Tailoring the Use of Central Pancreatectomy Through Prediction Models for Major Morbidity and Postoperative Diabetes
Eduard Antonie van Bodegraven, Sanne Lof, L. Jones, Béatrice Aussilhou, Gao Yong, Wei Jishu, Rosa Klotz, Darío M. Rocha-Castellanos, Ippei Matsumato, Charles de Ponthaud, Kimitaka Tanaka, Esther Biesel, Emmanuele Kauffmann, Traian Dumitraşcu, Yuichi Nagakawa, Pablo Martí‐Cruchaga, Geert Roeyen, Alessandro Zerbi, Mara Göetz, Vincent E. de Meijer, Patrick Pessaux, Povilas Ignatavičius, İhsan Ekin Demir, Mario Giuffrida, Bobby Tingstedt, Marco V. Marino, Sotiris Mastoridis, Maximilian Brunner, Isabel Mora, Cecilia Bortolato, Aistė Gulla, T. Apers, H. Hermand, Yusuke Mitsuka, Irinel Popescu, Ugo Boggi, Uwe A. Wittel, Satoshi Hirano, Sébastien Gaujoux, Keiko Kamei, Carlos Fernández-del Castillo, Thilo Hackert, Jiang Kuirong, Yi Miao, Alain Sauvanet, Marc G. Besselink, Mohammad Abu Hilal, Safi Dokmak
Abstract
OBJECTIVE: To develop a prediction model for major morbidity and endocrine dysfunction after central pancreatectomy (CP) which could help in tailoring the use of this procedure. BACKGROUND: CP is a parenchyma-sparing alternative to distal pancreatectomy for symptomatic benign and premalignant tumors in the body and neck of the pancreas CP lowers the risk of new-onset diabetes and exocrine pancreatic insufficiency compared with distal pancreatectomy but it is thought to increase the risk of short-term complications, including postoperative pancreatic fistula (POPF). METHODS: International multicenter retrospective cohort study including patients from 51 centers in 19 countries (2010-2021). The primary endpoint was major morbidity. Secondary endpoints included POPF grade B/C, endocrine dysfunction, and the use of pancreatic enzymes. Two risk models were designed for major morbidity and endocrine dysfunction utilizing multivariable logistic regression and internal and external validation. RESULTS: A total of 838 patients after CP were included [301 (36%) minimally invasive] and major morbidity occurred in 248 (30%) patients, POPF B/C in 365 (44%), and 30-day mortality in 4 (1%). Endocrine dysfunction in 91 patients (11%) and use of pancreatic enzymes in 108 (12%). The risk model for major morbidity included male sex, age, Body Mass Index, and American Society of Anesthesiologists score ≥3. The model performed acceptably with an area under the curve of 0.72 (CI: 0.68-0.76). The risk model for endocrine dysfunction included higher Body Mass Index and male sex and performed well [area under the curve: 0.83 (CI: 0.77-0.89)]. CONCLUSIONS: The proposed risk models help in tailoring the use of CP in patients with symptomatic benign and premalignant lesions in the body and neck of the pancreas (readily available through www.pancreascalculator.com ).