Abstract
BACKGROUND:
Sleep and endocrine disruptions are prevalent after traumatic brain injury (TBI) and are likely to contribute to morbidity.
OBJECTIVE:
To describe the interaction between sleep and hormonal regulation following TBI and elucidate the impact that alterations of these systems have on cognitive responses during the posttraumatic chronic period.
METHODS:
Review of preclinical and clinical literature describing long-lasting endocrine dysregulation and sleep alterations following TBI. The bidirectional relationship between sleep and hormones is described. Literature describing co-occurrence between sleep-wake disturbances and hormonal dysregulation will be presented. Review of literature describing cognitive effects of seep and hormones. The cognitive and functional impact of sleep disturbances and hormonal dysregulation is discussed within the context of TBI.
RESULTS/CONCLUSIONS:
Sleep and hormonal alterations impact cognitive and functional outcome after TBI. Diagnosis and treatment of these disturbances will impact recovery following TBI and should be considered in the post-acute rehabilitative setting.
Introduction
Traumatic brain injury (TBI) is a major cause of disability (Faul, Xu, & Wald, 2010; Selassie et al., 2008). Individuals with a history of TBI will show long-lasting symptoms of disturbed affect, hypervigilance, cognitive deficits, fatigue and autonomic dysregulation (Agha et al., 2004; Ashman et al., 2008; Ashman, Gordon, Cantor, & Hibbard, 2006; Rapoport & Feinstein, 2000). Even in those TBI cases with no overt anatomical damage, these common chronic complaints are not unusual. All the above-mentioned symptoms are associated with neuroendocrine function and sleep. Hormones and sleep will influence each other. Previous literature has shown that insomnia is associated with neuroendocrine dysfunction (Buysse, 2013; Roth, 2007). In turn, sleep plays a critical role in hormonal regulation (Van Cauter et al., 2007). Here we will present literature describing the interplay between sleep and hormones. The bidirectional relationship between hormones and sleep may lead to the propagation of posttraumatic morbidity. We will further describe how sleep and neuroendocrine function are disrupted following TBI and the impact that this has on long-term recovery.
TBI will impact brain regions involved in neuroendocrine and sleep regulation
Biomechanical forces at the time of injury frequently lead to injury that involves the diencephalon and midbrain resulting in the disruption of neural circuits that are critical for both neuroendocrine and sleep regulation (Gennarelli et al., 1982; Holbourn, 1943; Meaney, Morrison, & Dale Bass, 2014). Particularly notable is the impact that TBI has on cholinergic and adrenergic pathways (Arciniegas et al., 1999; Dixon, Hamm, Taft, & Hayes, 1994; Kobori, Hu, & Dash, 2011). The orexin/hypocretin system, has also been found to be affected following TBI (Baumann et al., 2009; Baumann, Werth, Stocker, Ludwig, & Bassetti, 2007). These pathways are intimately involved in sleep-wake regulation (Saper, Fuller, Pedersen, Lu, & Scammell, 2010). Impaired arousal after severe TBI has been associated with loss of serotonergic and noradrenergic neurons within the dorsal raphe nucleus and locus coeruleus (Valko et al., 2016). Posttraumatic disruption of arousal circuitry may involve multiple areas such as cerebral cortex, thalamus, hypothalamus and nuclei of the ascending arousal system. This will impact regions such as the ventrolateral preoptic areas which play a pivotal role in sleep regulation (Dyken, Afifi, & Lin-Dyken, 2012; Schwartz & Kilduff, 2015). Circuitry disruptions are more likely to contribute to sleep-wake disturbances after non-severe TBI; where diffuse injury, in contrast to cell loss, is prevalent. Disruptions in circuitry will also impact neuroendocrine function after TBI, which in turn is likely to influence monoamine regulation (Figlewicz, 1999; Nicolaides, Charmandari, Chrousos, & Kino, 2014; Sotomayor-Zarate, Cruz, Renard, Espinosa, & Ramirez, 2014). Astrocytic damage after TBI (Bartnik, Lee, Hovda, & Sutton, 2007; Kou & VandeVord, 2014; Raghupathi, 2004) is also likely to contribute to adenosine dysregulation (Halassa et al., 2009). Adenosine is an endogenous sleep promoting factor (Porkka-Heiskanen et al., 1997).
Endocrine anomalies following TBI
TBI can result in hypothalamic pituitary dysfunction, which may delay or interfere with functional recovery (Kornblum & Fisher, 1969; Brent E. Masel & Urban, 2015). Because of its vascular and anatomical characteristics, the hypothalamic-pituitary system is particularly vulnerable to TBI. Hypothalamic-pituitary disruption has been reported after TBI, including mild TBI. Indeed, dysregulation of neuroendocrine function can take place without overt histological damage (Benvenga, Campenni, Ruggeri, & Trimarchi, 2000; Griesbach, Hovda, Tio, & Taylor, 2011; Harper, Doyle, Adams, & Graham, 1986). Multiple studies have demonstrated a notable incidence of posttraumatic chronic neuroendocrine dysfunction (Bondanelli et al., 2004; Lieberman, Oberoi, Gilkison, Masel, & Urban, 2001; Tanriverdi, Unluhizarci, & Kelestimur, 2010; Tritos, Yuen, & Kelly, 2015). Variability in studies addressing posttraumatic pituitary dysfunction can be attributed to multiple factors either intrinsic or extrinsic to the injury. Extrinsic factors encompass methodology utilized for diagnosis and confounding effects of concurrent medications. While acknowledging that multiple neuroendocrine systems are affected by TBI, we will focus on hormonal alterations that are recurrently observed during the chronic posttraumatic period and are likely to have a considerable impact on cognitive function.
