Comprehensive radiomics nomogram for predicting survival of patients with combined hepatocellular carcinoma and cholangiocarcinoma

Abstract

BACKGROUND Combined hepatocellular carcinoma (HCC) and cholangiocarcinoma (cHCC-CCA) is defined as a single nodule showing differentiation into HCC and intrahepatic cholangiocarcinoma and has a poor prognosis. AIM To develop a radiomics nomogram for predicting post-resection survival of patients with cHCC-CCA. METHODS Patients with pathologically diagnosed cHCC-CCA were randomly divided into training and validation sets. Radiomics features were extracted from portal venous phase computed tomography (CT) images using the least absolute shrinkage and selection operator Cox regression and random forest analysis. A nomogram integrating the radiomics score and clinical factors was developed using univariate analysis and multivariate Cox regression. Nomogram performance was assessed in terms of the C-index as well as calibration, decision, and survival curves. RESULTS CT and clinical data of 118 patients were included in the study. The radiomics score, vascular invasion, anatomical resection, total bilirubin level, and satellite lesions were found to be independent predictors of overall survival (OS) and were therefore included in an integrative nomogram. The nomogram was more strongly associated with OS (hazard ratio: 8.155, 95% confidence interval: 4.498-14.785, P < 0.001) than a model based on the radiomics score or only clinical factors. The area under the curve values for 1-year and 3-year OS in the training set were 0.878 and 0.875, respectively. Patients stratified as being at high risk of poor prognosis showed a significantly shorter median OS than those stratified as being at low risk (6.1 vs 81.6 mo, P < 0.001). CONCLUSION This nomogram may predict survival of cHCC-CCA patients after hepatectomy and therefore help identify those more likely to benefit from surgery.

Magnetic Resonance Imaging Radiomics‐Based Nomogram From Primary Tumor for Pretreatment Prediction of Peripancreatic Lymph Node Metastasis in Pancreatic Ductal Adenocarcinoma: A Multicenter Study

“…Previous studies have indicated that radiomic features are closely linked to patients' prognosis and underlying genomic phenotyping across a wide range of cancer types. 11,12 A few studies have reported that radiomics analysis on the basis of contrast enhanced CT has a convincing predictive power in predicting primary pancreatic tumor and lymph nodes pathological status. 13,14 However, these radiomics signatures were built with few radiomics features or without external validation, making their use in clinical practice very limited.…”

“…Radiomics has recently emerged, a technique that can potentially transform original monochrome images into quantitative features using high‐throughput extraction methods. Previous studies have indicated that radiomic features are closely linked to patients' prognosis and underlying genomic phenotyping across a wide range of cancer types 11,12 . A few studies have reported that radiomics analysis on the basis of contrast enhanced CT has a convincing predictive power in predicting primary pancreatic tumor and lymph nodes pathological status 13,14 .…”

Differential diagnosis of hepatocellular carcinoma and intrahepatic cholangiocarcinoma by ultrasonography combined with multiphase enhanced computed tomography

“…In recent years, there has been significant progress in the application of radiomics in cancer research, which has led to improved non-invasive characterization of lesions 27 . A number of studies have focused on utilizing radiomics to analyze liver tissue and tumors in imaging studies 28 , 29 . Specifically, researchers have extensively explored the potential of radiomics features in discriminating between HCC and benign liver lesions.…”

Radiomics features from 3D‐MPRAGE imaging can differentiate temporal‐plus epilepsy from temporal lobe epilepsy

“…However, the accuracy of manual segmentation depends on the reader's experience. Furthermore, most ROIs are defined by at least two experienced radiologists and the results are compared, which is time‐consuming and may lead to inter‐observer reliability issues 23,24 . Segmentation is the most critical, challenging, and contentious component of radiomics 9 .…”

“…Furthermore, most ROIs are defined by at least two experienced radiologists and the results are compared, which is time-consuming and may lead to inter-observer reliability issues. 23,24 Segmentation is the most critical, challenging, and contentious component of radiomics. 9 In our study, we used an automated segmentation method to define ROIs, which has shown greater reproducibility in feature extraction than manual segmentation.…”