Alpha radioimmunotherapy using 225Ac-proteus-DOTA for solid tumors - safety at curative doses

This is the initial report of an α-based pre-targeted radioimmunotherapy (PRIT) using 225Ac and its theranostic pair, 111In. We call our novel tumor-targeting DOTA-hapten PRIT system “proteus-DOTA” or “Pr.” Herein we report the first results of radiochemistry development, radiopharmacology, and stoichiometry of tumor antigen binding, including the role of specific activity, anti-tumor efficacy, and normal tissue toxicity with the Pr-PRIT approach (as α-DOTA-PRIT). A series of α-DOTA-PRIT therapy studies were performed in three solid human cancer xenograft models of colorectal cancer (GPA33), breast cancer (HER2), and neuroblastoma (GD2), including evaluation of chronic toxicity at ~20 weeks of select survivors. Methods: Preliminary biodistribution experiments in SW1222 tumor-bearing mice revealed that 225Ac could not be efficiently pretargeted with current DOTA-Bn hapten utilized for 177Lu or 90Y, leading to poor tumor uptake in vivo. Therefore, we synthesized Pr consisting of an empty DOTA-chelate for 225Ac, tethered via a short polyethylene glycol linker to a lutetium-complexed DOTA for picomolar anti-DOTA chelate single-chain variable fragment (scFv) binding. Pr was radiolabeled with 225Ac and its imaging surrogate, 111In. In vitro studies verified anti-DOTA scFv recognition of [225Ac]Pr, and in vivo biodistribution and clearance studies were performed to evaluate hapten suitability and in vivo targeting efficiency. Results: Intravenously (i.v.) administered 225Ac- or 111In-radiolabeled Pr in mice showed rapid renal clearance and minimal normal tissue retention. In vivo pretargeting studies show high tumor accumulation of Pr (16.71 ± 5.11 %IA/g or 13.19 ± 3.88 %IA/g at 24 h p.i. for [225Ac]Pr and [111In]Pr, respectively) and relatively low uptake in normal tissues (all average ≤ 1.4 %IA/g at 24 h p.i.). Maximum tolerated dose (MTD) was not reached for either [225Ac]Pr alone or pretargeted [225Ac]Pr at administered activities up to 296 kBq/mouse. Single-cycle treatment consisting of α-DOTA-PRIT with either huA33-C825 bispecific anti-tumor/anti-DOTA-hapten antibody (BsAb), anti-HER2-C825 BsAb, or hu3F8-C825 BsAb for targeting GPA33, HER2, or GD2, respectively, was highly effective. In the GPA33 model, no complete responses (CRs) were observed but prolonged overall survival of treated animals was 42 d for α-DOTA-PRIT vs. 25 d for [225Ac]Pr only (P < 0.0001); for GD2, CRs (7/7, 100%) and histologic cures (4/7, 57%); and for HER2, CRs (7/19, 37%) and histologic cures (10/19, 56%) with no acute or chronic toxicity. Conclusions: [225Ac]Pr and its imaging biomarker [111In]Pr demonstrate optimal radiopharmacologic behavior for theranostic applications of α-DOTA-PRIT. For this initial evaluation of efficacy and toxicity, single-cycle treatment regimens were performed in all three systems. Histologic toxicity was not observed, so MTD was not observed. Prolonged overall survival, CRs, and histologic cures were observed in treated animals. In comparison to RIT with anti-tumor IgG antibodies, [225Ac]Pr has a much improved safety profile. Ultimately, these data will be used to guide clinical development of toxicity and efficacy studies of [225Ac]Pr, with the goal of delivering massive lethal doses of radiation to achieve a high probability of cure without toxicity.

