Feasibility of Vascular Endothelial Growth Factor Imaging in Human Atherosclerotic Plaque Using 89 Zr-Bevacizumab Positron Emission Tomography

Abstract

Intraplaque angiogenesis is associated with the occurrence of atherosclerotic plaque rupture. Cardiovascular molecular imaging can be used for the detection of rupture-prone plaques. Imaging with radiolabeled bevacizumab, a monoclonal anti-vascular endothelial growth factor (VEGF)-A, can depict VEGF levels corresponding to the angiogenic status in tumors. We determined the feasibility of ⁸⁹Zr-bevacizumab imaging for the detection of VEGF in carotid endarterectomy (CEA) specimens. Five CEA specimens were coincubated with ⁸⁹Zr-bevacizumab and aspecific ¹¹¹In-labeled IgG to determine the specificity of bevacizumab accumulation. In 11 CEA specimens, ⁸⁹Zr-bevacizumab micro-positron emission tomography (PET) was performed following 2 hours of incubation. Specimens were cut in 4 mm wide segments and were stained for VEGF and CD68. In each segment, the mean percent incubation dose per gram of tissue (%Inc/g) and tissue to background ratio were determined. A 10-fold higher accumulation of ⁸⁹Zr-bevacizumab compared to ¹¹¹In-IgG uptake was demonstrated by gamma counting. The mean %Inc/g_{hot spot} was 2.2 ± 0.9 with a hot spot to background ratio of 3.6 ± 0.8. There was a significant correlation between the segmental tissue to background uptake ratio and the VEGF score (ρ = .74, p < .001). It is feasible to detect VEGF tissue concentration within CEA specimens using ⁸⁹Zr-bevacizumab PET. ⁸⁹Zr-bevacizumab accumulation in plaques is specific and correlates with immunohistochemistry scores.

CARDIOVASCULAR IMAGING is an invaluable tool for evaluating atherosclerosis and provides clinical information needed for assessment of plaque burden, decision making, and evaluation of therapeutic efforts. Conventional imaging modalities are able to detect anatomic abnormalities within the structure of the vessel wall and luminal stenosis. However, most cardiovascular events do not correlate with stenosis severity but with rupture in nonstenotic plaques.¹ Thus, the accurate detection of plaques that are prone to rupture will improve the identification of individuals with a higher risk of vascular complications. Molecular imaging can be applied for targeting and quantification of ongoing biologic processes in atherosclerotic plaques leading to plaque rupture.

Intraplaque release of vascular endothelial growth factor (VEGF) is known to be an important process linked to plaque vulnerability.² Both VEGF and local hypoxia result predominantly in chemotactic processes within plaques. This process leads to the formation of immature blood vessels and subsequent intraplaque hemorrhage, thereby causing plaque instability. Clinical detection of intraplaque concentration of VEGF may be used for the assessment of plaque vulnerability.

The role of molecular imaging techniques in the detection of angiogenesis has been widely investigated in animal and experimental human studies targeting different molecules such as VEGF,³ VEGF receptors,⁴ and integrins.⁵ Radiolabeling of the humanized monoclonal antibody bevacizumab for targeting of all VEGF-A isoforms can provide a nuclear imaging tracer and can be used to quantify VEGF levels in tumors,⁶ to show evidence of angiogenesis, and to monitor therapeutic efforts targeting VEGF.^7,8 Previously, we showed specific uptake of ⁸⁹Zr-bevacizumab in a high-VEGF-secreting human SKOV-3 ovarian tumor xenograft in mice.³ Angiogenesis imaging in patients with melanoma lesions demonstrated a strong correlation between semiquantitative ¹¹¹In-bevacizumab uptake and VEGF expression in tumor.⁶

In this study, we investigated whether ⁸⁹Zr-bevacizumab VEGF imaging is feasible to detect VEGF in human carotid endarterectomy (CEA) specimens. We applied a method for ex vivo nuclear imaging of human CEA specimens that was recently developed by our group and takes advantage of high-resolution small-animal positron emission tomography (PET) technology in the visualization of molecular and cellular processes within atherosclerotic plaques.⁹

Materials and Methods

Study Design and Specimens

This study was designed according to previous work by our group that demonstrated the feasibility of ex vivo molecular imaging of CEA specimens after incubation in a solution containing tracer followed by high-resolution microPET scan to visualize tracer uptake with great detail. The study was approved by the institutional ethics review board of the University Medical Center Groningen, Groningen, the Netherlands. Eleven CEA specimens were included from 11 patients who underwent CEA because of significant symptomatic carotid artery stenosis between June 2009 and August 2010. Risk factors and diagnostic data for individual patients are shown in Table 1. The samples contained the carotid bifurcation, the distal segment of common carotid artery, and proximal segments of both the internal and external carotid arteries. Excised nonatherosclerotic renal artery obtained from a patient undergoing a kidney transplantation procedure was used as a negative control.

