Introduction

Only few pairs of radionuclides have been proposed for theranostics applications. These are iodine radioisotopes 123I and 124I for imaging, as well as 131I for therapy. Other examples are 86Y, 61Cu and 64Cu for PET imaging and β− emitters 90Y and 67Cu for therapy. Recently, scandium radioisotopes were also proposed for theranostics application. Significant attention in 44Sc and 43Sc as tracers in positron emission tomography imaging has been observed. 44Sc was first proposed by Rösch as a potential alternative for 68Ga in clinical PET diagnosis [1, 2]. 44Sc decays by the emission of low-energy positrons (E β+ = 1.47 MeV) with the half-life T 1/2 = 3.97 h which is almost four-fold longer than that of 68Ga. 44Sc can be obtained from the 44Ti/44Sc generator [3] or produced in the 44Ca(p,n)44Sc reaction on small or medium medical cyclotrons that currently supply 18F to hospitals [4,5,6,7,8,9]. However, the co-emission of high-energy γ-rays (E γ = 1157, 1499 keV), has to be taken into consideration with regard to radiation dose to the patients and clinical staff. Emission of high-energy γ-rays generates also radiolytic decomposition of biomolecules, which is thought to be accompanied by formation of free radicals [10]. Another radionuclide of scandium-43Sc, shows properties similar to 44Sc, but emits much lower energy of concurrent gamma. 43Sc can be produced either by the 43Ca(p,n) [11], or by 42Ca(d,n) [12] reaction, but unfortunately the cost of enriched calcium targets is very expensive. The more promising method of 43Sc production is alpha irradiation of natural calcium target through the 40Ca(α,p) and 40Ca(α,n) channels has been mentioned in recently published paper [13].

Another scandium radioisotope, i.e., 47Sc (T 1/2 = 3.4 days, E β-(av) = 162 keV, main E γ = 159.4 keV, I = 68.3%) is a promising low-energy β− emitter for targeted radiotherapy [14,15,16,17], thereby the β+ emitting radionuclides 44Sc or 43Sc together with the β− emitting 47Sc represent ideal theranostic pairs as mentioned above regarding iodine, copper and yttrium radioisotopes.

The method of producing highly active 47Sc in a nuclear reactor was described by Mausner et al., Kolsky et al. and Srivastava et al. [18,19,20]. An enriched 47TiO2 target was irradiated with high energy neutrons (E n > 1 MeV) to produce 47Sc via the 47Ti(n, p)47Sc reaction. The second method of 47Sc production is thermal neutron irradiation of 46Ca enriched target in the 46Ca(n,γ) → 47Ca(T 1/2 = 4.54 days) (β−) → 47Sc nuclear reaction. This method was described recently by Muller et al. [21]. The former nuclear reaction requires E n > 1 MeV, while the latter reaction uses more available thermal neutrons. The second advantage of this method is the use of 47Ca/ 47Sc generator system to supply 47Sc activity, but the disadvantage is the requirement of an enriched target. 46Ca is presently available with only a 30% enrichment and at a very high price, which makes the target cost prohibitive.

Other ways of producing 47Sc have been based on proton irradiation of natTi [22] and γ irradiation of 48Ti in an electron linear accelerator (LINAC) [23]. The production efficiency was lower than in the former cases. Both production routes require also radiochemical separation of 47Sc from the Ti targets. Recently, new cyclotron method for 47Sc production by alpha irradiation of 44Ca enriched target was reported [24]. Unfortunately due to low cross section of the 44Ca(α,p)47Sc reaction production by this method makes difficult to obtain GBq quantities, which are necessary to carry out clinical trials.

Various methods of 47Sc separation from TiO2 targets based on tributyl phosphate (TBP) extraction, extraction chromatography or on cation and anion exchange processes have been reported [11, 12]. Dissolution in hot concentrated H2SO4 and evaporation of the solution were the most difficult and time-consuming steps in the case of the TiO2 target, what significantly limits the use of this type of targets. In the case of calcium targets many methods of scandium radionuclide separation have been reported [4, 5, 8, 22]. All proposed methods are simple, fast and allow for high percent of scandium radionuclides and for calcium recovery.

In the present work we propose an alternative way of 47Sc production through the 48Ca(p,2n) reaction at medium size cyclotrons (proton energy below 30 MeV). During 48Ca irradiation the 48Sc (T 1/2 = 43.67 h) and 46Sc (T 1/2 = 83.79 days) are also co-produced, therefore the goal of our studies was optimization of the parameters of 48Ca irradiation for maximization the 47Sc production with minimal 48Sc and 46Sc impurities.

Experimental

Materials and methods

Hydrochloric acid (HCl), 37%, glacial acetic acid (CH3COOH) and citric acid were purchased from Sigma-Aldrich, sodium hydroxide (NaOH) from Merck, sodium acetate from POCH S.A. (Gliwice, Poland), ammonia and ethanol from Fluka. All chemicals were analytical grade and were used without further purification. Chelating ion exchange resin Chelex 100, (Na+-form, mesh size 100–200, bed size 0.8 × 4.0 cm, suspended in water) was purchased from Bio-Rad Laboratories.

