Production, Purification and Availability of ²¹¹At: Near Term Steps Towards Global Access

Abstract

The promising characteristics of the 7.2-h radiohalogen ²¹¹At have long been recognized; including having chemical properties suitable for labeling targeting vectors ranging from small organic molecules to proteins, and the emission of only one α-particle per decay, providing greater control over off-target effects. Unfortunately, the impact of ²¹¹At within the targeted α-particle therapy domain has been constrained by its limited availability. Paradoxically, the most commonly used production method – via the ²⁰⁹Bi(α,2n)²¹¹At reaction – utilizes a widely available natural material (bismuth) as the target and straightforward cyclotron irradiation methodology. On the other hand, the most significant impediment to widespread ²¹¹At availability is the need for an accelerator capable of generating ≥28 MeV α-particles with sufficient beam intensities to make clinically relevant levels of ²¹¹At. In this review, current methodologies for the production and purification of ²¹¹At – both by the direct production route noted above and via a ²¹¹Rn generator system – will be discussed. The capabilities of cyclotrons that currently produce ²¹¹At will be summarized and the characteristics of other accelerators that could be utilized for this purpose will be described. Finally, the logistics of networks, both academic and commercial, for facilitating ²¹¹At distribution to locations remote from production sites will be addressed.

Graphical abstract

1. Introduction

The directions taken in the evolution of nuclear imaging and targeted radiotherapy have been influenced to a major degree by radionuclide availability. In the diagnostic realm, the inherent advantages of PET were recognized years before this modality became a practical reality. Although other factors certainly played a role, PET could not have had a significant impact in the clinical domain without the development of an effective infrastructure for ¹⁸F production and distribution. In analogous fashion, one might consider targeted α-particle therapy (TAT) to be the radionuclide therapy parallel of PET – an approach whose potential advantages were appreciated long before it could be applied to even early phase clinical studies. Likewise, a major impediment to the clinical translation of TAT has been the limited availability of the most promising radionuclides for this purpose.

As is the case for PET, the number of radionuclides with advantageous properties is small, perhaps five or less for TAT [1]. These can be divided into two classes – those that are sourced from natural decay chains and generating multiple α-particles per decay, exemplified by ²²⁵Ac, and those emitting a single α-particle per decay epitomized by ²¹¹At. Both ²²⁵Ac and ²¹¹At have promising properties for TAT and suffer from poor availability albeit for quite different reasons. With ²²⁵Ac, natural source material limitations would constrain its use to less than 1,000 patients per year leading to the pursuit of alternative production routes that are hindered by either radionuclidic purity (²²⁷Ac) issues or challenging targetry (²²⁶Ra) [2]. With ²¹¹At, significant hurdles also exist but they are quite different than those encountered with ²²⁵Ac. Production methods are available from abundant target material; however, only a few accelerators with the beam characteristics required for producing meaningful levels of ²¹¹At currently are available.

In this review, we shall first describe the nuclear and radiochemical properties of ²¹¹At and then discuss present and emerging approaches for the production and purification of this 7.2-h half-life α-emitter. Finally, we shall speculate on the prospects for the creation of network infrastructures for ²¹¹At supply.

2. Nuclear and Radiochemical Properties of ²¹¹At and their Implications for TAT

An important characteristic of ²¹¹At that is different than most other α-emitters of relevance to TAT is that it yields one α-particle per decay, which offers certain translational advantages including simplification of radiation dosimetry calculations for ²¹¹At-labeled TAT agents. The double-branched decay scheme for ²¹¹At is illustrated in Fig. 1. Fortuitously, about 58.2% of ²¹¹At decays occur via electron capture to ²¹¹Po, producing 77–92 keV polonium x-rays that permit counting ²¹¹At activity with conventional gamma detectors and quantification of ²¹¹At distributions in vivo by planar and SPECT imaging [3]. These x-rays allow measurement of ²¹¹At biokinetics in patients, which can be used for safety and stability monitoring, as well as organ-level assessment of radiation dosimetry of actual treatment doses. The ²¹¹Po then decays with a 0.516-s half-life by emission of a 7.45 MeV α-particle to ²⁰⁷Pb, which is stable. The second ²¹¹At decay branch (41.8%) involves emission of 5.87 MeV α-particles to 31.55-y ²⁰⁷Bi, which likewise decays (by electron capture) to ²⁰⁷Pb.

Figure 1.

Simplified ²¹¹At decay scheme illustrating double-branched pathway; by direct alpha decay to ²⁰⁷Bi and by electron capture to ²¹¹Po followed by alpha decay to ²⁰⁷Pb.

It should be noted that there are two aspects of the ²¹¹At decay scheme that potentially could be problematic – the ²¹¹Po intermediate and the long-lived ²⁰⁷Bi intermediate noted above. With regard to the first, one must consider the effects of an initial nuclear decay event on the fate of a subsequent α-particle emission. If the first decay event is also by α-emission, then the daughters would not only undergo chemical transformation but also the α-particle recoil energy would lead to escape and migration from the original decay site [4]. On the other hand, in ²¹¹At decay, the ²¹¹Po intermediate is the progeny of electron capture decay, which involves chemical transformation but insignificant daughter recoil energy. Even with worst possible case assumptions – that this chemical transformation results in instantaneous release of ²¹¹Po from the cell surface and transport by unimpeded thermal diffusion - nearly 100% of ²¹¹Po atoms should decay within two cell diameters from the original cell surface [5]. Give that electron capture results in a highly charged daughter nucleus [6], which could impede diffusion [7, 8], the diffusion distance from the original ²¹¹At decay site might be even shorter. In any case, except in the rare instance where cancer presents for treatment as single-cell distributed disease, the diffusion of the ²¹¹Po daughter from the original decay site can be ignored. And second, with regard to the long half-life of the ²⁰⁷Bi daughter, this is also not of concern because nearly 100,000 decays of ²¹¹At are needed to produce a single decay of ²⁰⁷Bi [9]. Thus, a 370 MBq (10 mCi) hypothetical patient dose of an ²¹¹At-labeled TAT agent would generate ~0.1 μCi of ²⁰⁷Bi, a level that is only 0.1% of the 100 μCi Annual Limit of Intake (ALI) recommended for ²⁰⁷Bi by the Nuclear Regulatory Commission (US Nuclear Regulatory Commission Report NRC, 10CFR. https://www.nrc.gov/reading-rm/doc-collections/cfr/part020/appb/bismuth-207.html accessed 26 May 2021).

