PARP-1–Targeted Radiotherapy in Mouse Models of Glioblastoma

General

Unless specified otherwise, all reagents were purchased from Sigma-Aldrich and used as received. ¹³¹I-Na in 0.1N NaOH with a specific activity more than 4,600 TBq/g was purchased from Nordion. 4-(4-fluoro-3-(4-(3-iodobenzoyl)piperazine-1-carbonyl)benzyl)phthalazin-1(2H)-one was synthesized as described previously (14). Olaparib (AZD-2281) was purchased from LC Laboratories. PARPi-FL was synthesized as previously described (16,17). Water (>18.2 MΩcm⁻¹ at 25°C) was obtained from an Alpha-Q Ultrapure water system (Millipore) and acetonitrile (AcN) as well as ethanol (EtOH) were of high-performance liquid chromatography (HPLC) grade purity. Sterile 0.9% saline solution (Hospira) was used for all in vivo injections. HPLC purification and analysis were performed on a Shimadzu UFLC HPLC system equipped with a DGU-20A degasser, a SPD-M20A UV detector, a LC-20AB pump system, and a CBM-20A Communication BUS module. A LabLogic Scan-RAM radio–thin-layer chromatography/HPLC detector was used to detect activity. HPLC solvents (buffer A: water, buffer B: AcN) were filtered before use. HPLC analysis and purification were performed on a reversed-phase C18 Waters Atlantis T3 column (C18-RP, 5 μm, 6 mm, 250 mm). Purification of the iodinated benzoic acid was performed with method 1 (flow rate: 1 mL/min; gradient: 20 min 5%-95% B; 25 min 100% B; 26 min 100%-5% B); quality control analysis was performed with method 1. Purification of the final product was performed on a C6 Waters Spherisorb column (C6, 5 μm, 4.6 mm × 250 mm) with method 2 (flow rate: 1.5 mL/min; isocratic: 0–30 min 35% B). SPECT imaging was performed on a NanoSPECT/CT from Mediso Medical Imaging Systems. PET imaging experiments were conducted on a Focus 120 MicroPET (Concorde Microsystems). Digital phosphor autoradiography was obtained using a Typhoon FLA 7000 laser scanner from GE Healthcare. γ-counting and biodistributions were performed using a WIZARD² automatic γ-counter (PerkinElmer).

Synthesis of ¹²⁷I-PARPi

4-(4-fluoro-3-(piperazine-1-carbonyl)benzyl)phthalazin-1(2H)-one (20 mg, 54.5 μmol) was dissolved in 1 mL of AcN and added to 10 mg (41.3 μmol) of 3-iodobenzoic acid, followed by 25 mg (66 μmol) of N,N,N′,N′-tetramethyl-O-(1H-benzotriazol-1-yl)uronium hexafluorophosphate (HBTU) and 20 μL (145.5 μmol) of Et₃N. The reaction mixture was stirred for 30 min and purified by HPLC (method 2) to yield the title compound (13.4 mg, 22.3 μmol, 67%). ¹H-NMR (CDCl₃) δ  = 10.48 (s, 1H), 8.40–8.39 (m, 1H), 7.74–7.66 (m, 5H), 7.27–7.26 (d, 2H), 7.09–7.07 (d, 2H), 4.22 (s, 2H), 3.73–3.14 (m, 8H). LC-ESI-MS (+) m/z = 597.1 [M+H⁺]⁺. HRMS-ESI [M-H⁺]⁻m/z calculated for [C₂₇H₂₂FIN₄O₃]⁻595.0642, found 595.0640.

