Skip to main content

Poster Session 2025

Perturbed Gut Viral Ecology in Inflammatory Bowel Disease: a Multi-cohort Study

Presented By: Jiaxian Shen

Background: The burden of inflammatory bowel disease (IBD) is rising globally. While prior studies have uncovered the critical role of gut dysbiosis in IBD and its subtypes, Crohn’s disease (CD) and ulcerative colitis (UC), most have focused on its bacterial determinants and few have explored gut viral ecology.
Methods: Leveraging the recent development of a next-generation method for viral profiling, we uniformly processed and harmonized 2,574 IBD shotgun metagenomes from 580 individuals (320 CD, 124 UC, 136 non-IBD controls; Fig. 1A) enrolled in eight independent international cohorts from the Human Microbiome Bioactives Resource.
Results: In total, we detected 5,391 unique viral genome bins (VGBs, akin to bacterial species-level clades). The majority of these VGBs represented uncharacterized viral “dark matter,” of which 78% were completely unclassified across all taxonomic levels. This represents a relative overrepresentation of novel viruses in the IBD gut when compared to the 7% unclassified background of our reference database. The gut virome was 1.5x more associated with disease status (i.e., IBD vs. control) than gut bacteria (R2=2.1%, p=0.001, Fig. 1B). Aligning viral and bacterial profiles, we found that they were significantly coupled with similar clustering patterns (correlation=0.89, p=0.001). Interestingly, this coupling may be disrupted in CD and UC vs. non-IBD (pFDR≤0.01, Fig. 1C). Additionally, viral and bacterial richness and evenness decreased monotonically from non-IBD to UC to CD (Fig. 1D). Using generalized multivariable linear models, we identified 343 differentially abundant VGBs (15.6% of those detected) in IBD compared to non-IBD (69% greater than the count for differentially abundant bacteria, Fig. 1E). As expected, most of these significantly altered viruses were uncharacterized (75%), indicating that these novel viruses may be a critical yet underexplored factor in IBD. Among characterized VGBs, Vimunumvirus ST147VIM1phi71, Peduovirus P2, Felixounavirus felixO1, Lambdavirus lambda, and Punavirus P1—several of which are putative phage of the IBD-associated bacteria, Escherichia coli—were significantly enriched in IBD (Fig. 2A-B). Felixounavirus felixO1 and Fohxhuevirus gastrointestinalis were differentially abundant in CD vs. UC (Fig. 2C). Finally, using a machine learner, we found comparable accuracy in classifying IBD vs. non-IBD when using viral features compared to bacteria (AUC>0.95).
Conclusion: Our study represents the largest virally-targeted investigation of the IBD gut to-date. Next, we will expand our study to include comparatively understudied RNA viruses and will interrogate the interplay between microbial domains.