Microbiota-gut-brain axis in autism spectrum disorder: integrating brain structure, function, and transcriptomics

Abstract

Growing evidence indicates that disruption of the microbiota–gut–brain (MGB) axis is a key factor in autism spectrum disorder (ASD), affecting neurodevelopment, neural circuit function, and behavior. This review synthesizes multidisciplinary data to clarify mechanistic links between the MGB axis and ASD and to evaluate microbiota-targeted therapeutic strategies. We conducted a narrative synthesis of clinical and translational studies, including human cohort and case-control investigations, animal models, multi-omics analyses, immune profiling, multimodal brain assessments (structural/functional MRI and transcriptomics), and interventional trials. Emphasis was placed on evidence of microbiota–brain associations, intervention outcomes, and methodological limitations. Key findings reveal that individuals with ASD commonly exhibit gut microbiome dysbiosis and altered metabolomic signatures that can influence central nervous system function; three core bidirectional signaling routes link gut microbes to brain outcomes—microbial metabolite production (e.g., short-chain fatty acids, tryptophan metabolites), immune-mediated pathways, and neuroendocrine-vagal communication. Accumulating data associate ASD-related microbial profiles with changes in brain structure, functional connectivity, and transcriptomic patterns, supporting a mechanistic role for the MGB axis in ASD phenotypes. Microbiota-targeted interventions show promising effects on gastrointestinal symptoms, metabolic biomarkers, and selected behavioral measures in small studies, but results are heterogeneous and current evidence is insufficient for widespread clinical application. Integrating multimodal neuroimaging with multi-omics and machine learning provides a promising framework to identify reproducible microbial–brain biomarkers for early detection, clinical subtyping, and stratified treatment. Key challenges include methodological heterogeneity, limited causal inference, small and heterogeneous cohorts, and ethical/safety concerns for pediatric interventions. The MGB axis is a plausible pathogenic mechanism and therapeutic target in ASD. Translating mechanistic insights into precision clinical applications requires standardized, multicenter, longitudinal deep-phenotyping studies that combine multimodal imaging, comprehensive multi-omics, rigorous randomized trials, and careful ethical oversight.

Introduction

Autism spectrum disorder (ASD) is a neurodevelopmental condition characterized by persistent impairments in social communication and the presence of restricted, repetitive patterns of behavior. Its global prevalence has been steadily increasing, currently affecting more than 1% of the population worldwide, with recent estimates in the United States indicating a diagnosis rate of approximately 1 in 31 individuals [1]. Although the precise etiology remains incompletely understood, ASD is thought to arise from complex interactions between genetic susceptibility and environmental factors. Current diagnostic practice primarily relies on the behavioral criteria outlined in the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5), which are inherently subjective and are not fully applicable to very young infants and toddlers who cannot be reliably assessed behaviorally. The lack of objective biomarkers further complicates both accurate diagnosis and the development of targeted treatment strategies. In recent years, advancements in the study of the microbiota-gut-brain axis have opened new windows for the early diagnosis and treatment of ASD [2, 3]. Multi-omics studies, including neuroimaging, have provided insights into the potential for precision medicine. These studies suggest that combining gut microbiota interventions with brain-targeted therapies could offer a comprehensive treatment approach for ASD. This review examines microbiome–gut–brain (MGB) axis alterations in individuals with ASD, emphasizing changes in brain structure, function, and transcriptional profiles. As delineated in the conceptual framework (Fig. 1), we also integrate brain-directed interventions—such as transcranial magnetic stimulation (TMS) and deep brain stimulation (DBS)—into a broader translational perspective, which is further explored in Section "Interventional therapies targeting the gut microbiota in ASD".

Fig. 1. Overview of the microbiota-gut-brain axis in ASD.

A Brain-directed therapies, such as transcranial magnetic stimulation and deep brain stimulation, which are discussed in detail in Section 6. B Microbiota-targeted treatments, including probiotics, fecal microbiota transplantation, and microbiota transfer therapy, as well as dietary or pharmacologic interventions such as prebiotics, synbiotics, dietary modification, and antibiotics. C Assessing the impact of gut microbiota and their metabolites on the brain through metabolic, immune, and neurotransmitter pathways, using imaging techniques that include brain structure, function, networks, brain wave spectra, and transcription. D Evaluating the therapeutic efficacy of gut microbiota - targeted brain interventions through behavioral assessments and neuroimaging examinations.

