Proposal of a Preoperative CT-Based Score to Predict the Risk of Clinically Relevant Pancreatic Fistula after Cephalic Pancreatoduodenectomy

Abstract

Background and Objectives: Postoperative pancreatic fistula after cephalic pancreatoduodenectomy (CPD) is still the leading cause of postoperative morbidity, entailing long hospital stay and costs or even death. The aim of this study was to propose the use of morphologic parameters based on a preoperative multisequence computer tomography (CT) scan in predicting the clinically relevant postoperative pancreatic fistula (CRPF) and a risk score based on a multiple regression analysis. Materials and Methods: For 78 consecutive patients with CPD, we measured the following parameters on the preoperative CT scans: the density of the pancreas on the unenhanced, arterial, portal and delayed phases; the unenhanced density of the liver; the caliber of the main pancreatic duct (MPD); the preoperatively estimated pancreatic remnant volume (ERPV) and the total pancreatic volume. We assessed the correlation of the parameters with the clinically relevant pancreatic fistula using a univariate analysis and formulated a score using the strongest correlated parameters; the validity of the score was appreciated using logistic regression models and an ROC analysis. Results: When comparing the CRPF group (28.2%) to the non-CRPF group, we found significant differences of the values of unenhanced pancreatic density (UPD) (44.09 ± 6.8 HU vs. 50.4 ± 6.31 HU, p = 0.008), delayed density of the pancreas (48.67 ± 18.05 HU vs. 61.28 ± 16.55, p = 0.045), unenhanced density of the liver (UDL) (44.09 ± 6.8 HU vs. 50.54 ± 6.31 HU, p = 0.008), MPD (0.93 ± 0.35 mm vs. 3.14 ± 2.95 mm, p = 0.02) and ERPV (46.37 ± 10.39 cm3 vs. 34.87 ± 12.35 cm3, p = 0.01). Based on the odds ratio from the multiple regression analysis and after calculating the optimum cut-off values of the variables, we proposed two scores that both used the MPD and the ERPV and differing in the third variable, either including the UPD or the UDL, producing values for the area under the receiver operating characteristic curve (AUC) of 0.846 (95% CI 0.694–0.941) and 0.774 (95% CI 0.599–0.850), respectively. Conclusions: A preoperative CT scan can be a useful tool in predicting the risk of clinically relevant pancreatic fistula.