Skip to main content
NCBI home page
Wiley Open Access Collection logoLink to Wiley Open Access Collection
. 2025 Mar 22;66(7):2268–2284. doi: 10.1111/epi.18367

Medication‐resistant epilepsy is associated with a unique gut microbiota signature

Antonella Riva 1,2,✉, Eray Sahin 3, Greta Volpedo 1, Noemi Teresa Catania 1, Isabel Venara 4, Valentina Biagioli 1, Ganna Balagura 1, Elisabetta Amadori 1,5, Carmen De Caro 6, Emanuele Cerulli Irelli 7, Carlo Di Bonaventura 7, Federico Zara 1,2, Osman Ugur Sezerman 3, Emilio Russo 6, Pasquale Striano 1,5
PMCID: PMC12291004  PMID: 40119849

Abstract

Objective

Dysfunction of the microbiota–gut–brain axis is emerging as a new pathogenic mechanism in epilepsy, potentially impacting on medication response and disease outcome. We investigated the composition of the gut microbiota in a cohort of medication‐resistant (MR) and medication‐sensitive (MS) pediatric patients with epilepsy.

Methods

Children with epilepsy of genetic and presumed genetic etiologies were evaluated clinically and subgrouped into MR and MS. Age‐matched healthy controls (HCs) were also recruited. A food diary was used to evaluate nutritional habits, and the Rome IV questionnaire was used to record gastrointestinal symptoms. The microbiota composition was assessed in stool samples through 16S rRNA. α‐Diversity (AD) and β‐diversity (BD) were calculated, and differential abundance analysis was performed using linear multivariable models (significance: p.adj < .05).

Results

Forty‐one patients (MR:MS = 20:21) with a mean age of 7.2 years (±4.6 SD) and 27 age‐matched HCs were recruited. No significant differences in AD were found when comparing patients and HCs. Significant positive correlation was found between AD and age (Chao1 p.adj = .0004, Shannon p.adj = .0004, Simpson p.adj = .0028). BD depicted a different bacterial profile in the epilepsy groups compared to HCs (MS vs. HC: Bray–Curtis F = 1.783, p = .001; Jaccard F = 1.24, p = .001; MR vs. HC: Bray–Curtis F = 2.24, p = .001; Jaccard F = 1.364, p = .001). At the genus level, the epilepsy groups were characterized by a significant increase in Hungatella (MS vs. HC: +4.95 log2 change; MR vs. HC: +6.72 log2 change); the [Eubacterium] siraeum group changed between the MR and MS subgroups.

Significance

Epileptic patients display unique gut metagenomic signatures compared to HCs. Moreover, a different ratio of the butyrate‐producing [Eubacterium] siraeum group suggests dissimilarities between patients based on the response to antiseizure medications.

Keywords: epilepsy, gut microbiota signature, microbiota–gut–brain axis, pediatrics

Key points.

  • α‐Diversity correlates with age in patients with epilepsy.

  • In children, Hungatella differentiates between epilepsy patients and healthy controls.

  • [Eubacterium] siraeum group differentiates between MR and MS epilepsy patients.

1. INTRODUCTION

The gut microbiota consists of thousands of bacteria and, to a lesser extent, yeasts, parasites, and viruses, colonizing the human gastrointestinal (GI) tract. 1 This ecosystem plays a crucial role in maintaining immune and metabolic homeostasis and protecting against pathogens. Furthermore, it was shown that dysfunction of intestinal tight junctions can promote the direct passage of substances into the bloodstream, directly impacting neurotransmitter/precursor synthesis and influencing the outcome of neurological and psychiatric disorders. 2 , 3 This complex network of communication is commonly referred to as the microbiota–gut–brain axis (MGBA). Thanks to the recent advent of next generation sequencing (NGS) and multiomic integration approaches, the MGBA has been consistently studied, particularly in preclinical models of epilepsy, due to the more reproducible condition in which rodents are housed and fed. 4

In contrast, the human gut microbiota has not yet been completely defined due to the host's individual uniqueness and the high variability of bacterial population and extensive genomic intraspecies variations. 5 Furthermore, exposure to environmental factors and host genetics play a major role in determining microbial diversity between individuals. Starting from intrauterine life, the human gut microbiota is exposed to various “modifying” events (e.g., mode of delivery, breastfeeding, weaning), leading to different and individual microbiome signatures even among healthy individuals. 6 Nevertheless, some studies have attempted to investigate at a taxonomic level the microbiota composition in subjects affected by different neurological/neuropsychiatric disorders (e.g., epilepsy and autism), successfully characterizing different signatures between patients and controls. 3 , 7 In particular, Türay and colleagues have shown that the gut microbiota profile of children with epilepsy of unknown etiology was characterized by the presence of Megamonas and Coriobacterium, whereas healthy controls presented different metagenomic signatures. 8 Another study in medication‐naïve pediatric epilepsy patients showed an increased abundance in Akkermansia spp. and Proteobacteria, and a decreased relative abundance in Faecalibacterium spp. 9 Although these studies demonstrate a shift in the gut microbiota composition in pediatric epilepsy, additional research is needed to fully characterize these metagenomic signatures, due to the heterogeneity of the results obtained, the limited sample size, and the different epileptic syndromes investigated.

According to the International League Against Epilepsy (ILAE), 10 the intra‐ and intersyndromic variability in epileptic phenotypes is reflected in the range of possible etiologies. In approximately 30% of cases, the underlying cause is known or presumed to be genetic. 11 Thanks to NGS, the gap between diagnosed and unsolved cases has been partially filled. New causative genes have emerged and have been associated with diverse electroclinical phenotypes. 12 Moreover, even if a primary epileptogenic focus is identified, genetics should not be ruled out. For example, in tuberous sclerosis complex mutated TSC1 and TSC2 provide epileptogenic and systemic features that go beyond the finding of cortical tubers. 11 Consequently, in these cases managing seizures can be troublesome. Medication‐resistance is defined as the failure of two “adequately chose, used, and well tolerated” antiseizure medications (ASMs) in reducing seizures. 13 No definite mechanisms have been identified; however, different hypotheses have been adduced, including interneuron and cortical network alterations, as well as polymorphisms in drug‐metabolizing enzymes. 14

Interestingly, the gut microbiota can interact bidirectionally with orally administered medications, which encounter different populations of gut microbes while descending through the GI tract. Whereas on one hand medications and their metabolites can affect the GI microenvironment by modulating bacterial growth and composition, on the other they can be enzymatically transformed by gut resident bacteria, thereby altering their bioavailability, activity, or toxicity. 15 Due to these interactions, it is possible that the gut flora might influence the response to ASMs. A recent study showed changes in the gut microbiota composition of children with focal epilepsy compared to healthy controls. 16 Interestingly, these differences were attenuated posttreatment in the same patients, highlighting a possible interaction between gut microbiota and ASMs. In this study, we investigated gut microbiota differences between medication‐resistant and medication‐sensitive pediatric epileptic patients with genetic and presumed genetic etiology.

2. MATERIALS AND METHODS

2.1. Study population

Pediatric patients with epilepsy due to a definite or presumed genetic condition were recruited at IRCCS Istituto Giannina Gaslini between April 2020 and November 2022. Diagnosis and classification of the epilepsy syndrome was based on the ILAE guidelines. 10 Genetic investigations included array‐based comparative genomic hybridization, epilepsy panel, or whole exome sequencing as deemed appropriate as per clinical practice. Clinical data were retrieved through our local database. The inclusion criteria were as follows: (1) age < 18 years, (2) confirmed clinical diagnosis of epilepsy, and (3) written informed consent to participate in the study provided by the patient or by the caregivers/legal guardian. The exclusion criteria (EC) were as follows: (1) age ≥ 18 years, (2) antibiotic treatment started <3 months from recruitment, (3) autoimmune systemic comorbidities or ongoing infections at the time of recruitment, (4) any GI chronic disorder (i.e., Crohn disease), and (5) significant feeding problems (i.e., dysphagia).

