The Development of [¹²⁴I]Iodinated-VG76e: A Novel Tracer for Imaging Vascular Endothelial Growth Factor in Vivo Using Positron Emission Tomography¹

Solid tumors are unable to develop beyond 2–3 cubic millimeters in volume without establishing an adequate blood supply. This state of tumor dormancy arises because the cells are only able to grow within the diffusion boundaries of oxygen and other nutrients (1). Angiogenesis, or the development of new blood vessels from preexisting vasculature, is therefore essential for continued tumor growth. This process also promotes the progression of the primary lesion by allowing invasion of the cancer cells into the circulation and the growth of metastatic lesions at distant sites in the body (2).

Angiogenesis is a highly regulated process involving many positive and negative regulators. The degree of angiogenesis in the tumor depends on the relative balance of these factors. Various cells within the solid lesion, including tumor cells, macrophages, and fibroblasts, have the ability to produce angiogenic factors (2, 3). VEGF⁴ is a central cytokine in this process and mediates the development of new vasculature during embryogenesis, neovascularization, and tumorigenesis. There are five isoforms resulting from alternative mRNA splice sequences. VEGF₁₄₅ and VEGF₁₂₁ are freely diffusible, whereas VEGF₂₀₆ and VEGF₁₈₉ remain cell associated (4). The major isoform found in tumors, however, is VEGF₁₆₅. Although this isoform is also diffusible, it carries a heparin-binding domain that can result in cell association (5). Vascular endothelial cells express two receptors (VEGFR-1 or flt-1, VEFGR-2 or KDR/flk-1), and lymphatic endothelial cells express an additional receptor (VEGFR-3) to which the various isoforms of VEGF can bind (4). Binding activates a transduction pathway that not only stimulates endothelial cell proliferation but also stimulates an increase in vascular permeability and the extravasation of proteins necessary to allow the development of a fibrin matrix that facilitates the invasion of stromal cells into the tumor tissue (2).

Blocking tumor angiogenesis has the potential to provide a universal approach to prevent tumor establishment and metastasis. There are currently a number of promising strategies in both preclinical development and early-phase clinical trials (1). Early results suggest that antiangiogenesis agents in combination with existing chemo- and radiotherapeutic regimes may prove to be highly efficacious (6, 7). A method of noninvasively measuring VEGF in humans could provide useful prognostic information (8), allow the selection of patients most likely to benefit from antiangiogenesis agents, and facilitate efficacy monitoring of antiangiogenesis strategies clinically. Nearly all clinical studies on the expression and prognostic importance of VEGF have been carried out on resected primary tumors, and little is known about the variability between metastases or metastatic sites, the main target for such therapy. It could also provide a much-needed, quantitative, noninvasive in vivo method of evaluating VEGF levels preclinically.

We have developed a novel radiotracer for use with PET imaging that enables noninvasive assessment of the levels of VEGF in vivo. The technique relies on the use of a positron-emitting radionuclide with relatively medium half-life (¹²⁴I = 4.2 days). In this report, we show that a monoclonal antibody (VG76e) that recognizes the 121, 165, and 189 isoforms of human VEGF can be labeled with either iodine-125 (a gamma-emitting radionuclide) or iodine-124 (a positron-emitting radionuclide). We also show, by using two human xenograft tumor models established on the basis of VEGF production, that the iodinated antibody can be used to detect and image differential expression of VEGF in vivo.

All chemicals were of analytical or high-performance liquid chromatography grade and, unless noted otherwise, were purchased from Sigma (Dorset, United Kingdom), Fisher Scientific (Leicestershire, United Kingdom), or Merck (Dorset, United Kingdom). [¹²⁴I]Iodine was produced by the ¹²⁴Te(p,n)¹²⁴I reaction using irradiation of a [¹²⁴I]tellurium(IV) oxide target (98% enrichment; diameter, 10 mm; depth, 0.5–1.0 mm) with a proton beam (9.8–12.3 MeV; 10 μA) from a Scanditronix MC40 Mk II cyclotron. [¹²⁴I]Iodine was recovered from the target by dry distillation using a quartz apparatus (9). The [¹²⁴I]iodine was trapped in a NaOH solution (300 μ1, 8 mm), and the volume was reduced using a stream of nitrogen. [¹²⁵I]NaI (specific radioactivity, >2430 Ci/mol) and [¹²⁵I]iodinated N-succinimidyl 3-(4-hydroxy-5-iodophenyl) propionate ([¹²⁵I]-SHPP; specific radioactivity, ∼2000 Ci/mol) were obtained from Amersham Pharmacia Biotech (Buckinghamshire, United Kingdom).

