Decision-Making Scoring System for the Repetition of Conventional Transarterial Chemoembolization in Patients With Inoperable Hepatocellular Carcinoma

Abstract

Introduction: Patients with unresectable hepatocellular carcinoma treated with conventional transarterial chemoembolization (cTACE) have heterogeneous tumor burden and liver function. Therefore, the selection of patients for repeated cTACE is challenging owing to different outcomes. This study aimed to establish a decision-making scoring system for repeated cTACE to guide further treatment.

Methods: All patients with hepatocellular carcinoma who underwent cTACE between 2008 and 2019 were included and randomly assigned into training (n = 324) and validation (n = 162) cohorts. Tumor Size, number of Masses, Albumin-bilirubin score, baseline Alpha-fetoprotein level, and Response to initial cTACE session were selected to generate a "SMAART" score in the training cohort. Patients were stratified according to the SMAART score: low risk, 0-2; medium risk, 3-4; and high risk, 5-8. Prediction error curves based on the integrated Brier score and the Harrell C-index validated the SMAART scores and compared them with the Assessment for Retreatment with Transarterial chemoembolization (ART) score.

Results: The low-risk group had the longest median overall survival of 39.0 months, followed by the medium-risk and high-risk groups of 21.2 months and 10.5 months, respectively, with significant differences (P < 0.001). The validation cohort had similar results. The high-risk group had 63.1% TACE refractory cases. The Harrell C-indexes were 0.562 and 0.665 and the integrated Brier scores were 0.176 and 0.154 for ART and SMAART scores, respectively.

Discussion: The SMAART score can aid clinicians in selecting appropriate candidates for subsequent cTACE. A SMAART score of ≥5 after the first cTACE session identified patients with poor prognosis who may not benefit from additional cTACE sessions.

Figure 1.

Flow diagram of patient inclusion and exclusion. cTACE, conventional transarterial chemoembolization; HCC, hepatocellular carcinoma; PEI, percutaneous ethanol injection.

Figure 2.

Histogram of median overall survival according to the SMAART and ART scoring systems in all, training, and validation cohorts.

Figure 3.

Survival analyses of all (a), training (b), and validation (c) cohorts following the SMAART scoring system. A similar predictive performance for overall survival was observed between the validation and training cohorts. In subgroup analysis, the overall survival was significantly different in all enrolled BCLC-B patients (d) and in those with Child-Pugh class A (e) or B (f) according to the SMAART scoring system.

Figure 4.

Prediction error curve and integrated Brier scores (IBS) for Kaplan-Meier estimates based on the SMAART score with 3 groups (blue), the SMAART score with 2 groups (green), and the ART score (red), compared with the unstratified sample (black).