Development of Targeted Alpha Particle Therapy for Solid Tumors

Targeted alpha-particle therapy (TAT) aims to selectively deliver radionuclides emitting α-particles (cytotoxic payload) to tumors by chelation to monoclonal antibodies, peptides or small molecules that recognize tumor-associated antigens or cell-surface receptors. Because of the high linear energy transfer (LET) and short range of alpha (α) particles in tissue, cancer cells can be significantly damaged while causing minimal toxicity to surrounding healthy cells. Recent clinical studies have demonstrated the remarkable efficacy of TAT in the treatment of metastatic, castration-resistant prostate cancer. In this comprehensive review, we discuss the current consensus regarding the properties of the α-particle-emitting radionuclides that are potentially relevant for use in the clinic; the TAT-mediated mechanisms responsible for cell death; the different classes of targeting moieties and radiometal chelators available for TAT development; current approaches to calculating radiation dosimetry for TATs; and lead optimization via medicinal chemistry to improve the TAT radiopharmaceutical properties. We have also summarized the use of TATs in pre-clinical and clinical studies to date.


Introduction
Over the past two decades, radioimmunotherapy (RIT) has proven to be an effective treatment for non-solid tumors (reviewed in [1,2]); e.g., radiolabeled anti-CD20 monoclonal antibodies for the Figure 1. 68 Ga-PSMA-11 PET/CT images of a treatment-naïve patient with extensive bone metastasis at primary diagnosis. A complete remission was observed after three cycles of 225 Ac-PSMA-617 with de-escalating activities of 8/7/6 MBq. The patient remained symptom-free with undetectable serum PSA and a negative 68 Ga-PSMA-11 PET/CT at 11-month follow-up evaluation. This figure and legend were reproduced from Sathekge, et al. [6].
There are differences in the mechanism of tumor cell killing when comparing β-emission to αemission therapies. β-particles scale to the size of electrons, travel over a relatively longer range (0.5-12 mm) in tissues in comparison to α-particles, have relatively lower LET and generate hydroxyl freeradicals by breaking covalent bonds of water molecules in the tissue [9]. These free radicals result in oxidative damage to the cellular DNA macromolecules, causing double-strand breaks [10]. In contrast, α-particles are heavier (size of He atom), travel over a much shorter range (40-90 µm), and thus, have hundreds fold higher LET (α = 100 keV/µm versus β = 0.2 keV/µm) [11][12][13][14]. Hence, αemissions deposit large amounts of energy in a smaller volume relative to β-emissions and result in the direct breaking of covalent bonds; e.g., DNA double-strand breaks. In either case (α or β), the DNA damage can activate DNA damage checkpoints and double-strand break repair [15]. Where the damage is significant or if there are defects in the checkpoint or repair pathways such that repairs cannot be made, i.e., irreparable damage, programmed cell death (apoptosis) is initiated [16]. In apoptosis-deficient tumor cells, the resulting damage to the cellular machinery eventually results in necrotic cell death [16]. Hence, α-emission therapy potentially has several distinct advantages relative Figure 1. 68 Ga-PSMA-11 PET/CT images of a treatment-naïve patient with extensive bone metastasis at primary diagnosis. A complete remission was observed after three cycles of 225 Ac-PSMA-617 with de-escalating activities of 8/7/6 MBq. The patient remained symptom-free with undetectable serum PSA and a negative 68 Ga-PSMA-11 PET/CT at 11-month follow-up evaluation. This figure and legend were reproduced from Sathekge, et al. [6].
There are differences in the mechanism of tumor cell killing when comparing β-emission to α-emission therapies. β-particles scale to the size of electrons, travel over a relatively longer range (0.5-12 mm) in tissues in comparison to α-particles, have relatively lower LET and generate hydroxyl free-radicals by breaking covalent bonds of water molecules in the tissue [9]. These free radicals result in oxidative damage to the cellular DNA macromolecules, causing double-strand breaks [10]. In contrast, α-particles are heavier (size of He atom), travel over a much shorter range (40-90 µm), and thus, have hundreds fold higher LET (α = 100 keV/µm versus β = 0.2 keV/µm) [11][12][13][14]. Hence, α-emissions deposit large amounts of energy in a smaller volume relative to β-emissions and result in the direct breaking of covalent bonds; e.g., DNA double-strand breaks. In either case (α or β), the DNA damage can activate DNA damage checkpoints and double-strand break repair [15]. Where the damage is significant or if there are defects in the checkpoint or repair pathways such that repairs cannot be made, i.e., irreparable damage, programmed cell death (apoptosis) is initiated [16]. In apoptosis-deficient tumor cells, the resulting damage to the cellular machinery eventually results in necrotic cell death [16]. Hence, α-emission therapy potentially has several distinct advantages relative to β therapy. First, the shorter range of α-emissions result in lower off-target damage but will still allow for killing of adjacent tumor cells. Since tumors are known to be heterogeneous in marker expression [17], nearby cells that do not express the target can also be damaged and killed, potentially eliminating a mechanism of resistance; i.e., killing of marker expressing cells with the clonal expansion of non-marker expressing cells. Second, the greater energy deposition will result in greater DNA damage and a correspondingly greater level of cell killing. Third, α-emission-mediated DNA damage does not rely on the generation of free radicals, potentially eliminating the development of a major mechanism of resistance to radiation therapy; e.g., upregulation of superoxide dismutase, etc. [3].
There are many α-emitting radionuclides available for use in the development of TAT, and factor, including nuclear characteristics, availability, chemistry, specific activity, synthesis yield, chemical and biological stability of conjugates, and costs, need to be considered. Choice of targeting scaffold is also a major consideration in TAT development. Antibodies, antibody fragments, peptides and passive targeting strategies have been employed and the type of targeting moiety chosen will have a bearing on the type of radionuclide attachment used, the route of clearance and blood pharmacokinetics (PK), tissue biodistribution (BD) and radiation dosimetry. The relative merits of available radionuclides, attachment chemistries and targeting moieties are discussed herein. Radiation dosimetry (RD) is an area of key importance in development of TATs. The discrete decay chains of some α-particle-emitting radionuclides involve the generation of daughter products with individual radioactive emission properties. Hence, the biological fate of the daughter products and resulting tissue exposure to ionizing radiation is a major concern that must be evaluated and the current status of RD studies in the context of TAT development are also discussed. Strategies for lead optimization via medicinal chemistry to improve the TAT radiopharmaceutical properties and methods of current good manufacturing practice (cGMP) production are discussed. Also included are brief reviews of preclinical and clinical TAT studies conducted to date.

