Novel therapeutics in autism spectrum disorder

Abstract

Autism spectrum disorder (ASD) is a heterogeneous neurodevelopmental condition with limited treatment options that address its full complexity. This review critically evaluates novel therapeutics across five key domains: behavioral and psychosocial interventions, psychopharmacology, digital and AI-driven tools, neuromodulation, and genetic therapies. The broad heterogeneity of ASD complicates clinical research, often obscuring treatment effects and limiting generalizability. Challenges across domains include identifying reliable biomarkers, defining meaningful outcomes, and ensuring equitable access to evidence-based care and research participation. Behavioral and psychosocial therapies, considered first line interventions, are shifting toward naturalistic, developmentally informed, and neurodiversity-affirming models that reflect individual needs across the spectrum and across the lifespan. There continues to be no medications that are currently FDA approved for core autism symptoms, though ongoing pharmacologic research increasingly emphasizes biologically informed, stratified approaches to develop biomarkers that identify subsets of individuals who may benefit from a specific treatment. Digital and AI-driven tools promise greater personalization and expanded access but require safeguards, validation, and attention to equity. Neuromodulation techniques, including Transcranial Magnetic Stimulation (TMS) and Transcranial Direct Current Stimulation (tDCS), remain experimental yet highlight the importance of personalized protocols and ethical oversight. Genetics has seen advancements in gene-targeted therapies for syndromic forms of ASD marking a pivotal move toward precision medicine in autism, though ongoing challenges regarding safety, efficacy, and equitable access persist. This review informs clinicians, researchers, individuals with lived experiences and other important stakeholders by appraising current evidence, identifying limitations, and outlining future directions to advance rigorous, inclusive, and collaborative autism therapeutics.

Introduction

Autism Spectrum Disorder (ASD) is a neurodevelopmental condition with early childhood onset, defined by persistent difficulties in social communication and interaction, alongside restrictive, repetitive patterns of behavior, interests, or activities. As of 2021, ASD affected an estimated 61.8 million individuals globally (95 % uncertainty interval: 52.1–72.7), equating to approximately 1 in every 127 people. In the United States, prevalence estimates in children range from 1 to 3%, underscoring the substantial and growing public health relevance of the condition (Global Burden of Disease Study 2021 Autism Spectrum Collaborators [1,2].

The clinical presentation of ASD is highly heterogeneous. In addition to the core features, individuals often exhibit sensory processing differences and co-occurring intellectual or language impairments. These challenges often translate into significant functional impairments across the lifespan. Moreover, ASD frequently coexists with a range of psychiatric and medical comorbidities—such as anxiety, depression, epilepsy, and sleep disorders—that further complicate care and heighten the need for individualized, multidisciplinary treatment approaches.

Despite substantial advances in our understanding of the neurobiology and genetic architecture of ASD, pharmacologic innovation has lagged. No new medications have received regulatory approval for core or associated features of ASD since risperidone and aripiprazole were approved by the U.S. Food and Drug Administration for irritability and aggression in youth with ASD aged 5–18 years in 2003 and 2006, respectively. However, the landscape is rapidly evolving. There are currently hundreds of active recruiting clinical trials for autism spectrum disorder (ASD) worldwide, according to ClinicalTrials.gov. The range of ongoing studies demonstrates increasing diversity in therapeutic approaches and highlights the global commitment to developing new treatments. Many trials focus on children and explore both biological and behavioral strategies, with some investigating novel interventions based on recent advances in neuroscience and personalized medicine. Advances in precision medicine, genomics, neuromodulation, and computational approaches, including artificial intelligence and machine learning—have catalyzed a new wave of translational research, raising the possibility of more targeted and mechanistically-informed treatments.

In this review, we synthesize emerging evidence across five key domains of innovation in ASD therapeutics: behavioral and psychosocial interventions, pharmacological agents, neuromodulation, digital and AI-driven tools, and gene therapies. Together, these developments herald a transformative era in autism care, grounded in a deeper understanding of the underlying neurobiology, and tailored to the diverse needs of autistic individuals Fig. 1.

Fig. 1.

The diagram illustrates therapeutic modalities including genetics, pharmaceuticals, neuromodulation, digital/AI, and behavioral and psychological interventions through different mechanisms, collectively aim to optimize functioning and clinical improvement.

Special Note on Language: In preparing this review, we carefully considered the use of person-first language (e.g., “individuals with a diagnosis of autism”) versus identity-first language (e.g., “autistic individuals”). We recognize the importance of language in shaping perceptions of autism and honoring the diversity of lived experiences. Acknowledging that preferences vary across communities and individuals—including between self-advocates, families, and professionals, we have chosen to use a blended approach throughout this article. Our intention is to be inclusive of all voices in the autism community and to reflect the range of perspectives that inform contemporary autism discourse.

Behavioral and Psychological Interventions

Behavioral

Behavioral interventions in individuals with a diagnosis of autism stem from Applied Behavior Analysis (ABA), originating with Ivar Lovaas [3]. ABA-based therapies are highly structured interventions based on the science of learning, utilizing operant conditioning to increase positive behavior and decrease maladaptive behavior. Substantial evidence indicates that early and intensive ABA-based intervention (EIBI) improves cognitive skills scores, with some evidence also supporting improvements in social skills, communication, and adaptive behavior [4,5]. In recent years, there has been much critique of ABA from the autistic community and neurodiversity advocates, especially regarding Lovaas and others’ use of aversive consequences and attempts to dampen autistic behavior such as stimming [6,7].Subsequently, there has been a growing emphasis on more naturalistic interventions, termed Naturalistic Developmental Behavioral Interventions (NDBIs) [8]. NDBIs combine principles of learning with developmental science, rendering a more child-directed set of intervention components. Studies of NDBIs have generally shown efficacy, such as Early Start Denver Model [9,10], Project ImPACT [11,12], among others. This research area has grown immensely under the new umbrella term of NDBIs, with enough evidence that the research emphasis is shifting from the assessment of efficacy in academic settings to the evaluation of effectiveness when interventions are implemented in other settings.

In recent years, in addition to a focus on more naturalistic approaches that incorporate ABA principles, adaptations have been made to many behavioral interventions to be more community-based. The specific methods of intervention have generally remained consistent, but with priority on increasing access to care. There are efforts to adapt to different languages and cultures [[13], [14], [15], [16]], though more rigorous research is needed to better understand how behavioral interventions can reach a wider range of individuals. Prior to COVID-19, there were already efforts to expand access to care to more remote environments [17]; however, the onset of the global pandemic drastically increased remote service delivery by necessity, now accompanied by research to evaluate how to effectively deliver behavioral interventions via telehealth. Emerging evidence suggests that NDBIs and/or caregiver-mediated interventions for social communication or behavior management can yield benefit via telehealth [[17], [18], [19], [20]], including adaptations of manualized interventions (e.g., Project ImPACT) [[21], [22], [23]] ([22,23]). Research also suggests that brief training of community providers (e.g., early childhood educators, allied health professionals) to implement a behavioral intervention may be effective [24]. Overall, recent novel adaptations of behavioral interventions have primarily focused on care access, with growing use of technology to facilitate remote service delivery.

Another growing area of behavioral intervention in individuals with autism is focused on improving executive functioning. For instance, Unstuck and on Target is an executive functioning program for individuals with autism and others. Briefly, this program focuses on increasing cognitive flexibility and self-regulation in the school setting and was designed for delivery by school professionals [25]. Similar to adaptations of other behavioral interventions, there has been more recent focus on increasing access to care via remote service delivery [26], as well as training community providers to deliver the intervention [27,28].

Cognitive behavioral therapy

Initial behavioral treatments for autism focused on the core features of autism (e.g. social function and repetitive behaviors), as well as cognitive performance [29,30], communication, and adaptive behavior. Growing research has shown, however, that autistic children and adults also have higher rates of co-occurring mental illness than their neurotypical counterparts [[31], [32], [33]]. There is therefore an urgent need to develop and disseminate evidence-based psychotherapy for autistic individuals. To date, researchers and clinicians have primarily adapted existing cognitive behavioral therapy approaches to treat co-occurring mental illness in the autistic population. Applying cognitive behavioral therapy using treatment manuals for the co-occurring condition (the established pediatric OCD manual, for example) can be effective in improving core symptoms of autism as well as psychiatric comorbidities in both children, adolescents, and adults [29,30,34]. Traditional CBT is an effective treatment for autistic individuals with co-occurring anxiety disorders, but a manualized CBT adapted for autism was more effective in a head-to-head comparison RCT [29,30,35]. The Facing Your Fears Program [35] is an evidenced-based manualized CBT protocol for autistic youth ages 8 through 14, that was initially designed to be implemented in a clinic setting. Pilot trial data have demonstrated that the curriculum can be effectively implemented in a school setting [36] and with autistic teens with co-occurring IDD [37].

Mindfulness-based interventions

Research on mindfulness-based interventions in autistic children has shown promising evidence in decreasing anxiety, improving social skills, and reducing aggressive behaviors [38]. Mindfulness-based interventions have included movement-based interventions, such as yoga [39], as well as non-movement-based interventions, such as mindfulness-based CBT [40]. Treatment protocols have been developed specifically for autistic individuals (The Emotional Awareness and Skills Enhancement (EASE) program) [40], and modifications have been made to treatment manuals, such as Dialectical Behavioral Therapy, which was initially developed by Marsha Linehan to treat suicidality in women and was then extended to treat borderline personality disorder [41]. Two RCTs of DBT in autistic individuals show significant benefits for suicidal ideation and depression, although neither trial used an active control condition. The first used a waitlist control and the other using a ‘treatment as usual’ approach that was equivalent to a waitlist, with access to the DBT treatment upon trial completion [[42], [43], [44]]. Modifications to DBT have included increased use of visual materials, concrete examples to increase motivation, and utilizing a gaming format to align with patients' interests [44].