Table 1.

Individual Patient Data, Risk Factor, and Diagnostic Data

Patient No.	Sex	Age (yr)	Symptom	Degree Stenosis	Dyslipidemia	Diabetes	Hypertension	Smoking
1	M	69	Amaurosis fugax	50–70	0	0	0	0
2	M	78	Amaurosis fugax	70–99	1	1	1	1
3	M	74	TIA	70–99	0	0	1	0
4	M	65	CVA	80–99	1	0	1	1
5	M	80	CVA	80–99	0	0	0	0
6	M	67	CVA	80–99	1	0	1	0
7	F	75	CVA	80–99	0	0	1	0
8	M	79	TIA	80–99	1	1	1	0
9	M	66	TIA	70–99	1	0	0	0
10	M	44	CVA	70–99	1	0	0	0
11	M	82	TIA	70–99	1	0	1	0

CVA = cerebrovascular accident; TIA = transient ischemic attack.

Dyslipidemia: 0 = cholesterol and triglyceride within normal range, 1 = controlled with treatment; diabetes: 0 = nondiabetic, 1 = diagnosed diabetes mellitus; hypertension: 0 = none, 1 = hypertension; smoking: 0 = nonsmoker, 1 = current smoker (includes abstinence < 1 year).

MicroPET Procedure

Bevacizumab (Avastin, 25 mg/mL; Roche, Mijdrecht, the Netherlands) conjugation and labeling with ⁸⁹Zr (half-life = 3.27 days), from IBA Molecular, was performed as described previously.¹⁰ For the preparation of incubation buffer, ⁸⁹Zr-bevacizumab (6.9 ± 4.0 MBq, 30 μg) was diluted in 10 mL phosphate-buffered saline (PBS). Immediately after excision, the specimens were transported under sterile conditions and subsequently incubated with ⁸⁹Zr-bevacizumab for 2 hours at room temperature. Before the microPET procedure, the plaques were flushed three to five times with PBS and fixed in a humid box to prevent dehydration. All specimens were fixed on a bed and scanned using a microPET focus 220 camera (Siemens Preclinical Solutions, Knoxville, TN) for 30 minutes. Thereafter, a micro-computed tomographic (CT) scan was performed using a microCAT II system (Siemens Preclinical Solutions) using the same fixed bed so that the stereotactic position was maintained. MicroPET images were corrected for scatter and reconstructed applying an iterative reconstruction algorithm (two-dimensional ordered subset expectation maximization [OSEM 2D]).

Determination of Specific Uptake

¹¹¹In (Covidien, Petten, the Netherlands) -labeled human IgG (Sanquin, Amsterdam, the Netherlands) was used as an aspecific control antibody as described previously.¹⁰ A mixed dose of ⁸⁹Zr-bevacizumab and ¹¹¹In-IgG (5 ± 0.5 MBq, 0.3 mL diluted in 10 mL PBS was prepared, and five control specimens were incubated. After 2 hours, specimens were washed with PBS and were counted in a calibrated well-type LKB-1282-Compu-gamma-system (LKB Wallac, Turku, Finland) using a dual-isotope counting program for both isotopes. The overlap of ⁸⁹Zr counts in ¹¹¹In channel was corrected. The activities of ⁸⁹Zr and ¹¹¹In were calculated as percent incubation dose per gram of tissue (%Inc/g).

Immunohistochemistry

After the scan, specimens were oriented on a flat surface and were serially cross-sectioned in 4 mm slices and numbered from proximal to distal. The sections were formalin fixed, paraffin embedded, and sectioned at 5 μm. The tissue slides were blocked with 0.3% H₂O₂, followed by blocking endogenous avidin and biotin (SP-2001, Vector Laboratories, Burlingame, CA), and were incubated with rabbit anti-VEGF antibody A-20 (diluted 1:50; Santa Cruz Biotechnology, Santa Cruz, CA) for 1 hour at room temperature. Subsequently, the slides were incubated with biotinylated goat antirabbit antibody (diluted 1:300; DAKO), followed by ABC Elite (diluted 1:100; DAKO). Antibody detection was performed with the diaminobenzidine (DAB) kit (DAKO). The sections were counterstained with hematoxylin for 2 minutes. In five sections, macrophage content was evaluated by CD68 immunostaining and antibody was detected with the DAB kit to discover colocalization between VEGF and macrophage presence. All sections were examined on a blind basis using a light microscope with the 40 × objective.

