Predicting the outcome of transarterial chemoembolization combined with targeted immunotherapy for unresectable hepatocellular carcinoma based on MRI radiomics.
AI interpretation is pending for this paper.
Open original publication →What the AI sees
Not AI summarized yet.
Research significance
Pending deeper interpretation.
Source abstract
To establish and validate a multi-sequence magnetic resonance (MR) imaging-based radiomics model for predicting the objective response rate (ORR) of hepatocellular carcinoma (HCC) treated with transarterial chemoembolization (TACE) in combination with TKI and PD-1 inhibitors, aiming to maximize the efficacy of combination therapy. A total of 151 patients with unresectable HCC who received TACE combined with TKI and PD-1 inhibitors were included from two institutions between January 2019 and November 2022. Of these, 119 patients from Center 1 were randomized into the training group (n = 71) and the test group (n = 48). The 32 patients from Center 2 were used as an external validation set. The region-of-interest (ROI) and feature extraction from preoperative T2-weighted imaging (T2WI), arterial phase, and venous portal phase enhanced MRI images were manually extracted using 3D-slicer software. The most relevant radiomic features were selected using the least absolute shrinkage and selection operator (LASSO) regression. These features were then integrated with predictive clinical features into a logistic regression (LR) model to build a combined model for predicting tumor response. The predictive performance of each model was evaluated using the area under the curve (AUC) of the receiver operating characteristic (ROC). Model performance was further assessed using the calibration curve and the decision curve. Additionally, we also analyzed the relationship between LR-predicted objective response rate (preORR) and progression-free survival (PFS), as well as overall survival (OS) in patients with intermediate and advanced HCC. Univariate and multivariate survival analyses of PFS and OS were conducted to identify independent predictive factors in patients with unresectable HCC. The BCLC stage was identified as an independent factor affecting the ORR in the clinical model (P < 0.05). The AUC values based on the combined model in the training, test, and validation groups was 0.893(95%CI:0.813-0.973),0.862(95%CI: 0.755-0.965), and 0.804(95%CI: 0.614-0.971), respectively. Calibration curves and decision curves demonstrate the good calibration performance and clinical utility of the combined model. In the survival analysis, PFS and OS were significantly different between LR-predicted ORR + and LR-predicted ORR- (P < 0.05). Multivariate survival analysis showed that BCLC stage, tumor size, and radscore were independent predictors of PFS, while BCLC stage and Child-Pugh grade were independent predictors of OS. The combined model based on MRI radiomics can effectively predict the ORR in unresectable HCC with combination therapy.