The α-particle emission characteristics of ²¹¹At are similar to those of other TAT-candidate radionuclides [1], exhibiting a mean linear energy transfer of about 100 keV μm.⁻¹ This results in high relative biological effectiveness because they can create more than 10 ionizations in a 100 Å diameter × 3 nm column [10, 11], an ionization density close to the diameter of the DNA double helix [12]. Even though ²¹¹At emits fewer α-particles per decay than some other radionuclides under investigation for TAT, 211At-labeled targeted radiotherapeutics are exquisitely cytotoxic, with effective killing achievable with less than 10 atoms bound per cell [13]. Moreover, if uptake of ²¹¹At in the cell nucleus can be achieved, the fact that ²¹¹At also generates an average of 6.2 Auger electrons emitted per decay (comparable to ⁶⁷Ga) [14] also might be of therapeutic benefit.

One of the most attractive properties of ²¹¹At that contributes to the emerging demand for this radionuclide is the spectrum of targeting agents that are compatible with its labeling chemistry and physical half-life. In contrast to other α-emitters, astatine is a halogen and shares similar chemical properties with iodine, albeit with more metalloid properties [15]. Astatine can exist in several oxidation states [16] providing multiple synthetic options but also contributing to its sometimes confounding, capricious behavior [17]. Although significant differences between astatine and iodine have been observed in labeling chemistry, carbon-halogen bond strength, lipophilicity and the in vivo stability of carbon-halogen bond, the most common strategy to develop ²¹¹At-labeled TAT agents is to build on previous studies with radioiodine. Provided that one can demonstrate similar in vivo behavior for the two labeled compounds, this suggests the possibility of using a radioiodinated analogue (¹²⁴I for PET, ¹²³I for SPECT) as an imaging theranostic partner for the corresponding ²¹¹At-labeled therapeutic. Astatine-211 can readily be incorporated by direct substitution into small organic molecules, a potential advantage compared with radiometals, which require somewhat bulky polydentate ligands for stable incorporation. Two of the widely investigated approaches for ²¹¹At labeling are the electrophilic demetallation of tin and silicon precursors [18], and the use of carboranyl precursors [19]. Recently, a novel approach for ²¹¹At labeling that involves Cu-catalysed astatination of boronic esters was demonstrated to have broad applicability, including to the labeling of a PARP inhibitor [20]. Another method used a sulfonyl precursor for labeling neopentyl derivatives, providing high in vivo stability against nucleophilic substitution or Cytochrome P450 (CYP) metabolism [21].

As is the case with radiometals, biomolecules including monoclonal antibodies, affibodies, diabodies and nanobodies can be labeled with ²¹¹At using a variety of procedures [22]. For example, this can be accomplished via either the prototypical acylation agent N-succinimidyl 3-[²¹¹At]astatobenzoate [23], using thiol-Michael addition for site-specific conjugation [24], or N-succinimidyl 3-[²¹¹At]astato-5-guanidinomethyl benzoate, a prosthetic agent that provides intracellular trapping of radioactivity after internalization of receptor-targeted vectors [25, 26]. Finally, the remarkable affinity of ²¹¹At for gold has permitted the direct and nearly quantitative ²¹¹At labeling of gold nanoparticles, which can be used either alone or with targeting vectors decorating their surface [27]. Additional information about many of the most promising ²¹¹At-labeled radiopharmaceuticals that have been investigated for TAT applications, including those that have been evaluated in patients, can be found in several reviews [13, 22].

3. Production and Availability of ²¹¹At: Current State of the Art

Although the potential advantages of ²¹¹At for TAT are well recognized, the availability of ²¹¹At has been compromised by the limited number of cyclotrons with the beam characteristics required for its efficient production. By far the most commonly used route for ²¹¹At production is to irradiate natural bismuth with α-particles via the the ²⁰⁹Bi(α,2n)²¹¹At nuclear reaction. The incident energy (E_α) threshold for this reaction is about 20 MeV, reaching a maximum of ca. 900 mb at 30 MeV [28]. Unfortunately, exploiting much of the energy range where ²¹¹At production is most efficient is not advisable because once E_α is above 28.1 MeV, production of 8.1-h ²¹⁰At can occur via the ²⁰⁹Bi(α,3n)²¹⁰At reaction (Fig. 2). Particularly when patient studies are envisioned, production of ²¹⁰At must be avoided because it decays to ²¹⁰Po, a 138-day α-emitter with well-documented toxicity to bone marrow at very low doses. For this reason, E_α needs to be carefully optimized to maximize the production yield of ²¹¹At while avoiding producing significant levels of ²¹⁰At. With that goal in mind, Henriksen et al. [29] found that the content of co-produced ²¹⁰At at E_α = 27.6 and 28.6 MeV was undetectable and <0.01% (²¹⁰At activity relative to ²¹¹At; defined as the atomic ratio at the end of bombardment), respectively, while the ²¹¹At yield increased by about 40% from 27.6 to 28.6 MeV. Furthermore, the ²¹¹At yield increased 70% from 27.6 to 29.1 MeV with only 0.023% atomic ratio of ²¹⁰At present. Increasing E_α further (29.6 MeV) increased the ²¹⁰At atomic ratio almost tenfold, indicating an optimized E_α of 29 MeV. This was confirmed in a subsequent study reporting that ²¹¹At could be produced in high yield at 29 MeV with no evidence of co-produced ²¹⁰At [30]. Nonetheless, most of the 13 accelerators believed to have produced ²¹¹At during the past 5 years (Table 1) have utilized an E_α ≤28.4 MeV, in some cases, because required activity levels were readily obtainable at a lower E_α [31, 32].

Figure 2.

Cross sections for the production of ²¹¹At and ²¹⁰At by bombardment of natural bismuth targets via the ²⁰⁹Bi(α,2n)²¹¹At and ²⁰⁹Bi(α,3n)²¹⁰At nuclear reactions, respectively, as a function of incident α-particle beam energy (E_α). Data from Experimental Nuclear Reaction Data (EXFOR https://www-nds.iaea.org/exfor/), reported by Hermanne et al., 2005) and Lambrecht and Mirzadeh, 1985.

Table 1.

Current ²¹¹At production sites. - Facilities that have reported production of ²¹¹At during the last 5 years.

