Comparative analysis of positron emitters for theranostic applications based on small bioconjugates highlighting ⁴³Sc, ⁶¹Cu and ⁴⁵Ti

Abstract

Background

Targeted radionuclide therapy with ¹⁷⁷Lu-labelled small conjugates is expanding rapidly, and its success is linked to appropriate patient selection. Companion diagnostic conjugates are usually labelled with ⁶⁸Ga, offering good imaging up to ≈2 h post-injection. However, the optimal tumor-to-background ratio is often reached later. This study examined promising positron-emitting radiometals with half-lives between 3 h and 24 h and β⁺ intensity (I_β+) ≥ 15% and compared them to ⁶⁸Ga. The radiometals included: ⁴³Sc, ⁴⁴Sc, ⁴⁵Ti, ⁵⁵Co, ⁶¹Cu, ⁶⁴Cu, ⁶⁶Ga, ^85mY, ⁸⁶Y, ⁹⁰Nb, ¹³²La, ¹⁵⁰Tb and ¹⁵²Tb. ¹³³La (7.2% I_β+) was also examined because it was recently discussed, in combination with ¹³²La, as a possible diagnostic match for ²²⁵Ac.

Methods

Total electron and photon doses per decay and per positron; possibly interfering γ-ray emissions; typical activities to be injected for same-day imaging; positron range; and available production routes were examined.

Results

For each annihilation process useful for PET imaging, the total energy released (MeV) is: ⁴⁵Ti (1.5), ⁴³Sc (1.6), ⁶¹Cu and ⁶⁴Cu (1.8), ⁶⁸Ga (1.9), ⁴⁴Sc and ¹³³La (2.9), ⁵⁵Co (3.2), ^85mY (3.3), ¹³²La (4.8), ¹⁵²Tb (6.5), ¹⁵⁰Tb (7.1), ⁹⁰Nb (8.6), and ⁸⁶Y (13.6). Significant amounts (≥ 10%) of ≈0.5 MeV photons that may fall into the acceptance window of PET scanners are emitted by ⁵⁵Co, ⁶⁶Ga, ^85mY, ⁸⁶Y, ¹³²La, and ¹⁵²Tb. Compton background from more energetic photons would be expected for ⁴⁴Sc, ⁵⁵Co, ⁶⁶Ga, ⁸⁶Y, ⁹⁰Nb, ¹³²La,¹⁵⁰Tb, and ¹⁵²Tb. The mean positron ranges (mm) of ⁶⁴Cu (0.6), ^85mY (1.0), ⁴⁵Ti (1.5), ¹³³La (1.6), ⁴³Sc and ⁶¹Cu (1.7), ⁵⁵Co (2.1), ⁴⁴Sc and ⁸⁶Y (2.5), and ⁹⁰Nb (2.6) were lower than that of ⁶⁸Ga (3.6). DOTA chelation is applicable for most of the radiometals, though not ideal for ⁶¹Cu/⁶⁴Cu. Recent data showed that chelation of ⁴⁵Ti with DOTA is feasible. ⁹⁰Nb requires different complexing agents (e.g., DFO). Finally, they could be economically produced by proton-induced reactions at medical cyclotrons.

Conclusion

In particular, ⁴³Sc, ⁴⁵Ti, and ⁶¹Cu have overall excellent β⁺ decay-characteristics for theranostic applications complementing ¹⁷⁷Lu-labelled small conjugates, and they could be sustainably produced. Like Lu, ⁴³Sc, ⁴⁵Ti and to a lesser extent ⁶¹Cu could be labelled with DOTA.

Supplementary Information

The online version contains supplementary material available at 10.1186/s40658-024-00699-z.

Introduction

The field of targeted radionuclide therapy (TRT) is undergoing significant expansion. Many successful TRT applications are based on small conjugates such as radiopeptides and other small radioligands. Impressive results were recently obtained with the somatostatin analog [¹⁷⁷Lu]Lu-DOTATATE in patients with midgut neuroendocrine tumors [1] and with ¹⁷⁷Lu-labeled PSMA ligands in prostate cancer [2, 3]. The application of TRT is rapidly expanding thanks to the synthesis of novel conjugates aimed at various targets on tumor cells or in the tumor microenvironment [4, 5]. The radionuclide is often a radiometal linked to the targeting molecule by a chelating agent (e.g. DOTA). β^--emitters such as ¹⁷⁷Lu are the most widely used, while α-emitters (e.g., ²²⁵Ac) are also currently being studied.

Selecting appropriate patients for TRT is achieved by using a companion molecule labeled with a positron-emitting radionuclide for imaging with positron emission tomography (PET). Imaging provides information on disease extension and receptor expression. It can also inform the pharmacokinetics of the targeting molecule, which predicts both the success of TRT and the radiation dose to critical organs. The choice of the positron-emitting radionuclide is thus important because its chemical properties affect both labeling and biological behavior, and its physical properties determine which information PET imaging can provide or which radionuclide enables novel perspectives such as positronium imaging based on triple coincidences².