Patients were stratified into different subgroups (medication‐resistant [MR] and medication‐sensitive [MS]) based on their response to ASMs at the time of observation. Patients taking one ASM were taking monotherapy, whereas patients taking two or more ASMs were taking polytherapy. Although these groups differ from the ILAE definition of medication resistance, 13 this grouping was chosen due to challenges in assessing seizure response, tolerability, adequate dosage, and appropriate choice of the ASM for the type of seizures.

The healthy control (HC) group included age‐matched neurotypical children with the same EC as above, recruited at the IRCCS Istituto Giannina Gaslini Transfusional Center.

The study was conducted following the Helsinki protocol and approved by our local ethics committee, Comitato Unico Regionale Regione Liguria.

2.2. Study procedures

A stool sample was collected from each participant at the time of the enrollment by spontaneous or enema defecation using a self‐collection stool kit (CeGat). The sampling kit contained a liquid‐based DNA preservation solution, and samples were stored at room temperature (20–25°C) until DNA extraction.

At the time of stool sampling, any GI symptom was assessed by referring to the validated Rome IV Diagnostic Questionnaire (Rome Foundation, https://theromefoundation.org), which investigates the frequency and intensity of abdominal pain, constipation, diarrhea, reflux, bloating, dyspepsia, nausea, and vomiting.

All the participants were under free dietary regimens without any nutritional restrictions. The participants and their caretakers were also guided by a nutrition expert in the completion of a semiquantitative, specifically built, dietary questionnaire. Food types were broken down into macronutrient categories: proteins, fats, carbohydrates, and fibers. For fat sources (oil and butter), a reference weight value of 10 g (approximately 1 tablespoon) was considered. A pediatric food atlas (Istituto Scotti Bassani) was shown to the patients and caregivers to more easily assess the most frequently eaten portion sizes. Macronutrient intake was provided per day and converted to frequency per 7 days (daily frequency × 7). When there was uncertainty between two daily frequencies (e.g., 1 or 2 times/day), the mean was calculated.

Descriptive statistical analyses were performed to evaluate demographic and clinical characteristics, with continuous data presented as mean ± SD or median (interquartile range) as appropriate, and ordinal data expressed as number (percentage). GraphPad Prism 7.0 software was used for statistical analysis.

2.3. DNA extraction and sequencing

Total DNA was extracted using the Zymo BiomicsDNA/Rna Miniprep Kit (Zymo Research), and quality was checked with Qubit dsDNA BR or HS (Thermo Fisher Scientific), as per manufacturer's protocol. Twenty nanograms of DNA was used for library preparation by amplifying the V3–V4 hypervariable regions of the bacterial 16S rRNA gene using the Zymo Quick 16S NGS library preparation kit (Zymo Research). The final libraries were pooled and sequenced with MiSeq Sequencing System (Illumina) for 2 × 250 base paired‐end reads (90.71% bases ≥ Q30). Demultiplexing of the sequencing reads was performed with Illumina bcl2fastq (2.20). Reads were filtered for reads containing Ns and low‐quality bases and adapter's trimming was performed with Skewer (version 0.2.2). 17 Merging of the reads was performed using R package Divisive Amplicon Denoising Algorithm 2 (DADA2). 18 The quality of the FASTQ files was analyzed with FastQC (version 0.11.5‐cegat). 19

2.4. Bioinformatic analysis

The quality trimming of the raw 16S amplicon sequencing reads was carried out using fastp 20 with “qualified_quality_phred” set to 20 and “length_required” set to 200. Trimmed reads were imported into Qiime2 (version Amplicon 2023.9). 21 341F and 806R primer sequences were removed from reads using the Cutadapt plugin. 22 A denoising step including quality control of the sequences through trimming and filtering, removal of chimeras, and paired‐end reads merging was carried out using the DADA2 plugin. 18 The phylogenetic tree was constricted using the q2‐phylogeny plugin. To increase classification accuracy, the weighted taxonomic classifier was trained. 23 First, RESCRIPt plugin 24 was used to process the SILVA 138.1 SSURef NR99 full‐length database. 25 Using the q2‐clawback plugin, taxonomic weights were assembled for human stool. Then, the full‐length SILVA classifier was retrained using the assembled weights. For taxonomic assignments of the amplicon sequence variants (ASVs) obtained from DADA2, the classify‐sklearn option of the q2‐feature‐classifier plugin 26 was used with the trained weighted classifier. Rooted phylogenetic tree, ASV count table, and taxonomic assignment of ASVs were exported and used in the downstream analysis with R (Table S1).

2.5. KEGG pathway prediction

Phylogenetic Investigation of Communities by Reconstruction of Unobserved States 2 (PICRUSt2) is a tool used to infer the metabolic potential of the microbial communities in an environment using obtained amplicon sequences. Obtained ASV sequences were exported from Qiime2 and used in PICRUSt2 (version 2.5.2) 27 analysis using default settings to obtain Kyoto Encyclopedia of Genes and Genomes (KEGG) Ortholog abundances. KEGG pathway counts were normalized by calculating relative abundance percentages, and pathways with at least .001% abundance in at least half of the subjects in any group were kept for downstream differential analysis. Microbiome Multivariable Associations With Linear Models (MaAsLin2) was used for statistical testing using the same settings as in taxonomic analysis.

2.6. Diversity analysis

The phyloseq (version 1.46.0) package was used for the diversity analysis. A phyloseq object was created using metadata, ASV count and taxonomy tables, and rooted phylogenetic tree.

For α‐diversity (AD), rarefying was carried out using the minimum sample size of 13 287, and all indices were calculated by the estimate_richness function. Chao1, Shannon Diversity, and Simpson's Index were used in statistical tests. Chao1 reflects the richness only through the number of species and singletons, whereas the Shannon and Simpson diversity indices take both richness and evenness into account, with the first one being more biased toward dominant bacteria in the community. Generalized linear models (GLMs) with gamma error distribution and log link function were used for all three indices. For each model, one of the AD indices was used as a response variable, and group and covariates (age, sex, carbohydrate, protein, fat, fiber) were used as predictor variables. Two different models were used for comparison of epilepsy group (MR + MS) versus HCs, and comparison of MR versus MS. Probability values obtained from GLMs of three indices of each group comparison were adjusted using false discovery rate. Adjusted p‐value of .05 was used as a significance threshold (Table S1).

β‐Diversity (BD) analysis was conducted using Bray–Curtis and Jaccard dissimilarity indices. Both indices are used to estimate dissimilarities between communities; however, the Bray–Curtis index uses quantitative information, giving more weight to dominant taxa, whereas the Jaccard index uses presence/absence information of taxa and gives equal weight to dominant and rare taxa. BD distances were calculated using not rarefied data. Principle coordinate analysis (PCoA) ordination was performed on each index, and plots were obtained using the first two axes that were sorted in descending order based on the relative eigenvalues, which shows how much of the variation is explained. Each distance matrix was used in permutation‐based multivariate analysis of variance (PERMANOVA) to test the significance of the difference of bacterial compositions between groups (MR + MS vs. HC or MR vs. MS). Age, sex, and dietary intakes were used as confounding factors. Significance threshold for p‐values was set to .05 (Table S1).

2.7. Differential abundance analysis

Differential abundance (DA) analysis was carried out at the phylum, family, and genus levels (Table S1). First, ASV percentile relative abundances were calculated using the ASV counts in each sample.

