47Sc as useful β–-emitter for the radiotheragnostic paradigm: a comparative study of feasible production routes

Background Radiotheragnostics makes use of the same molecular targeting vectors, labeled either with a diagnostic or therapeutic radionuclide, ideally of the same chemical element. The matched pair of scandium radionuclides, 44Sc and 47Sc, satisfies the desired physical aspects for PET imaging and radionuclide therapy, respectively. While the production and application of 44Sc was extensively studied, 47Sc is still in its infancy. The aim of the present study was, therefore, to investigate and compare two different methods of 47Sc production, based on the neutron irradiation of enriched 46Ca and 47Ti targets, respectively. Methods 47Sc was produced by thermal neutron irradiation of enriched 46Ca targets via the 46Ca(n,γ)47Ca → 47Sc nuclear reaction and by fast neutron irradiation of 47Ti targets via the 47Ti(n,p)47Sc nuclear reaction, respectively. The product was compared with regard to yield and radionuclidic purity. The chemical separation of 47Sc was optimized in order to obtain a product of sufficient quality determined by labeling experiments using DOTANOC. Finally, preclinical SPECT/CT experiments were performed in tumor-bearing mice and compared with the PET image of the 44Sc labeled counterpart. Results Up to 2 GBq 47Sc was produced by thermal neutron irradiation of enriched 46Ca targets. The optimized chemical isolation of 47Sc from the target material allowed formulation of up to 1.5 GBq 47Sc with high radionuclidic purity (>99.99%) in a small volume (~700 μL) useful for labeling purposes. Three consecutive separations were possible by isolating the in-grown 47Sc from the 46/47Ca-containing fraction. 47Sc produced by fast neutron irradiated 47Ti targets resulted in a reduced radionuclidic purity (99.95–88.5%). The chemical purity of the separated 47Sc was determined by radiolabeling experiments using DOTANOC achievable at specific activities of 10 MBq/nmol. In vivo the 47Sc-DOTANOC performed equal to 44Sc-DOTANOC as determined by nuclear imaging. Conclusion The production of 47Sc via the 46Ca(n,γ)47Ca nuclear reaction demonstrated significant advantages over the 47Ti production route, as it provided higher quantities of a radionuclidically pure product. The subsequent decay of 47Ca enabled the repeated separation of the 47Sc daughter nuclide from the 47Ca parent nuclide. Based on the results obtained from this work, 47Sc shows potential to be produced in suitable quality for clinical application. Electronic supplementary material The online version of this article (doi:10.1186/s41181-017-0024-x) contains supplementary material, which is available to authorized users.


