Tozuleristide Fluorescence-Guided Surgery of Brain Tumors

14 Tozuleristide Fluorescence-Guided Surgery of Brain Tumors

Harish Babu, Dennis M. Miller, Julia E. Parrish-Novak, David Scott Kittle, Pramod Butte, and Adam N. Mamelak

Abstract
Tozuleristide is a near-infrared (NIR) tumor-targeting molecule. The tozuleristide molecule is a combination of the tumor-targeting peptide chlorotoxin and the NIR fluorophore indocyanine green. Tozuleristide demonstrates excellent binding to multiple tumor types including high- and low-grade gliomas. It has been tested in a number of human phase I trials, with no appreciable toxicity or side effects. Tozuleristide is now being tested for its capacity for fluorescence-guided removal of gliomas. In conjunction with a novel highly sensitive NIR imaging device (Synchronized InfraRed Imaging System [SIRIS]), which is required to detect the low levels of NIR light generated from tumor-specific targeting of tozuleristide in the brain, phase I trials demonstrated reliable uptake and the ability to differentiate tumor from normal brain tissue. Surgeon acceptance of the drug/device combination has also been very good. This chapter describes the preclinical and clinical experience with tozuleristide for gliomas and the associated imaging devices.

Keywords: fluorescence, near-infrared, tumor specific, intraoperative, imaging, chlorotoxin, tozuleristide, glioma, imaging device, surgery

14.1 Introduction

An estimated 25,000 surgical resections or biopsies are performed for brain tumors each year in the United States. Several studies have shown that the extent of resection is the single most important factor in determining survival.1,2 Balancing maximal tumor cytoreduction with preservation of adjacent brain parenchyma is often limited due to the invasive nature of gliomas and their spread along the connection between cortical and subcortical structures within the brain. As gliomas recur locally at the edge of the previous resection margin, efforts to maximize the extent of resection correlates with increased tumor control and improved survival and quality of life. Current use of the operating microscope relies on visual inspection of the subjective tumor margin, but often tumor deposits are not appreciated with these methods. MRI-based frameless neuronavigation assists in evaluating the tumor margins, but accuracy is lost once a surgical resection leads to “brain shift.”³ Advanced technologies such as confocal microscopy,⁴ MRI spectroscopy, time-resolved fluorescence spectroscopy,⁵ and Raman spectroscopy^6,7 remain research tools at present.

Imaging techniques can be divided between those that work on untreated tissue (endogenous contrast) and those requiring administration of extrinsic contrast agents. Endogenous contrast can be achieved through modifications in autofluorescence, infrared reflectivity, Raman scattering, and microanatomical cytoarchitecture. Exogenous contrast agents allow improved contrast between diseased and normal tissue. Fluorescence-guided neurosurgery requires agents that not only specifically target glioma tissue, but also can do so while the tissue is within the living patient, rather than following tissue excision. Simultaneous evolution of imaging technologies along with fluorescent dyes has generated a powerful tool that allows a surgeon to easily identify and remove tumor while sparing normal tissue. Fluorescent-labeled probes with unique molecular targets can provide intraoperative real-time distinction of cellular edge between glioma and adjacent normal brain. This technique of “fluorescence-guided surgery” (FGS) could revolutionize glioma surgery, allowing better delineation of tumor and thus improved extent of resection.

14.2 Agents for Fluorescence-Guided Glioma Surgery

There are three principal approaches used to generate a relatively strong fluorescent signal in brain tumors to guide resection: (1) passive—labeling occurs when the damaged blood–brain barrier allows exogenous agents to accumulate at the tumor site (e.g., fluorescein, indocyanine green [ICG], etc.); (2) metabolic—endogenous fluorescent agents are internalized, metabolized, and accumulated intracellularly within the tumor cells (e.g., 5-aminolevulinic acid [5-ALA]); and (3) molecular—targeted agents that bind to molecules on the cell surface of the tumor cell or are internalized into tumor cells (e.g., chlorotoxin [CTX]).

Here we focus on the specific imaging properties of tozuleristide, a promising tumor-specific fluorescent probe for FGS, and related imaging device requirements. Approaches for tumor resection using other fluorescent agents, such as the use of 5-ALA, and unconjugated fluorophores like sodium fluorescein and ICG will be addressed in other chapters in the book.