All ASVs assigned to any kingdom other than bacteria or assigned to mitochondria were filtered out. For each taxonomic rank of interest, ASV abundances were aggregated, and “unidentified” and (for the family and genus levels only) “uncultured” portions were removed prior to the DA analysis. Taxa with a relative abundance higher than .1 for phylum or .01 for family and genus in at least half of the subjects for each group were kept for DA analysis. MaAsLin2 28 was used to test differential relative abundance of bacteria at different taxonomic levels between groups by adjusting for covariates (i.e., age, carbohydrate, protein, fat, and fibers). Two different models were used for the comparison between the epilepsy and control groups (MR + MS vs. HC; set reference group = HC), and comparison within the epilepsy group (MR vs. MS; set reference group = MS). For each MaAsLin2 model, normalized and filtered data in phylum, family, or genus level were used; no further normalization and transformation were selected. Lastly, “CPLM” was used as an analysis method, and maximum significance was set at .1.

Based on our results, dominance‐based analysis was carried out specifically for Hungatella (Hung) and the [Eubacterium] siraeum group (Eub). First, samples in which neither of the two genera were detected were excluded from the analysis. For the rest of the samples, the relative abundances of the two genera were summed and used as an isolated community. Based on their proportional abundance in that isolated community, a binary score (0 or 1) was assigned as Hung/Eub dominance: if the proportional abundance of Hung was ≥50% in the isolated community, the dominance score was assigned as “1”, otherwise “0” was assigned to describe the dominant abundance of Eub. For significance of this dominance score between groups, two GLM functions with family parameter set to “binomial” were used. Group and covariates (age, sex, carbohydrate, protein, fat, and fiber) were used as independent variables. Raw p‐value of .05 was used as a significance threshold.

Relative abundances of significant genera were centered log‐ratio (clr) transformed, after replacing zero values with half of the minimum relative abundance value. Log2 fold changes were calculated by subtracting log2‐transformed median relative abundances in the studied groups. Significance values were adjusted p‐values obtained from the respective MaAsLin2 model. A heatmap was plotted using the ComplexHeatmap package (version 2.14.0). 29 Correlation plots were obtained using ggplot2 package (version 3.4).

Rooted tree was imported into R, and phyloseq object was created as described above. Significant genera‐associated ASVs were divided using the subset_taxa function of the phyloseq package. Trimmed phylogenetic tree was exported using phy_tree of the phyloseq package and write.tree function of the ape (version 5.7.1) package. 30 Genus relative abundance percentiles were clr transformed after replacing zero values with half of minimum abundance value in the genus abundance table. The visualization of trimmed phylogenetic tree encircled by a median clr‐transformed genus abundance heatmap was carried out with the iTOL website tool. 31

3. RESULTS

3.1. Demographic and clinical characteristics of the studied cohort

The epilepsy groups (MR + MS) accounted for 41 children (23 females) with a mean age of 7.2 years (±4.6 SD). Twenty‐six (63%) patients were born by vaginal delivery, 14 (34%) by cesarean section, and the route of delivery for one child was unknown. Nutrition during infancy was reported as exclusive breast‐feeding in 18 cases, mixed (i.e, breast‐fed for few months then formula‐fed) in seven cases, exclusive formula‐feeding in seven cases, and unknown in nine cases (Table 1). The mean ± SD age at weaning was 6 ± 4 months for 18 patients.

TABLE 1.

Demographic, clinical, and dietary data of the studied cohort.

Patient Sex Age at enrollment, years Sz types (current) Sz frequency DD ID Epilepsy syndrome/etiology ASMs Other treatments Rome IV Proteins a Carbohydrates a Fibers a Fats a Type of delivery Feeding type/age at weaning MR status
1 F 11 Tonic Sz with drop Daily Y Y DEE/idiopathic (WES performed) CLB, FBM ALA, FOS, inulin, thiamine Nonretentive FI 14.5 42 24.5 10.5 Vaginal, eutocic Artificial/6th month Y
2 F 17 Focal Sz Daily Y Y DEE/idiopathic (WES performed) CLB, VPA, CBD, NTZ ALA, FOS, inulin, baclofen, trihexyphenidyl Negative 21 11.5 3.5 20 CS NA/NA Y
3 F 7 Focal Sz Daily Y Y DEE/idiopathic (WES performed) CLB, VPA, LCM NA Negative 12 20.5 5 29 Vaginal, eutocic BF/6th month Y
4 F 7 PGTCS Weekly Y Y DEE/idiopathic (WES performed) LTG, CBD, DZP NA Negative 9 36 3.5 15.5 CS Mixed/6th month Y
5 F 5 PGTCS Daily Y Y DEE/idiopathic (WES performed) RFM, CZP, VPA, FBM, DZP, CLB ALA, FOS, inulin Negative 18 39 7 25 Vaginal, eutocic NA/NA Y
6 F 8 PGTCS Daily Y Y DEE/idiopathic (WES performed) TPM, CLB, VPA NA Negative 17.5 20.5 7 12.5 CS NA/6th month Y
7 F 9 PGTCS SF < 1 year Y Y DEE/idiopathic (WES performed) VPA, ESX NA Negative 19.5 18.5 24.5 7.5 Vaginal, eutocic Mixed/NA Y
8 F 4 Focal Sz, SGTCS Daily Y NA DEE/TSC2 (c.5238_5255del; p.His1746_Arg1751del) VPA, CBD, TPM, DZP NA Negative 17 27.5 0 15.5 Vaginal, eutocic Mixed/2nd month Y
9 F 1 SGTCS, absences Daily Y NA DEE/idiopathic (WES performed) LEV, VGB NA Functional constipation 10 17 14 7.5 CS BF/5th month Y
10 M 12 Myoclonic Sz Daily Y Y DEE/idiopathic (WES performed) RFM, CLB, CBD Melatonin Negative 16 41.5 7 6 Vaginal, eutocic BF/4th month Y
11 F 3 Myoclonic Sz Daily Y NA DEE/NEXMIF (c.1326delTp; p.Asp442Glufs*58) VPA, CBD ALA, FOS, inulin Negative 20.5 25 0 3 CS Artificial/6th month Y
12 F 3 Focal Sz Weekly N Y, mild FE/hypothalamic hamartoma (operated) + idiopathic (WES performed) CBZ, LEV NA Negative 11 36 7 12.5 CS Mixed/5th month Y
13 M 4 Absences, focal Sz Daily Y NA DEE/idiopathic (WES performed) VPA, CLB, CBD Melatonin Nonretentive FI 18 26 17.5 20 Vaginal, eutocic NA/NA Y
14 M 4 Epileptic spams Daily Y NA DEE/DYNC1H1 (c.10247_10279dup; p.Leu3416_Asn342dup) VPA, CBZ, TPM NA Functional constipation, aerophagia 9.5 14 19 7.5 CS Mixed/6th month Y
15 M 6 Tonic Sz, absences SF < 1 year Y Y DEE/idiopathic (WES performed) VPA, TPM Melatonin Nonretentive FI 23.5 22.5 7 7.5 Vaginal, eutocic Mixed/NA Y
16 M 14 Focal Sz Monthly Y Y DEE/idiopathic (WES performed) LTG, CBZ ALA, FOS, inulin Negative 25.5 34 8.5 12.5 Vaginal, eutocic BF/3rd month Y
17 M 1 Focal Sz Daily Y NA DEE/idiopathic (a‐CGH performed) VPA, LEV NA Negative 14.5 3 3.5 7.5 Vaginal, eutocic NA/NA Y
18 F 15 PGTCS, absences Daily Y Y DEE/de novo BRAF pathogenic variant VPA, CBD, LCM, LZP ALA, FOS, inulin Negative 35 15 7 15 Vaginal, eutocic Artificial/6th month Y
19 F 9 Focal Sz Daily N N FE/idiopathic LEV, CLB ALA, FOS, inulin Negative 21.5 44 4 18 Vaginal, eutocic NA/NA Y
20 M 13 Myoclonic Sz Daily Y Y DEE/idiopathic (WES performed) ESX, PER, VPA, LEV, CLB, CBD NA Negative 46.5 25 7 6 Vaginal, eutocic BF/NA Y
21 M 6 PGTCS SF > 1 year N Y IGE/idiopathic (WES performed) VPA NA Negative 18.5 32.5 1 17 Vaginal, eutocic BF/6th month N
22 M 8 Focal Sz SF > 1 year N N FE/idiopathic (WES performed) LEV NA Negative 20.5 24 7 17 Vaginal, eutocic BF/5th month N
23 F 13 Focal Sz Daily N N FE/idiopathic (epi panel performed) CBZ ALA, FOS, inulin Negative 19.5 22.25 4 10.5 CS BF/NA N
24 F 8 Absences SF > 1 year N N IGE/idiopathic ESX NA Negative 9 24 1.5 7.5 Vaginal, eutocic BF/6th month N
25 M 12 Focal Sz Daily N N FE/idiopathic CBZ ALA, FOS, inulin Negative 9.5 22 17.5 15.5 Vaginal, eutocic BF/NA N
26 M 5 Tonic Sz Yearly Y Y DEE/idiopathic (a‐CGH negative) VPA, DZP NA Negative 27.5 35.25 7 15.5 Vaginal, eutocic BF/6th month N
27 M 4 Absences, clonic Sz Daily Y NA DEE/idiopathic (a‐CGH performed) VPA ALA, FOS, inulin Negative 31 32.5 5 15.5 Vaginal, eutocic BF/2nd year N
28 F 1 Focal Sz Monthly Y NA DEE/Xp22 deletion (13804Kb) including CDKL5 VPA NA Negative 17 31.5 3.5 7.5 CS BF/NA N
29 F 4 PGTCS SF > 1 year N NA IGE/idiopathic (a‐CGH negative) CLB ALA, FOS, inulin Negative 11.5 25.5 14 25 Vaginal, eutocic Artificial/NA N
30 M 6 PGTCS, SE SF > 1 year Y Y DEE/idiopathic (a‐CGH performed) ESX NA Functional constipation 16 27 3 20 CS NA/NA N
31 F 12 PGTCS SF > 1 year N N IGE/idiopathic (a‐CGH negative) VPA NA Aerophagia 13.5 19 7 17 CS NA/NA N
32 F 0 Focal Sz SF < 1 year Y NA DEE/idiopathic (WES performed) LEV NA

