A Radium-223 Microgenerator from Cyclotron-Produced Trace Actinium-227

Abstract

The alpha particle emitter Radium-223 dichloride (²²³RaCl₂) has recently been approved for treatment of late-stage bone metastatic prostate cancer. There is considerable interest in studying this new agent outside of the clinical setting, however the supply of ²²³Ra is limited and expensive. We have engineered a ²²³Ra microgenerator using traces of ²²⁷Ac previously generated from cyclotron-produced ²²⁵Ac. Radiochemically pure ²²³RaCl₂ was made, characterized, evaluated in vivo, and the source was recovered in high yield for regeneration of the microgenerator.

Introduction

Carcinoma of the prostate is diagnosed in one in seven men (Siegel et al., 2014). When detected early, treatment options including surgery and external beam radiotherapy are highly effective. However, when the disease recurs or if diagnosed after metastasis, there is no definitive treatment. Prostate cancer has a strong tropism for colonizing the skeleton, and bone metastases occur in up to 90% of patients with castrate resistant disease (Ottewell et al., 2014). Recently, advances have led to the approval of alpha-particle radiopharmaceutical therapy Radium-223 Dichloride (²²³RaCl₂) for treatment of osseous prostate cancer metastases. This bone-seeking radionuclide delivers potent cytotoxic alpha-particles to bone metastasis sites (Bruland et al., 2006; Henriksen et al., 2002) improving overall median survival of patients by nearly 4 months (Parker et al., 2013).

There are several routes of ²²³Ra procurement. Historically, first described by Pierre and Marie Curie in the 1900’s, radium purification from pitchblende ore containing fissile uranium yielded several milligrams per ton of material (Adloff, 1999). Currently, clinical and commercial production of ²²³RaCl₂ (Bayer HealthCare Pharmaceuticals) involves ²²⁷Ac and ²²⁷Th isolation from a ²³¹Pa source (half-life of 3.28×10⁴ y) (Larsen et al., 2003). Access to this material is limited, costly and the long half-life of the parent makes use prohibitive. Alternatively, legacy sources of ²²⁷Ac exist, and innovative approaches have been leveraged to convert this material to medical and research grade ²²³Ra and ²²⁷Th (Soderquist et al., 2012).

Radiochemical separation of GBqs of ²²³Ra starting from sources of ultra-pure ²²⁷Ac and ²²⁷Th has previously been described (Mokhodoeva et al., 2015; Soderquist et al., 2012). Briefly, this purification utilized ionic exchange resins to separate ²²⁷Ac and ²²⁷Th from ²²³Ra-nitrate complexes based on charge discrepancies (Boll et al., 2005). Following ²²³Ra elution, the parent ²²⁷Ac and ²²⁷Th radionuclides may be resorbed on a new column for repeated purification. The decay chain for ²²⁷Ac isotopes is shown in Figure 1A as well as that for ²²⁵Ac.

Fig 1.

A) Decay chains of ²²⁷Ac and ²²⁵Ac: ²²⁷Ac is predominantly a beta-emitter decaying to ²²⁷Th. ²²⁷Th produces a total of 5 alpha particles to the stable ²⁰⁷Pb. ²²³Ra is the first daughter of ²²⁷Th, and the desired product of the purification. The initial source contains residual amounts of ²²⁵Ac decaying into 4 alpha-emitting daughters including ²¹³Bi detected in the recovered acid wash; B) Anion-exchange column (Dowex 1×8 resin) was utilized to purify the source of ²²⁷Ac and ²²⁷Th using a mix of HNO₃ and methanol. ²²³Ra was purified as nitrates, ²²³Ra solution was evaporated until dryness and resuspended in sodium citrate (0.03 M) – saline (150 mM) solution for in vivo experiments. Ultimately, the residual of column radioactivity was washed off using 0.5 M HNO₃ to recover ²²⁷Ac and ²²⁷Th.

