Systematic Review of Sleep Disturbances and Circadian Sleep Desynchronization in Autism Spectrum Disorder: Toward an Integrative Model of a Self-Reinforcing Loop

Background: A compelling number of studies, conducted in both children and adults, have reported an association between sleep disturbances/circadian sleep alterations and autism spectrum disorder (ASD); however, the data are sparse and the nature of this link is still unclear. The present review aimed to systematically collect the literature data relevant on sleep disturbances and circadian sleep dysrhythmicity related to ASD across all ages and to provide an integrative theoretical framework of their association. Methods: A systematic review of the MEDLINE, PubMed, and Cochrane databases was conducted from November 2018 to February 2019. The search strategies used were MeSH headings and keywords for “sleep–wake circadian rhythms” OR “circadian sleep disorders” OR “sleep–wake pattern” OR “sleep disorders” OR “melatonin” AND “autism spectrum disorder” OR “autism”. Results: One hundred and three studies were identified, 15 regarded circadian sleep dysrhythmicity, 74 regarded sleep disturbances, and 17 regarded melatonin alterations in children and adults with ASD. Our findings suggested that autistic subjects frequently present sleep disturbances in particular short sleep duration, low sleep quality/efficiency, and circadian sleep desynchronization such as delayed phases and/or eveningness. Sleep disturbances and circadian sleep alterations have been related to the severity of autistic symptoms. Genetic studies have shown polymorphisms in circadian CLOCK genes and in genes involved in melatonin pathways in subjects with ASD. Conclusions: Sleep disturbances and circadian sleep alterations are frequent in subjects with autistic symptoms. These subjects have shown polymorphisms in clock genes expression and in genes involved in melatonin production. The impairment of circadian sleep regulation may increase the individual’s vulnerability to develop symptoms of ASD by altering the sleep regulation in toto, which plays a key role in normal brain development. Even though controversies and “research gaps” are present in literature at this point, we may hypothesize a bidirectional relation between circadian sleep dysfunction and ASD. In particular, circadian sleep dysrhythmicity may predispose to develop ASD symptoms and vice versa within a self-reinforcing feedback loop. By targeting sleep disturbances and circadian sleep dysrhythmicity, we may improve treatment strategies for both children and adults with ASD.

Background: A compelling number of studies, conducted in both children and adults, have reported an association between sleep disturbances/circadian sleep alterations and autism spectrum disorder (ASD); however, the data are sparse and the nature of this link is still unclear. The present review aimed to systematically collect the literature data relevant on sleep disturbances and circadian sleep dysrhythmicity related to ASD across all ages and to provide an integrative theoretical framework of their association.
Methods: A systematic review of the MEDLINE, PubMed, and Cochrane databases was conducted from November 2018 to February 2019. The search strategies used were MeSH headings and keywords for "sleep-wake circadian rhythms" OR "circadian sleep disorders" OR "sleep-wake pattern" OR "sleep disorders" OR "melatonin" AND "autism spectrum disorder" OR "autism".
Results: One hundred and three studies were identified, 15 regarded circadian sleep dysrhythmicity, 74 regarded sleep disturbances, and 17 regarded melatonin alterations in children and adults with ASD. Our findings suggested that autistic subjects frequently present sleep disturbances in particular short sleep duration, low sleep quality/efficiency, and circadian sleep desynchronization such as delayed phases and/or eveningness. Sleep disturbances and circadian sleep alterations have been related to the severity of autistic symptoms. Genetic studies have shown polymorphisms in circadian CLOCK genes and in genes involved in melatonin pathways in subjects with ASD.
Conclusions: Sleep disturbances and circadian sleep alterations are frequent in subjects with autistic symptoms. These subjects have shown polymorphisms in clock genes expression and in genes involved in melatonin production. The impairment of circadian sleep regulation may increase the individual's vulnerability to develop symptoms of ASD by altering the sleep regulation in toto, which plays a key role in normal brain development.

