Abstract
BACKGROUND: Very early recurrence (VER), defined as recurrence within one year after curative resection of hepatocellular carcinoma (HCC), significantly impacts long-term survival. This study aimed to develop and validate the ANT Score, a novel prognostic model integrating nutrition, inflammation, and tumor burden to refine VER prediction. METHODS: A retrospective cohort of HCC patients undergoing curative liver resection was analyzed. Key predictors were identified using least absolute shrinkage and selection operator (LASSO) regression and multivariate logistic regression, forming the ANT Score. Model performance was evaluated through receiver operating characteristic (ROC) curve analysis, DeLong's test, calibration curves, and decision curve analysis (DCA). The prognostic value of capsule integrity was also assessed. RESULTS: Among 459 included patients, 118 (25.7%) experienced VER. Patients were randomly assigned to training (70%) and test (30%) cohorts. The ANT Score, comprising albumin-to-alkaline phosphatase ratio (AAPR), neutrophil-to-albumin ratio (NPAR), and tumor burden score (TBS), demonstrated superior predictive performance (area under the curve [AUC] = 0.751, 95% confidence interval: 0.669-0.832, P < 0.05) compared to conventional markers. Capsule incompleteness was an independent risk factor but did not significantly enhance predictive accuracy (AUC = 0.76 vs 0.751, P > 0.05, DeLong test). The ANT Score-based nomogram exhibited excellent calibration and clinical utility in DCA. CONCLUSION: The ANT Score is an independent and superior predictor of VER after curative HCC resection. The ANT Score-based nomogram showed promising predictive value, offering a practical tool for individualized risk assessment. External validation and prospective studies are warranted to further assess its clinical applicability.