The level of VEGF staining was scored from 0 to 3, determined by semiquantitative combined assessment of the percentage of stained cells and the staining intensity as described before.¹¹ CD68 staining was categorized semi-quantitatively according to a highly used and reproducible method as described previously.⁹

Data Analysis

Because of the short time difference between incubation and scanning, radioactive decay was considered negligible for both ⁸⁹Zr and ¹¹¹In. MicroPET and microCT images were registered using AMIDE software (version 0.9.1, Stanford University), and a whole-specimen three-dimensional region of interest (ROI) was drawn manually in microCT images and was applied to microPET images. The mean %Inc/g in total specimen was calculated. The maximal uptake of ⁸⁹Zr-bevacizumab within the specimen was determined and the hot spot was defined using a three-dimensional isocontour ROI with a threshold of 70% maximal uptake. %Inc/g in hot spot was calculated (%Inc/g_{hot spot}). To obtain a background value for ⁸⁹Zr-bevacizumab uptake, %Inc/g was measured in a manually drawn ROI within a specimen adjacent to the hot spot. Segmental ROI were drawn manually in accordance with immunohistochemical slices to compare tracer accumulation with immunohistochemistry. In each segment, a segmental %Inc/g_mean was calculated and the tissue to background ratio was calculated by dividing segmental %Inc/g_mean by background tracer accumulation of the same specimen, as described above. Categorical measures of VEGF and CD68 staining were compared using the chi-square test.

Statistical Analysis

Quantitative results were shown as mean ± SD. Regression analysis was used to determine relationships between mean %Inc/g in specimen and specimen size. %Inc/g_mean and tissue to background ratio of ⁸⁹Zr-bevacizumab uptake in each segment were compared to semiquantitative measures of VEGF-A staining and the results were tested by use of Spearman correlation coefficients (p). Per specimen analysis was performed by calculating average VEGF-A staining scores, %Inc/g, and the tissue to background ratio of the stained segments of each specimen.

Results

Feasibility

The mean length of the specimens was 2.4 ± 0.7 cm. The specimens were transported from the operating room to the laboratory within 10 minutes after excision. In all CEA specimens, clear ⁸⁹Zr-bevacizumab uptake within the arterial wall was seen (Figure 1). ⁸⁹Zr-bevacizumab uptake in all plaques was heterogeneously distributed. The mean %Inc/g_{hot spot} was 2.2 ± 0.9, the mean %Inc/g in the specimen was 0.9 ± 0.4 (ranging from 0.44 to 1.99), and the mean %Inc/g in the background was 0.6 ± 0.3. The average hot spot to background ratio within the CEA specimens was 3.6 ± 0.8. No correlation was found between plaque size and ⁸⁹Zr-bevacizumab uptake (r = .41; p = .2). The mean %Inc/g in the negative control arterial tissue was 0.18%.

Figure 1.

A, MicroPET image of ⁸⁹Zr-bevacizumab uptake in a human carotid endarterectomy (CEA) specimen. B, MicroCT of a human CEA specimen. C, Fused microPET/CT. D, Transverse view of a microPET image shows high tracer accumulation within the arterial wall.

Specific Uptake

In five specimens, the specificity of ⁸⁹Zr-bevacizumab uptake was evaluated. For this purpose, coincubation was performed using both ⁸⁹Zr-bevacizumab and control ¹¹¹In-IgG. Quantitative measurements of both tracers by gamma counter showed that ⁸⁹Zr-bevaciumab uptake was 10-fold higher when compared to the control ¹¹¹In-IgG (1.9 ± 0.9 %Inc/g and 0.2 ± 0.1 %Inc/g, respectively; p = .01).

