Circadian rhythms regulate various behavioral, physiological, psychological, and endocrine functions in humans (Froy, 2007; Ribas-Latre and Eckel-Mahan, 2016; Allada and Bass, 2021; Kinouchi et al., 2021; Thosar and Shea, 2021; Zhang and Jain, 2021). One can imagine that circadian dysfunction would cause multiple negative impacts, both short term and long term, which lead to the increased susceptibility to many diseases, decreased quality of life, and even reduced life expectancy (Froy and Miskin, 2007; Jagannath et al., 2013; Roenneberg and Merrow, 2016; Valenzuela et al., 2016; Logan and McClung, 2019; Xu and Lu, 2019; Allada and Bass, 2021). Interestingly, the onsets and symptoms of many diseases, such as stroke, asthma, and depression, also display clear circadian characteristics (Jagannath et al., 2013; Hsieh et al., 2018; Cederroth et al., 2019; Dobrek, 2021; Ruan et al., 2021), which are called as the circadian pathology signs (Cederroth et al., 2019). Speaking of pharmacology, circadian rhythms affect the absorption, distribution, metabolism, and excretion (ADME) or called the pharmacokinetic processes as well as the efficacy and adverse effects of many drugs, which is well known as the chronopharmacology or chronotherapy (Dallmann et al., 2016; Ohdo et al., 2019; Dong et al., 2020; Dobrek, 2021; Nahmias and Androulakis, 2021). Given the importance of circadian rhythms, three researchers who discovered the basic of biological clock in studies of Drosophila were awarded the Nobel Prize in 2017 (Dobrek, 2021; Ruan et al., 2021).

Back in the 1990s, the discovery of several circadian clock genes, such as Clock, Bmal1, Per, and Cry (Takahashi, 2004), proved that almost all human cells express these genes and have the capacity to generate circadian oscillations (Du Pre et al., 2014; Takahashi, 2017), which thwarted the previous neuro-centric view that the master clock is located only in the brain (Takahashi, 2017). As is generally believed and well understood, at the systemic level, the human circadian system consists of the inputs, circadian oscillators, and outputs (Figure 1) (Takahashi, 2017; Cederroth et al., 2019; Huang et al., 2020; Ruan et al., 2021), while at the cellular level, it consists of several cell-autonomous molecular oscillators that composed of three transcriptional-translational feedback loops that are widespread throughout the body (Figure 2) (Du Pre et al., 2014; Takahashi, 2017; Logan and McClung, 2019; Huang et al., 2020; Sumova and Cecmanova, 2020; Ruan et al., 2021).

As early as 1975, a rat study (Deguchi, 1975) found, for the first time, that the mammalian fetal clock oscillators could be detected already at or before birth and be entrained by the mother. Subsequent studies have revealed that the fetus of rat, hamster, sheep, baboon, and other mammalians exhibited the circadian rhythms of metabolic activity (Reppert, 1992; Serón-Ferré et al., 1993; Mirmiran and Lunshof, 1996; Seron-Ferre et al., 2012) and the expressions of canonical clock genes (Seron-Ferre et al., 2007; Du Pre et al., 2014; Sumova and Cecmanova, 2020).

In human fetus, circadian rhythms in several physiological and endocrine functions, including heart rate (Lunshof et al., 1998), breathing patterns (Patrick and Challis, 1980), limb movements (Einspieler et al., 2021), sleep-wake states (Peirano et al., 2003; Bennet et al., 2018), and hormone levels (Serón-Ferré et al., 2001a) have been detected at different stages of pregnancy (Seron-Ferre et al., 2007; Du Pre et al., 2014; Wong et al., 2022). Impressively, Frigato et al. (2009), first observed the rhythmic expression of clock genes such as Per2 in the HTR-8/SVneo cells derived from human extravillous trophoblast. As part of a series of important discoveries, Perez et al. (2015) went on to find the rhythmic expression of various circadian genes, including Clock, Bmal1, Per2, and Cry1 in human full-term placenta.

It is incredible that no obvious circadian rhythms were found in the anencephalic fetus despite an intact maternal circadian rhythms were detected through the 24-h period fetal heart rate monitoring for anencephaly (Mirmiran and Lunshof, 1996), which demonstrated that the fetal brain, especially in the SCN, is required for the generation of fetal circadian rhythms (Mirmiran and Lunshof, 1996). It is still unclear when the fetal SCN clock first appeared morphologically, yet through the in vitro autoradiography by ¹²⁵I-labeled melatonin, the SCN is apparent as discrete nuclei in the human fetus and already has melatonin receptors at 18 weeks of gestational age (GA) (Reppert et al., 1988). Besides, it has been demonstrated that the VIP and AVP neurons were first observed at 31 weeks of GA in the ventrolateral part of the fetal SCN (Swaab et al., 1990; Swaab et al., 1994). Therefore, it is currently recognized that the circadian rhythms in humans are formed and developed during the perinatal period (Rivkees, 1997; Sumova and Cecmanova, 2020), while the components of the circadian system like the SCN are established and functional early in human fetus (Serón-Ferré et al., 1993).

