LAUSR

Editorial for “Multiparametric MRI-Based Radiomic Signature for Preoperative Evaluation of Overall Survival in Intrahepatic Cholangiocarcinoma After Partial Hepatectomy” Intrahepatic cholangiocarcinoma (iCCA) is the second most common primary hepatic malignancy, with a poor prognosis. In addition, iCCA is a heterogeneous neoplasm in terms of etiology, location of origin, histopathologic and molecular characteristics, postoperative outcomes, and radiological characteristics. Although surgical resection is considered the only curative treatment option for iCCA, patients who undergo resection also have a poor prognosis, with a 5-year survival rate of 5%–56%. Currently, the most widely used staging system for patients with iCCA is the tumor–node–metastasis (TNM) staging from the American Joint Committee on Cancer, which incorporates pathological data such as tumor size, number of tumor, vascular invasion, visceral peritoneal perforation, invasion of extrahepatic structures, lymph node metastasis, and distant metastasis. Along with TNM staging, numerous staging systems have been proposed, the majority of which are based on pathological data and have insufficient discriminatory performance for overall survival. As these pathology-based staging systems cannot be evaluated prior to surgery, they cannot be used to guide preoperative clinical decision-making. Another problem of the pathology-based staging systems is the evaluation of the pathological nodal status of iCCA. Lymph node metastasis has been recognized as a significant prognostic factor, and most pathology-based staging systems incorporate nodal status. Therefore, lymph node dissection is recommended for all patients undergoing hepatic resection for proper prognostication. However, due to the unclear survival benefit and concerns regarding surgical complications, approximately 50% patients with iCCA undergoing hepatic resection do not undergo lymph node dissection, and pathological nodal status is not available in such patients. Pathological staging systems can only be applied in some patients with iCCA treated with hepatic resection. Thus, accurate preoperative prediction of postoperative survival is an unmet need for patients with iCCA. If the survival of patients with iCCA can be predicted prior to surgery using radiological findings, patients with an expected poor prognosis can be offered neoadjuvant therapy to improve survival or nonsurgical treatment to improve their quality of life. The radiomics is emerging technique for quantitative image analysis can be used for tumor characterization and prognostication. Recent studies have indicated that radiomics is useful in selecting candidates for transarterial radioembolization, predicting lymph node metastasis, and predicting recurrence after hepatic resection in patients with iCCA, implying that radiomics may also be useful in predicting overall survival after hepatic resection. The paper published in this issue of JMRI titled “Multiparametric MRI-based radiomic signature for preoperative evaluation of overall survival in intrahepatic cholangiocarcinoma after partial hepatectomy” attempted to develop and validate the preoperative MRI-based radiomic model for predicting the postoperative outcome in patients with iCCA. The authors developed a dichotomized radiomics signature, which can be calculated using radiomic features from T1-weighted, T2-weighted, and portal venous phase images. The radiomics signature was significantly associated with overall survival and disease-free survival in the training and validation sets. The combination of the radiomics signature and TNM staging showed significantly improved prognostic accuracy compared with TNM staging alone (training set, 0.701 vs. 0.579, P < 0.001; validation set, 0.745 vs. 0.649, P = 0.039). In addition, the authors developed a clinicopathological and MR radiographic model, namely the “CPR model,” and compared its discriminatory performance with that of the radiomics signature. There was no significant difference between the CPR model and radiomics signature in both the training and validation sets (training set, 0.680 vs. 0.705, P = 0.486; validation set, 0.698 vs. 0.674, P = 0.863). The authors concluded that the radiomics signature