manufacturer’s protocol. The synthetic RNA, UniSp6 RNA (spike-in RNA), which has a structure resembling miRNA, was provided by the miRCURY LNA RT Kit and added to the solution as an exogenous control. This served to normalize miRNA expression as well as any potential differences in reaction efficiency between samples, as explained by Cuomo-Haymour et al [27]. This approach is widely used in circulating miRNA studies, particularly when robust endogenous references are unavailable, and is considered an appropriate strategy for minimizing technical variability in plasma-based miRNA quantification [28–30]. Reverse transcription was conducted using the following thermal cycling parameters: 60 min at 42 °C, 5 min at 95 °C and on hold at 4 °C. Afterward, the cDNA samples were stored at -20 °C.

Quantitative real-time polymerase chain reaction (qPCR)

The measurement of miRNA expression level was described previously [31]. In brief, the expressions of miRNAs were detected by qPCR using miRCURY LNA SYBR Green PCR Kit (Qiagen) according to the manufacturer’s protocol in a 7900HT Fast Real-Time PCR System. The cDNA samples were diluted 1:30 by adding 8 µL cDNA to 232 µL RNAse free water, immediately before qPCR. The qPCR amplification mixtures contained 3 µL diluted template cDNA, 5 µL 2x miRCURY SYBR Green master mix, 0.5 µL ROX reference dye, 0.5 µL RNAse free water and 1 µL miRCURY LNA miRNA PCR Assay (Qiagen). The assays used were as follows: UniSp6 (GeneGlobe ID - YP00203954), hsa-miR-122-5p (YP00205664), hsa-miR-107 (YP00204468), hsa-miR-155-5p (YP02119311), hsa-miR-21-5p (YP00204230), hsa-miR-103a-2-5p (YP00204609). Samples from the two centers were run together on the qPCR plates and the samples were randomized. For each miRNA, samples were run in triplicate. The thermal cycling parameters of qPCR begin with an initial heat activation at 95 °C for 2 min, followed by 45 cycles of a 2-step process consisting of denaturation at 95 °C for 10 s and a combined annealing/extension step at 56 °C for 60 s.

Covariates

In addition to information on night shift exposure, and outcome data on miRNA relative expression levels, we also incorporated additional data on: age (continuous years), sex (male, female), body mass index [(BMI) continuous using in-person measurements of height and weight], smoking (never, former, current), alcohol use (never, former, current), physical activity [metabolic equivalents of task (METs) assessed using patient reported information from the International Physical Activity Questionnaire [32], and study center (Spain, Sweden). These covariates were chosen a priori as confounders of the relationship between night shift work and miRNA relative expression level and therefore were included in adjusted models. We also had information on hypertensive medication use and blood pressure values (mm Hg). Using this information, we defined hypertension if the participant met any of the following criteria: (a) systolic blood pressure greater than 120 mmHg, (b) diastolic blood pressure greater than 90 mm Hg, or (c) hypertensive medication use based on information from the baseline questionnaire. Using the BMI values, we applied the WHO cut-off guidelines to categorize BMI into underweight, normal, overweight and obese, and defined a binary BMI outcome which was underweight/normal vs. overweight/obese.

Statistical analyses

We described characteristics of the study population stratified by night shift status. We also described expression of miRNAs for the complete sample with miRNA data and stratified by (a) night shift status and (b) center.

We calculated the relative expression levels for each of the 5 miRNAs. Normalized quantification cycle values [ referred to as Δ cycle threshold (Ct)] were calculated as the difference between the mean Ct of the candidate miRNAs and the Ct of Unisp6. The relative expression levels of miRNAs were then calculated using the 2-ΔΔCt method [33, 34]. In determining the Ct from PCR, we excluded any of the triplicates which were undetermined or did not amplify in parallel to the other triplicates. Then, from the remaining values per sample, we calculated the mean Ct. For all samples the differences between max and min Ct per miRNA was below 1. Because of the skew in relative expression values, we log-transformed the values using the natural logarithm prior to performing regression models.

We then performed linear regression models to examine the associations between night shift exposure and relative expression of each miRNA, adjusting for the aforementioned covariates. We additionally ran regression models to examine the associations between (a) consecutive night shifts/week (reference day shift workers) as well as (b) number of night shifts/week (intensity, reference day shift workers) to examine whether a greater exposure to night shift work was associated with differences in relative expression levels.