An IgG1-type mouse monoclonal anti-VEGF, VG76e, was raised by immunization with recombinant human VEGF₁₈₉. The production, purification, and characterization of this antibody are described elsewhere (10). Briefly, the antibody was produced from hybridomas grown in culture, and a highly purified preparation was obtained using protein A-Sepharose columns. Hybridomas were screened by Western blotting of VEGF₁₈₉. Using recombinant proteins, it was demonstrated that VG76e recognized the 121, 165, and 189 isoforms of VEGF. The antibody was iodinated with either [¹²⁴I]iodine or [¹²⁵I]iodine. These two radioisotopes can be used interchangeably without altering the biological activity of an iodinated protein. VG76e was iodinated with [¹²⁴I]iodine for PET imaging studies. For all other studies, VG76e was iodinated with [¹²⁵I]iodine.

Three iodination strategies were evaluated. VG76e was labeled directly with [¹²⁵I]iodine using the IodoGen method (Ref. 11; Fig. 1a) or labeled indirectly using either the Bolton-Hunter method (Ref. 12; Fig. 1,b) or the m-N-succinimidyl [¹²⁵I]iodobenzoic acid (m-[¹²⁵I]-SIB) approach of Zalutsky and colleagues (Refs. 13, 14; Fig. 1 c). The IodoGen and m-[¹²⁵I]-SIB methods of protein iodination are well established and were performed according to the procedures outlined in the cited references. For indirect iodination using the Bolton-Hunter method, VG76e (10 μl, 72 μg, in PBS) was mixed with sodium borate buffer (5 μl, pH 8.5, 100 mm) and incubated with [¹²⁵I]-SHPP at 0°C for 1 h. The labeled antibody was purified by gel filtration (PD-10 column; Amersham Pharmacia Biotech; mobile phase = PBS), and radiochemical purity was determined using Instant Thin Layer Chromatography (Gelman, Ann Arbor, MI; mobile phase = methanol/H₂O at 4:1 v/v).

An adaptation of the Bolton-Hunter method was used to label the antibody with [¹²⁴I]iodine. The preparation of [¹²⁴I]-SHPP was carried out in a conical reaction vessel (0.3 ml; Supelco, Dorset, United Kingdom). A solution of NaI carrier (5 μl, 89.2 μm) was added to [¹²⁴I]iodine (14 μl, 924 μCi). HCl (14 μl, 100 mm), sodium phosphate buffer (10 μl, 250 mm, pH 7.4), and SHPP (1,4-dioxane, 2 μl, 3.8 mm) were added with mixing. After adding chloramine-T (10 μl, 22 mm) dissolved in sodium phosphate buffer (250 mm, pH 7.4), the reaction was terminated immediately with a sodium phosphate buffered solution (250 mm, pH 7.4) of Na₂S₂O₅ (10 μl, 63 mm). The reaction product was quickly extracted with benzene/dimethylfluoride (100 μl, 40:1 v/v), dried with Na₂SO₄ (30 mg), and transferred into a ReactiVial (100 μl; Pierce, Oxfordshire, United Kingdom). The solvent was then evaporated with a stream of nitrogen and VG76e (10 μl, 27 μg, in PBS), mixed with sodium borate buffer (5 μl, pH 8.5, 100 mm), and incubated with [¹²⁴I]-SHPP at 0°C for 1 h. The labeled antibody was purified by gel filtration (PD-10 column; Amersham Pharmacia Biotech) using PBS as eluent.