Location		Facility	CyclotronManufacturer	Model and Target	ProductionParameters	Current Production Status
NorthAmerica	Durham,USA	Duke University Medical Center	CTI	CS-30 cyclotron, Internal target system	28 MeV, 100 μA	Max 9.3 GBq in 4-h
	Seattle, USA	University of Washington Medical Center	Scanditronix	MP-50, External target system	29.0 MeV, 58 μA	Max 4.3 GBq in 4-h
	Philadelphia, USA	University of Pennsylvania	Japan Steel Works (JSW)	BC3015, External Target	28.4 MeV, 10 μA	Max 395 MBq in 5-h
	Bethesda, USA	National Institutes of Health	CTI	CS-30 cyclotron, Internal target system	29.8 MeV, 43 μA	Max 1.71 GBq in 1-h
	College Station, USA	Texas A&M University	In house	K150 variable energy cyclotron	28.8 MeV, 7 μA	1.5 GBq in 9-h
Europe	Copenhagen, Denmark	Copenhagen University Hospital	Scanditronix	MC-32, Internal target system	29 MeV, 20 μA	Max 3–4 GBq in 4-h
	Nantes, France	Arronax	IBA	Cyclone 70	28 MeV, 15 μA	Production since 2020, 0.5 – 1 GBq capacity
Asia	Osaka, Japan	RCNP-Osaka University	In house	K140 AVF cyclotron	28.2 MeV	3 GBq expected after upgrade
	Chengdu, China	Sichuan University	CTI	CS-30	28 MeV, 15–20 μA	Max 200 MBq in 2-h
	Takasaki, Japan	QST-Takasaki, (TIARA)	In house	AVF (K110)	28.1 MeV, 4.5 μA	300 MBq in 3 h
	Chiba	QST-NIRS	In house	AVF-930	28.5 MeV, 10–13 μA	0.74–1.11GBq in 5-h
	Wako Saitama, Japan	IPCR Riken	In house	AVF	29 MeV, 40 μA	1.3 GBq in 1-h
	Fukushima City, Japan	Fukushima Medical University	Sumitomo	CYPRIS MP-30, External target system	29 MeV, 20 μA	Max 2 GBq in 4-h

The fact that this production route involves the irradiation of an inexpensive and abundant natural material is a significant advantage. This allows safe handling and fabrication of targets, and obviates the need for procedures to recover and purify the target material. However, another property of bismuth can limit efficient ²¹¹At production, particularly when scale up to high beam currents is envisioned. Bismuth has poor thermal conductivity (7.97 W·K⁻¹·m⁻¹) and a low melting point (272°C), making it essential to provide good target cooling, particularly for high-current production runs. Target preparation is straightforward - metallic bismuth is simply melted or vaporized onto an aluminum backing plate; the surface is often machined to provide a uniform surface. Copper backing plates also have been evaluated to improve thermal conductivity; however, this strategy was not effective because it lowered ²¹¹At isolation yields and led to the production of longer-lived radionuclidic impurities [33]. Thick target yields reported for ²¹¹At production after 28–29.5 MeV α-particle bombardments range from 16.3 to 41 MBq/μA-h (electric μA) [29, 33–37].

Several factors need to be considered to optimize ²¹¹At production yield with the thickness of the bismuth target perhaps being the most critical aspect. Assuming an α-particle beam of 28 MeV, its kinetic energy would be degraded to 20 MeV after penetrating 80 μm in a metallic Bi target at a perpendicular incident angle, and thus below the threshold for the ²⁰⁹Bi(α,2n)²¹¹At reaction (calculated using SRIM http://www.srim.org accessed 26 May 2021). In this case, the optimal thickness of the Bi target is 80 μm at a perpendicular incident angle because a thinner target will not fully utilize the productive energy window while a thicker target will generate excess heat in the target with no increase in usable cross section. Placing the Bi target at a tangential incident angle relative to the beam direction (grazing angle) will increase the contact area of beam on the target, allowing better heat dissipation. This also will reduce the effective thickness of the Bi, permitting use of a thinner target, thereby allowing higher heat dissipation on the Al backing plate. Moreover, use of a thinner target should facilitate subsequent extraction of the ²¹¹At from the target either by target distillation or dissolution (vide infra). Both the front of the bismuth target and the aluminum backing plate need to be cooled sufficiently to prevent the bismuth from melting during the irradiation, which can lead to ²¹¹At escape, compromising production yield and potentially complicating handling [30, 38]. Although not as effective as other potential backing materials like copper, the thermal conductivity of aluminum is σ=237 W/(m*K), much larger than that of bismuth. Thus, efficient heat transfer at the interface of both the Bi and Al, and the Al and the coolant, will be necessary to allow maximum bean current for high ²¹¹At production rates.

As a frame of reference for the sections that follow, the irradiation parameters currently utilized at the three cyclotron facilities that have had the most active ²¹¹At production programs in recent years are compared in Fig. 3. In each case, ²¹¹At production runs have occurred on about a weekly basis in support of radiochemistry, preclinical and clinical studies. The target location, incident angle between the beam and the target, and the bismuth target thickness all vary among these cyclotron facilities. Nonetheless, each has been a reliable source of ²¹¹At production for more than 10 years.

Figure 3.

Irradiation parameters utilized at the 3 institutes currently having the highest ²¹¹At production capability.

3.1. Internal target systems for ²¹¹At production

At least three institutions have developed internal targets for ²¹¹At production [33–35]. One of these is at the leading ²¹¹At production facility in Europe - the PET and Cyclotron Unit at Copenhagen University Hospital in Denmark [32]. It is utilized on a Scanditronix MC32 cyclotron that was installed in the late 1980s. The internal target system consists of a bismuth target with an aluminum backing that is mounted on a water-cooled probe and placed at a 15° grazing angle for better heat dissipation (Fig. 4) [39]. An α-particle beam energy of 29 MeV is used for irradiation and GBq levels of ²¹¹At have been produced routinely, enabling many preclinical studies and at least one clinical trial [40].

Figure 4.

Production of ²¹¹At at the PET and Cyclotron Unit, Rigshospitalet, Copenhagen, Denmark. The Bi target is placed on a water-cooled probe at a grazing angle on beam contact. Following transport to Gothenberg, Sweden, a dry distillation apparatus is used for the purification of ²¹¹At.

The CS-30 cyclotron at Duke University Medical Center (DUMC) is equipped with an internal target system optimized for ²¹¹At production [33]. The MIT-1 internal target system was installed in the late 1990s and designed in collaboration with George Hendry (Cyclotron, Inc.) (Fig. 5) [33]. An aluminum target backing was fabricated with a curved face for bismuth target deposition to maximize the beam-target contact area and with fins machined into the back of the target to improve cooling efficiency. The target is placed at a 4.7° grazing angle, providing a higher surface area and reducing the heat density generated by the beam leading to improved heat dissipation. The target system also includes two graphite beam-current monitors located at the leading- and trailing-edge to optimize target positioning. These design elements have enabled the use of higher currents and longer irradiations without significantly increasing the surface temperature of the bismuth target. This internal target system has allowed weekly production of ²¹¹At at an α-beam energy of 28 MeV and greater than 65 μA beam current (electrical current), with a yield of 41 ± 7 MBq/μA-h; the saturation yield calculated based on this is 400 MBq/μA, close to the theoretical saturation yield - 420 MBq/μA [29, 34]. The highest published level of ²¹¹At produced with this system is 6.6 GBq at end of bombardment after a 4-h irradiation at 55 μAp (particle μA) [9]; although the maximum produced to date is 9.3 GBq. For recent clinical-level production runs, the irradiation parameters have been to run for 3 h at 48 μAp (particle μA), which has reliably provided an average of 6.8 GBq at the end of bombardment.

Figure 5.

Illustration of the ²¹¹At production and purification methods utilized at the Duke University Medical Center. The production of ²¹¹At is accomplished using an internal target system and ²¹¹At is isolated from the bismuth target by dry distillation. After elution from PTFE in the required solvent, the ²¹¹At is available for use in radiolabeling procedures.