We also performed two sensitivity analyses. First, we performed center-stratified analyses. Second, we re-ran analyses only among female participants, as they accounted for the vast majority of the total study population.

Pathway analyses

We predicted target genes of the key miRNAs under study using the miRTarBase [35] and TargetScanHuman [36] databases. Presumed target genes from miRTarBase were selected if they were validated by either reporter assay, Western blot analysis or qPCR. Presumed target genes from TargetScanHuman were selected if the Cumulative weighted context + + score (represents the predicted overall repression strength of a miRNA on a target gene by summing the effects of all its binding sites) was lower than 0.08 [37]. All genes meeting these criteria were kept for use in further analyses.

Gene Ontology (GO) category analysis provides a comprehensive annotation for functional attributes of genes, encompassing their (a) biological process, (b) cellular component and (c) molecular function [38]. In addition, the Kyoto Encyclopedia of Genes and Genomes (KEGG) employs a pathway-centric approach to gene classification, where genes with similar functions are organized into distinct pathways [39]. The GO and KEGG enrichment analyses were performed using an online platform SRplot (https://www.bioinformatics.com.cn/SRplot) [40]. The thresholds for enrichment analyses were set as an adjusted P-value less than 0.05. Bubble maps and bar charts were generated by Bioinformatics.

Finally, because of the highly correlated nature of the 5 miRNAs, to reduce dimensionality, we performed a secondary analysis using principal components analysis (PCA) methods to reduce dimensionality and summarize the miRNAs into a single component.

Results

This study included 189 participants, of whom 106 (56%) were night shift workers and 178 (94%) were women. Compared to day shift workers, night shift workers were more likely to currently smoke (13% vs. 6%), although less likely to currently use alcohol (62% vs. 74%). Night shift workers were more likely to have hypertension (32% vs. 23%). Night shift workers in our sample worked on average 4 shifts/week (standard deviation [SD] 1), and worked on average 3 consecutive shifts/week (SD 1) (Table 1).

Table 1.

Characteristics of study population based on shift work exposure for N = 189 participants

	Day shift (n = 83)	Night shift (n = 106)
Center
Spain	34 (41)	55 (52)
Sweden	49 (59)	51 (48)
Age (mean (SD))	43 (12)	45 (11)
Female (%)	79 (95)	99 (93)
Education (%)
Secondary studies	23 (28)	31 (29)
University or more	60 (72)	75 (71)
Smoking (%)
Never	67 (81)	81 (76)
Former	11 (13)	11 (10)
Current	5 (6)	14 (13)
Alcohol use (%)
Never	2 (2)	5 (5)
Former	20 (24)	35 (33)
Current	61 (74)	66 (62)
Total METS (mean (SD))	2334 (1794)	2275 (2189)
BMI (mean (SD))	26 (5)	27 (6)
Hypertension (%)	19 (23)	34 (32)
Average consecutive night shifts/week (mean (SD))	N/A	3 (1)
Average night shifts/week (mean (SD))	N/A	4 (1)
Average duration of prior night shift work history (mean (SD))	1.7 (3.0)	14 (10)

We examined the expression levels stratified by center, and observed that the transformed relative expression levels were higher in the Swedish subsample for all of the miRNAs (Fig. 1).

Fig. 1.

Histograms showing different relative expression levels by center (relative expression levels are log transformed)

Adjusted regression results were suggestive of higher expression levels of the miRNAs in night shift workers, for all of the miRNAs, however results were not statistically significant (Table 2). This trend was most notable for miR-155-5p (coefficient 0.46, 95%CI -0.03, 0.94) and miR-103a-2-5p (coefficient 0.45, 95%CI -0.06, 0.96), as well as for miR-22-5p (coefficient 0.13, 95%CI -0.31, 0.58), miR-107 (coefficient 0.37, 95%CI -0.12, 0.85) and miR-21-5p (coefficient 0.38, 95%CI -0.11, 0.87). The center-stratified results were more variable (Supplemental Table 2), but still suggested an overall pattern of higher miRNA expression for the majority of miRNAs.

Table 2.

