Prediction of Hepatocellular Carcinoma Response to Transcatheter Arterial Chemoembolization: A Real-World Study Based on Non-Contrast Computed Tomography Radiomics and General Image Features

Abstract

To construct a predictive model of short-term response and overall survival for transcatheter arterial chemoembolization (TACE) treatment in hepatocellular carcinoma (HCC) patients based on non-contrast computed tomography (NC-CT) radiomics and clinical features. Methods: Ninety-four HCC patients who underwent CT scanning 1 week before the first TACE treatment were retrospectively recruited and divided randomly into a training group (n = 47) and a validation group (n = 47). NC-CT radiomics data were extracted using MaZda software, and the compound model was calculated from radiomics and clinical features by logistic regression. The performance of the different models was compared by examining the area under the receiver operating characteristic curve (AUC). The prediction of prognosis was evaluated using survival analysis. Results: Thirty NC-CT radiomic features were extracted and analyzed. The compound model was formed using four NC-CT run-length matrix (RLM) features and general image features, which included the maximum diameter (cm) of the tumor and the number of tumors (n). The AUCs of the model for TACE response were 0.840 and 0.815, whereas the AUCs of the six-and-twelve grade were 0.754 and 0.750 in the training and validation groups, respectively. HCC patients were divided into two groups using the cutoff value of the model: a group in which the TACE-response led to good survival and a group in which TACE-nonresponse caused poor prognosis. Conclusion: Radiomic features from NC-CT predicted TACE-response. The compound model generated by NC-CT radiomics and clinical features is effective and directly predicts TACEresponse and overall survival. The model may be used repeatedly and is easy to operate.

A Novel Radiomics Approach for Predicting TACE Outcomes in Hepatocellular Carcinoma Patients Using Deep Learning for Multi-organ Segmentation

“…This was likely attributable to suboptimal organ segmentation in non-contrast CT. This contrasts with a recent study in which a combination of non-contrast CT radiomics and clinical variables proved to be effective in predicting overall survival (OS) in HCC patients following TACE 41 .…”

MRI-based radiomics model and nomogram for predicting the outcome of locoregional treatment in patients with hepatocellular carcinoma

“…Previous studies have shown that radiomics models with or without clinical factors had good performance in predicting the first TACE response in patients with HCC (AUC 0.815-0.900) [ 26 – 28 ]. We assessed the treatment response at 6 months, which possessed more clinical value than that after the first treatment.…”

“…Recent report used MRI radiomics nomogram to predict recurrence after ablation at 1, 2, and 3 years in HCC patients, the AUCs of which (0.72, 0.61 and 0.64) were lower than our study (0.867) [ 29 ]. Moreover, most published studies used CT-based radiomics [ 21 , 22 , 24 , 26 – 28 , 30 ], but MRI is known to have higher soft-tissue contrast than CT and is more commonly used in treated HCC viability evaluation. Based on the reports that postoperative CT and MRI were of predictive value for outcome of LRT [ 23 – 25 ], our study used initial postoperative MRI.…”

CT texture analysis in predicting treatment response and survival in patients with hepatocellular carcinoma treated with transarterial chemoembolization

“…A series of studies have been published using texture analysis or radiomic techniques to predict the response of TACE in HCC across different modalities, such as non-contrast CT [21], contrast CT [22], and MRI [23], with good performance (AUC ranging from 0.884-0.960). Tumor heterogeneity and size were identi ed as critical prognostic features in the Nested multiparametric decision tree [24].…”