Skip to main content

Production of scandium radionuclides for theranostic applications: towards standardization of quality requirements


In the frame of “precision medicine”, the scandium radionuclides have recently received considerable interest, providing personalised adjustment of radiation characteristics to optimize the efficiency of medical care or therapeutic benefit for particular groups of patients. Radionuclides of scandium, namely scandium-43 and scandium-44 (43/44Sc) as positron emitters and scandium-47 (47Sc), beta-radiation emitter, seem to fit ideally into the concept of theranostic pair. This paper aims to review the work on scandium isotopes production, coordination chemistry, radiolabeling, preclinical studies and the very first clinical studies. Finally, standardized procedures for scandium-based radiopharmaceuticals have been proposed as a basis to pave the way for elaboration of the Ph.Eur. monographs for perspective scandium radionuclides.


In the quest for new radionuclides providing personalised adjustment of radiation characteristics to optimize the efficiency of medical care or therapeutic benefit for particular groups of patients (the so called “precision medicine”), the scandium radionuclides have recently received considerable interest. In the last two decades, several new radionuclides for diagnostic imaging and therapy have been successfully introduced to clinical practice. Additionally, using the same targeting vectors and the combination of positron emitter gallium-68 (68Ga) for diagnostic imaging and matching therapeutic counterpart beta-emitting radionuclides such as lutetium-177 (177Lu) and yttrium-90 (90Y) or, recently, alpha emitters bismuth-213 (213Bi) and actinium-225 (225Ac), for therapeutic use, has been recognized as a clear advantage over currently available treatment options. From this perspective, radionuclides of scandium, namely scandium-43 and scandium-44 (43/44Sc) as positron emitters and scandium-47 (47Sc), beta-radiation emitter, seem to fit ideally into the concept of theranostic pair (Mausner and Srivastava 1993; Huclier-Markai et al. 2018). Scandium-47 is also a low energy γ-emitter and allows SPECT and planar imaging. Therapeutic potential of 47Sc was studied at the Brookhaven National Laboratory already in the 1990s (Mausner and Srivastava 1993; Pietrelli et al. 1992). However, the rapid growth of the scandium radionuclides applications started only after the introduction of 68Ga-labelled compounds for PET diagnosis in the early 2000s. Scandium-44 was proposed as a potential alternative to 68Ga for clinical applications in 2010 by the group of Rösch in Mainz, Germany (Roesch 2012; Pruszynski et al. 2010). Similarities in chemistry between 68Ga and 44Sc and their different physical properties have opened a wider avenue for applications of other Sc radionuclides.

Scandium radioisotopes can be produced in accelerators and in nuclear reactors (Huclier-Markai et al. 2018; Muller et al. 2018; Mikolajczak et al. 2019). In particular, due to their large number, biomedical cyclotrons accelerating protons up to 20 MeV are expected to provide 43Sc and 44gSc in quantities that would allow their wider use in diagnostic imaging (Roesch 2012; Synowiecki et al. 2018). Similarly, cyclotrons accelerating protons to higher energies, nuclear reactors and electron linear accelerators (linacs) might be a source of 47Sc for therapy (Domnanich et al. 2017a; Jalilian et al. 2020; Qaim 2019). However, due to several production routes possible, the unification of quality parameters of Sc radionuclides has not been yet attempted.

Scandium isotopes with medical potential

Natural scandium has only one stable isotope, scandium-45 (45Sc). Several other scandium isotopes can be produced artificially, however, most of them are short-lived, in seconds range ( Those with atomic mass smaller than 45 decay by emission of positrons, while those with atomic mass greater than 45 emit electrons. Characteristics of Sc radioisotopes with longer half-lives are given in Table 1 (adapted from Pawlak et al. 2019). Among them, 43Sc, 44,44mSc and 47Sc are suitable for medical applications.

Table 1 Physical characteristics of relevant Sc radioisotopes

Nuclear data and production methodologies

Production routes for 43Sc, 44g, 44mSc and 47Sc use either calcium, titanium or vanadium targets, typically enriched for the desired isotope. A summary of nuclear reactions, decay data, cross-sections, and targetry is given in Table 2, together with the chemical form, purity, enrichment and cost assessment of the target material (where available). For each of the radionuclides, we also report the production mode (i.e. accelerator, nuclear reactor, radionuclide generator) and examples of production facilities that have already implemented their production.

Table 2 Nuclear Data (NUDAT2), production methodologies and irradiation sites of scandium radionuclides

43Sc production

$$ {}^{43}\mathrm{Sc}\ \Big({\mathrm{T}}_{1/2}=3.89\;\mathrm{h},{{\mathrm{E}}_{\upbeta}}^{+}=476\;\mathrm{keV}\ \left(88.1\%\right),\mathrm{branching}\ \mathrm{ratio}\ {\upbeta}^{+}:88\%,\mathrm{E}\upgamma =372\;\mathrm{keV}\ \left(22.5\%\right) $$

Several production routes are possible for 43Sc, involving proton, deuteron or alpha beams (Chaple and Lapi 2018; Braccini 2016). These differents production routes are described in more detail below.

Production of 43Sc using alpha beams

Reasonable activities of 43Sc can be produced by irradiating a natCa target. It is obtained both directly throught 40Ca(α,p)43Sc and 40Ca(α,n)43Ti (T1/2 = 509 ms) → 43Sc nuclear reactions, using accelerators. Effective production of 43Sc using alpha beam was presented by Walczak et al. (Walczak et al. 2015). Szkliniarz et al. (Szkliniarz et al. 2016) irradiated natural calcium as calcium carbonate with 20 MeV alpha particle beam. Resulting 43Ti decays to 43Sc during the irradiation and no chemistry is needed to separate 43Sc from 43Ti due to its very short half-life. Impurities 44mSc and 44gSc are formed due to the 42Ca content (0.65%) in the natural calcium target as well as 46Sc and 47Sc due to the 44Ca content (2% of natural calcium). The 42Ca(α,np + pn)46Sc nuclear reaction has a maximum cross-section around 30 MeV, thus its formation can be decreased by using lower beam energy. Conversely, 47Sc formation cannot be avoided, as it has a similar maximum (17 MeV) as the 40Ca(α,p)43Sc reaction (14 MeV). The radionuclidic purity of the produced 43Sc was 99.95% at the end of bombardment (EOB) and 98.9% at 20 h after EOB. The radionuclidic purity decreases with time as 43Sc half life is lower than that of 44mSc, 46Sc and 47Sc. The use of enriched 40CaCO3 target material decreased dramatically the level of radionuclidic impurities to 1.5 × 10− 5%. Thick target yield for irradiation of calcium metal was found to be 240 MBq/μAh, from which 15 GBq production yield was extrapolated for 4 h irradiation with a 25 μA alpha particle beam. Based on the experimental results, 1 GBq yield of 44mSc is expected for high beam current alpha irradiation (12 h, 25 μA) of enriched 42CaCO3 target (Szkliniarz et al. 2016). Using alpha beam irradiation, Minegishi et al. (Minegishi et al. 2016) reported a remote method for 43Sc production on an unsolidified, powder calcium oxide (CaO) target. The powdery CaO target material was dissolved in situ in HCl in the target box and remotely recovered as a radio-Sc solution as it is done when using liquid targets. The yield of 43Sc following isolation via a precipitation method with a typical 0.22 μm sterile filter was 54.8 MBq/μAh at the end of separation (approximately 1.5 h from the EOB).

Production of 43Sc using protons

The number of cyclotrons providing regular and intense alpha beams is limited, therefore, production methods using medical cyclotrons (proton beams) are gaining more interest. Domnanich et al. (Domnanich et al. 2017b) demonstrated that 43Sc can be produced at a medical cyclotron via proton irradiation of enriched 43Ca or 46Ti oxide target. The production via the 46Ti(p,α)43Sc nuclear reaction yielded a 43Sc activity around 200 MBq and high radionuclidic purity (> 98%) at the end of a 7 h irradiation when using 97% enriched 46Ti. The production via the 43Ca(p,n)43Sc nuclear reaction resulted in higher quantities of 43Sc, but the product consisted of a mixture of 43Sc and 44gSc and an activity ratio of 2:1 when using 57.9% enriched 43CaCO3. This may be increased with higher enrichment of the 43Ca target (max available is 90% at the moment). However, the question remains whether 44gSc is a real problem as it decays with almost the same half-life by emitting a positron, as 43Sc.

Production of 43Sc using deuterons

43Sc can be also produced by deuteron bombardment of an enriched 42Ca oxide target, although the method has not been yet practically utilized. Recently, measurements of the 43Sc production cross-section in the reaction 42Ca(d,n)43Sc with a deuteron beam have been reported by Carzaniga et al. (Carzaniga et al. 2019; Carzaniga and Braccini 2019). The authors also studied practical aspects of producing 43Sc via this route using commercially available targets. Yet, the limited number of medical cyclotrons currently in operation offering deuteron beams prevents the wider application of this method.

