Mintziras_2023_Langenbecks.Arch.Surg_408_359

PURPOSE: To compare the predictive value of serum amylase and lipase regarding the occurrence of clinically relevant postoperative pancreatic fistula (cr-POPF) after partial pancreaticoduodenectomy (PD). METHODS: Data from 228 consecutive patients undergoing PD were obtained from a prospective database. Serum amylase and lipase were measured on postoperative days (PODs) 0-2. Receiver-operating characteristics analysis was performed and cutoff values were tested using logistic regression. RESULTS: Serum amylase had a larger area under the curve (AUC) on POD1 (AUC 0.89, p <0.001) than serum lipase. For serum amylase POD 1, a cutoff value of 70 U/l showed sensitivity and specificity of 100% and 70% for the diagnosis of cr-POPF. Serum amylase POD 1 > 70 U/l (OR 9.815, 95% CI 3.683-26.152, p < 0.001), drain amylase POD 1 > 300 U/l (OR 2.777, 95% CI 1.071-7.197, p= 0.036), and a small (>= 3mm) pancreatic duct diameter (OR 3.705, 95% CI 1.426-9.627, p= 0.007) were significant predictors of cr-POPF in the multivariable analysis. Patients were divided into three risk groups based on serum amylase POD 1 and pancreatic duct diameter. This model had a good performance in discriminating cr-POPF (AUC 0.846, 95% CI 0.793-0.898). The sensitivity, specificity, and negative predictive value for the combination of serum amylase POD 1 <70 U/l and pancreatic duct diameter >3 mm were 100%, 70%, and 100%. CONCLUSION: Serum amylase POD 1 was superior to serum lipase in predicting cr-POPF after PD. The proposed risk prediction model had a sensitivity and negative predictive value of 100%, allowing for early identification of cr-POPF.