Background
Over the past few years, the concept of personalized medicine, where patient treatment is performed according to an individually tailored treatment regime, has gained much recognition (Kraeber-Bodere and Barbet 2014). In nuclear medicine, this approach is realized by exploiting diagnostic techniques, such as non-invasive imaging by means of Positron Emission Tomography (PET) and Single Photon Emission Computed Tomography (SPECT), together with individualized radiotherapeutic treatment (Velikyan 2012). The resulting combination became known as the theragnostic approach and comprises the use of the same molecular targeting vectors, labeled either with a diagnostic or therapeutic radionuclide (Baum and Kulkarni 2012). Ideally, the employed radionuclides represent a matched pair, where both are radioisotopes of the same chemical element. Only a limited number of matching radionuclides entail suitable decay characteristics for radiotheragnostic application (Rösch and Baum 2011); of those the radionuclides of scandium, 44 Sc/ 43 Sc and 47 Sc, are interesting candidates. Based on the physical and chemical characteristics, the β − -emitter 47 Sc is particularly interesting for radionuclide therapy, while the decay characteristics of 44 Sc and 43 Sc are wellsuited for diagnostic PET imaging (Table 1) (Rösch 2012;Müller et al. 2014aMüller et al. , 2014bWalczak et al. 2015).
This matched pair would present an attractive alternative to 68 Ga and 177 Lu, which are currently used in clinics for PET imaging and therapy, respectively (Oh et al. 2011). Ga(III) and Lu(III) can be coordinated by 1,4,7,10-tetraazacyclododecane-1,4,7,10-tetraacetic acid (DOTA)-complexes (Majkowska-Pilip and Bilewicz 2011), however, they do not share the same coordination chemistry. Lu(III) is coordinated by all carboxyl groups of the octadentate DOTA (Viola-Villegas and Doyle 2009; Parus et al. 2015), while Ga(III) has a preference for the coordination number six, leaving two uncoordinated carboxyl groups in the Ga-DOTA-complex (Viola-Villegas and Doyle 2009; Majkowska-Pilip and Bilewicz 2011). As a result, these structural differences may have an influence on the radioconjugate's chemical properties and, consequently, on the in vivo kinetics and receptor binding affinity (Reubi et al. 2000;Majkowska-Pilip and Bilewicz 2011). By using chemically identical radionuclides such as 44 Sc/ 47 Sc-known to form stable complexes with DOTA-this limitation could be addressed.
The emission of low-energy β − -particles from 47 Sc (Eβ − av = 162 keV, Table 1) is particularly interesting for targeted radionuclide therapy of small tumors and cancer metastases, similar to the clinically-established 177 Lu (Eβ − av = 134 keV, T 1/2 = 6.65 d, Table 1). Moreover, the shorter half-life of 47 Sc (T 1/2 = 3.35 d) would encourage its use, in conjunction with small molecules, with relatively fast pharmacokinetic profiles. In analogy to 177 Lu, the decay of 47 Sc is characterized by the co-emission of γ-rays with an ideal energy (Eγ = 159 keV, Table 1) for SPECT imaging (Müller et al. 2014a).
The availability of high 47 Sc activity with adequate purity becomes a crucial issue for the realization of more detailed preclinical investigations and future clinical applications. So far, successful production of 47 Sc was described by two different neutron induced reactions: 47 Ti(n,p) 47 Sc and 46 Ca(n,γ) 47 Ca → 47 Sc ( Fig. 1) (Bartoś et al. 2012;Müller et al. 2014a). To produce 47 Sc from 47 Ti, fast neutrons (E n > 1 MeV) are required, while the 46 Ca(n,γ) 47 Ca reaction is induced by thermal neutrons (E n = 0.025 eV) (Bartoś et al. 2012). Proton irradiation of enriched 48 Ti targets made 47 Sc available via the 48 Ti(p,2p) 47 Sc nuclear reaction, however, too much of the long-lived 46 Sc was coproduced (Srivastava 2012). An alternative 47 Sc production route considers photonuclear reactions on enriched 48 Ti and 48 Ca targets, respectively (Yagi and Kondo 1977;Mamtimin et al. 2015;Rane et al. 2015;Starovoitova et al. 2015). So far only the former route was studied in detail with enriched targets (Yagi and Kondo 1977), while for the latter only natural target material was used for initial benchmark experiments. The aim of the present study was to optimize the parameters of the previously-reported production process for 47 Sc from enriched 46 Ca targets (Müller et al. 2014a) in order to reproducibly obtain increased 47 Sc yields in a formulation allowing direct preclinical application after radiolabeling and to compare it with the method using 47 Ti as target material. 47 Sc labeling experiments with DOTANOC were performed as part of the quality control. In the second part of the study the stability of 47 Sc-labeled DOTANOC was investigated and SPECT/CT imaging studies were performed in tumor-bearing mice to compare the performance of 47 Sc-DOTANOC with the previously obtained PET images of the 44 Sc-labeled counterpart. Moreover, comparison of SPECT images obtained with mice injected with 177 Lu-DOTANOC were also performed.  47 Ti metal and used as target material for fast neutron irradiation. Prior to irradiation, a precursory scan for trace metals by ICP-OES (Perkin Elmer Optima 3000) was performed.

Production of 47 Sc from enriched 47 Ti
The reduction of 47 TiO 2 was performed at Helmholtz Center for Heavy Ion Research (GSI) in Darmstadt as described elsewhere (Lommel et al. 2014). Briefly, the enriched 47 TiO 2 was combined with 40% surplus of calcium hydride and the reduction process was performed under constant argon flow at 900°C for 1 h. Dilute acetic acid was used for the isolation of the reduced 47 Ti metal from the co-produced calcium oxide.
To prepare the targets, 0.6-19.9 mg reduced 47 Ti powder was placed in a quartz glass ampoule (Suprasil, Heraeus, Germany) and sealed. The targets were irradiated with neutrons at the spallation-induced neutron source, SINQ, at Paul Scherrer Institut (PSI) at a fast neutron flux (>1 MeV) of 3.3-3.5 × 10 11 n cm −2 s −1 for 1.5-18.9 days and in the BR2 reactor at SCK.CEN, Mol, Belgium in a reflector channel at a fast neutron flux (>1 MeV) of 5.7 × 10 13 n cm −2 s −1 for 7 days. 47 Sc was formed via the 47 Ti(n,p) 47 Sc nuclear reaction with fast neutrons.

