Automated, cassette-based isolation and formulation of high-purity [61Cu]CuCl2 from solid Ni targets

Background A need for improved, cassette-based automation of 61Cu separation from irradiated Ni targets was identified given the growing interest in theranostics, and generally lengthy separation chemistries for 64Cu/64Ni, upon which 61Cu chemistry is often based. Methods A method for separating 61Cu from irradiated natNi targets was therefore developed, with provision for target recycling. Following deuteron irradiation, electroplated natNi targets were remotely transferred from the cyclotron and dissolved in acid. The dissolved target solution was then transferred to an automated FASTlab chemistry module, where sequential TBP and TK201 (Triskem) resins isolated the [61Cu]CuCl2, removed Ni, Co, and Fe, and concentrated the product into a formulation suitable for anticipated radiolabelling reactions. Results 61Cu saturation yields of 190 ± 33 MBq/μA from energetically thick natNi targets were measured. The average, decay-corrected, activity-based dissolution efficiency was 97.5 ± 1.4% with an average radiochemical yield of 90.4 ± 3.2% (N = 5). The isolated activity was collected approximately 65 min post end of bombardment in ~ 2 mL of 0.06 M HCl (HCl concentration was verified by titration). Quality control of the isolated [61Cu]CuCl2 (N = 5) measured 58Co content of (8.3 ± 0.6) × 10− 5% vs. 61Cu by activity, Ni separation factors ≥ (2.2 ± 1.8) × 106, EoB molar activities 85 ± 23 GBq/μmol and NOTA-based EoB apparent molar activities of 31 ± 8 MBq/nmol and 201 MBq/nmol for the 30 min and 3.3 h (N = 1) irradiations, respectively. Conclusion High purity 61Cu was produced with the developed automated method using a single-use, cassette-based approach. It was also applicable for 64Cu, as demonstrated with a single proof-of-concept 64Ni target production run.


