Glypican-3-Targeted Alpha Particle Therapy for Hepatocellular Carcinoma

Glypican-3 (GPC3) is expressed in 75% of hepatocellular carcinoma (HCC), but not normal liver, making it a promising HCC therapeutic target. GC33 is a full-length humanized monoclonal IgG1 specific to GPC3 that can localize to HCC in vivo. GC33 alone failed to demonstrate therapeutic efficacy when evaluated in patients with HCC; however, we posit that cytotoxic functionalization of the antibody with therapeutic radionuclides, may be warranted. Alpha particles, which are emitted by radioisotopes such as Actinium-225 (Ac-225) exhibit high linear energy transfer and short pathlength that, when targeted to tumors, can effectively kill cancer and limit bystander cytotoxicity. Macropa, an 18-member heterocyclic crown ether, can stably chelate Ac-225 at room temperature. Here, we synthesized and evaluated the efficacy of [225Ac]Ac–Macropa–GC33 in mice engrafted with the GPC3-expressing human liver cancer cell line HepG2. Following a pilot dose-finding study, mice (n = 10 per group) were treated with (1) PBS, (2) mass-equivalent unmodified GC33, (3) 18.5 kBq [225Ac]Ac–Macropa–IgG1 (isotype control), (4) 9.25 kBq [225Ac]Ac–Macropa–GC33, and (5) 18.5 kBq [225Ac]Ac–Macropa–GC33. While significant toxicity was observed in all groups receiving radioconjugates, the 9.25 kBq [225Ac]Ac–Macropa–GC33 group demonstrated a modest survival advantage compared to PBS (p = 0.0012) and 18.5 kBq [225Ac]Ac–IgG1 (p = 0.0412). Hematological analysis demonstrated a marked, rapid reduction in white blood cells in all radioconjugate-treated groups compared to the PBS and unmodified GC33 control groups. Our studies highlight a significant disadvantage of using directly-labeled biomolecules with long blood circulation times for TAT. Strategies to mitigate such treatment toxicity include dose fractionation, pretargeting, and using smaller targeting ligands.


Introduction
Liver cancer is the second leading cause of cancer-related death worldwide, with a total of 840,000 cases and as many deaths each year [1,2]. Hepatocellular carcinoma (HCC) accounts for approximately 85% of liver cancer cases globally [3]. In the U.S., there are 40,000 new HCC cases diagnosed each year, with an average 5 year survival of 19.6% [4]. Moreover, HCC incidence and mortality are increasing, with significant contribution from obesity-related non-alcoholic fatty liver disease (NAFLD), also referred to as metabolicassociated fatty liver disease (MAFLD) [5,6].
The majority of patients with HCC present with advanced disease [4]. Until recently, first-line treatment options have included tyrosine kinase inhibitors sorafenib and lenvatinib [7]. In the IMbrave 150 phase 3 clinical trial of treatment-naïve patients with HCC comparing sorafenib to combined atezolizumab (anti-PD-L1 antibody) and bevacizumab (anti-VEGF antibody), the experimental arm showed a clear survival benefit. These results led to FDA approval of the combination, offering a new first-line treatment option for patients with unresectable HCC [8]. However, while immunotherapy combined with antiangiogenic agents offers a new modality that can help improve the outcomes of patients with HCC, more work to identify and exploit the unique vulnerabilities of HCC is needed. For example, in the era of precision oncology, no HCC-targeted treatments currently exist as part of the standard of care.
The HCC-selective target, glypican-3 (GPC3), is a known pathological marker of HCC and is overexpressed in approximately 75% of HCC [9][10][11][12]. GC33 (a.k.a. codrituzumab) is a humanized monoclonal IgG1 specific for GPC3, which has been evaluated in a phase II clinical trial in patients with advanced HCC who had previously progressed on sorafenib. Surprisingly, despite having subnanomolar affinity for GPC3, demonstrating excellent in vivo target binding and therapeutic efficacy in preclinical studies [5,13,14], the effect of GC33 was not different from placebo with respect to progression-free and overall survival [15]. These findings show that, for GC33, antibody-dependent cellular cytotoxicity is not sufficient for an anti-tumor effect in patients and suggest that a cytotoxic payload may be needed to maximize its therapeutic effect.
We posited that coupling GC33 to an alpha particle-emitting radionuclide could enhance therapeutic effect in a preclinical model of HCC. Alpha particles inherently deposit a high amount of ionizing radiation over a distance on the order of cellular diameters. If localized to tumors, alpha particles can cause irreparable double-strand DNA breaks while minimizing dose to neighboring healthy tissues. Actinium-225 is a clinically relevant radionuclide that has been used in patients to deliver alpha particles to leukemia [16], prostate [17,18], and neuroendocrine cancers [19].
Conventionally, the metal chelator DOTA is conjugated to tumor-targeted ligands (e.g., peptides or antibodies) to create the targeted alpha particle therapeutic (TAT) agent. Chelation of Ac-225 by DOTA requires temperatures in the 60-80 • C range, which can denature many biomolecules, including antibodies. Recently, Thiele et al. showed that Macropa, an 18-member heterocyclic crown ether, can stably chelate Ac-225 at room temperature [20]. Here, we synthesized and characterized a [ 225 Ac]Ac-Macropa-GC33 radioconjugate and assessed its therapeutic effect and toxicity in mice using xenografts of the liver cancer cell line HepG2.

