In vivo tumour imaging employing regional delivery of novel gallium radiolabelled polymer composites

Background Understanding the regional vascular delivery of particles to tumour sites is a prerequisite for developing new diagnostic and therapeutic composites for treatment of oncology patients. We describe a novel imageable 67Ga-radiolabelled polymer composite that is biocompatible in an animal tumour model and can be used for preclinical imaging investigations of the transit of different sized particles through arterial networks of normal and tumour-bearing organs. Results Radiolabelling of polymer microspheres with 67Ga was achieved using a simple mix and wash method, with tannic acid as an immobilising agent. Final in vitro binding yields after autoclaving averaged 94.7%. In vivo stability of the composite was demonstrated in New Zealand white rabbits by intravenous administration, and intrahepatic artery instillations were made in normal and VX2 tumour implanted rabbit livers. Stability of radiolabel was sufficient for rabbit lung and liver imaging over at least 3 hours and 1 hour respectively, with lung retention of radiolabel over 91%, and retention in both normal and VX2 implanted livers of over 95%. SPECT-CT imaging of anaesthetised animals and planar imaging of excised livers showed visible accumulation of radiolabel in tumours. Importantly, microsphere administration and complete liver dispersal was more easily achieved with 8 μm diameter MS than with 30 μm MS, and the smaller microspheres provided more distinct and localised tumour imaging. Conclusion This method of producing 67Ga-radiolabelled polymer microspheres is suitable for SPECT-CT imaging of the regional vascular delivery of microspheres to tumour sites in animal models. Sharper distinction of model tumours from normal liver was obtained with smaller MS, and tumour resolution may be further improved by the use of 68Ga instead of 67Ga, to enable PET imaging.

Background Understanding the regional vascular delivery of particles to tumour sites is a prerequisite for developing new diagnostic and therapeutic composites for treatment of oncology patients. Clearly this is important in order to predict organ distribution prior to any treatment with particles carrying cytotoxic radioisotopes or toxins. This applies especially to radioisotope treatments with longer half-lives emitting relatively high energy beta emission, such as in selective internal radiation therapy (SIRT) of metastatic colorectal tumours present in the liver [1][2][3].
We have previously demonstrated liver tumour imaging in animal models using cationised carbon nanoparticles, comprised of a core of 99m Tc encapsulated in graphitic carbon, as a stable electrostatic radiolabel for polymer microspheres (MS) [4]. However, the Gallium isotopes already in clinical use have significant advantages, offering both single-photon emission computed tomography (SPECT) imaging with 67 Ga and higher resolution positron emission tomography (PET) imaging with 68 Ga. Gallium-67 has a physical half-life of 78.3 h and decays by electron capture, emitting gamma radiation suitable for SPECT imaging. The soluble citrate salt of 67 Ga is currently used clinically as an intravenous injection for imaging the extent of Hodgkin's disease, lymphomas and bronchogenic carcinoma [5]. Gallium-68 is a positron emitter with a 1.13 h half-life, but despite this short half-life, the radioisotope can be very useful for PET imaging when produced at the point of use from a 68 Ge/ 68 Ga generator [6][7][8].
In this report we describe a novel composite comprising 67 Ga immobilised on a polymer surface with tannic acid (TA), that has favourable stability in vitro and in vivo for preclinical animal imaging studies. Using SPECT-CT of intact animals, and planar imaging of excised livers, we demonstrate accumulation of MS at tumour sites in livers following arterial administration. We show that tumour accumulation is related to MS size, with smaller 8 μm MS producing superior tumour imaging to 30 μm MS.