⁶⁸Ga, which jump-started the development of theranostic applications, has advantages as well as limitations. Its 68-minute half-life is only adequate for imaging at 1–2 h post-injection, even though more extended imaging may be useful for increasing detection sensitivity and tumor-to-background ratio, even for small conjugates [6]. Later imaging is also needed when pre-therapy dosimetry to tumors and critical organs (e.g. kidneys, liver, bone marrow) would be useful. The half-life of a ⁶⁸Ga-labeled companion diagnostic molecule is too short to correctly model the retention and clearance kinetics of the therapy conjugate [7]. In addition, not all nuclear medicine departments are equipped with ⁶⁸Ge/⁶⁸Ga generators. Storage and long-distance transit coverage of ⁶⁸Ga-labeled tracers are difficult because of its relatively short half-life. Finally, due to differences in chemistry ⁶⁸Ga-labelling does not always offer a good match with the ¹⁷⁷Lu-counterpart.

Because alternatives to ⁶⁸Ga could help the field, this study investigated PET radiometals that might be suited for imaging small conjugates for theranostic applications.

Materials and methods

A systematic selection of positron emitters was made based on the nuclear data available in NuDat (https://www.nndc.bnl.gov/nudat); all nuclides with positron intensity per decay (I_β+) of at least 15% and half-lives between 3 h and 24 h were selected. From this list, the following were removed: the non-metals ⁷³Se, ⁷⁶Br, and ¹²⁰I; the alkali metals ⁸¹Rb and ^82mRb because in-vivo-stable chelation would be challenging; and ⁹⁰Mo because it decays to another positron emitter, ⁹⁰Nb, with the risk of loss of labelling or change in biodistribution. Although it only exhibits ~ 7% I_β+, ¹³³La was also added because it has been discussed, together with ¹³²La, as a possible diagnostic match for ²²⁵Ac. All the investigated radiometals were compared to ⁶⁸Ga.

The following variables were examined for all nuclides: total electron and photon emissions; possibly interfering γ-ray emissions; positron range; energy released per decay (taken from https://www.nndc.bnl.gov/mird) and per positron (normalization to I_β+); typical activities to be injected for same day imaging; and corresponding energy. Production options for the radionuclides that best fit these criteria were then carefully analyzed. Table 1 lists the most promising nuclides (⁴³Sc, ⁴⁵Ti, ⁶¹Cu, ⁶⁴Cu). The other nuclides (⁴⁴Sc, ⁵⁵Co, ⁶⁶Ga, ^85mY, ⁸⁶Y, ⁹⁰Nb, ¹³²La, ¹³³La, ¹⁵⁰Tb, and ¹⁵²Tb) are listed in Table S1. Of note, some of the production routes of ⁴³Sc lead to concomitant production of ⁴⁴Sc (⁴³Sc + ⁴⁴Sc; 2:1 mixture) which was also examined.

Table 1.

Physical properties of the selected positron-emitting radionuclides

	68 Ga	43 Sc ¥	43 Sc + 44 Sc (2:1 mixture) £	45 Ti	61 Cu	64 Cu
T 1/2  (h)	1.13	3.89	3.9	3.08	3.34	12.70
β +  branching ratio (% emission)	88.9	88.0	90	84.8	61	17.5
E β+ mean (keV)	830	476	580	439	500	278
E β+ max (keV)	1899	1199	1474	1028	1216	653
Mean positron range (Rmean, mm)Þ	3.6	1.7	2.0	1.5	1.7	0.56
Dominant γ rays E (keV) [I γ  (%)]	1077 [3.22]	373 [22.5]	373 [15] 1157 [33]	-	283 [12.7] 656 [9.7]	-
511 keV ± 10% (%)	0.0	0.0	0.0	0.0	0.4	0.0
511 keV ± 20% (%)	0.0	0.0	0.0	0.0	1.5	0.0
1022 keV ± 10% (%)	3.2	0.0	0.0	0.0	0.5	0.0
1022 keV ± 20% (%)	3.2	0.0	33	0.0	5.6	0.0
> 613 keV (Compton background, %)	3.6	0.0	34	0.3	16.7	0.5
> 1500 keV (pair creation background, %)	0.1	0.0	0.1	0.0	0.2	0.0
H10 (at 1 m, µSv/h/GBq)	158	165	220	146	136	31
Dose rate after 5 cm Pb (at 1 m, µSv/h/GBq)	0.36	0.05	3.3	0.07	0.65	0.09

All nuclides decay by β⁺ and electron capture. ⁶⁴Cu emits additional β⁻ radiation (39.0%). ^¥Pure ⁴³Sc when produced using α-beams. ^£43Sc + ⁴⁴Sc (2:1 mixture is obtained when ⁴³Sc is produced using proton beams. ^ÞCalculated with a Monte Carlo code for ⁶⁸Ga, ⁴³Sc, ⁴⁵Ti and ⁶⁴Cu and estimated using the equation of Cole and colleagues for other radionuclides [9]

Energy release and gamma ray emissions

The intensities of gamma ray emissions were extracted from the ENSDF database and summed over different intervals of interest: 460–562 keV (i.e., falling into a ±10% wide energy window around the 511 keV annihilation rays), and 409–613 keV (i.e., falling into a ±20% wide window around 511 keV). Similarly, γ-rays in the energy windows 920–1124 keV and 818–1226 keV (i.e., within ± 10% or ± 20% of 2 × 511 keV) could give rise to Compton background when backscattering in a detector, leaving roughly half of their energy there. Any γ-ray above the 511 keV window could still fall into this window after undergoing Compton scattering. To address this issue, all γ-rays above 613 keV were integrated. Finally, pair creation for very energetic γ-rays may generate virtual secondary positron sources far from the primary source; because the probability of pair creation rises strongly with energy, all γ-rays above 1500 keV were considered.