Introduction
Within the field of positron emission tomography (PET), radiometals research has increased during the past decade (e.g. publications per year with the keyword 68 Ga have increased from 69 to 506 between 2009 and 2019, Scopus). These radiometals complement the traditional PET nuclide 18 F thanks to differences in their chemical and radioactive decay properties. This opens a path toward more personalised medicine as an array of biomolecules and biodistribution mechanisms can be used to adapt a treatment to a specific disease case. Radiometals are widely used for radiopharmaceuticals, in part due to available chelation chemistry and labelling with biomolecules (Krasikova et al. 2016;Price and Orvig 2014;Aluicio-Sarduy et al. 2018;Vivier et al. 2018;Tsai and Wu 2018;Liu 2008). Among the radiometals, following Ga, Cu is one of the most extensively investigated for PET radiopharmaceutical purposes (Mikolajczak et al. 2019;Brandt et al. 2018;Wadas et al. 2006;McCarthy et al. 1999). One reason for this is the well-understood coordination chemistry and biodistribution of Cu (Wadas et al. 2006;Wadas et al. 2010;Wallhaus et al. 1998;Jalilian et al. 2009;Woo et al. 2019), which has resulted in a multitude of chelators and biomolecule options being available for Cu isotopes.
While 64 Cu has been suggested for theranostic applications, including pairing with 177 Lu (Song et al. 2016), several Cu radioisotopes are suitable for both imaging and therapy. This creates an opportunity for Cu to be used as a "true" (i.e. identical element) theranostic pair: 61 Cu (t ½ = 3.34 h, 61% β + , E Max = 1216 keV) is suitable for PET imaging; only 5.9 and 2.1% of its two major gammas (282.956 KeV, I γ = 12.2%; 656.008 keV, I γ = 10.8%, respectively), are coincident with β + decay (IAEA 2020a). 64 Cu (t ½ = 12.7 h) is more commonly used in PET imaging (18% β + , E Max = 653 keV) (Singh et al. 2017;Follacchio et al. 2017) with negligible gamma emissions (1345.77 keV, I γ = 0.475%), however, applications in β − and Auger emission therapy have been reported (39% β − , E Max = 580 keV) (Obata et al. 2005;Lewis et al. 1999Lewis et al. , 2001Gutfilen et al. 2018), and 67 Cu (t ½ = 61.83 h, 100% β − , E max = 562 keV) is a therapeutic radionuclide suitable for SPECT imaging (E γ = 184.577 keV, I γ = 48.7%). 61 Cu and 64 Cu share several physical properties; they are imageable with PET, have half-lives that allow for regional distribution and can both be made from Ni starting material. However, several production paths exist for these isotopes, and depending on the enrichment of the starting material, the production cost will vary significantly. Several 61 Cu production routes start from Ni target materials, including the nat Ni(d,x) 61 Cu, 60 Ni(d,n) 61 Cu, and 61 Ni(p,n) 61 Cu reactions, as seen in Table 1  Cu provide an opportunity to study biodistributions of larger molecules with slower kinetics, such as peptides or antibodies (with 61 Cu perhaps better suited for same-day imaging and 64 Cu allowing for later time-point imaging). However, their application in PET imaging will depend on the purpose of the study, as 61 Cu has a higher sensitivity, i.e. 3.43 vs 0.98 cps/Bq/mL (Williams et al. 2005), but slightly lower spatial resolution compared with 64 Cu (Williams et al. 2005). The physical decay properties of 64 Cu can also result in a relatively larger effective dose compared with 61 Cu. E.g. the effective dose for [ 64 Cu]Cu-PTSM and [ 61 Cu]Cu-PTSM as perfusion imaging agents were 3.8 vs 2.5 mSv per 100 MBq respectively, according to Williams et al. (2005). Thus, 61 Cu and 64 Cu are both valuable diagnostic radionuclides whose applications should be tailored to their physical decay characteristics. However, relatively more papers are published on 64 Cu than on 61 Cu; 1288 vs 113 hits between 2009 and 2019 for the keywords 64 Cu and 61 Cu respectively (Scopus)which likely arises from simplified, distribution-friendly 64 Cu logistics. Nevertheless, the increased availability of cyclotron production facilities, increasing solid target infrastructure, and automated radiochemistry systems compel reconsideration of the utility of 61 Cu.
Additionally, regardless of the production route, several Co radioisotopic contaminants will be produced, with 55 Co, 57 Co and 58 Co being of greatest interest due to their relatively long half-lives ( 55 Co t ½ = 17.5 h, 57 Co t ½ = 272 d, 58 Co t ½ = 71 d). Their quantities will depend on the selected reaction, irradiation conditions, target thickness, and isotopic abundance of Ni in the target material. Consequently, in addition to separating Cu from the stable Ni target material, efficient separation of Co is also necessary. As enriched Ni may be cost-prohibitive to implement as single-use, especially 61 Ni, target recycling is imperative. Efficient (> 96%) recovery and re-plating processes have been described (Avila-Rodriguez et al. 2007). For this reason, though 61 Ni targets were not employed in this study, target recovery/recycling was also investigated, including a preliminary production using 64 Ni. This paper focuses on obtaining high-purity 61 Cu via a cassette-based automated separation method using a two-column approach implemented on the FASTlab chemistry platform. A proof-of-concept 64 Cu production (N = 1) was performed to demonstrate applicability of this method to the 64 Ni(p,n) 64 Cu reaction. This is interesting as [ 64 Cu]Cu-DOTA-TATE has recently been granted FastTrack review by the US FDA 1 and has been determined to produce more true-positive lesion detections for neuroendocrine tumours than [ 68 Ga]Ga-DOTA-TOC (Johnbeck et al. 2017). Regardless of radioisotope, the Cu product must consistently be of high radionuclidic purity with a high apparent molar activity (AMA). Thus, it is important to have a robust separation chemistry and rigorous quality control (QC) process to ensure the high quality of the product.

Bench-top pre-studies
Prior to irradiating electroplated Ni, target dissolution and chemical separation processes were investigated at the bench using a heater block and FASTlab. Dissolution studies, initially on Ni foil, probed the effects of acid concentration, H 2 O 2 to HCl ratio, and temperature on dissolution efficiency, speed, and compatibility with downstream separation chemistry. Separation studies focused on minimizing Ni, Co, and Fe in the final product. These separation studies were performed by spiking dissolved stable Ni with ppm Cu, Co, and Fe, and analysing samples via semi-quantitative colorimetric tests (such as Merck's Mquant® colorimetric Ni kit, part number 1.14420, which allow for sub-ppm analysis), or, with low activity (kBq) spikes of 61 Cu and 55 Co, produced by proton (≤ 5 μAmin) and deuteron (≤ 5 μAmin) irradiations of nat Ni foil. During the benchtop tests, the relative distribution of 61 Cu (and similarly 55 Co) were determined for the collected fractions with an Ortec LaBr (digibase, brilliance 380) gamma spectrometer.