Negative

 13.5 24 3 3 Vaginal, eutocic Artificial/NA N
33 M 5 Focal Sz Yearly N N FE/idiopathic (epi panel performed) VPA NA Negative 18.5 61 14 12.5 CS Artificial/NA N
34 F 13 PGTCS Yearly N N IGE/idiopathic (epi panel negative) ESX NA Functional vomiting 20 24 14 7.5 CS Mixed/NA N
35 M 12 Absences, PGTCS SF > 1 year Y Y DEE/idiopathic (epi panel negative) VPA, DZP NA Functional constipation 19 30 7 10.5 NA NA/NA N
36 F 0 Epileptic spams Daily Y NA DEE/idiopathic (epi panel performed) LEV NA Negative 38 22.5 7 10.5 Vaginal, eutocic BF/NA N
37 M 4 FS, PGTCS Monthly N NA IGE/idiopathic (SCN1A and a‐CGH negative) VPA NA Negative 21.5 40 7 2.5 CS BF/NA N
38 F 2 FS, absences Monthly N NA IGE/idiopathic (a‐CGH performed) DZP ALA, FOS, inulin, levothyroxine Negative 14 38.5 14 15.5 Vaginal, eutocic BF/6th month N
39 M 12 Myoclonic Sz Yearly N Y DEE/idiopathic (WES performed) LEV RR Functional constipation 14.5 46.5 14 7.5 Vaginal, eutocic BF/NA N
40 M 3 Focal Sz Monthly Y Y DEE/idiopathic (a‐CGH negative) CLB NA Negative 7 28 0 10.5 Vaginal, eutocic Artificial/NA N
41 F 12 Focal Sz Daily Y Y DEE/idiopathic (a‐CGH negative) LEV NA Negative 21.5 39.5 5 6 Vaginal, eutocic BF/NA N
Open in a new tab

Abbreviations: a‐CGH, array‐based comparative genomic hybridization; ALA, α‐lactalbumin; ASM, antiseizure medication; BF, breast‐feeding; CBD = cannabidiol; CBZ, carbamazepine; CLB, clobazam; CS, cesarean section; CZP, clonazepam; DD, developmental delay; DEE, developmental and epileptic encephalopathy; DZP, diazepam; epi, epilepsy; ESX, ethosuximide; F, female; FBM, felbamate; FE, focal epilepsy; FI, fecal incontinence; FOS, fruttooligosaccharides; FS, febrile seizures; ID, intellectual disability; IGE, idiopathic generalized epilepsy; LCM, lacosamide; LEV, levetiracetam; LTG, lamotrigine; M, male; MR, medication‐resistant; N, no; NA, not available; NTZ, nitrazepam; PGTCS, primarily generalized tonic–clonic seizures; PER, perampanel; RFM, rufinamide; RR, risperidone; SE, status epilepticus; SF, seizure‐free; SGTCS, secondarily generalized tonic–clonic seizures; Sz, seizure(s); TPM, topiramate; VGB, vigabatrin; VPA, valproic acid; WES, whole exome sequencing; Y, yes.

a

For macronutrients, the reference is times eaten per week.

The diagnosis was developmental and epileptic encephalopathy in 28 (68%) patients, idiopathic generalized epilepsy in seven (17%) patients, and focal epilepsy in six (15%) patients (Table 1). The MR‐to‐MS ratio was 20:21. The mean number of concomitant ASMs in the MR group was 3.

For the HCs, 27 children with a mean age of 9.3 years (±4.1 SD) were included in the study (Table S1).

The pairwise Mann–Whitney U‐test showed no significant difference between the groups in respect to age (MR vs. MS: p = .57, MR vs. HC: p = .19, MS vs. HC: p = .06; Figure S1).

3.2. Gut microbiota richness increases with age

Taxonomic richness and diversity within each sample was evaluated through AD by focusing on three different indices; Chao1 reflects the richness through the number of species and singletons, whereas the Shannon and Simpson indices take into account both richness and evenness, with the first one being more biased toward dominant bacteria in the community.

All samples (n = 68) were included for evaluating the differences in the bacterial community composition between the epilepsy groups (MS and MR) and the HC group. Similarly, all the patient samples (n = 41) were used for comparing the MS versus MR subgroups. We found no significant differences in AD between the groups using the three indices (MR vs. HC: Chao1 p.adj = .75, Shannon p.adj = .799, Simpson p.adj = .898; MS vs. HC: Chao1 p.adj = .898, Shannon p.adj = .898, Simpson p.adj = .799; MS vs. MR: Chao1 p.adj = .668, Shannon p.adj = .456, Simpson p.adj = .586).

Age was the only confounding factor to be significantly correlated with increased bacterial AD in all groups using all the three indices tested (Chao1 p.adj = .0004, Shannon p.adj = .0004, Simpson p.adj = .0028; Figure 1A). No significant association was found with other confounding factors (sex: Chao1 p.adj = .416, Shannon p.adj = .973, Simpson p.adj = .898; carbohydrates: Chao1 p.adj = .275, Shannon p.adj = .159, Simpson p.adj = .168; protein: Chao1 p.adj = .416, Shannon p.adj = .068, Simpson p.adj = .099; fats: Chao1 p.adj = .416, Shannon p.adj = .898, Simpson p.adj = .954; fibers: Chao1 p.adj = .968, Shannon p.adj = .954, Simpson p.adj = .777).

FIGURE 1.

FIGURE 1

Open in a new tab

α‐Diversity (AD) correlates with aging, whereas β‐diversity depicts a different bacterial profile in the epilepsy groups compared to healthy controls (HCs). (A) Correlation plots illustrating the associations between three AD indices (Chao1, Shannon Diversity, Simpon's Index) and age. Dots represent individual subjects and are colored based on the group. Black line is the regression line, with a gray area corresponding to the 95% confidence interval. Coefficient and adjusted p‐values are obtained from respective generalized linear models. (B) Principle coordinate analysis (PCoA) plots of Bray–Curtis and Jaccard dissimilarity indices were obtained using the first two axes. The percentage of total variation explained by each axis is shown next to the axis label. Points represent each individual. Ellipses correspond to 95% confidence intervals for each group. MR, medication‐resistant; MS, medication‐sensitive. Pr(>F) is the p value of the F statistic.