Positron range

A higher tissue penetration of positrons, from emission until annihilation with electrons, degrades the resolution of the PET image. Positron range values in water (tissue equivalent) were obtained for ⁶⁸Ga, ⁴³Sc, ⁴⁴Sc, ⁴⁵Ti, ⁶¹Cu, ⁶⁴Cu, ⁸⁶Y, and ¹⁵²Tb with a Monte Carlo code described in [8]. Our code takes into account the whole positron energy spectrum, has very low transport cut-off energy (7.4 eV), and included modeling of positronium formation and diffusion. Positron ranges of other radionuclides were estimated using the empirical formula described in [9].

Imaging/measurement possibilities at specific time points

In order to assess the possibility of same-day imaging, a body content of activity was assumed at the time of imaging, providing 50 million annihilations/s needed for high-sensitivity imaging. The activity in GBq necessary at injection to achieve this rate of events at a given time point was calculated, considering the positron branching ratio (I_β+) of the radionuclide and the biological half-life of PSMA-617 [10]. The corresponding energy released in J (dosimetry estimate) was also calculated.

External dose rates

The gamma dose rates associated with different positron emitters are generally not considered a primary selection criterion. Nevertheless, for large-scale deployment of novel radionuclides, gamma dose rate could impact the collective dose of personnel involved in the chain of production to administration, and ALARA aspects should thus be considered. Gamma dose rates (in µSv/h) were calculated at 1 m distance from a point source, assuming two locally-created annihilation photons per emitted positron. The calculations were performed with the Nucleonica online tool (https://www.nucleonica.com) based on nuclear data from the ENDF8.0 library. These results were compared with Nucleonica calculations based on nuclear data from the JEFF3.3 library, as well as with H₁₀ values [11], which in turn were calculated with FLUKA based on nuclear data from ICRP107. No suitable nuclear data were available from JEFF3.3 for ^85mY, nor from ENDF8.0 and JEFF3.3 for ¹⁵⁰Tb. γ-dose rates are given for 1 m distance, once without photon shielding and once behind a 5 cm lead shield (broad beam geometry including buildup, representative for lead-shielded transport containers or hot cells).

Production routes for selected radionuclides

Production yields were calculated with the IAEA Medical Isotope Browser (https://www-nds.iaea.org/relnsd/isotopia/isotopia.html). Typical energy ranges (incident projectile energy and energy when leaving the target) were chosen to cover roughly the full-width-half-maximum of the excitation curve for the radionuclide of interest. When radionuclide impurities produced in other reaction channels needed to be minimized, the energy range was constrained accordingly. A typical production run was assumed to last for one half-life of the nuclide of interest (i.e., reaching 50% of saturation activity) at an incident beam current of 50µA (or 50 electrical µA for α beams, respectively). A delay of 4 h between the end of irradiation and injection was assumed, allowing for chemical separation, labelling, and shipment. A combined separation and labelling yield of 80% was assumed.

Caveats related to our methodology are described in the Supplement.

Results

As described in greater detail below, radionuclides were selected based on γ-ray and electron emissions, late imaging possibilities, dose rates, and positron range. Data for the four selected PET radionuclides (⁴³Sc (pure or as mixture), ⁴⁵Ti, ⁶¹Cu, ⁶⁴Cu) are shown in Table 1, and data for the additional 10 PET radionuclides (⁴⁴Sc, ⁵⁵Co, ⁶⁶Ga, ^85mY, ⁸⁶Y, ⁹⁰Nb, ¹³²La, ¹³³La, ¹⁵⁰Tb, ¹⁵²Tb) are available in Tables S1 and S2.

Energy released per decay or per emitted Positron and Gamma Ray Emissions

The total amount of energy released from electron or photon radiation was calculated, normalized per annihilation process. The values for the four selected radionuclides were: ⁴³Sc (~ 1.6 MeV), ⁴⁵Ti (~ 1.5 MeV), ⁶¹Cu (~ 1.8 MeV), and ⁶⁴Cu (~ 1.8 MeV); in comparison, the value for ⁶⁸Ga is ~ 1.9 MeV (Table 2). Separate values are provided for electron radiation (ß⁺, ß^-, conversion electrons, Auger electrons), considered to be fully absorbed in the patient, and for photons (partially absorbed) (Table 2). Most of the radionuclides that were not selected had higher values (⁸⁶Y had the highest value (~ 13.6 MeV) (Table S2)). Importantly, ⁴³Sc, ⁴⁵Ti, and ⁶⁴Cu had no γ-emissions that fell into the 511 keV window of PET machines. In addition, ⁴³Sc had no, and ⁴⁵Ti and ⁶⁴Cu had almost no (≤ 0.5%), higher energy γ-emissions capable of interfering with imaging after Compton scattering. ⁶¹Cu had a slightly increased potential Compton background compared to ⁶⁸Ga. A mixture of ⁴³Sc and ⁴⁴Sc resulting from certain production routes would have similar drawbacks as ⁴⁴Sc—that is, the 1157 keV γ-ray source of Compton background but at much lower intensity than pure ⁴⁴Sc. The other radionuclides are discussed in the Supplement.

Table 2.