Target preparation and recycling
Electroplated targets ranging from 70 to 120 mg were prepared by first dissolving natural Ni powder (Alfa Aesar, 99.8%, 325 mesh) in 2 mL of concentrated HNO 3 (Optima Grade, Fisher Chemical) and drying down under N 2 gas flow at 85°C. Next, the dried Ni was prepared into an electroplating solution following the method of Piel et al. (1992) to electrodeposit Ni onto gold plating substrates. We adapted this method to electrodeposit onto 99.9% silver plating substrates (10 mm deposited diameter). Briefly, the dried Ni was reconstituted in 2.3 mL of 2.4 M H 2 SO 4 (made from concentrated, Optima Grade H 2 SO 4 , Fisher Scientific, and 18 MΩ-cm milli-Q water) and the pH of the solution was brought to~9.1 using2 .5 mL concentrated NH 4 OH (28%, Optima Grade, Fisher Scientific). To the pHadjusted solution 250-300 mg of (NH 4 ) 2 SO 4 (99.9999%, Puratronic, Alfa Aesar) was added, and the solution was quantitatively transferred to an electrolytic cell. With a platinum wire cathode and laboratory DC power supply, constant currents of 40-90 mA were tested for optimization purposes, with a voltage of 6-7.5 V applied through the static electrochemical cell for 1-4 days.
A similar setup was used for re-plating of targets following irradiation. Namely, the Ni collection fraction following purification was then dried down under N 2 gas flow. As above with the fresh target material, 2 mL of concentrated HNO 3 was added and the solution dried again. The dried, recycled Ni was then electroplated on a silver substrate as above for subsequent irradiation. The 64 Ni (84.8 mg) was electroplated on a gold plating substrate, according to the method of Piel et al. (1992) directly.

Target irradiation
Electroplated, nat Ni targets were irradiated with 8.4 MeV deuterons on a PETtrace 800 cyclotron (GE Healthcare) equipped with a QIS (ARTMS) automated target handling system with typical beam currents of 20-30 μA, and typical irradiation times being either ≤30 min for initial tests (N = 3), or, up to 3.3 h (i.e. one half-life) for scaled-up demonstration (N = 1). To enable recycling comparison, both 1× recycled targets (N = 3) and 2× recycled targets (N = 2) were evaluated. For the preliminary enriched target production, the 64 Ni was irradiated with nominally 13.1 MeV protons at 20 μA for 1 hour. These experiments are summarized in Table 2.
For optimisation purposes, nat Ni targets were additionally irradiated for approximately 1 min with 1 μA of protons to produce a radiocobalt tracer via the nat Ni(p, x) 55 Co reaction. However, for quantitative analysis, 58 Co was used due to its longer half-life compared to 55 Co ( 58 Co t ½ = 71 d, 55 Co t ½ = 17.5 h), as 58 Co will be predominantly formed through the 60 Ni(d,α) 58 Co reaction.