VEGF Staining, Tracer Uptake, and Macrophage Content

VEGF staining was performed in 35 slices of the CEA specimens (Figure 2). Segmental %Inc/g_mean and tissue to background ratio in each group of VEGF staining are shown in Table 2. In total stained segments, the segmental %Inc/g_mean was 1.05 ± 0.39 and the tissue to background ratio was 2.05 ± 0.6. An association between VEGF score and segmental %Inc/g_mean (ρ = .29, p < .05) and tissue to background ratio of ⁸⁹Zr-bevacizumab uptake (ρ = .74, p < .001) was determined by Spearman correlation analysis (Figure 3A). Per specimen analysis was performed in nine specimens. Average VEGF scores, %Inc/g, and the tissue to background ratio of each specimen are shown in Table 3. Spearman correlation analysis showed an excellent correlation between average tissue to background ratio and average VEGF score (ρ = .91, p < .01, Figure 3B); however, no correlation was found between average %Inc/g and average VEGF score in CEA specimens (data not shown).

Figure 2.

A, Coronal view of a carotid endartertectomy (CEA) specimen. Dashed lines show the levels of transverse views. B1, Immunohistochemistry of a slide with high bevacizumab uptake shows intense staining of vascular endothelial growth factor in CEA specimens. B2, Immunohistochemistry of a slide with low bevacizumab uptake shows weak staining.

Figure 3.

A, The correlation between vascular endothelial growth factor (VEGF) score with segmental percent incubation dose per gram of tissue (%Inc/g_mean) (black bars; ρ = .29, p < .05) and segmental tissue to background ratio (dotted bars; ρ = .74, p < .001). B, Per specimen correlation between average tissue to background ratio and VEGF scores in the stained specimens (Spearman rank correlation).

Table 2.

Mean Segmental ⁸⁹Zr-Bevacizumab Uptake Values in Segments with Different VEGF Staining Scores

	VEGF Score
	Negative Control	1	2	3
Number (total)	1	6	15	13
Tissue to background ratio	—	1.54	2.00	2.41
%Inc/gmean	0.18	0.85	1.03	1.21

%Inc/g = percent incubation dose per gram of tissue; VEGF = vascular endothelial growth factor.

VEGF score was determined according to semiquantitative combined assessment of the percentage of stained cells and the staining intensity as described in Kitamura and colleagues.¹¹

Table 3.

Average VEGF Score, %Inc/g, and Average Tissue to Background Ratio of VEGF-Stained Segments in Each Specimen

Plaque Number	Mean VEGF Score	Mean %Inc/g	Mean Tissue to Background Ratio
1	1.80	1.13	1.94
2	1.33	1.02	1.51
3	2.00	1.15	1.83
4	2.33	0.86	2.40
5	1.40	0.81	1.64
6	2.75	0.72	2.00
7	3.00	1.23	3.16
8	2.67	0.84	2.10
9	3.00	2.25	2.81

%Inc/g = percent incubation dose per gram of tissue; VEGF = vascular endothelial growth factor.

To evaluate the cellular basis of VEGF release, CD68 staining was performed in eight sections and showed a colocalization between subendothelial macrophage content and VEGF staining (Figure 4). The chi-square test showed that the sections did not differ in VEGF and CD68 staining scores (ρ = .45).

Figure 4.

A, Four times magnification of vascular endothelial growth factor (VEGF)-stained cells. B, Four times magnification of CD68 staining in the same piece showing intense subendothelial macrophage content with a pattern similar to that of VEGF staining. C, Ten times magnification of VEGF staining and, D, ten times magnification of CD68 staining in the same slides (black arrows). The colocalization of VEGF and CD68 can be clearly identified. This finding suggests a combined role of both biomarkers in the inflammatory process of the vulnerable plaque.

Discussion

In the present study, we showed for the first time the feasibility of ex vivo ⁸⁹Zr-bevacizumab VEGF imaging in human CEA specimens. Here we report the promising features of radiolabeled anti-VEGF PET in atherosclerotic plaques. Accumulation of the tracer was size independent and heterogeneously distributed within the plaque with a high signal to background ratio. The average mean %Inc/g in CEA specimens was five times higher than the mean %Inc/g in a normal artery. Moreover, a high specific uptake was demonstrated as we showed a 10-fold higher accumulation of ⁸⁹Zr-bevacizumab compared to control ¹¹¹In-IgG using dual-isotope gamma counting and size-independent uptake. Furthermore, we showed that the mean activity accumulation in the plaque correlates with the tissue VEGF abundance, determined by immunohistochemistry. Our finding suggests the possibility of designing further clinical studies to assess the role of a VEGF-targeting tracer to obtain PET images in stratifying the risk of patients with atherosclerotic disease.