Because curative therapy is a major need in solid human tumors of adults and children, we have focused our efforts on PRIT for common solid tumors, using clinically applicable antibody-based radiotargeting systems [3]. We call this development "DOTA-PRIT." DOTA-PRIT utilizes anti-tumor antigen/anti-1,4,7,10-tetraazacyclododecane-N, N', N'', N'''-tetraacetic acid (DOTA) hapten BsAbs (IgGsingle-chain variable fragment (scFv) format [12]) based on the pioneering PRIT "2D12.5" anti-metal chelate antibody for nanomolar binding to DOTA chelates of all lanthanides [13,14]. We utilize an affinity maturated anti-DOTA antibody scFv "C825" for picomolar binding to DOTA-radiohaptens for improved in vivo radioactive bound-lifetimes of tumor-targeted radioisotopes during PRIT [15]. In this technique, the non-radioactive BsAb formulation is infused and, owing to long circulation times, it takes a couple of days to reach high a tumor-to-background ratio [12]. Once the clearing agent (CA) [16] has reduced unbound circulating anti-tumor antibody, the radiohapten in the form of radiometals carried by a DOTA-hapten [2] (molecular weight (MW) ~1 kDa) is infused and rapidly taken up by the pretargeted scFv on the BsAb. Any non-tumor-bound radiometal-DOTA complex is efficiently and rapidly cleared through the kidney, resulting in exceptionally high TI.
We have successfully applied DOTA-PRIT to a variety of solid tumor-associated antigens for high-TI targeting of β-emitting isotopes, including GD2 [17], GPA33 [18,19], and HER2 [20] (β-DOTA-PRIT). In our hands, β-DOTA-PRIT now reliably leads to cures in animal tumors with minimal radiotoxicity. Also, for hematological cancers, β-DOTA-PRIT with 90 Y has also been shown to be curative and safe in preclinical models [21,22]. One important limitation of DOTA-PRIT is that the methodology is confined to the β-emitting radiolanthanides, 177 Lu, and 90 Y, due to the hapten-binding scFv C825 antibody specificity [15]. In the present manuscript, we report a novel set of radiohaptens that permits the use of other forms of therapeutic radionuclides -in particular, α-emitters for histologic cure. 225 Ac is a long-lived (t 1/2 = 9.92 d) α-particle emitter that has been extensively investigated as an 225 Ac-radiolabelled anti-tumor antigen IgG ( 225 Ac-IgG) drug platform for α-radioimmunotherapy (α-RIT; [3,[23][24][25][26][27]). However, for solid tumors, with the long serum half-life of IgG in blood and tissues, α-RIT with 225 Ac-IgG has a limited TI, a common characteristic of large carriers with long biologic half-lives, such as IgG. In fact, 225 Ac-IgG has been shown to be quite radiotoxic to kidneys of mice [28]. To resolve this problem, we reasoned that a radiohapten-with its short serum half-life and complete renal excretion, without retention-would reduce in vivo radiation exposure and be the ideal carrier for the 225 Ac as part of DOTA-PRIT (α-DOTA-PRIT) [2].
We evaluated the hypothesis that a novel radiohapten precursor that we call "proteus-DOTA" (Pr) is a suitable chelator for the α-emitter 225 Ac, with pharmacodynamics that mimic our "gold standard" β-radiohapten [ 177 Lu]LuDOTA-Bn. We also determined if Pr was suitable with 111 In, a radioisotope that has been used as an imaging biomarker for 225 Ac-IgG [29]. We tested both the activity and toxicity of α-DOTA-PRIT in mouse xenografts of human colorectal cancer (GPA33expressing SW1222), breast cancer (HER2-expressing BT-474), and neuroblastoma (GD2-expressing IMR-32) as models for α-therapy of human solid tumors.

Methods
Please see Supplemental Information (SI) for additional details regarding: animal models, Pr synthesis, cell culture, in vitro binding assays with [ 225 Ac]Pr and BsAb, in vivo studies [ 225 Ac]Pr-BsAb complex, single-photon emission computed tomography/computed tomography (SPECT/CT) image analysis, macroscopic post-mortem examination and tissue sample collection, and histopathology (microscopic evaluation).