Synthesis of 67 GaTA-MS and in vitro stability to autoclaving
Polystyrene sulfonate MS of the same type as used in clinical SIRT [2] were labelled with the radioisotope 67 Ga were prepared in a simple, convenient, mix and wash method utilising TA as a stable complexing agent (Fig. 1A). With the strong anion exchange properties of the polystyrene sulfonate, it was straight forward to bind either the chloride or citrate salts of 67 Ga in high yields, by mixing with the polymer microsphere in the presence of HCl (0.2 M final concentration). The MS were mixed with the radioisotope at room temperature and then washed 3 times with water using centrifugation to collect the MS. The radioisotope was then fixed by mixing with TA (3 mM) at room temperature for 1 h, followed by a further washing step. The MS could then be autoclaved under standard conditions. The bound radioactivity on the MS following each of the binding, fixing and autoclaving steps was high (Fig. 1B), with final yields after autoclaving averaging 94.7% for both 8 and 30 μm diameter MS. Note that this process can also be applied to other metal radioisotopes such as 111 Indium, 201 Thallium, 177 Lutetium, and 90 Yttrium, with high binding yields (Fig. 1C, unoptimised yields).
Leach tests of 67 GaTA-MS stability by entrapment in normal rabbit lungs The in vivo stability of the 67 GaTA-MS preparations was tested by intravenous injection into the lungs of rabbits. Using less than 5 mg MS per kg of body weight, there was no apparent change in breathing over the 3 h postinjection test period. Suspensions of 67 GaTA-MS30 and 67 GaTA-MS8 were retained well in the vascular network of the lungs over 3 h, as shown by gamma camera imaging of the anaesthetised animals ( Fig. 2A and B, respectively), compared with intravenous injection of soluble 67 Ga (Fig. 2C). After 3 h the animals were euthanised, and the lungs, livers and body were imaged separately for more quantitative radioactivity biodistribution data (Table 1). For 67 GaTA-MS30, the average proportion of radioactivity retained in the lungs was 97.3 ± 0.4% of the total body activity, and only 0.61 ± 0.1% was present in the excised liver. For 67 GaTA-MS8, the average proportion of radioactivity retained in the lungs was slightly lower at 91.1 ± 0.4% of the total body activity, and with 1.5 ± 0.3% present in the liver. For comparison, rabbits injected with soluble 67 Ga had low lung and liver radioactivity with only 5.3 and 9.9%, respectively, of the total body radioactivity ( Fig. 2C and Table 1).

Biodistribution of 67 GaTA-MS in normal rabbit livers
Suspensions of both the 67 GaTA-MS30 and 67 GaTA-MS8 were administered to rabbit livers via the hepatic artery. Some administration failures were encountered with the larger 67 GaTA-MS30, when the MS remained predominantly at a central part of the liver and did not disperse widely in the organ. There were no administration failures when using the smaller 67 GaTA-MS8. Planar gamma imaging of anaesthetised animals showed virtually complete retention of the radiolabel in the liver, and after 1 h following administration the animals were euthanised and the lungs and liver were dissected to obtain more quantitative radioactivity biodistribution ( Table 2). Normal rabbit liver radioactivity when instilled with the 67 GaTA-MS30 composites was an average 98.9 ± 0.3% of the total body radioactivity, and lung activity was only 0.07 ± 0.03%. With 67 GaTA-MS8, the mean liver activity was slightly lower at 95.2 ± 2.4%, and the lung activity 0.57 ± 0.35% of the total body radioactivity. In comparison, soluble 67 Ga displayed no appreciable affinity for either the liver or lungs, with mean radioactivity of 15.4 and 4.7%, respectively, of the total body radioactivity. Images of the excised livers revealed a pronounced difference in the distribution for the 2 different diameters of MS, 67 GaTA-MS30 and 67 GaTA-MS8 ( Fig. 3A and B, respectively). Livers instilled with 67 GaTA-MS30 did not have radioactivity extending throughout the organ, but instead it appeared to have been arrested at limiting diameters of the arterial network, displaying a course tree-like structure. The image shown in Fig. 3A was the best liver dispersal obtained for 67 GaTA-MS30. In contrast, the smaller diameter 67 GaTA-MS8 routinely dispersed well within the liver and displayed a much finer liver distribution (Fig. 3B), being able to disperse to the finer arterial vessels throughout the whole tissue, while still not being able to pass through the microcapillaries to the venous collection. The soluble 67 Ga instilled intra-arterially in the liver displayed no appreciable retention in the liver, and was distributed systemically throughout the rabbit ( Fig.  3C and Table 2).

