MiCrobiota-gut-brain Axis in Resistant Epilepsy

Specific Aims and Experimental Design Accumulating evidence demonstrates strong associations between the GM and epilepsy, but no clear trend is emerging yet, which may be due to the small cohorts and environmental confounding factors. Whether microbial dysbiosis contributes to seizure triggering or is caused by anti-seizure medications (ASMs) remains to be elucidated. To better understand the interplay between MGBA - Microbiota-Gut-Brain-Axis and epilepsy, CARE project will enroll and longitudinally follow up to one year subjects at seizure onset and subjects with DRE, before and after procedures aimed at reducing the burden of such a chronic disabling condition (i.e., resective surgery, VNS, and KD). AIM 1 specific tasks will be: 1.1 To enroll subjects at the seizure onset (naïve), and DRE patients divided in three populations according to the treatment procedure: DRE undergoing resective surgery (DRE-surgery), DRE undergoing VNS (DRE-VNS), and DRE on diet therapy (DRE-KD) 1.2 To set up a detailed database including clinical, biochemical, neurophysiological, neuroradiological and genetic records 1.3 Data analysis The enrolment goal has been set as at least 50 subjects at the seizure onset (naïve), 50 DRE-surgery, 10 DRE DRE-VNS, and 10 DRE-KD. Patients will visit the clinical center as an inpatient/outpatient regimen and remain in the study for about 15 months (from screening to last follow-up visit). Study time points are as follows: t0, at the enrolment (before ASMs for naïve and before any procedures for DRE); t1, at 1 month; t2, at 6 months; and t3, at 12 months (after ASM starting for naïve and after surgery, VNS, or KD). The longitudinal nature of the study will compensate for well-known interindividual differences in gut microbiome features. At t0, each participant will undergo a comprehensive clinical parameter evaluation: neurological and physical examination (number of seizures/month through the evaluation of seizure diaries, EEG, neurobehavioral comorbid symptoms by dedicated scales, and detailed drug regimen assessment) assessment of gastrointestinal (GI) symptoms (by a modified 6-item GI severity index -GISI- and the Bristol stool scale) a dietary survey, by a combined 3-day food diary and 24-hour recall, and anthropometric evaluation (body-mass index -BMI- and BMI z-score for children) quality of life assessment for children with Pediatric Quality of Life Inventory TM for Epilepsy (PedsQL TM) and for adults with the QOLIE-31 (Quality of Life in Epilepsy Inventory- 31 item) stool and blood sample collection. For seizure frequency, enrolled DRE will be asked to fill the seizure diary during the 30 days before entering the study. Data will be recorded as baseline. After 1 month (t1), we will contact each subject to record possible rapid changes of outcome measures. Clinical parameters will be assessed by online call by a clinician from each enrolling center, thus facilitating the compliance to the protocol. Fecal samples will be collected and stored. At t2 (6 months) and t3 (12 months), each subject will undergo a complete assessment as depicted for t0. For clinical and nutritional data, descriptive analyses will be conducted. Continuous variables will be analyzed with parametric (normally distributed variables) and non-parametric (non normally distributed) tests, while categorical variables will be analyzed using the Chi-square test or Fisher's test. Spearman correlations will be applied to estimate the strength of association between GM features (AIM 2 and 3), DNAmet and metabolites (AIM 3) and clinical response parameters (AIM 1). Two-tailed P \< 0.05 will be considered statistically significant. Specific AIM 2: We aim at assessing the interrelation between clinical outcome and GM composition in longitudinal cohorts of patients with epilepsy (detailed in AIM 1) before and after ASMs and surgery or VNS, when DRE. GM will be characterized at all the planned time points with the aim at identifying pioneering microbial changes (t0) that could represent biomarkers for therapeutic outcomes as well as common changes in responder or non responder patients (t1, t2, and t3). Our preliminary findings suggest that ASM regimens can act on GM, inducing statistically significant microbial changes in a small cohort of adult subjects with epilepsy taking valproic acid compared with healthy controls. Similarly, children with refractory epilepsy display a reduction in richness and show a group-dependent clusterization when compared to healthy siblings/relatives. Despite the MGBA impact on central nervous system (CNS) diseases has been quite delineated, the contribution of each player within the axis is still to be fully elucidated. The CARE project will dissect the axis in resistant epilepsy by acting on CNS (surgery), top-down communication (VNS), and the environmental factor most affecting the microbiota, i.e., the diet (KD). AIM 2 specific tasks are: 2.1 To evaluate the GM composition changes occurring soon after the seizure onset before ASM starting and during drug therapy in naïve epilepsy patients; 2.2 To evaluate the impact of different therapies (surgery, VNS, and KD) on the GM composition of DRE patients; 2.3 To evaluate the association of baseline GM composition vs. indicators of response to drug therapy in naïve patients, and to surgery, VNS and KD in DRE patients. Eligible participants will be clinically characterized (see AIM 1) and stool and blood samples collected and stored at -80°C. A fraction of collected fecal samples will be used for bacterial DNA extraction by the commercial kit DNeasy® PowerSoil® Pro (Qiagen,Germany). The hypervariable V3-V4 regions of 16S Svedberg unit rRNA - ribosomal RNA gene will be amplified with a two-step barcoding approach and sequenced with an Illumina MiSeq platform. Amplicon sequence variants (ASVs) will be identified from 16S paired-end sequencing using the Divisive Amplicon Denoising Algorithm pipeline (DADA2, version 1.18.0) and taxonomy will be assigned through a naive Bayesian classifier based on the 11.8 release of the RDP - Ribosomal Database Project database using the DADA2-formatted Genome Taxonomy Database 16S rRNA database. Specific AIM 3 We aim at investigating the differential impact of the diverse epilepsy therapies on the GM metabolic functions and the rearrangement of DNAmet profiles. In addition, monitoring the baseline status and the changes occurring in the GM metaproteome and in the host methylome in longitudinal cohort studies might enable us to hypothesize the role of microbial (GM) and host functions (DNAmet) on clinical outcomes. The human GM metabolic status is heavily affected by dietary intervention. Recently (unpublished preliminary data) metaproteomic analyses enabled us to measure specific changes in microbial pathways following KD at 3 and 4 weeks, returning to baseline 10 days after KD suspension. VNS is expected to have a significant impact on GM metabolic functions, given the VN control over gut motility, gut secretions, liver functions, and mucosal barrier. In turn, GM metabolic activities can modify the drug bioavailability to their cell/tissue targets. Moreover, the GM can synthesize a large variety of metabolites serving as epigenetic substrates, co-factors or regulators of epigenetic enzyme activity, including DNA methylation or histone modifications. In addition, intrinsic molecules associated to bacterial taxa (LPS, peptidoglycan, flagellin, etc) have proinflammatory roles and host immune response to them is also accompanied to DNAmet changes. Finally, pairing the functional analyses of both GM and host DNAmet can highlight the role of microbiota-sensitive epigenetic variations. To obtain data on GM metabolic functions and on DNAmet serving in longitudinal clinical studies, we will use the same epilepsy patients' cohorts described in AIM 1. AIM 3 specific tasks are: 3.1 To evaluate the GM functional features and DNAmet profile changes occurring soon after the start and during drug therapy in naïve patients. 3.2 To evaluate the impact of different therapies (surgery, VNS, and KD) on the GM functions and DNAmet profile of DRE patients. 3.3 To evaluate the association of baseline GM functions and DNAmet profile vs. indicators of response to drug therapy in naïve patients, and to surgery, VNS and KD in DRE patients. A fraction of stool samples (as collected in aim 1) will be used to obtain the GM metaproteome profiles. Fecal samples will be subjected to functional profiling through shotgun metaproteomic analysis. Gut metaproteomic changes will be monitored at the combination of specific taxon/function levels, to assign modified metabolic pathways and/or protein (i.e., enzyme) functions to their respective taxonomic group, including taxa belonging to either bacteria or fungi. Blood samples will be used to perform genome-wide DNA methylation measurement with the Illumina EPIC array. Bisulfite converted DNA will be subjected to DNAm profiling using the Illumina Infinium Methylation EPIC Beadchip array, consisting of around 850,000 CpG - Cytosine Guanine (linear dinucleotide) sites over the whole genome. Alpha diversity will be calculated for metagenomic and metaproteomic data as richness and Shannon's index. The Principal Component Partial R-Squared (PCPR2) method will be applied to calculate the proportion of variability in the whole omic datasets explained by the variable "therapies" or "before" versus "ongoing" therapy. The set of peptides most discriminating between groups will be identified through discriminant analyses (DA) on log-transformed peptide intensities using sparse Partial Least Squared regression (sPLS) implemented in