P1330 The gut microbiome at the onset of inflammatory bowel disease: A systematic review of the literature and unified bioinformatic synthesis of sequencing data from >1000 treatment naïve patients

Abstract

Background
There are few studies of the gut microbiome in new-onset IBD. We present a novel secondary bioinformatic re-analysis (BA) of sequence outputs mapped to latest microbial taxonomy derived from systematic review.

Methods
MEDLINE and Embase searches were performed for microbiome studies in pre-treatment IBD. Appraisal was completed with ROBINS-E. Available 16S rRNA sequence datasets were downloaded and missing datasets requested. Integrated datasets were run through a unified pipeline based in QIIME2 supplemented by phyloseq and DESeq2.

Results
From 12,264 studies, 10,804 abstracts and 217 full texts were screened. 31 studies were eligible; 18 met the requirements for BA. Most were low risk for bias, except uncontrolled confounding. Methodology varied extensively. Considering only amplicon data; samples originated from 1,743 unique individuals (including 678 Crohn’s disease (CD), 399 ulcerative colitis (UC), 123 ‘other’ IBD, 130 healthy, 405 symptomatic, and 8 familial controls) derived from 15 countries. DNA was extracted using 15 different kits, sequencing performed on 4 separate platforms targeting 10 different 16S rRNA hypervariable regions. There was a paucity of adult data, particularly mucosal biopsies.

Bacterial alpha diversity was reduced: in faeces in paediatric UC vs symptomatic controls (SC), in adult CD and UC vs HC and in paediatric SC vs HC (Figure 1).

In paediatric biopsies in CD vs SC, CD vs UC, SC vs UC (Figure 2).

Beta-diversity BA by Bray-Curtis demonstrated clear distinctions between faecal and mucosal biopsy communities, least evident in UC. Faecal communities separated by geography, though less evidently in biopsies.

Differential abundance analysis revealed enrichment of oral genera across sample types (Fusobacterium, Peptostreptococcus, Neisseria in CD; Haemophilus, Aggregatibacter, Peptostreptococcus, Porphyromonas, Neisseria, Gemella in UC). Short-chain fatty acid (SCFA) producers were diminished but less consistently across IBD phenotype and sample type.

Conclusion
This work demonstrates the utility of a combined systematic review/secondary data analysis approach to published microbiome datasets. Whilst synthesis remains challenging due to the methodological inconsistencies within these studies, this approach better accounts for these variations and presents a new way to consider this data. A reduction in bacterial diversity remains core to the IBD microbiome, though with more complexity than previously considered (SC<UC in paediatric biopsies). Microbial community structure changes by sample type and geography. Enrichment of oral bacteria is consistent across IBD subtypes and sample types suggesting their importance in the disease. A deficit in SCFA producers may also be important.