Mechanism of Action/Tumor Cell Killing
Since the discovery of radioactive materials, the effect of radiation on the properties of matter has been of significant interest in the disciplines of material science, geology and astrophysics. For example, the first large-scale effect of radiation on solid material was observed by E.P. Wigner via exposure to a nuclear fission reactor, and that was thus named the Wigner effect, or "Wigner's disease" [1]. Since then, the large spread of medical technology involving sources of ionization radiation [2] and the development of nuclear weapons has caused a spike in studies on the effect of radiation on living tissue. That work increased our understanding of the concept of radiation risk and created new fields of scientific study; for example, radiation health safety, radiation dosimetry and radiation oncology.
Because of the high energy of α-particles and stochastic nature of ionization radiation, their effects may be observed on all levels of a biological system. Any molecule, cell, tissue or organ can display α-decay radiation damage, and such damage can be localized, or occur throughout the entire body of any multicellular organism [3].
The first step in producing radiation effects is the generation of a primary recoil atom and α-particles by a radioactive decay. Such events take place very rapidly in much less than 1 fs [4]. In the case of α-decay of radiopharmaceutical isotopes of interest, the average kinetic energy per recoil atom is~100 keV, and the average kinetic energy deposited within the range of a single α-particle is 5 MeV. It is clear that the relatively high levels of energy deposited by the combination of the fast-moving heavy ions and high energy α-particles can cause large amounts of damage to solid matter. There are many methods to estimate the effect of radiation in solid materials; e.g., the stopping theory based on coulombic interactions, molecular dynamics and transport theory. However, unlike solid materials, biological tissues do not consist of solid crystalline structures, and this significantly increases the complexity of estimating the effects of α-decay (as well as other ionization radiation), making the application of the aforementioned methods impossible or extremely difficult numerically. To overcome such limitations, a semi-quantitative approach has been applied. First, the biological effects observed in irradiated subjects were separated into one of two categories [3,6]: deterministic effects, which have a practical threshold absorption dose below which effects are negligible or not evident; and stochastic effects, where the relationship between dosage and severity of effect is either less evident or absent.
Maintaining the integrity of many different types of macromolecular structures is important to cell viability and all cellular organic molecules are subject to damage by ionizing radiation. However, the genomic DNA molecules are considered to be the most critical targets for the biological effects of ionizing radiation because intact DNA is required for cellular replication and damaged but repaired DNA can result in the fixation of genetic mutations that can affect normal cellular function and viability [18]. Ionizing radiation interacts with DNA either by directly transferring energy to the biological material or indirectly by creating reactive free radicals from the radiolysis of water. These interactions result in damage to the DNA's structure via broken covalent bonds. Linear energy transfer (LET) is an approach to describe the spatial distribution of ionization and excitation produced by direct or indirect effects of different types of radiation along a linear path [15]. Alpha (α) particles have high LET radiation because they create dense ionizations and excitations in matter due to coulombic interactions with atoms. Being a heavy charged particle, an α-particle will continuously slow down along its track with minimal deflection. Through the process of slowing down, the interaction cross-section towards the end of travel increases, resulting in increased LET, which is known as the Bragg Peak ( Figure 2) [19,20].
Relative biological effectiveness (RBE) is the ratio of the dose of a reference radiation and the dose of a test radiation that produces the same biological effect. Some of the most common biological effect measurements are necrotic and apoptotic (programmed) cell death, DNA damage, chromosomal aberrations and genetic mutations. The RBE of α-particles can range from 3.5 to 4 for cell killing or 6 to 12 for mutation, and be up to 10 for cell transformation [15]. As a comparison, the RBE for low LET electrons and photons is 1.
An important biological endpoint is cell killing. Damage to cells can be classified as either sub-lethal events or lethal events. Sub-lethal events are due to the accumulation of damage that has the potential to be repaired, typically as a result of exposure to lower doses, and lethal events typically result from irreparable damage due to exposure to higher doses [3]. The ability to repair these events is seen as a shoulder on the cell survival curve and is characteristic of low LET radiation ( Figure 3) [21]. A single event of high LET radiation can be lethal. The cell survival curve for a lethal event does not have a shoulder, indicating the inability to repair [22].
The DNA double-strand break is the most biologically significant type of damage, which occurs as a result of two single strand breaks in close proximity or a rupture of the double-strand at the site of interaction [23]. Cell survival is highly dependent on the spatial distribution of double-strand breaks [15]. Given the same dose, high LET radiation can create up to four times more double-strand breaks of low LET radiation. Additionally, the formation of high LET double-strand breaks are more complex compared to low LET in that they are less randomly distributed and form clustered DNA damage to multiple base-pairs [19]. It has been widely accepted that high oxygen levels play a large role in a cell's sensitivity to ionizing radiation, and hence, tumor hypoxia is an established factor in resistance to radiation therapy [24]. This is due to enhancement of free radical production by the presence of oxygen. Free radical production occurs as a result of indirect action, or low LET interactions. Since α-particles interact directly with the DNA, the level of oxygen becomes irrelevant; hence, eliminating a major mechanism of resistance to therapy [3]. produced by direct or indirect effects of different types of radiation along a linear path [15]. Alpha (α) particles have high LET radiation because they create dense ionizations and excitations in matter due to coulombic interactions with atoms. Being a heavy charged particle, an α-particle will continuously slow down along its track with minimal deflection. Through the process of slowing down, the interaction cross-section towards the end of travel increases, resulting in increased LET, which is known as the Bragg Peak ( Figure 2) [19,20]. LET of α-particles in water was calculated using stopping-power and range tables (continuous slowing down approximation range) for electrons, protons and helium ions from the National Institute of Standards and Technology (NIST). (B) The deposition of heavy ion energy as a function of penetrating depth of (a) a pristine beam and (b) a modulated beam with widened stopping region (spread out Bragg peaks). This figure and legend were reproduced from [19]. lethal events or lethal events. Sub-lethal events are due to the accumulation of damage that has the potential to be repaired, typically as a result of exposure to lower doses, and lethal events typically result from irreparable damage due to exposure to higher doses [3]. The ability to repair these events is seen as a shoulder on the cell survival curve and is characteristic of low LET radiation ( Figure 3) [21]. A single event of high LET radiation can be lethal. The cell survival curve for a lethal event does not have a shoulder, indicating the inability to repair [22].  [21].
The DNA double-strand break is the most biologically significant type of damage, which occurs as a result of two single strand breaks in close proximity or a rupture of the double-strand at the site of interaction [23]. Cell survival is highly dependent on the spatial distribution of double-strand breaks [15]. Given the same dose, high LET radiation can create up to four times more double-strand breaks of low LET radiation. Additionally, the formation of high LET double-strand breaks are more complex compared to low LET in that they are less randomly distributed and form clustered DNA damage to multiple base-pairs [19]. It has been widely accepted that high oxygen levels play a large role in a cell's sensitivity to ionizing radiation, and hence, tumor hypoxia is an established factor in resistance to radiation therapy [24]. This is due to enhancement of free radical production by the presence of oxygen. Free radical production occurs as a result of indirect action, or low LET After exposure to radiation that results in DNA damage, the cell cycle can be stopped at cell-cycle checkpoints which allow the cell to repair the damage via multiple repair mechanisms in order to preserve genomic integrity [23]. In the case of irreparable damage, the cell will eventually undergo cell death by apoptosis or necrosis. The two main repair mechanisms of double-strand breaks are homologous recombination and non-homologous end joining. Homologous recombination occurs in the late S and G2 phases of DNA synthesis where an intact DNA template is available, resulting in more efficient and higher-fidelity repair. Non-homologous end joining occurs throughout the cell cycle but is the only means of repair in G1 and early S phases. In this error prone repair method, DNA ends are rejoined with no sister templates [22]. In this case, chromosomal aberrations can occur as a result of recombining incorrect DNA ends; i.e., combining a loose end to some other molecule, and the truncation of ends. If incorrect repair occurs prior to DNA replication, these errors can be replicated in daughter DNA which can lead to mitotic cell death or can lead to the generation of genomic mutations without cell death [22]. There is also the situation where double-strand breaks are not repaired and the dividing cell enters mitosis, leading to mitotic catastrophe and eventual cell death. A higher proportion of double-strand breaks remain un-rejoined after exposure to high LET radiation [15]. When it comes to damage from high LET α-particles in close proximity to the cells being irradiated, the main radiobiological effect is complex and irreparable DNA damage resulting in cell death by either apoptosis or necrosis.
In the last few years, successful attempts have been made to explain the bystander effect [24,25]. The bystander effect is defined as a group of effects that are observed in cells that have not been directly irradiated following the irradiation of other nearby cells. Two mechanisms were proposed. One is the transfer of genomic instability through p53-mediated pathways, and the other suggests that irradiated cells secrete cytokines or other factors that transit to other cells that are not irradiated and signal for increased levels of intracellular reactive oxygen species [26]. One of the defined sub-classes of the bystander effect is the "abscopal effect," in which radiation treatment of a tumor propagates to tumors outside the irradiated volume [26]. A more recent publication demonstrated an effect that might explain the abscopal effect [27]. It was demonstrated that α-particle treatment of prostate cancer cells generated an adaptive antitumoral immune response, as has been previously reported for other forms of radiation. Combinations of bystander effects and the abscopal (likely immune) response in vivo are potential mechanisms of the efficacy for tumors that are not venerable to the targeted α-emitter radiotherapy in a patient with heterogeneous target expression.

Alpha-Particle Emitting Radionuclides
Using radiation as a method of cancer therapy requires delivering the maximum dose to the tumor while minimizing the dose to healthy tissues. Targeted radionuclide therapy is advantageous in that it seeks molecular and functional targets within patient tumor sites [28]. Beta (β)-emitting radionuclides ( 90 Y, 131 I, 177 Lu and 186 Re) are used for cancer-targeted therapy but have problems with cross-fire irradiation of normal tissues due to their relatively long range in tissue, which is 0.5-12 mm. In contrast, α-particles deposit higher energy over a much shorter range (40-90 µm), potentially causing higher cytotoxicity to tumor cells while delivering a lower dose to normal, adjacent tissues [29]. Alpha (α)-emission is the process by which an unstable nucleus ejects a highly-energetic, heavy, charged particle consisting of two protons and two neutrons. Alpha (α)-particles have a higher LET (100 keV/µm) compared to β-particles (0.2 keV/µm) which results in a dense ionization track in matter. The short range of α-emission provides specificity to the target cell population with a minimal effect on surrounding normal cells, and the high LET leads to a high frequency of irreparable DNA double-strand breaks [18]. This limits cytotoxic effects to within a small distance from the location of decay. It has been estimated that only one cell traversal by an α-particle track is necessary to kill a cell while, thousands of β-particle traversals are required for the same effect [30]. Because of the long range of β-particles and the need of a high number of hits for cell killing, a large portion of the dose deposited is outside of the intended target. In addition to treatment of solid tumors, the use of α-particles for targeted treatment of circulating disease could be an improvement due to the potentially reduced damage to normal tissues. Use of TAT has been considered for killing isolated cancer cells in transit in the vascular and lymphatic systems, in regressing tumors by disruption of tumor capillary networks and in treatment of micrometastatic foci [31]. In particular, TAT may be ideal for treatment of solid metastases, as the short range will primarily kill tumor cells, with little deleterious impact on surrounding normal tissue. Additionally, heterogeneity of target expression has been observed within a given tumor or metastasis [32] and this is thought to be a mechanism of developing resistance to targeted therapies; non-target expressing cells survive treatment and clonally expand into a resistant population. In this case, α or β-emissions from a targeted cell will serve to kill surrounding untargeted cells within the effective range, potentially reducing the development of resistance.
A number of factors need to be considered when choosing an α-emitting radionuclide for therapy. These include proper nuclear characteristics, ease of radiochemical incorporation, specific activity, synthesis yields, chemical and biological stability, availability and cost [29].
The physical half-life of the radionuclide should be long enough to allow for radiosynthesis preparation and be compatible with the pharmacokinetics of tumor localization [33]. The decay pathway of the α-emitter should be carefully analyzed. Due to the conservation of energy and linear momentum, a daughter nuclide which subsequently decays by α-emission could detach from the radioimmunoconjugate; see for example [34]. These free products could travel away and deposit doses to healthy tissues. A decay chain that is long and complicated, having many different decay types, could also present an issue dosimetrically, especially if the daughter products are metabolized differently than the parent. A possible way to overcome this issue is to use 225 Ac as an in vivo generator in which the delivery system is designed to be internalized into the target cell where the toxic daughter elements may detach from the targeting vector but remain trapped in the cell [35].
Another important nuclear characteristic is having a large number of α-emissions per decay. The radiotoxicity associated with having multiple emissions could be high enough to kill a tumor cell in a single decay. An accompanying gamma (γ) photon emission with energy suitable for in vivo imaging is also beneficial for assisting with pharmacokinetic and dosimetric evaluations [33].
Another important element for radionuclide selection is availability. Alpha (α)-emitters are produced either by cyclotron bombardment or by reactor irradiations, are incorporated into a generator and are eluted from a parental source. Obtaining radionuclides in pure form with high specific activity and large quantities is essential for adequate therapeutic evaluation. High specific activity is important to avoid receptor saturation by the unlabeled targeting agent [36]. If membrane antigenic density is low, insufficient binding to tumor cells will occur [36]. Transportation of these radionuclides safely and economically is also a key issue in selection.
While there are over 100 α-emitting radionuclides, only several have been investigated in preclinical and clinical studies. This is mostly due to radionuclides lacking nuclear properties, the absence of viable chemistry, complicated decay chains and production and economic issues [37]. Therefore, radionuclides meeting the criteria for therapeutic use have been limited to 225 Ac, 211 At, 212 Bi, 213 Bi, 212 Pb, 223 Ra, 224 Ra, 149 Tb and 227 Th. The physical characteristics of these isotopes can be seen in Table 1. The first α-emitter to be used in human clinical trials for therapy was 213 Bi in 1997, when it was labeled to the anti-leukemia antibody HuM195 [38]. 213 Bi is available through generator based 225 Ac and decays with a 45.6 min half-life by emission of one α (8.37 MeV) and two β-particles. The generator is produced at Oak Ridge National Laboratory in the US and at the Institute for Transuranium Elements in Karlsruhe Europe. In the decay of 213 Bi, there is an emission of a 440 keV isomeric γ, which is beneficial for imaging studies. 211 At decays with a half-life of 7.2 h and emits two α-particles through a split decay pathway with energies of 5.87 and 7.45 MeV. One path is to 207 Bi by α-emission followed by electron capture to 207 Pb and the other is by electron capture to 211 Po followed by α-emission to 207 Pb. An advantage of this decay path is that 211 Po emits 77-92 keV characteristic X-rays which can be used for imaging [37]. The main disadvantages are availability and purity. Conventionally, the production of 211 At requires α-particle cyclotrons, which are only available in a few places worldwide, to produce the 209 Bi(α,2n) 211 At reaction with minimal 210 At contamination [39]. Astatine has significant metallic characteristics that lead to complications in standard antibody labeling and results in a rapid release of free 211 At [37,40,41].
To resolve this problem, approaches have been developed by several research groups based on small linker molecules that create an aryl carbon-astatine bond involving an astatodemetallation reaction using tin, silicon or mercury precursors [42]. Other alternative methods for astatination, e.g., boron-astatine, rhodium-astatine or nanoparticle encapsulation, are also being pursued [43]. 225 Ac is a radiometal with a half-life of 10.1 days and produces six radionuclide daughters in the decay path to stable 209 Bi. For each decay event of 225 Ac, there are successively, four α and two β emissions with high energy (α 8.38 MeV, β 1.42 MeV). In the decay of 225 Ac and its daughters there are several isomeric γ emissions with energy suitable for imaging studies. The relatively long half-life allows for a centralized production site that can ship 225 Ac to users [44]. The main method for generating 225 Ac for clinical studies is through the decay of 229 Th which originates from 233 U. In the world there are three main sources of 229 Th: Oak Ridge National Laboratory (USA), The Institute of Physics and Power Engineering (Russia) and The Institute for Transuranium Elements (Germany). The quantities produced are not enough for a global application of 225 Ac. To keep up with the increasing demand for 225 Ac for clinical applications, it has been found that large scale quantities can be produced through high-energy proton irradiation of 232 Th [45,46]. To address the shortage, the US Department of Energy formed a Tri-lab collaboration of Los Alamos (LANL), Brookhaven (BNL) and Oak Ridge (ORNL) National Laboratories with the goal of developing an alternative route for production of Ac [47]. Another limitation for this radionuclide has been with the radiochemical stability of the attachment to immunoconjugate. McDevitt et al. [34], state that the instability of these attachments is due to the high classical recoil energy (100-200 keV) of the daughter product which breaks the molecular bonds of the chelator. Significant advances have been made in developing chelators that form thermodynamically stable and kinetically inert complexes with 225 Ac. Khabibullin et al., recently calculated the chelation stability of 225 Ac and daughters in the 1,4,7,10-tetraazacyclododecane-1,4,7,10-tetraacetic acid (DOTA) chelator [48]. As one potentially abundant starting material, 212 Bi (1.01 h half-life) can be obtained from 228 Th and decays via a branched pathway to 208 Tl (36% α) and 212 Po (64% β); then, both decay to stable 208 Pb [33]. However, 212 Bi has several disadvantages that potentially limit its use. First, its short half-life can be problematic if the production and shipping processes are lengthy. This issue can be solved by using 224 Ra as a generator to locally produce 212 Bi. Another complication is the high energy γ emission (2.6 MeV) which requires a considerable amount of shielding to minimize exposure. This, along with the short half-life, makes shipping difficult, resulting in an availability problem. 223 Ra is found naturally in trace amounts following the decay of 235 U, but it is mainly produced artificially by the decay of 227 Th (T 1/2 = 18.7 days), which is produced from 227 Ac (T 1/2 = 21.77 years). Since 227 Ac is found only in traces in uranium and thorium ores, it is mainly synthesized by 226 Ra (T 1/2 = 1600 years) irradiation in a nuclear reactor [49]. 223 Ra has a half-life of 11.4 days and emits four α-particles, two β-particles and γ rays on the path to stable 207 Pb [34]. While the emission of four α-particles is advantageous for tumor toxicity, 219 Rn in gaseous form is also emitted, which can redistribute in the body and dose non-targeted cells. See Figure 3 for comparable energies of α-particles.
The major challenge using 223 Ra is finding a suitable ligand for in vivo sequestration. However, similarly to cations of the alkaline earth elements, radium has natural bone-seeking properties without the need for a carrier agent. As radium mimics calcium, when 223 Ra dichloride (Xofigo ® ) is injected intravenously, it forms complexes with the bone mineral hydroxyapatite at areas of increased metabolic bone activity, such as bone metastases, thereby exerting a highly localized antitumoral effect [7]. In addition, the 223 Ra daughter isotopes are also retained in the bone matrix [50].
Thanks to this characteristic, 223 Ra dichloride ( 223 RaCl 2 ) has demonstrated improvement in overall survival for the treatment of bone metastases in castration-resistant prostate cancer and has received FDA approval for this application [34], making it the first FDA approved TAT. 227 Th can be produced continuously from 227 Ac and decays with a half-life of 18.7 days to 223 Ra [51]. The long half-life is beneficial for radiolabeling and targeting. 227 Th also decays with accompanying γ emissions of 236 keV and 50 keV, which can be used for imaging but are low enough to avoid the need for patient shielding. 212 Pb has a 10.6 h half-life and is produced either from the decay chain of 228 Th or by the 224 Ra generator [39]. Because of its long half-life and the fact that 212 Pb is a β-emitter that decays to 212 Bi, one approach has been to use this radionuclide as an in vivo generator to compensate for the short half-life of 212 Bi [52]. 212 Pb can deliver over ten times the dose per unit of administered activity compared to 212 Bi or 213 Bi [53]. The major issue with 212 Pb is the electron capture and auger electron emissions which can cause significant recoil of the 212 Bi daughter [54]. The free 212 Pb has been shown to cause severe bone marrow toxicity [55], while the free Bi has shown to cause kidney toxicity. By co-injecting DTPA or EDTA chelating agents, rapid release of free 212 Bi can be achieved [56].

