Quantitative dual-energy computed tomography texture analysis predicts the response of primary small hepatocellular carcinoma to radiofrequency ablation
Jin-Ping Li,Sheng Zhao,Hui-Jie Jiang,Hao Jiang,Lin-Han Zhang,Ting-Ting Fan,Song Wang,Zhong-Xin Shi
DOI: https://doi.org/10.1016/j.hbpd.2022.06.003
2022-06-08
Abstract:Background Radiofrequency ablation (RFA) is one of the effective therapeutic modalities on patients with hepatocellular carcinoma (HCC). However, there is no proper method to evaluate the HCC response to RFA. This study aimed to establish and validate a clinical prediction model based on dual-energy computed tomography (DECT) quantitative-imaging parameters, clinical variables, and CT texture parameters. Methods We enrolled 63 patients with small HCC. Two to four weeks after RFA, we performed DECT scanning to obtain DECT-quantitative parameters and to record the patients' clinical baseline variables. DECT images were manually segmented, and 56 CT texture features were extracted. We used LASSO algorithm for feature selection and data dimensionality reduction; logistic regression analysis was used to build a clinical model with clinical variables and DECT-quantitative parameters; we then added texture features to build a clinical-texture model based on clinical model. Results A total of six optimal CT texture analysis (CTTA) features were selected, which were statistically different between patients with or without tumor progression ( P < 0.05). When clinical variables and DECT-quantitative parameters were included, the clinical models showed that albumin-bilirubin grade (ALBI) [odds ratio (OR) = 2.77, 95% confidence interval (CI): 1.35-6.65, P = 0.010], λ AP (40-100 keV) (OR = 3.21, 95% CI: 3.16-5.65, P = 0.045) and IC AP (OR = 1.25, 95% CI: 1.01-1.62, P = 0.028) were associated with tumor progression, while the clinical-texture models showed that ALBI (OR = 2.40, 95% CI: 1.19-5.68, P = 0.024), λ AP (40-100 keV) (OR = 1.43, 95% CI: 1.10-2.07, P = 0.019), and CTTA-score (OR = 2.98, 95% CI: 1.68-6.66, P = 0.001) were independent risk factors for tumor progression. The clinical model, clinical-texture model, and CTTA-score all performed well in predicting tumor progression within 12 months after RFA (AUC = 0.917, 0.962, and 0.906, respectively), and the C-indexes of the clinical and clinical-texture models were 0.917 and 0.957, respectively. Conclusions DECT-quantitative parameters, CTTA, and clinical variables were helpful in predicting HCC progression after RFA. The constructed clinical prediction model can provide early warning of potential tumor progression risk for patients after RFA.
gastroenterology & hepatology