Production of High-Purity Radium-223 from Legacy Actinium-Beryllium Neutron Sources†

244 Current Radiopharmaceuticals, 2012, 5, 244-252 Production of High-purity Radium-223 from Legacy Actinium-Beryllium Neutron Sources†

Chuck Z. Soderquist, Bruce K. McNamara and Darrell R. Fisher*

Isotope Sciences Program, Pacific Northwest National Laboratory, Richland, Washington, USA

Abstract: Radium-223 is a short-lived alpha-particle-emitting radionuclide with potential applications in cancer treatment. Research to develop new radiopharmaceuticals employing 223Ra has been hindered by poor availability due to the small quantities of parent actinium-227 available world-wide. The purpose of this study was to develop innovative and cost-effective methods to obtain high-purity 223Ra from 227Ac. We obtained 227Ac from two surplus actinium-beryllium neutron generators. We retrieved the actinium/beryllium buttons from the sources and dissolved them in a sulfuric-nitric acid solution. A crude actinium solid was recovered from the solution by coprecipitation with thorium fluoride, leaving beryllium in solution. The crude actinium was purified to provide about 40 milligrams of actinium nitrate using anion exchange in methanol-water-nitric acid solution. The purified actinium was then used to generate high-purity 223Ra. We extracted 223Ra using anion exchange in a methanol-water-nitric acid solution. After the radium was separated, actinium and thorium were then eluted from the column and dried for interim storage. This single-pass separation produces high purity, carrier-free 223Ra product, and does not disturb the 227Ac/227Th equilibrium. A high purity, carrier-free 227Th was also obtained from the actinium using a similar anion exchange in nitric acid. These methods enable efficient production of 223Ra for research and new alpha-emitter radiopharmaceutical development. Keywords: Alpha emitters, actinium-227, thorium-227, radium-223, anion exchange.

INTRODUCTION cays by energetic beta (Emax = 1.15 MeV) to 223Ra. This side chain grows in quickly and gives the actinium chain a Actinium is one of the rarest, naturally occurring ele- high beta dose-rate. Because of potentially high dose and ments, and sufficient quantities of actinium radioisotopes are 227 high airborne contamination, curie quantities of Ac (milli- difficult to obtain for research. Actinium-227 is the longest- grams) must be handled and processed in a hot cell rather lived isotope of actinium (half-life = 21.8 years) and is the than in a glove box. only isotope of actinium that can be isolated in milligram or greater amounts. Actinium-227 can be obtained in trace The daughter product 223Ra has important medical appli- amounts from uranium minerals because it belongs to the cations as a therapeutic radionuclide for cancer treatment. natural 235U decay chain. Pure 227Ac metal has a specific Commercial availability of 227Ac is poor [1], and only a few activity of 2.66x103 GBq per gram. Actinium-227 can be small sources of 227Ac are known to exist in the world. Both produced synthetically by neutron irradiation of 226Ra in an 223Ra and 227Th can be generated from highly purified 227Ac. isotope production reactor via the nuclear reaction Möller et al. [2] explored the use of crystalline hybrid or- 226Ra(n,)227Ra (half-life = 42.2 minutes) 227Ac + -, and ganic/inorganic ion exchangers based on zirconium phos- by subsequent separation and purification of 227Ac. The ther- phate and phosphonate compounds for constructing a con- mal neutron cross section for the 226Ra(n,)227Ra reaction is venient 227Ac/223Ra generator system. In contrast, we inves- about 13 barns (National Nuclear Data Center, Brookhaven tigated radiochemical separations to produce 223Ra from National Laboratory; http://www.nndc.bnl.gov/). 227Ac using anion exchange chromatography in methanol- water-nitric acid solutions. Actinium-227 decays by beta-emission to 227Th (half-life = 18.7 days), which in turn decays by alpha-emission to 223Ra (half-life = 11.4 days). The remaining members of the Study Objectives decay chain have shorter half-lives and grow-in completely The purpose of this study was to develop improved ra- within a few hours. The first daughter of 223Ra is 219Rn, a diochemical separation methods for obtaining high-purity noble gas that must be carefully handled since it can cause 223Ra from 227Ac. Medical research on new applications for airborne alpha contamination in the work space. Most of the 223Ra has been hampered by poor availability of the parent members of the 227Ac decay series have gamma emissions 227Ac. To improve the availability of 223Ra, we recovered and that may cause high radiation fields. Actinium-227 also de- purified 227Ac from legacy neutron sources. The long-lived cays by alpha decay through a side chain (1.4% abundance) parent 227Ac decays through a chain of short-lived members to 223Fr (half-life = 21.8 minutes). Francium-223 de- (Fig. 1). The short-lived decay products, after 223Ra decay, contribute alpha and beta particle radiation and increase the 223 therapeutic effectiveness of Ra in medical applications. *Address correspondence to this author at the Isotope Sciences Program, Pacific Northwest National Laboratory, 902 Battelle Blvd., P7-27, Richland, Alpha-emitter Attributes for Cancer Treatment Washington 99354 USA; Tel: 509-375-5098; Fax: 509-375-5099; E-mail: [email protected]. In many instances, alpha-particle radiation excels for †Pacific Northwest National Laboratory in Richland, Washington, is oper- cancer treatment [3] compared to beta emitters and gamma- ated by Battelle for the U.S. Department of Energy. emitters, since short-range alpha particles impart straight,