Trauma focused interventions

Both rates of traumatic experiences and rates of Post-Traumatic Stress Disorder are more common in autistic individuals, but psychotherapy approaches are only beginning to be tested. Eye Movement Desensitization and Reprocessing (EMDR) is an evidence-based treatment for post-traumatic stress symptoms in the non-autistic population [45]. There is emerging work utilizing EMDR to help autistic individuals cope with negative past even. In an open trial without a control group, Leuning et al. [46] demonstrated that EMDR can lead to a decrease in perceived stress and an improvement in daily functioning. Adapting EMDR to autistic patients includes being more flexible, adapting treatment to the individual's needs, and clear communication [47]. Some initial recommendations have also been published for adapting trauma-focused cognitive behavioral therapy (TF-CBT) for autistic individuals [48]. No RCTs have been conducted for either EMDR or TF-CBT adapted for autism.

Social skills therapy

Traditional social skills curricula were rooted in the medical model of disability that focused on teaching autistic individuals how to ‘overcome’ their social impairments [49]. There have been several systemic reviews and meta-analyses done on the effectiveness of group social skills programs for autistic individuals [[50], [51], [52]]. Group social skills interventions have shown modest effects in improving social responsiveness compared to control conditions. These reviews have demonstrated that social skill training is a viable and easily implemented treatment [51] but skills do not always generalize to real-life settings [52].

In the mid-1990's, autistic advocates began to encourage a shift in thinking about autistic people, and in 1998, autistic sociologist Judy Singer and journalist Hervey Blume coined the term ‘neurodiversity’ [53,54]. The neurodiversity perspective asserts that typical development is neither superior nor inferior to divergent neurodevelopment, and that all people deserve respect and dignity [49]. In recent years, there has been an increase in the use of groups to develop genuine social connections for autistic individuals [55]. Research has shown that groups focused on specific skill sets in autistic adults, such as dating [56,57] or employment [58,59] can improve treatment outcomes and motivation [60]. While these studies all used a randomized control design, they all had small sample sizes. There has been an increase in research on incorporating individuals' interests into intervention. LEGO-based therapy has been shown to effectively build social interactions in autistic children while engaging in play and using materials that interest them [4,5]. While many studies have been non-randomized studies of interventions (NRSI) or case reports, there was one repeated-measured wait list control study that showed LEGO therapy improved social motivation and social interactions [5]. Tabletop role-playing games (TTRPGs) have also been incorporated into treatment and have been shown to increase meaningful social interactions in autistic adults [61]. Research to date has shown that TTRPGs can create a safe space for autistic individuals to connect, but more research is needed to understand the long-term impacts of these interventions using randomized control trials. Similarly to other behavioral interventions, there has been an increased focus on using technology to support treatment and virtual reality-based training to enhance social skills interventions [62,63]. Further RCTs using larger sample sizes are needed to test these technology-enhanced approaches.

Limitations and future directions

There have been modest changes to the behavioral/psychological intervention landscape in recent years, mostly emphasizing neurodiversity and promoting the use of technology. The majority of studies did not utilize a randomized control design, and most had smaller sample sizes. Studies of social skills and psychotherapeutic interventions have focused on individuals without intellectual disability. There is a need to provide tailored interventions to autistic individuals throughout the lifespan and across the spectrum of support needs and verbal abilities. Given the growing need for autism services and the limited resources, future treatment and development should focus on leveraging technology to help disseminate autism interventions.

Emerging Targets in Autism Pharmacotherapy

Decades of research have suggested dysregulation across multiple neurotransmitter systems in autism spectrum disorder (ASD), providing neurobiological hypotheses for targeted pharmacologic testing. Aberrations in excitatory–inhibitory balance, particularly involving glutamatergic and GABAergic signaling, are among the most consistently reported findings in both preclinical and clinical studies, although the direction of the imbalance is less consistent. Changes in neural connectivity related to increased glutamate activity has been proposed due to its role in synaptic pruning in early brain development [64]. In parallel, alterations in monoaminergic systems—such as serotonin and norepinephrine—as well as dysregulation of neuropeptides involved in social behavior, including oxytocin and vasopressin, have been explored as mechanistic contributors and potential therapeutic targets. While existing pharmacological treatments primarily address associated symptoms such as irritability or hyperactivity, recent efforts have shifted toward disease-modifying strategies that engage core neurochemical pathways implicated in social communication or cognitive flexibility. This section reviews emerging and investigational agents that modulate key neurotransmitter systems.

Norepinephrine

Some evidence suggests alterations in the norepinephrine system in autism. Norepinephrine (NE) is involved in stress response, attention, and arousal. Stress responses are elevated in some individuals with autism and are associated with increased noradrenergic activity [65]. In children with autism, Kim et al. [66]observed increased baseline (tonic) activity—measured as larger resting pupil diameter—and reduced phasic responses during tasks, relative to age- and intelligence-matched peers. Plasma-based assays in older studies [67,68] report elevated norepinephrine levels, suggesting chronic sympathetic hyperactivity. Likewise, Makris et al. [69] documented altered diurnal patterns of salivary alpha-amylase, an indirect marker of norepinephrine activity. These findings point to abnormalities in the NE system, but they do not reveal whether altered NE function contributes to autism or whether it simply reflects the stressful experience for autistic people living in a non-autistic world. Regardless, heightened tonic NE with diminished phasic response could underlie attentional and arousal differences observed in some people with a diagnosis of autism [70,71].

Several adrenergic drugs have been investigated in ASD, including atomoxetine, guanfacine, and propranolol. Atomoxetine inhibits the presynaptic norepinephrine transporter (NET), preventing the reuptake of NE throughout the brain and inhibiting dopamine reuptake (via NET) in specific brain regions such as the prefrontal cortex. As the most studied medication in this group, it has several RCTs and one systematic review and has been associated with improvements in attention deficit hyperactivity disorder (ADHD) symptoms in ASD, with reported effect sizes ranging from 0.89 to 1.27 and response rates around 21 % [72]. However, the response rates for ATX in this population are consistently lower than those reported among children without ASD 47 % compared to 60 % [73]. With the focus on ADHD symptoms, there is no indication of whether atomoxetine impacts core ASD symptoms.

The selective alpha-2 receptor agonist guanfacine acts on NE auto receptors to decrease norepinephrine release. Scahill and colleagues led a multisite randomized placebo-controlled trial (RCT), of guanfacine extended release (ER) in 62 autistic children with ADHD symptoms [74]. They found a 50 % rate of response on the Clinical Global Impression of Improvement scale (CGI-I) vs 9.4 % for placebo in an eight-week trial. While they saw a significant change in parent ratings of hyperactivity/noncompliance on the Aberrant Behavior Checklist (ABC-H), they did not see an improvement in the irritability/agitation subscale (ABC-I) that previously showed benefit for risperidone and aripiprazole. In a follow-up paper, Politte and colleagues reported that guanfacine ER was not superior to placebo for anxiety or sleep; however, secondary measures showed a reduction in both oppositional and repetitive behaviors was observed [75].

Propranolol blocks central and peripheral beta-2 NE receptors and is effective in reducing social performance anxiety, in addition to its original use as an antihypertensive drug. Its potential role as a calming agent has generated interest for reduction of anxiety or agitation in individuals with a diagnosis of ASD. Following a supportive retrospective chart review, London and colleagues conducted a small randomized, placebo-controlled cross-over trial and found that high-dose propranolol (600–900 mg/day) significantly reduced severe behaviors compared to placebo, with modest cardiovascular effects. [76,77]. A larger 12-week RCT of propranolol for severe aggression and disruptive behaviors in adolescents and adults with autism is currently underway (NCT07091279). Interestingly, a few small studies have evaluated propranolol's impact beyond anxiety and agitation in ASD, with potential benefits for cognitive performance and social functioning [78,79].

With multiple studies pointing to increased norepinephrine activity in ASD and some benefit reported for medications that target the NE transporter or norepinephrine NE receptors individually, some have hypothesized that decreasing NE production could even improve core symptoms in autism [80]. Based on this idea, a placebo-controlled crossover study to assess the efficacy and safety of L1-79 (a tyrosine hydroxylase inhibitor that blocks the synthesis of both NE and dopamine) on core symptoms of social-communication deficits in ASD was recently completed, but results have not yet been published (NCT05067582). To date, therefore, studies of the norepinephrine system in ASD show benefit for hyperactivity, with some evidence for improvements in oppositional or disruptive behaviors, and little evidence thus far for core autism symptoms.

Serotonin

In addition to its well-known roles in mood, anxiety, social affiliation, and cognition, the serotonin (5-hydroxytryptamine, 5-HT) system is critically involved in multiple neurodevelopmental processes, including neuronal proliferation, migration, and projection, making it well-poised to play a role in autism. Further, biomarker data have identified changes in the serotonin system in autism, although these data are not well-connected to therapeutics to date. The most consistent and well replicated biochemical finding in autism is elevated levels of serotonin in the blood, found in approximately 25–30 % of individuals with a diagnosis of autism [81,82]. The causes of hyperserotonemia are not well-understood, although blood serotonin levels are highly heritable [81,83]. Decreased binding of the serotonin 5-HT_2A receptors in both platelet and brain in individuals with a diagnosis of autism is observed although this does not clearly relate to hyperserotonemia itself [84]. Serotonin transporter (SERT) binding is also diminished in the brain of autistic adults; although it is unclear whether this represents a decrease in serotonergic fibers or diminished availability of SERT within the fibers [81]. One additional clue is that depletion of the serotonin 5-HT precursor, tryptophan, leads to worsening repetitive behaviors and mood in autistic individuals [85].

