Effects of ketogenic diet on the classification and functional composition of intestinal flora in children with mitochondrial epilepsy

The ketogenic diet (KD) is a low-carbohydrate, high-fat diet that induces ketosis and has been used since the 1920s for intractable epilepsy. KD is thought to enhance mitochondrial biogenesis and function and has shown efficacy in mitochondrial epilepsy and in reducing oxidative stress. Proposed mechanisms include modulation of neuronal metabolism, interruption of glutamatergic transmission, reduced glycolysis, and activation of ATP-sensitive potassium channels. KD has also been explored in other neurological conditions (e.g., Alzheimer’s disease and autism). However, the precise mechanisms by which KD exerts neurological benefits remain unclear. Medium-chain triglycerides can rapidly induce ketosis and may contribute to neuroprotection. Diet is a major regulator of gut microbiota, which, alongside genetics, age, and region, shapes microbial composition. Rapid and persistent microbiota shifts occur with dietary change. Microbiota alterations have been implicated in obesity, NAFLD, and type 2 diabetes. Antibiotics can increase seizure risk, and KD-induced microbiota changes may affect metabolic and neural pathways relevant to epilepsy. Prior studies in epilepsy report KD-associated decreases in Firmicutes and increases in Bacteroidetes, but findings vary due to differences in fat sources, timing, age, race, and epilepsy etiology. This study aimed to explore relationships and altered pathways among ketosis, gut microbiota, and mitochondrial epilepsy via 16S rRNA profiling before and after KD, assessing community changes and predicted functional pathway enrichment.

Multiple studies have linked KD to shifts in gut microbiota in epilepsy. Xie et al. observed significant differences between epileptic children and healthy controls, with KD reducing pathogenic bacteria (e.g., Escherichia coli, Salmonella, Vibrio) and increasing beneficial taxa (e.g., Bacteroides). Zhang et al. reported reduced α-diversity post-KD, decreased Firmicutes, and increased Bacteroidetes in refractory epilepsy. Lindefeldt et al. noted no significant α-diversity change after 3 months of KD in drug-resistant pediatric epilepsy but a decrease in Bifidobacterium and an increase in Escherichia coli. Differences across studies likely reflect heterogeneity in KD protocols, patient demographics, and etiologies. Animal models suggest KD-induced microbiota changes are necessary for seizure protection, and case evidence indicates fecal microbiota transplantation can alleviate seizures. Literature also implicates purine metabolism and adenosine signaling in KD’s neuroprotection, and ABC transporters have been associated with drug-resistant epilepsy.

Design and participants: Prospective controlled study across 11 clinical centers (Jan 2019–Dec 2020) with 15 patients (newborn to 16 years) diagnosed with mitochondrial disease with epilepsy via genetic testing (mtDNA or nDNA pathogenic variants) and abnormal biomarkers (lactate, ketone bodies). Exclusion: prior KD, other genetic metabolic disorders, immunodeficiency, severe gastrointestinal/cardiovascular/respiratory/hepatic/urogenital diseases. Randomization and interventions: Participants randomized to control (regular diet + antiepileptic drugs, AEDs) or KD (KD + AEDs). Control group provided baseline fecal samples during enrollment for starting KD. KD initiated at 2:1 fat:non-fat ratio without fluid restriction or fasting. Energy needs calculated by a nutritionist based on age, height, weight. Blood ketones and glucose monitored every 6 h for first 4 days; adverse events (e.g., hyperketonemia, hypoglycemia) managed promptly. From day 5, energy adjusted to daily requirements. The first month (titration) allowed adjustments to KD ratio, energy, meal timing/frequency. At 12 weeks, second fecal sample collected. Control group followed regular diet + AEDs for 4 weeks, then underwent 12 weeks of KD alongside the KD group. Withdrawal permitted for poor efficacy/tolerance; physicians could terminate KD for side effects or worsening disease. Sample collection: Parents collected fecal samples using sterile swabs/containers before KD initiation and after 3 months of KD; samples stored at home cold, transported chilled, and stored at −80°C. DNA extraction and sequencing: DNA extracted with E.Z.N.A. Soil DNA kit. 16S rRNA gene V3–V4 region amplified with primers 338F/806R. Amplicons purified (2% agarose gel, AxyPrep DNA Gel Extraction Kit), libraries prepared with NEXTFLEX Rapid DNA-Seq Kit, sequenced on Illumina NovaSeq PE250. Bioinformatics: Raw reads filtered (Trimmomatic v0.33), primers removed (cutadapt 1.9.1), denoised/merged/chimeras removed via dada2 in QIIME2 2020.6, yielding non-chimeric reads. α-diversity (Shannon, Chao1) and β-diversity (weighted Bray-Curtis; PCoA) computed in QIIME. Differential taxa identified by LEfSe with LDA > 3 as significant. Functional prediction via Tax4Fun using KEGG. Statistics: SPSS used for analyses. Student’s t-test for means, chi-square for categorical variables. Measurement data as mean ± SD. Spearman correlation to evaluate microbiota-epilepsy relationships. Significance threshold p < 0.05.