Dissolution
Following automated transportation of the irradiated nat Ni target from the cyclotron to the hot cell docking station, the target capsule was transferred to the QIS dissolution unit with tongs. Based on preliminary bench experiments, dissolution was in 3 mL 1:1 7 M HCl (Ultrapur, Merck): 30% H 2 O 2 (ultratrace analysis, Merck) whereby H 2 O 2 was added both to improve the dissolution and to oxidize Fe ions to Fe 3+ . These two reagents were mixed on-line and circulated over the target surface at 2 mL/min for~23 min at an estimated solution temperature of~60°C (based on a heater sleeve set point of 111°C and probing of the heated capsule exterior with a thermocouple). Finally, 3 mL 11.1 M HCl was automatically added to the dissolution solution, with 90 s bubbling with air to mix. Prior to transferring the solution to the FASTlab at 1 mL/min, the sequence was momentarily paused (N = 5), and~200 μL of the nominal 6 mL solution was removed for pre-purification analysis. The irradiated 64 Ni target was manually dissolved in a benchtop dissolution block and transferred to the FASTlab using a peristaltic pump at 1 mL/min, as with the automated target handling.
[ 61 Cu]CuCl 2 separation Separation was implemented on a cassette-based FASTlab platform using a 1 mL TBP column (a tributyl-phosphate-based resin, particle size 50-100 μm, pre-packed, Triskem, Britany, Fr) and 2 mL TK201 column (a tertiary-amine-based weak ionic exchange resin containing a small amount of a long-chained alcohol, particle size 50-100 μm, pre-packed, Triskem, Brittany, Fr) automatically conditioned in series with 7 mL H 2 O and 6 mL 11.1 M HCl. The cassette reagent vials were prepared using concentrated HCl (Optima Grade, Fischer Scientific), NaCl (ACS, Fischer Scientific) and milli-Q water (Millipore system, 18 MΩ-cm resistivity). A general schematic of the resin loading, washing, and elution steps is given in Fig. 1, with detailed process steps described below: Process steps: 1) The acid-adjusted dissolution solution (approx. 6 mL) was loaded over both columns in series and directed into a "Ni collection vial". The TBP resin acted as a guard column as it quantitatively retained Fe 3+ ions, while the Cu and Co chloride complexes were quantitatively retained on the TK201 resin. 2) Both columns were washed with 4 mL 6 M HCl to maximize Ni recovery for future recycling.
3) The TK201 column was washed with 5.5 mL 4.5 M HCl to elute the majority of Co (Waste #1) 4) The TK201 column was washed with 4 mL of 5 M NaCl in 0.05 M HCl to decrease residual acid on the resin and further remove any Co (Waste #2). In the longer term, Waste 1 and Waste 2 will be combined, but were separated for this study for analytical/optimization purposes. 5) The TK201 column was washed with 2 mL of 0.05 M HCl to quantitatively elute the [ 61 Cu]CuCl 2

Gamma ray spectrometry
For the electroplated target irradiations, 61 Cu and 58 Co activities were quantified by high purity germanium (HPGe) gamma spectrometry with an Al-windowed Canberra Model GC1519 (15% relative efficiency, full-width at half-maximum at 1173 keV = 1.8 keV) and used to determine the distribution in the samples and fractions. The gammas used for analysis were: 58 Co (810.759 keV, I γ = 99.45%) and 61 Cu (primarily 282.956 keV, I γ = 12.2% and 656.008 keV, I γ = 10.8%). Samples were counted at a range of distances to the front face of the cylindrical detector body and were selected to maintain the dead time to ≤15%. The energy and efficiency calibration of the detector were performed using a 5-source method: 241 Am, 133 Ba, 137 Cs, 152 Eu, and 60 Co.

Microwave plasma atomic emission spectrometry
Trace metal standards for Co, Cu, Fe, Ni, and Zn (1000 mg/L) were purchased from Sigma Aldrich. Trace metal analysis was conducted on aliquots of the collected sample fractions using an Agilent Technologies Model 4200 Microwave Plasma Atomic Emission Spectrometer (MP-AES). The concentration of HCl in each analysed sample aliquot was adjusted to 0.5 M. Calibration standards of 10, 50, 100, and 500 ppb and 1, 5, 10, and 50 ppm concentrations containing Co, Cu, Fe, Ni, and Zn were prepared in 0.5 M HCl and quantified using two atomic emission wavelengths for analysis of each element. The optimum wavelengths, depending on signal intensities, limits of detection, standard deviations and obvious interferants, were determined for the different elements post analysis. The molar activity (MA) and separation factors (SF) of/for the [ 61 Cu]CuCl 2 product can be calculated from MP-AES quantified stable Cu and Ni. However, when assessing the separation factors, some samples contained Ni in concentrations below the method detection limit (MDL). When this was the case, our calculations assumed the MDL of 10 ppb as a conservative estimate of the Ni quantity.