Pregnancy presents an unusual circadian physiology pattern in which the fetal circadian system is completely embodied within that of the mother (Mark et al., 2017), and the two systems are connected by the placenta and interact with each other through this interface (Mark et al., 2017; Astiz and Oster, 2020; Bates and Herzog, 2020). Generally, placenta is responsible for the bidirectional transference of nutrients, hormones, metabolites, and gases (i.e., oxygen and carbon dioxide) between the mother and fetus (Seron-Ferre et al., 2012; Valenzuela et al., 2015; Astiz and Oster, 2020). Meanwhile, the placenta conveys the maternal circadian timing cues, such as physical activity, feeding behavior, temperature, heart rate, blood pressure, and hormonal levels, to the fetus (Serón-Ferré et al., 2001a; Seron-Ferre et al., 2012). In particular, multiple hormones produced by the mother, including melatonin, dopamine, glucocorticoids, estrogen, and progesterone, have profound effects on the development and entrainment of the fetal circadian rhythms (Mirmiran and Lunshof, 1996; Rivkees, 1997; Seron-Ferre et al., 2007; Mark et al., 2017). In addition, hormones such as human chorionic gonadotropin (hCG), secreted by the placenta, also exhibit obvious circadian characteristics (Waddell et al., 2012; Mark et al., 2017; Bates and Herzog, 2020). It will be very interesting to know how those hormones affect the formation of the fetal circadian rhythms.

After birth, neonates immediately begin to establish their own physical and physiological independence while losing the protect of the maternal-placental barrier (Joseph et al., 2015; Wong et al., 2022). From now on, the ontogenesis of the newborn begins to be greatly affected by the external environment (Brooks and Canal, 2013; Hazelhoff et al., 2021). Increasing evidence indicates that human postnatal circadian rhythms gradually mature along with the ontogenesis (Rivkees and Hao, 2000; Rivkees, 2007; Bueno and Menna-Barreto, 2016), in which the external environment, especially the light, plays an important role in the development and maturation (Mirmiran and Ariagno, 2000; Nishihara et al., 2002; Challet, 2007). Particularly, it should be pointed out that, in early infancy, the maternal entrainment factors and maternal-fetal interactions retained during pregnancy are more important than the external environment (Löhr and Siegmund, 1999; Rivkees, 2001; Nishihara et al., 2002; Sumova et al., 2012).

It is well established that the sleep is essential for normal brain development and health throughout the whole life (Peirano et al., 2003; Gogou et al., 2019). Premature newborns spend more than 70% of their first several weeks sleeping after birth (Ardura et al., 1995; Wong et al., 2022), thereby maintaining the proper sleep homeostasis is even more important for their neurological development and functional maturation (Bennet et al., 2018; Uchitel et al., 2021). The direct behavioral observations, parental sleep questionnaires, video recordings, polysomnography, actigraphy, and electroencephalography (EEG) (Table 1) have been developed to investigate the sleep-wake states of neonates (Mirmiran et al., 2003a; Collins et al., 2015; Gogou et al., 2019).

Based on the behavioral, cardiopulmonary, and EEG patterns (Darnall et al., 2006; Dereymaeker et al., 2017), the sleep states of preterm infants are generally classified as: active sleep (AS), the precursor of adult rapid eye movement (REM) sleep; quiet sleep (QS), the precursor of adult non-REM sleep; and indeterminate sleep (IS), the transition between AS and QS patterns (Mirmiran et al., 2003a; Liao et al., 2018). More specifically, the AS could promote the synapse formation, neuronal differentiation and migration, and the development of brain functional connectivity networks (Kurth et al., 2017; Gogou et al., 2019), whilst the QS promote the myelination, replenishment of energy reserves, and cognitive development in premature infants (Liao et al., 2018; Gogou et al., 2019).

As summarized in Table 1, Curzi-Dascalova et al. (1993), found the AS and QS states can be discerned in preterm infants as early as 27 weeks of GA. The results varied due to the different GA of the enrolled cases, but most studies revealed that preterm infants experienced more total sleep time and AS, while less QS than term ones (Anders and Keener, 1985; Ardura et al., 1995; Sahni et al., 1995; Hoppenbrouwers et al., 2005; Guyer et al., 2015; Georgoulas et al., 2021), which might reflect the accelerated neurological maturation of preterm infants (Mirmiran et al., 2003a; Bennet et al., 2018). Besides, preterm infants had fewer total arousals and, more specifically, fewer arousals in the AS (Guyon et al., 2022), which seemed to cause a higher risk of sudden infant death syndrome (Mirmiran et al., 2003a; Bennet et al., 2018).

