Alpha-PET for Prostate Cancer: Preclinical investigation using 149Tb-PSMA-617

In this study, it was aimed to investigate 149Tb-PSMA-617 for targeted α-therapy (TAT) using a mouse model of prostate-specific membrane antigen (PSMA)-expressing prostate cancer. 149Tb-PSMA-617 was prepared with >98% radiochemical purity (6 MBq/nmol) for the treatment of mice with PSMA-positive PC-3 PIP tumors. 149Tb-PSMA-617 was applied at 1 × 6 MBq (Day 0) or 2 × 3 MBq (Day 0 & Day 1 or Day 0 & Day 3) and the mice were monitored over time until they had reached a pre-defined endpoint which required euthanasia. The tumor growth was significantly delayed in mice of the treated groups as compared to untreated controls (p < 0.05). TAT was most effective in mice injected with 2 × 3 MBq (Day 0 & 1) resulting in a median lifetime of 36 days, whereas in untreated mice, the median lifetime was only 20 days. Due to the β+-emission of 149Tb, tumor localization was feasible using PET/CT after injection of 149Tb-PSMA-617 (5 MBq). The PET images confirmed the selective accumulation of 149Tb-PSMA-617 in PC-3 PIP tumor xenografts. The unique characteristics of 149Tb for TAT make this radionuclide of particular interest for future clinical translation, thereby, potentially enabling PET-based imaging to monitor the radioligand’s tissue distribution.

Herein, we propose 149 Tb as a potential alternative α-emitter for targeted radioligand therapy, based on several attractive features: (i) 149 Tb decays with a half-life of 4.1 h, which is relatively short as compared to 225 Ac, but more than four-fold longer than the half-life of 213 Bi. This situation makes 149 Tb particularly interesting in combination with small molecules that are characterized by fast accumulation in the tumor lesions and efficient clearance from healthy tissue. (ii) 149 Tb emits low-energy α-particles (E α = 3.97 MeV; I = 17%), but the decay does not involve relevant α-emitting daughter nuclides, which is advantageous over 213 Bi and 225 Ac (Fig. 1b). (iii) The co-emission of β + -particles (positrons) is a unique feature of 149 Tb, making it suitable to trace 149 Tb-labeled radioligands using positron emission tomography (PET). This has recently been exemplified in a preclinical pilot study, in which we demonstrated the feasibility of visualizing 149 Tb using PET and referred to this approach as "alpha-PET 12 ". (iv) 149 Tb, as a radiolanthanide, can be stably coordinated with a DOTA chelator and, hence, be used with any established tumor-targeting agent that is also applied for 177 Lu-therapy. (v) Finally, it is important to recognize that additional, medically-interesting Tb radioisotopes exist, among those 161 Tb, which has similar characteristics to 177 Lu but co-emits conversion and Auger electrons that were shown to potentiate the therapeutic efficacy in a preclinical setting [13][14][15][16] . This situation could enable using chemically-identical radioligands for either β − -/ Auger electron therapy or TAT, respectively.
The potential of 149 Tb was demonstrated for the first time in a preclinical therapy study more than a decade ago 17 . It was shown that 149 Tb-rituximab was able to specifically kill circulating cancer cells and small cell clusters in a leukemia mouse model. The therapeutic efficacy of 149 Tb was also investigated by our own group using a 149 Tb-labeled DOTA-folate conjugate in a therapy study with KB tumor-bearing mice 12 . In this study, 149 Tb was used for the labeling of PSMA-617 and tested in a preclinical setting. 149 Tb-PSMA-617 was investigated in a therapy experiment with tumor-bearing mice using variable application schemes and for the visualization of PSMA-positive tumor xenografts using preclinical PET/CT.

Results
preparation of 149 Tb-PSMA-617. Directly after separation from zinc and isobar impurities, the final product ( 149 Tb in HCl 0.05 M) was used for the labeling of PSMA-617. 149 Tb-PSMA-617 was obtained at a molar activity of up to 6 MBq/nmol, with a radiochemical purity of >98%. The retention time of the product (t R = 8.7 min) was equivalent to previous data obtained with 177 Lu-PSMA-617 ( Supplementary Information, Fig. S1) 18 .