3.3. BD depicts a different bacterial profile in the epilepsy groups

BD measures intersample species richness and evenness. When comparing the three groups, the pairwise PERMANOVA test showed significant differences based on both dissimilarity indices for MS versus HC (Bray–Curtis F = 1.783, p = .001; Jaccard F = 1.24, p = .001), and MR versus HC (Bray–Curtis F = 2.24, p = .001; Jaccard F = 1.364, p = .001), indicating that epilepsy is associated with a unique gut bacterial profile. Conversely, no significant differences in the intersample bacterial composition were found between the MR and MS subgroups either with the Bray–Curtis (F = 1.241, p = .104) or with the Jaccard index (F = 1.081, p = .1; Figure 1B).

3.4. Taxonomic analysis reveals a unique bacterial population in the MR subgroup

DA analysis was carried out to investigate differences in the abundance of each taxon at three taxonomic ranks: phylum, family, and genus.

No significant changes between groups were found at the phylum or family level. However, two different genera were obtained from the pairwise comparison of the three groups (Figure 2). Specifically, Hung was more abundant in both the MS and MR groups, compared to HCs (4.95 and 6.72 log2 fold changes, respectively), whereas Eub further differentiated between the MR and MS groups (−4.87 log2 changes).

FIGURE 2.

FIGURE 2

Open in a new tab

Heatmap illustrating the relative abundance distribution of the two significant genera obtained from the differential abundance analysis of the three pairwise comparisons. The relative abundance (rel. abund.) percentiles were transformed into centered log‐ratio (clr) and visualized by color gradient. Samples were clustered based on group information. Right annotation columns show the significant changes obtained from each of the three pairwise comparisons, and the log2 fold change values with the significance stars. *.01 ≤ p.adj < .05, **.001 ≤ p.adj <.01. HC, healthy control; MR, medication‐resistant; MS, medication‐sensitive.

We then constructed a phylogenetic tree showing the ASVs annotated to the statistically significant genera described above (Figure 3). Our results show a similar profile in the MS and HC groups with regard to Eub. On the other hand, the MR group showed lower relative abundance in all the analyzed ASVs. Furthermore, the relative abundance of Hung ASV 1–5 was lowest in the HC group, intermediate in the MS group, and highest in the MR group, suggesting that these populations might play a role in refractory epilepsy.

FIGURE 3.

FIGURE 3

Open in a new tab

Phylogenetic tree showing the amplicon sequence variants (ASVs) annotated to the genera found to significantly change in our population. Rooted tree constructed by alignment of the obtained ASVs was trimmed to include 24 ASVs annotated as Hungatella or [Eubacterium] siraeum group. Heatmap encircling the phylogenetic tree shows the median of centered log‐ratio (clr)‐transformed genus relative abundances in each study group. HC, healthy control; MR, medication‐resistant; MS, medication‐sensitive.

To explore the relationship between different bacterial populations, we performed the ratio between Bacteroidota and Firmicutes, widely accepted to play an important role in maintaining normal intestinal homeostasis. Our data did not reveal any significant differences between the groups (Figure 4A). At the genus level, we investigated whether epilepsy and medication resistance were associated with a dominance in Hung abundance over Eub. Due to the nature of the data, we were not able to perform a straightforward ratio between Hung and Eub. Instead, we treated Hung and Eub as an isolated community and assigned a binary score based on their dominance. Our results show a significantly increased dominance in Hung in the MR versus HC and in the MS versus HC groups (Figure 4B).

FIGURE 4.

FIGURE 4

Open in a new tab

Ratio of relevant phyla and statistically significant genera. (A) Scatterplot of the ratio between abundance of Bacteroidota and Firmicutes phyla in the three groups. (B) Scatterplot of Hungatella (Hung) dominance over [Eubacterium] siraeum group (Eub). (C) Hung/Eub ratio at each sample level in the three groups studied. The relative abundance of Hung and Eub were summed and used as an isolated community. A binary score was assigned based on their proportional abundance: “1” indicated Hung dominance (≥50% abundance) and “0” indicated Eub dominance. *p ≤ .05, ***p ≤ .0001. HC, healthy control; MR, medication‐resistant; MS, medication‐sensitive.

3.5. Distinctive KEGG pathways identified in MR epilepsy

The gut microbiota produces metabolites that can travel systemically in the circulation and enter the central nervous system (CNS) to regulate brain activity. To gain insight on the functional potential of the populations characterized, we performed metabolic pathway prediction using PICRUSt2. The results revealed an upregulation of D‐glutamine and D‐glutamate metabolism in MS, compared to MR patients (p = .03, p.adj = .358). Furthermore, the MS group showed downregulation of fatty acid biosynthesis compared to the HC group (p = .046, p.adj = .62). Lastly, fructose and mannose metabolism was highest in MR patients, compared to MS (p = .007, p.adj = .346) and HC (p = .014, p.adj = .455; Figure 5).

FIGURE 5.

FIGURE 5

Open in a new tab

Significant changes in Kyoto Encyclopedia of Genes and Genomes pathway relative abundances between groups. Individual samples are represented in dots on left‐hand side, whereas the boxplots on the right side depict the median line in black and the interquartile interval with minimum and maximum values. The triangle shapes show outlier data points. Raw and adjusted p values are shown for significant changes. HC, healthy control; MR, medication‐resistant; MS, medication‐sensitive.

4. DISCUSSION

MGBA dysfunction is emerging as a new pathogenic mechanism in neurologic disorders. In particular, patients with epilepsy present intestinal dysbiosis, which can modulate disease outcome. Both nutritional habits and drug treatment can alter the gut microbiota composition and affect brain activity, excitotoxicity, and seizures. Here, we investigated gut microbiota populations in pediatric epileptic patients and the role of confounding factors, such as age, sex, and dietary intake, on the resistance to ASMs.

One third of all epilepsy patients suffer from medication‐resistant epilepsy, which has been correlated with genetic variation, therapeutic target, drug transporter type, drug pharmacokinetics, and severity of seizures. 13 , 14 The most common mechanism of action for ASMs is via interaction with ion channels and neurotransmitters to modulate neuronal transmission. Interestingly, new evidence has recently shown that the gut microbiota might also be involved in mediating ASM effects via a bidirectional communication system. Preclinical studies revealed that availability of certain ASM active metabolites can influence gut microbiota growth and composition. 32 In turn, the gut flora can directly affect treatment response by enzymatically transforming the structure of the drug and thereby altering its bioavailability, bioactivity, and toxicity. 15 Furthermore, a recent study found significant differences between the gut microbiota in MS and MR adult epilepsy patients, 33 suggesting a relationship between gut flora composition and medication response status. To our knowledge, this is the first study to investigate this relationship in pediatric patients. Gut microbiota plasticity is highest in children and incrementally diminishes throughout the course of life. 34 Implementing noninvasive interventions in children, such as promoting healthy nutritional habits, might have profound and long‐lasting beneficial effects on disease outcome and medication response.

Our results show similar bacterial richness (AD) between the studied groups, with a linear increase correlated with age. This finding is consistent with the literature; starting from early intrauterine life, the human gut microbiota is shaped by maternal and environmental factors. 6 Early reports suggests that around the 3rd year of age, our core microbiota becomes similar to that of adults. However, complete maturation might take longer in some cases, 6 and the gut microbiota is shaped by endogenous and external factors (e.g., diet, physical activity) up to adolescence. Hence, future studies may further subgroup patients into infants (<3 years old), preschool children (3–6 years old), school‐aged children (6–12 years old), and lastly teenagers (12–18 years old) to fully uncover the variability of this interesting window for the gut microbiota.

