Changes in artificial light exposure

Obesity is an increasing health problem and is associated with the development of type 2 diabetes and CVD⁽ Reference Schmidt, Johannesdottir and Lemeshow ^{1 )}. The pathophysiology of obesity is multifactorial, with the major contributions from overconsumption of high-energy highly palatable food and an inactive lifestyle⁽ Reference Lyznicki, Young and Riggs ^{2 )}. One modern environmental factor that contributes to changes in eating behaviour is the widespread use of artificial light. The relative independence from the natural light–dark (LD) cycle, allows people to eat and engage in activities until late in the evening and at night. Artificial light has also led to an increase in nighttime sky glow and to the transformation of nightscapes. More than 99% of the US and EU population, and about two-thirds of the world population lives in areas where the night sky is illuminated above the threshold for light pollution (artificial sky brightness greater than 10% of the natural night sky brightness above 45° elevation). Moreover, satellite data show that 70% of the US population and 50% of the European population can no longer see the Milky Way, even under the best conditions⁽ Reference Cinzano, Falchi and Elvidge ^{3 )}. Cinzano et al.⁽ Reference Cinzano, Falchi and Elvidge ^{3 )} calculated that only 40% of Americans live in a location where it becomes sufficiently dark at night for the human eye to make a complete transition from cone to rod vision. Despite the benefits for socio-economic development, changes in LD environment may have adverse effects on human subjects and wildlife⁽ Reference Kyba, Tong and Bennie ^{4 )}. In animals, light pollution leads to behavioural and physiological adaptations, such as alterations in orientation, survivorship, reproductive success and visual communication⁽ Reference Lewanzik and Voigt ^{5 ,} Reference Navara and Nelson ^{6 )}.

Interestingly, in human subjects, the increase in artificial light exposure parallels the increase in obesity prevalence with substantial evidence for additional adverse metabolic effects of increased exposure to artificial light⁽ Reference Fonken and Nelson ^{7 )}. Availability of artificial light enables people to eat at unusual feeding times, and since metabolic responses to a meal are time-of-day-dependent, this might negatively affect metabolism⁽ Reference Garaulet and Gomez-Abellan ^{8 )}. Furthermore, light exposure at inappropriate times itself may have adverse consequences for energy metabolism via changes in the biological clock and enhance the negative effects of eating at the wrong time of day⁽ Reference Cagampang and Bruce ^{9 )}. In addition to greater exposure to artificial light, daytime natural light exposure is often decreased since people tend to stay inside with lower light intensities.

Light synchronises the central circadian clock

For most organisms, a day is characterised by two distinct behavioural phases: one phase with activity and feeding behaviour and one phase with resting/sleeping and fasting behaviour. During the active period, ingested nutrients provide fuel for energy production and excess energy is stored. During the resting period, energy stores are mobilised to sustain metabolic homeostasis. The hypothalamus controls a vast array of the behavioural and physiological processes that alternate between the behavioural phases, including feeding, but also sleep and arousal, thermoregulation and energy expenditure⁽ Reference Saper, Scammell and Lu ^{10 ,} Reference Dietrich and Horvath ^{11 )}. These activity/feeding and resting/fasting periods are defined by a molecular mechanism in the central clock that is located in the suprachiasmatic nuclei (SCN) of the hypothalamus⁽ Reference Green and Gillette ^{12 )}. This central clock generates a biological rhythm of approximately 24 h (hence ‘circadian’ from ‘circa diem’, approximately 1 d) and lesions of the SCN result in loss of all circadian rhythms, including those in locomotor activity, food intake and drinking activity⁽ Reference Strubbe, Prins and Bruggink ^{13 ,} Reference Stephan and Zucker ^{14 )}.

