This systematic review comprehensively evaluated the efficacy and underlying mechanisms of both pharmacological and microbiota-based therapeutic strategies in preclinical models of ASD. We specifically focused on behavioral outcomes, which highlights the functional significance of the findings. The reviewed studies have demonstrated that a variety of interventions can be effectively utilized to ameliorate typical ASD behavioral deficits. Despite heterogeneity in ASD animal models and experimental design, most studies reported significant improvement in at least one key ASD behavioral trait. Together, these findings highlight the diverse nature of ASD pathophysiology, in which inflammatory, metabolic, and synaptic mechanisms all contribute to the behavioral symptom presentation (Fig. 4). This mechanistic diversity strengthens the growing recognition that treatment strategies extending beyond the CNS, particularly those involving the gut microbiota or peripheral metabolic pathways, may offer complementary therapeutic benefits.

Among all the pharmacological and microbiota-based interventions reviewed, treatment with oxytocin has been the most extensively investigated. Eight studies have assessed its direct impact on the amelioration of core ASD symptoms. The rationale for this strategy lies in both the physiological profile of oxytocin and clinical observations indicating lower circulating oxytocin levels in children with ASD compared to neurotypical individuals. However, these differences diminish during adolescence and adulthood [103, 104]. This age-dependent variability in oxytocin signaling may delineate a developmental window in which such therapy could be most effective.

In the preclinical studies, exogenous oxytocin administration proved to be a promising therapeutic approach, effectively enhancing social behaviors while reducing repetitive behaviors in rats prenatally exposed to VPA [31, 41] and BALB/cByJ mice [39]. In other ASD models, such as 15q dup mice [34], Oprm1-KO mice [42], Oxtr-KO mice [33], and C58/J mice [38], oxytocin selectively improved social deficits, without influencing other ASD symptoms. Notably, in models including 16p11.2^+/-, Fmr1-KO, and Shank3-KO mice, oxytocin failed to rescue ASD-like phenotypes [40]. An intriguing observation was reported in Oprm1-null mice, in which prior studies identified oxytocinergic dysfunction. Chronic intranasal oxytocin administration improved social behaviors, whereas the same regimen in wild-type mice induced pronounced and lasting social impairments [42, 105]. This suggests that a pre-existing deficit in the oxytocinergic system may be a prerequisite for beneficial outcomes of oxytocin treatment. Supporting this, VPA-exposed rats characterized by reduced cerebrospinal oxytocin levels and a decreased number of oxytocin-immunoreactive cells in the hypothalamus exhibited normalization of these parameters as well as rescue of the ASD-like phenotype following treatment [31].

Other important factors influencing oxytocin’s therapeutic efficacy may include the developmental stage, treatment schedule, and environmental context. Acute intranasal oxytocin improved social behavior in Oprm1-KO mice within 5 min after administration [42]. Nevertheless, acute exogenous oxytocin supplementation produced only a temporary prosocial effect [106, 107], likely due to short half-life of oxytocin in the brain [108, 109]. In contrast, treatment during the early postnatal period had a long-term benefit on sociability, which was observed in VPA-exposed rats during adolescence and in adult 15q dup mice [31, 34]. These findings confirm that repeated oxytocin exposure during sensitive periods may upregulate endogenous oxytocinergic function and remodel social neural circuits [31, 38]. Furthermore, social-contextual administration enhanced and prolonged oxytocin’s prosocial effects in Oprm1-KO mice [42]. Sex differences represent another crucial aspect in the optimization of ASD therapies. While in male rats in the VPA model, exogenous oxytocin was effective in treating core ASD symptoms due to deficits in the oxytocin pathway [31], in females in the same model (characterized by increased hippocampal and cortical levels of oxytocin and OXTR), the OXTR antagonist Atosiban demonstrated therapeutic efficacy [43]. Liu et al. [43] linked this effect to restoration of synaptic plasticity impairments in females. These findings suggest sex-specific pathophysiology and the utility of oxytocin antagonists in selected subgroups [43].

The pharmacological limitations of oxytocin, including its short half-life and poor blood-brain barrier penetration, have prompted the search for more bioavailable synthetic analogues, as well as research into its active metabolites [38, 39]. Moy et al. [39] reported that two synthetic OXTR agonists, TC-OT-39 and carbetocin, despite showing activity in vitro, did not show prosocial effects in BALB/cByJ mice, although TC-OT-39 reduced marble-burying behavior. In contrast, the oxytocin metabolite OT(4–9) enhanced sociability in BALB/cByJ mice in a dose-dependent manner, with effects persisting up to 12 days after treatment, though without changes in repetitive behaviors [39]. This dissociation suggests that some prosocial effects of oxytocin may be mediated by downstream metabolites acting through noncanonical mechanisms beyond OXTR activation.

