Evaluation of a Wet Chemistry Method for Isolation of Cyclotron Produced [²¹¹At]Astatine

Abstract

A “wet chemistry” approach for isolation of ²¹¹At from an irradiated bismuth target is described. The approach involves five steps: (1) dissolution of bismuth target in conc. HNO₃; (2) removal of the HNO₃ by distillation; (3) dissolution of residue in 8 M HCl; (4) extraction of ²¹¹At from 8 M HCl into DIPE; and (5) extraction of ²¹¹At from DIPE into NaOH. Results from 55 “optimized” ²¹¹At isolation runs gave recovery yields of approximately 78% after decay and attenuation corrections. An attenuation-corrected average of 26 ± 3 mCi in the target provided isolated (actual) yields of 16 ± 3 mCi of ²¹¹At. A sixth step, used for purification of ²¹¹At from trace metals, was evaluated in seven runs. In those runs, isolated ²¹¹At was distilled under reductive conditions to provide an average 71 ± 8% recovery. RadioHPLC analyses of the isolated ²¹¹At solutions, both initial and after distillation, were obtained to examine the ²¹¹At species present. The primary species of ²¹¹At present was astatide, but astatate and unidentified species were also observed. Studies to determine the effect of bismuth attenuation on ²¹¹At were conducted to estimate an attenuation factor (~1.33) for adjustment of ²¹¹At readings in the bismuth target.

1. Introduction

Several research groups are evaluating the use of α-particle-emitting radionuclide astatine-211 (²¹¹At) for targeted radionuclide therapy of cancer [1,2,3]. While there have been few clinical studies with ²¹¹At-labeled radiopharmaceuticals [4,5], preclinical studies have been directed at treatment of several different cancer types, including ovarian cancer [6,7], glioma [8], leukemia [9,10], lymphoma [11], breast cancer [12,13,14], squamous cell cancer [15,16], and prostate cancer [17]. The high interest in use of targeted α-emitters comes from potential advantages of their very high local cytotoxicity brought about by the α-particle’s short path length and its high linear energy transfer (LET) [18]. These and other properties of α-emitters, such as high cytotoxicity independent of tissue oxygen concentration and dose rate [19,20], provide an advantage over β-emitting radionuclides for some cancer therapy applications. Of the α-emitting radionuclides, ²¹¹At is particularly attractive, as it has negligible high-energy photon emissions and does not have any long-lived α-emitting daughter radionuclide(s). Further, its half-life (7.21 h) is long enough to allow ample time for preparation, quality control and administration before an appreciable amount has decayed, but short enough for its use in an outpatient setting. Most importantly, due to its physical and radiobiological properties, ²¹¹At is very attractive for use in targeted treatment of disseminated (metastatic) cancer, blood-borne cancers, minimal residual disease, and disease isolated in compartmental spaces (e.g., ovarian and pancreatic cancers in the peritoneal cavity). An effective treatment of these conditions could greatly improve the five-year survival rates in advanced stages of cancers [21].

Production of ²¹¹At can be readily accomplished by irradiating a natural abundance bismuth target (i.e., 100% ²⁰⁹Bi) with a cyclotron-produced α-particle beam [22]. We have used this ²⁰⁹Bi(α,2n) reaction to produce the ²¹¹At in all of our studies. Isolation of the ²¹¹At from the irradiated bismuth target is generally accomplished by dry distillation [22,23,24,25,26]. The dry distillation method for isolation of ²¹¹At can be relatively efficient, providing up to 70% recovery if parameters such as temperature of oven, inert gas flow rates, and size and design of glassware are optimized. We optimized the parameters of the dry distillation isolation approach [27,28] and used that method of isolation for nearly two decades. Even with process optimization, the recovered yields from the dry distillation varied greatly at times. In addition to that problem, more recently our bismuth target was increased in size to produce higher quantities of ²¹¹At required to undertake clinical studies [29]. The increase in the size of bismuth target (and its aluminum backing) resulted in having to greatly increase the size of quartz tube used in the high temperature oven. Attempts to remove the bismuth from the aluminum backing mechanically or through heat were deemed too dangerous. An alternate method of cutting the aluminum backing with a small band saw in a charcoal filtered Plexiglas glovebox was used for a short time, but this was also deemed too dangerous for routine application.

An alternate method for isolation of ²¹¹At from bismuth targets is referred to as the “wet chemistry” isolation method [22]. In this method, the bismuth target is dissolved in nitric acid, and in subsequent steps, the ²¹¹At is separated from the bismuth. The wet chemistry approach was used in early studies for isolation of ²¹¹At [30,31] and has been employed more recently by other research groups [32,33]. This approach was attractive for our purposes as it appeared to: (a) be an approach where there was control over the ²¹¹At recovery at each step, such that more consistent yields might be obtained; (b) provide high recovery yields; (c) be scalable to clinical and commercial levels of ²¹¹At production; and (d) ultimately allow automation of the isolation process. For these reasons, we undertook an investigation of the wet chemistry approach for isolation of ²¹¹At from bismuth targets. The results from our investigation are reported herein.

2. Experimental Section

2.1. General

Reagents used were obtained from commercial sources as analytical grade or better and were used without further purification, except for diisopropyl ether (Sigma-Aldrich, St. Louis, MO, USA), which was distilled within 24 h of use to purify and remove any stabilizers. Standard methods for safely handling radioactive samples were employed. Handling and processing of the irradiated target was conducted in a glovebox (Innovative Technologies, Inc., radioisotope glovebox, Amesbury, MA, USA) vented through a charcoal filter on the glovebox exhaust, and subsequently vented through tubing into a second charcoal-filtered Plexiglas enclosure (12” × 16” × 21”; Biodex 112-038) within a radiochemical fume hood. A general purpose rate meter (model 3A) with an alpha detection probe (model 43-2) and a second general purpose rate meter (model 3A) with a gamma detection probe (model 44-3) were placed on a shelf at the inside back of the glovebox for monitoring use during the isolation procedures.

