Biodistribution of free Francium-221 and Bismuth-213 in Tumour-bearing SCID mice after successful development of Actinium-225/Francium-221 radionuclide generator Set-up

Abstract

Purpose

Despite the clinical evidence of actinium-225 (²²⁵Ac)-based targeted alpha therapies (TαT) efficacy, optimized treatment regimens are needed to improve overall clinical response rates and decrease toxicities. The nuclear recoil effect of ²²⁵Ac and its resulting daughter nuclides have been hypothesized to contribute to non-targeted damage. However, a lack of generator concepts for radionuclidically pure francium-221 (²²¹Fr), involvement of strong acids for elution, and its short half-life (4.8 min), has limited in vivo studies. Here, we report on a successful application of an ²²⁵Ac/²²¹Fr generator concept and the in vivo distribution of ²²¹Fr and bismuth-213 (²¹³Bi).

Methods

The immobilization of ²²⁵Ac and elution of ²²¹Fr was performed on a LN2 resin column. The biodistribution of ²²¹Fr and ²¹³Bi was investigated in male SCID mice with LNCaP tumors at 5 and 15 min p.i.

Results

Our results indicate that LN2 resin is a highly efficient resin for selective separation of ²²⁵Ac and ²²¹Fr. The use of 0.1 M NaOAc enabled continuous elution at a constant pH. The biodistribution study revealed a fast distribution of ²²¹Fr and ²¹³Bi already 5 min p.i. We observed a strong accumulation of ²²¹Fr to the kidneys, salivary glands and small intestine. In ²¹³Bi-injected mice, the highest accumulation was in kidney and liver.

Conclusion

We present an unprecedented concept utilizing LN2 resin in ²²⁵Ac/²²¹Fr generator applications. Successfully eluted and injected ²²¹Fr fractions showed strong accumulation of ²²¹Fr and ²¹³Bi in key organs. Our data provide preliminary evidence of the potential contribution of recoiled progeny radionuclides to side-effects in non-targeted organs.

Supplementary information

The online version contains supplementary material available at 10.1007/s00259-025-07427-4.

Introduction

Since the first U.S. Food and Drug Administration (FDA) approval of radium-223 dichloride [²²³Ra]RaCl₂, Xofigo^®, Bayer, Leverkusen, Germany) in 2013 [1], targeted alpha therapies (TαT) have attracted a continuous interest in the field of modern nuclear medicine. Actinium-225 (²²⁵Ac) has recently shown exciting clinical data when used in combination with prostate-specific membrane antigen (PSMA)-targeting radioligands such as [²²⁵Ac]Ac-PSMA-617 (Action, NCT04597411); [²²⁵Ac]Ac-PSMA I&T (TATCIST, NCT05219500); and [²²⁵Ac]Ac-PSMA-R2 (SatisfAction, NCT05983198). In contrast to β⁻-emitters, α-emitters release high energy (several MeV) over short tissue range (< 0.1 mm) resulting in high linear energy transfer (LET), leading to a much higher killing efficiency of cancerous cells without damaging wider surrounding healthy tissue [2]. On the other hand, unfavorable half-lives, complicated decay pathways, production, and availability issues render only a limited number of α-emitters realistically suitable for TαT [3]. Among these, thorium-227 (²²⁷Th, T_1/2 = 18.7 d, E_α = 6.0 MeV, E_γ = 236 keV, I = 12.9%), ²²⁵Ac (T_1/2 = 9.9 d, E_α = 5.8 MeV), ²²⁴Ra (T_1/2 = 3.6 d, E_α = 5.7 MeV, E_γ = 241 keV, I = 4.1%), ²²³Ra (T_1/2 = 11.4 d, E_α = 5.7 MeV, E_γ = 269 keV, I = 13.3%), bismuth-213 (²¹³Bi, T_1/2 = 45.6 min, E_α = 5.9 MeV, E_γ = 440 keV, I = 25.9%), lead-212 (²¹²Pb, T_1/2 = 10.6 h, E_β,max = 0.57 MeV, E_γ = 239 keV, I = 43.6%), and astatine-211 (²¹¹At, T_1/2 = 7.2 h, E_α = 5.9 MeV) have been thoroughly investigated [2, 4–6].

In TαT, ²²⁵Ac, a so-called alpha in vivo nanogenerator, is currently considered to be the most promising α-emitter for pre-clinical and clinical applications. In its decay chain, a total of 28 MeV is released [7] through four dominant α-disintegrations (five in total) and two dominant β^–-disintegrations (three in total) until the stable bismuth-209 (²⁰⁹Bi) is reached (Fig. 1). To make best use of ²²⁵Ac as a therapeutic agent, a thorough understanding of the in vivo distribution of ²²⁵Ac and its daughter radionuclides is essential for accurate dosimetry calculations. The ²²⁵Ac decay provides two useful γ-emissions for detection, 218 keV (I = 11.4%, francium-221 (²²¹Fr), 99.9% α-branching and 0.01% β⁻-emission followed by α-decay) and 440 keV (I = 25.9%, ²¹³Bi, 2.1% α-branching and 97.9% β⁻-emission followed by α-decay).

