Autism spectrum disorder is a neurodevelopmental condition defined by difficulties in social communication and interaction together with restricted, repetitive patterns of behavior. Its prevalence has climbed sharply, from roughly 1 in 150 people in 2000 to 1 in 31 by 2022, a rise driven partly by changing diagnostic criteria and greater awareness. Beyond genetic susceptibility, environmental factors are increasingly implicated, contributing to metabolic abnormalities, oxidative stress, and immune dysregulation. Many children with autism also suffer gastrointestinal problems such as diarrhea and reflux, which have been linked to impaired epithelial barrier function and gut inflammation. Because the gut microbiota interacts with these environmental factors and shapes immune and metabolic pathways, researchers have come to view the intestinal ecosystem as a promising therapeutic target, even though no pharmacological treatment currently exists for the core symptoms of autism.

The two interventions tested in the trial represent opposite ends of the microbiome-manipulation spectrum. Fecal microbiota transplantation, or FMT, transfers standardized gut bacteria from a carefully screened healthy donor into the patient, aiming to rebuild the intestinal microenvironment. The ketogenic diet, by contrast, reshapes the existing microbiota indirectly through a high-fat, low-carbohydrate nutritional regimen, producing more moderate shifts. Earlier work had shown that microbiota transfer therapy can improve both gastrointestinal and core autism symptoms with benefits lasting up to two years, while ketogenic diets have been shown to alter microbial composition rapidly and to reduce inflammatory and oxidative stress markers in animal models of autism. What remained unclear was whether the two approaches were clinically interchangeable, and whether they left distinct microbial fingerprints behind.

The trial enrolled 30 children, 26 boys and 4 girls aged 19 to 97 months, all diagnosed according to DSM-5 criteria and confirmed with the gold-standard ADOS-2 observation schedule. Participants were excluded if they had secondary forms of autism, metabolic or genetic disease, serious gastrointestinal conditions, immunodeficiency, or significant malnutrition. Because both interventions lack standardized protocols and recruitment of young children is difficult, the team powered the study for preliminary efficacy, assigning 15 children to each arm. Allocation was pragmatic rather than randomized: children who could swallow capsules received FMT, while those who could not, or whose parents declined the procedure, were placed on the ketogenic diet. All participants continued their usual rehabilitation training throughout, so any measured changes came on top of standard care.

The FMT protocol spanned nine weeks, with oral capsules of lyophilized donor bacteria administered in weeks one and five, preceded by a residue-free semi-liquid diet and a polyethylene glycol bowel preparation to encourage colonization. The donor was a single screened 11-year-old male, tested for a comprehensive panel of pathogens, including SARS-CoV-2, multidrug-resistant organisms, and parasites. The dietary arm used a Modified Atkins Diet with a macronutrient ratio of 60 percent fat, 30 percent protein, and 10 percent carbohydrate. Caregivers received systematic training in carbohydrate counting and adverse-event management, and blood glucose and ketone levels were monitored intensively at first, tapering as metabolic adaptation stabilized. Symptom severity was assessed at baseline and at each group’s endpoint using the Childhood Autism Rating Scale, the Autism Behavior Checklist, and ADOS-2 calibrated severity scores.

The clinical results were encouraging for both groups. Mean CARS scores, which grade autism severity across 15 behavioral domains, fell significantly from 34.87 to 33.53 in the FMT group and from 35.13 to 33 in the ketogenic diet group. The ABC checklist dropped from 79.93 to 69.33 after FMT, a change that narrowly missed statistical significance, and from 63.07 to 42.73 after the diet, a substantial and significant reduction. The pattern of improvement differed between the arms: FMT produced greater gains in relation, imitation, emotional response, and object use on the more objective CARS scale, while the ketogenic diet yielded stronger improvements in sensory, relating, and language domains on the more subjective ABC. Safety labs covering blood cells and liver and kidney function showed no abnormalities, and reported adverse events were mostly transient gastrointestinal complaints, though five children on the diet experienced events including hypoglycemia and vomiting.