2.2. Dose Calibrator Measurements and Calibration

Measurements of ²¹¹At were conducted on either a Capintec CRC-15R or a CRC-55t Radioisotope Calibrator (Capintec Inc., Ramsey, NJ, USA) using the calibration number of 046 for ²¹¹At (calibration # calculated by KG). Initial measurements were conducted by insertion of the irradiated target into the dose calibrator, aligning the target (in a plastic zipper bag) into the same position for run-to-run comparisons. All other measurements were obtained by placing glass or plastic vials containing ²¹¹At into the dose calibrator using the Plexiglas vial/syringe dipper.

The Capintec dose calibrator calibration setting for ²¹¹At was empirically determined by cross calibration against a high purity germanium (HPGe) detector (AMETEK, Oak Ridge, TN, USA). The HPGe detector, coupled to a PC-based multichannel analyzer, provided spectra that were analyzed using Maestro-32 software (ORTEC, Oak Ridge, TN, USA). As the HPGe detector efficiency varies substantially in the ~80 keV range (i.e., the most abundant photons of ²¹¹At), the less abundant 687.0 keV (I_γ = 0.261%) photon of ²¹¹At, and the 596.65 (I_γ = 0.311%) and 897.8 (I_γ = 0.320%) photons of the short-lived ²¹¹Po daughter (branching ratio of 58.20% applied above) were selected for quantification. The HPGe detector was calibrated using ¹³³Ba, ¹³⁷Cs, and ⁶⁰Co at a sample-detector distance of 17 cm in the presence of a 2 mm sheet of copper and a 3 mm sheet of steel placed adjacent to the sample. This material was applied to allow for assaying of a substantial quantity of ²¹¹At while maintaining reasonable dead-time by attenuating the high abundance ~80 keV photons of ²¹¹At.

To obtain the cross calibration, 100 μL of purified ²¹¹At was aliquoted into the tip of a microcentrifuge tube and assayed for approximately 2 min on the HPGe detector. The dead-time was <16%, the statistical error of each photopeak was <2%, and the activity at the start of counting was determined to be 26.6 ± 0.7 MBq (~718 µCi) when calculating the average and standard deviation of the three photopeaks. Within the hour, the same microcentrifuge tube was re-assayed in Capintec models CRC-55tR and CRC-15R dose calibrators, whereby a calibration number of 046 was found to be necessary to display the HPGe-based decay-corrected ²¹¹At activity.

2.3. Preparation and Irradiation of Bismuth Targets

The University of Washington Medical Center (UWMC) houses a Scanditronix MC-50 positive-ion source cyclotron capable of accelerating and extracting α-particles with an energy range of 27.0–47.3 MeV and currents of up to 70 μA. A 0° beam extraction line and external target station were designed and installed to scale-up to clinical production levels of ²¹¹At [29]. Irradiation of a bismuth target (34 mm × 160 mm × 7 mm) placed on a 10° incline with a 50 μA, 29 MeV α-beam routinely produces 19+ mCi of ²¹¹At (as measured in the dose calibrator) in roughly a 45 min irradiation period.

The ²⁰⁹Bi metal target (99.999%, Aldrich) was prepared by melting >5 g of Bi into a machined cavity in the aluminum target body (18 mm × 120 mm × 0.30 mm) on a hot plate at 330 °C. During the melting process, a porcelain spatula was used to constantly scratch the aluminum surface below to eliminate air bubbles and improve adherence of Bi. After cooling to room temperature, the uneven bismuth surface was machined flat to yield a bismuth (and surrounding aluminum) thickness of 0.17 mm. At the specified depth (0.17 mm), width (18 mm) and length (120 mm) a mass of bismuth of ~3 g can be calculated, but the tolerances for milling the cavity in the aluminum backing can change the mass significantly. The bismuth target was installed on the beamline as previously described [29].

2.4. Procedure for Wet Chemistry Isolation of Na[²¹¹At]At

The wet chemistry isolation process is conducted in a charcoal-filtered radiochemical glovebox designed to prevent any release of ²¹¹At into the atmosphere. A picture of the glovebox is shown in the Supplemental Material (Figure S1). Most reagents, glassware and equipment (e.g., heater-stirrers, lab jacks, etc.) required for the isolation are placed in the glovebox prior to introduction of the irradiated target. A list of those items is provided in the Supplemental Material (List S2). Prior to the introduction of the plastic bag containing the irradiated target into the glovebox, the fan removing air through the charcoal filter is turned on to provide negative pressure in the glovebox. The exhaust from the fan passes through a charcoal filter within the glovebox and then through tubing, which is vented into a Plexiglas box that has two stacked charcoal filters. The exhaust-facing surfaces of both charcoal filters are examined to determine if ²¹¹At has been released from the glovebox. A picture of the Plexiglas enclosure with the vent tubing is shown in the Supplemental Material (Figure S3).

Just prior to beginning the procedure, the irradiated bismuth target, inside of a plastic zipper bag used while transporting from the cyclotron vault to the radiochemistry laboratory, was placed in a dose calibrator to estimate the amount of ²¹¹At produced. It was then passed through the antechamber into the glovebox. There is a second dose calibrator in the glovebox, so the target is read in that dose calibrator as well. Any ²¹¹At measurements performed on material within the glovebox are compared with the target reading obtained from the dose calibrator within the glovebox.

Step 1: Dissolution of Bi target in HNO₃. The irradiated bismuth target was removed from the plastic zipper bag and placed bismuth face down in a 1L modified polypropylene bottle. A 10 mL aliquot of conc. HNO₃ (ThermoFisher, Waltham, MA, USA) was then added by pipet to the bottle. The dissolution of the bismuth target was allowed to proceed for 10 min. The HNO₃ solution was transferred by a pipet to a 50 mL round bottom flask. An additional 5 mL of HNO₃ was added to the dissolution bottle to rinse the target and container. After gently shaking the flask, the wash HNO₃ was also transferred to the round bottom flask.

Step 2: Removal of HNO₃. The 50 mL flask containing Bi and ²¹¹At dissolved in conc. HNO₃ was connected to a short-path distillation head. That glassware was lowered into a 50 mL round bottom aluminum heating block on a hot plate at ~300 °C. The distillation flask and neck of distillation head were covered with aluminum foil. After rapidly heating the HNO₃, it was distilled with gentle stirring over a 20–30 min period. The distillation was stopped when no further HNO₃ distillate drops were visible. At that time the distilling glassware was raised from the heating block, and the white residue in the distillation flask was allowed to cool for 10 min.