In addition to these biomarker data, well-replicated benefits of serotonin reuptake inhibitors (SRIs) in obsessive compulsive disorder, which also includes prominent repetitive behavior, has driven substantial interest in serotonergic agents in autism [83]. Unfortunately, sizable studies of the SRIs citalopram [86] and fluoxetine [87] did not show significant benefit for repetitive behavior in children with a diagnosis of ASD, although one RCT in adults did show benefit for OCD symptoms in autistic adults [88]. There is a dearth of studies designed to study the effect of SRIs on co-occurring mental health conditions in autistic children, despite both depression and anxiety being more common in ASD [4]. Both risperidone and aripiprazole act as serotonin 5-HT_2A receptor antagonists, but the similar effects of haloperidol [89] lead to the expectation that their common antagonism at the dopamine D₂ receptor is responsible for improvements in irritability/agitation in individuals with ASD. A recently completed study of pimavanserin, a selective serotonin 5-HT_2A receptor inverse agonist (an antagonist that also blocks constitutive receptor activity), sought to evaluate whether improved irritability/agitation could be seen in the absence of action on dopamine receptors, with results pending (NCT05523895).

There are other agents that have attracted attention. Even without biomarker findings in autism, the well-known pro-social effects of the recreational drug MDMA (3,4-methylenedioxymethamphetamine) have led investigators to consider its use in autism, although only a single small trial has been published to date [90]. Recognizing that MDMA has a crude mechanism of action including reverse transport of serotonin as well as actions on the dopamine system, the Malenka lab evaluated the molecular mechanisms underlying its prosocial effects. In mice, they were able to narrow the impact on sociability to effects on the serotonin 5-HT_1B receptor in the nucleus accumbens, often referred to as the brain's reward center [91]. A follow-up study found that serotonin 5-HT_1B agonism increased sociability in multiple genetic mouse models of autism ([92]. These findings motivated an ongoing multi-center RCT to evaluate the efficacy of a serotonin 5-HT_1B receptor agonist compared with placebo for the improvement of social difficulties in participants diagnosed with autism (NCT05081245).

Oxytocin

Oxytocin (OXT) is a 9-amino-acid neuropeptide that is synthesized in the hypothalamus and facilitates attachment and social recognition in some contexts ([93] in both animals and humans. In non-autistic adults, intranasal OXT has been shown to boost trust and dampen anxiety in socially stressful situations; however, subsequent studies have shown inconsistent effects, potentially explained in part by sex differences, with larger brain and behavior effects in females [[94], [95], [96]]. Several studies have assessed plasma oxytocin levels in children and adults with autism with little consistency. Building on reports of changes in social attention and emotional recognition in non-autistic adults [97], initial studies in autism reported significant improvement on emotion comprehension tasks with one-time OXT infusion in a small adult sample [98] and with intranasal OXT in a small youth sample [99]. Multiple follow-up RCTs with small to medium sample sizes found inconsistent results with extended use of intranasal OXT, raising questions about the appropriate age range, dosing, duration of treatment, and the relationship of treatment response to plasma oxytocin levels [[100], [101], [102], [103]].

Finally, a large RCT in 290 children and adolescents assessed the effects of 24 weeks of escalating dose treatment with intranasal OXT compared to placebo in a heterogeneous population designed to span the full autism spectrum [104]. No difference was observed on any outcome measure, including when accounting for baseline plasma OXT levels, although this study did not have sufficient funding to incorporate the original social cognition measures that sparked initial interest in OXT. It is difficult to justify further free-standing studies of OXT in the context of this large negative trial with only a modest placebo effect, but it may still be reasonable to consider studies that combine oxytocin in concert with behavioral therapy aiming to amplify potential benefits on social cognition ([105].

Vasopressin

Like OXT, arginine vasopressin (AVP) is a 9-amino-acid neuropeptide that modulates social and emotional circuits across species, with some evidence suggesting a larger effect size in males [106,107]. Unlike OXT, which has a single receptor, AVP exerts its effect via the G-protein receptors AVPR1a (V1a), V1b and V2. Of these, vasopressin V1a and V1b are thought to be involved in social emotional circuitry, with V1a being the dominant receptor in the brain. In contrast, V2 is expressed in the kidney and is responsible for the anti-diuretic effects of AVP. Animal studies point toward increased prosocial behaviors with higher CSF levels of vasopressin [108,109]. Small observational studies suggest that the CSF concentration of vasopressin may be lower in autistic children; however, these data need to be replicated [110].

Based on animal literature and the small autism CSF studies, Parker and colleagues conducted an RCT to test the efficacy and tolerability of a 4-week daily intranasal AVP treatment (max dose 32 IU) in 30 children with a diagnosis of autism. Compared to placebo, children who received intranasal AVP showed improvement on the primary outcome measure, the Social Responsiveness Scale (P = 0.0052), with a large effect size (Cohen's d = 1.40). AVP treatment also diminished anxiety symptoms and some repetitive behaviors [111].

In contrast, based on animal model findings pointing to AVP as a trigger for social defensiveness and V1a antagonism as improving sociability and easing anxiety and repetitive behavior, Roche pursued clinical trials of the selective V1a antagonist balovaptan in the largest autism clinical trial program to date. After pilot work with a broad array of social cognition and real-world measures, they found promising results in a large phase 2 adult VANILLA study using a hybrid measure of adaptive function in the social and communication domains [112]. Unfortunately, two follow-up autism trials did not show separation from placebo in adults or in 5–17-year-olds [113,114]. Unlike animal experiments, where such ambiguous results would likely trigger a new round of experiments, the cost of clinical trial programs doesn't allow for follow-up studies to understand whether a failed trial indicates that a drug does not work or whether a large placebo effect swamped the ability to detect a difference [115].

γ-Aminobutyric acid (GABA) and glutamatergic modulating agents

Research has suggested that excitatory and inhibitory signaling imbalance (E: I imbalance) may contribute to the pathogenesis of ASD in some children [64,116]. Based on this theory, several pharmacological agents that modulate the GABAergic (inhibitory) and glutamatergic (excitatory) systems have been studied as potential treatments for the core symptoms of ASD and ASD-associated behavioral disturbances. These are considered together in this section because they are broadly based upon the same underlying hypotheses.

The gamma-aminobutyric acid (GABA) system

GABA is the primary inhibitory neurotransmitter in the brain, and some studies of E:I balance have suggested diminished GABA function in ASD, although this has not been consistent across studies. When activated, GABA-A and GABA-C receptors serve as chloride channels that typically allow chloride to enter the neuron, leading to hyperpolarization and inhibition of firing. During early development, however, chloride levels are higher inside the neuron, driven by the activity of the chloride importer NKCC1, and GABA-A activation therefore causes chloride to leave the cell, leading to depolarization. A switch from NKCC1 to the chloride exporter KCC2 is responsible for converting GABA's effect from excitatory to inhibitory in early postnatal life. One theory is that this neonatal switch between NKCC1 and KCC2 is delayed or altered in autism, based upon findings in two rodent models of autism risk (Ben-Ari et al., 2012); although it is unclear whether this is relevant to humans. Based on the theory of a defective neonatal switch, several groups have conducted studies of bumetanide, a loop diuretic that inhibits sodium-potassium-chloride co-transporters in the kidney and the brain. By blocking NKCC1, Bumetanide prevents the import of chloride into the neurons, thereby decreasing intracellular chloride and promoting GABAergic inhibition. It is difficult to keep parents and clinicians blinded to a study of a medication that leads to increased urination and changes in blood electrolytes, and some studies included unblinded potassium supplementation or had the same physician assess side effects and potential benefits. Even without rigorous blinding in some studies, a meta-analysis [117] of nine randomized controlled trials, including 1036 children, found that bumetanide does not have evidence for generalized or lasting results across childhood, which is consistent with a larger systematic review [118], as well as the largest multi-site randomized trial [119].

Some have argued that a subset of children may benefit, potentially identified by changes in EEG [120] or magnetic resonance spectroscopy (MRS) [121,122]. In the context of the large number of participants studied, however, it may make more sense to try to identify patterns of responders within available data rather than initiating new trials of bumetanide [123] NCT07005414).

Unlike other GABA receptors, GABA-B is a G-protein-coupled receptor that has slower action that includes opening potassium channels and closing calcium channels. Arbaclofen (R-baclofen) is a GABA-B receptor agonist that is the active enantiomer contained in racemic baclofen, which is commonly used to treat muscle spasms in cerebral palsy and other conditions. Some GABA-B receptors are expressed on glutamatergic neurons, thereby allowing arbaclofen to block their activation and release of glutamate [124]. Initial open-label data for arbaclofen in autism and placebo-controlled crossover data in fragile X syndrome were hopeful, but the primary outcome measures were not significant in follow-up RCTs [125,126]. Some secondary analyses were statistically significant, though without correction for multiple comparisons, which encouraged further study. Two recently completed RCTs (NCT03682978, NCT03887676) further evaluated arbaclofen in a total of 204 autistic children and adolescents with the Vineland Adaptive [118,127]Scales social domain as the primary outcome measure [128], as well as a number of overlapping secondary measures. A joint analysis was presented in 2024 with the suggestion that a subset of participants benefited, again in secondary analyses, although the final results have not been published [118,127].

The glutamate system

Glutamate is the primary excitatory neurotransmitter in the brain, and some have hypothesized increased glutamate function based upon the theory of excess excitation versus inhibition [129]. The evidence is somewhat mixed, and some data actually support a decrease in glutamate function, which may be accompanied by decreased inhibition [130]. In addition to the prevailing hypothesis favoring excess excitation in autism, there is also a practical consideration for medications targeting the glutamate system, in that glutamate agonist drugs can trigger seizures, which are already more common in autism. Medications acting on the glutamate system could target AMPA, NMDA, or metabotropic (mGluR) receptors.

