This article is mentioned in:

Introduction

Angiogenesis is the vital physiological process involving the growth and remodeling of new blood vessels and is implicated in a number of diseases including cancer. Neoangiogenesis is essential for tumor growth as well as crucial for local and distant metastatization through both blood and lymphatic vessels (1,2). Therefore, many new targeted therapies have been developed and they are based on drugs able to bind vascular endothelial growth factor (VEGF) and its receptor (VEGFR), which have been shown to be upregulated in tumor and highly proliferating endothelial cells. Such overexpression has been associated with progression, metastatization and poor outcome in particularly aggressive cancers (3,4). Some of these drugs have been approved for human use and proved to be effective in many solid tumors (5).

The most widely used in clinical practice is the anti-VEGF monoclonal antibody (mAb) bevacizumab that binds the free VEGF. Others, like the tyrosine kinase inhibitors (TKIs) sorafenib and sunitinib, are able to target the VEGFR2 blocking the signaling cascade (6). It has been reported that the majority of patients benefits from targeted therapies, but a small fraction fails to show even initial benefits. The reasons may range from the involvement of parallel angiogenic pathways to the absence of the targets (7). Therefore, it would be important to predict which patients would benefit from a specific targeted therapy and several studies indicated the possibility to image angiogenic markers with the use of radiopharmaceuticals targeting VEGF or VEGFR (8).

In particular, since VEGFR is expressed also in some cancer cells, this technique will be useful in both early detection and cancer treatment monitoring (9–12). Radiopharmaceuticals to image tumor angiogenesis have been described in the literature, but many of them showed limitations that slowed or blocked the shift from preclinical to clinical trials (13). Among them the most common were poor or variable binding affinity and exaggerated liver uptake (14,15). In the present study, we optimized the radiolabeling of VEGF165 with 99mTechnetium (99mTc) using HYNIC as bifunctional chelator, obtaining a highly stable radiopharmaceutical with high in vitro receptor binding affinity. In vivo we used 99mTc-HYNIC-VEGF165 to image VEGFR expression in different tumor xenografts and correlated in vivo data with histological findings.

Materials and methods

Radiolabeling of VEGF-A165 with 99mTc

The human VEGF-A165 analogue with a molecular weight of 19 kDa was provided by Trophogen Inc. and radiolabeled with 99mTc through an indirect method after conjugation with the bifunctional chelator 6-hydrazinonicotinamide (HYNIC).

Radiolabeling was optimized by testing several labeling conditions including different HYNIC:VEGF ratios (1:1, 4:1 and 8:1) and different amounts of tricine (from 0.9 mg/ml to 200 mg/ml PBS) or SnCl2 (from 2 mg/ml to 20 mg/ml 0.1 M HCl). Briefly, VEGF165 (0.5 mg) was incubated with an excess of succinimidyl-6-hydrazinonicotinate hydrochloride (SHNH, SoluLink Inc., San Diego, CA, USA) for 2 h at room temperature in the dark. At the end of the incubation free SHNH was removed by size exclusion chromatography using a G-25 Sephadex PD10 column (GE Healthcare, Little Chalfont, Buckinghamshire, UK) and nitrogen-purged phosphate buffer saline (pH 7.4) as eluent.

The number of HYNIC groups bound per molecule of VEGF165 was determined by a molar substitution ratio (MSR) assay. Briefly, conjugated VEGF165 (2 μl) was added to a 0.5 mM solution (18 μl) of 2-sulfobenzaldehyde in 0.1 M 2-(N-morpholino) ethanesulfonic acid (MES) buffer (pH 5.0) and incubated at room temperature for 2 h. Phosphate buffer saline (PBS) alone was used as blank and duplicates were prepared. After 2 h the absorbance at 345 nm of each reaction was measured with a spectrophotometer and the number of HYNIC groups per molecule was calculated as indicated in the SoluLink data sheet. Radiolabeling was performed incubating 30 μg of VEGF165 (in 100 μl PBS) with 300 MBq of freshly eluted 99mTcO4− (100 μl), 100 μl of tricine (Sigma-Aldrich Chemicals, Dorset, UK) and 5 μl SnCl2 (Sigma-Aldrich Chemicals). Labeling efficiency (LE) and colloids percentage were measured up to 30 min of incubation. After labeling, an additional purification by size exclusion chromatography was performed using a Zeba Spin Column (Thermo Fisher Scientific, Waltham, MA, USA) to remove any free 99mTcO4−, tricine and SnCl2.

