The Central Clock and Molecular Clockwork

In 2017, the Nobel Prize in Physiology and Medicine was awarded to Jeffrey C. Hall, Michael Rosbash, and Michael W. Young for their discoveries which clarified the molecular mechanisms underlying self-sustained intracellular circadian oscillations (Young, 2018). The core of the molecular clockwork is composed of transcriptional autoregulated feedback loops and a set of clock genes whose transcription can generate and maintain circadian rhythms in the absence of environmental time cues (Honma, 2018; Cox and Takahashi, 2019). At the cellular level, the clock genes CLOCK and ARNTL (alias BMAL1) encode activators of the main feedback loop, whereas period circadian regulators 1 to 3 (PER1, PER2, PER3) and cryptochrome circadian regulators 1 to 2 (CRY1, CRY2) encode repressors. The processes of translation/transcription and accumulation/degradation of these core components form the basic circadian clock and cycles with a period of approximately 24 h. Besides these main core clock genes and associated feedback loops, additional components have been described, adding to the complexity and precision of the clockwork mechanisms. Clock-controlled genes are considered the molecular output of the circadian clock as they are the links between the core clockwork and observable rhythms in cells, tissues, functions, and behaviors.

Peripheral Clocks

Today, we know that the molecular clockwork underlying circadian rhythms is intrinsic to most cells and tissues in mammals (Yamazaki et al., 2000), including in humans (Bjarnason et al., 2001; Archer et al., 2008; Akashi et al., 2010; Cuesta et al., 2017; Kervezee et al., 2019b; Du and Brown, 2021). The suprachiasmatic nucleus (SCN) is considered the central clock, while other tissues generating self-sustained circadian rhythms are named peripheral clocks (Brown et al., 2019). Contrary to the SCN which can self-sustain circadian rhythms for weeks without environmental cues, the oscillations observed in peripheral tissues usually dampen after a few days in vitro at the tissue level (Yamazaki et al., 2000), but not necessarily at the individual cell level (Welsh et al., 2004).

The existence of peripheral clocks in humans was demonstrated in oral mucosa cells (Bjarnason et al., 2001), then in peripheral blood mononuclear cells (PBMCs) and other blood cells (Boivin et al., 2003; Kusanagi et al., 2004; James et al., 2007a), in hair follicles (Akashi et al., 2010), skin cells (Brown et al., 2005; Wu et al., 2018), and in adipose fat tissues (Gomez-Abellan et al., 2008; Garaulet et al., 2011). We and others have shown the existence of non-SCN clocks in human post-mortem brain tissues, with desynchronized or dampened rhythmicity in Alzheimer’s disease or major depressive disorders (Cermakian et al., 2011; Li et al., 2013; Lim et al., 2013). A comprehensive study of the transcriptome carried out in 64 tissues collected in 12 baboons (one animal per time point) revealed that more than 80% of the detected protein-coding genes exhibited a 24-h rhythm in at least one tissue, with only limited overlap between tissues (Mure et al., 2018). Interestingly, Ruben et al. (2018) used an open access database of post-mortem RNA-sequenced human donor samples, combined with an automatic ordering algorithm, to create a population-level atlas of gene expression in 13 human tissues. They showed that 44% of the protein-coding genes cycled in at least one tissue studied.

Resetting of Circadian Clocks

Light is the most powerful synchronizer of the central circadian pacemaker in humans. The light-dark information is captured by the retina via specialized intrinsically photosensitive retinal ganglion cells expressing the photopigment melanopsin (Provencio et al., 2000; Yamazaki et al., 2000; Berson et al., 2002; Panda et al., 2002; Ruby et al., 2002). Based on animal studies, rods and cones are not necessary to photoentrainment (Freedman et al., 1999), but probably modulate the response to light (Foster et al., 2020). The light information is then directly transmitted to the SCN via monosynaptic connections of the retinohypothalamic tract. In humans, light exposure in the morning causes a phase advance of the circadian system, whereas exposure in the evening/early night causes a phase delay (Czeisler et al., 1989; Minors et al., 1991; Czeisler and Buxton, 2011; Vetter et al., 2021). Light administered in the middle of the day exerts only a small or non-discernible effect. The resetting effect of a light stimulus is also influenced by its intensity (Boivin et al., 1996; Zeitzer et al., 2000), duration (Dewan et al., 2011) and spectral composition (Ruger et al., 2013). Exposure to light can displace not only rhythms regulated by the central circadian clock such as those of cortisol and melatonin secretion, but also those of peripheral clocks (Yamazaki et al., 2000; James et al., 2007b; Ackermann et al., 2009; Cuesta et al., 2017; Kervezee et al., 2019b). Initial studies have suggested that the central circadian clock can be shifted faster than peripheral clocks in humans (James et al., 2007b) and rats (Yamazaki et al., 2000). However, it was later shown that bright light exposure in humans can synchronize peripheral clock gene expression more rapidly than previously suspected (Cuesta et al., 2017).