Animal care and models
Female athymic nude mice (strain: Hsd:Athymic Nude-Foxn 1nu , Envigo, aged 6-8 weeks, average weight for six-week-old and eight-week-old animals: 18.6 and 21.0 g, respectively) were used for all experiments. The subcutaneous BT-474 tumor model [20], SW1222 tumor model [18,19], and IMR32 or luciferase gene reporter transfected IMR32 (IMR-32/ luc) tumor model [17] (~50-900 mm 3 by caliper measurement, assuming ellipsoid geometry for calculation of tumor volume) was used for targeting of HER2, GPA33, or GD2 antigens, respectively, with minor modifications (described in SI). All animal experiments were done in accordance with protocols approved by the Institutional Animal Care and Use Committee of Memorial Sloan Kettering Cancer Center following National Institutes of Health guidelines for animal welfare.

Biodistribution experiments
For biodistribution assay following radiotracer injection, mice were euthanized by CO 2 (g) asphyxiation and tumor and selected organs were harvested, rinsed with water and allowed to air dry, weighed, and radioassayed by gamma scintillation counting (Perkin Elmer Wallac Wizard 3"). Count rates were background-and decay-corrected, converted to activities using a system calibration factor specific for the isotope, normalized to the administered activity, and expressed as percent injected activity per gram of tissue (%IA/g). For 225 Ac, each sample was counted for up to 10 min (24 h after collection when secular equilibrium was reached) using a 150 to 600 keV energy window.  [17] in comparison to HER2 (TI = 28) [20] and GPA33 (TI = 73) [19]. Ten IMR-32 tumor-bearing mice were injected with BsAb, dextran-CA (100 µg, 0.2 nmol dextran, 29 nmol of (Y)DOTA), and then an equimolar amount of either [ 225 Ac]DOTA-Bn [31] or, for comparison, [ 177 Lu]LuDOTA-Bn [17] (1.85 MBq and 3.7 MBq of 177 Lu and 225 Ac, respectively, 8-10 pmol; n = 5/radiohapten). Mice were sacrificed at 24 h postinjection (p.i.) of radiotracer for biodistribution assay. Note: While these different administered radioactivity doses potentially result in different anti-tumor efficacies, within 24 h post-injection we did not anticipate any effects on tumor uptake and retention of the respective radiohaptens at equimolar doses in the established tumors.

Pr radiochemistry and assay of C825 binding of [ 225 Ac]Pr
The 225 Ac used in this research was supplied by the United States Department of Energy Office of Science by the Isotope Program in the Office of Nuclear Physics. Carrier-free 225 Ac (2.146 E6 GBq/g [5.80 E4 Ci/g]) was obtained from Oak Ridge National Laboratory as a dried nitrate residue. The [ 225 Ac]AcNO 3 was dissolved in 0.2 M Optima grade hydrochloric acid for subsequent radiochemistry. 225 Ac-activity measurements were made at secular equilibrium using a CRC-15R radioisotope calibrator (Capintec, Inc.) set at 775 and multiplied the displayed activity value by 5; samples were positioned at the bottom and center of the well for measurement. No-carrier-added [ 111 In]InCl 3 sterile solution was obtained from Nuclear Diagnostic Products, Inc. 111 In-radioactivity measurements were made using a CRC-15R radioisotope calibrator (Capintec, Inc.) with the manufacturer's recommended settings for the isotope. Water and buffers were rendered metal-free and sterile by passing them through a column of Chelex-100 resin, followed by filtration through a sterile-filter device (0.22-or 0.45-µM). Initially, Pr was suspended in chelexed water at 10 mg/mL and immediately transferred to a 1.8-mL Nunc vial, and any unused stock was promptly stored at -20°C.
A typical synthesis of [ 225 Ac]Pr is described, with a summary of [ 225 Ac]Pr preparations provided in Table S1. To prepare [ 225 Ac]Pr, 20 µL of [ 225 Ac]AcNO 3 (2.442 MBq) in 0.2M HCl was mixed with 100 µL of 10 mg/mL Pr (1 mg; 0.74 µmoles) in a 1.8-mL Nunc vial. Next, 15 µL of L-ascorbic acid solution (150 g/L) and 100 µL of 3M NH 4 OAc solution was added. The pH of the solution was verified to be ~5.5 by spotting 1 µL of the reaction mixture onto pHydrion pH paper (range, 5.0-9.0). The reaction was incubated at 60°C for 30 min, and then purified using a Sephadex C-25 column pre-equilibrated with 6 mL of NSS. The reaction mixture was added to the column and eluted with 4 mL of NSS to recover all removable activity ([ 225 Ac]Pr; note: the % activity that washed off the resin was the % 225 Ac that was complexed by Pr  Figure S1).