In summary, 43Sc can be produced with proton and deuteron beams that are easily available worldwide, the use of enriched (and expensive) target material, 43Ca or 44Ca, is mandatory due to their low natural abundance. When proton irradiation route is preferred, the issue of co-produced 44gSc in 43Sc needs to be addressed (Sitarz et al. 2018). On the other hand, irradiation with an alpha particle beam allows to use directly natural calcium as the nuclear reaction of interest involves 40Ca (natural abundance is 96.94%). However, such a kind of beam is only available in a few places in the world. Though, higher radionuclidic purity can be obtained using enriched 40Ca target, that is more affordable than other enriched calcium materials.

44gSc production

$$ {}^{44\mathrm{g}}\mathrm{Sc}\ \left({\mathrm{T}}_{1/2}=3.97\;\mathrm{h},{{\mathrm{E}}_{\upbeta}}^{+}=632\;\mathrm{keV}\ \left(94.27\%\right),\mathrm{branching}\ \mathrm{ratio}\ {\upbeta}^{+}:94.3\%,{\mathrm{E}}_{\upgamma}=1157\;\mathrm{keV}\ \left(99.9\%\right)\right) $$

44gSc can be obtained directly by irradiation of an enriched 44Ca target of different chemical forms or via the use of the 44Ti/44gSc generator. The main difference of the two methods is that in the first case, there is a co-production of 44mSc (T1/2 = 58.61 h) which is not present in the generator produced 44Sc.

44Ti/44gSc generators

44Ti/44gSc generator has been proposed as a source of 44gSc (Roesch 2012; Filosofov et al. 2010; Radchenko et al. 2016). Although the generator method has been extensively investigated, only a small number of facilities worldwide use these generators. The parent radionuclide, titanium-44 (T1/2 = 60 y) is produced through the nuclear reaction 45Sc(p,2n)44Ti. Production of mCi amount of 44Ti is difficult due to its long half-life and the low cross section (probability) of this nuclear reaction. As an example, irradiation of 1.5 g of Sc target material produced about 185 MBq (5 mCi) (Chaple and Lapi 2018).

Production of 44Sc using protons

It is thus easier to use an accelerator to produce 44gSc. Medical cyclotron can be used for proton irradiation of a target containing either natural (Severin et al. 2012) or 44Ca enriched target (Krajewski et al. 2013). Enriched targets are preferred to avoid the production of contaminants such as 46,48Sc (Domnanich et al. 2017a). The higher the 44Ca enrichment, the higher the final radionuclidic purity. However, during the irradiation, 44mSc is co-produced which may be either an advantage or a disadvantage depending on the application. Indeed, its long half life allows 44mSc (T1/2 = 58.6 h) to be used to image slow biological processes. This is possible since 44mSc/44gSc can act as an in-vivo generator (Alliot et al., 2015a). The production of 44gSc via the 44Ca(p,n)44gSc nuclear reaction has been implemented at the research cyclotron at Paul Scherrer Institute, Zurich, providing this radionuclide with high radionuclidic purity (> 99%) and at high activities (> 2 GBq) (van der Meulen et al. 2015). This would be the method of choice when one wants to produce cost effectively 44gSc. Other targets or projectiles have also been used, presenting some advantages for dedicated cases: higher contribution of 44mSc for example or lower cost when using natCa. The use of natural calcium metal provides cost-efficient access to 44gSc for preclinical experiments. It can be pressed into the cavity of an appropriate target holder (coin or shuttle) and irradiated with high currents, as it has good heat conductivity. However, contaminating radiometals prevent the use of this production route for human application. Exotic production routes have been also studied using 47Ti in a 47Ti (p,α)44Sc nuclear reaction (Loveless et al. 2021).

Production of 44Sc using deuterons

Deuteron beams can also be used for production of 44Sc in a nuclear reaction 44Ca(d,2n) 44Sc/44mSc (Alliot et al. 2015a). A proof of principle has been demonstrated at low beam current (0.3 μA). An activity of 90 MBq (4 h after EOB) can be obtained with a 3 h bombardment on a 500 μm thick 44CaCO3 target at 17 MeV (Alliot et al. 2015b). This study showed also that the use of deuterons allows to increase the production of 44mSc with respect to proton irradiation keeping 44gSc at the same level as for protons (Duchemin et al. 2015; Duchemin et al. 2016). However, to favor 44mSc production the use of alpha beam is recommended (Szkliniarz et al. 2016).

47Sc production

$$ {}^{47}\mathrm{Sc}\ \left({\mathrm{T}}_{1/2}=3.349\ \mathrm{d},{\mathrm{E}}_{\upbeta -}=162\;\mathrm{keV}\ \left(100\%\right),{\mathrm{E}}_{\upgamma}=159\;\mathrm{keV}\ \left(68.3\%\right)\right) $$

Scandium-47 can be produced via several different nuclear reactions using a nuclear reactor, a cyclotron, or a linac (Srivastava 2013).

Production of 47Sc using neutrons

In a nuclear reactor, practically n.c.a 47Sc is produced via 47Ti(n,p)47Sc nuclear reaction by irradiation of 47Ti target with fast neutrons (energy greater than 1 MeV). The 47Ti(n,p)47Sc route also requires an enriched target, though 47Ti oxide is available with very high enrichment and at a reasonable cost. Thus, using the 47Ti(n,p)47Sc nuclear reaction could provide quantities sufficient for therapy. For example, in a HFIR reactor (Oak Ridge National Laboratory) a 3.35 day (one half-life of 47Sc) irradiation of 10 g target could produce approximately 2800 GBq of 47Sc at EOB (Kolsky et al. 1998; Mausner et al. 1993). However, along with 47Sc, 46Sc is co-produced (Domnanich et al. 2017a; Bokhari et al. 2009; Bartoś et al. 2012). This method suffers from the limited number of facilities delivering fast neutrons.

Alternatively, 47Sc can be produced with thermal neutrons (Mausner 1998; Deilami-Nezhad et al. 2016) that are more widely available. Neutron capture on 46Ca produces 47Ca (T1/2 = 4.5 d), which decays into 47Sc by β emission: 46Ca(n,γ)47Ca → 47Sc, and the obtained 47Ca can be further exploited as the 47Ca/47Sc generator system (Mausner 1998). This method suffers mainly from the low natural abundance of 46Ca (0.004%). Still, 47Sc production from neutron irradiated natural Ca target was shown to be feasible (Gizawy et al. 2020). Nontheless, to obtain significant activity of 47Ca the target enriched in 46Ca must be used (presently 46Ca is available with a maximum 30% enrichment) which is rather expensive (Chakravarty et al. 2017) but recycling would allow to drastically reduce the cost. For comparison, when 0,97 mg of 46Ca (48.5 mg of 5% enriched [46Ca]CaCO3) was irradiated in a thermal neutron flux 1.2 × 1014 ns− 1 cm− 2 for 6 days, around 700 MBq of 47Ca and 350 MBq of 47Sc were produced at EOB (Pawlak et al. 2019). The produced 47Ca decays to 47Sc with a half-life of 4.5 days, which is longer than the half-life of 47Sc, enabling multiple separations of in-grown 47Sc in the generator-like system. Using this approach, Domnanich et al. (Domnanich et al. 2017a) demonstrated that up to 2 GBq 47Sc can be produced by thermal neutron irradiation of enriched 46Ca targets. The optimized chemical isolation of 47Sc from the target material allowed the formulation of up to 1.5 GBq 47Sc with high radionuclidic purity (> 99.99%) in a small volume (700 μL), which was useful for labeling purposes. Three consecutive separations within 1 week were possible by isolating the in-grown 47Sc (Domnanich et al. 2017a; Pawlak et al. 2019).

Production of 47Sc using protons

Scandium-47 can also be produced using high energy proton reaction on 48Ti (Srivastava 2013; Srivastava and Dadachova 2001). The nuclear reaction on 48Ti was quite popular as the natural abundance is quite high (73.72%) and the low cross section of the 48Ti(p,2p)47Sc nuclear reaction can be partly compensated by the use of thick targets. However, this method suffers from the co-production of 46Sc (T1/2 = 83.79 d) which emits high energy gamma rays abundantly. This impurity is a major concern in the production of 47Sc both for the risk of unnecessary radiation dose and the regulatory constraints when stored at the therapy wards (Jafari et al. 2019).