Within the glutamate system, NMDA receptor antagonist medications, including memantine, have received the most attention. A 2022 Cochrane review of memantine for autism spectrum disorder evaluated three RCTs comparing the effects of memantine to placebo on core ASD symptoms and on secondary outcomes (language, intelligence, memory, adaptive behavior, hyperactivity, irritability) in children and adolescents, and found no statistically significant difference [131]. The largest RCT of memantine to date used a randomized withdrawal design that may have complicated its evaluation. All participants started on memantine for at least 12 weeks, with 60 % showing a favorable response by parent ratings. The responders were then randomized to continue on the same dose of memantine or switch to a lower dose or to placebo for 12 weeks. Unfortunately, more than 2/3 of the participants in each randomized group showed a loss of treatment response, with no significant difference between groups [132]. This worsening with expectation of switching to placebo has been termed by some as a “nocebo” response. Similar to other medications, some researchers have argued that memantine should be tested in an RCT in a subgroup of participants, in this case with a biomarker that indicates E: I imbalance ([133].

Other NMDA receptor antagonists have also been evaluated in individuals with autism, including amantadine and dextromethorphan, which acts as a non-competitive NMDA agonist in addition to actions on the dopamine, serotonin, and norepinephrine systems, with the overall evidence not supporting benefit for core symptoms or for irritability/agitation in autism [134]. Finally, ketamine is a noncompetitive agonist of the NMDA receptor that is primarily used as a dissociative anesthetic, but intranasal ketamine has recently been approved for treatment-resistant depression. One small, randomized cross-over study of intranasal ketamine in adolescents and young adults with ASD (n = 21) found no significant impact on primary or secondary outcomes of core social symptoms [135].

Beyond compounds that act directly on glutamate receptors, N-acetylcysteine (NAC) has mixed activity that includes serving as an indirect glutamate modulator, in addition to acting as an antioxidant in restoring intracellular glutathione. Across published RCTs to date, NAC has mixed evidence with regard to improvement in irritability/agitation, hyperactivity/defiance, and stereotypic behavior on the ABC, though all studies found NAC to be well-tolerated and safe [[136], [137], [138], [139]], with one follow-up RCT that is currently underway (NCT05664789).

The cannabinoid system

The endocannabinoid system (ECS) is a neuromodulatory system that regulates mood, anxiety, and immune responses. The ECS consists of endocannabinoids (eCBs) such as anandamide (AEA) and 2 arachidonoyl glycerol (2-AG), enzymes for synthesis and degradation of the endocannabinoids, cannabinoid transporters, and cannabinoid receptors (CB1 and CB2) [140,141]. CB1 receptors are located in the central nervous system (CNS), peripheral nervous system, and peripheral organs [142], whereas CB2 receptors are primarily located on immune cells, though also found in the CNS [143]. A couple of studies suggest that peripheral endocannabinoid levels may be lower in ASD, although more data is needed to evaluate this possibility [144,145]. Based upon reports of benefit in the popular media, however, considerable interest has focused on cannabis and its two primary phytocannabinoids, delta-9-tetrahydrocannabidiol (THC), an agonist at CB1, and cannabidiol (CBD), an allosteric modulator of CB1 that antagonizes the effects of THC. Notably, CBD preparation (Epidiolex) is effective as an adjunctive medication for seizures in treatment-resistant Dravet syndrome, a severe form of epilepsy with a high rate of co-occurring autism, although there was no significant effect on behavioral outcomes [146].

A recent systematic review [147] looked at seven studies of CBD-rich preparations comprising 494 autistic individuals between the ages of 5–19 years. The review included three RCTs, two retrospective cohort studies, and two prospective cohort studies. Response to treatment was inconsistent across individuals and across studies, with some suggestion of benefit for sleep quality and anxiety symptoms but mixed results for core autism symptoms. The most common side effects were drowsiness, anxiety, restlessness, decreased appetite, and weight loss. Adverse events were generally mild and occurred at similar rates across cannabinoid and placebo groups, causing the interventions to be considered relatively safe. It is important to note that CBD inhibits cytochrome P450 iso-enzymes [148,149] and taking CBD alongside medications metabolized by these iso-enzymes can alter their clinical effects and increase adverse effects.

One randomized, placebo-controlled crossover study of the proprietary Epidiolex CBD preparation has recently been reported, including 39 autistic boys between the ages of 7–14 years old with cognitive limitations and moderate to severe self-injurious, aggressive, or other disruptive behaviors [150]. There was no significant benefit of CBD over placebo on the primary outcome measures, which were the RBS-R, CBCL, and ADOS-2, with a prominent placebo effect on all three measures. A significant placebo effect was observed on all three measures. Although the data to date do not support its use, there are ongoing RCTs of CBD preparations, including Epidiolex, that may clarify whether there are any associated benefits (NCT05015439, NCT04517799).

There has also been interest in targeting endocannabinoid metabolism. Fatty Acid Amide Hydroxylase (FAAH) is responsible for the synthesis and degradation of endocannabinoids, including AEA, and inhibiting FAAH can therefore increase eCB activity. One RCT evaluated the FAAH inhibitor JNJ-42165279 in 61 autistic adolescents and adults 13–35 years old [151]. The treatment group did not show a statistically significant reduction in ASD symptoms compared with the placebo group on the primary outcome measures of core ASD symptoms; however, some moderate changes were reported on secondary outcome measures of core symptoms, as well as anxiety. Those with the highest levels of study medication or higher levels of AEA had the greatest reduction/improvement in symptoms. No follow-up studies have been initiated to date.

The purinergic theory

Beyond neurotransmitter-based mechanisms, emerging metabolic and immunologic models of autism have prompted interest in other therapeutic targets. Some have proposed that autism arises from a persistent disruption of the metabolic stress-response system regulated through purinergic signaling, termed the Cell Danger Response (CDR) [152]. Under cellular stress, extracellular adenosine triphosphate (ATP) and adenosine diphosphate (ADP) act as “danger signals” that bind to P1, P2X, and P2Y receptors. While adaptive in acute injury, a chronically activated CDR—as some proposed in conditions such as Chronic Fatigue Syndrome and Fibromyalgia—could perpetuate metabolic dysfunction, inflammation, or impaired cellular communication [153].

Suramin, a non-selective P2X/Y receptor inhibitor used for African trypanosomiasis and river blindness (not FDA-approved), has gained attention as a candidate therapy for ASD. In preclinical models—including maternal immune activation and valproic acid exposure—single doses of suramin have been reported to change social behavior and some synaptic, metabolic, and immune markers [[154], [155], [156]]. Two small trials of suramin have been reported, both enrolling only male children. Naviaux and colleagues (2017) conducted a pilot study of a single 20 mg/kg intravenous (IV) dose with five participants per group. All participants who received suramin experienced a rash within a day of administration, which may have been a problem for study blinding. A single dose led to suramin levels between 1.5 and 15 μmol/L out to 6 weeks post IV infusion. The suramin group showed changes in many metabolic markers, particularly in amino acid-related pathways. The authors concluded that “data were incomplete and insufficiently powered for analysis” of potential benefit, but they did note improvement in ADOS scores in comparison to baseline in the suramin group but not in the placebo group [157]. Hough and colleagues (2023) conducted a phase II trial (n = 52) of repeated dosing of suramin at 10 mg/kg or 20 mg/IV versus placebo at weeks 0, 4, and 8. Neither dose showed a significant change on the primary outcome measure, the ABC-Core composite measure. One of four secondary outcomes, the CGI-I, was significant for the 10 mg/kg group compared to placebo (p = 0.016), though it was not significant in the 20 mg/kg group. Rash occurred in 8 of 19 participants in the 20 mg/kg group but in only 1 in the 10 mg/kg group. One participant with multiple pre-existing risk factors discontinued due to status epilepticus, which was coded as a serious adverse event, and one participant discontinued due to a rash [158,159].

The unclear dose-response pattern, uncertain treatment effects, and adverse events indicate the need for mechanistic studies linking exposure to target engagement, as well as larger trials stratified by phenotype. Currently, a follow-up study is underway: STAT-2A (Suramin for the Treatment of Autism Trial), a multi-site, randomized, double-blind, crossover 30-week study to evaluate proof of concept, safety, and pharmacokinetics of suramin sodium with repeated dosing by IV infusion in 5–14 yr olds with ASD (NCT06866275).

Limitations and future directions

Despite substantial advancement in basic neuroscience, only modest gains have been made in developing medications that impact core symptoms, ameliorate associated psychopathology, or significantly improve quality of life in those diagnosed with autism. Large research consortia in the US, the EU, and Canada have begun to collaborate to overcome barriers in new drug development. Several challenges have been identified, including symptom heterogeneity in ASD, the lack of validated objective biomarkers for diagnosis, treatment prediction, or target mechanism engagement, as well as the need for improved clinical endpoints [160]. In the context of substantial genetic and presumably biological heterogeneity, one key goal would be to identify biomarkers that could be used to identify subgroups for targeted treatment, such as a measure of E:I imbalance prior to treatment with an agent in the glutamate or GABAergic systems [161,162]. This could allow precision medicine approaches that extend beyond the rare genetic syndromes that individually constitute a very small minority (<1 % each) of individuals with a diagnosis of autism. Another important step is to include markers of target engagement in clinical trials so that investigators can be confident that a medication had the desired effect in the brain, even if the behavioral outcome may be less clear [160,163].