A negative control antibody was also iodinated using the Bolton-Hunter reagent. Twenty μg of CIP5, an IgG1-type mouse monoclonal antibody with an affinity for polypeptide E6 expressed by human papillomavirus types 16 and 18 (15), were labeled with [¹²⁵I]-SHPP using the same method as described for VG76e iodination.

The immunoreactivity of the radiolabeled antibodies was determined using a modification of the method for radiolabeled anti-erbB2 described by Adel Bakir et al. (16). Briefly, ELISA plates (Corning Limited, Buckingham, United Kingdom) were coated with 10 μg/ml of human VEGF₁₆₅ (R&D Systems; Oxfordshire, United Kingdom) overnight in bicarbonate coating buffer (15 mm Na₂CO₃, 35 mm NaHCO₃, pH 9.6) at 4°C. Thereafter, wells were blocked with 100 μl of 1% BSA (Sigma) in PBS. The wells were then washed three times with PBS + 0.1% Tween 80 (Sigma). Radiolabeled VG76e was diluted to 10 ng/ml and added to the wells and allowed to bind for 2 h at room temperature. After incubation, unbound antibody was removed, the wells were washed three times with PBS + 0.1% Tween 80, and the bound antibody solubilized with 0.2 m NaOH. The total radioactivity added to each well and radioactivity from bound antibody were measured in a gamma counter (Compugamma 1282; Pharmacia LKB Biotechnology). Immunoreactivity of the antibody was calculated as bound counts × 100/total counts. Experiments were repeated at least three times.

Two subclones of the HT1080 human fibrosarcoma cell line (HT1080-26.6 and HT1080-1/3C; a gift from Prof. B. R. Binder, Vienna University, Vienna, Austria) were grown in RPMI 1640 (Life Technologies, Inc.). Cells were cultured at 37°C in a humidified 5% CO₂ incubator.

HT1080-26.6 and HT1080-1/3C cells were seeded in 150-cm² tissue culture flasks at a density of 1 × 10⁶ cells. Aliquots of medium were removed from each flask at regular intervals and immediately frozen. A human VEGF immunoassay kit (Quantikine; R&D Systems, Oxfordshire, United Kingdom) was used to determine the quantity of VEGF in each sample. The assay was performed according to the manufacturer’s instructions. Absorbance was measured with a plate reader (Anthos Labtech Instruments) at 450 nm (reference = 540 nm). Background absorbance was determined using diluent alone and subtracted from all wells. Standards with known concentrations of VEGF were included in the assay to obtain a calibration curve. For each cell line, the total VEGF production at each time point was determined and averaged.

This work was performed by licensed investigators in accordance with the United Kingdom’s Guidance on the Operation of Animals (Scientific Procedures) Act 1986 (HMSO, London, United Kingdom, 1990). All experiments were performed in full compliance with government regulations and United Kingdom Co-ordinating Committee on Cancer Research guidelines on animal welfare. HT1080-26.6 and HT1080-1/3C human fibrosarcoma cells were grown as xenografts in female BALB/c nu/nu mice (Harlan, Oxfordshire, United Kingdom), 8–10 weeks of age. HT1080-26.6 and HT1080-1/3C tumors were implanted by injecting 2 × 10⁶ and 4 × 10⁶ cells, respectively, into the rear dorsum subcutis. Tumors were selected for treatment when they had reached 5–8 mm in diameter (100–300 mg).

Tumor-bearing mice were injected i.v. via the lateral vein with 0.1 ml of [¹²⁵I]-SHPP-VG76e or [¹²⁵I]-SHPP-CIP5 (2–6.2 μCi). One cohort of mice also received an i.v. injection of unlabeled VG76e 2 h before the [¹²⁵I]-SHPP-VG76e injection. The concentration of unlabeled VG76e in this case was 100-fold greater than that in the injectate of [¹²⁵I]-SHPP-VG76e. At selected times after injection (from 5 min to 48 h), mice were sacrificed by exsanguination via cardiac puncture under general anesthesia (isofluorane inhalation). A minimum of three mice was used for each time point/treatment combination. Aliquots of heparinized blood were rapidly centrifuged at 2000 × g for 5 min to obtain plasma. Blood, plasma, urine, and tissues obtained by dissection (tumor, liver, kidney, spleen, lungs, heart, stomach, small intestines, large intestines, brain, muscle, skin, bone, and feces) were weighed, and the radioactivity was measured using an automatic gamma counter with decay correction (Cobra; Packard Bioscience, Berkshire, United Kingdom). Aliquots of the dose solution were also counted to determine the total radioactivity administered. The counts were then standardized for tissue mass and expressed as a percentage of the total activity injected into each mouse (i.e., %ID/g).

