Correlation between positron emission tomography and Cerenkov luminescence imaging in vivo and ex vivo using 64Cu-labeled antibodies in a neuroblastoma mouse model

Antibody-based therapies gain momentum in clinical therapy, thus the need for accurate imaging modalities with respect to target identification and therapy monitoring are of increasing relevance. Cerenkov luminescence imaging (CLI) are a novel method detecting charged particles emitted during radioactive decay with optical imaging. Here, we compare Position Emission Tomography (PET) with CLI in a multimodal imaging study aiming at the fast and efficient screening of monoclonal antibodies (mAb) designated for targeting of the neuroblastoma-characteristic epitope disialoganglioside GD2. Neuroblastoma-bearing SHO mice were injected with a 64Cu-labeled GD2-specific mAb. The tumor uptake was imaged 3 h, 24 h and 48 h after tracer injection with both, PET and CLI, and was compared to the accumulation in GD2-negative control tumors (human embryonic kidney, HEK-293). In addition to an in vivo PET/CLI-correlation over time, we also demonstrate linear correlations of CLI- and γ-counter-based biodistribution analysis. CLI with its comparably short acquisition time can thus be used as an attractive one-stop-shop modality for the longitudinal monitoring of antibody-based tumor targeting and ex vivo biodistribution. These findings suggest CLI as a reliable alternative for PET and biodistribution studies with respect to fast and high-throughput screenings in subcutaneous tumors traced with radiolabeled antibodies. However, in contrast to PET, CLI is not limited to positron-emitting isotopes and can therefore also be used for the visualization of mAb labeled with therapeutic isotopes like electron emitters.

Strikingly, we found no significant difference in tumor uptake of 64 Cu-DOTA-ch14.18 in GD2-positive LS-and in GD2-negative HEK293 tumors. As the GD2binding specificity of ch14.18 has been proven in the past, we assessed deviations of the immunoreactive fraction as potential source for this phenomenon; the detailed results will be published in an upcoming publication focusing on GD2 antibodies and GD2-PET imaging. Briefly, chelatorconjugation and subsequent 64 Cu-labeling of the used GD2-specific mAb resulted in a dramatically reduced immunoreactive fraction (reduction to 6.4 %), albeit the afore via FACS, ELISA and IHC assessed GD2-specificity of the used mAb, ch14.18. Potential reasons for this dramatic reduction in immunoreactive fraction could be the presence of a crucial lysine residue in the specificity domain of ch14.18, which could be blocked by coupled chelator, and radiolysis of the mAb caused by 64 Cu.
Most important, the aim of our paper was the comparison of CLI and PET quantification. The investigated antibodies and cell lines were used to determine the correlation of CLI and PET quantification in one biological model system.

Advantages and limitations of CLI
Imaging brain tumors, brain metastasis or tumors deeply-located in healthy tissue with CLI is much less favorable in comparison to a subcutaneous xenograph model -as the neuroblastoma model used in this study. Attenuation and scattering by dense tissue like bone or by the vast tissue mass surrounding deeply-located lesions also causing high rates of scatter, attenuation and light diffusion are major limitations of CLI in comparison to PET. However, for fast and efficient specificity-screening of antibodies targeted at tumor-type characteristic epitopes, we suggest using CLI and subcutaneous xenograph models, with all limitations and benefits of this kind of tumor mouse model system and imaging modality. An additional limitation of CLI is also shown in our study, as in vivo whole-body biodistribution of signal intensity is clearly different between PET and CLI. As the signal intensity of PET can be considered as ground-truth in this comparative imaging setup, the altered signal intensity in CLI-images is most likely caused by tissue attenuation and scatter, and by the low tissue-penetration of light in the blue range of visible light. Only relying on in vivo CLI-imaging might thus result in an incorrect data interpretation regarding tracer biodistribution -ex vivo control experiments by both ex vivo-CLI of excised organs and additional γ-counting are thus crucial for correct conclusions about tracer-specificity, total uptake in tissue of interest or tracer-biodistribution, as shown in this study.
The demonstration of this principle is the major aim of this manuscript, imaging of further tumor-model systems is beyond the scope of this report. Spinelli et al. demonstrate in a recent study the merit of CLI during neurosurgery using 90 Y-DOTATOC as CLI-tracer [1]. During surgical resection of a meningioma, CLI was used as a fast control for the resecting surgeon, confirming the presence of tumor-tissue in the excised mass by luminescence signal caused by 90 Y-DOTATOC tumoruptake, while the surrounding healthy brain tissue showed no radiotracer uptake. In addition to visual inspection during surgery, and resection-planning aided by the use of PET, MRI or CT, CLI adds information that cannot be provided by the other methods in such a setup -thus, CLI has a true added value not only for imaging-based antibody screenings, but also for resection-control directly flanking surgery.

Author contributions
FCM and WMT designed the research. JS established tumor cell lines and animal models. RH provided antibodies. WMT and JS performed PET and MRI measurements. FCM and WMT performed CLI measurements, analyzed the data and conducted statistics. FCM and WMT wrote the manuscript. 64 Cu was provided by WE and GR, antibody labeling was performed by AM. FCM, WMT, KN, RH and BJP discussed data and manuscript. All authors edited the manuscript.