With developmental maturity, preterm infants have more sleep during nighttime but less during daytime (Antonini et al., 2000; Korte et al., 2001; Guyer et al., 2015; Lan et al., 2019; Guyon et al., 2022). Meanwhile, as the PMA increased, the AS proportion comes out of a decreasing trend, but it is not true for the QS, IS, wakefulness, and activity, which all experience an increasing trend (Anders and Keener, 1985; Curzi-Dascalova et al., 1988; Curzi-Dascalova et al., 1993; Borghese et al., 1995; Sahni et al., 1995; Ingersoll and Thoman, 1999; Mirmiran et al., 2003b; Holditch-Davis et al., 2004; Hoppenbrouwers et al., 2005; Foreman et al., 2008; Dorn et al., 2014; Guyer et al., 2015; Lan et al., 2019; Cailleau et al., 2020; Park et al., 2020; Georgoulas et al., 2021; Guyon et al., 2022). In addition, other factors like sex, illness severity, body weight, ventilatory support, maternal smoking, and ambient temperature also affect the sleep-wake patterns (Bach et al., 2000; Hoppenbrouwers et al., 2005; Foreman et al., 2008; Lan et al., 2019).

It is well understood that the sleep homeostasis in humans are regulated by two independent but synergistic processes (Borbély, 1982; Deboer, 2018): a Clock-dependent circadian process (Process C), controlled by the SCN circadian oscillator, determines the alternation of different sleep propensity (Cremer et al., 2016); and a Sleep-dependent homeostatic process (Process S) that is determined by the prior sleep pressure, which comes from the adenosine buildup in the basal forebrain during wakefulness (Deboer, 2018; Wong et al., 2022). However, due to the immature development of the central nervous system, especially the SCN, Process C and Process S are not stably present in preterm infants or even in term ones (Salzarulo and Fagioli, 1992; Schwichtenberg et al., 2016). As a result, preterm infants experience many sleep and wake episodes within the 24-h period, and those ultradian sleep-wake rhythms persist for several months until the Process C and Process S are gradually developed (Mirmiran et al., 2003a; Cremer et al., 2016).

As shown in Table 1, preterm infants exhibit ultradian or irregular sleep-wake rhythms with different periods in the early postnatal life (Mirmiran et al., 1990; Hayes et al., 1993; Borghese et al., 1995; Shimada et al., 1999; Scher et al., 2005; Dorn et al., 2014; Koch et al., 2021), which might be explained by the environmental factors, such as feeding patterns (Glotzbach et al., 1995; Thomas, 2000; Bueno and Menna-Barreto, 2016) and respiratory states (Palmu et al., 2013). As for when the sleep-wake rhythms begin to occur and entrain, Scher et al. (2005), observed the ultradian sleep-wake rhythms as early as 25 weeks of PMA. Mirmiran and Kok, (1991) found the circadian sleep-wake rhythms began to appear after 29 weeks of PMA. However, McMillen et al. (1991), found that the entrainment of circadian sleep-wake rhythms did not occur in 50% of the preterm infants at 47 weeks of PMA, and all cases did not begin to develop the circadian rhythms until approximately 54 weeks of PMA.

Besides, several studies also demonstrated that a definite sleep-wake cycling existed in preterm infants with the advanced GA and became more prominent as the PMA increased (Sisman et al., 2005; Soubasi et al., 2009; Lee et al., 2010). Therefore, it could be concluded that with the continuous development of the brain and neural functions, circadian sleep-wake rhythms in preterm infants are consolidated and eventually developed to a 24-h pattern, just as those in adults (Mirmiran et al., 2003a; Bennet et al., 2018).

Many physiological biomarkers of the cardiopulmonary system in adults, such as the heart rate, blood pressure, and respiratory rate, exhibit distinct circadian rhythms (Elstad et al., 2018). A complex network that composed of the brainstem respiratory center, autonomic nervous system, and a variety of central and peripheral chemoreceptors and mechanoreceptors is responsible for regulating the rhythmic oscillations of the cardiorespiratory system (Darnall et al., 2006; Longin et al., 2006). Due to the immaturity of this network (Hunt, 2006), cardiorespiratory events like apnea, periodic breathing, and bradycardia are common in premature infants (Hodgman et al., 1990; Darnall et al., 2006), which leads to the erratic cardiopulmonary rhythms with marked individual differences (Begum et al., 2006). Clinically, the incidence and duration of cardiorespiratory events are associated with the GA and PMA (Hellmeyer et al., 2012; Fairchild et al., 2016; Patel et al., 2016).