Synthesis of ¹⁸F-PARPi

¹⁸F-PARPi was synthesized as described by Carney et al. (18). Briefly, a QMA cartridge containing cyclotron-produced ¹⁸F-fluoride (n.c.a.) was eluted with a solution containing 9 mg of Kryptofix [2.2.2] (4,7,13,16,21,24-hexaoxa-1,10-diazabicyclo[8.8.8]hexacosane), 0.08 mL of 0.15 M K₂CO₃, and 1.92 mL of AcN into a 5-mL reaction vial. Then, ethyl 4-nitrobenzoate (500 μg in 100 μL of dimthyl sulfoxide [DMSO]) was added to the azeotropically dried ¹⁸F⁻ and heated to 150°C (15 min). After this time, 50 μL of 1M NaOH were added followed by HCl (1M, 50 μL) 1 min later. Then, 2 mg of 4-(4-fluoro-3-(piperazine-1-carbonyl)benzyl)phthalazin-1(2H)-one (in 100 μL of DMSO) were added followed by 10 mg of HBTU in 100 μL of DMSO. Finally, 20 μL of Et₃N were added and the reaction mixture was further diluted with 400 μL of AcN and 1 mL of H₂O. The crude mixture was purified by HPLC. The decay-corrected radiochemical yield for the final step was 38.4 ± 2.5% with a molar activity (MA) of 35.9 ± 15.2 GB/μmol. The purified final compound was formulated with 10% EtOH/0.9% NaCl 0.9%.

Synthesis of ¹³¹I-PARPi

¹³¹I-PARPi was obtained in a manner similar to synthetic procedures reported before (14). First, ¹³¹I-NHS-benzoate was obtained by adding N-succinimidyl-4-(tributylstannyl) benzoate (30 μg, 5.9 μmol) in 30 μL of AcN to a solution containing methanol (40 μL), chloramine T (6 μg, 30 nmol), acetic acid (2 μL), and ¹³¹I-NaI in NaOH 0.1 M (185–370 MBq [5–10 mCi]). After the reaction solution was driven for 15 min at room temperature, the reaction was purified by HLPC (method 1), and ¹³¹I-NHS-benzoate collected at 15.1 min. The collected purified fraction was concentrated to dryness in vacuum, reconstituted in a solution of 80 μL of AcN, and added to a solution of 4-(4-fluoro-3-(piperazine-1-carbonyl)benzyl)phthalazin-1(2H)-one (0.3 mg, 0.9 μmol) in 20 μL of DMSO and HBTU (0.3 mg, 0.8 μmol) in 20 μL of DMSO; 10 μL of 2,6-lutidine were further added and the reaction mixture stirred in an Eppendorf ThermoMixer overnight at 65°C (500 rpm). The following day, the reaction mixture was injected and purified by HPLC (method 2) and the final ¹³¹I-PARPi was collected at room temperature for 13.1 min (radiochemical yield > 70%; radiochemical purity > 99%) and concentrated to dryness under vacuum. ¹³¹I-PARPi was formulated with 30% PEG300/70% saline (0.9% NaCl) for both in vitro and in vivo assays (MA = 1.5 ± 0.3 GBq/μmol). Coelution with nonradioactive ¹²⁷I-PARPi reference compound confirmed the identity of the radiotherapeutic.

PARP Inhibitor Binding Assays

Assay was performed by BPS Bioscience. DMSO solutions (100 μM) of each inhibitor were provided at a 100 nM, 50 nM, or 10 nM concentration. Measurements for each enzyme combined with each inhibitor were performed in triplicate in accordance with the BPS assay kit protocol (19). Luminescence was measured using a BioTek Synergy 2 microplate reader. Data were reported by BPS as percentage enzyme activity. Triplicates were combined to get a mean ± SD, which was then grouped and plotted onto a heat map using GraphPad Prism 7 (GraphPad Software).

Cell Culture

The human glioblastoma cell line U251 MG was kindly provided by the Laboratory of Dr. Ronald Blasberg (Memorial Sloan Kettering Cancer Center [MSKCC]). Cell lines were grown in Eagle’s minimal essential medium (MEM), 10% (vol/vol) heat-inactivated fetal bovine serum, penicillin 100 IU₂, and streptomycin (100 μg/mL), purchased from the culture medium preparation facility at MSKCC.