14.3 Advantages of Near-Infrared Imaging for Fluorescence-Guided Surgery

Fluorescent probes that emit light in the near-infrared (NIR; 700- to 1,000-nm excitation and emission) range offer several unique advantages over agents that fluoresce in the ultraviolet or visible light ranges. These include less scatter of emitted light, low tissue autofluorescence, and superior penetration of incident photons. The emission in the NIR spectrum allows for better visualization of deeper structures than seen using the visible or ultraviolet wavelength range. This spectral window does not interfere with other tissue components such as water, hemoglobin, and deoxyhemoglobin that can autofluoresce, making it ideal for fluorescence imaging of living tissue. Several NIR fluorophores are commercially available, including ICG, IRDye800 (Licor), Alexa800, and Cy5.5. At present only ICG is Food and Drug Administration (FDA) approved for clinical use in humans. ICG has a peak excitation of 780 nm and it emits at 805 to 825 nm in human tissues. ICG fluorescence angiography has been used in vascular neurosurgery to identify and characterize vessel integrity. Tumors induce neoangiogenesis that lead to angiogenic hot spots. This property of gliomas can be exploited, allowing angiogenic hot spots to be visualized through ICG fluorescence angiography. The dye accumulates in glioma due to the enhanced permeability and retention (EPR) effect. This accumulation in tumor tissue is enhanced because of the hypoxic environment 8 in most gliomas. Studies have shown direct correlation between gadolinium enhancement and ICG fluorescence.⁸ It was recently shown that intensity of intraoperative fluorescence was directly proportional to the amount of tumor cells.⁹ Similar to other nonspecific fluorophores, ICG-based detection is far superior in high-grade tumors, and typically absent in low-grade gliomas. It has also demonstrated some utility for meningioma.¹⁰

14.4 Targeted Fluorescence

Targeted NIR fluorescence imaging is performed by conjugating a ligand that binds to tumor cells with a fluorophore to generate a fluorescent signal to guide surgical removal. Examples include a cathepsin-activated tumor ligand conjugated to ICG,^11,12 an NIR fluorescent alkylphosphocholine analog,¹³ and modified integrin receptor ligands conjugated to an NIR dye.^14,15 This strategy is appealing as the fluorescent signal is specific to the tumor cells themselves. Such specificity should theoretically lead to better and more confident delineation of tumor margins and regions of infiltrating tumor, while avoiding unintended removal of adjacent brain tissue. In exchange for this increased specificity, targeted fluorescence signals are often several orders of magnitude less bright than those generated by free, unbound fluorophores such as ICG or sodium fluorescein, and therefore require specialized instrumentation to reliably detect the fluorescence. To date, the most appealing strategy for glial tumor is tozuleristide (Blaze Bioscience), a covalent conjugate of a tumor-targeting peptide CTX and ICG.

14.5 Chlorotoxin

CTX, a peptide initially isolated from the venom of the scorpion Leiurus quinquestriatus, preferentially binds to gliomas and other tumors of neuroectodermal origin.¹⁶ CTX is a 36-amino acid peptide (4.7 kD) with four disulfide bridges. It has been shown to indirectly inhibit the ClC₃ chloride ion channel,¹⁷ and to bind matrix metalloproteinase 2 (MMP-2),¹⁸ and annexin A2.¹⁹ Annexin A2 when combined with S100A10 is the putative primary binding target. Upon binding, CTX is internalized into the cell via clathrin-mediated endocytosis (and possibly other mechanisms^20,21,22) in glioma cells. In normal cells, annexin A2 is generally expressed as an intracellular protein, but in cancerous cells is phosphorylated and combines with S100A10 to be expressed on the cell surface.²³ There are also reports of increased annexin A2 expression in gliomas.

