Abstract
Background: High technical complexity limits the wide use of transradial approach (TRA) chemoembolization in the management of liver cancer. We sought to construct a thoracoabdominal aorta CTA-based nomogram model to identify ideal candidates for TRA chemoembolization in patients with liver cancer. Methods: Patients who had received thoracoabdominal aorta CTA before TRA chemoembolization from 2018 to 2020 were retrospectively enrolled and randomly divided into a training set and a validation set. The clinical characteristics and CTA features were collected to build a clinical model. Univariate and multivariate analyses were used to identify significant clinical-radiological variables. A CTA-based nomogram model was constructed by using multivariate logistic regression analysis. The predictive performance, as well as discrimination efficacy of the model, was evaluated by ROC analysis and calibration plot. Results: Vascular variation (P=0.028), Myla classification (P=0.030), length from left subclavian artery to the left subclavian artery (P=0.017), and angle between common hepatic artery and abdominal aorta (P=0.017) were identified as important factors associated with the technical complexity of TRA chemoembolization, indicated by fluoroscopy time of the total procedure. The CTA-based nomogram model was established by these abovementioned variables, which demonstrated good predictive ability in both the training cohort (AUC=0.929) and validation cohort (AUC= 0.769), with a high C-index of 0.928 and 0.827 respectively. Moreover, satisfactory calibrations were confirmed by the Hosmer-Lemeshow test with P values of 0.618 and 0.299 in the training cohort and validation cohort. Conclusion: Our study constructs a novel CTA-based nomogram, which can serve as a useful tool to identify ideal candidates for TRA chemoembolization in patients with liver cancer.