Energy released from electron radiation (ß⁺, ß⁻, conversion electrons, Auger electrons) and photons including annihilation photons

Radionuclides (half life, positron branching)	Electron energy (MeV) (including β+, β−, CE, Auger)		Photon energy (MeV) (γ + X rays + annihilation photons)		Total electron + photon energy (MeV) (including annihilation photons)
	per disintegration	normalized per annihilation process ¤	per disintegration	normalized per annihilation process ¤	per disintegration	normalized per annihilation process ¤
68Ga (1.13 h, 88.9%)	0.7	0.8	0.9	1.1	1.6	1.9
43Sc (3.89 h, 88.0%)	0.4	0.5	1.0	1.1	1.4	1.6
43Sc +44Sc 2:1 mixture (3.4 h, 90%)	0.5	0.5	1.4	1.5	1.8	2.0
45Ti (3.08 h, 84.8%)	0.4	0.4	0.9	1.0	1.2	1.4
61Cu (3.34 h, 61%)	0.3	0.5	0.8	1.3	1.1	1.8
64Cu (12.70 h, 17.5%)	0.1	0.7	0.2	1.1	0.3	1.8

Positron range

As seen in Table 1, ⁴³Sc, ⁴⁵Ti, ⁶¹Cu, and ⁶⁴Cu outperformed ⁶⁸Ga in terms of smaller R_mean (mm): ⁶⁴Cu (~ 0.6), ⁴⁵Ti (~ 1.5), ⁴³Sc and ⁶¹Cu (~ 1.7), ⁴³Sc + ⁴⁴Sc as 2:1 mixture (~ 2.0), versus ⁶⁸Ga (~ 3.6). See Table S1 for the R_mean values of the other radionuclides.

Possibility of same-day imaging and energy released

For each of the selected radionuclides, the activity in GBq that needed to be injected for imaging at specific time points (considering that a rate of 50 million annihilations/s is required for high-sensitivity imaging) is shown in Table 3; Table S3 displays activity for the other radionuclides. The corresponding total energy released is also depicted in Table 3. ⁴³Sc and the ⁴³Sc + ⁴⁴Sc mixture enabled high resolution imaging using low activity (Table 3). Nevertheless, the higher activity needed for ⁴⁵Ti and ⁶¹Cu may not be prohibitive because the total energy released was comparable or even slightly lower than for ⁴³Sc and the ⁴³Sc + ⁴⁴Sc mixture (Table 3). In all cases, ⁶⁴Cu needed higher activity and released higher amounts of energy.

Table 3.

Required activity for same day imaging based on the biological half-life of PSMA-617 and total energy released

		Radionuclide, half-life; β+ branching ratio	68Ga, 1.13 h; 88.9%	43Sc, 3.89 h; 88.0%	43Sc +44Sc 2:1 mixture, 3.9 h, 90%	45Ti, 3.08 h; 84.8%	61Cu, 3.34 h; 61%	64Cu, 12.70 h; 17.5%
Activity needed (GBq) to obtain 50 million annihilation events/s for high sensitivity imaging		1 h	0.14	0.09	0.09	0.10	0.13	0.40
		2 h	0.33	0.14	0.13	0.16	0.21	0.55
		4 h	1.6	0.29	0.28	0.36	0.47	0.89
		6 h	7.1	0.53	0.51	0.72	0.91	1.3
		8 h	28	0.87	0.83	1.3	1.6	1.6
Energy released (J, total energy/electron energy) for same day imaging		1 h	0.2/0.1	0.4/0.1	0.5/0.1	0.3/0.1	0.4/0.1	1.3/0.5
		2 h	0.5/0.2	0.6/0.2	0.8/0.2	0.5/0.1	0.7/0.2	1.8/0.7
		4 h	2.6/1.1	1.3/0.4	1.7/0.4	1.1/0.3	1.5/0.4	2.9/1.2
		6 h	11/5	2.4/0.7	3.1/0.8	2.3/0.7	2.8/0.8	4.2/1.7
		8 h	44/19	3.9/1.2	5.0/1.3	4.1/1.2	4.9/1.4	5.4/2.1

External dose rates

In line with their γ-emission properties, pure ⁴³Sc, ⁴⁵Ti, and ⁶⁴Cu exhibited 4-7-fold lower dose rates than ⁶⁸Ga behind 5 cm lead shielding, while ⁶¹Cu had a slightly higher dose rate than ⁶⁸Ga (Table 1). The ⁴³Sc + ⁴⁴Sc 2:1 mixture had a 9-fold higher dose rate than ⁶⁸Ga. With the exception of ¹³³La, the other radionuclides exhibited much higher (22-fold to 150-fold) dose rates than ⁶⁸Ga behind 5 cm shielding (Table S1). Results from all three calculations (Nucleonica with ENDF8.0, Nucleonica with JEFF3.3, and FLUKA with ICRP107) were generally within ±2%, with the exception of some heavier radionuclides with discrepant γ-ray intensities in the libraries (¹³²La, ¹³³La, ¹⁵²Tb), which showed differences up to 10% unshielded and up to one-third with 5 cm lead shielding. It should be noted that, because recent detailed measurements of ⁸⁶Y decay [12] are not yet included in any of the databases used, the present calculations likely underestimate the γ-dose rate for ⁸⁶Y.

Production routes

Because ⁴³Sc, ⁴⁵Ti, and ⁶¹Cu performed better than ⁶⁸Ga in terms of γ-ray emissions, possibility of same-day imaging, dose rates, and positron range, their production routes were carefully analyzed and compared. ⁶¹Cu production was also compared to the well-known production routes of ⁶⁴Cu (Table 4). Additional information is provided below for each radionuclide.