Beyond behavioral outcomes, mechanistic investigations have linked oxytocin treatment to modulation of neurotransmitter systems, including serotonin and dopamine, and to neuroprotective effects, such as reduced expression of the necroptosis markers RIP3 and MLKL in the hippocampus and amygdala of VPA-exposed rats [41]. These molecular alterations may underlie improvements in cognitive function and social memory. Examples include a study indicating functional interactions between oxytocin and serotonin systems in shaping social behavior. Supporting this interaction, in 15q dup mice, postnatal activation of 5-HT_1A receptors normalized sociability in a manner blocked by OXTR antagonism, indicating an interaction between serotonin and oxytocin in the developmental programming of social circuits [34].

Despite mostly promising preclinical data, meta-analyses of clinical trials indicate that oxytocin treatment does not significantly improve impaired social and repetitive behaviors in adults with ASD [110], with only a moderate effect observed in children [111]. Collectively, these analyses indicate the need for more large-scale, rigorous, and multi-site randomized controlled trials to confirm the effectiveness of oxytocin as a treatment for ASD [110, 111]. Equally critical is the identification of biomarkers capable of defining ASD subgroups in whom intervention with oxytocin, a synthetic analog or even an antagonist of OXTR, might yield beneficial outcomes, given the heterogeneity in ASD etiology, phenotype, and treatment response [112, 113].

The therapeutic success of another large group of interventions often depends on correcting the underlying neurochemical disturbances observed in patients with ASD, in particular the excitation/inhibition (E/I) imbalance, and modulating key neurotransmitter systems [45, 114]. The efficacy of this approach has been demonstrated in studies showing improvements in social behaviors achieved by inhibiting excitatory activity through the use of NMDA receptor antagonists, e.g., subanesthetic doses of Xenon [50] and memantine in VPA-exposed rats [51] and NitroSynapsin in Mef2c^+/− mice [52], as well as through modulation of AMPA receptors by risperidone in a VPA model [53]. In addition to restoring glutamate/GABA balance, additional mechanisms may contribute to the observed behavioral effects. For example, therapy may work by enhancing CREB/BDNF signaling, increasing hippocampal mRNA level of Glt-1, thereby preventing neuronal apoptosis and reducing oxidative stress, as shown in the case of memantine treatment [51]. The results of a recent randomized clinical trial in which memantine was administered for 12 weeks confirmed the observations from preclinical studies, significantly improving social impairments in youths with ASD [115].

Beyond targeting ionotropic receptors, several studies have also reported the alleviation of core ASD symptoms through modulation of mGlu receptor activity, particularly mGlu4 and mGlu5 receptors [54–56]. Their involvement may result from their regulatory role in synaptic plasticity, which is impaired in numerous ASD models [116]. Similar to disturbances in glutamatergic signaling, preclinical and clinical data point to the primary inhibitory neurotransmitter of the CNS as important for a better understanding of the pathogenesis of ASD and for identifying new therapeutic strategies [117]. Thus, the use of PAMs of GABA-A receptors (MP-III-022 and 2–261) has been shown to alleviate symptoms characteristic of ASD in environmental and idiopathic models of ASD [53, 55].

In addition to modulating the two main neurotransmission pathways, other neurotransmitters, including dopamine, serotonin, and choline, should not be overlooked. The dopamine system, known for its role in reward, motivation, and memory, and serotonin, important in the regulation of anxiety and neurodevelopmental processes, have been implicated in ASD pathophysiology [118, 119]. Preclinical studies using intranasally administered dopamine have demonstrated therapeutic effects in two different mouse models of ASD - BTBR and Fmr1-KO mice [60]. Activation of serotonin receptor 7 (5-HT₇R) using its agonist LP-211 showed beneficial effects on anxiety-like behaviors, social and cognitive deficits, in the VPA model [62]. However, a specific limitation of studies focusing on monoaminergic signaling is the omission of tests assessing aggressive behaviors (characteristic of patients with ASD) that appear to be modulated by dopamine and serotonin in individuals with ASD [59]. Furthermore, allosteric modulation of α7 nAChR with AVL-3288 alleviated social approach deficits and repetitive grooming behaviors in a BTBR mouse model [56]. The efficacy of compounds with such a broad spectrum of mechanisms of action confirms the complexity of ASD as a disorder [120]. It highlights the enormous challenges of developing targeted and safe and pharmacotherapy.