Apparent molar activity
The apparent molar activity (AMA) was determined by titration with NOTA and adapted from the process described by McCarthy et al. (1997). Namely, 500 μL of [ 61 Cu]CuCl 2 was added to 0.6 mL 0.25 M Sodium acetate (anhydrous, 99% pure, Fisher Scientific), a solution pH of 4-5 was verified using Whatman pH strips, and the solution was vortexed. Next, 100 μL of this activity mixture was added to each of ten vials pre-loaded with both 40 μL 0.25 M sodium acetate (pH 4-5) and 100 μL of NOTA (~0.001-10 nmol). Samples were vortexed, individually assayed for activity, and incubated at room temperature for 15 min. Thin-layer chromatography (TLC) was performed by spotting each sample onto an aluminium-backed silica plate, developing in 1:1 MeOH:(10% w/v) NH 4 OAc and scanning on an OptiQuant autoradiography system (Perkin Elmer Cyclone Plus Storage Phosphor System). In plotting the sigmoidal curve of percent binding vs. NOTA concentration, the NOTA concentration required for 50% binding was identified. The AMA was then calculated as the average sample activity (decay corrected to EoB) and divided by 2× the NOTA concentration required for 50% binding.

Product HCl titration
To evaluate the suitability of the product formulation, and ensure residual HCl was minimized, titrations were performed to determine the HCl concentration of the product fraction. To this end, 0.5 mL of product fraction (N = 5) was added to approximately 10 mL milli-Q water with 100 μL of phenolphthalein in an Erlenmeyer flask with magnetic stir bar, and 5.8 mM NaOH was added dropwise from a burette until a faint pink colour visually persisted in the solution.

Yields: 61 Cu and [ 61 Cu]CuCl 2
The overall 61 Cu yield was assessed by assaying the activity of the isolated product, waste vials, Ni recovery vial, resins, and target plate post dissolution. This resulted in a saturation yield (± SD) of 207 ± 26 MBq/μA (N = 3) for the initial low current optimization runs and 190 ± 33 MBq/μA (N = 5) for the subsequent 1×/2 × −recycled material. No significant difference was noted in saturation yield of the 61 Cu between initial low current and recycled target irradiations. These saturation yields are also a conservative estimate due to the likely presence of unquantified residual activity in the lines and manifold, and possible fractional intercept of the 10 mm diameter Ni plating by the deuteron beam. Nevertheless these saturation yields are considered to be in reasonable agreement with the reported saturation yield (IAEA 2020b) of 248 MBq/μA for 8.4 MeV deuterons on nat Ni. With regards to chemical processing, the average activitybased dissolution efficiency was 97.5 ± 1.4% (N = 5) with an average radiochemical yield of 90.4 ± 3.2% (N = 5) from the separation and an average dissolution + separation processing time of 65 ± 3 min. Plating efficiencies of 96.0 ± 0.9% (N = 3) were demonstrated for the fresh nat Ni targets onto silver backings, as calculated from the nat Ni input and plating masses. For the recycled nat Ni targets, overall recycling efficiencies, i.e. the percent of Ni recovered and re-plated, were collectively determined to be 88% and 92% respectively for the first and second round of recycling.

Formulation
Select recent examples of [ 6x Cu]CuCl 2 purification methods are given in Table 3. Many suffer from lack of automation or have final formulations in large volumes or high acid This study (Ohya et al. 2016 concentrations. A product with a large volume or high acid concentration may need large buffer quantities or potentially time-consuming reformulation steps which are also subject to transfer losses. The motivation of this work was to address these concerns by developing a fast, efficient, automated process with attractive final formulation qualities. When stating formulation, one must not assume that the acid concentration of the product is identical to that used for elution. As such, we titrated the HCl concentration of the [ 61 Cu]CuCl 2 product to assess the formulation directly. The HCl concentration of the product fraction was assessed on a subset (N = 5) of the product Cu vials (i.e. from the recycled productions) and determined to be 0.057 ± 0.002 M. This low product HCl concentration may circumvent the need for further product processing, such as roto-evaporation and reconstitution. This, in combination with the small product volume (2 mL), facilitates downstream radiolabelling by reducing the need for buffering. As a result, we are able to plan radiolabelling chemistries on the same single-use cassette using the FASTlab radiochemistry system.
Compared with the works cited in Table 3, our separation method is relatively fastsurpassed only by that of Matarrese et al. (2010). However, to achieve adequate labelling conditions, the method of Matarrese et al. would require~40× the buffer than what was used in this work. Our method and the method presented by Ohya et al. (2016Ohya et al. ( , 2019 have similar radiochemical yields. However, their method requires evaporation and has a considerably longer preparation and processing time. Overall, compared with the other methods in Table 3, our automated cassette-based purification method generally has a shorter preparation and process time, lower reagent consumption and more suitable product formulation. (8.3 ± 0.6) × 10 − 5 %, resulting in a reduction of the 58 Co to 61 Cu ratio by more than a factor of 500 following purification.

