Prediction Of Survival Benefit In Patients With Hepatocellular Carcinoma After Transcatheter Arterial Chemoembolization Based On CT Radiomics Signatures

Purpose:To develop a Nomogram model combined general clinical risk factors with CT-based radiomics signatures,and evaluate its value in predicting one-year and two-year survival in patients with hepatocellular carcinoma(BCLC B stage)after underwenttranscatheter arterial chemoembolizationMethods:118 eligible patients were retrospectively collected and randomly divided into training groups(n=80)and validation groups(n=38).The training group was used to establish a nomogram model.The survival-related clinical risk factors were extracted from the general clinical data using Kaplan-Meier univariate analysis and Cox regression multivariate analysis.The patient's preoperative portal and arterial CT images were collected.The region of interest(ROI)was manually segmented on the itk-SNAP software.After extracted the radiomics features from high-throughput data,LASSO regressionwas used for dimension reduction and parameters selection Multivariate logistic regression was used to establish a nomogram model which combined radiomics signatures with clinical risk factors by R software.The predictive value of the model was assessed by C-index and receiver operating characteristic curve(ROC)Results:The maximum diameter of the largest lesion in the liver was regarded as the independent predictor(P=0.024).10 radiomics signatures were chosen from the 398 extracted radiomics features.A Nomogram model was established with the maximum diameter of the largest lesion in the liver and radiomics signatures.The area under the curve(AUC)of radiomics signatures for one-year was 0.598 and for two-year was 0.775..The C-index of Nomogram model in the training set was 0.627(95%Cl:0.557-0.697),and the validation set was 0.617(95%Cl:0.534-0.686).The AUC of Nomogram model for one-year was 0.676 and for two-year was 0.833Conclusions:The nomogram model combined maximum diameter of the largest lesion in the liver with preoperative radiomics signatures can help to predict the 1-year and 2-year survival of patients with hepatocellular carcinoma(BCLC B stage)after transcatheter arterial chemoembolization.