In vitro binding assays of [ 225 Ac]Pr were performed with anti-HER2 BsAb to evaluate C825 binding of [ 225 Ac]Pr, followed by in vivo evaluation in BT-474 tumor-bearing mice to assay HER2 binding of the [ 225 Ac]Pr-BsAb complex. We used the HER2 model because of the three antibody-antigen systems currently available to us for DOTA-PRIT, radiolabeled forms of trastuzumab have been studied most extensively in preclinical and clinical studies [3,24,32,33].   [18]. In addition, serial whole-body activity measurements were collected immediately following imaging and a biodistribution assay was performed at 24 h p.i. Whole-body activity and blood activity clearance parameters were determined by fitting the data to exponential model curves using MATLAB (Mathworks, Inc.).

DOTA-PRIT with [ 225 Ac]Pr or [ 111 In]Pr
For initial pretargeting of [ 225 Ac]Pr or [ 111 In]Pr, the GPA33 system was used as a model. We used the GPA33 model for preliminary testing of DOTA-PRIT with radiolabeled forms of Pr since we have conducted the most in vivo experiments with this antibody-antigen system and tumor model in comparison with GD2/IMR32 and HER2/BT-474, and thus consider it to be a reliable test system for the development of novel radiohaptens. Lu]LuDOTA-Bn) dose spanning two orders of magnitude (~170-26900 pmol) in order to determine the upper limit of tumor uptake. From the biodistribution data, a plot of tumor uptake at 24 h p.i. versus administered pmol of hapten was prepared, and nonlinear regression analysis (one site, specific binding) was performed using Prism 7 (GraphPad, Inc.) to estimate the Bmax.

Toxicity and therapy studies
An initial dose escalation toxicity study was performed with varying dose levels of [ 225 Ac]Pr (0, 9.25, 18.5, 37, 74, 148, or 296 kBq/mouse; 910 pmol-29.1 nmol) in groups (n = 5) of tumor-free mice to determine maximum tolerated dose (MTD). This dose range was based on a previous MTD study in BALB/c mice with the PRIT reagent [ 225 Ac]DOTA-Biotin, where lethal toxicity was observed at 110 d after treatment with 740 kBq/mouse [34]. Treated mice were monitored daily and weighed up to twice weekly for evidence of treatment-induced toxicity for 145 d (~20 w) p.i. All survivors were evaluated for radiation-induced histologic organ damage by board-certified veterinary pathologists.

Statistical analysis
Differences in tissue uptake between cohorts were statistically analyzed with the student t test for unpaired data using Excel. Two-sided significance levels were calculated and a P value of <0.05 was considered statistically significant. Survival analysis was performed using Prism 7 (GraphPad, Inc.). Kaplan-Meier survival curves were analyzed with Mantel-Cox test. Data is presented as mean and standard deviation (SD).

Preliminary DOTA-PRIT with [ 225 Ac]DOTA-Bn, Pr synthesis and radiochemistry, and verification of C825 recognition of Pr
As shown in Figure 1A (also see  10.30 ± 6.42 %IA/g; P = 0.0054), prompting the development of Pr. C825 has been shown to have metal specificity [15]; e.g., with picomolar binding for DOTA complexes of Lutetium and Yttrium and nanomolar affinity for Indium and Gallium chelates, leading to significant differences in tumor uptake of corresponding radiohapten during in vivo pretargeting [16]. Therefore, we concluded that the C825 binding of [ 225 Ac]DOTA-Bn was unsuitable for efficient in vivo α-DOTA-PRIT ( Figure 1B).
The chemical synthesis route of Pr is described in Figure 1C. Pr was prepared in very high purity (>98%) and with an overall yield of 34% (please see SI for LC/MS and NMR data). Furthermore, Pr was successfully radiolabeled with 225 Ac or 111 In, suggesting that the 175 Lu-DOTA-benzene moiety of Pr does not interfere with efficient 225 Ac-or 111 In-radiometal complexation by the DO3A.