MAs, AMAs, and SFs
For the selected optimum wavelengths (Co 350 nm, Cu 324 nm, Fe 260 nm, Ni 352 nm and Zn 213 nm), the results of the MP-AES analysis largely generated concentrations below the method detection limit (MDL). For the analysed [ 61 Cu]CuCl 2 samples, only Cu produced consistent signals above the MDL, and the average product solution's Cu concentration was determined to be 176 ± 37 ng/mL (N = 5). Co, Zn and Fe all produced signals < MDL (i.e. 100, 100, and 500 ppb respectively).
To assess the chemical purity and applicability of the product, MAs, AMAs, and Ni separation factors were determined and compiled, as reported in Table 5. Table 4 The normalized activity distribution data of 58 Co and 61 Cu (N = 5), which is visually represented in Fig. 2 For longer irradiations, the MA is anticipated to increase proportionally with the produced activity, as the amount of stable Cu would not be expected to change significantly. Although not explicitly measured, if similar stable Cu (to the 30 min irradiations) is assumed for the 3.3 h irradiation, an estimated MA of 479 GBq/μmol [12.9 Ci/μmol] (N = 1) is calculated from the measured activity of that run at EoB. Additionally, assuming similar starting Ni quality, the MA would be expected to increase nearly 4-fold for enriched 60 Ni targets. Therefore, MAs approaching 2 GBq/nmol [50 Ci/μmol] are not unrealistic for enriched 60 Ni targets, and even higher MAs may be possible for enriched 61 Ni irradiation. With this in mind, the results in Table 5 are promising, particularly when considering that the theoretical maximum MA for 61 Cu is 35 GBq/nmol [939 Ci/μmol].
While a high MA is certainly desirable, it does not guarantee a chemically pure product capable of labelling, as other contaminants could be present and compete with the 61 Cu for the chelator. A potentially more informative value is the AMA, which considers not only stable Cu, but also competing stable contaminants other than Cu. However, the AMA will depend on the chelator and labelling conditions, so comparing AMAs between different experimental setups is not straightforward.

Proof of concept -64 Cu
To demonstrate applicability to 64 Cu, a single test irradiation on enriched 64 Ni was additionally performed. We report herein a NOTA-based AMA at EoB of 179 GBq/ μmol [4.8 Ci/ μmol], and radiochemical yield of 93%, with an overall processing time of 65 min. Additionally, the 64 Ni was recovered post-dissolution and re-plated, resulting in a recycling efficiency of 95%. This was on par with our noted nat Ni recycling efficiencies between 88 and 92%.

Conclusion
An automated method capable of producing high purity 61 Cu from recycled nat Ni targets has been developed. In summary, the product formulation is 2 mL of < 0.06 M HCl and the entire process can be achieved in~65 min from EoB to EoS, with average radiochemical yields of 90.4 ± 3.2%, and AMAs > 5 Ci/μmol for NOTA when irradiating with 30 μA for 3.3 h. The [ 61 Cu]CuCl 2 isolation radiochemistry reported here has a measured separation factor ≥ (2.2 ± 1.8) × 10 6 for Ni and 58 Co radionuclidic impurity of < 0.0001% relative to 61 Cu activity. Our process is also directly applicable to the production of 64 Cu via the 64 Ni(p,n) 64 Cu reaction, as was demonstrated in a preliminary, enriched 64 Ni target irradiation and purification test. Additionally, the repeatable, single-use, cassette-based method is facile to introduce into GMP environments; its final 61 Cu radiochemical yield and HCl concentration have relative standard deviations of 3.2 and 3.6%, respectively (N = 5). Finally, the method and FASTlab platform offer the possibility to perform radiolabelling on the same cassette as the separation.