A High Separation Factor for 165Er from Ho for Targeted Radionuclide Therapy

Background: Radionuclides emitting Auger electrons (AEs) with low (0.02–50 keV) energy, short (0.0007–40 µm) range, and high (1–10 keV/µm) linear energy transfer may have an important role in the targeted radionuclide therapy of metastatic and disseminated disease. Erbium-165 is a pure AE-emitting radionuclide that is chemically matched to clinical therapeutic radionuclide 177Lu, making it a useful tool for fundamental studies on the biological effects of AEs. This work develops new biomedical cyclotron irradiation and radiochemical isolation methods to produce 165Er suitable for targeted radionuclide therapeutic studies and characterizes a new such agent targeting prostate-specific membrane antigen. Methods: Biomedical cyclotrons proton-irradiated spot-welded Ho(m) targets to produce 165Er, which was isolated via cation exchange chromatography (AG 50W-X8, 200–400 mesh, 20 mL) using alpha-hydroxyisobutyrate (70 mM, pH 4.7) followed by LN2 (20–50 µm, 1.3 mL) and bDGA (50–100 µm, 0.2 mL) extraction chromatography. The purified 165Er was radiolabeled with standard radiometal chelators and used to produce and characterize a new AE-emitting radiopharmaceutical, [165Er]PSMA-617. Results: Irradiation of 80–180 mg natHo targets with 40 µA of 11–12.5 MeV protons produced 165Er at 20–30 MBq·µA−1·h−1. The 4.9 ± 0.7 h radiochemical isolation yielded 165Er in 0.01 M HCl (400 µL) with decay-corrected (DC) yield of 64 ± 2% and a Ho/165Er separation factor of (2.8 ± 1.1) · 105. Radiolabeling experiments synthesized [165Er]PSMA-617 at DC molar activities of 37–130 GBq·µmol−1. Conclusions: A 2 h biomedical cyclotron irradiation and 5 h radiochemical separation produced GBq-scale 165Er suitable for producing radiopharmaceuticals at molar activities satisfactory for investigations of targeted radionuclide therapeutics. This will enable fundamental radiation biology experiments of pure AE-emitting therapeutic radiopharmaceuticals such as [165Er]PSMA-617, which will be used to understand the impact of AEs in PSMA-targeted radionuclide therapy of prostate cancer.

A recent publication of the biomedical cyclotron production and radiochemical isolation of 165 Er from holmium reported the production of 1.6 GBq 165 Er in a 10 h, 10 µA proton irradiation [24]. The 165 Er was isolated in a 10 h process through a CX/αHIB column using a proprietary resin with 12-22 µm particle size followed by an EXC column using LN3 resin to concentrate the product. While the isolation of 165 Er from macroscopic Ho was achieved, neither the residual mass of Ho in the 165 Er product, the Ho/Er SF, nor radiopharmaceutical labeling results were reported [24]. The present work aims to improve the irradiation intensity tolerance of holmium targets to allow for the production of GBq-scale 165 Er in shorter irradiations, develop a radiochemical isolation process that utilizes commercially available resins to achieve a high Ho/Er SF in shorter times, and demonstrate the radiopharmaceutical quality of the produced 165 Er through chelator-based titration apparent molar activity (AMA) measurements and labeling a clinically relevant DOTA-based radiopharmaceutical (PSMA-617).

Ho Target Preparation and Irradiation
Cyclotron targets were prepared from holmium metal foils. Initially, 300-640 µm thick holmium foils (99.9%, Alfa Aesar, Haverhill, MA, USA) were used. Based on certificates of analysis, two differing foil lots contained 0.06% (600 ppm) or <0.01% (<100 ppm) erbium. High-purity 0.5 mm-thick, 10 mm diameter holmium metal discs with 0.5 ppm Er were purchased from the U.S. Department of Energy (DOE) Ames Laboratory Materials Preparation Center (MPC), Ames, IA, USA. Based on proton energy loss calculations [44], a 300 µm holmium degrades 12.5 MeV protons to 7.1 MeV, below which the 165 Ho(p,n) 165 Er nuclear reaction cross-section vanishes [24]. Holmium foils were wrapped in 25 µm stainless steel foil and rolled to desired thickness (190-400 µm) using a commercial rolling mill. A commercial disc cutter (Pepetools) was used to punch holmium discs of desired diameter (3-9.5 mm). The holmium target disc was centered and spot-welded to a 0.5 mm-thick, 19 mm diameter tantalum disc using a variable transformer-controlled (60-75% power) commercial 115 V spot welder fitted with a copper or silver electrode as previously described [45]. Roughly 6-10 individual spot welds uniformly cover the 7 to 70 mm 2 holmium target.
