Prognostic impact and predictive model for early non-curative recurrence after liver resection for hepatocellular carcinoma: a retrospective cohort study.

BACKGROUND: The prognostic impact of early non-curative recurrence after liver resection for hepatocellular carcinoma (HCC) and its predictive factors remain unknown. We aimed to identify independent predictive factors for early non-curative recurrence, develop a predictive model, and evaluate its clinical utility. MATERIALS AND METHODS: We retrospectively analyzed 624 patients who underwent curative HCC resection between 2002 and 2020. Patients were randomly assigned to training ( n = 416) and validation ( n = 208) sets. Early recurrence was defined as recurrence within 2 years after surgery. Non-curative recurrence was defined as recurrence corresponding to the intermediate or advanced stage of the Barcelona Clinic Liver Cancer staging system based on tumor factors. A nomogram was constructed based on independent predictive factors identified via multivariate analysis. Model performance was assessed by the area under the receiver operating characteristic curve (AUC), calibration plots, and decision curve analysis. RESULTS: Early non-curative recurrence was an independent prognostic factor for overall survival (OS) after surgery and OS after recurrence (hazard ratio = 5.387 and 2.806, P <0.001). Median survival time after surgery was significantly shorter in the early non-curative recurrence group than in the other group (training set: 34.7 vs. 178.6 months, P <0.001; validation set: 27.7 vs. 156.1 months, P <0.001). Five factors were independently associated with early non-curative recurrence: alpha-fetoprotein, tumor size, number of tumors, portal invasion, and serosal invasion. The predictive model demonstrated high calibration performance in the training (AUC = 0.825) and validation (AUC = 0.792) sets. Decision curve analysis confirmed the clinical utility of the predictive model. CONCLUSION: We developed and validated a novel predictive model for early non-curative recurrence, which is significantly associated with poor prognosis after curative liver resection for HCC. Its application may enhance individualized postoperative surveillance and treatment strategies, potentially improving long-term outcomes.