In vitro binding experiments with tracer [ 225 Ac]Pr and excess BsAb confirmed efficient and stable scFv C825-mediated binding of [ 225 Ac]Pr ( Figure S2). Efficient tumor targeting in vivo was demonstrated using pre-formed BsAb complexed with [ 225 Ac]Pr, suggesting that radiolabeled forms of Pr did not interfere with the BsAb anti-tumor antibody domain-tumor antigen interaction ( Figure S3).
As revealed in the SPECT/CT images ( Figure  2C), in the absence of specific tumor binding of the [ 111 In]Pr to pretargeted BsAb, Pr traffics quickly and almost exclusively out of the body by renal clearance. The whole-body activity assays demonstrate rapid excretion of activity with ~90-94% removal by 6 h p.i. of tracer (whole-body clearance half-lives for mouse 1 and mouse 2: 2.71 h (R 2 = 0.912) and 2.03 h (R 2 = 0.992), respectively; Figure S4A) [2]. Image-derived volume-of-interest (VOI) analysis of tumor and select normal tissues also confirmed renal clearance and minimal normal tissue uptake and retention (Figures S4B and S4C).

Therapy with α-DOTA-PRIT
Single-cycle therapy was performed in the three tumor xenograft models for initial testing of efficacy and safety. There were no significant differences in  Figure 4C and 4D) Figure 4E). We observed BLI-positive (tumor volumes: 74 mm 3 and 474 mm 3 ) in 2/7 animals within 90 d of treatment ( Figure S5). Of these two BLI-positive mice, one required sacrifice due to tumor burden at 117 d after treatment, and the other was sacrificed for further analysis after the tumor failed to grow above 100 mm 3 within 210 d of treatment. We have observed this rare apparent spontaneous tumor regression previously during treatment studies of SW1222 with β-DOTA-PRIT ( 177 Lu) [19], and we are currently investigating the possibility of a senescent phenotype. A third mouse developed a recurrence and had to be sacrificed 133 d after treatment due to tumor burden. The remaining four mice were confirmed to be histologic cures 141-241 d after treatment. In summary, of 4/7 assessable mice: 4/4 were cured. Survival of mice treated with GD2 α-DOTA-PRIT compared with controls was significantly extended ( Figure 4F); the MS (in d after treatment start) was significantly different (i.e., >125 d) for GD2 α-DOTA-PRIT, compared to 18.5 d post-treatment for [ 225 Ac]Pr-only treatment controls (P = 0.0002).

Toxicity of [ 225 Ac]Pr and α-DOTA-PRIT
Prior studies with long circulating 225 Ac-IgG provide a valuable benchmark for toxicity comparisons to the target organ (kidney) for 225 Ac toxicity after parenteral injection in mice [28]. For example, radiation nephropathy was observed in 225 Ac-IgG (720 kBq/kg (12.95 kBq))-treated mice examined at 20 w (from [28], summarized in Table 1). In contrast, in our study, no acute toxicity (as weight loss >10% of baseline; Figure 5A), nor any radiation-induced histologic organ damage was observed at any [ 225 Ac]Pr dose level at necropsy performed at 145 d ( Table 1 and Table S6). Additional details are provided in SI. Treatment studies with either: GPA33 α-DOTA-PRIT, GD2 α-DOTA-PRIT, or [ 225 Ac]Pr only was well tolerated, with no unscheduled mortalities. Treatment with HER2 α-DOTA-PRIT was also well tolerated, but three unscheduled mortalities were noted (3/19, 16%; two mice at 97 d and a single mouse at 140 d after treatment), which may be related to estrogen supplementation in the BT-474 mouse model, which was observed during HER2 β-DOTA-PRIT [20]). No weight losses were recorded in any of the treatment groups (SW1222: Figure S7; BT-474: Figure 5B; IMR-32/luc: Figure S8). Complete blood count (CBC) measurements in the mice treated with GPA33 α-DOTA-PRIT ( Figure S9) or HER2 α-DOTA-PRIT ( Figure 5C) revealed no significant changes during the observation period (up to 36 d post-treatment), with the exception of WBC for HER2 α-DOTA-PRIT, which showed an average ~55% difference at 21 d post-treatment (3.82 K/uL versus 2.21 K/uL for -2 d and 21 d post-treatment, respectively; P = 0.00813), with rebound to average 2.97 K/uL at 36 d (~25% difference from -2 d post-treatment, P = 0.04488).
