Preoperative PET imaging and fluorescence-guided surgery of human glioblastoma using dual-labeled antibody targeting ETA receptors in a preclinical mouse model: A theranostic approach

Abstract

Rationale: Glioblastoma (GBM) poses significant challenges regarding complete tumor removal due to its heterogeneity and invasiveness, emphasizing the need for effective therapeutic options. In the last two decades, fluorescence-guided surgery (FGS), employing fluorophores such as 5-aminolevulinic acid (5-ALA) to enhance tumor delineation, has gained attraction among neurosurgeons. However, some low-grade tumors do not show any accumulation of the tracers, and the lack of patient stratification represents an important limitation. Since 2000, endothelin axis has been extensively investigated for its role in cancer progression. More specifically, our team has identified endothelin A receptors (ET_A), overexpressed in glioblastoma cancer stem cells, as a target of interest for GBM imaging. This study aims to evaluate the efficacy of a novel preclinical bimodal imaging agent, [⁸⁹Zr]Zr-axiRA63-MOMIP, as a theranostic approach to: i) detect ET_A ⁺ cells in an orthotopic model of human GBM, ii) achieve complete tumoral resection. Methods: Monomolecular multimodal imaging platform (MOMIP) - containing both a fluorophore (IRDye800CW) and a chelator for a positron-emitting radiometal (desferroxamine B, DFO) - was conjugated to the axiRA63 antibody targeting ET_A receptors, overexpressed on the surface of GBM stem cells. Mice bearing orthotopic human GBM were imaged 48 h post injection of [⁸⁹Zr]Zr-axiRA63-MOMIP via positron emission tomography (PET) and optical imaging. Subsequently, post-mortem proof-of-concept FGS was implemented as well as ex vivo analyses (H&E staining, autoradiography, serial block face imaging) on brains with resected or unresected tumor to assess the correlation between PET and fluorescence signals. Results: PET imaging of [⁸⁹Zr]Zr-axiRA63-MOMIP enabled a clear detection of ET_A ⁺ cells in an orthotopic model of human GBM. Intraoperative optical imaging allowed a near-complete tumor resection together with the visualization of a weak fluorescence signal, after a prolonged exposure time, that was attributed to residual tumor cells via H&E staining. Besides, a qualitative correlation between the signals of both modalities was observed. Conclusions: The use of [⁸⁹Zr]Zr-axiRA63-MOMIP provides an effective theranostic approach to detect and treat GBM by surgery in a preclinical mouse model. Thanks to the high correlation between PET and fluorescence signal allowing patients stratification, this bimodal agent should have a great potential for clinical translation and should present a significant advantage over non-targeted fluorophores already used in the clinic.

Keywords: Bimodal Imaging; Endothelin A (ETA); Fluorescence-Guided surgery; Glioblastoma; ImmunoPET.

Figure 1

Synthesis of the bimodal imaging agent. A. Structure of the MOMIP. B. Two-step process bioconjugation to the antibody: 1) Enzymatic conjugation with MTGase to introduce the TCO moiety; 2) addition of the MOMIP via an Inverse Electron Demand Diels-Alder (IEDDA) reaction.

Figure 2

Brain tumor uptake and immunoPET imaging of [⁸⁹Zr]Zr-axiRA63-MOMIP at 48 h p.i. A. PET and MRI images. Brain tumor coronal section by MRI with contrast agent obtained 18 days before injection and by PET imaging at 48 h p.i. for the three groups: surgery, blocking and control. B. Quantitative uptake of [⁸⁹Zr]Zr-axiRA63-MOMIP in the Gli7 model. Data are presented as mean ± SD; statistical comparisons were performed using a two-tailed paired Student's t test. **P < 0.01; *P < 0.05; ns = not significant; n (mice surgery group) = 5; n (mice blocking group) = 3; n (mice control group) = 3.

Figure 3

Post mortem fluorescence images during different dissection stages. Fluorescent images were acquired post mortem at 48 h p.i. of [⁸⁹Zr]Zr-axiRA63-MOMIP-IR800, with the KIS 800 imaging system (Kaer Labs). The first set, called "whole mouse”, includes an image of the mouse head with skin, skull and brain with an exposure time of 500 ms. Next, an image of the mouse head skull and brain, called "brain with skull", was taken on all three groups of mice, after skin removal. Finally, the last set of images, called "brain", was taken after removal of the upper cranial cavity. The last two conditions required an exposure time of 100 ms.

Figure 4

Ex vivo co-localization of the fluorescent and nuclear signals in the brain tumor 48h p.i. of [⁸⁹Zr]Zr-axiRA63-MOMIP. A. Comparison of the 3D fluorescence reconstruction using the kratoscope (Kaer Labs) and the sagittal PET nuclear imaging. B. Comparison of nuclear and fluorescent signal on corresponding brain tumor sections obtained by autoradiography and block face fluorescence imaging (scale bars represent 2.4 mm).

Figure 5

Example of an ex vivo FGS performed post mortem on the GBM preclinical model: Left image shows the brain tumor Gli7 ET_A⁺ before surgery without fluorescence signal. Middle image corresponds to the brain Gli7 ET_A⁺ tumor viewed by fluorescence with a time exposure of 100 ms. Right image shows an image of a resected brain guided with fluorescence via the KIS 800 system (Kaer Labs) with a time exposure of 100 ms.

Figure 6

Ex vivo assessment of the GBM's fluorescence-guided surgery quality: Fluorescent tumor residues imaged by the KIS 800 at 100 ms, 200 ms and H&E coloration of brain sections corresponding to these residues.