Fig. 1.

Schematic representation of the alpha in vivo nanogenerator ²²⁵Ac (T_1/2 = 9.9 d, E_α = 5.8 MeV). The most prominent daughter radionuclide is represented by ²¹³Bi (T_1/2 = 45.6 min, E_α = 5.9 MeV, E_γ = 440 keV, I = 25.9%) which is also commonly applied for TαT. Its first daughter radionuclide ²²¹Fr (T_1/2 = 4.8 min, E_α = 6.3 MeV, E_γ = 218 keV) is used for the characterization as ²²⁵Ac forms so called secular equilibrium with ²²¹Fr. The figure was taken from Roscher et al., 2020 [8]

In TαT, different chelation strategies are utilized to safely transport the radioactivity to designated tumor sites within a dedicated targeting ligand. Currently, 1,4,7,10-tetraazacyclododecane-1,4,7,10-tetraacetic acid (DOTA)- and N,N’-bis[(6-carboxy-2-pyridil)methyl]−4,13-diaza-18-crown-6 (macropa)-based constructs are recognized as gold standards [9]. However, possible in vivo decomplexation or nuclear recoil phenomena can impede the intended outcomes of applied radiopharmaceuticals [10]. In case of ²²⁵Ac, the first daughter radionuclide ²²¹Fr is not retained by the chelating agent and can diffuse away from the targeting moiety [11], especially prior to internalization. This phenomenon is even more pronounced for the downstream daughter radionuclides of ²²¹Fr, particularly ²¹³Bi, due to its longer half-life (45.6 min).

A biodistribution study using radionuclidically pure ²²¹Fr would enable quantification of its diffusion and redistribution. For this purpose, extraction chromatography has been suggested as a suitable purification method, in which highly selective extractants fixed on an inert support, e.g. a polymer matrix or inorganic materials, serve as stationary phase in applied columns. The mobile phase usually consists of an acidic solution, e.g. HNO₃ or HCl [12]. Previously, a new and fast generator concept has been presented to provide a ²²¹Fr source from ²²⁵Ac [13]. Diglycolamides (TODGA, isoTODGA) were immobilized on a polyacrylonitrile (PAN) matrix and subsequently evaluated in batch and column experiments. Column experiments revealed isoTODGA-PAN and 4 M HNO₃ to be an ideal combination of extractant and eluent for ²²⁵Ac sorption. Under these conditions, ²²¹Fr yields of > 65% were achieved in approximately 2.5 mL of eluate with ²²⁵Ac contaminations below 1%, but not with the desired pH nor with sufficient amounts and purities of ²²¹Fr for biodistribution studies.

In this work, we aimed to develop an ²²⁵Ac/²²¹Fr radionuclide generator set-up for direct in vivo applications to study the biodistribution of the two main ²²⁵Ac progenies. Understanding the behavior of non-chelated α-emitters is crucial, as they might redistribute within the body, potentially increasing radiation risks to healthy organs and tissues. Our study seeks to provide valuable insights into their effects, ultimately contributing to a more comprehensive assessment of dosimetry and radiation safety of TαT.

Materials and methods

Materials

LN2 resin (G71m resin loaded with 2-ethylhexylphosphonic acid mono-2-ethylhexyl ester (HEH[EHP])) was purchased from TrisKem International having the following physical and chemical properties: particle size = 500–100 µm, density = 0.37 g/mL, capacity = 0.16 mmol/mL resin, conversion factor D_W/k’ = 1.82).

²²⁵Ac was provided by Global Morpho Pharma (GMP, La Chapelle-sur-Erdre, France) as ²²⁵Ac(NO₃)₃ dry film. The full activity batch was subsequently diluted in 150 µL ultra-pure and metal-free 0.05 M HCl and transferred into a 1.5 mL Eppendorf tube (Protein LoBind^®).

Phosphate-buffered saline (PBS, 1X, pH 7.4) was purchased from Gibco, 0.15 M sodium chloride (NaCl, pH 5.5) and sodium acetate (NaOAc) from Sigma-Aldrich. TraceMetal Grade nitric acid (HNO₃, 67–69%) was purchased from Fischer Scientific, HIPERPUR-plus hydrochloric acid (HCl, 35%) from PanReac AppliChem, and TraceSELECT™ H₂O from Honeywell.

