IDDF2023-ABS-0132 Evaluation of the targeted pathological indicators related to the combining targeted therapy and immunotherapy in hepatocellular carcinoma based on multi-parameter magnetic resonance imaging

Background

We aimed to explore the feasibility of multi-parameter MRI in evaluating the relationships between MRI parameters and the targeted indicators (VEGFA, PD-L1) related to combining targeted therapy and immunotherapy (TIT) in different tumor regions.

Methods

MRI scans with liver-specific contrast were conducted on C57/BL6 mouse model. 3D printing technology was used to spatially match the planes of MRI with pathology (n_tumor=15, n_plane=115). In the corresponding planes of MRI and pathology, MRI parameters and the targeted indicators related to TIT were evaluated and compared in the regions of the plane, margin, center, and whole tumor. Based on MRI parameters, we constructed the evaluation model for the pathological indicator in different regions (plane, margin, center, and whole tumor) respectively. The combined targeted indicators related to TIT were grouped into V^highI^high (high VEGFA and high PD-L1), V^highI^low, V^lowI^high, V^lowI^low groups. MRI parameters with significant differences among these four groups were used to establish the evaluation model.

Results

VEGFA, PD-L1, and MRI parameters Ktrans, iAUC60, T1pos significantly differed between the regions of margin and center(P<0.05). MRI parameters significantly correlated to the targeted indicators in different regions. MRI parameters Ktrans, Kep, Ve, and α significantly differed between VEGFA high and low groups in different regions. The AUCs of the VEGFA prediction model at the regions of margin combining center (ROC^M+T) and margin (ROC^M) were 0.732(95%CI 0.663-0.793), and 0.715(95%CI 0.621-0.797), respectively. MRI parameters Kep, Ve, iAUC60, and T1pre, T1pos significantly differed between PD-L1 high and low groups in different regions. The AUCs of PD-L1 prediction models of ROC^M+T, ROC^M, ROC^plane were 0.707(95%CI 0.637-0.770), 0.647(95%CI 0.552-0.735), 0.648(95%CI 0.552-0.736), respectively. Ktrans, Kep, Ve, and T1pre significantly differed among V^highI^high, V^highI^low, V^lowI^high, and V^lowI^low groups. The prediction model was statistically significant (P=0.001) with the goodness-of-fit test P> 0.8.

Conclusions

Multi-parameter MRI has the potential to evaluate the expression and distribution of the targeted indicators (VEGFA, PD-L1) related to TIT.