The metagenomic analysis, performed on fecal samples from 12 children per group before and after treatment, revealed that overall community structure remained remarkably stable. Neither alpha diversity indices such as Chao1, Shannon, and Simpson, nor beta diversity measured by principal coordinate analysis and non-metric multidimensional scaling, changed significantly. Instead, the interventions left specific taxonomic signatures. FMT significantly enriched three butyrate-producing taxa: Wujia chipingensis, Eubacterium sp. MSJ-33, and Butyrivibrio crossotus. The ketogenic diet elevated Blautia massiliensis while reducing two propionate-metabolizing Veillonella species. Spearman correlations tied these shifts to clinical change: rising Eubacterium sp. MSJ-33 correlated negatively with ABC general impressions scores, and declining Veillonella abundances tracked with improvements in total ABC scores and multiple subdomains, while increased Blautia massiliensis correlated negatively with sensory symptoms.

Functional profiling sharpened the mechanistic contrast. The FMT group showed 25 significantly differential genes but no enriched KEGG pathways, whereas the diet group displayed 262 differential genes with significant enrichment of propionate metabolism and aromatic compound degradation pathways. Nine core genes in the propionate pathway, including pduW and pdl, were downregulated, indicating suppressed microbial conversion of lactate into propionate. This matters because propionic acid is neurotoxic in animal models: injections of the compound induce autism-like behaviors, compromise blood-brain barrier integrity, activate immune signaling through PTEN/Akt, MAPK/ERK, and mTOR pathways, and disrupt serotonin and dopamine transmission. Conversely, butyrate, the short-chain fatty acid boosted by FMT-associated bacteria, is known to strengthen the intestinal barrier, improve blood-brain barrier integrity, and suppress neuroinflammation through pathways such as NF-κB. The two therapies, in other words, may improve symptoms by pushing the gut’s chemical output in opposite, beneficial directions, one adding protective butyrate and the other subtracting harmful propionate.

The authors are careful to frame these findings as preliminary. The sample was small, allocation was not randomized and depended on swallowing ability and parental preference, the diet group was younger at baseline, and the two arms were assessed at different post-intervention time points. Short-chain fatty acid levels were never measured directly, so the butyrate and propionate mechanisms remain inferential associations rather than demonstrated causal pathways. Most strikingly, 10 of the 25 children originally assigned to the ketogenic diet discontinued the intervention, a 40 percent dropout rate driven largely by the difficulty of maintaining a restrictive diet in children with marked food selectivity, and because the efficacy analysis was not intention-to-treat, the diet results may overestimate the true effect. Still, the study is the first to compare these two interventions clinically and microbially in the same trial, and it suggests a practical decision framework: FMT for children who can tolerate capsules and show emotional and relational symptoms, and the ketogenic diet for those who cannot access FMT and struggle most with sensory and language challenges. Larger randomized trials incorporating metabolomics will be needed to confirm efficacy, establish causality, and turn the gut-brain axis from a promising hypothesis into routine clinical practice for autism.

Subject of Research: Comparing fecal microbiota transplantation and ketogenic diet effects on symptoms and gut microbiota in children with autism spectrum disorder

Article Title: Comparison of clinical efficacy and gut microbiota characteristics in children with ASD treated with fecal microbiota transplantation and ketogenic diet

Article References: Wang, Y., Wang, L., Cai, Z., Yu, L., Guo, Y., Zhang, L., Zhu, M., Liu, Z., Zhao, Y., Liu, L., & Cao, A. (2026). Comparison of clinical efficacy and gut microbiota characteristics in children with ASD treated with fecal microbiota transplantation and ketogenic diet. BMC Psychiatry, 26(1), Article 685. https://doi.org/10.1186/s12888-026-08547-3

Image Credits: AI Generated

DOI: 10.1186/s12888-026-08547-3

Keywords: autism spectrum disorder, fecal microbiota transplantation, ketogenic diet, gut microbiota, gut-brain axis, short-chain fatty acids, butyrate, propionate, metagenomics, clinical trial, neurodevelopmental disorders, BMC Psychiatry