Validation of original, alternative, and updated alternative fistula risk scores after open and minimally invasive pancreatoduodenectomy in an Asian patient cohort.

Abstract

This study aimed to validate and compare the performance of the original fistula risk scores (o-FRS), alternative (a-FRS), and updated alternative FRS (ua-FRS) after open pancreatoduodenectomy (OPD) and laparoscopic pancreatoduodenectomy (LPD) in an Asian patient cohort. Data of 597 consecutive patients who underwent PD (305 OPD, 274 LPD) were collected from two tertiary centers. Model performance was assessed using the area under the receiver operating curve (AUC). The overall AUC values of o-FRS, a-FRS, and ua-FRS were 0.67, 0.69, and 0.68, respectively, which were lower than those of the Western validation. Three FRS systems had similar AUC values in the overall and OPD groups, whereas ua-FRS had a higher AUC than o-FRS in the LPD group. The accuracy of ua-FRS (47.2%) was higher than that of o-FRS (39.0%) and a-FRS (19.5%) overall, but low specificity and low positive predictive value were observed regardless of the operative type across the three FRS systems. In the multivariate analysis, pathology, estimated blood loss, and body mass index were not independent risk factors for CR-POPF in the OPD and LPD groups. Current FRS systems have some limitations, including a relatively lower performance in an Asian cohort, low positive predictive values, and inclusion of insignificant risk factors.