The SCN comprises about 20 000 pacemaker neurons⁽ Reference Welsh, Logothetis and Meister ^{15 )}. The single-cell circadian oscillators are regulated by a molecular feedback mechanism that maintains a 24 h rhythm. The transcription factors CLOCK and ARNTL/BMAL1 represent the positive limb of this molecular clock and induce the transcription of the factors CRY and PER, representing the negative limb of the clock by inhibiting their own transcription⁽ Reference Takahashi, Hong and Ko ^{16 )}. Since the endogenous period of the SCN oscillation is not exactly 24 h, it must be synchronised to the external environment. Retinal light is the dominant environmental Zeitgeber for the phase entrainment of circadian oscillators⁽ Reference Moore and Lenn ^{17 )}. In addition to rods and cones, the retina consists of intrinsically photosensitive retinal ganglion cells that contain the photopigment melanopsin⁽ Reference Hattar, Liao and Takao ^{18 )}. These intrinsically photosensitive retinal ganglion cells directly innervate the SCN via the retinohypothalamic tract⁽ Reference Hattar, Liao and Takao ^{18 –} Reference Berson, Dunn and Takao ^{22 )}. The geniculohypothalamic tract, originating in the intergeniculate leaflet, provides a second route for photic information of the SCN clock⁽ Reference Harrington and Rusak ^{23 )}. Intrinsically photosensitive retinal ganglion cells are sensitive to a range of wavelengths, with a maximum sensitivity in the short-wavelength (blue) domain of visible light⁽ Reference Lucas, Douglas and Foster ^{24 )}. Animal studies have shown that one single light pulse shifted clock gene rhythms in the SCN and induced a behavioural phase shift⁽ Reference Miyake, Sumi and Yan ^{25 )}. In human subjects, one single pulse of bright light induced a phase advance or a phase delay in the plasma profile of the dark hormone melatonin, depending upon the circadian phase at which the light exposure occurred⁽ Reference Brainard, Lewy and Menaker ^{26 ,} Reference Khalsa, Jewett and Cajochen ^{27 )}. Exposure to early morning room light results in a phase advance of the endogenous core body temperature cycle, while late evening light before bedtime has a phase-delaying effect on the circadian pacemaker⁽ Reference Boivin, Duffy and Kronauer ^{28 )}. The relationship between light intensity and the circadian rhythm response follows a nonlinear function, with even low-intensity light (100 lux) being able to phase shift the circadian clock⁽ Reference Zeitzer, Dijk and Kronauer ^{29 )}. Zeitzer et al.⁽ Reference Zeitzer, Dijk and Kronauer ^{29 )} showed that exposure to a single episode of 100 lux of evening bright light generates half of the maximal phase-delaying response observed after a light stimulus of 9000 lux.

Suprachiasmatic nuclei regulates food intake and glucose metabolism

Feeding behaviour has a clear day/night rhythm, which is influenced by the LD cycle⁽ Reference Plata-Salaman and Oomura ^{30 ,} Reference Zucker ^{31 )} and disrupted in SCN-lesioned animals⁽ Reference Nagai, Nishio and Nakagawa ^{32 ,} Reference Nishio, Shiosaka and Nakagawa ^{33 )}. Different hypothalamic projection areas of the SCN are involved in regulating feeding behaviour, including the paraventricular nucleus of the hypothalamus (PVN), the lateral hypothalamus and the arcuate nucleus⁽ Reference Dibner, Schibler and Albrecht ^{34 )}. Within the arcuate nucleus, neuropeptide Y and α-melanocyte-stimulating hormone neurons are known to be involved in feeding behaviour⁽ Reference Guzman-Ruiz, Saderi and Cazarez-Marquez ^{35 )}. In the lateral hypothalamus, expression of the orexigenic neuropeptide orexin (also known as hypocretin) demonstrates a daily rhythm⁽ Reference Abrahamson, Leak and Moore ^{36 ,} Reference Girault, Yi and Fliers ^{37 )}. In addition, indirect projections from the SCN to cortico-limbic areas exist⁽ Reference Luo and Aston-Jones ^{38 )}. Since the cortico-limbic area is important for signalling reward, the rhythmicity of the dopamine system within the cortico-limbic system points to a role for the biological clock in food reward⁽ Reference Mendoza, Clesse and Pevet ^{39 )}.

