Fiber-Optic System for Dual-Modality Imaging of Glucose Probes 18F-FDG and 6-NBDG in Atherosclerotic Plaques

Background Atherosclerosis is a progressive inflammatory condition that underlies coronary artery disease (CAD)–the leading cause of death in the United States. Thus, the ultimate goal of this research is to advance our understanding of human CAD by improving the characterization of metabolically active vulnerable plaques within the coronary arteries using a novel catheter-based imaging system. The aims of this study include (1) developing a novel fiber-optic imaging system with a scintillator to detect both 18F and fluorescent glucose probes, and (2) validating the system on ex vivo murine plaques. Methods A novel design implements a flexible fiber-optic catheter consisting of both a radio-luminescence and a fluorescence imaging system to detect radionuclide 18F-fluorodeoxyglucose (18F-FDG) and the fluorescent analog 6-(N-(7-Nitrobenz-2-oxa-1,3-diazol-4-yl)amino)-6-Deoxyglucose (6-NBDG), respectively. Murine macrophage-rich atherosclerotic carotid plaques were imaged ex vivo after intravenous delivery of 18F-FDG or 6-NBDG. Confirmatory optical imaging by IVIS-200 and autoradiography were also performed. Results Our fiber-optic imaging system successfully visualized both 18F-FDG and 6-NBDG probes in atherosclerotic plaques. For 18F-FDG, the ligated left carotid arteries (LCs) exhibited 4.9-fold higher radioluminescence signal intensity compared to the non-ligated right carotid arteries (RCs) (2.6×104±1.4×103 vs. 5.4×103±1.3×103 A.U., P = 0.008). Similarly, for 6-NBDG, the ligated LCs emitted 4.3-fold brighter fluorescent signals than the control RCs (1.6×102±2.7×101 vs. 3.8×101±5.9 A.U., P = 0.002). The higher uptake of both 18F-FDG and 6-NBDG in ligated LCs were confirmed with the IVIS-200 system. Autoradiography further verified the higher uptake of 18F-FDG by the LCs. Conclusions This novel fiber-optic imaging system was sensitive to both radionuclide and fluorescent glucose probes taken up by murine atherosclerotic plaques. In addition, 6-NBDG is a promising novel fluorescent probe for detecting macrophage-rich atherosclerotic plaques.