Radionuclide separation

²²⁵Ac(NO₃)₃ (~ 50 MBq, 150 µL 0.05 M HCl) was diluted with 450 µL 0.01 M HNO₃ to a total volume of 600 µL and loaded onto 0.3 mL (Height: 19.8 mm x Diameter: 11.7 mm) LN2 column, which was pre-conditioned with at least three column volumes of 0.01 M HNO₃. Afterwards, the column was washed with 800 µL of 0.01 M HNO₃ (pH 2.0), 0.15 M NaCl (pH 5.5), PBS (pH 7.4), and 0.1 M NaOAc (pH 6.5). Subsequently, multiple ²²¹Fr elutions were performed with 800 µL 0.1 M NaOAc (pH 6.5) after sufficient ²²¹Fr in-growth time of 25–30 min. Four fractions of 150, 300, 300 and 150 µL were collected and measured immediately in a dose calibrator (ISOMED 2010, Nuvia Instruments, Dresden, Germany). Quality control was conducted retrospectively by time and energy-dependent measurements in a gamma counter (Wizard 2410, with 10 independent NaI(Tl) detectors, Perkin Elmer, Waltham, Massachusetts, USA) and a gamma spectrometer HPGe(Li) detector, Mirion Technologies, Atlanta, Georgia, USA). The activity of ²²¹Fr and ²¹³Bi was assessed from the photo-peaks of 218 and 440 keV, respectively.

Cell line

LNCaP cells, derived from a human lymph node metastatic lesion of prostatic adenocarcinoma, were obtained from the German Collection of Microorganisms and Cell Cultures (Leibniz Institute DSMZ, Braunschweig, Germany).

Animal model

For the biodistribution study, male SCID (CB-17/Icr-Prkdc^scid/scid/Rj) mice (20–25 g, 7-weeks-old, Janvier Labs, Le Genest-Saint-Isle, France) were implanted subcutaneously (s.c.) with testosterone pellets (12.5 mg, 4 mm, prepared in-house). One day later, mice were inoculated s.c. with 5 × 10⁶ LNCaP cells into the right flank. All animal experiments were conducted in accordance with the German Animal Welfare Law and approved by the local authorities.

Biodistribution of ²²¹Fr in vivo

On Day 35 after inoculation, the mice (n = 3/group) were injected with a single dose of ²²¹Fr and sacrificed at 5 min post-injection (p.i.). On Day 37 after inoculation, the mice (n = 3/group) were injected with a single dose of ²²¹Fr and sacrificed at 15 min p.i. The procedure for each mouse was as follows: 25 min after the previous ²²⁵Ac/²²¹Fr generator elution, a fresh elution of four fractions (150, 300, 300 and 150 µL) was taken and the second fraction was checked for pH (> 6.0 and < 7.0) and a 100 µL intravenous (i.v.) tail vein injection into one mouse followed within 2–3 min after elution. Standards of the injected volume were taken for each injected elution and measured in parallel. Throughout the study, a timer was continuously used to record exact time-points for all actions.

After 5–15 min the mouse was sacrificed and the following organs were collected: kidney, tumor, blood, liver, small intestine, salivary gland, large intestine, pancreas, spleen, muscle. The organs were halved or divided for pair organs to allow for two parallel samples for measurement 3–5 min after start of sectioning. One set was measured once by gamma spectrometry measurements HPGe(Li) detector. The other set of samples was measured several times in a gamma counter by NaI(Tl) crystal. The next mouse was injected ~ 50 min later.

Biodistribution of ²¹³Bi in vivo

Biodistribution of ²¹³Bi was performed analogically as for ²²¹Fr. On Day 35 after inoculation, the mice (n = 3/group) were injected with a single dose of ²¹³Bi and sacrificed at 5 min p.i. On Day 37 after inoculation, the mice (n = 3/group) were injected with a single dose of ²¹³Bi and sacrificed at 15 min p.i. For the ²¹³Bi biodistributions the elutions were approximately 2 h old to not contain any ²²¹Fr or ²¹⁷At. The three mice for each time-point were injected in 5 min intervals on one day. Standards of the injected volume were taken and measured in parallel. The collection of organs and measurements was analogous to the ²²¹Fr biodistribution, as described in the previous paragraph.

Measurement HPGe(Li) detector

The gamma spectrometry measurements were performed with a HPGe(Li) detector purchased from Mirion Technologies, Atlanta, Georgia, USA. Measuring time was 5–6 min per sample. The activity of ²²¹Fr and ²¹³Bi were assessed from the gamma lines specifically attributed to these radionuclides.

Measurement NaI(Tl) detector

Wizard 2410 employing 10 independent NaI(Tl) detectors was purchased from Perkin Elmer, Waltham, Massachusetts, USA. Each sample was measured six times over a period of 40–45 min, with each measurement lasting 1 min. The acquisitions were performed in the channels of 180–240 keV for ²²¹Fr and 380–520 keV for ²¹³Bi.

Quantification of ²¹³Bi and ²²¹Fr activities in ²¹³Bi- and ²²¹Fr-injected mice

To determine the distribution of ²¹³Bi and ²²¹Fr in organs, we analyzed data from two experimental setups: mice injected with ²¹³Bi, and mice injected with ²²¹Fr. We began by processing the data from the ²¹³Bi-injected mice, where no ²²¹Fr was present in the tissues. Thus, ²¹³Bi activity was calculated by fitting an exponential decay function [Eq. (1)] to the CPM values measured in the ²¹³Bi energy window, with constraints ensuring non-negative initial CPM and background.

