CT-based radiomics signature: a potential biomarker for preoperative prediction of early recurrence in hepatocellular carcinoma

Abstract

Purpose To develop a CT-based radiomics signature and assess its ability for preoperatively predicting the early recurrence (≤1 year) of hepatocellular carcinoma (HCC). Methods A total of 215 HCC patients who underwent partial hepatectomy were enrolled in this retrospective study, and all the patients were followed up at least within 1 year. Radiomics features were extracted from arterial- and portal venous-phase CT images, and a radiomics signature was built by the least absolute shrinkage and selection operator (LASSO) logistic regression model. Preoperative clinical factors associated with early recurrence were evaluated. A radiomics signature, a clinical model, and a combined model were built, and the area under the curve (AUC) of operating characteristics (ROC) was used to explore their performance to discriminate early recurrence. Results Twenty-one radiomics features were chosen from 300 candidate features to build a radiomics signature that was significantly associated with early recurrence (P < 0.001), and they presented good performance in the discrimination of early recurrence alone with an AUC of 0.817 (95% CI: 0.758–0.866), sensitivity of 0.794, and specificity of 0.699. The AUCs of the clinical and combined models were 0.781 (95% CI: 0.719–0.834) and 0.836 (95% CI: 0.779–0.883), respectively, with the sensitivity being 0.784 and 0.824, and the specificity being 0.619 and 0.708, respectively. Adding a radiomics signature into conventional clinical variables can significantly improve the accuracy of the preoperative model in predicting early recurrence (P = 0.01). Conclusions The radiomics signature was a significant predictor for early recurrence in HCC. Incorporating radiomics signature into conventional clinical factors performed better for preoperative estimation of early recurrence than with clinical variables alone.