the Bioconductor package mixOmics. The methylation data will be pre-processed with standard routines for quality control (low quality samples will be removed), batch correction, removal of sex chromosome, and normalization. For the main statistical analysis, the association between the methylation level at each CpG and the patient's metadata (including response to therapy) will be quantified by fitting logistic regression models including as predictor the pseudocontinuous M-value of methylation at each CpG. Methods of data collection Database We will create a web-based database to be filled for all the patients with epilepsy entering the study at each clinical unit. Anthropometric measures and physical examination will be recorded at all the time points. Regarding epilepsy, the database will include information about family history of epilepsy; previous neonatal, febrile, or acute seizures; age at seizure onset; the presence of neurologic, psychiatric and cognitive disorders, seizure type at onset and during follow-up, ictal and interictal EEG activity, and neuroimaging findings. Seizure frequency at baseline and during follow-up will be recorded with seizure diaries. In accordance with the terminology proposed by the ILAE - International League Against Epilepsy, the epilepsies due to specific etiologies consistent with the clinical presentation will be defined as symptomatic. Etiology of epilepsy will be reported whenever available. When etiology cannot be clearly identified, epilepsy will be classified as of unknown etiology. We will collect information about current ASM treatment and all of the ASMs received in the past. Any concomitant therapy taken during the study will be recorded. Regarding gastrointestinal function, we will enter data of the Gastrointestinal Severity Index (GISI) and Bristol Stool Form Scale (BSFS) for all the time points. For each patient, we will enter information about the specific treatment used (epilepsy surgery, VNS or KD) and eventual complications/adverse events. Dietary survey for patients not undergoing ketogenic diet The dietetic and nutritional evaluation will consist of a review of diet and compliance at t0, t1, t2, and t3 through the use of a 3-day food diary. Patients and caregivers will be provided with detailed instructions on how to fill the diary that will encompass two weekly days and one day of the weekend. Metadieta® software will be used to collect participants' nutritional data, dietary micronutrients, and macronutrients at baseline and at each time point. Scales for quality life assessment To assess the adult patient's quality of life, a scoring scale (QOLIE-31 VERSION 1.0) will be applied at baseline (t0) and during the follow-up (t3). The scale encompasses 31 questions present within quality of life inventory in epilepsy-31 items (QOLIE-31). The T-score range is from 0 to 100, and classified the quality of life in three groups - \<30 (poor), moderate (31-50), and better (\>50). The Pediatric Quality of Life Inventory 4.0 Generic Core Scales (PedsQL) is designed for patients aged 2-25. The Italian validated form, is a questionnaire consisting of 23 items, categorized into four subscales: physical, emotional, social, and school functioning as well as three summary scores (total score, physical health summary score, and psychosocial health summary score). Scaled scores are standardized and range from 0-100, with higher scores indicating better quality of life. Methodologies and statistical analyses Statistic plan The main objective of the current proposal is to investigate changes in microbiota characteristics in relation to seizure control in patients with DRE undergoing different therapeutic regimens. Besides seizure control, other parameters such as the improvement of associated comorbidities and the quality of life will be taken into consideration. We have defined primary and secondary endpoints and measurable parameters for assessing the study results. Timing of analysis data Spearman correlations will be applied to estimate the strength of association between gut microbiota and methylation features and objective response parameters. Categorical variables will be reported by number of cases and tested using the Chi-square test or Fisher's exact test. Continuous variables of clinical data will be described by mean SD - Standard Deviation and compared using Mann- Whitney U test or Kruskal-Wallis H nonparametric test. Analyses will be adjusted for age, sex, and potential clinical confounders. The association between microbial features and methylation will be tested through linear regression models with Methylation values measured at time 0, 1 and 12 as a dependent variable, the corresponding microbial feature as predictor and age, sex, BMI as confounders.The association between the methylation level at each CpG and the treatment