Targeting Moieties
In the burgeoning field of targeted radiopharmaceuticals, there is no universal answer to the question: what is the best targeting molecule? Rather, the abundance of target molecule classifications allow for a more customized approach to developing TATs. Globally, the requirements of a targeting ligand include the ability to concentrate at and bind to extracellular targets and the availability of chemical functional groups amenable to the attachment of linkers and chelators. A balance of many other factors like off-target binding, biodistribution and pharmacokinetics (particularly with respect to the decaying half-life of the chosen radionuclide) are all critical factors for the selection of a proper targeting ligand. Recently, small molecules, peptides, antibodies, antibody fragments and even some passive targeting strategies have been investigated to deliver radioisotope payloads. See Tables 2 and 3 for examples of TAT conjugates that have been studied in the pre-clinical and clinical environments. This section will explore the advantages and challenges of the aforementioned targeting ligand categories.

Small Molecules
An advantage of using low molecular weight ligands as targeting moieties for α-therapy include rapid penetration into the tumor and rapid clearance of unbound conjugates from circulation; thus, reducing toxicity. Internalizing ligands are particularly beneficial for 225 Ac's application in order to harness the multiple α-particles emitted in its decay chain. A successful example of using small molecules in TAT is 225 Ac-PSMA-617 for the therapy of metastatic, castration-resistant prostate cancer [57,58]. The favorable pharmacokinetics properties of the PSMA-617 small molecule include its fast tumor uptake, high internalization rate, extended tumor retention and rapid clearance of unbound ligand [58]. These features make the molecule highly desirable for labeling with an α-emitter with a half-life of several days and multiple α-emissions in its decay chain, like 225 Ac. The first report of the significant therapeutic efficacy of 225 Ac-PSMA-617 involved two patients with late-stage, metastatic, castration-resistant prostate cancer (mCRPC) and complete remissions [59]. Additional clinical studies have further revealed the remarkable anti-tumor activity and the promising duration of tumor-control of treatment with 225 Ac-PSMA-617 [6,59,60]. These studies have further demonstrated the significant potential of TAT (Table 3, Figure 1). Xerostomia (dry mouth syndrome) was the main side effect, indicating that further modifications in clinical trial design might be necessary to further enhance the therapeutic range.

Peptides
Peptides are oligomers of amino acids that may sometimes exhibit secondary structure; they may have branched or linear frameworks, and may be composed of varying amounts of non-canonical monomers. The polypeptide chains of peptides can have anywhere from 2 to 70 amino acids but more typical examples of targeting peptides are made of less than 10-15 amino acids (1000-1500 MW). Owing to this molecular weight and their capacity to be synthesized and modified with conventional organic synthesis techniques, peptides have long been utilized as targeting agents for radionuclide therapies and diagnostic applications.
Since the 1980s, analogs of the endogenous peptide hormone somatostatin have been developed as therapeutics for neuroendocrine disorders [61]. The FDA approved octreotide (OC, Figure 4), the cyclic octapeptide, upon which much of the early peptide receptor radionuclide therapy (PRRT) was based. Chelating molecules attached and various radionuclides for β and α-emission therapies have been reported. Figure 4 details the structure of OC and two of its commonly-used PRRT ligand analogs, DOTATOC and DOTATATE.
Nephrotoxicity is often one of the highest concerns in PRRT due to tubular reabsorption of the labeled peptide after glomerular filtration [62]. Seventy-four human patients were followed (>1 year) after PRRT with the β-emitter 177 Lu-Octreotate. Renal function was monitored by accessing glomerular filtration rate (GFR). Somewhat surprisingly, renal impairment was minimal, with only 43% of patients experiencing a decrease in GFR (<2 mL/min/m2 per year) and 11 patients actually saw increased GFR (>10 mL/min/m2 per year) [63]. The first published account of the α-emitter 213 Bi in an OC analog (DOTATOC, Figure 4) was reported in 2006. The radio-peptide ligand was shown to retain its affinity for the somatostatin receptor and inhibited tumor growth in a somatostatin receptor-positive rat pancreatic tumor model (CA20948) [64]. Importantly, the study followed major organ toxicities in rodents. Little to no nephrotoxicity was observed at the various administered activities (13-22.2 Mbq).
More recently, even tumors that had become radioresistant to β-therapy ( 90 Y and 177 Lu-DOTATOC) responded to 213 Bi-DOTATOC in a clinical model of eight human patients [65]. The treatment for each patient was individualized according to each particular disease state and each patient had substantially positive outcomes in terms of tumor regression and survival. Critically, the acute hematological toxicity normally associated with the analogous β-therapies was only moderate. This limited report suggested only a mild reduction in acute renal function. No other major acute toxicities were reported. This first-in-human report of this peptide-targeted alpha therapeutic may lay the ground work for future human TAT using peptide-targeted systems. More recently, even tumors that had become radioresistant to β-therapy ( 90 Y and 177 Lu-DOTATOC) responded to 213 Bi-DOTATOC in a clinical model of eight human patients [65]. The treatment for each patient was individualized according to each particular disease state and each patient had substantially positive outcomes in terms of tumor regression and survival. Critically, the acute hematological toxicity normally associated with the analogous β-therapies was only moderate. This limited report suggested only a mild reduction in acute renal function. No other major acute toxicities were reported. This first-in-human report of this peptide-targeted alpha therapeutic may lay the ground work for future human TAT using peptide-targeted systems.
It has been the common supposition that TAT would work well for small, disseminated tumors but not as well for large, solid tumors. Recently, DOTATATE ( Figure 4) was labeled with 213 Bi and used to investigate efficacy differences in tumor size among two tumor models, rat pancreatic (CA20948) and human small cell carcinoma (H69) with each containing a small (50 mm 3 ) and large (200 mm 3 ) tumor group [66]. Both tumor lines maintained high expression of SSTR2 and all tumor groups saw delayed tumor growth and higher survival over controls. Three mice from the smaller tumor groups were effectively put into remission through the end of the study. This report has implications for TAT in regard to tumor receptor heterogeneity, tumor size and perfusion. More recently, α-emitting radionuclides have been conjugated to melanocortin 1 receptor targeting peptides and efficacy determined in the pre-clinical treatment of mouse xenograft models of cutaneous and uveal melanomas [67][68][69].