The extent of CTX binding to glioma cells, as measured by immunohistochemistry, has been shown to correlate with histological grade. Essentially complete and uniform binding occurs with all WHO grade IV (glioblastoma [GBM]) tumors, 90% binding in WHO grade III tumors, and only 40 to 45% binding in WHO grade I to II (low-grade glioma).²⁴ This is likely attributable to the observation that the number of cell surface receptors increases with grade, but nonetheless does appear to be present in all gliomas to an extent. In mouse xenograft and genetically modified spontaneous tumor models, pathological tissues tagged as cancer using a CTX:Cy5.5 fluorescence conjugate were confirmed to be cancerous, while adjacent nonfluorescent tissues were histologically normal.²⁵ This was true for a variety of cancers in mouse models. Interestingly, in these mouse models, glioma tissue was distinguishable from normal tissue 14 days after injection of CTX-Cy5.5.²⁵ Negative imaging (i.e., no fluorescence signal) was seen in mice in which tumors failed to implant, indicating that the effect is tumor specific and not related to local blood–brain barrier disruption alone. Intravenously administered radioactive Iodine (¹³¹I) labeled CTX had a high concentration in tumor compared to surrounding normal tissue.²⁶ In a phase I clinical trial, a single dose of CTX labeled with radioactive iodine (TM601; Transmolecular Industries), delivered directly into the patient’s brain via an intracavitary Ommaya reservoir after surgical resection of glioma, was shown to be well tolerated.²⁷ While unbound TM601 was eliminated from the body within 24 to 48 hours after delivery,^27,28 the drug that bound to the tumor cavity could be detected up to 7 days after administration, with a dose indicating long-term binding. Comparison of tumor volumes as determined by single-photon emission computed tomography (SPECT) imaging and MRI showed that tumor volumes obtained by ¹³¹I-CTX closely paralleled the T2-, but not T1-weighted gadolinium contrast volume with MRI.²⁸ Unfortunately, these studies and subsequent phase II studies failed to demonstrate sufficient survival advantage to justify further trials. In vivo animal studies using systemically injected fluorescent dyes (Cy5.5, IRDye800) conjugated to CTX have confirmed the specific binding of CTX to GBM.^25,29,30

MRI-compatible contrast agents or nanoparticles conjugated to CTX have been used both for advanced imaging and as a therapeutic molecule in animal model.^31,32 CTX-conjugated nanoparticles have been used to deliver chemotherapeutic drugs to glioma cells.³³ Importantly, all reported human clinical trials of CTX have demonstrated negligible toxicity in humans.

14.6 Tozuleristide

Tozuleristide is a conjugate of a modified CTX covalently attached to ICG. Tozuleristide is a drug candidate being developed for the specific purpose of FGS in many tumors types, with an initial focus on brain cancers, including gliomas ( Fig. 14.1). Tozuleristide has undergone extensive preclinical toxicity testing in both small and large animal models, as well as testing in animals harboring tumors. These data demonstrated that to date tozuleristide has minimal toxicity, even at very high doses, and can reliably detect a variety of tumors.^34,35 Tozuleristide is administered as a single intravenous (IV) injection, and tumor imaging is typically performed the same day or the next day (~ 24 hours) after IV administration, unlike traditional ICG imaging performed minutes after injection. Tozuleristide imaging is more similar to the technique of “second window” ICG imaging, in which tumors such as meningioma or glioma can be detected 18 hours or more after initial ICG injection, by taking advantage of the EPR effect.10 Delayed imaging allows time for unbound tozuleristide in the circulation to be washed out, with specifically bound drug accumulating in the tumor tissue (accumulation in ancillary tissues such as dura do not affect its potential utility). This unique feature of targeted FGS results in a very low concentration (50 pM–50 nM) of the fluorophore to be visualized during surgical resection. These concentrations are typically 100 to 1,000 times less than is seen in typical vascular applications of ICG imaging. Tozuleristide is capable of binding to both low-grade glioma (LGG) and high-grade glioma (HGG) with no appreciable uptake in normal brain tissue.

Fig. 14.1 Schematic drawing of fluorescence-guided surgery (FGS) using tozuleristide. (a) Tozuleristide is formed by covalent bond of the knottin peptide chlorotoxin to a single indocyanine green (ICG; red dot). The chlorotoxin binds the tumor, while the ICG provides a fluorescent signal. (b) Tozuleristide is administered intravenously and travels through the blood stream to the tumor, where it is taken up. (c) The chlorotoxin moiety on tozuleristide binds to the extracellular membrane of glioma cells via annexin A2 receptors and is then internalized. (d) Excitation of the covalently bound ICG with a near-infrared (NIR) light source produced emission fluorescence that can be visualized with a sensitive NIR camera and used to guide tumor detection and removal.