For the ²²¹Fr-injected mice, we accounted for the fact that ²¹³Bi decay also contributes to the signal in the ²²¹Fr window. Additionally, the calculation of ²¹³Bi activity in these animals required fitting a more complex function derived from the Bateman equations.

Calibration was performed using ²²⁵Ac in secular equilibrium, which provided known activities for both ²¹³Bi and ²²¹Fr. Linear fits through the origin yielded calibration factors of f_Bi =1.649 +/- 0.062 Bq⁻¹ and f_Fr = 3.513 +/- 0.012 Bq⁻¹.

All fits were performed in OriginPro 2021 using the Levenberg–Marquardt algorithm, with instrumental weighting to account for CPM uncertainty. The calculation procedure is described in more detail in the Supplementary Material.

Results

Focusing our literature research on resins potentially showing good retention of only Ac-225 and none of its daughter radionuclides, we focused on the three structurally different variations (LN, LN2, LN3) of the LN resin series, offering different acidities (Fig. 2) in the following decreasing order: LN > LN2 > LN3 [14]. Out of these three resins, the LN2 resin exhibited the most optimal performance in terms of ²²⁵Ac retention under loading and elution conditions, as well as in terms of pH, elution volume and purity of ²²¹Fr.

Fig. 2.

Structural details of extractants impregnated onto an inert support in the LN resin series (LN, LN2 and LN3 resins). LN resin contains bis(2-ethylhexyl) hydrogen phosphate (HDEHP), LN2 resin contains 2-ethylhexylphosphonic acid mono-2-ethylhexyl ester (HEH[EHP]) and LN3 contains bis(2,4,4-trimethylpentyl)phosphinic acid (H[TMPeP])– Cyanex272)

Elution of ²²¹Fr from ²²⁵Ac/²²¹Fr radionuclide generator

²²⁵Ac(NO₃)₃ in 0.01 M HNO₃ (~ 50 MBq, 600 µL, pH 1.0) was manually loaded on a LN2 resin column (V_c = 300 µL) and subsequently washed with 0.01 M HNO₃ (900 µL) to clear the column of all ²²⁵Ac progeny radionuclides (²²¹Fr, ²¹⁷At, ²¹⁷Ra, ²¹³Bi, ²¹³Po, ²⁰⁹Tl, ²⁰⁹Pb, ²⁰⁹Bi). In order to obtain ²²¹Fr fractions with a more physiological pH which can be tolerated by mice upon injection, eluents with a pH range of 6.5 to 8.5 were chosen. Therefore, the column was manually washed with an elution volume (V_E) of 800 µL of each 0.15 M NaCl (pH 5.5), PBS (pH 7.4) and 0.1 M NaOAc (pH 6.5) manually after loading with ²²⁵Ac (Fig. 3).

Fig. 3.

Schematic representation of ²²⁵Ac/²²¹Fr radionuclide generator set-up. ²²⁵Ac(NO₃)₃ was loaded on a LN2 resin column and subsequently washed with 0.01 M HNO₃ and stabilized with 0.15 M NaCl, PBS, and 0.1 M NaOAc. Upon secular equilibrium, ²²¹Fr was eluted with 0.1 M NaOAc (pH 6.5) and applied for in vivo biodistribution studies without any re-formulation

In our experiments, 0.1 M NaOAc proved to be an ideal eluent, maintaining the column at a physiologically relevant pH of approximately 6.5, which was essential for downstream in vivo applications. This buffer system not only ensures chemical stability but also significantly minimizes early ²²⁵Ac “breakthrough” and “bleeding” phenomena, which are critical for achieving high radionuclidic purity of the eluted ²²¹Fr. After an in-growth time of 25 min, manual elution of ²²¹Fr fractions with 0.1 M NaOAc (V_E = 800 µL) was consistently feasible up to eight times per day, enabling a practical and reproducible workflow. Furthermore, we have assessed the purity of 24-hour-old eluates and found that in all but one instance (out of 18 eluates analyzed), both ²²¹Fr and ²¹³Bi were below the lower limit of quantification (LLOQ), indicating only minimal ²²⁵Ac breakthrough. However, a notable disadvantage of using NaOAc was a reduced elution efficiency, with a decrease in ²²¹Fr stripping yield of approximately 20% compared to stronger mineral acids as eluents.

Still, high ²²¹Fr elution yields of > 50% in comparison to immobilized ²²⁵Ac were possible with ²²⁵Ac contaminations remaining significantly low (< 0.1%). The eluted ²²¹Fr activity would have been sufficient for injecting three mice per time-point while obtaining enough signal for the measurement upon all planned biodistribution studies. Based on the short half-life of ²²¹Fr and resulting handling challenges, we decided to inject a single mouse per elution.

Biodistribution of ²²¹Fr in ²²¹Fr-injected mice

Due to its short half-life, measuring the biodistribution of ²²¹Fr presents significant challenges. Organ samples were divided in half and simultaneously measured with both the HPGe detector (Supplementary Material, Figure S1) and the NaI(Tl) detector (Fig. 4). This strategy enabled the acquisition of complementary data, combining the HPGe detector’s superior energy resolution for precise radionuclide identification with the NaI(Tl) detector’s higher detection efficiency.

