Published online Aug 21, 2022. doi: 10.3748/wjg.v28.i31.4376
Peer-review started: April 20, 2022
First decision: June 2, 2022
Revised: June 14, 2022
Accepted: July 20, 2022
Article in press: July 20, 2022
Published online: August 21, 2022
Hepatocellular carcinoma (HCC) is the most common primary liver malignancy. The prognosis of HCC patients remains poor. Radiomics is an artificial intelligent-based method for obtaining prognostic and predictive information which may contribute to clinical outcomes improvement.
Currently, a few studies have analysed the largest cross-sectional area of HCC tumour. In this study, we have analysed the entire-tumour to build a more comprehensive prognostic prediction model with clinical characteristics. We aimed to develop a radiomics model for predicting the overall survival of HCC patients after hepatectomy.
In this study, we aimed to develop a radiomics model based on contrast-enhanced computed tomo
A total of 150 HCC patients were enrolled and randomly divided into a training cohort (n = 107) and a validation cohort (n = 43) at ratio 2.5:1. Radiomics features were extracted from the CECT images. In training cohort, the least absolute shrinkage and selection operator algorithm was applied for radiomics features selection and radiomics signature construction. Univariate and multivariate Cox regression analyses were used to develop the predictive model. The accuracy of the model was assessed with the concordance index, receiver operating characteristic curve and calibration curve. The clinical practicality was evaluated by decision curve analysis. The survival between the low- and high-risk subgroups was compared using Kaplan–Meier methodology.
In total, seven radiomics features were selected to construct the radiomics signature. Alpha-fetoprotein, neutrophil-to-lymphocyte ratio and radiomics signature were identified as independent risk predictors to build the predictive model. The C-indices of the model in the training and validation cohorts were 0.736 and 0.774, respectively. In receiver operating characteristic curve for predicting 1-, 3-, and 5-year overall survival, area under the curve (AUC) = 0.850, 0.791 and 0.823, respectively in training cohort; AUC = 0.905, 0.884 and 0.911, respectively in validation cohort. The calibration curve analysis indicated a good agreement between the model-prediction and actual survival. Decision curve analysis suggested that the predictive model had more benefit than traditional staging system models. In Kaplan–Meier survival analysis, patients in the low-risk group had longer overall survival and disease-free survival.
The predictive model is a reliable tool for predicting the overall survival of HCC patients after radical hepatectomy.
More precise and reliable tool to predict the prognosis of HCC patients is urgently needed. Radiomics is a new method for obtaining prognostic and predictive information. In this study, we aimed to develop a predictive model based on CECT images and clinical-pathologic characteristics to predict the overall survival of HCC patients.