Introduction
According to the World Health Organization, approximately one third of all deaths in developed countries are related to coronary artery disease (CAD), characterized by the buildup and disruption of atherosclerotic plaques within the coronary wall [1]. Atherosclerosis is a progressive inflammatory condition that underlies CAD-the leading cause of death in the United States [2]. A subset of plaques is thought to be unstable and prone to rupture (''vulnerable''), which can cause life-threatening acute coronary syndromes. The most common type of vulnerable plaque is an inflamed thin-cap fibroatheroma (TCFA), which is thought to account for 60% to 70% of coronary events [3]. If we are able to identify these vulnerable plaques, we may be able to guide treatment to reduce the risk of heart attack and the debilitating aftermath, reducing the burden on individuals as well as society.
The current clinical paradigm for detecting CAD is angiography, which only evaluates the luminal encroachment of the disease, without providing information about plaque extent and content [4]. There is a wide range of molecular imaging modalities employed for detection and characterization of atherosclerosis, primarily in large vessels. Examples include positron emission tomography (PET) with 18 F-FDG [5], magnetic resonance imaging (MRI) using ultra-small super-paramagnetic iron oxide (USPIO) probes [6,7], single photon emission computed tomography (SPECT) using VCAM-1-specific 99m Tc-labeled peptidic sequences [8], and the iodinated nanoparticulate contrast agent N1177 for computed tomography (CT) [9]. However, application of these external imaging approaches to coronary atherosclerosis has been challenging due to small coronary size, cardiac motion, and detecting adequate signal and contrast. Recently, an intravascular near infrared fluorescence (NIRF) approach employing a protease-activatable NIRF agent has been applied to image plaque inflammation in the rabbit aorta [10].
Several studies have shown that 18 F-FDG can be a marker of metabolically active (''vulnerable'') plaques due to its uptake by inflammatory macrophages in the carotids and aorta [11][12][13]. 18 F-FDG has the major advantage of being FDA approved for cardiac and cancer imaging; thus, clinical transition may be more easily achieved. However, PET 18 F-FDG detection in coronary plaque is still challenging due to the small size of such plaques, signal blurring due to motion, and obscuring FDG uptake by adjacent myocardium [14]. An intravascular molecular imaging approach [15] has the potential to overcome these limitations.
As fluorescent probes may provide higher intravascular signal [16] than radionuclides without the need for ionizing radiation, they also merit study. Thus, to complement the study of 18 F-FDG, we have selected 6-(N-(7-Nitrobenz-2-oxa-1,3-diazol-4-yl)amino)-6-Deoxyglucose (6-NBDG), a fluorescent glucose analog. It has higher signal than 2-NBDG [17] and does not degrade to nonfluorescent products due to its inability to be phosphorylated by hexokinase [18]. To the best of our knowledge, 6-NBDG has not been studied for detecting inflammation in atherosclerosis. We developed the dual-imaging system as the ideal method to compare radionuclide and fluorescent agents in vivo.
Macrophage infiltration in atherosclerotic plaques plays a vital role in the progression of atherosclerosis [19]. As a result, macrophages have become widely recognized as a key target for atherosclerosis imaging, since they contribute significantly to the progression of atherosclerosis [20]. Thus, in this feasibility study we implement a novel dual-modality fiber-optic system to detect atherosclerotic plaque by targeting macrophages in inflamed metabolically active plaques using 18 F-FDG and 6-NBDG. Therefore, the aims of this study were to (1) develop this fiberoptic imaging system with a scintillator to detect 18 F and fluorescent glucose probes, and (2) validate the system on ex vivo macrophage-rich murine plaques with confirmatory external optical imaging and autoradiography.

Metabolic imaging with glucose probes
Two glucose analogs were studied-18 F-FDG and 6-NBDG. 18 F-FDG is a radionuclide consisting of the positron-emitting radioactive isotope fluorine-18 instead of hydroxyl group (OH) at the 29 position of the glucose molecule (MW 181.15 g/mol). The 29 hydroxyl group (OH) in normal glucose is needed for further glycolysis; thus, 18 F-FDG cannot be further metabolized and is retained in macrophages in atherosclerotic plaques [11].
The 6-NBDG molecule consists of the small NBD fluorochrome, a nitrobenzoxydiazoamino group (MW 342.26 g/mol) attached to D-glucose at the C-6 position and fluoresces at 540 nm when excited at 465 nm (Invitrogen, Life Technologies, Grand Island, NY USA). This fluorescent probe has been shown to be transported into the cell by the same glucose transporters (GLUT) as glucose, but does not undergo phosphorylation by hexokinase (unlike 18 F-FDG) due to the NBD fluorochrome and never enters the glycolytic pathway, a reaction that causes 6-NBDG to accumulate within the cell [18,21].