In humans, only a few nonphotic stimuli have been reported to modify the rhythmicity of the central circadian system under very dim light conditions or in free-running blind individuals, including exogenous melatonin (Lockley et al., 2000; Burgess et al., 2010) and exercise (Buxton et al., 2003; Barger et al., 2004). Social contacts have also been proposed to entrain the circadian system in mammals and humans (Mistlberger and Skene, 2004), but their resetting effects remain controversial as social interactions may rather modulate the daily pattern of light-dark exposure than directly shift the circadian system.

Circadian Misalignment

The various non-standard hours and rosters required of shift workers inevitably force abrupt changes in the sleep-wake and light-darkness schedules to which the central and peripheral clocks are usually entrained. These changes result in circadian misalignment, which describes a state of desynchronization between circadian clocks and the environment (Boivin and James, 2002a; Boudreau et al., 2013a; Skene et al., 2018; Kervezee et al., 2019a). Another immediate effect of night-shift work is the reduction in amplitude or distortion of circadian rhythms such as those of melatonin and cortisol secretion (Touitou et al., 1990; Dijk et al., 2012; Mirick et al., 2013). As schematically presented in Figure 1, simulated night-shift experiments and field-based studies with shift workers both indicate that the central clock is resistant to adaptation from a day-oriented to a night-oriented schedule, as determined by the magnitude of phase shifts in rhythms of melatonin, cortisol, and body temperature over multiple days (Crowley et al., 2004; Boudreau et al., 2013a; Jensen et al., 2016; Molzof et al., 2019; Resuehr et al., 2019; Jensen et al., 2020). Thus, the nadir of cortisol, peak of melatonin, and trough of body temperature, which normally occur in the first, middle, and last thirds of the nocturnal sleep period, coincide with wake periods during night shifts (Boivin and James, 2002a; Benloucif et al., 2005; Resuehr et al., 2019). This misalignment of endogenous rhythms with the shifted sleep-wake cycle means that shift workers must perform their tasks and sleep at incongruous biological times.

Open in a separate window

Figure 1.

Disruption of central and peripheral rhythms by night-shift work. Under a night-oriented schedule, group rhythms are misaligned relative to the shifted rest-activity cycle and dampened in amplitude. Yellow and gray rectangles represent the environmental light and dark cycles, respectively. Rhythms are adapted from Cuesta et al. (2017) J Biol Rhythms 23; Cuesta et al. (2017); Koshy et al. (2019); Hattammaru et al. (2019). Abbreviation: PBMC = peripheral blood mononuclear cell.

Internal Desynchrony

There is evidence that disruption of the circadian system caused by night-shift work results not only in a misalignment between the circadian system and the external light-dark cycle but also in a state of internal desynchronization between several levels of the circadian system. We demonstrated this between rhythms controlled by the central circadian pacemaker (e.g., core body temperature, melatonin, cortisol) and those expressed in peripheral tissues (see Figure 1, central vs. peripheral rhythms). The discovery that rhythmic clock gene expression could be observed in PBMCs led to the exploration of the impact of a night-oriented schedule on this peripheral clock (Boivin et al., 2003; James et al., 2007a, 2007b; Cuesta et al., 2017). Under controlled laboratory conditions, rhythms of PER1, PER2, PER3, and BMAL1 clock genes expression desynchronized from the sleep-wake cycle and from each other after 3 days on a night-oriented schedule (Cuesta et al., 2017). While PER1 and BMAL1 rhythms delayed by ~2.5 to 3 h, other clock genes and rhythms of cortisol and melatonin remained adjusted to a day-oriented schedule. Three cycles of 8-h bright light exposure at night induced significant phase delays of ~7 to 9 h for central and peripheral markers, except BMAL1 (advanced by +5 h 29 minutes), thus demonstrating their endogenous circadian nature.