Transduction Procedure of U87 Cell Line

U87 human glioblastoma cells were grown in MEM with 10% fetal calf serum. The retroviral transient producer cell line H29 was maintained in Dulbecco MEM supplemented with 10% fetal calf serum, puromycin (2 μg/mL), and tetracycline (1 μg/mL), transfected with Cis-p53/tdTomato-CBRluc-Neo and incubated for 48 h. The U87 cells were transduced with the Cis-p53/tdTomato-CBRluc-Neo later using H29-produced virus by incubating 30% confluent tumor cell culture with retroviral vector-containing medium for 48 h. Transduced U87-p53/tdTomato-CBRluc-Neo cells (short: U87-p53) were selected by culturing in MEM, supplemented with 10% fetal calf serum and G418 sulfate (0.5 mg/mL) (Corning, catalog no. 30-234-CR). Several fluorescence-activated cell sorting (FACS) sorting procedures were performed to enrich cells with a low background of p53 activity. Selected cells were additionally transduced with a second constitutively expressed SFG-Turquoise/rsRluc virus as described above.

Cell-Internalization, Targeting Specificity, and Binding Properties

U251 MG cells were adhered overnight in 6-well plates (1 million cells/well) and washed twice with phosphate-buffered saline (PBS) prior to addition of ¹³¹I-PARPi (222 kBq/well, 0.15 nmol) at 37°C for 0, 15, 30, 45, 60, 120, and 240 min in MEM to allow for binding. Binding specificity was challenged with a 50-fold excess of olaparib (3.26 μg, 7.5 nmol). Next, at each time point, cells were first washed twice with PBS 1× and then lysed by incubation with 1 M NaOH (10 min at 37°C). Finally, the resulting lysate in each well was collected and the radioactivity measured in an automated γ-counter. Targeting specificity was assessed by a competitive binding assay between ¹²⁷I-PARPi and PARPi-FL. U251 MG cells (50,000 cells/well) were plated 48 h before the experiment. On the day of the experiment, binding affinity was determined by coincubating the plated cells (37°C, 40 min) with serial dilutions of ¹²⁷I-PARPi, ranging from 0 to 10 μM and 0.5 μM PARPi-FL. After this time, cells were washed twice with PBS 1×, fixed, and counterstained with Hoechst DNA stain. PARPi-FL uptake and blocking effects were observed and measured by confocal microscopy.

Mice

Female athymic nude CrTac:NCr-Fo mice were purchased from Taconic Laboratories at age 6–8 wk. Xenograft-bearing mice were used to determine all pharmacokinetics, the binding/imaging properties, and the treatment efficacy of the radiotherapeutic (n_total = 98). During subcutaneous injections, mice were anesthetized using 2% isoflurane gas in 2 L/min medical air. During orthotopic injections, mice were anesthetized using a 150 mg/kg ketamine and 15 mg/kg xylazine cocktail (10 μL/g of body weight). For intravenous injections, the lateral tail vein was used. Mice were warmed with a heat lamp and placed in a restrainer, and the tail was sterilized with alcohol pads before injection. U251 MG or U87-p53 cells were implanted subcutaneously (5 × 10⁶ cells in 150 μL of 1:1 PBS/Matrigel [BD Biosciences]) into the right shoulder and allowed to grow for approximately 2 wk until the tumors reached about 8 mm in diameter (100 ± 8 mm). For orthotopic injections, U251 MG or U87-p53 cells (5 × 10⁵ cells in 2 μL of PBS) were injected 2 mm lateral and 1 mm anterior to the bregma using a Stoelting Digital New Standard Stereotaxic Device and a 5-μL Hamilton syringe and allowed to grow for approximately 3 wk. All mouse experiments were performed in accordance with protocols approved by the Institutional Animal Care and Use Committee of MSKCC and followed National Institutes of Health guidelines for animal welfare.

