PET Imaging Expedites Detection of Aberration in the Humanization of an Annexin A1 Targeting Antibody

Objectives: Annexin-A1 is a 37 kDa phospholipid-binding protein which is concentrated in a truncated 34 kDa form (AnnA1) in caveolae on the tumor vascular endothelial cell surface with expression in many tumor types. PRISM developed the monoclonal mouse antibody mAnnA1 against AnnA1 for evaluation of AnnA1 as a potential target for imaging and therapy in oncology. mAnnA1 was humanized to make hAnnA1 for translation to clinical studies. Both PRISM-produced mAnnA1 and cGMP contractor-produced hAnnA1 were investigated using noninvasive PET/CT imaging, and dosimetry was evaluated to enable clinical translation of this strategy and to investigate in vivo behavior of hAnnA1. Methods: Antibodies mAnnA1 and hAnnA1 (PRISM "hAnnA1-P" or contractor generated "hAnnA1-C") were conjugated with the chelator deferoxamine and evaluated for immunoreactivity with ELISA. Conjugated antibodies were radiolabeled with zirconium-89. Naïve mice, rats, and non-human primates (NHP) were injected with [⁸⁹Zr]mAnnA1 or [⁸⁹Zr]hAnnA1 and imaged with PET/CT up to 10 days post injection. After imaging, mice and rats were euthanized and organs were collected, weighed, and radioactivity was quantified using a gamma counter. Dosimetry in mice and NHPs were calculated using OLINDA. Results: [⁸⁹Zr]mAnnA1 showed similar biodistribution to other antibodies with slow clearance through the liver. Transition to [⁸⁹Zr]hAnnA1-C during the dosimetry studies revealed substantial uptake in the spleen (130 ± 48% ID/g at day 5 post injection in female BALB/c), which was not observed with [⁸⁹Zr]mAnnA1 (5.6 ± 1.7% ID/g at day 7 PI). Further studies in multiple strains of mice showed variable elevated splenic uptake of [⁸⁹Zr]hAnnA1-C across mouse strains, with the highest uptake observed in female BALB/c mice (118.4 ± 23.1% ID/g) and the lowest uptake observed in male CD1 mice (34.7 ± 10.2% ID/g). Additionally, splenic uptake of hAnnA1-C was observed in Fischer rats (2.8 ± 0.6% ID/organ) and NHPs (1.6 ± 0.6% ID/organ), although at lower levels than what was observed in BALB/c mice (8.8 ± 1.8% ID/organ). Dosimetry results showed similar values between estimates based on mouse and NHP data, with the largest difference seen in the spleen (5.2 vs. 2.6 mSv/MBq in females respectively). Sequencing of hAnnA1-C revealed a frameshift mutation in the antibody sequence introduced during cGMP manufacture. Restoration of the antibody sequence by PRISM returned the antibody distribution into alignment with mAnnA1. Conclusions: An aberration introduced during cGMP production of hAnnA1-C resulted in increased splenic uptake and alteration of the biodistribution in mice. PET imaging enabled quantitative detection of the immunogenic behavior of hAnnA1, which led to detection of the sequence error. Restoration of the sequence resulted in an antibody which was non-immunogenic to mice.