tional chelates and cell-targeted protein carriers [6]. A bifunctional chelate for 223Ra has been developed but has not been tested [7] in laboratory animals. Alpha emitters should have reasonably short physical half-lives, and should be compatible with targeting moiety biokinetics of the protein carriers to maximize radiation dose to cancer cells and minimize dose to the remainder of the body. Optimal physical half-lives for cancer uptake and clearance biokinetics may be on the order of one to six days, and very short half- lives (such as the 46-minute bismuth-213) may be too short for many cancer-treatment applications. For commercial preparation and shipment to hospitals, physical half-lives should be longer than two or three days. Decay-chain daughters should have half-lives of a few minutes or less to minimize daughter migration away from target sites. The alpha emitters should have imageable photon emissions to aid in Fig. (1). Actinium-227 decay chain. gamma-camera imaging and dosimetry for treatment- planning. Only a few alpha-emitters meet these criteria and short-range (40 to 70 μm) high-ionization-density (90-150 can be provided with excellent availability at reasonable keV/μm) tracks through targeted cells. In addition, one or cost. Toward these objectives, we have studied 223Ra and two direct alpha-particles traversing a cell nucleus may be 227Th as candidate radionuclides for cancer treatment. Ra- sufficient to cause cell death, and repair of sub-lethal damage dium-223 and 227Th from 227Ac may have advantages over is minimized. The practical advantages of these properties other possible alpha-emitter choices in terms of half-life, distinguish alpha-emitters from beta-particle emitters used in relative cost, availability, and chemistry (Table 1). radiopharmaceuticals for targeted radionuclide therapy [4]. Alpha emitters provide effective treatment of cancer me- Examples of Clinical Application tastases, which often lead to cancer progression. For exam- Radium-223 can be administered in the simple, un- ple, the more common beta-emitters used in radionuclide chelated chloride form (Alpharadin™, Algeta ASA; Oslo, therapy cannot as efficiently irradiate and sterilize single Norway) for treating skeletal metastases associated with ad- cells and small metastatic lesions compared to alpha- vanced prostate and breast cancer [8]. Thorium-227 can be emitters. Alpha-emitters also minimize the toxicity associ- complexed by DOTA and other ligands and linked to mono- ated with radiation therapy by focusing a more localized clonal antibodies for cell-targeted cancer therapy. Our Labo- energy distribution pattern within targeted tissues. By com- ratory is also investigating the use of new nanoparticle con- parison, radiation from beta/gamma emitters extends greater structs for 227Th and 223Ra. distances from targeted lesions, and to a greater extent, can 223 damage normal organs and tissues--thereby limiting the Acting as a calcium mimic, Ra-chloride targets new amount of activity that may be administered. Alpha emitters bone growth in and around skeletal metastases [9]. The are effective in cell-killing at low dose rates and in low tis- short-range alpha particles destroy nearby cancer cells but do sue-oxygen environments characteristic of cancerous not uniformly traverse pockets of active red marrow in trabe- growths [5]. These properties establish a scientific rationale cular bone. A treatment involves intravenous infusion of 50 for identifying and employing alpha-emitting radionuclides, KBq per kg body weight, once per month, for six months; having suitable chemical and physical properties, for appli- this protracted infusion scheme provides maximum treatment cations in cancer treatment. benefit without exceeding normal tissue (bone marrow) toxicity. More patients have been treated with 223Ra than with The chemical properties of alpha emitters must be such any other alpha-emitter during clinical trials for Al- that they can be administered to cancer patients using bifunc- pharadin™ [10]. About 3000 patients have been treated in 20

Table 1. Physical and Chemical Properties of thorium-227 and radium-223

Properties 227Th 223Ra Advantages

Physical half-life 18.7 days 11.4 days Shipping over great distances

Total decay energy 34 MeV 28 MeV High dose to target tissues (5 alphas) (4 alphas)

Imageable photons Lines at 84, 154, and 269 keV from 223Ra Dosimetry and treatment planning

Availability and production of Single neutron capture on 226Ra in a nuclear reactor, followed by radio- Lower cost and better availability parent 227Ac chemical separation and purification of 227Ac, from which both 227Th and than other alpha-emitters proposed 223Ra may be obtained for radioimmunotherapy