Relative expression levels for N = 189 participants comparing levels in night shift workers to the reference day shift worker levels^a

	Coeff.	95% CI Lower	95% CI Upper
miR-22-5p	0.13	-0.31	0.58
miR-107	0.37	-0.12	0.85
miR-155-5p	0.46	-0.03	0.94
miR-21-5p	0.38	-0.11	0.87
miR-103a-2-5p	0.45	-0.06	0.96

^aAdjusted for: age, sex, BMI, smoking, alcohol use, physical activity and center

We further examined how night shift characteristics including (a) # consecutive night shifts/week, (b) intensity of night shifts/week and (c) duration of prior night shift work history was related to relative expression levels. Our results suggest that for each additional consecutive night shift/week, there was an association with higher relative expression levels of miRNAs (except for miR-155-5p, which was null). Our results also suggested that each additional night shift/week (greater intensity) was associated with higher relative expression levels of all miRNAs (Table 3). We found statistically significant differences in our analyses examining how duration of prior night shift work history related to miRNA expression level. Our results suggest that for each additional year of prior night shift work duration history, the relative expression levels of miR-22-5p (coefficient 0.03, 95%CI 0.00, 0.05), miR-107 (coefficient 0.04, 95%CI 0.01, 0.06), miR-21-5p (coefficient 0.04, 95%CI 0.01, 0.07) and miR-103a-2-5p (coefficient 0.02, 95%CI 0.00, 0.05) were higher. In sensitivity analyses which restricted to only female participants, the results were similar to the main study findings (Supplemental Tables 3 and 4).

Table 3.

Examining relationships between (a) consecutive number of night shifts, (b) intensity of night shifts and (c) history of night shift work with miRNA relative expression levels^a

	Consecutive			Intensity			History
	Coefficient	95% CI lower	95% CI upper	Coefficient	95% CI lower	95% CI upper	Coefficient	95% CI lower	95% CI upper
miR-22-5p	0.07	-0.06	0.20	0.04	-0.13	0.20	0.03	0.00	0.05
miR-107	0.13	-0.02	0.29	0.12	-0.07	0.32	0.04	0.01	0.06
miR-155-5p	-0.01	-0.15	0.14	0.03	-0.15	0.20	0.02	-0.01	0.04
miR-21-5p	0.14	-0.02	0.30	0.12	-0.08	0.32	0.04	0.01	0.07
miR-103a-2-5p	0.06	-0.09	0.20	0.05	-0.13	0.23	0.02	0.00	0.05

^aAdjusted for: age, sex, BMI, smoking, alcohol use, physical activity and center

We also examined the correlation between each miRNA as well as between the miRNAs and cardiovascular endpoints including continuous systolic blood pressure and BMI as well as binary hypertension and overweight/obesity status (Fig. 2). We observed a strong positive correlation between miR-107, miR-155-5p, miR-21-5p, and miR-103a-2-5p, while miR-122-5p was moderately associated with the other 4 miRNAs. There was low correlation between the miRNA relative expression levels and blood pressure or body mass index. There was also low correlation between age and the miRNAs.

Fig. 2.

Heatmap of Spearman correlation coefficients between miRNAs and other covariates

We retrieved predicted target genes associated with the miRNAs under study. After removing duplicates, a total of 553 genes were associated with the five miRNAs (Supplemental Table 5). The potential functions of these genes were revealed by KEGG and GO enrichment analysis (Figs. 3 and 4). The KEGG pathway enrichment analysis indicated that the target genes identified were significantly enriched in actions related to cardiometabolic health including: AGE-RAGE signaling pathway in diabetic complications, MAPK signaling pathway, TNF signaling pathway and lipid and atherosclerosis (Fig. 3). Moreover, when summarizing the top ten GO terms for each category (1. biological processes, 2. cellular component and 3. molecular) (ranked by enrichment score), several key processes, components and molecules were highlighted which relate to regulation of gene expression, reflecting the role of miRNA in gene expression, as well as cardiometabolic risk and disease progression (e.g. response to lipopolysaccharide, response to molecule of bacterial origin, and protein serine/threonine kinase activity; Fig. 4).

Fig. 3.

The top 10 terms from the Kyoto Encyclopedia of Genes and Genome (KEGG) pathway analysis

Fig. 4.

Gene Ontology (GO) function enrichment analysis. GO analysis for the target genes of differentially expressed miRNAs (A). The top 10 GO terms of three categories were presented (biological processes, cellular component, molecular fraction. The left x-axis values are enrichment scores (-log 10 P-value). The right x-axis values are gene counts

Finally, when we performed the PCA analysis, PC1 alone explained 92% of the variance and was therefore used as a composite miRNA score. In regression analyses, night-shift exposure showed no significant association with this composite measure (β = -0.0057, p > 0.05, data not shown), suggesting that night-shift work did not meaningfully influence the overall miRNA expression pattern.