In vitro quality controls

Quality controls were performed using instant thin layer chromatography-silica gel (ITLC-SG) strip (Pall Life Sciences, Port Washington, NY, USA). Results were analyzed by a radio scanner (Bioscan Inc., Washington, DC, USA) to calculate the LE of 99mTc-HYNIC-VEGF165. The mobile phase for LE determination was a 0.9% NaCl solution, whereas the amount of colloids was determined using a NH3:H2O:EtOH (1:5:3) solution. Quality controls were performed before and after the purification with a Zeba Spin column. Additionally, reverse phase HPLC was carried out using a C8 Kinetex 4.6×250 mm column and a gradient of H2O (A) and acetonitrile (B) with 0.1 % TFA. The following gradient was used: 0–5 min 0–5% B, 5–20 min 5–95% B, 20–25 min 95% B and 25–30 min 95–5% B. Stability assays were performed adding 100 μl of 99mTc-HYNIC-VEGF165 to a vial containing 900 μl of fresh human blood serum and to another containing 900 μl of 0.9% NaCl solution. Both vials were incubated up to 24 h at 37°C. The radiochemical purity was measured at 1, 3, 6 and 24 h by ITLC analysis. A cysteine challenge assay was performed incubating the radiolabeled VEGF165 at 37°C for 60 min with different cysteine concentration, ranging from 1000:1 (cysteine:VEGF165) to 0.1:1 molar ratio. For each time point, radiochemical purity was evaluated by ITLC as described above. All known chemical forms of 99mTc-cysteine have Rf values between 0.5 and 1, when normal saline was used as mobile phase.

Integrity of the radiolabeled VEGF165 molecule was also checked by sodium dodecyl sulphate-polyacrylamide gel electrophoresis under non-reducing conditions, according to the method of Laemmli (16). Proteins were visualized by staining the gels with Coomassie Brilliant Blue (Thermo Fisher Scientific). Radioactivity associated with each band was determined scanning the gel with a radio scanner.

Cell lines

VEGFR+ cell line, human umbilical veins endothelial cells (HUVEC) were cultured in F-12K medium supplemented with 10% FCS, 100 IU/ml penicillin, 100 μg/ml streptomycin, 2 mM L-glutamine and EGM®-2 Bullet kit (Lonza, Walkersville, MD, USA) (17). The human anaplastic thyroid cancer cell line (ARO), the human colorectal cancer cell line (HT29) and the human poorly differentiated thyroid cancer cell line (K1) were grown in DMEM high glucose (Gibco, Carlsbad, CA, USA) supplemented with 10% FCS, 100 IU/ml penicillin, 100 μg/ml streptomycin and 2 mM L-glutamine (18–20).

In vitro binding studies

Measurements of cell uptake and retention of radiolabeled VEGF-A165, was performed in vitro using the semi-automatic system LigandTracer™ that allows to follow binding over time (Ridgeview Instruments AB, Vänge, Sweden) (21). Briefly, 106 HUVEC cells were seeded in a tilted Petri dish and incubated in a humidified incubator at 37°C and 5% CO2 for 24 h. The dish was then placed in the LigandTracer and allowed to rotate continuously for 15 min to induce the release of weakly attached cells. After one gentle wash, 2 ml of PBS containing radiolabeled VEGF165 (30 nM) were added to the dish and the rotation started, and the device was stopped when reaching maximal binding. Then the liquid was removed and replaced with culture medium without radiolabeled VEGF165 for calculating release of radioactivity from cells. Association and dissociation curves were obtained analyzing data by non-linear regression analysis with GraphPad Prism (GraphPad Software Inc., La Jolla, CA, USA) to calculate the kon, koff and Kd values.

