First In Vivo and Phantom Imaging of Cyclotron-Produced 133La as a Theranostic Radionuclide for 225Ac and 135La

Visual Abstract

Ther anostic pairs in nuclear medicine involve labeling molecular target vectors first with a diagnostic radionuclide, followed by a therapeutic particle-emitting radionuclide (1). Both radionuclides should have similar chemical properties, ideally being isotopes of the same element. Theranostics has strong potential in targeted radionuclide therapy, with a diagnostic positron or g-emitting radionuclide used in PET or SPECT being paired with a therapeutic radionuclide emitting a-particles, b 2 -electrons, or Auger electrons (2). Recently introduced 133 La (half-life [t1 =2 ], 3.9 h), 132 La (t1 =2 , 4.8 h), and 134 Ce (t1 =2 , 3.2 d)/ 134 La (t1 =2 , 6.5 min) PET radionuclides are uniquely suited as theranostic imaging partners for 225 Ac (t1 =2 , 9.9 d) in targeted a-therapy or with 135 La (t1 =2 , 19.5 h) in Auger electron therapy (AET) because of their chemical similarity to, and longer half-lives than, the ubiquitous PET radiometal 68 Ga (t1 =2 , 68 min) (2)(3)(4)(5)(6)(7). 225 Ac has shown considerable efficacy in clinical trials for treating metastatic cancers (2,8). 132 La has been proposed as a theranostic PET imaging surrogate for 225 Ac therapy and has displayed in vivo uptake characteristics similar to those of 225 Ac (6). However, there are fundamental imaging limitations inherent in 132 La because of its high maximum positron emission energy (E max ) and mean positron emission energy (E mean ) (E max /E mean , 3.67/1.29 MeV), which significantly reduces image spatial resolution and contrast compared with other PET radionuclides (e.g., 18 F E max /E mean , 0.634/0.250 MeV; 68 Ga E max /E mean , 1.90/0.829 MeV; 64 Cu E max /E mean , 0.653/0.278 MeV; 44 Sc E max /E mean , 1.47/0.632 MeV), and its high-energy and highintensity g-emissions, which are problematic from a dosimetric perspective (3,9). 133 La has a lower positron emission energy (E max / E mean , 1.02/0.461 MeV) than 132 La,68 Ga, or 44 Sc; energy comparable to 89 Zr (E max /E mean , 0.902/0.396 MeV), and lower energy and lower-intensity g-emissions than 89 Zr, 44 Sc, or 132 La (3). Here, as outlined in Figure 1, we describe a high-yield cyclotron production method for 133 La using periodic table mix inductively coupled plasma optical emission spectrometry (ICP-OES) elemental standards were obtained from Sigma-Aldrich. Oxalic acid dihydrate (99.5%) was purchased from Fisher Scientific. Aluminum disks were obtained from Michaels, and aluminum foil was purchased from Goodfellow Cambridge. Indium wire was purchased from AIM Specialty Materials. Branched diglycolamide resin was purchased from Eichrom. Eckert and Ziegler Isotopes National Institute of Standards and Technology-traceable g-ray sources were used for highpurity germanium (HPGe) detector energy and efficiency calibration. Thin-layer chromatography silica gel sheets were purchased from Merck. Water (18 MVÁcm) was obtained from a MilliporeSigma Direct-Q 3 ultraviolet system. 89 Zr was provided by the Washington University Cyclotron Facility. DOTA was purchased from Macrocyclics. Macropa was purchased from MedChemExpress. PSMA-I&T was obtained from ABX Advanced Biochemical Compounds. DCFPyL was synthesized in-house.

Instrumentation
Activity and radionuclidic purity were assessed using an Ortec GEM35P4-70-SMP HPGe detector running GammaVision software, with dead times below 25%. Elemental purity was assessed using an Agilent Technologies 720 Series ICP-OES. A NEPTIS Mosaic-LC synthesis unit (Optimized Radiochemical Applications) separated 133 La from the Ba target solution.
An Eckert and Ziegler AR-2000 radio-thinlayer chromatography imaging scanner quantified the fraction of chelator-bound 133 La after reaction. Solid targets were manufactured using a Carver model 6318 hydraulic press and an MTI Corp. 10-mm (internal diameter) EQ-Die-10D-B hardened steel die. A Carbolite 16/610-tube 3-zone furnace was used for 135 BaCO 3 recovery. X-ray powder diffraction (XRD) patterns were acquired on starting and recovered BaCO 3 and intermediate BaC 2 O 4 using a Rigaku Ultima IV x-ray diffractometer to confirm phase identity and purity.

