The brainstem and its rostral networks underlie a wide array of operations, ranging from autonomic functions such as respiration (Bianchi and Gestreau, 2009), cardiovascular activity (Dampney, 2016), and sleep-wake regulation (Scammell et al., 2017), through sensorimotor reactivity (Kobayashi and Isa, 2002), and even involvement in consciousness and self-awareness (Parvizi and Damasio, 2001).

Autism spectrum disorders (ASD) are a set of neurodevelopmental disorders manifested in deficits in social-communication abilities and restrictive and repetitive behaviors (American Psychiatric Association [APA], 2013). Despite the DSM nosology that classifies ASD as a unified construct, various findings suggest a high degree of heterogeneity in ASD phenomenological manifestation and genetic basis that nevertheless share common cellular and molecular features, including alterations in neurogenesis, synaptogenesis, and structural formation (Gilbert and Man, 2017). Recent accounts, some from our lab, emphasize the role of early brainstem functions in the epiphenomena of ASD (Dadalko and Travers, 2018; Delafield-Butt and Trevarthen, 2018); namely, in social attention (Geva et al., 2017), communication (Geva et al., 2013, 2014), and repetitive behaviors (Cohen et al., 2013; Gandhi and Lee, 2021)– all key features of ASD.

Social attention, communication, and adaptation of behavior have primary roles in the human core being already at birth. Hence, we suggest that brainstem circuits that mature in very early life and even during fetal stages to regulate vital autonomic functions (Zec and Kinney, 2003), have a central role in shaping social communication and adaptation of behavior. We suggest that given the early maturation of brainstem pathways, their pervasive role in functioning at multiple levels, and their specific involvement in social-communication deficits, brainstem functions enable a valuable window into the pathophysiology of ASD. Its neurobehavioral manifestations are already evident in the first phases of postnatal life.

Research thus far suggests brainstem involvement in ASD by typically denoting a unitary brainstem marker. Building upon an integration of the literature, in the current perspective, we propose a brainstem-informed autism framework (BIAF) that zooms in on the distinct paths by which compromised brainstem functions possibly stir development and increase the susceptibility for ASD-related symptomatology. We then suggest that zooming out to look at the full battery of brainstem-related expressions, rather than individual markers, may enable constructing a highly sensitive early risk neurobehavioral screening tool of ASD.

Principal morphological changes in the embryonic brainstem in multiple organisms buds during the first trimester of pregnancy (ten Donkelaar et al., 2014). Animal models indicate that the genesis of motoneurons in rhombomeres 7 and 8 commence approximately at the fourth week of fetal life; these neurons subsequently migrate and form the vagal nerve nuclei, including the dorsal motor nucleus, nucleus ambiguus, solitary tract nucleus, and spinal trigeminal nucleus (ten Donkelaar et al., 2014; Watson et al., 2019). The neural functionality of brainstem pathways is noted from early gestational stages (Glover et al., 2008; Marrs and Spirou, 2012). A post-mortem specimens study of the medulla in human fetuses indicated that a neural branching from and into the solitary tract nucleus is established and expedites cardiorespiratory control around gestational age (GA) of 20 weeks (Zec and Kinney, 2003). The development of vital parasympathetic functions is further secured from mid-gestation to parturition as myelination of the vagal nerve roots progresses (Tanaka et al., 1995). Importantly, myelination of efferent fibers from the nucleus ambiguus to the sinus nodes that regulate cardiac pace is accelerated (Porges, 2011) and stabilizes the parasympathetic activity when reaching term age as manifested by increased heart rate variability at the higher (i.e., above 0.2 Hz) frequencies (Longin et al., 2006). Similar neuro-maturational processes involving the birth of neurons in hindbrain rhombomeres and mesencephalic neuromeres, neurons migration, and axonal navigation contribute to the formation of the cranial nerves sensorimotor nuclei in the brainstem from Carnegie stage 12 (O’Rahilly and Müller, 2006; ten Donkelaar et al., 2014). These structures support auditory, ocular, tactile, gustatory, and olfaction development and shape motor reactivity in a progressive fashion. As such, early postnatal myelination of axons radically increases the rate and synchronicity of transmission through the auditory pathway, emanating from the cochlear nuclei, superior olive, lateral lemniscus, and inferior colliculus (Sano et al., 2007), alongside other sensorimotor paths that evolve in tight temporospatial constraints.