More recently, research efforts have been undertaken by the US Department of Energy to achieve cyclotron-production of ²²⁵Ac using high-energy proton irradiation of natural thorium targets (Radchenko et al., 2015a; Radchenko et al., 2015b; Weidner et al., 2012). However, ²²⁷Ac contamination is a side-product of this irradiation (approximately ≈0.3%). In this work, we have recycled the ²²⁷Ac by-product from the decayed source of ²²⁵Ac to generate ²²³Ra in high purity. Creation of a miniaturized production scale ²²³Ra generator has been accomplished utilizing traces of ²²⁷Ac (kBq). This source of ²²⁷Ac has generated an indefinite supply of ²²³Ra in amounts needed for preclinical studies, while recovering parent radionuclides ²²⁷Ac and ²²⁷Th in high yield.

Materials and Methods

Chemicals, radioisotope supplies, generator assembling and recollection

All reagents and solvents were purchased from Sigma-Aldrich and used directly, unless otherwise noted. Cyclotron-produced ²²⁵Ac-nitrate was supplied by the Department of Energy. The radioactive material was received dry, containing 296 MBq of ²²⁵Ac-nitrates and approximately 888 kBq of ²²⁷Ac-nitrates. This material was set-aside for over 10 half-lives of ²²⁵Ac, allowing for ²²⁵Ac to decay away. Following ²²⁵Ac decay, the remaining radioactive material is composed mostly of ²²⁷Ac and its daughters. This material was then dissolved in 0.5–1 mL of CH₃OH (80%) and 0.4 M HNO₃ (20%). All aqueous-based solutions were made from ultra-pure metal-free water purified with Chelex 100 resin (BioRad).

For purification, a conditioned strong base, type 1, anion exchange resin, Dowex 1×8 (100–200 mesh) was used. Briefly, 25 g of resin was hydrated and decanted at least 3 times to eliminate fine particulates. The gel was then conditioned with basic (1 M NaOH) and acid (0.05 M HCl/1 M NaCl) washes. Finally, the stock of processed resin was stored at 4 °C in 50 mL 20% aqueous ethanol until used.

The microgenerator column was formed using 2 mL of the anion exchange resin solution that was settled in an empty poly-prep chromatographic column equipped with a frit disk (porous 30 µm polyethylene support, BioRad). The column was then washed with at least 3 column volumes of water and 3 column volumes of CH₃OH (80%), 0.4 M HNO₃.

For ²²³Ra purification, (Figure 1B) the radioactive source was adsorbed onto the column resin. ²²³Ra was eluted with CH₃OH (80%)/0.4 M HNO₃ in 0.5 mL increments. The elution was conducted up to complete exhaustion of radioactivity (determined by measuring each of the eluted fractions using a dose calibrator). Each fraction was later analyzed using well-counter and gamma spectroscopy (shown in Figure 2A).

Fig 2.

A) Radioactive elution profile of ²²³Ra-microgenerator fractions was assessed using a NaI well-counter acquiring within the spectral region 240–300 keV. B) The time-dependence γ-counts for ²²³Ra sample eluted at 8.5 mL (black dots) is plotted with the fit of the complete decay chain (blue) and the expected behavior for pure ²²³Ra (red). C) The ²²³Ra eluted fractions (8–14.5 mL) were γ-counted over a period of 28 days, the plots were fit to a decay model of the decay chain and detector response to determine the isotopic content for each fraction (Table 1).

Following ²²³Ra purification, ²²⁷Ac and ²²⁷Th parents were washed off the column with 3–6 column volumes of 0.5 M HNO₃. (Figure 1B). To limit ²²⁷Th and ²²⁷Ac breakthrough, a new column was prepared for each ²²³Ra production.

For in vivo administration, ²²³Ra was evaporated at 90°C, using a hot plate, twice till dryness to eliminate nitric acid. The dry vial was suspended in saline/0.03 M sodium citrate for conversion to ²²³RaCl₂. The pH was verified to be 7.

Radioisotopic evaluation of microgenerator and products

Dose Calibration

The source, eluted fractions and column were measured using a NaI dose calibrator (CRC-127R; Capintec Inc) for radioactive recovery determination. An empirically derived calibrator setting of #277 was used to measure ²²³Ra, as described previously (Abou et al., 2016). Briefly, the calibrator parameter was determined by selecting a value confirming the decay-corrected National Institute of Standards and Technology standard calibrated dose of a commercial ²²³RaCl₂ source (Bayer HealthCare Pharmaceuticals) (Bergeron et al., 2010).