Spot-welded holmium metal targets were proton irradiated at the University of Wisconsin using two biomedical cyclotrons-an RDS-112 (CTI Cyclotron Systems, Knoxville, TN, USA) and a PETtrace (GE Healthcare, Uppsala, Sweden). For RDS-112 irradiations, the target disc was clamped to a water-jet-cooled target support fixture with a 12.7 mm apertured aluminum ring and irradiated with 10-20 µA of undegraded 11 MeV protons. For PETtrace irradiations, a commercial solid target irradiation and transfer system (ARTMS QIS, Vancouver, BC, Canada) was used. Holmium targets discs were assembled into the water-jet cooled transfer capsule, positioned 3.6 cm down-beam from a water-cooled 500 µm thick aluminum degrader, and irradiated with 20-40 µA of 12.5 MeV protons.
For low intensity irradiations, 165 Er was quantified by high-purity germanium (HPGe) gamma spectrometry (full width at half maximum at 1333 keV = 1.6 keV, Canberra Inc., Meriden, CT, USA) of the irradiated target disc, while correcting for the self-attenuation of the 46-55 keV x-rays. The HPGe low energy range (30-300 keV) was efficiency calibrated using 133 Ba and 241 Am calibration standards (Amersham, United Kingdom).

165 Er Quality Control
Holmium mass across the separation procedure was quantified using microwave plasma atomic emission spectrometry (MP-AES, MP4200, Agilent Technologies, Santa Clara, CA, USA). Holmium standard solutions of 0.1-50 ppm were made by dissolving holmium chloride hydrate (REaction ® 99.99% (REO), Alfa Aesar, Haverhill, MA, USA) in water or holmium metal (US DOE Ames Laboratory MPC, Ames, IA, USA) in 11 M HCl, followed by dilution in 0.1 M HCl. The limit of detection for holmium was determined to be~0.1 ppm in an undiluted sample solution. The MP-AES-analyzed samples were typically diluted by a factor of 2-10 in 0.1 M HCl. The Ho/Er SFs of the various separation steps and the overall separation procedure were calculated according to Equation (1) with m Ho,before/after being the MP-AES-quantified holmium mass before/after the separation The apparent molar activity (AMA) of the final 165 Er solution was determined by titration using tetraazacyclododecane-1,4,7,10-tetraacetic acid (DOTA, Macrocyclics Inc, Plano, TX, USA) and diethylenetriaminepentaacetic acid (DTPA, Acros Organics, Geel, Belgium) as previously described [46]. To polypropylene tubes, 0.5-3 MBq (20 µL) of 165 Er in 0.01 M HCl, 1 M NaOAc (100 µL, pH 4.7, 99.995% trace metals basis, Aldrich, St. Louis, MO), and DOTA or DTPA (100 µL, 0.03-3 µg/mL) in 18 MΩ·cm water were added. Following incubation at 85 • C for DOTA and 21 • C for DTPA for 1 h, each tube was assayed by thin layer chromatography (TLC) using silica-based stationary phase (J.T. Baker, Phillipsburg, NJ, USA) and 0.05 M disodium ethylenediaminetetraacetic acid (EDTA, Fisher Scientific Co., Pittsburgh, PA, USA) as mobile phase. Radioactivity distribution on the TLC plates was visualized using a Cyclone Plus phosphor plate reader (Perkin Elmer, Waltham, MA, USA). Free 165 Er had a retention factor (R f ) of~1, chelated 165 Er-DOTA had an R f of~0.2, and 165 Er-DTPA had an R f of~0.7. To compute the AMA, 165 Er activity in MBq was divided by twice the number of moles of DOTA/DTPA required to complex 50% of the radioactivity, and the value was reported as MBq of 165 Er per nmol of ligand, or MBq/nmol (mean ± standard deviation, SD).
