Gut microbiota predict the development of post-discharge diabetes mellitus in acute pancreatitis

Background: Post-discharge morbidity and mortality is high in acute pancreatitis (AP) and pathophysiological mechanisms remain poorly understood. Objectives: We aim to investigate the composition of gut microbiota and clinical long-term outcomes of prospectively enrolled AP patients to predict post-discharge complications. Design: In this long-term follow-up study, we analysed clinical and microbiome data of 277 patients from the prospective multi-centre P-MAPS trial. Primary endpoint was the association of the microbial composition with post-discharge mortality, recurrent AP (RAP), progression to chronic pancreatitis (CP), pancreatic exocrine insufficiency (PEI), diabetes mellitus (DM) and pancreatic ductal adenocarcinoma (PDAC). Results: Buccal (n=238) and rectal (n=249) swabs were analysed by 16S rRNA and metagenomics sequencing using Oxford Nanopore Technologies. Median follow-up was 2.8 years. Distance-based redundancy analysis (dbRDA) with canonical analysis of principle coordinates (CAP) showed significant differences for β-diversity (Bray-Curtis) for post-discharge mortality (p=0.04), RAP (p=0.02), and DM (p=0.03). A ridge regression model including 11 differentially abundant species predicted post-discharge DM with an area under the receiving operating characteristic (AUROC) of 94.8% and 86.2% in the matched and entire cohort, respectively. Using this classifier, a positive predictive value of 66.6%, a negative predictive value of 96% and an accuracy of 95% was achieved. Conclusion: Our data indicate that the admission microbiome of AP patients correlates with post-discharge complications independent from multiple risk factors such as AP severity, smoking or alcohol. Microbiota at admission show excellent discriminative capacity to predict post-discharge DM and may thus open new stratification tools for a tailored risk assessment in the future.