Alternative production routes have been explored to try to overcome this issue using 48Ca, natV, 44Ca targets or photonuclear reactions (Mausner and Srivastava 1993; Szkliniarz et al. 2016; Sitarz et al. 2018; Srivastava and Dadachova 2001). The cyclotron production of 47Sc via the 48Ca(p,2n)47Sc nuclear reaction with a proton energy range of 24 → 17 led to a radionuclidic purity of only around 87%, due to 48Sc co-production (Misiak et al. 2017). Using enriched 48Ca for irradiation with 20 MeV protons may be a feasible route for the production of GBq activity levels of 47Sc, however, the prohibitively high cost of enriched 48Ca has made it impossible to implement this production route to date (Misiak et al. 2017). Still, the production of 47Sc in medical cyclotrons providing proton beams at energy range of 15–20 MeV via the 48Ca(p,2n)47Sc nuclear reaction on 48Ca enriched calcium oxide target could potentially provide wide access to this radionuclide (Braccini 2016). However, the co-produced 48Sc undermines the 47Sc purity, and its content strongly depends on the energy of protons impinging the target and on the thickness of the target material, feasibility of this approach has been studied in detail (Carzaniga and Braccini 2019). Another production route proposes the use of natural vanadium targets, since natural vanadium consists of two isotopes: for more than 99.75% it is formed by stable 51V while the very long-lived 50V (T1/2 = 1.4 × 1017 y) occurs only in 0.25%. Experimental data on natV can hence be interpreted in broad energy range as coming from a monoisotopic 51V target and reaction cross-sections can be derived. The cross-sections of the natV(p,x)47Sc nuclear reaction were measured up to 70 MeV proton beam (Jafari et al. 2019; Pupillo et al. 2019; Ditroi et al. 2016), the low reaction yields are the disadvantage of this approach.

Production of 47Sc using alpha beams

The α-particle irradiation of 44Ca targets at a cyclotron, inducing the 44Ca(α,p)47Sc nuclear reaction has been considered, though with low yield and radionuclidic purity. The advantage of the 44Ca(α,p)47Sc reaction lies in the short range of α projectiles in Ca target, allowing the use of a relatively small amount of 44Ca target material for small scale studies with 47Sc. For example, it is reported that 200 mg of [44Ca]CaO prepared in a diameter of 10 mm would give a yield of approximately 11 MBq at 10 eμA for 2 h irradiation at the end of preparation (approximately 1.5 h from the EOB) in the energy range of 28 → 0 MeV (Minegishi et al. 2016).

Production of 47Sc in photonuclear reactions

Photonuclear reactions in electron linear accelerators (linacs) using titanium (Jafari et al. 2019; Yagi and Kondo 1977; Mamtimin et al. 2015; Rotsch et al. 2018) and calcium targets (Starovoitova et al. 2015; Rane et al. 2015) have been explored. However, the cross sections are very small. It may be more convenient to use neutrons and 46Ca. Mamtimin et al. (Mamtimin et al. 2015) studied the production of 47Sc via the 48Ti(γ,p)47Sc nuclear reaction by Monte Carlo simulations. Rotsch et al. (Rotsch et al. 2018) evaluated the production yields and purification of the photonuclear-produced 47Sc from natural titanium oxide targets. In the recent report on photonuclear production Loveless et al. (Loveless et al. 2019a) used eLINAC to produce 47Sc via 48Ti(γ,p)47Sc reaction. They irradiated a stack of natural titanium foils using bremsstrahlung radiation generated by impinging 22 MeV electrons onto a 0.762 mm thick tungsten radiator. Despite the long irradiation times (10.5–14 h) low activity (approx. 2 MBq) was produced with 90% 47Sc, 1.2% 46Sc and 8.3% 48Sc at EOB. The authors suggested the use of enriched titanium, available in oxide form for low cost. It is expected to improve the attainable radionuclidic purity, but not significantly increase the yield, as the natural abundance of 48Ti is 73.7%. Importantly, the use of oxide target material requires a special target design, to effectively dissipate the heat during irradiation. Accelerator-based photoproduction of 47Sc in 48Ca(γ,n)47Ca → 47Sc nuclear reaction was also reported by Starovoitova et al. (Starovoitova et al. 2015) and Rane et al. (Rane et al. 2015), the latter aiming to develop the 47Ca/47Sc generator obtained from irradiated 48Ca target.

To date there is no preferred 47Sc production route. The presence of radionuclide contaminants is expected in all cases. The choice depends on the availability of the irradiation sites and enriched target materials. For example, higher 46Ca enrichment may favour its neutron irradiation. At the sites operating linear electron accelerators the photonulear reactions will be preferred (Qaim 2019).


Calcium enriched in 44Ca, required to reach high radionuclidic purity of 44Sc, is available only in salt form (oxide, carbonate). The low heat conductivity of the target limits the beam current during irradiation. At high beam currents, the heat accumulation caused burnout and/or cracking of the irradiated target and may lead to gas production through the thermal dissociation of CaCO3 (Wojdowska 2019). In some application, the issue of low thermal conductivity was handled by pressing the calcium carbonate on top of graphite powder to facilitate heat transfer and to hold the calcium carbonate powder in position (van der Meulen et al. 2015). The improvment of the heat conducitivity is paid by a lower production as part of the projectile will interact with carbon atoms. However, this method yielded up to 2 GBq of 44Sc when using proton beam energies of near 11 MeV, but can not be automatized easily (see Table 2). Calcium oxide obtianed through thermal decomposition of calcium carbonate, both natural and enriched in 44Ca, was used for preparation of disk shaped pellets which were encapsulated into aluminum for irradiation. Thus increased density of calcium oxide improved the irradiation yield compared to carbonate, though due to the higroscopicity of CaO to the pellets needed to be protected from moisture during storage and when exposed to irradiation (van der Meulen et al. 2020).

Alternatively, magnesium or aluminium powder were used as an additive for target pellet preparation (Mikolajczak et al. 2018; Stolarz et al. 2018). Nuclear reactions on these atoms under 16 MeV proton energy lead to short lived isotopes mainly, resulting in relatively little radioactive contaminants at the end of irradiation. The magnesium pellet can be selectively dissolved from an aluminium target holder with 3 M hydrochloric acid. Aluminum is also easily dissolved after removal from the holder coin.

Introduction of the powdery target material (calcium oxide or calcium carbonate) directly into the cyclotron represents a high risk of contamination as the material may evaporate during irradiation. If a metal foil is used as a target material cover to prevent evaporation, the choice of metal needs careful consideration as it will also act as a degrader. When the foil is too thin, it may separate from the target material during irradiation and may burn out at higher beam currents, as it was reported at 27 μA 16 MeV on a 12.5 μm aluminum foil by Severin et al. (Severin et al. 2012). Conversely, if a higher degrader thickness is required to adjust optimal energy on the target, it has to be actively cooled (van der Meulen et al. 2015). In medical cyclotrons it is safer to use helium-cooled HAVAR foils to separate the irradiated target material from the vacuum system of the cyclotron (e.g., the ARTMS system). In the target system developed in Debrecen (Mikolajczak et al. 2018), the original parts from a water target were used as an interface to the beam port of the cyclotron (see Fig. 1). This enabled the circulation of helium cooling gas between the two foils. After several months of regular use, the contamination caused by calcium evaporation was clearly visible on the internal surfaces of the target system on the second foil. However, no carbonate migrated through the foils to the cooling cycle, or to the high vacuum side.

Fig. 1
figure 1

Solid target system with pneumatic target transfer, developed at the University of Debrecen

The high investment costs and the complicated installation are usually mentioned as major drawbacks of a solid target system. Due to the more complex operation, solid target handling systems will be always more expensive than a simple liquid target but will produce 10 times more radioactivity. Installation of a shuttle type solid target system (commercially available from several companies) requires 4–5 cm diameter passage in the walls of the cyclotron bunker and the hot lab, and some free space in the floor duct between them. However, this can be simply made at one end and nearly impossible at another in already existing facilites.

The challenges related to solid target systems prompted the investigation of 44Sc production in a liquid target. Hoehr et al. (Hoehr et al. 2014) irradiated approximately 1.5 g/mL natural calcium nitrate solution with 13 MeV proton beam in a relatively small volume liquid target. 28 MBq 44Sc was produced with a 20 μA beam current for 1 h, which can reach a gigabecquerel level, if enriched target material is used. However, it is questionable, whether the production can be managed in a cost-efficient way given that a high amount of enriched material has to be irradiated (approx. 100 times more, than for a solid target production from 44CaCO3). The use of liquid targets for radiometal production offers easier installation and lower hardware investment costs, but the presence of corrosive liquids on a cyclotron, which is producing 18F for daily FDG production is of high concern. IBA (Nirta Ga liquid) and GE (gallium target) also offer liquid targets for radiometal production, however, evidence showing their long-term use is needed. From the regulatory point of view, it might be necessary that a liquid target is used for the production of only one radiometal. In contrast, a solid target system can be used to irradiate targets for several radiometals using dedicated target holders.