To determine the metabolite profile of [¹²⁵I]-SHPP-VG76e, a separate group of HT1080-26.6 and HT1080-1/3C tumor-bearing mice were injected with the radiotracer. Plasma and tumor samples were obtained as described above at 24 h after injection and analyzed by size-exclusion liquid chromatography. Aliquots of plasma were injected onto a size-exclusion column (Superdex 200 HR 10/30; Amersham Pharmacia Biotech) and eluted with a mobile phase of PBS at a flow rate of 2 ml/min. The eluent was monitored sequentially for radioactivity and UV absorbance at 280 nm. The radioactive components were integrated, and the data were corrected for physical decay and background radioactivity.

Tumor samples were rapidly dissected and frozen. To assess the metabolite profiles, samples were thawed, cut into small pieces, and added to ice-cold PBS. Samples were then homogenized using an Ultra-Thurrax homogenizer (Janke & Kunkel, KG, IKA) for 2 min, and the resultant homogenate was centrifuged (2000 × g; 15 min). Pellets and duplicate aliquots of the supernatant were then taken for measurement of the total radioactivity. The supernatant was filtered through a 0.2-μm membrane, and 1 ml was directly injected onto the size-exclusion column. The conditions, acquisition, and data processing were the same as described above for plasma.

The tail veins of HT1080-26.6-bearing mice were cannulated after induction of anesthesia with isofluorane/N₂O/O_2. For each scan, a tumor-bearing animal was placed prone within a thermostatically controlled jig and positioned in the bore of the scanner. After a bolus injection of [¹²⁴I]-SHPP-VG76e in PBS (10–16 μCi) via the tail cannula, emission scans were acquired on a HIDAC scanner (Ref. 17; Oxford Positron Systems, Oxfordshire, United Kingdom) in list-mode format. Animals were scanned for 2 h at any one time. The acquired list-mode data were sorted into 0.5-mm sinogram bins for image reconstruction (three-dimensional image pixel size, 0.5 × 0.5 × 0.5 mm). From the dynamic frames, static frames were obtained, and the images were visualized using the image analysis software Analyze (18).

Errors were expressed as SE. Comparisons between data sets were determined using Student’s t test for two independent populations. P ≤ 0.05 was considered significant.

The specific activity of ¹²⁴I-iodide was determined as 27.6 GBq/μmol (n = 2). The distillation yield (not decay corrected) of [¹²⁴I]NaI was 5.4 ± 1.4 MBq/microAh (n = 3). The radiochemical purity was >99% [TLC:methanol-water (3:1 v/v) as mobile phase and a cellulose stationary phase (Polygram Cel 300 UV254; Macherey-Nagel, Middleton-Cheney, United Kingdom)]. Radioactivity was measured by autoradiography (PhosphorImager 445 SI; Molecular Dynamics, Sunnyvale, CA). Isotopic impurities: 11% iodide-123 and 0.05 ± 0.02% (n = 4) of iodide-125 at end of bombardment. The average radiochemical yield of [¹²⁴I]-SHPP was 38.3 ± 17.4% (n = 3) with a radiochemical purity of 89.9 ± 5.2% (n = 3). Four radioiodinated VG76e analogues were synthesized ([¹²⁵I]-VG76e, [¹²⁵I]-SIB-VG76e, [¹²⁵I]-SHPP-VG76e, and [¹²⁴I]-SHPP-VG76e). The immunoreactivities of the [¹²⁵I]-analogues are shown in Fig. 2. Although the IodoGen method is the most common labeling strategy, this resulted in considerable loss of immunoreactivity (3.1% binding efficiency). Modifications to the IodoGen reaction conditions did not improve the biological activity of the VG76e analogue (data not shown). However, the indirect labeling methods improved immunoreactivity. The m-[¹²⁵I]-SIB approach resulted in a binding efficiency of 10.9%, whereas radioiodination of VG76e using the Bolton-Hunter reagent resulted in a binding efficiency of 34.0%. The immunoreactivity of [¹²⁵I]-SHPP-VG76e was significantly greater than those measured for the other VG76e analogues (P = 0.001 for [¹²⁵I]-SHPP-VG76e versus [¹²⁵I]-VG76e and P = 0.011 for [¹²⁵I]-SHPP-VG76e versus [¹²⁵I]-SIB-VG76e). The average radiochemical yield for [¹²⁵I]-SHPP-VG76e was 28% ± 8 (n = 5). The average radiochemical purity of [¹²⁵I]-SHPP-VG76e was 94% ± 2 (n = 6), and the specific activity was 33.3 GBq/μmol (900 mCi/μmol). In comparison, the average radiochemical yield of [¹²⁴I]-SHPP-VG76e was 16.3 ± 8.7% (n = 3), and the specific radioactivity was 11.1 GBq/μmol (300 mCi/μmol).

