Predicting cancer-specific mortality in T1/2 hepatocellular carcinoma after radiofrequency ablation by competing risk nomogram: A population-based analysis.

ABSTRACT

BACKGROUND: Radiofrequency ablation (RFA) is one of the primary treatment methods for T1/2 hepatocellular carcinoma (HCC), but the risk factors after RFA remain controversial. This study aims to identify the key factors associated with cancer-specific mortality (CSM) in patients with T1/2 HCC after RFA using competing risk analysis and to establish a prognostic nomogram for improved clinical management. METHODS: A total of 2,135 T1/2 HCC patients treated with RFA were obtained from the Surveillance, Epidemiology, and End Results (SEER) database and randomly categorized into training and validation sets. Univariate and multivariable competing risk analyses were performed to identify risk factors associated with CSM and construct a competing risk nomogram. Receiver operating characteristic (ROC) curves, concordance indices (C-indexes), calibration plots, and decision curve analysis (DCA) were conducted to evaluate the predictive efficiency and clinical applicability of the nomogram in the training and validation sets. Patients were stratified according to their nomogram score, and the different risk groups were compared using cumulative incidence function (CIF) curves and Gray's validation . RESULTS: The 5-year CSM rate for HCC patients treated with RFA was 30.1 %. Grade, tumor size, tumor number, cirrhosis, and AFP level were identified as independent risk factors for CSM. A prognostic nomogram was developed based on these risk factors. The time-dependent C-indexes (0.65) were greater than those of the AJCC stage model (0.55) during the 12 to 60 months of follow-up. The calibration plots of the competing risk nomograms demonstrated excellent consistency between actual survival and nomogram predictions. ROC analyses showed that the 1-, 3-, and 5-year AUC values in both the training and validation cohorts were all greater than 0.63 and exceeded those of the AJCC stage model. DCA demonstrated the clinical usefulness of the nomogram. Patients were classified into low-, moderate-, and high-risk groups based on the nomogram scores, with the high-risk group showing significantly higher CSM rates after RFA compared to the other two groups. CONCLUSIONS: We identified Grade, AFP, cirrhosis, tumor size, and tumor number as independent risk factors associated with CSM. The competing risk nomogram exhibited high performance in predicting the probability of CSM for HCC patients undergoing RFA.