Optimal structuring of the brainstem has vast implications on neurocognitive sequelae, as the early structural building blocks of these early maturing networks influence the emerging operations of higher-order top-down limbic and neocortical systems. Eventually, brainstem networks affect functions from basic reception through multisensory and motor integration (Geva and Feldman, 2008), in ways that affect behavioral inhibition (Geva et al., 2014), higher-order social engagement (Geva et al., 2017), and social communication capacities (Geva et al., 2013). As such, evaluation of brainstem integrity offers multiple candidate markers for ASD. These markers are potentially diagnosable at term age and soon thereafter.

To date, the research has mostly treated each BIAF factor as a single primary marker. We review each one shortly and suggest that their integration presents a strong case for a cohesive BIAF. We shall focus on the hallmarks of brainstem functions: cardiac, respiratory, and arousal regulation; sleep-wake homeostasis; and primary sensorimotor operations. Following the exploration of their main effects and interactions, we will delineate a BIAF, focusing on how it first unfolds in gestation and the post-birth period.

The vertical hierarchical framework was formulated in our lab to delineate the development of self-regulation and positioned brainstem functions at the crux of the model (Geva and Feldman, 2008). According to this model, brainstem networks serve pivotal roles in regulating the young infant’s arousal responses to sensations. Recent notions accentuate that nascent autonomic conditioning of socioemotional reflexes and arousal responses occur at the brainstem and peripheral levels and prior to the top-down cortical navigation of arousal (Ludwig and Welch, 2020). Poor arousal regulation is one of the key features of ASD (Prince et al., 2017; Cuve et al., 2018; Corbett et al., 2019; de Vries et al., 2021), evident by hyper- or hypo-arousal reactivity of the autonomic system in response to sensory stimulation. Particularly in infants who are siblings of children with ASD (Zivan et al., 2021). The functional implications of autonomic dysfunctions in individuals with ASD are vividly apparent in an array of markers, including pupil diameter (Zivan et al., 2021), electrodermal activity (Prince et al., 2017), and the coordination of heart rate and breathing (Corbett et al., 2019). These autonomic functions are by and large regulated at first by brainstem networks that mature in late-term stages and have been suggested to serve social engagement purposes (Porges, 2001).

The inner workings of the mechanisms involved in this interplay are now under investigation using different models and an array of techniques. Here we note some of the literature that exemplifies the richness of data and intricate set of interlinked neurophysiological processes currently explored.

Possible causes for the abnormalities in autonomic functions in ASD are aberrations in cerebellar-brainstem white matter tracts, involving insufficient glial maturation and axonal growth differences noted in infancy and early childhood (Shukla et al., 2010; Yu et al., 2020), altered white matter connectivity of brainstem tracts found in tractography machine learning analysis (Zhang et al., 2018), and atypical structuring of the medullary arcuate nucleus which is involved in cardiorespiratory regulation (Bailey et al., 1998). The exact trajectory and localization of histogenesis and primal autonomic circuitries development in ASD remain to be further elucidated. Hopefully, future studies will clarify whether abnormal patterns of myelination, axonal navigation, and circuits formation of vagal nerve nuclei during embryonic and neonatal development are implicated in the autonomic sequelae of individuals with ASD. Even though the developmental pathophysiological course is not yet fully established, several cardiorespiratory indices were utilized to weigh the involvement of vagal functions in ASD research. We shall focus on respiratory sinus arrhythmia (RSA).