Table 4.

Production routes, yields, and natural abundance of precursor for the production of ⁴³Sc, ⁴⁵Ti, and ⁶¹Cu

	43Sc				45Ti	61Cu
Reaction	43Ca(p, n)43Sc	44Ca(p,2n)43Sc	46Ti(p,α)43Sc	40Ca(α,p)43Sc	45Sc(p, n)45Ti	61Ni(p, n)61Cu	62Ni(p,2n)61Cu	64Zn(p,α)61Cu	60Ni(d, n)61Cu
Ein > Eout (MeV)	13 > 8	30 > 18	18 > 12	20 > 10	13 > 8	13 > 8	30 > 18	18 > 10	13 > 8
Production yield (GBq/µAh)	0.68	0.17	0.14	0.20	1.0	1.0	2.2	0.27	0.33
Natural abundance of precursor (%)	0.14	2.1	8.3	96.9	100	1.1	3.6	49.2	26.2

⁴³Sc: ⁴³Sc can be produced by proton-irradiation of enriched calcium targets in ⁴³Ca(p,n) and ⁴⁴Ca(p,2n) reactions, respectively. The radiochemical Ca/Sc separation has previously been described [13]. Commercially available grades of enriched ⁴³Ca contain between 58% and 87% ⁴³Ca with significant ⁴⁴Ca admixture, inevitably leading to co-production of some ⁴⁴Sc via the ⁴⁴Ca(p,n)⁴⁴Sc reaction. Target material containing 58% ⁴³Ca and 12% ⁴⁴Ca leads to a ⁴³Sc/⁴⁴Sc activity ratio of about 2:1 [14]. The ⁴⁴Ca(p,2n)⁴³Sc reaction uses ⁴⁴Ca targets that are available with high enrichment (> 95%). However, any (p,2n) reaction is inevitably accompanied by the (p,n) channel. In the present case, the ⁴⁴Ca(p,n)⁴⁴Sc reaction leads to co-production of ⁴⁴Sc. Irradiation of 100% enriched ⁴⁴Ca over the energy range 30 to 18 MeV would lead to co-production of ⁴³Sc and ⁴⁴Sc at an activity ratio of about 4:1 and, at lower energies, the fraction of ⁴⁴Sc increases. Considering the batch yield and the lower cost of enriched ⁴⁴Ca (due to higher natural abundance), the ⁴⁴Ca(p,2n)⁴³Sc reaction appears preferential for centralized production of large quantities of ⁴³Sc compared to the ⁴³Ca(p,n)⁴³Sc reaction.

The ⁴⁶Ti(p,α)⁴³Sc reaction exploiting enriched titanium targets has a much lower yield than the calcium routes but provides purer ⁴³Sc, with less contribution of ⁴⁴Sc. Target material containing 97% ⁴⁶Ti and 0.44% ⁴⁷Ti leads to a ⁴³Sc/⁴⁴Sc activity ratio of about 65:1 (i.e., 98.2% pure ⁴³Sc) [14]. Because the maximum current density compatible with metallic titanium targets is higher than calcium oxide (or calcium carbonate) targets [13, 15], part of the lower yield could be made up with a higher current. Enriched ⁴⁶Ti targets are considerably cheaper than ⁴³Ca targets due to the significantly higher natural abundance than ⁴³Ca and because volatile titanium compounds are available for centrifuge enrichment; in contrast, calcium is enriched electromagnetically, which is more costly. Reducing the ⁴³Ca target mass to match the cost of an enriched ⁴⁶Ti target would invert the yield advantage, making the titanium route favorable. Thus, except for the challenge of efficiently recycling and reducing enriched Ti targets, the ⁴⁶Ti(p,α) route with 18 MeV or 19 MeV cyclotrons appears overall preferable to the ⁴³Ca(p,n) route—though not necessarily preferable to the ⁴⁴Ca(p,2n) route.

Two alternatives should be mentioned for cyclotrons that provide other projectiles than protons. First, the ⁴²Ca(d,n)⁴³Sc route using 9 MeV deuterons can provide local supply [16]. Second, the ⁴⁰Ca(α,p)⁴³Sc route provides extremely pure ⁴³Sc when (cheap) enriched ⁴⁰Ca targets are used and retains over 99.9% radionuclidic purity when Ca targets of natural isotopic composition are used [17]. This would indeed be the preferred method for producing “pure” ⁴³Sc if more accelerators providing suitable (> 20 MeV) α-beams were available.

⁶¹Cu: ⁶¹Cu can be produced by proton-irradiation of enriched nickel targets in ⁶¹Ni(p,n)⁶¹Cu and ⁶²Ni(p,2n)⁶¹Cu reactions, respectively. All aspects of targetry, radiochemical Cu separation and purification, and Ni target recycling are identical to the well-known ⁶⁴Ni(p,n)⁶⁴Cu reaction, and commercial solutions with automated synthesis modules are available for this purpose [18].