In vivo studies

Biodistribution and imaging studies

For animal experiments, approval of the local ethics committee was obtained and the institutional and national guide for the care and use of laboratory animals was followed. Imaging studies were performed with a previously described high-resolution portable mini-gamma camera (HRC), IP-Guardian (Li-Tech S.r.l., Italy) (22). For in vivo biodistribution studies, 5.5 MBq (190 MBq/nmol, 100 μl) of radiolabeled VEGF165 were injected in the tail vein of 12 nude CD-1 mice and static planar posterior images were acquired using the HRC at 1, 3, 6 and 24 h, under light ether anesthesia. At the end of each imaging point three mice were euthanized and major organs were collected and counted in a single well gamma-counter.

In vivo cell-targeting experiments were performed in 36 nude CD-1 mice that were divided in three groups. Each group was injected subcutaneously in the right thigh with respectively 106 ARO, HT29 and K1 cells mixed with BD Matrigel® (BD Biosciences, Franklin Lakes, NJ, USA) (1:1). After tumor growth (approximately 0.6–1 cm3, in 20 days), 5.5 MBq of radiolabeled VEGF165 were administered i.v. in the tail vein and static planar posterior HRC images were acquired at 1, 3, 6 and 24 h, under light ether anesthesia. At each time point 3 mice were euthanized for ex vivo counting. Major organs and tumors were collected, weighed and counted for radio activity with a single well gamma-counter (Gammatom, Italy).

Blocking studies

Blocking studies were performed in four mice injected with 1 million HT29 cells mixed with Matrigel in the right thigh. After tumor growth, 5.5 MBq of 99mTc-VEGF165 were injected in the tail vein and images were acquired with a portable mini-gamma camera at 1 and 3 h post-injection. After 3 days, 99mTc-HYNIC-VEGF165 was pre-incubated for 1 h with 3.5-fold molar excess of recombinant human VEGFR2-Fc chimera (BioLegend Inc., San Diego, CA, USA) that act as a soluble decoy receptor (TRAP) that has been proved to prevent the binding of VEGF to endothelial cells. After the incubation, a dose of 5.5 MBq was injected in the tail vein of 3 of the 4 mice previously imaged and images were acquired with a portable mini-gamma camera at 1 and 3 h post-injection. After 3 more days, a 100-fold molar excess of unlabeled VEGF165 (COLD) was injected in the tail vein of the same 3 mice and after 10 min 5.5 MBq of 99mTc-HYNIC-VEGF165 was injected. Images were acquired with a portable mini-gamma camera at 1 and 3 h post-injection. Region of interest were drawn on the tumor and on the contralateral leg in each image and target-to-background (T/B) ratios were calculated.

