The role of the VEGF inhibitor bevacizumab in the treatment of diffuse intrinsic pontine glioma (DIPG) is unclear. We aim to study the biodistribution and uptake of zirconium-89 (89Zr)-labeled bevacizumab in DIPG mouse models. Human E98-FM, U251-FM glioma cells, and HSJD-DIPG-007-FLUC primary DIPG cells were injected into the subcutis, pons, or striatum of nude mice. Tumor growth was monitored by bioluminescence imaging (BLI) and visualized by MRI. Seventy-two to 96 hours after 89Zr-bevacizumab injections, mice were imaged by positron emission tomography (PET), and biodistribution was analyzed ex vivo. High VEGF expression in human DIPG was confirmed in a publically available mRNA database, but no significant 89Zr-bevacizumab uptake could be detected in xenografts located in the pons and striatum at an early or late stage of the disease. E98-FM, and to a lesser extent the U251-FM and HSJD-DIPG-007 subcutaneous tumors, showed high accumulation of 89Zr-bevacizumab. VEGF expression could not be demonstrated in the intracranial tumors by in situ hybridization (ISH) but was clearly present in the perinecrotic regions of subcutaneous E98-FM tumors. The poor uptake of 89Zr-bevacizumab in xenografts located in the brain suggests that VEGF targeting with bevacizumab has limited efficacy for diffuse infiltrative parts of glial brain tumors in mice. Translating these results to the clinic would imply that treatment with bevacizumab in patients with DIPG is only justified after targeting of VEGF has been demonstrated by 89Zr-bevacizumab immuno-PET. We aim to confirm this observation in a clinical PET study with patients with DIPG. Mol Cancer Ther; 15(9); 2166–74. ©2016 AACR.

Recent advances in molecular and cellular cancer biology have resulted in the identification of critical molecular tumor targets involved in the different phases of tumor growth and spreading. This knowledge has boosted the rational design of novel drugs, especially monoclonal antibodies and tyrosine kinase inhibitors. However, broad application of these targeted therapies to treat brain tumors has lagged behind in daily clinical practice. By using molecular positron emission tomography (PET) imaging, target expression and biodistribution can be studied concurrently in a relatively noninvasive manner. It potentially allows the identification of those patients who may, or may not, benefit from targeted therapy.

A disease that may benefit from molecular imaging is diffuse intrinsic pontine glioma (DIPG), a childhood malignancy located in the brainstem. In the past 40 years, the outcome of patients with DIPG has remained unchanged, with less than 10% of the patients being alive 2 years from diagnosis (1, 2). Given the lack of gadolinium uptake on MRI in DIPG tumors (3, 4), it is plausible that the blood–brain barrier (BBB) in DIPG often remains intact which might explain the resistance to systemic chemotherapy in these patients.

A well-studied drug target in gliomas is VEGF, a signal protein-stimulating angiogenesis and increasing vessel permeability (5). Overexpression of VEGF-A, its receptor VEGFR2, or both, have been implicated as poor prognostic markers in various clinical studies (5, 6). Bevacizumab is a recombinant humanized monoclonal antibody that selectively binds with high affinity to all isoforms of human VEGF-A and neutralizes their biological activity (5, 7). Studies in adult patients with recurrent high-grade gliomas (HGG) reported high radiological response rates with bevacizumab treatment (7–9), but recently 2 large phase III randomized studies showed no improvement in overall survival with bevacizumab treatment in an upfront setting (10–12). Bevacizumab has been studied in a number of nonrandomized trials in pediatric brain tumor patients. Efficacy in these trials has been variable, with a subset of patients showing clear radiological and/or clinical improvement (13, 14). The role of bevacizumab in the treatment of patients with DIPG is even less clear but is currently studied in several trials (refs. 15–18; NCT00890786 and NCT01182350; clinicaltrials.gov; NTR2391 Trialregister.nl).

No validated methods are available to identify patients who may potentially benefit from bevacizumab treatment. Lack of clinical effect may be due to either poor transport of bevacizumab into the tumor microenvironment due to an intact BBB, or a lack of VEGF expression. In this study, we analyzed VEGF(R) expression in adult and childhood HGG, including DIPG tumors. Furthermore, we studied bevacizumab distribution in vivo using molecular PET imaging with 89Zirconium-labeled bevacizumab in murine DIPG models (19).

VEGF-A and VEGFR2 mRNA expression profiles

VEGF-A and VEGFR2 (KDR) mRNA expression in DIPG (n = 27) and pediatric HGG (pHGG; n = 53) were determined in silico, using publicly available datasets, and compared with a dataset of nonmalignant brain tissue (n = 44), low-grade brainstem glioma (n = 6), and adult HGG (n = 284). These datasets include tumor material from biopsy, resection, and autopsy (DIPG). Differences were analyzed by two-way ANOVA and P < 0.01 was considered significant. As a validation of these findings, VEGF-associated gene expression was studied in normal brain, low-grade brainstem glioma (LG-BSG), DIPG, and glioblastoma by creating a heatmap using K-means clustering. All expression analyses were performed using R2, a web-based microarray analysis and visualization platform (http://r2.amc.nl).

Immunohistochemistry and in situ hybridization

Formalin-fixed, paraffin-embedded slides were sectioned from xenograft tumors and brain tissue and subjected to immunohistochemical (IHC) staining. Briefly, after deparaffinization and heat-induced antigen retrieval, sections were incubated with primary mouse anti-Ki67 antibodies (clone MIB-1, DAKO) overnight at 4°C. Thereafter, slides were washed and incubated with HRP-conjugated EnVision (DAKO) and subsequently stained by DAB with hematoxylin counterstaining.

For in situ hybridization (ISH), tumors were cut in 5-μm slices and incubated with VEGF probes against the human VEGF coding sequence using a previously described protocol (20). Samples were evaluated by microscopy with a Zeiss Axioskip microscope (HBO100W/Z) and equipped with a Canon digital camera and imaging software (Canon PowerShot A640, Canon Utilities, ZoomBrowser Ex. 5.7, Canon Inc.).