INTRODUCTION
Autism spectrum disorder (ASD) is an early-onset neurodevelopmental disorder whose core features have been defined by the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5) (1). It is characterized as persistent difficulties in social interaction and communication, presence of stereotypic behaviors, restricted interests, and atypical sensory reactivity. ASD, in fact, encompasses a set of clinical phenotypes that includes previously described autistic disorder (AD) and Asperger syndrome (AS), from severe to mild variants as endpoints of a continuum upon a spectrum model approach (2,3). Intellectual disability is observed in more than half of ASD cases (4,5) and ASD autistic symptoms also affect social language skills and emotion regulation. In addition, ASD symptoms are often related to coexisting mental disorders and to other developmental disorders (6).
Within the last decades, the diagnosis rate of autism has increased dramatically, and it has been reported that cases of ASD have a rate of 0.6-0.8% in preschool children, 1.0% in school children and young adults, and around 1.0% in adults (7,8).
In the last few years, an increased interest has been developed for mild forms of ASD, which often remain undiagnosed or misdiagnosed until adulthood. Although ASD is defined as a developmental disorder because symptoms appear within the first 2 years of life, it is generally considered a lifelong disorder with negative consequences on scholastic, working, social, and economic performances and quality of life. Some studies have shown that mild forms of autism are related to high rates of psychiatric comorbidity in adulthood such as anxiety, mood disorders, psychosis, stress-related disorders, and suicidal behaviors (9,10).
In this framework, understanding the mechanisms involved in the development of ASD should be a priority for identifying early markers that could help improve early diagnosis with a significant impact on lifelong prognosis (3). The mechanisms underlying the development of ASD may be the result of the interaction between multiple gene arrangement and the environment that may lead to the alteration of brain structures and functions (11,12). This interaction may determine epigenetic alterations disrupting the regulation of gene expression with a negative impact on biological pathways relevant for brain development (13). Abnormalities during brain development in autistic subjects go beyond the "social brain" encompassing sensory processing and attentional control. In fact, current evidence strongly supports a model of brain-wide abnormalities during the early development of autistic children (12). Indeed, it remains an open question if a single mechanism or several independent factors can lead to the emergence of autism.
In the last few years, a new hypothesis has emerged, suggesting the role of circadian system desynchronization in the development of ASD (11,14). Human physiological and biochemical processes as well as behavioral patterns have a circadian rhythmicity orchestrated by the master biological clock of the hypothalamus: the suprachiasmatic nuclei (SCN). The expression of many genes changes rhythmically over 24 h and the specific circadian genes are responsible for the main SCN clock-working machinery as well as that of subsidiary clocks at the peripheral level [among them: circadian locomotor output cycles kaput (CLOCK), brain and muscle ARNT-like protein 1 (BMAL1), cryptochromes CRY1-2, and band period homolog (PER)] [ (15,16); for an overview see Ref. (17)]. The suprachiasmatic nuclei is daily synchronized by environmental signals such as light, food intake, activities, or social cues and exposure to stress/trauma (18)(19)(20), and while driving, secretion of the melatonin hormone regulates peripheral clock within feed-forward mechanism. Rhythmic clock gene expression regulates multiple monoaminergic brain regions that control mood and motivational behaviors, stress and inflammatory systems, reward circuits, arousal, and sleep by interacting with the homeostatic regulation of sleep and wake [for an overview, see Refs. (17,21)]. The circadian system is critical for the synchronization with the environment and allows a correct functioning of various internal physiological processes essential for the optimization of responses to environmental fluctuations and for the strengthening of homeostatic control mechanisms (21). Abnormalities in the maturation of the circadian system principally lead to alterations in the sleep-wake pattern (22), which may interest around 50-80% of subjects with ADs (23). Wimpory and colleagues (24) have hypothesized that timing and social timing deficits were relevant in subjects with AD symptoms and were related to pathological variations in the structure/ function of clock/clock-related genes: the authors hypothesized a key role for circadian sleep dysregulation in autistic symptoms. This hypothesis has been confirmed by later studies, which have shown circadian-relevant gene abnormalities in ADs (25,26). Mutations in Clock, Bmal1, Cry1, and Cry2 genes determine an alteration in the circadian system regulation as well as in sleep fragmentation (27,28). Sleep is increasingly recognized as a key process in neurodevelopment and in brain optimization processes [for an overview, see Ref. (29)]. Both humans and animals' data have shown that sleep is essential for maturation of fundamental brain functions, and epidemiological findings increasingly indicate that children with early sleep disturbances suffer from later cognitive, attentional, and psychiatric problems. Indeed, from birth throughout infancy and early childhood, sleep patterns undergo dramatic changes that include the gradual consolidation of sleep and waking cycles, the intensification of deep NREM sleep slow-wave activity (EEG power in the 1-4.5 Hz frequency range), and a progressive decrease of REM sleep proportion. It has been suggested that REM sleep is an inducer of brain development and of early myelination in the sensory processing areas in the fetus and the newborns and that it follows the maturational trajectory of the brain (29)(30)(31)(32)(33)(34). Animal studies have shown that REM sleep has multifaceted functions in brain development, including learning and memory consolidation by selectively eliminating and maintaining newly formed synapses (35). On the other hand, slowwave sleep and sleep spindles (10-14 Hz) seem to be even involved in synaptic remodeling, being important for synaptic strength and synchronized neuronal firing, and it has been shown that while following the trajectory of brain maturation, they may orchestrate synaptic plasticity and pruning during brain development (30,(36)(37)(38)(39)(40). Particularly, Ringli and Huber hypothesized that slowwave sleep may contribute to cortical maturation by playing a role in the balance of brain synaptic strengthening/formation weakening/elimination that is tilted during development (39). Sleep promotes myelination and oligodendrocyte precursor cell proliferation (41), enhances transcription of genes involved in synthesis and maintenance of membranes and myelin too (42), and modulates the neuronal membrane homeostasis (43). Since adequate sleep has been proposed to be fundamental for brain development (31,35,39), sleep has received considerable research attention, as it appears to be important in the study of neurodevelopmental psychopathology (31,39,(44)(45)(46). Hence, if sleep is fundamental for brain development, we may hypothesize that sleep disturbances, via alterations in brain development, may contribute to autistic symptoms. This idea has been previously developed for other psychiatric disorders. In fact, extensive data have shown that poor sleep during childhood and adolescence is related to alterations in brain development (39,44,(47)(48)(49)(50)(51)(52)(53) to problems in cognitive, attentional, emotional, and behavioral areas, including risk-taking and aggression; and to psychiatric conditions such as attention deficit hyperactivity disorder and mood disorders (54). Indeed, we may also hypothesize that alteration in brain development related to ASD may contribute to sleep disturbances within a self-reinforcing loop.

