| Part ?-Radiomics Nomogram Based on Diffusion-weighted Imaging for Predicting Early Recurrence and Adjuvant Chemotherapy Benefits in Intrahepatic Cholangiocarcinoma After Partial Hepatectomy Objects: To develop radiomics nomogram based on diffusion-weighted imaging for predicting early recurrence(ER)in patients with intrahepatic cholangiocarcinoma(ICC)after partial hepatectomy,and explore the potential value for survival stratification and the implications for individualized adjuvant chemotherapy.Methods: In this retrospective study,124 patients with ICC who underwent partial hepatectomy from August 2012 to May 2019 were randomly divided into the training set(n=87)and the validation set(n=37).Radiomics model was built using radiomics features extracted from diffusion-weighted imaging(DWI)with random forest(RF).The clinicopathologic-radiologic(CPR)model and the combined model integrating the radiomics score with the CPR risk factors were developed with multivariate logistic regression analysis.The performance of the above models was assessed by calibration,discrimination,and clinical usefulness.The Kaplan–Meier method and log-rank test were used to compare the difference of prognosis between different groups.Results: The rates of very early recurrence(VER)were 37.1%(n=46)and ER of 62.1%(n=77).The area under the curve(AUC)of the radiomics model in training and validation sets were 0.823(95%CI: 0.729-0.917)and 0.753(95%CI: 0.597-0.909),which was comparable to CPR model in both sets(AUC: 0.697,p=0.06;AUC: 0.621,p=0.274).The AUC of the radiomics nomogram combining the radiomics score and CPR risk factors were0.876(95%CI: 0.796-0.955)and 0.821(95%CI: 0.684-0.959)in training and validation sets,with significantly better performance than CPR model in both sets(p=0.001;p=0.01;respectively).Besides,the radiomics nomogram exhibited promising discrimination of VER(AUC: 0.753,95%CI: 0.667-0.839),compared with radiomics model and CPR model(p=0.008;p=0.027).The radiomics nomogram achieved significant stratification of overall survival with independent significance on multivariate analysis(HR:1.153,95%CI:1.012-1.313,p=0.033).High-risk patients predicted by the radiomics nomogram could benefit from adjuvant chemotherapy(OS: p=0.023;DFS: p=0.031),while the low-risk patients could not(OS: p=0.62;DFS: p=0.43).Conclusions: The radiomics nomogram based on diffusion-weighted imaging could predict ER and VER risk for patients with ICC after partial hepatectomy,and provide a potential indicator for individual adjuvant chemotherapy and survival stratification.Part ?-Multiparametric MRI-based radiomics signature for Predicting Overall Survival in Intrahepatic Cholangiocarcinoma After Partial HepatectomyObjects: To develop radiomics signature using multiparametric MRI images to predict overall survival in patients with intrahepatic cholangiocarcinoma(ICC)and to investigate its incremental value for disease stratification.Methods: In this retrospective study,quantitative radiomics features(n=4998)were extracted from the multiparametric MRI of 163 patients with ICC(allocated to a training and validation set,7:3 ratio).Radiomics model was built based on LASSO algorithm.The clinicopathologic-radiologic(CPR)model and the combined model integrating the radiomics signature with the CPR risk factors were developed with multivariate cox regression model.The Kaplan–Meier method and log-rank test were used to estimate the efficacy of risk stratification of the above models.Net Reclassification Index(NRI)and Integrate Discrimination Improvement(IDI)were used to analyze the reclassification ability of the combined model compared with CPR model.Results: The radiomics signature achieved significant stratification of overall survival(OS)and disease-free survival(DFS)in both training(p<0.0001,p<0.001,respectively)and validation set(p<0.001,p=0.0036,respectively).The C-index of the radiomics signature for predicting OS in training and validation sets were 0.680(95%CI 0.618-0.742)and 0.698(95%CI 0.583-0.814),which was comparable to CPR model(training set: 0.705,95%CI: 0.634-0.777;validation set: 0.674,95%CI: 0.555-0.792)(p=0.486,p=0.752,respectively).The C-index of the combined model integrating radiomics signature and CPR risk factors were 0.750(95%CI 0.680-0.819)and 0.723(95%CI 0.624-0.822),with no improvement in prognostic performance compared with that of CPR model in both sets(p=0.109,p=0.214;respectively).The combined model showed increased ability to stratify prognosis over the CPR model(validation set: p=0.028 vs p=0.058).In reclassification analyses,incorporating radiomics signature into CPR model significantly improved the prognostic accuracy(validation set: NRI and IDI of 1-year OS were 0.325,and 0.081;NRI and IDI of 2-year OS were 0.343 and 0.129;NRI and IDI of 3-year OS were 0.414 and 0.162).Conclusions: Multiparametric MRI-based radiomics signature is a potential biomarker for preoperative prognostic stratification. |
PDF Full Text Request