Xun Liu, Xiao-An Liu, Di Wang, Jing Wang, Han-Xuan Wang, Ren Lang, Anshi Wu
AGR showed the most consistent associations with OS and recurrence among the five indices. Its addition produced modest improvements in the development cohort, with changes in the same direction during external validation. AGR may complement, but should not replace, established tumor-related prognostic factors.
BACKGROUND: Prognosis after liver transplantation for hepatocellular carcinoma (HCC) is driven mainly by tumor burden, vascular invasion, and alpha-fetoprotein. Routine pretransplant nutritional indices may add information on recipient condition, but direct comparisons are scarce and recurrence is not always analyzed with competing mortality.
METHODS: The development cohort included 254 adults undergoing first liver transplantation for pathologically confirmed HCC. Overall survival (OS) was the primary outcome; time to recurrence (TTR) was analyzed with death before recurrence as a competing event. Five nutritional indices were compared using Kaplan-Meier estimates, Cox regression, and restricted cubic splines. AGR underwent subgroup and Fine-Gray analyses. Conventional staging, AGR alone, a pretransplant clinical model, and the clinical model plus AGR were compared by discrimination, calibration, and decision curves. Time-dependent ROC curves included bootstrap 95% confidence intervals. Fixed development-model parameters were evaluated in an independent external validation cohort of 58 recipients.
RESULTS: Eighty-five patients died and 51 experienced recurrence. In the fully adjusted Cox model, each 1-standard-deviation (SD) increase in AGR was associated with lower overall mortality [hazard ratio (HR), 0.64; 95% confidence interval (CI), 0.48-0.84; P = 0.002]; GNRI was also associated with OS (HR, 0.75; 95% CI, 0.58-0.98; P = 0.037). The AGR-OS association was nonlinear (overall P < 0.001; nonlinear P = 0.004). The adjusted recurrence subdistribution HR was 0.66 per 1-SD increase in AGR (95% CI, 0.47-0.94; P = 0.022). Adding AGR increased the OS C-index from 0.750 to 0.780 in the development cohort and from 0.645 to 0.686 externally. At 3 years, the external AUC was 0.713 (95% CI, 0.574-0.834) for the clinical model and 0.764 (95% CI, 0.614-0.888) after AGR was added.
CONCLUSIONS: AGR showed the most consistent associations with OS and recurrence among the five indices. Its addition produced modest improvements in the development cohort, with changes in the same direction during external validation. AGR may complement, but should not replace, established tumor-related prognostic factors.