A predictive model of incomplete response after transarterial chemoembolization for early or intermediate stage of hepatocellular carcinoma: consideration of hepatic angiographic and cross-sectional imaging

• Published in: Abdominal imaging 2021-02, Vol.46 (2), p.581-589
• Main Authors: Tao, Pei-Yao, Zhang, Zi-Shu, Wang, Tian-Cheng, Yu, Meng-Qi, Xiao, Yu-Dong
• Format: Article
• Language: English
• Subjects: Angiography; Arteries; Chemoembolization; Decision analysis; Decision trees; Gastroenterology; Hepatocellular carcinoma; Hepatology; Imaging; Interventional Radiology; Liver cancer; Medicine; Medicine & Public Health; Population studies; Prediction models; Radiology; Regression analysis; Solid tumors; Staining; Statistical analysis; Tumors
• ISSN: 2366-004X; 2366-0058
• DOI: 10.1007/s00261-020-02701-5

Objectives The purpose of the present study is to develop a predictive model for incomplete response (IR) after conventional transarterial chemoembolization (cTACE) for hepatocellular carcinoma (HCC) based on hepatic angiographic and cross-sectional imaging. Methods Sixty patients with 139 target HCC lesions who underwent cTACE from February 2013 to March 2019 were included in this retrospective study. Hepatic angiographic features were identified: the number of feeding arteries, vascularity of the tumor, tumor staining on angiography, vascular lake phenomenon, and hepatic arterio-portal shunt. Cross-sectional imaging features were also identified: tumor extent, location, size, and enhancement pattern. Treatment response was assessed by the modified Response Evaluation Criteria in Solid Tumors (mRECIST) criteria. Logistic regression analysis was performed to determine the potential predictive factors for treatment response. To validate the predictive value of potential factors, the means of a decision tree were also calculated by Classification and Regression Tree (CART). P < 0.05 was considered statistically significant. Results The IR rate was 43.2% (60/139) in the entire study population. Logistic regression analysis showed that a tumor size > 50 mm ( P = 0.005; odds ratio, 7.25; 95% CI 1.79–29.33), central location ( P = 0.007; odds ratio, 0.14; 95% CI 0.03–0.59), and nondense tumor staining ( P < 0.001; odds ratio, 0.08; 95% CI 0.02–0.28) were predictors of IR after cTACE. Decision tree analysis showed a good ability to classify treatment response with an accuracy of 78.4%. Conclusion Tumor size > 50 mm, central tumor location, and nondense tumor staining were predictors of IR after cTACE. These factors should be taken into consideration when performing cTACE.