Abstract
365
Objectives We recently reported a nonlinear immunokinetic model based on quantitative PET imaging for direct tumor targeting of radiolabeled antitumor antibodies [1]. In this report, we explore proof of concept for quantitative PET imaging in the pretargeted setting; namely, the tumor targeting efficacy and dosimetry of an anti-HER2/anti-DOTA(metal) pretargeted radiohapten diagnosis and therapy system (DOTA-PRIT) in a HER2-overexpressing ovarian cancer model. The model provided estimates of tumor cell antigen density, tumor uptake transport parameters, and absorbed doses from HER2-positive SKOV3 human ovarian carcinoma xenografts.
Methods Intravenous tail vein injections of bispecific antibody (BsAb) anti-HER2/anti-DOTA(metal) complex (anti-HER2-C825; IgG-scFv) and a dextran-based clearing agent were delivered in sequence prior to scanning. Injection of 86Y-radiolabeled DOTA hapten (iv administered activity: 427 µCi for mouse 1 and 386 µCi for mouse 2) was performed with the mouse in the preclinical PET/CT bore, immediately after the start of the initial acquisition (i.e. while acquiring PET data). Additional PET acquisitions were performed at 2, 24, and 48 hrs post-injection with corresponding CT images at each time point and reconstructed into static images. Additionally, the initial 15-20 min acquisition was reconstructed as a dynamic image set with 10s, 30s, and 1 min frames adjusted to cover the period from 10 s pre-injection to the end of the scan. In this way, the initial blood activity peak could be captured in the image data. Regions of interest (ROIs) were drawn over the tumor, heart, kidneys, and bladder on CT images at each time point and transferred to the corresponding co-registered PET images. Time-activity curves (TACs) were fit to a nonlinear compartmental model [1]. Absorbed radiation doses to tumor and kidneys were estimated using cumulated activity values derived from the ROI TACs, assuming complete local absorption of particulate radiations only.
Results Our PRIT therapy achieved tumor-to-blood ratios of 3.6 and 9.5 for mouse 1 and 2, respectively, at 48 hours post-injection. Data from the dynamic image sets were combined with data from serial static images to form 0-24 hour TACs. Fitting the TACs to the nonlinear compartment model using the image-derived heart TAC as the input function yielded estimated total apparent antigen densities of 0.40 and 0.48 µM in the 2 mice. The apparent association constants were 6.3 × 10^5 and 2.0 × 10^5 /M/h, while efflux rates were 0.037 and 0.061/h. The decay-corrected 86Y-hapten image data were used to estimate residence time of the radiolabeled hapten at different ROIs. Using 90Y decay and equilibrium dose constants, we projected absorbed dose based on the substitution of 90Y for 86Y in this theranostic system. Tumor dose was found to be 462 and 1021 cGy, for mouse 1 and 2, respectively (absorbed-dose ratio therapeutic index: 1.5 and 3.8). Doses to the left and right kidneys were calculated independently but found to be the same within each mouse: 357 cGy for mouse 1 and 183 cGy for mouse 2.
Conclusions The application of non-linear immunokinetic modeling to pretargeted radioimmunotherapy was successfully demonstrated in a preclinical xenografted mouse model bearing HER2 expressing SKOV3 tumors. Using a combination of dynamic and static PET imaging, TACs with 10 s precision were obtained, and rise and fall of heart uptake could be accurately measured over the entire time course of the experiment. This is especially important for deriving suitable input functions in kinetic modeling. We demonstrated the use of pretherapy image data for calculation of projected absorbed dose in a theranostic system using 86Y and 90Y. This methodology will prove useful in continuing evaluation and optimization of our pretargeted radioimmunotherapy paradigm in animals and man. $$graphic_27D0AF5B-FC14-4DF7-AA8C-240C41EBBC31$$