Radioimmunotherapy with [ 225 Ac]Ac-Macropa-GC33 Shows Tumor Reduction Compared to [ 225 Ac]Ac-Macropa-IgG1
To determine the optimal dose for [ 225 Ac]Ac-Macropa-GC33, a pilot study was completed in a HepG2 subcutaneous flank model to test multiple administered activities of this radioconjugate. Treatment with 18.5 kBq of [ 225 Ac]Ac-Macropa-GC33 was found to cause tumor growth delay and was well tolerated. Therefore, we used 18.5 kBq as the

Radioimmunotherapy with [ 225 Ac]Ac-Macropa-GC33 Shows Tumor Reduction Compared to [ 225 Ac]Ac-Macropa-IgG1
To determine the optimal dose for [ 225 Ac]Ac-Macropa-GC33, a pilot study was completed in a HepG2 subcutaneous flank model to test multiple administered activities of this radioconjugate. Treatment with 18.5 kBq of [ 225 Ac]Ac-Macropa-GC33 was found to cause tumor growth delay and was well tolerated. Therefore, we used 18.5 kBq as the highest amount of activity administered in our subsequent therapy study. We also tested 9.25 kBq in an attempt to mitigate toxicity while preserving therapeutic effect.
To evaluate the therapeutic efficacy of [ 225 Ac]Ac-Macropa-GC33, we recorded tumor volumes and assessed survival of all groups. PBS and unmodified GC33 controls approached 2000 mm 3 as early as 14 days post-injection, necessitating sacrifice. Compared to control cohorts treated with PBS and unmodified GC33, a modest tumor growth delay was observed for the [ 225 Ac]Ac-Macropa-IgG1 cohort. This observation signals therapeutic benefit of the radiolabeled non-specific control, likely due to the enhanced permeability and retention effect (EPR) [21]. Tumor reduction was also observed when comparing spider plots of tumor volumes for the 9.25 and 18.5 kBq [ 225 Ac]Ac-Macropa-GC33 cohorts to the [ 225 Ac]Ac-Macropa-IgG1 cohort. This dampening of tumor growth is evident when comparing the geometric means of the cohorts treated with either radioconjugate ( Figure 3A, solid lines). and retention effect (EPR) [21]. Tumor reduction was also observed when comparing spider plots of tumor volumes for the 9.25 and 18.5 kBq [ 225 Ac]Ac-Macropa-GC33 cohorts to the [ 225 Ac]Ac-Macropa-IgG1 cohort. This dampening of tumor growth is evident when comparing the geometric means of the cohorts treated with either radioconjugate ( Figure  3A, solid lines).
Mean tumor volumes for each group were plotted until the first recorded death in each group, resulting in an inability to assess long-term tumor reduction effect for Ac-225treated animals ( Figure 3B). Accordingly, to determine if tumor growth delay was significant for the groups treated with either radioconjugate, we performed a Chi-square analysis of the number of animals bearing tumor volumes exceeding 500 mm 3 (at least 2-fold larger than initial tumor volume) at 7 days p.i., the latest time point that all animals in each group were alive ( Figure S2B). The number of animals exceeding 500 mm 3 in tumor volume was significantly higher in PBS (p = 0.0010), unmodified GC33 (p < 0.0001), and 18.5 kBq [ 225 Ac]Ac-Macropa-IgG1 (p = 0.0253) when compared to 18.5 kBq [ 225 Ac]Ac-Macropa-GC33. There was also a greater number of animals reaching 500 mm 3 in tumor volume in the PBS cohort compared to the 9.25 kBq [ 225 Ac]Ac-Macropa-GC33 group (p = 0.0062). A Student's t-test was performed to compare the average tumor volumes between treatment groups at 7 days p.i. Animals that received 18.  