Because AD only takes into consideration the general richness of single samples, but not the types of microbial communities present, two samples may have a very different metagenomic profile, despite having a similar AD. Evaluating the intersample variability (BD), we showed a significant difference between the epilepsy and control groups, revealing that the disease status is associated with gut dysbiosis. Interestingly, there were no statistical differences between MS and MR epilepsy patients, suggesting that medication responsiveness may not be directly associated with a general status of dysbiosis.

To investigate whether specific bacterial populations were associated with medication responsiveness, we performed taxonomic analysis. No significant changes at the phylum level were found in our populations. However, at the genus level, specific bacterial species were specifically associated with the epilepsy and HC groups, as well as between the MR and MS subgroups. These observations suggest that medication resistance might be associated with minor alterations of specific bacterial populations rather than broader changes, and that fine‐tuning of the gut microbiota can have profound changes on health and disease. In particular, we found a significantly higher abundance of Hung in the epilepsy groups (more pronounced in the MR group) compared to the HC group, highlighting an association with the disease. Hung belongs to the Clostridiaceae family and is a gram‐positive, anaerobic bacteria known to be highly resistant to antibiotics. 35 Moreover, different strains of this genera are associated with neurodegenerative and progressive neurological disorders, such as multiple sclerosis, 36 suggesting that Hung could play a role in different neurological conditions. A recent study also demonstrated that patients with recurrent major depressive disorder presented a higher abundance of Hung. 37 Whether these changes could anticipate the disease onset or are a direct consequence of the disease remains elusive.

Furthermore, the abundance of Eub was decreased in the MR versus the MS group. The Eubacterium spp. are butyrate producers, 38 with the only exception being E. oxidoreducens. 39 Butyrate is one of the three main short‐chain fatty acids (SCFAs), which can promote health and are locally produced by gut bacteria. 38 Once they cross the gut barrier and enter the bloodstream, these compounds can reach the CNS and prevent/reverse neurological conditions. 40 SCFAs exert antiseizure effects by modulating the ratio between inhibitory and excitatory neurotransmitters in the brain. 40 SCFAs also act on a variety of cell types in the CNS to reduce neuroinflammation, increase fatty acid oxidation, promote blood–brain barrier integrity, and regulate cell–cell interactions, mitochondrial function, and gene expression. 41 , 42 Due to these functions, gut dysbiosis and the concomitant alteration of SCFA levels have been implicated as potential risk factors in many neurological conditions, including epilepsy. Furthermore, Eubacterium has also been identified as a producer of γ‐aminobutyric acid (GABA), an inhibitory neurotransmitter known to control seizures. GABA has also been proposed as a potential postbiotic mediator for neurological disorders. 43 A recent study has demonstrated that Eubacterium was one of the main genera to increase in pediatric patients with medication‐resistant epilepsy and positively respond to the ketogenic diet, 44 which in turn is known to increase beneficial species and mediate ASM response in treated individuals. 45 Taken together, these observations suggest that Eubacterium could play a role in medication response; however, longitudinal clinical trials are needed to establish causation.

In the gut, different microbial communities coexist in a multifaceted ecosystem engaging in complex interactions. 1 For these reasons, it is important to investigate the proportion of a population in relation to another. The ratio between Bacteroidota and Firmicutes, for instance, is known to play an important role in maintaining normal intestinal homeostasis and has been associated with different neurological disorders. 1 Our data did not reveal any significant differences between the groups at the phylum level. However, at the genus level we found a dominance in Hung compared to Eub in both epilepsy groups compared to the HCs, highlighting an association between these populations and epilepsy.

A limitation of the current study is the lack of metabolic correlates to our metagenomic findings. To gain insight on the functional potential of the populations characterized, we performed metabolic pathway prediction using PICRUSt2. The results revealed significant differences in D‐glutamine and D‐glutamate metabolism, fatty acid biosynthesis, and fructose and mannose metabolism, pathways shown to play a role in epilepsy and in medication resistance. 46 , 47 , 48 Interestingly, SCFAs have been shown to modulate lipid, glutamine, and glutamate metabolism, 49 highlighting a potential association between these metabolic pathways and the abundance of SCFA‐producers in the gut.

Furthermore, the cross‐sectional nature of the study and the choice of monotherapy versus polytherapy as a practical, although potentially imperfect, proxy for medication resistance limit its capability to disentangle the observed association between certain bacterial species and treatment response and establish causality.

Given the potential impact of most ASMs (e.g., valproate, topiramate, cannabidiol, ethosuximide, and clobazam) on the MGBA, future longitudinal studies with pretreatment timepoints could elucidate whether the differences found in this study are independent of treatment‐induced changes and can therefore act as true predictive biomarkers of therapeutic response. In our study, there were observable differences in the averages of clinical measurements between the groups, which may suggest potential trends that warrant further investigation. However, our study did not perform differentiation analysis to predict group classifications based on these data.

Modulating the MGBA with pre‐/probiotics can also be an effective strategy to improve medication responsiveness in epilepsy patients and should be explored in the future as a therapeutic approach. Lastly, although efforts have been made to try to detangle the possible confounding factors associated with the gut microbiota composition (e.g., age, diet), future studies should go deeper in the analysis, considering more homogenous genetic backgrounds.

Understanding the impact of the MGBA in epilepsy and medication response and exploring how to modulate this axis to improve the clinical outcome is crucial to aid the discovery of predictive biomarkers and develop early intervention plans to enhance medication responsiveness and control frequency and severity of seizures, ultimately ameliorating the quality of life of the affected individuals.

5. CONCLUSIONS

Epilepsy is a serious health condition, and existing therapies target symptoms, rather than addressing the underlying abnormal biological processes. Epilepsy patients often exhibit gut dysbiosis, but it is still unclear how this dysregulation affects seizure development and the response to ASMs. There is a need to advance knowledge of epilepsy biology to identify predictive biomarkers and guide the development of novel therapies and early interventions for this multifaceted disease. Our study shows a significantly lower abundance of Eub in medication‐resistant epilepsy patients. Interestingly, Eubacterium hallii has been recently identified as a candidate for next generation probiotics, 50 with exciting potential for the prevention and treatment of several dysbiosis‐associated diseases. Another study shows that ketogenic diet reversed the abundance of epilepsy‐associated genera in children with medication‐refractory epilepsy. 44

As dysbiosis is associated with epilepsy and medication resistance, modulating the gut microbiota composition through the use of probiotics and diet could provide an efficacious and noninvasive intervention plan to improve patient symptoms and quality of life. However, additional studies are needed to fully characterize the microbial signatures associated with refractory seizures and establish a causal relationship between dysbiosis and medication responsiveness.

AUTHOR CONTRIBUTIONS

Antonella Riva: Collection of data; drafting. Eray Sahin: Data analysis; drafting. Greta Volpedo: Drafting. Noemi Teresa Catania, Isabel Venara, Valentina Biagioli, Ganna Balagura, Elisabetta Amadori, Carmen De Caro, Emanuele Cerulli Irelli, Carlo Di Bonaventura: Support in collecting the data. Federico Zara, Osman Ugur Sezerman, Emilio Russo: Revision of the data. Pasquale Striano: Conception of the study; revision and final approval of the manuscript. All authors agree to be accountable for all aspects of the work.

CONFLICT OF INTEREST STATEMENT

None of the authors has any conflict of interest to disclose.

ETHICS STATEMENT

We confirm that we have read the Journal's position on issues involved in ethical publication and affirm that this report is consistent with those guidelines. All methods were performed in accordance with the ethical standards as laid down in the Declaration of Helsinki and its later amendments or comparable ethical standards.

PATIENT CONSENT STATEMENT

The present study was approved by the Comitato Unico Regionale Regione Liguria, and patients signed written informed consent (Protcol MOBi‐DEE).

