Evaluation and prediction of therapeutic response for patients with hepatocellular carcinoma receiving stereotactic body radiotherapy using serial computed tomography scans and radiomics

Background: Stereotactic body radiotherapy (SBRT) allows accurate high-energy dose delivery to tumors of interest and has been shown effective for unresectable hepatocellular carcinoma (HCC). In the era of personalized medicine, we aimed to predict therapeutic outcome of SBRT in HCC patients using computed tomography (CT) images and radiomic analyses. Methods: A total of 77 HCC patients undergoing SBRT were retrospectively analyzed. Five millimeter of peritumoral area was identified using semi-automatic method on 3D slicer, and overall 839 radiomic features were extracted, followed by selection with elastic net regularization (ENR). Treatment response was evaluated by CT follow-ups and was quantified by mRECIST. Multivariate logistic regression model was trained and the model performance was evaluated by receiver operating characteristic (ROC) curve analysis. Results: During 4 imaging follow-ups, 34 tumors (43.6%) achieved response, and 5 tumors had complete response (6.4%). Among the 34 tumors, most tumors achieved response at first follow-up (FU1) (N=21, 61.7%). Using logistic regression, we identified that wavelet high-high-lowpass filtering (HHL) GLCM (GLCM waveletHHL) was the most significant feature for response at FU1 (coefficient =0.6805, P=0.0373, 95% CI, 0.0401-1.3208). With this single feature, logistic regression model was built and the model accuracy was 0.83 (AUC =0.71, 95% CI, 0.45-0.81). We also observed responders at FU1 had a trend toward higher survival probability within 2 years (P=0.16). Conclusions: The therapeutic impact of SBRT in HCC could be addressed by the tumor response at FU1, which corresponded to the local control about 1 year after therapy.