Areas under the curve (AUc) and AUc ratios of 149 Tb-PSMA-617. Based on previous studies that showed equal distribution of 177 Lu-and 161/152 Tb-labeled tumor targeting agents (including DOTA-folate 14 , DOTANOC 19 and PSMA-617 16 ), it was assumed that 149 Tb-PSMA-617 and 177 Lu-PSMA-617 would distribute equally in the body. The distribution of 177 Lu-PSMA-617 showed fast accumulation in PC-3 PIP tumor xenografts, with the kidneys being the only healthy organs with substantial accumulation of activity (Supplementary Information, Table S1) 2 . The biodistribution data obtained with 177 Lu-PSMA-617 were transformed to non-decay-corrected data, using the half-life of 149 Tb, to obtain the time-dependent uptake of 149 Tb-PSMA-617 in the various tissues. This enabled the determination of the areas under the curves (AUCs) and the respective tumor-to-background AUC ratios. Due to the much shorter half-life of 149 Tb as compared to 177 Lu, the activity retention in the tumor xenograft was shorter resulting in low uptake values (0.66 ± 0.10% IA/g) at 24 h p.i. In any normal tissue and organ the retention of activity was <0.1% IA/g at 24 h after injection of 149 Tb-PSMA-617 (Supplementary Information, Table S2). The tumor-to-blood, tumor-to-kidney and tumor-to-liver AUC ratios of 149 Tb-PSMA-617 were determined as 74, 10 and 225, respectively (Table 1). Based on the pharmacokinetic properties of radiolabeled PSMA-617, characterized by high retention of tumor-accumulated activity but fast excretion from background organs, the AUC ratios correlated positively with the half-life of the respective radionuclide. Calculations of AUC ratios for 213 Bi-PSMA-617 -under the assumption that it would distribute the same as 177 Lu-PSMA-617 -revealed clearly lower values than determined for 149 Tb-PSMA-617 (Table 1). Accordingly, calculated AUC ratios of 225 Ac-PSMA-617 resembled more closely those of 177 Lu-PSMA-617. It has to be critically acknowledged, however, that the daughter nuclides and their uncontrollable decay, also potentially in non-targeted tissues, has not been taken into consideration for this estimation.
The calculations for 177 Lu-PSMA-617 were performed in analogy and revealed a mean specific absorbed dose to the tumors and kidneys of 3.2 Gy/MBq and 0.041 Gy/MBq, respectively (Supplementary Information). It should be noted that these dosimetry estimations are based on the assumptions of an average sphere size of 60 mm 3 , while interindividual differences in tumor sizes were not considered. The variation in absorbed doses due to tumor size variations was, however, less than 5% for 149 Tb and 177 Lu, respectively. therapy study. Mice from four groups were injected with either only saline or a cumulative activity of 6 MBq 149 Tb-PSMA-617 using variable injection schemes (Fig. 2a). Group A (control group; injected with saline) showed constant tumor growth and, as a result, the first mouse had to be euthanized at Day 12 due to an oversized tumor. Tumor growth of mice in Group B, which received one injection of 6 MBq 149 Tb-PSMA-617 at Day 0, was clearly reduced compared to the control group. The first three mice of Group B had to be euthanized at Day 22 due to loss of body weight, presumably as a consequence of the tumor burden. The tumor growth inhibition in mice from Group C and D that received 2 × 3 MBq 149 Tb-PSMA-617 at Day 0 and Day 1 or at Day 0 and Day 3, respectively, was comparable between the two groups. The first mouse from Groups C and D, respectively, was euthanized at Day 30 and Day 26, due to a combination of body weight loss and increased tumor volume ( Fig. 2b and Table 2).
Quantification of the therapeutic effect by means of calculating the tumor growth inhibition (TGI) revealed a significantly (p < 0.05) increased value for Groups B, C and D as compared to Group A. The same was found when calculating the tumor growth delay indices 2 and 5 (TGDI 2 and TGDI 5 ), which were significantly (p < 0.05) larger in treated mice (Groups B-D) as compared to the control group (Group A). Among the treated mice, these values were highest for mice from Groups C and D. The lifetime of mice was based on the day of euthanasia which was required according to pre-defined endpoints. When compared to the control mice (median lifetime: 20 days), the treated mice had an increased median lifetime of 26 days (Group B), 36 days (Group C) and 32 days (Group D), respectively ( Fig. 2c and Table 2).
