Development and Validation of an Individualized Prediction Model for Postoperative Late Recurrence After Hepatectomy for Hepatocellular Carcinoma (POLAR-HCC): A Multicenter Study.

backgroundPostoperative late recurrence (POLAR) after 2 years from the date of surgical resection of hepatocellular carcinoma (HCC) represents a unique surveillance and management challenge. Despite identified risk factors, individualized prediction tools to guide personalized surveillance strategies for recurrence remain scarce. The current study sought to develop a predictive model for late recurrence among patients undergoing HCC resection.

methodsThis multicenter study analyzed HCC patients who underwent resection across 10 Chinese hepatobiliary centers and remained recurrence-free at 2 years after hepatectomy. Patients were randomly assigned to development and validation cohorts (2:1 ratio). Independent predictors identified through multivariate Cox regression analysis were integrated into a nomogram and web-based calculator.

resultsAmong 849 recurrence-free patients at 2 years after hepatectomy for HCC, seven independent predictors of POLAR were identified: male (hazard ratio [HR] 1.37, p = 0.04), cirrhosis (HR 1.42, p = 0.008), multiple tumors (HR 1.56, p = 0.006), satellite nodules (HR 1.59, p = 0.004), large tumor size (HR 1.49, p = 0.009), macrovascular invasion (HR 4.63, p < 0.001), and microvascular invasion (HR 1.69, p = 0.001). The POLAR-HCC nomogram-based calculator demonstrated robust performance in both the development (area under the curve [AUC] 0.660) and validation (AUC 0.626) cohorts. Using the optimal cut-off value of 1.93, patients were accurately stratified into high- and low-risk groups with different risks of POLAR (p < 0.001).

conclusionsThe POLAR-HCC online calculator enables risk stratification for POLAR after HCC resection. By integrating tumor characteristics and host factors, this prediction tool identified high-risk patients who may benefit from intensified recurrence surveillance, potentially improving long-term survival through earlier detection of POLAR. The model represents an important step toward personalized surveillance strategies among patients undergoing HCC resection.