Supporting information

Data S1.

EPI-66-2268-s001.docx (114.8KB, docx)

Data S2.

EPI-66-2268-s002.xlsx (1.1MB, xlsx)

ACKNOWLEDGMENTS

Supported by PNRR‐MUR‐M4C2 PE0000006 Research Program “MNESYS—A Multiscale Integrated Approach to the Study of the Nervous System in Health and Disease” and by MDBR‐23‐004‐STXBP1 “Fingerprinting a Multiomics Biomarker Profile in Patients With STXBP1‐RD.” IRCCS Istituto G. Gaslini is a member of the European Reference Network (ERN) EpiCARE. Open access funding provided by BIBLIOSAN.

Riva A, Sahin E, Volpedo G, Catania NT, Venara I, Biagioli V, et al. Medication‐resistant epilepsy is associated with a unique gut microbiota signature. Epilepsia. 2025;66:2268–2284. 10.1111/epi.18367

Antonella Riva, Eray Sahin and Greta Volpedo share first authorship.

DATA AVAILABILITY STATEMENT

The data that support the findings of this study are available in the article. The sequencing data were uploaded to BioProject (ID PRJNA1133091) and are publicly available. The scripts used for the data analysis in R can be found in the GitHub repository: https://github.com/eray‐sahin/Medication‐resistant_epilepsy_is_associated_with_a_unique_gut_microbiota_signature. If additional data are required, they can be requested from the corresponding author.