Well-Counting

An automated NaI gamma counter (Wizard2, Perkin Elmer) was used to determine fraction radioactivity. The γ-counting energy window was set between 240 and 300 keV for a duration of 600 seconds. The main photopeak detected at 270 keV was provided by the emissions of ²²³Ra (269 keV; 13.7%) and ²¹⁹Rn (271 keV; 10.8%). The time dependence of the count-rate for this specific region of interest has been modeled and fit to determine the fraction of ²²⁷Ac and ²²⁷Th relative to ²²³Ra in the purified fractions.

Multi Channel Analyzer (MCA)

Source and fraction samples were analyzed using Ace-2K (EG&G Ortec) connected to a detector Bicron 2 inch by 2 in well-counter acquired for 4 minutes. Spectra were read using Maestro 32 software.

Gamma Spectroscopy

Spectroscopic analysis of eluted fractions was carried out using a calibrated battery powered Stirling-cycle cooled portable high purity germanium (HPGe) spectrometer (Detective-EX-100, Ortec). Column fractions in eppendorf tubes were placed on a platform at distances of between 2 and 16 cm from the detector and counted for approximately 300 seconds. Spectra were initially analyzed in the Maestro software package (Ortec). A background spectra (in the absence of a radioactive sample) was also acquired for 360 sec. Using the collected data, the known detector efficiency values and the γ-ray emission probabilities (from the National Nuclear Data Center) we determined the isotopic content of purified fractions, acidic washes and the column residual. The relative efficiency of our HPGe detector was previously determined at a number of fixed energies using the γ-rays of ¹⁵²Eu and ¹³³Ba measured at fixed distances. The detector efficiencies were determined using a linear interpolation from the previously determined efficiencies.

Animal Experimentation

All mouse experimentation was in accordance with institutional animal welfare protocols at The Johns Hopkins University School of Medicine and conformed with National Institutes of Health Guide for the Care and Use of Laboratory Animals (2011). Skeletally mature male C57Bl/6 mice (>14 weeks) (n=5 per group) received a 150 µL retro-orbital dose of ²²³RaCl₂ (10 kBq) diluted in saline with 0.03 M sodium citrate from either remnant activity from a clinical dose (Bayer HealthCare Pharmaceuticals) or as purified from the microgenerator. After 24 h, animals were sacrificed by CO₂ asphyxiation and cervical dislocation followed by prompt dissection. Excised tissues were weighed, and radioactivity was assessed by γ-counting (Wizard2, as above). Count data were background and decay-corrected to determine the tissue uptake as a percentage injected activity (%IA) to the time of injection.

Statistical Analysis and data modeling

Figures were prepared in Illustrator (Adobe) and Prism (Graphpad). Statistical significance was defined using the unpaired, two-tailed Student’s t-test. Differences at the 95% confidence level (P<0.05) were considered to be statistically significant.

The fitting of half-life decay was conducted using Wavemetrics Igor Pro. We used a weighted non-linear Levenberg-Marquadt fitting procedure to fit the measured count-rate data to the coupled differential equations listed in Equations 1S (representing the decay chain in Figure 1A), and the detector response shown in Equation 1. In this process the unknown variables were limited to the amounts of ²²⁷Ac, ²²⁷Th, ²²³Fr and ²²³Ra present at time t=0.

$$\mathrm{C}\mathrm{R}(\mathrm{t})=\sum _{\mathrm{j}}\sum _{\mathrm{k}=\mathrm{\gamma }}^{\mathrm{R}\mathrm{O}\mathrm{I}}{\mathrm{\epsilon }}_{\mathrm{k}}(\mathrm{E}){\mathrm{I}}_{\mathrm{j},\mathrm{k}}{\mathrm{\lambda }}_{\mathrm{j}}{\mathrm{N}}_{\mathrm{j}}(\mathrm{t})$$ Equation 1:

Here the N’s are the number of atoms of each isotope at time, t, and the λ’s are the decay coefficients of each of the isotopes with half-life T_j1/2 where λ_j= Ln(2)/T_j1/2, I_j,k is the fraction of decays that emit gamma-rays with energy E and ε(E) is the detection efficiency at energy E. The set of differential equations (Equations 1S) may be simplified by considering that ²²³Ra and its daughters are all in equilibrium at the time of the measurement (>10 h post elution). Using this information, we can reduce the Equations 1S to the solution of only the top 4 coupled differential equations (with ²²⁷Ac, ²²⁷Th, ²²³Fr, ²²³Ra). The starting amount of each isotope was integrated using the differential equations (Equation 1S) from t=0 until the measurement time.

This allowed us to determine the number of atoms of each isotope at a given time, N_j(t). These values are then combined with Equation 1 to calculate the number of gamma-rays detected by the well counter and then used to fit this to the data of sample decay over 28 days (Figure 2B and 2C). We note that the measured data at t=0 and t=15 days was not included in this fit. At t=0, the equilibrium was not yet reached with the daughters (of ²²³Ra). At t=15 days, the data are consistently low for all measured samples. The plot and fit to the data for the eluted fraction 8.5 mL is shown in Figure 2B, and has been repeated for all other fractions with the results summarized in Table1 and plotted in Figure 2C. We point out that because the purified elutions are mostly ²²³Ra and the gamma-rays emitted by the samples are mostly from ²²³Ra (and its daughter ²¹⁹Rn) one can simplify Equation 1 further and the detector efficiency is not needed in order to determine the relative activities of ²²⁷Ac, ²²⁷Th and ²²³Ra (plus daughters). Since the absolute detector efficiency was not determined in this study we confine our analysis of this time series data to determine the relative ratios of activities (at time t=0). Fortunately because of the narrow energy region of the counting and the fact that counts are dominated by the emissions from the gamma-rays of ²²³Ra and ²¹⁹Rn at 269 and 271 keV respectively, that are very close in energy we can assume the detector efficiency is the same for these emissions (to within a few percent error). With these assumptions we can model the decay data to determine the relative isotopic content of the eluted fractions to within at best a few percent (consistent with our lack of detailed information on the detector efficiency).

Table 1.

Isotopic ratios at time t=0 extracted from the time dependent γ-count measurements for each purified ²²³Ra fraction.

Eluted Volume (mL)	Ratio (227Th/223Ra)%	Ratio (227Ac/223Ra)%
8	1.8±0.7	0±0.6
8.5	1.6±0.7	0±0.6
9	1.5±0.6	0.7±0.4
9.5	1.5±0.7	0±0.6
10	2.0±0.2	0±0.5
10.5	1.1±0.7	0±0.7
12.5	0.98±0.2	0±1.0
13.5	1.2±0.8	0±0.7
14.5	0.95±0.6	0±0.6

HPGe measurements were conducted on the purified ²²³Ra fraction (8.5mL), the column wash off, the column residual 23 days post-elution and a background with no sources present. These measured spectra, and example spectra of ²²⁷Ac, ²²⁷Th, ²²³Fr, and ²²³Ra (with its daughters) are displayed in Figure 3A–3F, where example spectra were generated calculated using GADRAS (Mitchell, 1992) and previously determined detector response models. In these Figures the live-time corrected background has been subtracted from the measurement.

Fig 3.

