Concurrent and habitual diet differentially associate with microbial Multi-omic Profiles in Inflammatory Bowel Disease(IBD)

Background: Studies have linked diet to the risk and severity of inflammatory bowel disease (IBD) and its subtypes, Crohn’s disease (CD) and ulcerative colitis (UC). Similarly robust evidence has associated disease activity to characteristic alterations in gut microbial taxonomy (metagenomics, MGX), community functions (metatranscriptomics, MTX), and microbial metabolites (metabolomics, MBX). However, in IBD, explorations into how these multi-omic readouts are affected by concurrent/short-term vs. habitual/long-term diets are limited. Methods: Using the Integrative Human Microbiome Project in which 105 densely-phenotyped participants with IBD (n=38 with UC, 67 with CD) and 27 non-IBD controls collectively provided 1,638 stool metagenomes, 835 metatranscriptomes, and 546 metabolomes over one year, we conducted a longitudinal survey to compare the relative importance of concurrent and habitual diet in relation to MGX, MTX, MBX profiles using an abbreviated food frequency questionnaire with each stool sample. Concurrent diet was assessed by reported intake from the one-week preceding sample collection, while habitual diet was modeled using a decaying average of concurrent food records. We explored several different decaying formulas; for example, one formula assigned a weight of 2-n for the nth prior week. With these integrated time-series profiles, we linked diet and microbiome matrices via intra-individual mantel tests, quantifying associations within subjects only. We assessed differences in correlation coefficients using 4,999 permutations and evaluated robustness with 4,999 bootstraps.

Results: Diet-microbiome associations showed stronger correlations when accounting for within-person correlation compared with a subject-unaware omnibus association, supporting a personalized link between diet and multi-omic profiles. Compared to concurrent, habitual diets had a significantly stronger correlation with MGX taxonomy and MGX functional potential. Correlation increased as the decay became more gradual, and plateaued at weight=1/sqrt(time+1). We further explored the impact of diet on MTX functional activities (normalized by DNA gene copy number) and observed community-level patterns consistent with MGX. However, when functions were stratified by contributing microorganisms, no differences emerged between concurrent and habitual diets. This suggests that although habitual diet more significantly shapes community-level functions than concurrent diet, it does not predict which species contribute to specific functions. In contrast, neither habitual nor concurrent diets had additive effects on shaping MBX, which could be attributed to an actual biological pattern or potentially to the technical noise in MBX data, as the association tests showed less stability during bootstrapping.

Conclusion: This work offers a glimpse at how diet is differentially coupled to alterations in longer-term microbial growth, functional potential (MGX), and community functional activities (MTX) for which additional data on habitual diet is informative vs. comparatively transient multi-omics (MBX). However, the magnitude of the association is modest. Next, we will interrogate the bidirectional relationship between disease and dietary stability and whether these linkages differ among microbes or biochemical pathways canonically linked to IBD.