Mean tumor volumes for each group were plotted until the first recorded death in each group, resulting in an inability to assess long-term tumor reduction effect for Ac-225-treated animals ( Figure 3B). Accordingly, to determine if tumor growth delay was significant for the groups treated with either radioconjugate, we performed a Chi-square analysis of the number of animals bearing tumor volumes exceeding 500 mm 3 (at least 2-fold larger than initial tumor volume) at 7 days p.i., the latest time point that all animals in each group were alive ( Figure S2B). The number of animals exceeding 500 mm 3 in tumor volume was significantly higher in PBS (p = 0.0010), unmodified GC33 (p < 0.0001), and 18.5 kBq [ 225 Ac]Ac-Macropa-IgG1 (p = 0.0253) when compared to 18.5 kBq [ 225 Ac]Ac-Macropa-GC33. There was also a greater number of animals reaching 500 mm 3   In addition to monitoring weight loss, a complete blood count was performed at regular intervals following treatment to assess hematologic toxicity. While animals in the PBS and unmodified GC33 groups maintained average weights well above 90% of their starting weight, groups receiving any radioconjugate exhibited significant weight loss early in the study. At 7 days p.i., the weights of the 18.5 kBq [ 225 Ac]Ac-Macropa-IgG1 and 18.5 kBq [ 225 Ac]Ac-Macropa-GC33 groups fell to 87.3 ± 1.3% and 83.0 ± 3.2% (mean ± SEM) of their starting body weight, respectively. Further, animals receiving 9.25 kBq [ 225 Ac]Ac-Macropa-GC33 experienced less dramatic weight loss, dropping to to 90.4 ± 1.4% of their starting weight at 7 days p.i. ( Figure S2A). However, animals that were treated with either radioconjugate never returned to their initial body weight, suggesting persistent toxicity for the duration of the study.
Acute bone marrow toxicity was observed for animals treated with any radioconjugate 7 days p.i. as evidenced by a notable drop in white blood cell (WBC) count ( Figure 4A). Interestingly, while the WBC count dropped below normal range for all groups receiving any radioconjugate, hemoglobin and platelet values remained within normal ranges in animals treated with 9.  . White blood cell count, hemoglobin, and platelets were measured as part of CBC analysis of whole blood samples and plotted over time (mean ± SEM). Counts for all parameters drop dramatically 7 days p.i. for groups treated with either radioconjugate, followed by eventual recovery of majority of parameters to within normal range denoted by dotted lines (n = 5). (B). Measurement of white blood cells, hemoglobin, and platelets from day 0 and day 14 p.i. were plotted on box-and-whisker plots and compared. Unlike other radioconjugate-treated groups, there was a less dramatic decrease in white blood cell count for the 9.25 kBq group hemoglobin and platelet counts remained in normal range for the duration of the study. Error bars represent the minimum and maximum values observed, the length of the box represents the range from the 25th to 75th percentile, and the horizontal line within the box represents the median. Dotted lines represent the normal range of values.

Liver and Kidney Function Are Stable
Basic Metabolic Panel (BMP) analysis of liver function enzymes bilirubin and alanine-aminotransferase showed minimal fluctuations from baseline up to 28 days p.i. (Figure 5A), suggesting minimal hepatotoxicity with any treatment. In addition, no remarkable changes to kidney function as measured by creatinine and blood urea nitrogen were observed, suggesting that nephrotoxicity was minimal.