Two subclones of the HT1080 human fibrosarcoma were selected on the basis of their VEGF production. The amount of VEGF produced by both cell lines was measured during exponential growth. There was a highly significant difference between the amounts of VEGF produced by both cell lines at each time point (P ≪ 0.05). The HT1080-26.6 cell line produced 2.1-, 3.4-, and 3.9-fold more VEGF than the HT1080-1/3C line at 1, 2, and 3 days, respectively.

Fig. 3 shows the distribution of [¹²⁵I]-SHPP-VG76e to a range of tissues in mice bearing either HT1080-26.6 or HT1080-1/3C xenografts. There was a slow, time-dependent accumulation of radioactivity in the tissues, reaching peak levels at 24 and 48 h in HT1080-26.6 and HT1080-1/3C, respectively. At the time to reach peak levels for each tumor type, the %ID/g of tissue was significantly higher (1.6-fold) in the HT1080/26.6 tumors compared with the HT1080-1/3C tumors (P = 0.027). In the other tissues, the maximum radioactivity was generally seen within 5 min and rapidly cleared over the following 48 h. At the late time points (24 and 48 h), the levels of radiotracer in the most tissues (liver, kidney, spleen, lungs, heart, intestines, and brain) were less than that in tumors. The levels of [¹²⁵I]-SHPP-VG76e in blood, plasma, and urine were significantly higher than any other tissue type, and the rate of clearance from these sources was significantly faster than the rate at which [¹²⁵I]-SHPP-VG76e levels decreased in the tumors during the 48-h period.

To determine whether [¹²⁵I]-SHPP-VG76e was rapidly metabolized, 24-h plasma and tumor samples were analyzed by size-exclusion liquid chromatography. Radiochromatograms of plasma samples obtained from HT1080-1/3C tumor-bearing mice showed one major radioactive component (Fig. 4,a) with a retention time of ∼9 min. The UV absorbance of this component coeluted with an unlabeled reference standard of VG76e (data not shown). A similar peak, albeit much smaller, was obtained from the supernatant of HT1080-1/3C tumor homogenates (Fig. 4 b), indicating similar but significantly less radioactivity in tumor tissue relative to plasma. Similar results were obtained for plasma and tumor samples from HT1080-26.6 tumor-bearing mice (data not shown). Other than this single peak, no other radioactive components were observed in the plasma and tumor samples, suggesting that the radioactivity in these samples was attributable entirely to unmetabolized [¹²⁵I]-SHPP-VG76e.

