Development and validation of an individualized online calculator for early mortality risk after hepatectomy among...

Hepatocellular carcinoma (HCC) is a leading cause of cancer death globally. Despite improved surgical techniques, early post-hepatectomy mortality remains a critical concern. Current staging systems and liver function classifications fail to estimate early mortality risk to guide surgical decision-making. We aimed to develop and validate an individualized online calculator to predict early post-hepatectomy mortality for HCC.

Patients undergoing curative-intent hepatectomy for HCC from 2011 to 2021 at 11 Chinese centers were included. The training cohort comprised nine centers, while the external validation cohort included two centers. Multivariable logistic regression identified predictors of postoperative 90-day mortality, which were incorporated into an online calculator. Discrimination was assessed using the concordance index (C-index) and calibration by graphical plots.

Among 4966 patients, 90-day mortality was 4.1%. Predictors of 90-day mortality included patient performance status, prothrombin time, albumin-bilirubin (ALBI) grade, aspartate aminotransferase to platelet ratio index (APRI), tumor burden score and gross vascular invasion. The model demonstrated excellent discrimination in training and validation (C-index 0.816 and 0.801) cohorts. The proposed model outperformed staging systems (American Joint Committee on Cancer AJCC and Barcelona Clinic Liver Cancer BCLC) and liver function classifications (Child-Pugh, APRI, ALBI, and Fibrosis-4 Index FIB4) (p < 0.001). Calibration was accurate in both cohorts. The calculator achieved sensitivity of 79.0% and specificity of 71.8% to identify high-risk patients. Decision curve analysis demonstrated that the model had superior net benefit compared with staging systems and liver function classifications.

An individualized online calculator was developed and validated to predict early post-hepatectomy mortality for HCC. By identifying high-risk patients, this tool may guide surgical decision-making.