Labeling and quality control of 89Zr-bevacizumab

Bevacizumab was labeled with 89Zr using N-succinyl-desferrioxamine (N-suc-Df) as described previously (21). In short, the chelator, desferrioxamine, was succinylated to N-suc-Df. Next, the hydroxamate groups were blocked with iron and the succinyl group was activated as its TFP-ester (Fe-N-suc-Df-TFP ester). Bevacizumab (6 mg/mL) was reacted with 2 equivalents of Fe-N-suc-Df-TFP ester at pH 9 for 30 minutes at room temperature. Hereafter, iron was removed at pH 4.2 to 4.5 with an excess of EDTA for 30 minutes at 35°C, and N-suc-Df-bevacizumab was purified by size exclusion chromatography using a PD-10 column. Radiolabeling of N-suc-Df-bevacizumab was performed in HEPES buffer: to 200 μL 89Zr in 1 mol/L oxalic acid 90 μL 2M Na2CO3 was added. After 3 minutes, 300 μL 0.5 mol/L HEPES, N-suc-Df-bevacizumab, and 700 μL 0.5 mol/L HEPES were added. After 60-minute reaction time, 89Zr-N-suc-Df-bevacizumab was purified by PD10 using 5 mg/mL gentisic acid in 0.9% NaCl (pH 4.9–5.4) as the mobile phase.

Radiochemical purity and antibody integrity were determined using instant thin-layer chromatography (iTLC), high-performance liquid chromatography (HPLC), and SDS-PAGE followed by phosphor imager analysis. For analysis of immunoreactivity, an ELISA-based assay was used. iTLC analysis of 89Zr-bevacizumab was performed on TEC control chromatography strips (Biodex). As the mobile phase, citrate buffer (20 mmol/L, pH 5.0) containing 10% acetonitrile was used. HPLC analyses of bevacizumab modification and radiolabeling were performed using a Jasco HPLC system equipped with a Superdex 200 10/30 GL size exclusion column (GE healthcare Life sciences) using a mixture of 0.05 mol/L sodium phosphate, 0.15 mol/L sodium chloride (pH 6.8), and 0.01 mol/L NaN3 as the eluent at a flow rate of 0.5 mL/min. The radioactivity of the eluate was monitored using an inline NaI(Tl) radiodetector (RaytestSockett).

Cell lines and animal models

Animal experiments were performed in accordance with the European Community Council Directive 2010/63/EU for laboratory animal care and the Dutch Law on animal experimentation. The experimental protocol was validated and approved by the local committee on animal experimentation of the VU University Medical Center. Athymic nude-Foxn1nu mice (6 weeks old) were purchased from Harlan/Envigo and kept under filter top conditions and received food and water ad libitum.

The primary HSJD-DIPG-007 cell line was established from DIPG tumor material obtained at Hospital Sant Joan de Deu (Barcelona, Spain) after autopsy from a 6-year-old patient and was confirmed to have a H3F3A (K27M) and ACVR1 (R206H) mutation (22). The E98 cell line was obtained from Radboud University Nijmegen Medical Center (Nijmegen, The Netherlands; ref. 23), the U251 glioma cell line from ATCC. All cell lines were transduced in our laboratory to express firefly luciferase (FLUC) and/or mCherry (24). Cell lines were mycoplasma-negative and were authenticated by STR analysis modified from De Weger and colleagues (25).

E98-FM cells were injected subcutaneously in female athymic nude mice (7–9 weeks of age) to expand the number of cells (19). When the subcutaneous tumor reached a diameter of 1 cm, the tumor was removed, and a single-cell suspension was prepared by mechanical disruption through a 100-μm nylon cell strainer. HSJD-DIPG-007-FLUC was cultured in tumor stem medium (TSM; ref. 26), U251-FM in DMEM supplemented with 10% FCS and penicillin/streptomycin. Shortly before stereotactic injection, cells were washed once with PBS and concentrated to 1 × 105 cells per μL. Mice were stereotactically injected with 5 × 105 cells in a final volume of 5 μL into either the pons or striatum or injected subcutaneously with 3 × 106 cells in a final volume of 100 μL/flank. Coordinates used for intracranial injections were 1.0 mm X, −0.8 mm Y, 4,5 mm Z from the lambda for pontine tumors and 0.5 mm X, 2 mm Y, −2 mm Z from the bregma for the striatum tumors. Coordinates were previously validated and on the basis of “The mouse brain in stereotaxic coordinates” by Franklin and Paxinos (27). Tumor engraftment was monitored by bioluminescence measurement of the Fluc signal. For E98-FM, early-stage tumors in the pons (n = 8), striatum (n = 7), and subcutaneous xenografts (n = 3) were allowed to grow for 18 days and late-stage tumors (n = 9, striatum; n = 3, subcutaneous) for 35 days after xenograft injection. E98 xenograft tumors in the pons were not available 35 days postinjection because injection of E98FM cells in the pons would result in death of the mice due to tumor growth within 3 weeks. For HSJD-DIPG-007-FLUC and U251-FM, tumors (pons, striatum, subcutaneous) were evaluated, at days 78 and 22, respectively (Fig. 2A).

Two to 3 days before the endpoint of the study, mice were injected intraperitoneally with 89Zr-labeled bevacizumab (40 μg, 5 MBq for PET analysis and ex vivo biodistribution or 185 kBq for ex vivo biodistribution only).

PET imaging and ex vivo analysis

The distribution of 89Zr-bevacizumab was determined 72 hours (ex vivo only) or 96 hours (PET followed by ex vivo analysis) after administration, as a minimum of 72-hour interval between injection and scanning has previously been shown to achieve optimal tumor-to-nontumor ratios (28). Scanning of E98-FM tumors was performed using a LSO/LYSO double-layer ECAT High-Resolution Research Tomograph (HRRT, CTI/Siemens): a small-animal and human brain 3-dimensional (3D) scanner with high spatial resolution (2.3–3.4 mm full width at half maximum) and high sensitivity (29). Mice were anesthetized by isoflurane inhalation anesthesia (1.5 L O2/min and 2.5% isoflurane) before positioning in the HRRT. A static transmission scan (6 minutes) using a rotating 740 MBq 137Cs point source was performed. Before positioning the mice in the HRRT, a canula was placed intraperitoneally to enable later injection of 18F(−). Static images of 60-minute acquisition time were obtained. Immediately thereafter, 18F(−) was injected intraperitoneally (10 MBq/mouse) for colocalization of bone structures (static images of 60-minute acquisition). Body temperature was controlled with a heated platform (kept at 37°C).