A) Full view γ-ray spectra HPGe of purified ²²³Ra material produced from the microgenerator (fraction eluted at 8.5 mL) (black triangles). The relative components of the spectrum are ²²³Ra and daughters (blue), ²²⁷Th (plain black), ²²⁵Ac (dotted blue), ²²³Fr (green) and ²²⁷Ac (red). B) Fit of γ-ray ²²³Ra samples in the region from 220 to 280 keV. Measurement fit (red) and radioisotopes models (²²³Ra, ²²⁷Th, ²²³Fr/²²⁷Ac) are presented. The components, detector and emission efficiency, used to fit the isotopic models are shown Table 2. C) Full view γ-ray spectrum of the acid wash (black triangles). The relative components of the spectrum are ²²³Ra and daughters (blue), ²²⁷Th (black), ²²⁵Ac (dotted), ²²³Fr (green) and of ²²⁷Ac (red). The sample presents traces of ²¹³Bi, daughter of ²²⁵Ac. This is a strong indication of the parent radioisotopes elution and recovery of ²²⁷Ac and ²²⁷Th. D) Fit of γ-ray acid wash (red) in the region from 220 to 280 keV, ²²⁷Th and ²²³Ra models are matching the γ-ray measurements of the column effluent (round plots). E) Full view γ-ray spectrum of the column residual (black triangles). The relative components of the spectrum are ²²³Ra and daughters (blue), ²²⁷Th (black plain), ²²⁵Ac (dotted blue), ²²³Fr (green) and ²²⁷Ac (red). The column shows an apparent reduction of ²²⁵Ac. This is a strong indication that the ²²⁵Ac was removed by acid wash, however ²²⁷Th remains partially adsorbed on the resin. F) Fit of γ-ray column residual (red) in the region from 220 to 280 keV, ²²⁷Th (dash) and ²²³Ra (blue) models are matching the γ-ray measurements of the column effluent (round plots). It shows that both ²²³Ra and ²²⁷Th residual remains on the column after purification and wash off.

A detailed estimation of the isotopic contents were calculated based on fitting the peaks using a Gaussian model of the γ-ray peaks in the narrow region 220–280 KeV of the HPGe spectra. The equation used to fit this region is shown in Equation 2. The efficiencies, emission intensity and γ-ray energies used in the model are listed in Table 2. For Equation 2 below, r is the distance from the source to the detector; A is the cross-sectional area of the detector; ε(E) is the intrinsic efficiency of the detector at the energy of the emitted γ-ray; ²²⁷Th, ²²³Fr, and ²²³Ra are respective activities, I is the relative intensity of the emitted γ-ray for that decay and the sums of i, j and k are over the γ-rays emitted by the isotopes in the region (250–400 keV), finally σ is the Gaussian width of the peak (1.82keV). Also included are the terms for a linear background, where S is the slope and Y0 the y offset. It should be noted that the peaks widths are in principle dependent upon the gamma-ray energy as well. However over this narrow spectral region these peaks are well represented by a constant width, thereby simplifying the equations.

$$\mathrm{C}\mathrm{R}=\left(\frac{\mathrm{A}}{\mathrm{\sigma }\sqrt{2\mathrm{\pi }}4\mathrm{\pi }{\mathrm{r}}^2}\right)\{{}^{223}\mathrm{R}\mathrm{a}{\sum }_{\mathrm{i}}{\mathrm{\epsilon }}_{\mathrm{i}}{\mathrm{I}}_{\mathrm{i}}{\mathrm{e}}^{(-\frac{{({\mathrm{E}}_{\mathrm{i}}-\mathrm{E})}^2}{{\mathrm{\sigma }}^2})}+{}^{227}\mathrm{T}\mathrm{h}{\sum }_{\mathrm{j}}{\mathrm{\epsilon }}_{\mathrm{j}}{\mathrm{I}}_{\mathrm{j}}{\mathrm{e}}^{(-\frac{{({\mathrm{E}}_{\mathrm{j}}-\mathrm{E})}^2}{{\mathrm{\sigma }}^2})}+{}^{223}\mathrm{F}\mathrm{r}{\sum }_{\mathrm{k}}{\mathrm{\epsilon }}_{\mathrm{k}}{\mathrm{I}}_{\mathrm{k}}{\mathrm{e}}^{(-\frac{{({\mathrm{E}}_{\mathrm{k}}-\mathrm{E})}^2}{{\mathrm{\sigma }}^2})}\}+\mathrm{E}\mathrm{S}+\mathrm{Y}0$$ Equation 2:

Table 2.

Photon emissions associated with HPGe detector efficiency and emission intensity ranging between 220–280 keV.