Liver and Kidney Function Are Stable
Basic Metabolic Panel (BMP) analysis of liver function enzymes bilirubin and alanineaminotransferase showed minimal fluctuations from baseline up to 28 days p.i. (Figure 5A), suggesting minimal hepatotoxicity with any treatment. In addition, no remarkable changes to kidney function as measured by creatinine and blood urea nitrogen were observed, suggesting that nephrotoxicity was minimal.

Discussion
The GPC3-specific humanized monoclonal IgG1 GC33 (a.k.a. codrituzumab) has previously demonstrated excellent preclinical and clinical localization to HCC-expressing GPC3 in vivo [5,13]. However, in a phase II trial of patients with advanced HCC, when compared to placebo control-treated patients, GC33 did not change progression-free or overall survival [15]. We hypothesize that the putative mechanism of cell kill for GC33, antibody-dependent cellular cytotoxicity, was insufficient to exert a therapeutic benefit in these patients. Given that GC33 has been previously shown to localize to GPC3-expressing HCC and the limited treatment options available, we aimed to label GC33 with a cytotoxic alpha particle-emitting isotope to evaluate its therapeutic efficacy in an animal model of HCC [5,13]. Here, we functionalized GC33 with the Macropa chelate, and labeled the conjugate with Ac-225 to yield [ 225 Ac]Ac-Macropa-GC33. Importantly, we selected the Macropa chelate because of its ability to label Ac-225 at room temperature, as opposed to at 60-80 °C as is required for DOTA, which can prevent antibody denaturation during the radiolabeling process ( Figure 1A) [20]. We demonstrate that [ 225 Ac]Ac-Macropa-GC33 has a modest anti-tumor effect as measured by tumor growth and survival in HepG2 murine xenograft models, though caution is warranted because of significant treatment-related toxicity.
In this study, we tested two administered activities of [ 225 Ac]Ac-Macropa-GC33: 18.5 kBq, which was determined to be effective and well tolerated in a pilot study, and 9.25 kBq, a lower amount that was tested to assess whether we might achieve comparable antitumor benefit while mitigating potential radiotoxicity. Mice treated with PBS, mass-equivalent unmodified GC33, and 18.5 kBq [ 225 Ac]Ac-IgG1 served as controls. While we ob-