It should be noted that there are significant hurdles in implementing internal target systems to existing cyclotrons because they require the design and manufacture of target rams that can accurately place the target inside the cyclotron with reliable and reproducible mechanical control. Beam simulations have to be implemented to ensure an accurate α-particle beam energy, particularly on the high end to avoid unwanted ²¹⁰At production. Other optimization processes will be necessary to achieve high production rate and radionuclidic purity. Although current ²¹¹At activity level requirements have been readily achievable at DUMC using the internal target at an E_α=28 MeV, it seems likely that the system will be optimized in the near future for an E_α=29 MeV, which should increase ²¹¹At yields considerably [34].

3.2. External target systems for ²¹¹At production

Although lower ²¹¹At production yields have been reported using external targets at a few sites [33, 35], newer cyclotrons equipped with optimized external target systems are capable of producing ²¹¹At at levels comparable with those of internal target systems [30, 32, 41–43]. Notably, Gagnon et al. [30] at the University of Washington Medical Center reported a maximum ²¹¹At production of 4.3 GBq after a 4-h irradiation with a 29.0-MeV incident α-particle energy and a 58 μA beam current on a Scanditronix MC-50 cyclotron (Fig. 6). A slanted target geometry was used to optimize the incident angle at the beam-target contact to reduce beam heat density. By also utilizing a water cooling system, the peak target temperature was estimated to be only about 54°C when running at ~50 μA, well below the bismuth melting point [30]. This is the highest reported ²¹¹At production level for an external target system and has been sufficient to permit the clinical translation of an ²¹¹At-labeled ant-CD45 monoclonal antibody [44].

Figure 6.

The ²¹¹At production and purification process at the University of Washington Medical Center. An external target is placed at a slanted angle in the pneumatic target system. An automatic system was developed to dissolve the target material and purify ²¹¹At using a wet chemistry method.

4. Isolation of ²¹¹At from the Cyclotron Target

The efficiency of ²¹¹At isolation from the cyclotron target both in terms of yield and time depends in large part on the methods utilized for performing the cyclotron irradiation including the size, thickness and mass of the bismuth target and its backing plate, and the backing plate material. In addition, the length of the irradiation as well as the intensity of the beam current can also play a significant role [33]. Two strategies have been used successfully for ²¹¹At purification – dry distillation and wet harvesting – with the choice between the two depending on technical considerations such as target dimensions and subsequent labeling chemistry as well as on personal preference of the investigator [45].

4.1. Dry distillation

Currently the most commonly used approach, this procedure utilizes a distillation apparatus configured to accommodate the size and shape of the bismuth target/backing plate [39, 45, 46]. An exception is a method where the bismuth is separated from the backing by either scraping, cutting or heating before the beginning of the distillation [39, 45, 46]. Although this step allows the use of a still with a smaller footprint and can expedite the distillation process, the increased potential for contamination is a major concern particularly when high activity levels of ²¹¹At are present. All procedures involve heating the irradiated target above the melting point of bismuth (272°C) and the boiling point of astatine (302°C), with most using a temperature of 650–800°C for 30 min or less. A recent abstract reported that use of an induction heating method was able to achieve an average ²¹¹At recovery of 51% in 10 min [47]. Further refinements of this technology are expected and should provide a dry distillation procedure that will offer significant advantages compared with current methods in terms of convenience, speed and ²¹¹At recovery yields.

Effective trapping of ²¹¹At generally requires proper balance of carrier gas flow – either nitrogen [39], argon [9], or in some cases O₂ [48] – with a modest vacuum on the distillation apparatus. Trapping the ²¹¹At after release from the target has been accomplished using several approaches including silica columns [28] and bubbler traps [49]. In current practice, a capillary tubing cryotrap fabricated from PEEK has become the favored approach [39]. A recent study has utilized PTFE tubing for this purpose due to its greater flexibility and translucency, facilitating the monitoring of extractant flow [50]. Distillation efficiencies of 80–90% have been reported at lower ²¹¹At activity levels [46]; however, lower and more variable ²¹¹At recovery (51–83%) was observed for therapy-level production runs [9]. Whether radiolytic factors are at play here remains to be elucidated in future studies. Although the potential merits of wet harvesting methods, described below, are recognized, and dry distillation methods can be capricious, they do offer several important advantageous including: a) limited handling of radioactive materials; b) minimal exposure of ²¹¹At to chemicals, including those that can alter its chemical form, frequently in unpredictable ways, c) the ability to trap the ²¹¹At activity in relatively small volumes and in different media to meet the needs of subsequent radiochemical procedures; and d) amenability to automation within systems that also can include the synthesis of the required ²¹¹At-labeled therapeutic [46].

4.2. Wet chemistry extraction methods

The separation of ²¹¹At from bismuth target materials in liquid form was first introduced in the 1950s [51, 52] during early studies of ²¹¹At radiochemistry. Balkin et al. [45] subsequently developed an optimized wet chemistry method that was both relatively easy to conduct and provided 78 ± 11% recovery after corrections for decay. More recently, this method was further optimized in order to meet the requirements for clinical translation [44]. Building on this advance, Li et al. developed a semi-automated process using a tellurium-packed column for isolating the ²¹¹At, reporting an 88–95% recovery yield after a 2-h separation process [53]. Wet chemistry approaches risk the formation of many solvated species of ²¹¹At, which can result in a lower radiochemical purity of the ²¹¹At product, potentially affecting the subsequent radiolabeling yield. On the other hand, this approach generally provides higher and more consistent ²¹¹At separation yields than dry distillation [45]. Additional refinements have improved the radiochemical purity of the isolated ²¹¹At. For example, Li et al. [54] optimized the tellurium-packed column separation method by neutralizing the HNO₃ from the target dissolution process with aqueous NH₂OH·HCl and achieved high separation yields (88–95%) and >99% radiochemical purity for sodium [²¹¹At]astatide. Others have addressed this issue by utilizing a solid-state support to permit control over ²¹¹At speciation at the end of separation [55]. Wet harvesting methods are advantageous because they can provide higher and more reproducible yields; however, the ²¹¹At speciation at the end of purification is limited to [²¹¹At]astatide, which is not directly useful for electrophilic demetallation radiolabelling chemistry. Finally, wet harvesting approaches avoid the formation of volatile ²¹¹At⁰, which is advantageous from a radiation safety perspective.

In summary, high ²¹¹At recovery from irradiated bismuth targets have been reported for both dry-distillation and wet-chemistry isolation methods. These two approaches are complementary and provide the researcher with the ability to choose the type of procedure that best suits local facility constraints and the type of chemistry for which the isolated ²¹¹At is intended. Moreover, new ²¹¹At purification methods are currently being developed [56]; however, it remains to be seen whether they will supplant existing procedures, particularly for use with high-activity level targets.