OBJECTIVES
On this basis, this review was aimed to systematically collect the literature data relevant on sleep disturbances and circadian sleep dysrhythmicity related to ASD across all ages; a comprehensive review that has been conducted recently was limited to youth (55). In particular, we aimed to construct an integrative theoretical framework of their association: a comprehensive framework would be quite useful from a clinical and therapeutic point of view for identifying elements to evaluate and target in the clinical practice.
The hypothesis of this review was that abnormal maturation of the circadian sleep system may lead to the disruption of the sleep system in toto that, by impairing brain neurodevelopment and melatonin production, may have a key role in ASD. The investigation of such elements may be helpful to a better understanding of the neurodevelopmental pathways of ASD. Hence, the aim of this review was to systematically collect the literature data on sleep and circadian disturbances in children and adults with ASD.

Search Strategy
We performed a systematic review, based on Preferred Reporting Items for Systematic Review and Meta-analyses (PRISMA) guidelines, of articles published up till February 2019, and indexed in the following databases: MEDLINE, PubMed, and Cochrane Library. The search strategies used MeSH headings and keywords for "sleep-wake circadian rhythms" OR "circadian sleep disorders" OR "sleep-wake pattern" OR "sleep disorders" OR "melatonin" AND "autism spectrum disorder" OR "autism. " Inclusion criteria: The searches were limited to the English language studies conducted on human populations, including longitudinal, cross-sectional, or case-control studies, analyzing the relationship between sleep disturbances and/or circadian sleep disorders and ASD.
Exclusion criteria: Studies that did not have as principal focus the evaluation of sleep disturbances and/or circadian sleep rhythms related to autism were excluded. We also excluded studies that investigated subjects with other neurodevelopmental disorders or intellectual disabilities.

Data Extraction
The abstracts located from the search strategy were entered into EndNote. The searches were limited to the English language studies conducted on human populations, including longitudinal, cross-sectional, or case-control studies, analyzing the relationship between sleep disturbances and/or circadian sleep disorders and ASD. Title and abstracts of all non-duplicated papers were independently screened by two of the authors (DC and IM). Potential pertinent papers were retained and assessed for eligibility by screening the full text. Two senior authors (LP and CC) acted as arbitrators when there is disagreement in any screening stage. All the authors independently evaluated the quality of the included studies. Disagreements between authors were resolved by consensus. Initial literature search returned 1,443 records (among which 7 studies were searched manually from other sources), 652 after exclusion of duplicates. Following preliminary screening of the titles and the exclusion of reviews and case reports, 492 of the retrieved articles were excluded; further 57 records were excluded after reading abstracts and full texts based on the relevance of data for the aim of the present study (Figure 1).