Biodistribution of 67 GaTA-MS in rabbit livers with implanted VX2 tumours
Implantation of rabbit livers with small pieces of VX2 tumours was performed using sterile keyhole surgery on a single site on one lobe, and after 18-21 days of growth, tumours appeared macroscopically on dissection as a single oblate ellipsoid of up to 2 cm diameter. Intraarterial instillations of 67 GaTA-MS30 and 67 GaTA-MS8 were then made in these rabbit livers hosting grown VX2 tumour implants. SPECT-CT imaging of the intact animals showed good retention of both sizes of MS in the liver (Fig. 4). Within the liver, 67 GaTA-MS30 showed isotope activity mainly in the right lobe hosting the  Fig. 4A. However, there was significant activity in the rest of the liver, as seen in the right side of the coronal image and in the right side of the transaxial image in Fig. 4A. Importantly, using the smaller 67 GaTA-MS8, accumulation of isotope appeared more localised to the single tumour site in the implanted lobe, as seen in the corresponding images of Fig. 4B. This result was confirmed by excising the livers from the same rabbits and making planar images, when it was apparent that while both 67 GaTA-MS30 and 67 GaTA-MS8 displayed a noticeable accumulation of the label at the tumour site ( Fig. 5A and B respectively), the smaller diameter 67 GaTA-MS8 had a more pronounced tumour accumulation, appearing in planar imaging as a sharply defined circular feature of high activity (Fig. 5B). In lobes that did not contain the VX2 tumour, 67 GaTA-MS30 displayed a coarse tree-like structure for the distribution of the radiolabel, and some normal liver tissue areas had significant radioactivity (Fig. 5A). By contrast, livers instilled with the smaller 67 GaTA-MS8 had lower intensities of radioactivity in normal lobes Fig. 5B), presumably due to volumetric dilution by more extensive dispersal throughout the tissue via finer blood vessels. Importantly, this was not accompanied by escape of the 67 GaTA-MS8 from VX2 implanted livers into the systemic circulation; 1 h following intraarterial instillation 95.2 ± 2.4% of the total body radioactivity was still present in the liver ( Table  2). By contrast, instillation of soluble 67 Ga did not produce any discernible liver or tumour retention, with only 4.5 ± 1.0% of the total body radioactivity remaining in the organ ( Fig. 5C and Table 2).