Gut-microbiota-based interventions represent a promising strategy for the treatment of neuropsychiatric disorders, including ASD. Although mechanistic understanding is only emerging, the importance of gut microbiota in ASD is supported by lower microbiota diversity in ASD patients [121]. Our analysis included six studies utilizing both targeted (probiotics) and broad-spectrum (FMT) intervention strategies. FMT-based strategies proved effective in reducing key ASD symptoms and inducing extensive metabolomic changes [70, 71]. However, a number of limitations, such as safety concerns for different populations, unclear duration of positive changes, and difficult donor selection limit the clinical application [122]. Despite these limitations and differences in protocol, FMT demonstrated more consistent improvements in social behavior compared to targeted interventions. Furthermore, recent methodological improvements in FMT delivery, particularly the development of encapsulated formulations, have reduced procedural complexity and enhanced standardization and repeatability [123].

Targeted microbiota interventions have a more predictable safety profile and higher translational significance [124]. However using individual bacterial strains produced modest [67] or limited benefits to a single behavioral trait [68]. Selected studies focused on classical probiotics (Lactobacillus and Bifidobacteria spp.) and future studies utilizing next-generation probiotics with a higher potential to synthesize biologically relevant metabolites could have better treatment outcomes [125]. Combining prebiotics and probiotics to achieve a synergistic synbiotic effect appears to be more effective strategy. Administration of Lactobacillus reuteri C501 with inulin improved sociability, social novelty, and reduced repetitive self-grooming [69], indicating a more effective synergistic effect. Furthermore, specific bacterial and prebiotic interventions are generally well tolerated, easily standardized and repeatable, and present a lower practical and ethical barrier compared to FMT. Overall, microbiota-based interventions show promise in alleviating behavioral deficits associated with ASD though further mechanistic studies are required to establish molecular pathways underlying these beneficial effects.

The effectiveness of microbiota-based interventions highlights the importance of the host’s metabolic profile on the CNS. The consistent improvement in sociability and repetitive behaviors following metformin and pioglitazone treatment supports the effectiveness of metabolic modulators to treat both metabolic and behavioral symptoms of ASD. These effects are likely achieved through a reduction of oxidative stress, which is elevated in ASD patients [126]. Both metformin and pioglitazone were able to downregulate pro-inflammatory mediators in multiple brain regions, indicating a global anti-inflammatory effect. Additionally, both treatments had a positive effect on glutathione levels, suggesting regulation of redox activity. Overall, it can be theorized that alleviation of behavioral deficits was induced by a broad oxidative neuroprotective effect rather than modulation of neuronal activity or a single neurotransmitter pathway.

Metformin and pioglitazone exerted their beneficial effects without adverse metabolic effects, indicated by no changes in body weight during or after the treatment. This is particularly important since most of the antipsychotic medications used in ASD patients carry a substantial metabolic burden with well-known weight gain effects [127]. Naturally, the combination of metabolic agents and antipsychotic drugs becomes a potential strategy for reducing the metabolic side effects of current treatments. The synergistic effect of combined metformin and risperidone therapy in reducing levels of proinflammatory cytokines more effectively than either of these drugs alone, supports this theory [75]. These findings bring importance to systemic pathophysiology of ASD and the potential for metabolic drugs in a clinical setting.

Cannabinoid-based treatments represent another emerging strategy that targets the neuromodulatory and inflammatory components of ASD pathophysiology. The ECS regulates social behavior, synaptic plasticity, and neurodevelopment in general [128]. Current preclinical studies indicate that both phytocannabinoids (CBD, CBDV, THC) and complex cannabinoid mixtures (Avidekel, Erez oils, JZP541) can modulate ASD-like behaviors. However, these effects appear to be highly dose-dependent and cannabinoid-specific. Higher doses are reported to lower activity levels. The behavioral outcomes are caused by extensive transcriptional and neuroimmune changes that suggest a broad modulatory role on neuronal excitability and neuroinflammation. These transcriptional changes are especially prominent in the PFC and cerebellum with hundreds of affected genes. Most of these changes are connected to synaptic signaling and neural excitability, which are important molecular mechanisms affected in ASD. While promising, these findings do not address challenges in translating cannabinoid therapies to clinical practice, including regulatory and societal constraints, highlighting the need for changes in policy and public perception.