Data Analysis
Meta-analysis was not conducted due to heterogeneity in definition and measurement of outcomes.

RESULTS
The systematic review of literature returned 103 articles focused on the sleep-wake pattern in neurodevelopmental disorders on children and an adult population according to the evolution of nosography; these studies include both previously defined diagnoses (56), such as AD, AS, childhood disintegrative disorder (CDD), pervasive developmental disorder (PDD), and not otherwise specified (NAS) (2), and the most recent diagnosis of ASD (1).

Circadian Sleep Dysrhythmicity in Children With ASD
Seven studies have evaluated the alterations of circadian rhythms in children with ASD ( Table 1). The first study (57) reported no tendency for irregular sleep-wake pattern, delayed sleep phase syndrome, or advanced sleep phase syndrome in children with ASD. Similar results were found in another study and in addition, no circadian phase preference was detected (61). On the other hand, several studies have shown that children with ASD may

Sleep Disturbances in Children With Autism Spectrum Disorder
Sixty-five studies have been evaluated on the sleep pattern of children with ASD ( Table 2), and some of them reported a prevalence rate of sleep disturbances between 64% and 93% (65,71,95,108,109,114,125) underlining a great prevalence of sleep disturbances during early life (62,63,65,86,109,125) with no gender differences (62,89). In particular, total sleep time has shown to be reduced starting from 30 months of age toward adolescence (98,104,107). Researches based on parental reports and objective measurements (actigraphy and polysomnography) have suggested that autistic children are more likely to have sleep difficulties than children with other neurodevelopment disorder (69,70,74,75,78,80,83,90,107) and children with normal development (62,64,69,80,85,90,91,120). Among autistic subjects, some studies have reported that children with intellectual disabilities and/or with low levels of functioning may experience higher frequency of sleep disturbances (58,62), while other studies have reported that high-functioning autistics may experience more sleep problems (shorter sleep duration, early night awakenings, and longer sleep latencies) than low-functioning children and control groups (63,64). Some studies conducted in adolescents have also reported that those with high-functioning autism were three times more likely to have sleep problems than their peers with normal development (102). The most frequent sleep pattern observed across all the studies included the following: difficulties falling asleep (63-65, 72, 78, 89, 91, 109), restless sleep (64,72), difficulties falling asleep in own beds with late bedtime (58,104,119), Parent-reported difficulties initiating sleep and daytime sleepiness were more common in children with AS/HFA than in controls.
Children with insomnia showed more parentreported autistic and emotional symptoms, and more teacher-reported emotional and hyperactivity symptoms than those without insomnia.
Small sample size; use of medication.
Krakowiak et al. Both autism and ID groups showed poorer daytime performance and behaviors than the TD one. These significant differences persisted over 6 months; long night awakenings and lower sleep efficiency predicted daytime sleepiness in the ID group; parent-report sleep problems were associated with daytime sleepiness and behavior problems. In the ASD group, the prevalence of chronic insomnia was more than 10 times higher compared to controls. ASD was a strong predictor of sleep problems and emotional and behavioral problems explained a large proportion of this association. CSHQ. Sleep problems were significantly associated with physical aggression, irritability, inattention, and hyperactivity. Night awakenings had the most consistently strong association with daytime behavior problems, even after controlling for the effects of age and sex.
No evaluation of sleep parameters by standardized measure.