Discussion
The GPC3-specific humanized monoclonal IgG1 GC33 (a.k.a. codrituzumab) has previously demonstrated excellent preclinical and clinical localization to HCC-expressing GPC3 in vivo [5,13]. However, in a phase II trial of patients with advanced HCC, when compared to placebo control-treated patients, GC33 did not change progression-free or overall survival [15]. We hypothesize that the putative mechanism of cell kill for GC33, antibodydependent cellular cytotoxicity, was insufficient to exert a therapeutic benefit in these patients. Given that GC33 has been previously shown to localize to GPC3-expressing HCC and the limited treatment options available, we aimed to label GC33 with a cytotoxic alpha particle-emitting isotope to evaluate its therapeutic efficacy in an animal model of HCC [5,13]. Here, we functionalized GC33 with the Macropa chelate, and labeled the conjugate with Ac-225 to yield [ 225 Ac]Ac-Macropa-GC33. Importantly, we selected the Macropa chelate because of its ability to label Ac-225 at room temperature, as opposed to at 60-80 • C as is required for DOTA, which can prevent antibody denaturation during the radiolabeling process ( Figure 1A) [20]. We demonstrate that [ 225 Ac]Ac-Macropa-GC33 has a modest anti-tumor effect as measured by tumor growth and survival in HepG2 murine xenograft models, though caution is warranted because of significant treatment-related toxicity.
In this study, we tested two administered activities of [ 225 Ac]Ac-Macropa-GC33: 18.5 kBq, which was determined to be effective and well tolerated in a pilot study, and 9.25 kBq, a lower amount that was tested to assess whether we might achieve comparable anti-tumor benefit while mitigating potential radiotoxicity. Mice treated with PBS, massequivalent unmodified GC33, and 18.5 kBq [ 225 Ac]Ac-IgG1 served as controls. While we observed tumor growth delay in all 225 Ac conjugates, including the IgG1 isotype control, results were more dramatic for animals treated with GC33-derived radioconjugates. This enhanced tumor growth delay in animals treated with either amount of [ 225 Ac]Ac-Macropa-GC33 compared to [ 225 Ac]Ac-Macropa-IgG1 is supported by biodistribution data confirming the ability of [ 225 Ac]Ac-Macropa-GC33 to preferentially target GPC3 in vivo in the HepG2 tumor model. In fact, the tumor uptake was nearly identical to that of a GC33-derived immunoPET agent (~15% IA/g) [13]. Moreover, 9.25 kBq [ 225 Ac]Ac-Macropa-GC33 was better tolerated than 18.5 kBq and showed significant improvement in overall survival compared to PBS and unmodified GC33 controls. The fact that we observed substantial toxicity in the 18.5 kBq cohort in this study suggests that this amount of administered activity may be at the steep portion of the toxicity-activity curve.
Bone marrow suppression was observed within days of treatment for groups given any radioconjugate. These findings likely reflect the long blood half-life of full-length antibodies that, when coupled to alpha particle-emitting radionuclides, can cause nonselective irradiation of the blood pool and hematopoietic progenitors, phenomena that mirrors what is observed in clinical studies using these agents [16]. Furthermore, it is possible that the Fc receptors (FcR) present on macrophages and B-cells, which can bind the Fc of full-length antibodies such as GC33, may have also contributed to a decrease in white blood cell counts, as has been previously described [22]. Given the off-tumor accumulation of our radioconjugate in the liver and spleen, which bear resident macrophages, our findings are consistent FcR-associated toxicity as well.
Comparing the complete blood counts of animals receiving 9.25 and 18.5 kBq of [ 225 Ac]Ac-Macropa-GC33 provides valuable information for mitigating potential radiotoxicity in future studies. Notably, the hemoglobin and platelet counts remained in the normal range for the 9.25 kBq [ 225 Ac]Ac-Macropa-GC33 treatment cohort throughout the course of our study, unlike cohorts receiving 18.5 kBq. While these observations include an element of survivor bias, the shorter duration of marrow suppression in animals receiving a lower injected activity of radioconjugate may very well have allowed this group to achieve the modest survival benefit not observed in other groups. Importantly, liver and kidney function appear to have been minimally impacted throughout the treatment course.
There are several limitations to our current study. Given the toxicity observed, optimization of the amount of activity and the administration of our conjugate are needed. Several strategies to address these issues are available, including fractionation [23,24] and pretargeting, which decouples the long blood half-life targeting agent from the cytotoxic cargo [25]. The use of smaller targeting ligands (e.g., antibody fragments, peptides or small molecules), which may have more favorable pharmacokinetics than full-length antibodies, can also be considered. In addition, while the amount of administered activities used here were empirically determined in a pilot study using the same preclinical model, our results clearly indicate that further tuning is required given the toxicity observed. Such dose optimization is critical in this model prior to exploring outcomes in more complex systems (e.g., orthotopic) and for clinical translation. Moreover, we anticipate that employing models to estimate the absorbed dose of TAT agents to tumors and normal tissues may predict therapeutic efficacy and toxicity, which is currently an area of active investigation [26].

Cell Culture
HepG2, a human hepatoblastoma cell line (GPC3 + ), and A431, an epithelioid cancer line, were purchased from ATCC (Manassas, VA, USA) and cultured according to vendor's instructions. A431-GPC3 + was engineered to stably overexpress GPC3 as previously described [27]. Cells were maintained with DMEM media (Life Technologies, Carlsbad, CA, USA) supplemented with 10% FetalPlex (Gemini Bio-Products, Sacramento, CA, USA). Cells tested negative for mycoplasma and were kept at 37 • C and 5% CO 2 in a humified incubator. Cells were implanted within 15 passages.