In addition to daily rhythms in feeding behaviour, daily rhythms in glucose metabolism have also been described in both human subjects and rodents. Blood glucose concentrations and glucose tolerance fluctuate over the day/night cycle with a peak in circulating glucose shortly before awakening, just before the active period⁽ Reference Jarrett, Baker and Keen ^{40 ,} Reference La Fleur, Kalsbeek and Wortel ^{41 )}. In rodents, this rhythm is independent of food intake⁽ Reference La Fleur, Kalsbeek and Wortel ^{42 ,} Reference Kalsbeek and Strubbe ^{43 )}, depends on an intact SCN⁽ Reference La Fleur, Kalsbeek and Wortel ^{42 ,} Reference Yamamoto, Nagai and Nakagawa ^{44 )}, and has a 12 h difference between nocturnal and diurnal species. In addition, in healthy human subjects, glucose tolerance possesses a diurnal variation, with lower glucose tolerance in the afternoon compared with the morning⁽ Reference Jarrett, Baker and Keen ^{40 ,} Reference Lee, Ader and Bray ^{45 ,} Reference Carroll and Nestel ^{46 )}. This effect has been explained by the diurnal variation in insulin sensitivity and insulin secretion⁽ Reference Lee, Ader and Bray ^{45 ,} Reference Gibson and Jarrett ^{47 ,} Reference Morgan, Aspostolakou and Wright ^{48 )} with insulin sensitivity of peripheral tissues and insulin secretion both reduced in the evening⁽ Reference Jarrett, Baker and Keen ^{40 )}.

To generate these daily rhythms in glucose metabolism, the SCN influences both the autonomic nervous system (ANS) and secretion of glucoregulatory hormones. Anatomical tracing experiments revealed that there are neuronal connections between the SCN and the liver, and the SCN and the pancreas⁽ Reference La Fleur, Kalsbeek and Wortel ^{49 ,} Reference Buijs, Chun and Niijima ^{50 )}. These connections could be involved in the rhythms of glucose metabolism by affecting, for example, hepatic glucose production and (meal-induced) insulin secretion. The involvement of liver innervation in SCN-mediated rhythms in plasma glucose concentrations was demonstrated by hepatic sympathetic denervation studies, showing that the SCN needs an intact sympathetic input to the liver to generate a daily rhythm in plasma glucose concentrations⁽ Reference Kalsbeek, La Fleur and van Heijningen ^{51 )}. The SCN does not directly innervate autonomic motor neurons in the brainstem or spinal cord, but transmits its signal to other areas within the hypothalamus. One such example is the PVN, which receives signals from the SCN and has extensive projections to sympathetic and parasympathetic motor neurons in the spinal cord and brainstem, respectively⁽ Reference Teclemariam-Mesbah, Kalsbeek and Pevet ^{52 )}. The functional importance of this SCN–PVN connection in controlling plasma glucose concentrations was revealed by administering different SCN transmitter agonists and antagonists into the vicinity of the PVN⁽ Reference Kalsbeek, La Fleur and van Heijningen ^{51 )}. Another hypothalamic area receiving input from the SCN is the lateral hypothalamus, particularly the orexin neurons. Orexin affects both glucose production and insulin sensitivity⁽ Reference Yi, Serlie and Ackermans ^{53 ,} Reference Tsuneki, Tokai and Nakamura ^{54 )} and with its circadian rhythmicity could be an important mechanism for the SCN to influence glucose metabolism.

In addition to the involvement of the ANS, glucose metabolism can also be influenced by the release of hormones such as insulin, glucagon and corticosterone. The magnitude of the endocrine response to a glucose or exercise challenge varies over the activity/inactivity cycle. For example, a marked effect of time of day on neuroendocrine responses to prolonged moderate exercise was found in healthy volunteers⁽ Reference Galassetti, Mann and Tate ^{55 )} and an oral glucose load in the early morning hours produces a higher insulin response compared with the evening or afternoon⁽ Reference Lee, Ader and Bray ^{45 ,} Reference Carroll and Nestel ^{46 )}. Similarly, in rats with meals equally distributed over the LD cycle, the insulin responses varied based on the time of the day the meal was consumed, despite equal meal sizes⁽ Reference Kalsbeek and Strubbe ^{56 )}. As locomotor activity is not affected by equally distributing meals throughout the day and maintains its rhythmicity, it can be concluded that it is not a change in activity that affects insulin sensitivity and insulin responses⁽ Reference Kalsbeek and Strubbe ^{56 )}. In addition, SCN-lesion studies showed this variation in endocrine responses to be dependent on a functional SCN⁽ Reference La Fleur, Kalsbeek and Wortel ^{41 )}.