In groups treated with HER2 α-DOTA-PRIT (296 kBq), no changes in gross organ weights, apart from a moderate decrease in kidney weights ( Figure S6B), nor radiation-induced histologic organ damage was observed at necropsy performed at 150 d ( Figure 6 and Table S11). There was evidence of minimal to mild renal tubular injury and such changes may be observed spontaneously in aging mice, so we cannot definitively attribute this finding to treatment, although the group difference suggests a possible association with treatment. Remarkable tubuleinterstitial features included minimal to mild multifocal cortical tubular degeneration and atrophy in 5/6 (83%) treated animals. Tubulointerstitial features were all <5% ( Table 1). The CBC showed no significant group differences or deviation from reference ranges (Tables S12-14). A single significant effect on serum chemistry (Table S15) that could be attributed to treatment consisted of evidence of mild elevation of serum blood urea nitrogen (BUN) and creatinine (CREA) in the mice showing minimal to mild renal tubular injury. BUN was occasionally slightly higher than the reference range while the CREA was always within the reference ranges despite the slight increase compared to untreated mice.
Histopathology analysis of the survivors from the GD2 α-DOTA-PRIT (37 kBq) study at 141-241 d revealed no treatment-related findings (Table S16). The hematology and clinical chemistry values were also within normal reference limits (Tables S17-S19). The elevated aspartate transaminase in a single treated mouse had no histopathologic correlate and was considered incidental (Mouse #4; Table S19).

Discussion
α-radiation is potentially one of the most highly focused and radiotoxic anti-cancer agents known, wherein a few radioactive atoms can kill a cancer cell. α-particles have much greater compared to β-radiation, due to their high energy deposition (α-particles: 5-8 MeV; β-particles: 0.1-1 MeV) and far shorter penetration depth (α-particles: 50-80 µm, corresponding to 2-4 typical cell diameters; βparticles: 2-12 mm) [35]. Thus, α-emitters offer high-LET radiation with multiple lethal radiation events, potentially leading to a significant number of double DNA strand breaks. In addition, high selectivity is achieved once they are cancer cell-bound, which results in minimal collateral damage to normal tissue [35].  α-RIT with whole IgG combines highly cytotoxic α radiation with efficient tumor targeting. However, the combination of slow tumor uptake and wholebody clearance creates off-target radiotoxicities. 225 Ac decay results in 4 α-particles, each of which are associated with daughter radionuclides, which can be redistributed throughout the body via the circulation. Radiation nephropathy was reported in mice at administered activities of [ 225 Ac]-IgG, ranging from 12.95-14.80 kBq/mouse [28,36,37]; 37 kBq/mouse was 100% lethal [38]. As an alternative to IgG with long biologic half-lives, small-size carriers with short serum half-lives such as peptides should theoretically shorten radiation exposure, but the renal uptake of such peptides can lead to significant whole-body retention of 10-30% injected dose [2], accompanied by minor (e.g., negligible protein cases observed histologically) to major kidney damage (e.g, pathologic changes consistent with radiation-induced tubular necrosis observed) based on the administered 225 Ac-activity [39][40][41]. Lesions are labeled in red on the low magnification image. Histopathology of the kidneys revealed multifocal minimal to mild chronic tubular injury in treated mice, while the kidneys of the untreated animal was histologically normal. While these lesions were interpreted as probably caused by the treatment, they were not considered adverse, as they did not appear to significantly affect renal function or general health of the animals, based on the low percentage of the renal parenchyma affected (< 5%), normal serum CREA concentration and minimal elevation of BUN, and absence of clinical signs or change in body weight. Scale bars = 1000 µm (low magnification) and 50 µm (high magnification).