SPECT/CT

SPECT/CT scans were obtained using a small-animal NanoSPECT/CT from Mediso Medical Imaging Systems. For subcutaneous U251 MG xenografts, ¹³¹I-PARPi (11.1 MBq, 1.5 GBq/μmol in 20 μL of 30% PEG300 in 0.9% sterile saline) was administered intratumorally (n = 3). At 0 and 6 h after injection, the mice were anesthetized with 1.5%–2.0% isoflurane (Baxter Healthcare) at 2 mL/min in oxygen, and SPECT/CT data were acquired for 20 min. For orthotopic U251 MG tumor–bearing mice, ¹³¹I-PARPi (1.9 MBq, 1.5 GBq/μmol in 5 μL of 30% PEG300 in 0.9% sterile saline) was injected intracranially using the same coordinates as for tumor cell injection (2 mm lateral and 1 mm anterior to the bregma using a Stoelting Digital New Standard Stereotaxic Device and a 5-μL Hamilton syringe). At 24 h after injection, the mice were anesthetized with 1.5%–2.0% isoflurane (Baxter Healthcare) at 2 mL/min in oxygen, and SPECT/CT data were collected for 20 min in the head region.

Autoradiography and Hematoxylin and Eosin Staining

Twenty-four hours after intratumoral injection of ¹³¹I-PARPi or olaparib/¹³¹I-PARPi in U251 MG subcutaneous xenograft–bearing mice, mice were sacrificed and tumors were harvested. The collected organs were embedded in Tissue-Tek O.C.T. compound (Sakura Finetek), flash-frozen in liquid nitrogen, and cut into 20-μm sections using a Vibratome UltraPro 5000 Cryostat (Vibratome). A storage phosphor autoradiography plate (BAS-MS2325; Fiji Photo Film) was exposed to the tissue slices overnight at −20°C and read the following day. Relative count intensity of the sections in each image was quantified using ImageJ 1.47u processing software (NIH). Tumor-to-muscle and brain-to-muscle ratios were calculated using Prism 6.0c (GraphPad Software). Sections were subsequently subjected to hematoxylin and eosin staining for morphologic evaluation of tissue pathology.

PET/CT Imaging

Small-animal PET imaging data were recorded on a Focus 120 MicroPET (Concorde Microsystems) and reconstructed using AsiPro VM MicroPET Analysis software (Siemens Medical Solutions). ¹⁸F-PARPi (5.55 MBq, 35.89 ± 15.17 GBq/μmol in 300 μL, 10% EtOH in 0.9% sterile saline) alone or following an intratumoral injection of ¹²⁷I-PARPi (10 μg, 16.5 μmol, 20 μL, 30% PEG300/70% NaCl 0.9% 6 h before ¹⁸F-PARPi injection) was injected into U251 MG tumor–bearing mice (n = 3) via the tail vein. At 2 h after injection, the mice were anesthetized with 1.5%–2.0% isoflurane (Baxter Healthcare) at 2 mL/min in oxygen, and PET/CT imaging was performed over 15 min. Images were analyzed using AsiPro VM software (Concorde Microsystems). Then, mice were sacrificed and key organs harvested for quantification and autoradiography/hematoxylin and eosin staining.

Ex Vivo Biodistribution and Dosimetry (Subcutaneous Model)