Respiratory sinus arrhythmia measures the variations in heart rate as a function of the respiration cycle and is regarded as an applicable index of the vagal tone and its coordination by the nucleus ambiguus (Berntson et al., 1993). Porges’ polyvagal theory proposes that inner physiological experiences are innervated by socio-emotional sensations right from birth and that this interplay underlies the nascent steps of social development (Porges, 1995, 2011, 2021).

A recent comprehensive meta-analysis (Cheng et al., 2020) involving participants with ages spanning the first three decades of life revealed that diagnosis of ASD was associated with diminished baseline RSA and diminished RSA reactivity during social experiences. A previous prospective study including a cohort of very preterm children has shown that neonatal RSA indices predicted social competence at the age of three (Doussard-Roosevelt et al., 1997) and then at school age (Doussard-Roosevelt et al., 2001). Further, infants diagnosed with ASD in late childhood showed a blunted pattern of RSA development from the age of 18 months (Sheinkopf et al., 2019). The RSA findings imply that the alignment of vagal resources with the social environment, mostly those involving the adaptive switching between tranquil/non-engaged and charged/engaged states, scaffolds the building blocks of social development from birth. It further accentuates that cardiovascular hypo- and hyper-arousal reactivity might affect vigilance and impede the prospect of a durable engagement with parents, peers, and significant others in ASD.

Vigilance models have been proposed to explain a range of psychopathological processes from mania to attention deficits (Hegerl and Hensch, 2014). These models have noted links between poor arousal regulation, unstable vigilance, and sleep deficits. We suggest that these notions are highly relevant to the BIAF.

Primal sleep-wake substrates in the brainstem promote sleep rhythms long before the anterior limbic circuits gain dominance (Villablanca et al., 2001). Given the primary involvement of brainstem networks in sleep regulation, the BIAF suggests that congenital compromised brainstem functions could instigate sleep-wake dysregulations from the neonatal period. Then, it might perturb the brainstem-limbo-cortical connectivity and lead to long-term sleep deficits (Geva and Feldman, 2010; Blumberg et al., 2014).

The primal sleep-wake system consists of wake-promoting loci in the reticular formation along the brainstem, including the monoaminergic locus coeruleus (LC) and dorsal raphe nucleus (DRN), and the cholinergic laterodorsal tegmental nucleus and parabrachial nuclei; the primal GABAergic sleep-promoting structures include the nucleus pontis oralis, nucleus subcoeruleus and the Purkinje cells in the cerebellum (Phillips and Robinson, 2007; Blumberg et al., 2014; Sokoloff et al., 2015). These brainstem-cerebellar hubs are highly implicated in the ultradian cyclicity of sleep-wake bouts during the first weeks of extrauterine life (Geva and Feldman, 2010). Infant sleep is marked by high rates of REM sleep that have a vital neuroprotective role and is guided by the aforementioned brainstem loci (Heraghty et al., 2008). Accordingly, lesions and morphological abnormalities in both gray and white matter in pontine and adjacent regions are associated with reduced REM sleep in human adults (Landau et al., 2005; Scherfler et al., 2011). Further, sleep organization and a smooth transition between sleep stages seem important. Fragmented neonatal sleep has been noted to impede infant attention orienting at 18 months of age (Geva et al., 2016). One should keep in mind these phenomena when considering the trajectories of sleep integrity in populations with ASD.

Children, adolescents, and adults with ASD display various sleep abnormalities, including decreased sleeping time, delayed sleep latency, and less efficient sleep (Elrod and Hood, 2015; Lugo et al., 2020; Chen et al., 2021), and show a strikingly elevated risk for sleep disorders (Lai et al., 2019; Lugo et al., 2020). These findings suggest that sleep deficits and ASD possibly stem from a similar neuropathological disrupted circuitry that involves the brainstem in a major way.

