1. Introduction
Autism spectrum disorder (ASD) is a complex neurodevelopmental condition characterized by cognitive, behavioral, and social dysfunctions. Autistic children experience severe difficulties in social interactions and communication, accompanied by limited interests and stereotyped repetitive behavior. ASD onset occurs at early childhood and often results in severe lifelong impairments [
1], and thus is currently regarded as one of the most common childhood morbidities. Although the overall ASD prevalence is 23 per 1000 children aged eight years (one in 44) [
2] and constantly rising, the etiology of its development and underlying triggers are still unclear. ASD is considered a multifactorial disorder elicited by genetic [
3,
4] and environmental prenatal, perinatal, and postnatal factors [
5,
6,
7]. This large spectrum of putative triggers could derive from variations in methodologies (e.g., case definition, comparison groups, race and region, sample size, and exposure assessment) [
8]. Adding to the complexity is the fact that ASD is a heterogeneous neurodevelopmental disorder with a broad range of clinical presentations (e.g., delays in motor development, in speech and language; lack of eye contact; limited socializing ability) [
9].
Despite the need to identify a developing disorder during infancy, when brain plasticity enables manipulations and possible repair, ASD is mostly diagnosed in older children or in teenagers [
10]. Evidently, early detection of signs that may predict later development of ASD remain a challenge due to clinical and practical implications [
11], which improve prognoses [
12]. While parents of ASD children have often noticed developmental problems prior to their first birthday, the only approved screening tool that identifies infants at risk is M-CHAT-R/F, available under the age of two years [
13]. Barbaro and colleagues [
14] have shown that the Social Attention and Communication Surveillance-Revised (SACS-R) and SACS-Preschool (SACS-PR) tools could be employed with high diagnostic accuracy for the identification of autism in a community-based sample of infants, toddlers, and preschoolers, indicating the utility of early autism developmental surveillance from infancy to the preschool period. Numerous reviews of early detection screening tools have, however, indicated equivocal degrees of effectiveness as primary early developmental variables have not been clearly delineated [
15,
16,
17].
Although the core deficits of ASD involve social functioning [
18], retrospective studies indicate that ASD children exhibit disruption in other developmental domains during their first-year postpartum, including motor, attention, and temperament [
19]. Since retrospective studies are based on information derived from medical records, parent recall, and systematic observational recordings of home videotapes taken during the first or second year postpartum, and prior to a diagnosis of ASD [
20], prospective longitudinal studies might provide a better description of the early symptoms of a forthcoming ASD and timing of the appearance of frank symptoms. However, most prospective studies thus far have focused on infants at high risk or who have an older sibling diagnosed with ASD. These studies indicate that the development of motor, cognitive, language, and social domains were grossly intact at six months of age, yet, a slowing in development was observable when nearing their first birthday [
21,
22].
In a study on infants at risk for ASD, abnormal postural control was already documented at six months of age [
23], yet no single factor that could predict the development of ASD had been identified. These results prompted Wang et al. [
24] to examine the magnitude of prenatal, perinatal, and postnatal variables on ASD development in a meta-analysis. They collected data from 37,634 autistic children and 12,081,416 typically developing children (TDC) (control children) enrolled in 17 studies and suggested that during the prenatal period, the risk factors associated with ASD were maternal and paternal age (≥35 years), race, gestational hypertension, gestational diabetes, maternal and paternal education, threatened abortion, and antepartum hemorrhage. During the perinatal period, the factors that were putatively associated with ASD development were caesarian delivery, gestational age ≤ 36 weeks, parity ≥ 4, breech presentation, preeclampsia, and fetal distress. During the postnatal period, the risk factors were low birth weight, postpartum hemorrhage, male gender, and brain anomaly. Then, they examined all of these factors individually but were unable to conclude whether they were causal or secondary in the development of ASD.
In a study seeking risk factors in early infancy that could predict forthcoming neurobehavioral disorders (NBDs) [
25], analysis of the electronic health records (EHR) of 161 infants, who were later diagnosed with ASD, developmental coordination disorder (DCD) and attention deficit hyperactive disorder (ADHD), indicated deviations from trajectories on seven parameters (gestational age, birth weight, head circumference percentile, weight percentile, gross motor development, and difficulties in speech and communication). Collectively, these seven parameters have been suggested to predict forthcoming ASD with 85% probability. Most recently, Engelhard and colleagues [
26] demonstrated that EHR during the first 30 days of life identified 45.5% of the ASD children, while similar analysis during the first year of life identified 60% of the ASD children. On the basis of these results, the authors have suggested that such an automated approach integrated with caregiver surveys could improve the accuracy of early autism screening.