Methods

We conducted a structured narrative review of the microbiota–gut–brain axis in ASD, with particular emphasis on brain structure, function and transcriptomic mechanisms. Because the review integrated both human and preclinical evidence, the review question was framed using an adapted PICO/PECO-informed approach rather than a strict intervention-based PICO framework: population (individuals with ASD or ASD-relevant animal models), exposure/intervention (gut microbiota, microbiota-derived metabolites, or microbiome-related interventions), comparator (neurotypical/typically developing or wild-type controls, baseline measures, or comparator groups where applicable), and outcomes (brain-related structural, functional, transcriptomic, molecular, behavioral, and/or ASD-relevant clinical outcomes). A structured literature search was performed in PubMed for articles published from 1 January 2010 through 1 October 2025. The search combined MeSH terms and free-text keywords related to ASD (e.g., “Autism Spectrum Disorder,” autism*, ASD, neurodevelopmental disorder*), gut microbiota/metabolites/interventions (e.g., “Gastrointestinal Microbiome,” microbiota*, microbiome, gut microbiota, intestinal microbiome, fecal microbiota transplant*, short-chain fatty acid*, SCFA), and neurobiological measures (e.g., brain, magnetic resonance imaging, diffusion tensor imaging, neuroimaging, transcriptome, gene expression). The complete PubMed search strategy is provided in Supplementary Table S1. To reduce the likelihood of missing relevant studies, manual screening of reference lists and citation searching of key reviews and original articles were also performed. Records were screened in two stages (title/abstract screening followed by full-text review) against prespecified eligibility criteria. We included original human and preclinical studies directly relevant to ASD and the microbiota–gut–brain axis, with a focus on brain-related structural, functional, transcriptomic, or mechanistic outcomes. Studies without direct neuroimaging or transcriptomic measures were also retained when they reported ASD-relevant behavioral, clinical, or neurobiological molecular outcomes directly pertinent to the predefined scope of the review. Reviews, editorials, conference abstracts, and studies not directly relevant to these domains were excluded. Data extracted from included studies comprised study type, participant or model characteristics, sample size, microbiome assessment, brain or molecular assessment, and principal findings; these are summarized in Supplementary Table S2. Given the substantial heterogeneity across study designs, species, exposures, and outcome measures, findings were synthesized narratively by grouping studies according to major neurobiological domains, including brain structure/connectivity, brain function/activity, and transcriptomics/molecular mechanisms, while also integrating relevant microbiome and metabolite findings from both clinical and preclinical literature. Because of the broad and heterogeneous scope of the review, a formal risk-of-bias assessment was not performed; however, the evidence was interpreted with attention to study design, sample size, and consistency across studies. To improve reporting transparency, the study selection process is presented using an adapted PRISMA 2020 flow diagram (Fig. 2).

Fig. 2. Adapted PRISMA 2020 flow diagram of the study selection process.

The workflow illustrates the systematic identification, screening, eligibility assessment, and final inclusion of literature for this review.

Gut microbiota and metabolic alterations in ASD patients

Multiple studies indicate that children with ASD harbor altered gut microbial communities, with changes detectable early (particularly before age three) and in some cohorts correlating with symptom severity, suggesting potential early diagnostic value [4–8]. Taxonomic shifts reported include phylum‑level increases in Bacteroidetes and Proteobacteria and decreases in Actinobacteria (notably Bifidobacterium), a marked reduction of Bifidobacterium longum and paradoxical increases in late colonizers and butyrate producers such as Faecalibacterium prausnitzii in certain young cohorts [9]. At the genus and species levels, multiple studies and meta‑analyses have reported higher relative abundance of taxa such as Bacteroides, Parabacteroides, Clostridium, Desulfovibrio, Sutterella, and Lactobacillus, while taxa including Bifidobacterium, Prevotella, Blautia, Dialister, Veillonella, and Akkermansia have been reported as depleted in other studies [10–13]. Gastrointestinal phenotypes also co‑vary with microbiota: for example, constipation in ASD has been linked to elevated Escherichia/Shigella and Clostridium cluster, and greater fungal (Candida) dispersion has been observed in some cohorts [14]. In children with autism, metabolic changes associated with the gut microbiota have been observed in their blood, urine, and stool [15]. These alterations include fluctuations in short-chain fatty acids (SCFAs), indoles, and lipopolysaccharide-related signals [16]. Notably, fecal acetic acid levels have been linked to the severity of ASD in toddlers [17]. Impaired intestinal and blood–brain barriers may allow microbial metabolites to enter the systemic circulation, thereby affecting neurotransmitter-related pathways (e.g., serotonin, dopamine, gamma-aminobutyric acid [GABA]) and other neuroimmune and metabolic processes [18, 19]. Crucially, the impact of the gut microbiota on host neurotransmitter availability (including serotonin, dopamine, and GABA) is a multifaceted process. Rather than acting merely as peripheral ‘producers’ via direct de novo synthesis, gut microbes orchestrate host metabolic homeostasis by regulating biosynthetic precursors and signaling through specialized cells, such as enterochromaffin cells. Longitudinal evidence also suggests an atypical trajectory of microbiota maturation in ASD. Specifically, α‑diversity (a metric reflecting the richness and evenness of microbial species within a single sample) fails to exhibit the age-related increase typically observed in neurotypical children [20]. To provide a clearer synthesis of these complex findings, representative bacterial taxa recurrently altered in ASD, along with their predominant direction of change, are summarized in Table 1.

