Machine learning based prediction model for bile leak following hepatectomy for liver cancer

Abdullah Altaf; Muhammad M Munir; Muhammad Muntazir M Khan; Zayed Rashid; Mujtaba Khalil; Alfredo Guglielmi; Francesca Ratti; Luca Aldrighetti; Todd W Bauer; Hugo P Marques; Guillaume Martel; Sorin Alexandrescu; Matthew J Weiss; Minoru Kitago; George Poultsides; Shishir K Maithel; Carlo Pulitano; Vincent Lam; Irinel Popescu; Ana Gleisner; Tom Hugh; Feng Shen; François Cauchy; Bas G Koerkamp; Itaru Endo; Timothy M Pawlik

doi:10.1016/j.hpb.2024.12.015

Machine learning based prediction model for bile leak following hepatectomy for liver cancer

HPB (Oxford). 2024 Dec 20:S1365-182X(24)02463-8. doi: 10.1016/j.hpb.2024.12.015. Online ahead of print.

Authors

Abdullah Altaf¹, Muhammad M Munir¹, Muhammad Muntazir M Khan¹, Zayed Rashid¹, Mujtaba Khalil¹, Alfredo Guglielmi², Francesca Ratti³, Luca Aldrighetti³, Todd W Bauer⁴, Hugo P Marques⁵, Guillaume Martel⁶, Sorin Alexandrescu⁷, Matthew J Weiss⁸, Minoru Kitago⁹, George Poultsides¹⁰, Shishir K Maithel¹¹, Carlo Pulitano¹², Vincent Lam¹³, Irinel Popescu⁷, Ana Gleisner¹⁴, Tom Hugh¹⁵, Feng Shen¹⁶, François Cauchy¹⁷, Bas G Koerkamp¹⁸, Itaru Endo¹⁹, Timothy M Pawlik²⁰

Affiliations

¹ Department of Surgery, Division of Surgical Oncology, The Ohio State University Wexner Medical Center and James Comprehensive Cancer Center, Columbus, OH, United States.
² Department of Surgery, University of Verona, Verona, Italy.
³ Department of Surgery, San Raffaele Hospital, Milan, Italy.
⁴ Department of Surgery, University of Virginia School of Medicine, Charlottesville, VA, United States.
⁵ Department of Surgery, Curry Cabral Hospital, Lisbon, Portugal.
⁶ Department of Surgery, University of Ottawa, Ottawa, ON, Canada.
⁷ Department of Surgery, Fundeni Clinical Institute, Bucharest, Romania.
⁸ Department of Surgery, Johns Hopkins Medicine, Baltimore, MD, United States.
⁹ Department of Surgery, Keio University, Tokyo, Japan.
¹⁰ Department of Surgery, Stanford University School of Medicine, Stanford, CA, United States.
¹¹ Department of Surgery, Emory University School of Medicine, Atlanta, GA, United States.
¹² Department of Surgery, Royal Prince Alfred Hospital, University of Sydney, Sydney, NSW, Australia.
¹³ Department of Surgery, Westmead Hospital, Sydney, NSW, Australia.
¹⁴ Department of Surgery, University of Colorado, Aurora, CO, United States.
¹⁵ Department of Surgery, School of Medicine, The University of Sydney, Sydney, NSW, Australia.
¹⁶ Department of Surgery, Eastern Hepatobiliary Surgery Hospital, Shanghai, China.
¹⁷ Department of Surgery, AP-HP, Beaujon Hospital, Clichy, France.
¹⁸ Department of Surgery, Erasmus University Medical Centre, Rotterdam, Netherlands.
¹⁹ Department of Surgery, Yokohama City University School of Medicine, Yokohama, Japan.
²⁰ Department of Surgery, Division of Surgical Oncology, The Ohio State University Wexner Medical Center and James Comprehensive Cancer Center, Columbus, OH, United States. Electronic address: [email protected].

PMID: 39753458
DOI: 10.1016/j.hpb.2024.12.015

Abstract

Objective: We sought to develop a machine learning (ML) preoperative model to predict bile leak following hepatectomy for primary and secondary liver cancer.

Methods: An eXtreme Gradient Boosting (XGBoost) model was developed to predict post-hepatectomy bile leak using data from the ACS-NSQIP database. The model was externally validated using data from hepatocellular carcinoma (HCC) and intrahepatic cholangiocarcinoma (ICC) multi-institutional databases.

Results: Overall, 20,570 and 2253 patients were identified from the ACS-NSQIP and multi-institutional databases, respectively. The incidence rates of bile leak were 7.0 %, 6.3 % and 10.2 % in the ACS-NSQIP, HCC and ICC databases, respectively. The XGBoost model achieved areas under receiver operating characteristic curves (AUROC) of 0.748, 0.719 and 0.711 in the training, testing and external validation cohorts, respectively. The SHAP algorithm demonstrated that the factors most strongly predictive of postoperative bile leak were serum alkaline phosphatase, surgical approach and cancer diagnosis. An online tool was developed for ease-of-use and clinical applicability (https://altaf-pawlik-bileleak-calculator.streamlit.app/).

Conclusion: A novel ML model demonstrated strong discrimination power to preoperatively identify patients at high risk of developing bile leak post-hepatectomy. The online calculator may be used as a clinical tool to inform preoperative surgical planning, intraoperative decision-making, and postoperative recovery protocols for patients undergoing hepatectomy.