Discussion
This article describes a novel method for radiolabelling polymer MS with Gallium isotopes, employing a simple,  Gallium isotopes can be considered the preferred choice for making imageable polymer particles that can be used to predict biodistribution of the corresponding therapeutic particles in organs hosting tumours. Gallium-67 citrate with its relatively long half-life of 78.3 h is already a current clinical product for SPECT imaging of patients with Hodgkin's disease [5], lymphomas and bronchogenic carcinoma. In our method, by simply adjusting the pH with 1 M HCl, the 67 Ga 3+ ions could be dissociated from the citrate complex and bound readily to polystyrene sulfonate MS, which has the properties of a strong cation exchange resin. A stable radioisotope complex is then simply formed by addition of TA, and the method does not require the use of any special equipment. Clearly, it would be straight-forward to adapt this method to 68 Ga eluted from a commercially available germanium-68/Gallium-68 generator. The radioisotope eluted with hydrochloric acid could be directly bound to the MS in a single step enabling PET imaging with the final radiolabelled MS. Furthermore, we have applied this method to other metal radioisotopes such as 111 Indium, 201 Thallium, 177 Lutetium, and 90 Yttrium, with high binding yields.
Tannins, a class of astringent, polyphenolic biomolecules, are major constituents of our beverages and food stuffs [9], so they are known to be biocompatible. The metal ion chelation properties of TA are well known, and the phenolic-metal ion complexes are often coloured [10]. It is believed that the highly stable complexes of Fe (III) with TA are not bioavailable [9], and in   (5 mL) were instilled into the hepatic artery of anaesthetised rabbits. Instillation of soluble 67 Ga (avg. 110 MBq) was also performed to provide a comparison of the biodistribution. The rabbits were euthanized after 1 h for dissection and gamma camera imaging of lungs, liver, and body. The activities of all images were corrected for background and the activity of each organ expressed as a percentage of the total activity. Note pronounced liver retention of radiolabel using both 67 GaTA-MS30 and 67 GaTA-MS8 compared to liver retention of soluble 67 Ga. Retention of microsphere radiolabel in livers with VX2 tumour implants was indistinguishable from retention by normal livers Fig. 5 Planar imaging of excised livers with VX2 tumours. The same rabbits shown in Fig. 4A and B above were dissected to enable planar imaging of their livers. As in Fig. 3 above for normal livers, note the prominent vessel features and less extensive volume filling of the liver obtained with 67 GaTA-MS30 (a), compared to the more extensively dispersed fine network obtained with 67 GaTA-MS8 (b). Importantly, while the tumour site in (a) was identifiable, having increased activity compared to the rest of the liver, the use of 67 GaTA-MS8 in (b) revealed a more sharply distinct tumour site of high activity, with only weak activity in the rest of the liver. For comparison, the excised liver from a rabbit instilled with soluble 67 Ga citrate showed extensive loss of isotope from the organ, and although a tumour was present in this liver, there was no appreciable tumour retention of isotope (c) , but significant activity also apparent in the rest of the liver. In (b), the 67 GaTA-MS8 shows more selective accumulation of radioactivity at the tumour site in the implanted lobe, with only low activity in the rest of the liver fact consumption of large quantities of tannin-rich foods is sometimes associated with iron-deficiency anemia [11].
Free Gallium in the blood acts as an iron analogue, binding to the plasma protein transferrin. A strong interaction between Gallium and TA is therefore likely and important for minimising unwanted organ uptake by interfering with interactions of Gallium with plasma proteins. As was observed for a patient treated with deferoxamine prior to 67 Ga-citrate scintigraphy demonstrating an altered biodistribution of Gallium [12].
Previously described methods for immobilising radioisotopes such as 67 Ga have involved the preparation of covalently linked macrocyclic ligands such as DOTA and NOTA. While DOTA is the mostly used chelator, the smaller, tri-chelator, NOTA was found to bind Gallium better [13,14]. In this method the TA is not covalently bound to the MS, but instead forms a stable complex involving both the Gallium ion and the polystyrene sulfonate surface.
This complex is stable to autoclave conditions and the radiolabelled MS were stable for 3 h in vivo following intravenous delivery to the lungs of New Zealand white rabbits. Furthermore, hepatic artery administration of the 67 GaTA-MS showed very efficient retention of the radiolabel at limiting diameters of the liver arterial network. As was observed for the Tc99m labelled MS that we described earlier [4], the smaller diameter 67 GaTA-MS8 displayed a much finer liver distribution, dispersing more extensively into the vascular network than the larger 67 GaTA-MS30, which tended to define a jagged, tree-like pattern of the larger vessels.
For rabbits with implanted VX2 tumours, there was no significant evidence of hepatopulmonary or hepatogastric shunting, and retention of radiolabel within livers hosting a tumour was maintained at a very high level. Images of livers implanted with VX2 tumours demonstrated that the tumours had remodelled the local arterial supply, producing a bias of distribution and accumulation of radiolabelled MS in tumours, as seen before in computed tomography studies [15]; this was evident for both sizes of MS tested. However, the smaller MS produced a more defined image of tumours, presumably due to its ability to penetrate further into the plexus of abnormally tortuous angiogenic vessels in the tumour microenvironment [16], while still being large enough not to transit to the venous collection and enter the systemic circulation. Our study of 8 μm MS extends further downward the range of MS size reported previously [17], when more extensive imaging of VX2 vascular networks was obtained with 30 μm than with 100 μm MS. It is evident that the tumour plexus, consistent with the observed increase in flow resistance [18], tends to behave as a labyrinthine trap that limits transit of non-deformable polymer particles, even those comparable in size to deformable blood cells. Blood cells, especially erythrocytes, are well known [19] to traverse capillaries by deforming to a smaller crosssection, whereas non-deformable particles of similar size to blood cells will not traverse capillaries.