[¹⁸F]Fluorodeoxyglucose PET imaging has been proposed for detecting macrophage content and vulnerability in atherosclerotic plaques.¹² However, the fact that [¹⁸F]FDG accumulates in any highly metabolizing cell results in suboptimal specificity in depicting vessel wall inflammation.¹³ Detection of VEGF activation, a patho-physiologically relevant indicator of macrophages and an independent determinant of plaque vulnerability, might offer the opportunity to track both vessel wall inflammation and angiogenesis in atherosclerosis. This study has shown a colocalization and similar intensity scores of VEGF and CD68 (a marker of macrophages) in CEA specimens; however, in accordance with previous studies, VEGF was also abundant in areas with numerous smooth muscle cells and within extracellular matrix.¹⁴ This result emphasizes the link between inflammation and angiogenesis in atherosclerotic plaques and raises the possibility that

VEGF-targeted imaging could provide indirect information on plaque inflammation correlated to its vulnerability in vivo. The predictive role of ¹⁸FDG PET or VEGF-targeted imaging with plaque vulnerability and rupture needs to be further investigated and compared.

Although it has been suggested by some studies that VEGF may also have a potential beneficial role in atherosclerotic disease through regeneration of endothelium and endothelial function improvement,¹⁵ a large body of evidence underlines the substantial role of VEGF in the maintenance and destabilization of atherosclerotic plaques rather than its protective effects.¹⁶ Thus, the ability to image VEGF within plaques by PET can offer a promising tool for evaluating vulnerability and monitoring therapeutic efforts aimed at stabilization of atherosclerotic plaques. In previous investigations, the value of radiolabeled bevacizumab imaging was shown in the detection of VEGF in VEGF-secreting tumors.^3,6–8

Adequate preservation of binding affinity of ⁸⁹Zr-labeled bevacizumab was previously shown by our group using a VEGF-coated enzyme-linked immunosorbent assay (ELISA).³ In vitro evaluation of binding affinity showed 54% ± 3.7% binding of ⁸⁹Zr-bevacizumab to VEGF-coated wells. Additions of 500-fold excess unlabeled bevacizumab for competition assay resulted in almost complete blocking of radiolabeled bevacizumab binding (< 5%, ranging from 1 to 4%) comparable to nonspecific binding of radiolabeled IgG (4–8%).

The specificity of radiolabeled bevacizumab uptake was previously shown in a study by our group on human tumor xenografts³ and indicated VEGF-mediated ⁸⁹Zr-bevacizumab uptake. To verify the specificity of ⁸⁹Zr-bevacizumab uptake in CEA specimens, we were not able to execute the frequently used approach of blocking specimens using an excess of the unlabeled bevacizumab as it has been shown that administration of unlabeled bevacizumab could cause functional changes in tissue and influence the pathobiologic properties of CEA specimens.¹⁷ In this study, instead, we coincubated the specimens in equal doses of ⁸⁹Zr-bevacizumab and aspecific ¹¹¹In-IgG as a control. Samples were counted for radioactivity by using a dual-isotope gamma-counting program in a calibrated well-type gamma counter, the same method used in a previous study by our group.^18,19 The results showed 10-fold higher accumulation of ⁸⁹Zr-bevacizumab than that of ¹¹¹In-IgG. Moreover, normal arterial tissue was incubated in ⁸⁹Zr-bevacizumab and showed five times less accumulation of bevacizumab, which further confirms the specificity of the tracer.

Another approach in visualization of the VEGF pathway and evaluation of angiogenesis is molecular imaging of VEGF receptor. In vivo imaging with radiolabeled VEGF₁₂₁ (⁶⁴Cu-DOTA-VEGF₁₂₁) has successfully demonstrated the ability to depict VEGF receptor in a mouse model of human glioblastoma tumor.⁴ Recently, an easy-to-label ⁶⁸Ga-labeled tracer, based on engineered single-chain VEGF, has been developed and showed promising characteristics for imaging VEGF receptor in angiogenic vasculature.²⁰ Given that the correlation between VEGF tissue concentration and VEGF receptor expression and the corresponding role of their levels in the progression of angiogenesis is not well determined, the two methods for imaging and quantification of VEGF and VEGF receptor can be complementary in evaluating angiogenesis in vivo.