Step 3: Dissolution in HCl. The 50 mL distillation flask containing the white residue was removed from distillation head. To that flask was added 8 mL of 8 M HCl (ThermoFisher, Waltham, MA, USA), and the mixture was agitated until all solid was dissolved (5–10 min). The solution was transferred via pipet to a 20 mL glass scintillation vial. The distillation flask was rinsed with an additional 2 mL of 8 M HCl as a wash, and that solution was also transferred to the scintillation vial.

Step 4: DIPE Extraction and HCl Washes. A 4 mL aliquot of freshly distilled diisopropyl ether (DIPE) (Sigma-Aldrich, St. Louis, MO, USA) that had been saturated with 8 M HCl was added to the scintillation vial and the resulting biphasic mixture was stirred for 10 min. Following that, the aqueous layer was removed and counted in a dose calibrator. A 5 mL quantity of DIPE-saturated 8 M HCl was added to the DIPE in the scintillation vial, and that biphasic mixture was stirred for 5 min. Following agitation, the aqueous HCl layer was again removed via pipet and counted in a dose calibrator. The preceding two steps were repeated an additional three times (for a total of 4 washes). The DIPE solution in the scintillation vial was counted in a dose calibrator to determine the ²¹¹At activity remaining.

Step 5: Back-Extraction into NaOH. A 600–700 µL quantity of 4 M NaOH (ThermoFisher, Waltham, MA, USA) was added to the DIPE in the scintillation vial, and the vial was stirred for 10 min. The pH of the basic aqueous ²¹¹At solution was checked to confirm that it was strongly basic (pH > 13). The aqueous layer (containing the ²¹¹At) was then removed and both layers were counted separately in the dose calibrator to determine the total amount of ²¹¹At isolated. If the ²¹¹At was to be used without a further purification, the isolated NaOH solution was neutralized (brought to pH 6.8–7.0) by addition of aliquots of HCl or NaOH solutions (4 M, 1 M, 0.5 M and 0.1 M) as required.

Step 6: Distillation under reductive conditions. To a 25 mL round bottom flask containing 9.19 mL DI-water, 0.75 M ferrous sulfate heptahydrate (Sigma-Aldrich, St. Louis, MO, USA), in 1.5 M H₂SO₄ (ThermoFisher, Waltham, MA, USA) was added the basic ²¹¹At solution. The resultant ²¹¹At solution was stirred to re-dissolve the characteristic cloudy greenish precipitate, and the flask was connected to a cold-water jacketed distillation head. To the receiving side of the apparatus was attached a 10 mL round bottom flask pre-charged with 100 µL of 1 M NaOH. The distillation assembly was lowered into a 25 mL round bottom aluminum heating block (and an ice bath for the receiving flask), the distilling flask and neck of the distillation head were covered with aluminum foil, and the ²¹¹At solution was distilled at 200 °C (hot plate temperature) with gentle stirring. The collected ²¹¹At, free of bismuth metal and other contaminants from the wet chemistry procedure, was used for radiolabeling experiments. In some studies, the aqueous solution was reduced in volume while heating (e.g., 80 °C) under an argon stream to provide a higher concentration of ²¹¹At.

2.5. HPLC Analysis of Isolated Na[²¹¹At]At

Analyses of the identity of ²¹¹At species and purity of the isolated Na[²¹¹At]At solutions, both from the initial isolation and the solutions of distilled ²¹¹At were performed by radioHPLC. Two radioHPLC systems were used. The HPLC system used in the studies was a Hewlett-Packard model 1050 HPLC (Hewlett-Packard Company, Palo Alto, CA, USA) with a Beckman Model 170 Radioisotope Detector (Beckman-Coulter, Brea, CA, USA) to detect the radioactive species present. The radionuclide solution was evaluated with a YMC J’sphere ODS-M80 reversed-phase column (YMC America, Allentown, PA, USA) under isocratic conditions using a 1:1 mixture of 0.05 M TBAP (tetrabutylammonium phosphate) at pH 4.5 and acetonitrile [34]. A second analytical radioHPLC method was conducted on the HPLC equipment described using a Dionex IonPac AS-20 anion exchange column with a Dionex AG-20 guard column (Dionex, Sunnyvale, CA, USA) eluting with an isocratic solvent system consisting of 50 mM NaOH and a flow-rate of 1 mL/min.

2.6. Evaluation of Bismuth Attenuation in Quantification of ²¹¹At

The effect of bismuth attenuation on the dose calibrator reading of ²¹¹At for a fixed volume was examined by taking readings in the presence of varying amounts of Bi in HNO₃. In the experiment, an aliquot of isolated ²¹¹At was added to 5 mL of concentrated HNO₃ in a 20 mL glass vial. This vial was assayed (CRC-55tR), after which ~200 mg pellet of bismuth was dissolved, the solution re-assayed, and this process repeated until the total reached ~1.5 g total bismuth was dissolved (average pellet size: 219 ± 32 mg). In plotting the data, all readings were normalized to the no bismuth added measurement and decay-corrected to account for the time lapsed during bismuth dissolution.

The effect of sample volume was examined by increasing the volume of the above solution with addition of 1 mL aliquots to a total volume of 15 mL (assaying after each addition). Although an increase in measured activity was noted with increasing volume, it was not conclusive as to whether this change was attributed to a change in geometry and/or an overall decrease in bismuth concentration. To assess the effect of geometry, a 0.3 mL sample of isolated (i.e., no bismuth added) ²¹¹At was similarly brought to a total volume of ~15 mL (in 1 mL steps).

3. Results

3.1. Wet Chemistry Isolation of ²¹¹At

The steps involved in the wet chemistry ²¹¹At isolation approach are outlined in Figure 1. The first step in the isolation process is to remove the bismuth and ²¹¹At from the aluminum target backing by dissolution in concentrated HNO₃. This is a very efficient process, where >99% of the ²¹¹At activity is transferred to the round bottom flask used in the next (distillation) step. Pictures of the target (bismuth face down) placed in the modified 1 L bottle are provided in the Supplemental Material (Figure S4). In this step it can be noted that reddish-brown gas (NO_x fumes) is released upon dissolution of the bismuth. It was somewhat surprising to find that this process does not release significant quantities of volatile ²¹¹At. The release of the NO_x gas, however, is damaging to the stainless steel interior of the glovebox, requiring covering the surface with adhesive PTFE sheet (Bytac, Sigma-Aldrich, St. Louis, MO, USA). Studies were conducted to replace the open dissolution chamber with a closed chamber system in which a stream of air (or N₂) was used to sweep the NO_x fumes through a scrubbing bubbler system.