Previous manifestations of PRIT were investigated clinically with objective tumor response including CRs (e.g., strept(avidin)-biotin PRIT with Yttrium-90 for treatment of non-Hodgkin's lymphoma [42]), and proof of principle for targeting human tumors in patients was achieved. Actually, for a single dose, very good targeting was achieved to tumors at doses of antibody and radioactivity, which were in a practical range. Nonetheless, using these reagents, full success including cures was essentially out of reach due to a combination of the immunogenicity of strept(avidin), the complexity of dosing schedules, and TI, including renal dose due to retention of strept(avidin). It was clear at that time that achieving radiation doses necessary to effectively treat solid tumors would be associated with intolerable toxicity, especially to kidneys.
Fundamental improvement in reagents for PRIT, such as ultra-high affinity multivalent BsAb-based approaches to PRIT, including DOTA-PRIT [12,15], generated new enthusiasm for translation due to the potential for reduced immunogenicity via use of humanized BsAb and absence of endogenous biotin interference, as well as superior TI and efficacy to prior versions of PRIT [21,43]. α-DOTA-PRIT offers a potential solution: an anti-tumor antibody-based drug platform that harnesses radiohapten capture and glomerular filtration to significantly enhance therapeutic window and safety, possibly allowing dose escalation to cure for human tumors. Thus, we reasoned that this Pr chelator cage, with a MW of ~1.5 kDa (as [ 225 Ac]Pr or [ 111 In]Pr for pretargeting 225 Ac or 111 In, respectively), would have suitable uptake in tumor and clearances ideal for high-TI DOTA-PRIT. Furthermore, this Pr approach should, in principle, allow for highly efficient and specific α-particle RIT, irrespective of the tumor antigen or tumor type, a true "plug-and-play" platform.
In  [19]. These tissue uptake ratios translate into high TI and provide a basis for potentially curative regimens for both α-and β-emitters.
The purpose of this study was to introduce a new DOTA-PRIT hapten for α-PRIT that allows us to obtain significant tumor responses, including CRs and cures, without toxicity. Furthermore, while we establish therapeutic efficacy without observing adverse effects, these initial studies provide benchmarks for further optimization of α-DOTA-PRIT, including: (1) improved tracer dosing with higher specific or molar activity preparations of [ 225 Ac]Pr to balance tumor saturation with normal tissue uptake; (2) dosimetry studies to establish doseresponse correlates and account for the complexity of the 225 Ac-decay chain (i.e., in vivo fate of the recoiled daughter isotopes on α-emission), as well as the intra-cellular residence of [ 225 Ac]Pr; (3) utilize protein engineering innovations to simplify dosing; and (4) utilize fractionated treatment regimens to potentially improve therapeutic outcome. We consider these important experiments to be outside the scope of the current manuscript.
Using DOTA-PRIT for tumor targeting of [ 225 Ac]Pr in models of human colon cancer, breast cancer, or neuroblastoma, tested activities ranging from 37-296 kBq of 225 Ac/mouse were well tolerated, while tumors were ablated. α-DOTA-PRIT therapy was not limited by acute or chronic radiotoxicity to normal organs. We demonstrated that GPA33 α-DOTA-PRIT was effective in SW1222 tumor-bearing mice, observing marginal but significant therapeutic effects (tumor growth control with prolonged survival; P<0.0001; Mantel-Cox test), with no detectable hematopoietic toxicity. Tumor response was likely limited by a combination of factors including the non-internalizing huA33 BsAb-GPA33 antibody-antigen complex [44] and low [ 225 Ac]Pr specific activity, significantly impacting tumor localization of decay daughters and absolute tumor uptake of 225 Ac, respectively. Furthermore, treatment of solid tumors with α-particle RIT, including α-DOTA-PRIT, may be theoretically limited by the potential mismatch between limited antibody penetration in tumor [45] and the short penetration depth of α-particles, leading to intratumoral absorbed dose heterogeneity.