Mice with HSJD-DIPG-007-FLUC or U251-FM tumors and where tumor growth was confirmed by increase of BLI signal were imaged using another, preclinical, PET system, (Nanoscan PET-CT, MEDISO), 72 hours after 89Zr-bevacizumab injection. PET images were analyzed using AMIDE software (Amide's a Medical Image Data Examiner, version 1.0.1; ref. 30).

MRI

Mice with representative HSJD-DIPG-007-FLUC or U251-FM tumors were selected for MRI on the basis of the intensity of the bioluminescence signal. Gadolinium (750 μmol, Dotarem) was administered intravenously immediately before imaging. Mice were anesthetized by isoflurane inhalation anesthesia (1.5 L O2/min and 2.5% isoflurane), placed in a preclinical PET-MRI system (Nanoscan system, MEDISO), and T1- and T2-weighed images with gadolinium contrast were acquired. MR images were analyzed using MIPAV software (Medical Image Processing, Analysis, and Visualization, version 7.2.0).

Ex vivo analysis

Immediately after PET imaging, animals were sacrificed for ex vivo tissue distribution analysis. Blood, urine, tumor, and various tissues were excised, rinsed in PBS to remove residual blood, and weighed. Radioactivity in blood and tissues (in percentage injected dose per gram of tissue: % ID/g) was determined using an LKB 1282 gamma-counter (Compugamma, LKB Wallac). Differences in the amount of radioactivity in healthy brain regions versus subcutaneous-, pontine-, and striatal tumors were analyzed by Kruskal–Wallis test with Dunn multiple comparison posthoc testing. P < 0.05 was considered statistically significant.

VEGF-A-, VEGFR2-, and VEGF-associated gene expression in DIPG

A search in a publically available mRNA expression database curated by the Amsterdam Medical Center (R2.amc.nl) revealed that VEGF-A mRNA is overexpressed in DIPG compared with normal brain and compared with low-grade brainstem glioma (LG-BSG) and non-pontine adult and pediatric HGG (Fig. 1A; P < 0.01, ANOVA). VEGFR2 (KDR) mRNA expression was low in both pediatric and adult glioma, compared with normal brain (Fig. 1B; P < 0.01, ANOVA). A heatmap generated using K-means clustering of expression of VEGF-A–associated genes confirmed an aberrant VEGF-A pathway to be more prominent in DIPG compared with LG-BSG. Most DIPG (red) samples clustered together with pediatric GBM (green) and LG-BSG (blue) clustered with normal brain tissue (purple) (Supplementary Fig. S1).

Figure 1.

Expression profiles of VEGF-A and VEGFR2. A, box plots representing relative median mRNA expression in DIPG (n = 27) and pHGG (n = 53) versus datasets of normal brain tissue (n = 44, dark grey), low-grade brainstem glioma (n = 6), and adult HGG (n = 284). VEGF-A overexpression is shown in DIPG compared with normal brain and compared with adult glioma. B, VEGFR2 is not overexpressed in DIPG and pediatric glioma compared with normal brain. *, P < 0.01, ANOVA.

Figure 1.

Expression profiles of VEGF-A and VEGFR2. A, box plots representing relative median mRNA expression in DIPG (n = 27) and pHGG (n = 53) versus datasets of normal brain tissue (n = 44, dark grey), low-grade brainstem glioma (n = 6), and adult HGG (n = 284). VEGF-A overexpression is shown in DIPG compared with normal brain and compared with adult glioma. B, VEGFR2 is not overexpressed in DIPG and pediatric glioma compared with normal brain. *, P < 0.01, ANOVA.

Close modal

Biodistribution of 89Zr-bevacizumab

After tumor growth was confirmed by an increase in BLI signal, 89Zr-bevacizumab was injected in these animals. Seventy-two to 96 hours after intraperitoneal injection of 89Zr-bevacizumab, animals were imaged by PET; a schematic overview of the experiment is given in Fig. 2, with representative BLI figures of E98-FM–engrafted mice shown in Fig. 2B. 89Zr-bevacizumab uptake was not visible in the E98 gliomas located in the pons (only early stage, Fig. 3A) or in the striatum at early or late stages. The rest of the brain showed no uptake of 89Zr-bevacizumab either, independently of the presence of a tumor. However, subcutaneous E98 tumors showed high accumulation of 89Zr-bevacizumab indicated by a red hotspot (Fig. 3A). Ex vivo tissue distribution measurements confirmed that no significant 89Zr-bevacizumab uptake was detected in the brain or brain tumor at any stage of disease, whereas there was high uptake in the subcutaneous tumor (P < 0.01; Fig. 4A). Besides accumulation in the subcutaneous tumor with an average level of 50% ID/g, 89Zr-bevacizumab was observed in all animals (to a lesser extent) in the blood pool and in well-perfused organs, such as the liver, spleen, and lungs (Fig. 5). Experiments using the 2 other cell lines (U251FM and HSJD-DIPG-007-FLUC) showed comparable results. Using these cell lines, no uptake of 89Zr-bevacizumab was visualized on PET in any of the xenografts (Fig. 3B). Ex vivo biodistribution analysis showed higher 89Zr-bevacizumab uptake in subcutaneous HSJD-DIPG-007-FLUC tumors (P < 0.05) as compared with brain (areas) without tumor (Fig. 4B). The subcutaneous U251FM tumors showed no significant increase in 89Zr-bevacizumab uptake (Fig. 4C) as compared with non-tumor brain. Of note, both HSJD-DIPG-007-FLUC and U251FM cells formed only very small subcutaneous tumors during the time window in which mice baring intracranial tumors had reached human endpoints (78 vs. 22 days after tumor injection).

Figure 2.

A, schematic overview of the experiment. Cells are injected at day 0 in the pons, striatum, or subcutis (early-stage tumors) or the striatum and subcutis (late-stage E98-FM tumors). Tumor progression was monitored by bioluminescent imaging. The time point for 89Zr-bevacizumab injection was dependent on the growth speed of used cell lines. This injection was done 72 to 96 hours before PET scanning. MRI, PET, or PET/CT imaging was followed by ex vivo measurement of 89Zr-bevacizumab accumulation in tissues (biodistribution). B, charged couple device (CCD) camera images of mice bearing E98-FM tumors. BLI images were obtained at the study endpoint, day 18 (early stage, top) and day 35 (late stage, bottom).

Figure 2.