The disruptive effects of atypical work schedules extend beyond the expression profile of canonical circadian clock genes and affect other transcripts of the human transcriptome. In a simulated 4-night-shifts laboratory study of the transcriptome, we demonstrated that about 11.8% of transcripts in the human genome were rhythmic, and the majority of these rhythms did not adjust to a night schedule (Kervezee et al., 2018). In general, amplitudes of probe sets that were rhythmic in both conditions were significantly reduced in the night-shift condition compared with baseline. These results are consistent with those of Archer et al. (2014) who demonstrated, using a forced desynchrony protocol, that circadian disruption produced a 6-fold reduction in circadian transcripts compared with when sleeping in phase with the melatonin rhythm. In addition to circadian misalignment, sleep restriction can affect the expression of the human transcriptome and alter its circadian expression. It remains to be determined whether these rhythmic transcripts are under circadian control or are rather linked to the rest-activity cycle. The circadian nature of peripheral rhythmic transcripts is supported by their sensitivity to light-induced phase shifts (Möller-Levet et al., 2013; Archer et al., 2014; Arnardottir et al., 2014; Kervezee et al., 2019b).

Research conducted with real shift workers demonstrates a similar resistance of peripheral clocks to adapt to a night-oriented schedule (Akashi et al., 2010; Koshy et al., 2019). Akashi et al. (2010) found the phase of clock genes expression in hair follicle cells to be delayed by only ~2 h on late shifts (1500-0000 h) despite a ~7-h delay in behavioral rhythms relative to early shifts (0600-1500 h). Hattammaru et al. (2019) observed that PER3, Nr1D1 and Nr1D2 expression in facial hair follicle cells of night workers remained adjusted to a day-oriented schedule after one shift and that their phases were scattered after ≥3 consecutive nights, leading to dampened group rhythms. Koshy et al. (2019) studied 11 police officers before and after 7 days of night shifts. At baseline, central clock rhythms (urinary 6-sulfatoxymelatonin and salivary cortisol) and peripheral clock rhythms (clock genes expression in oral mucosa cells and PBMCs) were aligned to a day-oriented schedule. After seven night shifts, central group rhythms were partially adjusted and dampened, and individual rhythms were scattered (Koshy et al., 2019). In addition, rhythms of PER1-3 and REV-ERBα expression in oral mucosa cells were disrupted and the time-of-day variation in PBMCs clock genes PER1-3 was lost (Koshy et al., 2019). Recently, Resuehr et al. (2019) reported the rhythms of cortisol, melatonin, and clock gene expression in PBMCs of night nurses to be adjusted to a day-oriented schedule and more scattered, leading to a significantly dampened group rhythm. In comparison, these rhythms were clustered and well aligned to the sleep-wake cycle for day-shift nurses. Surprisingly, however, significant rhythms of the canonical circadian clock genes PER1 and PER3 were only detected in night nurses. More studies are needed on the disruption of the central and peripheral clocks of shift workers.

Altogether, these results further emphasize the all-encompassing impact of night-shift work, which not only affects rhythms controlled by the central clock but also those of peripheral clocks, and provides insight into molecular mechanisms affecting most of the entire genome.

Rate of Circadian Adaptation

Circadian adaptation to a night-oriented schedule is a gradual process requiring extended, consistent exposure to the altered work-rest cycle, and there is a high degree of variability in the capacity of night-shift workers to do so (Crowley et al., 2004; Boivin et al., 2012a, 2012b; Stone et al., 2018; Molzof et al., 2019). Without specific interventions to facilitate shifts in the central clock, it is estimated only ~25% of workers show circadian adaptation to night work (Folkard, 2008). Field-based studies indicate that most night workers are unlikely to demonstrate signs of substantial adaptation in melatonin or cortisol rhythms within three consecutive night shifts (Grundy et al., 2009; Dumont et al., 2012). However, these studies generally focused on comparing the daily patterns of hormone levels between work schedules rather than documenting changes in circadian phase (Hansen et al., 2006; Garde et al., 2009; Leung et al., 2016; Daugaard et al., 2017). While a couple of these found small reductions in melatonin levels during the night shift, none indicated phase shifts that would warrant a classification of even partial adaptation (Grundy et al., 2011; Leung et al., 2016; Daugaard et al., 2017; Stone et al., 2018). In another study, Molzof et al. (2019) monitored the core body temperature rhythm of nurses working three consecutive night or day shifts and found the temperature minimum was improperly aligned with daytime sleep. Work cycles comprising sequences of more than four or five consecutive night shifts are more likely than shorter sequences to show signs of circadian adaptation. However, even in these cases, changes in the profile of these rhythms are still highly variable between individuals and usually not large enough to represent complete adaptation of the central clock (Hansen et al., 2010; Harris et al., 2010; Ferguson et al., 2012).