In vitro stability of [ 165 Er]PSMA-617 was investigated in the presence of L-ascorbic acid and freshly prepared normal human serum prepared from lyophilized powder (009-000-001, Jackson ImmunoResearch Laboratories, Inc., West Grove, PA, USA) reconstituted using phosphate buffered saline (2.0 mL, PBS, Lonza Bioscience). Dry [ 165 Er]PSMA-617 (3.1 MBq) and L-ascorbic acid (0.6 mg) were dissolved in serum (300 µL) and incubated at 37 • C for 12 h. Aliquots of the [ 165 Er]PSMA-617 complex were assessed by analytical HPLC at t = 1 and 12 h. The chromatograms were analyzed by integration of the [ 165 Er]PSMA-617 peak compared to all radioactive peaks (free erbium or decomposition products) in the chromatogram.
The distribution coefficient (logD value) of [ 165 Er]PSMA-617 was determined using a 1:1 (v/v) solution of n-octanol (Alfa Aesar, Haverhill, MA, USA) and PBS according to previously reported methods [11]. A sample of [ 165 Er]PSMA-617 (8.2-51 MBq, 2.8-8.4 MBq/nmol) was dried and diluted with PBS and n-octanol (700 µL each). The solution was vigorously agitated for 5 min before being centrifuged at 1000 rpm for 5 min. An aliquot of PBS and n-octanol was analyzed by HPGe gamma spectrometry under identical geometries and the ratio of decay-corrected net counts per minute was used to determine the distribution coefficient. The PBS aliquot required 4-5 days of decay before acquisition of an HPGe spectrum with acceptable (<5%) dead-time. Uncertainty in the LogD value was calculated by propagation of error associated with HPGe counting statistics and in the half-life of 165 Er (10.36 ± 0.04 h [47]). The logD experiment was repeated for five independent preparations of [ 165 Er]PSMA-617 and logD reported as average and standard deviation of results.

Ho Target Preparation and Irradiation
Cold rolling, disc cutting, and spot-welding methods were well suited for the fabrication of cyclotron irradiation targets of tightly controlled dimensions and mass from a variety of commercial holmium metal sources. The malleability of holmium allowed for the dramatic thinning of metal foils through rolling. While thickness reduction by factors of two or three was readily achieved, when an eight-fold change in thickness was attempted, cracking around the edges was observed, as shown in supplementary Figure  S1. Spot-welded holmium was well adhered to the tantalum backing and withstood proton irradiation at all investigated intensities with only minor discoloration, as shown in Figure 1.
617 peak compared to all radioactive peaks (free erbium or decomposition products) in the chromatogram.
The distribution coefficient (logD value) of [ 165 Er]PSMA-617 was determined using a 1:1 (v/v) solution of n-octanol (Alfa Aesar, Haverhill, MA, USA) and PBS according to previously reported methods [11]. A sample of [ 165 Er]PSMA-617 (8.2-51 MBq, 2.8-8.4 MBq/nmol) was dried and diluted with PBS and n-octanol (700 µ L each). The solution was vigorously agitated for 5 min before being centrifuged at 1000 rpm for 5 min. An aliquot of PBS and n-octanol was analyzed by HPGe gamma spectrometry under identical geometries and the ratio of decay-corrected net counts per minute was used to determine the distribution coefficient. The PBS aliquot required 4-5 days of decay before acquisition of an HPGe spectrum with acceptable (<5%) dead-time. Uncertainty in the LogD value was calculated by propagation of error associated with HPGe counting statistics and in the half-life of 165 Er (10.36 ± 0.04 h [47]). The logD experiment was repeated for five independent preparations of [ 165 Er]PSMA-617 and logD reported as average and standard deviation of results.