A, schematic overview of the experiment. Cells are injected at day 0 in the pons, striatum, or subcutis (early-stage tumors) or the striatum and subcutis (late-stage E98-FM tumors). Tumor progression was monitored by bioluminescent imaging. The time point for 89Zr-bevacizumab injection was dependent on the growth speed of used cell lines. This injection was done 72 to 96 hours before PET scanning. MRI, PET, or PET/CT imaging was followed by ex vivo measurement of 89Zr-bevacizumab accumulation in tissues (biodistribution). B, charged couple device (CCD) camera images of mice bearing E98-FM tumors. BLI images were obtained at the study endpoint, day 18 (early stage, top) and day 35 (late stage, bottom).

Close modal
Figure 3.

A,89Zr-PET combined with 18F(-) imaging of mice with an E98 tumor located in the pons (top) or subcutaneous (bottom). White circles indicate expected location of intracranial tumors. While no uptake of 89Zr-bevacizumab is observed in the pontine tumor, the subcutaneous tumor shows a PET hotspot reflecting 89Zr-bevacizumab accumulation in the tumor (arrow). In all animals, high intensity (visualized by red color code) is seen in the heart and liver of the mice, reflecting 89Zr-bevacizumab in the blood pool. Images acquired with Siemens HRRT. B, immuno-PET scanning of mice with HSJD-DIPG-007-Fluc (top and middle) or U251FM tumors (bottom) confirmed the lack of bevacizumab uptake in intracranial tumors. Image acquired with MEDISO Nanoscan PET-CT system. White circles indicate expected location of intracranial tumors.

Figure 3.

A,89Zr-PET combined with 18F(-) imaging of mice with an E98 tumor located in the pons (top) or subcutaneous (bottom). White circles indicate expected location of intracranial tumors. While no uptake of 89Zr-bevacizumab is observed in the pontine tumor, the subcutaneous tumor shows a PET hotspot reflecting 89Zr-bevacizumab accumulation in the tumor (arrow). In all animals, high intensity (visualized by red color code) is seen in the heart and liver of the mice, reflecting 89Zr-bevacizumab in the blood pool. Images acquired with Siemens HRRT. B, immuno-PET scanning of mice with HSJD-DIPG-007-Fluc (top and middle) or U251FM tumors (bottom) confirmed the lack of bevacizumab uptake in intracranial tumors. Image acquired with MEDISO Nanoscan PET-CT system. White circles indicate expected location of intracranial tumors.

Close modal
Figure 4.

89Zr-bevacizumab measured ex vivo by a gamma-counter and normalized to counts found in healthy brain tissue (brain tissue of animals without xenografted brain tumors). A, uptake is significantly higher in the subcutaneous E98-FM tumors (***, P < 0.01), but there is no significant difference in uptake in pontine or striatal xenografts at any stage of the disease compared with normal brain. B, uptake in subcutaneous DIPG7-Fluc tumors is higher than in normal brain (*, P < 0.05). C, no significant differences between U251-FM tumors (subcutaneous or intracranial) versus normal brain.

Figure 4.

89Zr-bevacizumab measured ex vivo by a gamma-counter and normalized to counts found in healthy brain tissue (brain tissue of animals without xenografted brain tumors). A, uptake is significantly higher in the subcutaneous E98-FM tumors (***, P < 0.01), but there is no significant difference in uptake in pontine or striatal xenografts at any stage of the disease compared with normal brain. B, uptake in subcutaneous DIPG7-Fluc tumors is higher than in normal brain (*, P < 0.05). C, no significant differences between U251-FM tumors (subcutaneous or intracranial) versus normal brain.

Close modal
Figure 5.

89Zr-bevacizumab uptake measured ex vivo by radioactivity of the specific organs and the tumor after dissection, expressed as percentage of the injected dose per gram tissue in E98-FM tumor–bearing animals. Note the high uptake in subcutaneous tumors, compared with the negligible uptake in the intracerebral tumors.

Figure 5.

89Zr-bevacizumab uptake measured ex vivo by radioactivity of the specific organs and the tumor after dissection, expressed as percentage of the injected dose per gram tissue in E98-FM tumor–bearing animals. Note the high uptake in subcutaneous tumors, compared with the negligible uptake in the intracerebral tumors.

Close modal

ISH of VEGF

To determine whether the differences in 89Zr-bevacizumab uptake in the subcutaneous and the intracranial tumors were due to an impaired distribution of 89Zr-bevacizumab into the brain or to a differential expression of its target (or both), VEGF expression was analyzed in the different xenografts. ISH confirmed expression of VEGF in the subcutaneous E98FM tumor (Fig. 6A), whereas VEGF expression was absent in E98FM brain tumors (Fig. 6B) and brain tissue without a xenograft. Of note, in the subcutaneous tumors, VEGF was preferentially expressed in perinecrotic areas, whereas necrosis was absent in all of the striatal and pontine tumors. A Ki67 (Mib-1) staining was performed to confirm presence of proliferating tumor tissue (Fig. 6C and D). In HSJD-DIPG-007-FLUC and U251FM tumors, VEGF mRNA expression was not detectable in striatal and pontine gliomas. The subcutaneous tumors were too small to adequately perform VEGF ISH and therefore no conclusions could be drawn regarding VEGF expression in these subcutaneous tumors.

Figure 6.

VEGF expression in DIPG tumors as assessed by VEGF ISH. A, positive VEGF staining in subcutaneous E98-FM tumor cells (magnification, 20×, insert 2.5×). The tumor shows extensive necrosis surrounded by VEGF-positive cells. B, no VEGF mRNA is detected in the E98-FM tumor tissue in the striatum (magnification, 20×, insert 1.25×). C, Ki67 staining shows the presence of proliferating tumor cells in the subcutaneous tumor tissue (magnification 20×, insert 2.5×). D, Ki67-positive tumor cells in the striatal E98-FM tumor (magnification 20×, insert 10×).

Figure 6.

VEGF expression in DIPG tumors as assessed by VEGF ISH. A, positive VEGF staining in subcutaneous E98-FM tumor cells (magnification, 20×, insert 2.5×). The tumor shows extensive necrosis surrounded by VEGF-positive cells. B, no VEGF mRNA is detected in the E98-FM tumor tissue in the striatum (magnification, 20×, insert 1.25×). C, Ki67 staining shows the presence of proliferating tumor cells in the subcutaneous tumor tissue (magnification 20×, insert 2.5×). D, Ki67-positive tumor cells in the striatal E98-FM tumor (magnification 20×, insert 10×).

