CT-based Dosimetry in Yttrium-90 Radioembolization Performed in a Rabbit Liver Model

Submission Summary

Purpose :

Perform a comparison of PET- and CT-based approaches to dosimetry in Yttrium-90 (90Y) radioembolization (RE) following the administration of 90Y-infused radiopaque microspheres (MS) in a rabbit liver model.

Materials & Methods :

An experimental calibration phantom (Figure 1.A) containing multiple cylinders varying in diameter and MS concentration was designed to quantify the correlation between CT Hounsfield units (HU) and MS concentration. Mean HU were extracted from CT volumes-of-interest placed within 15 mm-diameter cylinders containing nominal MS concentrations of 0.5 mg/mL, 5.0 mg/mL, and 25.0 mg/mL. A calibration curve was determined through a linear least squares fit of the data (Figure 1.B). PET and CT imaging of a rabbit liver was performed following the intra-arterial administration of activated 90Y MS. CT voxel values [HU] were transformed into 90Y activity [Bq] through the application of the calibration curve and three scale factors: CT voxel volume [mL], MS number per milligram [MS/mg], and specific activity [Bq/MS]. PET and CT activity distributions were convolved with 90Y dose-voxel kernels generated with the GATE Monte Carlo toolkit to produce PET- and CT-based dose distributions, dd_PET and dd_CT, respectively. Within the liver, the mean dose ± standard deviation (Dµ ± σ) and dose maxima (Dmax) were compared between dd_PET and dd_CT.

Results :

The calibration curve demonstrated strong linearity (r2 > 0.999) between HU and MS concentration. Axial, sagittal, and coronal slices through a high-dose region in dd_CT and dd_PET are shown in Figure 2. In dd_PET, Dmax = 309 Gy and Dµ = 68 ± 63 Gy. In dd_CT, Dmax = 1230 Gy and Dµ = 45 ± 65 Gy. These statistics suggest that Dµ can be ascertained with reasonable confidence independent of the imaging modality while the true range of dose values is better characterized through CT-based dosimetry.

Conclusions :

CT imaging of radiopaque MS distributions in 90Y RE can provide increased confidence in characterizing the dose heterogeneity - relative to PET-based dosimetry - by capturing the true range of dose values within the tumour and healthy liver tissue. The accurate and precise knowledge of the dose distribution is essential in identifying undertreated tumour volumes, identifying radiation toxicity in adjacent liver tissue, and improving our estimates of dose metrics used in establishing a viable dose-response relationship in 90Y RE.