As shown in Table 2, some, but not all, preterm infants experienced circadian or ultradian rhythms for the heart rate, pulse rate, respiratory rate, blood pressure, and oxygen consumption at the first few weeks after birth (Begum et al., 2006; Mirmiran and Kok, 1991; Bauer et al., 2009; D'Souza et al., 1992; Updike et al., 1985). Interestingly, unlike the ultradian sleep-wake rhythms gradually grew into circadian rhythms after birth, these cardiopulmonary rhythms in premature infants appeared and disappeared erratically (Tenreiro et al., 1991; Dimitriou et al., 1999), e.g., presence on day 2 but absence on day 7 after birth for the heart rate rhythms, which might be caused by the residual of maternal effects (Dimitriou et al., 1999). Tenreiro et al. (1991) also proposed that the circadian components of these cardiopulmonary rhythms gradually and erratically came into phases with one another, while the regular light-dark and feeding patterns seemed to promote the presence of the dominant circadian rhythms, which developed as the increased coupling between the component oscillators.

In addition, the well-developed laryngeal reflexes and coordination of pharyngoesophageal-cardiorespiratory (PECR) responses are essential for the development and maintenance of cardiorespiratory rhythms (Gewolb and Vice, 2006; Hasenstab-Kenney et al., 2020). As shown in Table 2, pharyngeal stimulations cause a decrease of heart rate in premature infants with uncoordinated suck-swallow-respiration rhythms due to the immature laryngeal reflexes and PECR responses, which would aggravate the disturbance of cardiac and respiratory rhythms (Hasenstab et al., 2019; Hasenstab-Kenney et al., 2020). (Gewolb et al., 2001; Gewolb and Vice, 2006) found that the development and establishment of suck-swallow rhythms were associated with their PMA. The swallow rhythms appeared at 32 weeks of PMA first (Gewolb et al., 2001), followed by the stabilization of suck and suck-swallow rhythms between 36 and 40 weeks of PMA (Gewolb and Vice, 2006), then the suck-swallow-respiration rhythms began to coordinate and to integrate as the adaptation of feeding patterns and the maturation of neurodevelopment (Darnall et al., 2006).

The human body temperature is precisely regulated by a network that consists of the skin thermal sensors, hypothalamic thermoregulatory center, autonomic nervous system, and several thermoregulation effector systems including brown adipose tissue, peripheral vasomotricity, and sweat glands (Bach et al., 1996; Jost et al., 2017). Due to the immaturity of the regulatory network, especially the dysfunction of the autonomic nervous system, their body temperature during the first few days of life is susceptible to the rapidly changed external environment temperature (Jost et al., 2017). Therefore, premature infants are typically nursed in the incubators to treat the autonomic dysregulation of body temperature (Thomas, 2001). Interestingly, Bueno and Menna-Barreto (2016), found a positive correlation between the wrist temperature and environment temperature inside the incubator, but no significant association between the period or potency for them. Similarly, Thomas (2001) demonstrated that the circadian of incubator temperature did not appear to be the primary determinant of the body temperature rhythms.

As summarized in Table 3, due to the heterogeneity of the body temperature monitoring, the GA of preterm infants, and sample size, the body temperature rhythms have not yet been consistently described. Several studies observed the ultradian body temperature rhythms within the first few days of life (Glotzbach et al., 1995; Mirmiran et al., 2003b; Koch et al., 2021), and the circadian rhythms by approximately 1–3 months of PNA (Mirmiran et al., 2003b; Bueno and Menna-Barreto, 2016). Interestingly, Thomas and Burr (2002), found that the acrophase of circadian abdominal skin temperature rhythms was related to the parental co-sleeping and length of hospital stay for preterm infants at 44–46 weeks of PMA. However, some studies demonstrated that the body temperature rhythms were only found in some, but not all preterm infants (Mirmiran et al., 1990; Mirmiran and Kok, 1991; D'Souza et al., 1992; Updike et al., 1985; Thomas, 2001). For example, Tenreiro et al. (1991) found that the ultradian and circadian rhythms of skin temperature appeared and disappeared erratically during 6–17 weeks of PNA, which was similar to the cardiopulmonary rhythms.

As summarized in Table 4, due to the difficulties in sample collection and analysis, studies on hormonal rhythms in preterm infants are still very limited until now, and nearly all focused on the cortisol and melatonin rhythms. With regard to the cortisol, due to the immature of HPA axis (Bolt et al., 2002), no significant circadian or ultradian rhythms were observed during the early postnatal periods (Economou et al., 1993; Jett et al., 1997; Kidd et al., 2005; Dorn et al., 2014). Nevertheless, studies have found that healthy preterm infants had higher nighttime cortisol levels than daytime at birth, and that cortisol levels tended to decrease gradually after birth (Kidd et al., 2005; Dorn et al., 2014). Impressively, premature infants with perinatal stress like respiratory distress experienced higher cortisol levels at nighttime after birth compared with those healthy preterm and term neonates (Economou et al., 1993; Gunes et al., 2006).