Conjugation to radiopharmaceu- Binding to chelating agents (such as Administered as the salt or com- Multiple applications in cancer ticals DOTA) plexed with nanoparticles treatment 246 Current Radiopharmaceuticals, 2012, Vol. 5, No. 3 Soderquist et al. countries, including nine medical centers in the U.S. and Recovery of Actinium from the Neutron Sources several others in Europe. Patient imaging shows the accumu- 223 This section describes the method used to open, recover, lation of activity in the intestines after clearance of Ra 227 223 and purify Ac present in two neutron sources. Source de- from the blood, indicating that simple cationic Ra not scriptions, chemical forms, and capsule drawing were not bound to skeleton clears the patient via the small intestine 227 into bowel contents [11] and is excreted, which represents a available for either orphan source. The two Ac/beryllium neutron sources were first examined, weighed, photo- distinct advantage over other radionuclides that clear through graphed, and radiographed to ascertain their construction. the renal pathway and subject kidney tissues to localized, One capsule was clean, one was rusted, and both sources high-dose radiation. were welded. The x-ray images showed no details inside In contrast to the use of 223Ra chloride, the delivery of because the images were distorted by the high gamma radia- 223Ra to cancer cells in radioimmunotherapy for treating soft- tion dose from 227Ac and its decay-chain daughters. The den- tissue malignancies requires a suitable bi-functional chelat- sity of the radioactive sources was about 7 grams per cubic ing agent. Ionizable calixarene-crown ether ionophores were centimeter, implying that the sources were solid steel tightly found to exhibit high selectivity for radium over lighter alka- enclosing the 227Ac/Be source material. line earth metal ions [7]. In addition, calix[4]arene-crown-6 The intact neutron sources were transferred into a hot cell exhibited high kinetic stability for radium in the presence of competing, serum-abundant metal ions, including sodium, and were mounted in a metallurgical saw. The sources were cut open by successively taking slices off the end, so that the potassium, magnesium, calcium, and zinc [7]. The combina- saw blade would not cut into the beryllium inside. The tion of high selectivity and high binding constant predicted a sources were constructed of welded cylinders inside welded useful chelating agent for radium. However, further work is cylinders. One source was triply encapsulated; the other was needed to develop water-soluble bifunctional chelating quadruply encapsulated. agents having ionizable calixarene-crown moieties at the lower rim as a radium ionophore, together with an isothiocy- The first 227Ac/Be source contained two pellets in the anate group at the upper rim for linkage to monoclonal anti- core that were too small to mount in the metallurgical saw bodies. Current research at our Laboratory focuses on alter- using hot-cell manipulators. The two pellets weighed 6.7 native radium-binding inorganic nanoparticles employing grams. Instead of cutting these pellets, we dissolved them in bifunctional properties. Such constructs should enable fur- a sulfuric-nitric acid mixture. The mass of the iron (together ther development of 223Ra-labeled antibodies. with chromium and nickel) was small enough to handle in 223 ordinary analytical operations and laboratory equipment. The Thorium-227 has been considered an alternative to Ra 227 two pellets dissolved slowly at first, providing a dark, green- in various cancer-treatment applications because Th can colored solution. After the outer layer was breached, the ma- be linked to monoclonal antibodies using bifunctional terial inside dissolved within minutes, effervescing furiously. DOTA chelates. Thorium-227-labeled antibodies decay in 223 The beryllium inside these two pellets was metal, rather than situ to Ra and decay-chain daughters. Thorium-227- labeled antibodies have been considered for targeting and the oxide form. The two pellets were completely dissolved within one hour, leaving no shell. Sulfuric acid with nitric treating a variety of cancer types, such as non-Hodgkin’s acid will readily dissolve titanium, chromium, manganese, lymphoma [12] breast cancer and ovarian cancer [13]. Tho- iron, nickel, cobalt, zinc, copper, the lanthanides, and actin- rium-227-polyphosphonate compounds have been proposed ium without forming precipitates. Beryllium sulfate is highly [14] as bone-seeking agents for treating skeletal metastases soluble. The green solution looked like and behaved like from breast and prostate cancer. dissolved stainless steel. RADIOCHEMISTRY When we opened the second source, the final cut went through the end of the innermost capsule and grazed the con- Actinium-227 Recovery for Beneficial Re-use tents, exposing gray beryllium metal inside. The innermost capsule, with one end cut off, was added to the green-colored As an alternative to waste disposal, and on behalf of a sulfuric acid solution obtained from the first neutron source. small isotope company (AlphaMed, Inc., Acton, Massachu- The beryllium metal bubbled and dissolved in a few minutes. setts), Pacific Northwest National Laboratory (Richland, 227 We weighed the capsule before and after the beryllium metal Washington) acquired two excess Ac/Be neutron sources dissolved, and the difference was 0.782 g. from a U.S. oil company to recover 227Ac for beneficial re- use. The neutron sources were encased in multiple layers of Separation of Crude Actinium steel. The origin and pedigree of the sources were unavail- able and unknown. The source activities were expected a When water was added water to dilute the sulfuric acid to priori to contain about 44 GBq and 5.5 GBq, respectively, a larger volume, a heavy white precipitate formed. The green but lacking documentation, no specific information was solution was decanted, leaving the white precipitate. The white precipitate exhibited a high gamma dose-rate, indicat- available concerning method of encapsulation, encasement ing presence of an undetermined radioactive species. alloy materials or thicknesses, weld materials, or chemical forms of the 227Ac and beryllium contents. Stainless steel and any other alloys of iron, chromium, manganese, cobalt, and nickel will not form precipitates in In various calculations that follow in this paper, we have dilute sulfuric acid. However, lead and silver have sparingly assumed that the total inventory of actinium in the two neu- soluble white sulfates. If the pellets had been soldered with tron sources was about 50 GBq. However, we did not accu- silver solder, then a white precipitate would form when we rately assay the actinium in this phase of the work. dissolved the pellets in sulfuric acid. The presence of lead Production of High-purity Radium-223 from Legacy Actinium-Beryllium Current Radiopharmaceuticals, 2012, Vol. 5, No. 3 247 and silver would also explain the high dose-rate, since lead sulfate coprecipitates radium [15]. Kohltoff and Elving [16] reported that lead sulfate will coprecipitate actinium. Silver Actinium-227 plus decay and lead sulfates can be dissolved by conversion to acid- products, sodium, iron, soluble oxides and hydroxides by metathesis with strong and other contaminants base, then dissolving the product in an acid. Silver converts Load onto anion exchange resin to dark brown silver oxide, and lead converts to a white hy- in 2M HNO3 + 70% CH3OH droxide. With excess sodium hydroxide, lead dissolves as passes through loads onto column sodium plumbite. The sulfate anion goes into solution. As- suming that the precipitate was lead and silver sulfates, we Ac, Th, Ra, Pb, Bi warmed the precipitate with dilute sodium hydroxide solution. The solids turned a dark brown color. We assumed that Wash column with 80%CH3OH + 1M HNO3 the following chemical reactions occurred: Na+ 2+ 2+ 223 Ag2SO4 + 2NaOH Ag2O + Na2SO4 + H2O Fe (column volumes 3-10) Elute Ra Ra Al3+ (Ag2SO4 is white and Ag2O is dark brown) Po+4 Wash column PbSO4 + 2NaOH Pb(OH)2 + Na2SO4 + H2O other with 8M HNO3 (with dilute NaOH only) Elute Ac3+ 227Ac

The dilute solution was filtered, and the solids and the fil- Wash column trate were acidified with dilute nitric acid. The dark-brown with 0.5 M HCl solids dissolved on contact in dilute nitric acid. The first treatment with sodium hydroxide produced a small amount Elute Th4+ 227Th of white sulfates, and the treatment was repeated to remove the white sulfates. Fig. (2). Process flow-chart for actinium, radium, and thorium sepa- To remove the lead and silver from these two solutions ration and purification by anion-exchange in methanol-water-nitric- without removing actinium, we bubbled hydrogen sulfide acid solution. gas through the solution. A black sulfide precipitate formed perature to dissolve the aluminum, convert the sulfates to (Ag S and PbS), which we filtered out of solution. The black 2 acid-soluble compounds, and convert the insoluble fluorides sulfide precipitate had low activity after several days, and 227 to acid-soluble hydroxides. Thorium hydroxide is highly was discarded. We recovered Ac from the dark-green sul- insoluble under these conditions, and carries actinium. This furic acid solution and the two filtrates after removing silver solution was filtered through a membrane filter. The filter and lead by coprecipitation on thorium fluoride. Each solu- cake then contained all of the thorium and actinium (accom- tion was transferred to a polyethylene jar and about 20 milli- panied by a trace of silver). The sulfate, fluoride, aluminum, grams of thorium was added to each. Several mL of concen- and lead are soluble under these conditions and pass through trated hydrofluoric acid were added to each one, and the so- the filter with the filtrate. lution was allowed to stand for the precipitate to form. The ThF4 precipitates were filtered out of solution on cellulose Thorium carrier was removed by anion exchange. The nitrate-acetate membrane filters. This precipitation separates filter cake of thorium and actinium hydroxides was dissolved actinium from beryllium, since beryllium is soluble in hydro- in 8 M nitric acid and was passed through an anion exchange fluoric acid solution. The green color (presumably iron, column. Thorium loads onto the resin, but actinium does not. nickel, and chromium) stayed in solution and was separated The column effluent with actinium was collected and evapo- from the actinium. rated dry to yield a crude actinium nitrate. The crude actinium nitrate appeared as a white crust of Actinium Purification crystalline matter evaporated onto the bottom of a beaker. The crude actinium, coprecipitated with thorium fluoride We expected it to be contaminated with thorium from the on several membrane filters, was taken to a smaller, cleaner thorium carrier, incompletely removed by anion exchange, hot cell for final purification and production of 223Ra. The and with aluminum, iron, nickel, lead, silver, sulfate, and filters holding thorium fluoride cakes were placed in a fluoride. These contaminants are separated from the actinium beaker with aluminum nitrate and nitric acid to dissolve the by anion exchange in methanol-water-nitric acid. Sulfate was thorium fluoride. Aluminum nitrate-nitric acid solutions dis- reported to not interfere [36], but the effect of fluoride in the solve thorium fluoride, but failed to dissolve the activity on anion exchange is unknown. The aluminum would presuma- the filters. The remaining insoluble matter had a high dose- bly take up the free fluoride and prevent it from interfering rate, possibly because some of the white sulfate precipitate with actinium loading, if enough aluminum happened to be had been decanted with the green-colored solution. present. Thorium would likewise scavenge fluoride. The solution with the filters and insoluble debris were re- To produce a chemically pure actinium source, we sepa- combined in a beaker, evaporated to near-dryness, and boiled rated actinium by anion exchange in a methanol-water-nitric with concentrated nitric acid at length to destroy the filters acid solution (Fig. 2). The crude actinium nitrate was dis- and expel fluoride. The wet-ashed residue was dissolved in solved in nitric acid, evaporated to a damp residue, and taken 0.01M nitric acid, and was then changed to 1M in sodium up in 6 mL of 2M nitric acid in 80% methanol. This solution hydroxide. The solution was warmed to near-boiling tem- was loaded onto a column of AG1-X8, 100-200 mesh, nitrate 248 Current Radiopharmaceuticals, 2012, Vol. 5, No. 3 Soderquist et al.