Immunohistochemical analysis

For light microscope immunohistochemical analysis, small fragments of each excised tumor (ARO, HT29 and K1) were processed according to ABC/HRP technique (avidin-complexed with biotinylated peroxidase). These samples were washed in PBS, fixed in 10% formalin and embedded in paraffin according to a standard procedure. Serial 3-μm sections were cut using a rotating microtome, mounted on gelatin-coated slides and processed for immuno-histochemistry. These sections were de-paraffinized in xylene and dehydrated. They were immersed in citrate buffer (pH 6.0) and subjected to microwave irradiation twice for 5 min. Subsequently, all sections were treated for 30 min with 0.3% hydrogen peroxide in methanol to quench endogenous peroxidase activity. To block non-specific binding, the slides were incubated with M.O.M. Mouse Ig Blocking Reagent (Vector Laboratories Burlingame, Burlingame, CA, USA) for 1 h at room temperature. The slides were incubated overnight at 4°C with the following antibodies: i) mouse anti-VEGF monoclonal antibody (Santa Cruz Biotechnology, Santa Cruz, CA, USA); ii) mouse anti-VEGF receptor 1 (Flt-1/EWC) monoclonal antibody (ab9540; Abcam, Cambridge, UK); iii) mouse anti-VEGF Receptor 2 (KDR/EIC) monoclonal antibody (ab9530; Abcam). Optimal antisera dilutions and incubation times were assessed in a series of preliminary experiments. After exposure to the primary antibodies, slides were rinsed twice in phosphate buffer and incubated for 1 h at room temperature with the appropriate secondary biotinylated goat anti-mouse IgG (Vector Laboratories Burlingame, BA9200 and BA1000) and with peroxidase-conjugated avidin (Vectastain Elite ABC kit standard PK 6–100) for 35 min. After a further wash with phosphate buffer, slides were treated with 0.05% 3,3-diaminobenzidine (DAB) and 0.1% H2O2 (DAB substrate kit for peroxidase, Vector Laboratories SK-4100). Finally, sections were counterstained with Mayer's haematoxylin and observed using a light microscope.

Negative control experiments were carried out: i) by omitting the primary antibody; ii) by substituting the primary antibody with an equivalent amount of non-specific immunoglobulins; iii) by pre-incubating the primary antibody with the specific blocking peptide (antigen/antibody = 5 according to supplier's instructions). The staining assessment was made by two experienced observers in light microscopy. Immunoreactivity of VEGF, VEGFR1, and VEGFR2 was assessed in all samples. The intensity of the immune reaction was assessed micro-densitometrically using an IAS 2000 image analyzer (Delta Sistemi, Rome, Italy). The system was calibrated taking as zero the background obtained in sections exposed to non-immune serum. Ten 100 μm2 areas were delineated in each section using a measuring diaphragm. The quantitative data regarding the intensity of immune staining were analyzed statistically using analysis of variance (ANOVA) followed by Duncan's multiple range test as a post-hoc test.

Results

VEGF165 analogue can be efficiently radiolabeled with 99mTc

Highest labeling efficiency was obtained when the analogue was conjugated with a ratio HYNIC:VEGF165 of 8:1. Determination of molar substitution ratio of HYNIC-conjugated VEGF165 demonstrated that an average of 4.3 molecules of HYNIC groups were bound per molecule of analogue. Higher ratios were not selected to avoid over-conjugation of the hormone and possible structural modification. Optimization of the labeling procedure of the HYNIC-VEGF165 conjugate (30 μg) with 99mTc showed that, after 10 min of incubation, the use of 100 μl of tricine (0.5 mM) and 5 μl of SnCl2 (50 nM) allowed to obtain the highest LE (65%) and the lowest amount of colloids (<5%). After purification we were able to obtain a radiochemical purity of >95% as confirmed by both ITLC and HPLC analysis (Fig. 1). Specific activity of resulting 99mTc-HYNIC-VEGF165 was 190 MBq/nmol. Radiolabeled VEGF165 was stable up to 24 h in both in human serum and in a 0.9% NaCl solution at 37°C, as well as in solutions containing increasing cysteine concentrations (Fig. 2A). A slight decrease in the radiochemical purity was observed only at high cysteine concentrations (>500:1) (Fig. 2B). Gel electrophoresis of radiolabeled, conjugated and unconjugated analogue showed no significant differences and the absence of significant degradation or aggregation resulting from conjugation and/or labeling (Fig. 3). Poor resolution of the bands is due to the high glycosylation of the analogues.

Figure 1 HPLC profile showing UV (black) and radioactive (grey) chromatograms of 99mTc-HYNIC-VEGF165.

Figure 2 Stability assay performed in saline and human serum (A), and cysteine challenge (B) of 99mTc-HYNIC-VEGF165.

Figure 3 SDS-Page electrophoresis in non-reducing conditions showing native (2), conjugated (3) and radiolabeled VEGF165 (4). Lanes 1 and 5 represent the molecular weight marker and scan of well no. 4 with a linear radioactivity scanner.