Fig. 4.

Biodistribution data from injection of ²²¹Fr in LNCaP tumor bearing mice. Results obtained from measuring with a NaI(Tl) detector depicted in (A) percentage of injected dose per gram organ or tissue (B) and as percentage of injected dose per organ or tissue. The values are mean values with standard deviation (n = 3 per time-point) considering linear propagation of uncertainties

The data from the NaI(Tl) detector showed a very rapid distribution of free ²²¹Fr after injection and at 5 min p.i., only 3.2% of the injected dose per g tissue (ID/g) of ²²¹Fr was still detected in the blood, which decreased further to 1.2% ID/g at the 15-min time-point (Fig. 4A). The kidneys showed the highest activity accumulation, with 35.7% and 32.4% ID/g at 5 and 15 min p.i., respectively. The salivary glands exhibited 28.7% and 25.1% ID/g (5 and 15 min p.i.), while the small intestine showed an accumulation of 7.5% and 10.8% ID/g (5 and 15 min p.i.). Also, the large intestine showed an accumulation of 7.4% and 5.2% ID/g (5 and 15 min p.i.). Other organs and tissues displayed minimal to no accumulation.

Analyzing the biodistribution data considering the entire organ, a more complete picture of where the ²²¹Fr accumulated in a mouse can be observed (Fig. 4B). Specifically, 19.7% ID accumulated in the kidney after 5 min and retained 18.0% ID after 15 min. The small intestine initially retained 7.0% ID after 5 min, which increased to 15.6% ID after 15 min. The whole muscle calculated weight exhibited relatively high ²²¹Fr accumulation with 16.1% ID and 12.4% ID (5 and 15 min p.i.). The uptake values observed in muscle may primarily reflect the tissue’s vascularization rather than true cellular uptake by myocytes, especially considering the rapid blood clearance of ²²¹Fr and ²¹³Bi and their potential redistribution during circulation. Nevertheless, these findings are important, as they suggest a possible source of off-target toxicity that has not been previously considered in ²²⁵Ac-based TαT. Some bias is also given with higher accumulation in organs with higher mass, however, even a smaller organ, such as the salivary glands, accumulated 5.9% ID within 5 min of the whole injected activity of ²²¹Fr and retained 4.1% ID after 15 min. This trend was also confirmed by measurements using the HPGe detector (Supplementary Material, Figure S1).

Biodistribution of ²¹³Bi in vivo in ²¹³Bi-injected mice

Similarly, for the biodistribution of ²¹³Bi, the organ samples were divided in half and measured simultaneously using the HPGe detector (Supplementary Material, Figure S2) and the NaI(Tl) detector (Fig. 5). The distribution of ²¹³Bi in the mice was fast, with only 5.9% ID/g remaining in the blood at 5 min and 1.4% ID/g at 15 min (Fig. 5A). The kidneys exhibited 24.8% ID/g at 5 min and 18.3% ID/g at 15 min. The liver showed an accumulation of 10.9% ID/g at 5 min, which increased to 16.0% ID/g at 15 min. The spleen showed some accumulation, with 4.9% ID/g at 5 min and 6.0% ID/g at 15 min. No significant accumulation of ²¹³Bi was observed in the other organs and tissues.

Fig. 5.

Biodistribution data from injection of ²¹³Bi in LNCaP tumor bearing mice. Results obtained from measuring with a NaI(Tl) detector depicted in (A) percentage of injected dose per gram organ or tissue (B) and as percentage of injected dose per organ or tissue. The values are mean values with standard deviation (n = 3 per time-point) considering linear propagation of uncertainties

When analyzing the biodistribution data across the entire organ, higher accumulation was observed primarily in the blood, liver, and muscle (Fig. 5B). Specifically, 12.0% ID accumulated in the liver at 5 min and retained 15.4% ID after 15 min. An initial 13.4% ID detected in the kidneys at 5 min and decreased to 9.2% ID after 15 min. In the blood, 11.1% ID was present 5 min after injection, but declined to 2.5% ID after 15 min.

Biodistribution of ²¹³Bi in vivo in ²²¹Fr-injected mice

The following study focused on the accumulation of ²¹³Bi generated from ²²¹Fr injections into the mice. Subsequently, these results were compared with data of direct ²¹³Bi injections (Fig. 6). Only 6.5% ID/g of ²¹³Bi was detected in the blood at 5 min, which decreased to 2.5% ID/g at 15 min (Fig. 6A). The kidneys showed accumulation of 32% ID/g at 5 min, increasing to 44% ID/g at 15 min. The salivary glands exhibited some accumulation, with 7.7% ID/g at 5 min and 8.8% ID/g at 15 min. The liver showed 6.6% ID/g at 5 min and 5.6% ID/g at 15 min. The spleen demonstrated 4.4% ID/g at 5 min, which decreased to 2.9% ID/g at 15 min. The small intestine showed accumulation of 2.9% ID/g at 5 min, increasing to 4.2% ID/g at 15 min. No significant accumulation of ²¹³Bi was observed in the other organs and tissues.