activity in the exhaust stack monitor. The activity appeared about 30 minutes after evaporations started, and decayed with a 30-minute half life. The 219Rn has a half-life of only 4 seconds and would be expected to completely decay in the minute or so transit time for air to move from the hot cell to the HEPA filter banks. The 30-minute delay and 30-minute half life indicated that the activity was 211Pb and 211Bi.

Measurement of Actinium Actinium mass and activity are difficult to measure-- particularly in disequilibrium with its decay chain. The beta emission from 227Ac has a low (45 keV) end-point energy and cannot be measured in the presence of alphas and more energetic betas of the decay chain. Actinium-227 has low abundance gammas that are impossible to measure in the presence of the hundreds of gamma peaks from the other Fig. (3). Bottom of beaker (top-down view) containing 40 mg ultra- members of the decay chain. It is possible, though not easy, pure actinium nitrate (56 GBq). This quantity evaporated to dryness to measure 227Ac by measuring the gamma emission twice represents a substantial fraction of the world supply of purified and calculating 227Ac by ingrowth of the 227Th. Many of the actinium-227, perhaps one of the rarest and most valuable of all gamma abundances for 227Ac, 227Th, and 223Fr are not accu- chemical radioelements. rately known. Our stock of actinium nitrate amounted to tens of milligrams and could be weighed, but the amount of form anion resin, 10 mL column volume, in 2M nitric acid, bound water was unknown. A small amount of actinium in 80% methanol. After the load solution had passed through the presence of a large amount of 227Th and 223Ra can be the column, a wash solution of 1M nitric acid, 70% methanol measured by chemically separating the actinium, such as by was added in 1 column volume (10 mL) increments. Each solvent extraction or ion exchange. After separation, the column volume of effluent was collected separately and 227Ac can be measured by gamma by ingrowth of 227Th. evaporated dry for storage. The column developed gas bubbles during the separation, possibly caused by radiolysis 223 227 from the high activity. The bubbles slowed, but did not stop Radiochemical Separation of Ra from Ac column flow. After 14 column volumes of wash had passed Direct production of 223Ra from 227Ac requires that 227Th through, a clean beaker was placed under the column, and be left in equilibrium with 227Ac, because the ingrowth time actinium was stripped with 80 mL of 8M nitric acid. Tho- is about 50% longer if one waits for 223Ra ingrowth after rium was then stripped with 60 mL of 0.5M hydrochloric 227Th ingrowth. Various methods for radiochemical separa- acid. tions of 223Ra and 227Th have been published [17-20]. Prepa- 223 227 227 The separated actinium evaporated down to a thin white ration of high purity Ra and Th from Ac is chemically straightforward, but is mechanically complex due to crust of Ac(NO3)3 (presumably 6-hydrate) on the bottom of the beaker. Fig. (3) shows actinium nitrate photographed intense alpha and gamma emission. Radiolysis produces re- through a hot-cell window. The eluted thorium evaporated active species (solvated electrons, hydroperoxide ions, down to a viscous liquid which solidified into a white mass. atomic hydrogen, and free hydroxyl) that can quickly de- compose ion exchangers and solvents and interfere with ra- The first ion exchange separation produced 15 fractions diochemical separations. Ion exchange resins degrade under that could, in theory, be used to draw an elution curve for high alpha dose. Radiochemistry must usually be performed 223Ra. However, the high dose and high airborne alpha con- 219 in shielded facilities. Radon-219 must be contained to pre- tamination from Rn prevented us from handling any of the vent general-area work place contamination. vials outside the hot cell. The elution curve would have been distorted by gas bubbles that formed in the column and by Preparation of High Purity 223Ra the chemical contaminants which were present. We evapo- 223 rated all 15 fractions and the load solution dry for storage. To generate an actual Ra elution curve, separated ac- The residue in the load vial was blue-green. The first several tinium was allowed to stand for three days to allow the de- column volumes had decreasing amounts of black residue. cay-chain products to grow in, and then we performed a sec- The last two fractions (column volumes 13 and 14) had ond ion exchange separation identical to the first separation. 227 223 white solids, which may have been the start of the 227Ac elu- After only three days, little Th and Ra had ingrown, and tion peak. the radiation dose-rate was low enough to handle the radionuclides outside the hot cell. No gas bubbles formed in the The dried vials were capped and placed into a steel can ion exchange column. Figure 4 shows the elution curve with a tight fitting lid to prevent escape of radon, to allow measured by gamma spectrometry 58 days after the ion ex- 223Ra to decay off for several weeks. After the 223Ra had 227 change separation. Actinium-227 was not directly measur- mostly decayed off, we measured Ac in the fractions by able because its gammas are too weak in the presence of chemical separation. other gammas associated with the actinium decay chain. In- When we evaporated the methanol-water-nitric acid solu- stead, we measured 227Ac by ingrowth of 227Th. Since tho- tions from the ion exchange run, the 219Rn gas that evolved rium does not come off the column under the conditions during the evaporation caused measurable alpha and beta used, 227Th in the final product grew in from 227Ac. The 223Ra Production of High-purity Radium-223 from Legacy Actinium-Beryllium Current Radiopharmaceuticals, 2012, Vol. 5, No. 3 249