Of the different shift-working populations that have been studied, the largest rates of adaptation to night work have consistently been reported in offshore oil-rig workers, whose isolated working environments and operating schedules are conducive to facilitating and maintaining changes to circadian alignment (Barnes et al., 1998; Gibbs et al., 2007; Hansen et al., 2010). Barnes et al. (1998) found that the acrophase of urinary sulfatoxymelatonin rhythm shifted on average ~1.3 to 1.8 h per day, with 96% of workers having their final acrophase within the second half of their daytime sleep period. Gibbs et al. (2007) and Hansen et al. (2010) found similar rates of adaptation for oil-rig workers compared with other occupations, over a week of consecutive shifts. In contrast, shift workers on similar sequences of night shifts who do not experience large phase shifts (e.g., police officers, nurses, and doctors) often have to meet work and domestic responsibilities that can interfere with their circadian adaptation (Boudreau et al., 2013a; Stone et al., 2018).

Some of the discrepancies in circadian adaptation observed in naturalistic field studies may be a consequence of the different environmental and behavioral confounders of various biomarkers of the central clock (Harris et al., 2010; Ferguson et al., 2012; Jensen et al., 2016). Indeed, the rhythms of melatonin, cortisol, and body temperature can be affected by light exposure, work-related stressors, and activity levels, respectively (Gander et al., 1986; Boivin and James, 2002b; Duffy and Dijk, 2002; van Eekelen et al., 2003). Furthermore, different metrics for assessing circadian rhythms in naturalistic environments may also influence whether adaptation can be determined (Barnes et al., 1998; Kudielka et al., 2007; Grundy et al., 2011; Jensen et al., 2020).

Interventions involving judicial exposure to light and darkness have been used to varying success for facilitating adaptation to night work. In a recent meta-analysis, Lam and Chung (2021) examined the phase-shifting effects of light therapy from the pooled-results of 13 studies comprising both real and simulated shift workers. Consistent with experimental protocols (Boivin et al., 1996; Zeitzer et al., 2000), brighter light at night was found to result in larger phase shifts and greater suppression of melatonin in a dose-responsive manner (Lam and Chung, 2021). However, the phase-shifting effects of longer or shorter treatments were more ambiguous (Lam and Chung, 2021). Light attenuation with goggles or sunglasses during the morning commute after night shifts has been shown to facilitate phase shifting (Boivin et al., 2012a), and studies focusing on the use or avoidance of blue-enriched light have demonstrated greater suppression of melatonin with more exposure (Rahman et al., 2013; Motamedzadeh et al., 2017).

Effect of Chronotype in Night-Shift Adaptation

Chronotype is a behavioral trait that describes an individual’s habitual sleep-timing preferences in relation to the 24-h light-dark cycle (Juda et al., 2013b) and is associated with the ability to adapt to specific shifts. For instance, early chronotypes generally have earlier bedtimes and wake-times than later chronotypes and typically function best in the morning than during the afternoon or evening (van de Ven et al., 2016). In contrast, late chronotypes typically have later and more flexible bedtimes, are more resilient to the consequences of night work (higher shift work tolerance), and obtain less sleep when engaging in morning shifts (Juda et al., 2013a; van de Ven et al., 2016; Kervezee et al., 2021). Moreover, early chronotypes tend to sleep for shorter durations when engaging in night work (Juda et al., 2013a), although this effect disappears when the effect of napping is considered (Kervezee et al., 2021). Increased morningness and eveningness were correlated with longer sleep duration during series of consecutive morning and evening shifts, respectively (Kervezee et al., 2021). Interestingly, Vetter et al. (2015) implemented a shift system wherein work hours were adjusted to accommodate workers’ chronotypes: morning shifts were abolished for late chronotypes, and night shifts were abolished for early chronotypes. It was found that aligning work hours and chronotype was associated with longer sleep duration across the work schedule (Vetter et al., 2015).