Analysis of selected studies suggests that interventions targeting the purinergic system, inflammation, and vitamin metabolism can yield measurable behavioral benefits in models of ASD. The common denominator of these approaches is the goal of restoring balance between metabolic, immunological, and neuroplastic processes. These results support the increasingly well-established systems approach, which views ASD as a disorder encompassing interdependent biological networks, ranging from mitochondria to the immune system to the gut-brain axis [129, 130]. Antipurinergic interventions based on suramin administration appear to act by inhibiting the excessive signaling of ATP and related nucleotides released from stressed cells. This finding is consistent with the ”cell danger response” and ”hyperpurinergy” hypotheses as possible pathogenic mechanisms of ASD [129]. Experimental studies report that suramin treatment improves social behavior while modulating mitochondrial gene expression and P2-receptor activity, suggesting coordinated regulation between purinergic metabolism and mitochondrial function. In contrast, anti-inflammatory approaches have produced the most consistent outcomes, including reductions in pro-inflammatory cytokines (IL-1β and TNF-α), elevations in anti-inflammatory mediators (IL-10 and LXA4), and structural restoration within the hippocampus and cerebellum, particularly increased neuronal density and Purkinje cell number. These neurobiological improvements align closely with enhanced social and cognitive performance, underscoring neuroinflammation as a promising therapeutic target [93, 96, 97]. Effective therapeutic paradigms appear to converge on modulation of the mitochondrial function, purinergic signaling, and neuroinflammation, often supplemented by nutritional regulators such as vitamin D₃ and retinoids. This integrative perspective supports the notion that persistent disruption of metabolic, immune, and nutritional signaling may sustain the ASD phenotype. It is also worth considering the roles of the gut-brain axis and the microbiota-immune-brain axis in modifying inflammation, gut barrier integrity, and neuroimmune signaling [131]. Dysbiosis and increased intestinal permeability (“leaky gut”) may amplify neuroinflammatory state of the brain, further justifying the need for research into interventions combining metabolic and microbiota-based therapies.

The current evidence base supports the idea that multi-pronged therapies targeting cellular metabolism, inflammation, and vitamin metabolism simultaneously have the greatest potential for translating into effective treatments for ASD. However, multiple limitations in preclinical studies reduce the significance of current findings. Less than 6% of the evaluated studies included male and female subjects. The absence of a standardized approach to examining both sexes in parallel limits the understanding of sex-related differences underlying the varying prevalence of ASD in boys and girls, as well as the potential for sex-dependent individualization of therapy within the development of personalized medicine. Another limitation is that many studies included in this review rely on single-dose designs; incorporating multiple-dose regimens and dose–response analyses could reduce potential misinterpretations and improve the robustness and interpretability of the findings, as study design critically influences effect estimation and translational validity in animal research [132]. An important point to emphasize here, in the context of current preclinical research on ASD, is that many behavioral tests assess domains that are relevant but not specific to autism alone. Commonly used measures, such as deficits in social interaction or repetitive behaviors, are present in many neuropsychiatric disorders, which limits the construct and face validity of the models used [102, 133]. This lack of specificity complicates the interpretation of treatment effects and underscores the need to develop more refined, translation-oriented behavioral paradigms. Additionally, a lack of standardization was observed in selecting and applying behavioral tests to assess ASD-like symptoms, which limits the ability to properly address key ASD behavioral deficits. Addressing these challenges would improve scientific rigor and increase the translational significance of interventional studies. It should also be emphasized that the role of the gut microbiota in ASD remains the subject of lively debate, and recent studies point to significant conceptual and methodological limitations, including issues related to causality, reproducibility of results, and confounding variables, which should also be kept in mind when interpreting the existing research findings, particularly with respect to causal inference and reproducibility [134]. Addressing these challenges will likely require improved stratification of ASD models, increased use of cross-model validation, and a stronger focus on mechanistically informed and translationally relevant endpoints [¹³⁴]. While the presented preclinical findings constitute a significant contribution to the search for novel ASD therapies, it is crucial to exercise caution against premature clinical extrapolations that might detract from established evidence-based interventions, including behavioral and educational support [3, 135].