Fig. (4). Elution curve for nitrate-methanol anion exchange 58 days after separation (activity in MBq versus column volume eluted). This figure shows the number of column volumes (3 to 10) required to remove 223Ra from the column before 227Ac starts to elute (at column volume 16). The elution curve shows a constant, low-level of ingrowth 223Ra from 223Fr over all column volumes. peak is much smaller than the 227Ac peak (most of which lies Analysis of Actinium in Elutions off the right side of the elution curve), partly because the Elutions from each ion exchange separation were ana- ingrowth time was short before the ion exchange separation, 223 lyzed for actinium by extraction into thenoyltrifluoroacetone and because the Ra decayed through five half lives before (TTA) in xylene [21]. The dried effluents from the ion ex- counting. change fractions were dissolved in dilute nitric acid, com- Radium appeared between column volumes 3 and 10, bined to make fewer samples, and adjusted to pH 2. Each centered about column volume 6. Actinium began to appear sample was shaken with 0.25M TTA in xylene to extract in column volume 13, but the actinium peak was centered to thorium, and the xylene layer was then removed. The pH the right of column volume 20, off the chart and far from the was readjusted to 6.0 and fresh 0.25M TTA-xylene was 223Ra peak. If the trailing tail of the 223Ra were rejected, then added to extract actinium. The xylene layer was transferred pure 223Ra could be produced in good yield. The 223Ra was to a clean vessel and back-extracted with 0.1M nitric acid to free of 227Th and 227Ac. recover the actinium. The actinium extract was removed 227 223 from the hot cell for gamma counting. The Ac activity was The elution curve shows the Ra after five half-lives of 227 223 measured by ingrowth of Th. decay. The Ra peak area decay-corrected to the separation 227 date was 662 MBq. The leading tail of the Ac peak shown In the first ion exchange (to purify the crude Ac(NO3)3), in the plot (column volumes 13 to 19) represented 38.8 MBq we measured 2690 MBq 227Ac in the load solution (about 5% 227Ac, or about 0.09% of the total actinium. of the assumed actinium inventory), and 5610 MBq 227Ac in 223 column volumes 1 through 14, combined (about 11% of the The elution curve has a constant, low level of Ra assumed actinium inventory). Column volumes 13 and 14 (which is too small to show on this chart) from decay of 223 227 223 had white solids, which may have been the beginning of the Fr. About 1.4% of the Ac decays by alpha to Fr, actinium elution peak. We stopped the separation at column which has a 22-minute half-life, and which will grow in dur- 223 volume 14. Chemical contaminants in the column may have ing the ion exchange separation. The Fr decays by beta- caused the actinium to elute early. emission to 223Ra. In the TTA extraction of the second ion exchange prod- Radon gas quickly diffuses through plastics. Dried frac- uct, we obtained 0.0121 MBq 227Ac in column volumes 1 tions from the ion exchange run were stored in glass liquid through 10 (223Ra peak), and 27.1 MBq 227Ac in column vol- scintillation vials with polyethylene liners in polyvinyl chlo- umes 11 through 19 (227Ac peak). The second ion exchange ride caps. Radon will diffuse through the caps and cause produced highly pure 223Ra. The 0.0121 MBq of 227Ac found alpha contamination on the outside of the vials. Our standard in the 223Ra peak (column volumes 1 through 10) may have practice for gamma counting is to tape each vial around the been hot cell contamination. The TTA extraction yielded an cap with vinyl tape, place the vial in a small polyethylene incomplete actinium recovery. The integrated area of the bag, and tape the bag shut. When the 223Ra fractions were second peak (the 227Ac peak, column volumes 11 to 19) was counted this way, alpha activity appeared on the outside of 38.8 MBq 227Ac, but the TTA extraction recovered only 27 227 the glass, under the bag, within about a day. Counting vials MBq Ac. are stored in a steel slip-lid can. After storing these vials for a few days, alpha activity appeared on the inside of the steel Separation and Recovery of Thorium 219 can. Since Rn has only a 4-second half-life, but appeared The anion exchange separation that we used to produce outside the vial, its diffusion rate through polyethylene was 223Ra can also be used to produce 227Th, with small adjust- rapid. ments. If one omits methanol and adjusts the nitric acid con- 250 Current Radiopharmaceuticals, 2012, Vol. 5, No. 3 Soderquist et al. centration to 7.5M, then thorium will load onto the anion moved from the separated product (such as by wet-ashing in exchanger and will allow actinium and radium to pass concentrated nitric acid). Anion exchange in methanol with a through. We used this chemistry to remove thorium carrier low concentration of water and nitric acid gives a product from the actinium stock. This separation can also be used to that evaporates down to carrier-free, essentially massless, separate pure 227Th. The pure actinium can be dissolved in trace-level radionuclides. The concentration of nitric acid 7.5M nitric acid and passed through an anion exchanger as used in these separations is too low to oxidize the alcohol. described above. Thorium-227 alone loads onto the resin. The eluted product evaporates quietly dry without oxidation, The 227Ac accompanied by the remainder of the decay chain even boiling on a hot plate. appears in the column effluent. Thorium is eluted with 0.5M Investigators have tabulated distribution coefficients for HCl. The thorium separation can be repeated to generate 227 227 metals in alcohol-nitric acid solutions [45, 48-51]. Thorium high-purity Th, free of parent Ac. and the other tetravalent actinides have the highest distribution coefficients, followed by the trivalent actinides and the Anion Exchange in Methanol-Water-Nitric Acid early lanthanides. Lead and bismuth have distribution coeffi- Anion exchange in nitrate solution represents a versatile cients comparable to the early lanthanides. Other elements, separation for the light lanthanides and many of the acti- including aluminum, iron, cobalt, nickel, copper, zinc, and nides. This separation retains actinium and thorium on the silver, have much lower distribution coefficients, and can be column, but allows radium to be