Production of 47 Sc from enriched 46 Ca
To prepare the targets, 65-91 mg enriched 46 CaCO 3 powder was dissolved in concentrated nitric acid and evaporated to complete dryness at 60-70°C. The 46 Ca(NO 3 ) 2 residue was taken up in dilute nitric acid (~1 M HNO 3 ) and an aliquot of the aqueous solution (0.14-0.35 mg 46 Ca) transferred into a quartz glass ampoule, evaporated to dryness and sealed. 47 Sc was produced by the irradiation of the described 46 Ca targets with thermal neutrons at the high flux reactor of Institut Laue-Langevin (ILL) in Grenoble, France at a thermal neutron flux of 1.0-1.4 × 10 15 n cm −2 s −1 , for 4 to 11 days, and at the BR2 reactor at SCK.CEN, Mol, Belgium at a thermal neutron flux of 3.2 × 10 14 n cm −2 s −1 for 7 days, respectively. 47 Sc was generated by the decay of the formed 47 Ca (T 1/2 = 4.54 d) occurring during the irradiation, but also after removal of the ampoule from the reactor.

Separation of 47 Sc from 46 Ca and 47 Ca
The irradiated 46 Ca ampoules were delivered to PSI several days post-irradiation (2.6-12.4 d) and the 47 Sc separation was performed immediately, similarly to previously reported (Müller et al. 2014a). Each ampoule was transferred into a hot cell and the glass surface was cleaned twice with~20 mL 1.0 M HCl and rinsed twice with~20 mL MilliQ water. The crushing of the quartz glass ampoule was performed within a plastic target tube in a separate hot cell. Subsequently, the target tube containing the crushed ampoule was attached to the separation panel with the aid of manipulators. The design of the separation panel, including the adaptation of its operation inside the hot cell, was a crucial part of the method development (Fig. 2). The 46 Ca(NO 3 ) 2 (~10-25 mg) from the ampoule was dissolved in 4 mL 3.0 M HCl and transferred from the target tube to the reaction vessel. A system of syringes, peristaltic pumps and three-way valves (see schematic of the panel in Fig. 2) was used to transfer the reagents from outside into the hot cell. To ensure complete dissolution of the target material, the solution was pumped from the target tube to the reaction vial and back several times. The solution was loaded on a pre-conditioned DGA column (1 mL cartridge filled with 50-70 mg of DGA resin). A second rinse cycle of the crushed glass ampoule with 2.5 mL 3.0 M HCl ensured collection of final traces of the 47 Sc activity, which were subsequently sorbed onto the DGA resin column. Radioactivity detection probes were attached in the vicinity of the target tube and the DGA column to follow the transfer of the 47 Sc radioactivity. Further application of 2 mL 3.0 M HCl removed the stable 46 Ca and radioactive 47 Ca from the DGA resin. The entire Ca-containing effluent was collected in a separate vessel and kept for consecutive separation of further in-growing 47 Sc from the decaying 47 Ca. The sorbed 47 Sc was eluted from the resin column with 4 mL 0.1 M HCl and sorbed on a second column containing SCX cation exchange resin (Method A). Alternately, the 47 Sc-containing eluate was collected, acidified to yield a 3.0 M HCl solution and sorbed on a second, smaller DGA resin column (1 mL cartridge filled with 20-25 mg DGA resin) at a slow flow rate of~0.3 mL/min (Method B), as described by Domnanich et al. (Domnanich et al. 2016). The elution of 47 Sc from the second column was performed with 700 μL 4.8 M NaCl/0.1 M HCl (for Method A) and with 1.7 mL 0.05 M HCl (for Method B) via a separate valve. In order to collect was not necessary for Method A, as the highest proportion of the 47 Sc radioactivity was trapped in a low quantity of eluate.
The renewed generation of 47 Sc, by the decay of radioactive 47 Ca in the Ca-containing fraction ( 47 Ca and stable 46 Ca), enabled subsequent separations after a minimum in-growth time of 3 days.

Radionuclidic purity
To identify the nuclide inventory of the samples, γ-ray spectrometry with an N-type high-purity germanium (HPGe) coaxial detector (EURISYS MESURES, France) and the Ortec InterWinner 7.1 software were employed. The aliquot of 47 Sc eluate was in the range of 3-15 MBq, while the entire neutron irradiated glass ampoules containing the 47 Ti were used for the measurements. The counting time was determined by ensuring the measurement error was <4%. To determine small activities of long-lived radionuclidic impurities, γ-spectrometry measurements of the same samples were performed with an extensive counting time several days post-irradiation.

Radiolabeling for quality control of the produced 47 Sc
After quantitative determination of the 47 Sc activity in the eluate with a dose calibrator (ISOMED 2010, Nuclear-Medizintechnik Dresden, GmbH, Germany), the required activity for radiolabeling in 0.05 M HCl was withdrawn from the product vial and 0.5 M sodium acetate solution (pH 8) was added to the 47 Sc eluate to obtain a pH value of 4.5. DOTANOC (0.7 mM solution in MilliQ water) was added to the 47 Sc solution (~50 MBq) to obtain a specific activity of 10-25 MBq/nmol (2-5 nmol DOTANOC). The reaction mixture was incubated at 95°C for 15 min. The preparation of 177 Lu- High-performance liquid chromatography (HPLC, Merck Hitachi, LaChrom) with a C-18 reversed-phase column (Xterra TM MS, C18, 5 μm, 150 × 4.6 mm; Waters) was used for determination of the radiolabeled fraction of DOTANOC. The detection was performed with a UV (LaChrom L-7400) and radiodetector (Berthold, HPLC Radioactivity Monitor, LB 506B). The mobile phase consisted of MilliQ water containing 0.1% trifluoracetic acid (A) and acetonitrile (B) with a gradient of 95% A and 5% B to 20% A and 80% B, over a period of 15 min, at a flow rate of 1.0 mL/min.