The yield of the ⁶¹Ni(p,n)⁶¹Cu reaction is over two times as large as that of the ⁶⁴Ni(p,n)⁶⁴Cu reaction over the relevant energy range (13 MeV to 8 MeV). Combined with the 3.5-fold higher positron intensity of the ⁶¹Cu decay, this shorter-lived copper radioisotope is clearly advantageous in terms of production capacity (patient doses per production batch). However, for decay times exceeding 13 h (between end-of-target irradiation and imaging), the longer-lived ⁶⁴Cu is favorable due to lower decay losses. The costs of enriched ⁶¹Ni and ⁶⁴Ni targets are comparable because the natural abundance of both isotopes is comparable. Local or regional production of ⁶¹Cu can thus be cost-effective compared to centralized production of ⁶⁴Cu when used for same-day imaging.

The ⁶²Ni(p,2n)⁶¹Cu reaction provides another factor 2 gain in production yield with respect to ⁶¹Ni(p,n)⁶¹Cu, but it requires larger cyclotrons providing incident proton beams of 24 to 30 MeV. With rising demand for ⁶¹Cu, this pathway could provide a cost-efficient regional supply for same-day imaging.

The ⁶⁰Ni(d,n)⁶¹Cu route profits from the high natural abundance of the target isotope but requires deuteron beams only available with some cyclotrons. Although ≈13 MeV deuterons would provide the highest yields, 8–9 MeV deuteron beams may provide useful quantities for local or regional use.

The ⁶⁴Zn(p,α)⁶¹Cu reaction could be well served with 18 MeV cyclotrons but has about four-fold lower yield than the ⁶¹Ni(p,n)⁶¹Cu reaction. Its main advantage is the use of very affordable enriched ⁶⁴Zn targets that do not require recycling compared to the costly Ni targets. This reaction can also be used with cheap Zn targets of natural composition, leading to > 97% radionuclidic purity at the end of bombardment (co-production of about 2% ⁶⁴Cu and 0.02% ⁶⁷Cu activity), which remains acceptable for most applications.

In all these reactions, some co-production of short-lived ⁶²Cu and/or ⁶⁰Cu may occur, but a decay time of 1–2 h between end of bombardment and injection suffices to let these decay to negligible levels. In practice, the duration of processing and shipment will usually provide such a decay time.

⁴⁵Ti: ⁴⁵Ti is best produced by proton-irradiation of naturally monoisotopic scandium targets in ⁴⁵Sc(p,n)⁴⁵Ti reactions. Production and separation of GBq quantities are discussed in detail in [19]. Other proton- or α-induced reactions are less competitive in terms of yield, target, and accelerator costs.

Discussion

Although many positron-emitting radionuclides have become available in recent years, each with its own physical and chemical advantages and disadvantages, the optimal radiometals for theranostic applications remain uncertain. This comparative analysis sought to identify a preferred diagnostic radionuclide analog for ¹⁷⁷Lu for theranostic applications using small bioconjugates requiring same-day imaging. It also explored the utility of these positron-emitting radiometals for other diagnostic purposes.

Many diagnostic conjugates (e.g., PSMA ligands, somatostatin analogs, albumin-binding modified ligands, affibodies, nanobodies, etc.) achieve optimal tumor-to-background ratio at 3–4 h post-injection or later [20–22], but such imaging time requires substantially higher activities of ⁶⁸Ga (> 300 MBq). In contrast, the half-lives of ⁴³Sc, ⁴⁵Ti, and ⁶¹Cu (~ 3–4 h) are well-suited to later imaging times. The injected activities calculated here for optimal imaging were PSMA-617-specific and based on studies done with conventional PET scanners. New, more sensitive PET machines (digital PET, and more recently long-axial field-of-view PET) will need less activity for all radionuclides and/or would allow imaging at later time points [23]. Pending confirmatory studies in humans, the dose delivered to patients by ⁴³Sc, ⁴⁵Ti, and ⁶¹Cu is likely to be more advantageous than that of ⁶⁸Ga because of their lower positron energy and lower intensity of photon emissions (Table 1). Importantly, the energy absorbed by patients (electron energy) for activities required for imaging at any time point between 2 h and 8 h was lower with ⁴³Sc, ⁴⁵Ti, and ⁶¹Cu than with ⁶⁸Ga. A recent modelling study confirmed the lower mean effective dose of [⁴³Sc]Sc-DOTATATE vs. [⁶⁸Ga]Ga-DOTATATE as early as 2 h assuming injected activity reported in Table 3 [24]. [⁶¹Cu]Cu-NODAGA-PSMA I&T has also been recently applied in a patient with metastatic prostate cancer. Only 105 MBq were injected and high-contrast imaging was possible 3 h post-injection [25]. ⁶⁴Cu had higher energy released in the form of electrons than ⁶⁸Ga up to 4 h due to its competing β^- decay mode. Another specific advantage of ⁴³Sc and ⁶¹Cu was that the co-emitted photons (373 keV for ⁴³Sc and 283/656 keV for ⁶¹Cu) had moderate intensity and did not fall into the acceptable window of PET scanners (Table 1). Here, ⁴⁵Ti, as well as ⁶⁴Cu, might be even more suited as they had almost no co-emitted photons (≤ 0.5%). Lastly, as regards radiation safety, the dose rates for ⁴³Sc, ⁴⁵Ti, ⁶¹Cu, and ⁶⁴Cu behind 5 cm lead shielding were in the same order of magnitude as ⁶⁸Ga or even lower, suggesting easy implementation of these radionuclides to already equipped PET centers.

