Investigation into Use of Positron Emission Tomography (PET) as an In Vivo Imaging Tool to Quantify PD-L1 Tumor Expression Levels

Objectives PD-L1 (Programmed Death Ligand 1) is a 40 kDa immune regulatory ligand that binds to PD-1 (Programmed Death 1), which functions as an immune checkpoint and is expressed on activated immune cell types including T cells, B cells, natural killer (NK) cells and many tumor-infiltrating lymphocytes (TILs).[1] PD-L1 binding to PD-1 deactivates these cytotoxic T cells, and as a consequence the expression of PD-L1 in tumors is correlated to immune suppression and poor prognosis. Therefore it may be feasible to use tumor expression of PD-L1 as a predictive marker for anti-PD-1 therapy. The purpose of this work was to evaluate the feasibility of using PET as and in vivo tool to image PD-L1. For this, the human PD-L1 specific mAb 22C3 was chosen as our proof of concept molecule.

Methods The anti-human PD-L1 mAb 22C3 and isotype matched control mAb 27F11 were conjugated with DOTA and subsequently radiolabeled with [64Cu]. MicroPET imaging studies were carried out in SCID mice implanted with the LOX human malignant melanoma cell line, known to express PD-L1, 48 hr. after administering either [64Cu]DOTA-22C3 (n=4) or [64Cu]DOTA-27F11 (n=4). Blocking studies were also performed in which the tumor bearing mice received 200 µg of unlabeled 22C3 1 hr. prior to administration of [64Cu]DOTA-22C3 (n=4). In all cases following the 48 hr. imaging scan, the mice were euthanized and biodistribution studies performed. Along with imaging studies, an in vitro homogenate binding assay using LOX xenografts homogenates was performed to measure the binding potential for both [64Cu]DOTA-22C3 and [64Cu]DOTA-27F11. Finally, autoradiographic and PD-L1 IHC staining were performed on the LOX tumors to see if tracer uptake correlated to regions of high PD-L1 expression.

Results Displaceable and saturable binding of [64Cu]DOTA-22C3 was observed from the in vitro homogenate binding assay, with [64Cu]DOTA-22C3 binding to a single site with high affinity (Kd = 0.4 nM). For [64Cu]DOTA-27F11 binding in tumor tissue was minimal and not saturable, indicating that there was no measureable specific binding. ComparisonComparison of autoradiography and HC staining of tumor slices confirmed that PD-L1 expression patterns matched [64Cu]DOTA-22C3 binding patterns. A similar correlation was not observed for [64Cu]DOTA-27F11. From the PET Images, tumor uptake in the mice that received [64Cu]DOTA-22C3 was clearly visualized, with uptake higher than [64Cu]DOTA-27F11. In addition, pretreatment with the unlabeled 22C3 reduced tumor uptake of [64Cu]DOTA-22C3 mAb to levels comparable to [64Cu]DOTA-27F11. Biodistribution data at 48 hr. confirmed the PET imaging analysis with tumor uptake of [64Cu]DOTA-22C3 being significantly higher than [64Cu]DOTA-27F11 (p<0.001), 6.38±0.42 vs. 2.58±0.06 SUV respectively. While pretreatment of 22C3 significantly reduced tumor uptake of [64Cu]DOTA-22C3 to 2.30 ± 0.25 SUV (p=0.001).

Conclusions The results of this study demonstrate that a PET ligand can specifically target PD-L1-expressing tumors in a mouse model, and support the hypothesis that PD-L1 can be imaged in vivo in the clinic using PET.