Whereas hormonal alterations during the acute period can be attributed to adaptive responses to TBI, endocrine alterations persisting well beyond the subacute period are more likely to reflect hypothalamic pituitary dysfunction. For example, activation of the hypothalamic-pituitary-adrenal (HPA) axis is acutely observed as a trauma-induced stress response, which results in adrenocorticotropin hormone (ACTH) release and consequent elevation of cortisol concentrations in an injury severity dependent manner (Johnson, 2006; Woolf et al., 1990). During the subacute period, the post-injury glucocorticoid profile differs depending on the characteristics intrinsic to the injury. Plasma cortisol levels tend to increase after mild to moderate brain injury, whereas after a severe injury, fasting cortisol levels are depressed (Cernak, Savic, Lazarov, Joksimovic, & Markovic, 1999; Englander, Bushnik, Oggins, & Katznelson, 2010; Steinbok & Thompson, 1979). Similarly, animal studies have indicated severity-dependent alterations in HPA function after experimental models of TBI that result in a region of marked cell death (Griesbach & Hovda, 2015; Taylor et al., 2008). Chronic posttraumatic anomalies are observed over 1 or more hypothalamic-pituitary axes. Clinical studies in TBI patients have described long-lasting TBI induced HPA dysregulation as abnormal levels of ACTH and cortisol levels (Cohan et al., 2005; Marina, Klose, Nordenbo, Liebach, & Feldt-Rasmussen, 2015; Tanriverdi et al., 2006; Yuan & Wade, 1991). Glucocorticoid release is tightly regulated through a negative feedback to prevent further release of ACTH and cortisol. Hyper-responsiveness to stress and dysregulation of the HPA axis has been reported following experimental TBI (Griesbach et al., 2011; Reger et al., 2012). The duration of this hyper-responsiveness appears to be dependent on TBI severity.
Disruption of the hypothalamic pituitary somatrotopic (HPS) axis, following TBI, is predominantly observed as growth hormone (GH) deficiency. A recent study found that 45% of chronic TBI patients receiving post-acute rehabilitation presented with severe GH deficiency (Kreber, Griesnbach, & Ashley, 2016). GH is a pleiotropic hormone synthesized in the anterior pituitary gland that stimulates the production and release of insulin-like growth factor-1 (IGF-1). Patients with severe GH deficiency are more likely to have low levels of other hormones, suggestive of pituitary damage and/or altered circuitry regulating endocrine responses (Kreber et al., 2016).
Alterations in the hypothalamic pituitary gonadal (HPG) axes have also been reported following TBI. HPG axis dysregulation involves the gonadotropin releasing hormone (GnRH) which will lead to alterations in sex hormones such as estradiol, progesterone and testosterone. In men hypogonadism is typically associated with low levels of testosterone and in women is usually associated with decreases in estradiol. Male hypogonadism is acutely present after TBI and has been found to persist over time in 37% of male patients (Wagner et al., 2012). In women, HPG suppression has been attributed to hypercortosolemia (Ranganathan et al., 2016). Anovulation observed in response to changes in cortisol levels after TBI typically resolve.
Although not as frequently observed, persistent disruption of the hypothalamic pituitary thyroid (HPT) axis can have substantial consequences given the widespread metabolic effects of thyroid hormones (Bianco, Salvatore, Gereben, Berry, & Larsen, 2002). It should be noted that thyroid hormone activity is influenced by other endocrine signals and compounds, such as cortisol (Tohei, 2004; Zhang, Boelen, Bisschop, Kalsbeek, & Fliers, 2017) and gonadal hormones (Marassi et al., 2007).
Sleep disturbances following TBI
The prevalence of sleep disorders following TBI is notably higher compared to the general population (Fogelberg, Hoffman, Dikmen, Temkin, & Bell, 2012; Mathias & Alvaro, 2012; Mazwi, Fusco, & Zafonte, 2015; Sandsmark, Elliott, & Lim, 2017; Verma, Anand, & Verma, 2007; Williams, Lazic, & Ogilvie, 2008). Common sleep related complaints after TBI include sleep and wake disturbances (Chen et al., 2015; Ouellet, Beaulieu-Bonneau, & Morin, 2015). Insomnia, as defined by difficulty initiating or staying asleep, has been reported in up to 60% of the TBI population across all TBI severities (Orff, Ayalon, & Drummond, 2009; Ouellet & Morin, 2006). Daytime sleepiness is associated with insomnia. Hypersomnia, and fatigue are other frequently reported complaints according to both subjective and objective reports (Imbach et al., 2015; Lu et al., 2015; Mazwi et al., 2015). Hypersomnia has also been reported after experimental TBI at multiple posttraumatic periods (Skopin, Kabadi, Viechweg, Mong, & Faden, 2015). In humans, hypersomnia has been reported up to 3 years post TBI (Kempf, Werth, Kaiser, Bassetti, & Baumann, 2010; Masel, Scheibel, Kimbark, & Kuna, 2001). Excessive sleepiness and fatigue can result from an array of sleep disturbances ranging from circadian disruptions to breathing related disorders. Irregularities in circadian cycles are frequently observed and likely to contribute to daytime sleepiness and fatigue (Ayalon, Borodkin, Dishon, Kanety, & Dagan, 2007; Patten & Lauderdale, 1992). Posttraumatic circadian anomalies may be attributed to hypothalamic circuitry disruptions affecting suprachiasmatic nucleus and pineal gland input (Kumar, 2008).