In vitro stability of 47 Sc-and 177 Lu-labeled DOTANOC
The in vitro stability of 47 Sc-and 177 Lu-labeled DOTANOC (radiochemical purities >95%) was investigated in phosphate buffered saline (PBS,pH 7.4). An activity of 50 MBq of 47 Sc-or 177 Lu-DOTANOC was diluted with PBS (pH 7.4) to a total volume of 500 μL and incubated at room temperature for 3 days. Once every 24 h an aliquot was withdrawn to determine the integrity of the labeled compound by means of HPLC.

SPECT/CT imaging with 47 Sc-and 177 Lu-DOTANOC
In vivo experiments were approved by the local veterinarian department and conducted in accordance with the Swiss law of animal protection. Female, athymic nude mice (CD-1 nude) at the age of 5-6 weeks were obtained from Charles River Laboratories, Sulzfeld, Germany. AR42J cells (rat exocrine pancreatic tumor cells, European Collection of Cell Cultures ECACC, Salisbury, U.K.) were suspended in PBS (5 × 10 6 cells in 100 μL) and subcutaneously inoculated on each shoulder. SPECT/CT experiments were performed about 2 weeks after tumor cell inoculation, when the tumor reached a size of about 400 mm 3 .
Imaging studies were performed using a small-animal SPECT camera (NanoSPECT/CT™, Mediso Medical Imaging Systems, Budapest, Hungary) as previously reported (Müller et al. 2014b). The energy peaks for the camera were set at 159.4 keV (± 10%) for the scans with 47 Sc and 56.1 keV (± 10%), 112.9 keV (± 10%) and 208.4 keV (± 10%) for the scans with 177 Lu. SPECT/CT scans were followed by CT scans. The images were acquired using Nucline Software (version 1.02, Bioscan Inc., Poway, California, US). The reconstruction was performed iteratively with HiSPECT software (version 1.4.3049, Scivis GmbH, Göttingen, Germany). SPECT and CT data were automatically co-registered and the fused datasets were analyzed with the VivoQuant post-processing software (version 2.50, inviCRO Imaging Services and Software, Boston, USA).
The mice were injected intravenously with 47 Sc-DOTANOC (12 MBq, 1.2 nmol, 100 μL) and 177 Lu-DOTANOC (40 MBq, 1.2 nmol, 100 μL), respectively. The in vivo SPECT/CT scans of 35 min duration were acquired 3 h after injection of 47 Sc-DOTA-NOC. During the scans, the mice were anesthetized by inhalation of a mixture of isoflurane and oxygen. Post-mortem scans of 1.3-3.5 h were performed 6 h after injection of 47 Sc-and 177 Lu-DOTANOC. The SPECT acquisitions were performed in such a manner to obtain the same total number of counts for each scan.

Production of 47 Sc from 47 Ti via the (n,p) reaction
The irradiation of enriched 47 Ti targets at both SINQ and the BR2 reactor resulted in the formation of 0.07-4.9 MBq 47 Sc at the end of irradiation (EOI). The respective 47 Sc saturation yields were determined to be between 1.8 and 10.0 MBq 47 Sc/mg 47 Ti 10 −13 n cm −2 s −1 (summarized in Table 2) by taking the irradiation time, mass of enriched 47 Ti and fast neutron flux into consideration. γ-spectrometry measurements of the neutron-irradiated 47 Ti ampoules revealed that, other than 47 Sc, the long-lived radionuclidic impurity 46 Sc was formed. The amount of generated 46 Sc was influenced by the irradiation period and the neutron energy (Bokhari et al. 2010;Zerkin 2016) and ranged from 3.8 to 11.5% 46 Sc/for the irradiations at SINQ (Additional file 1: Figure S3). Considerably less 46 Sc (0.05%) was produced by the irradiation at the BR2 reactor, however. In view of the high percentage of co-produced 46 Sc and the relatively low 47 Sc production yield at both facilities, the production of sufficiently high 47 Sc activities for radiopharmaceutical applications was not considered feasible and, thus, chemical isolation of 47 Sc from neutron irradiated 47 Ti targets was not performed.