4.3. Radiolysis and potential effects on ²¹¹At radiochemistry

The chemical form and oxidation state of ²¹¹At at the end of the purification process is perhaps the most important aspect in the production of ²¹¹At that has rarely been addressed. Nevertheless, these factors form a vital bridge between the chemical environment during purification and its effects on subsequent radiolabeling chemistry, sometimes confounding the later, particularly as the time interval between purification and radiolabeling increases. This behavior, as well as the sometimes erratic behavior of ²¹¹At at high activity concentrations [17, 32], suggest that radiolysis induced by the high-energy deposited by ²¹¹At α-particles in small volumes likely plays an important role. Radiolysis can generate free radicals, ions and other molecules including oxidants and reductants. These species can alter the oxidation state of the radionuclide and generate competitive species that either react with the radionuclide or the substrate, diminishing labeling yields and effective molar activity.

The dry distillation method is advantageous in this regard because during distillation atomic ²¹¹At⁰ is released from the target material in its gaseous phase and subsequently trapped by low temperature, minimizing exposure to chemicals and the likelihood of oxidation state change. On the other hand, the choice of solvent to elute ²¹¹At from the cold trap is critical because during the interaction between ²¹¹At and solvent, the radiation dose quickly exceeds 1000 Gy [9]. Chloroform frequently has been used for this purpose; however, chlorine radicals can be generated in a dose dependent manner after exposure to ²¹¹At. These radicals can then compete for precursor in electrophilic demetallation reactions, leading to a recommendation to use methanol as the solvent for ²¹¹At trapping [57]. Subsequent studies demonstrated that in methanol, exposure to increasing doses of ²¹¹At could alter pH and the oxidation state of the ²¹¹At to a species, presumed to be [²¹¹At]astatide, with poor reactivity for electrophilic astatodestannylation [58, 59]. This difficulty can be circumvented by adding a small amount of N-chlorosuccinimide (NCS) to the methanol used to elute the ²¹¹At from the cold trap, which is effective in preserving reactivity except at very high radiation doses and dose rates [60].

Although electrophilic approaches have been the mainstay of ²¹¹At labeling chemistry, promising alternative labeling strategies using nucleophilic astatine have emerged including those utilizing carboranyl and boronic ester precursors [19, 20, 45, 61]. The generation of [²¹¹At]astatide at elevated radiation doses suggests that these methods might be well suited to ²¹¹At labeling at very high activity concentrations. In addition, eluting the cold trap with NaOH solutions after dry distillation might be preferable with these chemistries. More research is needed to investigate the speciation of ²¹¹At, especially at high radioactivity levels, in potential trapping media.

Unfortunately, little information is available about potential radiolysis effects during the separation of ²¹¹At from cyclotron targets via wet chemistry approaches, and as noted above, on nucleophilic radioastatination. It is speculated that radiolysis could play a role in aqueous media but the effects will be different that those in organic solvents and that α-particles will generate different species that those from other types of radiation [62]. For example, the radiolysis of water can produce, hydrogen, hydrogen peroxide, as well as radicals that may be important at elevated ²¹¹At activity levels [63]. Studies of these phenomena might be aided by a recently reported radio-chromatography method using thin layer chromatography to identify different ²¹¹At species in aqueous solution in the presence of oxidation or reducing agents [64] although considering the volatility of the At⁰ species, an HPLC-based method might be preferable in the future.

4.4. Automation of ²¹¹At purification procedures

The radiation emitted by ²¹¹At is relatively benign from a shielding perspective. However, with >10 GBq activity levels anticipated and 2.5-min positron-emitting ³⁰P (produced by α-irradiation of the aluminum target backing plate) delaying manual processing, automation of the ²¹¹At purification procedure needs to be considered. Unlike the case with PET radionuclides, there are no commercially available automated systems for the purification of ²¹¹At after cyclotron irradiation. At present, most institutions producing ²¹¹At use customized systems involving manual operations. A confounding factor for automation is the variety of target sizes and configurations and purification methods in use at the different sites. Nonetheless, efforts are ongoing to develop of automatic systems for ²¹¹At purification. Aneheim et al. [46, 65] reported an automatic process for dry distillation that involved software-controlled monitoring of ²¹¹At activity and the option of combining it with a synthesis module for radiolabeling. A scraping device was developed to remove the bismuth target material from the aluminum backing plate to be able to reduce the still footprint and shorten the distillation time. For use with wet chemistry approaches, Li et al. [53] developed an automatic system for ²¹¹At purification using a tellurium-packed column. In addition, O’Hara et al. [66] reported a similar system using wet chemistry method. Further research in the development of automated modules for ²¹¹At purification is needed to make ²¹¹At more amenable for routine availability.

5. Current ²¹¹At Production Capabilities

Although there is considerable interest in using ²¹¹At for TAT, the feasibility of translating these concepts into clinically impactful drugs is critically dependent on improving the supply of ²¹¹At. As a first step, we have attempted to provide a comprehensive overview of current capabilities worldwide. There is an ongoing effort to expand ²¹¹At production in the US, Europe and Japan either through the construction of new cyclotrons or by taking advantage of existing accelerators. With regard to the latter, several institutes stopped producing ²¹¹At during the last 10 years and their production capability could be revitalized, at least in principle, to meet a growing demand for this promising radionuclide. Herein, we have summarized the characteristics and capabilities of both the facilities that currently (defined as in the last 5 years) produce ²¹¹At (Table 1) as well as potential production sites for ²¹¹At that should/could be available in the future (Table 2).

Table 2.

Production sites for ²¹¹At that could be operational in the near future.

Location	Facility	Accelerator Model	Irradiation Parameters	Current Production Status
Davis, USA	Crocker Nuclear Laboratory, UC Davis	Crocker Cyclotron	Currently 29 MeV α beam at 30 μA, internal target station	Retrofitting in progress
Lansing, USA	Ionetix	CS-30 cyclotron	Expected to produce 28 MeV α beam, 100 μA current	Cyclotron subsystem updating underway; should be operational in 2022
Rez, Czech Republic	Nuclear Physics Institute of the CAS	U-120M	<40 MeV α beam 40/5 μA current at internal/external	Last production in 2003 but are ready and willing to restart
Warsaw, Poland	POLATOM	IBA Cyclone 30XP	30 MeV α beam at 50 μA current, external target station	Building and installation begin at the end of 2021
Birmingham, United Kingdom	University of Birmingham	Scanditronix MC-40	11.8–40 MeV α beam at >10 μA current, external target	Currently no production, but 1 GBq weekly could be expected on demand. Production can be optimized further.
Bucharest, Romania	IFIN-HH, Magurele	U-120	26 MeV α beam, 100/20 μA current at internal/external target stations	3 MBq produced in 2015, cyclotron reinvigoration underway
Julich, Germany	Forschungszentrum Julich	IBA Cyclone 30 XP	30 MeV α beam at 50 μA current, external target station	Production expected to start Q2 2021
Seoul, South Korea	KIRMS	Scanditronix MC-50	29.2 MeV α beam at 17 μA current, external target station	32 MBq/ μAh Production started Q1 2021
Wako, Japan	RIKEN	Heavy-ion Linac	Estimated GBq level production	Optimization underway