Figure 1. Steps involved in the “wet chemistry” isolation and purification of ²¹¹At from a ²⁰⁹Bi target. Note that the isolation process is usually stopped at step 5, but an additional distillation under reductive conditions can be conducted to obtain ²¹¹At solutions without bismuth and other chemical contaminants from the wet chemistry process.

Figure 1. Steps involved in the “wet chemistry” isolation and purification of ²¹¹At from a ²⁰⁹Bi target. Note that the isolation process is usually stopped at step 5, but an additional distillation under reductive conditions can be conducted to obtain ²¹¹At solutions without bismuth and other chemical contaminants from the wet chemistry process.

The HNO₃ distillation was part of the system, as NOx fumes are also released during distillation. A picture and scheme for the scrubber system is shown in the Supplemental Material (Figure S5). Ultimately, this approach was not used since it would occupy too much space in the glove box.

The second and third steps in the isolation process involve removal of the conc. HNO₃ and dissolution of the residue in 8M HCl. Although it had been shown that having some HNO₃ present did not cause problems in the subsequent DIPE extraction step, the fact that HNO₃ is efficiently extracted into DIPE [35] caused problems with loss of At so all of the HNO₃ was removed by distillation to give a colorless residue. To obtain an efficient HNO₃ distillation, a heat block temperature of ~300 °C is required and the distilling flask needs to be covered with aluminum foil. Once the HNO₃ is removed the white residue was readily dissolved in 8 M HCl. This process is also very efficient with 93%–98% of the activity transferred into the 8 M HCl. The glassware setup is shown in the Supplemental Material (Figure S6).

The fourth step in the isolation process is extraction of ²¹¹At from 8 M HCl into diisopropyl ether (DIPE). In this step the bismuth remains in the 8 M HCl allowing a separation of the bulk quantities of bismuth from the ²¹¹At. To further decrease the levels of bismuth, several washes with 8 M HCl are required. When carrying out the separations, there were difficulties with removing specified amounts of DIPE after mixing due to volume changes, as the DIPE and 8 M HCl are somewhat miscible making it difficult to quantitatively isolate either reagent. Therefore, as with many of the earlier investigators, the DIPE and 8 M HCl were mixed prior to their use to minimize the loss of reagent due to miscibility. The miscibility is problematic as some ²¹¹At is present in each wash (probably in the miscible DIPE). The number of washes required to obtain low levels of Bi was investigated. It was found that four washes were required to obtain a low level of bismuth in the isolated ²¹¹At solutions.

The fifth step in the isolation process is back-extraction of the ²¹¹At from DIPE into 4 M NaOH. This step is thought to hydrolyze the ²¹¹At species in the DIPE to obtain a different species that is soluble in aqueous base. Although it is likely that other species are intermediates, the hydrolysis in strong base ultimately leads to formation of [²¹¹At]astatide [36]. The efficiency of extraction from DIPE was high. Direct use of this isolated ²¹¹At solution for radiolabeling biomolecules was hampered by the fact that varying concentrations of NaOH were present due to varying amounts of 8 M HCl present in the DIPE. Although many attempts were made, including pre-saturating the DIPE with 8 M HCl and carefully separating and aliquoting the 8 M HCl, we were unsuccessful in obtaining consistent concentrations of NaOH. Therefore, the basic solution is generally brought to near neutral pH, so that the pH of the ²¹¹At-labeling reaction mixtures can be easily controlled. Because there are varying concentrations of NaOH, the neutralization can take varying amounts of time. Despite this, good recovered yields have been obtained. The production, isolation, and recovery yields for experiments, where ~45 min irradiations (108 isolation runs), a single 2 h irradiation and a single 4 h irradiation were conducted, are summarized in Table 1. The ~45 min irradiation experiments have been split into the first 53 experiments and the second 55 experiments, as there were small adjustments in the procedure used that improved the yields (labeled as “optimized”) in the later experiments. It should be noted that the time to isolation in the most recent 27 runs is 125 ± 9 min. Spreadsheets showing data from individual experiments used in obtaining the average yields in Table 1 are provided in the Supplemental Material (Table S7 and Table S8).

Table 1. Wet chemistry ²¹¹At Isolation times, activity and yields *.

211At Isolation Runs (most ~0.75 h irrad.)	Activity Produced (mCi) **	Time to Isolation (min)	Isolated Activity (mCi)	Decay Corrected Recovery (%) ***
Non-optimized (n = 53)	20.7 ± 1.7	171 ± 46	13.6 ± 2.3	87 ± 15
Optimized (n = 55)	19.6 ± 2.0	155 ± 50	15.9 ± 2.7	104 ±15
Preparative Run (2 h)	54.0	118	43.2	97
Preparative Run (4 h)	100.5	114	72.0	86

* Values for multiple runs (steps 1–5) are average ± standard deviation; ** The activity produced is a dose calibrator measurement of ²¹¹At in the bismuth target. This value is systematically low due to bismuth attenuation of the dose calibrator reading (see later section); *** The decay corrected recovery yields are high due to the systematically low measurement of ²¹¹At activity produced. See Table 3 for attenuation corrected yields.

Table 1. Wet chemistry ²¹¹At Isolation times, activity and yields *.

211At Isolation Runs (most ~0.75 h irrad.)	Activity Produced (mCi) **	Time to Isolation (min)	Isolated Activity (mCi)	Decay Corrected Recovery (%) ***
Non-optimized (n = 53)	20.7 ± 1.7	171 ± 46	13.6 ± 2.3	87 ± 15
Optimized (n = 55)	19.6 ± 2.0	155 ± 50	15.9 ± 2.7	104 ±15
Preparative Run (2 h)	54.0	118	43.2	97
Preparative Run (4 h)	100.5	114	72.0	86

* Values for multiple runs (steps 1–5) are average ± standard deviation; ** The activity produced is a dose calibrator measurement of ²¹¹At in the bismuth target. This value is systematically low due to bismuth attenuation of the dose calibrator reading (see later section); *** The decay corrected recovery yields are high due to the systematically low measurement of ²¹¹At activity produced. See Table 3 for attenuation corrected yields.