Breathing related sleep disorders are also frequently observed during the chronic TBI period (Castriotta & Lai, 2001; Webster, Bell, Hussey, Natale, & Lakshminarayan, 2001). Apnea is defined as periods of breathing cessation for fixed periods during sleep. Apnea in turn can also contribute to arousal and consequent fragmented sleep. Apnea can be categorized as obstructive, central or mixed. Obstructive apnea will involve collapse or obstruction of upper airway; whereas central apnea is defined as secondary to neurologic causes (Bresnitz, Goldberg, & Kosinski, 1994). All three forms have been reported after TBI. Damage to the medullary respiratory center may account for obstructive and central apnea (Dyken et al., 2012). Increases in adiposity after TBI may also account for obstructive apnea.
Besides being regulated by sleep-wake homeostasis and circadian cycles, sleep also follows a particular overnight pattern consisting of repeated sleep cycles (Fig. 1). Polysomnography (PSG) studies have revealed irregularities in sleep cycle pattern, i.e. architecture, following TBI (Sandsmark et al., 2017). Brain activity during sleep is generally distinguished between rapid eye movement (REM) and non-REM (NREM) (Silber et al., 2007). NREM includes different stages. N1, formerly known as “Stage 1”, is identified as the transitory stage from wake to sleep and is electroencephalographically characterized as diminishing alpha waves with theta wave intrusions. N2, formerly known as “Stage 2” is an intermediary stage leading to slow wave sleep (SWS) that was formerly identified as Stages 3 and 4. N2 is characterized by distinct activity such as K complexes that are associated with suppression of cortical arousal (Dang-Vu et al., 2011) and occasional sleep spindles. SWS is identified by delta waves which have a higher amplitude and slow frequency. SWS has been associated with memory enhancement endocrine regulation and metabolic “hygiene”. REM stage usually follows SWS and has also been associated with cognitive functioning. REM is identified by the resurgence of theta and beta waves, assimilating wakefulness with the notable absence of muscle movement, except for that detected via electro-oculography. These periods cycle throughout the night. As the night progresses, SWS sleep periods will shorten. In turn, the length of the REM periods increases as the wake time approaches (Colrain, 2011). Notable alterations and fragmentation of sleep cycles are observed following TBI. In effect, TBI subjects tend to show significantly more arousals compared to healthy controls (Cohen, Oksenberg, Snir, Stern, & Groswasser, 1992; Lu et al., 2014; Parcell, Ponsford, Redman, & Rajaratnam, 2008; Ron, Algom, Hary, & Cohen, 1980). Characteristics of these disturbances are highly variable within the TBI population and likely to be reflective of TBI heterogeneity.

Sleep Cycle. (A) Sleep follows an overnight cycle that consists of non-rapid eye movements (NREM) and rapid eye movement (REM) stages. Each sleep cycle has an approximate duration of 90 minutes in humans. (B) Sleep cycles throughout the night. During the first cycles of the night the duration of SWS is longer. The length of REM stages increases as the night progresses. Fragmentation and changes in overnight pattern are frequently observed after traumatic brain injury.
A recent meta-analysis of 16 PSG studies showed that TBI subjects had significantly more sleep disorders compared to healthy controls. TBI resulted in poorer sleep efficiency, shorter total sleep duration and increased wake after sleep onset. Data was also suggestive that TBI subjects spent less time in REM sleep (Grima, Ponsford, Rajaratnam, Mansfield, & Pase, 2016). An array of sleep architecture anomalies has been reported following TBI. These include a higher proportion of N1 (Ouellet & Morin, 2006), decreases in REM (Parcell et al., 2008; Schreiber et al., 2008), and disruptions associated to N2 (Schreiber et al., 2008; Urakami, 2012), as well as SWS (Shekleton et al., 2010). It should be noted that discrepancy in PSG is not uncommon when studying TBI. Limitations in these studies include limited number of study participants, differences in equipment, procedures and scoring methodology, comorbidities, presence of medications and multiple factors intrinsic to the injury that are were not properly controlled. For example, decreases in N2 has been found in patients with a moderate to severe TBI (Lu et al., 2015). N2 decreases may be more reflective of thalamocortical network damage, where spindles are generated. In contrast, these decreases are not reported in patients that suffered a mild TBI (Schreiber et al., 2008). Comorbidities such as anxiety and depression should also be taken into consideration given that these are commonly observed after TBI and have been associated with sleep disruptions and REM alterations (Buysse, 2013) (Cowdin, Kobayashi, & Mellman, 2014; Giles, Biggs, Rush, & Roffwarg, 1988).