Radiolabeled VEGF165 binds with high affinity to VEGFRs

Kinetic biding assay with LigandTracer showed an increasing uptake of radiolabeled VEGF165 from HUVEC cells that reached a plateau after 50 min (Fig. 4). Retention studies revealed a slow dissociation rate from membrane bound receptors in the following 2 h, with a Kd of 192 pM.

Figure 4 Kinetic binding assay of radiolabeled VEGF165 on HUVEC cells. Fitting was obtained by nonlinear regression analysis determining kon (1.45×108), koff (0.0279) and kD (192 pM).

99mTc-HYNIC-VEGF165 is able to image tumor xenografts in mice

Biodistribution studies with 99mTc-HYNIC-VEGF165 showed a high and persistent uptake by the liver and a moderate uptake by the kidneys with almost no signal from other organs and blood pool (Fig. 5). Single organ counting revealed a high %ID/g also in the lungs and spleen. In vivo targeting experiments showed a focal uptake in the right thigh of each group bearing tumor xenografts with a T/B ratio of 4.5 at 1 h p.i in mice bearing a HT29 xenograft. Animals bearing ARO and K1 cells showed a T/B of 3.5 and 2.3, respectively, that decreased over time (Fig. 6).

Figure 5 Single organ counts at 1 h (white bars), 3 h (squared bars), 6 h (grey bars), and 24 h (black bars). Data are shown as %ID/g (A) and %ID/organ (B) over time. Each point is the mean of three mice. Error bars represent standard deviation.

Figure 6 T/B ratios calculated at 1, 3 and 6 h in mice bearing an ARO (black bars), HT29 (squared bars) or K1 (white bars) xenograft. Each point is the mean of three mice. Error bars represent standard deviation.

In vivo binding of 99mTc-HYNIC-VEGF165 can be inhibited by an excess of unlabeled VEGF or VEGFR2-Fc

Blocking studies with 99mTc-HYNIC-VEGF165 confirmed the results of previous targeting experiments. After pre-incubation of the radiopharmaceutical with recombinant VEGFR2-Fc, a main liver and spleen uptake, with reduced signal from kidneys, was detected, resembling the typical biodistribution of a nonspecific radiolabeled antibody (Fig. 7). The overall uptake in tissues was lower and the uptake in the tumor was considerably reduced. Similar findings were obtained after the pre-injection of a 100-fold molar excess of unlabeled VEGF165 with the exception of the signal from kidneys, which was similar to the signal obtained with labeled VEGF only. Calculated T/B ratios for the 'HOT' group reflected the data obtained with the previous experiments with a maximum uptake reached at 1 h that slowly decreases with time (Fig. 8). The T/B ratio in the TRAP group was reduced by 70% due to the co-incubation with VEGFR2-Fc at 1 h and the T/B ratio in the 'COLD' group was reduced by 60% at 1 h. Minor blocking was evident at 3 h in both TRAP and 'COLD' group mainly due to the decreased activity in tumors of the control group.

Figure 7 Body images (top) and details of the back (bottom) of the same mouse injected with 5.5 MBq of 99mTc-VEGF (left), 5.5 MBq of 99mTc-VEGF following pre-incubation with a 3.5-fold molar excess of VEGFR2-Fc (middle) and 5.5 MBq of 99mTc-VEGF following pre-injection of a 100-fold molar excess of unlabeled VEGF (right).

Figure 8 T/B ratios at 1 and 3 h calculated on the images obtained from the HOT (99mTc-VEGF), COLD (99mTc-VEGF+unlabeled VEGF) and TRAP (99mTc-VEGF+VEGFR2-Fc) groups. Values are the mean of 3 mice per each group and images were acquired at 1 h p.i. Error bars represent standard deviation.