Discussion

Through the exploration of miRNA relative expression levels of 5 candidate genes in a cohort of night and day shift workers, we found indications of higher relative miRNA expression among the night shift workers compared to day shift workers. Our results suggested that for each additional consecutive night shift/week, there was an association with higher relative expression of most miRNAs, meanwhile each additional night shift/week was associated with higher relative expression of all miRNAs, and in particular each additional prior year of night shift work history was associated with a significantly higher relative expression of all but one miRNA, suggesting a likely dose-response relationship between increasing night shift exposure and higher relative expression levels.

Higher relative expression of miRNAs in night shift workers may suggest lower expression of downstream target genes encoding proteins, because miRNAs bind to messenger RNAs (mRNA) most often resulting in mRNA degradation and reduced protein expression [41]. There are so far very few and small studies on circadian disruption and miRNA alterations. One study on simulated rotary shift schedule among 15 individuals reported higher expression of miRNAs, targeting phosphate transport-encoding genes [15]. An additional study found lower expression of miR-92a, related to higher brown adipose tissue activity, in 15 night shift compared with 15 daytime workers [16]. Both of these studies examine miRNA mechanisms that are relevant for cardiometabolic health, which was also a key focus of the present study.

While we did not find a correlation between the relative expression levels of miRNAs and the blood pressure nor BMI-related outcomes, they may, as characterized in the enrichment analysis, be relevant for several pathways and cellular functions and in turn to poor cardiometabolic outcomes. In particular, the AGE-RAGE signaling pathway in diabetic complications, the MAPK signaling pathway, the TNF signaling pathway and lipid and atherosclerosis were identified. All pathways have been studied extensively for their role in cardiometabolic outcomes, particularly the role that lipids and atherosclerosis play in the pathogenesis of cardiovascular disease [42–44]. The GO analyses also described functions relevant for cardiovascular health including for instance response to lipopolysaccharide and response to molecule of bacterial origin, which are examples of inflammatory-related responses, a major risk factor for cardiometabolic disorders [45]. Further, several molecular functions (e.g. ubiquitin protein ligase binding, SMAD binding, protein serine/threonine kinase activity) are all relevant for insulin signaling or resistance and highly relevant in metabolic disorders [46, 47].

There are several explanations for why we observed that these miRNAs are related to cardiometabolic pathways and cellular functions while not finding a direct relationship between miRNA relative expression and blood pressure or obesity outcomes in our population. First, it is possible that the miRNA changes we are seeing are relatively acute, whereas clinical blood pressure or obesity outcomes are chronic in nature. MiRNAs often change rapidly in response to environmental exposures (could include shift work or other occupational exposures, as well as air pollution, chemicals in food and drinking water etc.) [48] while cardiovascular outcomes develop gradually, generally over months or years. Therefore, the miRNA differences seen may be early indicators, reflecting acute physiological responses before the onset of chronic disease. Alternatively, or additionally, we may have identified miRNA regulated genes/pathways which are upstream drivers of cardiometabolic disease outcomes (e.g. metabolism, inflammation), but not see a direct connection between miRNA expression levels and these outcomes because these phenotypes are additionally influenced by other genetic or behavioral factors, which we were unable to address here. Third, these miRNAs may be related to other cardiometabolic pathways entirely e.g. atherosclerosis or lipid levels, but not with blood pressure or obesity which we examined here.