3a. Ho target preparation and irradiation
Cold rolling, disc cutting, and spot-welding methods were well suited for the fabrication of cyclotron irradiation targets of tightly controlled dimensions and mass from a variety of commercial holmium metal sources. The malleability of holmium allowed for the dramatic thinning of metal foils through rolling. While thickness reduction by factors of two or three was readily achieved, when an eight-fold change in thickness was attempted, cracking around the edges was observed, as shown in supplementary Figure S1. Spot-welded holmium was well adhered to the tantalum backing and withstood proton irradiation at all investigated intensities with only minor discoloration, as shown in Figure  1. For the PETtrace cyclotron, a proton irradiation energy of 12.5 MeV centers the 165 Ho(p,n) 165 Er excitation function peak (see supplementary Figure S2 [22][23][24]) within the energy loss window of the protons traversing a 200-300 µ m-thick holmium target. Experimental end-of-bombardment (EoB) 165 Er physical yields were measured via attenuationcorrected HPGe of target discs or dose calibrator measurements of dissolved target aliquots, or CX elution fractions ( Table 2). The yields show a significant dependence on the Experimental 165 Er physical yields were significantly lower than these theoretical maxima, likely because the cyclotron-integrated charge includes protons impinging on the target outside the holmium diameter. This is especially problematic for the PETtrace cyclotron, which has an oblong beam spot with full width at half maxima of 11 and 8.7 mm, as measured by autoradiography of irradiated aluminum discs. The RDS-112 cyclotron provides a significantly smaller beam spot, resulting in higher overall 165 Er physical yield, despite the lower irradiation energy and smaller target diameter. However, the RDS-112 cyclotron is limited to maximum irradiation current of 20 µA, a factor of 2 below that routinely used with the PETtrace cyclotron.

165 Er Radiochemical Isolation
The holmium target was dissolved over 5 min at room temperature and evaporated to dryness in 30 min. The time of dissolution/evaporation/reconstitution/CX injection was 50 ± 20 min (n = 13).
Step 1: CX/αHIB The first step of the 165 Er isolation procedure accomplishes a bulk holmium/erbium separation while accommodating 180 mg holmium loading masses through CX/αHIB column chromatography with commercially available CX resin in a standard stainless steel semipreparative high pressure chromatography column housing. This 19.6 mL column has a theoretical capacity of 1.8 g of trivalent Ho 3+ , ten times larger than the intended holmium loading masses. Based on the Dy/Ho separation process of Mocko et al. [33], 70 mM αHIB (pH = 4.7) was used as mobile phase. When mobile phase was freshly prepared and carefully pH adjusted to within 0.05 pH units, consistent retention times were observed with 90% of the total 165 Er radioactivity eluting in a Gaussian-shaped peak 40-90 min after injection, as shown in the representative radiochromatogram ( Figure 2). Holmium elutes after 165 Er with its leading edge beginning at~350 mL as seen through the diminishing Ho/Er SF with increasing 165 Er fraction collection volume. Use of mobile phase that was not freshly prepared or incorrectly adjusted to too low of a pH resulted in significantly longer retention times and diminished separation of 165 Er and Ho. With 5 mL/min flow, the column pressure was routinely 17-19 MPa. Following 165 Er elution, the column was stripped with freshly prepared 0.5 M αHIB (100 mL, pH = 4.7), followed by water (250 mL). The column was re-used until the flow pressure significantly increased (>22 MPa), upon which it was disassembled and repacked with fresh resin slurry, every~10 uses. a reproducible, optimal balance between yield and SF for this step, an inline radiation detector was used to determine when to stop collecting the 165 Er fraction. Following loading ≤120 mg Ho, 165 Er fraction collection was ended when the radioactivity signal was ~1/10th maximum value, resulting in ~95% 165 Er recovery. Following loading ~180 mg Ho, 165 Er fraction collection was ended when the radioactivity signal was ~1/4th maximum value, resulting in ~90% 165 Er recovery. Step 2: LN2 EXC The second step of the 165 Er isolation procedure accomplishes a high Ho/Er SF while accommodating milligram quantity holmium masses through EXC using commercially available LN2 resin in a polypropylene column. When filled to maximum capacity (500 As determined by dose calibrator measurements of 165 Er and MP-AES measurements of Ho in the eluted CX fractions, the CX/αHIB column accommodated 180 mg Ho loading mass and effectively removed bulk Ho with acceptable 165 Er yield. Loading 111 ± 17 mg Ho and recovering 94.7 ± 2.5 % of 165 Er resulted in a SF of 320 ± 210 (n = 6). Loading 174 ± 8 mg Ho and recovering 95.3 ± 1.8 % of 165 Er resulted in a SF of 130 ± 60 (n = 5). For the larger loading masses, decreasing 165 Er recovery to 90.5 ± 1.4 % resulted in an SF of 250 ± 150 (n = 5) and further decreasing 165 Er recovery to 80% resulted in a SF of 1000 ± 400 (n = 1). These results indicate the sensitivity of the CX/αHIB separation to Ho loading mass and demonstrate the challenging balance between 165 Er recovery and Ho/Er SF. To ensure a reproducible, optimal balance between yield and SF for this step, an inline radiation detector was used to determine when to stop collecting the 165 Er fraction. Following loading ≤120 mg Ho, 165 Er fraction collection was ended when the radioactivity signal was~1/10th maximum value, resulting in~95% 165 Er recovery. Following loading 180 mg Ho, 165 Er fraction collection was ended when the radioactivity signal was~1/4th maximum value, resulting in~90% 165 Er recovery.