Participants: 15 mitochondrial epilepsy patients (8 KD, 7 control). No baseline differences in gender, age, disease duration, medication, or seizure number (p > 0.05). Diversity: Shannon index slightly higher in controls (ns), while Chao1 richness was significantly higher in controls than KD (p < 0.05). β-diversity (weighted Bray-Curtis PCoA) showed compositional separation between KD and control; clustering heatmap indicated clear group distinctions. Taxonomic shifts: - Phylum level: Firmicutes dominant in both but lower in KD (42.76%) vs control (48.13%). Bacteroidota increased in KD (36.93%) vs control (25.41%). Lower-abundance phyla higher in controls: Actinobacteriota (KD 1.66% vs control 7.64%), Fusobacteriota (0.68% vs 1.65%), Desulfobacterota (0.15% vs 0.50%). - Genus level: Bacteroides markedly increased in KD (28.78%) vs control (9.51%). LEfSe (LDA > 3): - Control enrichment: Actinobacteriota (phylum); genera Phascolarctobacterium, Subdoligranulum, Agathobacter, Erysipelotrichaceae_UCG_003. - KD enrichment: genus Bacteroides driven by species Bacteroides fragilis; increases also noted for Blautia (Blautia_sp__N6H1_15) and Anaerotignum lactatifermentans. Functional prediction (Tax4Fun/KEGG): - Level 2 pathways: Infectious diseases: Bacterial and Signal transduction significantly elevated in KD. - Level 3 pathways: 320 pathways differed, with 12 significantly altered. KD showed increased enrichment in Citrate cycle (TCA cycle), Pertussis, Phosphatidylinositol signaling system, Biofilm formation–Escherichia coli, Penicillin and cephalosporin biosynthesis, Lysosome, and Glycosphingolipid biosynthesis (lacto and neolacto series). Decreased pathways included Quorum sensing, Bacterial secretion system, Nicotinate and nicotinamide metabolism, Legionellosis, and Arginine biosynthesis. Additional notable differences included Purine metabolism, Phenylalanine metabolism (decreased), and Phenylalanine, tyrosine, and tryptophan biosynthesis (increased) after KD.

KD modified the gut microbiota composition in children with mitochondrial epilepsy, notably increasing Bacteroides and specifically Bacteroides fragilis while decreasing Firmicutes. These shifts align with prior studies suggesting KD-linked increases in Bacteroidetes and highlight B. fragilis as a potential protective taxon in seizure control, consistent with reports of lower B. fragilis in drug-resistant epilepsy and clinical benefit from B. fragilis supplementation. Functional predictions suggest broad metabolic reprogramming, with pronounced alterations in purine metabolism and shifts in pathways related to energy metabolism (TCA cycle), signaling, and microbial community behaviors (e.g., quorum sensing, biofilm formation). The purine/adenosine axis is a plausible mediator of KD’s neuroprotective effects via increased ATP/adenosine and A1 receptor activation. Observed decreases in ABC transporter-related functions in KD (non-significant) are directionally consistent with elevated ABC transporters in drug-resistant epilepsy. Overall, findings support a microbiota-mediated contribution to KD’s antiepileptic effects and identify candidate microbial and functional biomarkers (e.g., B. fragilis, purine metabolism) for therapeutic monitoring.

This study demonstrates that a 12-week ketogenic diet in children with mitochondrial epilepsy is associated with distinct alterations in gut microbiota composition and predicted function, including increased Bacteroides (notably B. fragilis), decreased Firmicutes, reduced richness, and significant changes in multiple KEGG pathways (e.g., TCA cycle, quorum sensing, purine metabolism). These microbiome changes may contribute to KD’s antiseizure effects and could serve as biomarkers for treatment response. Future work should incorporate targeted metabolomics and host signaling measurements (e.g., adenosine/A1 receptors), larger multicenter cohorts, and extended KD durations to validate and mechanistically link microbial changes to clinical outcomes.

Small sample size (n = 15) limits statistical power and generalizability. Implementing KD poses challenges with patient compliance and potential adverse effects. The study used 16S rRNA profiling with functional prediction (Tax4Fun) rather than direct metagenomics/metabolomics; thus, functional inferences are indirect. Short intervention duration (12 weeks) may not capture longer-term microbiome dynamics. Spearman correlations mentioned were not detailed in results.