Cyclotron Targeting and Irradiation
Figure 2 depicts nuclear reaction crosssections for the 13x Ba(p,xn) 13x La reactions of interest for 132/133/135 La production from the TENDL 2019 library, weighted for nat Ba and isotopically enriched 135 BaCO 3 target material (10). Cyclotron targets were prepared with 150-200 mg of nat Ba metal, nat BaCO 3 , or enriched 135 BaCO 3 , a roughened aluminum disk (24 mm in diameter, 1.35 mm thick), indium wire (1 mm in diameter), and aluminum foil (125 mm thick) in a manner similar to that previously described (3,11). Aluminum was shown to be an adequate substitute for silver, presenting a lower cost and activation. Target components are shown in Supplemental Figure 1 (supplemental materials are available at http://jnm.snmjournals.org). Targets were irradiated for 5-263 min at 11.9 and 23.3 MeV using an Advanced Cyclotron Systems Inc. TR-24 cyclotron, at proton beam currents of 10 mA incident on the target assembly. Higher energy runs (beam-extracted at 24 MeV, 23.3 MeV incident on target pellets, 20.2 MeV exiting Ba metal, and 19.4 MeV exiting BaCO 3 ) were performed with 200 mg of barium material with the aluminum target cover facing the beam, to maximize 133 La production based on TENDL 2019 cross-section simulation data (10). During higher-energy runs, a silver disk was placed behind the target to avoid 13 N production from the 16 O(p,a) 13 N reaction. For lower-energy runs (18.2-MeV extraction, 11.9 MeV incident on target pellets, 7.8 MeV exiting Ba metal, and 6.4 MeV exiting BaCO 3 ), performed to maximize 135 La production, 150 mg of barium material were used, and the target was installed in reverse with the aluminum disk acting as a degrader to reduce beam energy from 18.2 to 11.9 MeV, as calculated using SRIM (12).

133
La and BaCO 3 were separated using a process with aspects derived from previous studies (3,4). The target was opened by peeling back the aluminum cover and placed in a Teflon (DuPont) dissolution vessel. The vessel was filled with 10 mL of 18 MVÁcm water and  Flow rates were kept below 2 mLÁmin 21 to avoid 133 La loss from the resin. 133 LaCl 3 was eluted using 1 mL of 0.05 M HCl. After passing through the resin, the first 30 mL of process solution were diverted to a collection vial for subsequent BaCO 3 recovery. After separation, target components were sonicated in 18 MVÁcm water for reuse. Radionuclidic and elemental purity of 133 LaCl 3 was determined by HPGe g-ray spectroscopy and ICP-OES.

BaCO 3 Target Material Recovery
The 30 mL of barium recovery solution were neutralized to pH 6-8 with NH 4 OH. Ten milliliters of 0.8 M C 2 H 2 O 4 were added to the recovery solution to precipitate BaC 2 O 4 . The solution was passed through a fritted column to trap BaC 2 O 4 and washed with 50 mL of 18 MVÁcm water. BaC 2 O 4 was removed from the column and then heated to 550 C for 2 h in a sealed tube furnace with an airflow of 20 mL/min to decompose BaC 2 O 4 to BaCO 3 while avoiding conversion to BaO (13). Waste gases from decomposition were vented to a fume hood. Recovery was quantified by gravimetric analysis of dried samples and tracked by HPGe g-spectroscopy using g-emissions from 135m Ba (268 keV; t1 =2 , 28.7 h). Samples of purchased BaCO 3 , precipitated BaC 2 O 4 , and recovered BaCO 3 were analyzed by XRD to identify the product and evaluate its quality.

Phantom Imaging
Phantom imaging was performed using Derenzo and National Electrical Manufacturers Association (NEMA) image-quality phantoms on an Inveon PET/CT scanner (Siemens Preclinical Solutions), as described by Ferguson et al. (14). The Derenzo phantom, used to investigate image contrast and spatial resolution, consists of sections with rods of varying diameters (0.8, 1.0, 1.25, 1.5, 2.0, and 2.5 mm) that are filled with the radionuclide of interest diluted in 20-30 mL of water. The NEMA phantom, used to investigate image noise, spillover ratio, and recovery coefficient, consists of several fillable sections including two 7.5-mm-diameter cold-air and water cylindric volumes. NEMA and Derenzo phantom scans for 133 La,132 La, and 89 Zr were acquired in list mode, binned into sinograms, and reconstructed with the default filtered backprojection, ordered-subset expectation maximization, and maximum a posteriori estimation algorithms. Acquisition, data processing, and evaluation followed the same procedure as used by Ferguson et al. (14) for 18 F, 64 Cu, 68 Ga, and 44 Sc to enable direct comparison of the different radionuclides' imaging performance.
Radiolabeling of DOTA, PSMA-I&T, and Macropa with 133 La Similar to techniques in previous studies (3,4), the activity of a 500-mL 133 LaCl 3 aliquot was measured, and the solution pH was adjusted to 4.5 with 50 mL of NaOAc buffer (pH 9.0). A 100-mL volume of this 133 La solution (5-150 MBq) was reacted with 0.1-20 mg of DOTA, PSMA-I&T, and macropa dissolved in 50 mL of 18 MVÁcm water at 90 C for 30 min. Each solution was analyzed using radio-thinlayer chromatography on silica plates to determine radiochemical purity and incorporation with 0.1 M citric acid buffer as the mobile phase.

