P920 Gut microbiota at diagnosis is associated with clinical features and future disease course in Inflammatory Bowel Disease: Data from IBSEN III, a new large Norwegian population-based inception cohort

Abstract Background Inflammatory bowel disease (IBD) in South-Eastern Norway (IBSEN) III is a population-based inception cohort including patients with suspected IBD between 2017 and 2019. The present study aimed to evaluate the diagnostic and prognostic properties of baseline microbiota profiling, particularly regarding future disease course within the first year from diagnosis. Methods Stool samples were collected in preservative before index colonoscopy or as close after as possible. Patients with suspected IBD but no evidence of inflammation were categorized as symptomatic controls. Patients <18 years and patients treated with antibiotics the previous three months were excluded. A severe disease course was defined as the need for one or more of the following within one year of diagnosis: IBD-related surgery, hospitalizations due to IBD, exposure to ≥2 biologics, exposure to >2 steroid courses, or development of complicated disease behaviour. The V3-V4 region of the 16S rRNA gene was amplified and sequenced on an Illumina platform. Statistics were performed in R version 4.2.0, using MaAsLin2 with confounder correction to test for differential abundance of bacteria. XGBoost with 5-fold cross validation was used with area under the curve (AUC) analysis to produce predictive machine learning (ML) models. Results 970 patients were included: n=569 Ulcerative colitis (UC, 53% male), n=287 Crohn’s disease (CD, 41% male), n=114 symptomatic controls (51% male). Baseline microbiota was associated with a severe disease course in UC-patients when compared to those with an indolent disease course, characterized by a reduced α-diversity at baseline (Shannon diversity index, p = 0.0001, beta = -1.42 [-1.79, -1.05]) and a distinct microbial profile identified by differential abundance analyses (q < 0.05, 42 taxa). UC and CD had unique microbial profiles compared to each other (q < 0.05, 35 taxa), and both presented with decreased α-diversity when baseline disease severity increased. Further, UC and CD were different from symptomatic controls (q < 0.05, 12 and 4 taxa respectively), but only UC had a reduced α-diversity compared to systematic controls. Using ML-models and AUC analyses, microbial profiles were better prognostic markers than calprotectin and CRP for predicting both IBD subtype, i.e. UC vs CD, and a severe disease course in UC (p < 0.00001, Table). Conclusion In this large IBD inception cohort followed for one year, baseline microbial profiles show potential as both a prognostic and diagnostic tool. The microbial profile outperformed traditional biochemical markers in predicting diagnosis as well as a severe disease course, which may have clinical implications.