A prediction model for overall survival after transarterial chemoembolization for hepatocellular carcinoma invading the hepatic vein or inferior vena cava

Abstract

Objectives: An effective therapeutic option has not yet been established for hepatocellular carcinoma (HCC) invading the hepatic vein (HV) or inferior vena cava (IVC). This study aimed to determine the therapeutic effect of transarterial chemoembolization (TACE) in HCC patients with HV or IVC invasion, and to build a risk prediction model. Methods: Data from patients who underwent TACE as a first-line treatment for HCC invading the HV or IVC between 1997 and 2019 were retrospectively evaluated. Results: Data from 296 patients were included (1997?2006 comprised the training cohort, n = 174; 2007?2019 comprised the validation cohort, n = 122). The median post-TACE survival was 7.3 months and an objective tumor response was achieved in 34.1% of patients. Multivariable Cox analysis of the training cohort identified five pretreatment factors (maximal tumor size > 10 cm, infiltrative HCC, combined portal vein invasion, extrahepatic metastasis, and ECOG performance status 1), which were used to create predictive models for overall survival. Median overall survival times in the validation cohort were 14 and 4.2 months for the low (sum of risk score: 0?3)- and high-risk (sum of risk score: 4?7) groups, respectively (p < 0.001). Time-dependent ROC curves for the predictive models for overall survival applied to the validation cohort showed acceptable AUC values (0.723 and 0.667 at 6 months and 1 year). Conclusions: TACE seems effective for selected patients with HCC invading the HV or IVC. The predictive model may help to identify candidates most likely to benefit from TACE. Key Points: ? To develop a risk prediction model for patients with HCC with HV or IVC invasion treated with TACE, five factors were selected from a multivariate Cox regression model for overall survival. ? The combination of these factors helped to identify two prognostic categories: low- and high-risk. ? The predictive model can help to select candidates who will benefit most from TACE in this patient group.