It remains unclear when premature infants develop the circadian cortisol rhythms. Antonini et al. (2000) found the salivary cortisol circadian rhythms emerged and persisted at approximately 8–12 weeks of PNA, which was in line with term infants. However, Ivars et al. (2017) found that the cortisol rhythms were established by 1 month of corrected age, persisted throughout the first year of life, but delayed by topical corticosteroid medication. In addition, Ivars et al. (2017) also suggested that the establishment of cortisol rhythms was related to the GA rather than PNA, because the maturation of adrenal cortex was depend on the GA of preterm infants (Bolt et al., 2002).

Circadian melatonin rhythms could not be detected in preterm infants under different ambient illumination conditions during the early postnatal life (Commentz et al., 1996; Mantagos et al., 1996; Biran et al., 2019). Several studies demonstrated that the blood melatonin and urine 6-sulfatoxymelatonin levels were positively correlated with the GA (Biran et al., 2019) and birth weight of preterm infants (Muñoz-Hoyos et al., 2007), but the serum melatonin levels and urine 6-sulfatoxymelaton excretion increased during the first 7 days and even 52 weeks of PNA (Kennaway et al., 1992; Commentz et al., 1996; Muñoz-Hoyos et al., 2007), which might be attributed to the gradual maturation of the pineal gland where the melatonin is mainly synthesized (Commentz et al., 1997).

However, Commentz et al. (1996) found the urine melatonin and 6-hydroxymelatonin sulfate excretion in male preterm infants during 2–7 days of PNA were negatively associated with the GA, indicating that the melatonin levels might be related to the sex. As for the establishment of circadian melatonin rhythms, Kennaway et al. (1992) observed the appearance of urine 6-sulfatoxymelaton circadian rhythms was approximately at 18–21 weeks of PNA, which was delayed by 9 weeks than those term infants and 2–3 weeks after correcting for GA.

In humans, several clinical observational studies with small sample size have witnessed the alterations of circadian sleep-wake (Landolt et al., 1995a; Landolt et al., 1995b; McHill et al., 2014; Weibel et al., 2021), body temperature (Wright et al., 1997; Wright et al., 2000; McHill et al., 2014), blood pressure (Green and Suls, 1996; Guessous et al., 2014), heart rates (Green and Suls, 1996; Kohler et al., 2006; Crooks et al., 2019), melatonin (Wright et al., 1997; Wright et al., 2000; Burke et al., 2015), and cortisol rhythms (Lovallo et al., 2005; Rieth et al., 2016) in adults who consumed caffeine by comparison with placebo controls.

In rodents, caffeine disrupted the mesors, amplitudes, and acrophases of the circadian heart rate, temperature, motor activity, and sleep-wake rhythms (Pelissier et al., 1999; Pelissier-Alicot et al., 2002; Vivanco et al., 2013; Panagiotou et al., 2019). Caffeine also potentiated the light-induced phase shift, which responded to the rest-activity circadian rhythms, indicating that caffeine enhanced the clock sensitivity to light (Antle et al., 2001; Vivanco et al., 2013; van Diepen et al., 2014; Jha et al., 2017; Ruby et al., 2018). In addition, caffeine lengthened the period and amplitude of circadian clocks in mammalian cells in vitro and in mice ex vivo and in vivo (Oike et al., 2011; Narishige et al., 2014; Burke et al., 2015). At the cellular level, caffeine also altered the expression of circadian clock genes, such as Clock, Bmal1, and Per1 in the liver and jejunum of mice under ad libitum feeding conditions (Sherman et al., 2011).

Caffeine influences the circadian rhythms by modulating the endogenous cAMP/Ca²⁺ signaling pathway, the core components of the mammalian circadian pacemaker (Harvey et al., 2020; O'Neill et al., 2008), through a variety of complex mechanisms (Aguilar-Roblero et al., 2007; Narishige et al., 2014; Burke et al., 2015; Landolt, 2015; Jagannath et al., 2021) (Figure 3). Basically, caffeine antagonizes all types of adenosine receptors (A₁, A_2A, A_2B, and A₃ receptors) and mainly functions by non-specifically antagonizing the A₁ and A_2A receptors (Nehlig et al., 1992; Cappelletti et al., 2015; Rodak et al., 2021; Yang et al., 2021). The blockade of adenosine receptors indirectly regulates the production of cAMP by inhibition (A₁ and A₃ receptors) or stimulation (A_2A and A_2B receptors) of adenylate cyclase (Nehlig et al., 1992; Kumar and Lipshultz, 2019; Yang et al., 2021). Caffeine also prevents the degradation and increases the intracellular cAMP levels by non-selectively inhibiting phosphodiesterase (Nehlig et al., 1992; Cappelletti et al., 2015; Kumar and Lipshultz, 2019; Yang et al., 2021). In addition, caffeine mobilizes intracellular Ca²⁺ transmission from the endoplasmic reticulum through activating the ryanodine receptor channels (Aguilar-Roblero et al., 2007; Kumar and Lipshultz, 2019) and the inositol triphosphate receptors (Yang et al., 2021).