Fig. 5 shows the biodistribution of [¹²⁵I]-SHPP-VG76e when challenged with unlabeled VG76e at 100 times the equivalent protein concentration. The %ID/g for HT1080-26.6 was 2.6-fold greater than that measured for HT1080-1/3C tumors (P = 0.018). When mice with HT1080–26.6 tumors were given unlabeled VG76e 2 h prior to [¹²⁵I]-SHPP-VG76e and sacrificed 24 h later, the %ID/g in the tumor decreased significantly by∼50% compared with unchallenged mice (P = 0.019). A similar trend was seen for mice with HT1080-1/3C, although the decrease was modest (24.6%) and did not reach significance. The %ID/g for the other tissues in both the HT1080-26.6 and HT1080-1/3C tumor-bearing mice did not decrease following the challenge. In most tissues, [¹²⁵I]-SHPP-VG76e levels increased after pretreatment with unlabeled VG76e. Fig. 5 also shows the uptake of [¹²⁵I]-SHPP-CIP5. In contrast to the expected differential uptake of [¹²⁵I]-SHPP-VG76e in HT1080-26.6 and HT1080-1/3C tumors, the uptake of [¹²⁵I]-SHPP-CIP5 in the two tumor models was similar (P = 0.411). Furthermore, in the other tissue types, the uptake of [¹²⁵I]-SHPP-CIP5 was generally higher than [¹²⁵I]-SHPP-VG76e, and similar levels of uptake were measured in the major tissues (liver, kidney, spleen, lungs, and heart).

Fig. 6 shows PET images of a HT1080-26.6 tumor-bearing mouse at different time points after injection of [¹²⁴I]-SHPP-VG76e. The images clearly show localization of the radiotracer to the tumor at the latter time point. As with the biodistribution studies above (Figs. 3 and 5) localization to the tumor (normalized to contralateral flank) was significantly higher than to brain (normalized to contralateral flank) at 24 h (P = 0.0286; n = 4). Additional areas of high radioactivity are visible in the thoracic cavity, abdominal cavity, and the bladder.

Angiogenesis is a common feature of all metastatic tumors, and the development of antiangiogenic strategies aimed at halting cancer progression by inhibiting the growth of new blood vessels is a major area of interest in clinical oncology (1, 19). Analyses of tumors show large variations in VEGF production, and little is known about the variation in the VEGF content of metastases, particularly in living perfused tumors (20, 21, 22). Therefore, functional or molecular imaging methods may provide a useful way of determining mechanisms of response based on pretreatment characteristics and also resistance pathways. In this report, we describe the development of a novel radiotracer for detecting VEGF levels by positron emission tomography.

Initial studies focused on labeling an anti-VEGF antibody with either [¹²⁵I]iodine for laboratory-based tracer evaluation studies or [¹²⁴I]iodine for PET imaging. Direct radioiodination of VG76e resulted in a loss of immunoreactivity. Direct iodination is commonly used for labeling antibodies with [¹³¹I]iodine in radioimmunotherapy, but yields can be low. Nikula et al. (23) have shown that the non-uniform distribution of tyrosine residues in IgG-type immunoglobulins (VG76e is isotype IgG1), particularly the overabundance of tyrosine in the complementarity determining regions, can lead to hyperiodination in the heavy chain variable region, causing a significant loss of function. Using two alternative indirect labeling approaches that targeted the lysine residues, the immunoreactivity of the iodinated antibody was improved. Similar to the murine-derived monoclonal antihuman antibody 2C3 (24), VG76e binds to VEGF and blocks binding to its receptors and thus VEGF-induced proliferation (10). Indirect radiolabeling of VG76e had no untoward effect on VEGF-induced endothelial cell proliferation (data not shown), consistent with the improved immunoreactivity. The tumor uptake studies also highlight that labeling VG76e with an iodo-linker molecule does not significantly impair biological function. The improved immunoreactivity was probably attributable to the separation of the radioiodine from the VEGF binding site. To the best of our knowledge, there is only one abstract reported in the literature on the use of [¹²⁴I]-SHPP for protein labeling (25), and we are the first to describe the use of the Bolton-Hunter reagent to label an antibody with a positron emitting radionuclide for PET imaging. The radiochemical yield of [¹²⁴I]-SHPP was lower than the [¹²⁵I]-SHPP antibody. The most probable explanation for this could be the use of a [¹²⁴-I]-labeled Bolton-Hunter reagent of lower specific activity. NaI (unlabeled) was used as carrier in the preparation of [¹²⁴-I]-labeled Bolton-Hunter reagent (see “Materials and Methods”), which led to a product of lower specific activity (33.3 versus 11.1 GBq/μmol). Another possible explanation could be radiolytic damage during the coupling of antibody to the [¹²⁴-I]-labeled Bolton-Hunter reagent, although we did not observe significant levels of lower molecular weight products in the analysis of the labeled product to suggest that radiolysis had occurred.

