Background and Aims: Pancreaticoduodenectomy (PD) and distal pancreatectomy (DP) are common surgical approaches for pancreatic tumors. Postoperative pancreatic fistula (POPF) is one of the most serious complications following pancreatic surgery, if the occurrence of POPF can be reliably predicted that will be of great clinical significance. The fistula risk score (FRS) and alternative fistula risk score (a-FRS) are two widely used prediction models for POPF, and however, the predictive efficiencies of FRS and a-FRS for POPF need further validation. This study was conducted to compare the predictive value of the FRS and a-FRS for POPF following DP and PD, so as to provide the theoretical basis and reference for the selection of the appropriate prediction model in clinical practice. 
Methods: The clinical data of all patients undergoing pancreatic surgery between 2018 and 2019 in a single center were retrospectively collected, and the enrolled patients after screening for exclusion criteria were used as study subjects. The incidence rates of POPF in the whole group of patients and patients undergoing different surgery types (PD and DP) were counted, and the predictive efficiencies of the two scoring models in predicting POPF for the whole group and different surgery types were compared using area under the ROC curve (AUC).
Results: A total of 339 patients were included after exclusion of the ineligible cases, with 193 patients undergoing PD and 146 cases undergoing DP. The incidence of POPF was 17.4% in the entire group, and was 18.1% and 16.4% in PD group and DP group, respectively. FRS and a-FRS had a similar predictive ability for POPF in the whole group (AUC: 0.67 vs. 0.65, P=0.412), and the predictive value of FRS for POPF was better than that of a-FRS in PD group (AUC: 0.74 vs. 0.67, P=0.006), but FRS showed no predictive value for POPF in DP group (AUC=0.57, 95% CI=0.44–0.70, P=0.285), while the predictive ability of a-FRS for POPF was better than that of FRS in DP group (AUC: 0.66 vs. 0.57, P=0.048). Moreover, the incidence rates of POPF were increased in either the whole group, PD group or DP group with the increase of the risk grade classified by either FRS or a-FRS. Among the predictive factors of FRS, there were significant differences in intraoperative blood loss and diameter of the main pancreatic duct between DP group and PD group (both P<0.05).
Conclusion: Both FRS and a-FRS can be used for predicting POPF. However, FRS has a better predictive value than that of a-FRS for POPF following PD, but is unsuitable for POPF following DP, while a-FRS may be helpful for predicting POPF flowing DP. Blood loss and diameter of the main pancreatic duct may be responsible for the poor predictive ability of FRS for POPF following DP. However, these conclusions still need to be further verified due to the limitations of the study.