BackgroundAcute pancreatitis (AP) is a common clinical disease with varying severity. The Revised Atlanta Classification (RAC) categorises AP into mild, moderately severe, and severe (MAP, MSAP and SAP) respectively. Despite the availability of different scoring systems to triage patients, these are not always suitable for predicting the course and outcome of certain patients during admission. In this study, untargeted metabolomics was used to identify metabolic parameters that can potentially be used as prognostic markers for stratifying the risk profiles of patients for improved management and treatment.MethodsSerum isolated from blood samples collected from 30 AP patients (8 MAP, 14 MSAP, and 8 SAP) and 9 healthy control (HC) individuals was analysed using1H nuclear magnetic resonance (NMR) spectroscopy. Wilcoxon and Kruskal–Walli’s rank-sum tests were used to compare differences in numerical covariates. A liposcale test was used for lipoprotein characterization and the Spearman rank test was conducted for correlation of the data. P-values < 0.05 were considered significant.ResultsElevated levels of lactate, (rho = 0.67; p-value < 0.001, FDR = 0.001), 3-hydroxybutyrate (rho = 0.46; p-value < 0.003, FDR = 0.013), acetoacetate (rho = 0.63; p-value < 0.001, FDR < 0.001) and lipid alpha-CH2 (rho = 0.45; p-value = 0.004, FDR = 0.013) were associated with AP severity as was decreased levels of ascorbate (rho = 0.46; p-value < 0.003, FDR = 0.013), methanol (rho = 0.46; p-value < 0.003, FDR = 0.013), glutamine (rho = -0.55; p-value < 0.001, FDR = 0.002), ethanol (rho = 0.64; p-value < 0.001, FDR< 0.001), protein-NH (rho= -0.75; p-value < 0.001, FDR<0.001) among others. HDL-C decreased while IDL-C and VLDL-C increased across all the AP metabolic phenotypes compared to the HC.ConclusionDysregulated metabolites and lipids can potentially add to the understanding of the pathophysiological conditions of AP and can aid in the early prognosis and stratification of patients for specialist care.