Fig. 6.

Biodistribution of ²¹³Bi from injection of ²²¹Fr in LNCaP tumor bearing mice. Results obtained from measuring with a NaI(Tl) detector depicted in (A) percentage of injected dose per gram organ or tissue (B) and as percentage of injected dose per organ or tissue. The values are mean values with standard deviation (n = 3 per time-point) considering linear propagation of uncertainties

To elucidate the organ-specific retention of ²²¹Fr and ²¹³Bi, we compared their activity ratios in various mouse organs to the identical ratio of a standard sample undergoing physical decay only (Fig. 7). This comparison assumes that if ²²¹Fr and ²¹³Bi had the same biodistribution across all organs, the plotted activity ratios would equal 1.00 for all organs (²²¹Fr = ²¹³Bi). Ratios greater than 1.00 would indicate organs of a higher affinity for ²²¹Fr (²²¹Fr > ²¹³Bi), while ratios lower than 1.00 would suggest a preference for ²¹³Bi (²²¹Fr < ²¹³Bi).

Fig. 7.

²²¹Fr/²¹³Bi activity ratio in the organs of LNCaP tumor-bearing mice, measured with a NaI(Tl) detector, compared to a reference sample undergoing physical decay only

The highest ratios were observed in the small intestine (2.55 and 2.64 at 5 and 15 min), the large intestine (2.77 and 2.52 at 5 and 15 min), and the salivary glands (3.51 and 3.13 at 5 and 15 min), respectively. In these organs, the elevated ratios indicate a strong preferential retention of ²²¹Fr over ²¹³Bi. In contrast, a higher affinity for ²¹³Bi is observed in the blood and liver. However, the ratios observed in the blood are not indicative of ²²¹Fr or ²¹³Bi affinities. The retention in all organs significantly decreased the activity concentration in the blood. This suggests that rapid clearance from circulation, rather than strong binding to blood components [15], is the predominant factor influencing blood activity levels. For the remaining organs, no significant difference in affinity between ²²¹Fr and ²¹³Bi was observed.

Discussion

Our establishment of a new ²²⁵Ac/²²¹Fr radionuclide generator allows the elution of ²²¹Fr (and the follow up daughter ²¹³Bi) under physiological conditions suitable for injection into a living organism and represents a crucial step towards understanding the behavior and biodistribution of non-complexed radionuclides in mammals. The observation that daughter radionuclides produced by alpha emitters do not remain at the site of the mother radionuclide led to the hypothesis that daughter radionuclides contribute not only to the therapeutic efficacy but also appear to play a role in adverse events. A publication by Kratochwil et al. described high-resolution quantitative Positron Emission Tomography (PET) imaging with [⁶⁸Ga]Ga-PSMA-617 while extrapolating radiation dosimetry for [²¹³Bi]Bi-PSMA-617 and comparing its therapeutic index with [²²⁵Ac]Ac-PSMA-617 [16]. Under the assumption of homogenous dose distribution, equivalent doses for critical organs and tumor lesions exhibited notable differences in dose distributions to off-target organs. In contrast to [²¹³Bi]Bi-PSMA-617 (1.2 GBq), [²²⁵Ac]Ac-PSMA-617 (7.4 MBq) delivered higher tumor doses (42.1 vs. 7.6 Sv_RBE5), also resulting in significantly higher salivary gland exposure (17.0 vs. 9.7 Sv_RBE5). This near two-fold dose to salivary glands is hypothesized to stem from the complex decay chain of ²²⁵Ac.

Earlier studies investigated the biodistribution of ²²¹Fr and ²¹³Bi following treatment with an ²²⁵Ac-labeled antibody which acted as an in vivo generator for ²²¹Fr and ²¹³Bi. Song et al. used an anti-HER-2/neu monoclonal antibody in HER-2/neu transgenic mice and found not only very good therapeutic efficacy but also accumulation of ²²¹Fr and ²¹³Bi in the kidneys, which probably contributed to the long-term renal toxicity observed in the surviving mice [17]. Miederer et al. administered high activities of an ²²⁵Ac-labeled IgG3 antibody and showed redistribution of ²²¹Fr and ²¹³Bi to the kidney, small intestine and, to a lesser extent, the stomach [18]. Similarly, Schwartz et al. found that non-equilibrium ²¹³Bi from a ²²⁵Ac-labeled huM195 antibody contributes a significant fraction of the total radiation dose to the kidney [19].