Close modal

MRI

To visualize disruption of the BBB, mice with HSJD-DIPG-007-FLUC and U251-FM intracranial tumors were imaged by MRI after intravenous administration of gadolinium. In the diffusely growing HSJD-DIPG-007-FLUC tumor, gadolinium enhancement on T1-weighed images was limited (Fig. 7A, arrow), whereas gadolinium enhancement was clearly visible in the U251-FM tumor (Fig. 7C, arrow). On T2-weighed images, tumors were not clearly visible (Fig. 7B and D; arrow).

Figure 7.

MRI scans of pontine tumors. The 89Zr-bevacizumab uptake in the tumors of these 2 mice is presented as %ID/g. A, limited gadolinium contrast enhancement in the tumor area (arrow) on T1 scan. B, HSJD-DIPG-007-FLUC tumor is poorly visible on T2-weighed MRI image (arrow). C, clear gadolinium contrast enhancement in the U251-FM tumor area (T1-weighed MRI scan) indicates disruption of the BBB. D, U251-FM tumor is visible on T2-weighed MRI scan.

Figure 7.

MRI scans of pontine tumors. The 89Zr-bevacizumab uptake in the tumors of these 2 mice is presented as %ID/g. A, limited gadolinium contrast enhancement in the tumor area (arrow) on T1 scan. B, HSJD-DIPG-007-FLUC tumor is poorly visible on T2-weighed MRI image (arrow). C, clear gadolinium contrast enhancement in the U251-FM tumor area (T1-weighed MRI scan) indicates disruption of the BBB. D, U251-FM tumor is visible on T2-weighed MRI scan.

Close modal

The potential benefit of bevacizumab in the treatment of DIPG is unclear, as efficacy depends on expression of VEGF-A as well as appropriate drug distribution (31). We used molecular PET imaging to study the influence of location and stage of disease on biodistribution of 89Zr-bevacizumab in 3 glioma mouse models (pontine, striatal, subcutaneous) using 3 different cell lines. The E98-FM pontine and striatal and HSJD-DIPG-007 pontine xenograft models have previously been described to resemble the diffuse phenotype of human DIPG and other diffuse HGGs (19, 32, 33). U251 has been described as an intracranial murine tumor model that recapitulates most of the key figures of adult glioblastoma (34). We found no significant uptake of 89Zr-bevacizumab in the intracranial tumor models at any stage of the disease, nor in the normal/nonneoplastic surrounding brain. In contrast, high accumulation of 89Zr-bevacizumab was observed in the subcutaneous E98-xenograft and moderate uptake in the subcutaneous HSJD-DIPG-007-FLUC.

We initially hypothesized that lack of 89Zr-bevacizumab uptake could be explained solely by poor distribution into the brain, as large molecules like monoclonal antibodies may not be able to pass the BBB. This hypothesis is supported by the absence of enhancement of the tumor on MRI after administration of gadolinium in animals with HSJD-DIPG-007-FLUC pontine tumors. However, MRI analysis of U251-FM tumors in the brainstem of mice showed clear gadolinium enhancement, which is indicative of “leaky” blood vessels in the tumor. Furthermore, VEGF expression of the E98FM glioma cells—analyzed by ISH—also differed between tumor locations: E98FM gliomas in both striatum and pons appeared VEGF-negative, whereas the subcutaneous E98FM tumors were partly VEGF-positive. The differences in VEGF expression of the tumors in distinct locations originating from the same cell line confirm that the orthotopic microenvironment and the resulting growth pattern significantly influence gene expression in glioma cells, a phenomenon that has been described previously (19, 20). Moreover, it has been shown that in glioblastoma, VEGF is predominantly overexpressed in hypoxic, perinecrotic cells (35, 36). Indeed, in our study, the VEGF expression in subcutaneous E98FM tumors was especially present around areas of necrosis, whereas in intracranial E98FM tumors, necrosis and VEGF expression were lacking and this also coincided with lack of bevacizumab uptake studied ex vivo and by PET. Of note, bevacizumab does not bind to murine VEGF-A (37), but as typically, the neoplastic cells are upregulating VEGF expression in tumor angiogenesis, we consider it unlikely that stromal cell–derived mouse VEGF-A plays an important role in this particular xenograft model (33).

In contrast to our preclinical findings, the in silico analysis that we performed in this study indicates that human DIPG tumors have relatively high expression levels of VEGF mRNA. However, the majority of tumors (23 of 27) used for the microarray experiments were collected post-mortem (38, 39) and therefore these samples represent the end stage of the disease and are post-radiation therapy. In the end stage of the disease, DIPG is known to have necrotic areas with microvascular proliferations and BBB disruption, compatible with the histology of a glioblastoma, which is associated with high VEGF expression. Although the numbers are low, it is important to point out that VEGF-A expression levels in samples obtained during pretreatment were low compared with expression in post-mortem/end-stage samples (Supplementary Fig. S2). In addition, biopsy samples that were included in the analysis are frequently directed at contrast-enhancing regions, and this “biased sampling” may well lead to overestimation of the role of VEGF in the tumor as a whole. We are currently studying the differences in VEGF-A expression in autopsy-derived DIPG tissue between the perinecrotic areas and the more diffusely growing tumor parts without necrosis.

Experimental and clinical research in both adult and pHGG and DIPG has suggested that there is a complex relation between a histologically diffuse growth pattern of brain tumors, VEGF expression, and availability and BBB integrity (20, 40–44). Traditionally, VEGF is viewed as the main cause of increased BBB permeability in central nervous system (CNS) tumors as represented by contrast-enhancing lesions on MRI (45). More recently, anti-VEGF therapy is thought to potentially induce a more diffuse and distant spread of tumor cells (41, 42). In contrast to adult HGGs, gadolinium contrast enhancement on MRI in DIPGs at diagnosis is generally limited, with 50% of the patients showing no enhancement at all (3). The lack of gadolinium enhancement suggests an intact BBB, at least for large molecules, in a substantial percentage of the patients, coinciding with low VEGF expression and inability of bevacizumab to target the tumor (46).