Although mean segmental %Inc/g_mean and mean tissue to background ratio in different VEGF scores showed a trend to superior accumulation of ⁸⁹Zr-bevacizumab uptake in higher scores of VEGF (see Figure 3), there was a weak correlation between mean segmental %Inc/g_mean and VEGF immunohistochemistry scores. Per specimen analysis also showed that despite an excellent correlation between mean tissue to background ratio in the stained sections, there was no association between average %Inc/g and average VEGF scores in the specimens. This may partially be explained by semiquantitative measurements of immunohistochemistry results. Moreover, VEGF staining shows cytoplasmic and macrophage-bound VEGF, although it is not very sensitive in accounting for the detection of VEGF secreted into the extracellular matrix. On the contrary, ⁸⁹Zr-bevacizumab has a high affinity to all VEGF-A isoforms, including those secreted into the extracellular matrix. To overcome this difference in targeting VEGF, we compared semiquantitative immunohistochemistry results to semiquantitative microPET measures of tissue to background ratio. The results showed a good correlation between segmental tissue to background ratio and VEGF staining level (Spearman ρ = .74) and specimen average tissue to background ratio and specimen average VEGF staining score (Spearman ρ = .91). A better correlation between VEGF scores with relative measures of ⁸⁹Zr-bevacizumab accumulation (tissue to background ratio) rather than absolute measures of tracer accumulation could also be explained by the fact that variability in the tracer dose in incubation buffer (6.4 ± 4.0 MBq) can result in different accumulation of bevacizumab in normal tissue within the specimens. However, tissue to background ratio corrects the quantitative imaging data for possible variation in the pharmacokinetic behavior of ⁸⁹Zr-bevacizumab in normal tissue due to a wide range of tracer dose. In addition, the fact that segmental %Inc/g_mean was compared to VEGF immunohistochemistry score on a 5 μm section of a 4 mm wide segment, which cannot represent total VEGF level within the segment, is an acknowledged limitation of this study. Tissue extract ELISA can provide more precise information on VEGF levels in corresponding segments of microPET images in future studies.

It has been reported in tumor xenografts that due to the long half-life of bevacizumab in blood (17–21 days) and its slow pharmacokinetics, the optimal imaging time, which provides the highest target to blood ratio, is 4 to 7 days after tracer injection. In our study, we targeted VEGF ex vivo; therefore, a long incubation time was not necessary. Future studies should be designed to determine the optimal imaging time point after tracer injection that provides the highest target to blood ratio in humans. Furthermore, future application of fluorescent or radiolabeled bevacizumab and other VEGF-targeted tracers in coronary artery disease requires a high target to heart uptake ratio. In previous work by our group, we observed a specific uptake value of 3.43% ± 0.99% in mouse heart 24 hours after injection, which rapidly declined over time (2.31% ± 0.93% at 72 hours and 2.11% ± 0.66 % at 168 hours).³ In our study, the average %Inc/g_{hot spot} of ⁸⁹Zr-bevacizumab at 2 hours was 2.2 ± 0.9. We predict an increase in plaque uptake of ⁸⁹Zr-bevacizumab over time, which remains high at 72 to 168 hours, with the same pattern that we observed in tumors. However, this prediction needs to be tested explicitly.

Given that bevacizumab is a recombinant humanized monoclonal antibody and species specific, it is not possible to investigate the role of ⁸⁹Zr-bevacizumab in viable rodent models of atherosclerosis, namely apolipoprotein E-deficient and low-density lipoprotein receptor-deficient mice. However, in a recent study on immunodeficient mice transplanted with human coronary artery tissue, it was shown that bevacizumab treatment ameliorates vascular remodeling through VEGF blockade.²¹ ⁸⁹Zr-bevacizumab imaging in this animal model can provide a unique approach to investigate the feasibility of humanized monoclonal antibody imaging in atherosclerotic disease in vivo. The biodistribution of bevacizumab in mice was earlier investigated by our group and showed a low muscle accumulation of the tracer, which is a promising property for imaging carotid arteries and could offer the opportunity to depict lesions with a high target to muscle ratio.³ Additionally, in a clinical study by our group, accumulation of ⁸⁹Zr-bevacizumab was calculated in different organs in patients with renal cell carcinoma.²² The mean specific uptake value of ⁸⁹Zr-bevacizumab in human muscle tissue was 0.47 ± 0.15% injected dose per gram, which makes the tracer a promising probe for imaging carotid arteries in humans due to low tracer uptake in surrounding muscle tissue.