REFERENCES

  • 1. Thursby E, Juge N. Introduction to the human gut microbiota. Biochem J. 2017;474(11):1823–1836. 10.1042/BCJ20160510 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 2. Rodrigues‐Amorim D, Rivera‐Baltanás T, Regueiro B, Spuch C, de Las Heras ME, Vázquez‐Noguerol Méndez R, et al. The role of the gut microbiota in schizophrenia: current and future perspectives. World J Biol Psychiatry. 2018;19(8):571–585. 10.1080/15622975.2018.1433878 [DOI] [PubMed] [Google Scholar]
  • 3. Gong X, Liu XU, Chen C, Lin J, Li A, Guo K, et al. Alteration of gut microbiota in patients with epilepsy and the potential index as a biomarker. Front Microbiol. 2020;11(51):517797. 10.3389/fmicb.2020.517797 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 4. Citraro R, Lembo F, De Caro C, Tallarico M, Coretti L, Iannone LF, et al. First evidence of altered microbiota and intestinal damage and their link to absence epilepsy in a genetic animal model, the WAG/Rij rat. Epilepsia. 2021;62(2):529–541. 10.1111/epi.16813 [DOI] [PubMed] [Google Scholar]
  • 5. Lan R, Reeves PR. Intraspecies variation in bacterial genomes: the need for a species genome concept. Trends Microbiol. 2000;8(9):396–401. 10.1016/s0966-842x(00)01791-1 [DOI] [PubMed] [Google Scholar]
  • 6. Derrien M, Alvarez AS, de Vos WM. The gut microbiota in the first decade of life. Trends Microbiol. 2019;27(12):997–1010. 10.1016/j.tim.2019.08.001 [DOI] [PubMed] [Google Scholar]
  • 7. Liu CG, Liu S, Chen Z, Wu Y, Lou T, Wang J, et al. Gut microbiome distinguishes patients with epilepsy from healthy individuals. Front Microbiol. 2022;12:696632. 10.3389/fmicb.2021.696632 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8. Türay S, Cangür Ş, Kahraman G, Kayabaşı E, Çetiner ÖF, Aydın B, et al. Can the gut microbiota serve as a guide to the diagnosis and treatment of childhood epilepsy? Pediatr Neurol. 2023;145:11–21. 10.1016/j.pediatrneurol.2023.04.006 [DOI] [PubMed] [Google Scholar]
  • 9. Ceccarani C, Viganò I, Ottaviano E, Redaelli MG, Severgnini M, Vignoli A, et al. Is gut microbiota a key player in epilepsy onset? A longitudinal study in drug‐Naïve children. Front Cell Infect Microbiol. 2021;3(11):749509. 10.3389/fcimb.2021.749509 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 10. Fisher RS, Cross JH, French JA, Higurashi N, Hirsch E, Jansen FE, et al. Operational classification of seizure types by the International League Against Epilepsy: Position Paper of the ILAE Commission for Classification and Terminology. Epilepsia. 2017;58(4):522–530. 10.1111/epi.13670 [DOI] [PubMed] [Google Scholar]
  • 11. Rastin C, Schenkel LC, Sadikovic B. Complexity in genetic epilepsies: a comprehensive review. Int J Mol Sci. 2023;24(19):14606. 10.3390/ijms241914606 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 12. Symonds JD, McTague A. Epilepsy and developmental disorders: next generation sequencing in the clinic. Eur J Paediatr Neurol. 2020;24:15–23. 10.1016/j.ejpn.2019.12.008 [DOI] [PubMed] [Google Scholar]
  • 13. Kwan P, Arzimanoglou A, Berg AT, Brodie MJ, Allen Hauser W, Mathern G, et al. Definition of drug resistant epilepsy: consensus proposal by the ad hoc task force of the ILAE commission on therapeutic strategies. Epilepsia. 2010;51(6):1069–1077. 10.1111/j.1528-1167.2009.02397.x [DOI] [PubMed] [Google Scholar]
  • 14. Gesche J, Beier CP. Drug resistance in idiopathic generalized epilepsies: evidence and concepts. Epilepsia. 2022;63(12):3007–3019. 10.1111/epi.17410 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 15. Koppel N, Maini Rekdal V, Balskus EP. Chemical transformation of xenobiotics by the human gut microbiota. Science. 2017;356(6344):eaag2770. 10.1126/science.aag2770 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 16. Zhou C, Gong S, Xiang S, Liang L, Hu X, Huang R, et al. Changes and significance of gut microbiota in children with focal epilepsy before and after treatment. Front Cell Infect Microbiol. 2022;12:965471. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 17. Jiang H, Lei R, Ding SW, Zhu S. Skewer: a fast and accurate adpter trimmer for next‐generation sequencing paired‐end reads. BMC Bioinformatics. 2014;15:182. 10.1186/1471-2105-15-182 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 18. Callahan BJ, McMurdie PJ, Rosen MJ, Han AW, Johnson AJ, Holmes SP. DADA2: high‐resolution sample inference from Illumina amplicon data. Nat Methods. 2016;13(7):581–583. 10.1038/nmeth.3869 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 19. Andrews S. FastQC: a quality control tool for high throughput sequence data. 2010. http://www.bioinformatics.babraham.ac.uk/projects/fastqc
  • 20. Chen S, Zhou Y, Chen Y, Gu J. fastp: an ultra‐fast all‐in‐one FASTQ preprocessor. Bioinformatics. 2018;34(17):i884–i890. 10.1093/bioinformatics/bty560 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 21. Bolyen E, Rideout JR, Dillon MR, Bokulich NA, Abnet CC, al‐Ghalith GA, et al. Reproducible, interactive, scalable and extensible microbiome data science using QIIME 2. Nat Biotechnol. 2019;37(8):852–857. 10.1038/s41587-019-0209-9 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 22. Martin M. CUTADAPT removes adapter sequences from high‐throughput sequencing reads. EMBnet J. 2011;17(1). 10.14806/ej.17.1.200 [DOI] [Google Scholar]
  • 23. Kaehler BD, Bokulich NA, McDonald D, Knight R, Caporaso JG, Huttley GA. Species abundance information improves sequence taxonomy classification accuracy. Nat Commun. 2019;10(1):4643. 10.1038/s41467-019-12669-6 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 24. Robeson MS II, O'Rourke DR, Kaehler BD, Ziemski M, Dillon MR, Foster JT, et al. RESCRIPt: reproducible sequence taxonomy reference database management. PLoS Comput Biol. 2021;17(11):e1009581. 10.1371/journal.pcbi.1009581 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 25. Quast C, Pruesse E, Yilmaz P, Gerken J, Schweer T, Yarza P, et al. The SILVA ribosomal RNA gene database project: improved data processing and web‐based tools. Nucleic Acids Res. 2013;41(D590):6. 10.1093/nar/gks1219 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 26. Pedregosa F, Varoquaux G, Gramfort A, Michel V, Thirion B, Grisel O, et al. Scikit‐learn: machine learning in python. J Mach Learn Res. 2011;12:2825–2830. [Google Scholar]
  • 27. Douglas GM, Maffei VJ, Zaneveld JR, Yurgel SN, Brown JR, Taylor CM, et al. PICRUSt2 for prediction of metagenome functions. Nat Biotechnol. 2020;38(6):685–688. 10.1038/s41587-020-0548-6 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 28. Mallick H, Rahnavard A, Mclver LJ, Ma S, Zhang Y, Nguyen LH, et al. Multivariable association discovery in population‐scale meta‐omics studies. PLoS Comput Biol. 2021;17(11):e1009442. 10.1371/journal.pcbi.1009442 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 29. Gu Z, Eils R, Schlesner M. Complex heatmaps reveal patterns and correlations in multidimensional genomic data. Bioinformatics. 2016;32(18):2847–2849. 10.1093/bioinformatics/btw313 [DOI] [PubMed] [Google Scholar]
  • 30. Paradis E, Schliep K. Ape 5.0: an environment for modern phylogenetics and evolutionary analyses in R. Bioinformatics. 2019;35(3):526–528. 10.1093/bioinformatics/bty633 [DOI] [PubMed] [Google Scholar]
  • 31. Letunic I, Bork P. Interactive tree of life (iTOL) v5: an online tool for phylogenetic tree display and annotation. Nucleic Acids Res. 2021;49(W1):W293–W296. 10.1093/nar/gkab301 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 32. Holmes M, Flaminio Z, Vardhan M, Xu F, Li X, Devinsky O, et al. Cross talk between drug‐resistant epilepsy and the gut microbiome. Epilepsia. 2020;61(12):2619–2628. [DOI] [PubMed] [Google Scholar]
  • 33. Lee H, Lee S, Lee DH, Kim DW. A comparison of the gut microbiota among adult patients with drug‐responsive and drug‐resistant epilepsy: an exploratory study. Epilepsy Res. 2021;172:106601. [DOI] [PubMed] [Google Scholar]
  • 34. Thriene K, Michels KB. Human gut microbiota plasticity throughout the life course. Int J Environ Res Public Health. 2023;20(2):1463. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 35. Ramlachan N, Anderson RC, Andrews K, Laban G, Nisbet DJ. Characterization of an antibiotic resistant Clostridium hathewayi strain from a continuous‐flow exclusion chemostat culture derived from the cecal contents of a feral pig. Anaerobe. 2007;13(3–4):153–160. [DOI] [PubMed] [Google Scholar]
  • 36. iMSMS Consortium . Gut microbiome of multiple sclerosis patients and paired household healthy controls reveals associations with disease risk and course. Cell. 2022;185(19):3467–3486. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 37. Xie Z, Huang J, Sun G, He S, Luo Z, Zhang L, et al. Integrated multi‐omics analysis reveals gut microbiota dysbiosis and systemic disturbance in major depressive disorder. Psychiatry Res. 2024;334:115804. 10.1016/j.psychres.2024.115804 [DOI] [PubMed] [Google Scholar]
  • 38. Mukherjee A, Lordan C, Ross RP, Cotter PD. Gut microbes from the phylogenetically diverse genus Eubacterium and their various contributions to gut health. Gut Microbes. 2020;12(1):1802866. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 39. Krumholz LR, Bryant MP. Eubacterium oxidoreducens sp. Nov. requiring H2 or formate to degradate gallate, pyrogallol, phloroglucinol and quercetin. Arch Microbiol. 1986;144:8–14. [Google Scholar]
  • 40. Silva YP, Bernardi A, Frozza RL. The role of short‐chain fatty acids from gut microbiota in gut‐brain communication. Front Endocrinol (Lausanne). 2020;11:25. 10.3389/fendo.2020.00025 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 41. Yang LL, Millischer V, Rodin S, MacFabe DF, Villaescusa JC, Lavebratt C. Enteric short‐chain fatty acids promote proliferation of human neural progenitor cells. J Neurochem. 2020;154(6):635–646. 10.1111/jnc.14928 [DOI] [PubMed] [Google Scholar]
  • 42. Ney LM, Wipplinger M, Grossmann M, Engert N, Wegner VD, Mosig AS. Short chain fatty acids: key regulators of the local and systemic immune response in inflammatory diseases and infections. Open Biol. 2023;13(3):230014. 10.1098/rsob.230014 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 43. Braga JD, Thongngam M, Kumrungsee T. Gamma‐aminobutyric acid as a potential postbiotic mediator in the gut‐brain axis. NPJ Sci Food. 2024;8(1):16. 10.1038/s41538-024-00253-2 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 44. Gong X, Cai Q, Liu X, An D, Zhou D, Luo R, et al. Gut flora and metabolism are altered in epilepsy and partially restored after ketogenic diets. Microb Pathog. 2021;155:104899. [DOI] [PubMed] [Google Scholar]
  • 45. Olson CA, Vuong HE, Yano JM, Liang QY, Nusbaum DJ, Hsiao EY. The gut microbiota mediates the anti‐seizure effects of the ketogenic diet. Cell. 2018;173(7):1728–1741. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 46. Wang W, Wu Y, Li X, Li L, Sun K, Yan S. Altered plasma glutamate and lutamine levels in patients with drug‐resistant and drug‐responsive symptomatic focal epilepsy. Neurosciences (Riyadh). 2021;26(4):315–322. 10.17712/nsj.2021.4.20210041 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 47. Riva A, Sahin E, Volpedo G, Petretto A, Lavarello C, Di Sapia R, et al. Identification of an epilepsy‐linked gut microbiota signature in a pediatric rat model of acquired epilepsy. Neurobiol Dis. 2024;194:106469. 10.1016/j.nbd.2024.106469 [DOI] [PubMed] [Google Scholar]
  • 48. McDonald TS, Neal ES, Borges K. Fructose 1,6‐bisphosphate is anticonvulsant and improves oxidative glucose metabolism within the hippocampus and liver in the chronic pilocarpine mouse epilepsy model. Epilepsy Behav. 2021;122:108223. 10.1016/j.yebeh.2021.108223 [DOI] [PubMed] [Google Scholar]
  • 49. den Besten G, Van Eunen K, Groen AK, Venema K, Reijngoud DJ, Bakker BM. The role of short‐chain fatty acids in the interplay between diet, gut microbiota, and host energy metabolism. J Lipid Res. 2013;54(9):2325–2340. 10.1194/jlr.R036012 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 50. Udayappan S, Manneras‐Holm L, Chaplin‐Scott A, Belzer C, Herrema H, Dallinga‐Thie GM, et al. Oral treatment with Eubacterium hallii improves insulin sensitivity in db/db mice. NPJ Biofilms Microbiomes. 2016;2:16009. 10.1038/npjbiofilms.2016.9 [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Data S1.

EPI-66-2268-s001.docx (114.8KB, docx)

Data S2.

EPI-66-2268-s002.xlsx (1.1MB, xlsx)

Data Availability Statement

The data that support the findings of this study are available in the article. The sequencing data were uploaded to BioProject (ID PRJNA1133091) and are publicly available. The scripts used for the data analysis in R can be found in the GitHub repository: https://github.com/eray‐sahin/Medication‐resistant_epilepsy_is_associated_with_a_unique_gut_microbiota_signature. If additional data are required, they can be requested from the corresponding author.

Articles from Epilepsia are provided here courtesy of Wiley

RESOURCES