System setup
2.2.1. Catheter radionuclide imaging (CRI) system. The CRI system (Figure 1) consists of a wide-angle lens (4.9 mm63.6 mm), providing 180 degree view of an artery, with a focal length of 3.85 mm (Dowell Electronic Technology Co. Ltd, Minzhi Longhua Shenzhen, Guangdong, China), attached to the distal ferrule (diameter 2.19 mm) of a leached image fiber bundle (Schott North America, Inc. Southbridge, MA USA). The proximal end (diameter 2.46 mm) of the fiber bundle is connected to an 8 mm megapixel fixed-focal-length lens and the other end of this lens is attached to a filter wheel. The leached image bundle runs parallel to the excitation fiber of the CFI and is enclosed within a 10-French catheter. A 35 mm lens connects the filter wheel to a CMOS camera (NEO sCMOS, Andor Technology, CT, USA), and a computer.
The novelty of the CRI system is the use of an inorganic scintillating screen [22] (10610 mm with 0.5 mm thickness, MTI Corporation. Richmond, CA, USA) of Lutetium Oxyorthosilicate (LSO), placed in the front of the wide angle lens to convert the bparticles of the positron emission signal into visible light due to radioactive decay. The efficiency of LSO is 24,000 photons/MeV and emits primarily in green with an emission peak at 420 nm. This light was then captured with the camera using deep thermoelectric cooling at 240uC for minimizing the background signal from the temperature-dependent dark current, hot pixel blemishes, and vibration. The exposure time is set to 10 seconds, which enabled sufficient signal from the ex vivo plaques. Both the CRI and CFI system were placed in a light-tight black box to prevent ambient light.
2.2.2. Catheter fluorescence imaging (CFI) system. The CFI system ( Figure 1) requires several additional components compared to the CRI system. These include a white LS-1 tungsten halogen light source (Ocean Optics, Dunedin, FL, USA), then a 10 dichroic blue color low pass filter with transmission between 390 and 480 nm with cutoff at 505615 nm wavelength (FD1B Thorlabs, Inc. New Jersey, USA) is placed in the built-in slot of the LS-1 to transmit only blue light through the excitation fiber (core diameter 1 mm) and deliver it to the sample at a 45 degree angle. The resulting photo-stimulated green light emitted at 540 nm wavelength from the 6-NBDG-enriched sample is collected with the wide-angle lens and a bandpass emission filter BP540/50 mm (filter BP 540 nm610 nm OD4 50 mm, Edmond Optics, USA) placed inside the filter wheel. Then, the emission light is captured by the shared wide-angle lens and the CMOS camera cooled at 240uC. The exposure time for the fluorescence imaging was also set to 10 seconds, as with the CRI system.

External optical imaging system (IVIS-200).
The IVIS-200 system (Xenogen Corporation/Caliper Life Sciences, Alameda, CA, USA) provides for bioluminescence and fluorescence optical-based imaging and was used to validate the CRI/ CFI system. Samples were placed inside the light-tight imaging chamber and the field-of-view set to 1.5 cm for both 6-NBDG and 18 F-FDG samples, with the highly sensitive charge-coupled device (CCD) camera system cooled to 290uC. For both 6-NBDG and 18 F-FDG imaging the binning factor was set at medium (868 pixels). For 6-NBDG fluorescence imaging, Green Fluorescent Protein (GFP) (excitation: 445-490 nm; emission: 515-575 nm) and GFP-background (410-440 nm) filters were used to obtain the correct excitation and emission signal from the 6-NBDG-enriched samples. The fluorescence images were collected over 0.5-10 seconds and 0.5-30 seconds for in vitro and ex vivo imaging, respectively. During 18 F-FDG radionuclide imaging, a scintillator screen was placed on top of the tissue sample and the bioluminescence mode was used to image the radioluminescence Fiber-Optic System for Imaging Atherosclerotic Plaques PLOS ONE | www.plosone.org from the b-particles of the positron emission signal from 18 F-FDG radioactive decay.
2.2.4. Autoradiography. Autoradiography was used to directly image the radiation emitted from the 18 F-FDG tissue samples with high spatial resolution [23] as further validation. Samples were placed in contact with a super resolution storagephosphor screen (12.5625.2 cm, PerkinElmer Inc., MA, USA). Over a sufficient exposure time (48 hours), emitted beta particles were recorded as an image on the storage-phosphor screen. The film was then read out by a Cyclone reader using a spatial resolution of 600 dpi (PerkinElmer Inc., Waltham, MA, USA). Using these techniques, the spatial distribution of radioisotope uptake of the tissue sample was recorded [24].