Monitoring of mice during the therapy study. Monitoring of the mice also revealed body weight loss over time in all groups except Group C, which was observed as a consequence of increasing tumor burden (Fig. 2d). The analysis of blood plasma parameters at the time of euthanasia indicated no significant changes in any of the measured parameters between treated mice of Groups B-D and untreated control mice of Group A (Supplementary Information, Table S3). Moreover, the average body weight at the time of euthanasia, as well as the organ mass of kidneys, liver and brain, and the ratios thereof did not reveal any significant differences among the mice of the different groups (Table 3).

Discussion
In the present study, 149 Tb was produced at a quantity and quality that enabled the labeling of PSMA-617 at a specific activity and radiochemical purity suitable for a preclinical therapy study. The experiment was designed with four groups of six mice, namely, one group of untreated control animals and three groups of mice treated with 149 Tb-PSMA-617 according to different application schemes. The treated groups of mice that received   Table 3. Body weight and organ weight of mice in the therapy study and their corresponding ratios. Values indicated as average ± standard deviation (SD). a Data obtained at the day of euthanasia when an endpoint criterion was reached. www.nature.com/scientificreports www.nature.com/scientificreports/ According to the decay properties of the radionuclides in question, 149 Tb-PSMA-617 is likely to be more potent than 213 Bi-PSMA-617 and, therefore, most probably equally effective at lower activities.
The mean absorbed dose of 149 Tb-PSMA-617 to the kidneys was determined to be ~10-fold higher than that of 177 Lu-PSMA-617. In a previously-performed therapy study in mice performed with 177 Lu-folate, a dose level of ~23 Gy to the kidneys was well tolerated 21 . Should this renal dose limit be translatable to α-emitters, one could still apply 6 cycles safely with 149 Tb-PSMA-617 (using 6 MBq per mouse with a cumulative activity of 36 MBq) resulting in accumulative dose of ~23 Sv RBE5 to the kidneys.
Radionephrotoxicity in patients treated with 177 Lu-PSMA-617 has not been observed, due to the low renal uptake and, consequently, low mean aborbed dose to the kidneys (~0.6 Gy/GBq) 22,23 . It is, therefore, likely that the generally-accepted (conservative) threshold dose of ~23 Gy 24,25 would not be reached with 149 Tb-PSMA-617, since the quantity of injected activity would be significantly lower. It can even be expected that the renal dose would still be within the safety margins if it was increased by a factor of 10 (i.e. ~6 Sv RBE5 /GBq), as observed in this preclinical study. Importantly, the calculated absorbed kidney dose reported for 213 Bi-PSMA-617 in patients was determined to be in a similar range (~8 Sv RBE5 /Gy) 26 .
Our calculations of AUC ratios from preclinical data indicated the highest and, thus, most favorable ratios for the longer-lived 225 Ac. These results were in line with literature reports on theoretical dose estimations that considered 225 Ac-PSMA-617 to be superior to 213 Bi-PSMA-617, due to favorable dosimetry, with an increased therapeutic index and less off-target radiation 26 . It is, however, important to recognize that 225 Ac decays by several αand β − -disintegrations, which may add to the off-target dose. The fact that 149 Tb does not have relevant α-emitting daughters adds particular value to this radionuclide. As the tumor-to-background AUC ratios increase with the half-life of the applied radionuclide, 149 Tb would be a clearly more favorable α-emitter than 213 Bi for TAT. The four-fold increased half-life of 149 Tb, as compared to 213 Bi would not only improve the tumor-to-background dose ratios but also facilitate the logistics of radioligand preparation and distribution. These promising circumstances warrant the evaluation of new production sites to make 149 Tb routinely available at larger quantities.