Production of 47 Sc from 46 Ca via the (n,γ) reaction
The irradiation of 46 Ca targets with thermal neutrons resulted in the formation of 47 Ca, which decayed to 47 Sc and yielded 210-2140 MBq 47 Sc at the time the separation was performed. The 47 Sc saturation yield was determined by taking the mass of 46 Ca, the irradiation time (t irr ), the decay time after EOI (t wait ) and the thermal neutron flux (Φ th ) into account and was within the range of 85-98 MBq 47 Sc/mg 46 Ca 10 −13 n cm −2 s −1 , which is comparable with the calculated 47 Sc saturation yield of 92 MBq 47 Sc/mg 46 Ca 10 −13 n cm −2 s −1 (summarized in Table 3). 47 Sc is formed during the irradiation but also, however, for some time after the end of irradiation by the decay of 47 Ca. This implies that both the irradiation time (t irr = 7.0-11.0 d) and the elapsed post-irradiation time until the start of the separation (t wait = 3.0-12.5 d) are part of the yield-determining factors. The highest 47 Sc activities under the applied irradiation conditions become accessible at an optimal post irradiation waiting time (t opt ) and are represented as relative activity a( 47 Sc) opt in Table 3. The measured relative 47 Sc activities (a( 47 Sc) meas ) are lower than the optimal relative 47 Sc activities (a( 47 Sc) opt ), as the separations were performed several days after the optimal waiting time. The variable f( 47 Sc) describes the ratio of the 47 Sc activity at t opt and the 47 Ca activity at EOI. It can be considered as a measure for the maximal obtainable 47 Sc activity, since both activities are referred to the time point of their maximum. The formulae used for the calculation of A( 47 Sc) calc , t opt, a( 47 Sc) opt and f( 47 Sc) are given in Additional file 1: Figure S1 a-d. The measured relative 47 Sc activities (a( 47 Sc) meas ) of the irradiations ILL 1, ILL 5 and SCK.CEN are represented by single data points in Fig. 3. After an optimal post irradiation waiting time (t opt ) of 2.8 days, the relative 47 Sc activity (a( 47 Sc) opt ) reaches the maximum with 0.43. The separations ILL 1 and ILL 5 were performed after waiting times of 6.7 and 6.8 days, resulting in lower relative 47 Sc activities of 0.34 and 0.37, respectively. The waiting time after the irradiation at SCK.CEN (3.0 days) is close to t opt , thus the obtained relative 47 Sc activity of 0.44 is comparable with the optimum value.

Separation of 47 Sc from 46 Ca and 47 Ca
After the ampoule was crushed, it was moved to the hot cell containing the production panel (Fig. 2), where the target material was dissolved by repeated application of 6.5 mL 3.0 M HCl. The 47 Ca and 47 Sc activity was transferred from the crushed glass ampoule to the DGA column, leaving only 1.1 ± 0.5% 47 Ca and 3.3 ± 0.4% 47 Sc attached to the glass. Direct application of 2 mL 3.0 M HCl quantitatively removed the Ca ( 47 Ca and stable Ca isotopes) from the resin. The collected Ca fraction contained 99.8 ± 0.2% of the total 47 Ca activity. The 47 Sc activity was eluted from the DGA column with 4 mL 0.1 M HCl and  only 1.5 ± 0.6% of the 47 Sc activity remained on the column. Using Method A, the solution was concentrated on the SCX cation exchange resin (used as the second column) and eluted with 700 μL 4.8 M NaCl/0.1 M HCl solution (pH 0-0.5), collecting 94.8 ± 2.1% of the total 47 Sc activity. When using Method B, the molarity of the 47 Sc eluate was increased from 0.1 to 3.0 M HCl and the resulting solution adsorbed on a second smaller DGA column and eluted with 1.7 mL 0.05 M HCl. Fractionized collection revealed that about~90% of the eluted 47 Sc activity was obtained in the first 700 μL (pH~0).
With the installation of the chemical separation system in a hot cell, yields of up to 1.9 GBq 47 Sc could be isolated from the irradiated 46 Ca target. The renewed generation of 47 Sc from the β − -decay of 47 Ca (T 1/2 = 4.54 d) in the Ca-containing fraction, reached the maximum 47 Sc activity after an in-growth period of 5.6 days (Fig. 4) and, thus, enabled repeated separations. As a result of experimental conditions, separations were performed after an in-growth time of 3-7 days. The separation process was successfully repeated 2-4 times, until the eluted 47 Sc activity was~100 MBq.