Antibodies
Full length immunoglobulins (IgGs) are typically in the 150 kDa molecular weight range, and have been proven to have high binding affinity and specificity to a broad range of extracellular receptors. Developments in hybridoma cell line technology have opened the door to the production of monoclonal antibodies (mAbs) which can be labeled with chelating molecules, to which It has been the common supposition that TAT would work well for small, disseminated tumors but not as well for large, solid tumors. Recently, DOTATATE ( Figure 4) was labeled with 213 Bi and used to investigate efficacy differences in tumor size among two tumor models, rat pancreatic (CA20948) and human small cell carcinoma (H69) with each containing a small (50 mm 3 ) and large (200 mm 3 ) tumor group [66]. Both tumor lines maintained high expression of SSTR 2 and all tumor groups saw delayed tumor growth and higher survival over controls. Three mice from the smaller tumor groups were effectively put into remission through the end of the study. This report has implications for TAT in regard to tumor receptor heterogeneity, tumor size and perfusion. More recently, α-emitting radionuclides have been conjugated to melanocortin 1 receptor targeting peptides and efficacy determined in the pre-clinical treatment of mouse xenograft models of cutaneous and uveal melanomas [67][68][69].

Antibodies
Full length immunoglobulins (IgGs) are typically in the 150 kDa molecular weight range, and have been proven to have high binding affinity and specificity to a broad range of extracellular receptors. Developments in hybridoma cell line technology have opened the door to the production of monoclonal antibodies (mAbs) which can be labeled with chelating molecules, to which radionuclides can be added. This approach to specifically deliver ionizing radiation payloads is termed radioimmunotherapy (RIT). While many of the examples here will involve radioactive payloads directly conjugated to the protein, antibodies have also been employed to target macromolecular payloads, such as nanoparticles and liposomes, to cellular targets [70][71][72].
A research group based in the Memorial Sloan-Kettering Cancer Center (MSKCC) has reported several accounts of their work labeling trastuzumab with actinium 225. This mAb is the well-known and FDA approved HER2/ERRB2-targeting agent. Initially, they showed through a three cell line, spheroid in vitro model that their α-RIT scheme could penetrate spheroids, retard growth and prevent regrowth of colonies in a dose dependent manner [73]. While a promising start, this work underscored the importance of target expression and also suggested challenges of RIT due to extravasation of targeting agents in normal tissues and toxicities of released/free decaying daughter products [74,75].
The MSKCC Scheinberg group's work eventually led to the first clinical trial of an actinium 225 chelated targeting antibody (Clinical Trials.gov Identifier: NCT00672165). This α-RIT scheme utilized the previously-explored humanized anti CD33 antibody, limtuzumab, to target acute myeloid leukemia cells. Subsequently, that preliminary work led to the birth of Actinium Pharmaceuticals, Inc. and a portfolio of targeted 225 Ac conjugated constructs in both preclinical and clinical pipelines.
Astatine-211 has also been utilized for α-RIT. In 2013, Orozco and coworkers coupled a decaborate cage structure (B10, Figure 5) with 211 At to anti-CD45 antibody in an attempt to target acute myeloid leukemia [76]. More recently Green and coworkers reported an example using anti-CD20 mAb conjugated with a 211 At with a similar B10 labeling scheme [77]. This study sought to eliminate minimal residual disease (MRD) in a mantle cell lymphoma animal model. Their remarkable results showed 70% eradication of the MRD in animals bearing a disseminated model. However, the subcutaneous lymphoma xenograft group treated with the same agents only saw prolonged survival (two to three-fold) without any cured animals. This study is clear evidence of the need for tumor perfusion of the TAT.
Molecules 2019, 24 FOR PEER REVIEW 12 and FDA approved HER2/ERRB2-targeting agent. Initially, they showed through a three cell line, spheroid in vitro model that their α-RIT scheme could penetrate spheroids, retard growth and prevent regrowth of colonies in a dose dependent manner [73]. While a promising start, this work underscored the importance of target expression and also suggested challenges of RIT due to extravasation of targeting agents in normal tissues and toxicities of released/free decaying daughter products [74,75]. The MSKCC Scheinberg group's work eventually led to the first clinical trial of an actinium 225 chelated targeting antibody (Clinical Trials.gov Identifier: NCT00672165). This α-RIT scheme utilized the previously-explored humanized anti CD33 antibody, limtuzumab, to target acute myeloid leukemia cells. Subsequently, that preliminary work led to the birth of Actinium Pharmaceuticals, Inc. and a portfolio of targeted 225 Ac conjugated constructs in both preclinical and clinical pipelines.
Astatine-211 has also been utilized for α-RIT. In 2013, Orozco and coworkers coupled a decaborate cage structure (B10, Figure 5) with 211 At to anti-CD45 antibody in an attempt to target acute myeloid leukemia [76]. More recently Green and coworkers reported an example using anti-CD20 mAb conjugated with a 211 At with a similar B10 labeling scheme [77]. This study sought to eliminate minimal residual disease (MRD) in a mantle cell lymphoma animal model. Their remarkable results showed 70% eradication of the MRD in animals bearing a disseminated model. However, the subcutaneous lymphoma xenograft group treated with the same agents only saw prolonged survival (two to three-fold) without any cured animals. This study is clear evidence of the

Antibody Fragments
A limitation of full-size IgGs is their typical 1-3 week serum stability. While this kind of circulation may be advantageous for certain therapeutic applications, for radioimmunotherapy (RIT) it presents a serious liability. Excess antibody can continue to circulate, lowering the tumor-tonontumor (T/NT) ratio, particularly with respect to the tumor-to-blood (T/B) ratio. Finally, high levels of continuously-circulating α-emitters result in hematological toxicities, as well as extravasation in normal tissues.
In attempts to shorten the long plasma half-life of full-length antibodies, many groups have sought to reduce the sizes of the mAbs while retaining their remarkable binding characteristics. To those ends, antibodies have been cleaved into smaller sections through enzymatic digestion and also engineered ab initio [78,79]. For comparison, the engineered antibody fragments of diabodies (Db) and minibodies (Mb) have typical circulating half-lives of 2-5 and 5-12 h, respectively, while the circulating half-life of full-length IgGs is more typically 1-3 weeks. Monovalent and bivalent

Antibody Fragments
A limitation of full-size IgGs is their typical 1-3 week serum stability. While this kind of circulation may be advantageous for certain therapeutic applications, for radioimmunotherapy (RIT) it presents a serious liability. Excess antibody can continue to circulate, lowering the tumor-to-nontumor (T/NT) ratio, particularly with respect to the tumor-to-blood (T/B) ratio. Finally, high levels of continuously-circulating α-emitters result in hematological toxicities, as well as extravasation in normal tissues.
In attempts to shorten the long plasma half-life of full-length antibodies, many groups have sought to reduce the sizes of the mAbs while retaining their remarkable binding characteristics. To those ends, antibodies have been cleaved into smaller sections through enzymatic digestion and also engineered ab initio [78,79]. For comparison, the engineered antibody fragments of diabodies (Db) and minibodies (Mb) have typical circulating half-lives of 2-5 and 5-12 h, respectively, while the circulating half-life of full-length IgGs is more typically 1-3 weeks. Monovalent and bivalent affibodies (7 and 15 kDa respectively) have been developed with targeting specificity to HER-2 and they have plasma half-lives of roughly 45 min [80]. The science of antibodies is ever progressing and new classes are still being observed in nature; nanobodies (13-14 kDa) are heavy chain-only antibodies, that have been isolated from Camelidae and have lately been conjugated with theranostic radionuclides [81]. These new categories of targeting constructs are adding to the growing body of evidence that there may be a so-called 'Goldilocks' zone in terms of the size of targeting constructs.

Passive Targeting
Another strategy often employed for TAT is based on the natural selective uptake and accumulation of certain elements. For instance, the actively restructuring environment of osteoblastic bone metastases is conducive to calcium accumulation in the form of hydroxyapatite. Due to the resemblance of radium-223 to calcium, it can act as a surrogate and be incorporated into the rapidly-growing matrix [7,29]. Indeed, the recent approval of 223 Ra dichloride therapies for metastatic castrate-resistant prostate cancer has been followed in patients for over twelve months with moderately successful outcomes [82]. Other studies have looked to 223 Ra dichloride for the treatment of the bone metastases from hormone-refractory breast cancer [83,84].
The enhanced permeability and retention (EPR) effect is often regarded as a rationale for the passive targeting of large macromolecules like nanoparticles and liposomes. This accumulation of macromolecules through the leaky vasculature of quickly-dividing tumor cells also provides the opportunity to incorporate multiple radionuclides into the framework of a targeting molecule. Additionally, it has been proposed that nanoparticles or liposomes could be used to sequester decaying daughter products that would otherwise be released to freely circulate in the blood because of damage to molecular bonds due to α-decay or metabolism of the radiopharmaceutical [85,86].