Biodistribution studies were performed in subcutaneous U251 MG xenograft–bearing athymic nude mice (n = 44, 11 randomized cohorts, 4 mice per cohort). Mice were divided into 2 groups (blocked and unblocked), and ¹³¹I-PARPi (0.56 MBq, 1.5 GBq/μmol in 20 μL, 30% PEG300 in 0.9% sterile saline) was administered intratumorally. The blocked group was preinjected 30 min before ¹³¹I-PARPi with olaparib (8.15 μg, 18 nmol in 100 μL, 30% PEG300/70% NaCl 0.9%). Mice were sacrificed by CO₂ asphyxiation at 1, 3, 6, 12, 24, 48, 72, and 96 h after injection (for ¹³¹I-PARPi–only groups) and at 1, 6, and 24 h after injections (for blocking experiments groups), and major organs were collected, weighed, and counted in a WIZARD² automatic γ-counter (PerkinElmer). The ¹³¹I-PARPi uptake was expressed as a percentage of injected dose per gram (%ID/g) using the following formula: [(activity in the target organ/g of tissue)/injected dose] × 100% (Supplemental Figs. 5 and 6; supplemental materials are available at http://jnm.snmjournals.org). The average of the 4 organ specimens from each time point was fitted to a biexponential function. The best fit for most organs comprised a 2-phase exponential clearance model. For 2 organs (the thyroid and large intestine), there was measured uptake over the first 12 h, and therefore these organs were fitted to an uptake and clearance exponential model. Cumulative uptake was calculated from the areas under the tumor uptake curves. Absorbed doses to tumor and all normal organs were estimated assuming absorbed fractions of 1 for the β-emissions of ¹³¹I and 0 for the penetrating photon emissions and using an equilibrium dose constant of 0.111 g Gy/MBq h (0.405 g cGy/μCi h). The ratio of tumor to normal organ dose exceeded 20 for all organs except the thyroid (Supplemental Fig. 4). The high dose to the thyroid gland was a likely consequence of the deiodination followed by uptake and then slow clearance (evidence of trapping in the thyroid tissue) (Supplemental Fig. 7).

Bioluminescence

Bioluminescence studies were performed in subcutaneous U87-p53 xenograft–bearing mice (n = 8, 4 mice per cohort). Mice were divided into 2 groups (treated and untreated). For treatment, mice were administered ¹³¹I-PARPi (14.8 MBq, 1.5 ± 0.3 GBq/μmol in 20 μL, 30% PEG300 in 0.9% sterile saline) intratumorally. The untreated group received an intratumoral injection of a 30% PEG300 solution in 0.9% sterile saline (20 μL). Imaging was performed on an IVIS Spectrum preclinical in vivo imaging system from PerkinElmer the day before treatment (baseline) and on day 0, 1, and 2 after treatment. For detection of the constitutively expressed SFG-Turquoise/rsRluc reporter, mice were given a retroorbital injection of 10 μg of coelenterazine h ((2-(4-dehydroxy)coelenterazine from Biotium) dissolved in 5% ethanol in 0.9% sterile saline solution and imaged within 22 s of injections. For detection of p53 activation (Cis-p53/tdTomato-CBRluc-Neo), mice were imaged 20 min after an intraperitoneal injection of 2 mg of luciferin/100 μL of sterile 0.9% NaCl solution.

Therapeutic Efficacy

Therapeutic efficacy studies were performed in subcutaneous U87-p53 xenograft–bearing mice (n = 29, 3 randomized groups, 9–10 mice per cohort). Mice were divided into 3 cohorts (vehicle, nonradioactive/cold ¹²⁷I-PARPi, and 3 × 14.8 MBq of ¹³¹I-PARPi). The cold group received mass equivalent doses (6 μg) of ¹²⁷I-PARPi. Vehicle, ¹²⁷I-PARPi, and 3 × ¹³¹I-PARPi were administered on day 0, 3, and 6 via intratumoral injections. Body weights were recorded every Monday, Wednesday, and Friday until tumor burden exceeded the endpoint of 800 mm.

Additional supporting tables and figures can be found in the supplemental materials. All animal experiments were performed in accordance with protocols approved by the Institutional Animal Care and Use Committee and followed the National Institutes of Health guidelines for animal welfare.