Although it is clear that the SCN plays a key role in the regulation of glucose metabolism, circadian oscillators are not only localised in the SCN, but also in other brain regions and peripheral tissues involved in energy metabolism, including the pancreas⁽ Reference Sadacca, Lamia and deLemos ^{57 )}, gut⁽ Reference Bron and Furness ^{58 –} Reference Pan and Hussain ^{60 )}, liver⁽ Reference Balsalobre, Brown and Marcacci ^{61 –} Reference Oishi, Fukui and Ishida ^{63 )}, skeletal muscle⁽ Reference McCarthy, Andrews and McDearmon ^{64 )} and adipose tissue⁽ Reference Loboda, Kraft and Fine ^{65 –} Reference Davidson, Poole and Yamazaki ^{68 )}. Peripheral clocks do not receive light input directly, but are synchronised by the SCN. Although the precise mechanism remains to be elucidated, there are several pathways through which light exposure (via the SCN) could entrain peripheral organs and indirectly affect energy metabolism. Light signals transmitted to the SCN might be forwarded through the ANS⁽ Reference La Fleur, Kalsbeek and Wortel ^{49 ,} Reference Buijs, Chun and Niijima ^{50 ,} Reference Cailotto, La Fleur and van Heijningen ^{69 ,} Reference Cailotto, van Heijningen and van der Vliet ^{70 )}, circulating hormones or metabolic signals to entrain the peripheral clocks⁽ Reference Balsalobre, Brown and Marcacci ^{61 ,} Reference Buijs, Wortel and Van Heerikhuize ^{71 )}.

Effect of light on food intake, body weight and glucose metabolism in animals

Many studies have investigated the effect of chronically altered LD schedules on food intake, body weight and glucose metabolism in nocturnal rodents. In mice, continuous light exposure has been shown to cause obesity and impaired glucose tolerance⁽ Reference Fonken, Workman and Walton ^{72 ,} Reference Coomans, van den Berg and Houben ^{73 )}. In one study, increased body weight gain under constant light conditions was partly due to increased food intake, but also due to a reduction in energy expenditure⁽ Reference Coomans, van den Berg and Houben ^{73 )}. Another study also showed increased body weight with constant light, but without differences in total food intake or daily locomotor activity, and energy expenditure was not measured in this study⁽ Reference Fonken, Workman and Walton ^{72 )}. Interestingly, a recent study in mice found that continuous light exposure did not affect total body weight, but instead increased adiposity associated with reduced brown adipose tissue activity⁽ Reference Kooijman, van den Berg and Ramkisoensing ^{74 )}. In contrast to mice, the effect of continuous bright light on body weight in rats is moderate⁽ Reference Natelson, Servatius and Tapp ^{75 ,} Reference Wideman and Murphy ^{76 )} or absent⁽ Reference Dauchy, Dauchy and Tirrell ^{77 ,} Reference Gale, Cox and Qian ^{78 )}. However, in rats, continuous bright light exposure may reduce glucose-mediated pancreatic insulin secretion⁽ Reference Qian, Block and Colwell ^{79 )} and in diabetes-prone transgenic human islet amyloid polypeptide rats, constant bright light causes accelerated loss of β cell function and development of diabetes⁽ Reference Gale, Cox and Qian ^{78 )}. Taken together, these studies suggest that disturbing the endogenous timing system by exposure to continuous bright light causes insulin resistance by inducing obesity/adiposity in mice, while in genetically susceptible rats bright light causes diabetes by reducing pancreatic insulin secretion.

Obviously, continuous bright light exposure is not frequently encountered outside the laboratory. In real life, many human subjects and animals are exposed to dim light at night when the natural sky is dark, either via intentional illumination or unintentional artificial light pollution. Nelson's group reported that in Swiss Webster mice, exposure to 5 lux dim light at night caused obesity and diabetes despite similar or reduced total food intake compared with control animals⁽ Reference Fonken, Workman and Walton ^{72 ,} Reference Fonken, Lieberman and Weil ^{80 –} Reference Borniger, Maurya and Periasamy ^{82 )}. This was explained by increased daytime food intake⁽ Reference Fonken, Workman and Walton ^{72 )} and decreased whole body total energy expenditure⁽ Reference Borniger, Maurya and Periasamy ^{82 )}. The effect of dim light at night on body weight gain increased when mice were fed a high-fat diet⁽ Reference Fonken, Lieberman and Weil ^{80 )} and the metabolic disruptions were reversible when the mice returned to their normal LD cycle⁽ Reference Fonken, Weil and Nelson ^{83 )}. The metabolic effects of dim light at night were recently reviewed more extensively elsewhere⁽ Reference Fonken and Nelson ^{7 )}.