Radionuclidic purity of 47 Sc produced from 46 Ca via (n,γ) reaction
The γ-ray spectrum of the neutron-irradiated 46 Ca target material (Additional file 1: Figure S2a) showed, exclusively, the γ-lines of 47 Sc (159 keV) and the parent nuclide 47 Ca (489, 808 and 1297 keV). After chemical separation and concentration of 47 Sc on SCX resin (Method A), the radionuclidic purity of the final 47 Sc eluate was 99.6 ± 0.7%. When the second DGA column (Method B) was used, the radionuclidic purity increased to 99.99 ± 0.03% (Additional file 1: Figure S2b). The long-lived radionuclidic impurity 46 Sc was only present in the eluate obtained from the first separation at a maximum of 0.005% and could be only detected by performing long-term γspectrometry measurements several days post separation. The isolated 47 Sc which was generated by the decay of 47 Ca in the Ca-containing fraction did not contain any 46 Sc, due to its entire removal within the first separation.

Radiolabeling and stability of 47 Sc labeled DOTANOC
Radiolabeling of 47 Sc was reproducible at a specific activity of 10 MBq 47 Sc per nmol DOTANOC, with >96% radiochemical purity. Depending on the activity concentration of the 47 Sc solution, it was also possible to label at higher specific activity of up to 25 MBq/nmol.
The stability of 47 Sc-labeled DOTANOC in PBS (pH 7.4) was investigated over a period of 3 days and compared to the stability of the 177 Lu-labeled analogue (Fig. 5). Directly after the radiosynthesis of 47 Sc with DOTANOC, the amount of intact radiolabeled product was 96.6-99.0% with less than 2% of radiolysis products visible as pre-peaks on the HPLC chromatogram. After 1 day at room temperature, the amount of intact radiolabeled 47 Sc-DOTANOC decreased to 81.3%, while 18.3% were subjected to radiolysis. Over the whole investigation period of 3 days, the percentage of radiolysis products increased to 44.4%, however, the amount of free 47 Sc was always below 2.1%. The stability of 47 Sc-labeled DOTANOC was found to be comparable with the clinically-used analogue 177 Lu-DOTANOC. After 3 days, the amount of intact 47 Sc-DOTANOC (54.1%) was similar to the amount of intact 177 Lu-DOTANOC (43.2%).
Imaging with 47 Sc-DOTANOC in comparison to 44 Sc-DOTANOC and 177 Lu-DOTANOC SPECT/CT experiments performed with AR42J tumor-bearing mice allowed excellent visualization of the accumulated 47 Sc-DOTANOC in tumor xenografts, which express the somatostatin receptor (Fig. 6a). Activity accumulation was also observed in the kidneys, which was due to renal excretion of the radiopeptide. The SPECT/CT image obtained with 47 Sc-DOTANOC showed an equal activity distribution profile as was previously demonstrated by a PET/CT scan of an AR42J tumor-bearing mouse 3 h after injection of 44 Sc-DOTANOC (Fig. 6b) (Domnanich et al. 2016).
To compare the image quality of 47 Sc with the clinically-employed 177 Lu, mice with AR42J tumors were injected with either 47 Sc-or 177 Lu-labeled DOTANOC and scanned 6 h after injection (Fig. 7). Both images visualized the uptake of the radiopeptides in tumor xenografts located on each shoulder of the mouse and in the kidneys. The distribution pattern was equal for both radiopeptides, as expected, based on similar coordination of 47 Sc and 177 Lu when using DOTA.