eluted. In this method, the removed from the column by washing. Most authors do not actinium/thorium sample is dissolved in methanol with 20% mention actinium, with one exception [52]. We found that water and 2M in nitric acid. The sample is them loaded onto actinium behaves like the early lanthanides. Guseva et al. 223 a nitrate-form anion exchanger. The tetravalent actinides [52] used an ion exchange separation to construct a Ra (including thorium) load firmly onto the resin as anionic generator, similar to the process that we describe in the fol- nitrate complexes and do not move down the column. Actin- lowing section. ium and the light lanthanides also load, but not as tightly, The heavier alkaline earth elements also load onto the and move slowly down the column. Most other elements, column, but not as well as the early lanthanides. The order of including radium, pass through the column faster than actin- distribution coefficients is [Be, Mg] << Ca < Sr < Ba [51]. ium and the light lanthanides. The anion exchange method is This property has been used to separate magnesium from a convenient way to chromatographically separate the light calcium [53], and calcium from strontium [54]. We found lanthanides. One advantage of this method is that the load that radium elutes well before actinium. solution and eluting agent are completely volatile and leave no residue. Quality Assurance and Quality Control Soon after commercial anion exchange resins became Quality assurance and quality control are essential to en- available, thorium was observed to load from aqueous nitric suring delivery of safe radioisotope products for human use acid [22-26]. Other metals, particularly the lanthanides, were and for compounding or radiolabeling as the products are found to load onto an anion exchanger from a weakly acidic solution of a nitrate salt such as lithium nitrate or aluminum developing into human-use radiopharmaceuticals. The proc- nitrate [27-34]. A number of chemists developed chroma- ess chemistry described in this paper has not been reduced to tographic separations that used a nitrate salt solution. Danon practice in the form of a shielded, compact generator system for use by radiopharmacists. The relatively long half-lives of [28] used anion exchange in LiNO3 solution to separate ac- 227 223 tinium from lanthanum. The disadvantage of the lithium ni- Th (18.7 days) and Ra (11.4 days) argue for direct dis- trate method is the high concentration of nitrate salts in the tribution of these two radionuclides from centralized labora- eluted product. Other investigators found that metals load tories rather than from distributed compact generator sys- more tightly to the anion exchanger if a water-miscible or- tems. This paper shows that the process chemistry is straight- ganic solvent is added. Thorium was found to load more forward, but that the mechanical details and steps involved tightly from a light alcohol-nitric acid solution than from are more complex than those involved in simple generator aqueous nitric acid alone [35, 36]. Uranium was also found elution. A standard generator based on our process chemistry 219 to load more tightly to the resin if an alcohol is added [37]. would require Rn control, a means for removing solvent Metals frequently load more strongly from a mineral acid from the product, and a means for storing 227Ac parent be- mixed with any of many organic solvents [38-42]. The tween generator milkings. choice of organic solvent makes only a minor difference, and No industry standard for 227Ac breakthrough has been de- results are similar using any light alcohol or acetone. scribed in the literature. Actinium contamination in the 223Ra The lanthanides, especially the lighter elements, load product cannot be measured directly because the 223Ra gam- onto an anion exchanger from an ethanol- or methanol-nitric mas interfere with the measurement. However, one may con- acid solution with useful differences in distribution coeffi- trol product purity and maintain a very low level of 227Ac in cient from one lanthanide to the next [36, 43, 44]. The alco- the 223Ra product. To measure actinium, radium first must be hol-nitric acid solution permits convenient chromatographic removed. In the event that one were to find 227Ac in excess separation of the early lanthanides [45-47]. These separations of one part per million (radioactivity), an additional an ion- perform as well as other chromatographic separations for the exchange separation step would be performed. Conditions lanthanides, but use only volatile reagents. Other common for radium-actinium separation were selected to place the chromatographic methods for separating the lanthanides use radium and actinium elution peaks apart, so that the tails of cation exchange with non-volatile complexing agents such as the peaks had negligible overlap to prevent contamination of EDTA, citrate, or -hydroxyisobutrate, which must be re- 223Ra with parent actinium. The final ion exchange separa- Production of High-purity Radium-223 from Legacy Actinium-Beryllium Current Radiopharmaceuticals, 2012, Vol. 5, No. 3 251 tion is relatively fast and can be repeated to remove 227Ac to [4] Wilbur, D.S.; Potential use of alpha-emitting radionuclides in the non-detectable levels. Chemical contaminants in the final treatment of cancer. Antibody Immunoconj. Radiopharm., 1991, 4, 85-97. radium product are best controlled using high-purity reagents [5] Sgouros, G.; Song, H.; Cancer stem cell targeting using the alpha- and non-contaminating vessels, such as quartz glassware. particle emitter, 213Bi: mathematical modeling and feasibility analy- Thorium-227 and 223Ra produced by the process described in sis. Cancer Biother. Radiopharm., 2008, 23(1), 74-81. this paper were free of other radionuclides (other than natural [6] Jurcic, J.G.; Larson, S.M.; Sgouros G; McDevitt, M.R.; Finn, R.D.; decay products) and non-radioactive contaminants. Divgi, C.R.; Ballangrud, A.M.; Hamacher, K.A.; Ma, D.; Humm, J.L.; Brechbiel, M.W.; Molinet, R.; Scheinberg, D.A.; Targeted alpha particle immunotherapy for myeloid leukemia. Blood, 2002, 100(4), 1233-1299. SUMMARY AND CONCLUSIONS [7] Chen, X.; Ji, M.; Fisher, D. R.; Wai, C. M.; Ionizable calixarene- 223 227 crown ethers with high selectivity for radium over light alkaline The improved availability of Ra and Th will enable