Key points are not available for this paper at this time.
The early assessment of treatment outcomes in hepatocellular carcinoma (HCC) post-NK cell therapy remains challenging due to the lack of immediate observable changes in tumor size. In this study, we aim to investigate the potential of using machine learning models based on radiomic features for the precise evaluation of HCC treatment outcomes after combination therapy of intraarterial transcatheter NK cell delivery and Sorafenib administration. N1S1 tumors were implanted in Twenty-four Sprague-Dawley rats. The animal models were assigned to four groups: control group (n=6), intraarterial transcatheter NK cell delivery group (n=6), Sorafenib group (n=6) and intraarterial transcatheter NK cell plus sorafenib (combination) therapy group (n=6). Animals in the NK cell immunotherapy and combination groups underwent catheterization of the proper hepatic artery. The NK therapy treatments were performed through intra-hepatic artery (IHA) local NK cell delivery. For Sorafenib treatment, animals were administered sorafenib daily for a week. The animals underwent weekly MRI examinations for 3 weeks and multi-parametric MRI data were collected. T2-weighted MRI texture features were extracted using five approaches. Feature selection was subsequently performed, eliminating highly correlated features and employing Recursive Feature Elimination (RFECV). Kernel-based Support Vector Machine (SVM) and Random Forest (RF) were developed to differentiate the treatment response and their performance was evaluated via accuracy and area under curve (AUC) of receiver operating characteristic (ROC) curve. Six T2w MRI features were identified as key predictor for predictive model. Random Forest model demonstrated superior performance for identifying the therapeutic efficacy. In the training set, the model achieved an accuracy of 100% and AUC of 1.00, along with a sensitivity of 98.7% and specificity of 97.3%. For the validation set, the model exhibited an accuracy of 96.0%, an AUC of 1.00, a sensitivity of 95.6%, and a specificity of 91.1% for early prediction of therapeutic response. T2-weighted MRI radiomic features provide insights into the effects of combination of sorafenib and NK cell immunotherapy for HCC, emphasizing the potential of quantitative MRI analysis for dynamic monitoring and early assessment of combination therapies in HCC.
Yu et al. (Wed,) studied this question.