Irradiation of calcium target

Production yield of 47Sc and 48Sc radionuclides for proton induced reactions on CaCO3 with natural isotopic composition were measured as function of proton energy in the range of 60 → 0 MeV using activation method on stacks. The stack consisted of thin metallic foils with natural isotopic composition interleaved with CaCO3 targets. The stack was assembled from six groups of Al–Cu–CaCO3 and five groups of Al–Cu–Ti–CaCO3. The stack of targets and foils were mounted in a target holder which was made of aluminum. CaCO3 powder of analytical grade from POCH S.A. (Gliwice, Poland) were pressed with 232 MPa. Thickness of targets were about 0.35 g/cm2. The thickness of monitor reaction Al, Cu and Ti foils (purity 99.0–99.9% supplied by Goodfellow Cambridge Ltd., England) was 0.02 mm for Al, 0.01 mm for Cu and 0.01 mm for Ti. The thin metallic foils were used to monitor intensity and/or energy of proton beam.

Two stacks were irradiated at the extracted proton beam of the AIC-144 cyclotron of the Institute of Nuclear Physics Polish Academy of Sciences Cracow. Irradiations were carried out for 5 h with beam current of about 30 nA and with initial proton bombarding energy of 60 MeV.

Data analysis

The activities of the radioactive products of the targets and monitors were measured nondestructively using the gamma HPGe-detector coupled with Multichannel Analyzers 919E EtherNIM. The photo peak area of γ-ray spectra was determined by using MAESTRO Multichannel Analyzer Emulation. The decay data for the monitors and Sc radionuclides, such as half-life (T 1/2), γ-ray energy (E γ) and γ-ray emission probability (I γ), were taken from the Table of Radioactive Isotopes [25].

The proton flux intensity was determined through the monitor reactions: 27Al(p,x)22,24Na, natCu(p,x)56Co, 62,65Zn and natTi(p,x)48V from the measured activities induced in monitor foils at the front position of each CaCO3 target. The standard cross-sections for the monitor reactions were taken from [26].

The energy degradation along the stacks and effective particle energy in the middle of each foil was calculated using the computer program SRIM version 2008.04 [27]. The estimated uncertainty of the points representing the proton energy ranges from ±0.6 up to ±1.5 MeV.

The following sources of errors were considered to derive the summed up uncertainty in the yield values of 46Sc, 47Sc and 48Sc: statistical error (1–7%), error of the monitor flux (~6%), error due to the sample thickness determination (1–2.5%) and error of efficiency calibration of γ-rays spectrometer (~5%). The overall uncertainty in the determined yield was around 12%.

Separation of Sc from the target

Three methods previously elaborated for 43,44Sc production were tested for separation of 47Sc from natural Ca target. First of the methods based on application of chelating resin, elaborated in our group [4], consisted of dissolution of the target in 1 M HCl and adsorption of scandium radionuclides on chelating ion exchange resin Chelex 100 of bed size 0.8 × 4.0 cm and conditioned with 5 ml of 1 M HCl. After adsorption of Sc3+ and Ca2+, the column was washed with 30 ml of 0.01 M HCl in order to remove Ca2+. The scandium radionuclides were then eluted with 1 M HCl in 0.5 ml fractions. In the second method, described by Valdovinos et al. [8], the irradiated natCaCO3 target was dissolved in 1 ml of 9 M HCl solution. The dissolved target solution was passed through a column containing 50 mg of UTEVA resin and after adsorption of scandium radionuclides the column was washed with 5 ml of 9 M HCl. The scandium radionuclides were eluted with a 400 µl portion of H2O. The third method which used 47Sc separation on 0.2 µm filter was recently described by Minegishi et al. [24]. In this method calcium target was dissolved in 0.5 M HCl and next was neutralized by 25% NH3 solution to pH 10. The obtained Sc solution was then passed through a 0.2 μm filter (Whatmann) to trap Sc radioisotopes. Subsequently, 3 ml of pure water was passed through the filter to wash out residual Ca2+ and NH4 + cation. Scandium radionuclides trapped on the filter was eluted by 0.5 M HCl. The methods were tested on proton irradiated natural Ca targets containing 44Sc, 47Sc and 48Sc no carrier added radionuclides.

Results and discussion

Up to now for the production of 46,47,48Sc through the natCa(p,xn)46,47,48Sc reaction only two groups reported experimental data: first group reported cross section data using natCa in calcium formate as the target [28] and a second group a thick target yields for natural calcium metal [6]. The first paper published by Michel et al. describes cross section data over the proton energy only in the range 70 → 24 MeV for natCa(p,xn)46Sc and in the proton energy range 70 → 15 MeV for natCa(p,xn)48Sc. The second paper published by Severin et al. which described result of the 44Sc production on natural calcium target, contains also results of production 47Sc and 48Sc contaminants at 16 MeV proton energy. So far there are no systematic studies on 47Sc production in 48Ca(p,2n)47Sc reaction. In the present work results of 47Sc production via the 48Ca(p,2n) reaction in energy range 60–0 MeV have been described.