Radiosynthesis
The recent approval of 223 RaCl 2 has reinvigorated research into TAT development [87]. Several α-particle-emitting radionuclides have been described in the literature, and these include 211 At, 212/213 Bi, 212 Pb and 225 Ac [44,88,89]. With the exception of 211 At [41,42,90], all are radiometals that require a bifunctional chelator for the attachment of the therapeutic radioisotope to the targeting moiety. The chemistries of 212/213 Bi and 212 Pb have been covered extensively elsewhere [91][92][93][94]. The radiochemistry of radium has recently been reviewed [95]. This section will cover recent advances in 225 Ac chelation chemistry that have occurred since the publication of the excellent review by Scheinberg and McDevitt in 2011 [96]. 225 Ac, whose decay chain was not finally resolved until the middle of the 20th century is one of 29 radioactive isotopes of actinium, which is the first element of the actinide series [97]. Although the lack of non-radioactive actinium isotopes has hindered the exploration of 225 Ac chemistry, the pioneering work of Seaborg and others has provided an important understanding of the basic reaction chemistry of this element [97][98][99]. Actinium chemistry has been observed to closely follow that of lanthanum chemistry, since both exist as trivalent ions in solution. These observations have led to the belief that La 3+ can be an important non-radioactive surrogate for Ac 3+ , with the caveat that the differences in ionic radii must be considered carefully [100].
Despite the relatively limited knowledge of actinium chemistry, the medical community has seized upon its therapeutic potential in TAT. Using lanthanide chemistry as a guide, many attempts to attach this radioisotope to targeting molecules have been reported with a variety of acyclic and cyclic ligands that would effectively chelate this radioisotope [101][102][103][104][105][106]. To date, however, only 1,4,7,10-tetraazacyclododecane-1,4,7,10-tetraacetic acid (DOTA) has found widespread acceptance as an 225 Ac chelator in both preclinical and clinical applications [96], despite the low radiochemical yields and specific activities that have been reported [73,104,107]. Maguire et al. have described an improved, one-step 225 Ac radiolabeling procedure of monoclonal antibodies that resulted in a 10-fold increase in radiolabeling yield and a 30-fold increase in specific activity [108]. The authors concluded that the improvements in yield and specific activity resulted from greater control over the DOTA-mAb conjugation reaction and the improved chelation of 225 Ac during radiosynthesis. Additionally, the new procedure did not alter antibody activity or therapeutic efficacy in vivo, which further suggests the usefulness of this strategy. More recently, Poty et al. reported a single-step click chemistry approach to 225 Ac-radioimmunoconjugate synthesis [109]. The click approach generated superior radiochemical yields compared to the standard two-step strategy, suitable specific activities, good in vitro stability and excellent biodistribution profiles. The authors proposed that said novel click strategy could be applied when the targeting vector is unstable under the conditions used for the standard method.
Despite improving the radiochemical yield and specific activity of 225 Ac radiopharmaceuticals, important challenges, which include minimizing the detrimental effects of recoil energy and effectively sequestering the cytotoxic daughter products 221 Fr and 213 Bi after 225 Ac decay, remain to be overcome. While several reports have described cellular internalization as an effective method of sequestering 225 Ac and its daughter products [35], scientists are increasingly turning to nanotechnology to resolve this issue, and a variety of nanoparticles, including liposomes and carbon-based nanoparticles, have been explored [96,[110][111][112][113][114][115]. Recent reports have demonstrated that LaPO 4 nanoparticles hold promise as safe and effective nanocarriers for 225 Ac [71]. In these studies, the authors synthesized LaPO 4 (monazite) nanoparticles doped with 225 Ac. These particles were prepared with functionalized surfaces that facilitated the attachment of 201B mAb, which targets murine thrombomodulin on lung epithelium. Upon in vivo injection, targeted particles accumulated rapidly in the lung through specific interactions between the 201B mAb and its antigen. Moreover, retention of 225 Ac and half of the 221 Fr and 213 Bi daughter nuclei within the particles demonstrated the potential for trapping radionuclides after high energy emissions. McLaughlin et al. improved upon this work by examining 225 Ac-doped LaPO 4 nanoparticles that were encapsulated by GdPO 4 and Au shells [86]. These particles had an average diameter of 27 nm, and were easily conjugated to the 201b mAb using standard bioconjugation techniques. Biodistribution and small animal single-photon emission computed tomography (SPECT) studies revealed the specific targeting of murine thrombomodulin within the lung epithelium that was reduced upon the administration of non-radioactive 201b mAb. Interestingly, the multi-layered nanoparticle design seemed to increase the retention of 225 Ac, 221 Fr and 213 Bi within the particle, which reduced the radiation dose experienced by the kidney. Approximately four and five-fold reductions were observed at 1 and 24 h post-injection, respectively. These data demonstrate that multi-functional, layered nanoparticles have the potential to deliver and retain 225 Ac and its daughter radionuclides at the target site, while minimizing off-target toxicities that can occur from errant daughter products in the 225 Ac decay chain.

Linkers/Rational Design
The most utilized chelating molecules for targeted radionuclide therapies using biological molecules are 1,4,7,10-tetraazacyclododecane-1,4,7,10-tetraacetic acid (DOTA) and diethylene triamine pentaacetic acid (DTPA), and their related analogs ( Figure 5). Depending on the how the compound is linked to the targeting ligand, these organic structures feature 3 or 4 carboxylates that are negatively charged at physiological pH, and along with the lone pair electrons from each of the 3 or 4 nitrogens, they can coordinate metal ions like α and β-emitting radionuclides. Often, these chelating molecules are linked to the targeting molecule by forming a new amide bond between an amine on the targeting molecule and a carboxylate on the DOTA/DTPA. This strategy is particularly advantageous in the case of peptide synthesis since the reaction, characterization and purification of the linker addition can be part of the overall synthesis of the targeting ligand.
When DOTA or DTPA are linked to proteins like antibodies, amide bonds are typically created between a primary amine of surface-exposed lysine residue and an activated carboxylate on the DOTA or DTPA (Figure 6). Another widespread chemistry available to link protein lysines with DOTA or DTPA is through an isothiocyanate which yields a stable thiourea ( Figure 6) [53,116]. While these newly formed amide bonds create the same covalent structure as the peptide-synthesized version, there are important distinctions. In the stochastic reaction of any accessible lysine, many different combinations are possible, and there is little to no control over the regioselectivity of the conjugation. It is unlikely, but possible, that targeting ligands could block or hinder the antigen-binding sites on the antibody.
Molecules 2019, 24 FOR PEER REVIEW 15 different combinations are possible, and there is little to no control over the regioselectivity of the conjugation. It is unlikely, but possible, that targeting ligands could block or hinder the Figure 6. Common conjugation chemistries used to functionalize biomolecules. Countless other schemes have been well characterized and the reader is directed to Bioconjugates Techniques for a review and protocols [117].
The importance of site-specific modifications to biologic targeting motifs is only recently becoming clear. In a 2014 paper from UCLA [118], diabodies (Dbs) were conjugated both sitespecifically through reduced cysteines and non-specifically through accessible lysine ε-amines to DOTA chelators. While the tumor-to-blood ratio of the specifically labeled protein was moderately higher than the more heterogeneous product, the more striking result was the renal and hepatic distribution. Kidney uptake levels were almost doubled for the cysteine-labeled Db, and liver uptake levels were reduced for the non-specific amine-labeled Dbs.
Other important research regarding linkers has uncovered different methods to reduce kidney toxicity, via the renal reabsorption of radiolabeled peptides and antibody fragments, as they are filtered by the glomerulus. By taking advantage of renal brush border enzymes, the radionuclide can be cleaved from targeting ligand and excreted. By engineering an antibody fragment with a Cterminal lysine, and subsequently modifying the ε-amine with DOTA:indium-111, Li and coworkers demonstrated a 50%-60% reduction in the kidney-uptake of the radionuclide [119]. More recently, a Japanese group probed the brush border enzymes to understand more about their specificity and they were able to pinpoint a glycine-tyrosine linkage that specifically cleaved a radio-iodinated benzoate derivative from an antibody Fab fragment [120]. Presumably, a similar strategy could be used for the α-emitting halogen, 211 At.
Polyethylene glycol (PEG) linkers are made of repeating -CH2CH2Omonomers and are commonly employed by chemists to alter biodistribution and pharmacokinetics. PEGs can be prepared in two categories, polymeric and discrete oligomers. Bifunctionalized versions of these polymers can link targeting molecules with the chelating agent or just add bulk/solubility to lower molecular weight entities. Researchers wishing to reduce immunologic response to nanoparticles have been known to decorate their macromolecules with PEG [70], as was the case with McLaughlin and coworkers with PEG12 linker used to link 225 Ac containing nanoparticles to mAb [86].

Radiation Dosimetry
Dosimetry provides means for evaluating the efficacy of a radiation therapy modality [3]. Because the goal is to a deliver high dose to tumor cells with a minimal dose to normal tissues, it is of great importance to quantify accurately where, when and how a dose is being deposited. The Figure 6. Common conjugation chemistries used to functionalize biomolecules. Countless other schemes have been well characterized and the reader is directed to Bioconjugates Techniques for a review and protocols [117].
The importance of site-specific modifications to biologic targeting motifs is only recently becoming clear. In a 2014 paper from UCLA [118], diabodies (Dbs) were conjugated both site-specifically through reduced cysteines and non-specifically through accessible lysine ε-amines to DOTA chelators. While the tumor-to-blood ratio of the specifically labeled protein was moderately higher than the more heterogeneous product, the more striking result was the renal and hepatic distribution. Kidney uptake levels were almost doubled for the cysteine-labeled Db, and liver uptake levels were reduced for the non-specific amine-labeled Dbs.
Other important research regarding linkers has uncovered different methods to reduce kidney toxicity, via the renal reabsorption of radiolabeled peptides and antibody fragments, as they are filtered by the glomerulus. By taking advantage of renal brush border enzymes, the radionuclide can be cleaved from targeting ligand and excreted. By engineering an antibody fragment with a C-terminal lysine, and subsequently modifying the ε-amine with DOTA:indium-111, Li and coworkers demonstrated a 50%-60% reduction in the kidney-uptake of the radionuclide [119]. More recently, a Japanese group probed the brush border enzymes to understand more about their specificity and they were able to pinpoint a glycine-tyrosine linkage that specifically cleaved a radio-iodinated benzoate derivative from an antibody Fab fragment [120]. Presumably, a similar strategy could be used for the α-emitting halogen, 211 At.
Polyethylene glycol (PEG) linkers are made of repeating -CH 2 CH 2 O-monomers and are commonly employed by chemists to alter biodistribution and pharmacokinetics. PEGs can be prepared in two categories, polymeric and discrete oligomers. Bifunctionalized versions of these polymers can link targeting molecules with the chelating agent or just add bulk/solubility to lower molecular weight entities. Researchers wishing to reduce immunologic response to nanoparticles have been known to decorate their macromolecules with PEG [70], as was the case with McLaughlin and coworkers with PEG12 linker used to link 225 Ac containing nanoparticles to mAb [86].