Step 2: LN2 EXC The second step of the 165 Er isolation procedure accomplishes a high Ho/Er SF while accommodating milligram quantity holmium masses through EXC using commercially available LN2 resin in a polypropylene column. When filled to maximum capacity (500 mg), the 1.3 mL column has a theoretical capacity of 36 mg of trivalent lanthanides according to the Triskem product sheet. However, a significant decrease in chromatographic performance was observed when loading more than 5% theoretical capacity, limiting the LN2 column capacity to 1-2 milligrams of holmium (see Supplementary Material Section S1).
Following the CX/αHIB column, the acidified Ho/ 165 Er solution was loaded onto the LN2 column, trapping both 165 Er and Ho. Based on previously published studies [41], the column was rinsed with 0.4 M HNO 3 to affect the differential elution of holmium before erbium. As shown in Figure 3, elution with 0.4 M HNO 3 (50 mL) removed >99% of the holmium, along with a cumulated~20% of the 165 Er. The remaining 165 Er was rapidly eluted with 1 M HNO 3 (~5 mL). To avoid moderate to severe decrease in chromatographic performance, care was taken to prevent the resin bed from going dry during use and columns were freshly packed and conditioned prior to each experiment (see Supplementary Material Sections S2 and S3).
Molecules 2021, 26, x FOR PEER REVIEW 9 of 15 mg), the 1.3 mL column has a theoretical capacity of 36 mg of trivalent lanthanides according to the Triskem product sheet. However, a significant decrease in chromatographic performance was observed when loading more than 5% theoretical capacity, limiting the LN2 column capacity to 1-2 milligrams of holmium (see supplementary material Section S1). Following the CX/αHIB column, the acidified Ho/ 165 Er solution was loaded onto the LN2 column, trapping both 165 Er and Ho. Based on previously published studies [41], the column was rinsed with 0.4 M HNO3 to affect the differential elution of holmium before erbium. As shown in Figure 3, elution with 0.4 M HNO3 (50 mL) removed >99% of the holmium, along with a cumulated ~20% of the 165 Er. The remaining 165 Er was rapidly eluted with 1 M HNO3 (~5 mL). To avoid moderate to severe decrease in chromatographic performance, care was taken to prevent the resin bed from going dry during use and columns were freshly packed and conditioned prior to each experiment (see supplementary material Sections S2 and S3). In the optimized procedure, LN2 columns loaded with Ho (570 ± 370 µ g) and rinsed with 0.4 M HNO3 (52 ± 9 mL) resulted in 78 ± 6% 165 Er recovery and a Ho/Er SF of 1020 ± 320 (n = 4). The LN2 SF was estimated from the Ho mass in the final preparation of 165 Er, assuming no Ho/Er separation was achieved with the final bDGA column. Three addi- In the optimized procedure, LN2 columns loaded with Ho (570 ± 370 µg) and rinsed with 0.4 M HNO 3 (52 ± 9 mL) resulted in 78 ± 6% 165 Er recovery and a Ho/Er SF of 1020 ± 320 (n = 4). The LN2 SF was estimated from the Ho mass in the final preparation of 165 Er, assuming no Ho/Er separation was achieved with the final bDGA column. Three additional replicates where holmium was below the MP-AES detection limit and four additional replicates with deviations from the above-described experimental procedure were excluded from this analysis. In two of these excluded replicates, a 0.4 M HNO 3 (42 mL) rinse resulted in 96% 165 Er recovery and a Ho/Er SF of 290 and a 0.4 M HNO 3 (52 mL) rinse resulted in a 63% 165 Er recovery and an SF of >2000. These results demonstrate the sensitivity of the procedure to HNO 3 concentration/volume and the interplay between Ho/Er SF and 165 Er recovery. The latter of the two results indicates that simply using a set volume of 0.4 M HNO 3 to remove Ho may lead to irreproducible 165 Er recovery and Ho/Er SF. To ensure a reproducible, optimal balance between recovery and SF, the radioactivity in the 0.4 M HNO 3 eluant is quantified by a dose calibrator as it is collected in 1-5 mL fractions. After~20% of the loaded radioactivity had eluted, the mobile phase was changed to 1 M HNO 3 to elute the remaining high purity 165 Er with the optimized results reported above.