VEGF expression at IHC and T/B ratio of 99mTc-HYNIC-VEGF165 correlates inversely

IHC analysis on excised tumor showed the presence of VEGF, VEGFR1 and VEGFR2 on both the lesion and the surrounding vessels to different extent (Fig. 9). After semi-quantitative analysis of expression levels, a higher amount of free VEGF was present in lesions derived from K1 cell lines (33.2%), followed by HT29 (15.7%) and ARO cells (10.6%). VEGFR1 and -2 were present heterogeneously between tumor cells and blood vessels, revealing that even cancer may express VEGF receptors on the plasma membrane. IHC data were compared with the uptake of radioactive VEGF165 and an inverse correlation was observed between endogenous VEGF and T/B ratio (r2= 0.63; p= 0.03, Fig. 10A). On the contrary, a positive correlation was observed between radioactive VEGF165 uptake and VEGFR1 (r2= 0.64; p=0.03, Fig. 10B). In addition, tumor weight positively correlates with VEGF production (r2=0.65; p=0.03) and shows a trend to inversely correlate with radiolabeled VEGF uptake (r2= 0.31; p= 0.35).

Figure 9 IHC analysis of VEGF, VEGFR1 and VEGFR2 expression on HT29, K1 and ARO excised tumors.

Figure 10 Correlation between 99mTc-VEGF165 uptake and VEGFR1 positive cells (%) (A) and between 99mTc-VEGF165 uptake and VEGF positive cells (%) (B) in ARO (grey), HT29 (white) and K1 (black) cell lines.

Discussion

Imaging of tumor microenvironment has been described as a promising approach for non-invasive diagnosis of cancer metastases and to monitor the efficacy of new drugs (23). Given the role of angiogenesis in metastatization and tumor growth, VEGF and its receptors are optimal diagnostic and therapeutic targets (24). Their presence has been reported in many cancer types and it was correlated with clinical data. However, given the heterogeneity of VEGFR expression on cancer cells, their role in tumor dedifferentiation or signaling is still unclear (13). In undifferentiated thyroid cancer, the use of TKIs blocking the VEGF/VEGFR pathway showed its potential as a promising therapeutic approach (25). Unfortunately, severe side effects have been reported in some patients after long time treatment. Therefore, a non-invasive diagnostic tool to predict the response to therapy and evaluate drug efficacy is vitally needed. In the past many attempts have been made to develop radiopharmaceuticals to image angiogenesis with promising results. Among them, 111In, 89Zr or 64Cu radiolabeled bevacizumab was able to efficiently image xenografts from ovarian cancer, but the high radiation burden to the patient and the low availability of 89Zr and 64Cu were some of the drawbacks of its use (26,27).

Other groups tried to use recombinant human VEGF to overcome the long half-life of mAbs and used radioiodine, 99mTechnetium (99mTc), 64Cu or 68Ga as the isotopes of choice (8,13). In the present study, we followed the same approach to strengthen the hypothesis that the use of recombinant human VEGF to target angiogenesis is a promising methodology to develop non-invasive diagnostic tools and monitor novel targeted drug development. In addition we improved the radiolabeling method to produce a high specific activity and highly stable radiopharmaceutical to bind VEGFR with high binding affinity and avoid misinterpretation of in vivo studies. Furthermore, the use of picomolar amounts of radiolabeled VEGF for a scintigraphic study should not raise any concern about a potential biologic effect of such radio-pharmaceuticals and in particular on the pro-angiogenic effect that VEGF analogues may have on existing blood vessels. Thus, in vivo results allowed us to image tumor angiogenesis in xenografts from three different human cell lines with high T/B ratio between 1 and 3 h post-injection (max T/B at 1 h for HT29 was 4.5). Nevertheless a high liver uptake was observed in all mice till late time points, confirming previous findings from other groups (28).