Table 1.

Key bacterial taxa consistently dysregulated in ASD.

Taxon	Level	Predominant reported direction in ASD	Comment
Bacteroidetes	Phylum	↑	Frequently reported as increased at the phylum level
Proteobacteria	Phylum	↑	Often reported as enriched in ASD cohorts
Actinobacteria (notably Bifidobacterium)	Phylum / Genus	↓	Reduced abundance is commonly reported, with depletion of Bifidobacterium noted in several studies
Bifidobacterium longum	Species	↓	Marked reduction has been described in some younger cohorts
Faecalibacterium prausnitzii	Species	↑	Paradoxical increase reported in certain young cohorts
Bacteroides / Parabacteroides	Genus	↑	Recurrently reported as increased across multiple studies
Clostridium / Desulfovibrio / Sutterella / Lactobacillus	Genus	↑	Frequently reported as enriched, although findings vary across cohorts
Prevotella / Blautia / Dialister / Veillonella / Akkermansia	Genus	↓	Commonly reported as depleted in ASD in several studies
Escherichia/Shigella	Genus/group	↑	More consistently associated with GI phenotypes, particularly constipation, than with ASD overall

Directionality reflects the predominant pattern across the studies discussed in this review and should be interpreted cautiously. Findings remain heterogeneous across cohorts, age ranges, gastrointestinal phenotypes, diets, sequencing platforms, and analytical pipelines. This table is intended as a representative summary rather than an exhaustive catalog.

Despite recurring patterns, findings across studies are frequently inconsistent: reports conflict on α‑diversity trends [9, 16] and on the Firmicutes:Bacteroidetes ratio, and specific taxa implicated vary between cohorts [7, 21, 22]. These discrepancies likely reflect a combination of methodological heterogeneity (sample handling, sequencing and analysis pipelines), cohort differences in age, diagnostic criteria and comorbidities, inadequate adjustment for key confounders (antibiotic exposure, diet/selective eating, gastrointestinal symptoms, mode of delivery, infant feeding), geographic and cultural influences, and the predominance of small, cross‑sectional studies that limit causal inference [6, 13, 23–25]. Importantly, many human studies have not established robust links between fecal microbiota profiles and core ASD symptom domains [21, 26]. Nonetheless, the convergent conclusion is that the gut microbiota and metabolite profiles in many children with ASD differ from neurotypical controls. To move beyond association, the field needs standardized, well‑powered, longitudinal multi‑omic studies with rigorous phenotyping and careful control of confounders to identify microbial alterations that are reproducible, developmentally critical, and functionally relevant to ASD pathophysiology [18].