Step 3: bDGA EXC The final step of the 165 Er isolation procedure reduced trace metal impurities while concentrating the product into a small volume, low acidity solution suitable for radiolabeling using a commercially available bDGA resin. The loading (5 M HNO 3, 5 mL) and rinsing (3 M HNO 3, 15 mL) solution concentrations were chosen due to literature studies showing high erbium and low transition metal (Fe, Co, Ni, Cu, Zn) affinities on a similar DGA resin [48]. A small volume rinse (0.5 M HNO 3 , 2 mL) decreased column acidity and minimized subsequent 165 Er elution volume. Following this loading and rinsing routine, 97 ± 2 % of the loaded 165 Er was recovered in 0.01 M HCl (1.2 ± 0.2 mL) (n = 18). Smaller elution volumes were attainable by fractionation of the 0.01 M HCl elution solution, with the optimized (n = 5) elution profile shown in Table 3. Overall separation: The overall chemical isolation procedure from start-of-dissolution to final 165 Er isolation in 0.01 M HCl (~400 µL) was 4.9 ± 0.7 h (n = 13). The optimized procedure had a decay-corrected 165 Er recovery of 64 ± 2% and a Ho/Er SF of (2.8 ± 1.1)·10 5 and was validated for holmium target masses up to 180 mg, with the final 165 Er fraction containing 370 ± 180 ng of residual holmium and no detectable radionuclidic impurities (Supplementary Figure S6) after separation (n = 4).

Radiosynthesis and Characterization of [ 165 Er]PSMA-617
Following successful isolation, the [ 165 Er]PSMA-617 was synthesized with the labeling yields summarized in Table 5

Discussion
For application in receptor-targeted therapeutic radiopharmaceuticals, a high molar activity is necessary to achieve good radiolabeling yield of small pharmaceutical masses with large amounts of radioactivity. This high molar activity ensures that the biologically administered mass of radiopharmaceutical is sufficiently small to not saturate the targeted receptor on diseased cells. When prepared for human use, [ 177 Lu]PSMA-617 and [ 177 Lu]DOTATATE are radiolabeled with high yield at a molar activity of 60 MBq/nmol [26,27]. For cyclotronproduced 165 Er, achieving a high molar activity must begin with careful consideration of the holmium target material. Many commercial sources of holmium have significant erbium impurity, which, in turn, will limit the maximum attainable molar activity of 165 Er produced in the proton irradiation of an impure holmium target. Additionally, as shown in Table 2, the holmium target and irradiation parameters significantly affect the overall quantity of 165 Er that can be produced in a bombardment, with larger targets having greater overall yields compared to their smaller counterparts. However, larger holmium mass targets come with the added challenges of a more difficult 165 Er/Ho separation and a higher cold erbium and holmium burden in the purified 165 Er fraction.
In addition to sourcing Ho target material with low Er impurity content, an effective Ho/Er radiochemical isolation procedure is necessary. Because of the chemical similarity between the adjacent lanthanide elements, any residual holmium in the 165 Er final formulation will affect the molar activity of 165 Er. A high Ho/Er SF is accomplished in this work by a multi-step separation process utilizing cation exchange and extraction chromatography. Based on MP-AES analysis of the final 165 Er product, this 165 Er radiochemical isolation process gives a Ho/Er SF of (2.8 ± 1.1) · 10 5 . The 165 Er radiochemical yield was 64 ± 2% calculated by the ratio of the decay-corrected 165 Er activity in the final product and the 165 Er activity produced in the irradiated target.