Although not a requirement for efficacy, the application of internalizing mAb for improved α-RIT potency has been demonstrated [46], as there is a limited probability of α-particles originating on the cell surface to traverse a cell due to the high recoil energy of α-particle emission and daughter product release [47]. We show that using an internalizing HER2 BsAb for treatment of BT-474 tumor-bearing mice, we can achieve 7/19 (37%) CRs, including 10/19 (53%) with verified histologic cure at ~150 d post-treatment. This result was somewhat surprising, considering that we anticipated that the low molar activity of [ 225 Ac]Pr would potentially impair responses as we observed for GPA33 α-DOTA-PRIT. We cautiously attribute this high level of efficacy to a combination of factors, including increased probability of exposure of tumor to α-particles due to increased internalization [48], as well as possibly higher radiosensitivity compared with SW1222 based on our previous β-DOTA-PRIT studies. We establish benchmarks for kidney toxicity, as remarkable tubulointerstitial features included minimal to mild multifocal cortical tubular degeneration and atrophy in 5/6 (83%) HER2 α-DOTA-PRIT (296 kBq)-treated animals. These lesions were interpreted as probably caused by the treatment, but they were not considered adverse, as they did not appear to significantly affect renal function or general health of the animals, based on the low percentage of the renal parenchyma affected (< 5%), normal serum CREA concentration and minimal elevation of BUN, and absence of clinical signs or change in body weight.
In addition to therapy applications, the ability to readily image tumor with high contrast to normal tissue is significant for staging as well as dosimetry. The development of theranostic pairs for patientspecific dosimetry has been described as imperative for precision molecular radiotherapy (e.g., radiotherapy of prostate-specific membrane antigen using α-and β-emitters [50]). As a theranostic pair for 90 Y or 177 Lu β-DOTA-PRIT, 86 Y can also be used for positron emission tomography (PET) [19]. As a theranostic pair for [ 225 Ac]Pr, 111 In is more convenient due to its considerably lower cost and availability (compared to 86 Y), as well as its acceptance in routine nuclear medicine SPECT imaging, which is still more widely available than PET. In addition, 111 In has a half-life of 2.8 d, a better match to the half-life of 225 Ac (10 d) in comparison to 86 Y (14 h), for the purposes of dosimetry estimates. The use of 111 In as a surrogate for 225 Ac was previously described in an antibodyoligonucleotide/radiolabeled oligonucleotide PRIT system by Mulvey, et al. [51]. We demonstrate that [ 111 In]Pr is a suitable SPECT imaging surrogate for [ 225 Ac]Pr for dosimetry and treatment monitoring, with comparably high TI and low uptake in the key normal tissues, kidney, liver, and bone. Thus, as a theranostic radiohapten, [ 111 In]Pr will help establish optimized α-therapy regimens centered on quantitative dosimetry, emphasizing cures with minimal toxicity in target organs.

Conclusion
In summary, we have described a novel proteus-DOTA that can efficiently chelate 225 Ac for use as a small radioligand for DOTA-PRIT. In addition, we have developed [ 111 In]Pr as a SPECTimaging surrogate for [ 225 Ac]Pr for dosimetry and treatment monitoring. Finally, we have established therapeutic activity of α-DOTA-PRIT in models of human colon cancer, breast cancer, or neuroblastoma. Furthermore, these anti-tumor effects were comparable to those achieved with β-DOTA-PRIT, although with much lower levels of administered [ 177 Lu]LuDOTA-Bn radioactivity (for GPA33: 55 MBq/mouse [19]; for HER2: 167 MBq/mouse [20]; for GD2: 33 MBq/mouse [17]), consistent with recently reported preclinical α-therapy studies [41] including α-PRIT [52,53]. With pharmacodynamic control of radiotoxic non-tumor-bound [ 225 Ac]Pr, DOTA-PRIT offers a meaningful improvement in the selectivity of α-therapy. support from NIH grant R01 CA233896 (PI: S.M.C). Technical services provided by the MSK Small-Animal Imaging Core Facility and Laboratory of Comparative Pathology were supported by Cancer Center Support Grant P30 CA008748. A Shared Resources Grant from the MSKCC Metastasis Research Center (to P.B.Z), which provided funding for the purchase of the NanoSPECT/CT, is gratefully acknowledged. Dr. Ouerfelli is supported by NCI grant R50 CA243895-01.