CSHQ.
There was no significant association between GI problems and autism severity, but a significant positive correlation have been found between different indicators of sleep disorders and severity of autism. Plasma levels of serotonin were significantly high in autistic children, but no significant association with sleep problems.
Small sample size; no evaluation of sleep parameters by standardized measure.
Sleep habits. High scores on the ABC-C (irritability, stereotypy, and hyperactivity subscales) at admission were significantly associated with fewer minutes slept during the last five nights of hospitalization. There was no association between total awakenings and ABC-C scores or ADOS-2 comparison scores.
No evaluation of sleep parameters by standardized measures; absence of a control group.   (73) found a correlation between insomnia and the prevalence of communication problems due to an increased sensitivity to stimuli in the sleeping environment. Further studies (101,111) found a correlation between typical autistic behaviors and chronic insomnia. Children with ASD and poor sleep were characterized by a major representation of symptoms of autism such as restricted/ repetitive behaviors (91,113), language impairment, problems with reciprocal social interactions, low overall intelligence (80,100,103,109), and even physical aggression, irritability, and oppositional defiant disorder (83,84,116,122). Affective disturbances, particularly depression (80) and anxiety-related problems (71,95,103,105,109), have even been described in ASD with poor sleep. Some research groups (97,100,106) have analyzed the relation between insomnia, psychopathology, and daytime functioning in both children and adolescents with autism and reported a severity of depressed mood, anxiety, and daytime functioning in relation to sleep disturbances. Rzepecka and colleagues (95) reported a significant correlation between sleep problems, challenging behavior, and anxiety in children with intellectual disability and/or ASD. This study indicated that the 41.9% of the variance of challenging behavior was predicted by sleep problems and anxiety. The first study (110) that has analyzed the relationship between anxiety, sensory problems, and sleep disturbances in autism showed that children with anxiety and sensory overresponsivity may be particularly predisposed to develop sleep problems. A recent study of Veatch et al. (123) has demonstrated that shorter sleep duration was associated not only with depressive and obsessive-compulsive disorders but also with social impairment such as a failure to develop peer relationships and with maladaptive behaviors in ASD. Children who reported 7 h of sleep per night showed a severe impairment in social/ communication compared to ASD children who tended to sleep 10-11 h per night (123). Anxiety, autism symptom severity, sensory sensitivities, snoring, obstructive sleep apneas, pain, and gastrointestinal problems were also associated with sleep disturbances (103,108,112,118). In summary, sleep disturbances are frequent in children and adolescents with ASD. Insomnia and sleepiness were the most commonly reported symptoms together with long sleep latency, long time spent awake after sleep onset, early time awakening, and short sleep time. As shown in several studies, sleep disturbances have been related to the severity of ASD symptoms such as difficulties in social interaction and communication, presence of stereotypic behaviors, as well as anxiety and depressive symptoms.

Circadian Sleep Dysrhythmicity in Adults With Autism Spectrum Disorder
Five studies analyzed the circadian rhythmicity in adults with ASD ( Table 3). The most frequent circadian sleep disorders was the delayed sleep phase (126,(128)(129)(130). The most recent study (130) has investigated two groups (41 adults with ASD and intellectual disability versus 51 normal development) by means of ambulatory circadian monitoring, recording temperature, motor activity, body position, sleep, and light intensity; the circadian phase advanced was more common in the ASD group compared to controls. In summary, despite the fact that a few studies have investigated circadian sleep dysrhythmicity in adults with ASD, all the results showed that a delayed sleep phase is frequent in adults with ASD.

Sleep Disturbances in Adults With Autism Spectrum Disorder
To date, only nine studies have described the sleep pattern on adults with ASD ( AS adults reported frequent insomnia in all measures.
Small sample size.
2-nights PSG. Results of the second night recordings were included in the analysis.
AS subjects displayed a similar PSG profile compared with controls. Sleep periods were equal in both groups with a great amount of slow-wave sleep in the early part of the night. The only sign indicating decreased sleep continuity in autistic subjects was the greater proportion of wake after sleep onset.
Small sample size.
Clinical diagnosis. Actigraphy. People with AS did not differ from the controls regarding actigraphic sleep profiles.
Small sample size.

Limoges et al. (134) Canada
Cross-sectional study 27 adults with HFA (16-27 years); 78 TD subjects (16-30 years). Although the sleep questionnaires completed by parents revealed only a moderate degree of sleep problems, great sleep disturbances were recorded with actigraphy. Low sleep efficiency (below 85%) or long sleep latency (more than 30 min) have been found in 80% of the subjects. There was no early morning awakening.

DSM-IV
Small sample size; absence of a control group.

Limoges et al. (136) Canada
Case-control study 17 adults with ASD (9 with HFA and 8 with AS); 14 TD individuals.
DSM-IV diagnostic criteria. ADI-R; a battery of nonverbal tasks was administered, in the morning after a second night of sleep in the laboratory.