Preclinical PET Imaging
Animal studies using LNCaP tumor-bearing male nu/nu nude mice (Charles River Laboratories) were performed according to the guidelines of the Canadian Council on Animal Care and approved by the local Cross Cancer Institute Animal Care Committee. Static PET image scans (20-min duration) of 133 La-PSMA-I&T at 60 min after injection were performed on an Inveon PET/CT scanner (Siemens Preclinical Solutions). Blocking experiments were performed using the PSMA-targeting agent DCFPyL. Radiotracer (33-50 MBq of 133 La-PSMA-I&T in 80-120 mL of NaOAc/saline) and blocking compound (300 mg of DCFPyL, dosed 5 min beforehand) were injected into the tail vein of isoflurane-anesthetized mice (100% oxygen; gas flow, 1.5 L/min), the mice were placed in a prone position into the center of the field of view, and body temperature was kept constant at 37 C. A transmission scan for attenuation correction was not acquired. The frames were reconstructed using ordered-subset expectation maximization and maximum a posteriori algorithms. No correction for partialvolume effects was applied. The image files were processed using ROVER software (version 2.0.51; ABX GmbH).

Statistical Analysis
All data are given as mean 6 SD (n $ 3).

Cyclotron Targeting and Irradiation
Average end-of-bombardment activities (n 5 3) of 133 La and coproduced 135 La for 100 mAÁmin runs (10 mA for 10 min) at 11.9and 23.3-MeV beam energies with different barium target materials are summarized in Table 2 Table 3 contains ICP-OES elemental purity results for the 133 LaCl 3 product. After removal from the reactor after sonification, the aluminum target backing and cover contained no detectable 133           contrast. 133 La exhibits spatial resolution similar to that of 89 Zr, is an improvement over 44 Sc and 68 Ga, and is superior to 132 La. Figure 6 plots the contrast between the rods and background for each of the 6 triangular segments in the Derenzo phantom and the recovery coefficients as a function of rod size in the NEMA image-quality phantom. Additional comparisons of imaging performance metrics between radionuclides for different reconstruction algorithms are included in Supplemental Figure 5. 133 La exhibits contrast similar to that of 89 Zr and is superior to 68 Ga and 44 Sc for larger rod diameters. 132 La was not included in the contrast comparison because of the low contrast for each rod diameter. The rods could not be distinguished below 1.25 mm in diameter for the higher-energy positron emitters 44 Sc and 68 Ga and 1 mm for the lower-energy positron emitters 18 F and 64 Cu. This blurring is due to the extrinsic scanner resolution, which is significantly impacted by the positron energy and therefore range. The recovery coefficient comparison demonstrates that 133 La exhibits favorable performance compared with 68 Ga and 132 La.

Radiolabeling
Radiolabeling was performed at 90 C for 30 min and analyzed with radio-thin-layer chromatography using 0.1 M citric acid buffer as the mobile phase. The 133 La-DOTA, 133

