Identification of a prognostic signature model for predicting disease-free survival of hepatocellular carcinoma.

e16128 Background: Hepatocellular carcinoma (HCC) is a type of aggressive disease with poor prognosis. Although surgery is the most effective therapy for early liver cancer, recurrence rate is up to 50%. Therefore, it is crucial to predict the recurrence for improvement of prognosis of liver cancer. This study aim is to establish a prognostic model using for the prediction of recurrence in HCC. Methods: We collected the genomic and clinical data of 372 patients from the cancer genome atlas (TCGA) database. Mutational signature were established based on univariate Cox analysis and least absolute shrinkage and selection operation (LASSO) Cox regression analysis. A prognostic signature model were developed. Results: The patients with higher pathological T stage (p < 0.001) and old age (p = 0.05) were associated with a worse DFS. In addtion, patients with a HRD score over 24 had a worse DFS compared to those in patients with HRD score less than 24 (p < 0.001). Based on the 1 year DFS, 112 differential mutation gene (DMEs) and 326 differential expressed genes (DEGs) were identified. The univariate Cox and LASSO Cox regression models were employed to select 47 DMEs and 9 gene expression construct an prognosis model. The AUC is 0.612. A nomogram of DFS was established based on the prognosis model, HRD, and tumor stage. The AUC is 0.812 in training set and 0.842 in validation set. Conclusions: We established and validated a novel nomogram model based the genomic and clinical factors for predict DFS in Hepatocellular Carcinoma patients. This model has good predictive value for prognosis, which could improve the risk stratification and individual treatment of Hepatocellular Carcinoma patients.