Discussion
Recently, the use of 47 Sc for therapeutic purposes, as part of the 44 Sc/ 47 Sc theragnostic radionuclide pair, has attracted considerable attention in the field of nuclear medicine (Rösch and Baum 2011). In the present study we report, to our knowledge, the first reproducible production of MBq to GBq activities of 47 Sc by the irradiation of enriched 46 Ca target material with thermal neutrons. Concurrently, an alternative production route was investigated, using 47 Ti as target material and, by irradiation with fast neutrons, 47 Sc was formed via the (n,p) nuclear reaction. Irradiation of enriched 47 Ti at the spallation source SINQ (PSI, Switzerland) and the BR2 reactor (SCK.CEN, Mol, Belgium) resulted in the formation of 1.8-10.0 MBq 47 Sc/mg 47 Ti 10 −13 n cm −2 s −1 . The production of 47 Sc by neutron irradiation of enriched 47 Ti target material was performed previously by Mausner and Kolsky in the fission neutron spectrum of the HFIR reactor in Oak Ridge National Laboratory (ORNL) and Brookhaven National Laboratory (BNL) ) and by Bartoś et al. at the Maria reactor in Świerk, Poland (Bartoś et al. 2012). The reported activity of produced 47 Sc was within the range of 1.4-4.2 MBq 47 Sc/mg 47 Ti 10 −13 n cm −2 s −1 , which is comparable with the 47 Sc radioactivity obtained in our experiment at BR2. The irradiations at SINQ generated higher 47 Sc activities, due to the larger proportion of fast neutrons (> 1 MeV) at the spallation source (Lehmann 2016) than at the reactor Page 12 of 17 (Chrysanthopoulou et al. 2014). The production of 85-98 MBq 47 Sc/mg 46 Ca 10 −13 n cm −2 s −1 was feasible via the 46 Ca(n,γ) 47 Ca → 47 Sc route with thermal neutrons from nuclear reactors at ILL and BR2, however. The nearly ten-fold higher accessible 47 Sc activity from 46 Ca irradiations can be attributed to the higher nuclear cross section of the 46 Ca(n,γ) 47 Ca reaction (σ = 0.7 barn) (Magill et al. 2015) in comparison to the 47 Ti(n,p) 47 Sc nuclear reaction (Additional file 1: Figure S4), as well as to an increased flux of thermal neutrons in comparison to fast neutrons. Lower measured relative 47 Sc activities than the optimal activities were obtained for the irradiations ILL 1-ILL 5, because the post-irradiation waiting period was too long and the built up 47 Sc already started to decay. Shorter waiting periods were prevented due to logistical issues. The waiting period after the irradiation at SCK.CEN, however, was close to the optimum time period, thus the measured relative 47 Sc activity obtained from the irradiation at SCK.CEN is in good correlation with the optimal value. γ-spectra of the irradiated 46 Ca indicated the presence of 47 Ca and 47 Sc, however, due to high dead time a precise determination of both activities before separation was not possible. After chemical separation a product of high radionuclidic purity, containing only 0.005% 46 Sc, was obtained (Additional file 1: Figure S2). The acquired radionuclide inventory of 47 Ti targets, neutron irradiated at the spallation source SINQ, indicated a higher percentage of 46 Sc (Additional file 1: Figure S3), which increased proportionally with irradiation time (3.8-11.5% 46 Sc for 1.5-18.9 days irradiation), whereas a significantly smaller amount of 46 Sc (0.05%) was produced after 7 days of irradiation at the BR2 reactor (SCK.CEN). Trace activities of 122 Sb and 124 Sb were identified in the γ-ray spectrum of the irradiated 47 Ti ampoule at SINQ (Additional file 1: Figure S3), conceivably produced by neutron activation reactions on natural Sb impurities (Additional file 1: Table S5). The presence of 22 Na and 7 Be can be attributed to spallation reactions with the capsulation material. The scan for trace metals revealed impurities of Ca, Sr, Sb and Zr in the reduced 47 Ti metal (Additional file 1: Table S5), which were probably introduced by the reduction process. Neutron activation reactions were only observed with Sb, but not with any of the other determined impurities, however. The formation of 46 Sc from 47 Ti via the 47 Ti(n,n + p) 46 Sc nuclear reaction is known to be only induced by very fast neutrons above the threshold of 10.7 MeV (Additional file 1: Figure S4) (Zerkin 2016). The considerably decreased 46 Sc impurity of the sample irradiated at the BR2 reactor can, therefore, be attributed to the lower proportion of very energetic neutrons in the fission spectrum of the BR2 reactor (Chrysanthopoulou et al. 2014), compared to the spallation neutron spectrum in the SINQ (Lehmann 2016). The percentage of 46 Sc obtained from the 47 Ti irradiation at the BR2 reactor was in agreement with those from previous experiments at ORNL, BNL and the Maria reactor, which was reported to be 0.06-0.64% 46 Sc Mausner et al. 1998;Bartoś et al. 2012). With respect to radiopharmaceutical applications, the ten-fold higher 47 Sc production from 46 Ca targets, together with the absence of long-lived radionuclidic impurities, intensified our research towards the further development of the more attractive 46 Ca route.
In order to meet the requirements for radiopharmaceutical applications, the obtained 47 Sc eluate needed to be of high chemical purity and concentrated into a small volume of moderately acidic eluate to facilitate efficient radiolabeling and subsequent in vivo application. Initially, SCX cation exchange resin (Method A) was used and 94.8% ± 2.1% of the total 47 Sc activity was recovered in only 700 μL eluate (4.8 M NaCl/0.1 M HCl). The use of this resin is already established for the concentration of the 68 Ga eluate from the 68 Ge generator (Mueller et al. 2013); however, direct preclinical in vivo application is not feasible due to the high osmolarity of the obtained eluate. In a modification of the separation procedure, a second, smaller DGA column was used (Method B), allowing the elution of~90% using 700 μL 0.05 M HCl. This enabled labeling and preclinical application as previously shown with 44 Sc (Domnanich et al. 2016).
The labeling of DOTANOC with 47 Sc was performed to verify a consistent chemical purity of the obtained eluate. Our results demonstrated reproducible radiosynthesis of 47 Sc-DOTANOC at specific activities of 10 MBq/nmol, whereas radiolabeling at 25 MBq/nmol proved possible. The obtained results indicated good quality of the produced 47 Sc achieving radiolabeling yields, in agreement with the previouslyperformed 47 Sc-radiolabeling of a DOTA-folate conjugate (Müller et al. 2014a).
In PBS 47 Sc remained stably coordinated by the DOTA-chelator over 3 days (<6% release), a result which was comparable to the 177 Lu-labeled peptide. 47 Sc-and 177 Lu-DOTANOC were, however, affected by radiolytic decomposition, which decreased the amount of intact product over time. It is likely the radiolytic stability could be enhanced by the addition of radical scavengers, such as ascorbic or gentisic acids, which were previously successfully employed for the stabilization of 90 Y-and 177 Lu-labeled DOTA-peptides (Liu and Edwards 2001;Liu et al. 2003).
In a proof-of-concept study, 47 Sc-DOTANOC was utilized for SPECT/CT imaging of AR42J tumor-bearing mice. The equal distribution profile of 47 Sc-DOTANOC and 44 Sc-DOTANOC, previously demonstrated using PET/CT, demonstrated the successful realization of the "matched pair" principle using scandium radionuclides. Moreover, the in vivo distribution of 47 Sc-DOTANOC was comparable to 177 Lu-DOTANOC. Due to the higher percentage of emitted γ-radiation in the case of 47 Sc, it is expected that less activity of 47 Sc-labeled compounds would be necessary for clinical SPECT as compared to the activity necessary for 177 Lu-labeled counterparts.