Sleep-wake disturbances often coincide with hormonal dysregulation, as the endocrine system has a complex response to sleep. Research shows that some hormones increase secretion during sleep while others decrease. Furthermore, subsets of hormones’ levels are associated with specific stages of the sleep cycle. SWS is thought to be the most restorative period of sleep and is closely related to modulation of endocrine release. In addition, fragmented sleep is linked to abnormal glucose metabolism and increased incidence of diabetes (Stamatakis & Punjabi, 2010; Tasali, Leproult, Ehrmann, & Van Cauter, 2008). Thus it is reasonable that sleep disorders, such as apnea lead to disruptions in hormonal release (Harper, Kumar, Macey, Ogren, & Richardson, 2012). Accordingly, hormonal dysregulation is associated with sleep disturbances in survivors of severe TBI (Frieboes et al., 1999). The following section will describe the regulation of specific hormones during sleep.
GH release
GH is primarily secreted during the first few hours of sleep onset and is associated with SWS (Van Cauter et al., 2004). Studies have indicated a decrease in nocturnal GH release in sleep-deprived persons, but show a corresponding increase during daytime recovery sleep. Correspondingly, GH release is decreased in individuals who experience a number of awakenings during the night, thus interrupting the normal sleep pattern (Van Cauter et al., 2007). Indeed, patients suffering from PTSD, which is characterized by disturbed and fragmented sleep, exhibit lower nocturnal GH plasma levels (van Liempt, Vermetten, Lentjes, Arends, & Westenberg, 2011). In cases of chronic insomnia, moderate increases in GH levels are observed. These increases may be indicative of a circadian influence and/or short periods of sleep where SWS occurs due to sleep-wake homeostasis (Fig. 2A). Early sleep-onset GH secretion may be primarily regulated by growth hormone releasing hormone (GHRH). GH secretion may also be influenced by the peptide somatostatin and ghrelin, known as the hunger hormone, which are all, in turn, regulated by sleep (Takaya et al., 2000; Van Cauter et al., 2004). GHRH and somatostatin gene expression exhibit modulation by REM sleep deprivation in rats, which corresponds to levels of serum GH (Toppila et al., 1996). Ghrelin levels have been found to peak in the early part of the night and exogenous administration has been shown to increase SWS and growth hormone levels in humans and rodents (Ariyasu et al., 2005; Weikel et al., 2003). Additionally, studies have shown a decreased capacity for leptin and ghrelin to accurately signal caloric need after periods of sleep deprivation; leptin levels have been shown to decrease, while ghrelin levels increase after short sleep duration (Taheri, Lin, Austin, Young, & Mignot, 2004). Given the relationship between these satiety signals, sleep, and growth hormone levels, it is reasonable to assume that a disruption in one will affect the others.

Effects of sleep on secretion of growth hormone (GH), cortisol and thyroid stimulating hormone (TSH). (A) GH release is strongly regulated by the sleep cycle. Elevations are typically observed during the earlier part of the night and notably diminished with sleep deprivation. Elevations in GH are observed after chronic sleep restriction, although, to a lesser degree compared to GH increases with uninterrupted sleep. (B) Cortisol levels typically increase during the second half of the night. The circadian pattern is maintained with a night of sleep deprivation. Elevations in cortisol are evident when sleep disruption is persistent. (C) TSH is primarily secreted during the late evening and declines throughout the night. The circadian pattern of TSH is mostly maintained with temporary sleep deprivation. Chronic sleep restriction is associated with TSH decreases.
Cortisol is strongly associated with circadian rhythm. In healthy individuals, cortisol levels are low in the evenings, begin to rise during the second half of the night, peak in the morning and slowly decline throughout the day (Friess, Wiedemann, Steiger, & Holsboer, 1995). SWS has been associated with ACTH suppression; thus, cortisol increases during the latter half of the night may be associated with the normal overnight sleep pattern which involve decreases in SWS as the night progresses (Bierwolf, Struve, Marshall, Born, & Fehm, 1997). The circadian pattern of cortisol is more resilient to sleep restriction, compared to GH (Fig. 2B). However, individuals subjected to experimental sleep restriction display elevated levels of evening cortisol (Spiegel, Leproult, & Van Cauter, 2003). Similar increases are also seen in healthy individuals with increased nighttime awakenings (Leproult, Copinschi, Buxton, & Van Cauter, 1997; Spiegel et al., 2004). It has been observed that sleep begins when HPA axis activity is at its lowest point and its activity is increased with sleep deprivation. Thus, mechanisms promoting activation of the HPA axis likely inhibit/prevent SWS and the corresponding secretion of SWS associated hormones, like GH. Even partial sleep deprivation is enough to delay HPA axis recovery and could thus modify negative glucocorticoid feedback regulation. Overall, this could affect the normal stress response and hasten the development of metabolic and cognitive consequences of increased levels of glucocorticoids. The relationship between glucocorticoids and sleep is bidirectional and likely to be affected by brain injury. Correspondingly, insomnia following mild TBI has been associated with HPA dysregulation (Zhou et al., 2015). Although chronic insomniacs show preservation of the circadian pattern for cortisol, they present overall increases in ACTH and cortisol, indicative of HPA hyperarousal, that may lead to increased vulnerability of affective disorders (Abell, Shipley, Ferrie, Kivimaki, & Kumari, 2016; Vgontzas et al., 2001). Moreover, the amount of sleep restriction appears to be correlated with diurnal ACTH increases (Guyon et al., 2014). It should be mentioned that other hormones associated to the HPA axis, such as aldosterone, which regulates hydro-mineral balance and are linked to overall health, have also been shown to be affected by sleep deprivation (Charloux et al., 2001).