From an occupational health perspective, the impact of the chronotype on sleep duration and timing may mediate some of the adverse health effects associated with shift work (Kecklund and Axelsson, 2016; Kervezee et al., 2020). In a study of a large group of female hospital employees, it was shown that sleep duration is an important mediator of the relationship between shift work and metabolic syndrome (Korsiak et al., 2018). It remains to be determined whether chronotype affects this relationship. Using a cross-sectional design, Yu et al. (2015) reported that being an evening chronotype was associated with increased risk of metabolic syndromes in middle-aged adults. Similar results were reported in a case-control study, where metabolic syndrome cases were more often evening chronotypes (Assmann et al., 2020). The increased risk of metabolic syndromes in evening chronotypes would be related to modifiable lifestyle behavior rather than genetic factors (Vera et al., 2018). However, as detailed in the following section, the effect of chronotype on metabolic risks appears to differ in individuals’ working shifts.

Sleep Disturbances

Reduced sleep quality and duration, and symptoms of insomnia are frequent in shift workers (Kecklund and Axelsson, 2016; Wyse et al., 2017; Yong et al., 2017; Moreno et al., 2019), especially those working nights, early morning, and rotating shifts (Akerstedt and Wright, 2009). Typically, the daytime sleep periods of night-shift workers end prematurely after 4 to 6 h (Akerstedt and Wright, 2009; Kecklund and Axelsson, 2016) and workers are often unable to resume sleep afterward. Early day shifts can be associated with similar levels of sleep restriction (Ganesan et al., 2019), as workers have difficulty falling asleep at earlier bedtimes coinciding with the evening wake maintenance zone, and sleep duration is curtailed to comply with the early work start. A large UK population-based study of more than 277,000 workers found that shift workers report less sleep per 24-h day and poorer sleep quality than non-shift workers (Wyse et al., 2017). A 5-year longitudinal study of 2615 hospital night-shift workers showed that the odds of reporting insomnia (Insomnia Severity Index ≥ 15) were increased by ~2-fold when working consecutive nights (vs. only one) (Lee et al., 2021). A meta-analysis summarizing 11 cross-sectional studies of police officers revealed that about 50% of workers report poor sleep quality (Pittsburgh sleep quality index > 5; Garbarino et al., 2019). In a large meta-analysis, Pilcher et al. (2000) reported an effect of the shift system on sleep duration. Compared with permanent day workers (7.0 h), permanent evening or night workers reported sleeping more (7.6 h, effect size = 0.42) or less (6.6 h, effect size = 0.35), respectively. Rotating shift workers slept more after evening shifts and less after night shifts, and this effect was more pronounced for rapid rotating schedules (evening shifts: 8.1 h, night shifts: 5.7 h, effect size ≥ 0.93). As a result of sleep loss during days of work on atypical shifts, rest days are often used for recovery and are associated with longer sleep periods (Garde et al., 2009; Paech et al., 2010; Garde et al., 2020; Kervezee et al., 2021). Recovery from one night of total sleep deprivation usually takes about one to two nights of recovery sleep (Balkin et al., 2008), whereas chronic sleep restriction takes longer. Axelsson et al. (2008) demonstrated that 7 days of recovery sleep was not enough to completely restore performances to baseline levels following 5 days of 4-h sleep per night. In a study of Norwegian nurses, Eldevik et al. (2013) found that the duration of breaks between successive work shifts was important, with quick returns (< 11 h off between shifts) associated with insomnia, excessive sleepiness, and shift work disorder. As detailed later, the negative cognitive, metabolic, and health outcomes of sleep curtailment are numerous, and probably play a role in the adverse effects associated with shift work.