Results from our preclinical PET, MRI with gadolinium contrast, and ISH studies suggest that this relation is not so straightforward, pressing the need for studying VEGF targeting in patients treated with anti-VEGF therapy. One could however argue whether high local tumor accumulation of bevacizumab is at all needed to obtain potential therapeutic effects in DIPG. Bevacizumab is capable of decreasing VEGF levels in blood to undetectable range in less than 20 days in a large cohort of adult cancer patients (47), but only a subgroup of patients responded to anti-VEGF therapy. Also in DIPG, decreased phosphoVEGR2 levels in peripheral blood mononuclear cells (PBMC) did not correlate with treatment response (15). In the E98 xenograft model used in this study, treatment with bevacizumab did not increase survival, nor did it influence the growth pattern in the diffusely growing parts of the tumor (36). This suggest simply decreasing VEGF in the blood pool is, at least for the tumor types studied and our xenograft model, often insufficient for adequate tumor targeting and instead, local bevacizumab accumulation seems needed (47–49).

The results of immuno-PET imaging and VEGF-ISH in these DIPG models are in line with the poor clinical response rates thus far obtained with bevacizumab in children with DIPG (15). The data presented suggest that no adequate uptake of bevacizumab will occur in diffusely growing gliomas, which present with BBB disruption but without drastically increased VEGF expression. Therefore, we suggest that bevacizumab treatment is only justified if targeting of VEGF by bevacizumab has been visualized by immuno-PET scan. We aim to confirm this hypothesis in a clinical PET study with patients with DIPG.

Future directions

This study underlies the importance of using strong biological and biodistributional rationale before using any therapy in any patient. Following the results of this study, we developed a molecular drug imaging trial with 89Zr-bevacizumab in children with DIPG (study number NTR3518 www.trialregister.nl). This technique aims to further unravel the role of bevacizumab treatment in DIPG. Ideally, such molecular imaging is combined with VEGF-A and VEGFR2 expression analysis on tumor tissue originating from biopsies taken from several (contrast-enhancing and nonenhancing) parts of the tumor. Because DIPG can generally be diagnosed on the basis of its typical radiological presentation and the delicate nature of the brain involved, taking biopsies from DIPGs is still no common practice and sampling of multiple regions is even more cumbersome. In general, integrating molecular imaging with radiolabeled drugs (classic cytostatic agents, small molecules, other monoclonal antibodies) in the treatment of childhood brain cancer provides an insight in drug targeting and might help personalize treatment and thereby to avoid unnecessary side effects of drugs that do not reach the tumor.

No potential conflicts of interest were disclosed.

Conception and design: M.H.A. Jansen, T. Lagerweij, A.C.P. Sewing, D.G. van Vuurden, G.A.M.S. van Dongen, G.J.L. Kaspers, E. Hulleman

Development of methodology: M.H.A. Jansen, T. Lagerweij, A.C.P. Sewing, V. Caretti, G.A.M.S. van Dongen

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): M.H.A. Jansen, T. Lagerweij, A.C.P. Sewing, D.J. Vugts, C.F.M. Molthoff, V. Caretti, N. Petersen, A.M. Carcaboso, D.P. Noske, P. Wesseling

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): M.H.A. Jansen, T. Lagerweij, A.C.P. Sewing, D.J. Vugts, C.F.M. Molthoff, S.J.E. Veringa, P. Wesseling, G.A.M.S. van Dongen

Writing, review, and/or revision of the manuscript: T. Lagerweij, A.C.P. Sewing, D.J. Vugts, D.G. van Vuurden, C.F.M. Molthoff, A.M. Carcaboso, D.P. Noske, W.P. Vandertop, P. Wesseling, G.A.M.S. van Dongen, G.J.L. Kaspers, E. Hulleman

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): M.H.A. Jansen, T. Lagerweij

Study supervision: G.J.L. Kaspers, E. Hulleman

We are thankful to Ricardo Vos, Iris Mes and Alex Poot (Department of Radiology & Nuclear Medicine, VUmc) for technical support with labeling of PET tracers and PET imaging and Piotr Waranecki and Yanyan Veldman (Neuro-oncology Research Group, VUmc) for technical assistance and animal handling. DIPG research at the VUmc has been made possible by the invaluable support of the Semmy Foundation, Stichting Egbers and of Stichting Kika (Children-Cancer-free, project 69).