Mass Spectrometry Analysis of GC33-Macropa and IgG1-Macropa Conjugates
Mass spectrometry via an Exactive Plus Extended Mass Range benchtop Orbitrap system with a heated electrospray ionization source (HESI) was used to determine the number of DFO chelators coupled per antibody as previously described [27]. GC33-Macropa conjugates were compared with unmodified GC33 via mass spectrometry to determine the number of moieties covalently bound to GC33. The number of Macropa moieties per GC33 was estimated using the molecular weights of Macropa (590.23 g/mol) and unmodified GC33. Conjugates and unmodified proteins were analyzed on an Exactive Plus EMR Orbitrap system (Thermo Scientific, Waltham, MA, USA). An amount of 1 µg of protein was loaded onto a 4 µM bead size MAbPac RP column (3 × 50 mm, Thermo Scientific) utilizing a Vanquish UHPLC (Thermo Scientific) connected to an Exactive Plus EMR mass spectrometer. The proteins were eluted with a 2 to 100% gradient of 80% acetonitrile with 0.1% formic acid over 8 min with a flow rate of 0.5 mL/min. The EMR was operated with three sequential orbitrap scan methods that each acquired MS1 at 8750 resolution with a maximum injection time of 100 ms and an AGC target of 3e6. Scans looked at the range of 600-10,000 m/z. The first segment included no additional solvation energy, and had 3 summed microscans, the second segment used 30 eV additional solvation energy and 5 microscans and the third microscan with 50 eV additional solvation energy used 10 microscans. All data were processed in BioPharma Finder 2.0 using the respect algorithm with a maximum mass deviation of 20 ppm with sliding windows [23]. desalting column with PBS as the eluate. Radiochemical yield and purity were determined by radio-iTLC with silica gel-impregnated glass-microfiber paper strips (iTLC-SG, Varian, Lake Forest, CA, USA) using an aqueous solution of EDTA (10 mM, pH 5.5) as the mobile phase. Data were analyzed using an AR-2000 (Eckert-Ziegler, Wilmington, MA, USA) to calculate the percent of total activity at the origin. To assess the stability of the purified [ 225 Ac]Ac-Macropa-GC33 radioconjugate, the compound was incubated at 37 • C in PBS or human serum. At several time points following incubation, 1 µL aliquots were taken and run on radio-iTLC with silica gel-impregnated glass-microfiber paper strips, using as a mobile phase an aqueous solution of EDTA (10 mM, pH 5.5), which was then analyzed using an AR-2000. Percent of total activity at the origin versus total activity was used to determine the intact radioconjugate.

Immunofluorescence
Tumors derived from A431-GPC3 + , HepG2, and A431 subcutaneously engrafted in female athymic nude mice were excised, rinsed with PBS and submerged in molds containing TissueTek OCT medium (Sakura Fineteck Inc, Torrence, CA, USA) and frozen on dry ice. Consecutive 10 µm sections were cut using a cryostat (Leica Biosystems, Buffalo Grove, IL, USA) from prepared frozen tissue blocks and placed on slides (Azer Scientific Inc., Morgantown, PA, USA). Tissue sections were incubated at room temperature to dry for 30 min and fixed in 100% methanol at −20 • C for 10 min, then left to air dry for 10 min, and rinsed in PBS. Blocking buffer was composed of 0.5% IgG-free BSA (Jackson Labs, Bar Harbor, ME, USA), 2.2% glycine (Sigma-Aldrich, St. Louis, MO, USA), 0.1% Tween-20 (Sigma-Aldrich), 5% donkey serum (Jackson Labs, Bar Harbor, ME, USA) solution in PBS. Slides were blocked with blocking buffer for 30 min at room temperature before being incubated with 10 µg/mL of GC33-AF488 antibody (diluted in blocking buffer) for 1 h at room temperature. Slides were subsequently rinsed in PBS and counterstained with 1:5000 dilution of 4',6-diamidino-2-phenylindole dihydrochloride (DAPI, Invitrogen, Eugene, OR, USA) in deionized water for 3-5 min at room temperature. Slides were rinsed in deionized water, mounted with Fluoro-gel water-soluble mounting medium (Electron Microscopy Sciences, Hatfield, PA, USA), and cover slips (Thermo Fisher, Portsmouth, NH, USA) were placed. Stained slides were stored in the dark at 4 • C. Fluorescent microscopy images were acquired using a Nikon Ti2 widefield microscope.

