313 - Histopathological and Quantitative Micro-CT Analysis of Microsphere Clustering in an OncoPig Model with Imageable Eye90

Purpose: TARE can achieve curative outcomes with ablative radiation doses, but response varies between tumors. Understanding how procedural factors affect microsphere distribution could optimize treatment. We aim to develop a quantitative framework to analyze microsphere clustering in a preclinical liver cancer model and assess histopathological changes following Eye-90 treatment.

Materials and Methods: We performed selective administration of 2.4 GBq (~961000) Eye90 ® microspheres in three transgenic Oncopigs harboring Cre-inducible TP53^R167H and KRAS^G12D mutations, which had undergone liver tumor induction via ultrasound-guided percutaneous intrahepatic injection of Cre recombinase (AdCre)-incubated liver biopsies, to analyze the tumor-specific microsphere distribution. After euthanasia, the liver was harvested. The liver lobe containing the tumor was divided into sections, including tumor and normal tissue. Micro-CT was performed using a Zeiss Xradia Versa 630 scanner. A total of 1601 projections were acquired. The source-to-sample distance was 27.05 mm and the sample-to-detector distance was 173.59 mm, resulting in voxels of 19.7 mm size in each dimension. A 3-stage algorithm was developed to estimate the number and locations of microspheres in tissue regions: (1) thresholding and region filtering, (2) estimating the number of microspheres in each region using the intensity voxel values, and (3) estimating microsphere locations. A DBSCAN method grouped microspheres into clusters within a radius of ~100 μm. We statistically analyzed the cluster pattern of one Oncopig as a proof-of-concept.

Results: Histopathology demonstrated no viable tumors on explant analysis in all animals treated with Eye90. Imaging analysis of micro-CT data for one with an algorithmic protocol identified 941,452 microspheres across all scanned tissue sections in one Oncopig, with a relative error of 2%. Across 17 sections with 490682 beads, the median number of microspheres per cluster was 2, with a maximum of 21921, and a median nearest cluster distance of 0.349 mm. Overall, 18.4% and 6.4% of clusters contained more than 10 and 50 microspheres, respectively.

Conclusion: Histopathological analysis of all Oncopigs treated with Eye90 revealed no viable tumors. Micro-CT delivered high-resolution imaging of microspheres in both tumor and normal liver parenchyma. Our algorithmic approach successfully quantified biodistribution heterogeneity in tumor and normal tissue. Future research will use the enhanced spatial resolution of micro-CT imaging to perform tumor microdosimetry.