E. Hulleman received funding from Semmy Foundation and Stichting Kika (Children-Cancer-free, project 69). M.H. Jansen received funding from Stichting Egbers. A.M. Carcaboso acknowledges funding from ISCIII-FEDER (CP13/00189).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1.
Hargrave
D
,
Bartels
U
,
Bouffet
E
. 
Diffuse brainstem glioma in children: critical review of clinical trials
.
Lancet Oncol
2006
;
7
:
241
8
.
2.
Jansen
MHA
,
van Vuurden
DG
,
Vandertop
WP
,
Kaspers
GJL
. 
Diffuse intrinsic pontine gliomas: a systematic update on clinical trials and biology
.
Cancer Treat Rev
2012
;
38
:
27
35
.
3.
Hargrave
D
,
Chuang
N
,
Bouffet
E
. 
Conventional MRI cannot predict survival in childhood diffuse intrinsic pontine glioma
.
J Neurooncol
2008
;
86
:
313
9
.
4.
Jansen
MH
,
Veldhuijzen van Zanten
SE
,
Sanchez Aliaga
E
,
Heymans
MW
,
Warmuth-Metz
M
,
Hargrave
D
, et al
Survival prediction model of children with diffuse intrinsic pontine glioma based on clinical and radiological criteria
.
Neuro Oncol
2015
;
17
:
160
6
.
5.
Ferrara
N
. 
Vascular endothelial growth factor: basic science and clinical progress
.
Endocr Rev
2004
;
25
:
581
611
.
6.
Shibuya
M
. 
Differential roles of vascular endothelial growth factor receptor-1 and receptor-2 in angiogenesis
.
J Biochem Mol Biol
2006
;
39
:
469
78
.
7.
Friedman
HS
,
Prados
MD
,
Wen
PY
,
Mikkelsen
T
,
Schiff
D
,
Abrey
LE
, et al
Bevacizumab alone and in combination with irinotecan in recurrent glioblastoma
.
J Clin Oncol
2009
;
27
:
4733
40
.
8.
Vredenburgh
JJ
,
Desjardins
A
,
Herndon
JE
,
Dowell
JM
,
Reardon
DA
,
Quinn
JA
, et al
Phase II trial of bevacizumab and irinotecan in recurrent malignant glioma
.
Clin Cancer Res
2007
;
13
:
1253
9
.
9.
Vredenburgh
JJ
,
Desjardins
A
,
Herndon
JE
,
Marcello
J
,
Reardon
DA
,
Quinn
JA
, et al
Bevacizumab plus irinotecan in recurrent glioblastoma multiforme
.
J Clin Oncol
2007
;
25
:
4722
9
.
10.
Chinot
OL
,
Wick
W
,
Mason
W
,
Henriksson
R
,
Saran
F
,
Nishikawa
R
, et al
Bevacizumab plus radiotherapy-temozolomide for newly diagnosed glioblastoma
.
N Engl J Med
2014
;
370
:
709
22
.
11.
Gilbert
MR
,
Dignam
JJ
,
Armstrong
TS
,
Wefel
JS
,
Blumenthal
DT
,
Vogelbaum
MA
, et al
A randomized trial of bevacizumab for newly diagnosed glioblastoma
.
N Engl J Med
2014
;
370
:
699
708
.
12.
Khasraw
M
,
Ameratunga
MS
,
Grant
R
,
Wheeler
H
,
Pavlakis
N
. 
Antiangiogenic therapy for high-grade glioma
.
Cochrane Database Syst Rev
2014
;
9
:
CD008218
.
13.
Narayana
A
,
Kunnakkat
S
,
Chacko-Mathew
J
,
Gardner
S
,
Karajannis
M
,
Raza
S
, et al
Bevacizumab in recurrent high-grade pediatric gliomas
.
Neuro Oncol
2010
;
12
:
985
90
.
14.
Parekh
C
,
Jubran
R
,
Erdreich-Epstein
A
,
Panigrahy
A
,
Bluml
S
,
Finlay
J
, et al
Treatment of children with recurrent high grade gliomas with a bevacizumab containing regimen
.
J Neurooncol
2011
;
103
:
673
80
.
15.
Gururangan
S
,
Chi
SN
,
Young Poussaint
T
,
Onar-Thomas
A
,
Gilbertson
RJ
,
Vajapeyam
S
, et al
Lack of efficacy of bevacizumab plus irinotecan in children with recurrent malignant glioma and diffuse brainstem glioma: a pediatric brain tumor consortium study
.
J Clin Oncol
2010
;
28
:
3069
75
.
16.
Aguilera
DG
,
Mazewski
C
,
Hayes
L
,
Jordan
C
,
Esiashivilli
N
,
Janns
A
, et al
Prolonged survival after treatment of diffuse intrinsic pontine glioma with radiation, temozolamide, and bevacizumab: report of 2 cases
.
J Pediatr Hematol Oncol
2013
;
35
:
e42
6
.
17.
Zaky
W
,
Wellner
M
,
Brown
RJ
,
Blüml
S
,
Finlay
JL
,
Dhall
G
. 
Treatment of children with diffuse intrinsic pontine gliomas with chemoradiotherapy followed by a combination of temozolomide, irinotecan, and bevacizumab
.
Pediatr Hematol Oncol
2013
;
30
:
623
32
.
18.
Hummel
TR
,
Salloum
R
,
Drissi
R
,
Kumar
S
,
Sobo
M
,
Goldman
S
, et al
A pilot study of bevacizumab-based therapy in patients with newly diagnosed high-grade gliomas and diffuse intrinsic pontine gliomas
.
J Neurooncol
2016
;
127
:
53
61
.
19.
Caretti
V
,
Zondervan
I
,
Meijer
DH
,
Idema
S
,
Vos
W
,
Hamans
B
, et al
Monitoring of tumor growth and post-irradiation recurrence in a diffuse intrinsic pontine glioma mouse model
.
Brain Pathol
2011
;
21
:
441
51
.
20.
Roodink
I
,
van der Laak
J
,
Kusters
B
,
Wesseling
P
,
Verrijp
K
,
de Waal
R
, et al
Development of the tumor vascular bed in response to hypoxia-induced VEGF-A differs from that in tumors with constitutive VEGF-A expression
.
Int J Cancer
2006
;
119
:
2054
62
.
21.
Verel
I
,
Visser
GWM
,
Boellaard
R
,
Stigter-van Walsum
M
,
Snow
GB
,
van Dongen
GAMS
. 
89Zr immuno-PET: comprehensive procedures for the production of 89Zr-labeled monoclonal antibodies
.
J Nucl Med
2003
;
44
:
1271
81
.
22.
Taylor
KR
,
Mackay
A
,
Truffaux
N
,
Butterfield
YS
,
Morozova
O
,
Philippe
C
, et al
Recurrent activating ACVR1 mutations in diffuse intrinsic pontine glioma
.
Nat Genet
2014
;
46
:
457
61
.
23.
Bernsen
HJ
,
Rijken
PF
,
Peters
H
,
Bakker
H
,
van der Kogel
AJ
. 
The effect of the anti-angiogenic agent TNP-470 on tumor growth and vascularity in low passaged xenografts of human gliomas in nude mice
.
J Neurooncol
1998
;
38
:
51
7
.
24.
Mir
SE
,
De Witt Hamer
PC
,
Krawczyk
PM
,
Balaj
L
,
Claes
A
,
Niers
JM
, et al
In silico analysis of kinase expression identifies WEE1 as a gatekeeper against mitotic catastrophe in glioblastoma