Experimental procedures
In vitro validation of 6-NBDG macrophage uptake. To establish 6-NBDG fluorophore as a potential molecular marker for detecting macrophage-rich atherosclerotic plaque, we studied macrophage uptake in vitro prior to in vivo studies. See Methods S1 and Results S1 for supplementary information on 6-NBDG uptake by macrophages.
2.3.2. Ex vivo imaging of 18 F-FDG/6-NBDG uptake. Ex vivo experiments were conducted according to protocol approved by the Administrative Panel on Laboratory Animal Care (APLAC permit number 9941) of Stanford University. All surgery was performed under anesthesia (2% Isoflurane), and all efforts were made to minimize suffering. Macrophage-rich atherosclerotic lesions were created in the left carotid arteries (LCs) of diabetic, hyperlipidemic FVB/NJ mice, as studied extensively by our laboratory [25][26][27][28]. In brief, 8-week-old male mice (n = 6) were fed a high-fat diet containing 40% kcal fat, 1.25% (by weight) cholesterol and 0.5% (by weight) sodium cholate (D12109, Research Diets, Inc., New Brunswick, NJ, USA). After 1 month on the diet, diabetes was induced by 5 daily intra-peritoneal injections of streptozotocin (STZ, 40 mg/kg, Sigma-Aldrich). Two weeks after the initiation of diabetes, the left common carotid artery was ligated below the bifurcation with the use of 5-0 silk ligature (Ethicon) while the mice were anesthetized with 2% Isoflurane. The non-ligated right carotid serves as an internal negative control. Two weeks after carotid ligation, mice were injected intravenously (IV) with either 18 F-FDG (0.964-1.437 mCi) or 6-NBDG (1 mg/mL concentration of 500 mM, 177 mL in volume) after 6 hours of fasting (n = 3/group). One hour after injection, ligated LCs and non-ligated right carotid arteries (RCs) were harvested and cut open longitudinally and placed individually on black paper for ex vivo imaging, also with the heart sectioned as a positive control. Samples from mice injected with 18 F-FDG or 6-NBDG were imaged ex vivo with the fiber-optic imaging system. The scintillating screen was placed in contact with the 18 F-FDG-enriched tissue sample to capture all light converted from the b-particles of the emission signal of 18 F-FDG decay. The best image quality can be achieved when the scintillator is in contact with the tissue or #100 mm air gap, so energy can be deposited in the scintillator by all the b-particles. For 6-NBDG, imaging was performed with the scintillating screen removed. The exposure time was chosen to be 10 seconds based on prior experiments to achieve adequate signal intensity. To validate these findings, LCs and RCs, plus the heart, were also imaged using the IVIS-200 system (see 2.2.3 for system setup). For radionuclide imaging with both the fiber-optic and IVIS-200 systems, each sample was placed under the scintillator screen to convert b-particles of the emission signal into visible light due to radionuclide decay of 18 F-FDG. For the fiber-optic system, each sample (LC, RC, and heart) was imaged separately due to the small field-of-view (5 mm).
Both radioluminescence and fluorescence signals from the LCs and RCs were calculated based on the selected ROIs (same size for all samples). An average of mean signal intensity was then calculated in arbitrary units (A.U.) based on the CRI and CFI images. An average radiance was calculated based on images of 18 F-FDG enriched tissue by correcting for field flatness specific to the camera used in the IVIS-200 imaging system (Figure 2). For the fluorescence study, an average of mean radiant efficiency was calculated after a CF was applied to the IVIS-200 images of 6-NBDG enriched tissue.

Autoradiography of ex vivo tissue with 18 F-
FDG. LCs and RCs were placed side by side in contact with a storage-phosphor screen over 48 hours, with the heart placed at a distance to avoid overlap of the radioisotope decay due to the high emission from heart tissue.