Thyroid hormone release
Disruptions in sleep are likely to impact thyroid hormone regulation which also has a circadian and pulsatile component (Brabant et al., 1990). Thyroid hormones are secreted primarily in the late evening and progressively decline during the night and throughout the day. Normal regulation is disrupted following periods of poor sleep (Fig. 2C). For example, like cortisol, frequent night time awakenings increase levels of thyroid stimulating hormone (TSH) (Van Cauter et al., 2007). TSH will lead to the release of thyroxine (T4), the precursor molecule that is converted into triiodothyronine (T3). T3 effectively binds to nuclear thyroid hormone receptors and regulates thyroid hormone-dependent transcriptional activation (Gereben, Zeold, Dentice, Salvatore, & Bianco, 2008). The profile of sleep deprivation and TSH levels differs when it is extended. TSH levels have been found to significantly decline following periods of repeated sleep restriction (Spiegel et al., 2004; Spiegel et al., 2003). Similar findings have been observed in animal studies. Prolonged sleep deprivation in rodents decreased serum T4 levels (Everson & Nowak, 2002). It should be noted that serum TSH was not affected in this study. Alike the HPA axis, the HPT axis is also controlled by a negative feedback mechanism allowing for its tight regulation. Accordingly, this animal study found increased thyroid releasing hormone mRNA expression (Everson & Nowak, 2002). The adaptive responses of the thyroid system were more recently observed in a study comparing a rodent model of sleep restriction with one of sleep deprivation. In the rodent sleep deprivation model, REM was found to play a role in TSH secretion. In this study, REM deprivation resulted in hypothyroidism (Rodrigues et al., 2015). In humans, partial sleep restriction was accompanied by statistically significant reductions in TSH and free T4. These decreases were principally detected in female participants (Kessler, Nedeltcheva, Imperial, & Penev, 2010). These findings are illustrative of the complexity of hormonal influences. Similar to the above-mentioned hormones, thyroid hormones are also influenced by other endocrine signals and compounds, including cortisol and gonadal hormones (Marassi et al., 2007; Tohei, 2004). Hypothyroidism has been shown to decrease adrenal weights and corticosterone levels in rats, while also lowering the pituitary response to luteinizing hormone-releasing hormone, increasing concentrations of progesterone and prolactin in female rats and reducing copulatory behavior in male rats (Tohei, 2004). Ultimately, a feedback loop exists between thyroid hormones, various other endocrine signals and sleep. Given this, alterations in other hypothalamic pituitary axes, due to TBI, will indirectly affect thyroid hormone regulation. In addition, it is reasonable that disturbances in sleep will also be a contributing factor.
Gonadal hormones release
Gonadal hormone release is also highly associated with SWS. Secretion of various sex hormones, including testosterone, estrogen, progesterone and luteinizing hormone (LH) is increased throughout the sleep period in normal healthy adults. Moreover, interrupted and fragmented sleep seems to significantly affect the secretion of these hormones (Pietrowsky, Meyrer, Kern, Born, & Fehm, 1994; Schiavi, White, & Mandeli, 1992; Van Cauter et al., 2007). Testosterone secretion has been shown to rely on sleep, rather than circadian rhythms, and appears to be particularly affected by poor sleep (Axelson et al., 2013; Wittert, 2014). Normal rises in testosterone present with the appearance of first REM episode and have been shown to require a minimum of 3 hours of sleep with normal architecture. Fragmented sleep results in a significant delay in testosterone rise (Luboshitzky, Zabari, Shen-Orr, Herer, & Lavie, 2001). Estrogen levels which are associated with menstrual cycles, are also associated with electrophysiological changes during sleep. In turn, sleep spindles and SWS are altered by estrogen and progesterone, during the menstrual cycle (Driver, Dijk, Werth, Biedermann, & Borbely, 1996; Genzel et al., 2012). LH is also increased during both nocturnal and diurnal sleep (Pietrowsky et al., 1994; Rossmanith, 1998).
Impact of sleep and hormonal alterations
As indicated above, a notable portion of those that endured a TBI present with long-lasting neuroendocrine dysregulation across several hypothalamic-pituitary axes. In addition, hormonal anomalies frequently co-occur with sleep disturbances. Both, hormonal and sleep disturbances have cognitive and functional consequences that will impact recovery following TBI and should be considered in the post-acute rehabilitative setting. Sleep and neuroendocrine anomalies will influence each other and contribute to post TBI morbidity.