Before discussing institutional capabilities, it is worth noting efforts in several geographical areas to coordinate ²¹¹At production. Remarkably, during the last decade, five Japanese institutes have established routine production and purification of ²¹¹At, making a total of six ²¹¹At production sites, enabling many preclinical using ²¹¹At in Japan [48, 67–71]. In 2020, the European Cooperation in Science and Technology (COST) funded a multi-institutional program focused on ²¹¹At TAT including a goal to build an efficient ²¹¹At production network to address the needs of the European Union members and neighboring countries (COST Action CA19114, https://www.cost.eu/actions/CA19114/#tabs|Name:overview accessed 26 May 2021). In the US, the Department of Energy recently has provided funding for enhancing ²¹¹At production at existing cyclotrons with the objective of utilizing the DOE Isotope Program University Isotope Network (DOE IP UIN) as a conduit for ²¹¹At distribution. Hopefully, these cooperative efforts will expand the availability of ²¹¹At in the near future. Current production capabilities at selected cyclotrons in different geographical regions are presented below.

5.1. North America

Within North America, production of 211At using the most commonly used α-particle bombardment of bismuth approach has been limited to facilities within the US. At Duke University, high-level ²¹¹At is being produced weekly for radiochemistry and preclinical research. Although no clinical studies are currently ongoing, the typical production level has been about 8–10 GBq per month. With two ²¹¹At TAT agents expected to enter clinical trial in about a year, that level is expected to double, which is well within current capabilities. The cyclotron facility at the University of Washington is routinely producing high-level ²¹¹At in support of one clinical and several preclinical programs [44]. Since 2018, over 74 GBq of ²¹¹At has been produced on target per year. The University of Pennsylvania has recently enhanced their ²¹¹At production capability, reporting a maximum of 396 MBq produced from a single run on their JSW 3015 cyclotron [72]. In 2020, Texas A&M University reported the production of 890 ± 80 MBq of ²¹¹At after an 8 h run using a 28.8 MeV incident α-beam energy at 4–8 μA current [73]. Recently, they described a novel separation method for isolation of ²¹¹At from the target [56].

5.2. Europe

For more than a decade, the most active facility for ²¹¹At production in Europe has been the PET and Cyclotron Unit at the Copenhagen University Hospital. They are capable of producing a maximum of about 3.6 GBq of ²¹¹At after an 8-h run [32]. In Nantes, France, the ARRONAX facility has been operating since 2008 and is capable of accelerating 70 MeV α-particles with a beam current of 35 μA. The beam is degraded to about 28 MeV to avoid ²¹⁰At production, which has permitted the production of up to ~1 GBq of ²¹¹At for basic research and eventually, clinical studies [74, 75].

5.3. Asia

Although the production of ²¹¹At in Asia has just recently gained momentum with multiple production facilities being established in Japan and South Korea, ²¹¹At has a deep-rooted history in Asia. The Research Center for Nuclear Physics (RCNP) at Osaka University is equipped with a K140 AVF cyclotron and has been producing ²¹¹At for more than two decades [76]. Even before that, Sichuan University in China installed a CS-30 cyclotron in the 1980s, contemporaneous with the CS-30 cyclotron installed at Duke. Astatine-211 production, separation and labeling chemistry have been investigated at Sichuan University since that time [77] and production of research level of ²¹¹At has been reported in recent years [78]. Since 2011, many new production sites have emerged across Japan including those at the Japan Atomic Energy Agency (JAEA), the National Institutes for Quantum and Radiological Science and Technology (QST), RIKEN and Fukushima Medical University (FMU). As summarized in Table 1, GBq levels of ²¹¹At are being produced in Chiba, Wako and Fukushima City.

6. Additional ²¹¹At Production Sites Potentially Available in the Near Term

In addition to the 13 production sites that are currently supplying ²¹¹At, we have identified 9 other accelerator facilities that are likely to contribute to worldwide ²¹¹At supply in the near future. With the exception of the IBA Cyclone 30XP that will be located at POLATOM (Warsaw) expected to be operational in 2022, the remaining sites are existing machines (Table 2). These cyclotrons, some of which have made ²¹¹At in the past, are undergoing a varying degree of modification, upgrading and/or target fabrication in order to produce ²¹¹At. Anticipated activity levels at these facilities range from about 3 MBq in Bucharest, sufficient for chemistry and cell studies, to more than 10 GBq in Lansing, enough to treat multiple patients. Of considerably less certainty is the possible contribution of the 10 cyclotrons listed in Table 3 to future ²¹¹At supply. Most of these machines have made ²¹¹At successfully in the past; however, some of these cyclotrons have either ceased operation or are scheduled for shutdown soon. At least hypothetically, some of these machines could be revived if demand for ²¹¹At increased to a level that would make a restart financially viable.

Table 3.

Facilities that were capable of ²¹¹At production but are either shut down, shutting down or of unknown availability/capability status.

Location	Facility	Cyclotron Model	Cyclotron Parameters	Current Production Status
Oslo, Norway	University of Oslo	Scanditronix MC-35	Unknown	Cyclotron under repair
Orleans, France	CNRS—CEMHTI laboratory	THOMSON-CSF	<50 MeV α beam at 15 μA current	Machine scheduled for shutdown in 2023
Paris, France	Curium, Saclay	CGR-560	<40 MeV	Appears to be operational as commercial machine
Dubna, Russia	JINR—FLNR	U-120 U-150	Unknown	Unknown
Tomsk, Russia	Tomsk polytechnic University	U-120	28 MeV α beam at 50 μA current	Producing 211At for research
Warsaw, Poland	Heavy Ion Laboratory, University of Warsaw	U200-P	≤34 MeV α beam at ≤1 μA current	Last production in 2017; no plans for future 211 At because POLATOM will produce
Jyväskylä, Finland	University of Jyväskylä	Scanditronix K130	Capable of producing α beam	No license for 211 At production, currently research purposed only
Ispra, Italy	JRC	Scanditronix MC-40	Currently no production	Shut down in 2014
Brussels, Belgium	VUB	CGR-MeV model 560	<40 MeV α beam	Low level production in the past but cyclotron was shut down in 2020
Essen, Germany	Universitätsklinikum Essen	Cyclotron Corp CV-28	28 MeV α beam at 50 μA current	Shut down in 2016