Orthotopic CED Model

Mice received intracranial injections of U87-p53 cells (5 × 10⁴ cells in 1.5 μL of PBS). After 3 wk, bioluminescence imaging was used to determine presence of tumor. Then, we surgically implanted an ALZET Osmotic Pump into the mice, which slowly delivered an infusion into the brain using the same coordinates as for the tumor cell injection. Both healthy and diseased mice received an ALZET Osmotic Pump (Model 1003D) with Brain Infusion Kit 3 containing ¹³¹I-PARPi (40.7 ± 11.1 MBq in 101 ± 8 μL 30% PEG300/70% NaCl 0.9%). Pumps were weighed before and after filling to calculate added volume, as per the manufacturer’s instruction. After filling pumps and assembling catheter and cannula, infusion kits were incubated in 0.9 % saline at 37°C overnight to ensure flow was initiated (1.0 μL/h [±0.15 μL/h]) and air bubbles were removed. Immediately before implantation, pumps were measured in dose calibrators to measure activity in pumps at time of surgery. SPECT/CT images were acquired 1 h after surgery and again at 12 h. Mice were imaged every 24 h for 6 days. After the fifth day, the pumps were surgically removed, and the remaining dose was measured in dose calibrators to ensure payload had been delivered. The image volumes were segmented using a combination of manual and semiautomatic segmentation techniques in 3D Slicer, version 4.6 (www.slicer.org) for quantification of absolute activity in different organs at each time point. The presence of activity levels above background in the SPECT images was not observed except in the brain/tumor and the osmotic pump.

Dosimetry (CED)

Synthesis, Specificity, and In Vitro Performance of ¹³¹I-PARPi

Several research groups have synthesized PARP-targeted radiotracers in the past; some of which were attached to radioiodine (14,22–24). We showed that one of these tracers, ¹³¹I-PARPi (14), has a high affinity with a concentration at which 50% of binding has been inhibited (IC₅₀) in the nanomolar range (11 ± 3 nM) and a logP_CHI of 2.3. ¹³¹I-PARPi was synthesized with a final molar activity of 1.5 GBq/μmol. Radiochemical purity was 99.1% ± 0.9% for all prepared compounds. Pharmacologically, ¹³¹I-PARPi behaves in a manner similar to the Food and Drug Administration–approved PARP inhibitor olaparib (Fig. 1; Supplemental Fig. 1; (25)). We tested the inhibitory activity of ¹²⁷I-PARPi on a panel of 12 PARP enzymes, including tankyrase 1 and tankyrase 2 (Fig. 1B). We found near-perfect overlap with literature-known values for ¹⁸F-PARPi, PARPi-FL, and olaparib at 100 nM (12). ¹⁸F-PARPi is a PARP-targeted PET imaging agent with an IC₅₀ of 3 nM (Supplemental Fig. 1) (18). PARPi-FL is a fluorescent PARP inhibitor with an IC₅₀ value of 12.2 ± 1.1 (17). The 2 dominant targets for ¹²⁷I-PARPi were found to be PARP-1 and PARP-2, just as for ¹⁸F-PARPi, PARPi-FL, and olaparib (26).

The in vitro binding specificity and internalization of ¹³¹I-PARPi was shown by incubating U251-MG cells with ¹³¹I-PARPi in the presence and absence of 150 μL of 6.67 μg/μL olaparib solution suspended in 25 μL of DMSO, 25 μL of polyethylene glycol (PEG), and 100 μL saline (Fig. 1C). In both cases, binding reached a plateau at approximately 50 min and was 10-fold higher in cells exposed to only ¹³¹I-PARPi than in cells coincubated with olaparib. Additionally, we showed that the radiotherapeutic binds to its intended target within the nucleus by incubating U251-MG cells with serial dilutions of its nonradioactive counterpart, ¹²⁷I-PARPi, in the presence of 0.5 μM PARPi-FL. Overall, uptake of PARPi-FL retained in the nucleus at a concentration of 0.5 μM could be suppressed by addition of a 20-fold excess of ¹²⁷I-PARPi (Supplemental Fig. 2B).