The ability of tozuleristide to target brain tumors in vivo was evaluated using nude mice with an intracranial implanted glioblastoma cell line. Eighteen days after implantation, the mice received a single dose of tozuleristide. Two days after injection of tozuleristide into the tail veins, they were sacrificed and the brain was sectioned and mounted on glass slides. Fluorescence signal was determined using quantitative NIR scanning methods (Odyssey CLx, Licor). Histology confirmed the presence of tumor in tozuleristide fluorescent regions. No uptake was observed in surrounding brain tissue.29,36 Similar results were observed in studies on dogs³⁴ containing a variety of tumors, for which tozuleristide was administered prior to tumor excision and fluorescence uptake compared to tumor histology. Tozuleristide could also reliably detect squamous cell carcinoma in a hamster model of dysplasia and carcinogenesis, and also differentiate high-grade from low-grade dysplasia.³⁷ These studies reliably demonstrated tumor-specific uptake by tozuleristide as detected by quantitative NIR scanning. In situ imaging in these dogs was carried out using a prototype NIR detection camera (Denali—a prototype of the Solaris, Perkin-Elmer). This system demonstrated that in vivo detection during surgical resection was indeed feasible and that tozuleristide fluorescence signal could provide contrast between tumor tissue and surrounding normal tissues in many tumor types. Taken as a whole, the tumor selective uptake of tozuleristide provided the clear motivation to move tozuleristide into phase I clinical trials in humans.

14.6.1 Use of Tozuleristide in Human Subjects

A phase I dose escalation study in patients with suspected skin cancers (squamous cell carcinoma, basal cell carcinoma, and melanoma) was carried out to determine safety, tolerability, and pharmacokinetic (PK) distribution of escalating doses of tozuleristide (NCT02097875, Blaze Bioscience).38 In situ imaging was feasible as all suspected lesions were on the skin. In situ imaging on patients was performed with the Fluobeam 800 (Fluoptics, Grenoble, France) hand-held NIR imaging system. Doses of 1 to 18 mg of tozuleristide were administered to 21 patients. Neither dose-limiting toxicity nor severe adverse events were observed. PK data indicated peak levels of drug in serum within 30 minutes after injection, with a sharp fall off between hours 2 and 4 and almost complete elimination of serum drug within 12 to 24 hours. Tumor-specific uptake was documented by serial imaging at 2 to 24 hours after injection. At doses with adequate signal and contrast (3–12 mg), fluorescence correlated with the presence of tumor (based on subsequent pathology) approximately 90% of the time. These data suggested that tozuleristide was well tolerated up to 18 mg and adequate tumor fluorescence for imaging could be achieved. The subsequent experience with Fluobeam 800 in brain cancer (NCT02234297) pointed to a key need for clinical trials with tumor-specific FGS for glioma (as opposed to skin cancer). A surgeon-friendly, microscope-integrated yet highly sensitive imaging device to detect fluorescence in situ would be a critical feature required for clinical success. Subsequently, significant effort was directed toward developing such an imaging unit.

14.6.2 Development of Imaging System with Adequate Sensitivity for Tozuleristide Detection In Vivo

To successfully advance a molecule like tozuleristide through human trials for glioma tumors, a surgically relevant yet sufficiently sensitive fluorescence imaging unit is required. Several imaging systems have been developed for intraoperative detection of fluorescence.^{39,40,41,42,43,44,45,46,47} These include NIR imaging modes in several commercial surgical microscopes, such as IR800 incorporated into the Zeiss Pentero microscopes, or FL800 incorporated into Leica microscopes, and stand-alone devices such as the Fluobeam 800 (Fluoptics), Artemis (Quest), and SPY Elite (Novadaq) systems, as well as several endoscope-based systems. Each of these other systems was designed to primarily detect intravascular ICG, which has a typical concentration of 100 nM to 50 µM in blood during fluorescence angiography. Systems such as the Fluobeam 800 can detect low nanomolar concentrations of ICG, but cannot do so at real-time video rate, or overlay those images on a color visible light background. Finally, none of the systems with potentially adequately sensitive fluorescence detection capabilities are integrated into the surgical microscopes traditionally used for glioma surgery. The absence of such an imaging device motivated us to design and build such a device in anticipation of clinical trials of tozuleristide for glioma surgery.

