Comparative liver function models for ruptured hepatocellular carcinoma: A 10-year single center experience

Abstract

Previous studies have proposed several objective means for liver function assessment in hepatocellular carcinoma (HCC) patients; however, their efficiency in predicting survival of HCC rupture is unknown. Our study aims to confirm which is a better liver function model for ruptured HCC. A total of 230 patients with HCC ruptures at our center were included. Kaplan-Meier and Cox regression analyses were performed to compare long-term survival and short-term mortality. The 90-day mortality was compared with the area under the receiver characteristic curve. Logistic regression was used to determine the risk factors for 90-day deaths, and the discriminant ability of the model was measured. There were significant differences in predicting OS of the Child-Pugh (CP) score in all patients, the non-surgical subgroup, and the surgical subgroup (all P<0.0001). But no statistical significance was shown of the ALBI score in the surgical (P=0.8985) or non-surgical subgroup (P=0.0634). The CP score yielded a better performance among all patients (AUC=0.746 vs. 0.712), the surgical subgroup (AUC=0.558 vs. 0.530), and the non-surgical subgroup (AUC=0.715 vs. 0.634) compared to ALBI score in predicting ninety-day mortality. A similar result can be found in the subgroup of surgical and non-surgical treatment group. Moreover, the logistic model that included CP or MELD had a better discriminatory ability than ALBI in predicting ninety-day mortality. The CP or MELD rather than ALBI score should be used as a liver function classification criterion for HCC rupture. NCT03534843 (retrospectively).