The current study provides evidence that bevacizumab imaging for the depiction of tissue abundance of VEGF in atherosclerotic plaques is feasible. To expand on observations of this study in human in vivo imaging, blood clearance of the tracer should be measured and dosimetry should be performed to provide further information on the potential of ⁸⁹Zr-bevacizumab application in a clinical setting. Moreover, due to the high radiation burden associated with ⁸⁹Zr-labeled probes, it seems reasonable to label VEGF-targeted probes with isotopes with a shorter half-life and lower radiation energy such as ¹⁸F to impose less radiation burden to patients undergoing imaging. Bevacizumab is not feasible to carry PET tracers with a shorter half-life because of its long biologic half-life. Future pilot studies on in vivo imaging of radiolabeled or fluorescent VEGF-targeted probes can provide further information on the potential of bevacizumab radionuclide imaging in cardiovascular diseases.

It could also be of interest to use radiolabeled ranibizumab, a monoclonal antibody fragment derivative of bevacizumab that has a serum half-life of 2 to 6 hours, for the detection of VEGF signaling in human plaques. The latter could be clinically more favorable as ⁸⁹Zr-bevacizumab takes at least 2 to 3 days before imaging and is associated with an unacceptable high radiation burden. In a recent study on mouse xenograft models of human cancer, it was shown that ⁸⁹Zr-ranibizumab-PET allows rapid visualization of VEGF with high affinity to all VEGF-A isoforms.²³ Rapid blood clearance and maximal tumor uptake within 24 hours make ranibizumab a promising tracer for future human studies. Theoretically, ¹⁸F-labeled ranibizumab would result in more optimal tracer kinetics and a large decrease in the radiation burden because the radiation burden associated with ⁸⁹Zr-labeled probes is too high.

In addition to the use of radiolabeled bevacizumab, the use of targeted fluorescent imaging could be of interest for imaging VEGF in atherosclerotic plaques. In a recent study, we demonstrated that IRDye 800CW–labeled bevacizumab showed a high specificity and sensitivity in detecting VEGF in tumor models in vivo.²⁴ The possibility to detect submillimeter lesions with an intraoperative near-infrared camera matches the structural characteristics of carotid atherosclerosis. Moreover, anatomic exteriority of carotid artery provides the opportunity to overcome the drawback of fluorescent imaging in lack of penetration.

Conclusion

This study illustrates the potential of VEGF as a target for visualization of pathobiologic processes in the development of atherosclerosis. Specifically, heterogeneous ⁸⁹Zr-bevacizumab uptake within plaques was observed. Noninvasive imaging of VEGF could provide clinicians with a new and perhaps better tool in the determination of individuals with a higher risk of cardiovascular events.

Footnotes

Acknowledgments

We would like to acknowledge Nynke Jager for her contribution to this work.

Financial disclosure of authors: Reza Golestani's research was funded by Siemens Medical Systems.

Financial disclosure of reviewers: None reported.

References

Naghavi

Libby

Falk

. From vulnerable plaque to vulnerable patient: a call for new definitions and risk assessment strategies: part I. Circulation 2003;108:1664–72, doi:10.1161/01.CIR.0000087480.94275.97.

Virmani

Kolodgie

Burke

. Atherosclerotic plaque progression and vulnerability to rupture: angiogenesis as a source of intraplaque hemorrhage. Arterioscler Thromb Vasc Biol 2005;25:2054–61, doi:10.1161/01.ATV.0000178991.71605.18.

Nagengast

de Vries

Hospers

. In vivo VEGF imaging with radiolabeled bevacizumab in a human ovarian tumor xenograft. J Nucl Med 2007;48:1313–9, doi:10.2967/jnumed.107.041301.

Cai

Chen

Mohamedali

. PET of vascular endothelial growth factor receptor expression. J Nucl Med 2006;47:2048–56.

Wei

Wadas

. (64)Cu-labeled CB-TE2A and diamsar-conjugated RGD peptide analogs for targeting angiogenesis: comparison of their biological activity. Nucl Med Biol 2009;36:277–85, doi:10.1016/j.nucmedbio.2008.12.008.

Nagengast

Hooge

van Straten

. VEGF-SPECT with (111)in-bevacizumab in stage III/IV melanoma patients. Eur J Cancer 2011;47:1595–602, doi:10.1016/j.ejca.2011.02.009.

Nagengast

de Korte

Oude Munnink

. 89Zr-bevacizumab PET of early antiangiogenic tumor response to treatment with HSP90 inhibitor NVP-AUY922. J Nucl Med 2010;51:761–7, doi:10.2967/jnumed.109.071043.