Sleep and cognitive function
Deficits across a wide array of cognitive areas, including attention and memory, are greater in TBI patients with concurrent sleep disorders (Mahmood, Rapport, Hanks, & Fichtenberg, 2004; Wilde et al., 2007). There is persuasive evidence indicating that sleep supports long-term memory. Interestingly, not only has it been demonstrated that sleep is involved in memory consolidation but also in the process of forgetting (Feld & Born, 2017). Decreases in long-term-potentiation (LTP) like plasticity have been observed following sleep deprivation in healthy subjects (Kuhn et al., 2016). LTP is a cellular correlate of learning (Kandel, Dudai, & Mayford, 2014; Siegelbaum & Kandel, 1991). These decreases in LTP support the hypothesis that associative plasticity may be induced through homeostatic modifications of net synaptic strength occurring during the sleep-wake cycle (Tononi & Cirelli, 2014). According to this hypothesis, learning during the wake period requires a large amount of synaptic strengthening that is accompanied by an energetic cost, and during sleep, synaptic plasticity may be restored allowing for memory consolidation to take place.
Animal studies have demonstrated neural substrates of learning and memory during sleep (Sandsmark et al., 2017). Imaging of cortical neurons has demonstrated the formation of dendritic spines during sleep when sleep was preceded by learning a motor task. This neural substrate of learning occurred in a performance dependent manner (Yang et al., 2014). Important studies have shown that SWS is involved in memory consolidation (Diekelmann & Born, 2010; Feld & Diekelmann, 2015). Other studies recoded spiking activity of visual cortex and CA1 hippocampal cells during maze learning and obtained firing rate maps. The firing pattern of these cells was later replayed during SWS (Z. Chen, Grosmark, Penagos, & Wilson, 2016; Wilson & McNaughton, 1994). Other NREM stages may also contribute to learning, due to particular physiological components. Sleep spindles, which are observed during N2 have been linked with neural plasticity (Andrade et al., 2011; Boutin et al., 2018; Schabus et al., 2006).
REM sleep has also been found to play an important role in memory. Spine formation and pruning after motor training is associated with enhanced performance in the same motor task (Li, Ma, Yang, & Gan, 2017). Conversely, REM inhibition was found to have a negative impact on task performance. This same study also demonstrated that REM spine elimination facilitated the formation of new spines in adjacent areas. REM appears to reorganize hippocampal synaptic activity facilitating long-term memory; potentially allowing for active forgetting processes to take place through NMDA mediated mechanisms (Grosmark, Mizuseki, Pastalkova, Diba, & Buzsaki, 2012; Kuhn et al., 2016). Human studies supporting the concept of sleep promoting forgetting is limited. To our knowledge, there are a few studies that indirectly support this idea. In one study, infants showed decline in item memory during sleep while increasing learning of grammatical rules (Gomez, Bootzin, & Nadel, 2006). In another study, a capacity limit for replay during sleep is suggested. Healthy participants, were more likely to remember a list of words after sleep if the list was not too extensive (Feld, Weis, & Born, 2016).
It has been proposed that sleep-dependent memory and learning relies on the presence of both REM and NREM sleep (Diekelmann & Born, 2010; Sonni & Spencer, 2015). Moreover, sleep cycle organization appeared to be critical, in that learning benefits were greatest when REM followed NREM. Sleep fragmentation also was observed to have a negative impact on recall. However, this negative impact was greatest when fragmented sleep was disorganized. In other words, subjects that performed worse were those that who were awoken during NREM throughout the night (Ficca, Lombardo, Rossi, & Salzarulo, 2000).
Sleep is also known to contribute to emotional and psychological well-being. Excessive daytime sleepiness, as a consequence of apnea, has been found to affect mood and judgement (Wilde et al., 2007). Indeed, sleep has been observed to play a critical role in emotional processing. The necessity of both SWS and REM is apparent when considering the emotional aspects of cognition (Lerner et al., 2016). Consolidation of emotional memories is best after a full night of sleep (Hu, Stylos-Allan, & Walker, 2006; Payne & Kensinger, 2011; Payne, Stickgold, Swanberg, & Kensinger, 2008). REM has been associated with the processing of emotional stimuli and preservation of emotional reactivity (Baran, Pace-Schott, Ericson, & Spencer, 2012; Groch, Wilhelm, Diekelmann, & Born, 2013). Evaluation of emotional faces is correlated to the time spent in REM during the preceding night (Wagner, Fischer, & Born, 2002). Alternatively, SWS has been implicated in contextual processing and declarative memory. The amount of time spent in REM and SWS has been linked to emotional ambiguity as well as level of stimuli averseness (Gujar, McDonald, Nishida, & Walker, 2011). Disruptions in sleep staging, which are observed in the TBI population, are likely to have an influence on affect. In effect, sleep staging has been found to be predictive of mood when consolidation of emotional memories occurred (Jones, Schultz, Adams, Baran, & Spencer, 2016).
Hormones and cognitive function
Like sleep, hormone dysregulation may alter cognitive function and mood. Disruptions in these areas can have major impacts on healthy adults and may compound and exacerbate impairments from TBI. Hormones, along with distinct neurochemical patterns, will interact with sleep and may influence memory consolidation. For example, low levels of cortisol and acetylcholine are essential for memory consolidation during sleep (Gais & Born, 2004; Kelemen, Bahrendt, Born, & Inostroza, 2014; Mitra et al., 2016; Plihal & Born, 1999).
Thyroid hormones
Dysregulation of thyroid hormones can impair mood, memory, metabolism and energy levels. This has been demonstrated in humans, as well as animal models. In rodents, T3 administration increased mRNA expression of reelin and brain-derived neurotrophic factor (BDNF), which are both involved in neural plasticity (Sui, Ren, & Li, 2010). In addition, the hippocampus, crucial for learning and memory processes, is particularly vulnerable to thyroid imbalance, specifically to low thyroid hormone levels. Treatment with thyroid hormones in adult animals has been shown to modify dendritic spine density in hippocampal CA1 cells (Calza, Aloe, & Giardino, 1997).