Radiation Dosimetry
Dosimetry provides means for evaluating the efficacy of a radiation therapy modality [3]. Because the goal is to a deliver high dose to tumor cells with a minimal dose to normal tissues, it is of great importance to quantify accurately where, when and how a dose is being deposited. The Medical Internal Radiation Dose Committee (MIRD) developed an approach to determine the average absorbed dose from internal radionuclides [121]. The absorbed dose method accounts for low LET radiation where thousands of statistically independent deposition events across a cell are required to induce a biological effect [3]. With respect to an α-particle's interactions, the averaged absorbed dose will fail to characterize biological outcome. Because the α-particle range is only a few cell lengths, some cell nuclei will receive multiple traversals while others will receive zero traversals. Additionally, the location of the α-particle track will determine the energy absorption by the cell. These factors result in large statistical variability of energy deposition and stochastic effects become important. To account for those effects, Roeske developed a microdosimetric methodology based on the mean dose to target cells, probability distribution of specific energy absorbed by target cells, and the fraction of cells receiving zero hits [122]. Kellerer and Chmelevsky described the requirement that microdosimetry must be used when the relative deviation of local dose is greater than 20% [122]. Following this criterion, microdosimetric techniques are more important in analyzing non targeted cells where the local average dose is small, resulting in larger variations than targeted cells.
In most cases, the activity level and mean absorbed doses to the target are large, resulting in a low expected stochastic deviation, and therefore, there is no need to use microdosimetric techniques. The MIRD Pamphlet No. 22: Radiobiology and Dosimetry of Alpha-Particle Emitters for Targeted Radionuclide Therapy recommends using the conventional MIRD formalism [3]. The mean absorbed dose to the target region, r T , from a source region, r S , due to a particular emission type, x, over a dose integration period, T D , is given by: where A(r S , T D ) is the total number of nuclear transitions in the target region; ∆ x i is the mean energy emitted per disintegration for the i-th emission of type x; ϕ r T ← r S ; E x i is the fraction of energy emitted per nuclear transition in the source region that is absorbed in the target region by the i-th emission of type x that is emitted with initial energy E; and M(r T ) is the mass of the target region. The total number of nuclear transitions is determined by measuring activity levels in tissue samples at several points in time post administration. The activity levels plotted against time gives a time-activity curve which is integrated to obtain the total number of nuclear transitions. The mean energy emitted per emission ∆ i is a physical property of the radionuclide and can be obtained from nuclear decay tables. The absorbed fraction for each decay type is dependent on the reference phantom geometry and obtained through Monte Carlo calculation. The dose contribution from each emission type is then weighted by its relative biological effectiveness, and then summed: Several assumptions are made when using this formalism of dose calculation. It is assumed that the activity is uniformly distributed in the organ and that α-energy deposition is also uniformly distributed in the organ. It is also assumed that dose from both α and β-emissions are locally deposited. The calculation of the absorbed fractions is based on idealized phantoms that have standard geometries which cannot account for the individual anatomy of a patient. It is also important to note that radioactive daughters are not taken into account. Therefore, daughter decays must also be calculated, as well as accounting for daughters' biodistribution. The formalism was developed as an adequate method for the dosimetry of internal radionuclides used for diagnostic purposes and lacks the accuracy needed for therapeutic applications.
As a first method to improve on the original MIRD formalism, phantoms were developed to better match the standard human anatomy. For example, Christy and Eckerman developed phantoms that represented a male, a female and children [123]. Later, as computational power increased, voxelized phantoms like VoxelMan were created based off 3D imaging [124]. MIRD Pamphlet No. 17 describes a method to extend the S value formalism to the voxel level to account for nonuniform distributions of activity [125]. The most extensive software that has adapted the MIRD S factors is OLINDA/EXM, which has calculated an internal dose for over 200 radionuclides, including α-emitters, in 10 different phantoms [126].
The voxel S method is still a model-based approach to dosimetry. There is much less tolerance for inaccuracies in therapeutic applications, which calls upon the need for patient-specific dosimetry. In order to accomplish this, the patient's own anatomy, obtained from CT, is used in combination with SPECT imaging to obtain 3D spatial distribution of activity. It is advantageous for the α-emitting radionuclide to also have γ photon emission as it decays so that SPECT imaging can provide information about regional uptake. The method to calculate a 3D, imaged-based dose is as follows [127]. Serial SPECT studies taken over time post administration of the radiopharmaceutical can determine the pharmacokinetic variations within an organ on the voxel level. The SPECT/CT images are registered together and integrated voxel by voxel over time to obtain the 3D, time-integrated activity. The CT provides a map of tissue electron density for each voxel, which is necessary for dosage calculations. A Monte Carlo package is then employed using the activity distribution as the source definition and energy deposition is tallied in each voxel. Several software packages are available for 3D, patient-specific dosimetry, including MCID [128], OEDIPE [129] and SCMS [130] which are based on MCNP, 3D-RD [131] and DOSIMG [132] which are based on EGS [133], RAYDOSE [134] and RAPID [135] which are based on GEANT4 [136], and DPM [137], which is not a public domain code. Both MCNP and GEANT4 can simulate α-particles, while EGS can only transport photons and electrons.
While imaging-based dosimetry can account for non-uniformity and calculate dose at the voxel level, it is still a macroscopic quantity. The spatial resolution of current SPECT imaging ranges from 5 to 25 mm [31]. These dimensions are much larger than the 40-90 µm range of an α-particle. This results in the stochastic variations due to α-particle energy depositions not being taken into account and the voxel dose could be misleading. For α-particles, the ideal dosimetric targets are isolated cancer cells in transit in the vascular and lymphatic systems, micro-metastases and tumor capillary networks [31]. The only way to study the dosimetry of α-particles on a small scale is pre-clinically. These have been limited to in vitro measurements and microdosimetric Monte Carlo models. However, recently, several groups have developed alpha cameras, such as the Ionizing-Radiation Quantum Imaging Detector, which provides images of the ex vivo activity distribution with a spatial resolution on the order of tens of microns µm [138,139]. Monte Carlo methods can then be employed to calculate the dose absorbed on that scale. This is viewed as an important advancement in alpha dosimetry because no further modeling was needed to obtain the spatial and temporal activity distribution. neu-N transgenic mouse model with rat HER-2/neu expression and spontaneous lung metastases & NT2.5 mouse mammary fat pad xenografts with rat HER-2/neu. 225 Ac-TAT had greater efficacy but with renal toxicity relative to 213 Bi-TAT & 90 Y-TBT.
LS174T human colon & Shaw human pancreatic i.p. xenografts in nude mice.

Preclinical Therapeutics Studies
In recent years, a number of pre-clinical studies have been reported that provide strong evidence of the potential for use of TAT in the treatment of cancers. See Table 2 for a summary of the published in vivo preclinical TAT studies (n = 48) that have used animal models of cancer. Most of these studies (81%) have involved the use of immunoconjugates where monoclonal antibodies serve both as the targeting moiety and as the attachment scaffold for the α-emitting radionuclide. Two of these immunoconjugates involved the use of antibody fragments (scFv) [140,141] and one used a diabody [142]. Three studies involved pretargeting, where a cancer targeting antibody-streptavidin construct, or antibody-TCO (trans-cyclooctene) construct for click chemistry, was injected prior to delivery of an α-emitting radionuclide attached to biotin [141,143,144]. In this study, pre-targeting demonstrated increased efficacy over TAT using the targeting antibody alone. Peptide-based targeting moieties were used in 15% of these studies. In other individual studies, a protein [145], a small molecule ligand [146] and an inorganic compound [147] were used as targeting groups.
As is the case for most radioimmunotherapies in current use, that typically deliver β-emitting radionuclides, a number of the reported TAT studies (23%) targeted non-solid tumors; i.e., leukemias, lymphomas and multiple myeloma. A majority of these non-solid tumor studies reported significant efficacy and/or decreased systemic toxicity relative to controls. Since α-emissions have higher LET over a shorter range, there is significant interest in the use of TAT for treatment of solid tumors. This is evident in the preclinical studies reported, as 79% involved targeting of solid tumors and nearly all of these studies reported significant efficacy against solid tumors relative to controls, with nearly half also reporting low or manageable systemic toxicity. Four of these studies demonstrated increased efficacy using TAT relative to targeted beta (β)-emission therapy (TBT) using the same targeting moiety and tumor model [143,151,152,[163][164][165][166]180,185].

Preclinical Imaging
Whether developing a small molecule or antibody-based therapeutic, pharmaceutical companies traditionally rely upon molecular imaging to assist them in identifying the most promising molecules in the research pipeline to be carried forward into clinical trials [203][204][205][206]. Furthermore, this strategy has been particularly effective in developing targeted, systemic β-based radiotherapies when the β-emitting radioisotope can be paired with a diagnostic surrogate for SPECT or PET imaging. Several of these pairs include 64/67 Cu, 86/90 Y and 124/131 I [207]. However, with the exception of the 203/212 Pb system [68,191,208], researchers developing TATs with other α-emitters such as 213 Bi, 211 At and 225 Ac, have not been able to capitalize on this strategy. This limitation has hindered TAT development, but also stimulated attempts by the scientific community to observe the pharmacokinetics of TAT therapies using the decay of TAT radioisotopes in preclinical or clinical settings. For example, Sgouros et al. attempted to use the 440 keV γ emission of 213 Bi to monitor the biodistribution of 213 Bi-HuM195 in patients with leukemia [209]. Due to the short half-life of the radioisotope, data acquisition was limited to sixty minutes after patient injection, but this was sufficient to detect elevated levels of radioactivity in the red marrow, liver and spleen. While the data suggested that it would be possible to derive important pharmacokinetic and dosimetric data from the in vivo imaging of the 213 Bi emissions, the large amount of activity necessary for imaging has precluded its routine use in a clinical setting.
The decay scheme and extreme cytocidal potency of 225 Ac further complicates the TAT community's ability to develop preclinical and clinical imaging methodologies to monitor pharmacokinetics and estimate dosimetry. Since the 440 keV emission derived from 213 Bi decay is considered too energetic for preclinical imaging applications, attention has focused on the less energetic 218 keV γ ray emission of 221 Fr, which has been used to monitor the biodistribution of 225 Ac-doped nanoparticles in mice [71,86], and this approach has shown promise. However, in this imaging paradigm, animals are typically euthanized 1 h after receiving the radiopharmaceutical and imaged 24 h after euthanasia to ensure isotopic equilibrium. Hence, these experiments do not address the potential of longitudinal SPECT studies to inform development of 225 Ac-radiopharmaceuticals. Clearly, new approaches are needed, since the ability to image and visualize 225 Ac-based radiopharmaceutical biodistribution, metabolism, and clearance in animal models through longitudinal imaging studies would be advantageous [210,211].
Cerenkov luminescence imaging has emerged as an alternative to traditional nuclear medicine techniques for visualizing the delivery and biodistribution of TATs. Cerenkov luminescence is derived from the emission of ultraviolet light when certain charged particles exceed the phase velocity of light in a given medium [212,213], and this effect can be observed using standard optical imaging systems originally designed to detect bioluminescence and fluorescence. Recently, Cerenkov emissions were observed with the decay of several medically relevant isotopes, including 225 Ac, which was reported to yield the largest optical signal among all isotopes examined [211,214]. Alpha (α) particles however, do not travel with sufficient velocity to generate Cerenkov emissions, which led researchers to postulate that the emissions observed resulted from the β-decay of the 213 Bi, 209 Tl and 209 Pb daughter products. Subsequent publications, using theoretical and experimental means, described the association between Cerenkov radiation and 225 Ac decay [214,215]. While both research groups determined that Cerenkov luminescence imaging (CLI) could be accomplished with 225 Ac and other α-particle emitting radionuclides, it was determined that radionuclides' decays solely by α-particle emission do not produce sufficient Cerenkov radiation to be useful for imaging. Secondly, a time delay would be needed for equilibrium to be established between the parent radionuclide and its daughter products. For 225 Ac, this delay was determined to be ten hours. This precludes the imaging of small molecules and peptides at early time points, but allows one to monitor the biodistribution of 225 Ac-labeled antibodies and nanoparticles, which require extended circulation time for effective tumor targeting and blood clearance. Finally, the high recoil energy associated with 225 Ac and daughter product release from the original conjugate must be considered. The emission of Cerenkov radiation may not reflect the actual biodistribution of the radiopharmaceutical in question, but of the daughter products producing the Cerenkov emissions [216]. Sequestering the daughter products with the original conjugate is important if this technique is to provide any benefit to the TAT community. In proof of principle studies, Pandya and coworkers seized upon these recommendations and further tested the association between 225 Ac decay and Cerenkov radiation in living animals [217]. The authors synthesized 225 Ac-DOTA-c(RGDyK) and evaluated its stability, biodistribution and potential use as an imaging agent in CLI in a murine model expressing human glioblastoma U87mg tumors, which overexpress the α v β 3 integrin. Additionally, they exploited the well-documented ability of RGD-containing ligands to be internalized on integrin binding to sequester the 225 Ac and the daughter products within the tumor. This technique, often referred to as the nanogenerator approach, has been used successfully to sequester 225 Ac-based radiotherapeutics and daughter decay products within a cell to increase their therapeutic effectiveness [35,149,150]. Surprisingly, all animals demonstrated no signs of distress during these experiments. When compared to α v β 3 tissues, in vivo CLI revealed five-fold more luminescence from the α v β 3 + tumors. Luminescence was also observed in the liver and kidney tissues, which have been shown in related biodistribution studies to be involved in the clearance and excretion of this radiopharmaceutical [218][219][220]. Ex vivo image analysis also revealed a similar trend as the in vivo results, and the addition of a blockade reduced the luminescence emitted from the tumors by 80%, which suggested to the authors that activity delivered by the radiopharmaceutical and internalized through specific receptor-interactions was being retained in the tumor. To further validate the theranostic potential of their approach, they also performed preliminary therapy studies in U87mg tumor-bearing mice, which illustrated the cytocidal potency of their α-emitting radiotherapeutic [221]. While more studies are needed, these promising results suggest that this area of research may have broader implications for molecular imaging and may help to facilitate TAT development.