Desar

Stillebroer

Oosterwijk

. 111In-bevacizumab imaging of renal cell cancer and evaluation of neoadjuvant treatment with the vascular endothelial growth factor receptor inhibitor sorafenib. J Nucl Med 2010;51:1707–15, doi:10.2967/jnumed.110.078030.

Masteling

Zeebregts

Tio

. High-resolution imaging of human atherosclerotic carotid plaques with micro 18F-FDG PET scanning exploring plaque vulnerability. J Nucl Cardiol 2011;18:1066–75, doi:10.1007/s12350-011-9460-2.

10.

Verel

Visser

Boerman

. Long-lived positron emitters zirconium-89 and iodine-124 for scouting of therapeutic radioimmunoconjugates with PET. Cancer Biother Radiopharm 2003;18:655–61, doi:10.1089/108497803322287745.

11.

Kitamura

Hatano

Higashi

. Proliferative activity in hepatocellular carcinoma is closely correlated with glucose metabolism but not angiogenesis. J Hepatol 2011;55:846–57, doi:10.1016/j.jhep.2011.01.038.

12.

Rudd

Narula

Strauss

. Imaging atherosclerotic plaque inflammation by fluorodeoxyglucose with positron emission tomography: ready for prime time? J Am Coll Cardiol 2010;55:2527–35, doi:10.1016/j.jacc.2009.12.061.

13.

Kuehl

Eggebrecht

Boes

. Detection of inflammation in patients with acute aortic syndrome: comparison of FDG-PET/CT imaging and serological markers of inflammation. Heart 2008;94:1472–7, doi:10.1136/hrt.2007.127282.

14.

Ramos

Kuzuya

Esaki

. Induction of macrophage VEGF in response to oxidized LDL and VEGF accumulation in human atherosclerotic lesions. Arterioscler Thromb Vasc Biol 1998;18:1188–96, doi:10.1161/01.ATV.18.7.1188.

15.

Tio

Tan

Jessurun

. PET for evaluation of differential myocardial perfusion dynamics after VEGF gene therapy and laser therapy in end-stage coronary artery disease. J Nucl Med 2004;45:1437–43.

16.

Holm

Slart

Zeebregts

. Atherosclerotic plaque development and instability: a dual role for VEGF. Ann Med 2009;41:257–64, doi:10.1080/07853890802516507.

17.

Willett

Boucher

di Tomaso

. Direct evidence that the VEGF-specific antibody bevacizumab has antivascular effects in human rectal cancer. Nat Med 2004;10:145–7, doi:10.1038/nm988.

18.

Oude Munnink

Korte

Nagengast

. (89)Zr-trastuzumab PET visualises HER2 downregulation by the HSP90 inhibitor NVP-AUY922 in a human tumour xenograft. Eur J Cancer 2010;46:678–84, doi:10.1016/j.ejca.2009.12.009.

19.

Oude Munnink

Arjaans

Timmer-Bosscha

. PET with the 89Zr-labeled transforming growth factor-beta antibody fresolimumab in tumor models. J Nucl Med 2011;52:2001–8, doi:10.2967/jnumed.111.092809.

20.

Eder

Krivoshein

Backer

. ScVEGF-PEG-HBED-CC and scVEGF-PEG-NOTA conjugates: comparison of easy-to-label recombinant proteins for [68Ga]PET imaging of VEGF receptors in angiogenic vasculature. Nucl Med Biol 2010;37:405–12, doi:10.1016/j.nucmedbio.2010.02.001.

21.

Zhang

Silva

Yarovinsky

. VEGF blockade inhibits lymphocyte recruitment and ameliorates immune-mediated vascular remodeling. Circ Res 2010;107:408–17, doi:10.1161/CIRCRESAHA.109.210963.

22.

Oosting

Nagengast

Oude Munnink

, editors. 89Zr-bevacizumab PET imaging in renal cell carcinoma patients: feasibility of tumor VEGF quantification. Eur J Cancer Suppl 2010;7:72–3.

23.

Nagengast

Lub-de Hooge

Oosting

. VEGF-PET imaging is a noninvasive biomarker showing differential changes in the tumor during sunitinib treatment. Cancer Res 2011;71:143–53, doi:10.1158/0008-5472.CAN-10-1088.

24.

Terwisscha van Scheltinga

van Dam

Nagengast

. Intraoperative near-infrared fluorescence tumor imaging with vascular endothelial growth factor and human epidermal growth factor receptor 2 targeting antibodies. J Nucl Med 2011;52:1778–85, doi:10.2967/jnumed.111.092833.