Scientific Abstracts: Basic Science and Technology
Objective: To evaluate the feasibility of using 4D flow MRI to characterize hepatic vessel blood flow changes during combined transarterial embolization and microwave ablation, and correlate flow dynamics with final ablation zone volume.
Methods: Transarterial embolization with microwave ablations were performed on female domestic swine (n=5). The left lateral and left medial lobes were embolized to stasis with 300 - 500 um microspheres. Microwave ablations were created under ultrasound guidance within embolized (n=9) and non-embolized (n=11) lobes to target major hepatic vessels at 65 W for 5 min using a 2.45 GHz MW ablation system. 4D flow MRI was acquired pre-embolization and post-embolization to characterize liver hemodynamics. The liver was then excised en bloc and cross-sectional area and volumes were measured. Multivariate linear regression modeling was performed to evaluate the effect of blood flow changes on final ablation volume.
Results: Combined trans-arterial embolization and microwave ablation zones were significantly greater in size compared to microwave ablation only (volume: 16.7 vs 8.5 mL, p=0.0001) (Figure 1). There was a significant decrease in average post-embolization hepatic artery velocity in the main hepatic artery (43.29 +/- 11.1 vs 28.30 +/- 4.2 cm/sec; p = 0.04). Post-embolization hepatic vein flow decreased (7.50 +/- 1.8 vs 5.8 +/- 1.5 cm/sec, p=0.04), and there was a significant compensatory increase in flow seen in the portal vein of the embolized lobe (5.1 +/- 2.2 vs 7.1 +/- 1.1 cm/sec, p=0.02) (Figure 2).
Within the embolized lobes, pre-embolization measurements of vessel size and vessel-antenna spacing of portal veins (regression coefficient B = 0.76 +/- 0.17; p=0.01 and B = -0.45 +/- 0.18; p=0.06, respectively) and hepatic veins (B = -2.22 +/- 0.40; p=0.003 and B = 1.18 +/-027; p=0.005, respectively) contributed to a significantly more accurate model compared to univariate predictors (adjusted R^2 = 0.88 vs 0.30, p=0.017). The addition of hepatic vein 4D-flow led to a small but not significant improvement in model accuracy (adjusted R^2 = 0.89 vs 0.85, p=0.14). Within the non-embolized lobes, only the pre-embolization measurement of the vessel size was significantly correlated to final ablation volume (B = -2.68 +/- 0.40; p=0.046).
Conclusions: Perivascular microwave ablation zones created in liver lobes that have been previously embolized are significantly larger and more predictable in size compared to non-embolized liver lobes. Blood flow values obtained by 4D-flow MRI do not independently correlate with final ablation volume, but may improve the overall model fit when combined with vessel size and vessel-antenna distance.