DISCUSSION
This study presents cyclotron production of 133 La using natural and isotopically enriched barium target material, favorable fundamental PET phantom imaging characteristics of 133 La, and the first (to our knowledge) in vivo preclinical PET tumor imaging using 133 La-PSMA-I&T. The new target assembly is well suited to the irradiation and processing of barium metal and BaCO 3 target material. Using aluminum instead of silver target backings as used in previous studies (3,11) avoids production of long-lived 107 Cd, 109 Cd, and 106m Ag, thereby strongly reducing overall activation of the target, lowering operator exposure, and enabling rapid reuse. Using the target backing as an intrinsic degrader simplifies and enhances the available range of irradiation energies. The indium wire seal stayed 1-2 mm outside the target beam spot, avoiding activation and formation of radiotin isotopes. Sonicating used target disks in 18 MVÁcm water allowed repeated reuse to make additional targets, with the same seal and target backing reused over 5 times.
Irradiating enriched 135 BaCO 3 at 23.3 MeV produced far more 133 La than did other target materials, allowing production of clinically relevant 133 La activities with significantly shorter irradiation times than using nat Ba target material. Target separation gave a high 133 LaCl 3 yield in a 1-mL product volume, ready for radiolabeling. Recovery of BaCO 3 target material demonstrated feasibility for cost-effective recovery of expensive isotopically enriched 135 BaCO 3 . XRD analysis of recovered BaCO 3 showed complete conversion of the BaC 2 O 4 intermediate and a pure recovered product, validating target material recovery and highlighting the potential for substantially improved economics with a simple and inexpensive recovery process. Radiolabeling DOTA, PSMA-I&T, and macropa with 133 La achieved high apparent molar activities for fresh and recovered BaCO 3 target material, similar to radiolanthanum chelation in previous studies (3)(4)(5)16).
Using isotopically enriched 135 BaCO 3 target material permits selective production of 133    significantly increases 133 La production via the 135 Ba(p,3n) 133 La reaction and reduces 135 La production from the 135 Ba(p,n) 135 La reaction, which is ideal for PET imaging applications. Irradiating at 11.9 MeV with enriched 135 BaCO 3 is ideal for producing large activities of pure 135 La for AET. Using these 2 distinct reactions permits production of a variety of 133/135 La isotopic blends on a variable-energy cyclotron.
Another production route could use isotopically enriched 134 BaCO 3 target material to produce 133 La via the 134 Ba(p,2n) 133 La reaction. This would enable 133 La production on lower-energy cyclotrons because of the 134 Ba(p,2n) 133 La cross-section threshold at 12 MeV as opposed to the 20 MeV threshold for the 135 Ba(p,3n) 133 La reaction. The lower natural isotopic abundance of 134 Ba (2.4%) than of 135 Ba (6.6%) would result in a higher isotopic enrichment cost. However, this is a compelling option for PET centers with lower-energy cyclotrons because of the 95.4% recovery yield of BaCO 3 target material demonstrated in this study.
PET phantom imaging clearly showed that 133 La exhibits spatial resolution and contrast superior to those of 44 Sc, 68 Ga, 132 La but similar to those of 89 Zr. As expected, lower positron emission energy leads to improved spatial resolution (17) and results in superior image quality for 133  Ga, and 41.2% 132 La), the LNCaP tumor was clearly defined, reaching an SUV mean of 0.97 6 0.17 or 3.94 6 0.68 %ID/g at 60 min after injection. For 68 Ga-PSMA-I&T, 4.95 6 1.47 %ID/g uptake into LNCaP tumors was reported in an ex vivo biodistribution study (18). As discussed previously (3), in vivo studies involving retention and dosing of 133 La decay daughter 133 Ba would be useful to address this potential limitation; however, as shown by Newton et al. (19), most 133 Ba activity could be expected to be excreted within 10 d after injection.
Since lanthanum and actinium are group 3 elements with similar chemical properties, 133 La is highlighted as a strong candidate to become a clinical PET imaging surrogate for 225 Ac a-therapy, with PET imaging characteristics superior to those of 132 La. As established in this study and previously (3), compared with 132 La, 133 La has superior inherent cyclotron production characteristics, a lower positron energy that translates to a higher spatial resolution, and lowerenergy and lower-abundance g-emissions that would translate to a lower patient and operator dose. These characteristics suggest that 133 La represents an attractive candidate for diagnostic PET imaging and treatment monitoring of clinical 225 Ac targeted a-therapy and research involving 135 La AET.

CONCLUSION
This work demonstrates the strong potential of 133 La to serve as a theranostic PET imaging agent with 225 Ac targeted a-therapy or 135 La AET. The first preclinical in vivo PET imaging studies on LNCaP tumors resulted in high spatial resolution and contrast. Phantom imaging of 133 La demonstrated that fundamental PET imaging properties, including spatial resolution, contrast, and recovery coefficient, were superior to those of other PET radiometals such as 68 Ga, 44 Sc, and 132 La and similar to those of 89 Zr. With cyclotron production routes capable of generating clinically relevant 133 La activities, and with demonstrated feasibility for performing high-yield recovery of expensive isotopically enriched 135 BaCO 3 target material, 133 La appears to be a promising radiometal candidate for high-resolution PET imaging as a PET/targeted a-therapy theranostic pair with 225 Ac or a PET/AET theranostic pair with 135 La.

DISCLOSURE
The Dianne and Irving Kipnes Foundation supported this work. Bryce Nelson received graduate scholarship funding from Alberta Advanced Education and the Natural Sciences and Engineering Research Council of Canada. No other potential conflict of interest relevant to this article was reported.