This study has several limitations and strengths. First, our population included healthcare workers, who were predominantly female. Moreover, although we have detailed night shift work data, we cannot exclude confounding by other occupational factors, specific job tasks, staffing, or specific job type. Therefore, the findings may not generalize well to other work sectors, nor to males. In addition, despite that we used the same standardized sampling protocol we saw differences between the two sampling centers in the relative expression levels. In particular, the Spanish samples had lower concentrations than the Swedish samples. We cannot explain this difference by classical confounders such as socioeconomic status (SES) differences nor artifacts in the lab. In our sample, education (a marker of SES) is very similar both between the two centers as well as between shift types, and therefore it is unlikely to influence the study findings. When we examined the months elapsed from blood sample collection to analyses, which could influence degradation of samples and result in lower relative expression concentrations, this was 3.2 months longer on for samples from Spain. When we examined whether there was evidence of correlation between month of sampling and relative expression level using Pearson correlation, we saw low or negligible correlation (all p-values < 0.33), indicating that this alone is unlikely to explain center-differences. Furthermore, both the combined regression analyses as well as the center-stratified analyses showed a similar pattern of over expression among the night shift workers. In addition, we took a targeted gene approach, focusing on 5 candidate genes known to be associated with cardiovascular health, whereas an agnostic approach to examine miRNA alterations more widely across the genome would be expected to provide different and useful information to supplement these findings. There was strong correlation between four of the five miRNAs and we cannot exclude that this is due to higher overall miRNA yield for some participants. It should be noted that we extracted specifically small RNAs and conventional total RNA quantification is therefore not informative. Instead, we included multiple controls for extraction and reaction efficiency, which showed consistent performance across samples. In addition, the miRNAs selected for analysis in this study were not chosen at random, but were a priori selected based on prior evidence linking them to cardiovascular and cardiometabolic diseases. As such, these miRNAs are known to participate in overlapping biological pathways and may be subject to shared upstream regulatory mechanisms, including common transcriptional control, inflammatory signaling, metabolic stress responses, or coordinated release into circulation, rather than technical artifacts or differences in total RNA input. Finally, the timing of sampling matters when examining miRNA concentrations because miRNAs have their own circadian rhythms and are highly dynamic molecules that respond to daily-or hourly stimuli. In this study, all blood samples were collected 07:00 h (+/-1 h allowed) to ensure homogeneity for all shifts, and therefore these concerns were minimized. However, individual exposures that could influence miRNA levels are diverse and may have differed between individuals, particularly considering that 7:00 h corresponded to the end of a night shift after several hours of wakefulness, while for the day shift workers, this was relatively soon after awakening. Ideally, multiple samples throughout the day would be most informative to better capture miRNA dynamics.

In addition, this study has several strengths including its relatively large sample size of 189 individuals, which is much larger than the existing studies of miRNA expression among night shift workers. We also had detailed shift exposure information including number of consecutive night shifts/week and intensity of night shifts/week and therefore were able to examine if there was any evidence of dose-response relationships between our shift exposure. We also had access to cardiovascular outcome data from the EPHOR-NIGHT study, so we could examine potential associations between the miRNA expression levels and these outcomes in our population.

In conclusion, night shift work appeared to be associated with a pattern of elevated expression of circulating miRNAs in all of our analyses, including dose response relationships with night shift work duration. Higher miRNA relative expression levels suggest lower expression of downstream protein-encoding target genes, and our pathway analyses characterized that target genes of these miRNAs are related to a variety of cardiometabolic disease mechanisms. This study presented new molecular data linking night shift work exposure to cardiometabolic markers via epigenetic regulation, which should be replicated in additional studies. Future studies should incorporate longitudinal follow-up and validated cardiovascular event data to elucidate the causal links between epigenetic modifications and disease development.

Abbreviations

miRNA

microRNA

EPHOR

NIGHT-Exposome Project for Health and Occupational Research Night Shift Study

cDNA

complementary DNA

BMI

Body mass index

METs

Metabolic equivalents of task

CT

Cycle threshold

GO

Gene ontology

KEGG

Kyoto Encyclopedia of Genes and Genomes

SD

Standard deviation

Authors’ contributions

BH and KB conceptualized the research project and designed the study. BH, MK, MA and KB acquired funding to support the study. BH, ZJ, DP collected and generated data. BH and ZJ analyzed data. BH and KB drafted the manuscript text. All authors critically reviewed drafts and approved the manuscript for publication.

Funding

This project has received funding from the European Union’s Horizon 2020 research and innovation programme under grant agreement No 874703. The Swedish part of the project and data collection was additionally supported by funding from The Swedish Research Council FORTE under grant agreement 2022 − 00322, ALF Region Stockholm, and Afa försäkring under grant agreement 220075. BH has received funding from the Instituto de Salud Carlos III (ISCIII) and the European Union NextGenerationEU/PRTR (Grant IHMC22/00017).

Data availability

Access to study data is possible through a completed data request form, available at https://www.isglobal.org/en/-/ephor. Reasonable requests will be reviewed and approved by the steering committee.

Declarations

Ethics approval and consent to participate

The study received ethical approval from CEIm-Parc de Salut Mar approval #2021/9845 (Spain) and Stockholm Ethics Review Authority approval #2021–02386 (Sweden) and all participants gave written informed consent.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.