⁴³Sc, ⁴⁵Ti, and ⁶¹Cu had similar positron ranges (1.5–1.7 mm) with values less than half that of ⁶⁸Ga (3.6 mm). Smaller positron ranges and fewer photon emissions should result in better image quality and resolution with ⁴³Sc compared to ⁴⁴Sc, which has been documented on Derenzo phantoms imaged with small-animal PET [14, 26]. ⁴⁵Ti also provided excellent resolution down to 1.25 mm spheres from Derenzo phantoms [27]. ⁶¹Cu offered better spatial resolution than ⁶⁸Ga in Jaszczak phantoms [28]. Overall, ⁴³Sc (pure or mixed with ⁴⁴Sc), ⁴⁵Ti, and ⁶¹Cu are expected to provide images with higher spatial resolution than ⁶⁸Ga. Because the intrinsic resolution of small-animal PET scanners is now as low as ~ 1.5 mm, the positron range of ⁶⁸Ga has become the limiting factor for spatial resolution and activity recovery in small structures [29, 30]. The resolution of clinical whole-body PET systems is lower (~ 4–5 mm) [31], but with the advent of silicon-photomultiplier-based PET technology and other developments, the impact of positron range on clinical PET imaging is expected to be more prominent [32].

For certain radiometals the positron emission is followed within few picoseconds by the emission of prompt γ-rays. When the latter are of suitable energy and intensity this combination can be exploited for new imaging technologies such as positronium imaging [33], 3γ imaging [34], or multiplexed PET [35], which are based on triple coincidence events (2 annihilation photons + 1 prompt photon). Among the radiometals investigated in this study, in particular ⁴⁴Sc, ⁵⁵Co and ⁸⁶Y qualify for such an application. The quantification of positronium formation, a bound electron-positron pair, has not yet been utilized in routine clinical applications of PET due to the need to detect either the emission of a prompt γ-ray or the decay of higher-order coincident events. This point deserves specific considerations as positronium imaging might offer new horizons in PET imaging [36–39]. As described by Moskal et al. [36–38], positronium is copiously formed in the free molecular spaces in the patient’s body during PET imaging. Assessment of the lifetime of the formed positronium can potentially yield additional diagnostic information of the surrounding tissue. The positronium properties vary according to the size of inter- and intramolecular voids and the concentration of molecules in them such as, e.g., molecular oxygen, O₂; therefore, positronium imaging may provide information about disease progression during the initial stages of molecular alterations. Measurement of positronium lifetime is feasible using dedicated devices such as the Jagiellonian (J) -PET [36–38], but also on long-axial field-of-view PET scanners [39]. The use of the triple co-incidence imaging has been also used in simultaneous multi-tracer imaging [38, 40]. The use of radiometals emitting copious amount of prompt γ-rays is beneficial for these specific applications [41], the increase of the absorbed dose should, however, be taken into account when used outside these perspectives, on conventional PET instruments for clinical routine (Table S3).

From a chemical standpoint, ⁴³Sc-labeled and ¹⁷⁷Lu-labeled molecules displayed excellent chemical and pharmacokinetic similarities, with direct implications for theranostic applications. Indeed, DOTA, the most widely used chelator for ¹⁷⁷Lu (as well as for ⁹⁰Y, ¹⁶¹Tb, and ²²⁵Ac), is also an excellent chelator for scandium [42, 43]. Previous preclinical investigations confirmed that Sc³⁺ was more compatible with Lu³⁺-labeled DOTA-functionalized bioconjugates than Ga³⁺ as exemplified with radiolabeled PSMA-inhibitors [44]. ⁴³Sc-labelled molecules could thus be better suited to select patients who might benefit from ¹⁷⁷Lu therapy and would also be an excellent theranostic match for ⁴⁷Sc therapy when this β^--emitting radionuclide becomes more available [45]. As regards Cu²⁺, it is a labile metal ion, and the optimal chelators for copper radioisotopes differ from those for ¹⁷⁷Lu [42], although the DOTA chelator can be used for imaging between 1 and 3 h post-injection as reported with [⁶⁴Cu]Cu-DOTATATE [46]. A clinical study using [⁶⁴Cu]Cu-PSMA-617 (DOTA chelator) clearly showed lower tumor uptake and higher liver uptake on next-day images, most probably due to demetallation [47]. Nevertheless, ⁶¹Cu-labeled bioconjugates can be used for ¹⁷⁷Lu-pre-therapy imaging if the specific Cu-chelate does not too strongly impair the behavior of the radiopharmaceutical. This issue needs to be addressed in (pre)clinical studies. Indeed, ⁶¹Cu could be the ideal theranostic match for same-day PET imaging of the therapeutic β^--emitter ⁶⁷Cu [48]. ⁴⁵Ti is a hard metal with high oxophilicity, low hydrolytic stability, and high propensity in forming titanyl species in aqueous environments for which the chelation chemistry have been under-investigated. Recently, chelation of ⁴⁵Ti with DOTA became possible through the formation of an intermediate [⁴⁵Ti]Ti-guaiacolate complex thus opening the door to the development of [⁴⁵Ti]Ti-DOTA-radiopharmaceuticals [49]. ⁴⁵Ti does not have a perfect theranostic match. Overall, ⁴³Sc, ⁴⁵Ti and ⁶¹Cu can be easily implemented in existing workflow regarding radiopharmaceutical supply chain although copper-specific chelators should be used for high-sensitivity imaging at late time points (4 to 8 h post-injection).