To address the need for a practical way to perform FGS for glioma with tozuleristide and similar targeted fluorophores, we developed a prototype device that can detect very low concentrations of NIR fluorophores in tissue.^29,48 Our system has a small profile and allows the surgeon to use the operating microscope separate from the NIR imaging system. Traditional NIR systems use two separate sensors for visible and NIR channels. Although this allows for the use of high-sensitivity infrared camera, it also adds to the weight and size of the device. We designed a clinical prototype system (Synchronized InfraRed Imaging System [SIRIS]) that uses the same sensor for both visible and NIR channels.²⁹ We use a single high-definition (HD) charged-couple device camera for both channels. Both NIR and visible light are delivered to the surgical field from a custom-built light source that is programmed to deliver synchronized pulses of NIR and visible light. The NIR and visible light sources (Lumencor ASTRA light engine, Beaverton, OR) are alternatively pulsed and synchronized with frame capture rates of 300 frames per second. A fixed focal length lens (35 mm) is attached to a C-mount. The NIR light (785 nm) is pulsed from the laser (Coherent, Santa Clara, CA), while four LEDs (blue, cyan, green, and red) deliver cool balanced white light.³⁶ The pulse rate, fluency, timing, and width of the laser pulse are controlled via software. A 785-nm notch filter is attached to the front of the lens to filter out the excitation light from the return image. The 785-nm laser wavelength is near peak absorption spectrum for tozuleristide and ICG in Intralipid^49,50 and far away from the peak emission spectrum curves. The collected light is transmitted to a graphics processing unit (GPU) via a CameraLink cable for image processing. A GPU is required due to high data rate (718 Mb/s) and real-time requirements. The resultant HD quality visible light images are superimposed on pseudocolored fluorescent maps of tozuleristide distribution ( Fig. 14.2).

The system is draped during surgical procedure for sterile operating room environments. Our custom-built SIRIS system was found to detect tozuleristide fluorescence down to 50 pM at video frame rates (< 30 millisecond NIR exposures), enabling imaging of targeted fluorescent probes in the surgical setting. It was also found to provide 25 to 1,000 times greater NIR sensitivity than current commercially available systems for intraoperative NIR detection ( Fig. 14.3).

14.6.3 Clinical Results from Phase I Trials of Tozuleristide

Four phase I clinical trials with tozuleristide in humans have been initiated and as of February 2017 over 80 patients have received different doses of tozuleristide. Three trials (NCT02097875, NCT02234297, and NCT02496065) have been completed with one trial (NCT02462629 in pediatric central nervous system [CNS] tumor subjects) ongoing.

NCT02097875 was the first in human trial of tozuleristide in adult subjects with suspected skin cancers. Blaze Bioscience then initiated NCT02234297, a dose escalation study in adults with new or recurrent gliomas⁵¹ and a similar trial in pediatric CNS tumors (NCT02462629). Trial NCT02496065 initially examined tozuleristide in several solid tumor types, but ultimately focused on adult subjects with breast cancer. The endpoints of these dose escalation trials included safety, PK, and tumor fluorescence. Fluobeam 800 was initially used alone and then the SIRIS was incorporated into ongoing studies following institutional review board’s (IRB’s) approval. To date, over 80 patients have received tozuleristide in doses ranging from 3 to 30 mg, with no dose-limiting toxicity observed. SIRIS has been used for ex situ imaging in 52 patients and for in situ use in 30 patients. As the studies progressed, Fluobeam 800 was replaced by SIRIS at all trial sites. PK results from the adult glioma study paralleled the first-in-human data, with rapid fall off in serum levels a few hours after dosing. For the initial three dose levels, surgery was performed approximately 24 hours after injection. Subsequent dosing cohorts permitted surgery as early as 2 hours after injection.