Characterization of tumor-associated neovasculature in HCC

Objectives: Radio-labeled prostate specific membrane antigen (PSMA) ligands have been developed for clinical imaging of both primary and metastatic prostate cancers. Besides, although PSMA has been reported to be overexpressed in the neovasculature of most solid human tumors, the lack of animal models has prevented the assessment of PSMA-targeted radioligands for imaging and therapy of human tumors, other than prostate cancers. We have identified the woodchuck model of primary liver cancer such as hepatocellular carcinoma (HCC) with overexpressed PSMA on tumor neovasculature allowing a “re-purposing” of ⁶⁸Ga-PSMA ligands for PET imaging of HCC. Our preclinical studies have confirmed the initial clinical findings on the uptake of ⁶⁸Ga-PSMA in HCC. This study is to further validate tumor-associated vascular PSMA expression in HCC.

Methods: Both clinical grade ⁶⁸Ga-PSMA-11 and ⁶⁸Ga-PSMA I&T were given IV to 3 woodchucks bearing naturally occurring HCC after two-four years of chronic viral hepatitis infection. PET acquisition started upon ⁶⁸Ga-PSMA bolus injection in list-mode and lasted for 60 min. The PET acquisitions were binned into 10 X 0.5, 5 X 1, 2 X 5 and 4 X 10 min, 21 frames for reconstruction using CT-based attenuation correction. Patlak and Logan graphical analyses were performed, using weighted heart and portal vein activity curves as a composite input function. Bmax was estimated for PSMA target density in the tumor region at known probe specific activities. For some animals, multi-phase contrast-enhanced CT scans were also performed. Using harvested liver tissues, Western Blot using the anti-(human) PSMA antibody was performed, and enzymatic assay for folate hydrolase was conducted as well. Results: The woodchuck HCC showed comparable uptake between ⁶⁸Ga-PSMA-11 and ⁶⁸Ga-PSMA I&T despite that modeling by crystal structure showed slightly better docking for PSMA-11 to the target. Analysis of PET image revealed poor fit for Patlak analysis in tumor and liver regions while excellent fit was obtained from Logan analysis. PSMA probe Binding Potential reached above 0.65 in tumor while stayed less than 0.25 in liver. Western Blot confirmed the existence of a higher level woodchuck PSMA in its HCC, and enzyme assay also showed a higher folate hydrolase activity in HCC in comparison with the surrounding liver parenchyma. After 100 days, which is equivalent of one-and-half woodchuck years, the liver tumor progressed from 2.1 cm to 3.7 cm, but the SUV of PSMA-11 remained stable above 3.0 in HCC and the Bmax maintained around 5.0 pmol/ml in HCC, approximately 1X10⁵ PSMA binding sites per cell. Conclusions: The woodchuck model of spontaneous HCC is insofar the only animal model with imaging detectable endogenous woodchuck PSMA expression in HCC. Specifically, the ⁶⁸Ga-labeled peptides developed against human PSMA seemed to bind with the woodchuck native PSMA in its liver tumor-associated neovasculature. As the primary liver cancer progressed with time, the uptake of ⁶⁸Ga-PSMA in HCC remained steady. The tumor-associated vascular PSMA target density was also unchanged despite of the growing size of the tumor over time.