.
Cancer Cell
2010
;
18
:
244
57
.
25.
de Weger
RA
,
Tilanus
MG
,
Scheidel
KC
,
van den Tweel
JG
,
Verdonck
LF
. 
Monitoring of residual disease and guided donor leucocyte infusion after allogeneic bone marrow transplantation by chimaerism analysis with short tandem repeats
.
Br J Haematol
2000
;
110
:
647
53
.
26.
Grasso
CS
,
Tang
Y
,
Truffaux
N
,
Berlow
NE
,
Liu
L
,
Debily
M-A
, et al
Functionally defined therapeutic targets in diffuse intrinsic pontine glioma
.
Nat Med
2015
;
21
:
555
9
.
27.
Paxinos
GKF
. 
The mouse brain in stereotaxic coordinates, compact, 3rd Edition | George Paxinos, Keith Franklin | ISBN 9780123742445 [Internet]
; 
2008
.
Available from
: http://store.elsevier.com/product.jsp?isbn=9780123742445&pagename=search.
28.
Verel
I
,
Visser
GWM
,
Boellaard
R
,
Boerman
OC
,
van Eerd
J
,
Snow
GB
, et al
Quantitative 89Zr immuno-PET for in vivo scouting of 90Y-labeled monoclonal antibodies in xenograft-bearing nude mice
.
J Nucl Med
2003
;
44
:
1663
70
.
29.
de Jong
HWAM
,
van Velden
FHP
,
Kloet
RW
,
Buijs
FL
,
Boellaard
R
,
Lammertsma
AA
. 
Performance evaluation of the ECAT HRRT: an LSO-LYSO double layer high resolution, high sensitivity scanner
.
Phys Med Biol
2007
;
52
:
1505
26
.
30.
Loening
AM
,
Gambhir
SS
. 
AMIDE: a free software tool for multimodality medical image analysis
.
Mol Imaging
2003
;
2
:
131
7
.
31.
Papadopoulos
N
,
Martin
J
,
Ruan
Q
,
Rafique
A
,
Rosconi
MP
,
Shi
E
, et al
Binding and neutralization of vascular endothelial growth factor (VEGF) and related ligands by VEGF Trap, ranibizumab and bevacizumab
.
Angiogenesis
2012
;
15
:
171
85
.
32.
Boult
JKR
,
Taylor
KR
,
Vinci
M
,
Popov
S
,
Jury
A
,
Molinari
V
, et al
Abstract 3271: Novel orthotopic pediatric high grade glioma xenografts evaluated with magnetic resonance imaging mimic human disease
.
Cancer Res
2015
;
75
:
3271
3271
.
33.
Claes
A
,
Schuuring
J
,
Boots-Sprenger
S
,
Hendriks-Cornelissen
S
,
Dekkers
M
,
van der Kogel
AJ
, et al
Phenotypic and genotypic characterization of orthotopic human glioma models and its relevance for the study of anti-glioma therapy
.
Brain Pathol
2008
;
18
:
423
33
.
34.
Jacobs
VL
,
Valdes
PA
,
Hickey
WF
,
De Leo
JA
. 
Current review of in vivo GBM rodent models: emphasis on the CNS-1 tumour model
.
ASN Neuro
2011
;
3
:
e00063
.
35.
Johansson
M
,
Brännström
T
,
Bergenheim
AT
,
Henriksson
R
. 
Spatial expression of VEGF-A in human glioma
.
J Neurooncol
2002
;
59
:
1
6
.
36.
Navis
AC
,
Hamans
BC
,
Claes
A
,
Heerschap
A
,
Jeuken
JWM
,
Wesseling
P
, et al
Effects of targeting the VEGF and PDGF pathways in diffuse orthotopic glioma models
.
J Pathol
2011
;
223
:
626
34
.
37.
Yu
L
,
Wu
X
,
Cheng
Z
,
Lee
CV
,
LeCouter
J
,
Campa
C
, et al
Interaction between bevacizumab and murine VEGF-A: a reassessment
.
Invest Ophthalmol Vis Sci
2008
;
49
:
522
7
.
38.
Paugh
BS
,
Broniscer
A
,
Qu
C
,
Miller
CP
,
Zhang
J
,
Tatevossian
RG
, et al
Genome-wide analyses identify recurrent amplifications of receptor tyrosine kinases and cell-cycle regulatory genes in diffuse intrinsic pontine glioma
.
J Clin Oncol
2011
;
29
:
3999
4006
.
39.
Paugh
BS
,
Qu
C
,
Jones
C
,
Liu
Z
,
Adamowicz-Brice
M
,
Zhang
J
, et al
Integrated molecular genetic profiling of pediatric high-grade gliomas reveals key differences with the adult disease
.
J Clin Oncol
2010
;
28
:
3061
8
.
40.
Zhao
L
,
Yang
Z
,
Liu
Y
,
Ying
H
,
Zhang
H
,
Xue
Y
. 
Vascular endothelial growth factor increases permeability of the blood-tumor barrier via caveolae-mediated transcellular pathway
.
J Mol Neurosci
2011
;
44
:
122
9
.
41.
Pàez-Ribes
M
,
Allen
E
,
Hudock
J
,
Takeda
T
,
Okuyama
H
,
Viñals
F
, et al
Antiangiogenic therapy elicits malignant progression of tumors to increased local invasion and distant metastasis
.
Cancer Cell
2009
;
15
:
220
31
.
42.
Salloum
R
,
DeWire
M
,
Lane
A
,
Goldman
S
,
Hummel
T
,
Chow
L
, et al
Patterns of progression in pediatric patients with high-grade glioma or diffuse intrinsic pontine glioma treated with Bevacizumab-based therapy at diagnosis
.
J Neurooncol
2015
;
121
:
591
8
.
43.
Gomez-Manzano
C
,
Holash
J
,
Fueyo
J
,
Xu
J
,
Conrad
CA
,
Aldape
KD
, et al
VEGF Trap induces antiglioma effect at different stages of disease
.
Neuro Oncol
2008
;
10
:
940
5
.
44.
Wesseling
P
,
Ruiter
DJ
,
Burger
PC
. 
Angiogenesis in brain tumors; pathobiological and clinical aspects
.
J Neurooncol
1997
;
32
:
253
65
.
45.
Gagner
J-P
,
Law
M
,
Fischer
I
,
Newcomb
EW
,
Zagzag
D
. 
Angiogenesis in gliomas: imaging and experimental therapeutics
.
Brain Pathol
2005
;
15
:
342
63
.
46.
Jansen
MHA
,
Kaspers
GJ
. 
A new era for children with diffuse intrinsic pontine glioma: hope for cure?
Expert Rev Anticancer Ther
2012
;
12
:
1109
12
.
47.
Jubb
AM
,
Harris
AL
. 
Biomarkers to predict the clinical efficacy of bevacizumab in cancer
.
Lancet Oncol
2010
;
11
:
1172
83
.
48.
Jubb
AM
. 
Impact of vascular endothelial growth factor-A expression, thrombospondin-2 expression, and microvessel density on the treatment effect of bevacizumab in metastatic colorectal cancer
.
J Clin Oncol
2005
;
24
:
217
27
.
49.
Shih
T
,
Lindley
C
. 
Bevacizumab: an angiogenesis inhibitor for the treatment of solid malignancies
.
Clin Ther
2006
;
28
:
1779
802
.