Gelbart et al. (Gelbart and Johnson 2019) developed a hybrid target system to overcome the installation problems of shuttle type systems. The equipment, used for target dissolution is located in the cyclotron bunker behind the irradiated target, allowing the transport of the produced radioisotope in solution to the hot lab. It was designed for 99mTc and 68Ga production, but can be utilized for scandium as well.

Chemical processing

Several separation methods have been described in the literature, each strongly related to the form of the initial target material. Those proposed for separation of scandium comprised mostly solvent extraction (Zhang et al. 1997; Vibhute and Khopkar 1985; Rane and Bhatki 1966; Radhakrishnan and Owens 1972; Kalyanaraman and Khopkar 2002). However, for the radioactive isotopes of scandium extraction chromatography is preferred. Cartridge- or small column-based extraction chromatography methods can be automatized easily, and enable reproducible purification. Considering the further development of scandium radionuclides production, the technical issues associated with irradiation, target handling, dissolution and processing need to be solved. For routine application in clinical trials and future diagnostic use, these steps should be automated in order to facilitate the production and to meet the requirements of GMP and radiation protection. Pourmand et al. (Pourmand and Dauphas 2010) reported the strong Sc(III) retention on DGA resin and the negligible Ti(III) retention at HCl molarities below 6 M. After elution with 4.0 mL of HCl 0.1 M scandium is concentrated in a smaller volume by the use of second column containing SCX cation exchange resin and eluted with 4.8 M NaCl/0.13 M HCl effluent. The same approach was used by van der Meulen et al. at PSI, Villigen (van der Meulen et al. 2015). A summary of different approaches developed in the literature is given in Table 3.

Table 3 Separation methods related to the initial target material for getting scandium radionuclides

The purity of the resulting batches is of great importance. Most of the published works dealt with 44Sc. These studies have shown that a high chemical purity of the final 44Sc fraction is important, since the presence of other metals may interact with the DOTA-chelator (or any other chelator) and, hence, the radiolabeling yield would be compromised. The concentration of common environmental contaminants (Al(III), Cu(II), Pb(II), Zn(II), Fe(III)) in 44Sc is frequent. The most problematic is Fe3+ for which the stability with the DOTA ligand is greater than that for Sc3+. By contrast, the influence of divalent metal cations (considered as contaminants) is negligible due to the much lower stability of their DOTA complexes. In order to meet the requirements for radiopharmaceutical applications, the obtained final solution containing scandium radionuclide needs 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. Most of the times, authors do not indicate the final chemical purity of the resulting bacthes.

Target recovery

Calcium has six naturally occurring isotopes (40Ca, 42Ca, 43Ca, 44Ca, 46Ca, and 48Ca), where 40Ca is the most abundant, comprising about 97% of naturally occurring calcium. However, calcium doesn’t form gaseous compounds near room temperature and atmospheric pressure, hence commonly used enrichment schemes such as gas centrifuge or thermal diffusion are not strictly impossible, but challenging. On the other hand methods based on liquid calcium forms, such as ion exchange chromatography or electrophoresis are difficult to scale up. Hence, enriched calcium materials are very expensive and a recycling process of the target needs to be developed (Jalilian et al., 2020). The choice of method used in the recovery process is related to the form of the initial target material, and it is particularly important when large amounts of the highly enriched target material are used. The recycling will contribute to a significant cost reduction, considering that the target material alone costs around 20–25 € per 1 mge (mg element) of [44Ca]CaCO3 dependig on the purchased quantity (price for 2020). Additionally, the purchased material needs to be purified before use so the method for recycling applies here as well.

The published data on this topic is quite limited. Few publications deal with the calcium carbonate targets (natural or enriched in 44Ca) for production of 44Sc in cyclotrons. One of the early works by Krajewski et al. (Krajewski et al. 2013) reported the use of chelating resin Chelex 100 for separation of 44Sc and target recycling efficiency of only 60%. In the work by Alliot et al. (Alliot et al. 2015a), using the mixture bicarbonate/methanol, the rate of solvent evaporation was increased and the solubility of calcium carbonate was lowered (Kan et al. 2002). The recovery yield of enriched calcium of 90 ± 2% was obtained when starting with CaCO3 material and the HCl solutions from the extraction process. The solution was then loaded on a pre-conditioned AG1 × 8 column to retain all metallic impurities (Cu, Co, Fe…) and enriched 44Ca was rinsed in 9 M HCl. This solution was evaporated to dryness and the dry residue recovered in a mixture of bicarbonate buffer (pH = 10.33)/methanol, filtered and the residue dried in an oven at 105 °C to remove water and methanol. The recycled targets were irradiated, and no significant difference in production yield was observed (Alliot et al. 2015a). The improved recovery reaching 98% of the initial enriched material, significantly reducing the operational costs, was reported by Van der Muelen et al. (van der Meulen et al. 2015) who collected effluents from the DGA column and recycled enriched calcium through simple and fast precipitation as Ca-oxalate, which was then converted to carbonate by slow heating to 500 °C. Alternatively, the final step involved heating at 900 °C to ensure conversion to CaO. It was shown that recycled material could be used effectively for 44Sc production (van der Meulen et al. 2020). In the production of 47Sc with alpha beams Minegishi et al. (Minegishi et al. 2016) recycled enriched target, collected as 44Ca2+ in the waste fraction via the precipitation with ammonium carbonate and further dehydration to 44CaO by decomposition of 44CaCO3 at 940 °C for 2 h. The recovery rate of approximately 85–90% (by weight), with negligible loss.

When starting material is TiO2, the enriched titatnium may be recovered and reused. A simple procedure was developed that recovers around 98.5% of the oxide based on precipitation of titanium at basic pH followed by conversion of the oxide using higher temperature (Kolsky et al. 1998). Other processes for recovery of titanium from its acidic solutions use either HF and HNO3 or converting to nano-sized titanium dioxide (Kolsky et al. 1998). The use of TiO2 as a starting material and its recycling process have been reported by Loveless et al. (Loveless et al. 2019b). The HCl solution and HNO3 obtained from the extraction processes were collected and NH4OH at pH 8 was added to precipitate TiO2. The precipitate was then heated for 4 h at 400 °C. Optionally, the enriched titanium oxide can be reduced with calcium or calcium hydride, resulting in titanium pellets with low content of impurities (Lommel et al. 2013).

Chelators for Sc

Scandium with its ionic radius (ri) 74.5 pm (CN = 6) is chemically similar to Y3+ and the heaviest lanthanides. Similarly to them, scandium is almost exclusively present in its compounds in the trivalent state. Therefore, ligands developed for these cations should be also suitable for chelating Sc. However, the chemistry of trivalent scandium has some differences compared to lanthanides; it is smaller (thus, harder and has a higher preference for hard oxygen donor ligands), and prefers donor numbers from six to eight (Kerdjoudj et al. 2016).

The multi-dentate ligands, which were already used in Gd(III)-based MRI contrast agents as well as for radiolanthanides, that is derivatives of DTPA or DOTA, were also the first choice for scandium chelation for medical applications (Mausner et al. 1995). The stability constant of Sc-DOTA complex was comparable to that for Lu3+ and heavier lanthanides but higher than those for In3+ and Ga3+. The 13C NMR studies have shown that Sc(DOTA) similarly to Lu(DOTA) forms in solution complexes with eight-coordination geometry (Majkowska-Pilip and Bilewicz 2011). The stability constants of scandium(III) complexes DTPA and DOTA (log KScL 27.43 and 30.79 respectively) were determined from potentiometric and 45Sc NMR spectroscopic data. Both complexes were fully formed even below pH 2. Complexation of DOTA with the Sc3+ ion was much faster than with trivalent lanthanides. Proton-assisted decomplexation of the [Sc(dota)] complex (τ1/2 = 45 h; 1 M aq. HCl, 25 °C) was much slower than that for [Ln(dota)] complexes (Pniok et al. 2014). Therefore, DOTA and its derivatives were assumed to be very suitable ligands for scandium radioisotopes (Pniok et al. 2014; Huclier-Markai et al. 2015).