Intact antibodies generally have slow kinetics. The accumulation in tumor tissue and clearance from the circulation can take several (3, 4, 5, 6, 7, 8, 9, 10) days (16, 26). For instance, the maximum accumulation of [¹²⁵I]-SHPP-VG76e in HT1080-26.6 tumors occurred at 24 h, and 64% of the radioactivity was still present at 48 h. In addition, antibodies can bind to sites of inflammation including tumors (27). Thus, validation to determine specific labeling of target protein is required. We have addressed the issue of specific versus nonspecific binding in this report in three different ways: (a) binding of [¹²⁵I]-SHPP-VG76e in two tumors expressing different levels of VEGF; (b) binding of [¹²⁵I]-SHPP-VG76e per se and following presaturation of VEGF binding sites with cold unlabeled VG76e; and (c) comparison of binding of a control antibody of the same IgG class (IgG1) and similar molecular weight, [¹²⁵I]-SHPP-CIP5. The uptake of [¹²⁵I]-SHPP-VG76e was significantly higher in HT1080-26.6 compared with HT1080-1/3C tumors. This is in keeping with the in vitro studies in which HT1080-26.6 cells produced higher quantities of VEGF (2–4-fold) compared with HT1080-1/3C cells and demonstrates VEGF-specific uptake. Furthermore, the preinjection of an excess of unlabeled VG76e prior to [¹²⁵I]-SHPP-VG76e resulted in a significant reduction in tumor %ID/g, supporting the assertion of that [¹²⁵I]-SHPP-VG76e binds specifically to VEGF in vivo. An excess of unlabeled VG76e in vivo would reduce the availability of binding sites for [¹²⁵I]-SHPP-VG76e.

Interestingly, in both tumor models, the %ID/g of [¹²⁵I]-SHPP-VG76e was not reduced to zero by preadministration of unlabeled VG76e. The low level of [¹²⁵I]-SHPP-VG76e in the tumors after the challenge treatment was probably attributable to nonspecific uptake of the radiotracer. This assertion also explains the increase in [¹²⁵I]-SHPP-VG76e in normal tissues after challenge. Unlabeled VG76e would decrease the available binding sites in tumors, leading to a larger proportion of [¹²⁵I]-SHPP-VG76e in the circulation.

The utility of the radiotracer may be limited in certain tissues including the heart (plasma radioactivity) and bladder (urine radioactivity) because of high background in these tissues at 24 and 48 h (Fig. 5). In addition, high concentrations of [¹²⁵I]-SHPP-VG76e were observed in the liver, kidney, spleen, and lungs, especially at the early time-points. These organs normally have very good tissue perfusion and a significant blood volume. Taking in consideration the high levels of radiotracer measured in blood and plasma, it is reasonable to infer that “blood-pooling” effects are responsible for the levels of radioactivity measured in these tissues. Conversely, reduced tissue blood volume and an intact blood-brain barrier may explain the low uptake of radiotracer into the bone and brain, respectively. Importantly, no iodinated derivatives of the parent radiotracer were detected in samples of plasma and tumor tissue from both HT1080-26.6 and HT1080-1/3C tumor-bearing mice at 24 h after injection. Thus, the biodistribution data can be associated entirely with the uptake of intact [¹²⁵I]-SHPP-VG76e. It should be noted, however, that other techniques such as one-dimensional SDS-PAGE and isoelectric focusing may provide greater resolution to detect low levels of metabolism.