Sleepiness

Sleepiness, impaired cognition and performance are common in shift workers, and have been reported in numerous studies, including in nurses (Behrens et al., 2019; Ganesan et al., 2019; Wilson et al., 2019), medical residents (Basner et al., 2017), police officers (Boivin et al., 2012a; Boudreau et al., 2013a), miners (Ferguson et al., 2010, 2011), marine pilots (Boudreau et al., 2018), professional truck drivers (Anund et al., 2018), train drivers (Jay et al., 2006), and airline pilots (Ingre et al., 2014; Sallinen et al., 2017; Aljurf et al., 2018; Sallinen et al., 2020). Extended wakefulness, lack of adequate recovery sleep between shifts, and being awake during the circadian trough of alertness, at night or in the early morning, lead to excessive sleepiness in shift workers (Mullins et al., 2014). Adjustment to consecutive night shifts is only modest under normal conditions but can improve significantly if circadian adaptation occurs (Bjorvatn et al., 2006; Boudreau et al., 2013a).

In the transportation industry, sleepiness raises important safety concerns (Akerstedt, 2019). On-the-road studies using electroencephalogram recordings in professional drivers (Kecklund and Akerstedt, 1993; Mitler et al., 1997) and inappropriate line crossing in non-professional drivers (Sagaspe et al., 2008; Hallvig et al., 2014) revealed increased incidence of severe sleepiness at night. In a group of 54 professional truck drivers, the odds of severe sleepiness (Karolinska sleepiness scale ≥ 7) were nine times higher during the first night shift compared with day and evening shifts (Pylkkonen et al., 2015). Commuting home after a night shift was also associated with higher risk of excessive sleepiness and accidents (Lee et al., 2016; Anderson et al., 2018; Liang et al., 2019). Cross-sectional studies of airline pilots revealed that about one airline pilot out of two reported to have unintentionally felt asleep while flying (Marqueze et al., 2017; Aljurf et al., 2018), which was confirmed by electroencephalogram recordings during real flights (Wright and McGown, 2001).

Physical Health

As previously described, the circadian misalignment associated with working at night has been implicated in the increased risk of cardiometabolic disorders (Brum et al., 2015; Kecklund and Axelsson, 2016), including metabolic syndrome (Khosravipour et al., 2021; Wang et al., 2021), type 2 diabetes (Vetter et al., 2018; Gao et al., 2020), and cardiovascular heart diseases (Vetter et al., 2016; Kervezee et al., 2020). Other studies have reported different forms of cancer (Schernhammer et al., 2006; Mancio et al., 2018; Ward et al., 2019), various gastrointestinal and digestive complaints (Knutsson and Bøggild, 2010; Gupta et al., 2019), menstrual irregularities, dysmenorrhea, and difficulties with pregnancy (Labyak et al., 2002; Zhu et al., 2004; Hammer et al., 2018). These are likely to be at least partially due to the circadian disruption of internal physiological processes.

Many studies identify shift work as having an adverse effect on various risk factors for metabolic and cardiovascular diseases, including elevated glucose, insulin and triacylglyceride levels, and higher white blood cell counts (Sookoian et al., 2007; van Drongelen et al., 2011; Manodpitipong et al., 2017; Wirth et al., 2017). Longitudinal studies also provide evidence for an effect of shift work on impaired glucose tolerance, being overweight, and gaining weight (Proper et al., 2016). In a cross-sectional population-based study, Sookoian et al. (2007) found that rotating shift workers had a significantly higher odds ratio (OR) for metabolic syndrome than day workers, even after controlling for age and physical activity. In another study, Manodpitipong et al. (2017) reported that night work was associated with poorer glycemic control than day work after controlling for factors such as body mass index and sleep duration. It has been hypothesized that metabolic conditions common in shift work may partly be exacerbated by disturbances of healthy gut microbiota caused by sleep loss and circadian misalignment (Reynolds et al., 2017). Working at night may also lead to altered autonomous nervous system modulation of the heart when sleep occurs at adverse circadian phases (Scheer et al., 2009; Boudreau et al., 2013b; Morris et al., 2016). Combined, these findings give credence to an effect of circadian misalignment as a risk factor of cardiometabolic disturbances, independent of behavioral changes associated with night-shift work.