Thermodynamic data for scandium(III) complexes with polyamino-polycarboxylic ligands, such as NOTA, EDTA or TETA have been determined using potentiometric titration and free ion selective radiotracer extraction (FISRE) method and the values of stability constants were found to be in the order TETA<NOTA<EDTA<DTPA<DOTA. (Huclier-Markai et al. 2011) DOTA derivatives with phosphinic/methylphosphonic acid pendant arms (i.e. DO3A; DO3APPrA, DO3APABn) were also investigated showing that the stability constants of the monophosphinate analogues were somewhat lower than that of the Sc(DOTA) complex. (Kerdjoudj et al. 2016) The thermodynamic stability constant of recently developed chelating agent AAZTA, Sc(AAZTA) was reported to be lower than that of Sc(DOTA) but the striking difference was observed on the radiochemical yield at 25 °C indicating that AAZTA quickly incorporated 44Sc (Nagy et al. 2017a). The AAZTA derivative AAZTA5 (1,4-bis (carboxymethyl)-6-[bis (carboxymethyl)]amino-6-[pentanoic-acid]perhydro-1.4-diazepine) was synthesized representing a bifunctional version with a pentanoic acid at the carbon-6 atom. [44Sc]Sc-AAZTA5 complexes as well as [44Sc]Sc-AAZTA5-TOC were formed at room temperature within 5 min in the pH range 4 to 5.5 and were very stable (Sinnes et al. 2019). Another chelator, H4pypa (N5O4) has been shown to exhibit high complexation constant with Sc (log K = 27) as well as with In, Lu, Y and La, as determined by potentiometric titration and UV spectrometry (Li et al. 2019a; Li et al. 2019b). Its radiolabeling could be performed at room temperature and in a quite wide range of pH values within 10 min. Also, it has been conjugated to a Glu-urea-Lys based PSMA (prostate-specific membrane antigen, e.g. PSMA-617). When labelled with 44Sc (pH = 4.5, 30 min RT) this conjugate showed specific tumor uptake (4.86% ID/g from ex vivo biodistribution, 8 h p.i.) without significant off-target uptake, except in the kidney. (Li et al. 2019b) Further search on chelators enabling Sc chelation at room temperature led to the development of the small-cavity triaza-macrocycle-based, picolinate-functionalized chelator H3mpatcn. Spectroscopic and radiochemical studies established the [Sc(mpatcn)] complex as kinetically inert and appropriate for biological applications. As a proof-of-concept bifunctional conjugate targeting the prostate-specific membrane antigen (PSMA), picaga-DUPA, chelated 44Sc to form {44Sc}Sc(picaga)-DUPA at room temperature with an apparent molar activity of 60 MBq μmol− 1 and formation of inert RRR-Λ and SSS-Δ-twist isomers. Sc(picaga)-DUPA exhibited a Ki of 1.6 nM for PSMA and the 44Sc labelled Sc(picaga)-DUPA revealed high-quality images in prostate cancer-bearing animals. H3mpatcn and its bifunctional analogue picaga represent new additions to the chelator toolbox for the emerging 44/47Sc theragnostic isotope pair as reported by Vaughn et al. (Vaughn et al. 2020a). Nonetheless, the molar activity calculated from their experimental data was estimated to be 0.03 MBq/nmol and thus the metal to ligand ratio was ranging between 1: 485,500 to 1: 1,084,600. Hence, picagaDUPA seems exhibiting complexation/radiolabeling at RT but complexation constant must be established and the potential of this ligand to provide higher molar activity must be evaluated with sources of 44Sc of high purity.

Scandium also displays favorable physical properties and chemistry for conjugation to mAb-chelate systems. Since it is chemically similar to 90Y and close to 177Lu, the same ligands developed for 90Y or 177Lu can be used for chelating 47Sc.

Table 4 provides a summary of the main chelates that have been evaluated, with the corresponding formula and the measured value of the complexation constant with scandium. The table reports values reported in the original manuscripts, accuracy of which is not homogeneous across the data. For simplicity, the uncertainties (available only for selected studies) are not reported.

Table 4 Selected chelators for scandium and their complexation constants

Radiolabeling studies

Labeling protocols should allow high labeling yields, radiochemical purity and molar activity (Fani and Maecke 2012). Labelling efficiency of ligands is usually tested at various pH, temperature and ligand concentration (or more correctly, at different radiometal-to-ligand molar ratios) and is monitored as a function of time in order to optimize the radiolabeling. Herein, we discuss the development of labelling methods that were developed for scandium radionuclides. Neverthelles, the reported data are often difficult to compare. To overcome this problem, the recommendations on consensus nomenclature rules for radiopharmaceutical chemistry should be followed (Coenen et al. 2017).

Most of the published radiolabeling works have been done with 44Sc. All have highlighted the importance of reaching a high chemical purity of the final 44Sc fraction before radiolabeling. The presence of other metals may interact with the DOTA-chelator (or any other chelator) that can affect the radiolabeling yield. The content of environmental contaminants (i.e. Al(III), Cu(II), Pb(II), Zn(II), Fe(III)) in 44Sc is frequent. The most problematic is Fe3+, for which the stability with the DOTA ligand is greater than that for Sc3+. By contrast, the influence of divalent metal cations (considered as contaminants) is negligible due to the lower stability of DOTA-divalent metal complexes. Although DOTA forms very stable complexes, it exhibits slow formation kinetics at room temperature, that could be increased by heating. However, elevated temperatures remain an important obstacle for efficient labelling of heat-sensitive molecules such as antibodies. Click-chemistry or the new ligands permitting the formation of scandium complexes with faster kinetics, or at much lower temperatures than that required for DOTA, might circumvent this issue. In this context, AAZTA or H4pypa seem to be interesting alternatives (Nagy et al. 2017a; Li et al. 2019a), although their availability is limited.

Table 5 provides a summary of the radiolabeling studies published to date. Notably, there are many discrepancies in presentation of results and in most of the cases the quality control results of the batches are not reported. When available, radiochemical yield and molar activity are indicated. This inhomogeneity of the published data makes the direct comparison of radiotracers efficacy difficult. The data in Table 5 are reported as published in the original papers, pointing to the discrepancies in the measurement units used. Indications on the quality control procedure are also given as well as the conditions of post-radiolabeling purification, wherever reported. The use of suitable buffers, stabilizers/free radical scavengers etc. was addressed.

Table 5 Radiolabeling studies performed on scandium radionuclides

Pre-clinical studies

The use of radiolabeled compounds for in vitro and in vivo preclinical animal studies has steadily increased in recent years, becoming a much more widely used method for studying biochemistry, physiology and pharmacology (Kilbourn and Scott 2018). The increasing availability of 44Sc and its compatibility with numerous chelators, among those DOTA, has led to multiple studies in animals, especially imaging studies. The half-lives of 43Sc, 44Sc or 47Sc are compatible with the pharmacokinetics of a fairly wide range of targeting vectors. Several somatostatin analogs, either DOTA-derivatized (DOTATOC, DOTATATE, DOTANOC) or with NODAGA as a chelator, have been labeled with scandium radionuclides to prove the suitability of the obtained radionuclide solution for radiolabeling (Pawlak et al. 2019; Loveless et al. 2019a) or to compare under the same conditions the in vitro/in vivo imaging/therapeutic potential of scandium radiolabeled peptides to their 68Ga- or 177Lu (90Y) – radiolabeled counterparts (Walczak et al. 2015; Loveless et al. 2019a; Singh et al. 2017; Pruszynski et al. 2012). The scandium radionuclides labeled bombesin analogs (Koumarianou et al. 2012), RGD peptides (Domnanich et al. 2017c), folate derivatives (Muller et al. 2018; Muller et al. 2014; Siwowska et al. 2019) and PSMA ligands (Umbricht et al. 2017) have been evaluated pre-clinically. Few studies have also reported on scandium radiolabeled antibodies, antibody fragments or affibodies (Chakravarty et al. 2014; Honarvar et al. 2017; Moghaddam-Banaem 2012) and nanoparticles (Eppard et al. 2018). A summary is given in Table 6.

Table 6 Current state of pre-clinical studies performed on scandium radionuclides

The early study by Koumarianou et al. (Koumarianou et al. 2011) was conducted on “cold” complexes with natural scandium and gallium. The binding affinity of both natSc-DOTATATE and natGa-DOTATATE in AR42J cell line was in the sub-nanomolar range, but in favor of natGa-DOTATATE. The affinity to GRP (gastrin releasing peptide) receptors in the PC-3 cell line was higher for natGa-DOTA-BN[2–14]NH2 than that of natSc-DOTA-BN[2–14]NH2 (IC50(nM) values 0.85 ± 0.06 vs. 6.49 ± 0.13). Despite this difference, in another study both 68Ga- and 44Sc-labeled DOTA-BN[2–14]NH2 showed comparable biodistribution and micro-PET imaging. (Koumarianou et al. 2012). Similar observations were reported by Ferguson et al. who compared [44gSc]Sc-DOTA-Ava-BBN, a bombesin receptor antagonist, to the [68Ga]Ga-DOTA-Ava-BBN in breast and prostate cancer model (Ferguson et al. 2020). In more recent studies, the binding affinity of natSc-PSMA-617 was evaluated in comparison to natGa-PSMA-617, however using different methods and cell lines (PC-3 PIP cells and PSMA+ LNCaP cells, respectively) (Umbricht et al. 2017; Eppard et al. 2017). Though, the binding affinity to the target was found to be in the same molar range. Both [44Sc]Sc- and [68Ga]Ga-PSMA-617 exhibited similar in vivo behavior, with [44Sc]Sc-PSMA-617 displaying higher tumor-to-liver ratios at 15 and 30 min p.i. These values correlated more closely to [177Lu]Lu-PSMA-617 than to [68Ga]Ga-PSMA-617, and therefore were considered useful for pre-therapeutic dosimetry (Umbricht et al. 2017).