The phenomenon of placebo response in medicine is not limited to autism clinical trials but has certainly contributed to fewer successful trials in this area. Meta-analytic reviews of pharmacological and dietary-supplement randomized controlled trials in ASD demonstrate a robust placebo response, with pooled response rates showing up to ∼20–30 % of participants achieving “much improved” status despite inactive treatment [164]Strong placebo response may limit the ability to discern true treatment effects contributing to failed clinical trials. Predictors of heightened placebo effects include lower baseline severity, flexible dosing protocols, multi-site designs, and reliance on caregiver-rated outcomes [164]. For example, in the STAART citalopram trial, participants with lower initial symptom severity were more likely to respond to placebo, while greater baseline impairment was associated with differential treatment response [165]. In balovaptan trials, reduced placebo effects correlated with higher baseline adaptive functioning (Vineland-II) and longer study duration [166]. To mitigate these effects and improve signal detection, emerging best practices emphasize enrolling participants with adequate baseline symptom severity, using validated and objective outcome measures less prone to expectancy bias, and incorporating single-blind placebo lead-in phases to stabilize baseline ratings [74]. Such methodological refinements are essential to reduce assay noise, enhance replicability, and enable the field to move toward precise, mechanism-based therapeutics for ASD.

Another important step is the greater incorporation of perspectives of those with lived experiences across cognitive and language abilities, particularly in identifying meaningful endpoints that are predictors of functional outcomes and improved quality of life.

Novel Technology-Based Therapies for Autism Spectrum Disorder

Digital tools often provide structured, visually oriented, and predictable environments, which may align with the preferences of many individuals with autism spectrum disorder (ASD). These tools can be used to address social communication, behavioral regulation, and adaptive functioning skills. The field of technology-enhanced ASD interventions has matured beyond initial feasibility studies; evidence base varies in quality and depth. Moreover, practical barriers such as training demands, hardware and software costs, reimbursement gaps, and regulatory ambiguity continue to slow routine adoption.

To understand the overall effectiveness of technology-based interventions for autism, several comprehensive meta-analyses have examined outcomes across diverse platforms and populations. Project AIM (Autism Intervention Meta-analysis) is a comprehensive systematic review that aggregated 130 independent samples comprising 6240 participants aged 0–8 years and 1615 effect sizes; when the full set was analyzed, the overall effect was moderate (Hedges' d = 0.45), but once analyses were restricted to low-risk randomized controlled trials, effects contracted and largely clustered on outcomes directly targeted by the intervention [167]. Notably, technology-enhanced Naturalistic Developmental Behavioral Interventions still produced statistically reliable gains in social communication under these stricter standards (Hedges' g = 0.36, 95 % confidence interval 0.23–0.49). This indicates that technology can enhance the effectiveness of evidence-based behavioral interventions that are specifically designed to promote skill generalization, though questions remain about whether these improvements extend to broader developmental outcomes beyond the targeted skills [168].

Among specific technological approaches, robot-mediated interventions demonstrate both significant potential and important limitations in their current form. A meta-analysis of 40 studies (346 participants, 17 RCTs) reported moderate effects on social functioning (Hedges' g = 0.35, 95 % confidence interval 0.12–0.58) but minimal impact on emotion recognition (Hedges' g = 0.13) and motor outcomes (Hedges' g = 0.06) [169]. Most studies used humanoid robots such as NAO, QTrobot, or Kaspar for four to twenty weeks across homes, clinics, and schools. The pattern appears to be that robots can help with focused social behaviors in the short to medium term, but breadth of effect is currently narrow. A broader review of digital interventions across 19 randomized controlled trials with 815 participants reported a small overall effect (Hedges' d = 0.32, 95 % confidence interval 0.12–0.51), with social skills outcomes looking the strongest; however, heterogeneity was extreme (I² = 100 %), which likely reflects major differences in platforms, intensity, fidelity, targets, and outcome measurement [170].

Robot-assisted therapy: implementation and clinical outcomes

Moving beyond meta-analytic findings to examine specific implementations, individual randomized controlled trials provide important insights into how robot-assisted interventions perform in real-world clinical settings. A well-designed study compared Pivotal Response Treatment delivered with and without robot assistance (NAO) across 12 weeks in 87 children aged 5–8 years. Both groups improved significantly on the Pervasive Developmental Disorders Behavior Inventory social skills subscale, but robot assistance did not add measurable benefit over human-led therapy (between-group difference: 0.69, 95 % confidence interval −2.11 to 3.49; p = 0.63) [171]. Completion rates were high (96 %), and qualitative feedback suggested that children found robots engaging, but the lack of additive benefit raises questions about cost-effectiveness when skilled human therapists are available.

Other trials have focused on multi-robot environments and group-based interventions. A study using standardized protocols with three humanoid robots (Pepper, NAO, and QTpal) across 16 sessions demonstrated improvements in collaborative behaviors, with effect sizes ranging from medium to large (Cohen's d = 0.5–0.8) for turn-taking and joint attention tasks [172]. However, these gains were measured primarily through structured robot-interaction tasks, and generalization to human-human social contexts was not systematically assessed [173].

Virtual reality interventions: immersive skill development

Virtual reality platforms offer unique advantages for autism intervention, providing highly controlled, customizable environments where social and behavioral skills can be practiced safely and repeatedly. A RCT randomized 120 preschoolers with Intelligence Quotient above 70 to 20 weeks of non-wearable virtual reality therapy or wait-list control and found sizable improvements on the Aberrant Behavior Checklist (mean difference −12.4, 95 % confidence interval −15.8 to −9.0) and the Childhood Autism Rating Scale (mean difference −5.2, 95 % confidence interval −6.8 to −3.6) [174]. Interestingly, 68 % of the virtual reality group showed clinically significant reductions in attention-deficit/hyperactivity disorder symptoms versus 12 % of controls. That crossover improvement could reflect overlapping symptom domains or measurement artifacts, but it also raises the possibility that immersive, highly structured practice environments generalize to attentional regulation.

Wearable technologies and real-time feedback systems

Recent advances in wearable technology have enabled the development of interventions that provide immediate, context-sensitive feedback during real-world social interactions. “Superpower Glass,” which pairs augmented reality glasses with on-device emotion recognition and prompts, was tested in a randomized controlled trial of 71 children aged 6–12 years; 20 min per day over six weeks led to significant gains on the Vineland Adaptive Behavior Scales, Second Edition, socialization domain (mean change 4.58, 95 % confidence interval 2.21–6.95; p = 0.001), with high completion (89 %) [175]. The effect was comparable in magnitude to roughly three months of standard therapy, which is notable, although long-term durability is still poorly characterized.

Augmentative and Alternative Communication: evidence-based applications

For children with minimal verbal communication, Augmentative and Alternative Communication (AAC) technologies represent some of the most established and clinically validated technological interventions, though implementation quality and outcomes vary significantly across settings. A 12-week trial of voice output communication aids demonstrated clear language gains on the Preschool Language Scale, Fifth Edition, for receptive (mean increase 18.3 versus 7.2; p < 0.001) and expressive language (15.7 versus 6.8; p < 0.001), along with better emotion regulation (22.4 % versus 8.1 %; p < 0.01) [12]. When the Proloquo2Go application was combined with Joint Attention, Symbolic Play, Engagement, and Regulation and Enhanced Milieu Teaching over 24 sessions in 68 minimally verbal children, joint attention improved (Hedges' d = 0.67; p < 0.01) and social communicative utterances rose sharply (12.4 versus 3.2 per 10 min; p < 0.001), with maintenance at follow-up [13]. A meta-analysis across 10 randomized controlled trials also reported significant benefits on visual and fine motor skills (standardized mean difference 0.41, 95 % confidence interval 0.15–0.67), with larger effects when parents led the intervention—echoing a broader pediatric pattern in which trained caregivers extend therapy into daily life [14].

Artificial intelligence and adaptive learning systems

The integration of artificial intelligence into autism interventions represents a potentially transformative development, allowing for real-time customization of therapeutic content based on individual responses and learning patterns. Interim randomized controlled trial data from 48 children aged 2–6 years showed greater reductions on the Autism Diagnostic Observation Schedule, Second Edition (2.3 versus 0.4 points; p < 0.01) and larger parent-reported gains in receptive language (23 % versus 5 %) and social skills (31 % versus 8 %) relative to control [176]. These preliminary results suggest that real-time adjustment of dose and difficulty is likely to be a defining feature of future tools, although replication and long-term follow-up will be essential.

Comparative effectiveness and integration approaches

Understanding how technology-based interventions compare to traditional therapeutic approaches is crucial for clinical decision-making and resource allocation. Comparative effectiveness work is still sparse but informative. Across 18 studies and 1266 participants, face-to-face social skills training (Hedges' g = 0.81) did not differ significantly from technology-based interventions (Hedges' g = 0.93) [177]. Arguably, that is the practical headline: technology can match traditional efficacy and sometimes makes practice at home, standardization, or remote delivery easier. Whether it does so at lower cost, at scale, and with durable outcomes is another question.

Limitations and future directions

Despite promising clinical outcomes, the widespread adoption of technology-based interventions in autism care faces significant barriers. Technical failures remain common, affecting up to 30 % of sessions, and training demands vary widely depending on complexity. High upfront costs—ranging from $250 for AAC apps to over $15,000 for programmable robots—are compounded by the lack of reimbursement pathways, such as CPT codes for clinician time or hardware depreciation. Cultural and linguistic equity gaps, insufficient long-term follow-up data, and inconsistent regulatory oversight further limit integration into real-world settings. To bridge these gaps, future efforts should focus on establishing clinical guidelines surrounding the development of technology-based interventions, generating cost-effectiveness data through pragmatic trials, developing scalable training and support models, and prioritizing inclusive, culturally adaptable designs. Only then can these tools move from specialized use into sustainable, equitable clinical practice.

