Radiomics Based on Contrast-enhanced CT for Recognizing c-Met-Positive Hepatocellular Carcinoma: a Noninvasive Approach to Predict the Outcome of Sorafenib Resistance

Abstract

Abstract Objectives The purpose of our project was to investigate the effectiveness of radiomic features based on contrast-enhanced CT that can detect the expression of c-Met in hepatocellular carcinoma (HCC) and to validate its efficacy in predicting the outcome of sorafenib resistance. Materials and Methods In total, 130 patients (median age, 60 years) with pathologically confirmed HCC who underwent contrast material–enhanced CT from October 2012 to July 2020 were randomly divided into a training set (n = 91) and a test set (n = 39). Radiomic features were extracted from arterial phase (AP), portal venous phase (VP) and delayed phase (DP) images of every participant’s enhanced CT images. Results The entire group comprised 39 Met-positive and 91 Met-negative patients. The combined model, which included the clinical factors and the radiomic features, performed well in the training (area under the curve [AUC] = 0.878) and validation (AUC = 0.851) cohorts. The nomogram, which relied on the combined model, fit well in the calibration curves. Decision curve analysis (DCA) further confirmed that the clinical valuation of the nomogram achieved comparable accuracy in c-Met prediction. Among another 20 patients with HCC who had received sorafenib, the predicted high-risk group had shorter overall survival (OS) than the predicted low-risk group (p < 0.05). Conclusion A multivariate model acquired from three phases (AP, VP and DP) of enhanced CT, HBV-DNA, and GGT-II could be considered a satisfactory preoperative marker of the expression of c-Met in patients with HCC. This approach may help in overcoming sorafenib resistance in advanced HCC.