Development and Validation of a New Nomogram for Predicting Clinically Relevant Postoperative Pancreatic Fistula After Pancreatoduodenectomy.

BACKGROUND There lacks an ideal model for accurately predicting clinically relevant postoperative pancreatic fistula (CR-POPF) after pancreatoduodenectomy (PD). This study aimed at developing a nomogram with high accuracy in predicting CR-POPF after PD. METHODS A total of 1182 patients undergoing PD in the First Affiliated Hospital of Sun Yat-sen University (FAHSYSU, n = 762) and Fudan University Shanghai Cancer Center (FUSCC, n = 420) between January 2010 and May 2018 were enrolled. The patients from FAHSYSU were assigned as testing cohort, and those from FUSCC were used as external validation cohort. Univariate and multivariate logistic regression analyses were performed to determine the predictive factors for CR-POPF. Nomogram was developed on the basis of significant predictors. The performance of nomogram was evaluated by area under receiver operating characteristic (ROC) curve (AUC), calibration curve, and decision curve analysis. RESULTS In testing cohort, 87 out of 762 patients developed CR-POPF. Three predictors were significantly associated with CR-POPF, including body mass index ≥24.0 kg/m2, pancreatic duct diameter <3 mm, and drainage fluid amylase on postoperative day 1 ≥2484 units/L (all p ≤ 0.001). Prediction of nomogram was accurate with AUC of 0.934 (95% confidence interval [CI]: 0.914-0.950) in testing cohort and 0.744 (95% CI: 0.699-0.785) in external validation cohort. The predictive accuracy of nomogram was better than that of previously proposed fistula risk scores both in testing and external validation cohort (all p < 0.05). CONCLUSIONS The novel nomogram based on three easily available parameters could accurately predict CR-POPF after PD. It would have high clinical value due to its accuracy and convenience.