Imaging androgen receptor signaling with a radiotracer targeting free prostate-specific antigen

Abstract

Despite intense efforts to develop radiotracers to detect cancers or monitor treatment response, few are widely used as a result of challenges with demonstrating clear clinical use. We reasoned that a radiotracer targeting a validated clinical biomarker could more clearly assess the advantages of imaging cancer. The virtues and shortcomings of measuring secreted prostate-specific antigen (PSA), an androgen receptor (AR) target gene, in patients with prostate cancer are well documented, making it a logical candidate for assessing whether a radiotracer can reveal new (and useful) information beyond that conferred by serum PSA. Therefore, we developed (89)Zr-labeled 5A10, a novel radiotracer that targets "free" PSA. (89)Zr-5A10 localizes in an AR-dependent manner in vivo to models of castration-resistant prostate cancer, a disease state in which serum PSA may not reflect clinical outcomes. Finally, we demonstrate that (89)Zr-5A10 can detect osseous prostate cancer lesions, a context where bone scans fail to discriminate malignant and nonmalignant signals.

Significance: This report establishes that AR-dependent changes in PSA expression levels can be quantitatively measured at tumor lesions using a radiotracer that can be rapidly translated for human application and advances a new paradigm for radiotracer development that may more clearly highlight the unique virtues of an imaging biomarker.

Figure 1. ⁸⁹Zr-5A10 specifically localizes to multiple preclinical models of AR- and fPSA-positive PCa

(a) Biodistribution data of selected tissues from intact male mice bearing LNCaP-AR xenografts at multiple time points show that peak intratumoral uptake of ⁸⁹Zr-5A10 is observed at 24 hours. Over time, activity depleted from the blood pool, represented by the blood and heart, and like many mAbs, persistently high uptake was observed in the liver. (b) Representative transverse (Trans.) and coronal PET slices of intact male mice bearing LNCaP-AR xenografts shows localization of ⁸⁹Zr-5A10 to the tumor (T) and uptake in the murine liver (L). The tissue from these animals were incorporated into the biodistribution profile at the 120 hour time point (c) Biodistribution data showing tumor associated ⁸⁹Zr-5A10 in multiple subcutaneous PCa models and several treatment conditions in intact male mice. The localization of ⁸⁹Zr-5A10 to LNCaP-AR was entirely competed by co-injection with excess unlabeled 5A10 (1 mg unlabeled mAb). The non-specific radiotracer ⁸⁹Zr-IgG did not localize to LNCaP-AR, and ⁸⁹Zr-5A10 did not localize to PC3, an AR- and fPSA-null model of PCa. Intermediate localization of ⁸⁹Zr-5A10 to CWR22Rv1 xenografts was observed, consistent with the lower basal expression of fPSA in this model compared to LNCaP-AR. *P<0.01 compared to all conditions. **P<0.01 compared to PC3 (d) Surgical implantation of a subcutaneous testosterone pellet in castrate mice bearing LNCaP-AR tumors resulted in increased tumor-associated ⁸⁹Zr-5A10, while uptake in other organs was unchanged. Biodistribution data was acquired at 24 hours post injection. *P<0.01 compared to no treatment (No Tx). Error bars represent the standard deviation from mean.

Figure 2. ⁸⁹Zr-5A10 detects pharmacological inhibition of AR in vivo

(a) Biodistribution data from castrate male mice bearing LNCaP-AR xenografts shows that MDV3100 inhibits localization of ⁸⁹Zr-5A10 to tumor. Animals were treated with vehicle, or the indicated dose of MDV3100 for 7 days, at which time ⁸⁹Zr-5A10 was injected, and animals were harvested for biodistribution studies 24 hours post injection *P<0.01 for the 40 mg/kg and 80 mg/kg dose of MDV3100 compared to vehicle or 10 mg/kg MDV3100. (b) Representative transverse (Trans.) and coronal PET slices of intact male mice bearing LNCaP-AR xenografts on the right flank, and imaged with ⁸⁹Zr-5A10 24 h post injection after manipulation with an subcutaneous testosterone pellet, or a daily oral gavage of vehicle or MDV3100 (80 mg/kg) for 7 days. Clear visual differences in tumor-associated ⁸⁹Zr-5A10 can be seen between the groups. Arrows indicate the position of the tumor (T) and the murine liver (L). (c) Region-of-interest analysis of the tumors from the PET study shows statistically significant changes in tumor-associated ⁸⁹Zr-5A10. *P<0.01 compared to vehicle. **P<0.05 compared to vehicle. Error bars represent the standard deviation from mean.

Figure 3. ⁸⁹Zr-5A10 specifically targets prostate cancer in the bone microenvironment in vivo

(a) A PET/CT image shows that ⁸⁹Zr-5A10 localizes to an osseous LNCaP-AR graft located in the left tibia of an intact male mouse. The image is a three dimensional volume rendering of PET data overlaid onto surface rendered CT data. To generate this image, a weighted average of the intensities along projections through the PET is computed and this in turn is blended with the CT rendered image. The PET data is represented with a semi-quantitative blue (low)-green (high) color scale, and shows a region of contrast in the animal’s left (tumor-bearing) tibia compared to the right (normal) tibia. (b) A co-registered PET/MRI image shows the co-localization of positron emissions from ⁸⁹Zr-5A10 with the tumor-associated contrast detected by MRI (c) Intact male mice received a fracture in the tibia, and ten days post surgery, bone remodeling was evaluated with ¹⁸F-NaF. A clear region of contrast was identified in the fractured tibia by PET. Two days after the first image, animals received a co-injection of ^99mTc-MDP and ⁸⁹Zr-5A10. SPECT imaging showed that ^99mTc-MDP also localized to the region of healing bone, as expected. In contrast, PET imaging showed no detectable ⁸⁹Zr-5A10 at the wound site, pointing to the high specificity of this reagent for PCa compared to contemporary clinical radiotracers. The white line represents the location of the fracture.