PSG.
Signs of poor sleep in the autism group were significantly correlated with either normal performance (selective attention and declarative memory) or inferior performance than controls (sensory-motor and cognitive procedural memories). Both groups presented a significant negative correlation between slow-wave sleep and learning a sensory-motor procedural memory task.
Small sample size; large number of correlation.
Richdale et al. HFASD group reported significantly more general sleep disturbances, high scores on the PSQI, long sleep onset latencies (actigraphy), and poor sleep efficiency (diary) and these results remained significant after accounting for the false discovery rate. HFASD group reported significantly shorter total sleep time, poorer refreshment scores upon waking in the morning and higher scores on the daytime dysfunction due to sleepiness subscale of the PSQI compared to the TD group.
Small sample size; use of medications. motor output on nonverbal performance tasks, suggesting that sleep disturbances in ASD might affect attention and/or memory components in the nonverbal modality (136). In summary, the main sleep problems in adults with ASD were insomnia, low sleep efficiency, short sleep duration, long sleep latency, frequent nighttime awakenings, and daytime sleepiness. As shown in some studies, sleep disturbances were related to difficulties in social interaction, communication, and cognitive performance.

Melatonin and Autism Spectrum Disorder
According to our research methods, we found 17 studies on the role of melatonin in circadian rhythm dysregulation in ASD ( Table 5). Two different studies (138, 139) measured melatonin serum levels every 4 h for 24 h in autistic subjects compared to healthy controls. They found an abnormal melatonin circadian rhythm, low blood concentration during the night related to sleep disturbances, and a tendency to have high daytime melatonin levels in autistic subjects. A later study by Tordjman and colleagues (140) analyzes the overnight urinary excretion of the predominant melatonin metabolite, 6-sulfatoxymelatonin (6-SM). Sixty-three percent of post-pubertal subjects with ASD showed low 6-SM, most marked in males and prepubertal children; this finding was confirmed in a later study (145). Another research group (146) confirmed a reduction in urinary secretion of 6-SM in subjects with ASD. The previous results were confirmed in another study, and low nocturnal excretion of 6-SM was associated with a greater severity of autistic symptoms (verbal language and repetitive behaviors). In a study of Pagan et al., (150), patients with ASD showed higher serotonin and lower melatonin levels than healthy controls. Patients with melatonin deficit reported more frequent sleep-onset and sleep-maintenance insomnia than patients with normal melatonin levels. The same research group investigated, for the first time, melatonin synthesis in the pineal gland and in the gut of patients with ASD, reinforcing the hypothesis that the nocturnal increase in circulating melatonin was reduced in those subjects and caused by enzymatic disruption of both aralkylamine N-acetyltransferase (AANAT) and acetylserotonin O-methyltransferase (ASMT) involved in melatonin synthesis confirming a melatonin reduction (153). An Afghanistan study (152) confirmed a reduction of melatonin serum level that was related to the severity of ASD symptoms. Recently, melatonin salivary concentrations have been studied in ASD subjects with and without anxiety and/or depression compared to controls (154). The timing of the dim light melatonin onset (DLMO) did not differ between the two groups, but advances and delays of the melatonin rhythms were the circadian sleep disorder more frequently observed in ASD subjects. Regarding the genetic pathways of melatonin, several studies focused on the variations of genes that regulate the synthesis, metabolism, and mechanism of action of melatonin. A multicenter study, the Paris Autism Research International (Sib-pair study) (141), sequenced all ASMT exons and promoters in 250 individuals with ASD. The ASD group showed non-conservative variations of the protein sequence of ASMT, and only two ASD families presented a splicing mutation. Forty-three individuals with ASD revealed a highly significant decrease of ASMT activity and melatonin level compared to 48 controls. Another subsequent research from the Sib-pair study (142) analyzed 941 individuals: 295 patients with ASD, 362 controls, and 284 individuals from different ethnic backgrounds. They sequenced MTNR1A and MTNR1B (melatonin receptor 1A and 1B) genes coding for melatonin 1 (MT1) and melatonin 2 (MT2) receptors, and G protein-coupled receptor 50 (GPR50) gene, coding for the orphan melatoninrelated receptor GPR50. Authors found MTNR1A and MTNR1B non-synonymous mutations altering the functional properties of the human melatonin receptors. Regarding GPR50, they detected a significant association between ASD and two variations in affected males. In the same year, Sweden researchers (143) screened the genetic mutations of AANAT, ASMT, MTNR1A, MTNR1B, and GPR50, encoding both synthesis enzymes and the three main receptors of melatonin, in 109 patients with ASD and 188 controls. In this study, several rare variants were found in melatonin-related genes in patients with ASD, including the mutation in ASMT. A Chinese study (147) investigated all ASMT exons and the neighboring region in 398 individuals with ASD and 437 healthy controls. They detected new rare coding mutations of ASMT affecting the protein sequence only in six individuals of the ASD group, but the authors did not find significant differences of genotypic distribution and allele frequencies of the common SNPs in ASMT between groups. Only one study (148) investigated a large sample of 1,747 subjects by means of the analysis of the SNPs and the duplication of exons 2-8 in ASMT, in order to identify genes involved in autism psychopathology. The authors identified the association between one SNP in the last intron of ASMT and social interaction impairments in females, but they did not detect any relation with language impairment or restricted and repetitive behaviors. According to previous research, Veatch et al. (151) studied 15 ASD children with sleep disturbances, of whom 11 were treated with melatonin, and found that an association between sleep onset delay and dysfunctional variation in genes related to the melatonin pathway, especially with regard to cytochrome (CYP) 1A2. They also, observed a strong correlation between genotypes in ASMT and in CYP1A2, particularly in the subset of children who responded to treatment with supplemental melatonin. In summary, most of the studies that focused on autistic children have found abnormal melatonin circadian secretion and low blood and salivary melatonin concentrations during the night with a tendency to have high daytime melatonin levels. Regarding the genetic pathways of melatonin, several studies focused on the variations of genes that regulate the synthesis, metabolism, and mechanism of action of melatonin (AANAT, ASMT, MTNR1A, MTNR1B, and GPR50) showing several rare genetic variants of them. These findings have been related to sleep disturbances in ASD. Sequencing ASMT exons and promoters; biochemical analyses performed on blood platelets and/or cultured cells.