response (\>50% seizure reduction) at 6 and or 12 months will be quantified by fitting logistic regression models including as predictor the pseudo-continuous M-value of methylation at each CpG. Separate models will be fitted for methylation levels measured from blood collected at time 0, and from blood collected at 6 (t2) and 12 months (t2). To quantify the effect of changes in the methylation profile over time, generalized mixed effect models will be estimated with the individuals at random level. Statistical analyses will be performed using MedCalc® Statistical Software version 20.118 (MedCalc Software Ltd, Ostend, Belgium; https://www.medcalc.org; 2022). Two-tailed P \< .05 will be considered statistically significant. Data analyses will be performed at different times according to the development of the project. All clinical data, dietary and quality of life measures will be collected and managed for data entry to obtain the whole cohort dataset. Hence, data collected during enrollment will be analyzed all together at months 9-11 to provide D1. Clinical evaluation data collected at follow up (t2 and t3) will be analyzed at months 19-22 to provide D3. Data analyses concerning baselines of GM and metDNA - Metabolite annotation and Dysregulated Network Analysis values will be performed soon after M2 (enrollment completed) at months 10-11 to provide D2. Data analyses concerning changes of GM and metDNA values at t1, t2, and t3 will be performed after M2 (enrollment completed), with randomized batches including a set of sibling samples from t0 to avoid bias due to batch effects, at months 19-21 to provide D4. Final data analyses aimed at GM/metDNA features correlation with therapeutic options and response to therapy will be performed at months 21-24 to provide D6 and D7. Expected outcomes The application of a multidisciplinary approach, which includes clinical characterization, fecal and blood biomarkers, will allow us to obtain a comprehensive picture of the impact of gut ecology on epilepsy onset and therapeutic response. We expect to generate data enabling the discovery of microbial biomarkers with predictive values of response to ASM, VNS, KD, and epilepsy surgery, as well as potential therapeutic targets to enhance the response to treatment. Notably, this project will allow to obtain the first ever large collection of comprehensive and matched data, collected and analyzed longitudinally with a multidimensional approach, including i) clinical data at epilepsy onset and/or at beginning of therapy, ii) a deep and dynamic evaluation of structural and functional variations of the largest and more complex microbial community of the human body, and, further, iii) whole genome methylation analysis that, per sè, will allow to strengthen the biological links and significance of microbiota-mediated control over genomic functions (epigenetic) that, in turn, might explain individual specific response to treatments. In addition to discovery of new knowledge on the expected microbial and epigenetic signatures of epilepsy, the project results will pave the way to new approaches aimed at modulating microbial taxonomy and metabolism and at monitoring (and possibly affecting) the methylation status of these patients, according to new scenarios of precision and personalized medicine. Indeed, according to the CARE objectives, the proposed tasks will enable to gain insights on the possible role of GM in the clinical outcome of seizure control thank to the dynamic changes in microbiota features that we will observe during the 12-months follow-up upon therapeutic interventions (ASMs, VNS, KD, and epilepsy surgery). Beside seizure reduction, the identification of indicator species will provide clinicians with microbial biomarkers for guiding the choice of the intervention with the most likelihood of improving patient quality of life. Significance and innovation The beneficial role in seizure control of some of the therapeutic options to treat drug-resistant epilepsy, such as KD or VNS, probably involves the modulation of the gut microbiota. Nevertheless, we have only preliminary data on gut microbiota changes induced by KD. Despite the vagal nerve being a key player of MGBA, no information on gut microbiota shifts in people with epilepsy after both VNS implantation and for patients undergoing epilepsy surgery. is available. The CARE project aims at achieving information about gut microbiota modifications related to the different treatment outcomes in order to identify a microbial signature of seizure responsiveness/resistance, which could potentially be unique for all treatments or different in each group of intervention. The microbial profile would be a key biomarker to early identify patients at risk of developing drug-resistant seizures, but also to monitor patients after or during a specific treatment.