EPIGUT: EPILEPSY AND GASTROINTESTINAL MICROBIOTA: UNDERSTANDING THERAPY RESPONSE

Background: Clinical studies in humans on the role of the microbiota in promoting seizures and epilepsy are sparse and limited to small, heterogenous cohorts of epileptic patients already treated with ASMs. Larger, treatment-naïve cohorts followed longitudinally are needed to investigate the role of the microbiota in seizures and anti-seizure drug therapy. EpiGUT is building such a cohort in order to study both the oral and gut microbiota by analysis of saliva and fecal samples (primary outcomes). Blood and CSF samples for metabolic and inflammation profiling are collected to analyze how the microbiota may influence systemic and central inflammation and metabolism (secondary outcomes). In vitro, non-antibiotic drugs can inhibit growth of gut microbes and microbes can metabolize drugs to produce active or toxic metabolites. In vivo, almost nothing is known about the interactions of the microbes and ASMs. Yet these interactions can profoundly influence treatment outcome, as demonstrated for anti-PD1 therapy of patients with cancer as well as for many other treatments of various diseases. It is thus plausible that individual differences in the gut microbiota may be part of the reason why some individuals respond to certain ASMs, and others do not. By confirming this hypothesis and further in vitro validation of specific drug-microbe interactions we can create valuable clinical guidelines for personalizing anti-seizure treatments based on the composition of the gut microbiota, thus increasing the probability of a positive treatment response of the individual patient. Study design: EpiGUT studies individuals with newly diagnosed epilepsy and their microbiota in response to different ASMs both on an individual level and in comparison to healthy controls. We will analyze the gut and oral microbial composition and function at baseline, i.e., time of diagnosis and study associations with specific subtypes of epilepsies. Building a unique sample collection of blood and CSF enables mechanistic studies on how specific microbes may modulate seizures through metabolites or inflammatory signaling. We will also study correlations between components of the microbiota and ASM treatment efficacy, as well as drug therapy-associated alterations in the gut microbiota. In the initial part, we have two sampling time points, at baseline, i.e., time of diagnosis and at follow-up at the time of evaluation of monotherapeutic outcome at three months (or before start of additional ASM, if earlier). About 30% of patients will present with drug-resistent epilepsy. Those will be asked to provide additional samples and quetionnaire answers before and after alternative treatment which may include the ketogenic diet, vagus nerve stimulation or surgery. Methods: Our primary outcome variable is the composition of the oral and intestinal microbiota, both on a taxonomic as well as functional level. This will be measured by high throughput whole-metagenomic sequencing of saliva and fecal samples with a minimum of 10 million reads per sample which provides a detailed microbial profile. The role of the microbiota for both research questions will be analyzed using advanced metagenomics-specific bioinformatics tools as well as state-of-the-art machine learning algorithms to identify microbial predictors of seizures and/or ASM therapy outcome. Blood levels of ASMs are measured routinely during follow-up and will be correlated to prescribed dose and components of the gut microbiota to identify potential candidates of ASM metabolizing intestinal microbes. Clinical data such as patient history, epilepsy classification, etiology, seizure burden and current medications will be entered in an anonymized Redcap database. In addition, patients/caregivers will be asked to answer a questionnaire with standard questions relevant to microbiota studies including other diagnoses, dietary habits, oral health, birth mode, physical activity etc. PERMANOVA analysis will identify significant factors contributing to our main outcome, the microbial composition. Regression analysis will take these significant factors into account when analyzing the contribution of microbes to the diagnosis or ASM therapeutic outcome. In addition, we biobank blood samples collected at both time points and CSF at baseline (only if taken as part of the clinical workup) to analyze inflammatory and metabolomic profiles of the patients using the Olink® inflammation panel and untargeted metabolomics. These profiles will be analyzed both independently and in relation to the microbial profiles to identify host and microbiota-mediated disease- and therapy-relevant factors as secondary outcomes. Selection of participants: 1,000 patients will be recruited, 300 children and 700 adults from university and regional hospitals across Sweden A healthy control group will be recruited in a 1:2 ratio and matched by age. Data analysis and statistics: Data analysis will be performed using advanced metagenomics-specific bioinformatics tools. The diversity of the resulting baseline taxonomic and functional profiles will be analyzed for confounding and other influencing clinical and lifestyle factors and ranked by effect size (R2) and statistical significance (p) in relation to our variables of interest (e.g. epilepsy type, seizure burden and refractoriness, etiology) using Permutational Multivariate Analysis of Variance (PERMANOVA). We will apply state-of-the-art machine learning algorithms to identify microbial predictors of seizures and/or AMS therapy outcome. The most important microbial features of the best model will be extracted and tested in silico as potential biomarkers for diagnosis and/or therapeutic outcome.