The comparison of 44Sc and 68Ga for imaging of animals with melanocortin-1 receptor (MC1-R), positive tumors using DOTA-NAPamide, at 4 h p.i. revealed significantly higher tumor uptake of 44Sc (81.7 ± 7.7%ID/g) over 68Ga (17.3 ± 1.85%ID/g) (Nagy et al. 2017b). DOTA-puromycin was radiolabeled with the generator produced 44Sc and investigated for the potential imaging of protein synthesis in vivo. In μPET images of tumor-bearing rats significant tumor uptake of [44Sc]Sc-DOTA-puromycin and a clear-cut tumor visualization were demonstrated. In addition, the cellular uptake of [44Sc]Sc-DOTA-puromycin could be suppressed by blocking protein synthesis (Eigner et al. 2013).

In vitro, [47Sc]Sc-folate demonstrated effective reduction of folate receptor-positive ovarian tumor cell viability similar to 177Lu-folate, but 90Y-folate was more potent at equal activities due to the higher energy of emitted β-particles. Comparable tumor growth inhibition was observed in mice that obtained the same estimated absorbed tumor dose (~ 21 Gy) when treated with [47Sc]Sc-folate (12.5 MBq), [177Lu]Lu-folate (10 MBq), and [90Y]Y-folate (5 MBq), respectively. However, there were no statistically significant differences among the therapeutic effects observed in treated groups (Siwowska et al. 2019).

CHX-A”-DTPA has been successfully used for radiolabeling of the monoclonal antibody (mAb) Cetuximab. (Chakravarty et al. 2014) Another study evaluated DOTA-HPMA (N-(2-hydroxypropyl)methacrylamide) conjugates for labeling efficiency with 68Ga, 177Lu and 44Sc and showed that the 44Sc labeled polymer allowed for in vivo PET imaging and ex vivo measurements of organ distribution for up to 24 h (Eppard et al. 2018).

Clinical experiences

The validity, usefulness and advantages of 44Sc have been demonstrated by studies featuring 44Sc-radiolabeled targeting vectors, including 44Sc radiopharmaceuticals in early clinical studies. The first of them has been performed using [44Sc]Sc-DOTATOC prepared from 44Sc produced in a cyclotron (Singh et al. 2017). Two patients were included in this study after being treated by peptide receptor radionuclide therapy due to neuroendocrine neoplasms. The obtained image quality was comparable to that of 68Ga. This very encouraging proof-of-concept study showed no clinical adverse effects with normal hematology, nor with renal and hepatic profiles.

The other studies were conducted in patients with metastasized castrate-resistant prostate cancer (Eppard et al. 2017; Khawar et al. 2018a; Khawar et al. 2018b). A first-in-human investigation was carried out in a cohort of four patients (mean age 70.0 ± 1.8 years) registered for [177Lu]Lu-PSMA-617 therapy. Physiological tracer uptake was observed in kidneys, liver, spleen, small intestine, urinary bladder, and salivary glands and pathological uptake in both soft and skeletal metastases. SUV values were significantly lower in the kidneys (14.0) compared to [68Ga]Ga-PSMA-11 PET (30.5). All other measured SUV values did not show a statistically significant difference. Tumor-to-liver ratios were found to lie between 1.9 and 8.3 for [68Ga]Ga-PSMA-11 and between 2.5 and 8.8 for [44Sc]Sc-PSMA-617 after 120 min. For [44Sc]Sc-PSMA-617 the ratios were higher and no statistically significant differences were observed. Total and % activity were highest in the liver followed by kidneys, spleen, small intestine and salivary glands. Rapid wash-out was seen in liver and spleen, and gradually over time in kidneys. Kidneys received the highest radiation absorbed dose of 0.354 (0.180–0.488) mSv/MBq. No adverse pharmacological effects were observed. Still, the authors concluded that the clinical advantages for individual dosimetry or other applications like intraoperative applications need to be investigated in further studies (Eppard et al. 2017).

The impact of physical properties of 44gSc on image quality has been studied by Bunka et al. (Bunka et al. 2016) and recently evaluated by Rosar et al. (Rosar et al. 2020) in comparison to 68Ga with different imaging phantoms. The lower mean positron energy of 44gSc (0.63 MeV) compared to 68Ga (0.83 MeV) can result in better spatial image resolutions. However, high-energy γ-rays (1157 keV) are emitted at high rates (99.9%) during 44gSc decay, which can reduce image quality. Despite the presence of high-energy γ-rays in 44gSc decay, a higher image resolution of small structures was observed with 44gSc when compared to 68Ga. Structures as small as 1 mm could be visualized and analyzed using two different pre-clinical PET scanners. Recently, Lima et al. (Lima et al. 2020) evaluated quantitative capabilities of 44Sc-PET using a commercial PET scanner and concluded that that clinical 44Sc-PET imaging has the potential to provide signal recovery in lesions of different sizes comparable to current 18F-PET standards.

Biodistribution and radiation exposure to normal organs with [44Sc]Sc-PSMA-617 in metastatic castration-resistant prostate carcinoma patients were investigated by Khawar et al. (Khawar et al. 2018a). These authors extrapolated the pharmacokinetics of [44Sc]Sc-PSMA-617 to that of [177Lu]Lu-PSMA-617 and demonstrated that pharmacokinetics of [44Sc]Sc-PSMA-617 PET/CT imaging could be utilized with the intent of predicting normal organ-absorbed doses and maximum permissible activity in patients scheduled for therapy with [177Lu]Lu-PSMA-617. Physiological tracer uptake was seen in kidneys, liver, spleen, small intestine, urinary bladder, and salivary glands and metastases. Kidneys, with the highest radiation absorbed dose of 0.319 mSv/MBqm were the critical organs, followed by the urinary bladder wall, spleen, salivary glands, and liver. Red marrow dose was found to be 0.0331 mSv/MBq. The mean effective dose of 0.0389 mSv/MBq and effective dose of 1.95 mSv was estimated from 50 MBq (treatment planning dose) of [44Sc]Sc-PSMA-617 (Khawar et al. 2018a). These first clinical studies using [44Sc]Sc-DOTA-TOC and [44Sc]Sc-PSMA-617 confirmed the potential of 44Sc-labeled tracers as promising radiopharmaceuticals, their use is especially relevant in pre-therapeutic dosimetry (Kostelnik and Orvig 2019).

To date, very limited data is available on radiation doses from 47Sc labelled radiopharmaceuticals. The absorbed dose in human organs of 47Sc-EDTMP (ethylene-diamine-tetramethylene-phosphonic acid) has been extrapolated from animal biodistribution using the MIRDOSE formalism. However the animal biodistribution data in this study were collected using EDTMP radiolabeled with the reactor produced 46Sc of rather low specific activity 148 MBq/mg (Deilami-Nezhad et al. 2017). The radionuclide 47Sc is a low energy beta emitter, similarly to 177Lu, and the mean range of its radiation in bone is 0.20 mm, compared to 0.15 mm for 177Lu (as extrapolated from Bouchet et al. (Bouchet et al. 2000)).

Quality specifications

The chemically identical radiopharmaceuticals for diagnosis and therapy using the matched pair 43Sc or 44Sc with 47Sc are very appealing. Although several groups are now producing scandium radionuclides locally and a few centers demonstrated the potential of Sc radionuclides for medical applications, the benefit of their use in the clinical setting is yet to be shown. Given the amount of work and documentation needed to obtain the approval for a clinical trial of a new radiopharmaceutical, this review aims to present the production options for Sc radionuclides and related quality constraints. This information might be useful when implementing Sc radionuclides production technology in the existing or planned production facilities.

43Sc (β+ = 88.1%, Eβ+ = 476 keV) and 44gSc (β+ = 94.3%, Eβ+ = 632 keV) have half-lives of 3.89 h and 3.97 h, respectively (Singh and Chen 2015). Although properties of these two isotopes are very similar, the main difference is the high-energy γ-radiation (Eγ = 1157 keV; Iγ = 100%) associated with 44gSc. The high energy gamma radiation, similarly to that of 89Zr (Eγ909 keV; Iγ = 99%), may not hamper its clinical use. It is considerd advantageous for 44gSc as the radionuclide of choice for 3-γ photons imaging, a new modality of imaging (Sitarz 2020). Additionally, 44mSc, which is co-produced with 44gSc, has a half-life of 58.6 h and has been suggested for use as an in vivo PET generator for 44gSc (Alliot et al. 2015a). However, the longer half-life coupled with the high-energy γ-ray emitted from 44gSc may lead to the unfavorable dosimetry for shorter biodistribution studies (Alliot et al., 2015a).

