Establishing Albumin-Bilirubin Score Changes as Predictors of Radiation-Induced Liver Disease in Hepatocellular Carcinoma Patients Post–External Beam Radiation Therapy

Purpose: To determine albumin-bilirubin (ALBI) score change thresholds that indicate hepatotoxicity and predict poorer overall survival (OS) in hepatocellular carcinoma patients post–external beam radiation therapy (EBRT). Methods and Materials: The development cohort consisted of 329 hepatocellular carcinoma patients treated with liver-targeted EBRT across 2 centers from 2008 to 2023, with 71% classified as Child-Pugh (CP) A and 29% as CP-B/C. Recursive partitioning analysis identified thresholds associated with OS, which were evaluated using Cox regression, compared to a 2-point increase in CP score (CP2+) and externally validated using data from 2 independent centers (n = 248). The primary outcome was to establish ALBI score change thresholds associated with poorer OS. Secondary outcomes included comparing the prognostic accuracy of ALBI score change thresholds with the CP2+ criterion, analyzing the predictive value across different baseline liver functions, and examining the association between radiation dose to the normal liver and ALBI score changes. Results: ALBI changes of >0.25 (ALBI change grade 1) and >0.5 (ALBI change grade 2) were identified as optimal thresholds. CP2+ showed superior discriminative performance in the overall cohort; however, when stratifying by baseline liver status, ALBI change grade 1 outperformed CP2+ in CP-B/C patients (hazard ratio, 4.4; 95% CI, 2.4-7.8). Overall, these findings were confirmed in the external validation cohort. Multivariable Cox models, including CP2+ or ALBI changes and baseline liver status, demonstrated higher or similar concordance for ALBI score changes across all data sets. Mean liver dose correlated more strongly with ALBI score changes (logistic regression odds ratio = 1.09/1.07 for grade 1/2) than with CP2+ (odds ratio = 1.04). Conclusions: ALBI score changes of >0.25 and >0.5 are alternatives to CP2+ to evaluate radiation-induced hepatotoxicity, with the potential for more accurate assessment based on baseline liver function. Their objectivity and stronger correlation with normal liver dose make ALBI score changes more suitable for dose-response models aimed at minimizing the risk of liver complications following EBRT.