Conclusions
The reproducible production of activities of up to 2 GBq 47 Sc at high radionuclidic purity via the 46 Ca(n,γ) 47 Ca nuclear reaction in the thermal neutron flux of a reactor was demonstrated. The subsequent decay of 47 Ca to 47 Sc creates a "pseudo-generator" system, which enables the repeated separation of the 47 Sc daughter nuclide from the 47 Ca parent nuclide. Together with the high radionuclidic purity and the superior yield of the isolated 47 Sc activity, the 46 Ca production route bears significant advantages over the 47 Ti production route with fast neutrons. Even though the high price of enriched 46 Ca represents a drawback, implementation of a suitable recovery method will limit the expenses. Based on the results obtained from this proof-of-concept study, 47 Sc has the potential to be produced in a suitable quality for clinical applications, however, the quantity of radioactivity still needs to be expanded to meet the requirements for radionuclide therapy.

Additional file
Additional file 1: Figure S1a. Formula for the calculation of the 47 Sc activity in Bq (s −1 ), accessible under the applied irradiation conditions. σ = nuclear cross section of the 46 Ca(n,γ) 47 Ca reaction in cm −2 , N T = number of 46 Ca atoms, Φ th = thermal neutron flux in n * cm −2 * s −1 , λ Sc and λ Ca = decay constants of 47 Sc and 47 Ca in s −1 , t irr = irradiation time and t wait = post irradiation waiting time in s. b Formula for the calculation of the optimal post irradiation waiting time (t opt ) in s, accessible at the applied irradiation time (t irr in s). The decay constants of 47 Sc (λ Sc ) and 47 Ca (λ Ca ) are given in s −1 . c Formula for the calculation of the optimal relative 47 Sc activity (a( 47 Sc) opt ) (dimensionless), accessible under the applied irradiation conditions. σ = nuclear cross section of the 46 Ca(n,γ) 47 Ca reaction in cm −2 , N T = number of 46 Ca atoms, Φ th = thermal neutron flux in n * cm −2 * s −1 , λ Sc and λ Ca = decay constants of 47 Sc and 47 Ca in s −1 . d Formula for the maximal obtainable 47 Sc activity (dimensionless). The irradiation time (t irr ) is given in s and the decay constants of 47 Sc (λ Sc ) and 47 Ca (λ Ca ) in s −1 . Figure S2. γ-Ray spectra of 47 Sc and 47 Ca from the neutron-irradiated 46 Ca ampoule, obtained 71 h after the end of irradiation (measurement time: 10 s) (a) and of the pure 47 Sc eluate after separation (Method B), obtained 1 h after the end of separation (measurement time: 250 s) (b). Figure S3. γ-Ray spectrum of the neutron-irradiated 47 Ti ampoule at SINQ, obtained 21 d after the end of irradiation (measurement time: 9600 s). Figure S4. Measured cross section values (squares, retrieved from the EXFOR-database) (Zerkin 2016) as well as the theoretical calculations from the TENDL-2015 library (straight line) (Koning, Rochman et al. 2015) for the 47 Ti(n,p) 47 Sc (blue) and the 47 Ti(n,p + n) 46 Sc (black) nuclear reactions. Table S5. Trace metal analysis of the reduced 46 Ti metal by ICP-OES. Only the elements determined at a concentration higher than the detection limit are listed below. (DOCX 415 kb)