Nomograms to Predict the Disease-free Survival and Overall Survival after Radiofrequency Ablation for Hepatocellular Carcinoma

Objective The purpose of this study was to construct nomograms for the disease-free survival (DFS) and overall survival (OS) of post-radiofrequency ablation (RFA) patients with hepatocellular carcinoma (HCC). Furthermore, we compared the prognostic predictive ability of these nomograms for estimating per-patient outcomes with that of traditional staging systems. Methods We retrospectively enrolled 298 patients in the training set and 272 patients in the validation set who underwent RFA for HCC. The nomograms for the DFS and OS were constructed from the training set using the multivariate Cox proportional hazards model. The discriminatory accuracy of the models was compared with traditional staging systems by analyzing the Harrell's C-index. Results The DFS nomogram was developed based on the tumor size, tumor number, aspartate aminotransferase (AST), albumin, age, and α-fetoprotein. The OS nomogram was developed based on the tumor size, the model for end-stage liver disease, AST, and albumin. Our DFS and OS nomograms had good calibration and discriminatory abilities in the training set, with C-indexes of 0.640 and 0.692, respectively, that were greater than those of traditional staging systems. The C-indexes of our DFS and OS nomograms were also greater than those of traditional staging systems in the validation set, with C-indexes of 0.614 and 0.657, respectively. RFA patients were stratified into low- and high-risk groups based on the median nomogram scores. High-risk patients receiving surgical resection (SR) were associated with a better DFS and OS than those undergoing RFA. However, the DFS and OS were similar between the low-risk RFA and SR groups. Conclusion We constructed reliable and useful nomograms that accurately predict the DFS and OS after RFA for early-stage HCC patients. These graphical tools are easy to use and will assist physicians during the therapeutic decision-making process.