Intratumoral Injection and Tissue Perfusion of ¹³¹I-PARPi

Intratumoral injections were implemented with a clinical CED model in mind, as well as for the favorable biodistribution (Fig. 2). To determine how intratumorally injected ¹³¹I-PARPi/¹²⁷I-PARPi distributes throughout a tumor, we designed an in vivo competitive binding specificity experiment with ¹⁸F-PARPi, a previously described PET imaging agent for PARP-1. One cohort of animals was intratumorally injected with ¹²⁷I-PARPi, followed by ¹⁸F-PARPi 30 min later. The second cohort was injected with ¹⁸F-PARPi only (Fig. 2A). We found that intratumoral injection of ¹²⁷I-PARPi suppressed uptake of ¹⁸F-PARPi, indicating that the small molecule occupies ¹⁸F-PARPi binding sites (Figs. 2B and 2C), similar to what we showed in ¹⁸F-PARPi/olaparib blocking experiments previously (18). Ex vivo quantification at 1 h after injection of ¹⁸F-PARPi demonstrated that uptake dropped by 1.7 ± 0.1 %ID/g (1.9 ± 0.3 and 0.18 ± 0.07 %ID/g, P < 0.0001) between mice without and with intratumoral injection of ¹²⁷I-PARPi (Fig. 2C). When replicated with olaparib and ¹³¹I-PARPi, brain uptake was low between mice without and with intravenous injection of olaparib (0.056 ± 0.020 and 0.061 ± 0.026 %ID/g, respectively) (Supplemental Figs. 3 and 4). Using autoradiography, we found heterogeneity in the intratumoral distribution of ¹³¹I-PARPi (Supplemental Fig. 5A). Overall quantification, however, corroborated the ¹⁸F-PARPi PET experiments (Fig. 2), and ratios between animals receiving only ¹³¹I-PARPi versus animals receiving a preinjection of olaparib were 7.8 (144.3 ± 32.7 AU and 18.5 ± 1.2 AU, P < 0.0001 for mice without and with preinjection of olaparib) (Supplemental Fig. 5B).

¹³¹I-PARPi Pharmacokinetics and Dosimetry in Glioblastoma

CED delivery of therapeutics to the brain relies on a drug’s fast perfusion, tissue permeability, and clearance from areas not expressing the targeted biomarker. To simulate this setting, we tested retention of ¹³¹I-PARPi in 2 cohorts of U251-MG subcutaneous xenografts. In 1 cohort, PARP was made unable to retain the radiotherapeutic through intravenous preinjection of olaparib. NanoSPECT/CT showed that at 6 h after intratumoral injection of ¹³¹I-PARPi (11.1 MBq, 20 μL, 30% PEG/0.9% NaCl solution), the radiotherapeutic was retained in tumors not previously injected with olaparib, whereas it had almost completely cleared from the blocked group (Fig. 3A). This observation was further corroborated by biodistribution experiments with endpoints (each time point n = 4) between 0 and 96 h after injection (Figs. 3B and 3C; Supplemental Figs. 3, 4, and 6). On the basis of these data, we calculated the activity and dose retained in different tissues (Supplemental Fig. 7). Intuitively, the highest dose was delivered to tumor tissue, with a calculated absorbed dose of 134.1 cGy/MBq. This was followed by the thyroid (68.6 cGy/MBq), which metabolizes and stores iodine (Supplemental Fig. 6; (27)). Besides PARP expression in the thyroid, storage of metabolized iodine is likely one reason for this comparatively high calculated dose. This is supported by the finding that in this setting, physical decay was observed as the only mode of clearance. However, moving forward, thyroid uptake can efficiently be blocked by administration of nonradioactive iodide (28).

Bioluminescence and p53 Response

We designed and transduced a reporter cell line responding to p53 activation, a marker of cell death and radiation damage. The cells coexpress p53/luciferase, together with constitutively expressing SFG-Turquoise/rsRluc. These bioluminescent reporters allow p53 expression to be imaged after injection of luciferase, and cell density to be imaged after injection of coelenterazine. Using this setup, we used bioluminescence imaging to determine p53 expression in relation to cell density (29). To use this experimental setup for testing whether ¹³¹I-PARPi retention is sufficient for inducing cell death, mice bearing subcutaneous transfected U87-p53 xenografts were injected with 16.3 ± 1.5 MBq of ¹³¹I-PARPi or vehicle (Supplemental Fig. 8). Dual reporter bioluminescence was measured 1 d before, on the day of, and 24 and 48 h after ¹³¹I-PARPi injection. For the control groups, no statistical differences were observed for any time points, suggesting that vehicle alone did not induce cell death (Supplemental Fig. 8). For mice injected with ¹³¹I-PARPi, however, a statistically significant increase in p53 signal was observed 24 h after administration of ¹³¹I-PARPi, which persisted at 48 h after injection (P < 0.005 after 24 h and P < 0.05 after 48 h when compared with day 0).