Lead Optimization
As previously touched upon, the characteristics of various targeting ligands, like lipophilicity, molecular weight and ionization potential are all critical variables that help modulate radiopharmaceutical properties. For example, small, polar compounds may undergo rapid renal excretion. The addition of a PEG linker will add water solubility, bulk, and may help avoid kidney toxicities. Conversely, the removal of an ionizable group (e.g., sulfonate or protonated amines at physiologic pH) during medicinal chemistry design may make the targeting ligand more hydrophobic and shift clearance toward initial liver processing. Therefore, when working to reduce targeting ligand complexity or increase the ease of synthesis, these physical attributes of the compound must be taken into consideration.
Antibodies and engineered fragments can also benefit from medicinal chemistry optimization. Many chelators and linkers covalently attached to antibodies and their fragments produce heterogeneous products. In the most typical fashion of linker attachment, surface accessible lysine side chains are utilized to form covalent bonds with radionuclide chelating molecules ( Figure 6). Site-specific antibody modifications can yield more homogeneous products. The Rader group at Scripps Florida has made such progress by engineering a selenocysteine residue into targeting antibodies, allowing for complete control of regiochemistry during functionalization [222].
Another route to increasing T/NT ratios is through the notion of pretargeting. These strategies operate under the premise that nonradiolabeled, bifunctional targeting agents can be administered to the patient and allowed to converge to their target. Once the unbound portion has cleared, a fraction of radiolabeled molecule with a binding affinity for the bifunctional targeting agent is administered. Typically, the radiolabeled agent is designed to clear quickly if unavailable for binding; thus, lowering off-target radiation. Typical binding partners amenable to pretargeting schemes include avidin/streptavadin-biotin, DNA-DNA or antibody-hapten [223]. For a more in-depth analysis of pretargeting strategies please see the recent review article from Frampas and coworkers [224].
Novel combination therapies are also now being investigated. Combination therapies can be particularly useful in heterogeneous diseases like multiple myeloma. Anti CD138 mAb conjugated with 213 Bi has been tested synergistically with the chemotherapeutic melphalan to help eliminate residual disease in a multiple myeloma model [116]. Interestingly, this study saw no benefit to the combination therapy when compared with RIT alone. The speculation was that the follow-up treatment with CD138-targeted mAb was less successful due a swift reduction in CD138 antigen in cells progressing through apoptosis [225], illustrating the importance of the orthogonality in study design as it relates to combination treatments and target expression.

cGMP Production
Current good manufacturing practice (cGMP) standards for the production of the targeting ligand precursors before the incorporation of the α-emitting radioisotopes are similar to those used for the manufacturing of peptide, antibody or antibody fragment drug and drug candidates. There are numerous cGMP-compliant peptide synthesis companies, and often the preclinical grade peptides will be prepared by standard solid-phase synthesis and purified to ≥ 95% purity, but cGMP-grade peptides will often be prepared via routes that allow greater control over the purity and reliability of the synthesis. For instance, cGMP peptides are more likely to be prepared using a greater reliance on solution phase-synthesis procedures than preclinical grade peptides because the costs can be much lower on scaled-up batches of peptide, and more importantly, impurities can be removed as they build up during a solution-phase synthesis, whereas impurities just build up during standard solid-phase synthesis procedures. The recent development of a fully solid-phase synthesis of cGMP-quality Fuzeon ® is an exception to this trend that may lead to a greater reliance on fully solid-phase synthesis procedures for the production of cGMP peptides. Frequently, solid phase methods are used in combination with solution phase synthesis methods. As an example, preclinical grade linear peptides can be prepared and cyclized to the final product on resin, whereas cGMP grade peptides will often be prepared on resin and purified and then cyclized in solution, due to the easier analysis of reactions in solution.
The cGMP production of antibody and antibody fragments will be focused on for the development of robust methods of production often using hybridoma technologies and purification by at least two orthogonal separation methods, usually involving an affinity column purification step followed by a size exclusion type purification system. Often these biotechnology-based manufacturing methods produce antibody or antibody fragments that are chemically programmed to provide specific sites for metal chelate attachment. For a recent review of this topic please, see the two-part article from Adumeau et al. [226,227].
The results of these manufacturing processes are lyophilized powders or relatively-concentrated, sterile, aqueous buffer solutions of the targeting moieties that are ready to chelate the α-emitters. The main difference between cGMP radiosynthesis for clinical trials and research grade radiosynthesis in the United States is the implementation of an FDA-compliant audit trail. Since the development of α-emitting radiopharmaceuticals is happening globally, another major issue concerning production and distribution of these novel therapies is the consideration of the local regulatory environment. A discussion of the potential for harmonization of the various international regulations has recently been published [228].

Recent TAT Clinical Trials
Despite the potential therapeutic benefits of the α-particle emitting radioisotopes over β-particle emitters, the restricted availability of relevant radionuclides and limited number of phase III clinical trials has so far limited the clinical use of TAT. Table 3 summarizes the TAT clinical studies conducted to date. Clinical trials prior to 2014 were reviewed by Elgqvist et al. [34] and we will focus on the more recent trials herein. The use of short-lived α-emitters like 211 At (T 1/2 = 7.2 h), 212 Bi (T 1/2 = 60 min) and 212 Pb (T 1/2 = 10.6 h) has been subdued so far, due to production and distribution limitations. However, this is not the case for 213 Bi (T 1/2 = 46 min), which, going back over the past two decades, has been used in the largest number of TAT clinical trials to date. Recent 213 Bi studies have included the targeting of peptides for the treatment of glioblastoma [229,230] and neuroendocrine tumors [65], and the targeting of monoclonal antibodies for the treatment of bladder cancer [231,232]. Since the approval of 223 Ra-dichloride for the treatment of CRPC in 2013, 225 Ac has also emerged as a targeted therapeutic radionuclide of interest that has been used for targeting a small molecule [5,6,233], peptide [234] and monoclonal antibody [235] for the targeting of prostate cancer, neuroendocrine tumors, and acute myeloid leukemia (AML), respectively. Clinical studies of TATs using 225 Ac or 213 Bi as the therapeutic radionuclide have recently been reviewed [58]. Although the responses have been impressive for some, 225 Ac-PSMA-617 in particular, the completion of more randomized clinical trials comparing TAT to standard of care will be essential to determine the clinical utility of these new therapeutic approaches.