Selecting novel radionuclides only makes sense if production routes allow large-scale economic and sustainable availability. Only α beams can lead to pure ⁴³Sc production but accelerators producing suitable > 20 MeV α beams are scarcely available. Unless a subsequent mass separation step is added after the irradiation step, none of the considered proton-induced production paths for ⁴³Sc provides “pure” ⁴³Sc, where “pure” is defined in the usual sense of > 99% or > 99.9% radionuclidic purity. However, in this specific case, the very similar decay properties of both ⁴³Sc and ⁴⁴Sc must be considered—that is, the latter is not a normal impurity but equally useful for PET imaging. Two-to-one mixtures of ⁴³Sc and ⁴⁴Sc have successfully been used, and corresponding doses are proportional to the mixture used [26].

As noted above, several promising production pathways exist for ⁶¹Cu that could be supplied at production costs potentially lower than those for ⁶⁴Cu. Production and purification methods for GBq quantities of ⁴⁵Ti were also discussed. However, it was not our objective to provide a market access study of ⁴³Sc, ⁴⁵Ti, and ⁶¹Cu, but rather a comparative study of their characteristics. When a market-access study is needed before industrial development of these radionuclides, any such study would strongly depend on disease(s), vector(s), population characteristics, etc. that could eventually bias outcomes towards a shorter- or longer-lived radionuclide.

¹³³La has recently been proposed as a novel PET radionuclide and deserves specific consideration. At first glance, ¹³³La closely resembles ⁴³Sc; in particular, their half-lives and their average positron energies and ranges are identical within reported uncertainties, and interfering γ-rays are relatively weak for both. However, the differences between these nuclides are significant. Chemically, Sc has the smallest ionic radius of all rare earth metals and thus more closely resembles Lu. On the other hand, La has the largest ionic radius of all lanthanides and thus provides a better match for Ac. Due to the very low positron intensity of ¹³³La—more than an order of magnitude lower than that of ⁴³Sc—higher activities need to be produced and injected to obtain the same annihilation event statistics (e.g., 3.5 GBq ¹³³La versus 0.3 GBq ⁴³Sc for an image after 4 h). Due to weak co-emission of other radiation, this high activity is not necessarily prohibitive in terms of patient dose—that is, the total dose per positron remains reasonable (1.5-fold the total dose from ⁶⁸Ga; Table S2). However, one notable difference concerns long-lived residual activity; while ⁴³Sc, ⁴⁵Ti, and ⁶¹Cu decay to stable nuclei, ¹³³La decays to long-lived radioactive ¹³³Ba. It should be noted that the presence of long-lived residual activity is not unique to this specific radionuclide, given that ^99mTc decays to long-lived ⁹⁹Tc, ¹³¹I is accompanied by tiny amounts of long-lived ¹²⁹I, and all generator-derived radionuclides may be accompanied by a tiny but finite breakthrough of the mother isotope (⁶⁸Ge, ⁸²Sr, etc.). A detailed calculation of the contamination by ¹³³Ba of ¹³³La is provided in the Supplement.

Conclusion

This exploratory study suggests that ⁴³Sc, ⁴⁵Ti, and ⁶¹Cu would be excellent radionuclides for theranostic applications or even routine clinical imaging. All three have the appealing possibility of sustainable production and distribution of radiopharmaceuticals within about one lifetime distance. Their half-lives, close to 4 h, also enable imaging at optimal time points for small ligands, such as peptides and small proteins. For imaging performed between 2 h and 8 h post-injection, patients’ dosimetry when using ⁴³Sc, ⁶¹Cu, and ⁴⁵Ti would be more favorable than ⁶⁸Ga. Moreover, ⁴³Sc and ⁶¹Cu can be perfectly paired with the β^--emitters ⁴⁷Sc and ⁶⁷Cu. The low positron range and the absence of co-emitted photons that fall into the range detected by PET scanners suggest better image quality. Finally, production capability highlights that ⁴³Sc, ⁴⁵Ti, and ⁶¹Cu could make their way into routine clinical use in the near future. The chemistry of Sc-labeled bioconjugates is highly compatible with that of therapeutic ¹⁷⁷Lu-labeled counterparts. Complexation of ⁶¹Cu and ⁴⁵Ti is feasible with DOTA. The use of ¹³³La might make sense as a specific ²²⁵Ac surrogate or with its perfect therapeutic match ¹³⁵La (Auger electron emitter), but not as ¹⁷⁷Lu surrogate nor for general same day PET imaging of small ligands. Finally, perspectives regarding positronium imaging with radionuclides emitting a prompt γ-ray are a novel and exciting domain.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Author contributions

EH performed the modelling study of the positron ranges and was a major contributor in writing of the article. UK selected the positron-emitters, analyzed the production routes and external dose rates of the radionuclides and a major contributor in writing of the article. CC performed the modelling study of the positron ranges. PZF was a major contributor in writing the article and critically reviewed the manuscript. CM selected positron emitters, performed calculations of energy released and possibility of same day imaging and was a major contributor in writing the article. All authors read and approved the final manuscript.

Funding

This study was funded in part by the European Union’s Horizon 2020 research and innovation program under grant agreement No 101008571 (PRISMAP). This work was conducted within the frame of the NEWMOOM Impulsion network of Bordeaux University. The comparison of the sustainability of different production paths has been performed by U.K. in the frame of the PRISMAP project.

Data availability

The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

Declarations

Competing interests

The authors have no conflict of interest to disclose, financial or otherwise.