While these studies provide insights on the fate of the daughter radionuclides, they are hampered by the set-up of using an antibody as an in vivo generator for ²²¹Fr and ²¹³Bi. The antibody itself exhibited a specific pharmacokinetic distribution pattern, with the location of the in vivo generators/antibodies constantly changing. Additionally, the antibody also supplied a constant rate of ²²¹Fr and ²¹³Bi over the whole study making it hard to distinguish between an ²²¹Fr or ²¹³Bi generated in the organ vs. an ²²¹Fr or ²¹³Bi redistributed into the organ. From an experimental perspective, conducting biodistribution studies with ²²¹Fr is a significant challenge due to its short half-life. As an alternate approach, the less accessible ²²⁷Ac (T_1/2 = 21.7 years) could be employed instead, as its first progeny ²²³Fr has a half-life approximately four times longer than ²²¹Fr used in this study [20]. This extended half-life of 22 min would potentially improve the feasibility of experimental studies on francium’s biochemical properties. Another strategy involved the cesium-132 (¹³²Cs) isotope, a same-group element, with a reasonable half-life of T_1/2 = 6.48 h [21]. In mice, the uptakes of ¹³²Cs to kidneys and blood (33.0 ± 7.7; 0.48 ± 0.05% ID/g) were much lower compared to ²²¹Fr-based biodistribution studies (52.3 ± 8.4; 5.4 ± 0.3% ID/g) [22] indicating that ¹³²Cs is not biochemically analogous to ²²¹Fr.

To overcome these limitations and enable the investigation of the distribution characteristics of ²²¹Fr or ²¹³Bi a new type of a radionuclide generator was needed. However, the short half-life of ²²¹Fr (T_1/2 = 4.8 min) offers distinct operational challenges of such generators, additionally calling for novel and innovative concepts for proper ²²¹Fr retrieval.

Yuan et al. designed and set up one of the first ²²⁵Ac/²²¹Fr generator concepts [23]. For this purpose, a DOTA-biotin construct was radiolabeled with ²²⁵Ac, subsequently reacted with an immobilized avidin column and eluted with 2 mL of sterile saline to yield desired ²²¹Fr fractions. However, the utilization of such generator principles for short-lived radionuclides is not favored due to the difficult operation, complexity, and overall cost inefficiency.

Aside from amide-based resins such as diglycoamide (DGA), organophosphorous-based materials have proven efficient in the extraction and separation of actinides. Their phosphorus-oxygen (P = O) and phosphorus-sulfur (P = S) centers of high polarity and charge density allow the formation of complexes with desired f-block elements, even in complex mixtures [24]. Furthermore, the extractants can be easily tuned by side chain modifications for improved solubility, lipophilicity, and selectivity [25].

Extraction chromatographic resins based on carbamoylphosphine oxide (CMPO) dissolved in tri-n-butyl phosphate (TBP) [26] and di(2-ethylhexyl)orthophosphoric acid (HDEHP) [14] have proven to successfully immobilize ²²⁵Ac³⁺ ions, which have not been utilized in ²²¹Fr separation procedures yet. In extraction chromatographic studies of ²²⁵Ac and various recoil effect elements (REE), Ostapenko et al. have concluded that the optimal ²²⁵Ac sorption conditions in LN resins are achieved in diluted nitric acid (0.05 M HNO₃), as opposed to TRU (2–4 M HNO₃) and DGA resins (4–7 M HNO₃) [27]. With respect to these observations, we chose to work with the LN resin series for our ²²⁵Ac/²²¹Fr radionuclide generator studies.

Other than the above-mentioned TRU and DGA resins, which are based on neutral extractants– thus requiring high HNO₃ concentrations for strong retention of Ac– the LN resin series is based on liquid cation exchangers. Accordingly, the extraction of cations such as Ac depends on the pH of the aqueous solution, with higher pH values leading to stronger retention. This makes this type of resin particularly suitable in this case as the ²²⁵Ac needs to remain retained on the resin while ²²¹Fr is eluted under near neutral conditions for direct in vivo application.

We aimed to evaluate a novel proof-of-concept for a ²²⁵Ac/²²¹Fr generator and to investigate the biodistribution profiles of free ²²¹Fr and ²¹³Bi to potentially explain off-target effects in TαT. Overall, our data confirmed the accumulation of ²²¹Fr and ²¹³Bi in various organs. The first observation from our studies is the rapid clearance of the injected ²²¹Fr and ²¹³Bi from blood. Only 6.5% of the injected activity of ²²¹Fr (decay corrected) remains in the blood after 5 min, whereas ²¹³Bi is somewhat slower with 11.1% ID remaining in the blood. The summation of the remaining ²²¹Fr activity of all organs gives a value of 69.3% and showed that 30.7% of the ID has been cleared from the body within 5 min of application. For ²¹³Bi we could show that 51.7% ID are eliminated and 48.3% ID are retained in various organs. With these numbers in mind, it seems likely that a certain amount of ²²¹Fr and ²¹³Bi generated from ²²⁵Ac in the blood is eliminated rapidly from the body.

As previously suggested by studies using ²²⁵Ac as an in vivo generator, our data clearly show that the major organ of accumulation for ²²¹Fr as well as ²¹³Bi were the kidneys. Both radionuclides are cleared most likely via renal elimination, but seemed to be retained to a certain extent. Potentially ²¹³Bi could be retained in the kidney bound to a bismuth-metal binding protein, which also binds stable ²⁰⁹Bi [28]. If this protein is also binding and retaining ²²¹Fr or it is a different mechanism, this remains to be investigated.

