Abstracts

Rationale: Ketogenic diets (KDs) are routinely used to control drug-refractory seizures in children and adults. A major drawback of KD treatment is the difficulty associated with adherence to a strict diet combined with potential adverse metabolic effects. KDs often fail in patients for unknown reasons underscoring the importance of developing predictive biomarkers of success. To understand how KDs alter the gut microbiome and/or predict response to a KD in children with epilepsy, we enrolled children with medically intractable epilepsy who were considering KD for treatment of epilepsy. We collected stool samples at baseline and one month after initiation of the diet with the aim to study gut bacteria before and after diet initiation and in relation to treatment response.

Methods: A total of 39 children were screened and consented to take part in this pilot study. Sixteen who supplied both pre and post samples were included in analysis. Baseline stool was collected when patients came to clinic to discuss dietary therapy or on the first day of admission for in hospital diet initiation. Follow up samples were collected between weeks 4-12 after initiation of the diet or once diet compliance was assured and target goal ratio of fat to carbohydrates plus protein was achieved. Concomitant medications were kept constant at this time. Demographic data collected included age, type of diet (modified Atkins, classic ketogenic diet with various ratios, serum bicarbonate and betahydroxybutyrate levels to correlate compliance and degree of ketosis and acidosis achieved.) Dietary response was recorded as improved, neutral, or worsening based on the frequency of seizures. Fecal samples were subjected to sequencing of the v4 region of the 16S ribosomal RNA (rRNA) gene on a MiSeq instrument and subsequent analyses were conducted using QIIME 2.

Results: Microbiome composition relationships across samples were dominated by interpersonal variation, with individuals having a related composition before and after diet and a lack of overall convergence in microbiome composition with diet modification. Previously reported changes that occur with the KD diet, such as a loss of Bifidobacterium, were not replicated. Interestingly, we found a relationship between microbiome alpha diversity and KD diet response using a 2-way ANOVA; individuals with an improvement in seizures with the diet had significantly lower alpha diversity. Also, individuals who had an improvement with diet modification but not those with a no change or worsening symptoms had an increase in Enterobacteraceae with the KD diet.

Conclusions: Our study indicates that low microbiome alpha diversity may have potential for predicting which patients with refractory epilepsy may respond favorably to a KD, which is important because this diet is difficult to adhere to and has potential for adverse metabolic effects. We observed an increase in Enterobacteraceae with KD diets only in those with improved symptoms, but further work will be needed to determine if activity of these bacteria directly improve response and by what mechanism.

Funding: Please list any funding that was received in support of this abstract.: Supported by an Associate Dean for Research Grant (Skaggs School of Pharmacy CU Anschutz) to M.P. and C.J.