Neuromodulation in Autism Spectrum Disorder: Emerging Therapeutic Frontiers

As we have outlined, despite decades of research and clinical development, pharmacology and behavioral intervention options for autism spectrum disorder (ASD) remain limited in scope and efficacy [178]. Neuromodulation techniques offer a fundamentally different approach by directly targeting neural circuits [179,180]. For co-occurring depression, transcranial magnetic stimulation (TMS) follows protocols approved by the FDA for non-autistic people with depression [181]. However, applications targeting core ASD symptoms represent an emerging research domain requiring careful evaluation of both efficacy and safety considerations [179].

Transcranial magnetic stimulation (TMS)

Clinicians working with individuals with a diagnosis of autism often exhaust conventional behavioral and pharmacological options, leaving patients and families with persistent core symptoms and limited alternatives. TMS offers a non-pharmacological intervention that can target specific brain circuits, providing a potential new avenue when traditional approaches prove insufficient. However, the evidence for TMS in autism core symptoms remains limited and conflicting.

TMS delivers magnetic pulses to specific brain regions, creating targeted electrical currents that can modify neural activity. Stimulating accessible cortical areas can influence deeper brain structures and distributed networks throughout the brain, allowing potential modulation of circuits involved in social communication and repetitive behaviors. The technique builds on established clinical success: TMS is FDA-approved and widely used for treating major depressive disorder, providing a foundation of safety data and clinical experience directly relevant to autism populations who frequently experience co-occurring depression. The safety profile is generally favorable for patients with ASD, though not without concerns. Most side effects are minimal - typically limited to mild scalp discomfort or brief headaches, with an overall adverse event rate of approximately 25 % [182,183]. There is a potential seizure risk with TMS that could theoretically be higher in ASD, but seizures have been reported only in isolated case reports, with the estimated seizure risk for theta burst stimulation being 0.02 % [184].

For depression in patients with ASD, TMS can follow protocols established in non-autistic people. Small open-label studies show improvements in depressive symptoms using standard TMS approaches, with sustained remission rates comparable to those seen in neurotypical populations [181,185]. With shorter protocol duration and reduced session time using theta burst stimulation, TMS treatment may be better tolerated in individuals with ASD, where tactile hypersensitivity to scalp stimulation and auditory hypersensitivity to coil clicking sounds may be limiting factors for traditional repetitive TMS protocols. [185]. TMS could theoretically be applied to other areas of difficulty in ASD, such as executive function deficits, but initial studies are discouraging. A well-designed, double-blind, sham-controlled trial (n = 40) of high-frequency repetitive TMS targeting the dorsolateral prefrontal cortex found no overall efficacy for executive function deficits in autistic adults [186]. These findings align with a magnetic resonance spectroscopy study showing that while TMS can modulate cortical glutamate levels in ASD, those changes may not necessarily translate to meaningful functional improvements [187]. In one open label TMS study for depression in autism, secondary cognitive benefits, including improvements in fluid cognition and associated EEG changes, were observed in autistic individuals treated for depression, though it remains unclear if these are associated with mood improvement or direct cognitive effects [188].

For core ASD symptoms, the evidence for TMS is considerably weaker and also conflicting. While recent systematic reviews from 2024 suggest potential benefits for stereotyped behaviors, repetitive behaviors, and some social communication domains, most individual randomized controlled trials show limited or no superiority over sham stimulation. While some small open-label studies suggested benefits, the largest and most rigorous double-blind randomized controlled trial (N = 60) found no superiority of repetitive TMS over sham stimulation for any clinical or neuropsychological measures in ASD [189]. A recent comprehensive network meta-analysis of 16 randomized trials (N = 709) concluded that TMS interventions showed no significant improvements in social-communication symptoms or restricted/repetitive behaviors - the defining features of ASD [190]. However, expert consensus acknowledges that most studies to date have been limited by small sample sizes, heterogeneous protocols, and methodological concerns, making definitive conclusions premature.

Some researchers have explored alternative approaches, such as low-frequency TMS targeting gamma oscillation abnormalities, with preliminary positive results for repetitive behaviors and executive functions [191]. An open label study found that accelerated theta burst stimulation was highly effective and well-tolerated for treating refractory depression in autistic individuals with substantial and sustained improvement in depressive symptoms lasting through 12 weeks post-treatment, achieving full remission in 5 of 9 completing participants and partial remission in 3 participants. However, these open-label studies involve small samples and require study in larger controlled trials.

Transcranial direct current stimulation (tDCS)

tDCS modulates cortical excitability by delivering low-intensity electrical currents (typically 1–2 mA) through scalp electrodes. Anodal stimulation increases excitability via membrane depolarization, while cathodal stimulation reduces excitability through hyperpolarization. Unlike TMS, which delivers discrete magnetic pulses to produce immediate neural firing, tDCS provides sustained, subthreshold modulation of resting membrane potentials, making it theoretically suitable for enhancing learning and consolidating therapeutic gains during concurrent behavioral interventions. The distribution of current depends heavily on electrode application, with computational modeling now employed to optimize targeting of functionally relevant brain regions. High-definition tDCS, which uses multiple smaller electrodes, may enhance spatial precision—potentially valuable given the neural heterogeneity observed in autism.

The evidence for tDCS in ASD remains limited and mixed, with significant methodological differences across studies. A comprehensive network meta-analysis of 16 RCTs (N = 709) identified that only one specific protocol—anodal tDCS over the left dorsolateral prefrontal cortex paired with cathodal stimulation at an extracephalic location—showed statistically significant improvement in overall ASD symptoms compared to sham controls [190]. However, this finding comes with important limitations: the effect size, while large, had very wide confidence intervals (SMD = −1.40, 95 % CI = −2.67 to −0.14), indicating substantial uncertainty. Critically, this same analysis found that while one tDCS protocol significantly improved overall autism symptom severity, no non-invasive brain stimulation interventions significantly improved the core defining features of ASD—social-communication symptoms or restricted/repetitive behaviors—when these domains were analyzed separately.

Individual studies have shown more encouraging results for cognitive neuroscience tasks, which may serve as intermediate outcomes that could demonstrate successful engagement of brain targets. For example, a randomized controlled trial (n = 22) combining tDCS with emotion recognition training demonstrated improvements in dynamic emotion recognition in adolescents with autism, though benefits were limited to the trained task rather than broader social functioning [192]. A recent study of high definition tDCS (n = 72) found improvements in social awareness primarily in children with typical sensory processing, highlighting the importance of individual differences in treatment response [180].

Safety profiles for tDCS appear favorable, with adverse effects typically limited to transient scalp discomfort or mild headaches, and no serious adverse events have been reported across studies [190]. However, systematic safety data in children remain limited, and the substantial individual variability in brain anatomy and current distribution raises questions about optimal dosing and targeting.

Deep Brain Stimulation (DBS)

DBS represents the most invasive neuromodulation approach and is reserved for patients with ASD who present with severe, treatment-refractory self-injurious behavior or aggression that poses an immediate risk of serious harm. For these patients, who have exhausted all conventional therapeutic options, DBS may offer a potentially life-saving intervention when the severity of symptoms outweighs the inherent surgical risks.

The evidence base for DBS in ASD, while limited, demonstrates encouraging results in carefully selected cases. The most rigorous evidence comes from a Phase I clinical trial involving six children (ages 7–14) with severe self-injurious behavior who underwent open-label DBS targeting the nucleus accumbens [193]. This study demonstrated feasibility and showed significant reductions in self-injury alongside meaningful quality-of-life improvements and metabolic reductions in the thalamus, striatum, and temporoinsular cortex on neuroimaging. Safety events included one asymptomatic intracranial hemorrhage and transient behavioral changes in three children during stimulation parameter adjustments. While other positive case reports or case series have been reported, including targets in the hypothalamus, globus pallidus, and internal capsule, there are no RCTs of DBS to date to allow a rigorous evaluation of efficacy.

Adverse effects occur in approximately 20 % of DBS cases and include hardware-related complications (lead migration, battery issues), transient mood or behavioral changes during stimulation adjustments, and rare instances of intracranial hemorrhage. The choice of target may also influence safety profile, highlighting the importance of surgical expertise and careful patient-specific planning. While risks are substantial, they must be weighed against the often-life-threatening nature of severe self-injurious behavior in this population.

Essential components include multidisciplinary evaluation involving neurosurgery, psychiatry, behavioral psychology, and ethics consultation, documentation of treatment failure with all conventional approaches, comprehensive informed consent addressing both risks and potential benefits, and access to long-term specialized care for device management and optimization. While the evidence base remains limited by small sample sizes and study design limitations, the consistent reports of meaningful improvement in this otherwise treatment-resistant population support continued careful investigation and compassionate use in appropriately selected cases.

Vagus nerve stimulation (VNS)

VNS involves surgical implantation of a device that delivers electrical stimulation to the vagus nerve, engaging multiple biological mechanisms, including brainstem arousal and neurotransmitter modulation, which may theoretically benefit the multisystem dysfunction observed in autism. The clinical evidence for VNS in ASD comes exclusively from patients with co-occurring epilepsy, where ASD was a secondary consideration rather than the primary treatment target. The largest study examined 77 patients with both ASD and intractable epilepsy from the VNS therapy patient outcome registry, comparing them to 315 patients with epilepsy alone [194]. While both groups showed similar overall quality-of-life improvements following VNS implantation, the autism group demonstrated significantly better mood improvement at 12 months post-implant. Importantly, there were no differences in most other quality-of-life variables between groups, suggesting that ASD does not negatively impact VNS efficacy for seizure control. These findings must be interpreted cautiously, given the retrospective, uncontrolled nature of registry data and the potential for reporting bias.