earth metal ions. Inorg. Chem., 1999, 38(23), 5449-5452. research to develop new radiopharmaceuticals using these [8] Liepe, K.; Alpharadin, a 223Ra-based alpha-particle-emitting radionuclides. On this project, we recovered, isolated, and pharmaceutical for the treatment of bone metastases in patients purified approximately 50 GBq from 227Ac from two legacy with cancer. Curr. Opin. Investig. Drugs, 2009, 10(12), 1346-1358. actinium-beryllium neutron sources. [9] Bruland, Ø. S.; Nilsson, S.; Fisher, D. R.; ; Larsen, R.H.; High- linear energy transfer irradiation targeted to skeletal metastases by An ion exchange in methanol-water-nitric-acid method the alpha-emitter 223Ra: adjuvant or alternative to conventional mo- was demonstrated for fast, carrier-free separation of actin- dalities? Clin. Cancer Res., 2006, 12(20 Pt 2), 6250s-6257s. [10] Nilsson, S.; Franzén, L.; Parker, C.; Tyrrell, C.; Blom, R.; Len- ium, thorium, and radium. This anion-exchange method uses nemäs, B.; Pettersson, U.; Johannessen, D.C.; Sokai, M.; Pigott, K.; 223 completely volatile eluting agents so that carrier-free Ra Yachnin, J.; Garkavij, M.; Strang, P.; Harmenberg, J.; Bolstad, B.; 227 227 and Th can be readily separated from parent Ac. A high- Bruland, Ø.S.; Bone-targeted radium-223 in syhmptomatic, hor- purity, carrier-free 223Ra may be separated from the parent mone-refractory prostate cancer: a randomized, multicenter, pla- cebo-controlled phase II study. Lancet Oncol., 2007, 8(7), 587-594. 227Ac by anion exchange with little contamination from 227Th 227 [11] Abbas, N.; Heyerdahl, H.; Bruland, Ø.S.; Borrebæk, J.; Nesland, J.; or Ac. This separation constitutes an economical, single- and Dahle, J.; Experimental -particle radioimmunotherapy of pass 223Ra production method that does not disturb the breast cancer using 227Th-labeled p-benzyl-DOTA-trastuzmab. 227Th/227Ac equilibrium. After 223Ra separation, the 227Ac can EJNMMI Res., 2011, 1(1),18. then be eluted off the anion resin to avoid problems from [12] Melhus, K.B.; Larsen, R.H.; Stokke, T.; Kaalhus, O.;Selbo, P.K.; Dahle, J.; Evaluation of the binding of radiolabeled rituximab to high radiation dose to the resin. CD20-positive lymphoma cells: an in vitro feasibility study concerning low-dose-rate radioimmunotherapy with the alpha-emitter Methanol-water-nitric-acid anion-exchange uses gravity- 227 Th. Cancer Biother. Radiopharm., 2007, 22(4), 469-479. feed columns but no mechanical pumps or other highly spe- [13] Heyerdahl, H.; Krogh, C.; Borrebæk, J.; Larsen, Å.; Dahle, J.; cialized equipment. This approach is relatively simple and Treatment of HER2-expressing breast cancer and ovarian cancer less costly than other typical ion-exchange methods. How- cells with alpha particle-emitting 227Th-trastuzumab. Int. J. Radiat. ever, radon-219 control is essential to reduce the spread of Onco.l Biol. Phys., 2011, 79(2), 563-570. [14] Henriksen, G.; Bruland, Ø.S.; Larsen, R.H.; Thorium and actinium alpha contamination in the laboratory. High-purity 227Th may 227 polyphosphonate compounds as bone-seeking alpha particle- also be prepared by this method. Radium-223 and Th are emitting agents. Anticancer Res., 2004, 24(1), 101-105. now available from our Laboratory when purchased through [15] Kolthoff, I. M.; Elving, P. J.; editors, Treatise on Analytical Chem- the Department of Energy’s Isotope Program. istry, Interscience Publishers: New York, 1966, Vol. 4, p. 293. [16] Kolthoff, I. M.; Elving, P. J.; editors, Treatise on Analytical Chem- istry, Interscience Publishers: New York, 1964, Vol. 6, p. 455. [17] Boll, R.A.; Malkemus, D.; Mirzadeh, S.; Production of actinium- ACKNOWLEDGEMENTS 225 for alpha particle mediated radioimmunotherapy. Appl. Radiat. This research was supported by the U.S. Department of Isot., 2005, 62(5), 667-679. [18] Henricksen, G.; Hoff, P.; Alstad J.; Larsen, R.H.; 223Ra for endora- Energy, Office of Science, Office of Nuclear Physics, Iso- diotherapeutic applications prepared from an immobilized tope Development and Production for Research Applications 227Ac/227Th source. Radiochim. Acta, 2001, 89,661-666. Program under Contract DE-AC05-76RL01830. [19] Kirby, H.W.; Residue adsorption - III: Mutual separation of 227Ac,. 227Th and 223Ra. J. Inorg. Nucl. Chem, 1969, 31(11), 3375-3385. [20] Horowitz, E.E.; Bond, A. H.; Purification of radionuclides for CONFLICT OF INTEREST nuclear medicine: The multicolumn selectivity inversion generator concept. Czech. J. Phys., 2003, 53(Suppl 1, Part 2), A713-A716. Declared none. [21] Kolthoff, I. M.; Elving, P. J.; editors, Treatise on Analytical Chem- istry, Interscience Publishers: New York, 1964, Vol. 6, p. 459. [22] Danon, J.; Adsorption of thorium by anion-exchange resins from REFERENCES nitric acid media. J. Am. Chem. Soc., 1956, 78(22), 5953-5954. [23] Carswell, D.J.; Separation of thorium and uranium nitrates by anion [1] Fisher, D. R.; Commercial availability of alpha-emitting radionu- exchange. J. Inorg. Chem., 1957, 3, 384-387. clides for medicine. Current Radiopharm., 2008, 1(3), 127-134. [24] Danon, J.; Separation of thorium and rare-earth elements in nitric [2] Möller, T.; Bestaoui, N.; Wierzbicki, M.; Adams, T.; Clearfield A.; acid media by anion exchange. J. Inorg. Nucl. Chem., 1958, 5, 237- Separation of lanthanum, hafnium, barium and radiotracers yt- 239. trium-88 and barium-133 using crystalline zirconium phosphate [25] Bunney, L.E.; Ballou, N.E.; Pascual, J.; Foti, S.; Anion exchange and phosphonate compounds as prospective materials for a Ra-223 behavior of several metal ions in hydrochloric, nitric, and sulfuric radioisotope generator. Appl. Radiat. Isot., 2011, 69(7), 947-954. acid solutions. Anal. Chem., 1959, 31(3), 324-326. [3] Sgouros G.; Roeske, J. C.; McDevitt, M. R.; Palm, S.; Allen, B. J; [26] Fritz, J.S.; Garralda, B.B.; Anion exchange separation of thorium Fisher, D. R.; Brill, A. B.; Song H.; Howell, R. W.; Akabani, G.; using nitric acid. Anal. Chem., 1962 34(11), 1387-1389. MIRD Pamphlet No. 22 (abridged): radiobiology and dosimetry of [27] Ockenden, H.M.; Foreman, J.K; The separation of uranium from alpha-particle emitters for targeted radionuclide therapy. J. Nucl. large amounts of iron and aluminium by nion exchange in nitrate Med., 2010, 51(2), 311-328. media, Analyst, 1957, 82, 592-593. 252 Current Radiopharmaceuticals, 2012, Vol. 5, No. 3 Soderquist et al.

