Skip to main navigation Skip to search Skip to main content

Development and validation of a preoperative risk prediction model for severe complications and very early recurrence after liver resection for hepatocellular carcinoma

Jun Kawashima, Yutaka Endo, Mujtaba Khalil, Selamawit Woldesenbet, Miho Akabane, Andrea Ruzzenente, Francesca Ratti, Hugo Marques, Sara Oliveira, Jorge Balaia, François Cauchy, Vincent Lam, George Poultsides, Minoru Kitago, Irinel Popescu, Guillaume Martel, Ana Gleisner, Thomas J. Hugh, Luca Aldrighetti, Itaru EndoTimothy M. Pawlik*

*Corresponding author for this work

Research output: Contribution to journal › Article › peer-review

Abstract

Introduction: We sought to develop and externally validate a preoperative model for predicting the risk of severe complications and very early recurrence after liver resection for hepatocellular carcinoma. Methods: Patients who underwent curative-intent hepatectomy for hepatocellular carcinoma between 2000 and 2023 were identified using a multi-institutional international database. Preoperative characteristics were assessed in relation to severe complications (defined as complications with a Clavien-Dindo classification III or greater) and very early recurrence (recurrence within 6 months after surgery) using multivariable analysis. Predictive models for severe complications and very early recurrence were developed and externally validated. Results: Among 969 patients, 97 patients (10.0%) experienced severe complications, and 116 patients (12.0%) developed very early recurrence. On multivariable analysis, American Society of Anesthesiologists class >2 and greater albumin-bilirubin score were associated with severe complications. Meanwhile, a greater albumin-bilirubin score and higher tumor burden score were associated with very early recurrence. A predictive model for very early recurrence was able to stratify patients relative to their risk for recurrence: low-risk (6-month recurrence-free survival, 94.1%), medium-risk (6-month recurrence-free survival, 86.0%), and high-risk (6-month recurrence-free survival, 67.1%). A total of 74 patients (7.6%) had an unfavorable risk profile (severe complication risk ≥30% and high-risk for very early recurrence). The discriminative accuracy of the severe complications (training: area under the curve, 0.69; external validation: area under the curve, 0.80) and very early recurrence (training: C-index: 0.65; external validation: C-index: 0.71) models were favorable (see online calculator: https://junkawashima.shinyapps.io/HCC_comp_VER/). Conclusion: An easy-to-use online calculator stratified patients relative to short- and long-term risks, identifying a subset of patients at a high risk of severe complications and very early recurrence who were unlikely to benefit from surgical resection.

Original languageEnglish
Article number109527
Pages (from-to)1-10
Number of pages10
JournalSurgery (United States)
Volume185
Early online date27 Jun 2025
DOIs
Publication statusPublished - Sept 2025
Externally publishedYes

Fingerprint

Dive into the research topics of 'Development and validation of a preoperative risk prediction model for severe complications and very early recurrence after liver resection for hepatocellular carcinoma'. Together they form a unique fingerprint.

Cite this