The theoretical rationale for VNS in ASD extends beyond seizure control to include modulation of circuits involved in mood, attention, and social behavior. VNS activates ascending projections to brain regions relevant to ASD pathophysiology and may influence neuroimmune dysfunction. However, these mechanistic hypotheses require empirical validation in autism-specific studies.

Critical limitations of the current evidence include the absence of controlled trials specifically targeting ASD symptoms, restriction of all clinical data to patients with comorbid epilepsy, and uncertainty about whether observed improvements reflect seizure control benefits rather than direct effects on autism features. The invasive nature of VNS implantation, requiring surgical device placement with associated risks of infection, hardware complications, and potential voice changes, must be weighed against the limited and indirect evidence for autism-specific benefits.

Limitations and future directions

Neuromodulation for ASD currently offers limited but important therapeutic options that vary dramatically by clinical indication and evidence quality [179,190]. For clinicians today, the evidence supports three distinct applications. TMS for comorbid depression follows established FDA-approved protocols with demonstrated efficacy comparable to neurotypical populations [181,185]. DBS for life-threatening self-injurious behavior can be effective in carefully selected cases, though it requires specialized centers and ethics oversight [193]. All other applications—including TMS and tDCS for core autism symptoms—remain investigational with insufficient evidence for routine clinical use [189].

The core challenge is treatment heterogeneity. While network meta-analyses show no consistent benefits for social communication or repetitive behaviors across ASD populations [190], emerging data reveal that specific subgroups respond to targeted interventions. Traditional mean group comparisons mask individual variability, as even measures with large effect sizes show 45–63 % of ASD cases performing within the typical range, and biologically plausible subgroups may exist despite small mean differences [195]. Patients with more severe adaptive deficits, particular sensory profiles, or specific neurophysiological markers may benefit from personalized protocols [180], but we lack validated methods to identify these responders prospectively. The path forward requires precision medicine approaches that match interventions to individual neurobiological profiles. Large-scale trials integrating neuroimaging, genetics, and electrophysiology with standardized behavioral outcomes will enable development of predictive algorithms. Identifying biomarkers that stratify biologically homogeneous populations is essential to realize the promise of precision medicine in ASD [196].

Neuromodulation will not replace established ASD interventions but may enhance their effectiveness in carefully selected patients. The field is transitioning from broad efficacy questions to precision targeting—a shift that promises more effective treatments for the individuals most likely to benefit while avoiding futile interventions in those who will not respond. Current neuromodulation research for autism faces a critical gap: studies consistently fail to identify which patients will respond to which interventions. The field requires precision approaches that predict treatment response through validated biomarkers rather than generic protocols. Large-scale TMS trials must distinguish true therapeutic effects from placebo responses in core autism symptoms. Standardized tDCS protocols and identification of responsive patient phenotypes will determine whether theoretical cost advantages translate into clinical benefits. VNS research beyond epilepsy populations can establish autism-specific outcomes through extended follow-up studies. Direct comparisons between these neuromodulation approaches remain absent, leaving treatment selection without empirical guidance. Head-to-head studies examining optimal timing, duration, and integration with behavioral therapies will address this knowledge gap. Implementation through specialized research centers with robust infrastructure remains necessary until definitive efficacy data becomes available.

Limitations and future directions

Current neuromodulation approaches for autism offer critical but limited clinical applicability. While TMS for comorbid depression and DBS for severe self-injury offer promising results in select cases, other applications, including TMS and tDCS for core ASD symptoms, remain investigational. A major limitation is treatment heterogeneity and the lack of validated methods to predict individual response. Most studies use group-level analyses that obscure biologically meaningful subgroups. Future progress depends on precision medicine approaches such as biomarker-based interventions that can stratify patients based on neurobiological profiles and predict treatment response. Large-scale trials integrating neuroimaging, electrophysiology, genetics, and standardized behavioral metrics are urgently needed. Direct comparisons between different neuromodulation modalities, treatment integration, and longer-term follow-up are essential to guide clinical use. Implementation through specialized research centers with robust infrastructure remains necessary until definitive efficacy data becomes available.

Gene-based therapies in autism spectrum disorder

For 10–20 % of people with autism there is a known genetic etiology, such as a single gene mutation (i.e., SCN2A) or a copy number variation (i.e., dup15q) [197]. More than 200 genes are implicated, including genes associated with ion channels, chromosome organization, cell cycle regulation, metabolism, synaptic function, and other cellular functions [197]. For this group of “syndromic autism”, direct repair of the genetic abnormality is an emerging therapeutic avenue, broadly outlined by the following three approaches.

Antisense oligonucleotides (ASOs)

Anti-sense oligonucleotides (ASOs) are short pieces of DNA that bind to RNA in the cell to modulate expression of proteins. There are already FDA-approved products that use ASOs to treat genetic defects in the central nervous system, such as nusinersin for spinal muscular atrophy and tofersen for SOD1-related amyotrophic lateral sclerosis. These drugs are given intrathecally at periodic intervals (every three months for nusinersin, monthly for tofersen). Nusinersin, as an example, is now foundational to care for spinal muscular atrophy with dramatically better motor outcomes and survival compared with historical and placebo controls [198].

For genetic disorders associated with autism, no ASO has received FDA approval; however, several candidates are in active clinical trials. Dravet Syndrome is caused by mutations in a sodium channel SCN1A, and causes refractory epilepsy and developmental delay. A quarter of affected children also have autism. [199]. An ASO for Dravet Syndrome (zorevunersen) is in a phase 3 trial, based on phase 1/2 data indicating 75 % reduction in seizures six months after the last dose, among children who received multiple doses [200]. As a second example, Angelman Syndrome is caused by dysfunction of the gene UBE3A, and also causes refractory epilepsy and developmental delay. Two in five children with Angelman have autism [201]. Multiple ASOs are in trial for Angelman Syndrome, including phase 3 studies (e.g. NCT06617429, NCT06914609), based on early promising improvements in neurodevelopment for both therapies.

Gene replacement (AAV)

Adeno-associated virus (AAV) gene replacement therapies use an engineered, non-replicating virus to deliver DNA to neurons and/or glia. The virus can be injected directly into the central nervous system via any of several routes (thecal, cisterna magna, cerebroventricular, or parenchymal) [202]. Recent advances include viruses that can be delivered intravenously and efficiently cross the blood barrier by binding to a human transferrin receptor [203]. Once in the target cells, the DNA forms an episome in the target cells, which then expresses the wild type protein of interest.

Clinical experience with these and other trials demonstrate that gene replacement therapies have important risks. AAV9 viruses have tropism for the liver, heart, and dorsal root ganglia, and injury to these tissues has been reported. There is also a risk for hematological complications (thrombocytopenia, thrombotic microangiopathy) and an overwhelming immune response such as hemophagocytic lymphohistiocytosis. These complications have led to deaths [204]. Clinical trials have adopted aggressive immunotherapy regimens and frequent blood tests to monitor these potential complications.

For central nervous system diseases, the FDA has approved intrathecal onasemnogene abeparvovec-xioi (Zolgesma) for spinal muscular atrophy. However, as with ASOs, there are no AAV therapies for disorders associated with autism. Several are in active development, though not as far along the regulatory pathway as for ASOs. For example, there are open AAV9 phase 1/2 trials for Dravet Syndrome (intracerebroventricularly; NCT05419492)and Rett syndrome (intrathecal; NCT05898620, NCT06152237, NCT05606614).

Gene editing

“Gene editing” refers to molecularly engineered tools that directly alter living cells by precisely cutting or modifying DNA. The current technologies are based on CRISPR-Cas9 systems. CRISPR (Clustered Regularly Interspaced Short Palindromic Repeats) is a sequence of RNA that targets a specific section of DNA in the genome, which is then cut by Cas9 (CRISPR-associated protein 9), a nuclease. Multiple rounds of refinement have improved the accuracy of targeting (reduced off-target effects), method of delivery (vector and nanoparticle systems), and breadth of editing (base editors and prime editors) [205,206].

The FDA has already approved therapeutics based on gene editing technologies performed on human tissue. Exagamglogene autotemcel (brand name Casgevy) increases expression of fetal hemoglobin in harvested autologous cells, which are then transplanted back into the affected individuals. Phase 3 trials demonstrated substantial reductions in vaso-occlusive crises for people with sickle cell disease [207] and transfusion dependence in people with beta-thalassemia [208]. A recent case report demonstrates that pathogenic mutations can be directly repaired in a living person. Clinical researchers at Children's Hospital of Philadelphia developed a patient-specific base editor for a child with carbamoyl-phosphate synthetase 1 (CPS1) deficiency, an urea-cycle disorder, which they administered via a lipid nanoparticle delivery system [209].

For genetic disorders associated with autism, several preclinical gene editing efforts are underway. Fragile X Syndrome, as an example, is caused by downregulation of the FMR1 gene due to excessive CGG repeats in a noncoding region of exon 1. Two pre-clinical efforts have demonstrated that removal of CGG repeats leads to increase expression of FMR1([210,211]). Preclinical models of gene editing have also shown promise for Rett Syndrome [212]and Angelman Syndrome [213].

Autism symptoms as outcomes, link to DEEs

Current gene-based therapies do not focus on core autism symptoms as a primary outcome, and thus future research is needed to understand if these therapies substantially improve autism features. The developmental and epileptic encephalopathies (DEEs) are a group of disorders characterized by seizures, developmental delay, and abnormal electroencephalographic findings that may further inhibit developmental progress. Dravet, Rett, and Angelman Syndromes are example DEEs, all with autism as a common feature. Thus, we anticipate early indications of the possibility of improved autism core features on children with DEE who receive precision genetic therapies [214].