Conclusion
We have now demonstrated tumour imaging results for different sizes of 2 types of polymer MS preparations, radiolabelled by the use of 2 completely different labelling chemistries and 2 different radioisotopes, i.e. 99m Tc-nanoparticle radiolabelled MS [4] and 67 GaTA-MS. The consistent message is that smaller MS, even down to 8 μm diameter, produce better tumour definition in imaging, and without appreciable leakage to the systemic circulation. The increased ability of smaller radiolabelled MS to highlight tumours in imaging is informative for scoping the location, size and number of tumours in an organ, and potentially also useful as an indicator of reduced tumour perfusion achieved by angiogenesis inhibitors. All these results assist in evaluating accessibility of solid tumours to therapeutic MS and the likelihood of achieving a therapeutic impact. Considerable further work is required however at the preclinical level to specifically address the biocompatibility, toxicology and clearance of this novel composite.

Stable labelling of MS with 67 Ga
Polystyrene sulfonate MS of the same type as used in clinical SIRT [2] and of two different sizes were used in this study; 30 μm (MS30), and 8 μm (MS8) median diameter (Sirtex Medical Ltd., Sydney). The radioisotope 67 Ga was obtained as an aqueous citrate salt from Global Medical Solutions Australia. USP grade TA was purchased from Sigma-Aldrich (Sydney, Australia). Radiolabelled 67 Ga MS were prepared as previously described [20] and summarised in Fig. 1A. Briefly, MS (15-40 mg) were first washed 3 times with water (5 mL) using centrifugation at 350 g for 2 min. The MS were then mixed for 1 h with 67 Ga-citrate and 1 M HCl (1 mL) in a final volume of 5 mL water. After washing 3 times using centrifugation (350 g, 2 min), the MS were then resuspended in 3 mM TA (5 mL) for 1 h. After 3 further washes with water, the MS were resuspended in water (3 mL) and autoclaved at 120°C for 20 min to give the radiolabelled product.
Leach tests of 67 Ga labelled MS ( 67 GaTA-MS) in normal rabbit lungs All rabbit procedures adhered to the National Health and Medical Research Council's Australian code for the care and use of animals for scientific purposes (Australian Government, 8th Ed., 2013), and the experimental protocols were approved by the Australian National University (ANU) Animal Ethics Committee. Rabbit imaging studies were carried out using intubation to deliver ventilation anaesthesia with isoflurane, so that biodistribution of the radiolabel in live animals could be followed for up to 3 h. Imaging of the anaesthetised rabbits, and their excised organs after 3 h, was performed with a Hawkeye Infinia SPECT-CT camera (GE Healthcare). Images were obtained using a medium energy collimator with energy emission windows of 93 ± 13, 184 ± 10, and 300 ± 10 keV, and Compton scatter correction windows of 75 ± 7, and 140 ± 10 keV. To test the lung retention of the radiolabelled MS, suspensions of 67 GaTA-MS30 and 67 GaTA-MS8 (avg. 87 MBq on 15 mg MS in 5 mL 5% dextrose) were injected intravenously into an ear vein, so that the MS were mechanically arrested at limiting diameters in the arterial network of the lungs. Static 5 min acquisitions were made on a 1024 × 1024 matrix, approx. 10 min after injection of the MS and again every hour up to 3 h, before the rabbits were euthanized by lethal injection while still under anaesthesia. The lungs and liver were then excised after tying off blood vessels to prevent leakage of the radioisotope, and the excised organs were imaged separately using a 5 min acquisition on a 1024 × 1024 matrix and utilising the camera's zoom function (4×). Counts registered in the acquisitions were corrected for the background activity of the corresponding field, and the corrected counts were used for calculation of the percentage activity in the lungs, liver, and body. The DICOM image files were read into MATLAB (MathWorks, MA, USA) and were scaled such that the total counts of individual images were equal.

