P380 Integrated fecal metagenome and serum metabolome analysis revealed gut microbiome associated serum metabolites as diagnostic biomarkers for inflammatory bowel disease

Abstract Background Gut microbiomes are altered among patients with IBD. Certain microbiome species were used as biomarkers for IBD diagnosis, yet with limited sensitivity and specificity, especially when used for the classification of UC and CD. Gut microbiome reprogramming was accompanied with metabolic shift. Previous studies discovered that gut microbiome-associated serum metabolites (GMSM) were disease-specific biomarkers with high sensitivity. The aim of this study is to explore how gut microbiome reprogramming is reflected in serum metabolome shift in IBD, and the potential of GMSM for IBD diagnosis and classification of UC and CD. Methods Shotgun metagenomic of stool and ultraperformance liquid chromatography-mass spectrometry based untargeted metabolomic profiling of serum samples from a discovery cohort of IBD and healthy individuals were performed. Spearman correlation analysis was used to identify GMSM that were differentially abundant among IBD and healthy individuals. Candidate GMSM for model construction were selected using LASSO and logistic regression. The selected GMSM panel was then transferred to multiple reaction monitoring based targeted detection. Diagnostic model for IBD and classification model for UC and CD were developed based on the GMSM panel using a training cohort, and then validated using an independent cohort consisting of patients with IBD, healthy individuals and patients with colorectal adenoma or non-adenoma polyps. Results Metagenome sequencing of fecal samples from 77 patients with IBD and 74 healthy individuals revealed significant shifts in gut microbiome, such as decrease in F. prausnitzii. Untargeted metabolomic profiling of serum from 76 patients with IBD and 28 healthy individuals revealed 724 differentially abundant metabolites, of which 151 metabolites were found to be significantly associated with gut microbiomes detected from the fecal samples (figure). Candidate GMSM for model construction were established using the method described and the ultimate panel consisted of 17 metabolites, including 10 GMSM for diagnostic model for IBD and 7 GMSM for classification model for UC and CD. Eventually, the diagnostic model for IBD achieved an area under curve (AUC) of 0.98 in the training cohort, and 0.92 in the validation cohort. The classification model for UC and CD achieved AUC of 0.94 in the training cohort, and 0.9 in the validation cohort. In addition, the diagnostic model for IBD also effectively distinguished IBD from patients with colorectal adenoma or non-adenoma polyps (table). Conclusion Based on GMSM, we developed diagnostic and classification models for IBD with high specificity and sensitivity, providing an accurate and non-invasive approach for IBD diagnosis and classification of UC and CD.