Practical considerations of gadoxetate disodium contrast in MRI-guided liver SBRT for primary and secondary liver tumors: optimizing image quality and patient throughput.

PURPOSE/

Objectives

Optimal implementation of gadoxetate disodium injection (GDI) in liver-directed MR-guided online adaptive radiotherapy (MRgART) remains poorly characterized. Herein, we evaluated the impact of GDI on physician-reported imaging metrics and quantitative contouring variability during liver MRgART planning. MATERIALS/

Methods

Patients were scanned using a 0.35T MRI onboard the MR-guided LINAC. Pre- and post-contrast TRUFI images were captured at multiple timepoints up to 30 minutes (5-10-minute intervals) following GDI. Four radiation oncologists independently scored each tumor on a 5-point scale (1: poor; 5: excellent) evaluating imaging metrics, such as tumor visualization, border delineation, and image degradation from artifacts. Responses were stratified into pre-contrast, "early" post-contrast (1-15 minutes), and "late" post-contrast (20-30 minutes). Physicians then contoured gross tumor volume (GTV) for 6 patients with either optimal or non-optimal image characteristics. Inter- and intra-observer variability was assessed using Sørensen-Dice coefficient (DICE), Hausdorff distance (HD), volume of tumor not treated (VTNT), and volume of healthy liver erroneously treated (VHL). Comparisons were performed using Mann-Whitney U-test or Wilcoxon Signed-Rank test, as applicable.

Results

Twenty-three patients with 37 total liver tumors were treated using MRgART (median 50 Gy; 3-5 fractions). Late post-contrast imaging significantly improved tumor visualization (4.45) compared to early post-contrast (3.73; p<0.001) and pre-contrast (3.25; p<0.001). Late post-contrast also significantly improved border delineation (4.38) versus early post-contrast (3.58; p<0.001) and pre-contrast (3.06; p<0.001). In 6 patients selected for contouring analysis, DICE increased from 0.75 to 0.92 (p<0.01) and HD90 decreased from 4.18 mm to 2.00 mm (p<0.01) when GDI was helpful (n=3). Across all 6 patients, median VTNT significantly decreased from 3.68 cm3 to 0.68 cm3 (p<0.001).

Conclusions

GDI administration 20-30 minutes prior to MRgART planning significantly enhances both physician-reported image quality and quantitative contouring accuracy. Its use should be considered in MRgART workflows to minimize potential geographic miss and optimize treatment delivery.