Hematological Data Collection and Analysis
Whole blood samples were collected for mice in the TAT study in a maximum 100 µL volume retro-orbitally once weekly (n = 5, randomly selected). Blood was separately collected into ethylenediaminetetraacetic acid (EDTA)-lined and lithium heparin-lined collection tubes (Sarstedt, Inc., Newton, NC, USA) and, if not processed immediately after collection, kept overnight in 4 • C and processed once acclimated to room temperature. Blood samples were diluted to meet minimal volume sample requirements for blood analyzer systems. Complete Blood Count (CBC) blood samples collected in EDTA-lined tubes were diluted 1:5 in 1X PBS and analyzed using the Vetscan HM5 (Abaxis, Union City, CA, USA), which was calibrated to accommodate a 1:5 sample dilution using a synthetic whole blood standard (Abaxis, Union City, CA, USA). Basic Metabolic Panel (BMP) blood samples in heparin-lined tubes were diluted 1:3 in 1X PBS, and analyzed using the Vetscan VS2 (Abaxis, Union City, CA, USA) with Comprehensive Diagnostic rotors.

Biodistribution Study
The biodistribution of 9.25 kBq of [ 225 Ac]Ac-GC33 and [ 225 Ac]Ac-IgG1 was determined at 48 and 144 h post-injection using HepG2 subcutaneous (right flank) models (female athymic nude mice). Mice with tumor volume below 1500 mm 3 were selected and divided into GC33 and IgG1 groups for each time point (n = 3, except the [ 225 Ac]Ac-IgG1 144 h time point where n = 2). At 48 and 144 h post-injection, mice were euthanized via CO 2 asphyxiation. Twelve tissues, including the tumor, were collected. Each sample was weighed and then measured in a gamma counter (2480 Wizard 3 , Perkin Elmer Inc, Shelton, CT, USA) calibrated for an open window (20-2000 kEv). The counts from each sample were decay-and background-corrected, and counts were converted into activity using a calibration curve generated from Ac-225 standards of known activity. The percent injected dose per gram (%IA/g) was estimated by normalizing data to the total activity injected into the corresponding animal and accounting for the mass of each organ. In addition, tumor:blood, tumor:liver, and tumor:muscle ratios were calculated using the mean %IA/g per group.

Statistical Analysis
Statistical analysis was performed using GraphPad Prism (San Diego, CA, USA). Log-rank Mantel-Cox test was performed to compare survival curves. Tumor volume data were analyzed using a Chi-square test to find statistical significance between the number of animals exceeding 500 mm 3 in tumor volume. Whole blood and biodistribution analysis were performed using the Student's t-test for unpaired data to identify statistical significance between the mean values of two groups. Statistical significance was defined by p < 0.05 (*), 0.01 (**), 0.001 (***), and 0.0001 (****).

Conclusions
In this radioimmunotherapy study, we demonstrate that targeted alpha particle therapy using the radioconjugate [ 225 Ac]Ac-Macropa-GC33 is capable of producing both a modest anti-tumor effect and a clear survival benefit in preclinical models of HCC. However, hematological toxicity remains a significant problem that needs to be addressed. Figure S1: Mass spectrometry results show that GC33 and IgG1 were successfully chelated with Macropa prior to radiolabeling. (A). Mass spectrogram demonstrates multiple peaks with molecular weights ranging from 148,182.90 to 155,255.60 g/mol, representing conjugates with 7-12 Macropa per GC33 (148,184 g/mol). Radio instant thin-layer chromatography tracings show excellent radiochemical yields and purity following size-exclusion chromatography of (B), [ 225 Ac]Ac-Macropa-GC33 and (C), [ 225 Ac]Ac-Macropa-IgG1. Figure S2: Animals receiving any radioconjugate experienced weight loss following treatment, while tumor burden analysis at day 7 p.i. suggests preferential tumor response. (A). Decreases in weight were noted in all groups receiving any radioconjugate. Dashed line denotes starting weight. (B). Mean tumor volumes as of day 7 p.i. demonstrate that there is a notable treatment response in all animals receiving either