The increased cytosolic cAMP/Ca²⁺ activates the protein kinase A (PKA) and Ca²⁺/calmodulin-dependent protein kinase Ⅱ (CaMKⅡ), thereby leading to the phospho-dependent activation of cAMP response element binding protein (CREB), which in concert with its coactivators to activate the cAMP response element (CRE) (Narishige et al., 2014; Harvey et al., 2020; Reichert et al., 2022). Besides, the increased intracellular Ca²⁺ levels also result in the phosphorylation of extracellular regulated protein kinases (ERK), which drives to form the activator protein 1 (AP-1) transcription factor (Jagannath et al., 2021). Then, interestingly, CRE and AP-1 together drive the Per gene transcription (Narishige et al., 2014; Jagannath et al., 2021), which in turn participates in the transcriptional feedback loops that regulate circadian rhythms (Figure 3).

In addition, caffeine affects the release of neurotransmitters, such as γ-aminobutyric acid, dopamine, glutamate, acetylcholine, norepinephrine, and serotonin, between synaptic neurons in almost all brain areas by blocking the adenosine receptors (Nehlig et al., 1992; Cappelletti et al., 2015; Yang et al., 2021) (Figure 3), thereby significantly influencing the sleep-wake rhythms (Kumar and Lipshultz, 2019).

The well-studied effects of caffeine on sleep-wake rhythms in preterm infants are still limited as the sample sizes were small and the study designs were heterogeneous (Table 5). Some studies revealed that the sleep-wake patterns were not significantly changed after short-term treatment with caffeine or theophylline during short observation periods (Gabriel et al., 1978; Curzi-Dascalova et al., 2002; Chardon et al., 2004; Seppä-Moilanen et al., 2021).

However, some other studies observed significant effects of caffeine on the sleep-wake rhythms, although these effects were not entirely consistent (Dietrich et al., 1978; Thoman et al., 1985; Hayes et al., 2007; Hassanein et al., 2015; Koch et al., 2020). For example, Koch et al. (2020), found that the AS decreased while the wakefulness increased but QS unchanged as caffeine concentrations and the PNA increased over the first 5 days of life in preterm infants more than 28 weeks of GA, but no clear effects on the sleep-wake states were found in preterm infants less than 28 weeks of GA, and no such PNA effects were found in no-caffeine cohort. Hassanein et al. (2015) also detected significant decreases in the AS, QS, and drowsiness, while increases in the quite alert, active alert, and crying in preterm infants half an hour after caffeine administration. Similar methylxanthine-induced changes in the AS and wakefulness states were also observed in studies conducted by Hayes et al. (2007) and by Thoman et al. (1985). However, Dietrich et al. (1978) found the AS and wakefulness increased while the QS and IS decreased during theophylline therapy.

In addition, Lee et al. (2010) discovered that the appearance of sleep-wake cycling was associated with the aminophylline use and more prominent. However, in the prospective follow-up study of the CAP trial (Marcus et al., 2014), no significant differences in sleep states were found in preterm infants aged 5–12 years who had been treated with caffeine after birth compared with the placebo group, which possibly due to the apparent discrepancy in total recording and sleep time between the two groups.

This is also true for some animal studies. Denenberg et al. (1982) found that theophylline reduced the AS, while increased wakefulness, delayed the development of QS, and affected the intermediate states of sleep-wake and AS-QS transitions in newborn rabbits. Montandon et al. (2009) also discovered that the sleep time was reduced, sleep onset latency was increased, and non-REM sleep was fragmented in adult rats treated with caffeine compared to controls during the neonatal period.

Due to the heterogeneous designs and inconsistent results of the above studies, it is difficult to draw clear conclusions. Nonetheless, it can be summarized that caffeine affects the sleep patterns in preterm infants, especially the AS and wakefulness, and the effects might persist into the childhood and even the adulthood. If this hypothesis holds true, then the inhibition of adenosine receptors by caffeine would exactly explain the altered sleep-wake states in preterm infants, as the association between caffeine, adenosine, and sleep has been well documented in adults (Huang et al., 2011; Porkka-Heiskanen and Kalinchuk, 2011; Huang et al., 2014a; Urry and Landolt, 2015; Reichert et al., 2022). In addition, the alteration of sleep-wake patterns might be partially responsible for the caffeine-induced increase in cerebral cortical activity (Supcun et al., 2010; Hassanein et al., 2015) and decrease in apneic episodes (Dietrich et al., 1978; Montandon et al., 2009; Seppä-Moilanen et al., 2019; Seppä-Moilanen et al., 2021).