Due to the positron emission of 149 Tb, the accumulation of 149 Tb-PSMA-617 in PSMA-positive prostate tumor xenografts was readily visualized using preclinical PET. This approach was previously demonstrated with 149 Tb-DOTANOC 12 . The unique characteristic of 149 Tb to emit α-particles and positrons (previously referred to as the concept of "alpha-PET") would most likely allow the imaging of 149 Tb-based α-therapy in patients. This would give 149 Tb an advantage over existing α-emitters and provide a new dimension in view of its clinical translation. It would also allow accurate retrospective dose estimations to plan future applications and minimize off-target toxicity.
Potential limitations of this study include the fact that the PC-3 PIP tumor mouse model is based on PCa cells that were transduced to stably express PSMA at levels which are higher than in LNCaP tumor xenografts that express PSMA physiologically 27 . Moreover, tumor xenografts based on PC-3 PIP cells express PSMA homogeneously throughout the xenograft, which may not exactly reflect the situation of lesions in patients. Finally, a human xenograft only grows in immune-deficient (athymic nude) mice, hence, immunological reactions, which may have an impact on the therapy outcome, are not considered in this model.
The mice were treated when the tumor xenografts were still quite small, in order to enable monitoring of the tumor growth (delay) over a reasonable time period as commonly performed in preclinical settings 16,28,29 . This may be seen as a limitation, since tumor lesions in patients may have developed over several weeks. It is, however, important to mention that the patients suffering from metastatic disease with very small lesions would profit most from TAT. It is, thus, vital to show the therapy effect in small tumors since these smallest lesions are commonly the ones, which do not get sufficient dose when using the current generation of β − -emitting radionuclides such as 177 Lu 30 .
A further limitation of any preclinical study refers to the legal requirements of defining endpoints, when mice have to be euthanized, which do not necessarily reflect the situation of a cancer patient. In this study, the endpoints of mice were defined based on the tumor size and body weight loss according to ethical guidelines of the local law of animal protection.

conclusion
The interesting features of 149 Tb for "alpha-PET" make it attractive for in-depth preclinical follow-up investigations. Certainly, higher quantities of activity and/or more frequent injections of 149 Tb-PSMA-617 would be necessary to eradicate the tumors entirely. This was, however, not feasible in this study due to the still limited availability of 149 Tb. Beyond the application of 149 Tb-PSMA-617, 149 Tb could be employed in combination with a large variety of DOTA-functionalized, tumor-targeting ligands used in clinics or currently under development. A potential clinical translation of 149 Tb-based radionuclide therapy may, thus, become a realistic future perspective, provided that a significant scale-up of the current production capabilities can be achieved by establishing effective new production sites.

Materials and Methods
production and chemical separation of 149 tb. 149 Tb was produced by proton-induced spallation in a tantalum target, followed by ionization of the spallation products and online mass separation at the ISOLDE facility (CERN, Geneva, Switzerland), as previously reported 12,13,31 . The foils, containing the 149 isobars, were transported to PSI where the 149 Tb was chemically separated from the zinc, as well as from the isobar and pseudo-isobar impurities using chromatographic methods. The final product was obtained as 149 TbCl 3 in a small volume of 0.05 M HCl, which enabled its application for direct radiolabeling. A detailed description of the separation process will be published elsewhere.

In vivo studies.