Due to low availability and high cost of 48Ca we decided to work with natural calcium targets containing 0.187% 48Ca. Table 1 shows the isotopic compositions of the natural and 48Ca enriched calcium targets.

Table 1 Isotopic composition of natural abundance and of enriched calcium targets (Sigma-Aldrich and ISOFLEX, USA)
Full size table

Table 2 shows results of irradiation of natCaCO3 in the energy range 60–0 MeV.

Table 2 Measured production yields of the reactions natCa(p,xn)46,47,48Sc at EOB as a function of proton bombardment energy
Full size table

As shown in Table 2 the optimum energy range for the 48Ca(p, 2n)47Sc reaction is 24 → 17 MeV with the peak at about 20 MeV for CaCO3 target. The peak for 48Ca(p,n)48Sc reaction is about 11 MeV and is about 33 MeV for 48Ca(p,3n)46Sc reaction. It should be noted that 46Sc forms also in 46Ca(p,n)46Sc nuclear reaction, but due to the very low abundance of 46Ca in natCa (0.004%) the produced 46Sc activity is negligible. In the obtained 47Sc radionuclide the percent of 48Sc impurity at EOB is 14.7 and decreases after 80.38 h (half-life of 47Sc) to 7.4%. As shown in Table 2 in this energy range also 20.1 kBq/μA h of 47Ca (T 1/2 = 4.5 days) is produced which by the β− decay generates additional 47Sc activity, therefore the yield of 47Sc at EOB includes also the activity of 47Sc generated from 47Ca decay during the irradiation. The ratio of 46Sc impurity to 47Sc was only 0.2% at EOB. Irradiation of thiner CaCO3 targets could allow to increase the representing points in the proton energy range.

The yields estimated by using the cross section data from Michel et al. [28] are lower by about 25% than in our work. Irradiations of natural Ca metal at 16 MeV proton energy were performed by Severin et al. [6]. This group presented yields at EOB equal to 0.09 for 47Sc and 0.33 MBq/μA h for 48Sc. However, direct comparison of obtained experimental yield values is difficult because irradiated Ca targets had different chemical forms. For comparison of measured yields obtained in this work with data obtained by Severin et al. [6], a normalization for the number of Ca atoms in the target was performed. Our yield value 0.31 MBq/μA h for 48Sc, after the normalization, over the proton energy range 16.6 → 0 MeV is in good agreement with the yield value obtained by Severin et al. [6]. In the case of 47Sc we obtained about two times higher yield value than Severin et al. [6]. This difference may result from error in determination of input proton energy on the target. In this proton energy range a small change in energy causes a large change in the efficiency of (p,2n) nuclear reaction.

To find the best method for separation of 47Sc from the calcium targets three methods previously elaborated for 44Sc were tested. These methods were compared in respect to Sc separation yield, possibility of separation from other metallic impurities which could negatively affect the effectiveness of 47Sc bioconjugates labeling and recovery of calcium target (Table 3).

Table 3 Comparison of the methods for separation of scandium radionuclides from calcium targets
Full size table

All separation procedures studied are fast and simple. In the case of Chelex 100 and UTEVA resins the target dissolution and separation of scandium radioisotopes were performed in 30 min and the separation process on 0.2 µm filter needs only 15 min. The separation process on 0.2 µm filter gives also highest recovery of scandium isotopes and is much simpler and faster in comparison with other methods studied. It contains only two simple processes: adsorption Sc(OH)3 on the filter and then dissolution of the precipitate in hydrochloric acid.

Chemical purity of the Sc product is important, since the presence of other metals may interact with the DOTA-chelator. The most dangerous is Fe3+ for which forms stronger complexes with DOTA than Sc3+ [29]. Influence of other possible impurities like Zn2+, Mg2+, Sr2+ and Co2+ is negligible due to the much lower stability constants of their DOTA complexes [29]. The results of Fe concentration in dissolved calcium targets and in scandium fractions after separation processes are presented in Table 4.

Table 4 Iron concentration in dissolved calcium target in HCl solutions and in scandium fractions after separation processes
Full size table

For all the methods studied the concentration of Ca2+ in Sc fractions was less than 1 ppm.

Taking into account the obtained results, we recommend filtration process for the separation of Sc radionuclides from the calcium targets.

Conclusion

The production of 47Sc in (p,2n) nuclear reaction on natural CaCO3 target was successfully performed. During proton irradiation of natural calcium target radionuclides of interest 46Sc, 47Sc and 48Sc can be formed only in p,n, p,2n and p,3n reactions on 48Ca, therefore, the results obtained for natural calcium can be recalculated for enriched 48Ca targets. The obtained results of production efficiency creates the opportunity to produce GBq activity levels of 47Sc. The separation process based on precipitation of Sc(OH)3 and separation on 0.2 µm filter is simple, reliable, efficient and fast and can be easily adapted for remote operation.