High radionuclidic purity 44gSc production with proton irradiation necessitates the use of enriched target material and a recycling procedure to keep the cost reasonable. The determination of acceptable limits for each radionuclidic impurity requires careful consideration, as it can have an adverse effect on the implementation of new production methods. Potential contamination with 43Sc can be argued to have no impact on patient safety and quality of images since it is a positron emitter with no high energy gamma radiation component, in contrast to the expected radiation burden due to 44gSc contamination in 43Sc. 44mSc decays to 44Sc with low energy gamma (271 keV) emission, which does not cause any harm to the patient. The limit for 44mSc content should be determined to avoid unnecessary radiation doses to the patient, because of the longer effective half-life of the 44mSc/44Sc isotope pair. However, the biological half-life and critical organs should be determined for unchelated scandium isotopes before using worst-case dosimetry calculations for the determination of the limit of longer half-life impurities, as discussed earlier for 68Ga labelled peptides and influence of 68Ge breakthrough on dosimetry (Velikyan et al. 2013).

From that perspective, 43Sc has the most favorable radiation characteristics for conventional PET, however it can be produced efficiently only using alpha particle beams. 44Sc is in the most advanced state and has the highest potential for broad applications. Medical cyclotrons that currently supply 18F to hospitals can be used to produce 44gSc or 43Sc, with a solid target system. The use of liquid target might not be reasonable as the amount of enriched material nedded is high. With the near to 4 h half-life and reasonable production cross-section, there is a potential for regional distribution following mass production at a single cyclotron unit with solid target. With the increasing numbers of installed cyclotrons, the availability of 43/44Sc may increase. Aiming for the production of high activities, a solid target is needed. This requires the addition of specialized hardware on the cyclotron and in the processing hot-cell, which is available from cyclotron providers but not frequently installed on 18F production machines. The co-emission of a high-energy γ-ray similar to 89Zr has to be taken into consideration when planning for radiation protection. If not controlled, it may increase the radiation dose to the patient and staff. 44Ti/44Sc generator currently has a high production cost and requires a regular and efficient use of the generator over long periods. 44mSc half-life enables transportation of 44mSc-labeled radiopharmaceuticals to hospitals that are located quite far away from the radiopharmaceutical production site (centralized production). This is based on the in-vivo generator principle.

Both positron emitters, 43Sc and 44gSc can be used for theranostic studies with 177Lu or other lanthanides. Scandium has chemistry close to 177Lu and possibility of centralized production at the regional level, which makes it more desirable than 68Ga (local production only) for its use in a theranostic pair. Application of 47Sc, as an alternative radionuclide to 177Lu, was proposed in earlier works (Srivastava 2013). The advantage of 47Sc production compared to that of n.c.a. 177Lu, is the relatively easy isolation of the radionuclide from the target, but the disadvantage is the smaller cross-section of the nuclear reaction compared to 177Lu production. Nonetheless, the main concern lies with the co-produced impurities. Another limitation concerns the availability of the starting enriched material needed for the set-up of efficient separation chemistry. At the time of writing this article, we are aware of several research groups developing ways of overcoming these limitations, as presented by Jalilian et al. (Jalilian et al. 2020).

When it comes to the scandium-based radiopharmaceuticals, unlike with 68Ga, the scientific community involved is unfortunately far from following a standardized procedure. From Table 5, it is inferred that neither the metallic impurity contents nor the radionuclidic purity of the resulting radiolabeled compound is always given. There is also a lack of homogeneity in the expression of the molar activity of the resulting radiolabeled vector. Though, some compromise could be found on the radiolabeling protocol and quality control. This is certainly a good basis to pave the way for a monograph for the European Pharmacopeia, together with the use of the recently approved nomenclature guidelines for radiopharmaceuticals (Coenen et al. 2017).

Next to the radionuclide generators and kits, the European Union (EU) directive 2001/83 (The European Parliament and the Council of European Union 2001) dealing with radiopharmaceuticals, defines also the “radionuclide precursors” (EU Commission 2004). The radionuclide precursor is defined as “any other radionuclide produced for the radio-labeling of another substance prior to administration” and as such has to hold a marketing authorization when introduced to the market. Even if used in a hospital radiopharmacy with a cyclotron on-site, and having a status of starting material, as postulated in a position paper by Neels et al. (Neels et al. 2019), the radionuclide precursor needs to be controlled by performing several quality control tests (identity, purity, assay, etc.). To assure safe use in humans, these tests and limits should comply with certain quality standards, and be supported by a suitable quality management system. It is generally accepted that such quality standards are established in Pharmacopoeia monographs.

The expected final form of the radionuclide, regardless of the production method used, is the form of solution for radiolabeling. The Ph.Eur. General monograph on Radiopharmaceutical preparations (0125) can serve as the reference for establishing quality specifications for the final product solution of 43Sc, 44Sc or 47Sc (EDQM 2020a). One should bear in mind that the Ph.Eur. monographs on Lutetium (177Lu) solution for radiolabelling (mon. 2798) (EDQM 2020b), and Yttrium (90Y) chloride solution for radiolabelling (mon. 2803) (EDQM 2020c), can be also very informative, similarly as the monograph of Gallium-68 (68Ga) chloride solution for radiolabelling (mon. 2464) (EDQM 2020d).

One needs to remember, however, that the specified parameters will depend on the production route (target material, nuclear reaction, accompanying nuclear reactions, chemical processing etc.). For example, when using 44Sc obtained from the 44Ti/44Sc radionuclide generator, the potential breakthrough of 44Ti will be of concern as well as its impact on the quality (and safety) of the final radiopharmaceutical. The detailed specifications for quality control of generator eluate as well as for the quality assessement of 44Sc radiolabeled PSMA-617 prior to its administration to patients were developed by Eppard (Eppard 2018).

The Group 14 (radioactive compounds) of European Pharmacopoeia elaborated the Guide for the elaboration of monographs on radiopharmaceutical preparations, European Pharmacopoeia, EDQM Edition 2018, which should help to prepare the monographs for new radiopharmaceuticals ( Based on this Ph. Eur. Guide, exemplary general list of tested parameters and the involved methods for the potential scandium radionuclide precursors are briefly discussed below.

Title, definition and production sections

In general, the definition states that the monograph applies to the substance obtained by a certain route of production and in the case of a radionuclide precursor the name of the substance is completed by “for radiolabeling”. Therefore, in case of the new monographs, e.g. for gallium-68 and technetium-99 m, the title of the monograph itself contains a reference to the production method, like: Gallium (68Ga) Chloride (Accelerator-Produced) Solution For Radiolabelling (mon. 3109) (EDQM 2020e) or Sodium Pertechnetate (99mTc) Injection (Accelerator-Produced (mon. 2891) (EDQM 2020f).

Such titles and definitions are intended to make the recipient/reader aware that other methods of obtaining radionuclides may and, in fact do, result in a completely different profile of potential contaminants, mainly radionuclidic. These constraints apply to the radionuclides of scandium, therefore it is not possible to develop a single monograph even for one of the scandium radionuclides, but it would be necessary to develop several monographs for this radionuclide depending on the way the radionuclide is produced. In the following, a monograph-like characteristics are proposed.


Appearance: clear, colourless solution.

Half-life and nature of radiation of scandium radionuclides:

Scandium-47: T1/2 = 3.35 d, Eβav = 162 keV, Eγ = 0.159 MeV.

Scandium-43: T1/2 = 3.89 h, 88% Eβ+, Eβ+av. = 476 keV, Eγ = 0.373 MeV (23%).

Scandium-44: T1/2 = 4.04 h, 94.27% Eβ+_and 5.73% EC, Eβ+av. = 632 keV, Eγ = 1.157 MeV (99.9%).

Identification: (characteristic of type of radiation - spectra, approximate half-life, pH, chromatography)

Gamma-ray spectrometry.

The most prominent gamma photon of scandium-47 has an energy of 0.159 MeV.

The most prominent gamma photon of scandium-43 has an energy of 0.373 MeV.

The most prominent gamma photon of scandium-44 has an energy of 1.157 MeV.

pH: 1.0 to 2.0, using a pH indicator strip R.


Radionuclidic purity

For 90Y, 177Lu and 68Ga (generator produced) the Ph.Eur. recommended limit is minimum of 99.9% of the total radioactivity, however, for the accelerator-produced 68Ga (draft monograph 3109 published in Pharmeuropa 30.4) the radionuclidic purity limit is only 98% due to the expected and the inevitable presence of contaminants such as 66Ga and 67Ga.