The issue has been raised that VEGF-based probes uptake in the tumor area is highly heterogeneous, probably because of the combination of several mechanisms like non-uniform perfusion of tumor vasculature, differential receptor occupancy by host VEGF or differential accessibility of VEGF receptors on luminal and subluminal surfaces of the endothe-lium (29,30). Moreover, it has been reported by Chen et al that tumor size negatively influences the uptake of radiolabeled VEGF by the tumor, probably because of the presence of necrotic areas (31). To address, in particular, the role of necrosis and endogenous VEGF production, we performed histological and immunohistochemical analysis of each tumor imaged with 99mTc-VEGF165. Results confirmed variability in VEGFR1 and VEGFR2 receptor expression and ligand occupancy in both host endothelium and cancer cells. The presence of both receptors has been also confirmed in a recent study by Meyer et al, but it would be of interest to investigate the differential contribution of the different VEGFR subtypes (29). Moreover, we confirm that bigger tumors show lower uptake of radiolabeled VEGF, although we did not observe the presence of significant necrotic areas in any tumor. On the other hand, we observed a positive correlation between tumor size and production of endogenous VEGF (p=0.03), suggesting that the reduced tumor uptake of the radiopharmaceutical could depend on saturation of VEGFRs rather than size or necrosis, as previously suggested (30).

Therefore, this study highlights an important aspect that has not been considered before: the role of both endogenous VEGF production and VEGFR expression on imaging strategies. While for other ligand receptor systems, the endogenous production of the ligand may not be highly relevant for imaging the receptor ligand (i.e. IL-2 and IL-2 receptor) (32,33), herein we show that the high production of endogenous VEGF by tumor cells hampers the possibility to image its receptors. In light of our results we can better interpret previously published studies with radiolabeled VEGF (both VEGF121 and VEGF165) (34) that showed very poor tumor uptake, in contrast with studies with radiolabeled anti-VEGF mAb that showed high tumor uptake (35). Collectively, these data confirm that the presence of high VEGF levels in tumors, particularly those advanced with highly hypoxic tumor microenvironment and aggressive phenotypes, may saturate VEGF receptors, thus limiting the possibility to image receptors.

Overexpression of soluble VEGFR in some tumors and significant sequestration of VEGF on cell surface heparin-sulfate proteoglycans may also contribute to highly variable imaging results in different tumors. One possible limitation of our study is the limited number of animals used. However, in mice bearing K1 tumors, or ARO or HT29, we found highly consistent data supporting the very low variability within the same cell line. Nevertheless, different tumors showed different levels of VEGF/VEGFR. Another possible limitation could be that the semi-quantitative evaluation of VEGF189 that we performed by immunohistochemistry may not represent the real production of VEGF121 and VEGF165. However, these forms are splicing variants of the same molecule and are usually expressed in similar quantities (36). Overall we believe that the above considerations may have a limited impact on the final conclusion that VEGFR imaging in tumors by using radiolabeled VEGF is extremely variable, influenced by the presence of endogenous VEGF and unrelated to VEGFR receptor expression. It can be extrapolated that an accurate in vivo evaluation of tumor angiogenesis should include both VEGF and VEGFR imaging, unless the predominant clinical relevance of one over the other is demonstrated. Finally, the development of a superagonist VEGF analogue could allow the use of molecule with a greatly increased affinity for its receptor, overcoming the quenching effect due to endogenous VEGF (37).

In conclusion, imaging of angiogenesis by targeting VEGFR with radiolabeled VEGF analogues may be a complementary approach to evaluate angiogenic status of tumors. This approach may allow the evaluation of anti-angiogenic drugs at both preclinical and clinical stages in combination with VEGF imaging. Our results indicate that VEGFR expression is variable in both tumors and its imaging is hampered by endogenous VEGF production. Therefore, additional studies are required to fully understand the VEGF/VEGFR relationship in different cancers and establish a more accurate and angiogenic phenotype-determined imaging protocols.

Acknowledgments

This study was funded by grants from the Italian Association for Cancer Research (AIRC IG-2013 14151 and 13234), 'Sapienza' University research projects and Regione Lazio (FILAS-RU-2014-1020). We also wish to acknowledge the non-profit association Nuclear Medicine Discovery for support.

Abbreviations:

References