Arterial distribution of 67 GaTA-MS in normal rabbit livers
Intrahepatic artery instillations of 67 GaTA-MS30, 67 GaTA-MS8 (40 mg each) and soluble 67 Ga-citrate were performed by catheterisation of the cystic artery and using pulses of the particle suspension (total 5 mL 5% dextrose, avg. 110 MBq) interspaced with normal hepatic artery blood flow, so as to disperse the radiolabelled material throughout the liver with close to normal arterial blood perfusion conditions [21]. The MS were kept suspended by gentle agitation during instillation. Static 5 min acquisitions were made on a 1024 × 1024 matrix, approx. 10 min after administration of the imaging agent and again after 1 hour, before the rabbits were euthanised by lethal injection while still under anaesthesia. The lungs and liver were excised after tying off blood vessels to prevent leakage of the radioisotope, and the excised organs were imaged separately using a 5 min acquisition on a 1024 × 1024 matrix and utilising the zoom function. Counts registered in the acquisitions were corrected for the background activity of the corresponding field, and the corrected counts were used for calculation of the percentage activity in the organs. The DICOM image files were read into MATLAB (MathWorks, MA, USA) and for each of the images the individual pixel counts were divided by the total image counts and then scaled by a constant factor.
Imaging of the rabbit VX2 liver tumour model The transplantable rabbit VX2 tumour [22] was a kind gift of Dr. J Geschwind (Johns Hopkins University, Baltimore, USA) and was maintained as a serial transplant on the hind limbs of New Zealand white rabbits. Liver implants of small pieces of tumour tissue were made at a single site in one lobe by keyhole surgery under ventilation anaesthesia with isoflurane and allowed to grow for 18-21 days before use of the rabbit in imaging experiments. At this stage of growth, the tumour was an oblate ellipsoid of maximum diameter 2 cm, still contained within the liver lobe and not involving the body wall or other organs. Macroscopically, the tumour usually had a white necrotic centre, surrounded by a prominently vascularised peripheral growth zone. Intrahepatic artery instillations of 67 GaTA-MS30 and 67 GaTA-MS8 (40 mg each) were made in the livers hosting tumours as above and SPECT-CT imaging with a Hawkeye Infinia camera was performed 1 h post instillation, before the rabbits were euthanised by lethal injection while still under anaesthesia. The lungs and liver were excised after tying off blood vessels to prevent leakage of the radioisotope, and planar images of the excised organs were made separately using a 5 min acquisition on a 1024 × 1024 matrix and utilising the zoom function. Counts registered in the acquisitions were corrected for the background activity of the corresponding field, and the corrected counts were used for calculation of the percentage activity in the organs. The DICOM image files were read into MATLAB (MathWorks, MA, USA) and for each of the images the individual pixel counts were divided by the total image counts and then scaled by a constant factor.  The ANU authors acknowledge the collaborative research project support generously provided to ANU by Sirtex Medical Ltd. (Sydney), including donation of a GE Hawkeye Infinia SPECT/CT scanner and a Xeleris image processing system.

Availability of data and materials
The datasets generated during and/or analysed during the current study are available from the corresponding author on reasonable request.

Declarations
Ethics approval All rabbit procedures adhered to the National Health and Medical Research Council's Australian code for the care and use of animals for scientific purposes (Australian Government, 8th Ed., 2013), and the experimental protocols were approved by the ANU Animal Ethics Committee.

Consent for publication
The corresponding author (RWS) has the permission of all co-authors.