Current studies have confirmed that caffeine acts both peripherally and centrally to stimulate respiration mainly via inhibiting the adenosine A₁ and A_2A receptors (Abdel-Hady et al., 2015; Eichenwald et al., 2016; Dobson and Hunt, 2018). Caffeine activates the medullary respiratory center, improves sensitivity to carbon dioxide, increases respiratory muscle strength, enhances diaphragmatic contractility, and induces bronchodilation (Kassim et al., 2009; Parikka et al., 2015; Dekker et al., 2017; Sanchez-Solis et al., 2020; Williams et al., 2020), which synergistically cause the increased minute ventilation and oxygen consumption, while cause the decreased apnea, periodic breathing, and intermittent hypoxia (Seppä-Moilanen et al., 2021; Seppä-Moilanen et al., 2019; Dobson et al., 2017; Rhein et al., 2014; von Poblotzki et al., 2003; Bauer et al., 2001; Carnielli et al., 2000).

In addition, caffeine or theophylline therapy increases the cardiac output, stroke volume, and metabolic rate (Walther et al., 1986; Walther et al., 1990; Carnielli et al., 2000; Bauer et al., 2001; Soloveychik et al., 2009; Shivakumar et al., 2019), but decreases blood flow velocities in cerebral and intestinal arteries (Pryds and Schneider, 1991; McDonnell et al., 1992; Bucher et al., 1994; Chang and Gray, 1994; Govan et al., 1995; Lundstrøm et al., 1995; Lane et al., 1999; Hoecker et al., 2002; Hoecker et al., 2006; Dix et al., 2018; Hwang et al., 2018; Abdel Wahed et al., 2019) for preterm infants, which appeared to be related to the enhanced endothelial function through antagonism of adenosine receptors, inhibition of phosphodiesterase, and through promotion of intracellular calcium concentrations (Higashi, 2019). Although the clinical significance remains unclear, this reduced perfusion activity was a reminder that caffeine might have adverse effects on the developing brain and gastrointestinal tract (McDonnell et al., 1992; Lane et al., 1999; Hoecker et al., 2002; Hoecker et al., 2006; Atik et al., 2017; Abdel Wahed et al., 2019).

Unlike the cardiopulmonary system, the effects of caffeine on the cardiorespiratory rhythms in preterm infants have not been specifically studied. Nonetheless, the effects of caffeine on the heart rate, respiratory rate, blood pressure, and oxygen saturation have been examined. As summarized in Table 6, some studies found that a loading of caffeine or theophylline increases the heart rate (Hassanein et al., 2015; Dekker et al., 2017; Parikka et al., 2015; von Poblotzki et al., 2003; Carnielli et al., 2000; Soloveychik et al., 2009; Dix et al., 2018; Govan et al., 1995; Chang and Gray, 1994; Bucher et al., 1994; McDonnell et al., 1992; Saliba et al., 1989), blood pressure (Soloveychik et al., 2009; Supcun et al., 2010; Hassanein et al., 2015; Dix et al., 2018; Huvanandana et al., 2019), respiratory rate (Williams et al., 2020), and oxygen saturation (Hassanein et al., 2015), which were in line with those studies with multiple caffeine dosing (Walther et al., 1986; Walther et al., 1990; Hoecker et al., 2006; Shivakumar et al., 2019). Those findings reflected the complex effects, directly or indirectly like the enhanced autonomic nervous system responsiveness (Huvanandana et al., 2019), of caffeine on the cardiopulmonary system. However, several other studies did not find similar effects (Pryds and Schneider, 1991; Dani et al., 2000; Bauer et al., 2001; Hoecker et al., 2002; Ulanovsky et al., 2014).

Unfortunately, no research has touched this area yet in premature infants until now. It is worth mentioning that neonatal caffeine treatment upregulates adenosine receptors in cardiorespiratory related nuclei of the rat brain (Gaytan et al., 2006; Gaytan and Pasaro, 2012), and this effect persists into the adulthood (Bairam et al., 2009), which underscores the urgent to study the potential long-term effects of caffeine on the cardiorespiratory system in preterm infants (Montandon et al., 2008). In view of the complex and profound effects of caffeine in this field, systematic and in-depth research is still necessary.

Two studies recorded the body temperature of preterm infants and incubator temperature during short-term caffeine administration. Chardon et al. (2004) found that caffeine has no significant effect on the skin temperature and incubator temperature. However, Bauer et al. (2001) observed that a lower incubator temperature was sufficient to maintain a normal body temperature for preterm infants after caffeine treatment, which might be related to the increased metabolism caused by methylxanthines (Bucher et al., 1994; Carnielli et al., 2000; Bauer et al., 2001). However, the effects of caffeine on circadian body temperature rhythms have not been extensively studied. Similarly, although caffeine has been shown to affect melatonin (Wright et al., 1997; Wright et al., 2000; Burke et al., 2015) and cortisol (Lovallo et al., 2005; Rieth et al., 2016) rhythms in adults, these effects in premature infants still need to be addressed.

Collectively, the relevant research on the circadian rhythms in premature infants receiving caffeine therapy is still scarce. Although existing studies have suggested the possible effects of caffeine on the circadian rhythms, heterogeneity in study designs and inconsistency in conclusions weaken the power of those evidence. More research is needed in the future to confirm the effects of caffeine and the underlying mechanisms. The story should not end here.