In addition to regulating other hormones, thyroid hormones regulate systemic glucose metabolism and may also regulate brain glucose. Consequently, many metabolic disorders present with thyroid hormone imbalance (Jahagirdar & McNay, 2012). Glucose is required for optimal cognitive processing; thus, regulatory changes will have a deleterious impact on function (Gold, 2005; Holmes, Koepke, & Thompson, 1986; McCrimmon, Ryan, & Frier, 2012). Both the HPA and HPT axes are involved in glucose metabolism and are bi-directionally related to sleep (Petit & Magistretti, 2016; Tamisier, Weiss, & Pepin, 2018). TBI induced dysregulation of these axes will impact glucose metabolism and is likely to be a contributing factor in cognitive deficits following TBI. Indeed, alterations in glucose metabolism are observed following TBI (Bartnik-Olson, Harris, Shijo, & Sutton, 2013; Bergsneider et al., 1997; Hovda, Yoshino, Kawamata, Katayama, & Becker, 1991).
Cortisol
Both human and animal studies have shown an extensive link between HPA axis dysregulation and negative mood states. Despite an initial draw to the monoamine theory of depression, more evidence is gaining in favor of a theory that involves HPA axis dysfunction in depression and bipolar disorders, as well as panic disorders (Strohle & Holsboer, 2003; Vreeburg et al., 2009). In addition to negative mood, over-activation of the HPA axis, which facilitates cortisol release, has been shown to impair learning and memory. Animal studies have demonstrated that increases in corticosterone, the major stress hormone in rodents, equivalent to cortisol, leads to decreased levels of BDNF in the hippocampus (Nibuya, Takahashi, Russell, & Duman, 1999; Smith, 1995; Ueyama, 1997). This is notable, as BDNF plays a critical role in activity-dependent plasticity by facilitating the synapse (Tyler & Pozzo-Miller, 2001) and enhancing neurotransmitter release (Levine, Crozier, Black, & Plummer, 1998; Takei et al., 1997). There is also evidence suggesting that stress, via HPA axis hormones, may cause differential effects on learning and memory processes based on whether the stress is acute or chronic; where acute stress may facilitate learning/memory (Wolf, 2003).
Stress interacts in a bidirectional manner with sleep. Both involve similar neural pathways One recent study has demonstrated a link between pre-learning cortisol and sleep to improve memory consolidation (Bennion, Mickley Steinmetz, Kensinger, & Payne, 2015). Conversely, HPA dysregulation is likely to exacerbate sleep disturbances impacting mood and cognition. The interplay between stress and sleep is also evident in metabolism. Sleep deprivation has been associated with HPA axis hyperactivity, which in turn negatively affects glucose tolerance (Spiegel, Leproult, & Van Cauter, 1999). Alterations in HPA axis and sleep contribute to the prevalence of metabolic disorders, such as obesity and diabetes (Hirotsu, Tufik, & Andersen, 2015; Seetho & Wilding, 2014). As indicated above, the HPA axis is tightly regulated via a negative feedback. Disruptions in this feedback, associated with changes in the sleep-wake cycle, may negatively impact glucose metabolism (Spiegel, Knutson, Leproult, Tasali, & Van Cauter, 2005). Thus, over-activation of the HPA axis may be a risk factor in development of metabolic syndrome. Those with apnea are likely to be at a higher risk, as sleep related breathing disorders are associated with increased visceral obesity, insulin resistance, and sympathetic activity as well as alterations in leptin levels (Buckley & Schatzberg, 2005; Sandoval & Davis, 2003).
Gonadal hormones
Beyond being sex hormones, gonadal hormones have been identified as having crucial effects on normal brain development, cognition and mood. Gonadal hormones will also influence the cognitive effects of sleep. Fluctuations in estrogen and progesterone levels have been associated with electrophysiological changes in SWS and sleep spindles that have an impact on memory (Genzel et al., 2012; Maki, Rich, & Rosenbaum, 2002; Shin et al., 2018). Gonadal steroids include androgens, estrogens and progestogens. Estrogen has been found to alter neural activity that is essential for learning and memory in both preclinical and clinical studies (Luine, 2014). Estrogen facilitates hippocampal morphological changes such as hippocampal dendritic spine formation and neurogenesis (Frick, Kim, Tuscher, & Fortress, 2015). Estradiol, the predominant form of estrogen, has a dose-dependent effect on memory (Bean, Ianov, & Foster, 2014). Positive correlations between estradiol and memory have been found in older men and women (Boss, Kang, Bergstrom, & Leasure, 2015). While estradiol has been positively associated with spatial and working memory, especially in rodent models, there seems to be an inverse relationship between estradiol and LH, where high levels can impair spatial memory, and may induce such impairments through increased brain levels of amyloid beta species, which is implicated in cognitive decline associated with Alzheimer’s disease (Berry, Tomidokoro, Ghiso, & Thornton, 2008; Burnham, Sundby, Laman-Maharg, & Thornton, 2017). Similar to estrogen, positive effects of testosterone on mood and cognition have been reported. Testosterone has been found to alter neural activity that is essential for learning and memory processes, and may also serve as a neuroprotective agent throughout aging (Janowsky, 2006). Additionally, testosterone supplementation improves spatial, verbal, and working memory in healthy older men (Cherrier et al., 2001; Resnick et al., 2017). Testosterone deficiency has been tied to depressed mood, as well as overall cognitive and functional impairment (Rosario et al., 2013; A. K. Wagner et al., 2012; Wang et al., 2004). Finally, Progesterone has been identified as critical component of normal brain function, and has also shown to protect against damage to the brain after TBI. Pre-clinical studies have shown treatment with progesterone to attenuate many of the negative consequences of TBI in young and aged rats (Wright, Hoffman, Virmani, & Stein, 2008). In spite, of positive preclinical findings, as well as a couple of positive early-phase TBI trials, phase III clinical trials have failed to show successful treatment of TBI with progesterone (Howard, Sayeed, & Stein, 2017; Wright et al., 2007; Xiao, Wei, Yan, Wang, & Lu, 2008).