	Energy (keV)	Detector Efficiency	Emission Intensity (%)
223Ra+ daughters
	269.46	49.73	13.7
	271.23	49.5	10.8
227Th
	235.96	54.26	12.9
	250.27	52.2	0.45
	252.5	51.92	0.11
	254.63	51.64	0.7
	256.23	51.43	7
	262.87	50.4	0.11
	272.91	49.0	0.5
223Fr
	234.74	54.4	3.0

In this region of interest only γ-rays from ²²⁷Th, ²²³Ra and ²²³Fr have significant emission probabilities, and are included in Equation 2.

Results

Actinium and thorium belong to the actinide metals group, while radium is an alkaline earth metal. As such, they are characterized by different ionic charges: +3; +4 and +2 respectively. These charge discrepancies are beneficial to the separation and purification of each element. When associated to nitrates, ²²⁷Ac, ²²⁷Th and ²²³Ra form various negatively charged species. Utilizing charge affinity chromatography, each metal can be separated. When dissolved in a mixture of HNO₃ <0.5 M/Methanol >70% and adsorbed on strong anion exchange resin, the following radioisotopic retention affinity has been reported: ²²⁷Th-nitrates > ²²⁷Ac-nitrates > ²²³Ra-nitrates (Faris et al., 1964; Guseva et al., 2004). Here we performed chemical separation of ²²³Ra on the kBq scale from cyclotron-produced ²²⁷Ac (Figure 1B). Previous ²²³Ra purifications utilize this strategy to produce GBq of ²²³Ra starting from pure ²²⁷Ac source material (Mokhodoeva et al., 2015; Soderquist et al., 2012).

Sample quality control was assessed by physical and radiochemical evaluation, utilizing a dose calibrator, a well counter and a HPGe detector. The material was then compared with clinical grade ²²³RaCl₂ supplied by the Johns Hopkins University School of Medicine radiopharmacy to conduct an evaluation of the biological distribution in healthy living mice to confirm the high quality of the produced ²²³Ra.

The source, a mixture of ²²⁵Ac/²²⁷Ac/²²⁷Th/²²³Ra (740–925 kBq), was adsorbed onto the resin. The purification and acid wash off were conducted and checked for radioactive recovery immediately after collection using the dose calibrator. The radioactive recovery of all purified ²²³Ra fractions was estimated to be 71.3±5.7% (n=5) of the initial source deposited onto the column. We note this is roughly what is expected based on the dose-rates ratio of ²²³Ra (+daughters) to the original column activity. This is estimated to be 68.5% based on the data from the NNDC for these isotopes. The acidic wash HNO₃ (0.5 M) was later conducted to recover ²²⁷Ac, ²²⁵Ac and ²²⁷Th (Figure 1B).

The purified ²²³Ra fractions (8–15 mL elution, Figure 2A) were measured using the well counter over a period of 28 days (Figure 2B). This data was then fit to the time dependent decay (Equation 1S) and the detector response (Equation 2). The results of this fitting are shown as the blue line in Figure 2C. For comparison, we show as the red line in Figure 2 the expected decay if the sample were pure ²²³Ra. This comparison clearly indicates that the observed decay data is more accurately described by the more complete decay model. The results of the fitting of the time series data to the more complete decay model for all the elutions are shown in Table 1. The isotopic content of purified fractions at time t=0 indicates that all eluted fractions (8–15 mL) show less than 1% of ²²⁷Ac and ²²⁷Th contaminant, by activity, demonstrating limited breakthrough of the parent isotopes.

In addition to half-life determination, γ-energy spectra were compared between purified samples, acid wash and column residual (Figures 3A–F). In the purified ²²³Ra fractions (Figure 3A, and 3B), primarily ²²³Ra was detected without any significant contaminant. In Figure 3B we show the result of fitting the Gaussian peaks to the narrow spectral region and determine that this sample is primarily ²²³Ra with less than 1% by activity of ²²⁷Th. This is consistent with the time dependence analysis.