In addition to the effects of increased light exposure, repeated shifts of the LD cycle may also cause obesity⁽ Reference Voigt, Forsyth and Green ^{84 )} and diabetes⁽ Reference Oike, Sakurai and Ippoushi ^{85 )} in mice, without significant effect on total food intake or total locomotor activity. In rats, however, the effects of repeated LD shifts seem to be strain dependent; in Long Evans⁽ Reference Bartol-Munier, Gourmelen and Pevet ^{86 )} and Sprague Dawley⁽ Reference Gale, Cox and Qian ^{78 )} rats, repeated shifts do not affect body weight, whereas in F344⁽ Reference Tsai, Tsai and Hwang ^{87 )} and diabetes-prone human islet amyloid polypeptide rats⁽ Reference Gale, Cox and Qian ^{78 )}, repeated shifts do cause increased body weight gain. In sheep, representing a larger diurnal mammal, repeated LD shifts did not affect body weight or glucose tolerance⁽ Reference Varcoe, Gatford and Voultsios ^{88 )}. Currently, repeated LD shifts are often used as a rodent model for shift-work in human subjects, a condition known to affect body weight and energy metabolism. For a systematic review on rodent shiftwork models see⁽ Reference Opperhuizen, van Kerkhof and Proper ^{89 )}. Finally, although few studies have investigated the acute effects of light on metabolism, it is well established that rats respond to a light pulse during the dark period by directly decreasing food intake⁽ Reference Plata-Salaman and Oomura ^{30 )}. For a complete overview of the effect of light on food intake, body weight and glucose metabolism in animals see Table 1.

Table 1. Overview of studies on the effect of light on food intake, body weight and glucose metabolism in animals

LD, light/dark; HF, high fat; LF, low fat; HF–HS, high fat/high sugar; dLAN, dim light at night; HIP, human isles amyloid polypeptide; lx, lux.

Among the hormones affected by light are the glucoregulatory hormones corticosterone (i.e. cortisol in human subjects) and melatonin. SCN output modulates the secretion of corticosterone via a neuroendocrine pathway involving the release of adrenocorticotropic hormone from the pituitary (i.e. the hypothalamic–pituitary–adrenal axis) and via a neural pathway involving sympathetic innervation of the adrenal gland⁽ Reference Buijs, Wortel and Van Heerikhuize ^{71 )}. Plasma corticosterone levels have a strong diurnal rhythm, with a sharp peak near habitual wake time⁽ Reference Kalsbeek, van der Vliet and Buijs ^{90 )}. Light stimulates the secretion of corticosterone directly via sympathetic innervation⁽ Reference Ishida, Mutoh and Ueyama ^{91 ,} Reference Cailotto, Lei and van der Vliet ^{92 )}. Another hormone involved in energy metabolism is melatonin, which is secreted by the pineal gland and has a strong diurnal rhythm with a peak during the dark period⁽ Reference Reiter ^{93 )}. Nocturnal exposure to light suppresses plasma melatonin levels⁽ Reference Gooley, Chamberlain and Smith ^{94 )}. Daily treatment with melatonin reduces body weight increase in response to a high-fat diet, independent of total food consumption and improves plasma glucose levels, although data on energy expenditure were not reported⁽ Reference Terron, Delgado-Adamez and Pariente ^{95 –} Reference Prunet-Marcassus, Desbazeille and Bros ^{98 )}.

A direct effect of light on glucose metabolism is to be expected, given that: (1) light directly affects the activity of orexin neurons⁽ Reference Adidharma, Leach and Yan ^{99 )}; (2) pre-autonomic connections between the SCN and the PVN regulate hepatic glucose production and meal-induced insulin secretion through the ANS; (3) a light pulse acutely decreases efferent vagal activity to pancreas and liver in anaesthetised rats⁽ Reference Niijima, Nagai and Nagai ^{100 )}; (4) a light pulse acutely decreases the hepatic expression of phosphoenolpyruvate carboxykinase in rats⁽ Reference Cailotto, Lei and van der Vliet ^{92 )}; and (5) light directly affects glucocorticoid and melatonin secretion (as described earlier). However, the direct effects of light exposure on glucose metabolism have never been shown.

