Supplementary Material for: Predictive Capacity of Biomarkers for Severe Acute Pancreatitis
datasetposted on 03.03.2016, 00:00 by Sternby H., Hartman H., Johansen D., Thorlacius H., Regnér S.
Background: Early prediction of severe acute pancreatitis (SAP) substantially improves treatment of patients. A large amount of biomarkers have been studied with this objective. The aim of this work was to study predictive biomarkers using preset cut-off levels in an unselected population of patients with acute pancreatitis (AP). Methods: 232 patients (52.2% males, median age 66 years) with AP admitted to Skåne University Hospital, Malmö, were consecutively enrolled. Blood samples were collected upon admission and clinical data were gathered both prospectively at inclusion and through review of medical notes. Cut-off levels were defined based on the reports of prior studies, and through their results eight biomarkers (IL-1β, IL-6, IL-8, IL-10, TNF-α, MCP-1, procalcitonin and D-dimer) were selected for analysis. Results: Of the patients, 83.2% had mild AP and 16.8% had SAP. Levels of IL-1β, IL-6 and IL-10 were significantly (p < 0.05) higher upon admission in the group with SAP. When applying the preset cut-off levels on our material, sensitivity and specificity for prediction of severity were low. Receiver operating characteristic curves showed that selected cut-off levels were acceptable, but areas under the curves were inferior compared to other studies. The results did not improve when using the revised Atlanta 2012 classification. Conclusions: Previous studies on severity prediction of AP are difficult to compare due to large variations in setups and outcomes. Calculated cut-offs in our cohort were in acceptable range from preset levels, however areas under the curves were low, indicating suboptimal biomarkers for the unselected population investigated. For comparable results and possible clinical implementations, future studies need large consecutive series with a reasonable percentage of severe cases. Additionally, novel biomarkers need to be explored.