The gut microbiota-brain axis pathway in ASD patients

The gut microbiota affects brain development and function in ASD via multidirectional signaling within the microbiota–gut–brain (MGB) axis, encompassing interconnected neuronal/vagal, immune, chemical/metabolite, and endocrine mechanisms [2, 5, 27, 28]. Neuronal signals are conveyed by vagal afferents, the enteric nervous system, and enteroendocrine cells, with dysbiosis altering microbial production of neurotransmitters such as 5-hydroxytryptamine (5-HT) and GABA that modulate social and emotional circuits including the prefrontal cortex and nucleus accumbens, and specific strains like Lactobacillus reuteri can improve social behavior through vagus-dependent oxytocinergic pathways [28, 29]. Immune pathways involve microbes shaping intestinal and systemic immunity, driving microglial activation and elevated pro-inflammatory cytokines (e.g., (Interleukin) IL-6, (tumor necrosis factor) TNF-α, IL-1β) that compromise blood–brain barrier integrity and promote neuroinflammation, while maternal immune activation links the maternal microbiota to altered fetal neurodevelopment and microbiota colonization can even affect splicing of autism-related genes in animal studies [27, 28, 30]; Chemically, microbial metabolites (including SCFAs (butyrate, propionate), bile acids, and tryptophan derivatives) regulate blood–brain barriers integrity, neurotransmitter synthesis, excitatory/inhibitory balance, and gene expression through epigenetic modifications, acting directly on the brain or indirectly via enteroendocrine, vagal, or immune routes [5, 28]. Finally, the microbiota modulates hypothalamic–pituitary–adrenal (HPA) axis activity, linking stress-hormone regulation to neural and behavioral outcomes in ASD, and together these parallel, interacting pathways explain how gut microbial communities can profoundly influence neural function and behavior in ASD [30].

Gut microbiota-related brain structural, functional, and transcriptional alterations in ASD

Gut microbiota-related brain structural changes in ASD

Emerging human and animal evidence converges to link ASD‑associated gut microbiome alterations with specific brain structural and microstructural changes that map onto core social deficits: a Mendelian randomization study found higher Lachnoclostridium abundance was associated with increased left cerebellar white‑matter volume and a protective indirect effect on ASD risk, whereas Fusicatenibacter related to altered putamen signal and increased risk [31]. In young autistic children, reduced Lachnospiraceae and Ruminococcaceae correlated with worse autism diagnostic observation schedule (ADOS) social scores and with white‑matter alterations (notably in the left inferior longitudinal fasciculus) with mediation analyses indicating the microbiome partially explains the white‑matter–social deficit link [32, 33]. Complementary animal models reinforce these associations: prenatal valproic‑acid rodents exhibit early social deficits together with reduced fractional anisotropy, increased mean diffusivity in socio‑emotional circuits, gut dysbiosis, disrupted SCFA profiles, elevated peripheral proinflammatory cytokines, and glial activation [34], while propionic acid administration induces autism‑like behaviours with hippocampal synaptic ultrastructural damage and neuroinflammation [35]. Crucially, interventions that modify the microbiome or its downstream consequences can rescue both brain microstructure and behavior in models—for example, central thalamic nucleus deep brain stimulation reversed dysbiosis and restored diffusion kurtosis metrics and neuronal organization [36], and dietary solid‑lipid curcumin normalized diffusion tensor imaging (DTI) measures, myelin proteins, and microbiota composition [37]. Reviews synthesize these findings, highlighting microbiota‑associated alterations in regions subserving social and emotional processing and an expanding literature linking microbiome composition to white‑matter development and integrity in ASD [38, 39]. Mechanistically, gut microbes influence white matter and neuronal architecture via metabolite‑ and immune‑mediated pathways: microbial metabolites such as SCFAs (particularly butyrate) directly regulate oligodendrocyte function and myelination, whereas dysbiosis promotes neuroinflammation and barrier dysfunction that indirectly impair white‑matter integrity; altered metabolite signatures (reduced acetate/butyrate, elevated formate, presence of propionic acid) can disrupt fatty‑acid oxidation, provoke glial activation, and impair myelination, providing a plausible link from microbial shifts to the microstructural substrates of social deficits [33–35, 38]. The ability of microbiome‑targeted manipulations to reverse both microbial and brain abnormalities in preclinical models therefore supports causal pathways and identifies tractable therapeutic targets for ASD.

Gut microbiota-related brain functional alterations in ASD

Human and animal studies converge to implicate gut–brain mechanisms in core ASD features, particularly social‑emotional and sensory processing deficits: a direct human study found lower fecal kynurenate in adolescents with ASD and linked specific gut‑derived tryptophan metabolites to atypical insular and cingulate activity, with mediation analyses showing mid‑insula activity bridging indolelactate and ADOS severity and mid‑cingulate activity bridging tryptophan betaine and disgust sensitivity [40]. Complementing this, electrophysiological work in male Cntnap4 − /− mice showed that loss of Lactobacillus (notably L. reuteri) reduced spontaneous inhibitory postsynaptic current amplitude and GABAA receptor expression in the basolateral amygdala, correlating with social and fear‑memory deficits [41]. Broader animal studies link microbiota disruption to structural and molecular brain changes—enlarged amygdala, dendritic atrophy in hippocampal CA2/3, abnormal prefrontal myelination, dysregulated brain-derived neurotrophic factor, decreased NR2B, and altered serotonin signaling—that together impair synaptic plasticity and neurotransmitter balance [42]. Mechanistic rescue experiments further show that L. reuteri can signal via the vagus nerve to activate hypothalamic oxytocinergic neurons, thereby enhancing synaptic plasticity in ventral tegmental area dopaminergic reward circuits and improving social behavior in multiple ASD mouse models [29]. Taken together, a recent review synthesizes these findings by noting consistent effects of the gut microbiota on activity and connectivity within social‑emotional networks (including the amygdala, insula, anterior cingulate cortex, and default mode network) resulting in atypical neural responses to social stimuli and emotional dysregulation in ASD [39].