Although nocturnal rodents display a 12 h phase shift compared with human subjects, the function of the circadian timing system and mechanisms of the molecular clock are very similar. The daily rhythms of gene expression and electrophysiological activity as well as the substructure of the SCN are similar between nocturnal and diurnal species⁽ Reference Cohen, Kronfeld-Schor and Ramanathan ^{101 )}, but the downstream pathways involved in the functional output of the SCN are often reversed. For example, in nocturnal rodents, exposure to light at night reduces activity, but increases activity in diurnal species⁽ Reference Shuboni, Cramm and Yan ^{102 )}. At which level of the downstream pathways this 12 h switch is occurring is not clear yet, although for the corticosterone rhythm this may be at the level of the subPVN and dorsomedial hypothalamic nucleus⁽ Reference Kalsbeek, Foppen and Schalij ^{103 )}.

In conclusion, animal studies emphasise the intricate relationship between acute and chronic light exposures and daily rhythms of activity, food intake and glucose tolerance. Moreover, continuous bright light exposure (24 h) and dim light at night, as well as exposure to repeated LD shifts all affect body weight and energy metabolism.

In line with the results from animal studies, there are also data from studies in human subjects suggesting that light exposure affects food intake, body weight and glucose metabolism which will be discussed in the following section.

Effect of light on food intake, body weight and glucose metabolism in human subjects

A recent report demonstrated that evening bright light exposure increases appetite⁽ Reference AlBreiki, Middleton and Ebajemito ^{104 )}. Studying the SCN in human subjects is difficult, and thus melatonin activity is studied instead, as an indirect indicator of SCN activity. Notably, chronically reduced melatonin levels are associated with obesity and type 2 diabetes⁽ Reference Mantele, Otway and Middleton ^{105 )}. Little is known about the direct effects of melatonin treatment on food intake and body weight. In human subjects, however, one study found a negative association between melatonin supplements and BMI in obese women⁽ Reference Nachtigal, Patterson and Stratton ^{106 )}. In addition to possible effects on food intake, melatonin might play a role in the development of type 2 diabetes, since melatonin receptors are expressed on pancreatic β cells⁽ Reference Ramracheya, Muller and Squires ^{107 )} and polymorphisms in the melatonin receptor are associated with an increased risk of developing type 2 diabetes⁽ Reference Bouatia-Naji, Bonnefond and Cavalcanti-Proenca ^{108 )}. To our knowledge, until now no studies have yet investigated the direct effects of acute light exposure on human glucose metabolism.

Long-term light intervention studies in human subjects are difficult to perform and therefore most data on the relationship between light exposure, food intake and metabolism are derived from observational studies. In the home setting, bedroom light intensity had a positive correlation with the prevalence of obesity⁽ Reference McFadden, Jones and Schoemaker ^{109 ,} Reference Obayashi, Saeki and Iwamoto ^{110 )} and evening artificial light intensity showed a positive correlation with the incidence of type 2 diabetes⁽ Reference Obayashi, Saeki and Iwamoto ^{111 )}. Furthermore, daytime light exposure was positively correlated with BMI⁽ Reference Reid, Santostasi and Baron ^{112 )}.

Since the economic and industrial revolutions, more than 20 % of the working population performs shift work in order to optimise productivity and flexibility⁽ Reference Gordon, Cleary and Parker ^{113 )} and shift workers are at increased risk of developing obesity and type 2 diabetes⁽ Reference Karlsson, Knutsson and Lindahl ^{114 –} Reference Gan, Yang and Tong ^{117 )}. Although several observational studies found an association between shift work and metabolic disease, evidence for a causal relationship between light exposure at an inappropriate time of the day and metabolic disturbances is limited. Furthermore, in shift workers, several other factors involved in metabolism might be changed, such as diet composition, timing and frequency of food intake, exercise and sleep. For example, timing of meals rather than their total food intake was affected by shift works⁽ Reference Lennernas, Hambraeus and Akerstedt ^{118 )}, and night shift workers reported lower meal frequency, but increased prevalence to high-energy snacks⁽ Reference Takagi ^{119 ,} Reference de Assis, Kupek and Nahas ^{120 )}. Furthermore, shift workers showed problems maintaining physical fitness and reported increased general fatigue as the main reason⁽ Reference Atkinson, Fullick and Grindey ^{122 ,} Reference Atkinson and Reilly ^{123 )}. These data fit many studies showing reduced sleep and increased sleepiness in night shift workers⁽ Reference Sallinen and Kecklund ^{124 ,} Reference Scheen ^{125 )}. Nevertheless, data on light intensity were not reported in these studies. Since light is the dominant synchroniser for the central clock, the use of artificial light at an inappropriate time of the day could lead to chronodisruption: desynchronisation of the internal circadian rhythms and the 24 h environmental cycles. Chronodisruption is associated with metabolic disturbances and even permanent night workers showed only partial adaptation in their 24 h rhythm of plasma levels of glucose and insulin⁽ Reference Simon, Weibel and Brandenberger ^{126 )}. Detailed studies, however, on the effects of artificial light exposure at the home setting or the length of artificial light exposure of shift workers have not been performed.