In vivo experiments were approved by the local veterinarian department and conducted in accordance with the Swiss law of animal protection. The preclinical studies have been ethically approved by the Cantonal Committee of Animal Experimentation and permitted by the responsible cantonal authorities (license number 75668). Athymic BALB/c nude mice were obtained from Charles River Laboratories (Sulzfeld, Germany) at the age of 5-6 weeks.  28,29,32,[36][37][38][39] . It was previously reported that PC-3 PIP cells express PSMA at significantly higher levels than LNCaP cells 27,29 , hence, the PSMA expression level of PC-3 PIP tumor xenografts does not exactly reflect the expression level of lesions in a patient. therapy study and monitoring of mice. The therapy study was performed with 6 mice per group 7 days after inoculation of PC-3 PIP tumor cells (4 × 10 6 cells, 100 μL Hank's Balanced Salt Solution (HBSS)) on the right shoulder. At this stage, the tumors were still quite small (average ~60 mm 3 ; Table 4) Table 4). The mice were monitored by measuring body weights and the tumor size every other day until the end of the study. Mice were euthanized when a predefined endpoint criterion was reached or when the study was terminated at Day 40. Endpoint criteria were defined as (i) body weight loss of >15%, (ii) a tumor volume of >800 mm 3 , (iii) a combination of body weight loss of >10% and a tumor volume of >700 mm 3 , (iv) www.nature.com/scientificreports www.nature.com/scientificreports/ signs of unease and pain or (v) a combination thereof. The relative body weight (RBW) was defined as [BW x / BW 0 ], where BW x is the body weight (in grams) at a given Day X and BW 0 the body weight (in grams) at Day 0. The tumor dimensions were determined by measuring the longest tumor axis (L) and its perpendicular axis (W) with a digital caliper. The tumor volume (V) was calculated according to the equation [V = 0.5 * (L * W 2 )]. The relative tumor volume (RTV) was defined as [TV x /TV 0 ], where TV x was defined as the tumor volume in mm 3 at a given Day X and TV 0 the tumor volume in mm 3 at Day 0. The anti-tumor efficacy of 149 Tb-PSMA-617 was expressed as percentage tumor growth inhibition (% TGI), using the equation [(100 − (T/C)) × 100], where T is the mean RTV of treated mice and C is the mean RTV of control mice at the time of euthanasia of the first mouse of the control group. As an additional measure of the efficacy of the radionuclide therapy, the tumor growth delay indices were determined. The tumor growth delay (TGD x ) was the time required for the tumor volume to increase x-fold over the initial volume at the Day 0. The tumor growth delay index [TGDI x = TGD x (T)/TGD x (C)] was calculated as the TGD x ratio of treated mice (T) over control mice (C) for a 2-fold (x = 2, TGD 2 ) and 5-fold (x = 5, TGD 5 ) increase of the initial tumor volume. The median lifetime, based on euthanasia of the mice when they reached an endpoint, was calculated using GraphPad Prism software (version 7). After euthanasia, kidneys, liver and the brain were collected and weighed. The organ ratios (kidney-to-brain and liver-to-brain) were calculated using the organ masses obtained at the day of euthanasia. Organ data were analyzed for significance using a one-way ANOVA test with a Tukey's post correction (GraphPad Prism software, version 7). A p-value of <0.05 was considered as statistically significant.
Blood samples were taken at the time of euthanasia for the evaluation of a selection of clinical chemistry parameters of renal and hepatic function (creatinine, blood urea nitrogen, alkaline phosphatase, total bilirubin and albumin) (Supplementary information).
PET/CT imaging studies. In a separate experiment, PET/CT scans were performed using a small-animal bench-top PET/CT scanner 40 (G8, Perkin Elmer, Massachusetts, U.S), as previously reported, with a set energy window ranging from 150 keV to 650 keV 41 . Mice were subcutaneously inoculated with PC-3 PIP tumor cells (6 × 10 6 cells) and PC-3 flu tumors cells (5 × 10 6 cells) on the right and left shoulder, respectively, 7-10 days before the acquisition of the PET/CT scans. This mouse model with a PSMA-positive and a PSMA-negative tumor xenograft in one animal enables the determination of PSMA-specific radioligand uptake without the need for blocking studies using 2-(phosphonomethyl)-pentandioic acid (2-PMPA) as a PSMA inhibitor. During the scan, mice were anesthetized with a mixture of isoflurane and oxygen. Static whole-body PET scans of 10 min duration were performed at 30 min, 2 h and 4 h after injection of 149 Tb-PSMA-617 (5 MBq, 1.2 nmol, 200 µL), followed by a CT scan of 1.5 minutes. The aquistion of the data and their reconstruction was performed using the G8 PET/CT scanner software (version 2.0.0.10). All images were prepared using VivoQuant post-processing software (version 3.5, inviCRO Imaging Services and Software, Boston U.S.). A Gaussian post-reconstruction filter (full width at half maximum = 1 mm) was applied to the images and the scale was adjusted by cutting 5-10% of the lower signal intensity to make the tumors and kidneys readily visible. ethical approval. This study was performed in agreement with the national law and PSI-internal guidelines of radiation safety protection. In vivo experiments were approved by the local veterinarian department and conducted in accordance with the Swiss law of animal protection.