Concern that low levels of bismuth and trace levels of other metal ions might be problematic in some ²¹¹At-labeling reactions, inclusion of a final purification step, Isolation Step 6, was evaluated in several ²¹¹At isolation runs. Inclusion of a final distillation step was also of interest as a potential means for circumventing the time-consuming neutralization after back-extraction into NaOH. Initial distillations were conducted with ²¹¹At at pH 8.5, 6.5–7, and 3. The highest yield (~45%) was obtained with the solution at pH 6.5–7. The distillation at pH 8.5 provided ~24% recovery, and the distillation at pH 3 gave only ~16% recovery. Unfortunately, the amount of ²¹¹At that distilled under acidic pH appeared to be limited by the formation of ²¹¹AtO₃⁻ (by radioHPLC). This led to evaluation of several reductive conditions, all of which gave low recovery yields. Johnson et al. previously reported distillation of ²¹¹At under a variety of conditions, including distilling under reductive conditions (solution containing 0.05 M FeSO₄ in 0.5 M H₂SO₄) [37]. They found that only the distillation under reductive conditions provided high recovery (97%) of ²¹¹At. It seemed that this reducing agent would work for our purification, but since the isolated ²¹¹At solutions were highly basic (with varying amounts of NaOH for reasons previously explained), we chose to have a much higher quantity of H₂SO₄ to offset the base. After some initial studies with lower amounts of reductant and acid, it was found that a solution of 0.75 M FeSO₄ in 1.5 M H₂SO₄ routinely provided a highly acidic solution which resulted in good ²¹¹At recovery yields. The reductive conditions keep the ²¹¹At as astatide during the distillation process. It is presumed that hydroastatic acid is distilled. To assure that the distilled ²¹¹At remain as astatide, the distillate was collected in a vessel precharged with a volume of 0.1 mL of 1 M NaOH. A compilation of the results from the seven reductive distillations employing solutions containing 0.75 M FeSO₄ in 1.5 M H₂SO₄ is provided as Table 2. It was noted that over 3 mL of distillate was required to obtain >80% recovery of activity, but that the majority of activity >60% was obtained in the first 1.2 mL of distillate. By collecting only the first 1.5–2.0 mL of activity the time for the final distillation step was shortened considerably. This shortening of time came about from the shorter time for the distillation, but also from not having to concentrate the final ²¹¹At solution prior to use.

Table 2. Isolation yields after reductive distillation.

Run #	211At Isolate to Distill (mCi)	211At Activity Purified (mCi)	Volume Collected (mL)	% 211At Recovered
1	14.3	12.2	3.4	85
2	8.0	4.8	1.5	61
3	12.3	9.1	1.5	74
4	7.5	5.4	1.2	72
5	15.1	10.8	1.8	72
6	16.5	10.5	1.8	64
7	34.0	23.0	2.0	67

^# Average ± SD 70.7 ± 7.9.

Table 2. Isolation yields after reductive distillation.

Run #	211At Isolate to Distill (mCi)	211At Activity Purified (mCi)	Volume Collected (mL)	% 211At Recovered
1	14.3	12.2	3.4	85
2	8.0	4.8	1.5	61
3	12.3	9.1	1.5	74
4	7.5	5.4	1.2	72
5	15.1	10.8	1.8	72
6	16.5	10.5	1.8	64
7	34.0	23.0	2.0	67

^# Average ± SD 70.7 ± 7.9.

Reasonable average isolation yields were obtained when the first 1.5–2.0 mL of distillate are used, but higher overall recovery yields may be obtained if larger quantities of distillate are isolated. More rapid and efficient methods for removal of water from the isolated aqueous ²¹¹At solution are being investigated.

3.2. RadioHPLC of ²¹¹At Solutions

The nature of the ²¹¹At species produced by the wet chemistry approach was of high interest in our investigation, as the method of isolation may affect the species obtained for radiolabeling. The initial step of the wet chemistry ²¹¹At isolation approach is highly oxidizing, most likely resulting in highly oxidized form of ²¹¹At. The fact that the ²¹¹At species produced is not volatile indicates that may be the case. In addition to the oxidized forms of ²¹¹At, use of 8 M HCl may produce a number of interhalogen species [24,36,38,39]. There was concern that oxidized forms of ²¹¹At might be obtained in the final isolated solution (except where the distillation was conducted under reductive conditions). Therefore, a HPLC method for identifying the ²¹¹At species present was sought. Initial radioHPLC studies were conducted on a YMC J’sphere ODS-M80 C-18 reversed-phase column, previously reported for separation of periodate, iodate and iodide [34]. This column has a very high surface area with high coating (14%) on 4 µm size silica particles with 80Å pore sizes. It has a usable pH range of 2.0–7.5. As only a small amount of the injected ²¹¹At was recovered when the HPLC eluent was acidic (i.e., <10%), it was determined that the column should only be used when the samples were neutralized. Radiochromatograms showing retention of ²¹¹AtO₃⁻ (panel A), ²¹¹At⁻ (panel B), crude reaction product mixture at pH 8 (panel C), and the same product mixture after setting at room temperature for 2.5 h (panel D) are shown in Figure 2. The At⁻ (t_R = 5.4 min) was obtained under reductive conditions to assure that the species was astatide. As a HPLC standard, Na[125I]I eluted at t_R = 6.2 min. The ²¹¹AtO₃⁻ (t_R = 4.8 min) was produced from ²¹¹At^- using the following oxidation conditions; ²¹¹At⁻, NaIO₄, 0.1N H₂SO₄, 80 °C (microwave). The corresponding Na[125I]IO₃, produced under the same oxidation conditions (except at 120 °C) had a t_R = 3.7 min. A number of ²¹¹At species have been observed in the radiochromatograms that could not be identified. In many cases the unidentified species appear to be converted to either astatide or astatate with time (e.g., Figure 2, panel D). Although the retention times for the isolated ²¹¹At species were the same when the pH was constant, slight variations in pH resulted in small changes in peak retention times making it difficult to identify species run-to-run with certainty. This was problematic as the pH of the isolated product varied from batch-to-batch.

