LAUSR

Background The application of postoperative pancreatic fistula (POPF) risk stratification and mitigation strategies requires an update of the protocol for an early drain removal after pancreaticoduodenectomy (PD). The aim of the study is to highlight the unreliability of a single drain fluid amylase (DFA) cutoff-based protocol in the setting of a risk-based drain management. Methods The role of postoperative day one (POD1) DFA in predicting POPF was explored in the setting of both selective drain placement and early drain removal protocols. Receiver operating characteristics (ROC) curves were used to assess the POPF diagnostic performance in terms of negative predictive value (NPV) of several POD1 DFA cutoffs in different clinical scenarios according to POPF risk and mitigation strategies. Results The areas under the curve (AUCs) for POD1-DFA were 0.815 for intermediate risk and pancreaticojejunostomy (PJ) (best cutoff 1000 IU/L, NPV 92.9%), 0.712 for intermediate risk and PJ with external stent (best cutoff 1000 IU/L, NPV 88.8%), and 0.574 for high risk and external stent (best cutoff 250 IU/L, NPV 93.7%). Independent predictors of POPF were body mass index, pancreatic texture, and early drain removal, whereas POD1 DFA was not. Conclusion In the era of risk stratification and mitigation strategies, selective early drain removal still is associated with a reduced rate of POPF. However, a single protocol based on POD1-DFA is not suitable for all clinical scenarios after PD.