[28] Danon, J.; Anion exchange studies with actinium and lanthanides [42] Alstad, J.; Brunfelt, A.O.; Adsorption of the rare-earth elements on in nitrate solutions. J. Inorg. Nucl. Chem., 1958, 7, 422-424. an anion-exchange resin from nitric acid-acetone mixtures. Analt- [29] Foreman, J. K.; McGowan, I. R.; Smith, T. D.; Some aspects of the tica Chimica Acta, 1967 , 38, 185-192. anion-exchange behavior of uranyl nitrate in the presence of other [43] Edge, R.A.; The anion exchange behavior of yttrium, neodymium inorganic nitrates. J. Chem. Soc. (Resumed), 1959, 738-742. and lanthanum in dilute nitric acid solutions containing ethanol. J. [30] Marcus, Y.; Nelson, F; Anion-exchange studies. XXV. The rare Chromatog., 1961, 5, 526-530. earths in nitrate solutions. J. Physic. Chem., 1959, 63, 77-79. [44] Edge, R.A.; The removal of Ce, Nd, Pr, and La from Er, Dy, Gd, [31] Edge, R.A.; The adsorption of thorium on a strong base anion ex- Eu, and Sm by anion exchange. Analytical Chimica Acta, 1963, 29, change resin from nitrate media. J. Chromatog., 1961, 6, 276-277. 321-324. [32] Marcus, Y.; Abrahamer, I.; Anion exchange of metal complexes – [45] Roelandts, I.; Duyckaerts, G.; Anion-exchange isolation of rare- VII. The lanthanides-nitrate system. J. Inorg. Nucl. Chem., 1961, earth elements from apatite minerals in methanol-nitric acid 22,141-150. medium. Analytica Chimica Acta, 1974, 73, 141-148. [33] Adar. S.; Sjoblom, R.K.; Barnes, R.F.; Fields, P.R.; Hulet, E.K.; [46] Strelow, F.W.E.; Quantitative separation of samarium from Wilson, H.D.; Ion-exchange behaviour of the transuranium neodymium by anion exchange chromatography in dilute nitric elements in LiNO3 solutions. J. Inorg. Nucl. Chem., 1963, 25, 447- acid-methanol. Anal. Chem., 1980, 52(14), 2420-2422. 452. [47] ASTM International. Standard Test Method for Atom Percent Fis- [34] Marcus, Y.; Givon, M.; Choppin, G.R.; Anion exchange of metal sion in Uranium and Plutonium Fuel (Neodymium-148 Method); complexes –XIII. The actinide(III)-nitrate system. J. Inorg. Nucl. ASTM Method E321-96. ASTM International: West Consho- Chem., 1963, 25, 1457-1463. hocken, Pennsylvania: 2005. [35] Tera, F.; Korkisch, J.; Hecht, F.; Ion exchange in mixed solvents – [48] Faris, J.P.; Warton, J.W.; Anion exchange resin separation of the IV. The distribution of thorium between alcohol-nitric acid solu- rare earths, yttrium, and scandium in nitric acid-methanol solutions. tions and the strongly-based anion exchanger Dowex-1. Separation Anal. Chem., 1962, 34(9), 1077-1080. of thorium from uranium. J. Inorg. Nucl. Chem., 1961, 16, 345- [49] Korkisch, J.; Hazan, I.; Arrhenius, G.; Ion exchange in mixed 349. solvents. Adsorption behaviour of the rare earths and some other [36] Korkisch, J.; Tera, F.; Separation of thorium by anion exchange. elements on a strong-base anion-exchange resin from nitric acid- Anal. Chem., 1961, 33(9), 1264-1266. alcohol media. Talanta, 1963, 10, 865-877. [37] Korkisch, J.; Tera, F.; The distribution of hexavalent uranium be- [50] Fritz, J.S.; Greene, R.G.; Separation of rare earths from other metal tween alcohol-nitric acid solutions and the strongly basic anion ex- ions by anion exchange. Anal. Chem., 1964, 36(6), 1095-1097. changer Dowex-1. J. Chromatog., 1962, 7, 564-567. [51] Faris, J.P.; Buchanan, R.F.; Anion Exchange Characteristics of the [38] Fritz, J.S.; Pietrzyk, D. J.; Non-aqueous solvents in anion-exchange Elements in Nitric Acid and Nitrate Solutions and Application in separations. Talanta, 1961, 8, 143-162. Trace Element Analysis, 1964, ANL-6811. [39] Korkisch, J.; Janauer, G.E.; Ion exchange in mixed solvents. Ad- [52] Guseva, L.I.; Tikhomirova, G.S.; Dogadkin, N.N.; Anion-exchange sorption behaviour of uranium and thorium on strong-base anion- separation of radium from alkaline earth metals and actinides in 227 223 exchange resins from mineral acid-alcohol media. Talanta, 1962, 9, aqueous-methanol solutions of HNO3. Ac- Ra generator. Ra- 957-985. diochem., 2004, 46(1), 58-62. [40] Edge, R.A.; The adsorption of yttrium, neodymium and lanthanum [53] Fritz, J.S.; Waki, H.; Anion exchange separation of magnesium and on a strong base anion exchange resin from dilute hydrochloric acid calcium with alcohol-nitric acid. Anal. Chem., 1963, 35(8), 1079- solutions containing ethanol. J. Chromatog., 1961, 5, 539-540. 1083. [41] Korkisch, J.; Arrhenius, G.; Separation of uranium, thorium, and [54] Fritz, J.S.; Waki, H.; Garralda, B.B.; Anion exchange separation of the rare earth elements by anion exchange. Anal. Chem., 1964, calcium and strontium. Anal. Chem., 1964, 36(4), 900-903. 36,(4), 850-854.

Received: January 16, 2012 Revised: March 28, 2012 Accepted: April 12, 2012