Gut microbiota-related brain transcriptional changes in ASD

Gut microbial perturbations exert profound effects on brain gene regulation, linking taxa and metabolites to altered transcriptional programs, post‑transcriptional processing, and consequent circuit and behavioral changes. In Shank3 knockout mice, dysbiosis with reduced cecal acetate correlated with social deficits that were reversed by oral acetate; this rescue coincided with extensive transcriptional reprogramming in the medial prefrontal cortex (mPFC), selectively enriching synaptic and immune gene pathways and implicating metabolite‑driven modulation of cortical gene expression [43]. Transfer of microbiota from children with ASD into germ‑free mice produced widespread alternative splicing changes in prefrontal cortex and striatum genes tied to social and repetitive behaviors, pointing to microbiota‑dependent regulation at the post‑transcriptional level [44]. Taxon‑specific effects further illustrate regulatory specificity: early-life exposure to elevated Bacteroides, especially B. fragilis, induced in male mice a coordinated downregulation of genes associated with inhibitory neurons and neural crest differentiation and an upregulation of cholesterol biosynthetic pathways in the prefrontal cortex, perturbing excitation–inhibition balance and neuronal signaling in a sex‑dependent manner [45]. Enrichment of Clostridiales, Lachnospiraceae, and Ruminococcaceae has been linked to suppression of myelination‑related gene expression and reduced prefrontal myelin thickness, indicating microbiota‑driven transcriptional control of oligodendrocyte programs [33]. Conversely, probiotic Lactobacillus reuteri treatment markedly altered GABA receptor gene and protein expression in hippocampus and prefrontal cortex, restoring neurotransmitter receptor profiles alongside improved social behavior [46]. Together, these findings converge on a model in which microbial taxa and metabolites modulate transcriptional and splicing networks governing synaptic function, inhibitory neuron identity, myelination, and neurotransmitter receptor expression in developmentally and sex‑dependent ways, thereby shaping ASD‑relevant neural circuitry and behavior.

While research into gut microbiota-related brain transcriptional changes in ASD has seen substantial advancements, current evidence linking the microbiome to transcriptomic and splicing alterations is primarily predicated on preclinical models, such as genetic mouse strains and microbiota-transfer studies. Consequently, direct validation within human ASD brain tissue remains notably sparse, highlighting a critical need for future post-mortem analyses or advanced neuroimaging-transcriptomic studies to bridge this translational gap.

Interventional therapies targeting the gut microbiota in ASD

Preclinical and clinical research increasingly evaluates gut microbiota–based interventions for ASD, with animal models providing causal evidence that microbial communities and metabolites can drive autism‑like behaviors and neuronal changes, and that targeted manipulations can reverse these effects [44]. Therapeutic approaches under study include probiotics (often combinations of Bifidobacteria, Streptococci, and Lactobacilli, with Lactobacillus plantarum singled out in some reports), prebiotics, synbiotics, dietary modification, antibiotics, and microbiota transfer procedures such as fecal microbiota transplantation (FMT) or microbiota transfer therapy (MTT) [47–50]. Probiotic, prebiotic, and synbiotic treatments have been shown to modulate the gut ecosystem and metabolic activity in children with ASD (commonly increasing Lactobacillus or Bifidobacterium and reducing taxa like Lachnoclostridium) and some trials report improvements in gastrointestinal symptoms and, in subsets of patients, core behavioral features [48, 49]. Microbiota transfer approaches have produced some of the most striking and durable results reported: open‑label MTT/FMT studies documented substantial gastrointestinal and behavioral improvements, systemic metabolomic shifts toward typical profiles, and persistence of benefits at two‑year follow‑up in one cohort [51–53]. However, findings across studies are heterogeneous, trial designs are often limited (small, open‑label or varied in protocols), and some analyses report inconclusive or null probiotic effects. Consequently, systematic reviews characterize the current evidence as suggestive but not definitive [54, 55]. To translate promise into practice, well‑controlled, randomized, placebo‑controlled trials are needed to establish which interventions (strains, combinations, or broader microbiota modulation), doses, treatment windows, and safety profiles are effective, and to clarify mechanisms linking microbiota changes to neural and behavioral outcomes.