DISCUSSION
Non-conservative variations of ASMT including a splicing mutation present in two families with ASD, but not in controls. Two polymorphisms located in the promoter (rs4446909 and rs5989681) were more frequent in ASD compared to control and were associated with a decrease in ASMT transcripts in blood cell lines. Highly significant decrease in ASMT activity and melatonin level in individuals with ASD.
No evaluation of sleep parameters.

(Continued)
June 2019 | Volume 10 | Article 366 Frontiers in Psychiatry | www.frontiersin.org Several rare variants were identified in patients with ASD, including splice site mutation in ASMT (IVS5+2T > C). However, mutations were found in upstream regulatory regions in three of the genes investigated, ASMT, MTNR1A, and MTNR1B.
No evaluation of sleep parameters.
In the hyperserotonemic group, significant increases at trend level in urinary excretion of 5-HIAA and 5-HT and a significant decrease for 6-SM were found. The urinary 5-HIAA:5-HT ratio was similar in the normoserotonemic versus the hyperserotonemic groups.  (55,65,71,95,108,109,114,125). Difficulties in falling asleep, frequent nighttime awakenings, and short sleep duration were the sleep disturbances most frequently described in both children and adults with ASD (72,78,79,124,130). Accordingly, these results were also confirmed by means of actigraphic and polysomnographic registration, displaying specific irregular sleep-wake cycles, low sleep efficiency, long sleep latency, insomnia, daytime sleepiness (68,91,97,106,119,120,132,133,134,136,137), delayed circadian phases, and evening preference as chronotype associated with ASD. According to our searches, polymorphisms in CLOCK genes that regulate sleep may be associated with ASD (25,26,59,60). Evidence is not strong although there are four separate studies implicating the alteration in CLOCK gene expression in the dysregulation of circadian sleep rhythmicity in ASD (25,26,59,60). Melatonin dysregulation, which includes delay in melatonin peak, reduction in amplitude, and alteration in melatonin gene expression, may also contribute to circadian sleep desynchronization in ASD [for an overview, see Ref. (155)]. In some studies, blood, salivary, or urinary levels of melatonin or of its metabolites have been shown to be reduced in autistic subjects (138-140, 145, 150) and to be correlated directly or indirectly with severe autistic behaviors such as verbal communication as well as repetitive behaviors and daytime sleepiness (146,152). According to our results, abnormalities in melatonin-related genes (enzymes involved in melatonin synthesis, metabolism and⁄or melatonin receptor function) may be related to low melatonin levels in ASD or may lead to an altered response to melatonin in a proportion of individuals with ASD (141-143, 148, 153). Several studies pointed out the potential correlation between sleep disturbances and the severity of autistic symptoms above all repetitive behaviors and deficits in verbal communication and/or in social reciprocity (73,100,103,109,113,156,157). In summary the current data have shown several striking findings with regard to sleep/circadian sleep function and ASD, but the nature of the link remains unclear. We may hypothesize that polymorphisms in clock genes and alterations in melatonin pathways may contribute to the dysregulation of the circadian sleep rhythmicity and consequently to the dysregulation of sleep system in toto. According to the theories about the functions of sleep during development (30,(36)(37)(38)(39)(40), we may hypothesize that sleep disturbances may negatively influence brain maturation, contributing to autism symptoms. It has already been hypothesized that even a modest and initial impairment of circadian sleep rhythmicity may increase the individual's vulnerability to ASD [for an overview, see Ref. (11)]. Indeed, on the other hand, autistic symptoms may reinforce sleep disturbances, creating a self-reinforcing feedback loop ( Figure  2). This framework should be useful in order to identify elements to evaluate and target in the clinical practice. In particular, sleep disturbances and circadian sleep alterations may represent a novel therapeutic target in ASD. According to this model, by treating