Figure 2. Examples of radiochromatograms obtained by eluting ²¹¹At solutions on a YMC J’Sphere ODS-M80 column using isocratic conditions eluting with 1:1 CH₃CN:50 mM TBAP @ 1 mL/min. Panel A shows a radiochromatogram of Na[²¹¹At]AtO₃. Panel B shows a radiochromatogram of Na[²¹¹At]At. Panel C shows a radiochromatogram of ²¹¹At species from an isolation run (from step 5). Panel D shows a radiochromatogram of ²¹¹At species in the same solution as panel C, after setting for 2.5 h.

Figure 2. Examples of radiochromatograms obtained by eluting ²¹¹At solutions on a YMC J’Sphere ODS-M80 column using isocratic conditions eluting with 1:1 CH₃CN:50 mM TBAP @ 1 mL/min. Panel A shows a radiochromatogram of Na[²¹¹At]AtO₃. Panel B shows a radiochromatogram of Na[²¹¹At]At. Panel C shows a radiochromatogram of ²¹¹At species from an isolation run (from step 5). Panel D shows a radiochromatogram of ²¹¹At species in the same solution as panel C, after setting for 2.5 h.

Rössler et al. had previously shown that a strongly basic anion exchange column (BioRad Aminex A 27) could be used to separate ²¹¹At from ¹²⁵I [40]. More recently, Sabatié-Gogova et al. described the use of a Dionex IonPak AS-20 column (with AG-20 guard column) in their studies to characterize the astatine species in solution [41] and Champion et al. used the same column in their evaluation of the hydrolyzed At(III) species under oxidizing and acidic conditions [42]. Discussions with technical personnel at Dionex Corporation also led us to use this chromatographic column for separating At species. It had previously been shown that this column separates chloride from chlorate and perchlorate, bromide from bromate and iodide from iodate (ThermoFisher product guide) using a hydroxide eluent, so it seemed likely that it could be used for separation of ²¹¹At species. This column is a polymer-bound anion exchange resin that is very stable under the highly basic conditions that the ²¹¹At solutions are obtained in, making it particularly attractive for our use. Examples of chromatograms obtained with the Dionex column are shown in Figure 3. The radiochromatogram peak at ~14.3 min (Figure 3, panel A) is [²¹¹At]At⁻ and the peak at 3.3 min is an unidentified species. It is interesting to note that after neutralization of the NaOH solution, the peak at 3.3 min is not observed (Figure 3, panel B), and after setting at room temperature, the ²¹¹At species do not change further (Figure 3, panel C). Distillation in the presence of FeSO₄ in H₂SO₄ in some runs provided a single ²¹¹At species (see Supplemental Material, Figure S9), but in this example (Figure 3, panel D) two additional unknown species were present. Monoclonal antibody radiolabeling reactions using distillation purified ²¹¹At solutions from step 6 of the isolation generally gave slightly lower labeling yields than those obtained with the non-purified ²¹¹At solutions from step 5 (data will be presented elsewhere). It is not clear why this is the case, but it is thought to be due to a higher dilution of activity obtained after purification, not the presence of additional ²¹¹At species.

Figure 3. Radiochromatograms obtained by eluting ²¹¹At solutions on a Dionex IonPac AS20 column using isocratic conditions with 50 mM NaOH @ 1 mL/min. (A) radiochromatogram of ²¹¹At in NaOH solution after isolation; (B) radiochromatogram of neutralized solution; (C) radiochromatogram of neutralized ²¹¹At after setting at room temperature for 2 h; (D) radiochromatogram of ²¹¹At species after distillation.

Figure 3. Radiochromatograms obtained by eluting ²¹¹At solutions on a Dionex IonPac AS20 column using isocratic conditions with 50 mM NaOH @ 1 mL/min. (A) radiochromatogram of ²¹¹At in NaOH solution after isolation; (B) radiochromatogram of neutralized solution; (C) radiochromatogram of neutralized ²¹¹At after setting at room temperature for 2 h; (D) radiochromatogram of ²¹¹At species after distillation.

3.3. Quantification of ²¹¹At in the Presence of Bismuth

It is apparent that bismuth can attenuate the 76.9 and 79.3 keV X-rays used to quantify ²¹¹At in the target and in solution, but the extent of attenuation was not investigated until our average decay-corrected wet chemistry isolated yields became greater 100% (see Table 1: “Optimized” runs had an average of 104% ± 15% decay-corrected yields). Additionally, high variability in the dose-calibrator yields for samples within a particular run suggested that attenuation in samples containing bismuth was larger than we anticipated. Therefore, two sets of experiments were conducted to understand the variability by testing the hypothesis that the dose calibrator measurements observed during the extraction process may be susceptible to bismuth attenuation and/or geometric effects due to varying sample volumes.

In the first experiment, dose calibrator readings were taken for a set amount of ²¹¹At, when the quantity of bismuth was increased in a constant volume of HNO₃. The experiment was conducted by dissolving increasing amounts of bismuth metal (0–1.5 g) in 5 mL of conc. HNO₃ containing ²¹¹At. The data obtained is plotted in Figure 4, panel A. All dose calibrator readings were normalized to the measurement with no bismuth added and decay-corrected to account for the time lapsed during bismuth dissolution. It is clear that the dose calibrator readings were significantly decreased due to attenuation of bismuth. At the highest amount of bismuth added (1.5 g), where the concentration of bismuth in solution was 300 mg Bi/mL HNO₃, an attenuation of ~38% was noted.

Figure 4. Graphs depicting the change in dose calibrator reading of a fixed amount of ²¹¹At when the concentrations of bismuth are changed. (A) Dependence of bismuth concentration vs. dose calibrator reading for a fixed (5 mL) quantity of HNO₃; (B) Dependence of sample volume vs. dose calibrator reading in the presence and absence of bismuth. Values in each graph have been decay corrected and normalized to; (A) the no added bismuth reading, or (B) the 5 mL reading.

Figure 4. Graphs depicting the change in dose calibrator reading of a fixed amount of ²¹¹At when the concentrations of bismuth are changed. (A) Dependence of bismuth concentration vs. dose calibrator reading for a fixed (5 mL) quantity of HNO₃; (B) Dependence of sample volume vs. dose calibrator reading in the presence and absence of bismuth. Values in each graph have been decay corrected and normalized to; (A) the no added bismuth reading, or (B) the 5 mL reading.