GH and IGF
Deficiencies in GH have been associated with a number of cognitive and psychological sequelae, including deficits in memory, attention, information processing speed and diminished quality of life (Nyberg & Hallberg, 2013). While recognized primarily for its role in somatic growth and metabolism, animal and human studies have shown that GH is critical for normal brain development and cognition (Aberg, 2010; Dehkhoda, Lee, Medina, & Brooks, 2018). Rodent studies have shown that GH deficiency during early-life leads to spatial memory impairment in mid-life; this effect can be remediated by GH supplementation (Nieves-Martinez et al., 2010). Cognitive deficits have been seen in both children and adults with GH deficiency, presenting as deficits in IQ, verbal comprehension, processing speed, motor assessment, memory, emotion, and motivation (Rosen, Wiren, Wilhelmsen, Wiklund, & Bengtsson, 1994). GH deficiency following TBI has also been associated with depression and disability (Kreber et al., 2016). Another study observed that TBI patients presenting GH deficiency showed more impairments in attention, memory and executive function compared to non-deficient TBI patients (Leon-Carrion et al., 2007). Decreased fatigue and quality of life improvements have been associated with GH supplementation (McGauley, Cuneo, Salomon, & Sonksen, 1996; Mossberg et al., 2017). GH facilitates the production of IGF-1, which has been shown to mediate neuroplasticity mechanisms (Llorens-Martin, Torres-Aleman, & Trejo, 2009). Both, GH and IGF-1 have been clearly associated with modification at the receptor level. Pre-clinical studies have shown a clear association between GH, IGF-1, and N-methyl-D-aspartate (NMDA) receptor modulation, particularly in the hippocampus (Molina, Ariwodola, Weiner, Brunso-Bechtold, & Adams, 2013). GH and IGF-1 have also been shown to stimulate proliferation and differentiation of cells and promote myelination and neuronal arborization (Aberg, 2010; Arce, Devesa, & Devesa, 2013; Nieto-Estevez, Defterali, & Vicario-Abejon, 2016).
Conclusion
Neuroendocrine and sleep disturbances are both prevalent after TBI. Disruptions in one or both of these systems will impact recovery from TBI; particularly when the bi-directional relationship between sleep and hormones is considered. It is crucial to consider sleep and hormonal dysregulation as related problems that may ultimately affect cognitive and functional outcome. These disturbances should be properly evaluated after TBI. Given the plurality of posttraumatic disturbances it is recommended to objectively evaluate sleep. Proper treatment of these alterations is particularly critical as a TBI patient ages. Both endocrine function and sleep are affected by age, and TBI is likely to contribute to age related disturbances (Griesbach, Masel, Helvie, & Ashley, 2018).
Using data gathered from both preclinical and clinical sleep and neuroendocrine research, it is possible to improve the response to rehabilitation after TBI. Altered sleep architecture and hormonal imbalances are likely to contribute to cognitive issues, thereby diminishing responsiveness to the rehabilitative process. Due to the frequency of TBI patients being on multiple medications, whose side effects often include sleep-wake issues, the idea of non-pharmacological intervention is appealing. One particular intervention of interest is exercise due to its regulatory effects on both sleep and endocrine function (Chennaoui, Arnal, Sauvet, & Leger, 2015; Hackney & Lane, 2015). Exercise has been presented as an alternative form of treatment for insomnia (Chennaoui et al., 2015; Passos et al., 2014), which can, in turn, improve cognitive performance and emotional well-being by improving sleep quality. Additionally, exercise has been associated with improvements in health associated with endocrine system functioning (Ball, 2015; Cotman, Berchtold, & Christie, 2007; Stokes, Gilbert, Hall, Andrews, & Thompson, 2013). Exercise may also exert different effects at different post-injury times (Kreber & Griesbach, 2016). It is likely that the duration of sleep and hormonal disturbances is dependent upon multiple factors, including differences in the injury itself, along with individual genetic and environmental influences. TBI-induced hormonal alterations and sleep disturbances are likely to have a self-propagating effect. It is necessary to view TBI as a chronic disease in which pathophysiological processes will interact with sleep and neuroendocrine function.
Conflict of interest
None to report.