sleep and circadian sleep disorders in ASD, we should contribute to an improvement in ASD symptoms. There is strong evidence about the use of melatonin in ASD, indicating its beneficial effects on sleep and autistic symptoms. As suggested by some studies (158)(159)(160)(161)(162)(163)(164)(165), melatonin may improve sleep difficulties in ASD subjects even when administered in association with concurrent psychotropic medications (160) and/or cognitive behavior therapy (164). Treatment with melatonin seems to improve sleep disturbances in the majority of children and adults with autistic symptoms (166), showing effectiveness on sleep duration, sleep latency, and nocturnal and early morning awakenings (158, 159, 161-165, 167, 168). Most importantly, by treating sleep and circadian sleep disorders with melatonin, an improvement in typical autistic behaviors has been shown in some studies (158,163,168). Data in children with neurodevelopmental disorders failed to point to serious adverse events associated with the use of melatonin (160,162,153,165,169) except for morning sleepiness (165), headache, and fatigue (167). These findings shed light on a rising hypothesis of an integrative model that may explain the interference of circadian rhythms and its variables during critical periods of brain development in ASD (Figure 2). Evaluating and targeting sleep disturbances and circadian sleep disorders in ASD should be useful to improve the trajectory of ASD. Indeed, the current data have shown a "research gap" about this topic. Future research should focus on the systematic study of circadian sleep dysrhythmicity in children and adults with ASD in relation to clock gene expression and to melatonin production.

LIMITATIONS
Our findings should be considered in the light of several limitations. Primarily, although this review aimed to summarize the most relevant studies on the relation between circadian sleep disorders and ASD, the inclusion/exclusion of specific studies may reflect our individual point of view or expertise and training. Moreover, some studies included may have been underpowered (some had only small sample sizes and small numbers of subjects enrolled) and/or some of them did not include control groups. Another limitation was the heterogeneity of the studies regarding the characteristics of the sample such as age, scholarity, study design, diagnostic criteria and measures of assessment, severity of sleep, and autistic symptoms that are potentially very important and should be considered routinely in future studies.
Since the studies included in our research cover a wide temporal range (1984-2019), it is necessary to consider the nosography evolution of neurodevelopmental disorders over time, which could be a further study limitation. Psychiatric comorbidity and/or the use of psychopharmacological drugs may interfere with the duration and severity of both autistic symptoms and sleep disturbances. In addition, the potential roles of inert brain anomalies, the potential impact of brain injury during childbirth, and the potential influence of adverse environmental stressors in the early development of the child need to be addressed in further studies as being potential reasons for an apparent rise in the prevalence of ASD symptoms with later developmental stages.

CONCLUSION
Overall, results from this systematic review highlight the idea that sleep and circadian sleep disturbances are frequent in subjects with autistic symptoms who have shown polymorphisms in clock gene expression and in genes involved in melatonin production. The impairment of circadian sleep regulation may increase the individual's vulnerability to develop symptoms of ASD by impairing the sleep regulation in toto, which instead plays a key role in normal brain development. Even though controversies and "research gaps" are present in literature at this point, we may hypothesize a bidirectional relation between circadian sleep dysfunction and ASD. In particular, circadian sleep dysrhythmicity may predispose to develop ASD symptoms and vice versa within a self-reinforcing loop. An early identification and assessment of circadian sleep dysrhythmicity could be useful for improving treatment strategies in both children and adults with ASD.