Studies have illustrated several influences of the brainstem in sleep-wake dysregulation in ASD populations. Vis-à-vis inhibitory pathways, genetic findings suggest that ASD is associated with microduplications copy-number variations in the chromosome 15q11–q13 region responsible for coding the GABA_A receptor’s subunits (Sebat et al., 2007; Meguro-Horike et al., 2011; Sanders et al., 2012). Accordingly, post-mortem studies found a decrease in GABAergic Purkinje neurons in cerebellum specimens of deceased ASD patients (Bailey et al., 1998; Palmen et al., 2004). Functional alterations in the activity of excitatory networks involving the LC are found in children with ASD (Bast et al., 2018; Huang et al., 2021). Taken together, these suggest that an imbalance between the primary excitatory (e.g., the monoaminergic LC and DRN) and inhibitory (e.g., the GABAergic Purkinje cells) circuits (Coghlan et al., 2012) of the brainstem is involved in the pathogenesis of sleep disturbances in ASD patients, as well as affecting other arousal-related domains (Wintler et al., 2020).

The extent of sleep deficits could also be viewed as a distinct factor that affects social development. Several findings support this latter notion. First, abnormal sleep patterns in early life are associated with subsequent ASD diagnosis and symptoms (Humphreys et al., 2014; Saenz et al., 2015; Miike et al., 2020). Further, in children with ASD, shorter sleep duration exacerbates the severity of both repetitive behaviors and social-communication deficits (Schreck et al., 2004; Tudor et al., 2012; Veatch et al., 2017). The hazards posed by sleep disruptions in children with ASD urged clinicians to recommend that sleep-wake homeostasis issues be assessed and managed as a central feature in the therapeutic plan of ASD patients (Cohen et al., 2014; Abel et al., 2017; Souders et al., 2017). This agenda accentuates the primary role sleep possibly serves in the evolution of ASD and the neuroprotective role of sleep in its containment (Wintler et al., 2020).

The findings suggest that sleep and arousal play a major role in ASD. Sleep disturbances are plausibly a result of fetal, genetic, and epigenetic brainstem-mediated antecedents. At the same time, they stand by themselves as factors that might exacerbate the risk for ASD or lead to more severe symptoms by impeding brain development and its regulated reactivity to stimulation through the various senses.

Atypical behavioral responses to sensory stimulation are a ubiquitous characteristic of ASD (Marco et al., 2011). Research on unimodal sensory processing and multisensory integration using various neuroimaging techniques demonstrated significant alterations in sensory processing neural substrates (Marco et al., 2011). Here we briefly review some of the unimodal and multisensory processing findings that pertain to the BIAF.

It has been shown that the ABR alone has a noteworthy sensitivity for detecting infants who later develop ASD, with 70% accuracy (Miron et al., 2016). We suggest that integrating the individual markers points to the importance of a cohesive early life BIAF. The advent of such a framework offers advancements in our multidimensional understanding of autism. First, the BIAF proposes that the susceptibility to ASD develops during gestation and marks the neural networks that account for its early presentations, those that precede DSM symptoms oftentimes. Secondly, given the richness of the pathways traversing through the brainstem, the BIAF accounts for the heterogeneity in autism. Thirdly, the integrated BIAF may inform and promote the establishment of a clinical screener that will build upon prominent indices of brainstem functions to reach high sensitivity and reliability for weighing the risk for social development deficits. Such a screener could also ascertain the domain-specific impairments for each infant and reveal in which functions auxiliary support is necessitated (e.g., auditory or tactile processing, autonomic-arousal reactivity, sleep hygiene, etc.). Accordingly, an early life BIAF screener could provide clinicians substantial gains vis-à-vis early detection of susceptible populations using electrophysiological indices (Figure 1, depicted on the right) along with behavioral ones (Figure 1, depicted on the left). The BIAF screener may carve new opportunities for customized interventions for the specific infant’s needs.

The original contributions presented in the study are included in the article/supplementary material, further inquiries can be directed to the corresponding author.

Both authors wrote the manuscript together and contributed equally to this article, and approved the submitted version.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.