Of the cyclotrons listed in Table 2, several facilities are likely to begin ²¹¹At production during the next few years. In North America, Crocker Nuclear Laboratory at the University of California, Davis was recently funded by the US Department of Energy to establish ²¹¹At capability at their cyclotron. Significant progress has been made in target fabrication and beam current optimization, and production is estimated to begin within the year [79]. Ionetix has acquired a CS-30 cyclotron similar to the machine at Duke University and is in the process of updating the machine with the goal of producing at least 10 GBq batches of ²¹¹At by Q4 2022. In Europe, it is anticipated that the Network for Optimized Astatine labeled Radiopharmaceuticals (NOAR) under COST Action CA19114 will motivate and perhaps support the development of more accelerators for ²¹¹At production. Astatine-211 production is planned at the new IBA Cyclone 30XP cyclotron at the Radioisotope Centre POLATOM (Warsaw, Poland) and seems likely, pending demand, at existing cyclotrons at the Institute for Nuclear Chemistry (Jülich, Germany) and the University of Birmingham, UK. Finally, in Asia, ²¹¹At production began in Q1 2021 at the Korean Institute of Radiological and Medical Science (KIRMS) Scanditronix MC-50 cyclotron. It is anticipated that the activity levels required for clinical investigation of ²¹¹At-labeled TAT agents should be available in Seoul in about a year.

7. Current and Potential Future Commercially Available Accelerators for ²¹¹At Production

Unlike the case with other TAT radionuclides, ²¹¹At supply is not constrained by a dependency on nuclear stockpiles that are heavily regulated and in some cases, of limited availability, or on inconvenient target materials [2]. The situation with ²¹¹At is different - ²¹¹At can be produced at high activity levels using ordinary target material and existing technology. Currently, there are at least three commercially available cyclotrons suitable for ²¹¹At production and their characteristics are summarized in Table 4. Both the IBA Cyclone 70XP and the Sumitomo MP-30 cyclotrons have been used to make ²¹¹At and it is anticipated that the IBA Cyclone 30XP will be evaluated for this purpose soon. Also included in Table 4 are two accelerators at various stages of development. Nusano is working on a linear accelerator that is anticipated to produce 3 mA of 50 MeV α-particles. Ionetix is working on a superconducting cyclotron, conceptually similar to their [¹³N]ammonia unit-dose machines, with a design goal of 300-μA beam current at 29 MeV. Timelines for development of these accelerators are not available and it is also not clear whether these machines will be commercially available for purchase or operated by these companies as part of their own isotope supply networks. Either way, if they meet their design goals, these accelerators could have a major impact on worldwide ²¹¹At supply.

Table 4.

Current and potential future commercial sources of accelerators for ²¹¹At production.

Company	Accelerator	Irradiation Parameters	Features	Status
IBA	Cyclone 30XP	30 MeV, 50 μA current, External target station	Can produce proton and deuteron beams	Available
IBA	Cyclone 70XP	67.4 MeV, 70 μA at internal target and 15 μA at external target station	Fixed energy so energy degrader required	Available
Sumitomo	MP-30	32 MeV, 30 μA current, External target station	Automatic target transport system	Available
Ionetix	Superconductingcyclotron	Design goal is 29 MeV, 300 μA current	Small footprint	Underdevelopment
Nusano	LINAC	Expected to produce 3 mA α beam current at 50 MeV	High beam current	Underdevelopment

With regard to the ²¹¹At activity levels that might be available to meet future needs, as a first approximation, one could estimate the number of doses of a typical ²¹¹At-labeled TAT agent that could be produced given current and projected future accelerator capabilities. Based on a semi-automated procedure for the synthesis of ²¹¹At-labeled trastuzumab [32, 46], we assume decay-corrected radiochemical yields of 89% for ²¹¹At purification, 75% for ²¹¹At-TAT agent labeling and 1 h as the total time required for purification and synthesis (91% activity remaining after decay). Under these assumptions, 607 MBq of ²¹¹At-labeled TAT agent could be produced per GBq of ²¹¹At in the target. As noted in Section 3.1., we routinely produce 6.8 GBq of ²¹¹At from a 3-h irradiation with 28 MeV α-particles at our cyclotron; with a planned increase in α-particle beam energy to 29 MeV, production of at least 14 GBq is expected from a 3-h irradiation [29]. If one assumes that 370 MBq represents a typical patient dose, then a 3-h irradiation at 28 and 29 MeV could produce 11.2 and 23.0 doses, respectively. Thus, two 3-h 29-MeV production runs per day, 5 days a week, 50 weeks per year, could provide 11,500 patient doses per year with existing technology using one cyclotron, a level more than twice that estimated for existing ²²⁵Ac capabilities [2]. Under the admittedly optimistic assumptions that 1) next generation cyclotrons come close to meeting the design goals indicated in Table 4 2) purification and labeling chemistry is scalable, and 3) ²¹¹At accelerator production networks become financially viable, production of ≥100,00 patient doses per year might be possible.

8. The ²¹¹Rn→²¹¹At Generator

Although the 7.2-h half-life of ²¹¹At is an advantage for use in tandem with many of the most promising molecular vehicles for TAT, it presents a logistical hurdle for distribution of the radionuclide beyond its production site. An attractive approach to circumvent this problem is via the ²¹¹Rn/²¹¹At generator system. Radon-211 has a 14.6-h half-life and decays to ²¹¹At in transient equilibrium. A ²¹¹Rn/²¹¹At generator allows ²¹¹At to grow in and reach its maximum radioactivity level at 14.5 h. Moreover, over 80% of the maximum ²¹¹At activity level is still available on the generator after 24 h. This permits delivery of a generator to remote locations while ²¹¹At grows in during transportation, extending the time period for efficient ²¹¹At distribution significantly [80, 81].

Radon-211 can be produced via multiple nuclear reactions; however, the most relevant approaches are the bombardment of natural bismuth targets with 42-MeV ⁶Li ions via the ²⁰⁹Bi(⁶Li,4n)²¹¹Rn reaction [82], or with 60-MeV ⁷Li beams via the ²⁰⁹Bi(⁷Li,5n)²¹¹Rn reaction [83]. The main concern for these reactions is the co-production of ²¹⁰Rn via the ²⁰⁹Bi(⁷Li,6n)²¹⁰Rn reaction and ²¹⁰At via a break-up process [84, 85] because these radionuclidic impurities, along with their daughter radionuclides (²⁰⁶Po, ²¹⁰Po), complicate generator conditions and its utility for subsequent medical application. Maeda et al. [85] investigated excitation functions for these nuclear reactions and concluded that an incident energy between 50 and 60 MeV was optimal for the ²⁰⁹Bi(⁷Li,5n)²¹¹Rn nuclear reaction to maximize yield while minimizing radionuclidic impurities (Fig. 7). However, separation and purification of ²¹¹Rn from the target material, as well as astatine and polonium radionuclides is essential [85] and several purification methods for achieving this objective have been investigated [81, 82, 86, 87]. Nolen et al. [88] developed a porous bismuth oxide target material that would allow inline continuous release and trapping of ²¹¹Rn during the irradiation. The spallation of actinide targets is another approach for producing ²¹¹Rn with high yields [89]; however, it requires high energy protons (~480 MeV) so it is limited to a few facilities (e.g. TRIUMF) [90]. To date, the production of ²¹¹Rn and the development of the ²¹¹Rn/²¹¹At generator system are being investigated at only a few institutions (Table 5); however, the possibility of supplying ²¹¹At via this generator is an appealing strategy for helping meet a growing and more geographically disperse demand for this promising radionuclide.