Statistical analysis
A pairwise two-sample Student's t-test was performed to compare ex vivo signal intensity from ligated LCs and non-ligated RCs for both the 18 F-FDG and 6-NBDG experiments. The underlying distribution was found to be normally distributed according to QQ-plots. As all animals had similar weight and dose, these factors were not considered in the statistical analysis. Ex vivo analyses were performed using MATLAB software.
We used P,0.05 for statistical significance for all in vitro and ex vivo analyses. confirmed the high signal intensity in the LCs compared to RCs (6.18610 6 63.25610 5 vs. 9.88610 5 62.25610 5 , p/sec/cm2/sr, p = 0.005) and in the heart (8.75610 6 64.25610 5 ). For further confirmation, autoradiography was performed on the ex vivo carotids (Figure 4), with high 18 F-FDG signal observed in the LCs vs. RCs. All mice were imaged at the same day to avoid system variability, so the dose of 18 F-FDG was adjusted (0.964-1.437 mCi) for each mouse to account for decay over sequential IV injections of 18 F-FDG. This sequential dosing of 18 F-FDG was performed as absorbed dose in organ depends on the radionuclide, the amount of activity, geo-metrical distributions of the radionuclide and on the residence time in the tissue [29,30]. Thus, the mouse imaged last received the largest 18 F-FDG dose so the radioluminescence signal would be comparable with the first animal.

6-NBDG imaging
Fluorescence images of ex vivo plaques with the CFI system after 6-NBDG injection exhibited 4.3-fold greater fluorescence intensity (Figures 5a-c) from the ligated LCs than the non-ligated RCs (1.6610 2 62.7610 1 vs. 3.8610 1 65.9 A.U., P = 0.002; Figure 5e) with high signal from the heart. The LC is clearly visible with our system (Figure 5a) unlike the RC, which failed to emit sufficient fluorescence to overcome the sensitivity threshold of the CMOS camera. Thus, the RC provided an image (Figure 5b) that was similar to background (i.e., light source completely turned  Fiber-Optic System for Imaging Atherosclerotic Plaques PLOS ONE | www.plosone.org off). Our CFI system was also able to detect structural details of the LCs, such as location of the ligation (Figure 5a), which was not detectable through the IVIS-200 system. The IVIS-200 imaging system did confirm high signals from the LCs and heart, before and after CF applied (Figures 5d, 5d-1), which were 2.5-3.66 higher compared to the RCs (P = 0.004 for LC vs. RC, Figure 5f).