Therapeutic Efficacy of ¹³¹I-PARPi in Mouse Model of Glioblastoma

We next sought to determine the therapeutic efficacy of ¹³¹I-PARPi in a subcutaneous xenograft model of transduced U87-p53. For this purpose, tumor-bearing mice were randomly assigned to 3 different cohorts (n = 9–10 per cohort): vehicle (30% PEG-300/PBS), nonradioactive ¹²⁷I-PARPi (5.9 μg, 9.9 nmol), or fractionated doses of ¹³¹I-PARPi (3 × 14.8 MBq, 3 × 9.9 nmol). For the fractionated-dose cohort, vehicle and cold ¹²⁷I-PARPi intratumoral injections were performed on days 0, 3, and 6. Body weights and tumor volumes were recorded 3 times per week for the length of the study (Supplemental Fig. 9B); neither cohort experienced statistically significant weight loss, confirming that the administered activity did not lead to systemic toxicity (Supplemental Fig. 9B). We performed 2-sample t tests to compare tumor growth curves under dependent right censoring to account for sacrifices due to ethical guidelines and applied a correction for the multiple pairwise group comparisons (Supplemental Fig. 9D) (30). No statistical difference of tumor volume was observed between the vehicle cohort and the ¹²⁷I-PARPi cohort. The fractionated-dose treatment led to a statistically significant reduction in tumor volume compared with the vehicle group (P < 0.005). This is also reflected in the tumor growth rates, which were not different at the beginning of the study (week 0) but showed reduced growth rates in the 2 radiotherapy groups at week 2 compared with the control group (Fig. 4A). When comparing survival of mice, statistical significance was found for the treatment group versus the vehicle (P < 0.005). No statistical difference was found between the vehicle group and cold ¹²⁷I-PARPi. Mice receiving fractionated doses had the longest median survival (29 d). Median survival for the PBS group and cold ¹²⁷I-PARPi group was 22 and 20 d, respectively (Fig. 4B).

Orthotopic Model

Simulating a potential clinical treatment scenario, we tested the delivery of ¹³¹I-PARPi in an orthotopic U87-p53 mouse model of glioblastoma paired with an implantable osmotic pump and cannula (Fig. 5A). The osmotic pump mimicked slow, CED-type delivery of our radiotherapeutic, and we compared the retention of ¹³¹I-PARPi in tumor-bearing versus healthy mice (Supplemental Fig. 10). On the basis of calculations of the administered dose from SPECT-derived pump activity concentrations versus time, it was shown that administration of 30.7 MBq (49.2 MBq initially implanted) resulted in the highest doses in the brain/tumor itself (30.1622 ± 0.1946 cGy/MBq ID), induced overwhelmingly by tumor-bound ¹³¹I-PARPi (27.4405 ± 0.03784 cGy/MBq ID) versus photon dose from the pump contents (2.0768 ± 0.100 cGy/MBq ID; Supplemental Fig. 5). Other organs, affected primarily by γ-radiation from the pump contents, received comparatively small doses (Fig. 5B). Only the left lung, which was in direct proximity to the osmotic pump activity reservoir, received a somewhat elevated dose (11.08 cGy/MBq injected dose). Experimental findings corroborated these models, as strong, durable retention of ¹³¹I-PARPi was observed in tumor-bearing mice, whereas healthy brain tissue retained the radiotherapeutic to a much lower degree (Fig. 5C); after 6 d, 9.1 Gy were delivered to the brains of tumor-bearing mice, whereas healthy mice retained less than 1 Gy.