In the second experiment, the change in dose calibrator reading of ²¹¹At was examined when the volume was increased from 5 mL to 15 mL by 1 mL additions of H₂O. Although an increase in measured activity was noted with increasing volume, it was not conclusive as to whether this change was attributed to a change in geometry and/or an overall decrease in bismuth concentration. To further assess the effect of geometry, a 0.3 mL sample of isolated (i.e., no bismuth added) ²¹¹At was similarly brought to a total volume of ~15 mL (in 1 mL steps). The results are presented in Figure 4, panel B. That graph clearly shows that the change in reading is almost exclusively attributed to the decrease in bismuth concentration, and not to the change in sample volume. Combining the results of panel A (i.e., ~38% reduction in activity when adding 1.5 mg of Bi), along with the increase of ~21% when diluting this solution to 100 mg Bi/mL (1.5g in 15 mL), the dose calibrator reading for such a bismuth concentration and volume is expected to be ~75% of that achieved for the case where no bismuth is added (i.e., underestimated by ~33%).

The simple experiments to determine the attenuation effect that bismuth has on dose calibrator readings of ²¹¹At show that significant changes in readings are due to attenuation by bismuth. If an approximate attenuation value can be obtained, a more accurate estimate of the recovered yields can be made. Measurement of the amount of ²¹¹At activity in an irradiated target after the HNO₃ dissolution has shown that <1% of the activity remains, so it is possible to say that the amount of activity in the target is essentially the same as that of the conc. HNO₃. Calculations suggest that after milling to 0.17 mm thickness, the targets hold ~3 g of bismuth metal. Unfortunately, this value can vary considerably based on the tolerances of the machining (weights suggest variances may be as large as 1 g). Fortunately, in the dissolution process only part of the bismuth is removed, suggesting that the top portion of the bismuth contains the ²¹¹At. Again, it is not possible to know exactly how much bismuth metal is dissolved, but it seems likely that it may be in the range of 1.5–2.5 g. Also, differences in the weights of bismuth targets before being milled (removing some bismuth and aluminum) and after milling are highly variable. With these variances and the fact that there were no good estimates of the amount of bismuth dissolved from the target, it is not possible to use the data obtained in the graphs to estimate a single global attenuation factor.

Another way to get an approximation of the attenuation factor is to measure a solution of ²¹¹At in the isolation process that does not have bismuth in it. That measurement, corrected for loss of ²¹¹At activity in each step/transfer to that point, and corrected for decay during the steps, can provide a value that approximates the original amount of ²¹¹At activity in target. Once the number is calculated, division of the ²¹¹At target measurement by that number (×100) can provide an estimate of the percent the target reading is of the actual amount. The first solution in the isolation process that does not have bismuth in it is the DIPE solution (prior to NaOH extraction). Fortunately, measurement of the ²¹¹At in the DIPE solution is routinely taken. The calculations that are described above were performed for 13 recent isolation runs. Although it is likely that the bismuth mass was highly variable in the runs, the average percent of (approximated) original ²¹¹At in the 13 runs was 74% ± 4%, which is on par with data of Figure 4. Thus, a bismuth attenuation factor may be approximated to be 1.33. We believe that adjusting the target reading upward by 33% can provide a reasonable estimate of the amount of ²¹¹At activity in the irradiated target, and that value can provide a better estimate of the percent ²¹¹At recovered from the target than obtained without using an attenuation factor.

Adjustment of the amount of ²¹¹At produced and the decay- and attenuation-corrected recovery yields are provided in Table 3. Spreadsheet showing data from individual experiments used in obtaining the average yields in Table 3 are provided in the Supplemental Material (Table S10 and Table S11).

Table 3. Wet chemistry ²¹¹At isolation yields corrected for decay and attenuation *.

211At Isolation Runs (most ~0.75 h irrad.)	Corrected Act. Produced (mCi) **	Isolated Activity (mCi)	Est. Attenuation* & Decay Corrected Recovery (%) ***
Non-optimized (n = 53)	27.6 ± 2.3	13.6 ± 2.3	65 ± 11
Optimized (n = 55)	26.1 ± 2.7	15.9 ± 2.7	78 ± 11
Preparative Run (2 h)	72	43.2	73
Preparative Run (4 h)	134	72.0	65

* An attenuation correction of ~1.33 has been applied to the measured bismuth target value based on an average difference between the ²¹¹At reading in DIPE (corrected for decay and losses) and the initial target reading; ** Estimated mCi of ²¹¹At produced in bismuth target based on approximation that the reading was 75% of actual; *** Estimated recovery yields if decay and attenuation of ²¹¹At emissions are factored into calculations.

Table 3. Wet chemistry ²¹¹At isolation yields corrected for decay and attenuation *.

211At Isolation Runs (most ~0.75 h irrad.)	Corrected Act. Produced (mCi) **	Isolated Activity (mCi)	Est. Attenuation* & Decay Corrected Recovery (%) ***
Non-optimized (n = 53)	27.6 ± 2.3	13.6 ± 2.3	65 ± 11
Optimized (n = 55)	26.1 ± 2.7	15.9 ± 2.7	78 ± 11
Preparative Run (2 h)	72	43.2	73
Preparative Run (4 h)	134	72.0	65

* An attenuation correction of ~1.33 has been applied to the measured bismuth target value based on an average difference between the ²¹¹At reading in DIPE (corrected for decay and losses) and the initial target reading; ** Estimated mCi of ²¹¹At produced in bismuth target based on approximation that the reading was 75% of actual; *** Estimated recovery yields if decay and attenuation of ²¹¹At emissions are factored into calculations.

4. Discussion

This investigation of a wet chemistry approach for isolation of ²¹¹At was prompted by difficulties in our production of ²¹¹At brought about by a large increase in size of the bismuth target used. The increase in target size came about when designing and implementing an external target station for increasing our ²¹¹At production levels [29]. It should be noted that our cyclotron situation is somewhat unique, as an external target was required rather than an internal target [43,44,45] because the cyclotron had to remain in daily operation for fast neutron therapy of cancer patients. In addition to circumventing difficulties with the size increase of target, we were also interested in determining whether the wet chemistry method could provide more consistent ²¹¹At isolation yields than previously obtained with dry distillation. It is our belief that a robust isolation method will be required when conducting clinical studies.