Two main sources of erbium/holmium-cold erbium target impurity and residual holmium due to incomplete separation-are the two dominating terms in the denominator of supplementary Equation (S1), which calculates a maximum achievable EoB MA for a given 165 Er preparation. With the experimental 165 Er production and radiochemical separation results presented above, supplementary Equation (S1) yields a calculated EoB 165 Er MA of 9.9 ± 0.5 MBq/nmol for a 1 h, 12.5 MeV, 130 mg, 9.5 mm Ø, low purity (100 ppm Er) holmium target PETtrace irradiation. This calculated MA EoB is in good agreement with the measured DOTA/DTPA 165 Er AMAs (Table 4). In this case, the calculated EoB MA is nearly entirely driven by the cold erbium impurity present in the holmium target material (second term in supplementary Equation (S1) denominator). This underscores the fact that under the investigated irradiation conditions, utilization of holmium targets with high erbium impurity content will limit the molar activity of the resulting 165 Er to values lower than typically acceptable for therapeutic radiopharmaceutical research applications. Producing high MA 165 Er requires holmium with extremely low erbium content, such as target material that has been prepurified from erbium, or the DOE Ames Laboratory MPC metal used in this work.
For an identical irradiation of a high-purity (0.5 ppm Er) holmium target, supplementary Equation (S1) yields an 165 Er EoB MA of 240 ± 60 MBq/nmol, with this molar activity being driven by cold holmium remaining in the 165 Er preparation after their separation by a factor of (2.8 ± 1.1) · 10 5 (third term in supplementary Equation (S1) denominator). However, the EoB 165 Er AMAs (Table 4) Table 5) do not reflect this high value of MA. This disagreement may be a result of nanomole, sub-ppm trace metal (Zn, Fe, Cu) impurities in the radiolabeling reactions, which are not considered in supplementary Equation (S1). This hypothesis is supported by the fact that, for 165 Er isolated from high-purity holmium targets, the titration-based AMA measured using DTPA were 5-22 times higher than for DOTA (n = 3). This difference in DTPA/DOTA AMA values, which has also been observed for cyclotron-produced 86 Y [49], is likely due to the non-selective nature of DOTA's metal binding properties. Compared with DTPA, DOTA binds 10,000 times stronger to Fe 2+ (LogK FeDOTA = 20.22 ± 0.07 [50], LogK FeDTPA = 16.0 ± 0.1 [51]), 100 times stronger to Zn 2+ (LogK ZnDOTA = 20.8 ± 0.2, LogK ZnDTPA = 18.6 ± 0.1 [51]), and 10 times stronger to Cu 2+ (LogK CuDOTA = 22.3 ± 0.1, LogK CuDTPA = 21.5 ± 0.1 [51]), causing these three common trace metal impurities to be significantly more problematic in DOTA versus DTPA radiochemical labelings. Thus, a systematically higher DTPA-based AMA compared with DOTA-based AMA is indicative that Fe/Zn/Cu-based trace metal impurities in the radiolabeling solutions are significantly impacting the 165 Er AMA values. The impact of trace Zn, Fe, and Cu on the [ 165 Er]PSMA-617 radiolabeling experiments is supported by the presence of UV-absorbing impurities with analytical HPLC retention times equivalent to nat Zn-PSMA-617, nat Fe-PSMA-617, and nat Cu-PSMA-617 in the final radiopharmaceutical preparation (Supplementary Figure S7).
This work represents the first published radiosynthesis of [ 165 Er]PSMA-617, a radiopharmaceutical that could serve as a useful in vitro and in vivo tool that can be used to assess the role of AEs in the efficacy of PSMA-targeted radionuclide therapy of prostate cancer using [ 161 Tb]PSMA-617 [11]. The radiopharmaceutical is stable in serum for at least 12 h and has an octanol-water partition coefficient of LogD = −3.3 ± 0.3, less polar than [ 161 Tb]PSMA-617 (−3.9 ± 0.1) [11], [ 44

Conclusions
A 2 h biomedical cyclotron irradiation and 5 h radiochemical separation can produce GBq-scale 165 Er suitable for high-yield radiolabelings of DOTA-based radiopharmaceuticals at molar activities befitting investigations of targeted radionuclide therapeutics. The separation utilizes column chromatography with commercially available resins and is well suited for automation. This significant step forward in the production and high holmium/erbium SF radiochemical isolation of 165 Er will enable fundamental radiation biology experiments of pure AE-emitting therapeutic radiopharmaceuticals. Proof-ofconcept radiolabeling studies were successfully performed synthesizing [ 165 Er]PSMA-617, which will be utilized in vitro and in vivo to understand the role of AEs in PSMA-targeted radionuclide therapy of prostate cancer.