Using our ²²⁵Ac/²²¹Fr generator to study the distribution of ²¹³Bi, we could clearly show that ²¹³Bi accumulates in the liver. This is in contrast to other test systems, such as those using polymersomes as model carriers, which naturally accumulate in the liver [29]. A study of chronic ²⁰⁹Bi exposure in rats also showed accumulation of ²⁰⁹Bi in the liver of normal rats [30], suggesting a physiological mechanism responsible for ²¹³Bi accumulation in the liver of mammals.

The distribution of ²²¹Fr in the mouse showed, in addition to the kidney, the highest % ID/g in salivary glands, with elevated levels also in the small intestine and slightly lower levels in the large intestine. ²²¹Fr exhibited 2.5 times higher accumulation in the small and large intestines and more than 3 times higher accumulation in the salivary glands in comparison to ²¹³Bi. Specifically, the accumulation of ²²¹Fr in the salivary glands could be of importance for targeted radionuclide therapy. The currently used ²²⁵Ac-labeled low-molecular-weight PSMA ligands show per se salivary gland accumulation and this toxicity could be enhanced with non-PSMA-targeted ²²¹Fr accumulating in the salivary gland. If this ²²¹Fr accumulation could be translated into humans, it would have implications for any long-circulating ²²⁵Ac-labeled ligands, as this would act as a generator for additional ²²¹Fr.

In our research on both the distribution of ²²¹Fr and ²¹³Bi, we used the opportunity to also investigate the re-distribution of ²¹³Bi, which was generated from ²²¹Fr. Overall, this distribution pattern exhibited similarities that resembled those of ²²¹Fr. ²²¹Fr accumulated in the small intestine, large intestine, and salivary glands, whereas no direct accumulation of ²¹³Bi was observed in these organs. Instead, the presence of ²¹³Bi in these tissues is most likely due to the decay of accumulated ²²¹Fr. This is supported by observations that ²¹³Bi clears from the blood at a rate characteristic of ²²¹Fr. Both radionuclides accumulated in the kidneys, so no conclusion can be drawn from the kidney values.

A major challenge for the future will be the translatability of the in vivo generated mouse data to patients. Data generated with ²²⁵Ac-labeled antibodies in non-human primates have shown accumulation of ²¹³Bi in the kidneys [31]. Also, SPECT imaging undertaken in the course of clinical studies have been able to measure ²¹³Bi and ²²¹Fr in patients treated with ²²⁵Ac-labeled compounds. However, the signal for imaging is fairly low and from an experimental perspective, conducting studies of ²²¹Fr is a significant challenge due to its short half-life.

Conclusion

In this study, we gained valuable insights into the biodistribution patterns of ²²⁵Ac progenies, specifically ²²¹Fr and ²¹³Bi, in various organs and tissues. Our findings suggest that these progenies rapidly distribute from the blood into their preferred organ and may contribute to the increased side-effects observed in TαT. Notably, ²²¹Fr exhibited a strong affinity for kidneys, salivary glands and both small and large intestine, while ²¹³Bi selectively accumulated in the kidneys and liver. These observations are valid for the distribution of free ²²¹Fr and ²¹³Bi from the blood to organs and enhance our understanding of the potential additive side-effects of TαT and underscore the importance of accurate dosimetry in therapies involving complex alpha-emitting in vivo nanogenerators, such as ²²⁵Ac. Using ²²⁵Ac-labeled compounds with distinct pharmacokinetics will, of course, influence the distribution of free ²²¹Fr and ²¹³Bi depending on the organ in which they are generated. Therefore, this re-distribution has to be investigated individually for each compound.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Author contributions

Contributing to conception and design: All authors.

Acquiring data: SZK, YR, IM, MS.

Analyzing and interpreting data: SZK, YR, MS, BM, LM, MBS.

Drafting the manuscript: SZK, HT, BM, LM, MBS.

Critically contributing to or revising the manuscript: All authors.

Approving the final content of the manuscript: All authors.

Funding

Open Access funding enabled and organized by Projekt DEAL. Open Access funding enabled and organized by Projekt DEAL. Ingrid Moen, Frans Suurs, Christoph Schatz, Urs B. Hagemann are employees of Bayer AG or Bayer AS. During the time the study was conducted Sabine Zitzmann-Kolbe was also an employee of Bayer AG. Steffen Happel is an employee of TrisKem International SAS.

Data availability

The datasets generated during an/or analyzed during the current study are available from the corresponding author on reasonable request.

Declarations

Ethics declaration

All animal experiments were conducted in accordance with the Federal Republic of Germany Animal Welfare Law and were approved by the responsible local authority (Berlin State Office for Health and Social Affairs/Landesamt für Gesundheit und Soziales (LaGeSo), Germany).

Competing interests

No other potential conflicts of interest relevant to this article exist.