BackgroundAcute necrotizing pancreatitis (NP), a severe form of acute pancreatitis (AP), has higher mortality and worse outcome than non-necrotizing pancreatitis (non-NP). Infected NP is a devastating subgroup of NP. To date neither NP nor infected NP has robust prediction strategies, which may delay early recognition and timely intervention. Recent studies revealed correlations between disturbed gut microbiota and AP severity. Some features of intestinal microbiota have the potential to become biomarkers for NP prediction.MethodsWe performed 16S rRNA sequencing to analyze gut microbiota features in 20 healthy controls (HC), and 58 AP patients on hospital admission. The AP patients were later classified into NP and non-NP groups based on subsequent diagnostic imaging features. Random forest regression model and ROC curve were applied for NP and infected NP prediction. PIRCUSt2 was used for bacterial functional pathway prediction analysis.ResultsWe found that the three groups (HC, NP, and non-NP) had distinct microorganism composition. NP patients had reduced microbial diversity, higher abundance of Enterobacteriales, but lower abundance of Clostridiales and Bacteroidales compared with the non-NP group. Correlation analyses displayed that intestine bacterial taxonomic alterations were related to severity, ICU admission, and prognosis. By pathway prediction, species more abundant in NP patients had positive correlation with synthesis and degradation of ketone bodies, and benzoate degradation. Enterococcus faecium (ASV2) performed best in discriminating NP and non-NP patients. Finegoldia magna (ASV3) showed the maximal prediction capacity among all ASVs and had comparable accuracy with Balthazar CT to detect patients with infected NP.ConclusionsOur study suggests that NP patients have distinct intestinal microbiota on admission compared to non-NP patients. Dysbiosis of intestinal microbiota might influence NP progression through ketone body or benzoate metabolism. Enterococcus faecium and Finegoldia magna are potential predictors for NP and infected NP. Our findings explore biomarkers which may inform clinical decision-making in AP and shed light on further studies on NP pathophysiology and management.