A number of meta-analyses and systematic reviews have attempted to combine the various epidemiological studies that address the relationship between shift work and metabolic and cardiovascular health, providing evidence for an association between shift work and metabolic syndrome (Wang et al., 2021), diabetes mellitus (Gan et al., 2015; Gao et al., 2020), obesity (Sun et al., 2018), hypertension (Manohar et al., 2017), and cardiovascular disease (Torquati et al., 2018). Sex appears to moderate some of these relationships, as female shift workers were shown to have a higher risk of developing metabolic syndrome and diabetes mellitus compared with male shift workers (Gao et al., 2020; Wang et al., 2021), but a lower risk of hypertension (Manohar et al., 2017). However, Gan et al. (2015) reported an increased risk of diabetes mellitus in female compared with male shift workers. A recent meta-analysis of 21 studies with follow-up periods ranging from 4 to 24 years concluded that the pooled relative risk of diabetes mellitus was 1.10 (95% confidence interval [CI] [1.05, 1.14]) in shift workers compared with non-shift workers (Gao et al., 2020). A dose-response analysis comprising three cohorts of female shift workers indicated that relative risk increased by 1.05 (95% CI [1.03, 1.07]) per 5 years of exposure to shift work (Gao et al., 2020). More follow-up studies are required to assess the cumulative exposure risk of shift work and the modulating effect of sex and gender.

The type of shift schedule has also been shown to affect the risk of different health conditions. For instance, several meta-analyses indicated that a rotating shift schedule was associated with an increased risk of diabetes and hypertension compared with other types of shift schedules, even fixed night shifts (Gan et al., 2015; Manohar et al., 2017), whereas another analysis indicated permanent night-shift workers had a greater risk of developing obesity than workers on rotating shift schedules (Sun et al., 2018). In a large meta-analysis comprising 21 longitudinal and case-controls studies, Torquati et al. (2018) recently demonstrated a heightened risk of coronary heart disease and ischemic heart disease associated with shift work. Several meta-analyses have revealed dose-response effects in the relationship between health and exposure to shift work, such that more years as a shift worker is linked to poorer cardiometabolic health (Sun et al., 2018; Torquati et al., 2018; Gao et al., 2020; Wang et al., 2021). Nonetheless, there is no consensus on the definition of exposure, and more research is required to differentiate the effect of years of work and intensity of shift schedules on health outcomes (Kecklund and Axelsson, 2016). Improving understanding of the different means through which shift work affects health outcomes will facilitate the development of strategies to mitigate the burden of shift work.

A working group of the International Agency for Research on Cancer (IARC, 2020) concluded that night-shift work is probably carcinogenic to humans, based on extensive analysis of human research, experimental animal studies, and mechanistic evidence. There is evidence that night work may have a role in causing or exacerbating several specific cancers, including those of the breast (Manouchehri et al., 2021), prostate (Gan et al., 2018; Mancio et al., 2018), colon, and rectum (Wang et al., 2015). There is weak support from other meta-analyses of an increased relative risk of prostate cancer for rotating shift schedules compared with fixed daytime workers, but no increased risk for fixed night-shift work (Du et al., 2017; Gan et al., 2018; Mancio et al., 2018). A couple of recent meta-analyses found that shift work had little to no effect on breast cancer or other types of cancer risk, including prostate, pancreatic, and colorectal (Travis et al., 2016; Dun et al., 2020). These contradictory results might be explained by factors such as the lifetime duration of shift work exposure. Indeed, Wegrzyn et al. (2017) found that nurses with long-term rotating night work (≥20 years) experience had a higher risk of breast cancer, especially those who were younger when they began shift work. Support for a dose-responsive effect of night work on colorectal cancer has been demonstrated with a meta-analysis, describing an estimated 11% increase in risk for every 5 years of exposure (Wang et al., 2015). More recent epidemiological research by Papantoniou et al. (2018) also found that nurses with more than 15 years of shift work exposure had a higher risk of rectal cancer. In shift workers, it was hypothesized that photic suppression of melatonin at night may be a plausible mechanism for the increased risk of breast cancer (IARC, 2020). Some studies support a carcinogenic effect of light exposure at night (Yang et al., 2014), whereas others do not (Dun et al., 2020). Some studies in humans have also found an association between the risk of breast cancer and either the homozygous or heterozygous 5-repeat allele of PER3 (Zhu et al., 2005) or the Ala394Thr polymorphism of the NPAS2 gene (Zhu et al., 2008). However, other analyses which found marginal associations between polymorphisms of ARNTL and CRY1 and breast cancer revealed insignificant results after statistically adjusting for multiple comparisons (Grundy et al., 2013). While the data from these separate studies do not uniformly report effects of specific clock genes on cancer, they raise the possibility that desynchronized rhythms of clock gene expression in peripheral tissues provide a plausible mechanism for increased risk of cancer development.