The wet chemistry approach described herein is not new. In 1949, Johnson et al. reported that bismuth targets could be dissolved in conc. HNO₃ with little loss of ²¹¹At when the HNO₃ was distilled away [37]. Eight years later, Neumann described a similar procedure using extraction of ²¹¹At into DIPE and back-extracting into NaOH [30]. In 1973, Neirincks and Smit reported extracting ²¹¹At into DIPE, but used aqueous hydroxylamine to back-extract [31]. More recently Yordanov et al. described a similar approach using DIPE or butylether for the extraction of ²¹¹At [32], and Bourgeois et al. described an approach using the DIPE extract directly for radiolabeling of m-Me₃Snbenzoate N-hydroxysuccinimidyl ester [46]. Zona et al. also described a similar wet chemistry approach to obtain ²¹¹At on silver foils [33]. Interestingly, Alliot et al. reported on an investigation of the extraction mechanism when ²¹¹At is extracted into DIPE [35].

Our goals in investigating the wet chemistry approach were: (1) to evaluate previous approaches and optimize conditions to efficiently isolate ²¹¹At from the large amounts of bismuth (up to 3 g) in our targets; and (2) to demonstrate reproducibility of the optimal approach. The data presented herein support our achieving those goals. A secondary goal is to further refine the wet chemistry approach such that it can be automated. Isolation of ²¹¹At by an automated wet chemistry isolation system is currently under investigation. While all of the steps in the described isolation approach can be readily accomplished, the miscibility issues with DIPE separations requiring pre-saturation of reagents is not favorable for automation, so alternative separation approaches are being sought.

The chemistry of radiolabeling with ²¹¹At is not always straightforward [47,48]. Arguments can be made that some of the radiolabeling difficulties could be due to having species of ²¹¹At present that are not reactive or react differently than expected/desired. In this investigation we examined two radioHPLC methods for separation and identification of ²¹¹At species so that we could use them for identification of the desired Na[²¹¹At]At, and for quality control of the isolated ²¹¹At. Both a C-18 reversed-phase separation method and an ion exchange separation method were evaluated. The ion exchange radioHPLC method has the advantage that it provides good separation of ²¹¹At species when taking samples directly from the highly basic isolation solution, whereas the separations on reversed-phase were best conducted by injection of ²¹¹At solutions that had been neutralized. Our radioHPLC studies identified At species we believe were astatide (At⁻) and astatate (AtO₃⁻) based on using known reactions to obtain these species. Minor amounts of other unidentified species were also seen, but we have demonstrated that the wet chemistry approach can provide high radiochemical purity astatide as a final product.

In addition to knowing the nature of the species in the final product, it is of interest to know the nature of the astatine species obtained in each step of the isolation steps. Unfortunately, this has not been established and is difficult to ascertain. It is generally assumed that step 1, dissolution of the bismuth target in concentrated nitric acid provides At(0) in solution [33], but if that is the case it is surprising that no volatile astatine is released in this open-air vigorous dissolution. Irrespective of the initial At species in step 1, based on the fact that iodine (I₂) is converted to iodate (IO₃⁻) under heating in concentrated HNO₃ [49] it seems highly likely that in step 2 At(0) or other species present will be converted to astatate (AtO₃⁻). The fact that the HNO₃ distillation can be heated to dryness without release of volatile astatine species supports formation of a species such as astatate. Neumann suggested that the use of 8 M HCl to dissolve the residue from the HNO₃ distillation in step 3 provides either AtCl₂⁻ or AtCl₄⁻ [30]. The hypothesis was based on the fact that ICl₂⁻ and ICl₄⁻ are known species and it was known that ICl₂ could be extracted into ethers from HCl solutions. In experiments conducted it was demonstrated that the species formed was efficiently extracted DIPE (as in step 4), but was not extracted into benzene or CCl₄. The later fact ruled out the formation of neutral species such as AtX (where X is a halide). Norseyev and Khalkin later reported on the stability constants of chloride complexes of At formed from varying concentrations of HCl with oxidants present and concluded that AtCl₂⁻ was the likely species formed [50]. Interhalogen species, including AtCl₂⁻, were also prepared by Meyer and Rössler via distillation methods [24]. Back-extraction with 4 M NaOH in step 5 produces a species that has very different properties to the DIPE extracted species. Initial evaluation by reversed-phase and anion exchange radioHPLC can have a number of species present, but with time the minor species disappear leaving only one species, which appears to be astatide. It seems likely that the species in DIPE (perhaps AtCl₂⁻) is hydrolyzed with a strong base to intermediate species before obtaining astatide. It was not the purpose or focus of this investigation to study the At species formed, so the forgoing discussion on potential At species in each step of the isolation process is meant only for consideration. Definitive experiments to identify At species will have to be conducted as separate studies.

As stated before, one of our goals was to optimize the wet chemistry isolation procedure. We were pleased in our initial studies where the recovery yields were estimated to be >80%, but became concerned with our data when the average yields were 100% or higher. In hindsight, it was not too surprising that average optimized yields might be 100% or more, as we knew that the ²¹¹At readings of the irradiated target were low due to attenuation. Yet, it was surprising that the attenuation was so high (readings estimated to be only ~75% of actual). We believe these studies may be the first to evaluate the level of attenuation in the irradiated bismuth target, or in solutions of ²¹¹At that also contain bismuth ions. The fact that an attenuation factor was not used (to the best of our knowledge) in calculating some previously reported ²¹¹At isolated yields for dry distillations means that those reported yields may be higher than they actually were.

5. Conclusions

We have demonstrated with over 100 ²¹¹At isolation runs that the wet chemistry method for isolation of ²¹¹At described in this report is relatively easy to conduct, it provides good recovery yields, is robust and quite reproducible. The average time for an investigator to conduct the wet chemistry isolation (without final distillation) is ~2 h if he/she is familiar with the procedure. In our experience, the 2 h time period is similar to the time it takes to setup and conduct the dry distillation approach. The use of the wet chemistry ²¹¹At isolation rather than the dry distillation will be dependent on an investigator’s unique situation, and perhaps their personal preference, but we have shown that it is an alternative worth considering.