Radium-223 Dichloride ( 223 RaCl 2 ) Trials
In the following section, we will discuss the first and only clinically-approved TAT in the United States so far. After more than a 100 years since the discovery of the radium element in 1898 by Marie and Pierre Curie, on May 15, 2013, one of its radioisotopes, radium-223 (Ra-223) became the first α-emitting radionuclide to be approved by the Food and Drug Administration (FDA) in the form of radium-223 dichloride ( 223 RaCl 2 , Xofigo ® formerly Alpharadin, made by Bayer HealthCare Pharmaceuticals Inc.) for the treatment of patients with castration-resistant prostate cancer (CRPC), with symptomatic bone metastases, and no known visceral metastatic disease [252]. This was based on the improvement in overall survival following intravenous administration of 223 RaCl 2 in patients with advanced prostate cancer with metastases to the bone, demonstrated in a double-blind randomized placebo-controlled trial called the ALSYMPCA (alpharadin in symptomatic prostate cancer) trial [7].
The role of systemic radionuclide therapy to the bone is to deliver therapeutically-effective radiation doses to multiple foci, including microscopic disease, as opposed to external beam radiotherapy, that gives more focal treatment and affects both normal bone marrow and metastases [253]. The two most commonly used radionuclide bone seeking agents for bone pain palliation are 89 Sr (Metastron) and 153 Sm-ethylene diamine N,N -tetra(methylene) phosphonic acid ( 153 Sm-EDTMP; Quadramet) [254], but these β-particle therapies, although effective for pain palliation, frequently cause myelosuppression and have not been shown to affect patient survival [255].
Cations of the alkaline earth elements (strontium, barium, and radium) have natural bone-seeking properties without the need for a carrier agent. As radium mimics calcium, when Ra-223 dichloride is injected intravenously, it forms complexes with the bone mineral hydroxyapatite at areas of increased metabolic bone activity, such as bone metastases, thereby exerting a highly localized antitumoral effect [7,256]. In addition, the Ra-223 daughter isotopes are also retained in the bone matrix [50]. The shorter penetration range of Ra-223 causes less bone marrow toxicity than bone-seeking β-particles [256]. The peak skeletal uptake of Ra-223 occurs within 1 h of injection, and blood radioactivity levels are <1% after 24 h with almost no redistribution of daughter nuclides from bone [256]. The total skeletal uptake ranges between 40% and 60% of the administered activity [257], and the excretion is predominantly via the gastrointestinal tract, with much less renal clearance [258]. Due to their significantly higher potency, the activities administered of α-emitters are much lower than β emitters, with significantly less radiation exposure to hospital staff and family members [259]. Merely basic hygiene measures and routine contamination protection barriers, such as latex gloves, are enough to protect both health care personnel involved in the patient's radionuclide administration and the patient's family.
In the first-in-human single dosage phase I trial for patients with breast and prostate cancer skeletal metastases, single dose Ra-223 was well tolerated at all therapeutically relevant doses (46-250 kBq/kg intravenously [IV]) [257]. There was no dose-limiting hematological toxicity, only mild, transient myelosuppression (nadir 2 to 4 weeks after the injection) and mild transient diarrhea, with nausea and vomiting more frequently observed in the highest dosage group [257]. Subsequently, the phase I dose-escalation study in advanced prostate cancer with bone metastases, showed that repeated Ra-223 dosing was well tolerated to a total dose of 250 kBq/kg [50].
In the following phase II study of CRPC patients with symptomatic bone metastases, Ra-223 given at different doses (5, 25, 50 and 100 kBq/kg) was well tolerated with a positive effect on pain assessments (visual analogue scale [VAS] and analgesic use). At week 8, 56% and 71% of patients had reduced pain and stable analgesic consumption in the 50 and 100 kBq/kg groups, respectively. A statistically significant decrease in bone turnover markers, including alkaline phosphatase, was noted in the highest dose-group (p = 0.0067) [260].
The phase III, double-blind, randomized, placebo-controlled, multicenter, multinational ALSYMPCA trial enrolled patients from June 2008 through February 2011 [7]. Eligible patients had progressive metastatic symptomatic disease and regular use of analgesic medication or treatment with external beam radiation therapy (EBRT) for cancer-related bone pain within the previous 12 weeks. Patients were randomized 2:1 to receive six injections of radium-223 (50 kBq/kg IV) or a saline placebo infusion; a cycle was comprised of one injection every 4 weeks. The primary endpoint was overall survival (OS). Various secondary endpoints included time to first symptomatic skeletal events (SSE), time to alkaline phosphatase increase and time to PSA progression. A planned interim analysis was conducted, and on June 3, 2011, the Independent Data Monitoring Committee recommended stopping the trial early due to evidence of a significant treatment benefit. Median OS was 14.0 months for Ra-223 (95% CI (confidence interval): 12.1-15.8) versus 11.2 months for placebo (95% CI: 9.0-13.2) (p = 0.00185; HR (hazard ratio) = 0.695; 95% CI: 0.552-0.875).
An updated analysis included 528 events from 921 randomized patients prior to placebo patients crossing over to Ra-223 [7,261]. For this analysis, median OS was 14.9 months for Ra-223 arm (95% CI: 13.9-16.1) and 11.3 months for patients receiving placebo (95% CI: 10.4-12.8) (HR = 0.695; 95% CI: 0.581-0.832). OS favored Ra-223, with a 30% reduction in the risk of death compared with placebo. The OS benefit with Ra-223 was supported by a delay in time to first SSE compared to the placebo (15.6 versus 9.8 months; HR = 0.66; 95% CI, 0.52-0.83; p < 0.001). Improvement in survival was independent of both prior exposure to docetaxel [250] and the use of bisphosphonates [261]. Ra-223 significantly prolonged time to total ALP progression (HR = 0.167; 95% CI: 0.129-0.217; p < 0.00001); and time to PSA progression (HR = 0.643; 95% CI: 0.539-0.768; p < 0.00001) was associated with a beneficial effect on pain and quality of life (QOL) [262], and was associated with a highly favorable safety profile with a low rate of adverse hematological events. The most common (at least two percent) of the adverse reactions in patients receiving Ra-223 were nausea, diarrhea, vomiting, renal impairment and peripheral edema.
Based on these data, the National Comprehensive Cancer Network (NCCN) supports the use of 223 RaCl 2 (Xofigo ® ) as a first-line therapy for patients with symptomatic metastatic CRPC and no visceral metastases with a Category 1 recommendation [227]. The recommended dose for 223 RaCl 2 is 50 kBq/kg (1.35 microcuries/kg) administered by slow intravenous injection, over 1 min, every 4 weeks, for a total of six cycles.
Since completion of the ALSYMPCA trial, Ra-223 has been studied in combination with other, newly-approved CRPC therapies, such as abiraterone, enzalutamide, docetaxel, cabazitaxel, sipuleucel-T, olaparib and pembrolizumab [263][264][265]. A phase 3, randomized, double-blind study of radium-223 or placebo, each in combination with abiraterone plus prednisone in chemotherapy-naive patients with asymptomatic or mildly symptomatic mCRPC with bone metastases (ERA 223; NCT02043678) [266], was recently prematurely unblinded. The independent data monitoring committee (IDMC) recommended unblinding the trial due to the observation of more fractures and deaths in the combination treatment arm. The European Medicines Agency (EMA)'s Pharmacovigilance Risk Assessment Committee (PRAC) has reviewed the preliminary data from the ongoing clinical study. In the study 34.7%, of patients treated with Xofigo, Zytiga and prednisone/prednisolone have died so far, compared with 28.2% of patients given placebo, Zytiga and prednisone/prednisolone. Fractures have also occurred more frequently with the Xofigo combination than the placebo combination (26% versus 8.1%). Unblinded data from the study are currently being analyzed to confirm the preliminary findings of the IDMC. Given these results from ERA 223 trial, the current recommendation is not to combine radium-223 with concomitant abiraterone acetate and prednisone (https://www.ema.europa. eu/news/prostate-cancer-medicine-xofigo-must-not-be-used-zytiga-prednisoneprednisolone). Hence, the administration of concomitant chemotherapy with Xofigo is not recommended due to an increased risk of adverse reactions [267].
Immune modulation has been examined in vitro in human prostate, breast and lung carcinoma cells following exposure to sub-lethal doses of 223 Ra. Malamas and his colleagues showed that 223 Ra significantly enhanced T cell-mediated lysis of each tumor type by CD8+ cytotoxic T lymphocytes (CTLs) specific for MUC-1, brachyury, and CEA tumor antigens. Therefore, 223 Ra may be more effective in combination with immunotherapies [268].
In view of the increased long term risk of secondary malignancies, such as sarcomas and leukemias, that were seen in patients who were treated with high doses of another radium isotope, radium-224 (T 1/2 = 3.7 days) (primarily used in Europe) for bone pain associated with ankylosing spondylitis, there is concern for development secondary malignancy with the use of Ra-223. So far there has been no evidence of secondary malignancies following Ra-223 therapy [7,269]. However, it is important to note that patients with metastatic CRPC have a much shorter overall survival than patients with ankylosing spondylitis. Care should be taken when expanding the use of Ra-223 to other patient groups with longer expected overall survival times, since the rate of secondary tumor development may increase.
Ra-223 trials have been performed for patients with diseases other than CRPC. Ra-223 was found to be safe and well tolerated when studied in a phase II, open-label trial investigating the safety and short-term efficacy in patients with breast cancer who had bone-dominant disease [270], but the data are still limited. In contrast to prostate cancer, multiple myeloma bone lesions are predominantly lytic. It is not yet know if Ra-223 will have any role in this category of patients, but based on previous clinical trials using 153 Sm-EDTMP plus chemotherapy, the combination showed efficacy in some patients with far-advanced myeloma [271], so theoretically, these results could be replicated with Ra-223, with potentially fewer side effects. In view of the limited but positive experience in 153 Sm-EDTMP for osteosarcoma due to its osteoblastic nature [272], there is currently a phase I dose trial for the use of Ra-223 in patients with osteosarcoma at the M.D. Anderson Cancer Center (www.clinicaltrials.gov; NCT01833520).

Conclusions
There has recently been significant interest and progression toward the development of targeted alpha-particle therapy (TAT) for the treatment of solid tumors. Many pre-clinical and clinical studies have demonstrated promising anti-tumor efficacy results (Tables 2 and 3). The success of 223 RaCl 2 in the care of patients with bone metastases and the apparent efficacy of additional TAT agents in recent clinical studies, e.g., 213 Bi-DOTATOC and 225 Ac-PSMA, has brought attention to this area of radiopharmaceutical development and increased the potential for successful translation of TAT for treatment of solid tumors and metastases [65,233]. Three radioisotopes have seen the most use in preclinical and clinical studies, 213 Bi, 225 Ac and 211 At. Each of these radionuclides has distinct advantages and disadvantages regarding availability, cost, decay chain attributes, tissue distribution and stability of attachment. Of these, 225 Ac is readily available and has the distinct advantage of having a long, 10 d half-life, which allows for centralized production and distribution compared to the requirement of local production for the other radionuclides with shorter half-lives. Since TAT does not rely on the generation of free-radicals for the generation of DNA damage, as is the case for β-emission therapies, a major mechanism of resistance to radiation treatment can be bypassed; i.e., upregulation of superoxide dismutase. Some evidence of this was seen in a report stating that patients with tumors that have developed resistance to 90 Y/ 177 Lu-DOTATOC therapy were sensitive to subsequent treatment with 213 Bi-DOTATOC [65]. Additionally, the use of α-emitters enables solid-tumor targeted therapy with decreased peripheral tissue damage and the potential for decreased development of resistance due to the ability to kill surrounding tumor cells that do not express the target marker in a heterogeneous tumor microenvironment. Small molecules, peptides and monoclonal antibodies have been used for tumor targeting (Tables 2 and 3). However, due to the shorter clearance time (minutes to hours), small molecules and peptides may have an advantage over antibodies that can circulate for days in that they can accumulate in the target tumor and clear systemically prior to decay and deposition of energy, maximizing the beneficial effects of the ionizing radiation while minimizing the off-target exposure. This is especially true when using radionuclides with a long half-life; e.g., 225 Ac. Peptide-targeted agents may have a greater risk of generating renal toxicity [66], but there is evidence that lead optimization via medicinal chemistry could potentially improve the renal toxicity of a given peptide agent. For example, a recent preclinical study of two different gastrin-releasing peptide receptor (GRP-R)-targeting agents using two different 213 Bi-labeled GRP-R targeting peptides demonstrated differential renal toxicities [185]. TAT is currently emerging as a potentially new and effective approach toward the treatment of solid tumors, and the clinical translation of novel TAT radiopharmaceuticals appears to be on the near horizon. Since the abscopal effect has been observed in external beam therapy in animals, the ability to target internal radiation to tumors via TAT may have additional added value as a companion to the recently-successful immune checkpoint targeted therapies.