The expected radionuclide contaminants in 47Sc radionuclide precursor are 46Sc (T1/2 = 83.8 d) and 48Sc (T1/2 = 43.7 h). Although their presence has been reported, the limits for individual radionuclidic impurities were not yet determined.

In case of 43Sc, irradiation of 46Ti targets (97.0% enriched) with protons yielded a product of high radionuclidic purity, containing 98.2% 43Sc and only 1.5% 44Sc. Long-term γ-spectroscopy measurements determined low activity levels of 0.079% 44mSc, 46Sc, 47Sc, 48Sc and 0.34% 44mSc, 47Sc, 48Sc in the final products of irradiated 46Ti and 43Ca targets, respectively. (Domnanich et al. 2017b).

Chemical purity

The ions of copper, iron, zinc and lead and the traces of target material such as Ca or Ti.

Since DOTA is the most common chelator used for labeling with 47Sc, it can be assumed that the content of chemical impurities such as copper, iron, zinc and lead ions may have an impact on 47Sc quality. The content of chemical impurities may vary depending on the processing method and purity of reagents. The methods used for determination of chemical contaminants included: determination of Ti using ICP-MS or determination of metallic impurities by ICP atomic emission spectrometry. These instrumental techniques can be recommended to establish chemical impurity acceptance levels.

Although the influence of metallic impurities on the labelling yields of DOTA-chelated peptides has been studied in detail, the first report dealing with Sc labelled DOTATATE is just recently published by Walczak et al. (Walczak et al. 2019).

Radiochemical purity

In case of 68Ga and 111In, the Ph.Eur. monographs for diagnostic radionuclide precursors (2464, 1227) require the presence of gallium or indium min. 95% in + 3 ion form, whereas for therapeutic radionuclides like 177Lu (mon. 2798) and 90Y (mon. 2803) the limit is increased to min. 99% in + 3 ion form. It seems that these rules should be also adopted in case of 43Sc and 44Sc radionuclide precursors and 47Sc, respectively.

In the Ph.Eur. monographs for 177Lu and 90Y solutions for radiolabeling, for the assessment of radiochemical purity, in addition to the chemical determination of the radiometal form as trivalent cation, the TLC system recommends using complexes of these radiometals with pentetic acid as reference.

To test the radiochemical purity of obtained 47Sc when used for radiolabeling of selected common chelators such as DOTA, DTPA or EGTA, or peptide conjugates such as DOTATATE or DOTA-bombesin, with further TLC development to assess the radiochemical purity of 47Sc-labelled compound. Usually, the radiochemical purity should be not less than 98–99% to confirm the quality of 47Sc.

This approach can also serve as the indirect test of 47Sc suitability for radiolabeling, and the experiments can be designed to check the effective specific activity of obtained 47Sc solutions. Such tests provide very useful information during the process development, especially when instrumental methods for determination of chemical purity are not available on-site.

Bacterial endotoxins

General applicable limit for the presence of bacterial endotoxins in radiopharmaceutical preparations is less than 175 IU/V, V being the maximum volume to be used for the preparation of a single patient dose, if intended for use in the manufacture of parenteral preparations without a further appropriate procedure for the removal of bacterial endotoxins. Whereas, however, it should be borne in mind that the limit of 175 IU/V applies to the final injection preparation, therefore the determination of the limit of bacterial endotoxins for the radionuclide precursor should take into account the additional effect of the labelled substance and should therefore be established on the basis of actual results obtained for several batches of the radionuclide precursor intended for radiolabeling.


The test applies if intended for use in the manufacture of parenteral preparations without a further appropriate sterilization procedure.

From the regulatory point of view, labeling efficiency of ligands is usually tested at different pH, temperature and ligand concentrations (or more precisely, at different radiometal-to-ligand molar ratios), and monitored as a function of time to optimize the radiolabeling.

The preparation of 43/44/47Sc-radiolabeled peptides for patient administration is designed to be performed on a modular system in which the final radiotracer is typically purified on a cartridge and diluted in a physiological vehicle before injection. Using the same commercially-available modular entity as is commonly used for the preparation of 68Ga-radiopharmaceuticals has the advantage to allow the preparation of 43/44/47Sc-radiopharmaceuticals for clinics without further evaluation and GMP validation of the system.

All these elements combined with a larger availability of the isotope production sites and broader distribution network are expected to allow more clinical trials with scandium radionuclides labelled radiopharmaceuticals (Domnanich et al. 2017c). These radionuclides have attracted considerable interest in the last decade and their broader availability opens new avenues for investigations.


The 44Sc/47Sc or 43Sc/47Sc pairs are appealing true theranostic radionuclides for Nuclear Medicine. Reliable methodologies for the production of all discussed herein medical scandium radionuclides exist and are increasingly reliable. With the increasing availability of scandium PET isotopes, some preclinical studies have been conducted, but remain limited and mostly performed with peptides. Since scandium exhibits suitable conjugation chemistry to be coupled with MAbs, and 44mSc half life is adapted to long biodistribution time, it paves the way for bringing new developments in this area. Scandium-based vectors from diagnosis to therapy offer a great opportunity for dosimetric calculations and the development of personalized medicine.

Among scandium radionuclides, the 44Sc seems to be in the most advanced state. Several production and purification methods were developed and some hold the promise of their relatively easy adaptation to the locally available infrastructure. At present, the primary importance of 44Sc lies in its potential broad-scale availability based on the production in medical cyclotrons and the possibility to use it as the diagnostic match with 177Lu or 47Sc. However, efficient production of 47Sc is still not developed though the half life of 47Sc allows its production in a centralized facility, which might provide this radionuclide to several countries. Its widespread use will be questionable if one considers the current availability of 177Lu and the problems associated with the use of a therapeutic agent without centralised marketing authorisation. The shorter half-life of 44Sc can be beneficial considering the lower radiation dose for the patients. The main issue is the co production of 46Sc when Ti target is used, but that is not a problem when produced through Ca target.

The 44Sc is an alternative to 68Ga, a well-established radiometal, with a similar field of application. The availability of 68Ga from generator gave a boost of its widespread use, but it is clear now, that after the initial success in the introduction of new diagnostics (somatostatin analogues and PSMA), production capacity and cost efficiency will be the main factors determining the future of these examinations, and the interest of centralized production. Additionally, the short half-life of 68Ga may limit its use for some pharmakokinetics studies. These factors favor fluorine labeling over radiometals. 44Sc can find its place in the palette of medical cyclotrons, which could be capable of producing scandium before the daily 18F irradiation. Using the [18F]FDG transportation channels for shipping 44Sc solution for radiolabeling or the GMP produced 44Sc tracers to the nuclear medicine departments facilitates the use of new specialized tracers, currently relying on the 68Ge/68Ga generator supply.

The authors trust that the medical applications of scandium radionuclides will be growing, however standardization of their quality is still needed. The question remains whether there is enough experience already to draft the pharmacopoeia monograph for any of scandium radionuclides as radipharmaceutical precursors and which of the scandium radionuclides becomes economically feasible.

Availability of data and materials

Data sharing is not applicable to this article as no datasets were generated or analysed during the current study.


Download references


Not applicable.


This work was partly supported by the IAEA/CRP code F22053 and the Polish Ministry of Science and Higher Education, Grant No. 3639/FAO/IAEA/16/2017/0 and the CERAD project, financed under Smart Growth Operational Programme 2014–2020, Priority IV, Measure 4.2. POIR.04.02.00–14-A001/16. This work was also supported by grants from the French National Agency for Research called “Investissements d’Avenir”, Equipex Arronax-Plus (ANR-11-EQPX-0004), Labex IRON (ANR-11-LABX-18-01) and ISITE NExT (ANR-16-IDEX-0007).

Author information

Authors and Affiliations



Conceptualization, RM, SHM; methodology, RM, SHM; validation, SHM, CA, FH, PG; writing—original draft preparation, RM, SHM, CA, FH, DS, VF, PG, ....; writing—review and editing, RM, SHM, FH, PG.; funding acquisition, RM., SHM, FH. PG. All authors read and approved the final manuscript.

Corresponding author

Correspondence to S. Huclier-Markai.

Ethics declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare that they have no competing interests.

Additional information

Publisher’s Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Mikolajczak, R., Huclier-Markai, S., Alliot, C. et al. Production of scandium radionuclides for theranostic applications: towards standardization of quality requirements. EJNMMI radiopharm. chem. 6, 19 (2021).

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI:


  • Scandium radionuclides
  • Accelerator- and nuclear reactor production
  • Coordination
  • Radiolabeling
  • In vivo studies
  • Quality specifications