Discussion

This review critically evaluates emerging evidence across five key domains in autism therapeutics: psychopharmacology, behavioral and psychosocial interventions, digital and AI-driven tools, neuromodulation, and genetic therapies. These advances mark a critical turning point, as the field begins to confront the limitations of generalized treatments. Together, the advances discussed illustrate how the field is beginning to address the complex realities of autism with greater scientific rigor and nuance. While challenges remain, the rapid expansion of diagnostics, genetics, and related fields has begun to make early, precise diagnosis and targeted intervention increasingly feasible. As a result, there is a growing consensus that precision medicine, grounded in individualized profiles, represents the future of ASD care [215]. These fundamental developments have challenged each writer to not only consider recent advances but also how each innovation is grappling with diversity in clinical presentation, biological mechanisms, and treatment response.

At the same time, individuals with autism, their families, and advocates are playing a more active role in shaping autism research and care [216]. The rise of self-advocacy and the neurodiversity movement has brought renewed attention to language, identity, and the diversity of lived experiences. Simultaneously, the concept of “profound autism”, used to describe individuals with significant intellectual and communicative challenges, has drawn attention to the need for tailored approaches that reflect both biological and practical realities [217]. These developments highlight that addressing diversity in autism requires not only scientific exploration, but also a deeper understanding of stakeholder perspectives and lived experiences.

Behavioral and psychosocial interventions

While pharmacological strategies often target associated symptoms or underlying neurobiology, psychological and behavioral interventions remain foundational in autism care, especially for supporting core challenges in social communication, behavior, and adaptive functioning. While traditional approaches such as Applied Behavior Analysis (ABA) remain widely used and can be particularly important for individuals with profound autism and high support needs, ABA has faced growing criticism from autistic individuals and neurodiversity advocates, particularly regarding ethical concerns and the suppression of authentic behaviors, fueling a shift toward more person-centered and affirming models. Indeed, there has been movement in the field beyond highly structured, clinician-driven models toward more naturalistic, developmentally informed, and person-centered approaches. This shift reflects not only advances in behavioral science but also the influence of self-advocacy, neurodiversity perspectives, and an increased focus on individualized goals and quality of life. In this section, the authors examine the evolving landscape of behavioral and psychosocial interventions for ASD, with attention to both traditional evidence-based practices and emerging innovations that seek to address the diverse needs of autistic individuals and their families. Despite these shifts, few fundamentally new psychological or behavioral models have emerged in recent years. Most advances have centered on adapting established, evidence-based practices through new delivery modes, cultural adaptation, or a greater focus on person-centered and neurodiversity-affirming models. Expanding access to services and adopting person-centered models of care are now priorities, but more rigorous research is still needed to evaluate innovative interventions and clarify the role of augmentation with technology.

Psychopharmacology

Advances in psychopharmacology have long been at the forefront of efforts to address both core and associated features of ASD [218]. Notably, the FDA approval of risperidone and aripiprazole for the treatment of irritability associated with ASD has had a marked impact on a category of symptoms that can be profoundly disabling for many individuals and their families. Yet, these and other medications primarily target co-occurring symptoms. New research has spurred the search for disease-modifying treatments that move beyond symptomatic relief, with growing emphasis on stratifying patients by biological and clinical characteristics. Recent research is shifting toward mechanistically informed and stratified approaches, including trials of agents targeting specific neurotransmitter systems (e.g., glutamate, GABA) and pathways implicated in subgroups of autistic individuals. While several novel compounds and adjunctive therapies have shown promise for certain associated symptoms, no pharmacologic agent has yet shown consistent, clinically meaningful improvement in core autism features.

Technology, artificial intelligence, and digital therapeutics

The rapid rise of artificial intelligence and digital tools is reshaping autism research and care. Some innovations in this space, such as the use of eye-tracking devices for diagnostic biomarkers, were initially ahead of their time, but are now seeing broader adoption as technology advances and evidence grows. AI-driven approaches including machine learning-based behavioral analysis are making earlier and more precise diagnosis increasingly feasible. Virtual reality environments are being used to create immersive, safe settings for individuals to practice social and daily living skills. Meanwhile, app-based platforms and telehealth services have expanded significantly, particularly since the COVID-19 pandemic, improving access to care for many.

These tools are helping to reduce longstanding barriers, such as provider shortages and access gaps in rural or underserved areas, while offering new modes of personalized support, and are being integrated into both home and clinical settings to support autistic individuals and families across the lifespan. However, these advances also bring new challenges, including the need for rigorous scientific validation, AI safeguards, equitable access, and careful attention to privacy, data use, and user autonomy. As digital therapeutics continue to evolve, meaningful progress will depend on sustained collaboration among clinicians, technologists, autistic individuals, and families to ensure technology delivers real-world, measurable benefits.

Neuromodulation

As the search for effective autism interventions expands, neuromodulation has emerged as a promising frontier. Techniques such as transcranial magnetic stimulation (TMS) and transcranial direct current stimulation (tDCS) are being actively investigated for their potential to modulate neural circuits implicated in ASD. TMS, in particular, has a well-established role in treating depression and other neuropsychiatric conditions, and this clinical track record has prompted investigation into its therapeutic potential in autism, particularly for core features like social cognition. Deep Brain Stimulation (DBS), though far more invasive, is also under preliminary investigation for individuals with severe, treatment-resistant behaviors, offering a potential lifesaving therapy in the most extreme cases.

In ASD, these non-invasive brain stimulation approaches remain largely experimental, with research focused on moving beyond symptom management to directly address core features such as social communication and executive function. Studies increasingly emphasize personalized protocols and recognition of individual differences in treatment response, as well as safety and ethical considerations. While early results are promising for some outcomes, reproducibility, optimal dosing, and identifying the individuals most likely to benefit remain ongoing challenges. As evidence accumulates, neuromodulation serves as both an example of innovation in ASD therapeutics and a reminder of the complexities involved in translating neuroscientific advances into clinical care.

Genetic therapies

Recent advances in human genetics have transformed our understanding of autism as a highly heterogeneous condition shaped by both rare, high-impact mutations and a wide array of common risk variants. These insights have enabled the development of gene-targeted therapies for syndromic forms of ASD, such as Fragile X, Rett, and Phelan-McDermid syndromes. Approaches like gene replacement and antisense oligonucleotides are progressing from preclinical stages to early human trials, though many remain in proof-of-concept phases focused on safety and efficacy. For the broader autism population, where genetic risk is more diffuse, therapeutic translation is still in its infancy. While enthusiasm for genetic therapies is high, challenges including safety, efficacy, and equitable access persist. Even so, these efforts mark a pivotal move toward precision medicine in autism, with continued progress relying on collaboration among researchers, clinicians, families, and autistic individuals.

Conclusion and future directions

Across all therapeutic domains, common challenges persist. Identifying reliable biomarkers, developing sensitive and meaningful outcome measures, and ensuring equitable access remain significant obstacles to translating scientific progress into real-world impact. Overcoming these barriers will be crucial to translating scientific progress into meaningful, scalable solutions for individuals and families affected by autism.

Recent advances in pharmacology, genetics, neuroscience, behavioral science, and digital health are reshaping ASD therapeutics. Pharmacological research is increasingly focused on targeted and stratified interventions, while behavioral approaches are moving toward individualized, developmentally informed models. Technology-enabled solutions including AI-driven diagnostics and digital therapeutics are expanding access and supporting both assessment and intervention. Meanwhile, neuromodulation and gene-based therapies represent innovative, though still early-stage, strategies for addressing core symptoms and underlying etiologies.

Despite this progress, the profound heterogeneity of ASD continues to complicate clinical research and practice, often obscuring treatment effects and limiting generalizability. Integrating stakeholder perspectives especially those of autistic individuals, families, and those with profound autism remains essential for defining meaningful outcomes and guiding future directions.

Moving forward, the field must prioritize large-scale, stratified studies; invest in the development of robust biomarkers and outcome measures; and build systems to ensure equitable access and implementation. Achieving the promise of precision medicine in autism will require not only scientific rigor but also sustained collaboration across disciplines and true engagement with the autism community. Only through this comprehensive approach can future interventions reflect the complexity and diversity of autism and ensure meaningful, measurable outcomes for individuals and families across the autism spectrum.

Author Contributions

• 1.Behavioral & Psychological Interventions - Hallie Brown, PhD, Michelle Gorenstein, PsyD;
• 2.Emerging Targets in Autism Pharmacology – Pankhuree Vandana, MD; Sara Daniella Kevelson, MD
• 3.Novel Technology-Based Therapies for Autism Spectrum Disorder – Fenil Patel, MD; Ernie Pedapati, MD, PhD
• 4.Neuromodulation in Autism Spectrum Disorder: Emerging Therapeutic Frontiers – Rana Elmaghraby, MD, Ernie Pedapati, MD, PhD
• 5.Gene based therapies in autism spectrum disorder - Jennifer Bain, MD§, PhD, Zachary Michael Grinspan, MD MS
• 6.Ernie Pedapati contributed to conceptualization and drafting of the final discussion section.
• 7.Pankhuree Vandana, Sara Daniella Kevelson, and Jeremy Veenstra-VanderWeele contributed to the conceptualization, drafting, and editing of the entire article.

Declaration of generative AI and AI-assisted technologies in the writing process

During the preparation of this work the author(s) used Claude Opus 4.1(Anthropic PBC; San Francisco, CA), Elicit.com and ChatGPT (Open AI) to conduct systematic literature review, complete data extraction of relevant articles, general research assistance and exploratory concept development. Grammarly (Grammarly Inc; San Francisco, CA) was used to improve readability and to check grammar. These tools served as research assistants and did not replace critical thinking, clinical judgment, or original analysis After using this tool/service, the authors reviewed and edited the content as needed and take full responsibility for the content of the publication.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.