In contrast, the acid wash presented a mix of isotopic content (as identified by their characteristic γ-ray lines) including ²²³Ra, ²²⁷Th, ²²³Fr and ²²⁵Ac (through its daughter ²¹³Bi) (Figure 3C). Fitting the spectrum in the narrow region 220–280 keV to the Gaussian peak (Figure 3D) we estimated the isotopic ratio of ²²⁷Th/²²³Ra to be 41.6±2.5%. Additionally, the large ²¹³Bi signal in Figure 3C, is indicative of ²²⁵Ac, strongly suggests ²²⁷Ac washed off from the column.

Similarly, the column residual presented a mix of isotopic content (as identified by their characteristic γ-ray lines) including ²²³Ra, ²²⁷Th and ²²³Fr as seen shown in Figure 3E. However, unlike Figure 3C, there is little evidence of ²²⁵Ac (through it daughter ²¹³Bi), indicating that little ²²⁵Ac or ²²⁷Ac remains on the column. Fitting the spectrum in the narrow region 220–280 keV to the Gaussian peaks, as seen in Figure 3F, we estimated the isotopic ratio of ²²⁷Th/²²³Ra to be 702±8 %.

In order to verify the quality of the microgenerated material, we undertook biodistribution studies using the in-laboratory produced ²²³Ra and clinical grade ²²³RaCl₂. We compared uptake in mice of the same age and same breed. For both groups, mice were sacrificed and organs harvested 24 h post-injection. No statistical difference of organ distribution was noted. As expected, the bulk of ²²³Ra was absorbed into bones and mineral elements, 2.5–5% of the initially administered activity as shown in Figure 4. Significant accumulation in cecum and spleen were observed as well with 2.5% and 0.5% respectively. These values correlate with previously reported small-animal studies using ²²³Ra (Abou et al., 2016).

Fig 4.

Biodistribution: ²²³Ra localization in organs comparing clinically supplied and in laboratory generated material using C57Bl/6 mice (>14 weeks). The mice were sacrificed 24 h post-injection (p.i.). Following organs harvesting, tissue specimen were weighed and gamma counted to define percentage of injected activity.

Discussion

There are several advantages to alpha-particle radiotherapy. Radium-223 produces a total of 4 alpha-particles through its decay. When administered, the high linear-energy-transfer of these emissions results in the deposition of alpha particles within a vicinity of only several cell diameters. The short-range spares distant non-diseased surrounding tissue (Hobbs et al., 2012). The irradiation is also highly potent, causing irreversible DNA damage, fatal to the targeted cells (Wideroe, 1979). ²²³Ra is a mimetic of calcium, and as a result is a bone-seeker. The recently approved radiopharmaceutical ²²³RaCl₂ (tradename Xofigo) is of considerable research interest due to its noted survival benefit and palliative effect provided to patients with bone-metastatic castration resistant prostate cancer (Parker et al., 2013). However, ²²³Ra and associated source materials such as ²²⁷Ac or ²²⁷Th are difficult to access for the sole purpose of preclinical research. For these reasons alternate ²²³Ra production approaches are needed to make this material widely available for laboratory investigation. In this work we describe a process to produce laboratory-scale quantities of ²²³Ra (kBqs) with high separation capacity of the radiometal mixture for repeatable use of the parent radionuclides ²²⁷Th and ²²⁷Ac.

Previously, methods for ²²³Ra production have been conducted at industrial scale (GBqs) using parent radionuclide sources that are not widely available. A recent advance in the cyclotron production of ²²⁵Ac includes a small amount of ²²⁷Ac side-product. To that end, we undertook the recycling of this perceived contaminant for ²²³Ra production. Using this radioisotopic waste, may reroute unwanted ²²⁷Ac from a long-term storage-hazard towards high radiopurity ²²³RaCl₂ for research use. This approach is attractive to permanently supply the currently pursued radiopharmaceutical drug ²²³RaCl₂.