Challenges, future directions, and translational outlook

Neuroimaging should be central to future investigations of the ASD gut‑microbiome–brain axis because it provides objective, spatially resolved, and quantifiable measures that can link microbial signatures to neural circuits and thereby bridge preclinical mechanisms with human phenotypes. Large-scale, longitudinal, multi-center studies that integrate multimodal imaging techniques—such as structural MRI, diffusion imaging (including DTI, advanced diffusion, and myelin mapping), functional MRI, and magnetic resonance spectroscopy (focusing on GABA, glutamate, and N-acetylaspartate)—with metagenomics, metabolomics, and other omics approaches will facilitate the identification of robust “microbial–imaging” biomarkers. These biomarkers hold promise for early risk detection, precise subtyping, and tailored treatment stratification. Radiomic feature extraction and machine‑learning models trained on well‑phenotyped, geographically diverse cohorts can fuse microbial and imaging data to predict clinical trajectories and personalize interventions, while imaging provides mechanistic readouts to assess effects of microbiota‑targeted treatments (probiotics, FMT), dietary or pharmacologic manipulations, and brain‑directed therapies such as transcranial magnetic stimulation (TMS) or deep brain stimulation (DBS). In parallel, germ‑free and ASD model animals paired with translational neuroimaging can validate causal pathways suggested by human associations and clarify how microbial metabolites and immune mediators alter white‑matter integrity, synaptic organization, and neurotransmitter systems. Achieving this promise will require rigorous standardization of microbiome and imaging protocols, careful control for diet, medication, and comorbidity confounds, and explicit consideration of ethical issues around stratification and intervention; If these challenges are met, neuroimaging-anchored, multi-omic approaches represent one of the most promising routes toward a mechanistic understanding and precision therapeutics for ASD [56].

The translational utility of neuroimaging in ASD is currently constrained by practical and methodological bottlenecks. Methodological challenges, such as motion sensitivity and the sensory demands of the scanning environment, frequently result in an ascertainment bias against individuals with higher support needs or intellectual disabilities. Beyond these technical hurdles, the high cost and specialized infrastructure required create systemic barriers for socioeconomically disadvantaged families. To improve the ecological validity and inclusivity of MGB axis research, it is essential to implement ASD-specific adaptations (such as motion-robust acquisition and desensitization protocols) while incorporating lower-burden alternatives like functional near-infrared spectroscopy to capture a more diverse clinical spectrum.

Conclusion

Structural, functional, and transcriptomic alterations in the brain constitute central mediators of the MGB axis in ASD. Regional volumetric changes, disrupted functional connectivity, and white‑matter abnormalities link ASD‑associated gut microbial signatures to macroscale neural circuit reorganization, while brain transcriptomic shifts reveal molecular pathways—immune, metabolic, and synaptic—through which microbial signals may reshape neural development and function. Integrating multimodal neuroimaging with transcriptomics and complementary multi‑omics will be essential to identify reproducible microbial–brain biomarkers, clarify causal mechanisms, and enable stratified, mechanism‑driven interventions. Achieving this goal requires standardized, longitudinal, multicenter deep‑phenotyping studies and ethically rigorous interventional trials.

Author contributions

SZ conceptualized and designed the study, obtained funding, and edited drafts of the manuscript. GW drafted the manuscript, prepared the figures and tables, and prepared the manuscript for submission. WL conducted the literature search and drafted sections of the manuscript. YC critically revised the manuscript. All authors read and approved the final manuscript.

Funding

The work was supported by the PhD Research Foundation of Affiliated Hospital of Jining Medical University (2022-BS-06) and the Key Research and Development Plan of Jining (Grant No. 2025YXNS008).

Competing interests

The authors affirm that the research was carried out without any commercial or financial relationships that could be interpreted as a potential conflict of interest.

Supplementary information

The online version contains supplementary material available at https://doi.org/10.1038/s41398-026-04158-4.