The well-organized infrastructure of the well-baby clinics in Israel was exploited to examine the files of 1105 ASD children. The aim of the study was to derive detailed information from the Big Data repository of Maccabi Health Organization (MHO) in Israel to statistically analyze and compare the data obtained with that from TDC from the same population.
4. Discussion
This retrospective cohort study reveals deviations in growth and developmental parameters in infancy that may contribute to the overall variance in developing predictive models for ASD. These parameters include: (A) Extreme upward deviations from trajectories of weight, height, and head circumference percentiles during the first three months of life and continuous excessive increase in weight and height up to twelve months (represented in
Table 2,
Table 3 and
Table 7), which may suggest metabolic and hormonal changes in the body and in the brain leading to brain inflammation and dysfunction in neuronal connectivities [
38]. (B) As typical motor development during the first few years of life involves organization of neural networks and connectivities [
34], atypical motor development may lead to the formation of alternative neuronal circuits and impede the centers associated with the development of language and speech, coordination, communication, and cognition [
39,
40].
4.1. Metabolic Complications May Lead to ASD Development
The upward deviations from weight trajectories during the first year of life in ASD children may be initiated as a result of genetic and environmental factors [
41]. Indeed, a possible link between parental obesity and development of NBDs in their offspring has already been suggested [
42,
43]. Moreover, deletions in chromosome 16p11.2 were implicated in obesity as well as in NBD [
44,
45]. Epigenetic obesity was proposed to be linked to DNA modifications (e.g., methylation), affecting the regulation of gene expression [
46]. A study of a nationally representative sample of school-aged children linked lesions in the non-dominant right hemisphere and the parietal lobe of children with excessive weight gain to abnormal brain functioning, cognitive impairment in visuo-spatial organization, and general mental ability [
47]. Furthermore, Dhanasekara and colleagues [
48] have shown in a recent meta-analysis that the risks for cardiometabolic diseases, such as diabetes, hypertension, dyslipidemia, and heart disease were high among children with autism.
The incidence of the “metabolic syndrome” has been shown to parallel the incidence of obesity [
49], which, according to the CDC, has an increasing prevalence and is estimated to be 19.7% among children and adolescents aged 2–19 years in 2017–2020 [
50]. Human and animal studies have shown that the maternal prenatal “metabolic syndrome” includes increased adiposity and insulin resistance, resulting in an inflammatory state [
51,
52] with a significant impact on fetal neurodevelopment [
53]. Inflammation resulting from a metabolic condition could be a putative mechanism that contributes to the overall variance in ASD [
54]. The rapid increase in HCp, following the rapid increase in weight (
Table 7), suggests a metabolic process rendering brain inflammation and edema that may be associated with ASD development. This suggestion is corroborated by the studies of Rossignol and Frye [
55], as well as Angelidou et al. [
56], who have suggested that perinatal stress that induces brain inflammatory responses and increased extra-axial cerebrospinal-fluid (EA-CSF) are associated with a higher risk of developing ASD. The putative connection between excessive weight and brain damage could involve increased body status, such as low-degree inflammation of blood vessels in the brain, since excessive weight was identified as a pro-inflammatory state [
57]. In addition, hyperinsulinemia was shown to be associated with impeded glucose metabolism and insulin signaling in several brain regions (e.g., frontal lobes, hippocampus) involved in planning and organizing [
58]. Another hormone involved in obesity as well as in ASD development is leptin, found to be elevated in the plasma of obese people [
59], and upon placental dysfunction during pregnancy [
60]. Indeed, significantly elevated levels of plasma leptin were reported in children with ASD compared to TDC [
61].
CSF plays a critical role during brain development and functions throughout life [
62] to deliver nutrients and growth factors supporting healthy neural growth and removal of neurotoxins and metabolic waste from neuronal functions [
63]. Therefore, in early development and throughout life, CSF production and absorption tend to be balanced. However, perturbations in CSF circulation may impair the clearance of harmful substances that accumulate in the brain and lead to neuro-inflammation [
64]. An example of such a condition is idiopathic intracranial hypertension (IIH), a disorder typified by elevated intracranial pressure with an estimated incidence of 15–19 cases per 100,000 among overweight or obese 20–44-year-old women [
65,
66]. Overall, excess weight is an important risk factor signifying the development of IIH among post-pubertal children or adults [
67]. Moreover, the high percentage of IIH among girls that gain weight rapidly at menarche may explain, in part, its sudden increase. Recently, increased EA-CSF, originating in infancy and observed through preschool age, found in three independent cohort studies of children with ASD, was proposed as a putative ASD marker [
68,
69,
70].