As changes in duration and intensity of sunlight exposure are part of the defining features of the seasons, seasonal patterns in metabolism also suggest metabolic effects of light. The incidence of type 2 diabetes has a seasonal pattern with a peak in March and a trough in August⁽ Reference Doro, Benko and Matuz ^{127 )}. Moreover, healthy subjects possess a seasonal pattern in glycaemia with higher glucose levels in the winter⁽ Reference Suarez and Barrett-Connor ^{128 –} Reference MacDonald, Liston and Carlson ^{131 )} and patients with type 2 diabetes have a seasonal pattern of increased HbA1c levels and resulting insulin requirements in the winter⁽ Reference Tseng, Brimacombe and Xie ^{132 –} Reference Sohmiya, Kanazawa and Kato ^{134 )}. Secondary to direct effects of light exposure on glucose metabolism, these seasonal patterns may be partly explained by seasonal variations in temperature, levels of physical activity and food intake affecting body weight.

Taken together, these observational studies suggest that increased duration (but not intensity) of daytime light exposure is associated with metabolic health, whereas increased nighttime light exposure is associated with metabolic disease. Thus, these studies are consistent with rodent studies reporting adverse metabolic effects of light at night.

Interestingly, two case reports describe patients with seasonal affective disorder and insulin-dependent diabetes that showed a strong reduction in insulin requirements shortly after the initiation of light therapy⁽ Reference Nieuwenhuis, Spooren and Tilanus ^{135 ,} Reference Allen, Kerr and Smythe ^{136 )}. In addition, two small studies investigated the effects of long-term light treatment on body weight, although both had methodological challenges. A randomised controlled study in twenty-five obese subjects investigated the effect of adding 1 h of 5000 lux bright light therapy daily to a 6-week moderate exercise programme. Bright light therapy did not affect body weight, but induced a slight reduction in body fat mass as measured by bioelectrical impedance analysis⁽ Reference Dunai, Novak and Chung ^{137 )}. Another randomised controlled study in thirty-four obese female subjects investigated the effect of  3 weeks of 45 min of 1300 lux bright light therapy every morning on body weight and fat mass. Similarly, bright light therapy did not affect body weight, but induced a small reduction in fat mass. However, food intake was not recorded⁽ Reference Danilenko, Mustafina and Pechenkina ^{138 )}. For a complete overview of the effect of light on food intake, body weight and glucose metabolism in human subjects see Table 2.

Table 2. Overview of studies on the effect of light on food intake, body weight and glucose metabolism in human subjects

M, male; F, female; T2D, type 2 diabetes; SAD, seasonal affective disorder; VAS, visual analogue scale; LAN, light at night; lx, lux.

In addition to the long-term metabolic effects of light, it seems likely that light also has direct metabolic effects in human subjects, as light intensity directly affects ANS activity in human subjects⁽ Reference Scheer, Van Doornen and Buijs ^{139 –} Reference Cajochen, Munch and Kobialka ^{141 )}. Furthermore, light inhibits melatonin secretion through the ANS⁽ Reference Kalsbeek, Garidou and Palm ^{142 )} and light has been reported to affect glucocorticoid secretion, although some studies describe increased glucocorticoid levels due to bright light⁽ Reference Leproult, Colecchia and L'hermite-Baleriaux ^{143 ,} Reference Scheer and Buijs ^{144 )}, whereas another study describes decreased glucocorticoid levels⁽ Reference Jung, Khalsa and Scheer ^{145 ,} Reference Kostoglou-Athanassiou, Treacher and Wheeler ^{146 )}. These inconsistent findings might be related to the duration, intensity or timing of the light exposure.

In summary, human observational studies indicate that the duration of daytime light exposure is associated with blood glucose levels and insulin requirements, whereas exposure to light at night, as well as performing shift work, is associated with obesity and diabetes. Two small intervention studies suggest that bright light therapy may affect body composition.