Figure 7.

Optimization of ⁷Li beam energy and cooling period for the production of ²¹¹Rn. A) Estimated yields of ²¹¹Rn and ²¹⁰Rn in a thick Bi target after a 1-h irradiation at a beam intensity of 1 μAp. B) The radioactivity ratio of ²¹¹At and ²¹⁰At during the cooling period. Data kindly provided by Professor Akihiko Yokoyama (Kanazawa University, Japan). C) Possible impurities.

Table 5.

Facilities investigating the production of ²¹¹At via a ²¹¹Rn/²¹¹At generator.

Location	Institute	Production Means	Current Production Status
Lemont, USA	Argonne National Laboratory	ATLAS	Generator in development
Vancouver, Canada	TRIUMF	500 MeV TRIUMF cyclotron	40 MBq 211At per production
Caen, France	GANIL	Superconducting linear accelerator and Synchrocyclotrons	Production of 211Rn in development
Tokai, Japan	JAEA	Tandem accelerator facility	Generator in development

9. Future ²¹¹At Distribution Networks: Prospects and Considerations

If ²¹¹At-labeled TAT agents are to have any realistic prospect for evaluation in multicenter clinical trials and subsequent regulatory approval, methods must be devised for achieving their efficient distribution. As noted earlier, the challenge has similarities to that encountered by [¹⁸F]FDG, which was overcome by the installation of cyclotrons in multiple medical centers, and then when PET was on firmer financial footing, by commercial entities. Compared with ¹⁸F, ²¹¹At has a longer half-life and significantly less penetrating radiation requiring less shielding during transport; however, as discussed in Section 4.3., radiolysis can be a complicating issue.

With a ²¹¹At-labeled TAT agent, there are at least 3 choices for distribution – the irradiated bismuth target, ²¹¹At after isolation from the target, and the final ²¹¹At-labeled TAT agent. Shipping the target has been the approach taken to permit ²¹¹At produced in Copenhagen to be utilized in Gothenburg where it is distilled and labeling of the final product is performed [32, 40]. On the other hand, ²¹¹At was isolated from the cyclotron target before shipment from Duke University to Baltimore, MD in support of preclinical therapy studies. Although anecdotal, lower yields were obtained when ²¹¹At labeling was performed after the 6–8 h trip rather than before transport [91]. Further work needs to be done to optimize this strategy, which relieves the recipient institution of the burden of having the facilities and personnel required to perform ²¹¹At purification on site. On the other hand, shipping the target is attractive because it minimizes radiolysis issues by minimizing the time interval between exposure of ²¹¹At to solvents and initiation of the labeling reaction.

In one distribution model, the formulation of the ²¹¹At labeled TAT agent is centralized for use at treatment centers within a few hours’ drive of the production center (Fig. 8). This centralized distribution model is suitable for highly populated areas. In some cases, patients would travel to a hospital near the cyclotron not unlike what is done with patients undergoing proton therapy. In the de-centralized distribution model, irradiated bismuth targets could be transported to formulation hubs that have GMP-level radiochemistry labs staffed with skilled technicians to perform onsite ²¹¹At purification, labeling, formulation and quality control (Fig. 8). The formulated ²¹¹At-labeled TAT agents could then be delivered to locations around the multiple hubs, extending the coverage of the ²¹¹At production centers. Irradiated Bi targets are shipped instead of purified ²¹¹At or ²¹¹At-labeled compounds to avoid radiolysis issues and additional quality control steps, and also to allow different TAT agents or combinations to be synthesized at each site as needed. This de-centralized model would benefit by the development of automatic purification and synthesis systems to streamline radiopharmaceutical production at the distribution hubs.

Figure 8.

Two potential distribution models for ²¹¹At: A) Centralized distribution where formulated radiopharmaceuticals are distributed to hospitals within a feasible distance radius; B) De-centralized distribution where irradiated targets containing ²¹¹At are distributed to hub-facilities with a GMP radiochemistry lab and ²¹¹At radiopharmaceuticals are formulated there before distribution to hospitals.

Various combinations of the centralized and de-centralized models are possible depending on population density and available transport options. In the US, the delivery of ²¹¹At-labeled TAT agents to more than 90% of the population could be feasible with about 7–10 optimized ²¹¹At production centers. One could envision eventual involvement of entities such as PETNET Solutions Inc. and Cardinal Health because they already have countrywide networks of GMP distribution hubs and transport infrastructure for supplying PET radiopharmaceuticals. Moreover, the de-centralized distribution network model has been used in China for many years. China Isotope & Radiation Corporation, the largest radiopharmaceutical supplier in China, has built a network of GMP distribution hubs. Instead of distributing ⁹⁹Mo/^99mTc generators to end users, they are sent to distribution hubs (i.e. “milk stations”) and formulated ^99mTc radiopharmaceuticals are distributed to hospitals for patient use. With the emerging interest in ²¹¹At TAT and progress in ²¹¹At production and purification methodologies described herein, such initiatives for ²¹¹At distribution networks are not beyond the realm of reality. As a first step in that direction, Ionetix recently announced plans for an ²¹¹At distribution network (https://ionetix.com/why-alpha-therapy accessed 26 May 2021). Taking advantage of their existing GMP radiochemistry labs across the country, their CS-30 cyclotron production center in Lansing, Michigan will be able to send irradiated bismuth targets to these distribution hubs while also providing ²¹¹At for use within a 400 km radius of the cyclotron. It is anticipated that the first phase of this network – production of ²¹¹At for use locally - will be operational in 2022.

10. Conclusions

Astatine-211, with its one α-particle per decay and flexible chemistry, remains an attractive option for TAT, particularly for small molecular scaffolds. Despite the fact that ²¹¹At is produced from an abundant target material using straightforward irradiation and purification methods, its availability remains limited. It is encouraging that the need for improved accelerator infrastructure for ²¹¹At production recently has led to tangible action by government agencies in several parts of the world. In the US, the Department of Energy has funded enhancements and updates of five existing cyclotrons to improve capabilities for ²¹¹At production. Meanwhile, an EU funded COST Program was funded in 2020 and an accelerator network has been established in Japan with the same overall goal. Finally, recent interest in the commercial sector in ²¹¹At supply is an encouraging development. Hopefully, these initiatives will lead to substantive near-term improvements in ²¹¹At availability so the potential benefits of this radionuclide for TAT can be evaluated.