Discussion
We have shown that our novel fiber-optic imaging system was sensitive enough to detect murine plaque uptake, ex vivo, of both 18 F-FDG and 6-NBDG glucose probes. The clinically available 18 F-FDG showed almost 5-fold higher signal intensity from murine carotid plaques compared to control, with 6-NBDG showing greater than 4-fold increase. Importantly, this is also the first demonstration of 6-NDBG as a glucose probe for atherosclerotic plaques.
Intravascular molecular imaging approaches utilizing optical imaging have demonstrated usefulness for biological imaging of plaques in coronary-sized vessels [31]. Based on this feasibility study we demonstrated that our dual-modality CRI-CFI prototype fiber-optic system was sufficiently sensitive towards the detection of 18 F-FDG and 6-NBDG in atherosclerotic plaque with good resolution. Thus, our novel system in conjunction with radionuclide and fluorescence molecular probes may lead to intravascular imaging platforms for coronary arterial molecular imaging.
Coronary imaging remains challenging with PET or other currently available imaging modalities due to small size, constant motion, and in the case of 18 F-FDG, obscuring uptake by adjacent myocardium [14]. Although 18 F-FDG uptake is a marker for viable myocardium, this myocardial uptake makes it harder to image the adjacent small-sized coronaries [32], though there are some promising initial results [33]. This makes an intravascular imaging approach advantageous, minimizing the distance and maximizing the sensitivity to coronary plaque signal. While PET 18 F-FDG has been well studied, there are other PET agents being investigated, such as 18 F-labeld mannose [34], which could also be studied by our fiber-optic system.
Given the wealth of experimental and clinical data with 18 F-FDG, identifying a fluorescent analog for in vivo intravascular imaging would be valuable. Intravascular fluorescence imaging [31,35,36] can use molecular and cellular fluorescent probes that are sensitive to metabolic processes of interest, such as inflammation, to obtain information about the underlying biology. 6-NBDG appeared to be the most promising fluorescent glucose marker based on prior studies in other disease models [17,21,37]. Our positive results with 6-NBDG, both in macrophage cell culture and murine plaques, show that further studies of this fluorescent glucose probe for atherosclerosis are warranted. This may complement other studies of fluorescent-based probes of enzymatic activity [15,38].
The radionuclide imaging with the CRI system was designed to capture only b-particles. Although, we did not have any built-in cparticle detector in our system, we performed experiments to confirm that no c-particles interacted with the radioluminescence signal from the sample. Before imaging the LCs or RCs with the CRI system, we placed a slice of 18 F-FDG enriched heart in close proximity (0.5 cm) and observed no background signal (Figure 3b). Even with the scintillator in such close proximity to cparticles from the heart, we were still able to image the plaque location in the LCs without any background noise.
LSO (75% or 28,000 photons/MeV) was chosen as a scintillating material over BGO (15% or 8,200 photons/MeV) or plastic scintillator for its high relative light output or efficiency. Energy resolution that influences scatter rejection is also low for LSO (15-20%) compared to BGO (25%) and faster scintillating decay time (LSO: 40 nsec compared to BGO: 300 nsec). This allows fast coincidence timing with efficient rejection of random events to provide very high count rates [22]. As any LSO scintillating screen with a thickness as small as 1 mm provides an intrinsic efficiency close to 10% due to annihilation photons, we  Fiber-Optic System for Imaging Atherosclerotic Plaques PLOS ONE | www.plosone.org chose to use a scintillating screen with a thickness of 0.5 mm. Due to this extremely thin screen, the CRI system eliminates a significant amount of annihilation photons, which would have come from distant tissues and hinder the imaging of the local 18 F-FDG uptake in the plaque. As this was a study to investigate the feasibility of the CRI system, we used a higher radiation dose than typically used. Future experiments will be needed to assess lower doses.
While we studied both 18 F-FDG and 6-NBDG separately after intravenous delivery in an ex vivo atherosclerosis model, this study was limited as we did not perform in vivo intravascular imaging. Our system was built for dual-modality imaging to be advantageous for in vivo imaging, given that atherosclerosis can vary in extent and in location from subject to subject. The two modalities may also complement each other, for example in regards to signal depth and spatial localization. In the current study, 6-NBDG imaging appeared qualitatively to show more spatial detail, but this requires further study. In vivo animal studies comparing the two modalities will then inform whether a dual-modality system, or the superior single modality only, should be translated to future clinical imaging. The current manuscript demonstrates the feasibility of each component ex vivo, with further miniaturization/optimization needed to compare the modalities in vivo.
To miniaturize the system, an optimized wide-angle lens will be investigated, as well as further improvements in resolution/ sensitivity (e.g., higher number of optical imaging fibers, lower temperature CCD camera) and flexible scintillating screen. The leached image bundle of the CRI needs to be modified to accommodate the delivery of excitation light for 6-NBDG, turning off the fluorescence light source during the radionuclide imaging.
While our study used a small number of animals, one of the advantages of this animal model is that the non-ligated contralateral carotid serves as a negative control, plus heart tissue serves as a positive control, so we imaged 3 specimens per animal. With this sample size, we were able to demonstrate statistical significance between ligated and non-ligated carotid signal for both 18 F-FDG and 6-NBDG. A larger sample size will be needed for future in vivo studies comparing radionuclide and fluorescent agents.

Conclusion
A dual-modality fiber-optic molecular imaging approach successfully detected both radionuclide ( 18 F-FDG) and fluorescent (6-NBDG) glucose probes in murine atherosclerosis. This may provide new opportunities for investigating atherosclerosis and identifying high-risk coronary plaques. Figure S1 Fluorescence imaging of 6-NBDG uptake by macrophages. (a) In vitro fluorescence imaging of RAW264.7 macrophage cells (10 5 cells per well) with various concentrations of 6-NBDG fluorophore. (b) Significant positive correlation between the average radiant efficiency of all samples (n = 6) with exposure time 0.5-10 seconds (any exposure over 10 seconds saturated the in vitro fluorescence image taken with IVIS-200) and 6-NBDG concentrations. A quadratic relationship between signal and concentration was found to be the best fit and highly significant (p,0.0001 for all 3 exposure times).