As discussed above, the efficacy of caffeine appeared to interact with the circadian rhythms in premature infants. Studies have demonstrated the significant effects and underlying mechanisms of caffeine in adults and in animals (Landolt, 2015), but it remains unclear whether the similar mechanisms also exist in those preterm infants. The effects of caffeine on the circadian rhythms, especially the sleep-wake rhythms, are advised to be considered into the strategy of the caffeine therapy (Figure 4).

In addition, studies have revealed that several circadian-related problems like sleep, breathing, and blood pressure in premature infants may persist into childhood and even adulthood (Weisman et al., 2011; Huang et al., 2014b; Sipola-Leppanen et al., 2015; Caravale et al., 2017; Durankus et al., 2020). Based on the existing evidence, it is feasible to propose that caffeine’s effects on circadian rhythms may ameliorate those problems and promote the maturation of circadian rhythms in preterm infants to the level of normal term infants.

The concept of chronopharmacology holds that the ADME processes and the sensitivity of a biological target to a drug are determined by the endogenous biological circadian oscillations (Ohdo et al., 2019; Bicker et al., 2020; Dong et al., 2020; Dobrek, 2021). Variable efficacy and safety profiles would be exhibited for many drugs if they are administered at different times of the day (Dallmann et al., 2016; Cederroth et al., 2019; Nahmias and Androulakis, 2021). For preterm infants, interestingly, several circadian-related gene polymorphisms were found to be significantly associated with the response to caffeine therapy for AOP (Guo et al., 2022). It remains unclear whether caffeine administrated at different times of the day would cause changes in the ADME processes and the therapeutic effects, but it really opens a possibility to applicate the chronopharmacology in the NICU.

Although less research is currently available, there are rare but thought-provoking reports that arouse our strong interests (Smolensky et al., 1987; Pelissier-Alicot et al., 2002), which will lead us into a wonderland in the future. For examples, Pelissier-Alicot et al. (2002) found that the pharmacokinetic profiles of caffeine in rats, such as the clearance, volume of distribution, and area under the plasma concentration-time curve (AUC), depended strongly on the time of day of administration, while the daily rhythmicity of heart rate, body temperature, and locomotor activity in rats also changed with the dosing time of caffeine. Similarly, Smolensky et al. (1987) demonstrated that the pharmacokinetic profiles and therapeutic effects of theophylline in asthmatic children varied with the dosing time. These findings attract us that the circadian rhythms might play a critical role in the ADME processes as well as the efficacy and safety of caffeine therapy in preterm infants.

Currently, caffeine is now commonly administered once daily in preterm infants (Long et al., 2021). The question is whether we are willing to make positive attempts to tailor the dosing time according to the principles of chronopharmacology. If the significant association between circadian-related gene polymorphisms and response to caffeine therapy in preterm infants (Guo et al., 2022) were true and phenotypically manifested, then the administration at different time points of the day is more likely to witness those potentially altered pharmacokinetics of and clinical response to caffeine.

Maintaining normal circadian rhythms are necessary to stay health. Essentially, caffeine interferes with these rhythms to a certain extent, and its arousal effects are very important for the AOP management among various pharmacological mechanisms. Therefore, whether to apply caffeine in accordance with the circadian rhythms to maintain the stabilities of these rhythms as much as possible, or to subtly counteract these rhythms to amplify its arousal effect and achieve a better therapeutic effect, all these aspects deserve our in-depth consideration (Figure 4).

If the homeostasis of circadian rhythms were necessary for health, then correcting the possible adverse effects due to preterm birth is a matter that needs to be taken seriously in the NICU, including the effects on the treatment drugs being used. As discussed above, several external stimuli or known as zeitgebers, such as light, sound, temperature, nursing, and parental care, etc., play important roles in the maturation of circadian rhythms. Cycled light (Abraham et al., 2006; Ohta et al., 2006; Bode et al., 2011), music therapy (Arnon et al., 2006; Loewy et al., 2013), appropriate incubator temperature (Tourneux et al., 2008), comfortable nursing (Collins et al., 2015; Lan et al., 2018), and even the adequate parental care (Löhr and Siegmund, 1999; Nishihara et al., 2002; Park et al., 2020) are helpful for the development and maturation of the circadian rhythms in neonates.

Therefore, the beneficial effects of those external stimuli on the circadian rhythms for premature infants cannot be ignored in the NICU, taking the application of caffeine to manage the AOP for example (Figure 4). Coordinating all treatment strategies with the principles of circadian rhythms will be a constructive attempt to improve the disease management and care for premature infants. Assuredly, we have to admit that only rare evidence is available currently, and the realization of the therapeutic strategies cannot be achieved overnight. However, any kind of discussions, attempts, and efforts in this field should well be encouraged in the future.