Using a miniaturized purification system, our method resulted in the reliable production of high quality grade carrier-free ²²³Ra, with attractive yield and recovery of the parent isotopes. Five complete elution and column recovery cycles were conducted during the course of this study. We did not observe any signs of resin oxidation or damage and no noticeable microgenerator radiolysis was seen. This is likely due to the low radioactive content of our samples associated with a large chromatographic resin surface (<1000 kBq for 1 g of resin). Column conditioning was performed prior to each purification to limit particle desorption, yet the presence of ²²⁷Th or ²²⁷Ac-labeled mesh may not be excluded. Fortunately, negligible radioactive contaminants were detected in ²²³Ra samples, limiting the likelihood of this phenomenon.

Purity of in-laboratory produced ²²³Ra was confirmed by both physical and biological evaluations. In vivo examination of the produced ²²³Ra showed no significant difference in organ distribution as compared to the clinical grade sample. The radioactive decay fitting provided an overview of the sample quality. We estimate that no more than 1% (by activity) of either ²²⁷Th or ²²⁷Ac impurities at time t=0 is present in the purified ²²³Ra elutions. In addition the HPGe results indicate that ²²⁷Th and ²²⁷Ac were both retained by the column while purifying ²²³Ra. As a result, no apparent breakthrough occurred during the purification.

The HPGe results indicate that the column wash off was successfully conducted and contained ²²³Fr, ²¹³Bi, ²²⁷Th and ²²³Ra. Detection of ²¹³Bi revealed undecayed ²²⁵Ac, present in the wash off. Additionally, ²²⁵Ac detected in the column wash-off shows that actinium is desorbed upon acid wash of the column. The apparent concentration of ²¹³Bi appears to be significantly reduced in the column residual. This is additional evidence of ²²⁷Ac desorption when washing off. In contrast, ²²⁷Th was identified and quantified in both wash off and column residual. However given the time between the acid wash and the HPGE measurements (~2 days), and the time between the ²²³Ra elutions and the acid wash (23 days); the ²²³Ra measured in the acid wash is consistent with the in-growth of ²²³Ra from the parents ²²⁷Th and ²²⁷Ac. The ²²⁷Th content in the acid wash appears greater than expected from the ²²⁷Ac in-growth and is suggestive of at least partial desorption of ²²⁷Th from the column. ²²⁷Th and ²²³Ra detected in the column residual are consistent with ²²⁷Th remaining on the column after acid rinse and consequently the ²²³Ra in growth. In summary, the HPGe measurements show that significant amount of ²²⁷Ac is recovered ensuring a long-term supply of ²²³Ra for research studies.

The utility of this work has been demonstrated through several convenient aspects: (1) ²²⁷Ac originally perceived as a waste from cyclotron-produced ²²⁵Ac source, has been converted into a pure alpha particle emitting radiopharmaceutical drug; (2) With the complete recovery of ²²⁷Ac, ²²³Ra microscale production is repeatable; (3) ²²⁷Th was separated from ²²⁷Ac, opening another perspective for alpha-particle emitter production.

Conclusion

We have constructed a ²²³Ra microgenerator using recycled ²²⁷Ac from a ²²⁵Ac cyclotron-produced source. kBqs of pure ²²³Ra-nitrate were consistently generated and further converted into a readily injectable ²²³RaCl₂ for laboratory in vivo experiments. The parent radioisotopic source was recovered enabling repeated production. This opens the path for expanded pre-clinical investigation to better understand the distribution and radiobiological effects of ²²³RaCl₂ therapy in the context of prostate and breast cancer.

Table 3.

The ratio of ²²⁷Th to ²²³Ra activities extracted from HPGe fit (t=15 days post-elution) for purified ²²³Ra purified elutions (500 µL), acid wash (1 mL) and column residual (full column).

Sample	227Th/223Ra(%)
223Ra (8.5ml fraction)	0.0 ±1
Acid wash	41.6± 2.5
Column residual	702±8

Highlights.

• A ²²³Ra microgenerator was built using residual ²²⁷Ac from cyclotron-produced ²²⁵Ac.

• Following ²²⁵Ac decay, the residual ²²⁷Ac was processed into pure ²²³Ra.

• ²²⁷Ac and ²²⁷Th were recovered in high yield for a permanent supply of ²²³Ra.

• Clinically supplied and generator-produced ²²³Ra have equivalent in vivo distribution.

• Microdose column provides sufficient material for research use.