Courchesne et al. [
71] reported in a longitudinal study that 59% of the infants they studied had demonstrated accelerated HC growth trajectories (>2.0 SDs) from birth up to 14 months of age, whereas only 6% of the TDC infants followed these accelerated growth trajectories. In a recent prospective study using brain MRI scans of high-risk infants, Shen and co-workers concluded that increased EA-CSF at six months of age was a significant predictor for later developing ASD [
69]. Thus, early accelerated brain enlargement could be a contributing factor to ASD development. It is worth noting that upward deviations from weight and height growth trajectories of infants, who were later diagnosed for ASD, have been previously reported in a retrospective study [
25]. In an attempt to explain the link between growth dysregulation and ASD development, Green and colleagues [
30] have proposed two mechanisms: A) a connective tissue disorder, frequently associated with increased height and disproportionate body ratios and B) dysregulation of the hypothalamic-pituitary-adrenal (HPA) axis that regulates growth hormones. Notably, the clinical significance of larger brain volumes in ASD toddlers is still controversial. The present study reveals that, except for the initial rapid increase, no significant differences in HCp between the ASD and TDC groups could be observed after the first three months of life (
Table 7), suggesting that brain volume by itself is not a suitable predictor of ASD development, as also corroborated by a recent study [
72]. The present study supports the theory that ASD is associated more with dysregulated growth in general, rather than to merely a dysregulation of neuronal growth in the brain. Still, it remains unclear whether this non-regulated growth could serve as a biomarker of ASD development.
4.2. Atypical Motor Development May Be Associated with ASD
Motor difficulties are the principal reasons for toddlers’ visits to Child Developmental Centers, and so the first professional to meet the child is likely a physiotherapist. Surprisingly, however, no difference between the ASD and TDC groups was observed, implying that motor delay and referral to a physiotherapist cannot serve as a predictive factor specific for ASD. Instead, a dramatic increase in referrals for assessment and treatment of deformational plagiocephaly (DP), reported over the last two decades, might be related to the “Back to Sleep” guidelines as a preventive measure of Sudden Infant `death Syndrome (SIDS) [
73]. Although considered a medically benign condition, DP may be associated with developmental difficulties, suggesting that children with this condition should be routinely followed until school age for developmental delays or deficits [
74]. Numerous authors have theorized about the relation between sleeping position and ASD [
75,
76,
77]. Another prospective study on infants at risk has shown significantly poorer fine and gross motor scores at 36 months of age in children diagnosed at a later age for ASD [
78]. These results have suggested that early motor difficulties may predict a forthcoming NBD. A wide range of motor delays and deficits that constitute the core of s have also been reported in ASD and in ADHD individuals, and there is a link between early deviations from motor developmental trajectories and psychological outcomes in childhood [
25,
79]. Overall, although motor delay problems cannot serve as a single risk factor predicting forthcoming ASD, all babies may benefit from motor intervention for the prevention of asymmetry, plagiocephaly, torticollis and motor delay from the first month after birth.
4.3. Study Limitations
The study described here is a post hoc cohort study of data-mining that makes use of the outcomes of over 20 years of data-gathering from a database connected to one of the major HMOs in the State of Israel. The instant report does not cover all potentially influencing factors because they could not possibly have been identified at the outset. There were questions in the database about topics like vaginal vs. cesarian section delivery or breastfeeding vs. formula feeding. Although they were not accessible, potential confounding factors, such as more specific pregnancy information and racial or ethnic practice variations, will undoubtedly be the foundation for future research. Due to the fact that the MHO primarily served people from middle class backgrounds, concerns relating to poverty that were absent from the original data source are now incorporated. Another disadvantage is that variations in head circumference, weight, and height may be due to disparities between autistic and non-autistic people rather than problems that need to be rectified to “prevent” people from “becoming” autistic. Given the multifactorial nature of our current understanding of the etiology of ASD and the search for early markers, the small proportion of the variance still being significant can potentially contribute to the development of biomarkers to bring the age of identification lower and for interventions to commence earlier.
One possible approach to increase the variance is to divide the infants into three groups according to their birth weight (large (LGA), appropriate (AGA), and small (SGA) gestational age) and follow the changes in weight, height, and head circumference growth rates along the first year of life. The possibility that the LGA infant went through a rapid weight gain and increased HC during the last trimester should be raised. We could not tell from our study if the metabolic and motoric disturbances are primary or secondary to ASD, but they could be considered as two early predictive environmental factors. It is our intention to work with MHO big data sets in the near future to seek further early signs of predictive factors of ASD.