Background: EEN is able to induce remission in CD patients, but can be difficult to maintain. A novel dietary intervention that combines partial enteral nutrition with an exclusion diet, that excludes foods proposed to trigger dysbiosis and inflammation (CDED+PEN) has been shown effective. We aim to compare the gut microbiome of CD participants in a prospective RCT comparing these diets, and to assess whether microbiome profiles can identify subgroups that are able to sustain remission.
Methods: 24 pediatric patients received either CDED + 50% Modulen for 6 weeks, then CDED+25% Modulen for 6 weeks (Group 1) or EEN with Modulen for 6 weeks followed by free diet plus 25% Modulen (Group 2). Remission was present in 11 /14 (78.5%) in Group 1, and 8/10 (80%) in Group 2. One patient from Group 1 relapsed by week 12. DNA from fecal samples collected at three time points (baseline, w6 and w12) was sequenced for 16S and whole metagenome. Taxonomic composition was inferred from 16S (QIIME) and from metagenomes (Metaphlan) and inferred for function (Diamond/HUMAnN).
Results: Both interventions induced an increase in alpha diversity by w12. EEN patients experienced a transient reduction in diversity at w6, whereas CDED+PEN did not. Taxonomic and functional profiles were similar by w12. Pooling the results for both diets, the taxonomic composition of the 18 patients who sustained remission differed significantly from the 6 patients who did not. Twenty-two 16S-derived operational taxonomic units had different relative abundance (p≤0.05), with a subgroup of eleven identified according to a false discovery threshold (q-value) of 0.15. Similar taxonomic results were inferred from the metagenome. Analysis of the functional repertoire yielded a similar pattern; 1811 genes differed between patients who sustained remission and those who did not, with 711 identified according to q-value <0.05. To confirm that these results (taxonomic and functional) were not being driven by the signal associated with a single diet, we separately compared the remission patients for each diet to the pooled set of patients who did not maintain remission. Results were consistent for both diets, but with slightly smaller subgroups of genes. Supervised modeling is currently underway to investigate if functional and taxonomic profiles can be exploited to predict patient outcomes.
Conclusions: Microbiome changes induced by CDED+PEN 50% are comparable with EEN in a pediatric RCT with active CD. Remission achieved with either dietary intervention is associated with similar structural and functional profiles.