Studies with [¹²⁵I]-SHPP-CIP5 in mice bearing HT1080-26.6 or HT1080-1/3C tumors also confirmed the specificity of [¹²⁵I]-SHPP-VG76e for VEGF. CIP5 is a monoclonal antibody with a similar molecular weight to VG76e and an affinity for polypeptide E6 expressed by human papillomavirus types 16 and 18 (15). The similarities between the molecular weights of [¹²⁵I]-SHPP-CIP5 and [¹²⁵I]-SHPP-VG76e and the absence of E6 antigen in immune-deficient mice meant that iodinated CIP5 could be used as a negative control because its biodistribution would be dependent upon tissue perfusion and blood pooling. Although tumor localization of [¹²⁵I]-SHPP-CIP5 was high, there was no selective accumulation, and tumor levels were similar to that in the major organs at 24 h. The observation of high tumor localization of [¹²⁵I]-SHPP-CIP5 per se (comparable with [¹²⁵I]-SHPP-VG76e) was surprising because CIP5 is only known to bind to E6, which is not expressed in HT1080 cells (28). Of importance, however, this localization of [¹²⁵I]-SHPP-CIP5 was not VEGF dependent, as demonstrated by the similar levels of [¹²⁵I]-SHPP-CIP5 uptake in the two tumor types that express different levels of VEGF.

Whole-body PET demonstrated lack of tumor (HT1080-26.6) to normal tissue contrast at early time points after i.v. injection of [¹²⁴I]-SHPP-VG76e. High contrast was achieved, however, at 24 h in HT1080-26.6 tumors. The late images, therefore, illustrate in situ detection of VEGF and can be seen in all orthogonal planes. Analysis of regions of interest placed on the tumor, contralateral flank, and brain showed that tumor:flank ratios were significantly higher than brain:flank ratios, which is in agreement with the higher absolute %ID/g in tumor compared with the brain at 24 h (Fig. 3). In addition to tumor, the bladder, abdominal cavity, and thoracic cavity were also prominent in the images. This also compares favorably with the biodistribution data.

One other group has used an [¹²⁴I]iodine-labeled anti-VEGF monoclonal antibody in PET imaging (29, 30). Patients with advanced neoplasia participating in a Phase I clinical trial with an [¹²⁴I]-labeled humanized anti-VEGF antibody (HuMV833) that recognizes the 121 and 165 isoforms of VEGF⁵ underwent PET imaging. Unfortunately, the authors of these reports do not indicate whether HuMV833 showed specific targeting of the tumor tissue; in fact, the uptake values quoted for a range of tissues, including, tumor, aorta, spleen, liver, and kidney, are relatively similar. This is in contrast with our data in which we show specific targeting of tumor tissue. VG76e recognizes VEGF₁₂₁ (which is secreted), VEGF₁₈₉ (which is bound extracellularly to the cell surface), and VEGF₁₆₅ (which has intermediate properties). Binding to VEGF₁₈₉ would promote tissue-specific uptake of VG76e rather than an uptake influenced to some degree by local perfusion, but our study cannot distinguish which isoform is the major contributor. However, the HuMV833 studies do have some similarities with our findings, particularly the uptake of HuMV833 into the tumor tissue was similar at 24 and 48 h after injection, and the uptake values for bone marrow were low.

The development of PET or single-photon emission computed tomography methods of imaging angiogenesis-related factors or to monitor the effects of antiangiogenic agents is an emerging field of interest. Although there is little published work in this field, other approaches currently being explored for radiotracer development include radiolabeled antagonists of the α_Vβ₃ integrin expression (31), radiolabeled angiostatin and endostatin (32, 33), and radiolabeled squalamine (34). In the case of [¹²⁴I]-SHPP-VG76e, isotopic substitution of ¹²⁴I by ¹²³I or ¹³¹I will allow single-photon emission computed tomography imaging, which is more widely available, to be performed.

In summary, we have developed a novel positron-emitting radiotracer based upon a human monoclonal anti-VEGF antibody. The data show that the radiotracer specifically binds to VEGF. The further development of [¹²⁴I]-SHPP-VG76e as a radiotracer for measuring levels of tumor VEGF in vivo in patients is now warranted. This may provide a method to classify patients for therapy, understand pathways of angiogenesis in vivo, and assess resistance mechanisms.