Abstract Background The oral cavity is the second most complex human microbial niche and is the conduit through which initial gut microbial colonisation occurs during early life. While the gut microbiome is implicated in the pathogenesis of inflammatory bowel disease (IBD), the role of the oral microbiome remains underexplored. We aimed to characterise the oral-gut microbiome in IBD and its association with disease activity and complete mucosal healing (CMH), a key prognostic marker for long-term remission. Methods All patients with Crohn’s disease (CD) and Ulcerative Colitis (UC) were prospectively recruited via www.musicstudy.uk. We conducted a 3-stage experimental approach. (1) We profiled the oral microbiota in IBD vs healthy controls (HC) (CD = 34, UC = 24 and HC = 25). (2) We carried out trans-anatomical oral-ileal-stool microbial source tracking1 within the same patients at a single time-point (CD = 23; inflamed vs. non-inflamed mucosal samples), alongside longitudinal oral-stool sampling (CD = 32, UC = 17). (3) Our machine learning (ML) modelling integrated stool microbiota taxonomic profiles (CD = 53, UC = 11; sampled at 0, 3, and 6 months) with ~80 clinical variables to classify CMH using random forest, AdaBoost, XGBoost, logistic regression, and neural networks. Results 16S rRNA sequencing revealed significant divergence between IBD and HC oral microbiota (PCoA, Bray-Curtis; PERMANOVA, p=0.003). Active disease also differed from remission (PCoA, Bray-Curtis; PERMANOVA, p=0.01). Higher oral Veillonella abundances were associated with IBD; bacteria commonly enriched in the IBD gut. We determined within patient oral microbiota abundances in gut samples and found no difference between inflamed ileal mucosa compared to non-inflamed (p=0.52). Patients who did not achieve CMH exhibited significantly higher oral microbiota abundances in stool compared to those with CMH (β = 0.046, p=0.022, linear mixed-effects model). Oral microbiota abundances in stool correlated with clinical markers of disease activity, calprotectin (r = 0.23, p=0.009) and albumin (r = –0.28, p=0.002; Benjamini-Hochberg adjusted). ML models integrating stool microbiota taxonomic profiles and clinical variables outperformed clinical markers alone in predicting CMH (AUC: 0.79 vs. 0.55), indicating that the inclusion of stool microbiota features could be valuable in predicting CMH in the clinical setting. Conclusion Our data show distinct oral microbiota taxonomic profiles in IBD and active disease, with higher oral microbiota abundances in stool associated with lack of CMH. These findings highlight the oral microbiome as a potential upstream regulator of microbial-immune crosstalk from the site of IBD-inflammation.
Read full abstract