MRI-based Nomogram Predicts the Risk of Progression of Unresectable Hepatocellular Carcinoma After Combined Lenvatinib and anti-PD-1 Antibody Therapy

Combined immune and anti-angiogenic treatment has shown promising results for unresectable hepatocellular carcinoma (HCC), but with a high risk of early progression. In this study, we aimed to investigate whether pre-treatment magnetic resonance imaging (MRI) features and MRI-based nomogram could predict the risk of disease progression of unresectable HCC after first-line lenvatinib/anti-PD-1 antibody therapy.Thirty-seven HCC participants with qualified pre-treatment contrast-enhanced MRI were enrolled. All patients received combined lenvatinib and anti-PD-1 antibody treatment. Progression free survival rate was analyzed using the Kaplan-Meier method. Potential clinical-radiological risk factors for progression were analyzed using the log-rank tests and Cox regression model. The performance of MRI-based nomogram was evaluated based on C-index, calibration, and decision curve analyses.The 6-month and 12-month cumulative progression free survival rates were 59.5% (95% confidence interval (CI), 43.6%-75.4%) and 48.0% (95% CI, 31.7%-64.3%). On multivariate analysis, no or incomplete tumor capsule (hazard ratio (HR) = 15.215 [95% CI 2.707-85.529], p = 0.002), heterogeneous signal on T2-weighted imaging (HR = 28.179 [95% CI 2.437-325.838]; p = 0.008) and arterial contrast-to-noise ratio ≤95.45 (HR = 5.113 [95% CI 1.538-17.00]; p = 0.008) were independent risk factors for disease progression. Satisfactory predictive performance of the nomogram incorporating the three independent imaging features was obtained with a C-index value of 0.880 (95% CI 0.824-0.937), and the combined nomogram had more favorable clinical prediction performance than any single feature.MRI features can be considered effective predictors of disease progression for unresectable HCC with first-line lenvatinib plus anti-PD-1 antibody therapy, and the combined MRI-based nomogram achieved a superior prognostic model, which may help to identify appropriate candidates for the therapy.