Marco Angrisani1, Marta Sandini2, Marco Cereda3, Salvatore Paiella4, Giovanni Capretti5, Gennaro Nappo5, Linda Roccamatisi3, Fabio Casciani4, Riccardo Caccialanza6, Claudio Bassi4, Alessandro Zerbi5, Luca Gianotti3. 1. School of Medicine and Surgery, University of Milano-Bicocca and Department of Surgery, San Gerardo Hospital, Monza, Italy; Pancreatic Surgery Unit, Department of Surgery, Humanitas Research Hospital and University, Rozzano, Milan, Italy. 2. School of Medicine and Surgery, University of Milano-Bicocca and Department of Surgery, San Gerardo Hospital, Monza, Italy. Electronic address: m.sandini@campus.unimib.it. 3. School of Medicine and Surgery, University of Milano-Bicocca and Department of Surgery, San Gerardo Hospital, Monza, Italy. 4. Unit of General and Pancreatic Surgery, The Pancreas Institute, Policlinico G.B. Rossi, University of Verona Hospital Trust, Verona, Italy. 5. Pancreatic Surgery Unit, Department of Surgery, Humanitas Research Hospital and University, Rozzano, Milan, Italy. 6. Clinical Nutrition and Dietetics Unit, Fondazione IRCCS Policlinico San Matteo, Pavia, Italy.
Abstract
BACKGROUND: Anthropometric parameters have been associated with increased risk of clinically relevant postoperative pancreatic fistula (CR-POPF) after pancreaticoduodenectomy (PD). Nonetheless, conventional metrics to predict POPF do not include the assessment of body composition. We aimed to validate the most used Fistula Risk Score (FRS), and to assess whether the appraisal of adipose compartment at bioimpedance vector analysis (BIVA) improves the accuracy of FRS in CR-POPF prediction. METHOD: PD patients from 3 Italian academic institutions were prospectively included over a 2-year period. Patients with ASA score ≥3, heart failure, chronic kidney disease, or compartmentalized fluid collections were excluded. BIVA was performed on the day prior to surgery. CR-POPF occurrence and severity were classified per the ISGPS classification. RESULTS: Out of 148 PDs, 84 patients (56.8%) had pancreatic cancer, and 29 (19.6%) experienced CR-POPF. FRS elements, namely soft pancreatic texture (p = 0.009), small pancreatic duct diameter (p = 0.029), but not blood loss (p = 0.450), as well as high BMI (p = 0.004) were associated with CR-POPF. Also, the preoperative fat mass (FM) amount measured at BIVA was significantly higher in patients who developed CR-POPF, compared to those who did not (median FM = 19.4 kg/m2 vs. 14.4 kg/m2, respectively; p = 0.005). The predictive ability of a multivariate model adding FM to the FRS, assessed at the receiver operating characteristics curve showed a higher accuracy than the FRS alone (AUC = 0.774 and AUC = 0.738, respectively). CONCLUSIONS: Assessment of preoperative FM at BIVA can improve the accuracy of FRS in predicting CR-POPF following pancreatoduodenectomy.
BACKGROUND: Anthropometric parameters have been associated with increased risk of clinically relevant postoperative pancreatic fistula (CR-POPF) after pancreaticoduodenectomy (PD). Nonetheless, conventional metrics to predict POPF do not include the assessment of body composition. We aimed to validate the most used Fistula Risk Score (FRS), and to assess whether the appraisal of adipose compartment at bioimpedance vector analysis (BIVA) improves the accuracy of FRS in CR-POPF prediction. METHOD:PDpatients from 3 Italian academic institutions were prospectively included over a 2-year period. Patients with ASA score ≥3, heart failure, chronic kidney disease, or compartmentalized fluid collections were excluded. BIVA was performed on the day prior to surgery. CR-POPF occurrence and severity were classified per the ISGPS classification. RESULTS: Out of 148 PDs, 84 patients (56.8%) had pancreatic cancer, and 29 (19.6%) experienced CR-POPF. FRS elements, namely soft pancreatic texture (p = 0.009), small pancreatic duct diameter (p = 0.029), but not blood loss (p = 0.450), as well as high BMI (p = 0.004) were associated with CR-POPF. Also, the preoperative fat mass (FM) amount measured at BIVA was significantly higher in patients who developed CR-POPF, compared to those who did not (median FM = 19.4 kg/m2 vs. 14.4 kg/m2, respectively; p = 0.005). The predictive ability of a multivariate model adding FM to the FRS, assessed at the receiver operating characteristics curve showed a higher accuracy than the FRS alone (AUC = 0.774 and AUC = 0.738, respectively). CONCLUSIONS: Assessment of preoperative FM at BIVA can improve the accuracy of FRS in predicting CR-POPF following pancreatoduodenectomy.